From: Martin Quinson <martin.quinson@ens-rennes.fr>
Date: Thu, 15 Jun 2017 08:43:11 +0000 (+0200)
Subject: Merge pull request #181 from bcamus/master
X-Git-Tag: v3.16~86
X-Git-Url: http://info.iut-bm.univ-fcomte.fr/pub/gitweb/simgrid.git/commitdiff_plain/befbbbe1fbb31663a8f91e24ce12df271cf4ae79?hp=ae68749ec3d90fc5e403e8a4950d30b832a9629f

Merge pull request #181 from bcamus/master

Add tests for multi-core VM
---

diff --git a/.cproject b/.cproject
index 69f3f2f38d..78e5091149 100644
--- a/.cproject
+++ b/.cproject
@@ -22,7 +22,7 @@
 							<tool id="cdt.managedbuild.tool.gnu.archiver.base.374652938" name="GCC Archiver" superClass="cdt.managedbuild.tool.gnu.archiver.base"/>
 							<tool id="cdt.managedbuild.tool.gnu.cpp.compiler.base.738159103" name="GCC C++ Compiler" superClass="cdt.managedbuild.tool.gnu.cpp.compiler.base">
 								<option id="gnu.cpp.compiler.option.include.paths.216814103" name="Include paths (-I)" superClass="gnu.cpp.compiler.option.include.paths" valueType="includePath">
-									<listOptionValue builtIn="false" value="/usr/include/ns3.22"/>
+									<listOptionValue builtIn="false" value="/usr/include/ns3.26"/>
 									<listOptionValue builtIn="false" value="/opt/gtnets/include"/>
 									<listOptionValue builtIn="false" value="/usr/include/lua5.3"/>
 									<listOptionValue builtIn="false" value="/usr/lib/jvm/java-8-openjdk-amd64/include"/>
@@ -32,7 +32,7 @@
 							</tool>
 							<tool id="cdt.managedbuild.tool.gnu.c.compiler.base.1053916774" name="GCC C Compiler" superClass="cdt.managedbuild.tool.gnu.c.compiler.base">
 								<option id="gnu.c.compiler.option.include.paths.100642505" name="Include paths (-I)" superClass="gnu.c.compiler.option.include.paths" valueType="includePath">
-									<listOptionValue builtIn="false" value="/usr/include/ns3.22"/>
+									<listOptionValue builtIn="false" value="/usr/include/ns3.26"/>
 									<listOptionValue builtIn="false" value="/opt/gtnets/include"/>
 									<listOptionValue builtIn="false" value="/usr/include/lua5.3"/>
 									<listOptionValue builtIn="false" value="/usr/lib/jvm/java-8-openjdk-amd64/include"/>
diff --git a/.gitignore b/.gitignore
index f1b3b52d48..60839f9d05 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1051,6 +1051,7 @@ teshsuite/smpi/macro-partial-shared-communication/macro-partial-shared-communica
 teshsuite/smpi/type-struct/type-struct
 teshsuite/smpi/type-vector/type-vector
 teshsuite/s4u/actor/actor
+teshsuite/s4u/comm-pt2pt/comm-pt2pt
 teshsuite/s4u/concurrent_rw/concurrent_rw
 teshsuite/s4u/host_on_off_wait/host_on_off_wait
 teshsuite/s4u/listen_async/listen_async
diff --git a/ChangeLog b/ChangeLog
index 8562356777..f959057234 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -17,26 +17,41 @@ SimGrid (3.16) UNRELEASED
    - Remove attribute 'content_type' of <storage_type>: was never used
    - Make attribute 'model' of <storage_type> optional: for future usage
    - Remove Bconnection model property: was never (in)validated. Replaced by
-      the maximum of the read and write bandwidth as a resource constraint. 
+      the maximum of the read and write bandwidth as a resource constraint.
 
  SimDag
-  - New and Backwards Compatibility break: 
+  - New and Backwards Compatibility break:
     SD_simulate_with_update (double how_long, xbt_dynar_t changed_tasks_dynar)
     When one wants to get the list of tasks whose states have changed during a
-    simulation round, s/he has to allocate and free a dynar and use it as argument 
-    to this function. The former SD_simulate (double how_long) now returns void.
+    simulation round, s/he has to allocate and free a dynar and use it as
+    argument to this function. The former SD_simulate (double how_long)
+    now	returns void.
 
  Virtual Machines
   - Allow multicore VMs, along with the correct sharing computations
 
+ Energy
+  - New function to update the consumption of all hosts at once.
+
  MSG
-  - The netzone are now available from the MSG API. 
+  - The netzone are now available from the MSG API.
     The old names still work, but are now deprecated.
 
  SMPI
   - New algorithm to privatize globals: dlopen, with dynamic loading tricks
   - New option: smpi/keep-temps to not cleanup temp files
+  - New option : smpi/shared-malloc-blocksize . Relevant only when global shared 
+    mallocs mode is used, allows to change the size of the fake file used
+    (default 1MB), to potentially limit the number of mappings for large runs.
   - Support for sparse privatized malloc with SMPI_PARTIAL_SHARED_MALLOC()
+  - Fortran ifort and flang compilers support
+  - New RMA calls supported (experimental) :
+      - MPI_Win_allocate, MPI_Win_create_dynamic, MPI_Win_attach
+      - MPI_Win_detach, MPI_Win_set_info, MPI_Win_get_info
+      - MPI_Win_lock_all, MPI_Win_unlock_all, MPI_Win_flush
+      - MPI_Win_flush_local, MPI_Win_flush_all, MPI_Win_flush_local_all
+      - MPI_Op_commutative, MPI_Fetch_and_Op, MPI_Compare_and_swap
+      - MPI_Rput, MPI_Rget, MPI_Raccumulate, MPI_Rget_accumulate
 
  XBT
   - Replay: New function xbt_replay_action_get():
@@ -44,6 +59,7 @@ SimGrid (3.16) UNRELEASED
   - DROPPED FUNCTION: xbt_str_varsubst()
   - DROPPED MODULE: strbuff. We don't need it anymore.
   - DROPPED MODULE: matrix. We don't need it anymore.
+  - DROPPED MODULE: lib. We don't need it anymore.
 
  -- Release target: June 21 2017   -- Da SimGrid team <simgrid-devel@lists.gforge.inria.fr>
 
diff --git a/contrib/benchmarking_code_block/inject.h b/contrib/benchmarking_code_block/inject.h
index 57e2c8291b..c171b0925c 100644
--- a/contrib/benchmarking_code_block/inject.h
+++ b/contrib/benchmarking_code_block/inject.h
@@ -30,12 +30,12 @@
  * Histogram entry for each measured block
  * Each entry is guarded inside xbt dictionary which is read from the file */
 typedef struct xbt_hist {
-	int n;
-	int counts;
-	double mean;
-	double *breaks;
-	double *percentage;
-	char* block_id;
+  int n;
+  int counts;
+  double mean;
+  double* breaks;
+  double* percentage;
+  char* block_id;
 } xbt_hist_t;
 
 extern RngStream get_randgen(void);
@@ -54,159 +54,144 @@ static inline double xbt_hist_time(char *key);
 /* Initializing xbt dictionary for SMPI version, reading xbt_hist_t entries line by line */
 static inline void xbt_inject_init(char *inputfile)
 {
-	xbt_dict_t mydict = get_dict();
-	FILE* fpInput = fopen(inputfile, "r");
-	if (fpInput == NULL)
-		printf("Error while opening the inputfile");
-	fseek(fpInput, 0, 0);
-
-	char line[200];
-	char *key;
-	int i;
-	xbt_hist_t* data;
-
-	if (fgets(line, 200, fpInput) == NULL)
-		printf("Error input file is empty!");//Skipping first row
-	while (fgets(line, 200, fpInput) != NULL)
-	{
-		key = strtok(line, "\t");
-
-		data = xbt_dict_get_or_null(mydict, key);
-		if (data)
-			printf("Error, data with that block_id already exists!");
-
-		data = (xbt_hist_t *) xbt_new(xbt_hist_t, 1);
-
-		data->block_id = key;
-		data->counts = atoi(strtok(NULL, "\t"));
-		data->mean = atof(strtok(NULL, "\t"));
-		data->n = atoi(strtok(NULL, "\t"));
-
-		data->breaks = (double*) malloc(sizeof(double) * data->n);
-		data->percentage = (double*) malloc(sizeof(double) * (data->n - 1));
-		for (i = 0; i < data->n; i++)
-			data->breaks[i] = atof(strtok(NULL, "\t"));
-		for (i = 0; i < (data->n - 1); i++)
-			data->percentage[i] = atof(strtok(NULL, "\t"));
-
-		xbt_dict_set(mydict, key, data, NULL);
-	}
+  xbt_dict_t mydict = get_dict();
+  FILE* fpInput     = fopen(inputfile, "r");
+  if (fpInput == NULL)
+    printf("Error while opening the inputfile");
+  fseek(fpInput, 0, 0);
+
+  char line[200];
+  char* key;
+
+  if (fgets(line, 200, fpInput) == NULL)
+    printf("Error input file is empty!"); // Skipping first row
+  while (fgets(line, 200, fpInput) != NULL) {
+    key = strtok(line, "\t");
+
+    xbt_hist_t* data = xbt_dict_get_or_null(mydict, key);
+    if (data)
+      printf("Error, data with that block_id already exists!");
+
+    data = (xbt_hist_t*)xbt_new(xbt_hist_t, 1);
+
+    data->block_id = key;
+    data->counts   = atoi(strtok(NULL, "\t"));
+    data->mean     = atof(strtok(NULL, "\t"));
+    data->n        = atoi(strtok(NULL, "\t"));
+
+    data->breaks     = (double*)malloc(sizeof(double) * data->n);
+    data->percentage = (double*)malloc(sizeof(double) * (data->n - 1));
+    for (int i        = 0; i < data->n; i++)
+      data->breaks[i] = atof(strtok(NULL, "\t"));
+    for (int i            = 0; i < (data->n - 1); i++)
+      data->percentage[i] = atof(strtok(NULL, "\t"));
+
+    xbt_dict_set(mydict, key, data, NULL);
+  }
+  fclose(fInput);
 }
 
 /* Initializing xbt dictionary for StarPU version, reading xbt_hist_t entries line by line */
 static inline void inject_init_starpu(char *inputfile, xbt_dict_t *dict, RngStream *rng)
 {
-	*dict = xbt_dict_new_homogeneous(free);
-	*rng = RngStream_CreateStream("Randgen1");
-	unsigned long seed[] = {134, 233445, 865, 2634, 424242, 876541};
-	RngStream_SetSeed(*rng, seed);
-
-	xbt_dict_t mydict = *dict;
-	mydict = *dict;
-	FILE* fpInput = fopen(inputfile, "r");
-	if (fpInput == NULL)
-	{
-		printf("Error while opening the inputfile");
-		return;
-	}
-
-	fseek(fpInput, 0, 0);
-
-	char line[MAX_LINE_INJ];
-	char *key;
-	int i;
-	xbt_hist_t* data;
-
-	if (fgets(line, MAX_LINE_INJ, fpInput) == NULL)
-	{
-		printf("Error input file is empty!");//Skipping first row
-		return;
-	}
-
-
-	while (fgets(line, MAX_LINE_INJ, fpInput) != NULL)
-	{
-		key = strtok(line, "\t");
-
-		data = xbt_dict_get_or_null(mydict, key);
-		if (data)
-			printf("Error, data with that block_id already exists!");
-
-		data = (xbt_hist_t *) xbt_new(xbt_hist_t, 1);
-		data->block_id = key;
-		data->counts = atoi(strtok(NULL, "\t"));
-		data->mean = atof(strtok(NULL, "\t"));
-		data->n = atoi(strtok(NULL, "\t"));
-		data->breaks = (double*) malloc(sizeof(double) * data->n);
-		data->percentage = (double*) malloc(sizeof(double) * (data->n - 1));
-
-		for (i = 0; i < data->n; i++)
-			data->breaks[i] = atof(strtok(NULL, "\t"));
-			for (i = 0; i < (data->n - 1); i++)
-			{
-				data->percentage[i] = atof(strtok(NULL, "\t"));
-			}
-
-		xbt_dict_set(mydict, key, data, NULL);
-	}
+  *dict                = xbt_dict_new_homogeneous(free);
+  *rng                 = RngStream_CreateStream("Randgen1");
+  unsigned long seed[] = {134, 233445, 865, 2634, 424242, 876541};
+  RngStream_SetSeed(*rng, seed);
+
+  xbt_dict_t mydict = *dict;
+  FILE* fpInput     = fopen(inputfile, "r");
+  if (fpInput == NULL) {
+    printf("Error while opening the inputfile");
+    return;
+  }
+
+  fseek(fpInput, 0, 0);
+
+  char line[MAX_LINE_INJ];
+  char* key;
+
+  if (fgets(line, MAX_LINE_INJ, fpInput) == NULL) {
+    printf("Error input file is empty!"); // Skipping first row
+    return;
+  }
+
+  while (fgets(line, MAX_LINE_INJ, fpInput) != NULL) {
+    key = strtok(line, "\t");
+
+    xbt_hist_t* data = xbt_dict_get_or_null(mydict, key);
+    if (data)
+      printf("Error, data with that block_id already exists!");
+
+    data             = (xbt_hist_t*)xbt_new(xbt_hist_t, 1);
+    data->block_id   = key;
+    data->counts     = atoi(strtok(NULL, "\t"));
+    data->mean       = atof(strtok(NULL, "\t"));
+    data->n          = atoi(strtok(NULL, "\t"));
+    data->breaks     = (double*)malloc(sizeof(double) * data->n);
+    data->percentage = (double*)malloc(sizeof(double) * (data->n - 1));
+
+    for (int i        = 0; i < data->n; i++)
+      data->breaks[i] = atof(strtok(NULL, "\t"));
+    for (int i = 0; i < (data->n - 1); i++) {
+      data->percentage[i] = atof(strtok(NULL, "\t"));
+    }
+
+    xbt_dict_set(mydict, key, data, NULL);
+  }
+  fclose(fInput);
 }
 
 /* Injecting time */
 static inline double xbt_inject_time(char *key)
 {
-	return xbt_hist_time(key);
-	//return xbt_mean_time(key);
+  return xbt_hist_time(key);
+  // return xbt_mean_time(key);
 }
 
 /* Injecting mean value */
 static inline double xbt_mean_time(char *key)
 {
-	xbt_dict_t mydict = get_dict();
-	xbt_hist_t* data = xbt_dict_get_or_null(mydict, key);
+  xbt_dict_t mydict = get_dict();
+  xbt_hist_t* data  = xbt_dict_get_or_null(mydict, key);
 
-	if (!data)
-	{
-	  printf("Warning: element with specified key does not exist (%s)\n",key);
-	  return 0;
-	}
+  if (!data) {
+    printf("Warning: element with specified key does not exist (%s)\n", key);
+    return 0;
+  }
 
-	return data->mean;
+  return data->mean;
 }
 
 /* Injecting random value from the histogram */
 static inline double xbt_hist_time(char *key)
 {
-	int i, k = 0;
-	double left = 0, right = 1;
-	double timer = 0;
-	RngStream rng_stream;
-	double r, r2;
-
-	xbt_dict_t mydict = get_dict();
-	xbt_hist_t* data = xbt_dict_get_or_null(mydict, key);
-
-	if (!data)
-		{
-		  printf("Warning: element with specified key does not exist (%s)\n",key);
-		  return 0;
-		}
-
-	/* Choosing random interval of the histogram */
-	rng_stream = get_randgen();
-	r = RngStream_RandU01(rng_stream);
-	for (i = 0; i < (data->n - 1); i++)
-	{
-		left += (i == 0) ? 0 : data->percentage[i - 1];
-		right += data->percentage[i];
-		if (left < r && r <= right)
-			k = i;
-	}
-
-	/* Choosing random value inside the interval of the histogram */
-	r2 = RngStream_RandU01(rng_stream);
-	timer = data->breaks[k] + r2 * (data->breaks[k + 1] - data->breaks[k]);
-
-	return timer;
+  xbt_dict_t mydict = get_dict();
+  xbt_hist_t* data  = xbt_dict_get_or_null(mydict, key);
+
+  if (!data) {
+    printf("Warning: element with specified key does not exist (%s)\n", key);
+    return 0;
+  }
+
+  /* Choosing random interval of the histogram */
+  RngStream rng_stream = get_randgen();
+  double r             = RngStream_RandU01(rng_stream);
+  int k                = 0;
+  double left          = 0;
+  double right         = 1;
+  for (int i = 0; i < (data->n - 1); i++) {
+    left += (i == 0) ? 0 : data->percentage[i - 1];
+    right += data->percentage[i];
+    if (left < r && r <= right)
+      k = i;
+  }
+
+  /* Choosing random value inside the interval of the histogram */
+  double r2    = RngStream_RandU01(rng_stream);
+  double timer = data->breaks[k] + r2 * (data->breaks[k + 1] - data->breaks[k]);
+
+  return timer;
 }
 
 #endif // __INJECT_H__
diff --git a/doc/doxygen/inside.doc b/doc/doxygen/inside.doc
index d39e3d2784..03e3270ceb 100644
--- a/doc/doxygen/inside.doc
+++ b/doc/doxygen/inside.doc
@@ -36,4 +36,7 @@ git will say so, and provide a ready to use patch that you can apply
 to improve your commit. Just carefully read the error message you get
 to find the exact command with git-apply to fix your formating.
 
+If you find that for a specific commit, the formatter does a very bad
+job, then add --no-verify to your git commit command line.
+
 */
diff --git a/doc/doxygen/options.doc b/doc/doxygen/options.doc
index 6a94beb730..4739cf5def 100644
--- a/doc/doxygen/options.doc
+++ b/doc/doxygen/options.doc
@@ -52,18 +52,116 @@ SimGrid.
 @code
 #include <xbt/config.h>
 
-extern xbt_cfg_t _sg_cfg_set;
-
 int main(int argc, char *argv[]) {
      SD_init(&argc, argv);
 
      /* Prefer MSG_config() if you use MSG!! */
-     xbt_cfg_set_parse(_sg_cfg_set,"Item:Value");
+     xbt_cfg_set_parse("Item:Value");
 
      // Rest of your code
 }
 @endcode
 
+\section options_index Index of all existing configuration options
+
+\note
+  The full list can be retrieved by passing "--help" and
+     "--help-cfg" to an executable that uses SimGrid.
+
+- \c clean-atexit: \ref options_generic_clean_atexit
+
+- \c contexts/factory: \ref options_virt_factory
+- \c contexts/guard-size: \ref options_virt_guard_size
+- \c contexts/nthreads: \ref options_virt_parallel
+- \c contexts/parallel_threshold: \ref options_virt_parallel
+- \c contexts/stack-size: \ref options_virt_stacksize
+- \c contexts/synchro: \ref options_virt_parallel
+
+- \c cpu/maxmin-selective-update: \ref options_model_optim
+- \c cpu/model: \ref options_model_select
+- \c cpu/optim: \ref options_model_optim
+
+- \c exception/cutpath: \ref options_exception_cutpath
+
+- \c host/model: \ref options_model_select
+
+- \c maxmin/precision: \ref options_model_precision
+- \c maxmin/concurrency-limit: \ref options_concurrency_limit
+
+- \c msg/debug-multiple-use: \ref options_msg_debug_multiple_use
+
+- \c model-check: \ref options_modelchecking
+- \c model-check/checkpoint: \ref options_modelchecking_steps
+- \c model-check/communications-determinism: \ref options_modelchecking_comm_determinism
+- \c model-check/dot-output: \ref options_modelchecking_dot_output
+- \c model-check/hash: \ref options_modelchecking_hash
+- \c model-check/property: \ref options_modelchecking_liveness
+- \c model-check/max-depth: \ref options_modelchecking_max_depth
+- \c model-check/record: \ref options_modelchecking_recordreplay
+- \c model-check/reduction: \ref options_modelchecking_reduction
+- \c model-check/replay: \ref options_modelchecking_recordreplay
+- \c model-check/send-determinism: \ref options_modelchecking_comm_determinism
+- \c model-check/sparse-checkpoint: \ref options_modelchecking_sparse_checkpoint
+- \c model-check/termination: \ref options_modelchecking_termination
+- \c model-check/timeout: \ref options_modelchecking_timeout
+- \c model-check/visited: \ref options_modelchecking_visited
+
+- \c network/bandwidth-factor: \ref options_model_network_coefs
+- \c network/crosstraffic: \ref options_model_network_crosstraffic
+- \c network/latency-factor: \ref options_model_network_coefs
+- \c network/maxmin-selective-update: \ref options_model_optim
+- \c network/model: \ref options_model_select
+- \c network/optim: \ref options_model_optim
+- \c network/sender_gap: \ref options_model_network_sendergap
+- \c network/TCP-gamma: \ref options_model_network_gamma
+- \c network/weight-S: \ref options_model_network_coefs
+
+- \c ns3/TcpModel: \ref options_pls
+- \c path: \ref options_generic_path
+- \c plugin: \ref options_generic_plugin
+
+- \c storage/max_file_descriptors: \ref option_model_storage_maxfd
+
+- \c surf/precision: \ref options_model_precision
+
+- \c <b>For collective operations of SMPI, please refer to Section \ref options_index_smpi_coll</b>
+- \c smpi/async-small-thresh: \ref options_model_network_asyncsend
+- \c smpi/bw-factor: \ref options_model_smpi_bw_factor
+- \c smpi/coll-selector: \ref options_model_smpi_collectives
+- \c smpi/comp-adjustment-file: \ref options_model_smpi_adj_file
+- \c smpi/cpu-threshold: \ref options_smpi_bench
+- \c smpi/display-timing: \ref options_smpi_timing
+- \c smpi/grow-injected-times: \ref options_model_smpi_test
+- \c smpi/host-speed: \ref options_smpi_bench
+- \c smpi/IB-penalty-factors: \ref options_model_network_coefs
+- \c smpi/iprobe: \ref options_model_smpi_iprobe
+- \c smpi/iprobe-cpu-usage: \ref options_model_smpi_iprobe_cpu_usage
+- \c smpi/init: \ref options_model_smpi_init
+- \c smpi/keep-temps: \ref options_smpi_temps
+- \c smpi/lat-factor: \ref options_model_smpi_lat_factor
+- \c smpi/ois: \ref options_model_smpi_ois
+- \c smpi/or: \ref options_model_smpi_or
+- \c smpi/os: \ref options_model_smpi_os
+- \c smpi/papi-events: \ref options_smpi_papi_events
+- \c smpi/privatization: \ref options_smpi_privatization
+- \c smpi/send-is-detached-thresh: \ref options_model_smpi_detached
+- \c smpi/shared-malloc: \ref options_model_smpi_shared_malloc
+- \c smpi/shared-malloc-hugepage: \ref options_model_smpi_shared_malloc
+- \c smpi/simulate-computation: \ref options_smpi_bench
+- \c smpi/test: \ref options_model_smpi_test
+- \c smpi/wtime: \ref options_model_smpi_wtime
+
+- \c <b>Tracing configuration options can be found in Section \ref tracing_tracing_options</b>.
+
+- \c storage/model: \ref options_storage_model
+- \c verbose-exit: \ref options_generic_exit
+
+- \c vm/model: \ref options_vm_model
+
+\subsection options_index_smpi_coll Index of SMPI collective algorithms options
+
+TODO: All available collective algorithms will be made available via the ``smpirun --help-coll`` command.
+
 \section options_model Configuring the platform models
 
 \anchor options_storage_model
@@ -118,8 +216,8 @@ described in
 
 If you compiled SimGrid accordingly, you can use packet-level network
 simulators as network models (see \ref pls_ns3). In that case, you have
-two extra models, described below, and some \ref options_pls "specific
-additional configuration flags".
+two extra models, described below, and some 
+\ref options_pls "specific additional configuration flags".
   - \b NS3: Network pseudo-model using the NS3 tcp model
 
 Concerning the CPU, we have only one model for now:
@@ -544,7 +642,8 @@ In SimGrid, the user code is virtualized in a specific mechanism
 that allows the simulation kernel to control its execution: when a user
 process requires a blocking action (such as sending a message), it is
 interrupted, and only gets released when the simulated clock reaches
-the point where the blocking operation is done.
+the point where the blocking operation is done. This is explained
+graphically in the [relevant tutorial, available online](http://simgrid.gforge.inria.fr/tutorials/simgrid-simix-101.pdf).
 
 In SimGrid, the containers in which user processes are virtualized are
 called contexts. Several context factory are provided, and you can
@@ -553,7 +652,7 @@ configuration item. Some of the following may not exist on your
 machine because of portability issues. In any case, the default one
 should be the most effcient one (please report bugs if the
 auto-detection fails for you). They are approximately sorted here from
-the slowest to the most effient:
+the slowest to the most efficient:
 
  - \b thread: very slow factory using full featured threads (either
    pthreads or windows native threads). They are slow but very
@@ -1215,104 +1314,4 @@ hence, the output would mismatch, causing the test to fail.
 
 It can be done by using XBT. Go to \ref XBT_log for more details.
 
-\section options_index Index of all existing configuration options
-
-\note
-  The full list can be retrieved by passing "--help" and
-     "--help-cfg" to an executable that uses SimGrid.
-
-- \c clean-atexit: \ref options_generic_clean_atexit
-
-- \c contexts/factory: \ref options_virt_factory
-- \c contexts/guard-size: \ref options_virt_guard_size
-- \c contexts/nthreads: \ref options_virt_parallel
-- \c contexts/parallel_threshold: \ref options_virt_parallel
-- \c contexts/stack-size: \ref options_virt_stacksize
-- \c contexts/synchro: \ref options_virt_parallel
-
-- \c cpu/maxmin-selective-update: \ref options_model_optim
-- \c cpu/model: \ref options_model_select
-- \c cpu/optim: \ref options_model_optim
-
-- \c exception/cutpath: \ref options_exception_cutpath
-
-- \c host/model: \ref options_model_select
-
-- \c maxmin/precision: \ref options_model_precision
-- \c maxmin/concurrency-limit: \ref options_concurrency_limit
-
-- \c msg/debug-multiple-use: \ref options_msg_debug_multiple_use
-
-- \c model-check: \ref options_modelchecking
-- \c model-check/checkpoint: \ref options_modelchecking_steps
-- \c model-check/communications-determinism: \ref options_modelchecking_comm_determinism
-- \c model-check/dot-output: \ref options_modelchecking_dot_output
-- \c model-check/hash: \ref options_modelchecking_hash
-- \c model-check/property: \ref options_modelchecking_liveness
-- \c model-check/max-depth: \ref options_modelchecking_max_depth
-- \c model-check/record: \ref options_modelchecking_recordreplay
-- \c model-check/reduction: \ref options_modelchecking_reduction
-- \c model-check/replay: \ref options_modelchecking_recordreplay
-- \c model-check/send-determinism: \ref options_modelchecking_comm_determinism
-- \c model-check/sparse-checkpoint: \ref options_modelchecking_sparse_checkpoint
-- \c model-check/termination: \ref options_modelchecking_termination
-- \c model-check/timeout: \ref options_modelchecking_timeout
-- \c model-check/visited: \ref options_modelchecking_visited
-
-- \c network/bandwidth-factor: \ref options_model_network_coefs
-- \c network/crosstraffic: \ref options_model_network_crosstraffic
-- \c network/latency-factor: \ref options_model_network_coefs
-- \c network/maxmin-selective-update: \ref options_model_optim
-- \c network/model: \ref options_model_select
-- \c network/optim: \ref options_model_optim
-- \c network/sender_gap: \ref options_model_network_sendergap
-- \c network/TCP-gamma: \ref options_model_network_gamma
-- \c network/weight-S: \ref options_model_network_coefs
-
-- \c ns3/TcpModel: \ref options_pls
-- \c path: \ref options_generic_path
-- \c plugin: \ref options_generic_plugin
-
-- \c storage/max_file_descriptors: \ref option_model_storage_maxfd
-
-- \c surf/precision: \ref options_model_precision
-
-- \c <b>For collective operations of SMPI, please refer to Section \ref options_index_smpi_coll</b>
-- \c smpi/async-small-thresh: \ref options_model_network_asyncsend
-- \c smpi/bw-factor: \ref options_model_smpi_bw_factor
-- \c smpi/coll-selector: \ref options_model_smpi_collectives
-- \c smpi/comp-adjustment-file: \ref options_model_smpi_adj_file
-- \c smpi/cpu-threshold: \ref options_smpi_bench
-- \c smpi/display-timing: \ref options_smpi_timing
-- \c smpi/grow-injected-times: \ref options_model_smpi_test
-- \c smpi/host-speed: \ref options_smpi_bench
-- \c smpi/IB-penalty-factors: \ref options_model_network_coefs
-- \c smpi/iprobe: \ref options_model_smpi_iprobe
-- \c smpi/iprobe-cpu-usage: \ref options_model_smpi_iprobe_cpu_usage
-- \c smpi/init: \ref options_model_smpi_init
-- \c smpi/keep-temps: \ref options_smpi_temps
-- \c smpi/lat-factor: \ref options_model_smpi_lat_factor
-- \c smpi/ois: \ref options_model_smpi_ois
-- \c smpi/or: \ref options_model_smpi_or
-- \c smpi/os: \ref options_model_smpi_os
-- \c smpi/papi-events: \ref options_smpi_papi_events
-- \c smpi/privatization: \ref options_smpi_privatization
-- \c smpi/send-is-detached-thresh: \ref options_model_smpi_detached
-- \c smpi/shared-malloc: \ref options_model_smpi_shared_malloc
-- \c smpi/shared-malloc-hugepage: \ref options_model_smpi_shared_malloc
-- \c smpi/simulate-computation: \ref options_smpi_bench
-- \c smpi/test: \ref options_model_smpi_test
-- \c smpi/wtime: \ref options_model_smpi_wtime
-
-- \c <b>Tracing configuration options can be found in Section \ref tracing_tracing_options</b>.
-
-- \c storage/model: \ref options_storage_model
-- \c verbose-exit: \ref options_generic_exit
-
-- \c vm/model: \ref options_vm_model
-
-\subsection options_index_smpi_coll Index of SMPI collective algorithms options
-
-TODO: All available collective algorithms will be made available via the ``smpirun --help-coll`` command.
-
 */
diff --git a/examples/msg/app-chainsend/broadcaster.h b/examples/msg/app-chainsend/broadcaster.h
index d4413d4573..116f44fd03 100644
--- a/examples/msg/app-chainsend/broadcaster.h
+++ b/examples/msg/app-chainsend/broadcaster.h
@@ -32,7 +32,7 @@ typedef struct s_broadcaster {
 
 typedef s_broadcaster_t *broadcaster_t;
 
-xbt_dynar_t build_hostlist_from_hostcount(int hostcount); 
+xbt_dynar_t build_hostlist_from_hostcount(int hostcount);
 
 /* Broadcaster: helper functions */
 broadcaster_t broadcaster_init(xbt_dynar_t host_list, unsigned int piece_count);
diff --git a/examples/msg/app-chainsend/chainsend.c b/examples/msg/app-chainsend/chainsend.c
index 771e52177c..292abe8055 100644
--- a/examples/msg/app-chainsend/chainsend.c
+++ b/examples/msg/app-chainsend/chainsend.c
@@ -11,7 +11,7 @@
 #include "peer.h"
 
 /** @addtogroup MSG_examples
- * 
+ *
  *  - <b>chainsend: MSG implementation of a file broadcasting system, similar to Kastafior (from Kadeploy).</b>.
  */
 
diff --git a/examples/msg/app-chainsend/iterator.c b/examples/msg/app-chainsend/iterator.c
index a6f7e4448b..1bba4d552e 100644
--- a/examples/msg/app-chainsend/iterator.c
+++ b/examples/msg/app-chainsend/iterator.c
@@ -13,7 +13,7 @@
 xbt_dynar_iterator_t xbt_dynar_iterator_new(xbt_dynar_t list, xbt_dynar_t (*criteria_fn)(int))
 {
   xbt_dynar_iterator_t it = xbt_new(xbt_dynar_iterator_s, 1);
-  
+
   it->list = list;
   it->length = xbt_dynar_length(list);
   it->indices_list = criteria_fn(it->length); // Creates and fills a dynar of int
diff --git a/examples/msg/app-chainsend/peer.c b/examples/msg/app-chainsend/peer.c
index 63fa465069..f19df05eef 100644
--- a/examples/msg/app-chainsend/peer.c
+++ b/examples/msg/app-chainsend/peer.c
@@ -28,7 +28,7 @@ int peer_execute_task(peer_t peer, msg_task_t task)
 {
   int done = 0;
   message_t msg = MSG_task_get_data(task);
-  
+
   XBT_DEBUG("Peer %s got message of type %d\n", peer->me, msg->type);
   if (msg->type == MESSAGE_BUILD_CHAIN)
     peer_init_chain(peer, msg);
@@ -130,7 +130,7 @@ void peer_delete(peer_t p)
 
 void peer_print_stats(peer_t p, float elapsed_time)
 {
-  XBT_INFO("### %f %llu bytes (Avg %f MB/s); copy finished (simulated).", elapsed_time, p->bytes, p->bytes / 1024.0 / 1024.0 / elapsed_time); 
+  XBT_INFO("### %f %llu bytes (Avg %f MB/s); copy finished (simulated).", elapsed_time, p->bytes, p->bytes / 1024.0 / 1024.0 / elapsed_time);
 }
 
 /** Peer function  */
diff --git a/examples/msg/app-pingpong/app-pingpong.c b/examples/msg/app-pingpong/app-pingpong.c
index 9687ce7e2e..91a5b0fa18 100644
--- a/examples/msg/app-pingpong/app-pingpong.c
+++ b/examples/msg/app-pingpong/app-pingpong.c
@@ -30,7 +30,7 @@ static int pinger(int argc, char *argv[])
   XBT_INFO("Task received : %s", pong_task->name);
   xbt_free(pong_task->data);
   MSG_task_destroy(pong_task);
-  XBT_INFO("Pong time (bandwidth bound): %e", communication_time);
+  XBT_INFO("Pong time (bandwidth bound): %.3f", communication_time);
 
   return 0;
 }
@@ -51,14 +51,14 @@ static int ponger(int argc, char *argv[])
   XBT_INFO("Task received : %s", ping_task->name);
   xbt_free(ping_task->data);
   MSG_task_destroy(ping_task);
-  XBT_INFO(" Ping time (latency bound) %e", communication_time);
+  XBT_INFO(" Ping time (latency bound) %f", communication_time);
 
   /*  - ... Then send a 1GB pong back (bandwidth bound) */
   double now = MSG_get_clock();
   msg_task_t pong_task = MSG_task_create("large communication (bandwidth bound)", 0.0, 1e9, NULL);
   pong_task->data = xbt_new(double, 1);
   *(double *) pong_task->data = now;
-  XBT_INFO("task_bw->data = %e", *((double *) pong_task->data));
+  XBT_INFO("task_bw->data = %.3f", *((double*)pong_task->data));
   MSG_task_send(pong_task, argv[1]);
 
   return 0;
@@ -80,6 +80,6 @@ int main(int argc, char *argv[])
 
   msg_error_t res = MSG_main();             /* - Run the simulation */
 
-  XBT_INFO("Total simulation time: %e", MSG_get_clock());
+  XBT_INFO("Total simulation time: %.3f", MSG_get_clock());
   return res!=MSG_OK;
 }
diff --git a/examples/msg/app-pingpong/app-pingpong.tesh b/examples/msg/app-pingpong/app-pingpong.tesh
index d7e1a3a52a..a8108ff921 100644
--- a/examples/msg/app-pingpong/app-pingpong.tesh
+++ b/examples/msg/app-pingpong/app-pingpong.tesh
@@ -6,11 +6,11 @@ $ $SG_TEST_EXENV ${bindir:=.}/app-pingpong$EXEEXT ${srcdir:=.}/small_platform.xm
 > [  0.000000] (1:pinger@Tremblay) Ping -> Jupiter
 > [  0.000000] (2:ponger@Jupiter) Pong -> Tremblay
 > [  0.019014] (2:ponger@Jupiter) Task received : small communication (latency bound)
-> [  0.019014] (2:ponger@Jupiter)  Ping time (latency bound) 1.901449e-02
-> [  0.019014] (2:ponger@Jupiter) task_bw->data = 1.901449e-02
+> [  0.019014] (2:ponger@Jupiter)  Ping time (latency bound) 0.019014
+> [  0.019014] (2:ponger@Jupiter) task_bw->data = 0.019
 > [150.178356] (1:pinger@Tremblay) Task received : large communication (bandwidth bound)
-> [150.178356] (1:pinger@Tremblay) Pong time (bandwidth bound): 1.501593e+02
-> [150.178356] (0:maestro@) Total simulation time: 1.501784e+02
+> [150.178356] (1:pinger@Tremblay) Pong time (bandwidth bound): 150.159
+> [150.178356] (0:maestro@) Total simulation time: 150.178
 
 p Testing the deprecated CM02 network model
 
@@ -20,11 +20,11 @@ $ $SG_TEST_EXENV ${bindir:=.}/app-pingpong$EXEEXT ${srcdir:=.}/small_platform.xm
 > [  0.000000] (1:pinger@Tremblay) Ping -> Jupiter
 > [  0.000000] (2:ponger@Jupiter) Pong -> Tremblay
 > [  0.001462] (2:ponger@Jupiter) Task received : small communication (latency bound)
-> [  0.001462] (2:ponger@Jupiter)  Ping time (latency bound) 1.461663e-03
-> [  0.001462] (2:ponger@Jupiter) task_bw->data = 1.461663e-03
+> [  0.001462] (2:ponger@Jupiter)  Ping time (latency bound) 0.001462
+> [  0.001462] (2:ponger@Jupiter) task_bw->data = 0.001
 > [145.639041] (1:pinger@Tremblay) Task received : large communication (bandwidth bound)
-> [145.639041] (1:pinger@Tremblay) Pong time (bandwidth bound): 1.456376e+02
-> [145.639041] (0:maestro@) Total simulation time: 1.456390e+02
+> [145.639041] (1:pinger@Tremblay) Pong time (bandwidth bound): 145.638
+> [145.639041] (0:maestro@) Total simulation time: 145.639
 
 p Testing the surf network Reno fairness model using lagrangian approach
 
@@ -35,11 +35,11 @@ $ $SG_TEST_EXENV ${bindir:=.}/app-pingpong$EXEEXT ${srcdir:=.}/small_platform.xm
 > [  0.000000] (1:pinger@Tremblay) Ping -> Jupiter
 > [  0.000000] (2:ponger@Jupiter) Pong -> Tremblay
 > [  0.015200] (2:ponger@Jupiter) Task received : small communication (latency bound)
-> [  0.015200] (2:ponger@Jupiter)  Ping time (latency bound) 1.519994e-02
-> [  0.015200] (2:ponger@Jupiter) task_bw->data = 1.519994e-02
+> [  0.015200] (2:ponger@Jupiter)  Ping time (latency bound) 0.015200
+> [  0.015200] (2:ponger@Jupiter) task_bw->data = 0.015
 > [158.330528] (1:pinger@Tremblay) Task received : large communication (bandwidth bound)
-> [158.330528] (1:pinger@Tremblay) Pong time (bandwidth bound): 1.583153e+02
-> [158.330528] (0:maestro@) Total simulation time: 1.583305e+02
+> [158.330528] (1:pinger@Tremblay) Pong time (bandwidth bound): 158.315
+> [158.330528] (0:maestro@) Total simulation time: 158.331
 
 p Testing the surf network Vegas fairness model using lagrangian approach
 
@@ -50,8 +50,8 @@ $ $SG_TEST_EXENV ${bindir:=.}/app-pingpong$EXEEXT ${srcdir:=.}/small_platform.xm
 > [  0.000000] (1:pinger@Tremblay) Ping -> Jupiter
 > [  0.000000] (2:ponger@Jupiter) Pong -> Tremblay
 > [  0.015200] (2:ponger@Jupiter) Task received : small communication (latency bound)
-> [  0.015200] (2:ponger@Jupiter)  Ping time (latency bound) 1.519994e-02
-> [  0.015200] (2:ponger@Jupiter) task_bw->data = 1.519994e-02
+> [  0.015200] (2:ponger@Jupiter)  Ping time (latency bound) 0.015200
+> [  0.015200] (2:ponger@Jupiter) task_bw->data = 0.015
 > [158.330528] (1:pinger@Tremblay) Task received : large communication (bandwidth bound)
-> [158.330528] (1:pinger@Tremblay) Pong time (bandwidth bound): 1.583153e+02
-> [158.330528] (0:maestro@) Total simulation time: 1.583305e+02
+> [158.330528] (1:pinger@Tremblay) Pong time (bandwidth bound): 158.315
+> [158.330528] (0:maestro@) Total simulation time: 158.331
diff --git a/examples/msg/async-wait/async-wait.c b/examples/msg/async-wait/async-wait.c
index 286526c741..e13a66afb0 100644
--- a/examples/msg/async-wait/async-wait.c
+++ b/examples/msg/async-wait/async-wait.c
@@ -10,7 +10,7 @@ XBT_LOG_NEW_DEFAULT_CATEGORY(msg_async_wait, "Messages specific for this msg exa
 /* Main function of the Sender process */
 static int sender(int argc, char *argv[])
 {
-  xbt_assert(argc==7, "The sender function expects 6 arguments from the XML deployment file"); 
+  xbt_assert(argc==7, "The sender function expects 6 arguments from the XML deployment file");
   long number_of_tasks = xbt_str_parse_int(argv[1], "Invalid amount of tasks: %s");        /* - number of tasks */
   double task_comp_size = xbt_str_parse_double(argv[2], "Invalid computational size: %s"); /* - computational cost */
   double task_comm_size = xbt_str_parse_double(argv[3], "Invalid communication size: %s"); /* - communication cost */
diff --git a/examples/msg/async-waitall/async-waitall.c b/examples/msg/async-waitall/async-waitall.c
index 3bbfe40295..c828b13c6d 100644
--- a/examples/msg/async-waitall/async-waitall.c
+++ b/examples/msg/async-waitall/async-waitall.c
@@ -50,7 +50,7 @@ static int receiver(int argc, char *argv[])
 
   char mailbox[80];
   snprintf(mailbox,79, "receiver-%d", id);
-   
+
   MSG_process_sleep(10);
   while (1) {
     XBT_INFO("Wait to receive a task");
diff --git a/examples/msg/async-waitany/async-waitany.c b/examples/msg/async-waitany/async-waitany.c
index 01ebfc0635..95b92aa18d 100644
--- a/examples/msg/async-waitany/async-waitany.c
+++ b/examples/msg/async-waitany/async-waitany.c
@@ -30,7 +30,7 @@ static int sender(int argc, char *argv[])
     xbt_dynar_push_as(comms, msg_comm_t, comm);
     XBT_INFO("Send to receiver-%ld %s comm_size %f", i % receivers_count, taskname, task_comm_size / coef);
   }
-   
+
   /* Here we are waiting for the completion of all communications */
   while (xbt_dynar_is_empty(comms) == 0) {
     msg_comm_t comm;
@@ -63,7 +63,7 @@ static int receiver(int argc, char *argv[])
 
   char mailbox[80];
   snprintf(mailbox,79, "receiver-%d", id);
-   
+
   MSG_process_sleep(10);
   for (int i = 0; i < task_amount; i++) {
     XBT_INFO("Wait to receive task %d", i);
diff --git a/examples/msg/cloud-simple/cloud-simple.c b/examples/msg/cloud-simple/cloud-simple.c
index 7c633ad735..74d131432b 100644
--- a/examples/msg/cloud-simple/cloud-simple.c
+++ b/examples/msg/cloud-simple/cloud-simple.c
@@ -86,7 +86,7 @@ static void launch_communication_worker(msg_host_t tx_host, msg_host_t rx_host)
 {
   char *mbox = bprintf("MBOX:%s-%s", MSG_host_get_name(tx_host), MSG_host_get_name(rx_host));
   char **argv = NULL;
-  
+
   const char *pr_name_tx =  "comm_tx";
   argv = xbt_new(char *, 3);
   argv[0] = xbt_strdup(pr_name_tx);
@@ -95,7 +95,7 @@ static void launch_communication_worker(msg_host_t tx_host, msg_host_t rx_host)
 
   MSG_process_create_with_arguments(pr_name_tx, communication_tx_fun, NULL, tx_host, 2, argv);
 
-  const char *pr_name_rx =  "comm_rx";  
+  const char *pr_name_rx =  "comm_rx";
   argv = xbt_new(char *, 3);
   argv[0] = xbt_strdup(pr_name_rx);
   argv[1] = xbt_strdup(mbox);
diff --git a/examples/msg/dht-chord/dht-chord.c b/examples/msg/dht-chord/dht-chord.c
index 66faa97c3a..88096d24ba 100644
--- a/examples/msg/dht-chord/dht-chord.c
+++ b/examples/msg/dht-chord/dht-chord.c
@@ -134,7 +134,7 @@ static void print_finger_table(node_t node)
 }
 
 /* Sets a finger of the current node.
- * 
+ *
  * \param node the current node
  * \param finger_index index of the finger to set (0 to nb_bits - 1)
  * \param id the id to set for this finger
@@ -150,7 +150,7 @@ static void set_finger(node_t node, int finger_index, int id)
 }
 
 /* Sets the predecessor of the current node.
- * 
+ *
  * \param node the current node
  * \param id the id to predecessor, or -1 to unset the predecessor
  */
@@ -169,14 +169,14 @@ static void set_predecessor(node_t node, int predecessor_id)
 }
 
 /* Node main Function
- * 
+ *
  * Arguments:
  * - my id
  * - the id of a guy I know in the system (except for the first node)
  * - the time to sleep before I join (except for the first node)
  */
 /* This function is called when the current node receives a task.
- * 
+ *
  * \param node the current node
  * \param task the task to handle (don't touch it afterward: it will be destroyed, reused or forwarded)
  */
@@ -276,7 +276,7 @@ void create(node_t node)
 }
 
 /* Makes the current node join the ring, knowing the id of a node already in the ring
- * 
+ *
  * \param node the current node
  * \param known_id id of a node already in the ring
  * \return 1 if the join operation succeeded, 0 otherwise
@@ -340,7 +340,7 @@ void quit_notify(node_t node)
 }
 
 /* Makes the current node find the successor node of an id.
- * 
+ *
  * \param node the current node
  * \param id the id to find
  * \return the id of the successor node, or -1 if the request failed
@@ -358,7 +358,7 @@ int find_successor(node_t node, int id)
 }
 
 /* \brief Asks another node the successor node of an id.
- * 
+ *
  * \param node the current node
  * \param ask_to the node to ask to
  * \param id the id to find
@@ -447,7 +447,7 @@ int remote_find_successor(node_t node, int ask_to, int id)
 }
 
 /* Asks its predecessor to a remote node
- * 
+ *
  * \param node the current node
  * \param ask_to the node to ask to
  * \return the id of its predecessor node, or -1 if the request failed
@@ -526,7 +526,7 @@ int remote_get_predecessor(node_t node, int ask_to)
 }
 
 /* Returns the closest preceding finger of an id with respect to the finger table of the current node.
- * 
+ *
  * \param node the current node
  * \param id the id to find
  * \return the closest preceding finger of that id
diff --git a/examples/msg/dht-pastry/dht-pastry.c b/examples/msg/dht-pastry/dht-pastry.c
index b9cac1eb3a..517797e6d8 100644
--- a/examples/msg/dht-pastry/dht-pastry.c
+++ b/examples/msg/dht-pastry/dht-pastry.c
@@ -245,13 +245,13 @@ static void handle_task(node_t node, msg_task_t task) {
     /* Try to join the ring */
     case TASK_JOIN:
       next = routing_next(node, task_data->answer_id);
-      XBT_DEBUG("Join request from %08x forwarding to %08x", task_data->answer_id, next);      
+      XBT_DEBUG("Join request from %08x forwarding to %08x", task_data->answer_id, next);
       type = TASK_JOIN_LAST_REPLY;
 
       req_data = xbt_new0(s_task_data_t,1);
       req_data->answer_id = task_data->sender_id;
       req_data->steps = task_data->steps + 1;
-      
+
       // if next different from current node forward the join
       if (next!=node->id) {
         get_mailbox(next, mailbox);
@@ -263,8 +263,8 @@ static void handle_task(node_t node, msg_task_t task) {
           task_free(task_sent);
         }
         type = TASK_JOIN_REPLY;
-      } 
-      
+      }
+
       // send back the current node state to the joining node
       req_data->type = type;
       req_data->sender_id = node->id;
@@ -351,7 +351,7 @@ static void handle_task(node_t node, msg_task_t task) {
       print_node_namespace_set(node);
       int curr_namespace_set[NAMESPACE_SIZE];
       int task_namespace_set[NAMESPACE_SIZE+1];
-      
+
       // Copy the current namespace and the task state namespace with state->id in the middle
       i=0;
       for (; i<NAMESPACE_SIZE/2; i++){
@@ -360,7 +360,7 @@ static void handle_task(node_t node, msg_task_t task) {
       }
       task_namespace_set[i] = task_data->state->id;
       for (; i<NAMESPACE_SIZE; i++){
-        curr_namespace_set[i] = node->namespace_set[i];  
+        curr_namespace_set[i] = node->namespace_set[i];
         task_namespace_set[i+1] = task_data->state->namespace_set[i];
       }
 
@@ -465,8 +465,8 @@ static int join(node_t node){
 static int node(int argc, char *argv[])
 {
   double init_time = MSG_get_clock();
-  msg_task_t task_received = NULL;  
-  int join_success = 0;  
+  msg_task_t task_received = NULL;
+  int join_success = 0;
   double deadline;
   xbt_assert(argc == 3 || argc == 5, "Wrong number of arguments for this node");
   s_node_t node = {0};
@@ -556,9 +556,9 @@ static int node(int argc, char *argv[])
 int main(int argc, char *argv[])
 {
   MSG_init(&argc, argv);
-  xbt_assert(argc > 2, 
+  xbt_assert(argc > 2,
        "Usage: %s [-nb_bits=n] [-timeout=t] platform_file deployment_file\n"
-       "\tExample: %s ../msg_platform.xml pastry10.xml\n", 
+       "\tExample: %s ../msg_platform.xml pastry10.xml\n",
        argv[0], argv[0]);
 
   char **options = &argv[1];
diff --git a/examples/msg/maestro-set/maestro-set.cpp b/examples/msg/maestro-set/maestro-set.cpp
index c1ab4d27f2..3b37c1cb95 100644
--- a/examples/msg/maestro-set/maestro-set.cpp
+++ b/examples/msg/maestro-set/maestro-set.cpp
@@ -5,10 +5,10 @@
 
 /** @addtogroup MSG_examples
  *
- *  - <b>maestro-set/maestro-set.cpp: Switch the system thread hosting our maestro</b>. 
+ *  - <b>maestro-set/maestro-set.cpp: Switch the system thread hosting our maestro</b>.
  *    That's a very advanced example in which we move the maestro thread to another process.
  *    Not many users need it (maybe only one, actually), but this example is also a regression test.
- * 
+ *
  *    This example is in C++ because we use C++11 threads to ensure that the feature is working as
  *    expected. You can still use that feature from a C code.
  */
diff --git a/examples/msg/mc/bugged1_liveness.c b/examples/msg/mc/bugged1_liveness.c
index 47107f0fca..05da543dc7 100644
--- a/examples/msg/mc/bugged1_liveness.c
+++ b/examples/msg/mc/bugged1_liveness.c
@@ -21,7 +21,7 @@
 
 XBT_LOG_NEW_DEFAULT_CATEGORY(bugged1_liveness, "my log messages");
 
-int r=0; 
+int r=0;
 int cs=0;
 
 #ifdef GARBAGE_STACK
@@ -43,15 +43,15 @@ static int coordinator(int argc, char *argv[])
   xbt_dynar_t requests = xbt_dynar_new(sizeof(char *), NULL);
   char *req;
 
-  while(1){  
+  while(1){
     MSG_task_receive(&task, "coordinator");
-    const char *kind = MSG_task_get_name(task); 
-    if (!strcmp(kind, "request")) {    
+    const char *kind = MSG_task_get_name(task);
+    if (!strcmp(kind, "request")) {
       req = MSG_task_get_data(task);
-      if (CS_used) {           
+      if (CS_used) {
         XBT_INFO("CS already used. Queue the request.");
         xbt_dynar_push(requests, &req);
-      } else {               
+      } else {
         if(strcmp(req, "1") != 0){
           XBT_INFO("CS idle. Grant immediatly");
           answer = MSG_task_create("grant", 0, 1000, NULL);
@@ -130,7 +130,7 @@ static int client(int argc, char *argv[])
       r=0;
       XBT_INFO("Propositions changed : r=0, cs=0");
     }
-    
+
   }
   return 0;
 }
diff --git a/examples/msg/mc/bugged1_liveness.h b/examples/msg/mc/bugged1_liveness.h
index c24c967c49..eaf06f44fb 100644
--- a/examples/msg/mc/bugged1_liveness.h
+++ b/examples/msg/mc/bugged1_liveness.h
@@ -10,4 +10,4 @@
 int predR(void);
 int predCS(void);
 
-#endif 
+#endif
diff --git a/examples/msg/mc/bugged2_liveness.c b/examples/msg/mc/bugged2_liveness.c
index 2310f28163..67e91ac89d 100644
--- a/examples/msg/mc/bugged2_liveness.c
+++ b/examples/msg/mc/bugged2_liveness.c
@@ -28,7 +28,7 @@ static int coordinator(int argc, char *argv[])
     const char *kind = MSG_task_get_name(task); //is it a request or a release?
     if (!strcmp(kind, "request")) {     // that's a request
       char *req = MSG_task_get_data(task);
-      if (CS_used) { 
+      if (CS_used) {
         XBT_INFO("CS already used.");
         msg_task_t answer = MSG_task_create("not grant", 0, 1000, NULL);
         MSG_task_send(answer, req);
diff --git a/examples/msg/mc/bugged2_liveness.h b/examples/msg/mc/bugged2_liveness.h
index d9ad041d7e..787520a42f 100644
--- a/examples/msg/mc/bugged2_liveness.h
+++ b/examples/msg/mc/bugged2_liveness.h
@@ -9,4 +9,4 @@
 
 int predCS(void);
 
-#endif 
+#endif
diff --git a/examples/msg/mc/electric_fence.c b/examples/msg/mc/electric_fence.c
index b8b2c5313c..7729c38a9d 100644
--- a/examples/msg/mc/electric_fence.c
+++ b/examples/msg/mc/electric_fence.c
@@ -5,7 +5,7 @@
 
 /******************** Non-deterministic message ordering  *********************/
 /* This example implements one process which receives messages from two other */
-/* processes. There is no bug on it, it is just provided to test the soundness*/ 
+/* processes. There is no bug on it, it is just provided to test the soundness*/
 /* of the state space reduction with DPOR, if the maximum depth (defined with */
 /* --cfg=model-check/max_depth:) is reached.                                  */
 /******************************************************************************/
diff --git a/examples/msg/platform-properties/platform-properties.c b/examples/msg/platform-properties/platform-properties.c
index 9e1de6daae..a7b4d5963e 100644
--- a/examples/msg/platform-properties/platform-properties.c
+++ b/examples/msg/platform-properties/platform-properties.c
@@ -7,7 +7,7 @@
 
 XBT_LOG_NEW_DEFAULT_CATEGORY(test, "Property test");
 
-static void test_host(const char*hostname) 
+static void test_host(const char*hostname)
 {
   msg_host_t thehost = MSG_host_by_name(hostname);
   xbt_dict_t props = MSG_host_get_properties(thehost);
diff --git a/examples/msg/synchro-semaphore/synchro-semaphore.c b/examples/msg/synchro-semaphore/synchro-semaphore.c
index aa8d556631..d91e546767 100644
--- a/examples/msg/synchro-semaphore/synchro-semaphore.c
+++ b/examples/msg/synchro-semaphore/synchro-semaphore.c
@@ -11,7 +11,7 @@ XBT_LOG_NEW_DEFAULT_CATEGORY(msg_semaphore_example, "Messages specific for this
 msg_sem_t sem;
 
 static int peer(int argc, char* argv[]){
-  int i = 0; 
+  int i = 0;
   while(i < argc) {
     double wait_time = xbt_str_parse_double(argv[i],"Invalid wait time: %s");
     i++;
diff --git a/examples/s4u/io/s4u_io.cpp b/examples/s4u/io/s4u_io.cpp
index a62a5ebc54..90126b62a4 100644
--- a/examples/s4u/io/s4u_io.cpp
+++ b/examples/s4u/io/s4u_io.cpp
@@ -17,7 +17,7 @@ public:
 
     for (const auto&kv : mounts) {
       const char* mountpoint = kv.first.c_str();
-      simgrid::s4u::Storage &storage = *kv.second;
+      simgrid::s4u::Storage storage = *kv.second;
 
       // Retrieve disk's information
       sg_size_t free_size = storage.sizeFree();
@@ -55,7 +55,7 @@ public:
     write = file->write(100000);  // Write 100,000 bytes
     XBT_INFO("Write %llu bytes on %s", write, filename);
 
-    simgrid::s4u::Storage &storage = simgrid::s4u::Storage::byName("Disk4");
+    simgrid::s4u::Storage* storage = simgrid::s4u::Storage::byName("Disk4");
 
     // Now rename file from ./tmp/data.txt to ./tmp/simgrid.readme
     const char *newpath = "/home/tmp/simgrid.readme";
@@ -71,13 +71,13 @@ public:
     delete file;
 
     // Now attach some user data to disk1
-    XBT_INFO("Get/set data for storage element: %s",storage.name());
-    XBT_INFO("    Uninitialized storage data: '%s'", (char*)storage.userdata());
+    XBT_INFO("Get/set data for storage element: %s", storage->name());
+    XBT_INFO("    Uninitialized storage data: '%s'", (char*)storage->userdata());
 
-    storage.setUserdata(xbt_strdup("Some user data"));
-    XBT_INFO("    Set and get data: '%s'", (char*)storage.userdata());
+    storage->setUserdata(xbt_strdup("Some user data"));
+    XBT_INFO("    Set and get data: '%s'", (char*)storage->userdata());
 
-    xbt_free(storage.userdata());
+    xbt_free(storage->userdata());
   }
 };
 
diff --git a/examples/simdag/availability/sd_availability.c b/examples/simdag/availability/sd_availability.c
index 1d9c75c775..a0b20a03f9 100644
--- a/examples/simdag/availability/sd_availability.c
+++ b/examples/simdag/availability/sd_availability.c
@@ -83,7 +83,7 @@ int main(int argc, char **argv)
              sg_host_route_latency(hosts[0], hosts[1]));
     XBT_INFO("Jupiter: speed=%.0f", sg_host_speed(hosts[0])* sg_host_get_available_speed(hosts[0]));
     XBT_INFO("Tremblay: speed=%.0f", sg_host_speed(hosts[1])* sg_host_get_available_speed(hosts[1]));
-     
+
     unsigned int ctr;
     SD_task_t task;
     xbt_dynar_foreach(changed_tasks, ctr, task) {
diff --git a/examples/simdag/dag-dotload/sd_dag-dotload.c b/examples/simdag/dag-dotload/sd_dag-dotload.c
index 6baccb0fbc..176bd08ddd 100644
--- a/examples/simdag/dag-dotload/sd_dag-dotload.c
+++ b/examples/simdag/dag-dotload/sd_dag-dotload.c
@@ -39,7 +39,7 @@ int main(int argc, char **argv)
   char *tracefilename;
   char *last = strrchr(argv[2], '.');
   tracefilename = bprintf("%.*s.trace", (int) (last == NULL ? strlen(argv[2]) : last - argv[2]),argv[2]);
-  if (argc == 4) 
+  if (argc == 4)
     tracefilename = xbt_strdup(argv[3]);
 
   /* Display all the tasks */
diff --git a/examples/simdag/daxload/sd_daxload.c b/examples/simdag/daxload/sd_daxload.c
index e3af5411e6..ef3d029fcf 100644
--- a/examples/simdag/daxload/sd_daxload.c
+++ b/examples/simdag/daxload/sd_daxload.c
@@ -42,7 +42,7 @@ int main(int argc, char **argv)
   if (!dax){
     XBT_ERROR("A problem occurred during DAX parsing (cycle or syntax). Do not continue this test");
     free(tracefilename);
-  
+
     exit(255);
   }
 
diff --git a/examples/simdag/goal/goal_test.c b/examples/simdag/goal/goal_test.c
index 6dfc394840..b5d73b4cfc 100644
--- a/examples/simdag/goal/goal_test.c
+++ b/examples/simdag/goal/goal_test.c
@@ -1,6 +1,6 @@
-/* Example of scatter communication, accepting a large amount of processes. 
+/* Example of scatter communication, accepting a large amount of processes.
  * This based the experiment of Fig. 4 in http://hal.inria.fr/hal-00650233/
- * That experiment is a comparison to the LogOPSim simulator, that takes 
+ * That experiment is a comparison to the LogOPSim simulator, that takes
  * GOAL files as an input, thus the file name. But there is no actual link
  * to the GOAL formalism beside of this.
  */
diff --git a/examples/simdag/scheduling/sd_scheduling.c b/examples/simdag/scheduling/sd_scheduling.c
index 870583652f..bb53f2b888 100644
--- a/examples/simdag/scheduling/sd_scheduling.c
+++ b/examples/simdag/scheduling/sd_scheduling.c
@@ -65,14 +65,14 @@ static xbt_dynar_t get_ready_tasks(xbt_dynar_t dax)
 static double finish_on_at(SD_task_t task, sg_host_t host)
 {
   double result;
-  unsigned int i;
-  double data_available = 0.;
-  double redist_time = 0;
-  double last_data_available;
 
   xbt_dynar_t parents = SD_task_get_parents(task);
 
   if (!xbt_dynar_is_empty(parents)) {
+    unsigned int i;
+    double data_available = 0.;
+    double redist_time    = 0;
+    double last_data_available;
     /* compute last_data_available */
     SD_task_t parent;
     last_data_available = -1.0;
diff --git a/examples/smpi/NAS/DGraph.c b/examples/smpi/NAS/DGraph.c
index 530bf3ee94..ad4f682106 100644
--- a/examples/smpi/NAS/DGraph.c
+++ b/examples/smpi/NAS/DGraph.c
@@ -37,7 +37,7 @@ void nodeShow(DGNode* nd){
 /*
   if(nd->verified==1) fprintf(stderr,"%ld.%s\t: usable.",nd->id,nd->name);
   else if(nd->verified==0)  fprintf(stderr,"%ld.%s\t: unusable.",nd->id,nd->name);
-  else  fprintf(stderr,"%ld.%s\t: notverified.",nd->id,nd->name);   
+  else  fprintf(stderr,"%ld.%s\t: notverified.",nd->id,nd->name);
 */
 }
 
@@ -83,7 +83,7 @@ int AttachNode(DGraph* dg, DGNode* nd) {
       }
       memcpy( &(tmpnd->inArc[ tmpnd->inDegree]), nd->inArc, nd->inDegree*sizeof( DGArc *));
       tmpnd->inDegree += nd->inDegree;
-    }   
+    }
     if ( nd->outDegree > 0 ) {
       tmpnd->maxOutDegree += nd->maxOutDegree;
       ar =(DGArc **) calloc(tmpnd->maxOutDegree,sizeof(DGArc*));
@@ -96,7 +96,7 @@ int AttachNode(DGraph* dg, DGNode* nd) {
       memcpy( &(tmpnd->outArc[tmpnd->outDegree]),nd->outArc,nd->outDegree*sizeof( DGArc *));
       tmpnd->outDegree += nd->outDegree;
     }
-    free(nd); 
+    free(nd);
     return i;
   }
   nd->id = dg->numNodes;
diff --git a/examples/smpi/NAS/dt.c b/examples/smpi/NAS/dt.c
index 55b29dd5e8..7dacc29df8 100644
--- a/examples/smpi/NAS/dt.c
+++ b/examples/smpi/NAS/dt.c
@@ -1,10 +1,10 @@
 /*************************************************************************
- *                                                                       * 
+ *                                                                       *
  *    N  A  S   P A R A L L E L   B E N C H M A R K S  3.3     *
  *                                     *
  *                  D T       *
  *                                     *
- ************************************************************************* 
+ *************************************************************************
  *                                     *
  *   This benchmark is part of the NAS Parallel Benchmark 3.3 suite.   *
  *                                     *
@@ -31,7 +31,7 @@
  *     E-mail:  npb@nas.nasa.gov                   *
  *     Fax:   (650) 604-3957                     *
  *                                     *
- ************************************************************************* 
+ *************************************************************************
  *                                     *
  *   Author: M. Frumkin         *       *
  *                                     *
@@ -263,7 +263,7 @@ static DGraph *buildWH(const char cls){
     numPrevLayerNodes=numLayerNodes;
   }
   source=newNode((char*)"Source");
-  AttachNode(dg,source);   
+  AttachNode(dg,source);
   for(i=0;i<numPrevLayerNodes;i++){
     nd=dg->node[firstLayerNode+i];
     ar=newArc(source,nd);
@@ -323,7 +323,7 @@ static DGraph *buildBH(const char cls){
     numPrevLayerNodes=numLayerNodes;
   }
   sink=newNode((char*)"Sink");
-  AttachNode(dg,sink);   
+  AttachNode(dg,sink);
   for(i=0;i<numPrevLayerNodes;i++){
     nd=dg->node[firstLayerNode+i];
     ar=newArc(nd,sink);
@@ -428,7 +428,7 @@ static Arr* WindowFilter(Arr *a, Arr* b,int w){
   int i=0,j=0,k=0;
   double rms0=0.0,rms1=0.0,rmsm1=0.0;
   double weight=((double) (w+1))/(w+2);
- 
+
   w+=1;
   if(timer_on){
     timer_clear(w);
diff --git a/examples/smpi/NAS/ep.c b/examples/smpi/NAS/ep.c
index 8dde5cb4e9..24257dc1b2 100644
--- a/examples/smpi/NAS/ep.c
+++ b/examples/smpi/NAS/ep.c
@@ -24,7 +24,7 @@ int main(int argc, char **argv) {
   int    m;
   int    mk=16;
   int    nk = (int)(pow(2,mk)),
-         nq=10, 
+         nq=10,
          np, node, no_nodes, i, ik, kk, l, k, nit, no_large_nodes, np_add, k_offset;
   int    verified;
   char   size[500]; // mind the size of the string to represent a big number
diff --git a/examples/smpi/NAS/is.c b/examples/smpi/NAS/is.c
index e29b87c4ea..7314dee222 100644
--- a/examples/smpi/NAS/is.c
+++ b/examples/smpi/NAS/is.c
@@ -1,42 +1,42 @@
 /*************************************************************************
- *                                                                       * 
+ *                                                                       *
  *        N  A  S     P A R A L L E L     B E N C H M A R K S  3.3       *
- *                                                                       * 
- *                                  I S                                  * 
- *                                                                       * 
- ************************************************************************* 
- *                                                                       * 
+ *                                                                       *
+ *                                  I S                                  *
+ *                                                                       *
+ *************************************************************************
+ *                                                                       *
  *   This benchmark is part of the NAS Parallel Benchmark 3.3 suite.     *
- *   It is described in NAS Technical Report 95-020.                     * 
- *                                                                       * 
- *   Permission to use, copy, distribute and modify this software        * 
- *   for any purpose with or without fee is hereby granted.  We          * 
- *   request, however, that all derived work reference the NAS           * 
+ *   It is described in NAS Technical Report 95-020.                     *
+ *                                                                       *
+ *   Permission to use, copy, distribute and modify this software        *
+ *   for any purpose with or without fee is hereby granted.  We          *
+ *   request, however, that all derived work reference the NAS           *
  *   Parallel Benchmarks 3.3. This software is provided "as is"          *
- *   without express or implied warranty.                                * 
- *                                                                       * 
+ *   without express or implied warranty.                                *
+ *                                                                       *
  *   Information on NPB 3.3, including the technical report, the         *
- *   original specifications, source code, results and information       * 
- *   on how to submit new results, is available at:                      * 
- *                                                                       * 
- *          http://www.nas.nasa.gov/Software/NPB                         * 
- *                                                                       * 
- *   Send comments or suggestions to  npb@nas.nasa.gov                   * 
- *   Send bug reports to              npb-bugs@nas.nasa.gov              * 
- *                                                                       * 
- *         NAS Parallel Benchmarks Group                                 * 
- *         NASA Ames Research Center                                     * 
- *         Mail Stop: T27A-1                                             * 
- *         Moffett Field, CA   94035-1000                                * 
- *                                                                       * 
- *         E-mail:  npb@nas.nasa.gov                                     * 
- *         Fax:     (650) 604-3957                                       * 
- *                                                                       * 
- ************************************************************************* 
- *                                                                       * 
- *   Author: M. Yarrow                                                   * 
- *           H. Jin                                                      * 
- *                                                                       * 
+ *   original specifications, source code, results and information       *
+ *   on how to submit new results, is available at:                      *
+ *                                                                       *
+ *          http://www.nas.nasa.gov/Software/NPB                         *
+ *                                                                       *
+ *   Send comments or suggestions to  npb@nas.nasa.gov                   *
+ *   Send bug reports to              npb-bugs@nas.nasa.gov              *
+ *                                                                       *
+ *         NAS Parallel Benchmarks Group                                 *
+ *         NASA Ames Research Center                                     *
+ *         Mail Stop: T27A-1                                             *
+ *         Moffett Field, CA   94035-1000                                *
+ *                                                                       *
+ *         E-mail:  npb@nas.nasa.gov                                     *
+ *         Fax:     (650) 604-3957                                       *
+ *                                                                       *
+ *************************************************************************
+ *                                                                       *
+ *   Author: M. Yarrow                                                   *
+ *           H. Jin                                                      *
+ *                                                                       *
  *************************************************************************/
 
 #include "smpi/mpi.h"
@@ -221,7 +221,7 @@ static void rank( global_data* gd, int iteration )
   INT_TYPE    min_key_val, max_key_val;
   INT_TYPE    *key_buff_ptr;
 
-/*  Iteration alteration of keys */  
+/*  Iteration alteration of keys */
   if(gd->my_rank == 0){
     gd->key_array[iteration] = iteration;
     gd->key_array[iteration+MAX_ITERATIONS] = max_key - iteration;
@@ -579,7 +579,7 @@ int main( int argc, char **argv )
              1220703125.00 ),   /* Random number gen mult */
              1220703125.00 );   /* Random number gen mult */
 
-/*  Do one interation for free (i.e., untimed) to guarantee initialization of  
+/*  Do one interation for free (i.e., untimed) to guarantee initialization of
     all data and code pages and respective tables */
   rank(gd, 1 );
 
diff --git a/examples/smpi/mc/bugged1_liveness.c b/examples/smpi/mc/bugged1_liveness.c
index 5ec2c2b24a..317eecc530 100644
--- a/examples/smpi/mc/bugged1_liveness.c
+++ b/examples/smpi/mc/bugged1_liveness.c
@@ -24,13 +24,14 @@
 int r, cs;
 
 int main(int argc, char **argv){
-  int err, size, rank;
+  int size;
+  int rank;
   int recv_buff;
   MPI_Status status;
   xbt_dynar_t requests = xbt_dynar_new(sizeof(int), NULL);
 
   /* Initialize MPI */
-  err = MPI_Init(&argc, &argv);
+  int err = MPI_Init(&argc, &argv);
   if(err !=  MPI_SUCCESS){
     printf("MPI initialization failed !\n");
     exit(1);
diff --git a/examples/smpi/mc/bugged2.c b/examples/smpi/mc/bugged2.c
index dd529f8894..d3baebffc5 100644
--- a/examples/smpi/mc/bugged2.c
+++ b/examples/smpi/mc/bugged2.c
@@ -12,11 +12,14 @@
 
 int main(int argc, char **argv)
 {
-  int x,y, err, size, rank;
+  int x;
+  int y;
+  int size;
+  int rank;
   MPI_Status status;
 
   /* Initialize MPI */
-  err = MPI_Init(&argc, &argv);
+  int err = MPI_Init(&argc, &argv);
   if (err != MPI_SUCCESS) {
     printf("MPI initialization failed!\n");
     exit(1);
diff --git a/examples/smpi/mc/mutual_exclusion.c b/examples/smpi/mc/mutual_exclusion.c
index c63b9fd4b6..35f7b82205 100644
--- a/examples/smpi/mc/mutual_exclusion.c
+++ b/examples/smpi/mc/mutual_exclusion.c
@@ -13,14 +13,15 @@
 #define RELEASE_TAG 2
 
 int main(int argc, char **argv){
-  int err, size, rank;
+  int size;
+  int rank;
   int recv_buff;
   MPI_Status status;
   int CS_used = 0;
   xbt_dynar_t requests = xbt_dynar_new(sizeof(int), NULL);
 
   /* Initialize MPI */
-  err = MPI_Init(&argc, &argv);
+  int err = MPI_Init(&argc, &argv);
   if(err !=  MPI_SUCCESS){
     printf("MPI initialization failed !\n");
     exit(1);
diff --git a/examples/smpi/mc/non_termination1.c b/examples/smpi/mc/non_termination1.c
index 24f0593a0c..3ec31a2f33 100644
--- a/examples/smpi/mc/non_termination1.c
+++ b/examples/smpi/mc/non_termination1.c
@@ -6,7 +6,9 @@ int x = 5;
 int y = 8;
 
 int main(int argc, char **argv) {
-  int recv_buff, size, rank;
+  int recv_buff;
+  int size;
+  int rank;
   MPI_Status status;
 
   MPI_Init(&argc, &argv);
diff --git a/examples/smpi/mc/non_termination2.c b/examples/smpi/mc/non_termination2.c
index aafca4d250..60aa0a8bf2 100644
--- a/examples/smpi/mc/non_termination2.c
+++ b/examples/smpi/mc/non_termination2.c
@@ -5,7 +5,9 @@
 int x;
 
 int main(int argc, char **argv) {
-  int recv_buff, size, rank;
+  int recv_buff;
+  int size;
+  int rank;
   MPI_Status status;
 
   MPI_Init(&argc, &argv);
diff --git a/examples/smpi/mc/non_termination3.c b/examples/smpi/mc/non_termination3.c
index efdaa3cf13..0e0f56f349 100644
--- a/examples/smpi/mc/non_termination3.c
+++ b/examples/smpi/mc/non_termination3.c
@@ -6,7 +6,10 @@ int x = 0;
 int y = 0;
 
 int main(int argc, char **argv) {
-  int recv_x, recv_y, size, rank;
+  int recv_x;
+  int recv_y;
+  int size;
+  int rank;
   MPI_Status status;
 
   MPI_Init(&argc, &argv);
diff --git a/examples/smpi/mc/non_termination4.c b/examples/smpi/mc/non_termination4.c
index 5e7add0560..a0792ed812 100644
--- a/examples/smpi/mc/non_termination4.c
+++ b/examples/smpi/mc/non_termination4.c
@@ -5,7 +5,9 @@
 int x = 20;
 
 int main(int argc, char **argv) {
-  int recv_x = 1, size, rank;
+  int recv_x = 1;
+  int size;
+  int rank;
   MPI_Status status;
 
   MPI_Init(&argc, &argv);
diff --git a/examples/smpi/mc/only_send_deterministic.c b/examples/smpi/mc/only_send_deterministic.c
index 01188f1412..8c9108c2dd 100644
--- a/examples/smpi/mc/only_send_deterministic.c
+++ b/examples/smpi/mc/only_send_deterministic.c
@@ -12,11 +12,13 @@
 
 int main(int argc, char **argv)
 {
-  int recv_buff, err, size, rank;
+  int recv_buff;
+  int size;
+  int rank;
   MPI_Status status;
 
   /* Initialize MPI */
-  err = MPI_Init(&argc, &argv);
+  int err = MPI_Init(&argc, &argv);
   if (err != MPI_SUCCESS) {
     printf("MPI initialization failed!\n");
     exit(1);
diff --git a/examples/smpi/smpi_msg_masterslave/masterslave_mailbox_smpi.c b/examples/smpi/smpi_msg_masterslave/masterslave_mailbox_smpi.c
index 51cc923511..90091d7b25 100644
--- a/examples/smpi/smpi_msg_masterslave/masterslave_mailbox_smpi.c
+++ b/examples/smpi/smpi_msg_masterslave/masterslave_mailbox_smpi.c
@@ -72,7 +72,7 @@ static int alltoall_mpi(int argc, char *argv[])
   XBT_INFO("alltoall for rank %d", rank);
   int* out=malloc(1000*size*sizeof(int));
   int* in=malloc(1000*size*sizeof(int));
-  MPI_Alltoall(out, 1000, MPI_INT,in, 1000, MPI_INT, MPI_COMM_WORLD); 
+  MPI_Alltoall(out, 1000, MPI_INT,in, 1000, MPI_INT, MPI_COMM_WORLD);
 
   XBT_INFO("after alltoall %d", rank);
   free(out);
diff --git a/include/simgrid/forward.h b/include/simgrid/forward.h
index 82b6c583e9..96e8194eac 100644
--- a/include/simgrid/forward.h
+++ b/include/simgrid/forward.h
@@ -17,14 +17,28 @@ namespace simgrid {
 namespace kernel {
 namespace activity {
   class ActivityImpl;
+  using ActivityImplPtr = boost::intrusive_ptr<ActivityImpl>;
   XBT_PUBLIC(void) intrusive_ptr_add_ref(ActivityImpl* activity);
   XBT_PUBLIC(void) intrusive_ptr_release(ActivityImpl* activity);
+
+  class CommImpl;
+  using CommImplPtr = boost::intrusive_ptr<CommImpl>;
+  class ExecImpl;
+  using ExecImplPtr = boost::intrusive_ptr<ExecImpl>;
+  class IoImpl;
+  using IoImplPtr = boost::intrusive_ptr<IoImpl>;
+  class RawImpl;
+  using RawImplPtr = boost::intrusive_ptr<RawImpl>;
+  class SleepImpl;
+  using SleepImplPtr = boost::intrusive_ptr<SleepImpl>;
 }
 namespace routing {
   class NetPoint;
 }
 }
 namespace simix {
+  class ActorImpl;
+  using ActorImplPtr = boost::intrusive_ptr<ActorImpl>;
   class Host;
 }
 namespace surf {
@@ -43,8 +57,10 @@ namespace trace_mgr {
 typedef simgrid::s4u::Actor s4u_Actor;
 typedef simgrid::s4u::Host s4u_Host;
 typedef simgrid::s4u::Link s4u_Link;
+typedef simgrid::s4u::File s4u_File;
+typedef simgrid::s4u::Storage s4u_Storage;
 typedef simgrid::s4u::NetZone s4u_NetZone;
-typedef simgrid::kernel::activity::ActivityImpl* smx_activity_t;
+typedef boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> smx_activity_t;
 typedef simgrid::kernel::routing::NetPoint routing_NetPoint;
 typedef simgrid::surf::Resource surf_Resource;
 typedef simgrid::trace_mgr::trace tmgr_Trace;
@@ -54,6 +70,8 @@ typedef simgrid::trace_mgr::trace tmgr_Trace;
 typedef struct s4u_Actor s4u_Actor;
 typedef struct s4u_Host s4u_Host;
 typedef struct s4u_Link s4u_Link;
+typedef struct s4u_File s4u_File;
+typedef struct s4u_Storage s4u_Storage;
 typedef struct s4u_NetZone s4u_NetZone;
 typedef struct kernel_Activity* smx_activity_t;
 typedef struct routing_NetPoint routing_NetPoint;
@@ -65,14 +83,12 @@ typedef struct Trace tmgr_Trace;
 typedef s4u_NetZone* sg_netzone_t;
 typedef s4u_Host* sg_host_t;
 typedef s4u_Link* sg_link_t;
-
+typedef s4u_Storage* sg_storage_t;
+typedef s4u_File* sg_file_t;
 
 typedef routing_NetPoint* sg_netpoint_t;
 typedef surf_Resource *sg_resource_t;
 
-// Types which are in fact dictelmt:
-typedef struct s_xbt_dictelm *sg_storage_t;
-
 typedef tmgr_Trace *tmgr_trace_t; /**< Opaque structure defining an availability trace */
 
 typedef struct s_smx_simcall s_smx_simcall_t;
diff --git a/include/simgrid/msg.h b/include/simgrid/msg.h
index 98580d402a..d5bd22ed36 100644
--- a/include/simgrid/msg.h
+++ b/include/simgrid/msg.h
@@ -7,7 +7,6 @@
 #ifndef MSG_H
 #define MSG_H
 
-#include "xbt/lib.h"
 #include "simgrid/forward.h"
 #include "simgrid/simix.h"
 
@@ -58,46 +57,16 @@ typedef struct msg_task *msg_task_t;
 typedef msg_host_t msg_vm_t;
 
 /* ******************************** File ************************************ */
-
-typedef struct simdata_file* simdata_file_t;
-
-typedef struct msg_file_priv {
-  char *fullpath;
-  sg_size_t size;
-  char* mount_point;
-  char* storageId;
-  char* storage_type;
-  int desc_id;
-  void *data;
-  simdata_file_t simdata;
-} s_msg_file_priv_t;
-
-typedef struct msg_file_priv* msg_file_t;
+typedef sg_file_t msg_file_t;
 
 /* ******************************** Storage ************************************ */
-/* TODO: PV: to comment */
-
-extern int MSG_STORAGE_LEVEL;
 
 /** @brief Storage datatype.
  *  @ingroup msg_storage_management
  *
  *  You should consider this as an opaque object.
  */
-typedef xbt_dictelm_t msg_storage_t;
-
-struct msg_storage_priv  {
-  const char* name;
-  const char* hostname;
-  sg_size_t size;
-  void* data;
-};
-typedef struct msg_storage_priv  s_msg_storage_priv_t;
-typedef struct msg_storage_priv* msg_storage_priv_t;
-
-static inline msg_storage_priv_t MSG_storage_priv(msg_storage_t storage){
-  return (msg_storage_priv_t )xbt_lib_get_level(storage, MSG_STORAGE_LEVEL);
-}
+typedef sg_storage_t msg_storage_t;
 
 /**
  * \brief @brief Communication action.
diff --git a/include/simgrid/plugins/energy.h b/include/simgrid/plugins/energy.h
index 06d1de8fd4..bc1da109dc 100644
--- a/include/simgrid/plugins/energy.h
+++ b/include/simgrid/plugins/energy.h
@@ -13,6 +13,7 @@
 SG_BEGIN_DECL()
 
 XBT_PUBLIC(void) sg_host_energy_plugin_init();
+XBT_PUBLIC(void) sg_host_energy_update_all();
 XBT_PUBLIC(double) sg_host_get_consumed_energy(sg_host_t host);
 XBT_PUBLIC(double) sg_host_get_wattmin_at(sg_host_t host, int pstate);
 XBT_PUBLIC(double) sg_host_get_wattmax_at(sg_host_t host, int pstate);
diff --git a/include/simgrid/s4u/Activity.hpp b/include/simgrid/s4u/Activity.hpp
index aaa9e38dd4..57ab0371c0 100644
--- a/include/simgrid/s4u/Activity.hpp
+++ b/include/simgrid/s4u/Activity.hpp
@@ -61,7 +61,7 @@ public:
   void *getUserData() { return userData_; }
 
 private:
-  simgrid::kernel::activity::ActivityImpl *pimpl_ = nullptr;
+  simgrid::kernel::activity::ActivityImplPtr pimpl_ = nullptr;
   e_s4u_activity_state_t state_ = inited;
   double remains_ = 0;
   void *userData_ = nullptr;
diff --git a/include/simgrid/s4u/Actor.hpp b/include/simgrid/s4u/Actor.hpp
index 5517dc62bd..8cafe9fc36 100644
--- a/include/simgrid/s4u/Actor.hpp
+++ b/include/simgrid/s4u/Actor.hpp
@@ -175,6 +175,12 @@ public:
    */
   static ActorPtr createActor(const char* name, s4u::Host* host, std::function<void()> code);
 
+  static ActorPtr createActor(const char* name, s4u::Host* host, std::function<void(std::vector<std::string>*)> code,
+                              std::vector<std::string>* args)
+  {
+    return createActor(name, host, [code](std::vector<std::string>* args) { code(args); }, args);
+  }
+
   /** Create an actor using code
    *
    *  Using this constructor, move-only type can be used. The consequence is
@@ -256,7 +262,7 @@ public:
    * This blocks the calling actor until the actor on which we call join() is terminated
    */
   void join();
-  
+
   // Static methods on all actors:
 
   /** Ask kindly to all actors to die. Only the issuer will survive. */
diff --git a/include/simgrid/s4u/Comm.hpp b/include/simgrid/s4u/Comm.hpp
index 45db6eccb7..697f259089 100644
--- a/include/simgrid/s4u/Comm.hpp
+++ b/include/simgrid/s4u/Comm.hpp
@@ -26,18 +26,22 @@ public:
 
   virtual ~Comm();
 
-  /*! take a range of s4u::Comm* (last excluded) and return when one of them is finished. The return value is an
+  /*! take a range of s4u::CommPtr (last excluded) and return when one of them is finished. The return value is an
    * iterator on the finished Comms. */
   template <class I> static I wait_any(I first, I last)
   {
     // Map to dynar<Synchro*>:
-    xbt_dynar_t comms = xbt_dynar_new(sizeof(simgrid::kernel::activity::ActivityImpl*), NULL);
+    xbt_dynar_t comms = xbt_dynar_new(sizeof(simgrid::kernel::activity::ActivityImpl*), [](void*ptr){
+      intrusive_ptr_release(*(simgrid::kernel::activity::ActivityImpl**)ptr);
+    });
     for (I iter = first; iter != last; iter++) {
-      Comm& comm = **iter;
-      if (comm.state_ == inited)
-        comm.start();
-      xbt_assert(comm.state_ == started);
-      xbt_dynar_push_as(comms, simgrid::kernel::activity::ActivityImpl*, comm.pimpl_);
+      CommPtr comm = *iter;
+      if (comm->state_ == inited)
+        comm->start();
+      xbt_assert(comm->state_ == started);
+      simgrid::kernel::activity::ActivityImpl* ptr = comm->pimpl_.get();
+      intrusive_ptr_add_ref(ptr);
+      xbt_dynar_push_as(comms, simgrid::kernel::activity::ActivityImpl*, ptr);
     }
     // Call the underlying simcall:
     int idx = simcall_comm_waitany(comms, -1);
@@ -50,17 +54,21 @@ public:
     (*res)->state_ = finished;
     return res;
   }
-  /*! Same as wait_any, but with a timeout. If wait_any_for return because of the timeout last is returned.*/
+  /*! Same as wait_any, but with a timeout. If the timeout occurs, parameter last is returned.*/
   template <class I> static I wait_any_for(I first, I last, double timeout)
   {
     // Map to dynar<Synchro*>:
-    xbt_dynar_t comms = xbt_dynar_new(sizeof(simgrid::kernel::activity::ActivityImpl*), NULL);
+    xbt_dynar_t comms = xbt_dynar_new(sizeof(simgrid::kernel::activity::ActivityImpl*), [](void*ptr){
+      intrusive_ptr_release(*(simgrid::kernel::activity::ActivityImpl**)ptr);
+    });
     for (I iter = first; iter != last; iter++) {
-      Comm& comm = **iter;
-      if (comm.state_ == inited)
-        comm.start();
-      xbt_assert(comm.state_ == started);
-      xbt_dynar_push_as(comms, simgrid::kernel::activity::ActivityImpl*, comm.pimpl_);
+      CommPtr comm = *iter;
+      if (comm->state_ == inited)
+        comm->start();
+      xbt_assert(comm->state_ == started);
+      simgrid::kernel::activity::ActivityImpl* ptr = comm->pimpl_.get();
+      intrusive_ptr_add_ref(ptr);
+      xbt_dynar_push_as(comms, simgrid::kernel::activity::ActivityImpl*, ptr);
     }
     // Call the underlying simcall:
     int idx = simcall_comm_waitany(comms, timeout);
diff --git a/include/simgrid/s4u/File.hpp b/include/simgrid/s4u/File.hpp
index d9d0080ab1..278081f2db 100644
--- a/include/simgrid/s4u/File.hpp
+++ b/include/simgrid/s4u/File.hpp
@@ -8,13 +8,12 @@
 
 #include <xbt/base.h>
 
+#include "src/surf/StorageImpl.hpp"
 #include <simgrid/simix.h>
 
 namespace simgrid {
 namespace s4u {
 
-class Storage;
-
 /** @brief A simulated file
  *
  * Used to simulate the time it takes to access to a file, but does not really store any information.
@@ -71,6 +70,10 @@ public:
   XBT_PUBLIC(msg_error_t) MSG_file_rcopy(msg_file_t fd, msg_host_t host, const char* fullpath);
   XBT_PUBLIC(msg_error_t) MSG_file_rmove(msg_file_t fd, msg_host_t host, const char* fullpath);
   */
+  char* storage_type;
+  char* storageId;
+  char* mount_point;
+  int desc_id = 0;
 
 private:
   smx_file_t pimpl_ = nullptr;
diff --git a/include/simgrid/s4u/Host.hpp b/include/simgrid/s4u/Host.hpp
index ab1dd5e1a9..6f9d6ce59e 100644
--- a/include/simgrid/s4u/Host.hpp
+++ b/include/simgrid/s4u/Host.hpp
@@ -145,7 +145,7 @@ public class Host {
    * The external load (coming from an availability trace) is not taken in account.
    *
    * @return      The number of tasks currently running on a host.
-   */ 
+   */
   public native int getLoad();
 
 
diff --git a/include/simgrid/s4u/Mailbox.hpp b/include/simgrid/s4u/Mailbox.hpp
index 62c36fdd59..fdd13ddcf1 100644
--- a/include/simgrid/s4u/Mailbox.hpp
+++ b/include/simgrid/s4u/Mailbox.hpp
@@ -26,7 +26,7 @@ namespace s4u {
  * Rendez-vous point for network communications, similar to URLs on
  * which you could post and retrieve data. Actually, the mailboxes are
  * not involved in the communication once it starts, but only to find
- * the contact with which you want to communicate. 
+ * the contact with which you want to communicate.
 
  * Here are some mechanisms similar to the mailbox in other
  * communication systems: The phone number, which allows the caller to
@@ -93,8 +93,8 @@ namespace s4u {
  * starts to flow as soon as the sender posts it, even if the receiver
  * did not post its recv() yet. This can obviously lead to bad
  * simulation timings, as the simulated communications do not start at
- * the exact same time than the real ones. 
- * 
+ * the exact same time than the real ones.
+ *
  * If the simulation timings are very important to you, you can
  * declare a specific receiver to a given mailbox (with the function
  * setReceiver()). That way, any send() posted to that mailbox will
diff --git a/include/simgrid/s4u/Mutex.hpp b/include/simgrid/s4u/Mutex.hpp
index 0d1f2277a0..9e8e317e72 100644
--- a/include/simgrid/s4u/Mutex.hpp
+++ b/include/simgrid/s4u/Mutex.hpp
@@ -23,13 +23,13 @@ class ConditionVariable;
  *  @ingroup s4u_api
  *
  * It is strictly impossible to use a real mutex, such as
- * [std::mutex](http://en.cppreference.com/w/cpp/thread/mutex)
- * or [pthread_mutex_t](http://pubs.opengroup.org/onlinepubs/007908775/xsh/pthread_mutex_lock.html),
+ * <a href="http://en.cppreference.com/w/cpp/thread/mutex">std::mutex</a>
+ * or <a href="http://pubs.opengroup.org/onlinepubs/007908775/xsh/pthread_mutex_lock.html">pthread_mutex_t</a>,
  * because it would block the whole simulation.
  * Instead, you should use the present class, that is a drop-in replacement of
- * [std::mutex](http://en.cppreference.com/w/cpp/thread/mutex).
+ * <a href="http://en.cppreference.com/w/cpp/thread/mutex>std::mutex</a>.
  *
- * As for any S4U object, Mutexes are using the @ref "RAII idiom" s4u_raii for memory management.
+ * As for any S4U object, Mutexes are using the @ref s4u_raii "RAII idiom" for memory management.
  * Use createMutex() to get a ::MutexPtr to a newly created mutex and only manipulate ::MutexPtr.
  *
  */
diff --git a/include/simgrid/s4u/NetZone.hpp b/include/simgrid/s4u/NetZone.hpp
index b0accf1e76..fe9d878cbb 100644
--- a/include/simgrid/s4u/NetZone.hpp
+++ b/include/simgrid/s4u/NetZone.hpp
@@ -67,6 +67,8 @@ public:
                                    kernel::routing::NetPoint* gw_src, kernel::routing::NetPoint* gw_dst,
                                    std::vector<surf::LinkImpl*>* link_list)>
       onRouteCreation;
+  static simgrid::xbt::signal<void(NetZone&)> onCreation;
+  static simgrid::xbt::signal<void(NetZone&)> onSeal;
 
 protected:
   std::vector<kernel::routing::NetPoint*>
diff --git a/include/simgrid/s4u/Storage.hpp b/include/simgrid/s4u/Storage.hpp
index 305bbba87d..12bfe853a3 100644
--- a/include/simgrid/s4u/Storage.hpp
+++ b/include/simgrid/s4u/Storage.hpp
@@ -17,17 +17,18 @@
 namespace simgrid {
 namespace s4u {
 
+std::map<std::string, Storage*>* allStorages();
+
 XBT_PUBLIC_CLASS Storage
 {
   friend s4u::Engine;
-
-  Storage(std::string name, smx_storage_t inferior);
+  friend simgrid::surf::StorageImpl;
 
 public:
-  Storage() = default;
-  virtual ~Storage();
+  explicit Storage(surf::StorageImpl * pimpl) : pimpl_(pimpl) {}
+  virtual ~Storage() = default;
   /** Retrieve a Storage by its name. It must exist in the platform file */
-  static Storage& byName(const char* name);
+  static Storage* byName(const char* name);
   const char* name();
   const char* host();
   sg_size_t sizeFree();
@@ -38,22 +39,21 @@ public:
   const char* property(const char* key);
   void setProperty(const char* key, char* value);
   std::map<std::string, sg_size_t*>* content();
-  std::unordered_map<std::string, Storage*>* allStorages();
-
-protected:
-  smx_storage_t inferior();
 
 public:
   void setUserdata(void* data) { userdata_ = data; }
   void* userdata() { return userdata_; }
 
-private:
-  static std::unordered_map<std::string, Storage*>* storages_;
+  /* The signals */
+  /** @brief Callback signal fired when a new Link is created */
+  static simgrid::xbt::signal<void(s4u::Storage&)> onCreation;
+
+  /** @brief Callback signal fired when a Link is destroyed */
+  static simgrid::xbt::signal<void(s4u::Storage&)> onDestruction;
 
-  std::string hostname_;
+private:
   std::string name_;
-  sg_size_t size_      = 0;
-  smx_storage_t pimpl_ = nullptr;
+  surf::StorageImpl* const pimpl_ = nullptr;
   void* userdata_      = nullptr;
 };
 
diff --git a/include/simgrid/s4u/forward.hpp b/include/simgrid/s4u/forward.hpp
index cc29359308..237a5ebe33 100644
--- a/include/simgrid/s4u/forward.hpp
+++ b/include/simgrid/s4u/forward.hpp
@@ -26,6 +26,7 @@ using MailboxPtr = boost::intrusive_ptr<Mailbox>;
 class Mutex;
 class NetZone;
 
+class File;
 class Storage;
 
 XBT_PUBLIC(void) intrusive_ptr_release(Comm* c);
diff --git a/include/simgrid/simdag.h b/include/simgrid/simdag.h
index b317441903..27efe346a0 100644
--- a/include/simgrid/simdag.h
+++ b/include/simgrid/simdag.h
@@ -121,8 +121,8 @@ XBT_PUBLIC(void) SD_task_schedulel(SD_task_t task, int count, ...);
 
 /** @} */
 
-/** @addtogroup SD_task_dependency_api 
- * 
+/** @addtogroup SD_task_dependency_api
+ *
  *  This section describes the functions for managing the dependencies between the tasks.
  *
  *  @see SD_task_api
diff --git a/include/simgrid/simix.h b/include/simgrid/simix.h
index 64f9aac67b..2e98432974 100644
--- a/include/simgrid/simix.h
+++ b/include/simgrid/simix.h
@@ -95,9 +95,6 @@ typedef struct s_smx_sem *smx_sem_t;
 /********************************** File *************************************/
 typedef struct s_smx_file *smx_file_t;
 
-/********************************** Storage *************************************/
-typedef xbt_dictelm_t smx_storage_t;
-
 /* ****************************** Process *********************************** */
 
 typedef enum {
@@ -215,6 +212,8 @@ XBT_PUBLIC(int) SIMIX_process_has_pending_comms(smx_actor_t process);
 XBT_PUBLIC(void) SIMIX_process_on_exit_runall(smx_actor_t process);
 XBT_PUBLIC(void) SIMIX_process_on_exit(smx_actor_t process, int_f_pvoid_pvoid_t fun, void *data);
 
+SG_END_DECL()
+
 /****************************** Communication *********************************/
 XBT_PUBLIC(void) SIMIX_comm_set_copy_data_callback(void (*callback) (smx_activity_t, void*, size_t));
 XBT_PUBLIC(void) SIMIX_comm_copy_pointer_callback(smx_activity_t comm, void* buff, size_t buff_size);
@@ -249,6 +248,7 @@ XBT_PUBLIC(void) simcall_execution_set_bound(smx_activity_t execution, double bo
 XBT_PUBLIC(e_smx_state_t) simcall_execution_wait(smx_activity_t execution);
 
 /**************************** Process simcalls ********************************/
+SG_BEGIN_DECL()
 /* Constructor and Destructor */
 XBT_PUBLIC(smx_actor_t)
 simcall_process_create(const char* name, xbt_main_func_t code, void* data, sg_host_t host, int argc, char** argv,
@@ -278,6 +278,7 @@ XBT_PUBLIC(smx_actor_t) simcall_process_restart(smx_actor_t process);
 XBT_PUBLIC(void) simcall_process_join(smx_actor_t process, double timeout);
 /* Sleep control */
 XBT_PUBLIC(e_smx_state_t) simcall_process_sleep(double duration);
+SG_END_DECL()
 
 /************************** Comunication simcalls *****************************/
 
@@ -323,6 +324,7 @@ XBT_PUBLIC(int) simcall_comm_testany(smx_activity_t* comms, size_t count);
 XBT_PUBLIC(void) simcall_set_category(smx_activity_t synchro, const char *category);
 
 /************************** Synchro simcalls **********************************/
+SG_BEGIN_DECL()
 XBT_PUBLIC(smx_mutex_t) simcall_mutex_init();
 XBT_PUBLIC(smx_mutex_t) SIMIX_mutex_ref(smx_mutex_t mutex);
 XBT_PUBLIC(void) SIMIX_mutex_unref(smx_mutex_t mutex);
@@ -359,10 +361,9 @@ XBT_PUBLIC(xbt_dynar_t) simcall_file_get_info(smx_file_t fd);
 XBT_PUBLIC(sg_size_t) simcall_file_tell(smx_file_t fd);
 XBT_PUBLIC(int) simcall_file_seek(smx_file_t fd, sg_offset_t offset, int origin);
 XBT_PUBLIC(int) simcall_file_move(smx_file_t fd, const char* fullpath);
-/*****************************   Storage   **********************************/
-XBT_PUBLIC(xbt_dict_t) simcall_storage_get_properties(smx_storage_t storage);
 /************************** MC simcalls   **********************************/
 XBT_PUBLIC(int) simcall_mc_random(int min, int max);
 
 SG_END_DECL()
+
 #endif                          /* _SIMIX_SIMIX_H */
diff --git a/include/smpi/smpi.h b/include/smpi/smpi.h
index 11c67ee7de..39b548347f 100644
--- a/include/smpi/smpi.h
+++ b/include/smpi/smpi.h
@@ -561,8 +561,8 @@ MPI_CALL(XBT_PUBLIC(int), MPI_Put,( void *origin_addr, int origin_count, MPI_Dat
     MPI_Aint target_disp, int target_count, MPI_Datatype target_datatype, MPI_Win win));
 MPI_CALL(XBT_PUBLIC(int), MPI_Accumulate,( void *origin_addr, int origin_count, MPI_Datatype origin_datatype,
     int target_rank, MPI_Aint target_disp, int target_count, MPI_Datatype target_datatype, MPI_Op op, MPI_Win win));
-MPI_CALL(XBT_PUBLIC(int), MPI_Get_accumulate,( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, 
-    void* result_addr, int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, 
+MPI_CALL(XBT_PUBLIC(int), MPI_Get_accumulate,( void *origin_addr, int origin_count, MPI_Datatype origin_datatype,
+    void* result_addr, int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp,
     int target_count, MPI_Datatype target_datatype, MPI_Op op, MPI_Win win));
 
 MPI_CALL(XBT_PUBLIC(int), MPI_Rget,( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, int target_rank,
@@ -571,8 +571,8 @@ MPI_CALL(XBT_PUBLIC(int), MPI_Rput,( void *origin_addr, int origin_count, MPI_Da
     MPI_Aint target_disp, int target_count, MPI_Datatype target_datatype, MPI_Win win, MPI_Request* request));
 MPI_CALL(XBT_PUBLIC(int), MPI_Raccumulate,( void *origin_addr, int origin_count, MPI_Datatype origin_datatype,
     int target_rank, MPI_Aint target_disp, int target_count, MPI_Datatype target_datatype, MPI_Op op, MPI_Win win, MPI_Request* request));
-MPI_CALL(XBT_PUBLIC(int), MPI_Rget_accumulate,( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, 
-    void* result_addr, int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, 
+MPI_CALL(XBT_PUBLIC(int), MPI_Rget_accumulate,( void *origin_addr, int origin_count, MPI_Datatype origin_datatype,
+    void* result_addr, int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp,
     int target_count, MPI_Datatype target_datatype, MPI_Op op, MPI_Win win, MPI_Request* request));
 
 MPI_CALL(XBT_PUBLIC(int), MPI_Fetch_and_op,( void *origin_addr, void* result_addr, MPI_Datatype datatype,
@@ -603,9 +603,9 @@ typedef void MPI_Handler_function(MPI_Comm*, int*, ...);
 typedef void* MPI_Errhandler;
 
 typedef void MPI_Comm_errhandler_function(MPI_Comm *, int *, ...);
-typedef int MPI_Grequest_query_function(void *extra_state, MPI_Status *status); 
-typedef int MPI_Grequest_free_function(void *extra_state); 
-typedef int MPI_Grequest_cancel_function(void *extra_state, int complete); 
+typedef int MPI_Grequest_query_function(void *extra_state, MPI_Status *status);
+typedef int MPI_Grequest_free_function(void *extra_state);
+typedef int MPI_Grequest_cancel_function(void *extra_state, int complete);
 #define MPI_DUP_FN MPI_Comm_dup
 
 #define MPI_WIN_DUP_FN ((MPI_Win_copy_attr_function*)MPI_DUP_FN)
@@ -723,7 +723,7 @@ MPI_CALL(XBT_PUBLIC(int), MPI_Win_set_errhandler, (MPI_Win win, MPI_Errhandler e
 MPI_CALL(XBT_PUBLIC(int), MPI_Type_get_envelope,(MPI_Datatype datatype,int *num_integers,int *num_addresses,
                             int *num_datatypes, int *combiner));
 MPI_CALL(XBT_PUBLIC(int), MPI_Type_get_contents,(MPI_Datatype datatype, int max_integers, int max_addresses,
-                            int max_datatypes, int* array_of_integers, MPI_Aint* array_of_addresses, 
+                            int max_datatypes, int* array_of_integers, MPI_Aint* array_of_addresses,
                             MPI_Datatype *array_of_datatypes));
 MPI_CALL(XBT_PUBLIC(int), MPI_Type_create_darray,(int size, int rank, int ndims, int* array_of_gsizes,
                             int* array_of_distribs, int* array_of_dargs, int* array_of_psizes,
@@ -866,14 +866,14 @@ XBT_PUBLIC(void) smpi_sample_1(int global, const char *file, int line, int iters
 XBT_PUBLIC(int) smpi_sample_2(int global, const char *file, int line);
 XBT_PUBLIC(void) smpi_sample_3(int global, const char *file, int line);
 
-/** 
- * Need a public setter for SMPI copy_callback function, so users can define 
+/**
+ * Need a public setter for SMPI copy_callback function, so users can define
  * their own while still using default SIMIX_copy_callback for MSG copies.
  */
 XBT_PUBLIC(void) smpi_comm_set_copy_data_callback(void (*callback) (smx_activity_t, void*, size_t));
 
 
-/** 
+/**
  * Functions for call location tracing. These functions will be
  * called from the user's application! (With the __FILE__ and __LINE__ values
  * passed as parameters.)
@@ -961,7 +961,7 @@ static void __attribute__((destructor)) __postfini_##name(void) { \
 
 #define SMPI_VARGET_GLOBAL(name) name[smpi_process_index()]
 
-/** 
+/**
  * This is used for the old privatization method, i.e., on old
  * machines that do not yet support privatization via mmap
  */
diff --git a/include/surf/surf_routing.h b/include/surf/surf_routing.h
deleted file mode 100644
index 2d19df1798..0000000000
--- a/include/surf/surf_routing.h
+++ /dev/null
@@ -1,21 +0,0 @@
-/* Copyright (c) 2004-2015. The SimGrid Team.
- * All rights reserved.                                                     */
-
-/* This program is free software; you can redistribute it and/or modify it
- * under the terms of the license (GNU LGPL) which comes with this package. */
-
-#ifndef SURF_SURF_ROUTING_H
-#define SURF_SURF_ROUTING_H
-
-#include "xbt/lib.h"
-
-SG_BEGIN_DECL()
-
-// FIXME: this header file should die
-
-XBT_PUBLIC_DATA(xbt_lib_t) storage_lib;
-XBT_PUBLIC_DATA(int) SURF_STORAGE_LEVEL;  // Surf storage level
-
-SG_END_DECL()
-
-#endif                          /* _SURF_SURF_H */
diff --git a/include/xbt/automaton.h b/include/xbt/automaton.h
index 1ba708f6ed..2d0d94e837 100644
--- a/include/xbt/automaton.h
+++ b/include/xbt/automaton.h
@@ -78,7 +78,7 @@ XBT_PUBLIC(void) xbt_automaton_transition_set_source(xbt_automaton_transition_t
 XBT_PUBLIC(void) xbt_automaton_transition_set_destination(xbt_automaton_transition_t t, xbt_automaton_state_t dst);
 XBT_PUBLIC(xbt_dynar_t) xbt_automaton_state_get_out_transitions(xbt_automaton_state_t s);
 XBT_PUBLIC(xbt_dynar_t) xbt_automaton_state_get_in_transitions(xbt_automaton_state_t s);
-XBT_PUBLIC(xbt_automaton_state_t) xbt_automaton_state_exists(xbt_automaton_t a, char *id); 
+XBT_PUBLIC(xbt_automaton_state_t) xbt_automaton_state_exists(xbt_automaton_t a, char *id);
 XBT_PUBLIC(void) xbt_automaton_display(xbt_automaton_t a);
 XBT_PUBLIC(void) xbt_automaton_exp_label_display(xbt_automaton_exp_label_t l);
 
diff --git a/include/xbt/base.h b/include/xbt/base.h
index 108003c381..68ea79ea08 100644
--- a/include/xbt/base.h
+++ b/include/xbt/base.h
@@ -153,7 +153,7 @@
 #  define XBT_PUBLIC_CLASS            class __declspec(dllimport)
 #  define XBT_PRIVATE
 
-#elif defined(__ELF__) 
+#elif defined(__ELF__)
 #  define XBT_PUBLIC(type)            __attribute__((visibility("default"))) type
 #  define XBT_EXPORT_NO_IMPORT(type)  __attribute__((visibility("default"))) type
 #  define XBT_IMPORT_NO_EXPORT(type)  __attribute__((visibility("default"))) type
diff --git a/include/xbt/cunit.h b/include/xbt/cunit.h
index cf7a3ec212..c0064534dc 100644
--- a/include/xbt/cunit.h
+++ b/include/xbt/cunit.h
@@ -36,10 +36,10 @@ XBT_PUBLIC(void) xbt_test_suite_push(xbt_test_suite_t suite, const char *name, t
  * It is a coma (,) separated list of directives. They are applied from left to right.
  *
  * Each of them of form:
- * 
+ *
  * [-|+]suitename[:unitname[:testname]]
- * 
- * * First char: 
+ *
+ * * First char:
  *   if it's a '-', the directive disables something
  *   if it's a '+', the directive enables something
  *   By default, everything is enabled, but you can disable a suite and reenable some parts
@@ -54,24 +54,24 @@ XBT_PUBLIC(void) xbt_test_dump(char *selection);
 /* Cleanup the mess */
 XBT_PUBLIC(void) xbt_test_exit();
 
-/** 
+/**
  * @addtogroup XBT_cunit
  * @brief Unit testing implementation (see @ref inside_tests_add_units)
- *  
+ *
  * This module is mainly intended to allow the tests of SimGrid itself and may lack the level of genericity that you
  * would expect as a user. Only use it in external projects at your own risk (but it works rather well for us). We play
  * with the idea of migrating to an external solution for our unit tests, possibly offering more features, but having
  * absolutely no dependencies is a nice feature of SimGrid (and this code is sufficient to cover our needs, actually,
  * so why should we bother switching?)
- * 
+ *
  * Unit testing is not intended to write integration tests.
  * Please refer to \ref inside_tests_add_integration for that instead.
  *
- * @{ 
+ * @{
  */
 /** @brief Provide information about the suite declared in this file
  *  @hideinitializer
- * 
+ *
  * Actually, this macro is only used by the script extracting the test units, but that should be transparent for you.
  *
  * @param suite_name the short name of this suite, to be used in the --tests argument of testall afterward. Avoid
@@ -101,15 +101,15 @@ XBT_PUBLIC(void) xbt_test_exit();
 XBT_PUBLIC(void) _xbt_test_add(const char *file, int line, const char *fmt, ...) XBT_ATTRIB_PRINTF(3, 4);
 XBT_PUBLIC(void) _xbt_test_fail(const char *file, int line, const char *fmt, ...) XBT_ATTRIB_PRINTF(3, 4);
 XBT_PUBLIC(void) _xbt_test_log(const char *file, int line, const char *fmt, ...) XBT_ATTRIB_PRINTF(3, 4);
-/** @brief Declare that a new test begins (printf-like parameters, describing the test) 
+/** @brief Declare that a new test begins (printf-like parameters, describing the test)
  *  @hideinitializer */
 #define xbt_test_add(...)       _xbt_test_add(__FILE__, __LINE__, __VA_ARGS__)
-/** @brief Declare that the lastly started test failed (printf-like parameters, describing failure cause) 
+/** @brief Declare that the lastly started test failed (printf-like parameters, describing failure cause)
  *  @hideinitializer */
 #define xbt_test_fail(...)      _xbt_test_fail(__FILE__, __LINE__, __VA_ARGS__)
 /** @brief The lastly started test is actually an assert
- *  @hideinitializer 
- * 
+ *  @hideinitializer
+ *
  * - If provided a uniq parameter, this is assumed to be a condition that is expected to be true
  * - If provided more parameters, the first one is a condition, and the other ones are printf-like arguments that are
  *   to be displayed when the condition fails.
diff --git a/include/xbt/dynar.h b/include/xbt/dynar.h
index 747e3185a6..86751e859d 100644
--- a/include/xbt/dynar.h
+++ b/include/xbt/dynar.h
@@ -19,7 +19,7 @@ SG_BEGIN_DECL()
   * @brief DynArr are dynamically sized vector which may contain any type of variables.
   *
   * These are the SimGrid version of the dynamically size arrays, which all C programmer recode one day or another.
-  *  
+  *
   * For performance concerns, the content of DynArr must be homogeneous (in contrary to dictionnaries -- see the
   * \ref XBT_dict section). You thus have to provide the function which will be used to free the content at
   * structure creation (of type void_f_pvoid_t).
@@ -40,7 +40,7 @@ SG_BEGIN_DECL()
   * \until xbt_dynar_free
   *
   * \section XBT_dynar_exptr Example with pointed data
-  * 
+  *
   * \skip test_dynar_string
   * \skip dynar_t
   * \until s2
@@ -147,7 +147,7 @@ XBT_PUBLIC(void *) xbt_dynar_pop_ptr(xbt_dynar_t const dynar);
  *  @{
  */
 
-  /** @brief Quick retrieval of scalar content 
+  /** @brief Quick retrieval of scalar content
    *  @hideinitializer */
 #  define xbt_dynar_get_as(dynar,idx,type) \
           (*(type*)xbt_dynar_get_ptr((dynar),(idx)))
@@ -155,19 +155,19 @@ XBT_PUBLIC(void *) xbt_dynar_pop_ptr(xbt_dynar_t const dynar);
  *  @hideinitializer */
 #  define xbt_dynar_set_as(dynar,idx,type,val) \
          (*(type*)xbt_dynar_set_at_ptr((dynar),(idx))) = val
-  /** @brief Quick retrieval of scalar content 
+  /** @brief Quick retrieval of scalar content
    *  @hideinitializer */
 #  define xbt_dynar_getlast_as(dynar,type) \
           (*(type*)xbt_dynar_get_ptr((dynar),xbt_dynar_length(dynar)-1))
-  /** @brief Quick retrieval of scalar content 
+  /** @brief Quick retrieval of scalar content
    *  @hideinitializer */
 #  define xbt_dynar_getfirst_as(dynar,type) \
           (*(type*)xbt_dynar_get_ptr((dynar),0))
-  /** @brief Quick insertion of scalar content 
+  /** @brief Quick insertion of scalar content
    *  @hideinitializer */
 #  define xbt_dynar_insert_at_as(dynar,idx,type,value) \
           *(type*)xbt_dynar_insert_at_ptr(dynar,idx)=value
-  /** @brief Quick insertion of scalar content 
+  /** @brief Quick insertion of scalar content
    *  @hideinitializer */
 #  define xbt_dynar_push_as(dynar,type,value) \
           *(type*)xbt_dynar_push_ptr(dynar)=value
@@ -190,7 +190,7 @@ XBT_PUBLIC(void *) xbt_dynar_pop_ptr(xbt_dynar_t const dynar);
 
 XBT_PUBLIC(void) xbt_dynar_cursor_rm(xbt_dynar_t dynar, unsigned int *const cursor);
 
-/* 
+/*
  * \warning DO NOT USE THIS STRUCTURE DIRECTLY! Instead, use the public interface:
  *          This was made public to allow:
  *           - the inlining of the foreach elements
@@ -222,8 +222,8 @@ static inline int _xbt_dynar_cursor_get(const xbt_dynar_t dynar, unsigned int id
   return TRUE;
 }
 
-/** @brief Iterates over the whole dynar. 
- * 
+/** @brief Iterates over the whole dynar.
+ *
  *  @param _dynar what to iterate over
  *  @param _cursor an integer used as cursor
  *  @param _data
@@ -238,7 +238,7 @@ xbt_dynar_foreach (dyn,cpt,str) {
   printf("Seen %s\n",str);
 }
 \endcode
- * 
+ *
  * Note that underneath, that's a simple for loop with no real black  magic involved. It's perfectly safe to interrupt
  * a foreach with a break or a return statement.
  */
diff --git a/include/xbt/exception.hpp b/include/xbt/exception.hpp
index ea18bc08e3..6e801981d2 100644
--- a/include/xbt/exception.hpp
+++ b/include/xbt/exception.hpp
@@ -43,8 +43,7 @@ typedef std::vector<xbt_backtrace_location_t> Backtrace;
 class ThrowPoint {
  public:
   ThrowPoint() = default;
-  ThrowPoint(const char* file, int line, const char* function) :
-    file(file), line(line), function(function) {}
+  explicit ThrowPoint(const char* file, int line, const char* function) : file(file), line(line), function(function) {}
   const char* file = nullptr;
   int line = 0;
   const char* function = nullptr;
@@ -62,7 +61,7 @@ class ThrowPoint {
  *  You are not expected to inherit from it. Instead of you use should
  *  @ref XBT_THROW an exception which will throw a subclass of your original
  *  exception with those additional features.
- * 
+ *
  *  However, you can try `dynamic_cast` an exception to this type in order to
  *  get contextual information about the exception.
  */
diff --git a/include/xbt/lib.h b/include/xbt/lib.h
deleted file mode 100644
index 56a900786a..0000000000
--- a/include/xbt/lib.h
+++ /dev/null
@@ -1,80 +0,0 @@
-/* xbt/lib.h - api to a generic library                                     */
-
-/* Copyright (c) 2011, 2013-2014. The SimGrid Team.
- * All rights reserved.                                                     */
-
-/* This program is free software; you can redistribute it and/or modify it
- * under the terms of the license (GNU LGPL) which comes with this package. */
-
-#ifndef XBT_LIB_H
-#define XBT_LIB_H
-
-#include <xbt/dict.h>
-
-SG_BEGIN_DECL()
-
-/** Container for all the objects of a given type
- *
- *  * each item is identified by a string name/identifier;
- *
- *  * the lib itself is a dictionary from the element id to the element;
- *
- *  * the element itself is represented aby the xbt_dictelm_t;
- *
- *  * the element can store any number of associated facets/data structures (corresponding to the different layers of
- *    SimGrid or its extensions) in ((void**)dictelt->content)[level];
- *
- *  * each level is allocated in the lib with `xbt_lib_add_level`.
- *
- *  <pre>
- *  // Define a collection for the foo objects and two associated facets:
- *  typedef xbt_dictelm_t foo_t;
- *  xbt_lib_t foo_lib = xbt_lib_new();
- *  int BAR_FOO_LEVEL  = xbt_lib_add_level(foo_lib, free_bar);
- *  int AUTH_FOO_LEVEL = xbt_lib_add_level(foo_lib, free_auth);
- *
- *  // Store a bar:
- *  bar_t bar = bar_new();
- *  char* id = bar_name(bar);
- *  xbt_lib_set(id, id, BAR_FOO_LEVEL, bar);
- *
- *  // Find the corresponding foo and the facet again:
- *  foo_t foo = xbt_lib_get_elm_or_null(foo_lib, id);
- *  bar_t bar2 = (bar_t) xbt_lib_get_level(foo, BAR_FOO_LEVEL);
- *  assert(bar == bar2);
- *
- *  // Add authentication facet for the previous object:
- *  auth_t auth = auth_new();
- *  xbt_lib_set(foo_lib, id, AUTH_FOO_LEVEL, auth);
- *  </pre>
- */
-struct s_xbt_lib {
-  xbt_dict_t dict;
-  int levels;
-  void_f_pvoid_t *free_f;       /* This is actually a table */
-};
-typedef struct s_xbt_lib  s_xbt_lib_t;
-typedef struct s_xbt_lib* xbt_lib_t;
-
-#define xbt_lib_cursor_t xbt_dict_cursor_t
-
-XBT_PUBLIC(xbt_lib_t) xbt_lib_new(void);
-XBT_PUBLIC(void) xbt_lib_free(xbt_lib_t * lib);
-XBT_PUBLIC(int) xbt_lib_add_level(xbt_lib_t lib, void_f_pvoid_t free_f);
-XBT_PUBLIC(void) xbt_lib_set(xbt_lib_t lib, const char *name, int level, void *obj);
-XBT_PUBLIC(void) xbt_lib_unset(xbt_lib_t lib, const char *key, int level, int invoke_callback);
-XBT_PUBLIC(void *) xbt_lib_get_or_null(xbt_lib_t lib, const char *name, int level);
-XBT_PUBLIC(xbt_dictelm_t) xbt_lib_get_elm_or_null(xbt_lib_t lib, const char *key);
-XBT_PUBLIC(void *) xbt_lib_get_level(xbt_dictelm_t elm, int level);
-XBT_PUBLIC(void) xbt_lib_remove(xbt_lib_t lib, const char *key);
-
-#define xbt_lib_length(lib) xbt_dict_length((lib)->dict)
-
-/** @def xbt_lib_foreach
-    @hideinitializer */
-#define xbt_lib_foreach(lib, cursor, key, data)         \
-  xbt_dict_foreach((lib)->dict, cursor, key, data)
-
-SG_END_DECL()
-
-#endif                          /* _XBT_LIB_H */
diff --git a/include/xbt/log.h b/include/xbt/log.h
index a6be428cf5..be33e8664f 100644
--- a/include/xbt/log.h
+++ b/include/xbt/log.h
@@ -12,12 +12,12 @@
 
 /** \defgroup XBT_log_cats Existing log categories
  *  \ingroup XBT_log
- *  \brief (automatically extracted) 
+ *  \brief (automatically extracted)
  *
  *  This is the list of all existing log categories in SimGrid.
  *  This list is automatically extracted from the source code by the tools/doxygen/xbt_log_extract_hierarchy.pl utility.
  *
- *  It should thus contain every categories that are defined in the SimGrid library. 
+ *  It should thus contain every categories that are defined in the SimGrid library.
  *  If you want to see the one defined in your code in addition, provide `--help-logs` on the command line of your simulator.
  */
 
@@ -150,14 +150,14 @@ typedef enum {
  * \param desc string describing the purpose of this category
  * \hideinitializer
  *
- * Defines a new subcategory of the parent. 
+ * Defines a new subcategory of the parent.
  */
 #define XBT_LOG_NEW_SUBCATEGORY(catName, parent, desc)    \
   XBT_LOG_EXTERNAL_CATEGORY(parent);                      \
   XBT_LOG_NEW_SUBCATEGORY_helper(catName, parent, desc)   \
 
 /**
- * \ingroup XBT_log  
+ * \ingroup XBT_log
  * \param catName name of new category
  * \param desc string describing the purpose of this category
  * \hideinitializer
@@ -168,7 +168,7 @@ typedef enum {
    XBT_LOG_NEW_SUBCATEGORY_helper(catName, XBT_LOG_ROOT_CAT, desc)
 
 /**
- * \ingroup XBT_log  
+ * \ingroup XBT_log
  * \param cname name of the cat
  * \hideinitializer
  *
@@ -183,7 +183,7 @@ typedef enum {
 #endif
 
 /**
- * \ingroup XBT_log  
+ * \ingroup XBT_log
  * \param cname name of the cat
  * \param desc string describing the purpose of this category
  * \hideinitializer
@@ -196,7 +196,7 @@ typedef enum {
     XBT_LOG_DEFAULT_CATEGORY(cname)
 
 /**
- * \ingroup XBT_log  
+ * \ingroup XBT_log
  * \param cname name of the cat
  * \param parent name of the parent
  * \param desc string describing the purpose of this category
@@ -210,7 +210,7 @@ typedef enum {
     XBT_LOG_DEFAULT_CATEGORY(cname)
 
 /**
- * \ingroup XBT_log  
+ * \ingroup XBT_log
  * \param cname name of the cat
  * \hideinitializer
  *
@@ -283,7 +283,7 @@ struct xbt_log_event_s {
 XBT_PUBLIC(void) xbt_log_threshold_set(xbt_log_category_t cat, e_xbt_log_priority_t thresholdPriority);
 
 /**
- * \ingroup XBT_log_implem  
+ * \ingroup XBT_log_implem
  * \param cat the category (not only its name, but the variable)
  * \param app the appender
  *
@@ -291,7 +291,7 @@ XBT_PUBLIC(void) xbt_log_threshold_set(xbt_log_category_t cat, e_xbt_log_priorit
  */
 XBT_PUBLIC(void) xbt_log_appender_set(xbt_log_category_t cat, xbt_log_appender_t app);
 /**
- * \ingroup XBT_log_implem  
+ * \ingroup XBT_log_implem
  * \param cat the category (not only its name, but the variable)
  * \param lay the layout
  *
@@ -300,7 +300,7 @@ XBT_PUBLIC(void) xbt_log_appender_set(xbt_log_category_t cat, xbt_log_appender_t
 XBT_PUBLIC(void) xbt_log_layout_set(xbt_log_category_t cat, xbt_log_layout_t lay);
 
 /**
- * \ingroup XBT_log_implem  
+ * \ingroup XBT_log_implem
  * \param cat the category (not only its name, but the variable)
  * \param additivity whether logging actions must be passed to parent.
  *
@@ -309,7 +309,7 @@ XBT_PUBLIC(void) xbt_log_layout_set(xbt_log_category_t cat, xbt_log_layout_t lay
  */
 XBT_PUBLIC(void) xbt_log_additivity_set(xbt_log_category_t cat, int additivity);
 
-/** @brief create a new simple layout 
+/** @brief create a new simple layout
  *
  * This layout is not as flexible as the pattern one
  */
@@ -343,7 +343,7 @@ extern xbt_log_layout_t xbt_log_default_layout;
 /* ********************** */
 
 /**
- * \ingroup XBT_log 
+ * \ingroup XBT_log
  * \param catName name of the category
  * \param priority minimal priority to be enabled to return true (must be #e_xbt_log_priority_t)
  * \hideinitializer
diff --git a/include/xbt/mallocator.h b/include/xbt/mallocator.h
index e607e10ac2..d0b0f36130 100644
--- a/include/xbt/mallocator.h
+++ b/include/xbt/mallocator.h
@@ -15,7 +15,7 @@ SG_BEGIN_DECL()
 
 /** @addtogroup XBT_mallocator
  *  @brief The mallocator system
- * 
+ *
  *  This section describes the API to a mallocator.
  *  A mallocator allows you to recycle the objects you don't need anymore  instead of freeing them. A mallocator is a
  *  stack which stores the unused objects  or a given type. If you often need to malloc() / free() objects of a certain
diff --git a/include/xbt/parmap.h b/include/xbt/parmap.h
index 89a8434e65..905b12c16f 100644
--- a/include/xbt/parmap.h
+++ b/include/xbt/parmap.h
@@ -16,7 +16,7 @@
 SG_BEGIN_DECL()
 
 /** \addtogroup XBT_parmap
-  * \ingroup XBT_misc  
+  * \ingroup XBT_misc
   * \brief Parallel map.
   *
   * A function is applied to all elements of a dynar in parallel with n worker threads.
diff --git a/include/xbt/string.hpp b/include/xbt/string.hpp
index 9343ce600d..cafddcb5b8 100644
--- a/include/xbt/string.hpp
+++ b/include/xbt/string.hpp
@@ -165,10 +165,7 @@ public:
     return data()[i];
   }
   // Conversion
-  operator std::string() const
-  {
-    return std::string(this->c_str(), this->size());
-  }
+  operator std::string() const { return std::string(this->c_str(), this->size()); }
 
   // Iterators
   iterator begin()               { return data(); }
diff --git a/include/xbt/swag.h b/include/xbt/swag.h
index 5c1ff355ab..0f25e2b29b 100644
--- a/include/xbt/swag.h
+++ b/include/xbt/swag.h
@@ -14,10 +14,10 @@
 
 SG_BEGIN_DECL()
 
-/** 
+/**
  * @addtogroup XBT_swag
  * @brief a O(1) set based on linked lists
- * 
+ *
  *  Warning, this module is done to be efficient and performs tons of cast and dirty things. So make sure you know what
  *  you are doing while using it.
  *  It is basically a fifo but with restrictions so that it can be used as a set. Any operation (add, remove, belongs)
@@ -29,7 +29,7 @@ SG_BEGIN_DECL()
 /** @defgroup XBT_swag_type Swag types
     @ingroup XBT_swag
 
-    Specific set. 
+    Specific set.
 
     These typedefs are public so that the compiler can do his job but believe me, you don't want to try to play with
     those structs directly. Use them as an abstract datatype.
@@ -39,7 +39,7 @@ typedef struct xbt_swag_hookup {
   void *next;
   void *prev;
 } s_xbt_swag_hookup_t;
-/**< This type should be added to a type that is to be used in a swag. 
+/**< This type should be added to a type that is to be used in a swag.
  *
  *  Whenever a new object with this struct is created, all fields have to be set to NULL
  *
@@ -78,9 +78,9 @@ typedef struct xbt_swag* xbt_swag_t;
 /**< A typical swag */
 /* @} */
 
-/** @defgroup XBT_swag_func SWAG functions 
+/** @defgroup XBT_swag_func SWAG functions
  *  @ingroup XBT_swag
- 
+
  *  @{
  */
 
@@ -136,7 +136,7 @@ static inline void *xbt_swag_getFirst(xbt_swag_t swag)
  * \defgroup XBT_swag_curs Swag cursor
  * @ingroup XBT_swag
 
- * Iterates over the whole swag. 
+ * Iterates over the whole swag.
  *
  * @{ */
 
@@ -157,13 +157,13 @@ static inline void *xbt_swag_getFirst(xbt_swag_t swag)
        (obj)=(decltype(obj)) xbt_swag_getNext((obj),(swag)->offset))
 #endif
 /**
- * @brief A safe swag iterator 
+ * @brief A safe swag iterator
  * @param obj the indice of the loop
  * @param obj_next the object that is right after (if any) \a obj in the swag
  * @param swag what to iterate over
  * @hideinitializer
 
-    You can safely modify the \a swag while using this loop. 
+    You can safely modify the \a swag while using this loop.
     Well, safely... Err. You can remove \a obj without having any trouble at least.  */
 
 #ifndef __cplusplus
diff --git a/include/xbt/xbt_os_thread.h b/include/xbt/xbt_os_thread.h
index 889026389b..fa4c014184 100644
--- a/include/xbt/xbt_os_thread.h
+++ b/include/xbt/xbt_os_thread.h
@@ -17,10 +17,10 @@ typedef pthread_key_t xbt_os_thread_key_t;
 
 /** @addtogroup XBT_thread
  *  @brief Thread portability layer
- * 
+ *
  *  This section describes the thread portability layer. It defines types and functions very close to the pthread API,
  *  but it's portable to windows too.
- * 
+ *
  *  @{
  */
 
diff --git a/include/xbt/xbt_os_time.h b/include/xbt/xbt_os_time.h
index 47f33bec40..e7b2c244c7 100644
--- a/include/xbt/xbt_os_time.h
+++ b/include/xbt/xbt_os_time.h
@@ -13,7 +13,7 @@
 
 SG_BEGIN_DECL()
 
-/** @brief get time in seconds 
+/** @brief get time in seconds
  *
  * gives  the  number  of  seconds since the Epoch (00:00:00 UTC, January 1, 1970).
  */
diff --git a/src/bindings/java/jmsg.h b/src/bindings/java/jmsg.h
index 34ba2a24a9..c0e7e2d0e7 100644
--- a/src/bindings/java/jmsg.h
+++ b/src/bindings/java/jmsg.h
@@ -23,7 +23,7 @@ SG_BEGIN_DECL()
 /* end of eclipse-mandated pimple */
 
 extern int JAVA_HOST_LEVEL;
-static std::unordered_map<char*, jobject> java_storage_map;
+static std::unordered_map<msg_storage_t, jobject> java_storage_map;
 
 JavaVM *get_java_VM();
 JNIEnv *get_current_thread_env();
diff --git a/src/bindings/java/jmsg_host.cpp b/src/bindings/java/jmsg_host.cpp
index 2cf5b28ff7..3f41bfdc18 100644
--- a/src/bindings/java/jmsg_host.cpp
+++ b/src/bindings/java/jmsg_host.cpp
@@ -141,7 +141,7 @@ JNIEXPORT void JNICALL Java_org_simgrid_msg_Host_on(JNIEnv *env, jobject jhost)
 
 JNIEXPORT void JNICALL Java_org_simgrid_msg_Host_off(JNIEnv *env, jobject jhost) {
   msg_host_t host = jhost_get_native(env, jhost);
-  MSG_host_off(host); 
+  MSG_host_off(host);
 }
 
 JNIEXPORT jint JNICALL Java_org_simgrid_msg_Host_getCount(JNIEnv * env, jclass cls) {
@@ -336,6 +336,11 @@ JNIEXPORT void JNICALL Java_org_simgrid_msg_Host_setAsyncMailbox(JNIEnv * env, j
   env->ReleaseStringUTFChars((jstring) jname, name);
 }
 
+JNIEXPORT void JNICALL Java_org_simgrid_msg_Host_updateAllEnergyConsumptions(JNIEnv* env, jclass cls)
+{
+  sg_host_energy_update_all();
+}
+
 JNIEXPORT jdouble JNICALL Java_org_simgrid_msg_Host_getConsumedEnergy (JNIEnv *env, jobject jhost)
 {
   msg_host_t host = jhost_get_native(env, jhost);
diff --git a/src/bindings/java/jmsg_host.h b/src/bindings/java/jmsg_host.h
index ebbc41e868..0b9ed50145 100644
--- a/src/bindings/java/jmsg_host.h
+++ b/src/bindings/java/jmsg_host.h
@@ -59,6 +59,7 @@ JNIEXPORT jobjectArray JNICALL Java_org_simgrid_msg_Host_getStorageContent(JNIEn
 JNIEXPORT jobjectArray JNICALL Java_org_simgrid_msg_Host_all(JNIEnv *env, jclass cls);
 JNIEXPORT void JNICALL Java_org_simgrid_msg_Host_setAsyncMailbox(JNIEnv * env, jclass cls_arg, jobject jname);
 
+JNIEXPORT void JNICALL Java_org_simgrid_msg_Host_updateAllEnergyConsumptions(JNIEnv* env, jclass cls);
 JNIEXPORT jdouble JNICALL Java_org_simgrid_msg_Host_getConsumedEnergy (JNIEnv *env, jobject jhost);
 
 JNIEXPORT void JNICALL Java_org_simgrid_msg_Host_setPstate(JNIEnv* env, jobject jhost, jint pstate);
diff --git a/src/bindings/java/jmsg_storage.cpp b/src/bindings/java/jmsg_storage.cpp
index e38230cd64..2d3a02931c 100644
--- a/src/bindings/java/jmsg_storage.cpp
+++ b/src/bindings/java/jmsg_storage.cpp
@@ -69,7 +69,7 @@ JNIEXPORT jobject JNICALL Java_org_simgrid_msg_Storage_getByName(JNIEnv * env, j
   }
   env->ReleaseStringUTFChars(jname, name);
 
-  if (java_storage_map.find(storage->key) == java_storage_map.end()) {
+  if (java_storage_map.find(storage) == java_storage_map.end()) {
     /* Instantiate a new java storage */
     jstorage = jstorage_new_instance(env);
 
@@ -93,9 +93,9 @@ JNIEXPORT jobject JNICALL Java_org_simgrid_msg_Storage_getByName(JNIEnv * env, j
     /* the native storage data field is set with the global reference to the
      * java storage returned by this function
      */
-    java_storage_map.insert({storage->key, jstorage});
+    java_storage_map.insert({storage, jstorage});
   } else
-    jstorage = java_storage_map.at(storage->key);
+    jstorage = java_storage_map.at(storage);
 
   /* return the global reference to the java storage instance */
   return (jobject)jstorage;
@@ -215,8 +215,8 @@ JNIEXPORT jobjectArray JNICALL Java_org_simgrid_msg_Storage_all(JNIEnv * env, jc
 
   for (index = 0; index < count; index++) {
     storage = xbt_dynar_get_as(table,index,msg_storage_t);
-    if (java_storage_map.find(storage->key) != java_storage_map.end()) {
-      jstorage = java_storage_map.at(storage->key);
+    if (java_storage_map.find(storage) != java_storage_map.end()) {
+      jstorage = java_storage_map.at(storage);
     } else {
       jname = env->NewStringUTF(MSG_storage_get_name(storage));
       jstorage = Java_org_simgrid_msg_Storage_getByName(env, cls_arg, jname);
diff --git a/src/bindings/java/jtrace.cpp b/src/bindings/java/jtrace.cpp
index 26c64daf99..c33ba631d6 100644
--- a/src/bindings/java/jtrace.cpp
+++ b/src/bindings/java/jtrace.cpp
@@ -25,7 +25,7 @@ SG_BEGIN_DECL()
 JNIEXPORT void JNICALL Java_org_simgrid_trace_Trace_hostStateDeclare(JNIEnv * env, jclass cls, jstring js)
 {
   const char *s = env->GetStringUTFChars(js, 0);
-  TRACE_host_state_declare(s);  
+  TRACE_host_state_declare(s);
   env->ReleaseStringUTFChars(js, s);
 }
 
@@ -36,7 +36,7 @@ JNIEXPORT void JNICALL Java_org_simgrid_trace_Trace_hostStateDeclareValue (JNIEn
   const char *value = env->GetStringUTFChars(js_value, 0);
   const char *color = env->GetStringUTFChars(js_color, 0);
 
-  TRACE_host_state_declare_value(state, value, color);  
+  TRACE_host_state_declare_value(state, value, color);
 
   env->ReleaseStringUTFChars(js_state, state);
   env->ReleaseStringUTFChars(js_value, value);
@@ -50,7 +50,7 @@ JNIEXPORT void JNICALL Java_org_simgrid_trace_Trace_hostSetState (JNIEnv *env, j
   const char *state = env->GetStringUTFChars(js_state, 0);
   const char *value = env->GetStringUTFChars(js_value, 0);
 
-  TRACE_host_set_state(host, state, value);  
+  TRACE_host_set_state(host, state, value);
 
   env->ReleaseStringUTFChars(js_host, host);
   env->ReleaseStringUTFChars(js_state, state);
@@ -64,7 +64,7 @@ JNIEXPORT void JNICALL Java_org_simgrid_trace_Trace_hostPushState (JNIEnv *env,
   const char *state = env->GetStringUTFChars(js_state, 0);
   const char *value = env->GetStringUTFChars(js_value, 0);
 
-  TRACE_host_push_state(host, state, value);  
+  TRACE_host_push_state(host, state, value);
 
   env->ReleaseStringUTFChars(js_host, host);
   env->ReleaseStringUTFChars(js_state, state);
@@ -77,7 +77,7 @@ JNIEXPORT void JNICALL Java_org_simgrid_trace_Trace_hostPopState (JNIEnv *env, j
   const char *host = env->GetStringUTFChars(js_host, 0);
   const char *state = env->GetStringUTFChars(js_state, 0);
 
-  TRACE_host_pop_state(host, state);  
+  TRACE_host_pop_state(host, state);
 
   env->ReleaseStringUTFChars(js_host, host);
   env->ReleaseStringUTFChars(js_state, state);
diff --git a/src/bindings/java/org/simgrid/msg/Host.java b/src/bindings/java/org/simgrid/msg/Host.java
index 9165a10316..4676242247 100644
--- a/src/bindings/java/org/simgrid/msg/Host.java
+++ b/src/bindings/java/org/simgrid/msg/Host.java
@@ -133,7 +133,16 @@ public class Host {
 	/** This methods returns the list of storages (names) attached to an host */
 	public native String[] getAttachedStorage();
 
-	/** Returns the amount of Joules consumed by that host so far */
+	/** After this call, sg_host_get_consumed_energy() will not interrupt your process
+	 * (until after the next clock update).
+	 */
+	public static native void updateAllEnergyConsumptions();
+	/** Returns the amount of Joules consumed by that host so far 
+	 *
+	 * Please note that since the consumption is lazily updated, it may require a simcall to update it.
+	 * The result is that the actor requesting this value will be interrupted,
+	 * the value will be updated in kernel mode before returning the control to the requesting actor.
+	 */
 	public native double getConsumedEnergy();
 	
 	/** Returns the current pstate */
diff --git a/src/bindings/lua/lua_debug.cpp b/src/bindings/lua/lua_debug.cpp
index df54f3f135..c02d921fdf 100644
--- a/src/bindings/lua/lua_debug.cpp
+++ b/src/bindings/lua/lua_debug.cpp
@@ -1,5 +1,5 @@
 /* Copyright (c) 2010-2016. The SimGrid Team.
- * All rights reserved.                                                     
+ * All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms of the license (GNU LGPL) which comes with this package. */
@@ -40,7 +40,7 @@ const char* sglua_tostring(lua_State* L, int index) {
       snprintf(buff, 4, "nil");
       break;
 
-    case LUA_TNUMBER: 
+    case LUA_TNUMBER:
       snprintf(buff, 64, "%.3f", lua_tonumber(L, index));
       break;
 
@@ -84,7 +84,7 @@ static int sglua_dump_table(lua_State* L) {
     if (lua_istable(L, i)) {
       lua_pushnil(L); /* table nil */
 
-      //lua_next pops the topmost element from the stack and 
+      //lua_next pops the topmost element from the stack and
       //gets the next pair from the table
       while (lua_next(L, -1)) { /* table key val  */
         // we need to copy here, as a cast from "Number" to "String"
diff --git a/src/bindings/lua/simgrid_lua.cpp b/src/bindings/lua/simgrid_lua.cpp
index 91b854fd65..846d2b096a 100644
--- a/src/bindings/lua/simgrid_lua.cpp
+++ b/src/bindings/lua/simgrid_lua.cpp
@@ -69,7 +69,7 @@ static int critical(lua_State* L) {
  * @brief Dumps a lua table with XBT_DEBUG
  *
  * This function can be called from within lua via "simgrid.dump(table)". It will
- * then dump the table via XBT_DEBUG 
+ * then dump the table via XBT_DEBUG
  */
 static int dump(lua_State* L) {
   int argc = lua_gettop(L);
@@ -78,7 +78,7 @@ static int dump(lua_State* L) {
     if (lua_istable(L, i)) {
       lua_pushnil(L); /* table nil */
 
-      //lua_next pops the topmost element from the stack and 
+      //lua_next pops the topmost element from the stack and
       //gets the next pair from the table at the specified index
       while (lua_next(L, i)) { /* table key val  */
         // we need to copy here, as a cast from "Number" to "String"
diff --git a/src/include/mc/mc.h b/src/include/mc/mc.h
index 3f5864ce2f..0834db9516 100644
--- a/src/include/mc/mc.h
+++ b/src/include/mc/mc.h
@@ -24,7 +24,7 @@
 
 SG_BEGIN_DECL()
 
-/********************************** Configuration of MC **************************************/  
+/********************************** Configuration of MC **************************************/
 extern XBT_PUBLIC(int) _sg_do_model_check;
 extern XBT_PRIVATE int _sg_do_model_check_record;
 extern XBT_PRIVATE int _sg_mc_checkpoint;
diff --git a/src/include/surf/maxmin.h b/src/include/surf/maxmin.h
index 9a1ffb310b..7365bfaa41 100644
--- a/src/include/surf/maxmin.h
+++ b/src/include/surf/maxmin.h
@@ -18,10 +18,10 @@ class Action;
 }
 }
 
-/** @addtogroup SURF_lmm 
- * @details 
+/** @addtogroup SURF_lmm
+ * @details
  * A linear maxmin solver to resolve inequations systems.
- * 
+ *
  * Most SimGrid model rely on a "fluid/steady-state" modeling that simulate the sharing of resources between actions at
  * relatively coarse-grain.  Such sharing is generally done by solving a set of linear inequations. Let's take an
  * example and assume we have the variables \f$x_1\f$, \f$x_2\f$, \f$x_3\f$, and \f$x_4\f$ . Let's say that \f$x_1\f$
@@ -44,7 +44,7 @@ class Action;
  * This is called *max-min fairness* and is the most commonly used objective in SimGrid. Another possibility is to
  * maximize \f$\sum_if(x_i)\f$, where \f$f\f$ is a strictly increasing concave function.
  *
- * Constraint: 
+ * Constraint:
  *  - bound (set)
  *  - shared (set)
  *  - usage (computed)
@@ -56,7 +56,7 @@ class Action;
  *
  * Element:
  *  - value (set)
- * 
+ *
  * A possible system could be:
  * - three variables: `var1`, `var2`, `var3`
  * - two constraints: `cons1`, `cons2`
@@ -65,19 +65,19 @@ class Action;
  *  - `elem2` linking `var2` and `cons1`
  *  - `elem3` linking `var2` and `cons2`
  *  - `elem4` linking `var3` and `cons2`
- * 
+ *
  * And the corresponding inequations will be:
- * 
+ *
  *     var1.value <= var1.bound
  *     var2.value <= var2.bound
  *     var3.value <= var3.bound
  *     var1.weight * var1.value * elem1.value + var2.weight * var2.value * elem2.value <= cons1.bound
  *     var2.weight * var2.value * elem3.value + var3.weight * var3.value * elem4.value <= cons2.bound
- * 
+ *
  * where `var1.value`, `var2.value` and `var3.value` are the unknown values.
- * 
- * If a constraint is not shared, the sum is replaced by a max. 
- * For example, a third non-shared constraint `cons3` and the associated elements `elem5` and `elem6` could write as: 
+ *
+ * If a constraint is not shared, the sum is replaced by a max.
+ * For example, a third non-shared constraint `cons3` and the associated elements `elem5` and `elem6` could write as:
  *
  *     max( var1.weight * var1.value * elem5.value  ,  var3.weight * var3.value * elem6.value ) <= cons3.bound
  *
@@ -99,28 +99,28 @@ class Action;
  * (lmm_solve()) activates it when appropriate. It is possible that the variable is again disabled, e.g. to model the
  * pausing of an action.
  *
- * Concurrency limit and maximum 
- * 
- * We call concurrency, the number of variables that can be enabled at any time for each constraint. 
+ * Concurrency limit and maximum
+ *
+ * We call concurrency, the number of variables that can be enabled at any time for each constraint.
  * From a model perspective, this "concurrency" often represents the number of actions that actually compete for one
  * constraint.
  * The LMM solver is able to limit the concurrency for each constraint, and to monitor its maximum value.
- * 
+ *
  * One may want to limit the concurrency of constraints for essentially three reasons:
  *  - Keep LMM system in a size that can be solved (it does not react very well with tens of thousands of variables per
  *    constraint)
- *  - Stay within parameters where the fluid model is accurate enough.      
+ *  - Stay within parameters where the fluid model is accurate enough.
  *  - Model serialization effects
  *
  * The concurrency limit can also be set to a negative value to disable concurrency limit. This can improve performance
  * slightly.
- * 
+ *
  * Overall, each constraint contains three fields related to concurrency:
  *  - concurrency_limit which is the limit enforced by the solver
  *  - concurrency_current which is the current concurrency
- *  - concurrency_maximum which is the observed maximum concurrency 
+ *  - concurrency_maximum which is the observed maximum concurrency
  *
- * Variables also have one field related to concurrency: concurrency_share. 
+ * Variables also have one field related to concurrency: concurrency_share.
  * In effect, in some cases, one variable is involved multiple times (i.e. two elements) in a constraint.
  * For example, cross-traffic is modeled using 2 elements per constraint.
  * concurrency_share formally corresponds to the maximum number of elements that associate the variable and any given
@@ -130,7 +130,7 @@ class Action;
 XBT_PUBLIC_DATA(double) sg_maxmin_precision;
 XBT_PUBLIC_DATA(double) sg_surf_precision;
 XBT_PUBLIC_DATA(int) sg_concurrency_limit;
- 
+
 static inline void double_update(double *variable, double value, double precision)
 {
   //printf("Updating %g -= %g +- %g\n",*variable,value,precision);
@@ -172,7 +172,7 @@ XBT_PUBLIC(void) lmm_system_free(lmm_system_t sys);
  * @brief Create a new Linear MaxMin constraint
  * @param sys The system in which we add a constraint
  * @param id Data associated to the constraint (e.g.: a network link)
- * @param bound_value The bound value of the constraint 
+ * @param bound_value The bound value of the constraint
  */
 XBT_PUBLIC(lmm_constraint_t) lmm_constraint_new(lmm_system_t sys, void *id,double bound_value);
 
@@ -323,7 +323,7 @@ XBT_PUBLIC(double) lmm_get_cnst_weight_from_var(lmm_system_t sys, lmm_variable_t
 XBT_PUBLIC(int) lmm_get_number_of_cnst_from_var(lmm_system_t sys, lmm_variable_t var);
 
 /**
- * @brief Get a var associated to a constraint 
+ * @brief Get a var associated to a constraint
  * @details Get the first variable of the next variable of elem if elem is not NULL
  * @param sys The system associated to the variable (not used)
  * @param cnst A constraint
@@ -356,7 +356,7 @@ XBT_PUBLIC(lmm_constraint_t) lmm_get_first_active_constraint(lmm_system_t sys);
  * @brief Get the next active constraint of a constraint in a system
  * @param sys A system
  * @param cnst An active constraint of the system
- * 
+ *
  * @return The next active constraint
  */
 XBT_PUBLIC(lmm_constraint_t) lmm_get_next_active_constraint(lmm_system_t sys, lmm_constraint_t cnst);
diff --git a/src/include/surf/surf.h b/src/include/surf/surf.h
index 4a3c1094ba..0d0ced1f14 100644
--- a/src/include/surf/surf.h
+++ b/src/include/surf/surf.h
@@ -13,9 +13,7 @@
 #include "xbt/misc.h"
 #include "xbt/config.h"
 #include "src/internal_config.h"
-#include "surf/surf_routing.h"
 #include "surf/datatypes.h"
-#include "xbt/lib.h"
 #include "simgrid/datatypes.h"
 #include "simgrid/forward.h"
 
@@ -85,8 +83,6 @@ typedef surf_NetworkModel *surf_network_model_t;
 typedef surf_StorageModel *surf_storage_model_t;
 typedef surf_Storage* surf_storage_t;
 
-typedef xbt_dictelm_t surf_resource_t;
-
 /** @ingroup SURF_c_bindings
  *  \brief Action structure
  *
@@ -116,15 +112,6 @@ XBT_PUBLIC(void) model_help(const char *category, s_surf_model_description_t * t
 /* Generic model object */
 /***************************/
 
-static inline surf_storage_t surf_storage_resource_priv(const void* storage)
-{
-  return (surf_storage_t)xbt_lib_get_level((xbt_dictelm_t)storage, SURF_STORAGE_LEVEL);
-}
-
-static inline void *surf_storage_resource_by_name(const char *name){
-  return xbt_lib_get_elm_or_null(storage_lib, name);
-}
-
 /** @{ @ingroup SURF_c_bindings */
 
 /**
@@ -258,7 +245,7 @@ XBT_PUBLIC(int) surf_host_file_seek(sg_host_t host, surf_file_t fd, sg_offset_t
  * @param resource The surf storage
  * @return The size in bytes of the storage
  */
-XBT_PUBLIC(sg_size_t) surf_storage_get_size(surf_resource_t resource);
+XBT_PUBLIC(sg_size_t) surf_storage_get_size(surf_storage_t resource);
 
 /**
  * @brief Get the available size in bytes of a storage
@@ -266,7 +253,7 @@ XBT_PUBLIC(sg_size_t) surf_storage_get_size(surf_resource_t resource);
  * @param resource The surf storage
  * @return The available size in bytes of the storage
  */
-XBT_PUBLIC(sg_size_t) surf_storage_get_free_size(surf_resource_t resource);
+XBT_PUBLIC(sg_size_t) surf_storage_get_free_size(surf_storage_t resource);
 
 /**
  * @brief Get the size in bytes of a storage
@@ -274,10 +261,10 @@ XBT_PUBLIC(sg_size_t) surf_storage_get_free_size(surf_resource_t resource);
  * @param resource The surf storage
  * @return The used size in bytes of the storage
  */
-XBT_PUBLIC(sg_size_t) surf_storage_get_used_size(surf_resource_t resource);
+XBT_PUBLIC(sg_size_t) surf_storage_get_used_size(surf_storage_t resource);
 
 /** @brief return the properties set associated to that storage */
-XBT_PUBLIC(xbt_dict_t) surf_storage_get_properties(surf_resource_t resource);
+XBT_PUBLIC(xbt_dict_t) surf_storage_get_properties(surf_storage_t resource);
 
 /**
  * @brief [brief description]
@@ -320,8 +307,8 @@ XBT_PUBLIC(xbt_dict_t) surf_storage_action_get_ls_dict(surf_action_t action);
  * @return The host name
  * may not exist.
  */
-XBT_PUBLIC(const char * ) surf_storage_get_host(surf_resource_t resource);
-XBT_PUBLIC(const char * ) surf_storage_get_name(surf_resource_t resource);
+XBT_PUBLIC(const char*) surf_storage_get_host(surf_storage_t resource);
+XBT_PUBLIC(const char*) surf_storage_get_name(surf_storage_t resource);
 
 /** @} */
 
diff --git a/src/instr/instr_paje_trace.cpp b/src/instr/instr_paje_trace.cpp
index ae772d9106..6afb85878f 100644
--- a/src/instr/instr_paje_trace.cpp
+++ b/src/instr/instr_paje_trace.cpp
@@ -126,9 +126,9 @@ static void print_timestamp(PajeEvent* event) {
   /* prevent 0.0000 in the trace - this was the behavior before the transition to c++ */
   if (event->timestamp < 1e-12)
     stream << 0;
-  else 
+  else
     stream << event->timestamp;
-}  
+}
 
 /* internal do the instrumentation module */
 static void insert_into_buffer (PajeEvent* tbi)
@@ -845,6 +845,11 @@ void ResetStateEvent::print() {
   }
 }
 
+StartLinkEvent::~StartLinkEvent()
+{
+  free(value);
+  free(key);
+}
 StartLinkEvent::StartLinkEvent (double timestamp, container_t container,
     type_t type, container_t sourceContainer, const char *value, const char *key)
   : StartLinkEvent(timestamp, container, type, sourceContainer, value, key, -1)
@@ -904,7 +909,11 @@ EndLinkEvent::EndLinkEvent (double timestamp, container_t container, type_t type
   insert_into_buffer (this);
 }
 
-
+EndLinkEvent::~EndLinkEvent()
+{
+  free(value);
+  free(key);
+}
 void EndLinkEvent::print() {
   if (instr_fmt_type == instr_fmt_paje) {
     XBT_DEBUG("%s: event_type=%d, timestamp=%.*f", __FUNCTION__, (int)event_type, TRACE_precision(), timestamp);
diff --git a/src/instr/instr_paje_types.cpp b/src/instr/instr_paje_types.cpp
index acb89e3c47..2eb903041b 100644
--- a/src/instr/instr_paje_types.cpp
+++ b/src/instr/instr_paje_types.cpp
@@ -136,7 +136,7 @@ type_t PJ_type_container_new (const char *name, type_t father)
 }
 
 type_t PJ_type_event_new (const char *name, type_t father)
-{ 
+{
   if (name == nullptr){
     THROWF (tracing_error, 0, "can't create an event type with a nullptr name");
   }
diff --git a/src/instr/instr_private.h b/src/instr/instr_private.h
index d4da8faf72..9d10c4b1a3 100644
--- a/src/instr/instr_private.h
+++ b/src/instr/instr_private.h
@@ -101,7 +101,7 @@ class s_container;
 typedef s_container *container_t;
 
 class s_container {
-  public: 
+  public:
   sg_netpoint_t netpoint;
   char *name;     /* Unique name of this container */
   char *id;       /* Unique id of this container */
@@ -136,7 +136,7 @@ class DefineContainerEvent : public PajeEvent
 
 //--------------------------------------------------
 
-class DefineVariableTypeEvent : public PajeEvent 
+class DefineVariableTypeEvent : public PajeEvent
 {
   public:
   type_t type;
@@ -154,7 +154,7 @@ class DefineStateTypeEvent : public PajeEvent  {
 
 class DefineEventTypeEvent : public PajeEvent  {
   type_t type;
-  public: 
+  public:
   DefineEventTypeEvent(type_t type);
   void print() override;
 };
@@ -216,7 +216,7 @@ class SubVariableEvent : public PajeEvent  {
   public:
   container_t container;
   type_t type;
-  double value; 
+  double value;
   public:
   SubVariableEvent(double timestamp, container_t container, type_t type, double value);
   void print() override;
@@ -277,11 +277,12 @@ class StartLinkEvent : public PajeEvent  {
   char *key;
   int size;
   public:
-  StartLinkEvent (double timestamp, container_t container, type_t type, container_t sourceContainer,
-                                    const char *value, const char *key);
-  StartLinkEvent (double timestamp, container_t container, type_t type,
-                                            container_t sourceContainer, const char *value, const char *key, int size);
-  void print() override;
+    ~StartLinkEvent();
+    StartLinkEvent(double timestamp, container_t container, type_t type, container_t sourceContainer, const char* value,
+                   const char* key);
+    StartLinkEvent(double timestamp, container_t container, type_t type, container_t sourceContainer, const char* value,
+                   const char* key, int size);
+    void print() override;
 };
 
 class EndLinkEvent : public PajeEvent  {
@@ -293,6 +294,7 @@ class EndLinkEvent : public PajeEvent  {
   public:
   EndLinkEvent (double timestamp, container_t container, type_t type, container_t destContainer,
                                   const char *value, const char *key);
+  ~EndLinkEvent();
   void print() override;
 };
 
diff --git a/src/kernel/activity/ActivityImpl.cpp b/src/kernel/activity/ActivityImpl.cpp
index c732b28eea..6a9b6e83df 100644
--- a/src/kernel/activity/ActivityImpl.cpp
+++ b/src/kernel/activity/ActivityImpl.cpp
@@ -5,6 +5,8 @@
 
 #include "src/kernel/activity/ActivityImpl.hpp"
 
+XBT_LOG_EXTERNAL_CATEGORY(simix_process);
+
 namespace simgrid {
 namespace kernel {
 namespace activity {
@@ -12,31 +14,25 @@ namespace activity {
 ActivityImpl::ActivityImpl()  = default;
 ActivityImpl::~ActivityImpl() = default;
 
-void ActivityImpl::ref()
-{
-  // Atomic operation! Do not split in two instructions!
-  xbt_assert(refcount_ != 0);
-  refcount_++;
-}
-
-void ActivityImpl::unref()
-{
-  xbt_assert(refcount_ > 0,
-             "This activity has a negative refcount! You can only call test() or wait() once per activity.");
-  refcount_--;
-  if (refcount_ == 0)
-    delete this;
-}
-
 // boost::intrusive_ptr<Activity> support:
 void intrusive_ptr_add_ref(simgrid::kernel::activity::ActivityImpl* activity)
 {
-  activity->ref();
+  xbt_assert(activity->refcount_ >= 0);
+  activity->refcount_++;
+  XBT_CDEBUG(simix_process, "%p->refcount++ ~> %d", activity, (int)activity->refcount_);
+  if (XBT_LOG_ISENABLED(simix_process, xbt_log_priority_trace))
+    xbt_backtrace_display_current();
 }
 
 void intrusive_ptr_release(simgrid::kernel::activity::ActivityImpl* activity)
 {
-  activity->unref();
+  XBT_CDEBUG(simix_process, "%p->refcount-- ~> %d", activity, ((int)activity->refcount_) - 1);
+  xbt_assert(activity->refcount_ >= 0);
+  activity->refcount_--;
+  if (XBT_LOG_ISENABLED(simix_process, xbt_log_priority_trace))
+    xbt_backtrace_display_current();
+  if (activity->refcount_ <= 0)
+    delete activity;
 }
 }
 }
diff --git a/src/kernel/activity/ActivityImpl.hpp b/src/kernel/activity/ActivityImpl.hpp
index 1496c41717..2a50de8059 100644
--- a/src/kernel/activity/ActivityImpl.hpp
+++ b/src/kernel/activity/ActivityImpl.hpp
@@ -31,17 +31,12 @@ namespace activity {
     virtual void resume()=0;
     virtual void post() =0; // What to do when a simcall terminates
 
-    /** @brief Increases the refcount */
-    void ref();
-    /** @brief Reduces the refcount */
-    void unref();
-
-     // boost::intrusive_ptr<Activity> support:
+    // boost::intrusive_ptr<ActivityImpl> support:
     friend void intrusive_ptr_add_ref(ActivityImpl * activity);
     friend void intrusive_ptr_release(ActivityImpl * activity);
 
   private:
-    std::atomic_int_fast32_t refcount_{1};
+    std::atomic_int_fast32_t refcount_{0};
   };
 }}} // namespace simgrid::kernel::activity
 
diff --git a/src/kernel/activity/CommImpl.cpp b/src/kernel/activity/CommImpl.cpp
index 85ccbea5c1..33d31e7d03 100644
--- a/src/kernel/activity/CommImpl.cpp
+++ b/src/kernel/activity/CommImpl.cpp
@@ -17,7 +17,6 @@ simgrid::kernel::activity::CommImpl::CommImpl(e_smx_comm_type_t _type) : type(_t
   state    = SIMIX_WAITING;
   src_data = nullptr;
   dst_data = nullptr;
-  intrusive_ptr_add_ref(this);
   XBT_DEBUG("Create comm activity %p", this);
 }
 
@@ -62,7 +61,6 @@ void simgrid::kernel::activity::CommImpl::cancel()
   if (state == SIMIX_WAITING) {
     mbox->remove(this);
     state = SIMIX_CANCELED;
-    this->unref();
   } else if (not MC_is_active() /* when running the MC there are no surf actions */
              && not MC_record_replay_is_active() && (state == SIMIX_READY || state == SIMIX_RUNNING)) {
 
@@ -124,6 +122,5 @@ void simgrid::kernel::activity::CommImpl::post()
   /* if there are simcalls associated with the synchro, then answer them */
   if (not simcalls.empty()) {
     SIMIX_comm_finish(this);
-    this->unref();
   }
 }
diff --git a/src/kernel/activity/ExecImpl.cpp b/src/kernel/activity/ExecImpl.cpp
index 56c6436df8..aee76bc532 100644
--- a/src/kernel/activity/ExecImpl.cpp
+++ b/src/kernel/activity/ExecImpl.cpp
@@ -16,6 +16,7 @@ simgrid::kernel::activity::ExecImpl::ExecImpl(const char* name, sg_host_t host)
   if (name)
     this->name = name;
   this->state  = SIMIX_RUNNING;
+  XBT_DEBUG("Create exec %p", this);
 }
 
 simgrid::kernel::activity::ExecImpl::~ExecImpl()
@@ -24,6 +25,7 @@ simgrid::kernel::activity::ExecImpl::~ExecImpl()
     surf_exec->unref();
   if (timeoutDetector)
     timeoutDetector->unref();
+  XBT_DEBUG("Destroy exec %p", this);
 }
 void simgrid::kernel::activity::ExecImpl::suspend()
 {
diff --git a/src/kernel/activity/MailboxImpl.cpp b/src/kernel/activity/MailboxImpl.cpp
index 8c66d352d1..3f8cedc888 100644
--- a/src/kernel/activity/MailboxImpl.cpp
+++ b/src/kernel/activity/MailboxImpl.cpp
@@ -53,10 +53,10 @@ void MailboxImpl::setReceiver(s4u::ActorPtr actor)
 /** @brief Pushes a communication activity into a mailbox
  *  @param comm What to add
  */
-void MailboxImpl::push(activity::CommImpl* comm)
+void MailboxImpl::push(activity::CommImplPtr comm)
 {
-  this->comm_queue.push_back(comm);
   comm->mbox = this;
+  this->comm_queue.push_back(std::move(comm));
 }
 
 /** @brief Removes a communication activity from a mailbox
@@ -64,15 +64,18 @@ void MailboxImpl::push(activity::CommImpl* comm)
  */
 void MailboxImpl::remove(smx_activity_t activity)
 {
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(activity);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(activity);
 
+  xbt_assert(comm->mbox == this, "Comm %p is in mailbox %s, not mailbox %s", comm.get(),
+             (comm->mbox ? comm->mbox->name_ : "(null)"), this->name_);
   comm->mbox = nullptr;
   for (auto it = this->comm_queue.begin(); it != this->comm_queue.end(); it++)
     if (*it == comm) {
       this->comm_queue.erase(it);
       return;
     }
-  xbt_die("Cannot remove the comm %p that is not part of the mailbox %s", comm, this->name_);
+  xbt_die("Comm %p not found in mailbox %s", comm.get(), this->name_);
 }
 }
 }
diff --git a/src/kernel/activity/MailboxImpl.hpp b/src/kernel/activity/MailboxImpl.hpp
index ea9b272fba..8c4541d611 100644
--- a/src/kernel/activity/MailboxImpl.hpp
+++ b/src/kernel/activity/MailboxImpl.hpp
@@ -31,12 +31,12 @@ public:
   static MailboxImpl* byNameOrNull(const char* name);
   static MailboxImpl* byNameOrCreate(const char* name);
   void setReceiver(s4u::ActorPtr actor);
-  void push(activity::CommImpl* comm);
+  void push(activity::CommImplPtr comm);
   void remove(smx_activity_t activity);
   simgrid::s4u::Mailbox piface_; // Our interface
   char* name_;
 
-  boost::intrusive_ptr<simgrid::simix::ActorImpl> permanent_receiver; // process which the mailbox is attached to
+  simgrid::simix::ActorImplPtr permanent_receiver; // process which the mailbox is attached to
   boost::circular_buffer_space_optimized<smx_activity_t> comm_queue;
   boost::circular_buffer_space_optimized<smx_activity_t> done_comm_queue; // messages already received in the permanent receive mode
 };
diff --git a/src/kernel/activity/SynchroIo.cpp b/src/kernel/activity/SynchroIo.cpp
index cef2ced0f2..012165b614 100644
--- a/src/kernel/activity/SynchroIo.cpp
+++ b/src/kernel/activity/SynchroIo.cpp
@@ -7,19 +7,19 @@
 #include "src/surf/surf_interface.hpp"
 #include "src/simix/smx_private.h"
 
-void simgrid::kernel::activity::Io::suspend()
+void simgrid::kernel::activity::IoImpl::suspend()
 {
   if (surf_io)
     surf_io->suspend();
 }
 
-void simgrid::kernel::activity::Io::resume()
+void simgrid::kernel::activity::IoImpl::resume()
 {
   if (surf_io)
     surf_io->resume();
 }
 
-void simgrid::kernel::activity::Io::post()
+void simgrid::kernel::activity::IoImpl::post()
 {
   for (smx_simcall_t simcall : simcalls) {
     switch (simcall->call) {
diff --git a/src/kernel/activity/SynchroIo.hpp b/src/kernel/activity/SynchroIo.hpp
index b9c405daa3..7a229bf749 100644
--- a/src/kernel/activity/SynchroIo.hpp
+++ b/src/kernel/activity/SynchroIo.hpp
@@ -13,14 +13,15 @@ namespace simgrid {
 namespace kernel {
 namespace activity {
 
-  XBT_PUBLIC_CLASS Io : public ActivityImpl {
-  public:
-    void suspend() override;
-    void resume() override;
-    void post() override;
-
-    sg_host_t host = nullptr;
-    surf_action_t surf_io = nullptr;
+XBT_PUBLIC_CLASS IoImpl : public ActivityImpl
+{
+public:
+  void suspend() override;
+  void resume() override;
+  void post() override;
+
+  sg_host_t host        = nullptr;
+  surf_action_t surf_io = nullptr;
   };
 
 }}} // namespace simgrid::kernel::activity
diff --git a/src/kernel/activity/SynchroRaw.cpp b/src/kernel/activity/SynchroRaw.cpp
index 0f9e29f30d..509c41d5cd 100644
--- a/src/kernel/activity/SynchroRaw.cpp
+++ b/src/kernel/activity/SynchroRaw.cpp
@@ -9,21 +9,21 @@
 
 XBT_LOG_EXTERNAL_DEFAULT_CATEGORY(simix_synchro);
 
-simgrid::kernel::activity::Raw::~Raw()
+simgrid::kernel::activity::RawImpl::~RawImpl()
 {
   sleep->unref();
 }
-void simgrid::kernel::activity::Raw::suspend()
+void simgrid::kernel::activity::RawImpl::suspend()
 {
   /* The suspension of raw synchros is delayed to when the process is rescheduled. */
 }
 
-void simgrid::kernel::activity::Raw::resume()
+void simgrid::kernel::activity::RawImpl::resume()
 {
   /* I cannot resume raw synchros directly. This is delayed to when the process is rescheduled at
    * the end of the synchro. */
 }
-void simgrid::kernel::activity::Raw::post()
+void simgrid::kernel::activity::RawImpl::post()
 {
   XBT_IN("(%p)",this);
   if (sleep->getState() == simgrid::surf::Action::State::failed)
diff --git a/src/kernel/activity/SynchroRaw.hpp b/src/kernel/activity/SynchroRaw.hpp
index b85e06f638..f5ef90d2fc 100644
--- a/src/kernel/activity/SynchroRaw.hpp
+++ b/src/kernel/activity/SynchroRaw.hpp
@@ -14,14 +14,15 @@ namespace kernel {
 namespace activity {
 
   /** Used to implement mutexes, semaphores and conditions */
-  XBT_PUBLIC_CLASS Raw : public ActivityImpl {
-  public:
-    ~Raw() override;
-    void suspend() override;
-    void resume() override;
-    void post() override;
-
-    surf_action_t sleep = nullptr;
+XBT_PUBLIC_CLASS RawImpl : public ActivityImpl
+{
+public:
+  ~RawImpl() override;
+  void suspend() override;
+  void resume() override;
+  void post() override;
+
+  surf_action_t sleep = nullptr;
   };
 
 }}} // namespace simgrid::kernel::activity
diff --git a/src/kernel/context/ContextRaw.cpp b/src/kernel/context/ContextRaw.cpp
index fa39f6cc3a..057d46e8da 100644
--- a/src/kernel/context/ContextRaw.cpp
+++ b/src/kernel/context/ContextRaw.cpp
@@ -3,7 +3,7 @@
 /* This program is free software; you can redistribute it and/or modify it
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
-#include "src/internal_config.h" 
+#include "src/internal_config.h"
 
 #include "xbt/parmap.h"
 
@@ -23,12 +23,12 @@ class RawContextFactory;
 
 /** @brief Fast context switching inspired from SystemV ucontexts.
   *
-  * The main difference to the System V context is that Raw Contexts are much faster because they don't 
+  * The main difference to the System V context is that Raw Contexts are much faster because they don't
   * preserve the signal mask when switching. This saves a system call (at least on Linux) on each context switch.
   */
 class RawContext : public Context {
 protected:
-  void* stack_ = nullptr; 
+  void* stack_ = nullptr;
   /** pointer to top the stack stack */
   void* stack_top_ = nullptr;
 public:
diff --git a/src/kernel/routing/FatTreeZone.cpp b/src/kernel/routing/FatTreeZone.cpp
index 70a928d07e..a364180f7a 100644
--- a/src/kernel/routing/FatTreeZone.cpp
+++ b/src/kernel/routing/FatTreeZone.cpp
@@ -11,8 +11,6 @@
 #include "src/kernel/routing/NetPoint.hpp"
 #include "src/surf/network_interface.hpp"
 
-#include "xbt/lib.h"
-
 #include <boost/algorithm/string/classification.hpp>
 #include <boost/algorithm/string/split.hpp>
 
diff --git a/src/kernel/routing/NetZoneImpl.cpp b/src/kernel/routing/NetZoneImpl.cpp
index dc21bfffa1..dc907922b4 100644
--- a/src/kernel/routing/NetZoneImpl.cpp
+++ b/src/kernel/routing/NetZoneImpl.cpp
@@ -58,7 +58,7 @@ simgrid::s4u::Host* NetZoneImpl::createHost(const char* name, std::vector<double
     for (auto kv : *props)
       res->setProperty(kv.first.c_str(), kv.second.c_str());
 
-  simgrid::s4u::Host::onCreation(*res);
+  simgrid::s4u::Host::onCreation(*res); // notify the signal
 
   return res;
 }
diff --git a/src/mc/AddressSpace.hpp b/src/mc/AddressSpace.hpp
index 050650c0c9..c4985e286f 100644
--- a/src/mc/AddressSpace.hpp
+++ b/src/mc/AddressSpace.hpp
@@ -50,7 +50,7 @@ class ReadOptions {
 public:
   constexpr ReadOptions() : value_(0) {}
 
-  constexpr operator bool() const { return value_ != 0; }
+  explicit constexpr operator bool() const { return value_ != 0; }
   constexpr bool operator!() const { return value_ == 0; }
 
   constexpr ReadOptions operator|(ReadOptions const& that) const
@@ -111,7 +111,7 @@ class AddressSpace {
 private:
   Process* process_;
 public:
-  AddressSpace(Process* process) : process_(process) {}
+  explicit AddressSpace(Process* process) : process_(process) {}
   virtual ~AddressSpace() = default;
 
   /** The process of this addres space
diff --git a/src/mc/LocationList.hpp b/src/mc/LocationList.hpp
index da1af221ce..58b9fe24b8 100644
--- a/src/mc/LocationList.hpp
+++ b/src/mc/LocationList.hpp
@@ -38,9 +38,7 @@ public:
   LocationListEntry(DwarfExpression expression, range_type range)
     : expression_(std::move(expression)), range_(range)
   {}
-  LocationListEntry(DwarfExpression expression)
-    : expression_(std::move(expression)), range_({0, UINT64_MAX})
-  {}
+  LocationListEntry(DwarfExpression expression) : expression_(std::move(expression)), range_({0, UINT64_MAX}) {}
 
   DwarfExpression& expression()
   {
@@ -67,9 +65,8 @@ private:
   void* memory_;
   int register_id_;
 public:
-  Location(void* x) :memory_(x) {}
-  Location(int register_id) :
-    memory_(nullptr), register_id_(register_id) {}
+  explicit Location(void* x) : memory_(x) {}
+  explicit Location(int register_id) : memory_(nullptr), register_id_(register_id) {}
   // Type of location:
   bool in_register() const { return memory_ == nullptr; }
   bool in_memory()   const { return memory_ != nullptr; }
diff --git a/src/mc/ObjectInformation.cpp b/src/mc/ObjectInformation.cpp
index 64fc01e284..015a60b345 100644
--- a/src/mc/ObjectInformation.cpp
+++ b/src/mc/ObjectInformation.cpp
@@ -106,7 +106,7 @@ void ObjectInformation::remove_global_variable(const char* name)
     int cmp = current_var.name.compare(name);
 
     if (cmp == 0) {
-  
+
       // Find the whole range:
       size_type first = cursor;
       while (first != 0 && this->global_variables[first - 1].name == name)
@@ -115,12 +115,12 @@ void ObjectInformation::remove_global_variable(const char* name)
       size_type last = cursor;
       while (last != size - 1 && this->global_variables[last + 1].name == name)
         last++;
-  
+
       // Remove the whole range:
       this->global_variables.erase(
         this->global_variables.begin() + first,
         this->global_variables.begin() + last + 1);
-  
+
       return;
     } else if (cmp < 0)
       first = cursor + 1;
diff --git a/src/mc/ObjectInformation.hpp b/src/mc/ObjectInformation.hpp
index 61739713e6..bfebc40bbf 100644
--- a/src/mc/ObjectInformation.hpp
+++ b/src/mc/ObjectInformation.hpp
@@ -83,7 +83,7 @@ public:
    *
    * We need to efficiently find the function from any given instruction
    * address inside its range. This index is sorted by low_pc
-   * 
+   *
    * The entries are sorted by low_pc and a binary search can be used to look
    * them up. In order to have a better cache locality, we only keep the
    * information we need for the lookup in this vector. We could probably
diff --git a/src/mc/PageStore.cpp b/src/mc/PageStore.cpp
index feef524b97..56c33d47f9 100644
--- a/src/mc/PageStore.cpp
+++ b/src/mc/PageStore.cpp
@@ -224,7 +224,7 @@ XBT_TEST_SUITE("mc_page_store", "Page store");
 XBT_TEST_UNIT("base", test_mc_page_store, "Test adding/removing pages in the store")
 {
   using simgrid::mc::PageStore;
-  
+
   xbt_test_add("Init");
   std::size_t pagesize = (size_t) getpagesize();
   std::unique_ptr<PageStore> store
diff --git a/src/mc/Process.hpp b/src/mc/Process.hpp
index bb0a85d3a5..19c2e5eb4e 100644
--- a/src/mc/Process.hpp
+++ b/src/mc/Process.hpp
@@ -153,7 +153,7 @@ public:
       this->refresh_malloc_info();
     return this->heap_info.data();
   }
-  
+
   void clear_cache()
   {
     this->cache_flags_ = Process::cache_none;
diff --git a/src/mc/RegionSnapshot.cpp b/src/mc/RegionSnapshot.cpp
index 7884740527..3fbfe3e54e 100644
--- a/src/mc/RegionSnapshot.cpp
+++ b/src/mc/RegionSnapshot.cpp
@@ -145,6 +145,6 @@ RegionSnapshot sparse_region(RegionType region_type,
   region.page_data(std::move(page_data));
   return region;
 }
-  
+
 }
 }
diff --git a/src/mc/RegionSnapshot.hpp b/src/mc/RegionSnapshot.hpp
index cc135f432e..8fd9e60490 100644
--- a/src/mc/RegionSnapshot.hpp
+++ b/src/mc/RegionSnapshot.hpp
@@ -210,7 +210,7 @@ public:
     page_numbers_.clear();
     privatized_regions_.clear();
   }
-  
+
   void flat_data(Buffer data)
   {
     storage_type_ = StorageType::Flat;
diff --git a/src/mc/VisitedState.hpp b/src/mc/VisitedState.hpp
index 3ac5e7b15f..3665060744 100644
--- a/src/mc/VisitedState.hpp
+++ b/src/mc/VisitedState.hpp
@@ -24,7 +24,7 @@ struct XBT_PRIVATE VisitedState {
   int num          = 0; // unique id of that state in the storage of all stored IDs
   int original_num = 0; // num field of the VisitedState to which I was declared equal to (used for dot_output)
 
-  VisitedState(unsigned long state_number);
+  explicit VisitedState(unsigned long state_number);
   ~VisitedState();
 };
 
diff --git a/src/mc/checker/Checker.hpp b/src/mc/checker/Checker.hpp
index 3b7079f7b3..ce27ab2f43 100644
--- a/src/mc/checker/Checker.hpp
+++ b/src/mc/checker/Checker.hpp
@@ -34,7 +34,7 @@ namespace mc {
 class Checker {
   Session* session_;
 public:
-  Checker(Session& session);
+  explicit Checker(Session& session);
 
   // No copy:
   Checker(Checker const&) = delete;
diff --git a/src/mc/checker/CommunicationDeterminismChecker.cpp b/src/mc/checker/CommunicationDeterminismChecker.cpp
index 2410b9d1dc..cb723fb60e 100644
--- a/src/mc/checker/CommunicationDeterminismChecker.cpp
+++ b/src/mc/checker/CommunicationDeterminismChecker.cpp
@@ -185,7 +185,7 @@ void CommunicationDeterminismChecker::get_comm_pattern(xbt_dynar_t list, smx_sim
   if (call_type == MC_CALL_TYPE_SEND) {
     /* Create comm pattern */
     pattern->type = simgrid::mc::PatternCommunicationType::send;
-    pattern->comm_addr = simcall_comm_isend__get__result(request);
+    pattern->comm_addr = static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_isend__getraw__result(request));
 
     simgrid::mc::Remote<simgrid::kernel::activity::CommImpl> temp_synchro;
     mc_model_checker->process().read(temp_synchro,
@@ -223,7 +223,7 @@ void CommunicationDeterminismChecker::get_comm_pattern(xbt_dynar_t list, smx_sim
     }
   } else if (call_type == MC_CALL_TYPE_RECV) {
     pattern->type = simgrid::mc::PatternCommunicationType::receive;
-    pattern->comm_addr = simcall_comm_irecv__get__result(request);
+    pattern->comm_addr = static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_irecv__getraw__result(request));
 
     simgrid::smpi::Request mpi_request;
     mc_model_checker->process().read(&mpi_request,
diff --git a/src/mc/checker/CommunicationDeterminismChecker.hpp b/src/mc/checker/CommunicationDeterminismChecker.hpp
index f14abb171b..aa07b43554 100644
--- a/src/mc/checker/CommunicationDeterminismChecker.hpp
+++ b/src/mc/checker/CommunicationDeterminismChecker.hpp
@@ -22,7 +22,7 @@ namespace mc {
 
 class XBT_PRIVATE CommunicationDeterminismChecker : public Checker {
 public:
-  CommunicationDeterminismChecker(Session& session);
+  explicit CommunicationDeterminismChecker(Session& session);
   ~CommunicationDeterminismChecker();
   void run() override;
   RecordTrace getRecordTrace() override;
diff --git a/src/mc/checker/LivenessChecker.hpp b/src/mc/checker/LivenessChecker.hpp
index f57a1ff7d2..2722fc5a37 100644
--- a/src/mc/checker/LivenessChecker.hpp
+++ b/src/mc/checker/LivenessChecker.hpp
@@ -37,7 +37,7 @@ struct XBT_PRIVATE Pair {
   int depth = 0;
   bool exploration_started = false;
 
-  Pair(unsigned long expanded_pairs);
+  explicit Pair(unsigned long expanded_pairs);
   ~Pair() = default;
 
   Pair(Pair const&) = delete;
diff --git a/src/mc/checker/SafetyChecker.cpp b/src/mc/checker/SafetyChecker.cpp
index 4256b56a26..36d4952cfd 100644
--- a/src/mc/checker/SafetyChecker.cpp
+++ b/src/mc/checker/SafetyChecker.cpp
@@ -335,6 +335,6 @@ Checker* createSafetyChecker(Session& session)
 {
   return new SafetyChecker(session);
 }
-  
+
 }
 }
diff --git a/src/mc/mc_base.cpp b/src/mc/mc_base.cpp
index ee5098ee8b..f1b7d8054c 100644
--- a/src/mc/mc_base.cpp
+++ b/src/mc/mc_base.cpp
@@ -79,7 +79,7 @@ bool request_is_enabled(smx_simcall_t req)
   {
     /* FIXME: check also that src and dst processes are not suspended */
     simgrid::kernel::activity::CommImpl* act =
-        static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__get__comm(req));
+        static_cast<simgrid::kernel::activity::CommImpl*>(&*simcall_comm_wait__get__comm(req));
 
 #if SIMGRID_HAVE_MC
     // Fetch from MCed memory:
@@ -107,7 +107,7 @@ bool request_is_enabled(smx_simcall_t req)
   case SIMCALL_COMM_WAITANY: {
     xbt_dynar_t comms;
     simgrid::kernel::activity::CommImpl* act =
-        static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__get__comm(req));
+        static_cast<simgrid::kernel::activity::CommImpl*>(&*simcall_comm_wait__get__comm(req));
 
 #if SIMGRID_HAVE_MC
     s_xbt_dynar_t comms_buffer;
diff --git a/src/mc/mc_checkpoint.cpp b/src/mc/mc_checkpoint.cpp
index cc77840d18..e4fa6e445a 100644
--- a/src/mc/mc_checkpoint.cpp
+++ b/src/mc/mc_checkpoint.cpp
@@ -439,7 +439,7 @@ static std::vector<s_mc_snapshot_stack_t> take_snapshot_stacks(simgrid::mc::Snap
 static void snapshot_handle_ignore(simgrid::mc::Snapshot* snapshot)
 {
   xbt_assert(snapshot->process());
-  
+
   // Copy the memory:
   for (auto const& region : mc_model_checker->process().ignored_regions()) {
     s_mc_snapshot_ignored_data_t ignored_data;
@@ -612,7 +612,7 @@ void restore_snapshot_fds(simgrid::mc::Snapshot* snapshot)
   xbt_die("FD snapshot not implemented in client/server mode.");
 
   for (auto const& fd : snapshot->current_fds) {
-    
+
     int new_fd = open(fd.filename.c_str(), fd.flags);
     if (new_fd < 0)
       xbt_die("Could not reopen the file %s fo restoring the file descriptor", fd.filename.c_str());
diff --git a/src/mc/mc_comm_pattern.cpp b/src/mc/mc_comm_pattern.cpp
index c6eede3fb3..dfba80e066 100644
--- a/src/mc/mc_comm_pattern.cpp
+++ b/src/mc/mc_comm_pattern.cpp
@@ -88,7 +88,9 @@ void MC_handle_comm_pattern(
     {
     simgrid::mc::RemotePtr<simgrid::kernel::activity::CommImpl> comm_addr = nullptr;
     if (call_type == MC_CALL_TYPE_WAIT)
-      comm_addr = remote(static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__get__comm(req)));
+      comm_addr = remote(static_cast<simgrid::kernel::activity::CommImpl*>(
+          simgrid::simix::unmarshal<simgrid::kernel::activity::ActivityImpl*>(req->result)));
+
     else {
       simgrid::kernel::activity::CommImpl* addr;
       // comm_addr = REMOTE(xbt_dynar_get_as(simcall_comm_waitany__get__comms(req), value, smx_synchro_t)):
diff --git a/src/mc/mc_request.cpp b/src/mc/mc_request.cpp
index 06e09a48d6..efa8aa7dd0 100644
--- a/src/mc/mc_request.cpp
+++ b/src/mc/mc_request.cpp
@@ -23,9 +23,9 @@ static inline simgrid::kernel::activity::CommImpl* MC_get_comm(smx_simcall_t r)
 {
   switch (r->call ) {
   case SIMCALL_COMM_WAIT:
-    return static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__get__comm(r));
+    return static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__getraw__comm(r));
   case SIMCALL_COMM_TEST:
-    return static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_test__get__comm(r));
+    return static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_test__getraw__comm(r));
   default:
     return nullptr;
   }
@@ -257,7 +257,7 @@ std::string simgrid::mc::request_to_string(smx_simcall_t req, int value, simgrid
 
   case SIMCALL_COMM_WAIT: {
     simgrid::kernel::activity::CommImpl* remote_act =
-        static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__get__comm(req));
+        static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__getraw__comm(req));
     char* p;
     if (value == -1) {
       type = "WaitTimeout";
@@ -290,7 +290,7 @@ std::string simgrid::mc::request_to_string(smx_simcall_t req, int value, simgrid
 
   case SIMCALL_COMM_TEST: {
     simgrid::kernel::activity::CommImpl* remote_act =
-        static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_test__get__comm(req));
+        static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_test__getraw__comm(req));
     simgrid::mc::Remote<simgrid::kernel::activity::CommImpl> temp_synchro;
     simgrid::kernel::activity::CommImpl* act;
     if (use_remote_comm) {
@@ -329,7 +329,7 @@ std::string simgrid::mc::request_to_string(smx_simcall_t req, int value, simgrid
       read_element(mc_model_checker->process(),
         &remote_sync, remote(simcall_comm_waitany__get__comms(req)), value,
         sizeof(remote_sync));
-      char* p = pointer_to_string(remote_sync);
+      char* p = pointer_to_string(&*remote_sync);
       args = bprintf("comm=%s (%d of %lu)",
         p, value + 1, xbt_dynar_length(&comms));
       xbt_free(p);
@@ -404,25 +404,22 @@ namespace mc {
 
 bool request_is_enabled_by_idx(smx_simcall_t req, unsigned int idx)
 {
-  smx_activity_t remote_act = nullptr;
+  simgrid::kernel::activity::ActivityImpl* remote_act = nullptr;
   switch (req->call) {
 
   case SIMCALL_COMM_WAIT:
     /* FIXME: check also that src and dst processes are not suspended */
-    remote_act = simcall_comm_wait__get__comm(req);
+    remote_act = simcall_comm_wait__getraw__comm(req);
     break;
 
   case SIMCALL_COMM_WAITANY: {
-    read_element(
-      mc_model_checker->process(), &remote_act,
-      remote(simcall_comm_waitany__get__comms(req)),
-      idx, sizeof(remote_act));
+    read_element(mc_model_checker->process(), &remote_act, remote(simcall_comm_waitany__getraw__comms(req)), idx,
+                 sizeof(remote_act));
     }
     break;
 
   case SIMCALL_COMM_TESTANY:
-    remote_act = mc_model_checker->process().read(remote(
-      simcall_comm_testany__get__comms(req) + idx));
+    remote_act = mc_model_checker->process().read(remote(simcall_comm_testany__getraw__comms(req) + idx));
     break;
 
   default:
@@ -489,7 +486,7 @@ std::string request_get_dot_output(smx_simcall_t req, int value)
       else
         label = simgrid::xbt::string_printf("[(%lu)] WaitTimeout", issuer->pid);
     } else {
-      smx_activity_t remote_act = simcall_comm_wait__get__comm(req);
+      simgrid::kernel::activity::ActivityImpl* remote_act = simcall_comm_wait__getraw__comm(req);
       simgrid::mc::Remote<simgrid::kernel::activity::CommImpl> temp_comm;
       mc_model_checker->process().read(temp_comm,
                                        remote(static_cast<simgrid::kernel::activity::CommImpl*>(remote_act)));
@@ -511,7 +508,7 @@ std::string request_get_dot_output(smx_simcall_t req, int value)
   }
 
   case SIMCALL_COMM_TEST: {
-    smx_activity_t remote_act = simcall_comm_test__get__comm(req);
+    simgrid::kernel::activity::ActivityImpl* remote_act = simcall_comm_test__getraw__comm(req);
     simgrid::mc::Remote<simgrid::kernel::activity::CommImpl> temp_comm;
     mc_model_checker->process().read(temp_comm, remote(static_cast<simgrid::kernel::activity::CommImpl*>(remote_act)));
     simgrid::kernel::activity::CommImpl* comm = temp_comm.getBuffer();
diff --git a/src/mc/mc_state.cpp b/src/mc/mc_state.cpp
index e399fa2979..f606492d7b 100644
--- a/src/mc/mc_state.cpp
+++ b/src/mc/mc_state.cpp
@@ -130,7 +130,7 @@ static inline smx_simcall_t MC_state_get_request_for_process(
 
       case SIMCALL_COMM_WAIT: {
         simgrid::mc::RemotePtr<simgrid::kernel::activity::CommImpl> remote_act =
-            remote(static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__get__comm(&actor->simcall)));
+            remote(static_cast<simgrid::kernel::activity::CommImpl*>(simcall_comm_wait__getraw__comm(&actor->simcall)));
         simgrid::mc::Remote<simgrid::kernel::activity::CommImpl> temp_act;
         mc_model_checker->process().read(temp_act, remote_act);
         simgrid::kernel::activity::CommImpl* act = temp_act.getBuffer();
@@ -176,10 +176,9 @@ static inline smx_simcall_t MC_state_get_request_for_process(
   switch (req->call) {
   case SIMCALL_COMM_WAITANY: {
     state->internal_req.call = SIMCALL_COMM_WAIT;
-    smx_activity_t remote_comm;
-    read_element(mc_model_checker->process(),
-      &remote_comm, remote(simcall_comm_waitany__get__comms(req)),
-      state->transition.argument, sizeof(remote_comm));
+    simgrid::kernel::activity::ActivityImpl* remote_comm;
+    read_element(mc_model_checker->process(), &remote_comm, remote(simcall_comm_waitany__getraw__comms(req)),
+                 state->transition.argument, sizeof(remote_comm));
     mc_model_checker->process().read(state->internal_comm,
                                      remote(static_cast<simgrid::kernel::activity::CommImpl*>(remote_comm)));
     simcall_comm_wait__set__comm(&state->internal_req, state->internal_comm.getBuffer());
@@ -191,8 +190,8 @@ static inline smx_simcall_t MC_state_get_request_for_process(
     state->internal_req.call = SIMCALL_COMM_TEST;
 
     if (state->transition.argument > 0) {
-      smx_activity_t remote_comm = mc_model_checker->process().read(
-        remote(simcall_comm_testany__get__comms(req) + state->transition.argument));
+      simgrid::kernel::activity::ActivityImpl* remote_comm = mc_model_checker->process().read(
+          remote(simcall_comm_testany__getraw__comms(req) + state->transition.argument));
       mc_model_checker->process().read(state->internal_comm,
                                        remote(static_cast<simgrid::kernel::activity::CommImpl*>(remote_comm)));
     }
@@ -202,15 +201,15 @@ static inline smx_simcall_t MC_state_get_request_for_process(
     break;
 
   case SIMCALL_COMM_WAIT:
-    mc_model_checker->process().read_bytes(&state->internal_comm ,
-      sizeof(state->internal_comm), remote(simcall_comm_wait__get__comm(req)));
+    mc_model_checker->process().read_bytes(&state->internal_comm, sizeof(state->internal_comm),
+                                           remote(simcall_comm_wait__getraw__comm(req)));
     simcall_comm_wait__set__comm(&state->executed_req, state->internal_comm.getBuffer());
     simcall_comm_wait__set__comm(&state->internal_req, state->internal_comm.getBuffer());
     break;
 
   case SIMCALL_COMM_TEST:
-    mc_model_checker->process().read_bytes(&state->internal_comm,
-      sizeof(state->internal_comm), remote(simcall_comm_test__get__comm(req)));
+    mc_model_checker->process().read_bytes(&state->internal_comm, sizeof(state->internal_comm),
+                                           remote(simcall_comm_test__getraw__comm(req)));
     simcall_comm_test__set__comm(&state->executed_req, state->internal_comm.getBuffer());
     simcall_comm_test__set__comm(&state->internal_req, state->internal_comm.getBuffer());
     break;
diff --git a/src/mc/mc_state.h b/src/mc/mc_state.h
index e4eb0d9d30..d6ee81450e 100644
--- a/src/mc/mc_state.h
+++ b/src/mc/mc_state.h
@@ -24,7 +24,7 @@ enum class PatternCommunicationType {
 
 struct PatternCommunication {
   int num = 0;
-  smx_activity_t comm_addr;
+  simgrid::kernel::activity::CommImpl* comm_addr;
   PatternCommunicationType type = PatternCommunicationType::send;
   unsigned long src_proc = 0;
   unsigned long dst_proc = 0;
diff --git a/src/msg/instr_msg_task.cpp b/src/msg/instr_msg_task.cpp
index 65b69cc8e3..fb5862813c 100644
--- a/src/msg/instr_msg_task.cpp
+++ b/src/msg/instr_msg_task.cpp
@@ -8,6 +8,8 @@
 #include "src/instr/instr_private.h"
 #include "src/msg/msg_private.h"
 
+#include <atomic>
+
 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(instr_msg, instr, "MSG instrumentation");
 
 void TRACE_msg_set_task_category(msg_task_t task, const char *category)
@@ -31,10 +33,10 @@ void TRACE_msg_set_task_category(msg_task_t task, const char *category)
 /* MSG_task_create related function*/
 void TRACE_msg_task_create(msg_task_t task)
 {
-  static long long counter = 0;
+  static std::atomic_ullong counter{0};
   task->counter = counter++;
   task->category = nullptr;
-  
+
   if(MC_is_active())
     MC_ignore_heap(&(task->counter), sizeof(task->counter));
 
diff --git a/src/msg/msg_deployment.cpp b/src/msg/msg_deployment.cpp
index e188b59254..a5089f6258 100644
--- a/src/msg/msg_deployment.cpp
+++ b/src/msg/msg_deployment.cpp
@@ -15,7 +15,7 @@ SG_BEGIN_DECL()
  *
  *     \include simgrid.dtd
  *
- * Here is a small example of such a platform 
+ * Here is a small example of such a platform
  *
  *     \include msg/masterslave/deployment_masterslave.xml
  *
@@ -30,8 +30,8 @@ void MSG_launch_application(const char *file)
 /** \ingroup msg_simulation
  * \brief Registers the main function of a process in a global table.
  *
- * Registers a code function in a global table. 
- * This table is then used by #MSG_launch_application. 
+ * Registers a code function in a global table.
+ * This table is then used by #MSG_launch_application.
  * \param name the reference name of the function.
  * \param code the function (must have the same prototype than the main function of any C program: int ..(int argc, char *argv[]))
  */
diff --git a/src/msg/msg_environment.cpp b/src/msg/msg_environment.cpp
index fc067f7130..1a62573380 100644
--- a/src/msg/msg_environment.cpp
+++ b/src/msg/msg_environment.cpp
@@ -25,7 +25,7 @@ SG_BEGIN_DECL()
  *
  *     \include simgrid.dtd
  *
- * Here is a small example of such a platform 
+ * Here is a small example of such a platform
  *
  *     \include small_platform.xml
  *
@@ -36,17 +36,6 @@ void MSG_create_environment(const char *file)
   SIMIX_create_environment(file);
 }
 
-void MSG_post_create_environment() {
-  xbt_lib_cursor_t cursor;
-  void **data;
-  char *name;
-
-  /* Initialize MSG storages */
-  xbt_lib_foreach(storage_lib, cursor, name, data) {
-    __MSG_storage_create(xbt_dict_cursor_get_elm(cursor));
-  }
-}
-
 msg_netzone_t MSG_zone_get_root()
 {
   return simgrid::s4u::Engine::instance()->netRoot();
diff --git a/src/msg/msg_global.cpp b/src/msg/msg_global.cpp
index 2dd6fc906b..7c2e367bf1 100644
--- a/src/msg/msg_global.cpp
+++ b/src/msg/msg_global.cpp
@@ -51,8 +51,6 @@ void MSG_init_nocheck(int *argc, char **argv) {
     SIMIX_function_register_process_create(MSG_process_create_from_SIMIX);
     SIMIX_function_register_process_cleanup(MSG_process_cleanup_from_SIMIX);
 
-    simgrid::s4u::onPlatformCreated.connect(MSG_post_create_environment);
-
     simgrid::MsgHostExt::EXTENSION_ID = simgrid::s4u::Host::extension_create<simgrid::MsgHostExt>();
     simgrid::s4u::Host::onCreation.connect([](simgrid::s4u::Host& host) {
       host.extension_set<simgrid::MsgHostExt>(new simgrid::MsgHostExt());
@@ -64,8 +62,6 @@ void MSG_init_nocheck(int *argc, char **argv) {
     MC_ignore_heap(&(msg_global->sent_msg), sizeof(msg_global->sent_msg));
   }
 
-  XBT_DEBUG("ADD MSG LEVELS");
-  MSG_STORAGE_LEVEL = xbt_lib_add_level(storage_lib, (void_f_pvoid_t) __MSG_storage_destroy);
   if (xbt_cfg_get_boolean("clean-atexit"))
     atexit(MSG_exit);
 }
diff --git a/src/msg/msg_gos.cpp b/src/msg/msg_gos.cpp
index 3e2eda2192..7f2e4b4bb8 100644
--- a/src/msg/msg_gos.cpp
+++ b/src/msg/msg_gos.cpp
@@ -5,6 +5,7 @@
 
 #include <xbt/ex.hpp>
 
+#include "src/kernel/activity/ExecImpl.hpp"
 #include "src/msg/msg_private.h"
 #include "src/simix/smx_private.h" /* MSG_task_listen looks inside the rdv directly. Not clean. */
 
@@ -66,12 +67,13 @@ msg_error_t MSG_parallel_task_execute_with_timeout(msg_task_t task, double timeo
     simdata->setUsed();
 
     if (simdata->host_nb > 0) {
-      simdata->compute = static_cast<simgrid::kernel::activity::ExecImpl*>(simcall_execution_parallel_start(
-          task->name, simdata->host_nb, simdata->host_list, simdata->flops_parallel_amount,
-          simdata->bytes_parallel_amount, 1.0, -1.0, timeout));
-      XBT_DEBUG("Parallel execution action created: %p", simdata->compute);
+      simdata->compute =
+          boost::static_pointer_cast<simgrid::kernel::activity::ExecImpl>(simcall_execution_parallel_start(
+              task->name, simdata->host_nb, simdata->host_list, simdata->flops_parallel_amount,
+              simdata->bytes_parallel_amount, 1.0, -1.0, timeout));
+      XBT_DEBUG("Parallel execution action created: %p", simdata->compute.get());
     } else {
-      simdata->compute = static_cast<simgrid::kernel::activity::ExecImpl*>(
+      simdata->compute = boost::static_pointer_cast<simgrid::kernel::activity::ExecImpl>(
           simcall_execution_start(task->name, simdata->flops_amount, simdata->priority, simdata->bound));
     }
     simcall_set_category(simdata->compute, task->category);
@@ -124,13 +126,13 @@ msg_error_t MSG_process_sleep(double nb_sec)
   }
   catch(xbt_ex& e) {
     if (e.category == cancel_error) {
-      XBT_DEBUG("According to the JAVA API, a sleep call should only deal with HostFailureException, I'm lost."); 
+      XBT_DEBUG("According to the JAVA API, a sleep call should only deal with HostFailureException, I'm lost.");
       // adsein: MSG_TASK_CANCELED is assigned when someone kills the process that made the sleep, this is not
       // correct. For instance, when the node is turned off, the error should be MSG_HOST_FAILURE, which is by the way
       // and according to the JAVA document, the only exception that can be triggered by MSG_Process_sleep call.
       // To avoid possible impacts in the code, I just raised a host_failure exception for the moment in the JAVA code
       // and did not change anythings at the C level.
-      // See comment in the jmsg_process.c file, function JNIEXPORT void JNICALL Java_org_simgrid_msg_Process_sleep(JNIEnv *env, jclass cls, jlong jmillis, jint jnanos) 
+      // See comment in the jmsg_process.c file, function JNIEXPORT void JNICALL Java_org_simgrid_msg_Process_sleep(JNIEnv *env, jclass cls, jlong jmillis, jint jnanos)
       status = MSG_TASK_CANCELED;
     } else
       throw;
@@ -315,7 +317,7 @@ static inline msg_comm_t MSG_task_isend_internal(msg_task_t task, const char *al
   /* Send it by calling SIMIX network layer */
   smx_activity_t act = simcall_comm_isend(myself->getImpl(), mailbox->getImpl(), t_simdata->bytes_amount, t_simdata->rate,
                                          task, sizeof(void *), match_fun, cleanup, nullptr, match_data,detached);
-  t_simdata->comm = static_cast<simgrid::kernel::activity::CommImpl*>(act);
+  t_simdata->comm = boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(act);
 
   msg_comm_t comm = nullptr;
   if (not detached) {
@@ -489,7 +491,6 @@ int MSG_comm_test(msg_comm_t comm)
     if (finished && comm->task_received != nullptr) {
       /* I am the receiver */
       (*comm->task_received)->simdata->setNotUsed();
-      comm->s_comm->unref();
     }
   }
   catch (xbt_ex& e) {
@@ -522,7 +523,7 @@ int MSG_comm_testany(xbt_dynar_t comms)
   int finished_index = -1;
 
   /* Create the equivalent array with SIMIX objects: */
-  std::vector<simgrid::kernel::activity::ActivityImpl*> s_comms;
+  std::vector<simgrid::kernel::activity::ActivityImplPtr> s_comms;
   s_comms.reserve(xbt_dynar_length(comms));
   msg_comm_t comm;
   unsigned int cursor;
@@ -557,7 +558,6 @@ int MSG_comm_testany(xbt_dynar_t comms)
     if (status == MSG_OK && comm->task_received != nullptr) {
       /* I am the receiver */
       (*comm->task_received)->simdata->setNotUsed();
-      comm->s_comm->unref();
     }
   }
 
@@ -586,7 +586,6 @@ msg_error_t MSG_comm_wait(msg_comm_t comm, double timeout)
 {
   try {
     simcall_comm_wait(comm->s_comm, timeout);
-    comm->s_comm->unref();
 
     if (comm->task_received != nullptr) {
       /* I am the receiver */
@@ -635,11 +634,14 @@ int MSG_comm_waitany(xbt_dynar_t comms)
   int finished_index = -1;
 
   /* create the equivalent dynar with SIMIX objects */
-  xbt_dynar_t s_comms = xbt_dynar_new(sizeof(smx_activity_t), nullptr);
+  xbt_dynar_t s_comms = xbt_dynar_new(sizeof(smx_activity_t), [](void*ptr){
+    intrusive_ptr_release(*(simgrid::kernel::activity::ActivityImpl**)ptr);
+  });
   msg_comm_t comm;
   unsigned int cursor;
   xbt_dynar_foreach(comms, cursor, comm) {
-    xbt_dynar_push(s_comms, &comm->s_comm);
+    intrusive_ptr_add_ref(comm->s_comm.get());
+    xbt_dynar_push_as(s_comms, simgrid::kernel::activity::ActivityImpl*, comm->s_comm.get());
   }
 
   msg_error_t status = MSG_OK;
@@ -671,7 +673,6 @@ int MSG_comm_waitany(xbt_dynar_t comms)
   if (comm->task_received != nullptr) {
     /* I am the receiver */
     (*comm->task_received)->simdata->setNotUsed();
-    comm->s_comm->unref();
   }
 
   return finished_index;
@@ -710,7 +711,8 @@ msg_task_t MSG_comm_get_task(msg_comm_t comm)
  */
 void MSG_comm_copy_data_from_SIMIX(smx_activity_t synchro, void* buff, size_t buff_size)
 {
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
 
   SIMIX_comm_copy_pointer_callback(comm, buff, buff_size);
 
@@ -797,9 +799,8 @@ msg_error_t MSG_task_send_with_timeout(msg_task_t task, const char *alias, doubl
                               t_simdata->rate, task, sizeof(void *), nullptr, nullptr, nullptr, task, 0);
     if (TRACE_is_enabled())
       simcall_set_category(comm, task->category);
-    t_simdata->comm = static_cast<simgrid::kernel::activity::CommImpl*>(comm);
+    t_simdata->comm = boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(comm);
     simcall_comm_wait(comm, timeout);
-    comm->unref();
   }
   catch (xbt_ex& e) {
     switch (e.category) {
@@ -868,7 +869,8 @@ int MSG_task_listen(const char *alias)
 int MSG_task_listen_from(const char *alias)
 {
   simgrid::s4u::MailboxPtr mbox = simgrid::s4u::Mailbox::byName(alias);
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(mbox->front());
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(mbox->front());
 
   if (not comm)
     return -1;
diff --git a/src/msg/msg_host.cpp b/src/msg/msg_host.cpp
index 43a6b24aac..f8e8cd7ecd 100644
--- a/src/msg/msg_host.cpp
+++ b/src/msg/msg_host.cpp
@@ -20,7 +20,7 @@ int sg_storage_max_file_descriptors = 1024;
 
 /** @addtogroup m_host_management
  * (#msg_host_t) and the functions for managing it.
- *  
+ *
  *  A <em>location</em> (or <em>host</em>) is any possible place where  a process may run. Thus it may be represented
  *  as a <em>physical resource with computing capabilities</em>, some <em>mailboxes</em> to enable running process to
  *  communicate with remote ones, and some <em>private data</em> that can be only accessed by local process.
@@ -271,7 +271,7 @@ xbt_dict_t MSG_host_get_storage_content(msg_host_t host)
   xbt_dict_t storage_list = host->mountedStoragesAsDict();
 
   xbt_dict_foreach(storage_list,cursor,mount_name,storage_name){
-    storage = static_cast<msg_storage_t>(xbt_lib_get_elm_or_null(storage_lib,storage_name));
+    storage            = simgrid::s4u::Storage::byName(storage_name);
     xbt_dict_t content = MSG_storage_get_content(storage);
     xbt_dict_set(contents,mount_name, content,nullptr);
   }
diff --git a/src/msg/msg_io.cpp b/src/msg/msg_io.cpp
index ef67c5befd..eeb421e424 100644
--- a/src/msg/msg_io.cpp
+++ b/src/msg/msg_io.cpp
@@ -4,7 +4,9 @@
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
 #include "../surf/StorageImpl.hpp"
+#include "simgrid/s4u/File.hpp"
 #include "simgrid/s4u/Host.hpp"
+#include "simgrid/s4u/Storage.hpp"
 #include "src/msg/msg_private.h"
 #include <numeric>
 
@@ -18,21 +20,6 @@ SG_BEGIN_DECL()
  *  \see #msg_file_t
  */
 
-/********************************* File **************************************/
-void __MSG_file_get_info(msg_file_t fd){
-
-  xbt_dynar_t info = simcall_file_get_info(fd->simdata->smx_file);
-  sg_size_t *psize;
-
-  fd->storage_type = xbt_dynar_pop_as(info, char*);
-  fd->storageId    = xbt_dynar_pop_as(info, char*);
-  fd->mount_point  = xbt_dynar_pop_as(info, char*);
-  psize            = xbt_dynar_pop_as(info, sg_size_t*);
-  fd->size         = *psize;
-  xbt_free(psize);
-  xbt_dynar_free_container(&info);
-}
-
 static int MSG_host_get_file_descriptor_id(msg_host_t host)
 {
   simgrid::MsgHostExt* priv = host->extension<simgrid::MsgHostExt>();
@@ -59,7 +46,7 @@ static void MSG_host_release_file_descriptor_id(msg_host_t host, int id)
  */
 msg_error_t MSG_file_set_data(msg_file_t fd, void *data)
 {
-  fd->data = data;
+  fd->setUserdata(data);
   return MSG_OK;
 }
 
@@ -71,7 +58,7 @@ msg_error_t MSG_file_set_data(msg_file_t fd, void *data)
  */
 void *MSG_file_get_data(msg_file_t fd)
 {
-  return fd->data;
+  return fd->userdata();
 }
 
 /** \ingroup msg_file
@@ -80,9 +67,6 @@ void *MSG_file_get_data(msg_file_t fd)
  * \param fd is a the file descriptor
  */
 void MSG_file_dump (msg_file_t fd){
-  /* Update the cached information first */
-  __MSG_file_get_info(fd);
-
   XBT_INFO("File Descriptor information:\n"
            "\t\tFull path: '%s'\n"
            "\t\tSize: %llu\n"
@@ -90,7 +74,7 @@ void MSG_file_dump (msg_file_t fd){
            "\t\tStorage Id: '%s'\n"
            "\t\tStorage Type: '%s'\n"
            "\t\tFile Descriptor Id: %d",
-           fd->fullpath, fd->size, fd->mount_point, fd->storageId, fd->storage_type, fd->desc_id);
+           fd->path(), fd->size(), fd->mount_point, fd->storageId, fd->storage_type, fd->desc_id);
 }
 
 /** \ingroup msg_file
@@ -104,18 +88,17 @@ sg_size_t MSG_file_read(msg_file_t fd, sg_size_t size)
 {
   sg_size_t read_size;
 
-  if (fd->size == 0) /* Nothing to read, return */
+  if (fd->size() == 0) /* Nothing to read, return */
     return 0;
 
   /* Find the host where the file is physically located and read it */
-  msg_storage_t storage_src           = static_cast<msg_storage_t>(xbt_lib_get_elm_or_null(storage_lib, fd->storageId));
-  msg_storage_priv_t storage_priv_src = MSG_storage_priv(storage_src);
-  msg_host_t attached_host            = MSG_host_by_name(storage_priv_src->hostname);
-  read_size                           = simcall_file_read(fd->simdata->smx_file, size, attached_host);
+  msg_storage_t storage_src           = simgrid::s4u::Storage::byName(fd->storageId);
+  msg_host_t attached_host            = MSG_host_by_name(storage_src->host());
+  read_size                           = fd->read(size); // TODO re-add attached_host;
 
-  if (strcmp(storage_priv_src->hostname, MSG_host_self()->cname())) {
+  if (strcmp(attached_host->cname(), MSG_host_self()->cname())) {
     /* the file is hosted on a remote host, initiate a communication between src and dest hosts for data transfer */
-    XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", storage_priv_src->hostname, read_size);
+    XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", attached_host->cname(), read_size);
     msg_host_t m_host_list[] = {MSG_host_self(), attached_host};
     double flops_amount[]    = {0, 0};
     double bytes_amount[]    = {0, 0, static_cast<double>(read_size), 0};
@@ -150,13 +133,12 @@ sg_size_t MSG_file_write(msg_file_t fd, sg_size_t size)
     return 0;
 
   /* Find the host where the file is physically located (remote or local)*/
-  msg_storage_t storage_src           = static_cast<msg_storage_t>(xbt_lib_get_elm_or_null(storage_lib, fd->storageId));
-  msg_storage_priv_t storage_priv_src = MSG_storage_priv(storage_src);
-  msg_host_t attached_host            = MSG_host_by_name(storage_priv_src->hostname);
+  msg_storage_t storage_src = simgrid::s4u::Storage::byName(fd->storageId);
+  msg_host_t attached_host  = MSG_host_by_name(storage_src->host());
 
-  if (strcmp(storage_priv_src->hostname, MSG_host_self()->cname())) {
+  if (strcmp(attached_host->cname(), MSG_host_self()->cname())) {
     /* the file is hosted on a remote host, initiate a communication between src and dest hosts for data transfer */
-    XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", storage_priv_src->hostname, size);
+    XBT_DEBUG("File is on %s remote host, initiate data transfer of %llu bytes.", attached_host->cname(), size);
     msg_host_t m_host_list[] = {MSG_host_self(), attached_host};
     double flops_amount[]    = {0, 0};
     double bytes_amount[]    = {0, static_cast<double>(size), 0, 0};
@@ -176,9 +158,8 @@ sg_size_t MSG_file_write(msg_file_t fd, sg_size_t size)
     }
   }
   /* Write file on local or remote host */
-  sg_size_t offset     = simcall_file_tell(fd->simdata->smx_file);
-  sg_size_t write_size = simcall_file_write(fd->simdata->smx_file, size, attached_host);
-  fd->size             = offset + write_size;
+  // sg_size_t offset     = fd->tell();
+  sg_size_t write_size = fd->write(size); // TODO readd attached_host;
 
   return write_size;
 }
@@ -193,15 +174,8 @@ sg_size_t MSG_file_write(msg_file_t fd, sg_size_t size)
  */
 msg_file_t MSG_file_open(const char* fullpath, void* data)
 {
-  msg_file_t fd         = xbt_new(s_msg_file_priv_t, 1);
-  fd->data              = data;
-  fd->fullpath          = xbt_strdup(fullpath);
-  fd->simdata           = xbt_new0(s_simdata_file_t, 1);
-  fd->simdata->smx_file = simcall_file_open(fullpath, MSG_host_self());
+  msg_file_t fd         = new simgrid::s4u::File(fullpath, MSG_host_self());
   fd->desc_id           = MSG_host_get_file_descriptor_id(MSG_host_self());
-
-  __MSG_file_get_info(fd);
-
   return fd;
 }
 
@@ -213,14 +187,9 @@ msg_file_t MSG_file_open(const char* fullpath, void* data)
  */
 int MSG_file_close(msg_file_t fd)
 {
-  if (fd->data)
-    xbt_free(fd->data);
-
-  int res = simcall_file_close(fd->simdata->smx_file, MSG_host_self());
-  MSG_host_release_file_descriptor_id(MSG_host_self(), fd->desc_id);
-  __MSG_file_destroy(fd);
+  delete fd;
 
-  return res;
+  return MSG_OK;
 }
 
 /** \ingroup msg_file
@@ -232,12 +201,10 @@ int MSG_file_close(msg_file_t fd)
 msg_error_t MSG_file_unlink(msg_file_t fd)
 {
   /* Find the host where the file is physically located (remote or local)*/
-  msg_storage_t storage_src           = static_cast<msg_storage_t>(xbt_lib_get_elm_or_null(storage_lib, fd->storageId));
-  msg_storage_priv_t storage_priv_src = MSG_storage_priv(storage_src);
-  msg_host_t attached_host            = MSG_host_by_name(storage_priv_src->hostname);
-  int res                             = simcall_file_unlink(fd->simdata->smx_file, attached_host);
-  __MSG_file_destroy(fd);
-  return static_cast<msg_error_t>(res);
+  // msg_storage_t storage_src           = simgrid::s4u::Storage::byName(fd->storageId);
+  // msg_host_t attached_host            = MSG_host_by_name(storage_src->host());
+  fd->unlink(); // simcall_file_unlink(fd->simdata->smx_file, attached_host);
+  return MSG_OK;
 }
 
 /** \ingroup msg_file
@@ -247,7 +214,7 @@ msg_error_t MSG_file_unlink(msg_file_t fd)
  * \return the size of the file (as a #sg_size_t)
  */
 sg_size_t MSG_file_get_size(msg_file_t fd){
-  return simcall_file_get_size(fd->simdata->smx_file);
+  return fd->size();
 }
 
 /**
@@ -264,7 +231,8 @@ sg_size_t MSG_file_get_size(msg_file_t fd){
  */
 msg_error_t MSG_file_seek(msg_file_t fd, sg_offset_t offset, int origin)
 {
-  return static_cast<msg_error_t>(simcall_file_seek(fd->simdata->smx_file, offset, origin));
+  fd->seek(offset); // TODO re-add origin
+  return MSG_OK;
 }
 
 /**
@@ -277,12 +245,12 @@ msg_error_t MSG_file_seek(msg_file_t fd, sg_offset_t offset, int origin)
  */
 sg_size_t MSG_file_tell(msg_file_t fd)
 {
-  return simcall_file_tell(fd->simdata->smx_file);
+  return fd->tell();
 }
 
 const char *MSG_file_get_name(msg_file_t fd) {
   xbt_assert((fd != nullptr), "Invalid parameters");
-  return fd->fullpath;
+  return fd->path();
 }
 
 /**
@@ -292,7 +260,8 @@ const char *MSG_file_get_name(msg_file_t fd) {
  */
 msg_error_t MSG_file_move (msg_file_t fd, const char* fullpath)
 {
-  return static_cast<msg_error_t>(simcall_file_move(fd->simdata->smx_file, fullpath));
+  fd->move(fullpath);
+  return MSG_OK;
 }
 
 /**
@@ -306,11 +275,10 @@ msg_error_t MSG_file_move (msg_file_t fd, const char* fullpath)
 msg_error_t MSG_file_rcopy (msg_file_t file, msg_host_t host, const char* fullpath)
 {
   /* Find the host where the file is physically located and read it */
-  msg_storage_t storage_src = static_cast<msg_storage_t>(xbt_lib_get_elm_or_null(storage_lib, file->storageId));
-  msg_storage_priv_t storage_priv_src = MSG_storage_priv(storage_src);
-  msg_host_t attached_host = MSG_host_by_name(storage_priv_src->hostname);
+  msg_storage_t storage_src = simgrid::s4u::Storage::byName(file->storageId);
+  msg_host_t attached_host  = MSG_host_by_name(storage_src->host());
   MSG_file_seek(file, 0, SEEK_SET);
-  sg_size_t read_size = simcall_file_read(file->simdata->smx_file, file->size, attached_host);
+  sg_size_t read_size = file->read(file->size());
 
   /* Find the real host destination where the file will be physically stored */
   xbt_dict_cursor_t cursor   = nullptr;
@@ -329,25 +297,22 @@ msg_error_t MSG_file_rcopy (msg_file_t file, msg_host_t host, const char* fullpa
     if (not strcmp(file_mount_name, mount_name) && strlen(mount_name) > longest_prefix_length) {
       /* The current mount name is found in the full path and is bigger than the previous*/
       longest_prefix_length = strlen(mount_name);
-      storage_dest          = (msg_storage_t)xbt_lib_get_elm_or_null(storage_lib, storage_name);
+      storage_dest          = simgrid::s4u::Storage::byName(storage_name);
     }
     xbt_free(file_mount_name);
   }
   xbt_dict_free(&storage_list);
 
-  char* host_name_dest = nullptr;
   if(longest_prefix_length>0){
     /* Mount point found, retrieve the host the storage is attached to */
-    msg_storage_priv_t storage_dest_priv = MSG_storage_priv(storage_dest);
-    host_name_dest = (char*)storage_dest_priv->hostname;
-    host_dest = MSG_host_by_name(host_name_dest);
+    host_dest = MSG_host_by_name(storage_dest->host());
   }else{
     XBT_WARN("Can't find mount point for '%s' on destination host '%s'", fullpath, host->cname());
     return MSG_TASK_CANCELED;
   }
 
-  XBT_DEBUG("Initiate data transfer of %llu bytes between %s and %s.", read_size, storage_priv_src->hostname,
-            host_name_dest);
+  XBT_DEBUG("Initiate data transfer of %llu bytes between %s and %s.", read_size, attached_host->cname(),
+            storage_dest->host());
   msg_host_t m_host_list[] = {attached_host, host_dest};
   double flops_amount[]    = {0, 0};
   double bytes_amount[]    = {0, static_cast<double>(read_size), 0, 0};
@@ -359,7 +324,7 @@ msg_error_t MSG_file_rcopy (msg_file_t file, msg_host_t host, const char* fullpa
 
   if(transfer != MSG_OK){
     if (transfer == MSG_HOST_FAILURE)
-      XBT_WARN("Transfer error, %s remote host just turned off!", host_name_dest);
+      XBT_WARN("Transfer error, %s remote host just turned off!", storage_dest->host());
     if (transfer == MSG_TASK_CANCELED)
       XBT_WARN("Transfer error, task has been canceled!");
 
@@ -388,40 +353,11 @@ msg_error_t MSG_file_rmove (msg_file_t file, msg_host_t host, const char* fullpa
   return res;
 }
 
-/**
- * \brief Destroys a file (internal call only)
- */
-void __MSG_file_destroy(msg_file_t file)
-{
-  xbt_free(file->fullpath);
-  xbt_free(file->simdata);
-  xbt_free(file);
-}
-
 /********************************* Storage **************************************/
 /** @addtogroup msg_storage_management
  * (#msg_storage_t) and the functions for managing it.
  */
 
-msg_storage_t __MSG_storage_create(smx_storage_t storage)
-{
-  msg_storage_priv_t storage_private = xbt_new0(s_msg_storage_priv_t, 1);
-
-  storage_private->name     = surf_storage_get_name(storage);
-  storage_private->hostname = surf_storage_get_host(storage);
-  storage_private->size     = surf_storage_get_size(storage);
-
-  xbt_lib_set(storage_lib, storage_private->name, MSG_STORAGE_LEVEL, storage_private);
-  return xbt_lib_get_elm_or_null(storage_lib, storage_private->name);
-}
-
-/**
- * \brief Destroys a storage (internal call only)
- */
-void __MSG_storage_destroy(msg_storage_priv_t storage) {
-  free(storage);
-}
-
 /** \ingroup msg_storage_management
  *
  * \brief Returns the name of the #msg_storage_t.
@@ -430,8 +366,7 @@ void __MSG_storage_destroy(msg_storage_priv_t storage) {
  */
 const char *MSG_storage_get_name(msg_storage_t storage) {
   xbt_assert((storage != nullptr), "Invalid parameters");
-  msg_storage_priv_t priv = MSG_storage_priv(storage);
-  return priv->name;
+  return storage->name();
 }
 
 /** \ingroup msg_storage_management
@@ -440,7 +375,7 @@ const char *MSG_storage_get_name(msg_storage_t storage) {
  * \return the free space size of the storage element (as a #sg_size_t)
  */
 sg_size_t MSG_storage_get_free_size(msg_storage_t storage){
-  return simgrid::simix::kernelImmediate([storage] { return surf_storage_resource_priv(storage)->getFreeSize(); });
+  return storage->sizeFree();
 }
 
 /** \ingroup msg_storage_management
@@ -449,7 +384,7 @@ sg_size_t MSG_storage_get_free_size(msg_storage_t storage){
  * \return the used space size of the storage element (as a #sg_size_t)
  */
 sg_size_t MSG_storage_get_used_size(msg_storage_t storage){
-  return simgrid::simix::kernelImmediate([storage] { return surf_storage_resource_priv(storage)->getUsedSize(); });
+  return storage->sizeUsed();
 }
 
 /** \ingroup msg_storage_management
@@ -460,7 +395,7 @@ sg_size_t MSG_storage_get_used_size(msg_storage_t storage){
 xbt_dict_t MSG_storage_get_properties(msg_storage_t storage)
 {
   xbt_assert((storage != nullptr), "Invalid parameters (storage is nullptr)");
-  return (simcall_storage_get_properties(storage));
+  return storage->properties();
 }
 
 /** \ingroup msg_storage_management
@@ -472,7 +407,7 @@ xbt_dict_t MSG_storage_get_properties(msg_storage_t storage)
  */
 void MSG_storage_set_property_value(msg_storage_t storage, const char* name, char* value)
 {
-  xbt_dict_set(MSG_storage_get_properties(storage), name, value, nullptr);
+  storage->setProperty(name, value);
 }
 
 /** \ingroup m_storage_management
@@ -484,7 +419,7 @@ void MSG_storage_set_property_value(msg_storage_t storage, const char* name, cha
  */
 const char *MSG_storage_get_property_value(msg_storage_t storage, const char *name)
 {
-  return static_cast<char*>(xbt_dict_get_or_null(MSG_storage_get_properties(storage), name));
+  return storage->property(name);
 }
 
 /** \ingroup msg_storage_management
@@ -494,23 +429,16 @@ const char *MSG_storage_get_property_value(msg_storage_t storage, const char *na
  */
 msg_storage_t MSG_storage_get_by_name(const char *name)
 {
-  return static_cast<msg_storage_t>(xbt_lib_get_elm_or_null(storage_lib,name));
+  return simgrid::s4u::Storage::byName(name);
 }
 
 /** \ingroup msg_storage_management
  * \brief Returns a dynar containing all the storage elements declared at a given point of time
  */
 xbt_dynar_t MSG_storages_as_dynar() {
-  xbt_lib_cursor_t cursor;
-  char *key;
-  void **data;
   xbt_dynar_t res = xbt_dynar_new(sizeof(msg_storage_t),nullptr);
-
-  xbt_lib_foreach(storage_lib, cursor, key, data) {
-    if(xbt_lib_get_level(xbt_lib_get_elm_or_null(storage_lib, key), MSG_STORAGE_LEVEL) != nullptr) {
-      xbt_dictelm_t elm = xbt_dict_cursor_get_elm(cursor);
-      xbt_dynar_push(res, &elm);
-    }
+  for (auto s : *simgrid::s4u::allStorages()) {
+    xbt_dynar_push(res, &(s.second));
   }
   return res;
 }
@@ -522,8 +450,7 @@ xbt_dynar_t MSG_storages_as_dynar() {
  */
 msg_error_t MSG_storage_set_data(msg_storage_t storage, void *data)
 {
-  msg_storage_priv_t priv = MSG_storage_priv(storage);
-  priv->data = data;
+  storage->setUserdata(data);
   return MSG_OK;
 }
 
@@ -536,8 +463,7 @@ msg_error_t MSG_storage_set_data(msg_storage_t storage, void *data)
 void *MSG_storage_get_data(msg_storage_t storage)
 {
   xbt_assert((storage != nullptr), "Invalid parameters");
-  msg_storage_priv_t priv = MSG_storage_priv(storage);
-  return priv->data;
+  return storage->userdata();
 }
 
 /** \ingroup msg_storage_management
@@ -548,8 +474,7 @@ void *MSG_storage_get_data(msg_storage_t storage)
  */
 xbt_dict_t MSG_storage_get_content(msg_storage_t storage)
 {
-  std::map<std::string, sg_size_t*>* content =
-      simgrid::simix::kernelImmediate([storage] { return surf_storage_resource_priv(storage)->getContent(); });
+  std::map<std::string, sg_size_t*>* content = storage->content();
   xbt_dict_t content_dict = xbt_dict_new_homogeneous(nullptr);
 
   for (auto entry : *content) {
@@ -566,8 +491,7 @@ xbt_dict_t MSG_storage_get_content(msg_storage_t storage)
  */
 sg_size_t MSG_storage_get_size(msg_storage_t storage)
 {
-  msg_storage_priv_t priv = MSG_storage_priv(storage);
-  return priv->size;
+  return storage->size();
 }
 
 /** \ingroup msg_storage_management
@@ -578,8 +502,7 @@ sg_size_t MSG_storage_get_size(msg_storage_t storage)
  */
 const char *MSG_storage_get_host(msg_storage_t storage) {
   xbt_assert((storage != nullptr), "Invalid parameters");
-  msg_storage_priv_t priv = MSG_storage_priv(storage);
-  return priv->hostname;
+  return storage->host();
 }
 
 SG_END_DECL()
diff --git a/src/msg/msg_private.h b/src/msg/msg_private.h
index d83517bd3f..2a45111d2d 100644
--- a/src/msg/msg_private.h
+++ b/src/msg/msg_private.h
@@ -30,9 +30,6 @@ public:
 typedef struct simdata_task {
   ~simdata_task()
   {
-    if (this->compute)
-      this->compute->unref();
-
     /* parallel tasks only */
     xbt_free(this->host_list);
   }
@@ -42,8 +39,8 @@ typedef struct simdata_task {
     this->isused = false;
   }
 
-  simgrid::kernel::activity::ExecImpl* compute = nullptr; /* SIMIX modeling of computation */
-  simgrid::kernel::activity::CommImpl* comm = nullptr; /* SIMIX modeling of communication */
+  simgrid::kernel::activity::ExecImplPtr compute = nullptr; /* SIMIX modeling of computation */
+  simgrid::kernel::activity::CommImplPtr comm    = nullptr; /* SIMIX modeling of communication */
   double bytes_amount = 0.0; /* Data size */
   double flops_amount = 0.0; /* Computation size */
   msg_process_t sender = nullptr;
@@ -104,8 +101,6 @@ XBT_PUBLIC_DATA(MSG_Global_t) msg_global;
 /*************************************************************/
 
 XBT_PRIVATE msg_host_t __MSG_host_create(sg_host_t host);
-XBT_PRIVATE msg_storage_t __MSG_storage_create(smx_storage_t storage);
-XBT_PRIVATE void __MSG_storage_destroy(msg_storage_priv_t host);
 XBT_PRIVATE void __MSG_file_destroy(msg_file_t file);
 
 XBT_PRIVATE void MSG_process_cleanup_from_SIMIX(smx_actor_t smx_proc);
@@ -114,8 +109,6 @@ XBT_PRIVATE smx_actor_t MSG_process_create_from_SIMIX(const char* name, std::fun
                                                       smx_actor_t parent_process);
 XBT_PRIVATE void MSG_comm_copy_data_from_SIMIX(smx_activity_t comm, void* buff, size_t buff_size);
 
-XBT_PRIVATE void MSG_post_create_environment();
-
 XBT_PRIVATE void MSG_host_add_task(msg_host_t host, msg_task_t task);
 XBT_PRIVATE void MSG_host_del_task(msg_host_t host, msg_task_t task);
 
diff --git a/src/msg/msg_task.cpp b/src/msg/msg_task.cpp
index 98752c1c29..42a69af2f3 100644
--- a/src/msg/msg_task.cpp
+++ b/src/msg/msg_task.cpp
@@ -265,7 +265,7 @@ void MSG_task_set_bytes_amount(msg_task_t task, double data_size)
  */
 double MSG_task_get_remaining_communication(msg_task_t task)
 {
-  XBT_DEBUG("calling simcall_communication_get_remains(%p)", task->simdata->comm);
+  XBT_DEBUG("calling simcall_communication_get_remains(%p)", task->simdata->comm.get());
   return task->simdata->comm->remains();
 }
 
diff --git a/src/s4u/s4u_comm.cpp b/src/s4u/s4u_comm.cpp
index 547e39c424..7729d3e6f9 100644
--- a/src/s4u/s4u_comm.cpp
+++ b/src/s4u/s4u_comm.cpp
@@ -23,8 +23,6 @@ Comm::~Comm()
       XBT_INFO("pimpl_ is null");
     xbt_backtrace_display_current();
   }
-  if (pimpl_)
-    pimpl_->unref();
 }
 
 s4u::CommPtr Comm::send_init(s4u::MailboxPtr chan)
@@ -90,6 +88,7 @@ void Comm::start() {
         matchFunction_, cleanFunction_, copyDataFunction_,
         userData_, detached_);
   } else if (dstBuff_ != nullptr) { // Receiver side
+    xbt_assert(not detached_, "Receive cannot be detached");
     pimpl_ = simcall_comm_irecv(receiver_, mailbox_->getImpl(), dstBuff_, &dstBuffSize_,
         matchFunction_, copyDataFunction_,
         userData_, rate_);
@@ -117,8 +116,6 @@ void Comm::wait() {
     }
   }
   state_ = finished;
-  if (pimpl_)
-    pimpl_->unref();
 }
 
 void Comm::wait(double timeout) {
@@ -127,7 +124,6 @@ void Comm::wait(double timeout) {
   if (state_ == started) {
     simcall_comm_wait(pimpl_, timeout);
     state_ = finished;
-    pimpl_->unref();
     return;
   }
 
@@ -143,8 +139,6 @@ void Comm::wait(double timeout) {
         userData_, timeout, rate_);
   }
   state_ = finished;
-  if (pimpl_)
-    pimpl_->unref();
 }
 
 void Comm::send_detached(MailboxPtr dest, void* data, int simulatedSize)
@@ -177,23 +171,23 @@ s4u::CommPtr Comm::recv_async(MailboxPtr dest, void** data)
 
 void Comm::cancel()
 {
-  simgrid::kernel::activity::CommImpl* commPimpl = static_cast<simgrid::kernel::activity::CommImpl*>(pimpl_);
+  simgrid::kernel::activity::CommImplPtr commPimpl =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(pimpl_);
   commPimpl->cancel();
 }
 
 bool Comm::test() {
   xbt_assert(state_ == inited || state_ == started || state_ == finished);
-  
-  if (state_ == finished) 
+
+  if (state_ == finished)
     xbt_die("Don't call test on a finished comm.");
-  
+
   if (state_ == inited) {
     this->start();
   }
-  
+
   if(simcall_comm_test(pimpl_)){
     state_ = finished;
-    pimpl_->unref();
     return true;
   }
   return false;
diff --git a/src/s4u/s4u_conditionVariable.cpp b/src/s4u/s4u_conditionVariable.cpp
index 98b4e6b37f..709276c58f 100644
--- a/src/s4u/s4u_conditionVariable.cpp
+++ b/src/s4u/s4u_conditionVariable.cpp
@@ -74,14 +74,14 @@ std::cv_status ConditionVariable::wait_until(std::unique_lock<Mutex>& lock, doub
     timeout = timeout_time - now;
   return this->wait_for(lock, timeout);
 }
-  
+
 /**
  * Notify functions
  */
 void ConditionVariable::notify_one() {
    simcall_cond_signal(cond_);
 }
- 
+
 void ConditionVariable::notify_all() {
   simcall_cond_broadcast(cond_);
 }
diff --git a/src/s4u/s4u_engine.cpp b/src/s4u/s4u_engine.cpp
index d895beb099..9ab100720c 100644
--- a/src/s4u/s4u_engine.cpp
+++ b/src/s4u/s4u_engine.cpp
@@ -53,7 +53,6 @@ Engine *Engine::instance() {
 void Engine::shutdown() {
   delete s4u::Engine::instance_;
   s4u::Engine::instance_ = nullptr;
-  delete s4u::Storage::storages_;
 }
 
 double Engine::getClock()
diff --git a/src/s4u/s4u_file.cpp b/src/s4u/s4u_file.cpp
index fafc21dd17..0f26d5903c 100644
--- a/src/s4u/s4u_file.cpp
+++ b/src/s4u/s4u_file.cpp
@@ -12,20 +12,26 @@
 #include "simgrid/s4u/File.hpp"
 #include "simgrid/s4u/Host.hpp"
 #include "simgrid/s4u/Mailbox.hpp"
+#include "simgrid/s4u/Storage.hpp"
 
 XBT_LOG_NEW_DEFAULT_CATEGORY(s4u_file,"S4U files");
 
-
 namespace simgrid {
 namespace s4u {
 
 File::File(const char* fullpath, void* userdata) : path_(fullpath), userdata_(userdata)
 {
   // this cannot fail because we get a xbt_die if the mountpoint does not exist
-  pimpl_ = simcall_file_open(fullpath, Host::current());
+  pimpl_           = simcall_file_open(fullpath, Host::current());
+  xbt_dynar_t info = simcall_file_get_info(pimpl_);
+  storage_type     = xbt_dynar_pop_as(info, char*);
+  storageId        = xbt_dynar_pop_as(info, char*);
+  mount_point      = xbt_dynar_pop_as(info, char*);
+  xbt_dynar_free(&info);
 }
 
 File::~File() {
+  //  Host::current()->extension<simgrid::MsgHostExt>()->file_descriptor_table->push_back(desc_id_);
   simcall_file_close(pimpl_, Host::current());
 }
 
diff --git a/src/s4u/s4u_host.cpp b/src/s4u/s4u_host.cpp
index c36d9ec715..8ee524a1b7 100644
--- a/src/s4u/s4u_host.cpp
+++ b/src/s4u/s4u_host.cpp
@@ -182,7 +182,7 @@ boost::unordered_map<std::string, Storage*> const& Host::mountedStorages() {
     char *mountname;
     char *storagename;
     xbt_dict_foreach(dict, cursor, mountname, storagename) {
-      mounts->insert({mountname, &Storage::byName(storagename)});
+      mounts->insert({mountname, Storage::byName(storagename)});
     }
     xbt_dict_free(&dict);
   }
@@ -266,8 +266,8 @@ xbt_dict_t Host::mountedStoragesAsDict()
  */
 void Host::attachedStorages(std::vector<const char*>* storages)
 {
-  simgrid::simix::kernelImmediate([this, storages] { 
-     this->pimpl_->getAttachedStorageList(storages); 
+  simgrid::simix::kernelImmediate([this, storages] {
+     this->pimpl_->getAttachedStorageList(storages);
   });
 }
 
diff --git a/src/s4u/s4u_netzone.cpp b/src/s4u/s4u_netzone.cpp
index f539136108..1eec476116 100644
--- a/src/s4u/s4u_netzone.cpp
+++ b/src/s4u/s4u_netzone.cpp
@@ -20,6 +20,8 @@ simgrid::xbt::signal<void(bool symmetrical, kernel::routing::NetPoint* src, kern
                           kernel::routing::NetPoint* gw_src, kernel::routing::NetPoint* gw_dst,
                           std::vector<surf::LinkImpl*>* link_list)>
     NetZone::onRouteCreation;
+simgrid::xbt::signal<void(NetZone&)> NetZone::onCreation;
+simgrid::xbt::signal<void(NetZone&)> NetZone::onSeal;
 
 NetZone::NetZone(NetZone* father, const char* name) : father_(father), name_(xbt_strdup(name))
 {
diff --git a/src/s4u/s4u_storage.cpp b/src/s4u/s4u_storage.cpp
index d010801b6c..23c7e603f3 100644
--- a/src/s4u/s4u_storage.cpp
+++ b/src/s4u/s4u_storage.cpp
@@ -6,73 +6,55 @@
 #include "../surf/StorageImpl.hpp"
 #include "simgrid/s4u/Storage.hpp"
 #include "simgrid/simix.hpp"
-#include "xbt/lib.h"
 #include <unordered_map>
 
-extern xbt_lib_t storage_lib;
-
 namespace simgrid {
 namespace s4u {
-
-std::unordered_map<std::string, Storage*>* Storage::storages_ = new std::unordered_map<std::string, Storage*>();
-
-Storage::Storage(std::string name, smx_storage_t inferior) :
-    name_(name), pimpl_(inferior)
+std::map<std::string, Storage*>* allStorages()
 {
-  hostname_ = surf_storage_get_host(pimpl_);
-  size_     = surf_storage_get_size(pimpl_);
-  storages_->insert({name, this});
-}
+  std::unordered_map<std::string, surf::StorageImpl*>* map = surf::StorageImpl::storagesMap();
+  std::map<std::string, Storage*>* res                     = new std::map<std::string, Storage*>;
+  for (auto s : *map)
+    res->insert({s.first, &(s.second->piface_)}); // Convert each entry into its interface
 
-Storage::~Storage() = default;
-
-smx_storage_t Storage::inferior()
-{
-  return pimpl_;
+  return res;
 }
 
-Storage& Storage::byName(const char* name)
+Storage* Storage::byName(const char* name)
 {
-  s4u::Storage* res = nullptr;
-  try {
-    res = storages_->at(name);
-  } catch (std::out_of_range& e) {
-    smx_storage_t inferior = xbt_lib_get_elm_or_null(storage_lib,name);
-    if (inferior == nullptr)
-      xbt_die("Storage %s does not exist. Please only use the storages that are defined in your platform.", name);
-
-    res = new Storage(name,inferior);
-  }
-  return *res;
+  surf::StorageImpl* res = surf::StorageImpl::byName(name);
+  if (res == nullptr)
+    return nullptr;
+  return &res->piface_;
 }
 
 const char* Storage::name()
 {
-  return name_.c_str();
+  return pimpl_->cname();
 }
 
 const char* Storage::host()
 {
-  return hostname_.c_str();
+  return pimpl_->attach_;
 }
 
 sg_size_t Storage::sizeFree()
 {
-  return simgrid::simix::kernelImmediate([this] { return surf_storage_resource_priv(pimpl_)->getFreeSize(); });
+  return simgrid::simix::kernelImmediate([this] { return pimpl_->getFreeSize(); });
 }
 
 sg_size_t Storage::sizeUsed()
 {
-  return simgrid::simix::kernelImmediate([this] { return surf_storage_resource_priv(pimpl_)->getUsedSize(); });
+  return simgrid::simix::kernelImmediate([this] { return pimpl_->getUsedSize(); });
 }
 
 sg_size_t Storage::size() {
-  return size_;
+  return pimpl_->size_;
 }
 
 xbt_dict_t Storage::properties()
 {
-  return simcall_storage_get_properties(pimpl_);
+  return simgrid::simix::kernelImmediate([this] { return pimpl_->getProperties(); });
 }
 
 const char* Storage::property(const char* key)
@@ -87,13 +69,14 @@ void Storage::setProperty(const char* key, char* value)
 
 std::map<std::string, sg_size_t*>* Storage::content()
 {
-  return simgrid::simix::kernelImmediate([this] { return surf_storage_resource_priv(this->pimpl_)->getContent(); });
+  return simgrid::simix::kernelImmediate([this] { return pimpl_->getContent(); });
 }
 
-std::unordered_map<std::string, Storage*>* Storage::allStorages()
-{
-  return storages_;
-}
+/*************
+ * Callbacks *
+ *************/
+simgrid::xbt::signal<void(s4u::Storage&)> Storage::onCreation;
+simgrid::xbt::signal<void(s4u::Storage&)> Storage::onDestruction;
 
 } /* namespace s4u */
 } /* namespace simgrid */
diff --git a/src/simdag/dax_dtd.c b/src/simdag/dax_dtd.c
index cc41601d23..e5e368aaed 100644
--- a/src/simdag/dax_dtd.c
+++ b/src/simdag/dax_dtd.c
@@ -50,7 +50,7 @@
 #if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L
 
 /* C99 says to define __STDC_LIMIT_MACROS before including stdint.h,
- * if you want the limit (max/min) macros for int types. 
+ * if you want the limit (max/min) macros for int types.
  */
 #ifndef __STDC_LIMIT_MACROS
 #define __STDC_LIMIT_MACROS 1
@@ -67,7 +67,7 @@ typedef uint32_t flex_uint32_t;
 typedef signed char flex_int8_t;
 typedef short int flex_int16_t;
 typedef int flex_int32_t;
-typedef unsigned char flex_uint8_t; 
+typedef unsigned char flex_uint8_t;
 typedef unsigned short int flex_uint16_t;
 typedef unsigned int flex_uint32_t;
 
@@ -193,7 +193,7 @@ extern FILE *dax_in, *dax_out;
 
     /* Note: We specifically omit the test for yy_rule_can_match_eol because it requires
      *       access to the local variable yy_act. Since yyless() is a macro, it would break
-     *       existing scanners that call yyless() from OUTSIDE dax_lex. 
+     *       existing scanners that call yyless() from OUTSIDE dax_lex.
      *       One obvious solution it to make yy_act a global. I tried that, and saw
      *       a 5% performance hit in a non-dax_lineno scanner, because yy_act is
      *       normally declared as a register variable-- so it is not worth it.
@@ -212,7 +212,7 @@ extern FILE *dax_in, *dax_out;
                     if ( *p == '\n' )\
                         --dax_lineno;\
             }while(0)
-    
+
 /* Return all but the first "n" matched characters back to the input stream. */
 #define yyless(n) \
   do \
@@ -269,7 +269,7 @@ struct yy_buffer_state
 
     int yy_bs_lineno; /**< The line count. */
     int yy_bs_column; /**< The column count. */
-    
+
   /* Whether to try to fill the input buffer when we reach the
    * end of it.
    */
@@ -1237,13 +1237,13 @@ static yyconst flex_int16_t yy_chk[2195] =
 /* Table of booleans, true if rule could match eol. */
 static yyconst flex_int32_t yy_rule_can_match_eol[141] =
     {   0,
-0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 
-    0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 
-    1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 
-    0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 
+0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0,
+    0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1,
+    1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0,
+    0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1,
     0,     };
 
 static yy_state_type yy_last_accepting_state;
@@ -1266,13 +1266,13 @@ char *dax_text;
  * FleXML is Copyright (C) 1999-2005 Kristoffer Rose.  All rights reserved.
  * FleXML is Copyright (C) 2003-2013 Martin Quinson.  All rights reserved.
  * (1.9.6).
- * 
+ *
  * There are two, intertwined parts to this program, part A and part B.
  *
  * Part A
  * ------
- * 
- * Some parts, here collectively called "Part A", are found in the 
+ *
+ * Some parts, here collectively called "Part A", are found in the
  * FleXML package.  They are Copyright (C) 1999-2005 Kristoffer Rose
  * and Copyright (C) 2003-2013 Martin Quinson. All rights reserved.
  *
@@ -1290,20 +1290,20 @@ char *dax_text;
  * Notice that these are explicit rights granted to you for files
  * generated by the FleXML system.  For your rights in connection with
  * the FleXML system itself please consult the GNU General Public License.
- * 
+ *
  * Part B
  * ------
- * 
- * The other parts, here collectively called "Part B", and which came 
- * from the DTD used by FleXML to generate this program, can be 
+ *
+ * The other parts, here collectively called "Part B", and which came
+ * from the DTD used by FleXML to generate this program, can be
  * distributed (or not, as the case may be) under the terms of whoever
- * wrote them, provided these terms respect and obey the two conditions 
+ * wrote them, provided these terms respect and obey the two conditions
  * above under the heading "Part A".
  *
  * The author of and contributors to FleXML specifically disclaim
- * any copyright interest in "Part B", unless "Part B" was written 
+ * any copyright interest in "Part B", unless "Part B" was written
  * by the author of or contributors to FleXML.
- * 
+ *
  */
 
 /* Version strings. */
@@ -1316,7 +1316,7 @@ const char dax__flexml_version[] = "1.9.6";
 #include <assert.h>
 #include <stdarg.h>
 #include <ctype.h>
-     
+
 #if defined(_WIN32) || defined(__WIN32__) || defined(WIN32) || defined(__TOS_WIN__)
 # ifndef __STRICT_ANSI__
 #  include <io.h>
@@ -1325,7 +1325,7 @@ const char dax__flexml_version[] = "1.9.6";
 #else
 # include <unistd.h>
 #endif
-     
+
 #ifndef FLEXML_INDEXSTACKSIZE
 #define FLEXML_INDEXSTACKSIZE 1000
 #endif
@@ -1525,7 +1525,7 @@ static void pushbuffer(int p)
 {
     ck_ilimit();
     indexstack[inext++] = p;
-    indexstack[inext++] = bnext;    
+    indexstack[inext++] = bnext;
 }
 
 static int popbuffer(void)
@@ -1663,7 +1663,7 @@ extern int dax_wrap (void );
 #endif
 
 #ifndef YY_NO_UNPUT
-    
+
 #endif
 
 #ifndef yytext_ptr
@@ -1687,11 +1687,11 @@ static int input (void );
         static int yy_start_stack_ptr = 0;
         static int yy_start_stack_depth = 0;
         static int *yy_start_stack = NULL;
-    
+
     static void yy_push_state (int _new_state );
-    
+
     static void yy_pop_state (void );
-    
+
 /* Amount of stuff to slurp up with each read. */
 #ifndef YY_READ_BUF_SIZE
 #ifdef __ia64__
@@ -1799,7 +1799,7 @@ YY_DECL
   yy_state_type yy_current_state;
   char *yy_cp, *yy_bp;
   int yy_act;
-    
+
   if ( !(yy_init) )
     {
     (yy_init) = 1;
@@ -1939,7 +1939,7 @@ yy_find_action:
       yy_size_t yyl;
       for ( yyl = 0; yyl < dax_leng; ++yyl )
         if ( dax_text[yyl] == '\n' )
-             
+
     dax_lineno++;
 ;
       }
@@ -2004,7 +2004,7 @@ SKIP;
 case 11:
 /* rule 11 can match eol */
 YY_RULE_SETUP
-SET(ROOT_dax__adag); 
+SET(ROOT_dax__adag);
   YY_BREAK
 case 12:
 /* rule 12 can match eol */
@@ -3099,7 +3099,7 @@ static int yy_get_next_buffer (void)
 {
   yy_state_type yy_current_state;
   char *yy_cp;
-    
+
   yy_current_state = (yy_start);
 
   for ( yy_cp = (yytext_ptr) + YY_MORE_ADJ; yy_cp < (yy_c_buf_p); ++yy_cp )
@@ -3163,7 +3163,7 @@ static int yy_get_next_buffer (void)
 
 {
   int c;
-    
+
   *(yy_c_buf_p) = (yy_hold_char);
 
   if ( *(yy_c_buf_p) == YY_END_OF_BUFFER_CHAR )
@@ -3225,7 +3225,7 @@ static int yy_get_next_buffer (void)
   (yy_hold_char) = *++(yy_c_buf_p);
 
   if ( c == '\n' )
-       
+
     dax_lineno++;
 ;
 
@@ -3235,12 +3235,12 @@ static int yy_get_next_buffer (void)
 
 /** Immediately switch to a different input stream.
  * @param input_file A readable stream.
- * 
+ *
  * @note This function does not reset the start condition to @c INITIAL .
  */
     void dax_restart  (FILE * input_file )
 {
-    
+
   if ( ! YY_CURRENT_BUFFER ){
         dax_ensure_buffer_stack ();
     YY_CURRENT_BUFFER_LVALUE =
@@ -3253,11 +3253,11 @@ static int yy_get_next_buffer (void)
 
 /** Switch to a different input buffer.
  * @param new_buffer The new input buffer.
- * 
+ *
  */
     void dax__switch_to_buffer  (YY_BUFFER_STATE  new_buffer )
 {
-    
+
   /* TODO. We should be able to replace this entire function body
    * with
    *    dax_pop_buffer_state();
@@ -3297,13 +3297,13 @@ static void dax__load_buffer_state  (void)
 /** Allocate and initialize an input buffer state.
  * @param file A readable stream.
  * @param size The character buffer size in bytes. When in doubt, use @c YY_BUF_SIZE.
- * 
+ *
  * @return the allocated buffer state.
  */
     YY_BUFFER_STATE dax__create_buffer  (FILE * file, int  size )
 {
   YY_BUFFER_STATE b;
-    
+
   b = (YY_BUFFER_STATE) dax_alloc(sizeof( struct yy_buffer_state )  );
   if ( ! b )
     YY_FATAL_ERROR( "out of dynamic memory in dax__create_buffer()" );
@@ -3326,11 +3326,11 @@ static void dax__load_buffer_state  (void)
 
 /** Destroy the buffer.
  * @param b a buffer created with dax__create_buffer()
- * 
+ *
  */
     void dax__delete_buffer (YY_BUFFER_STATE  b )
 {
-    
+
   if ( ! b )
     return;
 
@@ -3351,7 +3351,7 @@ static void dax__load_buffer_state  (void)
 
 {
   int oerrno = errno;
-    
+
   dax__flush_buffer(b );
 
   b->yy_input_file = file;
@@ -3367,13 +3367,13 @@ static void dax__load_buffer_state  (void)
     }
 
         b->yy_is_interactive = file ? (isatty( fileno(file) ) > 0) : 0;
-    
+
   errno = oerrno;
 }
 
 /** Discard all buffered characters. On the next scan, YY_INPUT will be called.
  * @param b the buffer state to be flushed, usually @c YY_CURRENT_BUFFER.
- * 
+ *
  */
     void dax__flush_buffer (YY_BUFFER_STATE  b )
 {
@@ -3402,7 +3402,7 @@ static void dax__load_buffer_state  (void)
  *  the current state. This function will allocate the stack
  *  if necessary.
  *  @param new_buffer The new state.
- *  
+ *
  */
 void dax_push_buffer_state (YY_BUFFER_STATE new_buffer )
 {
@@ -3432,7 +3432,7 @@ void dax_push_buffer_state (YY_BUFFER_STATE new_buffer )
 
 /** Removes and deletes the top of the stack, if present.
  *  The next element becomes the new top.
- *  
+ *
  */
 void dax_pop_buffer_state (void)
 {
@@ -3456,7 +3456,7 @@ void dax_pop_buffer_state (void)
 static void dax_ensure_buffer_stack (void)
 {
   yy_size_t num_to_alloc;
-    
+
   if (!(yy_buffer_stack)) {
 
     /* First allocation is just for 2 elements, since we don't know if this
@@ -3469,9 +3469,9 @@ static void dax_ensure_buffer_stack (void)
                 );
     if ( ! (yy_buffer_stack) )
       YY_FATAL_ERROR( "out of dynamic memory in dax_ensure_buffer_stack()" );
-                  
+
     memset((yy_buffer_stack), 0, num_to_alloc * sizeof(struct yy_buffer_state*));
-        
+
     (yy_buffer_stack_max) = num_to_alloc;
     (yy_buffer_stack_top) = 0;
     return;
@@ -3499,13 +3499,13 @@ static void dax_ensure_buffer_stack (void)
 /** Setup the input buffer state to scan directly from a user-specified character buffer.
  * @param base the character buffer
  * @param size the size in bytes of the character buffer
- * 
- * @return the newly allocated buffer state object. 
+ *
+ * @return the newly allocated buffer state object.
  */
 YY_BUFFER_STATE dax__scan_buffer  (char * base, yy_size_t  size )
 {
   YY_BUFFER_STATE b;
-    
+
   if ( size < 2 ||
        base[size-2] != YY_END_OF_BUFFER_CHAR ||
        base[size-1] != YY_END_OF_BUFFER_CHAR )
@@ -3534,14 +3534,14 @@ YY_BUFFER_STATE dax__scan_buffer  (char * base, yy_size_t  size )
 /** Setup the input buffer state to scan a string. The next call to dax_lex() will
  * scan from a @e copy of @a str.
  * @param yystr a NUL-terminated string to scan
- * 
+ *
  * @return the newly allocated buffer state object.
  * @note If you want to scan bytes that may contain NUL values, then use
  *       dax__scan_bytes() instead.
  */
 YY_BUFFER_STATE dax__scan_string (yyconst char * yystr )
 {
-    
+
   return dax__scan_bytes(yystr,strlen(yystr) );
 }
 
@@ -3549,7 +3549,7 @@ YY_BUFFER_STATE dax__scan_string (yyconst char * yystr )
  * scan from a @e copy of @a bytes.
  * @param yybytes the byte buffer to scan
  * @param _yybytes_len the number of bytes in the buffer pointed to by @a bytes.
- * 
+ *
  * @return the newly allocated buffer state object.
  */
 YY_BUFFER_STATE dax__scan_bytes  (yyconst char * yybytes, yy_size_t  _yybytes_len )
@@ -3558,7 +3558,7 @@ YY_BUFFER_STATE dax__scan_bytes  (yyconst char * yybytes, yy_size_t  _yybytes_le
   char *buf;
   yy_size_t n;
   yy_size_t i;
-    
+
   /* Get memory for full buffer, including space for trailing EOB's. */
   n = _yybytes_len + 2;
   buf = (char *) dax_alloc(n  );
@@ -3644,16 +3644,16 @@ static void yy_fatal_error (yyconst char* msg )
 /* Accessor  methods (get/set functions) to struct members. */
 
 /** Get the current line number.
- * 
+ *
  */
 int dax_get_lineno  (void)
 {
-        
+
     return dax_lineno;
 }
 
 /** Get the input stream.
- * 
+ *
  */
 FILE *dax_get_in  (void)
 {
@@ -3661,7 +3661,7 @@ FILE *dax_get_in  (void)
 }
 
 /** Get the output stream.
- * 
+ *
  */
 FILE *dax_get_out  (void)
 {
@@ -3669,7 +3669,7 @@ FILE *dax_get_out  (void)
 }
 
 /** Get the length of the current token.
- * 
+ *
  */
 yy_size_t dax_get_leng  (void)
 {
@@ -3677,7 +3677,7 @@ yy_size_t dax_get_leng  (void)
 }
 
 /** Get the current token.
- * 
+ *
  */
 
 char *dax_get_text  (void)
@@ -3687,18 +3687,18 @@ char *dax_get_text  (void)
 
 /** Set the current line number.
  * @param _line_number line number
- * 
+ *
  */
 void dax_set_lineno (int  _line_number )
 {
-    
+
     dax_lineno = _line_number;
 }
 
 /** Set the input stream. This does not discard the current
  * input buffer.
  * @param _in_str A readable stream.
- * 
+ *
  * @see dax__switch_to_buffer
  */
 void dax_set_in (FILE *  _in_str )
@@ -3729,7 +3729,7 @@ static int yy_init_globals (void)
 
     /* We do not touch dax_lineno unless the option is enabled. */
     dax_lineno =  1;
-    
+
     (yy_buffer_stack) = 0;
     (yy_buffer_stack_top) = 0;
     (yy_buffer_stack_max) = 0;
@@ -3759,7 +3759,7 @@ static int yy_init_globals (void)
 /* dax_lex_destroy is for both reentrant and non-reentrant scanners. */
 int dax_lex_destroy  (void)
 {
-    
+
     /* Pop the buffer stack, destroying each element. */
   while(YY_CURRENT_BUFFER){
     dax__delete_buffer(YY_CURRENT_BUFFER  );
@@ -3789,7 +3789,7 @@ int dax_lex_destroy  (void)
 #ifndef yytext_ptr
 static void yy_flex_strncpy (char* s1, yyconst char * s2, int n )
 {
-    
+
   int i;
   for ( i = 0; i < n; ++i )
     s1[i] = s2[i];
@@ -3814,7 +3814,7 @@ void *dax_alloc (yy_size_t  size )
 
 void *dax_realloc  (void * ptr, yy_size_t  size )
 {
-    
+
   /* The cast to (char *) in the following accommodates both
    * implementations that use char* generic pointers, and those
    * that use void* generic pointers.  It works with the latter
diff --git a/src/simdag/dax_dtd.h b/src/simdag/dax_dtd.h
index 5552c33ca7..b300c56987 100644
--- a/src/simdag/dax_dtd.h
+++ b/src/simdag/dax_dtd.h
@@ -4,13 +4,13 @@
  * FleXML is Copyright (C) 1999-2005 Kristoffer Rose.  All rights reserved.
  * FleXML is Copyright (C) 2003-2013 Martin Quinson.  All rights reserved.
  * (1.9.6).
- * 
+ *
  * There are two, intertwined parts to this program, part A and part B.
  *
  * Part A
  * ------
- * 
- * Some parts, here collectively called "Part A", are found in the 
+ *
+ * Some parts, here collectively called "Part A", are found in the
  * FleXML package.  They are Copyright (C) 1999-2005 Kristoffer Rose
  * and Copyright (C) 2003-2013 Martin Quinson. All rights reserved.
  *
@@ -28,20 +28,20 @@
  * Notice that these are explicit rights granted to you for files
  * generated by the FleXML system.  For your rights in connection with
  * the FleXML system itself please consult the GNU General Public License.
- * 
+ *
  * Part B
  * ------
- * 
- * The other parts, here collectively called "Part B", and which came 
- * from the DTD used by FleXML to generate this program, can be 
+ *
+ * The other parts, here collectively called "Part B", and which came
+ * from the DTD used by FleXML to generate this program, can be
  * distributed (or not, as the case may be) under the terms of whoever
- * wrote them, provided these terms respect and obey the two conditions 
+ * wrote them, provided these terms respect and obey the two conditions
  * above under the heading "Part A".
  *
  * The author of and contributors to FleXML specifically disclaim
- * any copyright interest in "Part B", unless "Part B" was written 
+ * any copyright interest in "Part B", unless "Part B" was written
  * by the author of or contributors to FleXML.
- * 
+ *
  */
 
 #ifndef _FLEXML_dax_H
diff --git a/src/simdag/sd_daxloader.cpp b/src/simdag/sd_daxloader.cpp
index eae376c635..e642431eea 100644
--- a/src/simdag/sd_daxloader.cpp
+++ b/src/simdag/sd_daxloader.cpp
@@ -154,7 +154,7 @@ static void dax_task_free(void *task)
 }
 
 /** @brief loads a DAX file describing a DAG
- * 
+ *
  * See https://confluence.pegasus.isi.edu/display/pegasus/WorkflowGenerator for more details.
  */
 xbt_dynar_t SD_daxload(const char *filename)
diff --git a/src/simdag/sd_dotloader.cpp b/src/simdag/sd_dotloader.cpp
index 2fbb3d3a7d..619fcb13f4 100644
--- a/src/simdag/sd_dotloader.cpp
+++ b/src/simdag/sd_dotloader.cpp
@@ -23,7 +23,7 @@ static void dot_task_p_free(void *task) {
 }
 
 /** @brief loads a DOT file describing a DAG
- * 
+ *
  * See http://www.graphviz.org/doc/info/lang.html  for more details.
  * The size attribute of a node describes:
  *   - for a compute task: the amount of flops to execute
diff --git a/src/simdag/sd_task.cpp b/src/simdag/sd_task.cpp
index e57a89d8d2..8ea5cc145a 100644
--- a/src/simdag/sd_task.cpp
+++ b/src/simdag/sd_task.cpp
@@ -133,7 +133,7 @@ SD_task_t SD_task_create_comp_par_amdahl(const char *name, void *data, double fl
 
 /** @brief create a complex data redistribution task that can then be  auto-scheduled
  *
- * Auto-scheduling mean that the task can be used with SD_task_schedulev(). 
+ * Auto-scheduling mean that the task can be used with SD_task_schedulev().
  * This allows to specify the task costs at creation, and decouple them from the scheduling process where you just
  * specify which resource should communicate.
  *
diff --git a/src/simgrid/sg_config.cpp b/src/simgrid/sg_config.cpp
index e617c48091..ff40e0cb75 100644
--- a/src/simgrid/sg_config.cpp
+++ b/src/simgrid/sg_config.cpp
@@ -12,7 +12,6 @@
 #include "xbt/log.h"
 #include "xbt/mallocator.h"
 #include "xbt/str.h"
-#include "xbt/lib.h"
 #include "xbt/sysdep.h"
 #include "surf/surf.h"
 #include "surf/maxmin.h"
@@ -469,11 +468,11 @@ void sg_config_init(int *argc, char **argv)
     xbt_cfg_register_string("smpi/lat-factor",
         "65472:11.6436;15424:3.48845;9376:2.59299;5776:2.18796;3484:1.88101;1426:1.61075;732:1.9503;257:1.95341;0:2.01467", nullptr, "Latency factors for smpi.");
     xbt_cfg_register_alias("smpi/lat-factor","smpi/lat_factor");
-    
+
     xbt_cfg_register_string("smpi/IB-penalty-factors", "0.965;0.925;1.35", nullptr,
         "Correction factor to communications using Infiniband model with contention (default value based on Stampede cluster profiling)");
     xbt_cfg_register_alias("smpi/IB-penalty-factors","smpi/IB_penalty_factors");
-    
+
 #if HAVE_SMPI
     xbt_cfg_register_double("smpi/host-speed", 20000.0, nullptr, "Speed of the host running the simulation (in flop/s). Used to bench the operations.");
     xbt_cfg_register_alias("smpi/host-speed","smpi/running_power");
diff --git a/src/simix/ActorImpl.cpp b/src/simix/ActorImpl.cpp
index 2c59aab6d3..5af5070581 100644
--- a/src/simix/ActorImpl.cpp
+++ b/src/simix/ActorImpl.cpp
@@ -81,7 +81,8 @@ int SIMIX_process_has_pending_comms(smx_actor_t process) {
  */
 void SIMIX_process_cleanup(smx_actor_t process)
 {
-  XBT_DEBUG("Cleanup process %s (%p), waiting synchro %p", process->name.c_str(), process, process->waiting_synchro);
+  XBT_DEBUG("Cleanup process %s (%p), waiting synchro %p", process->name.c_str(), process,
+            process->waiting_synchro.get());
 
   process->finished = true;
   SIMIX_process_on_exit_runall(process);
@@ -93,26 +94,22 @@ void SIMIX_process_cleanup(smx_actor_t process)
   xbt_os_mutex_acquire(simix_global->mutex);
 
   /* cancel non-blocking communications */
-  smx_activity_t synchro = static_cast<smx_activity_t>(process->comms.front());
+  smx_activity_t synchro = process->comms.front();
   while (not process->comms.empty()) {
-    simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+    simgrid::kernel::activity::CommImplPtr comm =
+        boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
 
     /* make sure no one will finish the comm after this process is destroyed,
      * because src_proc or dst_proc would be an invalid pointer */
 
     if (comm->src_proc == process) {
-      XBT_DEBUG("Found an unfinished send comm %p (detached = %d), state %d, src = %p, dst = %p",
-          comm, comm->detached, (int)comm->state, comm->src_proc, comm->dst_proc);
+      XBT_DEBUG("Found an unfinished send comm %p (detached = %d), state %d, src = %p, dst = %p", comm.get(),
+                comm->detached, (int)comm->state, comm->src_proc, comm->dst_proc);
       comm->src_proc = nullptr;
 
-      /* I'm not supposed to destroy a detached comm from the sender side, */
-      if (comm->detached)
-        XBT_DEBUG("Don't destroy it since it's a detached comm and I'm the sender");
-      else
-        comm->unref();
     } else if (comm->dst_proc == process) {
-      XBT_DEBUG("Found an unfinished recv comm %p, state %d, src = %p, dst = %p",
-          comm, (int)comm->state, comm->src_proc, comm->dst_proc);
+      XBT_DEBUG("Found an unfinished recv comm %p, state %d, src = %p, dst = %p", comm.get(), (int)comm->state,
+                comm->src_proc, comm->dst_proc);
       comm->dst_proc = nullptr;
 
       if (comm->detached && comm->src_proc != nullptr) {
@@ -120,10 +117,10 @@ void SIMIX_process_cleanup(smx_actor_t process)
         comm->src_proc->comms.remove(comm);
       }
     } else {
-      xbt_die("Communication synchro %p is in my list but I'm not the sender nor the receiver", synchro);
+      xbt_die("Communication synchro %p is in my list but I'm not the sender nor the receiver", synchro.get());
     }
     process->comms.pop_front();
-    synchro = static_cast<smx_activity_t>(process->comms.front());
+    synchro = process->comms.front();
     comm->cancel();
   }
 
@@ -278,7 +275,7 @@ smx_actor_t SIMIX_process_create(const char* name, std::function<void()> code, v
   if (host->extension<simgrid::simix::Host>() == nullptr)
     host->extension_set<simgrid::simix::Host>(new simgrid::simix::Host());
 
-  /* Add the process to it's host process list */
+  /* Add the process to its host process list */
   xbt_swag_insert(process, host->extension<simgrid::simix::Host>()->process_list);
 
   XBT_DEBUG("Start context '%s'", process->name.c_str());
@@ -287,6 +284,7 @@ smx_actor_t SIMIX_process_create(const char* name, std::function<void()> code, v
   simix_global->process_list[process->pid] = process;
   XBT_DEBUG("Inserting %s(%s) in the to_run list", process->cname(), host->cname());
   xbt_dynar_push_as(simix_global->process_to_run, smx_actor_t, process);
+  intrusive_ptr_add_ref(process);
 
   /* Tracing the process creation */
   TRACE_msg_process_create(process->cname(), process->pid, process->host);
@@ -421,19 +419,20 @@ void SIMIX_process_kill(smx_actor_t process, smx_actor_t issuer) {
   process->exception = nullptr;
 
   /* destroy the blocking synchro if any */
-  if (process->waiting_synchro) {
-
-    simgrid::kernel::activity::ExecImpl* exec =
-        dynamic_cast<simgrid::kernel::activity::ExecImpl*>(process->waiting_synchro);
-    simgrid::kernel::activity::CommImpl* comm =
-        dynamic_cast<simgrid::kernel::activity::CommImpl*>(process->waiting_synchro);
-    simgrid::kernel::activity::SleepImpl* sleep =
-        dynamic_cast<simgrid::kernel::activity::SleepImpl*>(process->waiting_synchro);
-    simgrid::kernel::activity::Raw *raw = dynamic_cast<simgrid::kernel::activity::Raw*>(process->waiting_synchro);
-    simgrid::kernel::activity::Io *io = dynamic_cast<simgrid::kernel::activity::Io*>(process->waiting_synchro);
+  if (process->waiting_synchro != nullptr) {
+
+    simgrid::kernel::activity::ExecImplPtr exec =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::ExecImpl>(process->waiting_synchro);
+    simgrid::kernel::activity::CommImplPtr comm =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::CommImpl>(process->waiting_synchro);
+    simgrid::kernel::activity::SleepImplPtr sleep =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::SleepImpl>(process->waiting_synchro);
+    simgrid::kernel::activity::RawImplPtr raw =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::RawImpl>(process->waiting_synchro);
+    simgrid::kernel::activity::IoImplPtr io =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::IoImpl>(process->waiting_synchro);
 
     if (exec != nullptr) {
-      exec->unref();
 
     } else if (comm != nullptr) {
       process->comms.remove(process->waiting_synchro);
@@ -442,13 +441,11 @@ void SIMIX_process_kill(smx_actor_t process, smx_actor_t issuer) {
       auto i = boost::range::find(process->waiting_synchro->simcalls, &process->simcall);
       if (i != process->waiting_synchro->simcalls.end())
         process->waiting_synchro->simcalls.remove(&process->simcall);
-      comm->unref();
     } else if (sleep != nullptr) {
       SIMIX_process_sleep_destroy(process->waiting_synchro);
 
     } else if (raw != nullptr) {
       SIMIX_synchro_stop_waiting(process, &process->simcall);
-      delete process->waiting_synchro;
 
     } else if (io != nullptr) {
       SIMIX_io_destroy(process->waiting_synchro);
@@ -485,21 +482,21 @@ void SIMIX_process_throw(smx_actor_t process, xbt_errcat_t cat, int value, const
   /* cancel the blocking synchro if any */
   if (process->waiting_synchro) {
 
-    simgrid::kernel::activity::ExecImpl* exec =
-        dynamic_cast<simgrid::kernel::activity::ExecImpl*>(process->waiting_synchro);
+    simgrid::kernel::activity::ExecImplPtr exec =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::ExecImpl>(process->waiting_synchro);
     if (exec != nullptr) {
       SIMIX_execution_cancel(process->waiting_synchro);
     }
 
-    simgrid::kernel::activity::CommImpl* comm =
-        dynamic_cast<simgrid::kernel::activity::CommImpl*>(process->waiting_synchro);
+    simgrid::kernel::activity::CommImplPtr comm =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::CommImpl>(process->waiting_synchro);
     if (comm != nullptr) {
       process->comms.remove(comm);
       comm->cancel();
     }
 
-    simgrid::kernel::activity::SleepImpl* sleep =
-        dynamic_cast<simgrid::kernel::activity::SleepImpl*>(process->waiting_synchro);
+    simgrid::kernel::activity::SleepImplPtr sleep =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::SleepImpl>(process->waiting_synchro);
     if (sleep != nullptr) {
       SIMIX_process_sleep_destroy(process->waiting_synchro);
       if (not xbt_dynar_member(simix_global->process_to_run, &(process)) && process != SIMIX_process_self()) {
@@ -508,12 +505,14 @@ void SIMIX_process_throw(smx_actor_t process, xbt_errcat_t cat, int value, const
       }
     }
 
-    simgrid::kernel::activity::Raw *raw = dynamic_cast<simgrid::kernel::activity::Raw*>(process->waiting_synchro);
+    simgrid::kernel::activity::RawImplPtr raw =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::RawImpl>(process->waiting_synchro);
     if (raw != nullptr) {
       SIMIX_synchro_stop_waiting(process, &process->simcall);
     }
 
-    simgrid::kernel::activity::Io *io = dynamic_cast<simgrid::kernel::activity::Io*>(process->waiting_synchro);
+    simgrid::kernel::activity::IoImplPtr io =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::IoImpl>(process->waiting_synchro);
     if (io != nullptr) {
       SIMIX_io_destroy(process->waiting_synchro);
     }
@@ -694,7 +693,8 @@ void simcall_HANDLER_process_join(smx_simcall_t simcall, smx_actor_t process, do
   simcall->issuer->waiting_synchro = sync;
 }
 
-static int SIMIX_process_join_finish(smx_process_exit_status_t status, smx_activity_t synchro){
+static int SIMIX_process_join_finish(smx_process_exit_status_t status, void* synchro)
+{
   simgrid::kernel::activity::SleepImpl* sleep = static_cast<simgrid::kernel::activity::SleepImpl*>(synchro);
 
   if (sleep->surf_sleep) {
@@ -716,7 +716,6 @@ static int SIMIX_process_join_finish(smx_process_exit_status_t status, smx_activ
     sleep->surf_sleep->unref();
     sleep->surf_sleep = nullptr;
   }
-  sleep->unref();
   // intrusive_ptr_release(process); // FIXME: We are leaking here. See comment in SIMIX_process_join()
   return 0;
 }
@@ -724,14 +723,14 @@ static int SIMIX_process_join_finish(smx_process_exit_status_t status, smx_activ
 smx_activity_t SIMIX_process_join(smx_actor_t issuer, smx_actor_t process, double timeout)
 {
   smx_activity_t res = SIMIX_process_sleep(issuer, timeout);
-  static_cast<simgrid::kernel::activity::ActivityImpl*>(res)->ref();
+  intrusive_ptr_add_ref(res.get());
   /* We are leaking the process here, but if we don't take the ref, we get a "use after free".
    * The correct solution would be to derivate the type SynchroSleep into a SynchroProcessJoin,
    * but the code is not clean enough for now for this.
    * The C API should first be properly replaced with the C++ one, which is a fair amount of work.
    */
   intrusive_ptr_add_ref(process);
-  SIMIX_process_on_exit(process, (int_f_pvoid_pvoid_t)SIMIX_process_join_finish, res);
+  SIMIX_process_on_exit(process, (int_f_pvoid_pvoid_t)SIMIX_process_join_finish, &*res);
   return res;
 }
 
@@ -757,7 +756,7 @@ smx_activity_t SIMIX_process_sleep(smx_actor_t process, double duration)
 
   simgrid::kernel::activity::SleepImpl* synchro = new simgrid::kernel::activity::SleepImpl();
   synchro->host = host;
-  synchro->surf_sleep                       = host->pimpl_cpu->sleep(duration);
+  synchro->surf_sleep                           = host->pimpl_cpu->sleep(duration);
   synchro->surf_sleep->setData(synchro);
   XBT_DEBUG("Create sleep synchronization %p", synchro);
 
@@ -766,13 +765,13 @@ smx_activity_t SIMIX_process_sleep(smx_actor_t process, double duration)
 
 void SIMIX_process_sleep_destroy(smx_activity_t synchro)
 {
-  XBT_DEBUG("Destroy synchro %p", synchro);
-  simgrid::kernel::activity::SleepImpl* sleep = static_cast<simgrid::kernel::activity::SleepImpl*>(synchro);
+  XBT_DEBUG("Destroy sleep synchro %p", synchro.get());
+  simgrid::kernel::activity::SleepImplPtr sleep =
+      boost::dynamic_pointer_cast<simgrid::kernel::activity::SleepImpl>(synchro);
 
   if (sleep->surf_sleep) {
     sleep->surf_sleep->unref();
     sleep->surf_sleep = nullptr;
-    sleep->unref();
   }
 }
 
@@ -958,6 +957,7 @@ void SIMIX_segment_index_set(smx_actor_t proc, int index){
  * \param properties the properties of the process
  * \param auto_restart either it is autorestarting or not.
  */
+extern "C"
 smx_actor_t simcall_process_create(const char* name, xbt_main_func_t code, void* data, sg_host_t host, int argc,
                                    char** argv, xbt_dict_t properties)
 {
diff --git a/src/simix/ActorImpl.hpp b/src/simix/ActorImpl.hpp
index 4a53a46f6b..b15d40ef10 100644
--- a/src/simix/ActorImpl.hpp
+++ b/src/simix/ActorImpl.hpp
@@ -70,7 +70,8 @@ public:
 
   /* Refcounting */
 private:
-  std::atomic_int_fast32_t refcount_{1};
+  std::atomic_int_fast32_t refcount_{0};
+
 public:
   friend void intrusive_ptr_add_ref(ActorImpl* process)
   {
@@ -131,17 +132,13 @@ XBT_PRIVATE void SIMIX_process_empty_trash();
 XBT_PRIVATE void SIMIX_process_yield(smx_actor_t self);
 XBT_PRIVATE void SIMIX_process_exception_terminate(xbt_ex_t * e);
 XBT_PRIVATE void SIMIX_process_change_host(smx_actor_t process, sg_host_t dest);
-XBT_PRIVATE smx_activity_t SIMIX_process_suspend(smx_actor_t process, smx_actor_t issuer);
 XBT_PRIVATE void SIMIX_process_resume(smx_actor_t process);
 XBT_PRIVATE int SIMIX_process_get_PID(smx_actor_t self);
 XBT_PRIVATE void SIMIX_process_set_data(smx_actor_t process, void *data);
 XBT_PRIVATE smx_actor_t SIMIX_process_get_by_name(const char* name);
 XBT_PRIVATE int SIMIX_process_is_suspended(smx_actor_t process);
 XBT_PRIVATE xbt_dict_t SIMIX_process_get_properties(smx_actor_t process);
-XBT_PRIVATE smx_activity_t SIMIX_process_join(smx_actor_t issuer, smx_actor_t process, double timeout);
-XBT_PRIVATE smx_activity_t SIMIX_process_sleep(smx_actor_t process, double duration);
 
-XBT_PRIVATE void SIMIX_process_sleep_destroy(smx_activity_t synchro);
 XBT_PRIVATE void SIMIX_process_auto_restart_set(smx_actor_t process, int auto_restart);
 XBT_PRIVATE smx_actor_t SIMIX_process_restart(smx_actor_t process, smx_actor_t issuer);
 
@@ -150,4 +147,9 @@ extern void (*SMPI_switch_data_segment)(int dest);
 
 SG_END_DECL()
 
+XBT_PRIVATE void SIMIX_process_sleep_destroy(smx_activity_t synchro);
+XBT_PRIVATE smx_activity_t SIMIX_process_suspend(smx_actor_t process, smx_actor_t issuer);
+XBT_PRIVATE smx_activity_t SIMIX_process_join(smx_actor_t issuer, smx_actor_t process, double timeout);
+XBT_PRIVATE smx_activity_t SIMIX_process_sleep(smx_actor_t process, double duration);
+
 #endif
diff --git a/src/simix/libsmx.cpp b/src/simix/libsmx.cpp
index a3aad4a842..5c1ee870bf 100644
--- a/src/simix/libsmx.cpp
+++ b/src/simix/libsmx.cpp
@@ -459,7 +459,8 @@ smx_activity_t simcall_comm_iprobe(smx_mailbox_t mbox, int type, int src, int ta
 void simcall_comm_cancel(smx_activity_t synchro)
 {
   simgrid::simix::kernelImmediate([synchro] {
-    simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+    simgrid::kernel::activity::CommImplPtr comm =
+        boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
     comm->cancel();
   });
 }
@@ -746,18 +747,6 @@ int simcall_file_move(smx_file_t fd, const char* fullpath)
   return simcall_BODY_file_move(fd, fullpath);
 }
 
-/**
- * \ingroup simix_storage_management
- * \brief Returns a dict of the properties assigned to a storage element.
- *
- * \param storage A storage element
- * \return The properties of this storage element
- */
-xbt_dict_t simcall_storage_get_properties(smx_storage_t storage)
-{
-  return simcall_BODY_storage_get_properties(storage);
-}
-
 void simcall_run_kernel(std::function<void()> const& code)
 {
   simcall_BODY_run_kernel(&code);
diff --git a/src/simix/popping_accessors.h b/src/simix/popping_accessors.h
index 8ce7f9f2ad..5e15eb5ed3 100644
--- a/src/simix/popping_accessors.h
+++ b/src/simix/popping_accessors.h
@@ -18,6 +18,10 @@
 static inline smx_actor_t simcall_process_kill__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_kill__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_kill__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
@@ -25,6 +29,10 @@ static inline void simcall_process_kill__set__process(smx_simcall_t simcall, smx
 static inline int simcall_process_killall__get__reset_pid(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[0]);
 }
+static inline int simcall_process_killall__getraw__reset_pid(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[0]);
+}
 static inline void simcall_process_killall__set__reset_pid(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[0], arg);
 }
@@ -32,6 +40,10 @@ static inline void simcall_process_killall__set__reset_pid(smx_simcall_t simcall
 static inline smx_actor_t simcall_process_cleanup__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_cleanup__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_cleanup__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
@@ -39,6 +51,10 @@ static inline void simcall_process_cleanup__set__process(smx_simcall_t simcall,
 static inline smx_actor_t simcall_process_suspend__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_suspend__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_suspend__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
@@ -46,6 +62,10 @@ static inline void simcall_process_suspend__set__process(smx_simcall_t simcall,
 static inline smx_actor_t simcall_process_resume__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_resume__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_resume__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
@@ -53,12 +73,20 @@ static inline void simcall_process_resume__set__process(smx_simcall_t simcall, s
 static inline smx_actor_t simcall_process_set_host__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_set_host__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_set_host__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline sg_host_t simcall_process_set_host__get__dest(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_host_t>(simcall->args[1]);
 }
+static inline sg_host_t simcall_process_set_host__getraw__dest(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_host_t>(simcall->args[1]);
+}
 static inline void simcall_process_set_host__set__dest(smx_simcall_t simcall, sg_host_t arg) {
     simgrid::simix::marshal<sg_host_t>(simcall->args[1], arg);
 }
@@ -66,12 +94,20 @@ static inline void simcall_process_set_host__set__dest(smx_simcall_t simcall, sg
 static inline smx_actor_t simcall_process_is_suspended__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_is_suspended__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_is_suspended__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline int simcall_process_is_suspended__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_process_is_suspended__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_process_is_suspended__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -79,18 +115,30 @@ static inline void simcall_process_is_suspended__set__result(smx_simcall_t simca
 static inline smx_actor_t simcall_process_join__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_join__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_join__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline double simcall_process_join__get__timeout(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[1]);
 }
+static inline double simcall_process_join__getraw__timeout(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[1]);
+}
 static inline void simcall_process_join__set__timeout(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[1], arg);
 }
 static inline int simcall_process_join__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_process_join__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_process_join__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -98,12 +146,20 @@ static inline void simcall_process_join__set__result(smx_simcall_t simcall, int
 static inline double simcall_process_sleep__get__duration(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[0]);
 }
+static inline double simcall_process_sleep__getraw__duration(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[0]);
+}
 static inline void simcall_process_sleep__set__duration(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[0], arg);
 }
 static inline int simcall_process_sleep__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_process_sleep__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_process_sleep__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -111,131 +167,249 @@ static inline void simcall_process_sleep__set__result(smx_simcall_t simcall, int
 static inline const char* simcall_execution_start__get__name(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<const char*>(simcall->args[0]);
 }
+static inline const char* simcall_execution_start__getraw__name(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<const char*>(simcall->args[0]);
+}
 static inline void simcall_execution_start__set__name(smx_simcall_t simcall, const char* arg) {
     simgrid::simix::marshal<const char*>(simcall->args[0], arg);
 }
 static inline double simcall_execution_start__get__flops_amount(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[1]);
 }
+static inline double simcall_execution_start__getraw__flops_amount(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[1]);
+}
 static inline void simcall_execution_start__set__flops_amount(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[1], arg);
 }
 static inline double simcall_execution_start__get__priority(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[2]);
 }
+static inline double simcall_execution_start__getraw__priority(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[2]);
+}
 static inline void simcall_execution_start__set__priority(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[2], arg);
 }
 static inline double simcall_execution_start__get__bound(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[3]);
 }
+static inline double simcall_execution_start__getraw__bound(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[3]);
+}
 static inline void simcall_execution_start__set__bound(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[3], arg);
 }
-static inline smx_activity_t simcall_execution_start__get__result(smx_simcall_t simcall){
-    return simgrid::simix::unmarshal<smx_activity_t>(simcall->result);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_execution_start__get__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result);
+}
+static inline simgrid::kernel::activity::ActivityImpl* simcall_execution_start__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->result);
 }
-static inline void simcall_execution_start__set__result(smx_simcall_t simcall, smx_activity_t result){
-    simgrid::simix::marshal<smx_activity_t>(simcall->result, result);
+static inline void
+simcall_execution_start__set__result(smx_simcall_t simcall,
+                                     boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> result)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result, result);
 }
 
 static inline const char* simcall_execution_parallel_start__get__name(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<const char*>(simcall->args[0]);
 }
+static inline const char* simcall_execution_parallel_start__getraw__name(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<const char*>(simcall->args[0]);
+}
 static inline void simcall_execution_parallel_start__set__name(smx_simcall_t simcall, const char* arg) {
     simgrid::simix::marshal<const char*>(simcall->args[0], arg);
 }
 static inline int simcall_execution_parallel_start__get__host_nb(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[1]);
 }
+static inline int simcall_execution_parallel_start__getraw__host_nb(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[1]);
+}
 static inline void simcall_execution_parallel_start__set__host_nb(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[1], arg);
 }
 static inline sg_host_t* simcall_execution_parallel_start__get__host_list(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_host_t*>(simcall->args[2]);
 }
+static inline sg_host_t* simcall_execution_parallel_start__getraw__host_list(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_host_t*>(simcall->args[2]);
+}
 static inline void simcall_execution_parallel_start__set__host_list(smx_simcall_t simcall, sg_host_t* arg) {
     simgrid::simix::marshal<sg_host_t*>(simcall->args[2], arg);
 }
 static inline double* simcall_execution_parallel_start__get__flops_amount(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double*>(simcall->args[3]);
 }
+static inline double* simcall_execution_parallel_start__getraw__flops_amount(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double*>(simcall->args[3]);
+}
 static inline void simcall_execution_parallel_start__set__flops_amount(smx_simcall_t simcall, double* arg) {
     simgrid::simix::marshal<double*>(simcall->args[3], arg);
 }
 static inline double* simcall_execution_parallel_start__get__bytes_amount(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double*>(simcall->args[4]);
 }
+static inline double* simcall_execution_parallel_start__getraw__bytes_amount(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double*>(simcall->args[4]);
+}
 static inline void simcall_execution_parallel_start__set__bytes_amount(smx_simcall_t simcall, double* arg) {
     simgrid::simix::marshal<double*>(simcall->args[4], arg);
 }
 static inline double simcall_execution_parallel_start__get__amount(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[5]);
 }
+static inline double simcall_execution_parallel_start__getraw__amount(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[5]);
+}
 static inline void simcall_execution_parallel_start__set__amount(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[5], arg);
 }
 static inline double simcall_execution_parallel_start__get__rate(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[6]);
 }
+static inline double simcall_execution_parallel_start__getraw__rate(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[6]);
+}
 static inline void simcall_execution_parallel_start__set__rate(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[6], arg);
 }
 static inline double simcall_execution_parallel_start__get__timeout(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[7]);
 }
+static inline double simcall_execution_parallel_start__getraw__timeout(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[7]);
+}
 static inline void simcall_execution_parallel_start__set__timeout(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[7], arg);
 }
-static inline smx_activity_t simcall_execution_parallel_start__get__result(smx_simcall_t simcall){
-    return simgrid::simix::unmarshal<smx_activity_t>(simcall->result);
-}
-static inline void simcall_execution_parallel_start__set__result(smx_simcall_t simcall, smx_activity_t result){
-    simgrid::simix::marshal<smx_activity_t>(simcall->result, result);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_execution_parallel_start__get__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result);
 }
-
-static inline smx_activity_t simcall_execution_cancel__get__execution(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]);
+static inline simgrid::kernel::activity::ActivityImpl*
+simcall_execution_parallel_start__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->result);
 }
-static inline void simcall_execution_cancel__set__execution(smx_simcall_t simcall, smx_activity_t arg) {
-    simgrid::simix::marshal<smx_activity_t>(simcall->args[0], arg);
+static inline void
+simcall_execution_parallel_start__set__result(smx_simcall_t simcall,
+                                              boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> result)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result, result);
 }
 
-static inline smx_activity_t simcall_execution_set_priority__get__execution(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_execution_cancel__get__execution(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]);
+}
+static inline simgrid::kernel::activity::ActivityImpl*
+simcall_execution_cancel__getraw__execution(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->args[0]);
+}
+static inline void
+simcall_execution_cancel__set__execution(smx_simcall_t simcall,
+                                         boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> arg)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0], arg);
 }
-static inline void simcall_execution_set_priority__set__execution(smx_simcall_t simcall, smx_activity_t arg) {
-    simgrid::simix::marshal<smx_activity_t>(simcall->args[0], arg);
+
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_execution_set_priority__get__execution(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]);
+}
+static inline simgrid::kernel::activity::ActivityImpl*
+simcall_execution_set_priority__getraw__execution(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->args[0]);
+}
+static inline void
+simcall_execution_set_priority__set__execution(smx_simcall_t simcall,
+                                               boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> arg)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0], arg);
 }
 static inline double simcall_execution_set_priority__get__priority(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[1]);
 }
+static inline double simcall_execution_set_priority__getraw__priority(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[1]);
+}
 static inline void simcall_execution_set_priority__set__priority(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[1], arg);
 }
 
-static inline smx_activity_t simcall_execution_set_bound__get__execution(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]);
-}
-static inline void simcall_execution_set_bound__set__execution(smx_simcall_t simcall, smx_activity_t arg) {
-    simgrid::simix::marshal<smx_activity_t>(simcall->args[0], arg);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_execution_set_bound__get__execution(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]);
+}
+static inline simgrid::kernel::activity::ActivityImpl*
+simcall_execution_set_bound__getraw__execution(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->args[0]);
+}
+static inline void
+simcall_execution_set_bound__set__execution(smx_simcall_t simcall,
+                                            boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> arg)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0], arg);
 }
 static inline double simcall_execution_set_bound__get__bound(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[1]);
 }
+static inline double simcall_execution_set_bound__getraw__bound(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[1]);
+}
 static inline void simcall_execution_set_bound__set__bound(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[1], arg);
 }
 
-static inline smx_activity_t simcall_execution_wait__get__execution(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_execution_wait__get__execution(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]);
+}
+static inline simgrid::kernel::activity::ActivityImpl* simcall_execution_wait__getraw__execution(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->args[0]);
 }
-static inline void simcall_execution_wait__set__execution(smx_simcall_t simcall, smx_activity_t arg) {
-    simgrid::simix::marshal<smx_activity_t>(simcall->args[0], arg);
+static inline void
+simcall_execution_wait__set__execution(smx_simcall_t simcall,
+                                       boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> arg)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0], arg);
 }
 static inline int simcall_execution_wait__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_execution_wait__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_execution_wait__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -243,18 +417,30 @@ static inline void simcall_execution_wait__set__result(smx_simcall_t simcall, in
 static inline smx_actor_t simcall_process_on_exit__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_on_exit__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_on_exit__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline int_f_pvoid_pvoid_t simcall_process_on_exit__get__fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int_f_pvoid_pvoid_t>(simcall->args[1]);
 }
+static inline int_f_pvoid_pvoid_t simcall_process_on_exit__getraw__fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int_f_pvoid_pvoid_t>(simcall->args[1]);
+}
 static inline void simcall_process_on_exit__set__fun(smx_simcall_t simcall, int_f_pvoid_pvoid_t arg) {
     simgrid::simix::marshal<int_f_pvoid_pvoid_t>(simcall->args[1], arg);
 }
 static inline void* simcall_process_on_exit__get__data(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[2]);
 }
+static inline void* simcall_process_on_exit__getraw__data(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[2]);
+}
 static inline void simcall_process_on_exit__set__data(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[2], arg);
 }
@@ -262,12 +448,20 @@ static inline void simcall_process_on_exit__set__data(smx_simcall_t simcall, voi
 static inline smx_actor_t simcall_process_auto_restart_set__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_auto_restart_set__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_auto_restart_set__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline int simcall_process_auto_restart_set__get__auto_restart(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[1]);
 }
+static inline int simcall_process_auto_restart_set__getraw__auto_restart(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[1]);
+}
 static inline void simcall_process_auto_restart_set__set__auto_restart(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[1], arg);
 }
@@ -275,12 +469,20 @@ static inline void simcall_process_auto_restart_set__set__auto_restart(smx_simca
 static inline smx_actor_t simcall_process_restart__get__process(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_process_restart__getraw__process(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_process_restart__set__process(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline smx_actor_t simcall_process_restart__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<smx_actor_t>(simcall->result);
 }
+static inline smx_actor_t simcall_process_restart__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->result);
+}
 static inline void simcall_process_restart__set__result(smx_simcall_t simcall, smx_actor_t result){
     simgrid::simix::marshal<smx_actor_t>(simcall->result, result);
 }
@@ -288,103 +490,176 @@ static inline void simcall_process_restart__set__result(smx_simcall_t simcall, s
 static inline smx_mailbox_t simcall_comm_iprobe__get__mbox(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[0]);
 }
+static inline smx_mailbox_t simcall_comm_iprobe__getraw__mbox(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mailbox_t>(simcall->args[0]);
+}
 static inline void simcall_comm_iprobe__set__mbox(smx_simcall_t simcall, smx_mailbox_t arg) {
     simgrid::simix::marshal<smx_mailbox_t>(simcall->args[0], arg);
 }
 static inline int simcall_comm_iprobe__get__type(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[1]);
 }
+static inline int simcall_comm_iprobe__getraw__type(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[1]);
+}
 static inline void simcall_comm_iprobe__set__type(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[1], arg);
 }
 static inline int simcall_comm_iprobe__get__src(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[2]);
 }
+static inline int simcall_comm_iprobe__getraw__src(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[2]);
+}
 static inline void simcall_comm_iprobe__set__src(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[2], arg);
 }
 static inline int simcall_comm_iprobe__get__tag(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[3]);
 }
+static inline int simcall_comm_iprobe__getraw__tag(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[3]);
+}
 static inline void simcall_comm_iprobe__set__tag(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[3], arg);
 }
 static inline simix_match_func_t simcall_comm_iprobe__get__match_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[4]);
 }
+static inline simix_match_func_t simcall_comm_iprobe__getraw__match_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_match_func_t>(simcall->args[4]);
+}
 static inline void simcall_comm_iprobe__set__match_fun(smx_simcall_t simcall, simix_match_func_t arg) {
     simgrid::simix::marshal<simix_match_func_t>(simcall->args[4], arg);
 }
 static inline void* simcall_comm_iprobe__get__data(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[5]);
 }
+static inline void* simcall_comm_iprobe__getraw__data(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[5]);
+}
 static inline void simcall_comm_iprobe__set__data(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[5], arg);
 }
-static inline smx_activity_t simcall_comm_iprobe__get__result(smx_simcall_t simcall){
-    return simgrid::simix::unmarshal<smx_activity_t>(simcall->result);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_comm_iprobe__get__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result);
+}
+static inline simgrid::kernel::activity::ActivityImpl* simcall_comm_iprobe__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->result);
 }
-static inline void simcall_comm_iprobe__set__result(smx_simcall_t simcall, smx_activity_t result){
-    simgrid::simix::marshal<smx_activity_t>(simcall->result, result);
+static inline void
+simcall_comm_iprobe__set__result(smx_simcall_t simcall,
+                                 boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> result)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result, result);
 }
 
 static inline smx_actor_t simcall_comm_send__get__sender(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_comm_send__getraw__sender(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_comm_send__set__sender(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline smx_mailbox_t simcall_comm_send__get__mbox(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]);
 }
+static inline smx_mailbox_t simcall_comm_send__getraw__mbox(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mailbox_t>(simcall->args[1]);
+}
 static inline void simcall_comm_send__set__mbox(smx_simcall_t simcall, smx_mailbox_t arg) {
     simgrid::simix::marshal<smx_mailbox_t>(simcall->args[1], arg);
 }
 static inline double simcall_comm_send__get__task_size(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[2]);
 }
+static inline double simcall_comm_send__getraw__task_size(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[2]);
+}
 static inline void simcall_comm_send__set__task_size(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[2], arg);
 }
 static inline double simcall_comm_send__get__rate(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[3]);
 }
+static inline double simcall_comm_send__getraw__rate(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[3]);
+}
 static inline void simcall_comm_send__set__rate(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[3], arg);
 }
 static inline void* simcall_comm_send__get__src_buff(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[4]);
 }
+static inline void* simcall_comm_send__getraw__src_buff(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[4]);
+}
 static inline void simcall_comm_send__set__src_buff(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[4], arg);
 }
 static inline size_t simcall_comm_send__get__src_buff_size(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<size_t>(simcall->args[5]);
 }
+static inline size_t simcall_comm_send__getraw__src_buff_size(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<size_t>(simcall->args[5]);
+}
 static inline void simcall_comm_send__set__src_buff_size(smx_simcall_t simcall, size_t arg) {
     simgrid::simix::marshal<size_t>(simcall->args[5], arg);
 }
 static inline simix_match_func_t simcall_comm_send__get__match_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[6]);
 }
+static inline simix_match_func_t simcall_comm_send__getraw__match_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_match_func_t>(simcall->args[6]);
+}
 static inline void simcall_comm_send__set__match_fun(smx_simcall_t simcall, simix_match_func_t arg) {
     simgrid::simix::marshal<simix_match_func_t>(simcall->args[6], arg);
 }
 static inline simix_copy_data_func_t simcall_comm_send__get__copy_data_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[7]);
 }
+static inline simix_copy_data_func_t simcall_comm_send__getraw__copy_data_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_copy_data_func_t>(simcall->args[7]);
+}
 static inline void simcall_comm_send__set__copy_data_fun(smx_simcall_t simcall, simix_copy_data_func_t arg) {
     simgrid::simix::marshal<simix_copy_data_func_t>(simcall->args[7], arg);
 }
 static inline void* simcall_comm_send__get__data(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[8]);
 }
+static inline void* simcall_comm_send__getraw__data(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[8]);
+}
 static inline void simcall_comm_send__set__data(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[8], arg);
 }
 static inline double simcall_comm_send__get__timeout(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[9]);
 }
+static inline double simcall_comm_send__getraw__timeout(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[9]);
+}
 static inline void simcall_comm_send__set__timeout(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[9], arg);
 }
@@ -392,127 +667,215 @@ static inline void simcall_comm_send__set__timeout(smx_simcall_t simcall, double
 static inline smx_actor_t simcall_comm_isend__get__sender(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_comm_isend__getraw__sender(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_comm_isend__set__sender(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline smx_mailbox_t simcall_comm_isend__get__mbox(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]);
 }
+static inline smx_mailbox_t simcall_comm_isend__getraw__mbox(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mailbox_t>(simcall->args[1]);
+}
 static inline void simcall_comm_isend__set__mbox(smx_simcall_t simcall, smx_mailbox_t arg) {
     simgrid::simix::marshal<smx_mailbox_t>(simcall->args[1], arg);
 }
 static inline double simcall_comm_isend__get__task_size(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[2]);
 }
+static inline double simcall_comm_isend__getraw__task_size(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[2]);
+}
 static inline void simcall_comm_isend__set__task_size(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[2], arg);
 }
 static inline double simcall_comm_isend__get__rate(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[3]);
 }
+static inline double simcall_comm_isend__getraw__rate(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[3]);
+}
 static inline void simcall_comm_isend__set__rate(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[3], arg);
 }
 static inline void* simcall_comm_isend__get__src_buff(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[4]);
 }
+static inline void* simcall_comm_isend__getraw__src_buff(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[4]);
+}
 static inline void simcall_comm_isend__set__src_buff(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[4], arg);
 }
 static inline size_t simcall_comm_isend__get__src_buff_size(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<size_t>(simcall->args[5]);
 }
+static inline size_t simcall_comm_isend__getraw__src_buff_size(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<size_t>(simcall->args[5]);
+}
 static inline void simcall_comm_isend__set__src_buff_size(smx_simcall_t simcall, size_t arg) {
     simgrid::simix::marshal<size_t>(simcall->args[5], arg);
 }
 static inline simix_match_func_t simcall_comm_isend__get__match_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[6]);
 }
+static inline simix_match_func_t simcall_comm_isend__getraw__match_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_match_func_t>(simcall->args[6]);
+}
 static inline void simcall_comm_isend__set__match_fun(smx_simcall_t simcall, simix_match_func_t arg) {
     simgrid::simix::marshal<simix_match_func_t>(simcall->args[6], arg);
 }
 static inline simix_clean_func_t simcall_comm_isend__get__clean_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_clean_func_t>(simcall->args[7]);
 }
+static inline simix_clean_func_t simcall_comm_isend__getraw__clean_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_clean_func_t>(simcall->args[7]);
+}
 static inline void simcall_comm_isend__set__clean_fun(smx_simcall_t simcall, simix_clean_func_t arg) {
     simgrid::simix::marshal<simix_clean_func_t>(simcall->args[7], arg);
 }
 static inline simix_copy_data_func_t simcall_comm_isend__get__copy_data_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[8]);
 }
+static inline simix_copy_data_func_t simcall_comm_isend__getraw__copy_data_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_copy_data_func_t>(simcall->args[8]);
+}
 static inline void simcall_comm_isend__set__copy_data_fun(smx_simcall_t simcall, simix_copy_data_func_t arg) {
     simgrid::simix::marshal<simix_copy_data_func_t>(simcall->args[8], arg);
 }
 static inline void* simcall_comm_isend__get__data(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[9]);
 }
+static inline void* simcall_comm_isend__getraw__data(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[9]);
+}
 static inline void simcall_comm_isend__set__data(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[9], arg);
 }
 static inline int simcall_comm_isend__get__detached(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[10]);
 }
+static inline int simcall_comm_isend__getraw__detached(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[10]);
+}
 static inline void simcall_comm_isend__set__detached(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[10], arg);
 }
-static inline smx_activity_t simcall_comm_isend__get__result(smx_simcall_t simcall){
-    return simgrid::simix::unmarshal<smx_activity_t>(simcall->result);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_comm_isend__get__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result);
 }
-static inline void simcall_comm_isend__set__result(smx_simcall_t simcall, smx_activity_t result){
-    simgrid::simix::marshal<smx_activity_t>(simcall->result, result);
+static inline simgrid::kernel::activity::ActivityImpl* simcall_comm_isend__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->result);
+}
+static inline void simcall_comm_isend__set__result(smx_simcall_t simcall,
+                                                   boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> result)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result, result);
 }
 
 static inline smx_actor_t simcall_comm_recv__get__receiver(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_comm_recv__getraw__receiver(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_comm_recv__set__receiver(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline smx_mailbox_t simcall_comm_recv__get__mbox(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]);
 }
+static inline smx_mailbox_t simcall_comm_recv__getraw__mbox(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mailbox_t>(simcall->args[1]);
+}
 static inline void simcall_comm_recv__set__mbox(smx_simcall_t simcall, smx_mailbox_t arg) {
     simgrid::simix::marshal<smx_mailbox_t>(simcall->args[1], arg);
 }
 static inline void* simcall_comm_recv__get__dst_buff(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[2]);
 }
+static inline void* simcall_comm_recv__getraw__dst_buff(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[2]);
+}
 static inline void simcall_comm_recv__set__dst_buff(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[2], arg);
 }
 static inline size_t* simcall_comm_recv__get__dst_buff_size(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<size_t*>(simcall->args[3]);
 }
+static inline size_t* simcall_comm_recv__getraw__dst_buff_size(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<size_t*>(simcall->args[3]);
+}
 static inline void simcall_comm_recv__set__dst_buff_size(smx_simcall_t simcall, size_t* arg) {
     simgrid::simix::marshal<size_t*>(simcall->args[3], arg);
 }
 static inline simix_match_func_t simcall_comm_recv__get__match_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[4]);
 }
+static inline simix_match_func_t simcall_comm_recv__getraw__match_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_match_func_t>(simcall->args[4]);
+}
 static inline void simcall_comm_recv__set__match_fun(smx_simcall_t simcall, simix_match_func_t arg) {
     simgrid::simix::marshal<simix_match_func_t>(simcall->args[4], arg);
 }
 static inline simix_copy_data_func_t simcall_comm_recv__get__copy_data_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[5]);
 }
+static inline simix_copy_data_func_t simcall_comm_recv__getraw__copy_data_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_copy_data_func_t>(simcall->args[5]);
+}
 static inline void simcall_comm_recv__set__copy_data_fun(smx_simcall_t simcall, simix_copy_data_func_t arg) {
     simgrid::simix::marshal<simix_copy_data_func_t>(simcall->args[5], arg);
 }
 static inline void* simcall_comm_recv__get__data(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[6]);
 }
+static inline void* simcall_comm_recv__getraw__data(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[6]);
+}
 static inline void simcall_comm_recv__set__data(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[6], arg);
 }
 static inline double simcall_comm_recv__get__timeout(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[7]);
 }
+static inline double simcall_comm_recv__getraw__timeout(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[7]);
+}
 static inline void simcall_comm_recv__set__timeout(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[7], arg);
 }
 static inline double simcall_comm_recv__get__rate(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[8]);
 }
+static inline double simcall_comm_recv__getraw__rate(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[8]);
+}
 static inline void simcall_comm_recv__set__rate(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[8], arg);
 }
@@ -520,118 +883,210 @@ static inline void simcall_comm_recv__set__rate(smx_simcall_t simcall, double ar
 static inline smx_actor_t simcall_comm_irecv__get__receiver(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]);
 }
+static inline smx_actor_t simcall_comm_irecv__getraw__receiver(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_actor_t>(simcall->args[0]);
+}
 static inline void simcall_comm_irecv__set__receiver(smx_simcall_t simcall, smx_actor_t arg) {
     simgrid::simix::marshal<smx_actor_t>(simcall->args[0], arg);
 }
 static inline smx_mailbox_t simcall_comm_irecv__get__mbox(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]);
 }
+static inline smx_mailbox_t simcall_comm_irecv__getraw__mbox(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mailbox_t>(simcall->args[1]);
+}
 static inline void simcall_comm_irecv__set__mbox(smx_simcall_t simcall, smx_mailbox_t arg) {
     simgrid::simix::marshal<smx_mailbox_t>(simcall->args[1], arg);
 }
 static inline void* simcall_comm_irecv__get__dst_buff(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[2]);
 }
+static inline void* simcall_comm_irecv__getraw__dst_buff(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[2]);
+}
 static inline void simcall_comm_irecv__set__dst_buff(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[2], arg);
 }
 static inline size_t* simcall_comm_irecv__get__dst_buff_size(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<size_t*>(simcall->args[3]);
 }
+static inline size_t* simcall_comm_irecv__getraw__dst_buff_size(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<size_t*>(simcall->args[3]);
+}
 static inline void simcall_comm_irecv__set__dst_buff_size(smx_simcall_t simcall, size_t* arg) {
     simgrid::simix::marshal<size_t*>(simcall->args[3], arg);
 }
 static inline simix_match_func_t simcall_comm_irecv__get__match_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[4]);
 }
+static inline simix_match_func_t simcall_comm_irecv__getraw__match_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_match_func_t>(simcall->args[4]);
+}
 static inline void simcall_comm_irecv__set__match_fun(smx_simcall_t simcall, simix_match_func_t arg) {
     simgrid::simix::marshal<simix_match_func_t>(simcall->args[4], arg);
 }
 static inline simix_copy_data_func_t simcall_comm_irecv__get__copy_data_fun(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[5]);
 }
+static inline simix_copy_data_func_t simcall_comm_irecv__getraw__copy_data_fun(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simix_copy_data_func_t>(simcall->args[5]);
+}
 static inline void simcall_comm_irecv__set__copy_data_fun(smx_simcall_t simcall, simix_copy_data_func_t arg) {
     simgrid::simix::marshal<simix_copy_data_func_t>(simcall->args[5], arg);
 }
 static inline void* simcall_comm_irecv__get__data(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<void*>(simcall->args[6]);
 }
+static inline void* simcall_comm_irecv__getraw__data(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<void*>(simcall->args[6]);
+}
 static inline void simcall_comm_irecv__set__data(smx_simcall_t simcall, void* arg) {
     simgrid::simix::marshal<void*>(simcall->args[6], arg);
 }
 static inline double simcall_comm_irecv__get__rate(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[7]);
 }
+static inline double simcall_comm_irecv__getraw__rate(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[7]);
+}
 static inline void simcall_comm_irecv__set__rate(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[7], arg);
 }
-static inline smx_activity_t simcall_comm_irecv__get__result(smx_simcall_t simcall){
-    return simgrid::simix::unmarshal<smx_activity_t>(simcall->result);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_comm_irecv__get__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result);
 }
-static inline void simcall_comm_irecv__set__result(smx_simcall_t simcall, smx_activity_t result){
-    simgrid::simix::marshal<smx_activity_t>(simcall->result, result);
+static inline simgrid::kernel::activity::ActivityImpl* simcall_comm_irecv__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->result);
+}
+static inline void simcall_comm_irecv__set__result(smx_simcall_t simcall,
+                                                   boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> result)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->result, result);
 }
 
 static inline xbt_dynar_t simcall_comm_waitany__get__comms(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<xbt_dynar_t>(simcall->args[0]);
 }
+static inline xbt_dynar_t simcall_comm_waitany__getraw__comms(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<xbt_dynar_t>(simcall->args[0]);
+}
 static inline void simcall_comm_waitany__set__comms(smx_simcall_t simcall, xbt_dynar_t arg) {
     simgrid::simix::marshal<xbt_dynar_t>(simcall->args[0], arg);
 }
 static inline double simcall_comm_waitany__get__timeout(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[1]);
 }
+static inline double simcall_comm_waitany__getraw__timeout(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[1]);
+}
 static inline void simcall_comm_waitany__set__timeout(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[1], arg);
 }
 static inline int simcall_comm_waitany__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_comm_waitany__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_comm_waitany__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
 
-static inline smx_activity_t simcall_comm_wait__get__comm(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_comm_wait__get__comm(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]);
+}
+static inline simgrid::kernel::activity::ActivityImpl* simcall_comm_wait__getraw__comm(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->args[0]);
 }
-static inline void simcall_comm_wait__set__comm(smx_simcall_t simcall, smx_activity_t arg) {
-    simgrid::simix::marshal<smx_activity_t>(simcall->args[0], arg);
+static inline void simcall_comm_wait__set__comm(smx_simcall_t simcall,
+                                                boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> arg)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0], arg);
 }
 static inline double simcall_comm_wait__get__timeout(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[1]);
 }
+static inline double simcall_comm_wait__getraw__timeout(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[1]);
+}
 static inline void simcall_comm_wait__set__timeout(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[1], arg);
 }
 
-static inline smx_activity_t simcall_comm_test__get__comm(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_comm_test__get__comm(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]);
+}
+static inline simgrid::kernel::activity::ActivityImpl* simcall_comm_test__getraw__comm(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->args[0]);
 }
-static inline void simcall_comm_test__set__comm(smx_simcall_t simcall, smx_activity_t arg) {
-    simgrid::simix::marshal<smx_activity_t>(simcall->args[0], arg);
+static inline void simcall_comm_test__set__comm(smx_simcall_t simcall,
+                                                boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> arg)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0], arg);
 }
 static inline int simcall_comm_test__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_comm_test__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_comm_test__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
 
-static inline smx_activity_t* simcall_comm_testany__get__comms(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_activity_t*>(simcall->args[0]);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>*
+simcall_comm_testany__get__comms(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>*>(simcall->args[0]);
 }
-static inline void simcall_comm_testany__set__comms(smx_simcall_t simcall, smx_activity_t* arg) {
-    simgrid::simix::marshal<smx_activity_t*>(simcall->args[0], arg);
+static inline simgrid::kernel::activity::ActivityImpl** simcall_comm_testany__getraw__comms(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl**>(simcall->args[0]);
+}
+static inline void simcall_comm_testany__set__comms(smx_simcall_t simcall,
+                                                    boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>* arg)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>*>(simcall->args[0], arg);
 }
 static inline size_t simcall_comm_testany__get__count(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<size_t>(simcall->args[1]);
 }
+static inline size_t simcall_comm_testany__getraw__count(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<size_t>(simcall->args[1]);
+}
 static inline void simcall_comm_testany__set__count(smx_simcall_t simcall, size_t arg) {
     simgrid::simix::marshal<size_t>(simcall->args[1], arg);
 }
 static inline int simcall_comm_testany__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_comm_testany__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_comm_testany__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -639,6 +1094,10 @@ static inline void simcall_comm_testany__set__result(smx_simcall_t simcall, int
 static inline smx_mutex_t simcall_mutex_init__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<smx_mutex_t>(simcall->result);
 }
+static inline smx_mutex_t simcall_mutex_init__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mutex_t>(simcall->result);
+}
 static inline void simcall_mutex_init__set__result(smx_simcall_t simcall, smx_mutex_t result){
     simgrid::simix::marshal<smx_mutex_t>(simcall->result, result);
 }
@@ -646,6 +1105,10 @@ static inline void simcall_mutex_init__set__result(smx_simcall_t simcall, smx_mu
 static inline smx_mutex_t simcall_mutex_lock__get__mutex(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mutex_t>(simcall->args[0]);
 }
+static inline smx_mutex_t simcall_mutex_lock__getraw__mutex(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mutex_t>(simcall->args[0]);
+}
 static inline void simcall_mutex_lock__set__mutex(smx_simcall_t simcall, smx_mutex_t arg) {
     simgrid::simix::marshal<smx_mutex_t>(simcall->args[0], arg);
 }
@@ -653,12 +1116,20 @@ static inline void simcall_mutex_lock__set__mutex(smx_simcall_t simcall, smx_mut
 static inline smx_mutex_t simcall_mutex_trylock__get__mutex(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mutex_t>(simcall->args[0]);
 }
+static inline smx_mutex_t simcall_mutex_trylock__getraw__mutex(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mutex_t>(simcall->args[0]);
+}
 static inline void simcall_mutex_trylock__set__mutex(smx_simcall_t simcall, smx_mutex_t arg) {
     simgrid::simix::marshal<smx_mutex_t>(simcall->args[0], arg);
 }
 static inline int simcall_mutex_trylock__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_mutex_trylock__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_mutex_trylock__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -666,6 +1137,10 @@ static inline void simcall_mutex_trylock__set__result(smx_simcall_t simcall, int
 static inline smx_mutex_t simcall_mutex_unlock__get__mutex(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mutex_t>(simcall->args[0]);
 }
+static inline smx_mutex_t simcall_mutex_unlock__getraw__mutex(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mutex_t>(simcall->args[0]);
+}
 static inline void simcall_mutex_unlock__set__mutex(smx_simcall_t simcall, smx_mutex_t arg) {
     simgrid::simix::marshal<smx_mutex_t>(simcall->args[0], arg);
 }
@@ -673,6 +1148,10 @@ static inline void simcall_mutex_unlock__set__mutex(smx_simcall_t simcall, smx_m
 static inline smx_cond_t simcall_cond_init__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<smx_cond_t>(simcall->result);
 }
+static inline smx_cond_t simcall_cond_init__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_cond_t>(simcall->result);
+}
 static inline void simcall_cond_init__set__result(smx_simcall_t simcall, smx_cond_t result){
     simgrid::simix::marshal<smx_cond_t>(simcall->result, result);
 }
@@ -680,6 +1159,10 @@ static inline void simcall_cond_init__set__result(smx_simcall_t simcall, smx_con
 static inline smx_cond_t simcall_cond_signal__get__cond(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_cond_t>(simcall->args[0]);
 }
+static inline smx_cond_t simcall_cond_signal__getraw__cond(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_cond_t>(simcall->args[0]);
+}
 static inline void simcall_cond_signal__set__cond(smx_simcall_t simcall, smx_cond_t arg) {
     simgrid::simix::marshal<smx_cond_t>(simcall->args[0], arg);
 }
@@ -687,12 +1170,20 @@ static inline void simcall_cond_signal__set__cond(smx_simcall_t simcall, smx_con
 static inline smx_cond_t simcall_cond_wait__get__cond(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_cond_t>(simcall->args[0]);
 }
+static inline smx_cond_t simcall_cond_wait__getraw__cond(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_cond_t>(simcall->args[0]);
+}
 static inline void simcall_cond_wait__set__cond(smx_simcall_t simcall, smx_cond_t arg) {
     simgrid::simix::marshal<smx_cond_t>(simcall->args[0], arg);
 }
 static inline smx_mutex_t simcall_cond_wait__get__mutex(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mutex_t>(simcall->args[1]);
 }
+static inline smx_mutex_t simcall_cond_wait__getraw__mutex(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mutex_t>(simcall->args[1]);
+}
 static inline void simcall_cond_wait__set__mutex(smx_simcall_t simcall, smx_mutex_t arg) {
     simgrid::simix::marshal<smx_mutex_t>(simcall->args[1], arg);
 }
@@ -700,18 +1191,30 @@ static inline void simcall_cond_wait__set__mutex(smx_simcall_t simcall, smx_mute
 static inline smx_cond_t simcall_cond_wait_timeout__get__cond(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_cond_t>(simcall->args[0]);
 }
+static inline smx_cond_t simcall_cond_wait_timeout__getraw__cond(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_cond_t>(simcall->args[0]);
+}
 static inline void simcall_cond_wait_timeout__set__cond(smx_simcall_t simcall, smx_cond_t arg) {
     simgrid::simix::marshal<smx_cond_t>(simcall->args[0], arg);
 }
 static inline smx_mutex_t simcall_cond_wait_timeout__get__mutex(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_mutex_t>(simcall->args[1]);
 }
+static inline smx_mutex_t simcall_cond_wait_timeout__getraw__mutex(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_mutex_t>(simcall->args[1]);
+}
 static inline void simcall_cond_wait_timeout__set__mutex(smx_simcall_t simcall, smx_mutex_t arg) {
     simgrid::simix::marshal<smx_mutex_t>(simcall->args[1], arg);
 }
 static inline double simcall_cond_wait_timeout__get__timeout(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[2]);
 }
+static inline double simcall_cond_wait_timeout__getraw__timeout(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[2]);
+}
 static inline void simcall_cond_wait_timeout__set__timeout(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[2], arg);
 }
@@ -719,6 +1222,10 @@ static inline void simcall_cond_wait_timeout__set__timeout(smx_simcall_t simcall
 static inline smx_cond_t simcall_cond_broadcast__get__cond(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_cond_t>(simcall->args[0]);
 }
+static inline smx_cond_t simcall_cond_broadcast__getraw__cond(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_cond_t>(simcall->args[0]);
+}
 static inline void simcall_cond_broadcast__set__cond(smx_simcall_t simcall, smx_cond_t arg) {
     simgrid::simix::marshal<smx_cond_t>(simcall->args[0], arg);
 }
@@ -726,12 +1233,20 @@ static inline void simcall_cond_broadcast__set__cond(smx_simcall_t simcall, smx_
 static inline unsigned int simcall_sem_init__get__capacity(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<unsigned int>(simcall->args[0]);
 }
+static inline unsigned int simcall_sem_init__getraw__capacity(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<unsigned int>(simcall->args[0]);
+}
 static inline void simcall_sem_init__set__capacity(smx_simcall_t simcall, unsigned int arg) {
     simgrid::simix::marshal<unsigned int>(simcall->args[0], arg);
 }
 static inline smx_sem_t simcall_sem_init__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<smx_sem_t>(simcall->result);
 }
+static inline smx_sem_t simcall_sem_init__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_sem_t>(simcall->result);
+}
 static inline void simcall_sem_init__set__result(smx_simcall_t simcall, smx_sem_t result){
     simgrid::simix::marshal<smx_sem_t>(simcall->result, result);
 }
@@ -739,6 +1254,10 @@ static inline void simcall_sem_init__set__result(smx_simcall_t simcall, smx_sem_
 static inline smx_sem_t simcall_sem_release__get__sem(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_sem_t>(simcall->args[0]);
 }
+static inline smx_sem_t simcall_sem_release__getraw__sem(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_sem_t>(simcall->args[0]);
+}
 static inline void simcall_sem_release__set__sem(smx_simcall_t simcall, smx_sem_t arg) {
     simgrid::simix::marshal<smx_sem_t>(simcall->args[0], arg);
 }
@@ -746,12 +1265,20 @@ static inline void simcall_sem_release__set__sem(smx_simcall_t simcall, smx_sem_
 static inline smx_sem_t simcall_sem_would_block__get__sem(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_sem_t>(simcall->args[0]);
 }
+static inline smx_sem_t simcall_sem_would_block__getraw__sem(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_sem_t>(simcall->args[0]);
+}
 static inline void simcall_sem_would_block__set__sem(smx_simcall_t simcall, smx_sem_t arg) {
     simgrid::simix::marshal<smx_sem_t>(simcall->args[0], arg);
 }
 static inline int simcall_sem_would_block__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_sem_would_block__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_sem_would_block__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -759,6 +1286,10 @@ static inline void simcall_sem_would_block__set__result(smx_simcall_t simcall, i
 static inline smx_sem_t simcall_sem_acquire__get__sem(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_sem_t>(simcall->args[0]);
 }
+static inline smx_sem_t simcall_sem_acquire__getraw__sem(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_sem_t>(simcall->args[0]);
+}
 static inline void simcall_sem_acquire__set__sem(smx_simcall_t simcall, smx_sem_t arg) {
     simgrid::simix::marshal<smx_sem_t>(simcall->args[0], arg);
 }
@@ -766,12 +1297,20 @@ static inline void simcall_sem_acquire__set__sem(smx_simcall_t simcall, smx_sem_
 static inline smx_sem_t simcall_sem_acquire_timeout__get__sem(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_sem_t>(simcall->args[0]);
 }
+static inline smx_sem_t simcall_sem_acquire_timeout__getraw__sem(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_sem_t>(simcall->args[0]);
+}
 static inline void simcall_sem_acquire_timeout__set__sem(smx_simcall_t simcall, smx_sem_t arg) {
     simgrid::simix::marshal<smx_sem_t>(simcall->args[0], arg);
 }
 static inline double simcall_sem_acquire_timeout__get__timeout(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<double>(simcall->args[1]);
 }
+static inline double simcall_sem_acquire_timeout__getraw__timeout(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<double>(simcall->args[1]);
+}
 static inline void simcall_sem_acquire_timeout__set__timeout(smx_simcall_t simcall, double arg) {
     simgrid::simix::marshal<double>(simcall->args[1], arg);
 }
@@ -779,12 +1318,20 @@ static inline void simcall_sem_acquire_timeout__set__timeout(smx_simcall_t simca
 static inline smx_sem_t simcall_sem_get_capacity__get__sem(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_sem_t>(simcall->args[0]);
 }
+static inline smx_sem_t simcall_sem_get_capacity__getraw__sem(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_sem_t>(simcall->args[0]);
+}
 static inline void simcall_sem_get_capacity__set__sem(smx_simcall_t simcall, smx_sem_t arg) {
     simgrid::simix::marshal<smx_sem_t>(simcall->args[0], arg);
 }
 static inline int simcall_sem_get_capacity__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_sem_get_capacity__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_sem_get_capacity__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -792,24 +1339,40 @@ static inline void simcall_sem_get_capacity__set__result(smx_simcall_t simcall,
 static inline smx_file_t simcall_file_read__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_read__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_read__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline sg_size_t simcall_file_read__get__size(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_size_t>(simcall->args[1]);
 }
+static inline sg_size_t simcall_file_read__getraw__size(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_size_t>(simcall->args[1]);
+}
 static inline void simcall_file_read__set__size(smx_simcall_t simcall, sg_size_t arg) {
     simgrid::simix::marshal<sg_size_t>(simcall->args[1], arg);
 }
 static inline sg_host_t simcall_file_read__get__host(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_host_t>(simcall->args[2]);
 }
+static inline sg_host_t simcall_file_read__getraw__host(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_host_t>(simcall->args[2]);
+}
 static inline void simcall_file_read__set__host(smx_simcall_t simcall, sg_host_t arg) {
     simgrid::simix::marshal<sg_host_t>(simcall->args[2], arg);
 }
 static inline sg_size_t simcall_file_read__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<sg_size_t>(simcall->result);
 }
+static inline sg_size_t simcall_file_read__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_size_t>(simcall->result);
+}
 static inline void simcall_file_read__set__result(smx_simcall_t simcall, sg_size_t result){
     simgrid::simix::marshal<sg_size_t>(simcall->result, result);
 }
@@ -817,24 +1380,40 @@ static inline void simcall_file_read__set__result(smx_simcall_t simcall, sg_size
 static inline smx_file_t simcall_file_write__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_write__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_write__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline sg_size_t simcall_file_write__get__size(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_size_t>(simcall->args[1]);
 }
+static inline sg_size_t simcall_file_write__getraw__size(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_size_t>(simcall->args[1]);
+}
 static inline void simcall_file_write__set__size(smx_simcall_t simcall, sg_size_t arg) {
     simgrid::simix::marshal<sg_size_t>(simcall->args[1], arg);
 }
 static inline sg_host_t simcall_file_write__get__host(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_host_t>(simcall->args[2]);
 }
+static inline sg_host_t simcall_file_write__getraw__host(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_host_t>(simcall->args[2]);
+}
 static inline void simcall_file_write__set__host(smx_simcall_t simcall, sg_host_t arg) {
     simgrid::simix::marshal<sg_host_t>(simcall->args[2], arg);
 }
 static inline sg_size_t simcall_file_write__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<sg_size_t>(simcall->result);
 }
+static inline sg_size_t simcall_file_write__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_size_t>(simcall->result);
+}
 static inline void simcall_file_write__set__result(smx_simcall_t simcall, sg_size_t result){
     simgrid::simix::marshal<sg_size_t>(simcall->result, result);
 }
@@ -842,18 +1421,30 @@ static inline void simcall_file_write__set__result(smx_simcall_t simcall, sg_siz
 static inline const char* simcall_file_open__get__fullpath(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<const char*>(simcall->args[0]);
 }
+static inline const char* simcall_file_open__getraw__fullpath(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<const char*>(simcall->args[0]);
+}
 static inline void simcall_file_open__set__fullpath(smx_simcall_t simcall, const char* arg) {
     simgrid::simix::marshal<const char*>(simcall->args[0], arg);
 }
 static inline sg_host_t simcall_file_open__get__host(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_host_t>(simcall->args[1]);
 }
+static inline sg_host_t simcall_file_open__getraw__host(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_host_t>(simcall->args[1]);
+}
 static inline void simcall_file_open__set__host(smx_simcall_t simcall, sg_host_t arg) {
     simgrid::simix::marshal<sg_host_t>(simcall->args[1], arg);
 }
 static inline smx_file_t simcall_file_open__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<smx_file_t>(simcall->result);
 }
+static inline smx_file_t simcall_file_open__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->result);
+}
 static inline void simcall_file_open__set__result(smx_simcall_t simcall, smx_file_t result){
     simgrid::simix::marshal<smx_file_t>(simcall->result, result);
 }
@@ -861,18 +1452,30 @@ static inline void simcall_file_open__set__result(smx_simcall_t simcall, smx_fil
 static inline smx_file_t simcall_file_close__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_close__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_close__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline sg_host_t simcall_file_close__get__host(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_host_t>(simcall->args[1]);
 }
+static inline sg_host_t simcall_file_close__getraw__host(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_host_t>(simcall->args[1]);
+}
 static inline void simcall_file_close__set__host(smx_simcall_t simcall, sg_host_t arg) {
     simgrid::simix::marshal<sg_host_t>(simcall->args[1], arg);
 }
 static inline int simcall_file_close__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_file_close__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_file_close__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -880,18 +1483,30 @@ static inline void simcall_file_close__set__result(smx_simcall_t simcall, int re
 static inline smx_file_t simcall_file_unlink__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_unlink__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_unlink__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline sg_host_t simcall_file_unlink__get__host(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_host_t>(simcall->args[1]);
 }
+static inline sg_host_t simcall_file_unlink__getraw__host(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_host_t>(simcall->args[1]);
+}
 static inline void simcall_file_unlink__set__host(smx_simcall_t simcall, sg_host_t arg) {
     simgrid::simix::marshal<sg_host_t>(simcall->args[1], arg);
 }
 static inline int simcall_file_unlink__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_file_unlink__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_file_unlink__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -899,12 +1514,20 @@ static inline void simcall_file_unlink__set__result(smx_simcall_t simcall, int r
 static inline smx_file_t simcall_file_get_size__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_get_size__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_get_size__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline sg_size_t simcall_file_get_size__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<sg_size_t>(simcall->result);
 }
+static inline sg_size_t simcall_file_get_size__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_size_t>(simcall->result);
+}
 static inline void simcall_file_get_size__set__result(smx_simcall_t simcall, sg_size_t result){
     simgrid::simix::marshal<sg_size_t>(simcall->result, result);
 }
@@ -912,12 +1535,20 @@ static inline void simcall_file_get_size__set__result(smx_simcall_t simcall, sg_
 static inline smx_file_t simcall_file_tell__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_tell__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_tell__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline sg_size_t simcall_file_tell__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<sg_size_t>(simcall->result);
 }
+static inline sg_size_t simcall_file_tell__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_size_t>(simcall->result);
+}
 static inline void simcall_file_tell__set__result(smx_simcall_t simcall, sg_size_t result){
     simgrid::simix::marshal<sg_size_t>(simcall->result, result);
 }
@@ -925,24 +1556,40 @@ static inline void simcall_file_tell__set__result(smx_simcall_t simcall, sg_size
 static inline smx_file_t simcall_file_seek__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_seek__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_seek__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline sg_offset_t simcall_file_seek__get__offset(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<sg_offset_t>(simcall->args[1]);
 }
+static inline sg_offset_t simcall_file_seek__getraw__offset(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<sg_offset_t>(simcall->args[1]);
+}
 static inline void simcall_file_seek__set__offset(smx_simcall_t simcall, sg_offset_t arg) {
     simgrid::simix::marshal<sg_offset_t>(simcall->args[1], arg);
 }
 static inline int simcall_file_seek__get__origin(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[2]);
 }
+static inline int simcall_file_seek__getraw__origin(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[2]);
+}
 static inline void simcall_file_seek__set__origin(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[2], arg);
 }
 static inline int simcall_file_seek__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_file_seek__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_file_seek__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
@@ -950,12 +1597,20 @@ static inline void simcall_file_seek__set__result(smx_simcall_t simcall, int res
 static inline smx_file_t simcall_file_get_info__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_get_info__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_get_info__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline xbt_dynar_t simcall_file_get_info__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<xbt_dynar_t>(simcall->result);
 }
+static inline xbt_dynar_t simcall_file_get_info__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<xbt_dynar_t>(simcall->result);
+}
 static inline void simcall_file_get_info__set__result(smx_simcall_t simcall, xbt_dynar_t result){
     simgrid::simix::marshal<xbt_dynar_t>(simcall->result, result);
 }
@@ -963,63 +1618,86 @@ static inline void simcall_file_get_info__set__result(smx_simcall_t simcall, xbt
 static inline smx_file_t simcall_file_move__get__fd(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<smx_file_t>(simcall->args[0]);
 }
+static inline smx_file_t simcall_file_move__getraw__fd(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<smx_file_t>(simcall->args[0]);
+}
 static inline void simcall_file_move__set__fd(smx_simcall_t simcall, smx_file_t arg) {
     simgrid::simix::marshal<smx_file_t>(simcall->args[0], arg);
 }
 static inline const char* simcall_file_move__get__fullpath(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<const char*>(simcall->args[1]);
 }
+static inline const char* simcall_file_move__getraw__fullpath(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<const char*>(simcall->args[1]);
+}
 static inline void simcall_file_move__set__fullpath(smx_simcall_t simcall, const char* arg) {
     simgrid::simix::marshal<const char*>(simcall->args[1], arg);
 }
 static inline int simcall_file_move__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_file_move__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_file_move__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
 
-static inline smx_storage_t simcall_storage_get_properties__get__storage(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_storage_t>(simcall->args[0]);
-}
-static inline void simcall_storage_get_properties__set__storage(smx_simcall_t simcall, smx_storage_t arg) {
-    simgrid::simix::marshal<smx_storage_t>(simcall->args[0], arg);
-}
-static inline xbt_dict_t simcall_storage_get_properties__get__result(smx_simcall_t simcall){
-    return simgrid::simix::unmarshal<xbt_dict_t>(simcall->result);
-}
-static inline void simcall_storage_get_properties__set__result(smx_simcall_t simcall, xbt_dict_t result){
-    simgrid::simix::marshal<xbt_dict_t>(simcall->result, result);
-}
-
 static inline int simcall_mc_random__get__min(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[0]);
 }
+static inline int simcall_mc_random__getraw__min(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[0]);
+}
 static inline void simcall_mc_random__set__min(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[0], arg);
 }
 static inline int simcall_mc_random__get__max(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<int>(simcall->args[1]);
 }
+static inline int simcall_mc_random__getraw__max(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->args[1]);
+}
 static inline void simcall_mc_random__set__max(smx_simcall_t simcall, int arg) {
     simgrid::simix::marshal<int>(simcall->args[1], arg);
 }
 static inline int simcall_mc_random__get__result(smx_simcall_t simcall){
     return simgrid::simix::unmarshal<int>(simcall->result);
 }
+static inline int simcall_mc_random__getraw__result(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<int>(simcall->result);
+}
 static inline void simcall_mc_random__set__result(smx_simcall_t simcall, int result){
     simgrid::simix::marshal<int>(simcall->result, result);
 }
 
-static inline smx_activity_t simcall_set_category__get__synchro(smx_simcall_t simcall) {
-  return simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]);
+static inline boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_set_category__get__synchro(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]);
+}
+static inline simgrid::kernel::activity::ActivityImpl* simcall_set_category__getraw__synchro(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<simgrid::kernel::activity::ActivityImpl*>(simcall->args[0]);
 }
-static inline void simcall_set_category__set__synchro(smx_simcall_t simcall, smx_activity_t arg) {
-    simgrid::simix::marshal<smx_activity_t>(simcall->args[0], arg);
+static inline void simcall_set_category__set__synchro(smx_simcall_t simcall,
+                                                      boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> arg)
+{
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0], arg);
 }
 static inline const char* simcall_set_category__get__category(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<const char*>(simcall->args[1]);
 }
+static inline const char* simcall_set_category__getraw__category(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<const char*>(simcall->args[1]);
+}
 static inline void simcall_set_category__set__category(smx_simcall_t simcall, const char* arg) {
     simgrid::simix::marshal<const char*>(simcall->args[1], arg);
 }
@@ -1027,6 +1705,10 @@ static inline void simcall_set_category__set__category(smx_simcall_t simcall, co
 static inline std::function<void()> const* simcall_run_kernel__get__code(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<std::function<void()> const*>(simcall->args[0]);
 }
+static inline std::function<void()> const* simcall_run_kernel__getraw__code(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<std::function<void()> const*>(simcall->args[0]);
+}
 static inline void simcall_run_kernel__set__code(smx_simcall_t simcall, std::function<void()> const* arg) {
     simgrid::simix::marshal<std::function<void()> const*>(simcall->args[0], arg);
 }
@@ -1034,6 +1716,10 @@ static inline void simcall_run_kernel__set__code(smx_simcall_t simcall, std::fun
 static inline std::function<void()> const* simcall_run_blocking__get__code(smx_simcall_t simcall) {
   return simgrid::simix::unmarshal<std::function<void()> const*>(simcall->args[0]);
 }
+static inline std::function<void()> const* simcall_run_blocking__getraw__code(smx_simcall_t simcall)
+{
+  return simgrid::simix::unmarshal_raw<std::function<void()> const*>(simcall->args[0]);
+}
 static inline void simcall_run_blocking__set__code(smx_simcall_t simcall, std::function<void()> const* arg) {
     simgrid::simix::marshal<std::function<void()> const*>(simcall->args[0], arg);
 }
@@ -1046,18 +1732,36 @@ XBT_PRIVATE void simcall_HANDLER_process_suspend(smx_simcall_t simcall, smx_acto
 XBT_PRIVATE void simcall_HANDLER_process_set_host(smx_simcall_t simcall, smx_actor_t process, sg_host_t dest);
 XBT_PRIVATE void simcall_HANDLER_process_join(smx_simcall_t simcall, smx_actor_t process, double timeout);
 XBT_PRIVATE void simcall_HANDLER_process_sleep(smx_simcall_t simcall, double duration);
-XBT_PRIVATE smx_activity_t simcall_HANDLER_execution_start(smx_simcall_t simcall, const char* name, double flops_amount, double priority, double bound);
-XBT_PRIVATE void simcall_HANDLER_execution_wait(smx_simcall_t simcall, smx_activity_t execution);
+XBT_PRIVATE boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_HANDLER_execution_start(smx_simcall_t simcall, const char* name, double flops_amount, double priority,
+                                double bound);
+XBT_PRIVATE void
+simcall_HANDLER_execution_wait(smx_simcall_t simcall,
+                               boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution);
 XBT_PRIVATE smx_actor_t simcall_HANDLER_process_restart(smx_simcall_t simcall, smx_actor_t process);
-XBT_PRIVATE smx_activity_t simcall_HANDLER_comm_iprobe(smx_simcall_t simcall, smx_mailbox_t mbox, int type, int src, int tag, simix_match_func_t match_fun, void* data);
+XBT_PRIVATE boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_HANDLER_comm_iprobe(smx_simcall_t simcall, smx_mailbox_t mbox, int type, int src, int tag,
+                            simix_match_func_t match_fun, void* data);
 XBT_PRIVATE void simcall_HANDLER_comm_send(smx_simcall_t simcall, smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate, void* src_buff, size_t src_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double timeout);
-XBT_PRIVATE smx_activity_t simcall_HANDLER_comm_isend(smx_simcall_t simcall, smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate, void* src_buff, size_t src_buff_size, simix_match_func_t match_fun, simix_clean_func_t clean_fun, simix_copy_data_func_t copy_data_fun, void* data, int detached);
+XBT_PRIVATE boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_HANDLER_comm_isend(smx_simcall_t simcall, smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate,
+                           void* src_buff, size_t src_buff_size, simix_match_func_t match_fun,
+                           simix_clean_func_t clean_fun, simix_copy_data_func_t copy_data_fun, void* data,
+                           int detached);
 XBT_PRIVATE void simcall_HANDLER_comm_recv(smx_simcall_t simcall, smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff, size_t* dst_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double timeout, double rate);
-XBT_PRIVATE smx_activity_t simcall_HANDLER_comm_irecv(smx_simcall_t simcall, smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff, size_t* dst_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double rate);
+XBT_PRIVATE boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+simcall_HANDLER_comm_irecv(smx_simcall_t simcall, smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff,
+                           size_t* dst_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun,
+                           void* data, double rate);
 XBT_PRIVATE void simcall_HANDLER_comm_waitany(smx_simcall_t simcall, xbt_dynar_t comms, double timeout);
-XBT_PRIVATE void simcall_HANDLER_comm_wait(smx_simcall_t simcall, smx_activity_t comm, double timeout);
-XBT_PRIVATE void simcall_HANDLER_comm_test(smx_simcall_t simcall, smx_activity_t comm);
-XBT_PRIVATE void simcall_HANDLER_comm_testany(smx_simcall_t simcall, smx_activity_t* comms, size_t count);
+XBT_PRIVATE void simcall_HANDLER_comm_wait(smx_simcall_t simcall,
+                                           boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm,
+                                           double timeout);
+XBT_PRIVATE void simcall_HANDLER_comm_test(smx_simcall_t simcall,
+                                           boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm);
+XBT_PRIVATE void simcall_HANDLER_comm_testany(smx_simcall_t simcall,
+                                              boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>* comms,
+                                              size_t count);
 XBT_PRIVATE smx_mutex_t simcall_HANDLER_mutex_init(smx_simcall_t simcall);
 XBT_PRIVATE void simcall_HANDLER_mutex_lock(smx_simcall_t simcall, smx_mutex_t mutex);
 XBT_PRIVATE int simcall_HANDLER_mutex_trylock(smx_simcall_t simcall, smx_mutex_t mutex);
diff --git a/src/simix/popping_bodies.cpp b/src/simix/popping_bodies.cpp
index 758d8748d7..cafc761568 100644
--- a/src/simix/popping_bodies.cpp
+++ b/src/simix/popping_bodies.cpp
@@ -35,343 +35,383 @@ inline static R simcall(e_smx_simcall_t call, T const&... t)
   }
   return simgrid::simix::unmarshal<R>(self->simcall.result);
 }
-  
+
 inline static void simcall_BODY_process_kill(smx_actor_t process) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_process_kill(&SIMIX_process_self()->simcall, process);
     return simcall<void, smx_actor_t>(SIMCALL_PROCESS_KILL, process);
   }
-  
+
 inline static void simcall_BODY_process_killall(int reset_pid) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_process_killall(&SIMIX_process_self()->simcall, reset_pid);
     return simcall<void, int>(SIMCALL_PROCESS_KILLALL, reset_pid);
   }
-  
+
 inline static void simcall_BODY_process_cleanup(smx_actor_t process) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_process_cleanup(process);
     return simcall<void, smx_actor_t>(SIMCALL_PROCESS_CLEANUP, process);
   }
-  
+
 inline static void simcall_BODY_process_suspend(smx_actor_t process) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_process_suspend(&SIMIX_process_self()->simcall, process);
     return simcall<void, smx_actor_t>(SIMCALL_PROCESS_SUSPEND, process);
   }
-  
+
 inline static void simcall_BODY_process_resume(smx_actor_t process) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_process_resume(process);
     return simcall<void, smx_actor_t>(SIMCALL_PROCESS_RESUME, process);
   }
-  
+
 inline static void simcall_BODY_process_set_host(smx_actor_t process, sg_host_t dest) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_process_set_host(&SIMIX_process_self()->simcall, process, dest);
     return simcall<void, smx_actor_t, sg_host_t>(SIMCALL_PROCESS_SET_HOST, process, dest);
   }
-  
+
 inline static int simcall_BODY_process_is_suspended(smx_actor_t process) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_process_is_suspended(process);
     return simcall<int, smx_actor_t>(SIMCALL_PROCESS_IS_SUSPENDED, process);
   }
-  
+
 inline static int simcall_BODY_process_join(smx_actor_t process, double timeout) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_process_join(&SIMIX_process_self()->simcall, process, timeout);
     return simcall<int, smx_actor_t, double>(SIMCALL_PROCESS_JOIN, process, timeout);
   }
-  
+
 inline static int simcall_BODY_process_sleep(double duration) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_process_sleep(&SIMIX_process_self()->simcall, duration);
     return simcall<int, double>(SIMCALL_PROCESS_SLEEP, duration);
   }
-  
-inline static smx_activity_t simcall_BODY_execution_start(const char* name, double flops_amount, double priority, double bound) {
+
+  inline static boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+  simcall_BODY_execution_start(const char* name, double flops_amount, double priority, double bound)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_execution_start(&SIMIX_process_self()->simcall, name, flops_amount, priority, bound);
-    return simcall<smx_activity_t, const char*, double, double, double>(SIMCALL_EXECUTION_START, name, flops_amount, priority, bound);
+    return simcall<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, const char*, double, double, double>(
+        SIMCALL_EXECUTION_START, name, flops_amount, priority, bound);
   }
-  
-inline static smx_activity_t simcall_BODY_execution_parallel_start(const char* name, int host_nb, sg_host_t* host_list, double* flops_amount, double* bytes_amount, double amount, double rate, double timeout) {
+
+  inline static boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+  simcall_BODY_execution_parallel_start(const char* name, int host_nb, sg_host_t* host_list, double* flops_amount,
+                                        double* bytes_amount, double amount, double rate, double timeout)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_execution_parallel_start(name, host_nb, host_list, flops_amount, bytes_amount, amount, rate, timeout);
-    return simcall<smx_activity_t, const char*, int, sg_host_t*, double*, double*, double, double, double>(SIMCALL_EXECUTION_PARALLEL_START, name, host_nb, host_list, flops_amount, bytes_amount, amount, rate, timeout);
+    return simcall<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, const char*, int, sg_host_t*, double*,
+                   double*, double, double, double>(SIMCALL_EXECUTION_PARALLEL_START, name, host_nb, host_list,
+                                                    flops_amount, bytes_amount, amount, rate, timeout);
   }
-  
-inline static void simcall_BODY_execution_cancel(smx_activity_t execution) {
+
+  inline static void
+  simcall_BODY_execution_cancel(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_execution_cancel(execution);
-    return simcall<void, smx_activity_t>(SIMCALL_EXECUTION_CANCEL, execution);
+    return simcall<void, boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(SIMCALL_EXECUTION_CANCEL,
+                                                                                        execution);
   }
-  
-inline static void simcall_BODY_execution_set_priority(smx_activity_t execution, double priority) {
+
+  inline static void
+  simcall_BODY_execution_set_priority(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution,
+                                      double priority)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_execution_set_priority(execution, priority);
-    return simcall<void, smx_activity_t, double>(SIMCALL_EXECUTION_SET_PRIORITY, execution, priority);
+    return simcall<void, boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, double>(
+        SIMCALL_EXECUTION_SET_PRIORITY, execution, priority);
   }
-  
-inline static void simcall_BODY_execution_set_bound(smx_activity_t execution, double bound) {
+
+  inline static void
+  simcall_BODY_execution_set_bound(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution,
+                                   double bound)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_execution_set_bound(execution, bound);
-    return simcall<void, smx_activity_t, double>(SIMCALL_EXECUTION_SET_BOUND, execution, bound);
+    return simcall<void, boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, double>(
+        SIMCALL_EXECUTION_SET_BOUND, execution, bound);
   }
-  
-inline static int simcall_BODY_execution_wait(smx_activity_t execution) {
+
+  inline static int simcall_BODY_execution_wait(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_execution_wait(&SIMIX_process_self()->simcall, execution);
-    return simcall<int, smx_activity_t>(SIMCALL_EXECUTION_WAIT, execution);
+    return simcall<int, boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(SIMCALL_EXECUTION_WAIT,
+                                                                                       execution);
   }
-  
+
 inline static void simcall_BODY_process_on_exit(smx_actor_t process, int_f_pvoid_pvoid_t fun, void* data) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_process_on_exit(process, fun, data);
     return simcall<void, smx_actor_t, int_f_pvoid_pvoid_t, void*>(SIMCALL_PROCESS_ON_EXIT, process, fun, data);
   }
-  
+
 inline static void simcall_BODY_process_auto_restart_set(smx_actor_t process, int auto_restart) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_process_auto_restart_set(process, auto_restart);
     return simcall<void, smx_actor_t, int>(SIMCALL_PROCESS_AUTO_RESTART_SET, process, auto_restart);
   }
-  
+
 inline static smx_actor_t simcall_BODY_process_restart(smx_actor_t process) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_process_restart(&SIMIX_process_self()->simcall, process);
     return simcall<smx_actor_t, smx_actor_t>(SIMCALL_PROCESS_RESTART, process);
   }
-  
-inline static smx_activity_t simcall_BODY_comm_iprobe(smx_mailbox_t mbox, int type, int src, int tag, simix_match_func_t match_fun, void* data) {
+
+  inline static boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+  simcall_BODY_comm_iprobe(smx_mailbox_t mbox, int type, int src, int tag, simix_match_func_t match_fun, void* data)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_iprobe(&SIMIX_process_self()->simcall, mbox, type, src, tag, match_fun, data);
-    return simcall<smx_activity_t, smx_mailbox_t, int, int, int, simix_match_func_t, void*>(SIMCALL_COMM_IPROBE, mbox, type, src, tag, match_fun, data);
+    return simcall<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, smx_mailbox_t, int, int, int,
+                   simix_match_func_t, void*>(SIMCALL_COMM_IPROBE, mbox, type, src, tag, match_fun, data);
   }
-  
+
 inline static void simcall_BODY_comm_send(smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate, void* src_buff, size_t src_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double timeout) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_send(&SIMIX_process_self()->simcall, sender, mbox, task_size, rate, src_buff, src_buff_size, match_fun, copy_data_fun, data, timeout);
     return simcall<void, smx_actor_t, smx_mailbox_t, double, double, void*, size_t, simix_match_func_t, simix_copy_data_func_t, void*, double>(SIMCALL_COMM_SEND, sender, mbox, task_size, rate, src_buff, src_buff_size, match_fun, copy_data_fun, data, timeout);
   }
-  
-inline static smx_activity_t simcall_BODY_comm_isend(smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate, void* src_buff, size_t src_buff_size, simix_match_func_t match_fun, simix_clean_func_t clean_fun, simix_copy_data_func_t copy_data_fun, void* data, int detached) {
+
+  inline static boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+  simcall_BODY_comm_isend(smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate, void* src_buff,
+                          size_t src_buff_size, simix_match_func_t match_fun, simix_clean_func_t clean_fun,
+                          simix_copy_data_func_t copy_data_fun, void* data, int detached)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_isend(&SIMIX_process_self()->simcall, sender, mbox, task_size, rate, src_buff, src_buff_size, match_fun, clean_fun, copy_data_fun, data, detached);
-    return simcall<smx_activity_t, smx_actor_t, smx_mailbox_t, double, double, void*, size_t, simix_match_func_t, simix_clean_func_t, simix_copy_data_func_t, void*, int>(SIMCALL_COMM_ISEND, sender, mbox, task_size, rate, src_buff, src_buff_size, match_fun, clean_fun, copy_data_fun, data, detached);
+    return simcall<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, smx_actor_t, smx_mailbox_t, double,
+                   double, void*, size_t, simix_match_func_t, simix_clean_func_t, simix_copy_data_func_t, void*, int>(
+        SIMCALL_COMM_ISEND, sender, mbox, task_size, rate, src_buff, src_buff_size, match_fun, clean_fun, copy_data_fun,
+        data, detached);
   }
-  
+
 inline static void simcall_BODY_comm_recv(smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff, size_t* dst_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double timeout, double rate) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_recv(&SIMIX_process_self()->simcall, receiver, mbox, dst_buff, dst_buff_size, match_fun, copy_data_fun, data, timeout, rate);
     return simcall<void, smx_actor_t, smx_mailbox_t, void*, size_t*, simix_match_func_t, simix_copy_data_func_t, void*, double, double>(SIMCALL_COMM_RECV, receiver, mbox, dst_buff, dst_buff_size, match_fun, copy_data_fun, data, timeout, rate);
   }
-  
-inline static smx_activity_t simcall_BODY_comm_irecv(smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff, size_t* dst_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double rate) {
+
+  inline static boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>
+  simcall_BODY_comm_irecv(smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff, size_t* dst_buff_size,
+                          simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double rate)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_irecv(&SIMIX_process_self()->simcall, receiver, mbox, dst_buff, dst_buff_size, match_fun, copy_data_fun, data, rate);
-    return simcall<smx_activity_t, smx_actor_t, smx_mailbox_t, void*, size_t*, simix_match_func_t, simix_copy_data_func_t, void*, double>(SIMCALL_COMM_IRECV, receiver, mbox, dst_buff, dst_buff_size, match_fun, copy_data_fun, data, rate);
+    return simcall<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, smx_actor_t, smx_mailbox_t, void*,
+                   size_t*, simix_match_func_t, simix_copy_data_func_t, void*, double>(
+        SIMCALL_COMM_IRECV, receiver, mbox, dst_buff, dst_buff_size, match_fun, copy_data_fun, data, rate);
   }
-  
+
 inline static int simcall_BODY_comm_waitany(xbt_dynar_t comms, double timeout) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_waitany(&SIMIX_process_self()->simcall, comms, timeout);
     return simcall<int, xbt_dynar_t, double>(SIMCALL_COMM_WAITANY, comms, timeout);
   }
-  
-inline static void simcall_BODY_comm_wait(smx_activity_t comm, double timeout) {
+
+  inline static void simcall_BODY_comm_wait(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm,
+                                            double timeout)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_wait(&SIMIX_process_self()->simcall, comm, timeout);
-    return simcall<void, smx_activity_t, double>(SIMCALL_COMM_WAIT, comm, timeout);
+    return simcall<void, boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, double>(SIMCALL_COMM_WAIT, comm,
+                                                                                                timeout);
   }
-  
-inline static int simcall_BODY_comm_test(smx_activity_t comm) {
+
+  inline static int simcall_BODY_comm_test(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_test(&SIMIX_process_self()->simcall, comm);
-    return simcall<int, smx_activity_t>(SIMCALL_COMM_TEST, comm);
+    return simcall<int, boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(SIMCALL_COMM_TEST, comm);
   }
-  
-inline static int simcall_BODY_comm_testany(smx_activity_t* comms, size_t count) {
+
+  inline static int simcall_BODY_comm_testany(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>* comms,
+                                              size_t count)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_comm_testany(&SIMIX_process_self()->simcall, comms, count);
-    return simcall<int, smx_activity_t*, size_t>(SIMCALL_COMM_TESTANY, comms, count);
+    return simcall<int, boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>*, size_t>(SIMCALL_COMM_TESTANY,
+                                                                                                comms, count);
   }
-  
+
 inline static smx_mutex_t simcall_BODY_mutex_init() {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_mutex_init(&SIMIX_process_self()->simcall);
     return simcall<smx_mutex_t>(SIMCALL_MUTEX_INIT);
   }
-  
+
 inline static void simcall_BODY_mutex_lock(smx_mutex_t mutex) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_mutex_lock(&SIMIX_process_self()->simcall, mutex);
     return simcall<void, smx_mutex_t>(SIMCALL_MUTEX_LOCK, mutex);
   }
-  
+
 inline static int simcall_BODY_mutex_trylock(smx_mutex_t mutex) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_mutex_trylock(&SIMIX_process_self()->simcall, mutex);
     return simcall<int, smx_mutex_t>(SIMCALL_MUTEX_TRYLOCK, mutex);
   }
-  
+
 inline static void simcall_BODY_mutex_unlock(smx_mutex_t mutex) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_mutex_unlock(&SIMIX_process_self()->simcall, mutex);
     return simcall<void, smx_mutex_t>(SIMCALL_MUTEX_UNLOCK, mutex);
   }
-  
+
 inline static smx_cond_t simcall_BODY_cond_init() {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_cond_init();
     return simcall<smx_cond_t>(SIMCALL_COND_INIT);
   }
-  
+
 inline static void simcall_BODY_cond_signal(smx_cond_t cond) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_cond_signal(cond);
     return simcall<void, smx_cond_t>(SIMCALL_COND_SIGNAL, cond);
   }
-  
+
 inline static void simcall_BODY_cond_wait(smx_cond_t cond, smx_mutex_t mutex) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_cond_wait(&SIMIX_process_self()->simcall, cond, mutex);
     return simcall<void, smx_cond_t, smx_mutex_t>(SIMCALL_COND_WAIT, cond, mutex);
   }
-  
+
 inline static void simcall_BODY_cond_wait_timeout(smx_cond_t cond, smx_mutex_t mutex, double timeout) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_cond_wait_timeout(&SIMIX_process_self()->simcall, cond, mutex, timeout);
     return simcall<void, smx_cond_t, smx_mutex_t, double>(SIMCALL_COND_WAIT_TIMEOUT, cond, mutex, timeout);
   }
-  
+
 inline static void simcall_BODY_cond_broadcast(smx_cond_t cond) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_cond_broadcast(cond);
     return simcall<void, smx_cond_t>(SIMCALL_COND_BROADCAST, cond);
   }
-  
+
 inline static smx_sem_t simcall_BODY_sem_init(unsigned int capacity) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_sem_init(capacity);
     return simcall<smx_sem_t, unsigned int>(SIMCALL_SEM_INIT, capacity);
   }
-  
+
 inline static void simcall_BODY_sem_release(smx_sem_t sem) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_sem_release(&SIMIX_process_self()->simcall, sem);
     return simcall<void, smx_sem_t>(SIMCALL_SEM_RELEASE, sem);
   }
-  
+
 inline static int simcall_BODY_sem_would_block(smx_sem_t sem) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_sem_would_block(&SIMIX_process_self()->simcall, sem);
     return simcall<int, smx_sem_t>(SIMCALL_SEM_WOULD_BLOCK, sem);
   }
-  
+
 inline static void simcall_BODY_sem_acquire(smx_sem_t sem) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_sem_acquire(&SIMIX_process_self()->simcall, sem);
     return simcall<void, smx_sem_t>(SIMCALL_SEM_ACQUIRE, sem);
   }
-  
+
 inline static void simcall_BODY_sem_acquire_timeout(smx_sem_t sem, double timeout) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_sem_acquire_timeout(&SIMIX_process_self()->simcall, sem, timeout);
     return simcall<void, smx_sem_t, double>(SIMCALL_SEM_ACQUIRE_TIMEOUT, sem, timeout);
   }
-  
+
 inline static int simcall_BODY_sem_get_capacity(smx_sem_t sem) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_sem_get_capacity(&SIMIX_process_self()->simcall, sem);
     return simcall<int, smx_sem_t>(SIMCALL_SEM_GET_CAPACITY, sem);
   }
-  
+
 inline static sg_size_t simcall_BODY_file_read(smx_file_t fd, sg_size_t size, sg_host_t host) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_read(&SIMIX_process_self()->simcall, fd, size, host);
     return simcall<sg_size_t, smx_file_t, sg_size_t, sg_host_t>(SIMCALL_FILE_READ, fd, size, host);
   }
-  
+
 inline static sg_size_t simcall_BODY_file_write(smx_file_t fd, sg_size_t size, sg_host_t host) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_write(&SIMIX_process_self()->simcall, fd, size, host);
     return simcall<sg_size_t, smx_file_t, sg_size_t, sg_host_t>(SIMCALL_FILE_WRITE, fd, size, host);
   }
-  
+
 inline static smx_file_t simcall_BODY_file_open(const char* fullpath, sg_host_t host) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_open(&SIMIX_process_self()->simcall, fullpath, host);
     return simcall<smx_file_t, const char*, sg_host_t>(SIMCALL_FILE_OPEN, fullpath, host);
   }
-  
+
 inline static int simcall_BODY_file_close(smx_file_t fd, sg_host_t host) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_close(&SIMIX_process_self()->simcall, fd, host);
     return simcall<int, smx_file_t, sg_host_t>(SIMCALL_FILE_CLOSE, fd, host);
   }
-  
+
 inline static int simcall_BODY_file_unlink(smx_file_t fd, sg_host_t host) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_file_unlink(fd, host);
     return simcall<int, smx_file_t, sg_host_t>(SIMCALL_FILE_UNLINK, fd, host);
   }
-  
+
 inline static sg_size_t simcall_BODY_file_get_size(smx_file_t fd) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_get_size(&SIMIX_process_self()->simcall, fd);
     return simcall<sg_size_t, smx_file_t>(SIMCALL_FILE_GET_SIZE, fd);
   }
-  
+
 inline static sg_size_t simcall_BODY_file_tell(smx_file_t fd) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_tell(&SIMIX_process_self()->simcall, fd);
     return simcall<sg_size_t, smx_file_t>(SIMCALL_FILE_TELL, fd);
   }
-  
+
 inline static int simcall_BODY_file_seek(smx_file_t fd, sg_offset_t offset, int origin) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_seek(&SIMIX_process_self()->simcall, fd, offset, origin);
     return simcall<int, smx_file_t, sg_offset_t, int>(SIMCALL_FILE_SEEK, fd, offset, origin);
   }
-  
+
 inline static xbt_dynar_t simcall_BODY_file_get_info(smx_file_t fd) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_get_info(&SIMIX_process_self()->simcall, fd);
     return simcall<xbt_dynar_t, smx_file_t>(SIMCALL_FILE_GET_INFO, fd);
   }
-  
+
 inline static int simcall_BODY_file_move(smx_file_t fd, const char* fullpath) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_file_move(&SIMIX_process_self()->simcall, fd, fullpath);
     return simcall<int, smx_file_t, const char*>(SIMCALL_FILE_MOVE, fd, fullpath);
   }
-  
-inline static xbt_dict_t simcall_BODY_storage_get_properties(smx_storage_t storage) {
-    /* Go to that function to follow the code flow through the simcall barrier */
-    if (0) SIMIX_storage_get_properties(storage);
-    return simcall<xbt_dict_t, smx_storage_t>(SIMCALL_STORAGE_GET_PROPERTIES, storage);
-  }
-  
+
 inline static int simcall_BODY_mc_random(int min, int max) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) simcall_HANDLER_mc_random(&SIMIX_process_self()->simcall, min, max);
     return simcall<int, int, int>(SIMCALL_MC_RANDOM, min, max);
   }
-  
-inline static void simcall_BODY_set_category(smx_activity_t synchro, const char* category) {
+
+  inline static void simcall_BODY_set_category(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> synchro,
+                                               const char* category)
+  {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_set_category(synchro, category);
-    return simcall<void, smx_activity_t, const char*>(SIMCALL_SET_CATEGORY, synchro, category);
+    return simcall<void, boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>, const char*>(
+        SIMCALL_SET_CATEGORY, synchro, category);
   }
-  
+
 inline static void simcall_BODY_run_kernel(std::function<void()> const* code) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_run_kernel(code);
     return simcall<void, std::function<void()> const*>(SIMCALL_RUN_KERNEL, code);
   }
-  
+
 inline static void simcall_BODY_run_blocking(std::function<void()> const* code) {
     /* Go to that function to follow the code flow through the simcall barrier */
     if (0) SIMIX_run_blocking(code);
diff --git a/src/simix/popping_enum.h b/src/simix/popping_enum.h
index e533a242cd..4e59ccf9af 100644
--- a/src/simix/popping_enum.h
+++ b/src/simix/popping_enum.h
@@ -71,7 +71,6 @@ typedef enum {
   SIMCALL_FILE_SEEK,
   SIMCALL_FILE_GET_INFO,
   SIMCALL_FILE_MOVE,
-  SIMCALL_STORAGE_GET_PROPERTIES,
   SIMCALL_MC_RANDOM,
   SIMCALL_SET_CATEGORY,
   SIMCALL_RUN_KERNEL,
diff --git a/src/simix/popping_generated.cpp b/src/simix/popping_generated.cpp
index 45bd19d646..1a27a3acd1 100644
--- a/src/simix/popping_generated.cpp
+++ b/src/simix/popping_generated.cpp
@@ -77,7 +77,6 @@ const char* simcall_names[] = {
     "SIMCALL_FILE_SEEK",
     "SIMCALL_FILE_GET_INFO",
     "SIMCALL_FILE_MOVE",
-    "SIMCALL_STORAGE_GET_PROPERTIES",
     "SIMCALL_MC_RANDOM",
     "SIMCALL_SET_CATEGORY",
     "SIMCALL_RUN_KERNEL",
@@ -86,7 +85,7 @@ const char* simcall_names[] = {
 
 /** @private
  * @brief (in kernel mode) unpack the simcall and activate the handler
- * 
+ *
  * This function is generated from src/simix/simcalls.in
  */
 void SIMIX_simcall_handle(smx_simcall_t simcall, int value) {
@@ -138,33 +137,51 @@ case SIMCALL_PROCESS_SLEEP:
       break;
 
 case SIMCALL_EXECUTION_START:
-      simgrid::simix::marshal<smx_activity_t>(simcall->result, simcall_HANDLER_execution_start(simcall, simgrid::simix::unmarshal<const char*>(simcall->args[0]), simgrid::simix::unmarshal<double>(simcall->args[1]), simgrid::simix::unmarshal<double>(simcall->args[2]), simgrid::simix::unmarshal<double>(simcall->args[3])));
-      SIMIX_simcall_answer(simcall);
-      break;
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(
+      simcall->result,
+      simcall_HANDLER_execution_start(simcall, simgrid::simix::unmarshal<const char*>(simcall->args[0]),
+                                      simgrid::simix::unmarshal<double>(simcall->args[1]),
+                                      simgrid::simix::unmarshal<double>(simcall->args[2]),
+                                      simgrid::simix::unmarshal<double>(simcall->args[3])));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_EXECUTION_PARALLEL_START:
-      simgrid::simix::marshal<smx_activity_t>(simcall->result, SIMIX_execution_parallel_start(simgrid::simix::unmarshal<const char*>(simcall->args[0]), simgrid::simix::unmarshal<int>(simcall->args[1]), simgrid::simix::unmarshal<sg_host_t*>(simcall->args[2]), simgrid::simix::unmarshal<double*>(simcall->args[3]), simgrid::simix::unmarshal<double*>(simcall->args[4]), simgrid::simix::unmarshal<double>(simcall->args[5]), simgrid::simix::unmarshal<double>(simcall->args[6]), simgrid::simix::unmarshal<double>(simcall->args[7])));
-      SIMIX_simcall_answer(simcall);
-      break;
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(
+      simcall->result,
+      SIMIX_execution_parallel_start(
+          simgrid::simix::unmarshal<const char*>(simcall->args[0]), simgrid::simix::unmarshal<int>(simcall->args[1]),
+          simgrid::simix::unmarshal<sg_host_t*>(simcall->args[2]), simgrid::simix::unmarshal<double*>(simcall->args[3]),
+          simgrid::simix::unmarshal<double*>(simcall->args[4]), simgrid::simix::unmarshal<double>(simcall->args[5]),
+          simgrid::simix::unmarshal<double>(simcall->args[6]), simgrid::simix::unmarshal<double>(simcall->args[7])));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_EXECUTION_CANCEL:
-      SIMIX_execution_cancel(simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]));
-      SIMIX_simcall_answer(simcall);
-      break;
+  SIMIX_execution_cancel(
+      simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_EXECUTION_SET_PRIORITY:
-      SIMIX_execution_set_priority(simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]), simgrid::simix::unmarshal<double>(simcall->args[1]));
-      SIMIX_simcall_answer(simcall);
-      break;
+  SIMIX_execution_set_priority(
+      simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]),
+      simgrid::simix::unmarshal<double>(simcall->args[1]));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_EXECUTION_SET_BOUND:
-      SIMIX_execution_set_bound(simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]), simgrid::simix::unmarshal<double>(simcall->args[1]));
-      SIMIX_simcall_answer(simcall);
-      break;
+  SIMIX_execution_set_bound(
+      simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]),
+      simgrid::simix::unmarshal<double>(simcall->args[1]));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_EXECUTION_WAIT:
-      simcall_HANDLER_execution_wait(simcall, simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]));
-      break;
+  simcall_HANDLER_execution_wait(
+      simcall,
+      simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]));
+  break;
 
 case SIMCALL_PROCESS_ON_EXIT:
       SIMIX_process_on_exit(simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]), simgrid::simix::unmarshal<int_f_pvoid_pvoid_t>(simcall->args[1]), simgrid::simix::unmarshal<void*>(simcall->args[2]));
@@ -182,43 +199,75 @@ case SIMCALL_PROCESS_RESTART:
       break;
 
 case SIMCALL_COMM_IPROBE:
-      simgrid::simix::marshal<smx_activity_t>(simcall->result, simcall_HANDLER_comm_iprobe(simcall, simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[0]), simgrid::simix::unmarshal<int>(simcall->args[1]), simgrid::simix::unmarshal<int>(simcall->args[2]), simgrid::simix::unmarshal<int>(simcall->args[3]), simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[4]), simgrid::simix::unmarshal<void*>(simcall->args[5])));
-      SIMIX_simcall_answer(simcall);
-      break;
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(
+      simcall->result, simcall_HANDLER_comm_iprobe(simcall, simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[0]),
+                                                   simgrid::simix::unmarshal<int>(simcall->args[1]),
+                                                   simgrid::simix::unmarshal<int>(simcall->args[2]),
+                                                   simgrid::simix::unmarshal<int>(simcall->args[3]),
+                                                   simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[4]),
+                                                   simgrid::simix::unmarshal<void*>(simcall->args[5])));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_COMM_SEND:
       simcall_HANDLER_comm_send(simcall, simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]), simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]), simgrid::simix::unmarshal<double>(simcall->args[2]), simgrid::simix::unmarshal<double>(simcall->args[3]), simgrid::simix::unmarshal<void*>(simcall->args[4]), simgrid::simix::unmarshal<size_t>(simcall->args[5]), simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[6]), simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[7]), simgrid::simix::unmarshal<void*>(simcall->args[8]), simgrid::simix::unmarshal<double>(simcall->args[9]));
       break;
 
 case SIMCALL_COMM_ISEND:
-      simgrid::simix::marshal<smx_activity_t>(simcall->result, simcall_HANDLER_comm_isend(simcall, simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]), simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]), simgrid::simix::unmarshal<double>(simcall->args[2]), simgrid::simix::unmarshal<double>(simcall->args[3]), simgrid::simix::unmarshal<void*>(simcall->args[4]), simgrid::simix::unmarshal<size_t>(simcall->args[5]), simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[6]), simgrid::simix::unmarshal<simix_clean_func_t>(simcall->args[7]), simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[8]), simgrid::simix::unmarshal<void*>(simcall->args[9]), simgrid::simix::unmarshal<int>(simcall->args[10])));
-      SIMIX_simcall_answer(simcall);
-      break;
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(
+      simcall->result,
+      simcall_HANDLER_comm_isend(
+          simcall, simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]),
+          simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]),
+          simgrid::simix::unmarshal<double>(simcall->args[2]), simgrid::simix::unmarshal<double>(simcall->args[3]),
+          simgrid::simix::unmarshal<void*>(simcall->args[4]), simgrid::simix::unmarshal<size_t>(simcall->args[5]),
+          simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[6]),
+          simgrid::simix::unmarshal<simix_clean_func_t>(simcall->args[7]),
+          simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[8]),
+          simgrid::simix::unmarshal<void*>(simcall->args[9]), simgrid::simix::unmarshal<int>(simcall->args[10])));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_COMM_RECV:
       simcall_HANDLER_comm_recv(simcall, simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]), simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]), simgrid::simix::unmarshal<void*>(simcall->args[2]), simgrid::simix::unmarshal<size_t*>(simcall->args[3]), simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[4]), simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[5]), simgrid::simix::unmarshal<void*>(simcall->args[6]), simgrid::simix::unmarshal<double>(simcall->args[7]), simgrid::simix::unmarshal<double>(simcall->args[8]));
       break;
 
 case SIMCALL_COMM_IRECV:
-      simgrid::simix::marshal<smx_activity_t>(simcall->result, simcall_HANDLER_comm_irecv(simcall, simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]), simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]), simgrid::simix::unmarshal<void*>(simcall->args[2]), simgrid::simix::unmarshal<size_t*>(simcall->args[3]), simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[4]), simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[5]), simgrid::simix::unmarshal<void*>(simcall->args[6]), simgrid::simix::unmarshal<double>(simcall->args[7])));
-      SIMIX_simcall_answer(simcall);
-      break;
+  simgrid::simix::marshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(
+      simcall->result, simcall_HANDLER_comm_irecv(simcall, simgrid::simix::unmarshal<smx_actor_t>(simcall->args[0]),
+                                                  simgrid::simix::unmarshal<smx_mailbox_t>(simcall->args[1]),
+                                                  simgrid::simix::unmarshal<void*>(simcall->args[2]),
+                                                  simgrid::simix::unmarshal<size_t*>(simcall->args[3]),
+                                                  simgrid::simix::unmarshal<simix_match_func_t>(simcall->args[4]),
+                                                  simgrid::simix::unmarshal<simix_copy_data_func_t>(simcall->args[5]),
+                                                  simgrid::simix::unmarshal<void*>(simcall->args[6]),
+                                                  simgrid::simix::unmarshal<double>(simcall->args[7])));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_COMM_WAITANY:
       simcall_HANDLER_comm_waitany(simcall, simgrid::simix::unmarshal<xbt_dynar_t>(simcall->args[0]), simgrid::simix::unmarshal<double>(simcall->args[1]));
       break;
 
 case SIMCALL_COMM_WAIT:
-      simcall_HANDLER_comm_wait(simcall, simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]), simgrid::simix::unmarshal<double>(simcall->args[1]));
-      break;
+  simcall_HANDLER_comm_wait(
+      simcall,
+      simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]),
+      simgrid::simix::unmarshal<double>(simcall->args[1]));
+  break;
 
 case SIMCALL_COMM_TEST:
-      simcall_HANDLER_comm_test(simcall, simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]));
-      break;
+  simcall_HANDLER_comm_test(
+      simcall,
+      simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]));
+  break;
 
 case SIMCALL_COMM_TESTANY:
-      simcall_HANDLER_comm_testany(simcall, simgrid::simix::unmarshal<smx_activity_t*>(simcall->args[0]), simgrid::simix::unmarshal<size_t>(simcall->args[1]));
-      break;
+  simcall_HANDLER_comm_testany(
+      simcall,
+      simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>*>(simcall->args[0]),
+      simgrid::simix::unmarshal<size_t>(simcall->args[1]));
+  break;
 
 case SIMCALL_MUTEX_INIT:
       simgrid::simix::marshal<smx_mutex_t>(simcall->result, simcall_HANDLER_mutex_init(simcall));
@@ -336,20 +385,17 @@ case SIMCALL_FILE_MOVE:
       SIMIX_simcall_answer(simcall);
       break;
 
-case SIMCALL_STORAGE_GET_PROPERTIES:
-      simgrid::simix::marshal<xbt_dict_t>(simcall->result, SIMIX_storage_get_properties(simgrid::simix::unmarshal<smx_storage_t>(simcall->args[0])));
-      SIMIX_simcall_answer(simcall);
-      break;
-
 case SIMCALL_MC_RANDOM:
       simgrid::simix::marshal<int>(simcall->result, simcall_HANDLER_mc_random(simcall, simgrid::simix::unmarshal<int>(simcall->args[0]), simgrid::simix::unmarshal<int>(simcall->args[1])));
       SIMIX_simcall_answer(simcall);
       break;
 
 case SIMCALL_SET_CATEGORY:
-      SIMIX_set_category(simgrid::simix::unmarshal<smx_activity_t>(simcall->args[0]), simgrid::simix::unmarshal<const char*>(simcall->args[1]));
-      SIMIX_simcall_answer(simcall);
-      break;
+  SIMIX_set_category(
+      simgrid::simix::unmarshal<boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>>(simcall->args[0]),
+      simgrid::simix::unmarshal<const char*>(simcall->args[1]));
+  SIMIX_simcall_answer(simcall);
+  break;
 
 case SIMCALL_RUN_KERNEL:
       SIMIX_run_kernel(simgrid::simix::unmarshal<std::function<void()> const*>(simcall->args[0]));
diff --git a/src/simix/popping_private.h b/src/simix/popping_private.h
index 389f89b407..90ac87f707 100644
--- a/src/simix/popping_private.h
+++ b/src/simix/popping_private.h
@@ -71,6 +71,20 @@ SG_END_DECL()
 
 #ifdef __cplusplus
 
+/* Defines the marshal/unmarshal functions for each type of parameters.
+ *
+ * They will be used in popping_accessors.h to define the functions allowing
+ * to retrieve/set each parameter of each simcall.
+ *
+ * There is a unmarshal_raw() function, which is exactly similar to unmarshal()
+ * for all types but boost::intrusive_ptr(T). For that type, the unmarshal()
+ * function builds a new intrusive_ptr wrapping the pointer (that is stored raw
+ * within the simcall) while the unmarshal_raw retrieves the raw pointer.
+ *
+ * This is used in <simcall>_getraw_<param> functions, that allow the
+ * model-checker, to read the data in the remote memory of the MCed.
+ */
+
 namespace simgrid {
 namespace simix {
 
@@ -85,15 +99,11 @@ class type {
 };
 
 template<typename T> struct marshal_t {};
-#define SIMIX_MARSHAL(T, field) \
-  inline void marshal(type<T>, u_smx_scalar& simcall, T value) \
-  { \
-    simcall.field = value; \
-  } \
-  inline T unmarshal(type<T>, u_smx_scalar const& simcall) \
-  { \
-    return simcall.field; \
-  }
+#define SIMIX_MARSHAL(T, field)                                                                                        \
+  inline void marshal(type<T>, u_smx_scalar& simcall, T value) { simcall.field = value; }                              \
+  inline T unmarshal(type<T>, u_smx_scalar const& simcall) { return simcall.field; }                                   \
+  inline T unmarshal_raw(type<T>, u_smx_scalar const& simcall)                                                         \
+  { /* Exactly same as unmarshal. It differs only for intrusive_ptr */ return simcall.field; }
 
 SIMIX_MARSHAL(char, c);
 SIMIX_MARSHAL(short, s);
@@ -109,8 +119,12 @@ SIMIX_MARSHAL(float, d);
 SIMIX_MARSHAL(double, d);
 SIMIX_MARSHAL(FPtr, fp);
 
-inline
-void unmarshal(type<void>, u_smx_scalar const& simcall) {}
+inline void unmarshal(type<void>, u_smx_scalar const& simcall)
+{
+}
+inline void unmarshal_raw(type<void>, u_smx_scalar const& simcall)
+{
+}
 
 template<class T> inline
 void marshal(type<T*>, u_smx_scalar& simcall, T* value)
@@ -122,19 +136,31 @@ T* unmarshal(type<T*>, u_smx_scalar const& simcall)
 {
   return static_cast<T*>(simcall.dp);
 }
+template <class T> inline T* unmarshal_raw(type<T*>, u_smx_scalar const& simcall)
+{
+  return static_cast<T*>(simcall.dp);
+}
 
 template <class T>
 inline void marshal(type<boost::intrusive_ptr<T>>, u_smx_scalar& simcall, boost::intrusive_ptr<T> value)
 {
-  intrusive_ptr_add_ref(&*value);
-  simcall.dp = static_cast<void*>(&*value);
+  if (value.get() == nullptr) { // Sometimes we return nullptr in an intrusive_ptr...
+    simcall.dp = nullptr;
+  } else {
+    intrusive_ptr_add_ref(&*value);
+    simcall.dp = static_cast<void*>(&*value);
+  }
 }
 template <class T> inline boost::intrusive_ptr<T> unmarshal(type<boost::intrusive_ptr<T>>, u_smx_scalar const& simcall)
 {
+  // refcount was already increased during the marshaling, thus the "false" as last argument
   boost::intrusive_ptr<T> res = boost::intrusive_ptr<T>(static_cast<T*>(simcall.dp), false);
-  intrusive_ptr_release(&*res);
   return res;
 }
+template <class T> inline T* unmarshal_raw(type<boost::intrusive_ptr<T>>, u_smx_scalar const& simcall)
+{
+  return static_cast<T*>(simcall.dp);
+}
 
 template<class R, class... T> inline
 void marshal(type<R(*)(T...)>, u_smx_scalar& simcall, R(*value)(T...))
@@ -146,6 +172,10 @@ auto unmarshal(type<R(*)(T...)>, u_smx_scalar simcall) -> R(*)(T...)
 {
   return (R(*)(T...)) simcall.fp;
 }
+template <class R, class... T> inline auto unmarshal_raw(type<R (*)(T...)>, u_smx_scalar simcall) -> R (*)(T...)
+{
+  return (R(*)(T...))simcall.fp;
+}
 
 template<class T> inline
 void marshal(u_smx_scalar& simcall, T const& value)
@@ -157,6 +187,10 @@ typename std::remove_reference<T>::type unmarshal(u_smx_scalar& simcall)
 {
   return unmarshal(type<T>(), simcall);
 }
+template <class T> inline typename std::remove_reference<T>::type unmarshal_raw(u_smx_scalar& simcall)
+{
+  return unmarshal(type<T>(), simcall);
+}
 
 template<std::size_t I>
 inline void marshalArgs(smx_simcall_t simcall) {}
diff --git a/src/simix/simcalls.in b/src/simix/simcalls.in
index 31bc2788e5..ad9ed06060 100644
--- a/src/simix/simcalls.in
+++ b/src/simix/simcalls.in
@@ -46,26 +46,26 @@ int  process_is_suspended(smx_actor_t process) [[nohandler]];
 int  process_join(smx_actor_t process, double timeout) [[block]];
 int  process_sleep(double duration) [[block]];
 
-smx_activity_t execution_start(const char* name, double flops_amount, double priority, double bound);
-smx_activity_t execution_parallel_start(const char* name, int host_nb, sg_host_t* host_list, double* flops_amount, double* bytes_amount, double amount, double rate, double timeout) [[nohandler]];
-void          execution_cancel(smx_activity_t execution) [[nohandler]];
-void          execution_set_priority(smx_activity_t execution, double priority) [[nohandler]];
-void          execution_set_bound(smx_activity_t execution, double bound) [[nohandler]];
-int           execution_wait(smx_activity_t execution) [[block]];
+boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution_start(const char* name, double flops_amount, double priority, double bound);
+boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution_parallel_start(const char* name, int host_nb, sg_host_t* host_list, double* flops_amount, double* bytes_amount, double amount, double rate, double timeout) [[nohandler]];
+void          execution_cancel(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution) [[nohandler]];
+void          execution_set_priority(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution, double priority) [[nohandler]];
+void          execution_set_bound(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution, double bound) [[nohandler]];
+int           execution_wait(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> execution) [[block]];
 
 void          process_on_exit(smx_actor_t process, int_f_pvoid_pvoid_t fun, void* data) [[nohandler]];
 void          process_auto_restart_set(smx_actor_t process, int auto_restart) [[nohandler]];
 smx_actor_t   process_restart(smx_actor_t process);
 
-smx_activity_t comm_iprobe(smx_mailbox_t mbox, int type, int src, int tag, simix_match_func_t match_fun, void* data);
+boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm_iprobe(smx_mailbox_t mbox, int type, int src, int tag, simix_match_func_t match_fun, void* data);
 void           comm_send(smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate, void* src_buff, size_t src_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double timeout) [[block]];
-smx_activity_t comm_isend(smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate, void* src_buff, size_t src_buff_size, simix_match_func_t match_fun, simix_clean_func_t clean_fun, simix_copy_data_func_t copy_data_fun, void* data, int detached);
+boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm_isend(smx_actor_t sender, smx_mailbox_t mbox, double task_size, double rate, void* src_buff, size_t src_buff_size, simix_match_func_t match_fun, simix_clean_func_t clean_fun, simix_copy_data_func_t copy_data_fun, void* data, int detached);
 void           comm_recv(smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff, size_t* dst_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double timeout, double rate) [[block]];
-smx_activity_t comm_irecv(smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff, size_t* dst_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double rate);
+boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm_irecv(smx_actor_t receiver, smx_mailbox_t mbox, void* dst_buff, size_t* dst_buff_size, simix_match_func_t match_fun, simix_copy_data_func_t copy_data_fun, void* data, double rate);
 int            comm_waitany(xbt_dynar_t comms, double timeout) [[block]];
-void           comm_wait(smx_activity_t comm, double timeout) [[block]];
-int            comm_test(smx_activity_t comm) [[block]];
-int            comm_testany(smx_activity_t* comms, size_t count) [[block]];
+void           comm_wait(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm, double timeout) [[block]];
+int            comm_test(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> comm) [[block]];
+int            comm_testany(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl>* comms, size_t count) [[block]];
 
 smx_mutex_t mutex_init();
 void        mutex_lock(smx_mutex_t mutex) [[block]];
@@ -96,10 +96,8 @@ int         file_seek(smx_file_t fd, sg_offset_t offset, int origin);
 xbt_dynar_t file_get_info(smx_file_t fd);
 int         file_move(smx_file_t fd, const char* fullpath);
 
-xbt_dict_t storage_get_properties(smx_storage_t storage) [[nohandler]];
-
 int        mc_random(int min, int max);
-void       set_category(smx_activity_t synchro, const char* category) [[nohandler]];
+void       set_category(boost::intrusive_ptr<simgrid::kernel::activity::ActivityImpl> synchro, const char* category) [[nohandler]];
 
 void       run_kernel(std::function<void()> const* code) [[nohandler]];
 void       run_blocking(std::function<void()> const* code) [[block,nohandler]];
diff --git a/src/simix/simcalls.py b/src/simix/simcalls.py
index 68635e5dd2..f846463b9c 100755
--- a/src/simix/simcalls.py
+++ b/src/simix/simcalls.py
@@ -80,14 +80,21 @@ class Simcall(object):
     def accessors(self):
         res = []
         res.append('')
+        regex = re.compile(r"^boost::intrusive_ptr<(.*?)>(.*)$") # to compute the raw type
         # Arguments getter/setters
         for i in range(len(self.args)):
             arg = self.args[i]
+            rawtype = regex.sub(r'\1*\2', arg.rettype())
             res.append('static inline %s simcall_%s__get__%s(smx_simcall_t simcall) {' % (
                 arg.rettype(), self.name, arg.name))
             res.append(
                 '  return simgrid::simix::unmarshal<%s>(simcall->args[%i]);' % (arg.rettype(), i))
             res.append('}')
+            res.append('static inline %s simcall_%s__getraw__%s(smx_simcall_t simcall) {' % (
+                rawtype, self.name, arg.name))
+            res.append(
+                '  return simgrid::simix::unmarshal_raw<%s>(simcall->args[%i]);' % (rawtype, i))
+            res.append('}')
             res.append('static inline void simcall_%s__set__%s(smx_simcall_t simcall, %s arg) {' % (
                 self.name, arg.name, arg.rettype()))
             res.append('    simgrid::simix::marshal<%s>(simcall->args[%i], arg);' % (arg.rettype(), i))
@@ -95,10 +102,15 @@ class Simcall(object):
 
         # Return value getter/setters
         if self.res.type != 'void':
+            rawtype = regex.sub(r'\1*\2', self.res.rettype())
             res.append(
                 'static inline %s simcall_%s__get__result(smx_simcall_t simcall){' % (self.res.rettype(), self.name))
             res.append('    return simgrid::simix::unmarshal<%s>(simcall->result);' % self.res.rettype())
             res.append('}')
+            res.append(
+                'static inline %s simcall_%s__getraw__result(smx_simcall_t simcall){' % (rawtype, self.name))
+            res.append('    return simgrid::simix::unmarshal_raw<%s>(simcall->result);' % rawtype)
+            res.append('}')
             res.append(
                 'static inline void simcall_%s__set__result(smx_simcall_t simcall, %s result){' % (self.name, self.res.rettype()))
             res.append('    simgrid::simix::marshal<%s>(simcall->result, result);' % (self.res.rettype()))
diff --git a/src/simix/smx_context.cpp b/src/simix/smx_context.cpp
index b9911d1635..5906e9d9d1 100644
--- a/src/simix/smx_context.cpp
+++ b/src/simix/smx_context.cpp
@@ -33,10 +33,10 @@
 #include <sys/mman.h>
 #endif
 
-#ifdef __MINGW32__ 
-#define _aligned_malloc __mingw_aligned_malloc 
-#define _aligned_free  __mingw_aligned_free 
-#endif //MINGW
+#ifdef __MINGW32__
+#define _aligned_malloc __mingw_aligned_malloc
+#define _aligned_free  __mingw_aligned_free
+#endif /*MINGW*/
 
 #if HAVE_VALGRIND_H
 # include <valgrind/valgrind.h>
@@ -271,7 +271,7 @@ int SIMIX_context_get_nthreads() {
  * \param nb_threads the number of threads to use
  */
 void SIMIX_context_set_nthreads(int nb_threads) {
-  if (nb_threads<=0) {  
+  if (nb_threads<=0) {
      nb_threads = xbt_os_get_numcores();
      XBT_INFO("Auto-setting contexts/nthreads to %d",nb_threads);
   }
diff --git a/src/simix/smx_global.cpp b/src/simix/smx_global.cpp
index a3c309a3c5..39c1c41507 100644
--- a/src/simix/smx_global.cpp
+++ b/src/simix/smx_global.cpp
@@ -38,7 +38,7 @@
 #include "src/mc/remote/Client.hpp"
 #include "src/mc/remote/mc_protocol.h"
 #include <stdlib.h>
-#endif 
+#endif
 
 #include "src/mc/mc_record.h"
 
@@ -235,12 +235,11 @@ void SIMIX_global_init(int *argc, char **argv)
     });
 
     simgrid::surf::storageCreatedCallbacks.connect([](simgrid::surf::StorageImpl* storage) {
-      const char* name = storage->cname();
-      // TODO, create sg_storage_by_name
-      sg_storage_t s = xbt_lib_get_elm_or_null(storage_lib, name);
-      xbt_assert(s != nullptr, "Storage not found for name %s", name);
+      sg_storage_t s = simgrid::s4u::Storage::byName(storage->cname());
+      xbt_assert(s != nullptr, "Storage not found for name %s", storage->cname());
     });
   }
+
   if (not simix_timers)
     simix_timers = xbt_heap_new(8, [](void* p) {
       delete static_cast<smx_timer_t>(p);
@@ -357,7 +356,7 @@ static void SIMIX_wake_processes()
     XBT_DEBUG("Handling the processes whose action failed (if any)");
     while ((action = surf_model_extract_failed_action_set(model))) {
       XBT_DEBUG("   Handling Action %p",action);
-      SIMIX_simcall_exit((smx_activity_t) action->getData());
+      SIMIX_simcall_exit(static_cast<simgrid::kernel::activity::ActivityImpl*>(action->getData()));
     }
     XBT_DEBUG("Handling the processes whose action terminated normally (if any)");
     while ((action = surf_model_extract_done_action_set(model))) {
@@ -365,7 +364,7 @@ static void SIMIX_wake_processes()
       if (action->getData() == nullptr)
         XBT_DEBUG("probably vcpu's action %p, skip", action);
       else
-        SIMIX_simcall_exit((smx_activity_t) action->getData());
+        SIMIX_simcall_exit(static_cast<simgrid::kernel::activity::ActivityImpl*>(action->getData()));
     }
   }
 }
@@ -654,19 +653,19 @@ void SIMIX_display_process_status()
 
       const char* synchro_description = "unknown";
 
-      if (dynamic_cast<simgrid::kernel::activity::ExecImpl*>(process->waiting_synchro) != nullptr)
+      if (boost::dynamic_pointer_cast<simgrid::kernel::activity::ExecImpl>(process->waiting_synchro) != nullptr)
         synchro_description = "execution";
 
-      if (dynamic_cast<simgrid::kernel::activity::CommImpl*>(process->waiting_synchro) != nullptr)
+      if (boost::dynamic_pointer_cast<simgrid::kernel::activity::CommImpl>(process->waiting_synchro) != nullptr)
         synchro_description = "communication";
 
-      if (dynamic_cast<simgrid::kernel::activity::SleepImpl*>(process->waiting_synchro) != nullptr)
+      if (boost::dynamic_pointer_cast<simgrid::kernel::activity::SleepImpl>(process->waiting_synchro) != nullptr)
         synchro_description = "sleeping";
 
-      if (dynamic_cast<simgrid::kernel::activity::Raw*>(process->waiting_synchro) != nullptr)
+      if (boost::dynamic_pointer_cast<simgrid::kernel::activity::RawImpl>(process->waiting_synchro) != nullptr)
         synchro_description = "synchronization";
 
-      if (dynamic_cast<simgrid::kernel::activity::Io*>(process->waiting_synchro) != nullptr)
+      if (boost::dynamic_pointer_cast<simgrid::kernel::activity::IoImpl>(process->waiting_synchro) != nullptr)
         synchro_description = "I/O";
 
 
@@ -682,7 +681,7 @@ void SIMIX_display_process_status()
 */
 
       XBT_INFO("Process %lu (%s@%s): waiting for %s synchro %p (%s) in state %d to finish", process->pid,
-               process->cname(), process->host->cname(), synchro_description, process->waiting_synchro,
+               process->cname(), process->host->cname(), synchro_description, process->waiting_synchro.get(),
                process->waiting_synchro->name.c_str(), (int)process->waiting_synchro->state);
     }
     else {
diff --git a/src/simix/smx_host.cpp b/src/simix/smx_host.cpp
index e4592493c0..751916292b 100644
--- a/src/simix/smx_host.cpp
+++ b/src/simix/smx_host.cpp
@@ -161,20 +161,21 @@ smx_activity_t SIMIX_execution_start(smx_actor_t issuer, const char *name, doubl
                                     double bound){
 
   /* alloc structures and initialize */
-  simgrid::kernel::activity::ExecImpl* exec = new simgrid::kernel::activity::ExecImpl(name, issuer->host);
+  simgrid::kernel::activity::ExecImplPtr exec =
+      simgrid::kernel::activity::ExecImplPtr(new simgrid::kernel::activity::ExecImpl(name, issuer->host));
 
   /* set surf's action */
   if (not MC_is_active() && not MC_record_replay_is_active()) {
 
     exec->surf_exec = issuer->host->pimpl_cpu->execution_start(flops_amount);
-    exec->surf_exec->setData(exec);
+    exec->surf_exec->setData(exec.get());
     exec->surf_exec->setPriority(priority);
 
     if (bound > 0)
       static_cast<simgrid::surf::CpuAction*>(exec->surf_exec)->setBound(bound);
   }
 
-  XBT_DEBUG("Create execute synchro %p: %s", exec, exec->name.c_str());
+  XBT_DEBUG("Create execute synchro %p: %s", exec.get(), exec->name.c_str());
 
   return exec;
 }
@@ -184,7 +185,8 @@ smx_activity_t SIMIX_execution_parallel_start(const char* name, int host_nb, sg_
 {
 
   /* alloc structures and initialize */
-  simgrid::kernel::activity::ExecImpl* exec = new simgrid::kernel::activity::ExecImpl(name, nullptr);
+  simgrid::kernel::activity::ExecImplPtr exec =
+      simgrid::kernel::activity::ExecImplPtr(new simgrid::kernel::activity::ExecImpl(name, nullptr));
 
   /* set surf's synchro */
   sg_host_t *host_list_cpy = xbt_new0(sg_host_t, host_nb);
@@ -201,21 +203,22 @@ smx_activity_t SIMIX_execution_parallel_start(const char* name, int host_nb, sg_
   /* set surf's synchro */
   if (not MC_is_active() && not MC_record_replay_is_active()) {
     exec->surf_exec = surf_host_model->executeParallelTask(host_nb, host_list_cpy, flops_amount, bytes_amount, rate);
-    exec->surf_exec->setData(exec);
+    exec->surf_exec->setData(exec.get());
     if (timeout > 0) {
       exec->timeoutDetector = host_list[0]->pimpl_cpu->sleep(timeout);
-      exec->timeoutDetector->setData(exec);
+      exec->timeoutDetector->setData(exec.get());
     }
   }
-  XBT_DEBUG("Create parallel execute synchro %p", exec);
+  XBT_DEBUG("Create parallel execute synchro %p", exec.get());
 
   return exec;
 }
 
 void SIMIX_execution_cancel(smx_activity_t synchro)
 {
-  XBT_DEBUG("Cancel synchro %p", synchro);
-  simgrid::kernel::activity::ExecImpl* exec = static_cast<simgrid::kernel::activity::ExecImpl*>(synchro);
+  XBT_DEBUG("Cancel synchro %p", synchro.get());
+  simgrid::kernel::activity::ExecImplPtr exec =
+      boost::static_pointer_cast<simgrid::kernel::activity::ExecImpl>(synchro);
 
   if (exec->surf_exec)
     exec->surf_exec->cancel();
@@ -223,22 +226,25 @@ void SIMIX_execution_cancel(smx_activity_t synchro)
 
 void SIMIX_execution_set_priority(smx_activity_t synchro, double priority)
 {
-  simgrid::kernel::activity::ExecImpl* exec = static_cast<simgrid::kernel::activity::ExecImpl*>(synchro);
+  simgrid::kernel::activity::ExecImplPtr exec =
+      boost::static_pointer_cast<simgrid::kernel::activity::ExecImpl>(synchro);
   if(exec->surf_exec)
     exec->surf_exec->setPriority(priority);
 }
 
 void SIMIX_execution_set_bound(smx_activity_t synchro, double bound)
 {
-  simgrid::kernel::activity::ExecImpl* exec = static_cast<simgrid::kernel::activity::ExecImpl*>(synchro);
+  simgrid::kernel::activity::ExecImplPtr exec =
+      boost::static_pointer_cast<simgrid::kernel::activity::ExecImpl>(synchro);
   if(exec->surf_exec)
     static_cast<simgrid::surf::CpuAction*>(exec->surf_exec)->setBound(bound);
 }
 
 void simcall_HANDLER_execution_wait(smx_simcall_t simcall, smx_activity_t synchro)
 {
-  simgrid::kernel::activity::ExecImpl* exec = static_cast<simgrid::kernel::activity::ExecImpl*>(synchro);
-  XBT_DEBUG("Wait for execution of synchro %p, state %d", synchro, (int)synchro->state);
+  simgrid::kernel::activity::ExecImplPtr exec =
+      boost::static_pointer_cast<simgrid::kernel::activity::ExecImpl>(synchro);
+  XBT_DEBUG("Wait for execution of synchro %p, state %d", synchro.get(), (int)synchro->state);
 
   /* Associate this simcall to the synchro */
   synchro->simcalls.push_back(simcall);
@@ -256,7 +262,7 @@ void simcall_HANDLER_execution_wait(smx_simcall_t simcall, smx_activity_t synchr
     SIMIX_execution_finish(exec);
 }
 
-void SIMIX_execution_finish(simgrid::kernel::activity::ExecImpl* exec)
+void SIMIX_execution_finish(simgrid::kernel::activity::ExecImplPtr exec)
 {
   for (smx_simcall_t simcall : exec->simcalls) {
     switch (exec->state) {
@@ -294,9 +300,6 @@ void SIMIX_execution_finish(simgrid::kernel::activity::ExecImpl* exec)
     simcall_execution_wait__set__result(simcall, exec->state);
     SIMIX_simcall_answer(simcall);
   }
-
-  /* We no longer need it */
-  exec->unref();
 }
 
 void SIMIX_set_category(smx_activity_t synchro, const char *category)
@@ -304,13 +307,15 @@ void SIMIX_set_category(smx_activity_t synchro, const char *category)
   if (synchro->state != SIMIX_RUNNING)
     return;
 
-  simgrid::kernel::activity::ExecImpl* exec = dynamic_cast<simgrid::kernel::activity::ExecImpl*>(synchro);
+  simgrid::kernel::activity::ExecImplPtr exec =
+      boost::dynamic_pointer_cast<simgrid::kernel::activity::ExecImpl>(synchro);
   if (exec != nullptr) {
     exec->surf_exec->setCategory(category);
     return;
   }
 
-  simgrid::kernel::activity::CommImpl* comm = dynamic_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::dynamic_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
   if (comm != nullptr) {
     comm->surf_comm->setCategory(category);
   }
diff --git a/src/simix/smx_host_private.h b/src/simix/smx_host_private.h
index 91149d1479..5a953dd552 100644
--- a/src/simix/smx_host_private.h
+++ b/src/simix/smx_host_private.h
@@ -48,20 +48,21 @@ XBT_PRIVATE void SIMIX_host_add_auto_restart_process(sg_host_t host,
                                          int auto_restart);
 
 XBT_PRIVATE void SIMIX_host_autorestart(sg_host_t host);
-XBT_PRIVATE smx_activity_t SIMIX_execution_start(smx_actor_t issuer, const char *name,
-    double flops_amount, double priority, double bound);
-XBT_PRIVATE smx_activity_t SIMIX_execution_parallel_start(const char* name, int host_nb, sg_host_t* host_list,
-                                                          double* flops_amount, double* bytes_amount, double amount,
-                                                          double rate, double timeout);
 XBT_PRIVATE void SIMIX_execution_cancel(smx_activity_t synchro);
 XBT_PRIVATE void SIMIX_execution_set_priority(smx_activity_t synchro, double priority);
 XBT_PRIVATE void SIMIX_execution_set_bound(smx_activity_t synchro, double bound);
 
-XBT_PRIVATE void SIMIX_execution_finish(simgrid::kernel::activity::ExecImpl* exec);
+XBT_PRIVATE void SIMIX_execution_finish(simgrid::kernel::activity::ExecImplPtr exec);
 
 XBT_PRIVATE void SIMIX_set_category(smx_activity_t synchro, const char *category);
 
 SG_END_DECL()
 
+XBT_PRIVATE smx_activity_t SIMIX_execution_start(smx_actor_t issuer, const char* name, double flops_amount,
+                                                 double priority, double bound);
+XBT_PRIVATE smx_activity_t SIMIX_execution_parallel_start(const char* name, int host_nb, sg_host_t* host_list,
+                                                          double* flops_amount, double* bytes_amount, double amount,
+                                                          double rate, double timeout);
+
 #endif
 
diff --git a/src/simix/smx_io.cpp b/src/simix/smx_io.cpp
index 38659cae08..a420fe13df 100644
--- a/src/simix/smx_io.cpp
+++ b/src/simix/smx_io.cpp
@@ -33,7 +33,7 @@ smx_activity_t SIMIX_file_read(smx_file_t fd, sg_size_t size, sg_host_t host)
   if (host->isOff())
     THROWF(host_error, 0, "Host %s failed, you cannot call this function", host->cname());
 
-  simgrid::kernel::activity::Io *synchro = new simgrid::kernel::activity::Io();
+  simgrid::kernel::activity::IoImpl* synchro = new simgrid::kernel::activity::IoImpl();
   synchro->host = host;
   synchro->surf_io = surf_host_read(host, fd->surf_file, size);
 
@@ -56,7 +56,7 @@ smx_activity_t SIMIX_file_write(smx_file_t fd, sg_size_t size, sg_host_t host)
   if (host->isOff())
     THROWF(host_error, 0, "Host %s failed, you cannot call this function", host->cname());
 
-  simgrid::kernel::activity::Io *synchro = new simgrid::kernel::activity::Io();
+  simgrid::kernel::activity::IoImpl* synchro = new simgrid::kernel::activity::IoImpl();
   synchro->host = host;
   synchro->surf_io = surf_host_write(host, fd->surf_file, size);
   synchro->surf_io->setData(synchro);
@@ -78,7 +78,7 @@ smx_activity_t SIMIX_file_open(const char* fullpath, sg_host_t host)
   if (host->isOff())
     THROWF(host_error, 0, "Host %s failed, you cannot call this function", host->cname());
 
-  simgrid::kernel::activity::Io *synchro = new simgrid::kernel::activity::Io();
+  simgrid::kernel::activity::IoImpl* synchro = new simgrid::kernel::activity::IoImpl();
   synchro->host = host;
   synchro->surf_io = surf_host_open(host, fullpath);
   synchro->surf_io->setData(synchro);
@@ -100,7 +100,7 @@ smx_activity_t SIMIX_file_close(smx_file_t fd, sg_host_t host)
   if (host->isOff())
     THROWF(host_error, 0, "Host %s failed, you cannot call this function", host->cname());
 
-  simgrid::kernel::activity::Io *synchro = new simgrid::kernel::activity::Io();
+  simgrid::kernel::activity::IoImpl* synchro = new simgrid::kernel::activity::IoImpl();
   synchro->host = host;
   synchro->surf_io = surf_host_close(host, fd->surf_file);
   synchro->surf_io->setData(synchro);
@@ -176,17 +176,12 @@ int SIMIX_file_move(smx_actor_t process, smx_file_t file, const char* fullpath)
   return  surf_host_file_move(host, file->surf_file, fullpath);
 }
 
-xbt_dict_t SIMIX_storage_get_properties(smx_storage_t storage){
-  return surf_storage_get_properties(storage);
-}
-
 void SIMIX_io_destroy(smx_activity_t synchro)
 {
-  simgrid::kernel::activity::Io *io = static_cast<simgrid::kernel::activity::Io*>(synchro);
-  XBT_DEBUG("Destroy synchro %p", synchro);
+  simgrid::kernel::activity::IoImplPtr io = boost::static_pointer_cast<simgrid::kernel::activity::IoImpl>(synchro);
+  XBT_DEBUG("Destroy synchro %p", synchro.get());
   if (io->surf_io)
     io->surf_io->unref();
-  delete io;
 }
 
 void SIMIX_io_finish(smx_activity_t synchro)
diff --git a/src/simix/smx_io_private.h b/src/simix/smx_io_private.h
index 0f9bfd328a..0d48df10c3 100644
--- a/src/simix/smx_io_private.h
+++ b/src/simix/smx_io_private.h
@@ -22,8 +22,6 @@ XBT_PRIVATE xbt_dynar_t SIMIX_file_get_info(smx_actor_t process, smx_file_t fd);
 XBT_PRIVATE int SIMIX_file_seek(smx_actor_t process, smx_file_t fd, sg_offset_t offset, int origin);
 XBT_PRIVATE int SIMIX_file_move(smx_actor_t process, smx_file_t fd, const char* fullpath);
 
-XBT_PRIVATE xbt_dict_t SIMIX_storage_get_properties(smx_storage_t storage);
-
 XBT_PRIVATE void SIMIX_io_destroy(smx_activity_t synchro);
 XBT_PRIVATE void SIMIX_io_finish(smx_activity_t synchro);
 
diff --git a/src/simix/smx_network.cpp b/src/simix/smx_network.cpp
index 528ceda881..07f4c3c8e6 100644
--- a/src/simix/smx_network.cpp
+++ b/src/simix/smx_network.cpp
@@ -26,18 +26,14 @@ XBT_LOG_NEW_DEFAULT_SUBCATEGORY(simix_network, simix, "SIMIX network-related syn
 
 static void SIMIX_waitany_remove_simcall_from_actions(smx_simcall_t simcall);
 static void SIMIX_comm_copy_data(smx_activity_t comm);
-static void SIMIX_comm_start(smx_activity_t synchro);
-static simgrid::kernel::activity::CommImpl*
-_find_matching_comm(boost::circular_buffer_space_optimized<smx_activity_t>* deque, e_smx_comm_type_t type,
-                    int (*match_fun)(void*, void*, smx_activity_t), void* user_data, smx_activity_t my_synchro,
-                    bool remove_matching);
+static void SIMIX_comm_start(simgrid::kernel::activity::CommImplPtr synchro);
 
 /**
  *  \brief Checks if there is a communication activity queued in a deque matching our needs
  *  \param type The type of communication we are looking for (comm_send, comm_recv)
  *  \return The communication activity if found, nullptr otherwise
  */
-static simgrid::kernel::activity::CommImpl*
+static simgrid::kernel::activity::CommImplPtr
 _find_matching_comm(boost::circular_buffer_space_optimized<smx_activity_t>* deque, e_smx_comm_type_t type,
                     int (*match_fun)(void*, void*, smx_activity_t), void* this_user_data, smx_activity_t my_synchro,
                     bool remove_matching)
@@ -46,19 +42,19 @@ _find_matching_comm(boost::circular_buffer_space_optimized<smx_activity_t>* dequ
 
   for(auto it = deque->begin(); it != deque->end(); it++){
     smx_activity_t synchro = *it;
-    simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+    simgrid::kernel::activity::CommImplPtr comm =
+        boost::dynamic_pointer_cast<simgrid::kernel::activity::CommImpl>(std::move(synchro));
 
     if (comm->type == SIMIX_COMM_SEND) {
       other_user_data = comm->src_data;
     } else if (comm->type == SIMIX_COMM_RECEIVE) {
       other_user_data = comm->dst_data;
     }
-    if (comm->type == type && (match_fun == nullptr || match_fun(this_user_data, other_user_data, synchro)) &&
+    if (comm->type == type && (match_fun == nullptr || match_fun(this_user_data, other_user_data, comm)) &&
         (not comm->match_fun || comm->match_fun(other_user_data, this_user_data, my_synchro))) {
-      XBT_DEBUG("Found a matching communication synchro %p", comm);
+      XBT_DEBUG("Found a matching communication synchro %p", comm.get());
       if (remove_matching)
         deque->erase(it);
-      comm->ref();
 #if SIMGRID_HAVE_MC
       comm->mbox_cpy = comm->mbox;
 #endif
@@ -67,7 +63,7 @@ _find_matching_comm(boost::circular_buffer_space_optimized<smx_activity_t>* dequ
     }
     XBT_DEBUG("Sorry, communication synchro %p does not match our needs:"
               " its type is %d but we are looking for a comm of type %d (or maybe the filtering didn't match)",
-              comm, (int)comm->type, (int)type);
+              comm.get(), (int)comm->type, (int)type);
   }
   XBT_DEBUG("No matching communication synchro found");
   return nullptr;
@@ -96,36 +92,34 @@ XBT_PRIVATE smx_activity_t simcall_HANDLER_comm_isend(smx_simcall_t simcall, smx
                                   void (*copy_data_fun)(smx_activity_t, void*, size_t),// used to copy data if not default one
                           void *data, int detached)
 {
-  XBT_DEBUG("send from %p", mbox);
+  XBT_DEBUG("send from mailbox %p", mbox);
 
   /* Prepare a synchro describing us, so that it gets passed to the user-provided filter of other side */
-  simgrid::kernel::activity::CommImpl* this_comm = new simgrid::kernel::activity::CommImpl(SIMIX_COMM_SEND);
+  simgrid::kernel::activity::CommImplPtr this_comm =
+      simgrid::kernel::activity::CommImplPtr(new simgrid::kernel::activity::CommImpl(SIMIX_COMM_SEND));
 
   /* Look for communication synchro matching our needs. We also provide a description of
    * ourself so that the other side also gets a chance of choosing if it wants to match with us.
    *
    * If it is not found then push our communication into the rendez-vous point */
-  simgrid::kernel::activity::CommImpl* other_comm =
+  simgrid::kernel::activity::CommImplPtr other_comm =
       _find_matching_comm(&mbox->comm_queue, SIMIX_COMM_RECEIVE, match_fun, data, this_comm, /*remove_matching*/ true);
 
   if (not other_comm) {
-    other_comm = this_comm;
+    other_comm = std::move(this_comm);
 
-    if (mbox->permanent_receiver!=nullptr){
+    if (mbox->permanent_receiver != nullptr) {
       //this mailbox is for small messages, which have to be sent right now
       other_comm->state   = SIMIX_READY;
       other_comm->dst_proc=mbox->permanent_receiver.get();
-      other_comm->ref();
       mbox->done_comm_queue.push_back(other_comm);
-      XBT_DEBUG("pushing a message into the permanent receive list %p, comm %p", mbox, other_comm);
+      XBT_DEBUG("pushing a message into the permanent receive list %p, comm %p", mbox, other_comm.get());
 
     }else{
-      mbox->push(this_comm);
+      mbox->push(other_comm);
     }
   } else {
     XBT_DEBUG("Receive already pushed");
-    this_comm->unref();
-    this_comm->unref();
 
     other_comm->state = SIMIX_READY;
     other_comm->type = SIMIX_COMM_READY;
@@ -185,14 +179,14 @@ smx_activity_t SIMIX_comm_irecv(smx_actor_t dst_proc, smx_mailbox_t mbox, void *
     void (*copy_data_fun)(smx_activity_t, void*, size_t), // used to copy data if not default one
     void *data, double rate)
 {
-  simgrid::kernel::activity::CommImpl* this_synchro = new simgrid::kernel::activity::CommImpl(SIMIX_COMM_RECEIVE);
-  XBT_DEBUG("recv from %p %p. this_synchro=%p", mbox, &mbox->comm_queue, this_synchro);
+  simgrid::kernel::activity::CommImplPtr this_synchro =
+      simgrid::kernel::activity::CommImplPtr(new simgrid::kernel::activity::CommImpl(SIMIX_COMM_RECEIVE));
+  XBT_DEBUG("recv from mbox %p. this_synchro=%p", mbox, this_synchro.get());
 
-  simgrid::kernel::activity::CommImpl* other_comm;
+  simgrid::kernel::activity::CommImplPtr other_comm;
   //communication already done, get it inside the list of completed comms
   if (mbox->permanent_receiver != nullptr && not mbox->done_comm_queue.empty()) {
 
-    this_synchro->unref();
     XBT_DEBUG("We have a comm that has probably already been received, trying to match it, to skip the communication");
     //find a match in the list of already received comms
     other_comm = _find_matching_comm(&mbox->done_comm_queue, SIMIX_COMM_SEND, match_fun, data, this_synchro,
@@ -204,14 +198,11 @@ smx_activity_t SIMIX_comm_irecv(smx_actor_t dst_proc, smx_mailbox_t mbox, void *
       mbox->push(this_synchro);
     } else {
       if (other_comm->surf_comm && other_comm->remains() < 1e-12) {
-        XBT_DEBUG("comm %p has been already sent, and is finished, destroy it",other_comm);
+        XBT_DEBUG("comm %p has been already sent, and is finished, destroy it", other_comm.get());
         other_comm->state = SIMIX_DONE;
         other_comm->type = SIMIX_COMM_DONE;
         other_comm->mbox = nullptr;
-        other_comm->unref();
       }
-      other_comm->unref();
-      this_synchro->unref();
     }
   } else {
     /* Prepare a comm describing us, so that it gets passed to the user-provided filter of other side */
@@ -223,19 +214,15 @@ smx_activity_t SIMIX_comm_irecv(smx_actor_t dst_proc, smx_mailbox_t mbox, void *
     other_comm = _find_matching_comm(&mbox->comm_queue, SIMIX_COMM_SEND, match_fun, data, this_synchro,
                                      /*remove_matching*/ true);
 
-    if (not other_comm) {
-      XBT_DEBUG("Receive pushed first %zu", mbox->comm_queue.size());
+    if (other_comm == nullptr) {
+      XBT_DEBUG("Receive pushed first (%zu comm enqueued so far)", mbox->comm_queue.size());
       other_comm = this_synchro;
       mbox->push(this_synchro);
     } else {
-      XBT_DEBUG("Match my %p with the existing %p", this_synchro, other_comm);
-
-      other_comm = static_cast<simgrid::kernel::activity::CommImpl*>(other_comm);
+      XBT_DEBUG("Match my %p with the existing %p", this_synchro.get(), other_comm.get());
 
       other_comm->state = SIMIX_READY;
       other_comm->type = SIMIX_COMM_READY;
-      this_synchro->unref();
-      this_synchro->unref();
     }
     dst_proc->comms.push_back(other_comm);
   }
@@ -272,15 +259,15 @@ smx_activity_t SIMIX_comm_iprobe(smx_actor_t dst_proc, smx_mailbox_t mbox, int t
                               int tag, int (*match_fun)(void *, void *, smx_activity_t), void *data)
 {
   XBT_DEBUG("iprobe from %p %p", mbox, &mbox->comm_queue);
-  simgrid::kernel::activity::CommImpl* this_comm;
+  simgrid::kernel::activity::CommImplPtr this_comm;
   int smx_type;
   if(type == 1){
-    this_comm = new simgrid::kernel::activity::CommImpl(SIMIX_COMM_SEND);
+    this_comm = simgrid::kernel::activity::CommImplPtr(new simgrid::kernel::activity::CommImpl(SIMIX_COMM_SEND));
     smx_type = SIMIX_COMM_RECEIVE;
   } else{
-    this_comm = new simgrid::kernel::activity::CommImpl(SIMIX_COMM_RECEIVE);
+    this_comm = simgrid::kernel::activity::CommImplPtr(new simgrid::kernel::activity::CommImpl(SIMIX_COMM_RECEIVE));
     smx_type = SIMIX_COMM_SEND;
-  } 
+  }
   smx_activity_t other_synchro=nullptr;
   if (mbox->permanent_receiver != nullptr && not mbox->done_comm_queue.empty()) {
     XBT_DEBUG("first check in the permanent recv mailbox, to see if we already got something");
@@ -293,17 +280,13 @@ smx_activity_t SIMIX_comm_iprobe(smx_actor_t dst_proc, smx_mailbox_t mbox, int t
       (e_smx_comm_type_t) smx_type, match_fun, data, this_comm,/*remove_matching*/false);
   }
 
-  if(other_synchro)
-    other_synchro->unref();
-
-  this_comm->unref();
   return other_synchro;
 }
 
 void simcall_HANDLER_comm_wait(smx_simcall_t simcall, smx_activity_t synchro, double timeout)
 {
   /* Associate this simcall to the wait synchro */
-  XBT_DEBUG("simcall_HANDLER_comm_wait, %p", synchro);
+  XBT_DEBUG("simcall_HANDLER_comm_wait, %p", synchro.get());
 
   synchro->simcalls.push_back(simcall);
   simcall->issuer->waiting_synchro = synchro;
@@ -318,7 +301,8 @@ void simcall_HANDLER_comm_wait(smx_simcall_t simcall, smx_activity_t synchro, do
       if (timeout < 0.0)
         THROW_IMPOSSIBLE;
 
-      simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+      simgrid::kernel::activity::CommImplPtr comm =
+          boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
       if (comm->src_proc == simcall->issuer)
         comm->state = SIMIX_SRC_TIMEOUT;
       else
@@ -335,9 +319,10 @@ void simcall_HANDLER_comm_wait(smx_simcall_t simcall, smx_activity_t synchro, do
     SIMIX_comm_finish(synchro);
   } else { /* if (timeout >= 0) { we need a surf sleep action even when there is no timeout, otherwise surf won't tell us when the host fails */
     surf_action_t sleep = simcall->issuer->host->pimpl_cpu->sleep(timeout);
-    sleep->setData(synchro);
+    sleep->setData(&*synchro);
 
-    simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+    simgrid::kernel::activity::CommImplPtr comm =
+        boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
     if (simcall->issuer == comm->src_proc)
       comm->src_timeout = sleep;
     else
@@ -347,7 +332,8 @@ void simcall_HANDLER_comm_wait(smx_simcall_t simcall, smx_activity_t synchro, do
 
 void simcall_HANDLER_comm_test(smx_simcall_t simcall, smx_activity_t synchro)
 {
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
 
   if (MC_is_active() || MC_record_replay_is_active()){
     simcall_comm_test__set__result(simcall, comm->src_proc && comm->dst_proc);
@@ -370,8 +356,8 @@ void simcall_HANDLER_comm_test(smx_simcall_t simcall, smx_activity_t synchro)
   }
 }
 
-void simcall_HANDLER_comm_testany(
-  smx_simcall_t simcall, simgrid::kernel::activity::ActivityImpl* comms[], size_t count)
+void simcall_HANDLER_comm_testany(smx_simcall_t simcall, simgrid::kernel::activity::ActivityImplPtr comms[],
+                                  size_t count)
 {
   // The default result is -1 -- this means, "nothing is ready".
   // It can be changed below, but only if something matches.
@@ -382,7 +368,7 @@ void simcall_HANDLER_comm_testany(
     if(idx == -1){
       SIMIX_simcall_answer(simcall);
     }else{
-      simgrid::kernel::activity::ActivityImpl* synchro = comms[idx];
+      simgrid::kernel::activity::ActivityImplPtr synchro = comms[idx];
       simcall_comm_testany__set__result(simcall, idx);
       synchro->simcalls.push_back(simcall);
       synchro->state = SIMIX_DONE;
@@ -392,7 +378,7 @@ void simcall_HANDLER_comm_testany(
   }
 
   for (std::size_t i = 0; i != count; ++i) {
-    simgrid::kernel::activity::ActivityImpl* synchro = comms[i];
+    simgrid::kernel::activity::ActivityImplPtr synchro = comms[i];
     if (synchro->state != SIMIX_WAITING && synchro->state != SIMIX_RUNNING) {
       simcall_comm_testany__set__result(simcall, i);
       synchro->simcalls.push_back(simcall);
@@ -405,21 +391,18 @@ void simcall_HANDLER_comm_testany(
 
 void simcall_HANDLER_comm_waitany(smx_simcall_t simcall, xbt_dynar_t synchros, double timeout)
 {
-  smx_activity_t synchro;
-  unsigned int cursor = 0;
-
   if (MC_is_active() || MC_record_replay_is_active()){
     if (timeout > 0.0)
-      xbt_die("Timeout not implemented for waitany in the model-checker"); 
+      xbt_die("Timeout not implemented for waitany in the model-checker");
     int idx = SIMCALL_GET_MC_VALUE(simcall);
-    synchro = xbt_dynar_get_as(synchros, idx, smx_activity_t);
+    smx_activity_t synchro = xbt_dynar_get_as(synchros, idx, smx_activity_t);
     synchro->simcalls.push_back(simcall);
     simcall_comm_waitany__set__result(simcall, idx);
     synchro->state = SIMIX_DONE;
     SIMIX_comm_finish(synchro);
     return;
   }
-  
+
   if (timeout < 0.0){
     simcall->timer = NULL;
   } else {
@@ -429,8 +412,11 @@ void simcall_HANDLER_comm_waitany(smx_simcall_t simcall, xbt_dynar_t synchros, d
       SIMIX_simcall_answer(simcall);
     });
   }
-  
-  xbt_dynar_foreach(synchros, cursor, synchro){
+
+  unsigned int cursor;
+  simgrid::kernel::activity::ActivityImpl* ptr;
+  xbt_dynar_foreach(synchros, cursor, ptr){
+    smx_activity_t synchro = simgrid::kernel::activity::ActivityImplPtr(ptr);
     /* associate this simcall to the the synchro */
     synchro->simcalls.push_back(simcall);
 
@@ -444,11 +430,13 @@ void simcall_HANDLER_comm_waitany(smx_simcall_t simcall, xbt_dynar_t synchros, d
 
 void SIMIX_waitany_remove_simcall_from_actions(smx_simcall_t simcall)
 {
-  smx_activity_t synchro;
   unsigned int cursor = 0;
   xbt_dynar_t synchros = simcall_comm_waitany__get__comms(simcall);
 
-  xbt_dynar_foreach(synchros, cursor, synchro) {
+  simgrid::kernel::activity::ActivityImpl* ptr;
+  xbt_dynar_foreach(synchros, cursor, ptr){
+    smx_activity_t synchro = simgrid::kernel::activity::ActivityImplPtr(ptr);
+
     // Remove the first occurence of simcall:
     auto i = boost::range::find(synchro->simcalls, simcall);
     if (i !=  synchro->simcalls.end())
@@ -460,21 +448,19 @@ void SIMIX_waitany_remove_simcall_from_actions(smx_simcall_t simcall)
  *  \brief Starts the simulation of a communication synchro.
  *  \param synchro the communication synchro
  */
-static inline void SIMIX_comm_start(smx_activity_t synchro)
+static inline void SIMIX_comm_start(simgrid::kernel::activity::CommImplPtr comm)
 {
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
-
   /* If both the sender and the receiver are already there, start the communication */
-  if (synchro->state == SIMIX_READY) {
+  if (comm->state == SIMIX_READY) {
 
     simgrid::s4u::Host* sender   = comm->src_proc->host;
     simgrid::s4u::Host* receiver = comm->dst_proc->host;
 
     comm->surf_comm = surf_network_model->communicate(sender, receiver, comm->task_size, comm->rate);
-    comm->surf_comm->setData(synchro);
+    comm->surf_comm->setData(comm.get());
     comm->state = SIMIX_RUNNING;
 
-    XBT_DEBUG("Starting communication %p from '%s' to '%s' (surf_action: %p)", synchro, sender->cname(),
+    XBT_DEBUG("Starting communication %p from '%s' to '%s' (surf_action: %p)", comm.get(), sender->cname(),
               receiver->cname(), comm->surf_comm);
 
     /* If a link is failed, detect it immediately */
@@ -508,7 +494,8 @@ static inline void SIMIX_comm_start(smx_activity_t synchro)
  */
 void SIMIX_comm_finish(smx_activity_t synchro)
 {
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
 
   while (not synchro->simcalls.empty()) {
     smx_simcall_t simcall = synchro->simcalls.front();
@@ -546,7 +533,7 @@ void SIMIX_comm_finish(smx_activity_t synchro)
       switch (comm->state) {
 
         case SIMIX_DONE:
-          XBT_DEBUG("Communication %p complete!", synchro);
+          XBT_DEBUG("Communication %p complete!", synchro.get());
           SIMIX_comm_copy_data(synchro);
           break;
 
@@ -577,7 +564,7 @@ void SIMIX_comm_finish(smx_activity_t synchro)
         case SIMIX_LINK_FAILURE:
           XBT_DEBUG("Link failure in synchro %p between '%s' and '%s': posting an exception to the issuer: %s (%p) "
                     "detached:%d",
-                    synchro, comm->src_proc ? comm->src_proc->host->cname() : nullptr,
+                    synchro.get(), comm->src_proc ? comm->src_proc->host->cname() : nullptr,
                     comm->dst_proc ? comm->dst_proc->host->cname() : nullptr, simcall->issuer->cname(), simcall->issuer,
                     comm->detached);
           if (comm->src_proc == simcall->issuer) {
@@ -666,7 +653,8 @@ void SIMIX_comm_set_copy_data_callback(void (*callback) (smx_activity_t, void*,
 
 void SIMIX_comm_copy_pointer_callback(smx_activity_t synchro, void* buff, size_t buff_size)
 {
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
 
   xbt_assert((buff_size == sizeof(void *)), "Cannot copy %zu bytes: must be sizeof(void*)", buff_size);
   *(void **) (comm->dst_buff) = buff;
@@ -674,7 +662,8 @@ void SIMIX_comm_copy_pointer_callback(smx_activity_t synchro, void* buff, size_t
 
 void SIMIX_comm_copy_buffer_callback(smx_activity_t synchro, void* buff, size_t buff_size)
 {
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
 
   XBT_DEBUG("Copy the data over");
   memcpy(comm->dst_buff, buff, buff_size);
@@ -690,14 +679,15 @@ void SIMIX_comm_copy_buffer_callback(smx_activity_t synchro, void* buff, size_t
  */
 void SIMIX_comm_copy_data(smx_activity_t synchro)
 {
-  simgrid::kernel::activity::CommImpl* comm = static_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
 
   size_t buff_size = comm->src_buff_size;
   /* If there is no data to copy then return */
   if (not comm->src_buff || not comm->dst_buff || comm->copied)
     return;
 
-  XBT_DEBUG("Copying comm %p data from %s (%p) -> %s (%p) (%zu bytes)", comm,
+  XBT_DEBUG("Copying comm %p data from %s (%p) -> %s (%p) (%zu bytes)", comm.get(),
             comm->src_proc ? comm->src_proc->host->cname() : "a finished process", comm->src_buff,
             comm->dst_proc ? comm->dst_proc->host->cname() : "a finished process", comm->dst_buff, buff_size);
 
diff --git a/src/simix/smx_private.h b/src/simix/smx_private.h
index f600d3544a..602dfdfa8a 100644
--- a/src/simix/smx_private.h
+++ b/src/simix/smx_private.h
@@ -73,10 +73,8 @@ XBT_PUBLIC(void) SIMIX_clean();
   _smx_throw_issuer->exception = std::make_exception_ptr(e); \
   } else ((void)0)
 
-/* ******************************** File ************************************ */
 typedef struct s_smx_file {
   surf_file_t surf_file;
-  void* data;                   /**< @brief user data */
 } s_smx_file_t;
 
 
diff --git a/src/simix/smx_synchro.cpp b/src/simix/smx_synchro.cpp
index 2095a4c0b1..c4d9731312 100644
--- a/src/simix/smx_synchro.cpp
+++ b/src/simix/smx_synchro.cpp
@@ -27,9 +27,10 @@ static smx_activity_t SIMIX_synchro_wait(sg_host_t smx_host, double timeout)
 {
   XBT_IN("(%p, %f)",smx_host,timeout);
 
-  simgrid::kernel::activity::Raw *sync = new simgrid::kernel::activity::Raw();
+  simgrid::kernel::activity::RawImplPtr sync =
+      simgrid::kernel::activity::RawImplPtr(new simgrid::kernel::activity::RawImpl());
   sync->sleep                          = smx_host->pimpl_cpu->sleep(timeout);
-  sync->sleep->setData(sync);
+  sync->sleep->setData(&*sync);
   XBT_OUT();
   return sync;
 }
@@ -67,7 +68,7 @@ void SIMIX_synchro_stop_waiting(smx_actor_t process, smx_simcall_t simcall)
 
 void SIMIX_synchro_finish(smx_activity_t synchro)
 {
-  XBT_IN("(%p)",synchro);
+  XBT_IN("(%p)", synchro.get());
   smx_simcall_t simcall = synchro->simcalls.front();
   synchro->simcalls.pop_front();
 
@@ -89,7 +90,6 @@ void SIMIX_synchro_finish(smx_activity_t synchro)
 
   SIMIX_synchro_stop_waiting(simcall->issuer, simcall);
   simcall->issuer->waiting_synchro = nullptr;
-  delete synchro;
   SIMIX_simcall_answer(simcall);
   XBT_OUT();
 }
@@ -157,7 +157,7 @@ bool MutexImpl::try_lock(smx_actor_t issuer)
 
 /** Unlock a mutex for a process
  *
- * Unlocks the mutex and gives it to a process waiting for it. 
+ * Unlocks the mutex and gives it to a process waiting for it.
  * If the unlocker is not the owner of the mutex nothing happens.
  * If there are no process waiting, it sets the mutex as free.
  */
@@ -175,7 +175,6 @@ void MutexImpl::unlock(smx_actor_t issuer)
   if (xbt_swag_size(this->sleeping) > 0) {
     /*process to wake up */
     smx_actor_t p = (smx_actor_t) xbt_swag_extract(this->sleeping);
-    delete p->waiting_synchro;
     p->waiting_synchro = nullptr;
     this->owner = p;
     SIMIX_simcall_answer(&p->simcall);
@@ -293,14 +292,14 @@ static void _SIMIX_cond_wait(smx_cond_t cond, smx_mutex_t mutex, double timeout,
   synchro = SIMIX_synchro_wait(issuer->host, timeout);
   synchro->simcalls.push_front(simcall);
   issuer->waiting_synchro = synchro;
-  xbt_swag_insert(simcall->issuer, cond->sleeping);   
+  xbt_swag_insert(simcall->issuer, cond->sleeping);
   XBT_OUT();
 }
 
 /**
  * \brief Signalizes a condition.
  *
- * Signalizes a condition and wakes up a sleeping process. 
+ * Signalizes a condition and wakes up a sleeping process.
  * If there are no process sleeping, no action is done.
  * \param cond A condition
  */
@@ -313,12 +312,11 @@ void SIMIX_cond_signal(smx_cond_t cond)
 
   XBT_DEBUG("Signal condition %p", cond);
 
-  /* If there are processes waiting for the condition choose one and try 
+  /* If there are processes waiting for the condition choose one and try
      to make it acquire the mutex */
   if ((proc = (smx_actor_t) xbt_swag_extract(cond->sleeping))) {
 
     /* Destroy waiter's synchronization */
-    delete proc->waiting_synchro;
     proc->waiting_synchro = nullptr;
 
     /* Now transform the cond wait simcall into a mutex lock one */
@@ -431,7 +429,6 @@ void SIMIX_sem_release(smx_sem_t sem)
 
   XBT_DEBUG("Sem release semaphore %p", sem);
   if ((proc = (smx_actor_t) xbt_swag_extract(sem->sleeping))) {
-    delete proc->waiting_synchro;
     proc->waiting_synchro = nullptr;
     SIMIX_simcall_answer(&proc->simcall);
   } else {
@@ -496,7 +493,7 @@ void simcall_HANDLER_sem_acquire(smx_simcall_t simcall, smx_sem_t sem)
 void simcall_HANDLER_sem_acquire_timeout(smx_simcall_t simcall, smx_sem_t sem, double timeout)
 {
   XBT_IN("(%p)",simcall);
-  _SIMIX_sem_wait(sem, timeout, simcall->issuer, simcall);  
+  _SIMIX_sem_wait(sem, timeout, simcall->issuer, simcall);
   XBT_OUT();
 }
 int simcall_HANDLER_sem_would_block(smx_simcall_t simcall, smx_sem_t sem) {
diff --git a/src/smpi/colls/allgather/allgather-2dmesh.cpp b/src/smpi/colls/allgather/allgather-2dmesh.cpp
index 786b0af042..724373c86e 100644
--- a/src/smpi/colls/allgather/allgather-2dmesh.cpp
+++ b/src/smpi/colls/allgather/allgather-2dmesh.cpp
@@ -101,7 +101,7 @@ static int is_2dmesh(int num, int *i, int *j)
  * comm: communication
  * Descrp: Function realizes the allgather operation using the 2dmesh
  * algorithm. Allgather ommunication occurs first in the x dimension then in
- * the y dimension.  The communication in each dimension follows 
+ * the y dimension.  The communication in each dimension follows
  * "simple"
  * Auther: Ahmad Faraj
 ****************************************************************************/
@@ -144,7 +144,7 @@ Coll_allgather_2dmesh::allgather(void *send_buff, int send_count, MPI_Datatype
 
   req_ptr = req;
 
-  // do local allgather/local copy 
+  // do local allgather/local copy
   recv_offset = rank * block_size;
   Datatype::copy(send_buff, send_count, send_type, (char *)recv_buff + recv_offset,
                  recv_count, recv_type);
diff --git a/src/smpi/colls/allgather/allgather-3dmesh.cpp b/src/smpi/colls/allgather/allgather-3dmesh.cpp
index b97d22a5a1..58927c9dbb 100644
--- a/src/smpi/colls/allgather/allgather-3dmesh.cpp
+++ b/src/smpi/colls/allgather/allgather-3dmesh.cpp
@@ -137,12 +137,12 @@ int Coll_allgather_3dmesh::allgather(void *send_buff, int send_count,
 
   req_ptr = req;
 
-  // do local allgather/local copy 
+  // do local allgather/local copy
   recv_offset = rank * block_size;
   Datatype::copy(send_buff, send_count, send_type, (char *)recv_buff + recv_offset,
                  recv_count, recv_type);
 
-  // do rowwise comm 
+  // do rowwise comm
   for (i = 0; i < Y; i++) {
     src = i + my_row_base;
     if (src == rank)
diff --git a/src/smpi/colls/allgather/allgather-NTSLR-NB.cpp b/src/smpi/colls/allgather/allgather-NTSLR-NB.cpp
index ed4744e7d3..c4785d2c1b 100644
--- a/src/smpi/colls/allgather/allgather-NTSLR-NB.cpp
+++ b/src/smpi/colls/allgather/allgather-NTSLR-NB.cpp
@@ -33,9 +33,9 @@ Coll_allgather_NTSLR_NB::allgather(void *sbuf, int scount, MPI_Datatype stype,
 
   // irregular case use default MPI fucntions
   if (scount * sextent != rcount * rextent) {
-    XBT_WARN("MPI_allgather_NTSLR_NB use default MPI_allgather.");  
+    XBT_WARN("MPI_allgather_NTSLR_NB use default MPI_allgather.");
     Coll_allgather_default::allgather(sbuf, scount, stype, rbuf, rcount, rtype, comm);
-    return MPI_SUCCESS;    
+    return MPI_SUCCESS;
   }
 
   // topo non-specific
diff --git a/src/smpi/colls/allgather/allgather-NTSLR.cpp b/src/smpi/colls/allgather/allgather-NTSLR.cpp
index 72ecd14f67..a4a4b5e226 100644
--- a/src/smpi/colls/allgather/allgather-NTSLR.cpp
+++ b/src/smpi/colls/allgather/allgather-NTSLR.cpp
@@ -30,9 +30,9 @@ Coll_allgather_NTSLR::allgather(void *sbuf, int scount, MPI_Datatype stype,
 
   // irregular case use default MPI fucntions
   if (scount * sextent != rcount * rextent) {
-    XBT_WARN("MPI_allgather_NTSLR use default MPI_allgather.");  
+    XBT_WARN("MPI_allgather_NTSLR use default MPI_allgather.");
     Coll_allgather_default::allgather(sbuf, scount, stype, rbuf, rcount, rtype, comm);
-    return MPI_SUCCESS;    
+    return MPI_SUCCESS;
   }
 
   // topo non-specific
diff --git a/src/smpi/colls/allgather/allgather-SMP-NTS.cpp b/src/smpi/colls/allgather/allgather-SMP-NTS.cpp
index 366f74f4fd..91eb582fae 100644
--- a/src/smpi/colls/allgather/allgather-SMP-NTS.cpp
+++ b/src/smpi/colls/allgather/allgather-SMP-NTS.cpp
@@ -47,7 +47,7 @@ int Coll_allgather_SMP_NTS::allgather(void *sbuf, int scount,
   if (comm_size <= num_core) {
     XBT_WARN("MPI_allgather_SMP_NTS use default MPI_allgather.");
     Coll_allgather_default::allgather(sbuf, scount, stype, rbuf, rcount, rtype, comm);
-    return MPI_SUCCESS;    
+    return MPI_SUCCESS;
   }
 
   // the last SMP node may have fewer number of running processes than all others
@@ -78,7 +78,7 @@ int Coll_allgather_SMP_NTS::allgather(void *sbuf, int scount,
 
   }
 
-  // INTER-SMP-ALLGATHER 
+  // INTER-SMP-ALLGATHER
   // Every root of each SMP node post INTER-Sendrecv, then do INTRA-Bcast for each receiving message
   // Use logical ring algorithm
 
diff --git a/src/smpi/colls/allgather/allgather-bruck.cpp b/src/smpi/colls/allgather/allgather-bruck.cpp
index f2fd8b8b9f..f9b53fe4cb 100644
--- a/src/smpi/colls/allgather/allgather-bruck.cpp
+++ b/src/smpi/colls/allgather/allgather-bruck.cpp
@@ -65,7 +65,7 @@ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  *         algorithm.
  * Auther: MPICH
  * Comment: Original bruck algorithm from MPICH is slightly modified by
- *          Ahmad Faraj.  
+ *          Ahmad Faraj.
  ****************************************************************************/
 
 namespace simgrid{
@@ -92,7 +92,7 @@ int Coll_allgather_bruck::allgather(void *send_buff, int send_count,
   char *send_ptr = (char *) send_buff;
   char *recv_ptr = (char *) recv_buff;
 
-  // get size of the communicator, followed by rank 
+  // get size of the communicator, followed by rank
   num_procs = comm->size();
   rank = comm->rank();
 
diff --git a/src/smpi/colls/allgather/allgather-mvapich-smp.cpp b/src/smpi/colls/allgather/allgather-mvapich-smp.cpp
index 740c911f74..e30e455f18 100644
--- a/src/smpi/colls/allgather/allgather-mvapich-smp.cpp
+++ b/src/smpi/colls/allgather/allgather-mvapich-smp.cpp
@@ -44,7 +44,7 @@ int Coll_allgather_mvapich2_smp::allgather(void *sendbuf,int sendcnt, MPI_Dataty
 {
     int rank, size;
     int local_rank, local_size;
-    int leader_comm_size = 0; 
+    int leader_comm_size = 0;
     int mpi_errno = MPI_SUCCESS;
     MPI_Aint recvtype_extent = 0;  /* Datatype extent */
     MPI_Comm shmem_comm, leader_comm;
@@ -55,7 +55,7 @@ int Coll_allgather_mvapich2_smp::allgather(void *sendbuf,int sendcnt, MPI_Dataty
 
   if (not comm->is_uniform() || not comm->is_blocked())
     THROWF(arg_error,0, "allgather MVAPICH2 smp algorithm can't be used with irregular deployment. Please insure that processes deployed on the same node are contiguous and that each node has the same number of processes");
-  
+
     if (recvcnt == 0) {
         return MPI_SUCCESS;
     }
@@ -65,7 +65,7 @@ int Coll_allgather_mvapich2_smp::allgather(void *sendbuf,int sendcnt, MPI_Dataty
 
     /* extract the rank,size information for the intra-node communicator */
     recvtype_extent=recvtype->get_extent();
-    
+
     shmem_comm = comm->get_intra_comm();
     local_rank = shmem_comm->rank();
     local_size = shmem_comm->size();
@@ -82,20 +82,20 @@ int Coll_allgather_mvapich2_smp::allgather(void *sendbuf,int sendcnt, MPI_Dataty
     /*If there is just one node, after gather itself,
      * root has all the data and it can do bcast*/
     if(local_rank == 0) {
-        mpi_errno = Colls::gather(sendbuf, sendcnt,sendtype, 
-                                    (void*)((char*)recvbuf + (rank * recvcnt * recvtype_extent)), 
+        mpi_errno = Colls::gather(sendbuf, sendcnt,sendtype,
+                                    (void*)((char*)recvbuf + (rank * recvcnt * recvtype_extent)),
                                      recvcnt, recvtype,
                                      0, shmem_comm);
     } else {
-        /*Since in allgather all the processes could have 
+        /*Since in allgather all the processes could have
          * its own data in place*/
         if(sendbuf == MPI_IN_PLACE) {
-            mpi_errno = Colls::gather((void*)((char*)recvbuf + (rank * recvcnt * recvtype_extent)), 
-                                         recvcnt , recvtype, 
+            mpi_errno = Colls::gather((void*)((char*)recvbuf + (rank * recvcnt * recvtype_extent)),
+                                         recvcnt , recvtype,
                                          recvbuf, recvcnt, recvtype,
                                          0, shmem_comm);
         } else {
-            mpi_errno = Colls::gather(sendbuf, sendcnt,sendtype, 
+            mpi_errno = Colls::gather(sendbuf, sendcnt,sendtype,
                                          recvbuf, recvcnt, recvtype,
                                          0, shmem_comm);
         }
@@ -130,7 +130,7 @@ int Coll_allgather_mvapich2_smp::allgather(void *sendbuf,int sendcnt, MPI_Dataty
 
             mpi_errno = Colls::allgatherv(sendbuf,
                                        (recvcnt*local_size),
-                                       recvtype, 
+                                       recvtype,
                                        recvbuf, recvcnts,
                                        displs, recvtype,
                                        leader_comm);
@@ -138,10 +138,10 @@ int Coll_allgather_mvapich2_smp::allgather(void *sendbuf,int sendcnt, MPI_Dataty
             xbt_free(recvcnts);
         } else {
         void* sendtmpbuf=((char*)recvbuf)+recvtype->get_extent()*(recvcnt*local_size)*leader_comm->rank();
-        
-          
 
-            mpi_errno = Coll_allgather_mpich::allgather(sendtmpbuf, 
+
+
+            mpi_errno = Coll_allgather_mpich::allgather(sendtmpbuf,
                                                (recvcnt*local_size),
                                                recvtype,
                                                recvbuf, (recvcnt*local_size), recvtype,
diff --git a/src/smpi/colls/allgather/allgather-ompi-neighborexchange.cpp b/src/smpi/colls/allgather/allgather-ompi-neighborexchange.cpp
index cb350c6df9..4f2707e700 100644
--- a/src/smpi/colls/allgather/allgather-ompi-neighborexchange.cpp
+++ b/src/smpi/colls/allgather/allgather-ompi-neighborexchange.cpp
@@ -12,21 +12,21 @@
  * Returns:      MPI_SUCCESS or error code
  *
  * Description:  Neighbor Exchange algorithm for allgather.
- *               Described by Chen et.al. in 
- *               "Performance Evaluation of Allgather Algorithms on 
+ *               Described by Chen et.al. in
+ *               "Performance Evaluation of Allgather Algorithms on
  *                Terascale Linux Cluster with Fast Ethernet",
- *               Proceedings of the Eighth International Conference on 
+ *               Proceedings of the Eighth International Conference on
  *               High-Performance Computing inn Asia-Pacific Region
  *               (HPCASIA'05), 2005
- * 
+ *
  *               Rank r exchanges message with one of its neighbors and
  *               forwards the data further in the next step.
  *
  *               No additional memory requirements.
- * 
+ *
  * Limitations:  Algorithm works only on even number of processes.
  *               For odd number of processes we switch to ring algorithm.
- * 
+ *
  * Example on 6 nodes:
  *  Initial state
  *    #     0      1      2      3      4      5
@@ -61,13 +61,13 @@
  *         [4]    [4]    [4]    [4]    [4]    [4]
  *         [5]    [5]    [5]    [5]    [5]    [5]
  */
- 
+
  #include "../colls_private.h"
 
 namespace simgrid{
 namespace smpi{
 
-int 
+int
 Coll_allgather_ompi_neighborexchange::allgather(void *sbuf, int scount,
                                                  MPI_Datatype sdtype,
                                                  void* rbuf, int rcount,
@@ -88,7 +88,7 @@ Coll_allgather_ompi_neighborexchange::allgather(void *sbuf, int scount,
 
    if (size % 2) {
       XBT_DEBUG(
-                   "coll:tuned:allgather_intra_neighborexchange WARNING: odd size %d, switching to ring algorithm", 
+                   "coll:tuned:allgather_intra_neighborexchange WARNING: odd size %d, switching to ring algorithm",
                    size);
       return Coll_allgather_ring::allgather(sbuf, scount, sdtype,
                                                   rbuf, rcount, rdtype,
@@ -112,7 +112,7 @@ Coll_allgather_ompi_neighborexchange::allgather(void *sbuf, int scount,
    if (MPI_IN_PLACE != sbuf) {
       tmpsend = (char*) sbuf;
       Datatype::copy (tmpsend, scount, sdtype, tmprecv, rcount, rdtype);
-   } 
+   }
 
    /* Determine neighbors, order in which blocks will arrive, etc. */
    even_rank = not(rank % 2);
@@ -134,8 +134,8 @@ Coll_allgather_ompi_neighborexchange::allgather(void *sbuf, int scount,
 
    /* Communication loop:
       - First step is special: exchange a single block with neighbor[0].
-      - Rest of the steps: 
-        update recv_data_from according to offset, and 
+      - Rest of the steps:
+        update recv_data_from according to offset, and
         exchange two blocks with appropriate neighbor.
         the send location becomes previous receve location.
    */
@@ -157,15 +157,15 @@ Coll_allgather_ompi_neighborexchange::allgather(void *sbuf, int scount,
 
    for (i = 1; i < (size / 2); i++) {
       const int i_parity = i % 2;
-      recv_data_from[i_parity] = 
+      recv_data_from[i_parity] =
          (recv_data_from[i_parity] + offset_at_step[i_parity] + size) % size;
 
       tmprecv = (char*)rbuf + recv_data_from[i_parity] * rcount * rext;
       tmpsend = (char*)rbuf + send_data_from * rcount * rext;
-      
+
       /* Sendreceive */
-      Request::sendrecv(tmpsend, 2 * rcount, rdtype, 
-                                     neighbor[i_parity], 
+      Request::sendrecv(tmpsend, 2 * rcount, rdtype,
+                                     neighbor[i_parity],
                                      COLL_TAG_ALLGATHER,
                                      tmprecv, 2 * rcount, rdtype,
                                      neighbor[i_parity],
diff --git a/src/smpi/colls/allgather/allgather-rdb.cpp b/src/smpi/colls/allgather/allgather-rdb.cpp
index 44dc246a7d..1c028e8178 100644
--- a/src/smpi/colls/allgather/allgather-rdb.cpp
+++ b/src/smpi/colls/allgather/allgather-rdb.cpp
@@ -32,7 +32,7 @@ Coll_allgather_rdb::allgather(void *sbuf, int send_count,
   char *send_ptr = (char *) sbuf;
   char *recv_ptr = (char *) rbuf;
 
-  // get size of the communicator, followed by rank 
+  // get size of the communicator, followed by rank
   unsigned int num_procs = comm->size();
   unsigned int rank = comm->rank();
 
diff --git a/src/smpi/colls/allgather/allgather-rhv.cpp b/src/smpi/colls/allgather/allgather-rhv.cpp
index 2311385451..1c50923087 100644
--- a/src/smpi/colls/allgather/allgather-rhv.cpp
+++ b/src/smpi/colls/allgather/allgather-rhv.cpp
@@ -28,7 +28,7 @@ Coll_allgather_rhv::allgather(void *sbuf, int send_count,
   unsigned int mask;
   int curr_count;
 
-  // get size of the communicator, followed by rank 
+  // get size of the communicator, followed by rank
   unsigned int num_procs = comm->size();
 
   if((num_procs&(num_procs-1)))
@@ -45,10 +45,10 @@ Coll_allgather_rhv::allgather(void *sbuf, int send_count,
   recv_chunk = r_extent * recv_count;
 
   if (send_chunk != recv_chunk) {
-    XBT_WARN("MPI_allgather_rhv use default MPI_allgather.");  
+    XBT_WARN("MPI_allgather_rhv use default MPI_allgather.");
     Coll_allgather_default::allgather(sbuf, send_count, send_type, rbuf, recv_count,
                               recv_type, comm);
-    return MPI_SUCCESS;        
+    return MPI_SUCCESS;
   }
 
   // compute starting offset location to perform local copy
diff --git a/src/smpi/colls/allgather/allgather-smp-simple.cpp b/src/smpi/colls/allgather/allgather-smp-simple.cpp
index 030733b7ea..42dad102d3 100644
--- a/src/smpi/colls/allgather/allgather-smp-simple.cpp
+++ b/src/smpi/colls/allgather/allgather-smp-simple.cpp
@@ -67,7 +67,7 @@ int Coll_allgather_smp_simple::allgather(void *send_buf, int scount,
 
   }
 
-  // INTER-SMP-ALLGATHER 
+  // INTER-SMP-ALLGATHER
   // Every root of each SMP node post INTER-Sendrecv, then do INTRA-Bcast for each receiving message
 
 
@@ -86,7 +86,7 @@ int Coll_allgather_smp_simple::allgather(void *send_buf, int scount,
       src = ((inter_rank - i + inter_comm_size) % inter_comm_size) * num_core;
       //send_offset = (rank * sextent * scount);
       recv_offset = (src * sextent * scount);
-      //      Request::sendrecv((recv_buf+send_offset), (scount * num_core), stype, dst, tag, 
+      //      Request::sendrecv((recv_buf+send_offset), (scount * num_core), stype, dst, tag,
       //             (recv_buf+recv_offset), (rcount * num_core), rtype, src, tag, comm, &status);
       //MPIC_Isend((recv_buf+send_offset), (scount * num_core), stype, dst, tag, comm, req_ptr++);
       *(req_ptr++) = Request::irecv(((char *) recv_buf + recv_offset), (rcount * num_core), rtype,
@@ -98,7 +98,7 @@ int Coll_allgather_smp_simple::allgather(void *send_buf, int scount,
       //src = ((inter_rank-i+inter_comm_size)%inter_comm_size) * num_core;
       send_offset = (rank * sextent * scount);
       //recv_offset = (src * sextent * scount);
-      //      Request::sendrecv((recv_buf+send_offset), (scount * num_core), stype, dst, tag, 
+      //      Request::sendrecv((recv_buf+send_offset), (scount * num_core), stype, dst, tag,
       //             (recv_buf+recv_offset), (rcount * num_core), rtype, src, tag, comm, &status);
       *(req_ptr++) = Request::isend(((char *) recv_buf + send_offset), (scount * num_core), stype,
                 dst, tag, comm);
diff --git a/src/smpi/colls/allgatherv/allgatherv-ompi-bruck.cpp b/src/smpi/colls/allgatherv/allgatherv-ompi-bruck.cpp
index ff68e9d1e1..3874163fcc 100644
--- a/src/smpi/colls/allgatherv/allgatherv-ompi-bruck.cpp
+++ b/src/smpi/colls/allgatherv/allgatherv-ompi-bruck.cpp
@@ -33,7 +33,7 @@
  *                in Multiport Message-Passing Systems"
  * Note:         Unlike in case of allgather implementation, we relay on
  *               indexed datatype to select buffers appropriately.
- *               The only additional memory requirement is for creation of 
+ *               The only additional memory requirement is for creation of
  *               temporary datatypes.
  * Example on 7 nodes (memory lay out need not be in-order)
  *   Initial set up:
@@ -55,7 +55,7 @@
  *         [ ]    [ ]    [ ]    [ ]    [5]    [5]    [ ]
  *         [ ]    [ ]    [ ]    [ ]    [ ]    [6]    [6]
  *   Step 1: send message to (rank - 2^1), receive message from (rank + 2^1).
- *           message contains all blocks from (rank) .. (rank + 2^2) with 
+ *           message contains all blocks from (rank) .. (rank + 2^2) with
  *           wrap around.
  *    #     0      1      2      3      4      5      6
  *         [0]    [ ]    [ ]    [ ]    [0]    [0]    [0]
@@ -66,7 +66,7 @@
  *         [ ]    [ ]    [5]    [5]    [5]    [5]    [ ]
  *         [ ]    [ ]    [ ]    [6]    [6]    [6]    [6]
  *   Step 2: send message to (rank - 2^2), receive message from (rank + 2^2).
- *           message size is "all remaining blocks" 
+ *           message size is "all remaining blocks"
  *    #     0      1      2      3      4      5      6
  *         [0]    [0]    [0]    [0]    [0]    [0]    [0]
  *         [1]    [1]    [1]    [1]    [1]    [1]    [1]
@@ -83,7 +83,7 @@ namespace smpi{
 int Coll_allgatherv_ompi_bruck::allgatherv(void *sbuf, int scount,
                                            MPI_Datatype sdtype,
                                            void *rbuf, int *rcounts,
-                                           int *rdispls, 
+                                           int *rdispls,
                                            MPI_Datatype rdtype,
                                            MPI_Comm comm)
 {
@@ -100,31 +100,31 @@ int Coll_allgatherv_ompi_bruck::allgatherv(void *sbuf, int scount,
 
    XBT_DEBUG(
                 "coll:tuned:allgather_ompi_bruck rank %d", rank);
-   
+
    sdtype->extent(&slb, &sext);
 
    rdtype->extent(&rlb, &rext);
 
    /* Initialization step:
-      - if send buffer is not MPI_IN_PLACE, copy send buffer to block rank of 
+      - if send buffer is not MPI_IN_PLACE, copy send buffer to block rank of
         the receive buffer.
    */
    tmprecv = (char*) rbuf + rdispls[rank] * rext;
    if (MPI_IN_PLACE != sbuf) {
       tmpsend = (char*) sbuf;
-      Datatype::copy(tmpsend, scount, sdtype, 
+      Datatype::copy(tmpsend, scount, sdtype,
                             tmprecv, rcounts[rank], rdtype);
    }
-   
+
    /* Communication step:
       At every step i, rank r:
       - doubles the distance
       - sends message with blockcount blocks, (rbuf[rank] .. rbuf[rank + 2^i])
         to rank (r - distance)
-      - receives message of blockcount blocks, 
-        (rbuf[r + distance] ... rbuf[(r+distance) + 2^i]) from 
+      - receives message of blockcount blocks,
+        (rbuf[r + distance] ... rbuf[(r+distance) + 2^i]) from
         rank (r + distance)
-      - blockcount doubles until the last step when only the remaining data is 
+      - blockcount doubles until the last step when only the remaining data is
       exchanged.
    */
    new_rcounts = (int*) calloc(4*size, sizeof(int));
@@ -139,7 +139,7 @@ int Coll_allgatherv_ompi_bruck::allgatherv(void *sbuf, int scount,
 
       if (distance <= (size >> 1)) {
          blockcount = distance;
-      } else { 
+      } else {
          blockcount = size - distance;
       }
 
@@ -152,7 +152,7 @@ int Coll_allgatherv_ompi_bruck::allgatherv(void *sbuf, int scount,
           new_rcounts[i] = rcounts[tmp_rrank];
           new_rdispls[i] = rdispls[tmp_rrank];
       }
-      Datatype::create_indexed(blockcount, new_scounts, new_sdispls, 
+      Datatype::create_indexed(blockcount, new_scounts, new_sdispls,
                                     rdtype, &new_sdtype);
       Datatype::create_indexed(blockcount, new_rcounts, new_rdispls,
                                     rdtype, &new_rdtype);
diff --git a/src/smpi/colls/allgatherv/allgatherv-ompi-neighborexchange.cpp b/src/smpi/colls/allgatherv/allgatherv-ompi-neighborexchange.cpp
index 3a4b71d279..e88c1882d0 100644
--- a/src/smpi/colls/allgatherv/allgatherv-ompi-neighborexchange.cpp
+++ b/src/smpi/colls/allgatherv/allgatherv-ompi-neighborexchange.cpp
@@ -11,23 +11,23 @@
  * Accepts:      Same arguments as MPI_Allgatherv
  * Returns:      MPI_SUCCESS or error code
  *
- * Description:  Neighbor Exchange algorithm for allgather adapted for 
+ * Description:  Neighbor Exchange algorithm for allgather adapted for
  *               allgatherv.
- *               Described by Chen et.al. in 
- *               "Performance Evaluation of Allgather Algorithms on 
+ *               Described by Chen et.al. in
+ *               "Performance Evaluation of Allgather Algorithms on
  *                Terascale Linux Cluster with Fast Ethernet",
- *               Proceedings of the Eighth International Conference on 
+ *               Proceedings of the Eighth International Conference on
  *               High-Performance Computing inn Asia-Pacific Region
  *               (HPCASIA'05), 2005
- * 
+ *
  *               Rank r exchanges message with one of its neighbors and
  *               forwards the data further in the next step.
  *
  *               No additional memory requirements.
- * 
+ *
  * Limitations:  Algorithm works only on even number of processes.
  *               For odd number of processes we switch to ring algorithm.
- * 
+ *
  * Example on 6 nodes:
  *  Initial state
  *    #     0      1      2      3      4      5
@@ -62,13 +62,13 @@
  *         [4]    [4]    [4]    [4]    [4]    [4]
  *         [5]    [5]    [5]    [5]    [5]    [5]
  */
- 
+
  #include "../colls_private.h"
- 
+
 namespace simgrid{
 namespace smpi{
 
-int 
+int
 Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
                                                   MPI_Datatype sdtype,
                                                   void* rbuf, int *rcounts, int *rdispls,
@@ -78,7 +78,7 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
     int line = -1;
     int rank, size;
     int neighbor[2], offset_at_step[2], recv_data_from[2], send_data_from;
-  
+
     int i, even_rank;
     int err = 0;
     ptrdiff_t slb, rlb, sext, rext;
@@ -90,10 +90,10 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
 
     if (size % 2) {
         XBT_DEBUG(
-                     "coll:tuned:allgatherv_ompi_neighborexchange WARNING: odd size %d, switching to ring algorithm", 
+                     "coll:tuned:allgatherv_ompi_neighborexchange WARNING: odd size %d, switching to ring algorithm",
                      size);
         return Coll_allgatherv_ring::allgatherv(sbuf, scount, sdtype,
-                                                     rbuf, rcounts, 
+                                                     rbuf, rcounts,
                                                      rdispls, rdtype,
                                                      comm);
     }
@@ -108,16 +108,16 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
     if (MPI_SUCCESS != err) { line = __LINE__; goto err_hndl; }
 
     /* Initialization step:
-       - if send buffer is not MPI_IN_PLACE, copy send buffer to 
+       - if send buffer is not MPI_IN_PLACE, copy send buffer to
        the appropriate block of receive buffer
     */
     tmprecv = (char*) rbuf + rdispls[rank] * rext;
     if (MPI_IN_PLACE != sbuf) {
         tmpsend = (char*) sbuf;
-        err = Datatype::copy(tmpsend, scount, sdtype, 
+        err = Datatype::copy(tmpsend, scount, sdtype,
                               tmprecv, rcounts[rank], rdtype);
         if (MPI_SUCCESS != err) { line = __LINE__; goto err_hndl;  }
-    } 
+    }
 
     /* Determine neighbors, order in which blocks will arrive, etc. */
     even_rank = !(rank % 2);
@@ -139,8 +139,8 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
 
     /* Communication loop:
        - First step is special: exchange a single block with neighbor[0].
-       - Rest of the steps: 
-       update recv_data_from according to offset, and 
+       - Rest of the steps:
+       update recv_data_from according to offset, and
        exchange two blocks with appropriate neighbor.
        the send location becomes previous receve location.
        Note, we need to create indexed datatype to send and receive these
@@ -148,16 +148,16 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
     */
     tmprecv = (char*)rbuf + rdispls[neighbor[0]] * rext;
     tmpsend = (char*)rbuf + rdispls[rank] * rext;
-    Request::sendrecv(tmpsend, rcounts[rank], rdtype, 
+    Request::sendrecv(tmpsend, rcounts[rank], rdtype,
                                    neighbor[0], COLL_TAG_ALLGATHERV,
-                                   tmprecv, rcounts[neighbor[0]], rdtype, 
+                                   tmprecv, rcounts[neighbor[0]], rdtype,
                                    neighbor[0], COLL_TAG_ALLGATHERV,
                                    comm, MPI_STATUS_IGNORE);
 
 
 
-  
-   
+
+
     /* Determine initial sending counts and displacements*/
     if (even_rank) {
         send_data_from = rank;
@@ -169,7 +169,7 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
         MPI_Datatype new_rdtype, new_sdtype;
         int new_scounts[2], new_sdispls[2], new_rcounts[2], new_rdispls[2];
         const int i_parity = i % 2;
-        recv_data_from[i_parity] = 
+        recv_data_from[i_parity] =
             (recv_data_from[i_parity] + offset_at_step[i_parity] + size) % size;
 
         /* Create new indexed types for sending and receiving.
@@ -177,12 +177,12 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
            We are receiving data from ranks (recv_data_from[i_parity]) and
            (recv_data_from[i_parity]+1).
         */
-        
+
         new_scounts[0] = rcounts[send_data_from];
         new_scounts[1] = rcounts[(send_data_from + 1)];
         new_sdispls[0] = rdispls[send_data_from];
         new_sdispls[1] = rdispls[(send_data_from + 1)];
-        err = Datatype::create_indexed(2, new_scounts, new_sdispls, rdtype, 
+        err = Datatype::create_indexed(2, new_scounts, new_sdispls, rdtype,
                                       &new_sdtype);
         if (MPI_SUCCESS != err) { line = __LINE__; goto err_hndl; }
         new_sdtype->commit();
@@ -191,14 +191,14 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
         new_rcounts[1] = rcounts[(recv_data_from[i_parity] + 1)];
         new_rdispls[0] = rdispls[recv_data_from[i_parity]];
         new_rdispls[1] = rdispls[(recv_data_from[i_parity] + 1)];
-        err = Datatype::create_indexed(2, new_rcounts, new_rdispls, rdtype, 
+        err = Datatype::create_indexed(2, new_rcounts, new_rdispls, rdtype,
                                       &new_rdtype);
         if (MPI_SUCCESS != err) { line = __LINE__; goto err_hndl; }
         new_rdtype->commit();
-      
+
         tmprecv = (char*)rbuf;
         tmpsend = (char*)rbuf;
-      
+
         /* Sendreceive */
         Request::sendrecv(tmpsend, 1, new_sdtype, neighbor[i_parity],
                                        COLL_TAG_ALLGATHERV,
@@ -207,7 +207,7 @@ Coll_allgatherv_ompi_neighborexchange::allgatherv(void *sbuf, int scount,
                                        comm, MPI_STATUS_IGNORE);
 
         send_data_from = recv_data_from[i_parity];
-      
+
         Datatype::unref(new_sdtype);
         Datatype::unref(new_rdtype);
     }
diff --git a/src/smpi/colls/allgatherv/allgatherv-ring.cpp b/src/smpi/colls/allgatherv/allgatherv-ring.cpp
index ee91331a64..8f4e17e397 100644
--- a/src/smpi/colls/allgatherv/allgatherv-ring.cpp
+++ b/src/smpi/colls/allgatherv/allgatherv-ring.cpp
@@ -94,7 +94,7 @@ Coll_allgatherv_ring::allgatherv(void *send_buff, int send_count,
   for (i = 1; i < num_procs; i++) {
     src = (rank - i + num_procs) % num_procs;
     dst = (rank + i) % num_procs;
-    
+
     Request::sendrecv(sendptr, send_count, send_type, dst, tag,
                  recvptr + recv_disps[src] * extent, recv_counts[src], recv_type,
                  src, tag, comm, &status);
diff --git a/src/smpi/colls/allreduce/allreduce-lr.cpp b/src/smpi/colls/allreduce/allreduce-lr.cpp
index 1b9f5afff8..c9e10abef4 100644
--- a/src/smpi/colls/allreduce/allreduce-lr.cpp
+++ b/src/smpi/colls/allreduce/allreduce-lr.cpp
@@ -6,8 +6,8 @@
 
 #include "../colls_private.h"
 
-/* IMPLEMENTED BY PITCH PATARASUK 
-   Non-topoloty-specific all-reduce operation designed bandwidth optimally 
+/* IMPLEMENTED BY PITCH PATARASUK
+   Non-topoloty-specific all-reduce operation designed bandwidth optimally
    Bug fixing by Xin Yuan, 04/04/2008
 */
 
@@ -41,10 +41,10 @@ Coll_allreduce_lr::allreduce(void *sbuf, void *rbuf, int rcount,
   if (rcount < size) {
     XBT_WARN("MPI_allreduce_lr use default MPI_allreduce.");
     Coll_allreduce_default::allreduce(sbuf, rbuf, rcount, dtype, op, comm);
-    return MPI_SUCCESS; 
+    return MPI_SUCCESS;
   }
 
-  /* when communication size is not divisible by number of process: 
+  /* when communication size is not divisible by number of process:
      call the native implementation for the remain chunk at the end of the operation */
   if (rcount % size != 0) {
     remainder = rcount % size;
@@ -62,7 +62,7 @@ Coll_allreduce_lr::allreduce(void *sbuf, void *rbuf, int rcount,
   /* our ALL-REDUCE implementation
      1. copy (partial of)send_buf to recv_buf
      2. use logical ring reduce-scatter
-     3. use logical ring all-gather 
+     3. use logical ring all-gather
    */
 
   // copy partial data
@@ -95,7 +95,7 @@ Coll_allreduce_lr::allreduce(void *sbuf, void *rbuf, int rcount,
                  ((rank + size - 1) % size), tag + i, comm, &status);
   }
 
-  /* when communication size is not divisible by number of process: 
+  /* when communication size is not divisible by number of process:
      call the native implementation for the remain chunk at the end of the operation */
   if (remainder_flag) {
     return Colls::allreduce((char *) sbuf + remainder_offset,
diff --git a/src/smpi/colls/allreduce/allreduce-mvapich-two-level.cpp b/src/smpi/colls/allreduce/allreduce-mvapich-two-level.cpp
index c20f819271..730307b0c1 100644
--- a/src/smpi/colls/allreduce/allreduce-mvapich-two-level.cpp
+++ b/src/smpi/colls/allreduce/allreduce-mvapich-two-level.cpp
@@ -52,8 +52,8 @@ extern int (*MV2_Allreduce_intra_function)( void *sendbuf,
     int count,
     MPI_Datatype datatype,
     MPI_Op op, MPI_Comm comm);
-    
-    
+
+
 namespace simgrid{
 namespace smpi{
 static  int MPIR_Allreduce_reduce_p2p_MV2( void *sendbuf,
@@ -76,7 +76,7 @@ static  int MPIR_Allreduce_reduce_shmem_MV2( void *sendbuf,
   return MPI_SUCCESS;
 }
 
-    
+
 /* general two level allreduce helper function */
 int Coll_allreduce_mvapich2_two_level::allreduce(void *sendbuf,
                              void *recvbuf,
@@ -95,11 +95,11 @@ int Coll_allreduce_mvapich2_two_level::allreduce(void *sendbuf,
       MV2_Allreduce_intra_function = Coll_allreduce_mpich::allreduce;
     if(MV2_Allreducection==NULL)
       MV2_Allreducection = Coll_allreduce_rdb::allreduce;
-    
+
     if(comm->get_leaders_comm()==MPI_COMM_NULL){
       comm->init_smp();
     }
-  
+
     if (count == 0) {
         return MPI_SUCCESS;
     }
@@ -122,7 +122,7 @@ int Coll_allreduce_mvapich2_two_level::allreduce(void *sendbuf,
 
     /* Doing the shared memory gather and reduction by the leader */
     if (local_rank == 0) {
-        if ((MV2_Allreduce_intra_function == &MPIR_Allreduce_reduce_shmem_MV2) || 
+        if ((MV2_Allreduce_intra_function == &MPIR_Allreduce_reduce_shmem_MV2) ||
               (MV2_Allreduce_intra_function == &MPIR_Allreduce_reduce_p2p_MV2) ) {
         mpi_errno =
         MV2_Allreduce_intra_function(sendbuf, recvbuf, count, datatype,
@@ -151,7 +151,7 @@ int Coll_allreduce_mvapich2_two_level::allreduce(void *sendbuf,
         }
     } else {
         /* insert the first reduce here */
-        if ((MV2_Allreduce_intra_function == &MPIR_Allreduce_reduce_shmem_MV2) || 
+        if ((MV2_Allreduce_intra_function == &MPIR_Allreduce_reduce_shmem_MV2) ||
               (MV2_Allreduce_intra_function == &MPIR_Allreduce_reduce_p2p_MV2) ) {
         mpi_errno =
         MV2_Allreduce_intra_function(sendbuf, recvbuf, count, datatype,
diff --git a/src/smpi/colls/allreduce/allreduce-ompi-ring-segmented.cpp b/src/smpi/colls/allreduce/allreduce-ompi-ring-segmented.cpp
index f4f2357b79..2736daf63e 100644
--- a/src/smpi/colls/allreduce/allreduce-ompi-ring-segmented.cpp
+++ b/src/smpi/colls/allreduce/allreduce-ompi-ring-segmented.cpp
@@ -62,24 +62,24 @@
  *   Accepts:        Same as MPI_Allreduce(), segment size
  *   Returns:        MPI_SUCCESS or error code
  *
- *   Description:    Implements pipelined ring algorithm for allreduce: 
+ *   Description:    Implements pipelined ring algorithm for allreduce:
  *                   user supplies suggested segment size for the pipelining of
  *                   reduce operation.
- *                   The segment size determines the number of phases, np, for 
- *                   the algorithm execution.  
- *                   The message is automatically divided into blocks of 
+ *                   The segment size determines the number of phases, np, for
+ *                   the algorithm execution.
+ *                   The message is automatically divided into blocks of
  *                   approximately  (count / (np * segcount)) elements.
- *                   At the end of reduction phase, allgather like step is 
+ *                   At the end of reduction phase, allgather like step is
  *                   executed.
  *                   Algorithm requires (np + 1)*(N - 1) steps.
  *
- *   Limitations:    The algorithm DOES NOT preserve order of operations so it 
+ *   Limitations:    The algorithm DOES NOT preserve order of operations so it
  *                   can be used only for commutative operations.
- *                   In addition, algorithm cannot work if the total size is 
+ *                   In addition, algorithm cannot work if the total size is
  *                   less than size * segment size.
  *         Example on 3 nodes with 2 phases
  *         Initial state
- *   #      0              1             2 
+ *   #      0              1             2
  *        [00a]          [10a]         [20a]
  *        [00b]          [10b]         [20b]
  *        [01a]          [11a]         [21a]
@@ -88,9 +88,9 @@
  *        [02b]          [12b]         [22b]
  *
  *        COMPUTATION PHASE 0 (a)
- *         Step 0: rank r sends block ra to rank (r+1) and receives bloc (r-1)a 
+ *         Step 0: rank r sends block ra to rank (r+1) and receives bloc (r-1)a
  *                 from rank (r-1) [with wraparound].
- *    #     0              1             2  
+ *    #     0              1             2
  *        [00a]        [00a+10a]       [20a]
  *        [00b]          [10b]         [20b]
  *        [01a]          [11a]       [11a+21a]
@@ -98,20 +98,20 @@
  *      [22a+02a]        [12a]         [22a]
  *        [02b]          [12b]         [22b]
  *
- *         Step 1: rank r sends block (r-1)a to rank (r+1) and receives bloc 
+ *         Step 1: rank r sends block (r-1)a to rank (r+1) and receives bloc
  *                 (r-2)a from rank (r-1) [with wraparound].
- *    #     0              1             2  
+ *    #     0              1             2
  *        [00a]        [00a+10a]   [00a+10a+20a]
  *        [00b]          [10b]         [20b]
  *    [11a+21a+01a]      [11a]       [11a+21a]
  *        [01b]          [11b]         [21b]
  *      [22a+02a]    [22a+02a+12a]     [22a]
- *        [02b]          [12b]         [22b] 
+ *        [02b]          [12b]         [22b]
  *
  *        COMPUTATION PHASE 1 (b)
- *         Step 0: rank r sends block rb to rank (r+1) and receives bloc (r-1)b 
+ *         Step 0: rank r sends block rb to rank (r+1) and receives bloc (r-1)b
  *                 from rank (r-1) [with wraparound].
- *    #     0              1             2  
+ *    #     0              1             2
  *        [00a]        [00a+10a]       [20a]
  *        [00b]        [00b+10b]       [20b]
  *        [01a]          [11a]       [11a+21a]
@@ -119,22 +119,22 @@
  *      [22a+02a]        [12a]         [22a]
  *      [22b+02b]        [12b]         [22b]
  *
- *         Step 1: rank r sends block (r-1)b to rank (r+1) and receives bloc 
+ *         Step 1: rank r sends block (r-1)b to rank (r+1) and receives bloc
  *                 (r-2)b from rank (r-1) [with wraparound].
- *    #     0              1             2  
+ *    #     0              1             2
  *        [00a]        [00a+10a]   [00a+10a+20a]
  *        [00b]          [10b]     [0bb+10b+20b]
  *    [11a+21a+01a]      [11a]       [11a+21a]
  *    [11b+21b+01b]      [11b]         [21b]
  *      [22a+02a]    [22a+02a+12a]     [22a]
- *        [02b]      [22b+01b+12b]     [22b] 
+ *        [02b]      [22b+01b+12b]     [22b]
+ *
  *
- *         
  *        DISTRIBUTION PHASE: ring ALLGATHER with ranks shifted by 1 (same as
  *         in regular ring algorithm.
  *
  */
- 
+
 #define COLL_TUNED_COMPUTED_SEGCOUNT(SEGSIZE, TYPELNG, SEGCOUNT)        \
     if( ((SEGSIZE) >= (TYPELNG)) &&                                     \
         ((SEGSIZE) < ((TYPELNG) * (SEGCOUNT))) ) {                      \
@@ -144,7 +144,7 @@
         if( residual > ((TYPELNG) >> 1) )                               \
             (SEGCOUNT)++;                                               \
     }                                                                   \
-    
+
 #define COLL_TUNED_COMPUTE_BLOCKCOUNT( COUNT, NUM_BLOCKS, SPLIT_INDEX,       \
                                        EARLY_BLOCK_COUNT, LATE_BLOCK_COUNT ) \
     EARLY_BLOCK_COUNT = LATE_BLOCK_COUNT = COUNT / NUM_BLOCKS;               \
@@ -157,16 +157,16 @@
 
 namespace simgrid{
 namespace smpi{
-int 
+int
 Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
                                                MPI_Datatype dtype,
                                                MPI_Op op,
-                                               MPI_Comm comm) 
+                                               MPI_Comm comm)
 {
    int ret = MPI_SUCCESS;
    int line;
    int k, recv_from, send_to;
-   int early_blockcount, late_blockcount, split_rank; 
+   int early_blockcount, late_blockcount, split_rank;
    int segcount, max_segcount;
    int num_phases, phase;
    int block_count;
@@ -191,7 +191,7 @@ Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
       }
       return MPI_SUCCESS;
    }
-   
+
    /* Determine segment count based on the suggested segment size */
    extent = dtype->get_extent();
    if (MPI_SUCCESS != ret) { line = __LINE__; goto error_hndl; }
@@ -205,27 +205,27 @@ Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
    /* Special case for count less than size * segcount - use regular ring */
    if (count < size * segcount) {
       XBT_DEBUG( "coll:tuned:allreduce_ring_segmented rank %d/%d, count %d, switching to regular ring", rank, size, count);
-      return (Coll_allreduce_lr::allreduce(sbuf, rbuf, count, dtype, op, 
+      return (Coll_allreduce_lr::allreduce(sbuf, rbuf, count, dtype, op,
                                                    comm));
    }
 
    /* Determine the number of phases of the algorithm */
    num_phases = count / (size * segcount);
-   if ((count % (size * segcount) >= size) && 
+   if ((count % (size * segcount) >= size) &&
        (count % (size * segcount) > ((size * segcount) / 2))) {
       num_phases++;
    }
 
-   /* Determine the number of elements per block and corresponding 
+   /* Determine the number of elements per block and corresponding
       block sizes.
       The blocks are divided into "early" and "late" ones:
-      blocks 0 .. (split_rank - 1) are "early" and 
+      blocks 0 .. (split_rank - 1) are "early" and
       blocks (split_rank) .. (size - 1) are "late".
       Early blocks are at most 1 element larger than the late ones.
       Note, these blocks will be split into num_phases segments,
       out of the largest one will have max_segcount elements.
     */
-   COLL_TUNED_COMPUTE_BLOCKCOUNT( count, size, split_rank, 
+   COLL_TUNED_COMPUTE_BLOCKCOUNT( count, size, split_rank,
                                   early_blockcount, late_blockcount )
    COLL_TUNED_COMPUTE_BLOCKCOUNT( early_blockcount, num_phases, inbi,
                                   max_segcount, k)
@@ -250,7 +250,7 @@ Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
       ptrdiff_t phase_offset;
       int early_phase_segcount, late_phase_segcount, split_phase, phase_count;
 
-      /* 
+      /*
          For each of the remote nodes:
          - post irecv for block (r-1)
          - send block (r)
@@ -269,7 +269,7 @@ Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
       */
       send_to = (rank + 1) % size;
       recv_from = (rank + size - 1) % size;
-      
+
       inbi = 0;
       /* Initialize first receive from the neighbor on the left */
       reqs[inbi] = Request::irecv(inbuf[inbi], max_segcount, dtype, recv_from,
@@ -277,8 +277,8 @@ Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
       /* Send first block (my block) to the neighbor on the right:
          - compute my block and phase offset
          - send data */
-      block_offset = ((rank < split_rank)? 
-                      (rank * early_blockcount) : 
+      block_offset = ((rank < split_rank)?
+                      (rank * early_blockcount) :
                       (rank * late_blockcount + split_rank));
       block_count = ((rank < split_rank)? early_blockcount : late_blockcount);
       COLL_TUNED_COMPUTE_BLOCKCOUNT(block_count, num_phases, split_phase,
@@ -286,39 +286,39 @@ Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
       phase_count = ((phase < split_phase)?
                      (early_phase_segcount) : (late_phase_segcount));
       phase_offset = ((phase < split_phase)?
-                      (phase * early_phase_segcount) : 
+                      (phase * early_phase_segcount) :
                       (phase * late_phase_segcount + split_phase));
       tmpsend = ((char*)rbuf) + (block_offset + phase_offset) * extent;
       Request::send(tmpsend, phase_count, dtype, send_to,
                               666, comm);
-      
+
       for (k = 2; k < size; k++) {
          const int prevblock = (rank + size - k + 1) % size;
-         
+
          inbi = inbi ^ 0x1;
-         
+
          /* Post irecv for the current block */
          reqs[inbi] = Request::irecv(inbuf[inbi], max_segcount, dtype, recv_from,
                                666, comm);
          if (MPI_SUCCESS != ret) { line = __LINE__; goto error_hndl; }
-         
+
          /* Wait on previous block to arrive */
          Request::wait(&reqs[inbi ^ 0x1], MPI_STATUS_IGNORE);
-         
+
          /* Apply operation on previous block: result goes to rbuf
             rbuf[prevblock] = inbuf[inbi ^ 0x1] (op) rbuf[prevblock]
          */
          block_offset = ((prevblock < split_rank)?
                          (prevblock * early_blockcount) :
                          (prevblock * late_blockcount + split_rank));
-         block_count = ((prevblock < split_rank)? 
+         block_count = ((prevblock < split_rank)?
                         early_blockcount : late_blockcount);
          COLL_TUNED_COMPUTE_BLOCKCOUNT(block_count, num_phases, split_phase,
                                        early_phase_segcount, late_phase_segcount)
          phase_count = ((phase < split_phase)?
                         (early_phase_segcount) : (late_phase_segcount));
          phase_offset = ((phase < split_phase)?
-                         (phase * early_phase_segcount) : 
+                         (phase * early_phase_segcount) :
                          (phase * late_phase_segcount + split_phase));
          tmprecv = ((char*)rbuf) + (block_offset + phase_offset) * extent;
          if(op!=MPI_OP_NULL) op->apply( inbuf[inbi ^ 0x1], tmprecv, &phase_count, dtype);
@@ -326,25 +326,25 @@ Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
          Request::send(tmprecv, phase_count, dtype, send_to,
                               666, comm);
       }
-      
+
       /* Wait on the last block to arrive */
       Request::wait(&reqs[inbi], MPI_STATUS_IGNORE);
 
-      
-      /* Apply operation on the last block (from neighbor (rank + 1) 
+
+      /* Apply operation on the last block (from neighbor (rank + 1)
          rbuf[rank+1] = inbuf[inbi] (op) rbuf[rank + 1] */
       recv_from = (rank + 1) % size;
       block_offset = ((recv_from < split_rank)?
                       (recv_from * early_blockcount) :
                       (recv_from * late_blockcount + split_rank));
-      block_count = ((recv_from < split_rank)? 
+      block_count = ((recv_from < split_rank)?
                      early_blockcount : late_blockcount);
       COLL_TUNED_COMPUTE_BLOCKCOUNT(block_count, num_phases, split_phase,
                                     early_phase_segcount, late_phase_segcount)
       phase_count = ((phase < split_phase)?
                      (early_phase_segcount) : (late_phase_segcount));
       phase_offset = ((phase < split_phase)?
-                      (phase * early_phase_segcount) : 
+                      (phase * early_phase_segcount) :
                       (phase * late_phase_segcount + split_phase));
       tmprecv = ((char*)rbuf) + (block_offset + phase_offset) * extent;
       if(op!=MPI_OP_NULL) op->apply( inbuf[inbi], tmprecv, &phase_count, dtype);
@@ -356,15 +356,15 @@ Coll_allreduce_ompi_ring_segmented::allreduce(void *sbuf, void *rbuf, int count,
    for (k = 0; k < size - 1; k++) {
       const int recv_data_from = (rank + size - k) % size;
       const int send_data_from = (rank + 1 + size - k) % size;
-      const int send_block_offset = 
+      const int send_block_offset =
          ((send_data_from < split_rank)?
           (send_data_from * early_blockcount) :
           (send_data_from * late_blockcount + split_rank));
-      const int recv_block_offset = 
+      const int recv_block_offset =
          ((recv_data_from < split_rank)?
           (recv_data_from * early_blockcount) :
           (recv_data_from * late_blockcount + split_rank));
-      block_count = ((send_data_from < split_rank)? 
+      block_count = ((send_data_from < split_rank)?
                      early_blockcount : late_blockcount);
 
       tmprecv = (char*)rbuf + recv_block_offset * extent;
diff --git a/src/smpi/colls/allreduce/allreduce-rab-rdb.cpp b/src/smpi/colls/allreduce/allreduce-rab-rdb.cpp
index 43a65f1e6d..9cfbf1ad4d 100644
--- a/src/smpi/colls/allreduce/allreduce-rab-rdb.cpp
+++ b/src/smpi/colls/allreduce/allreduce-rab-rdb.cpp
@@ -39,10 +39,10 @@ int Coll_allreduce_rab_rdb::allreduce(void *sbuff, void *rbuff, int count,
   // processes of rank < 2*rem send their data to
   // (rank+1). These even-numbered processes no longer
   // participate in the algorithm until the very end. The
-  // remaining processes form a nice power-of-two. 
+  // remaining processes form a nice power-of-two.
 
   if (rank < 2 * rem) {
-    // even       
+    // even
     if (rank % 2 == 0) {
 
       Request::send(rbuff, count, dtype, rank + 1, tag, comm);
@@ -59,12 +59,12 @@ int Coll_allreduce_rab_rdb::allreduce(void *sbuff, void *rbuff, int count,
       // the operation is commutative or not.
        if(op!=MPI_OP_NULL) op->apply( tmp_buf, rbuff, &count, dtype);
 
-      // change the rank 
+      // change the rank
       newrank = rank / 2;
     }
   }
 
-  else                          // rank >= 2 * rem 
+  else                          // rank >= 2 * rem
     newrank = rank - rem;
 
   // If op is user-defined or count is less than pof2, use
@@ -74,12 +74,12 @@ int Coll_allreduce_rab_rdb::allreduce(void *sbuff, void *rbuff, int count,
   // datatypes on one process and derived on another as long as
   // the type maps are the same. Breaking up derived
   // datatypes to do the reduce-scatter is tricky, therefore
-  // using recursive doubling in that case.) 
+  // using recursive doubling in that case.)
 
   if (newrank != -1) {
     // do a reduce-scatter followed by allgather. for the
     // reduce-scatter, calculate the count that each process receives
-    // and the displacement within the buffer 
+    // and the displacement within the buffer
 
     cnts = (int *) xbt_malloc(pof2 * sizeof(int));
     disps = (int *) xbt_malloc(pof2 * sizeof(int));
@@ -97,7 +97,7 @@ int Coll_allreduce_rab_rdb::allreduce(void *sbuff, void *rbuff, int count,
     last_idx = pof2;
     while (mask < pof2) {
       newdst = newrank ^ mask;
-      // find real rank of dest 
+      // find real rank of dest
       dst = (newdst < rem) ? newdst * 2 + 1 : newdst + rem;
 
       send_cnt = recv_cnt = 0;
@@ -115,31 +115,31 @@ int Coll_allreduce_rab_rdb::allreduce(void *sbuff, void *rbuff, int count,
           recv_cnt += cnts[i];
       }
 
-      // Send data from recvbuf. Recv into tmp_buf 
+      // Send data from recvbuf. Recv into tmp_buf
       Request::sendrecv((char *) rbuff + disps[send_idx] * extent, send_cnt,
                    dtype, dst, tag,
                    (char *) tmp_buf + disps[recv_idx] * extent, recv_cnt,
                    dtype, dst, tag, comm, &status);
 
       // tmp_buf contains data received in this step.
-      // recvbuf contains data accumulated so far 
+      // recvbuf contains data accumulated so far
 
       // This algorithm is used only for predefined ops
       // and predefined ops are always commutative.
       if(op!=MPI_OP_NULL) op->apply( (char *) tmp_buf + disps[recv_idx] * extent,
                         (char *) rbuff + disps[recv_idx] * extent, &recv_cnt, dtype);
 
-      // update send_idx for next iteration 
+      // update send_idx for next iteration
       send_idx = recv_idx;
       mask <<= 1;
 
       // update last_idx, but not in last iteration because the value
-      // is needed in the allgather step below. 
+      // is needed in the allgather step below.
       if (mask < pof2)
         last_idx = recv_idx + pof2 / mask;
     }
 
-    // now do the allgather 
+    // now do the allgather
 
     mask >>= 1;
     while (mask > 0) {
@@ -149,7 +149,7 @@ int Coll_allreduce_rab_rdb::allreduce(void *sbuff, void *rbuff, int count,
 
       send_cnt = recv_cnt = 0;
       if (newrank < newdst) {
-        // update last_idx except on first iteration 
+        // update last_idx except on first iteration
         if (mask != pof2 / 2)
           last_idx = last_idx + pof2 / (mask * 2);
 
@@ -186,9 +186,9 @@ int Coll_allreduce_rab_rdb::allreduce(void *sbuff, void *rbuff, int count,
   // participate above.
 
   if (rank < 2 * rem) {
-    if (rank % 2)               // odd 
+    if (rank % 2)               // odd
       Request::send(rbuff, count, dtype, rank - 1, tag, comm);
-    else                        // even 
+    else                        // even
       Request::recv(rbuff, count, dtype, rank + 1, tag, comm, &status);
   }
 
diff --git a/src/smpi/colls/allreduce/allreduce-rab1.cpp b/src/smpi/colls/allreduce/allreduce-rab1.cpp
index 4b99a247f8..8f26894bb8 100644
--- a/src/smpi/colls/allreduce/allreduce-rab1.cpp
+++ b/src/smpi/colls/allreduce/allreduce-rab1.cpp
@@ -63,7 +63,7 @@ int Coll_allreduce_rab1::allreduce(void *sbuff, void *rbuff,
       if(op!=MPI_OP_NULL) op->apply( tmp_buf, (char *) recv + recv_idx * extent, &recv_cnt,
                      dtype);
 
-      // update send_idx for next iteration 
+      // update send_idx for next iteration
       send_idx = recv_idx;
       mask >>= 1;
     }
@@ -97,7 +97,7 @@ int Coll_allreduce_rab1::allreduce(void *sbuff, void *rbuff,
       if(op!=MPI_OP_NULL) op->apply( tmp_buf, (char *) rbuff + recv_idx * extent, &recv_cnt,
                      dtype);
 
-      // update send_idx for next iteration 
+      // update send_idx for next iteration
       send_idx = recv_idx;
       mask >>= 1;
     }
diff --git a/src/smpi/colls/allreduce/allreduce-rdb.cpp b/src/smpi/colls/allreduce/allreduce-rdb.cpp
index 910ed8314b..552698a0d8 100644
--- a/src/smpi/colls/allreduce/allreduce-rdb.cpp
+++ b/src/smpi/colls/allreduce/allreduce-rdb.cpp
@@ -47,10 +47,10 @@ int Coll_allreduce_rdb::allreduce(void *sbuff, void *rbuff, int count,
   // processes of rank < 2*rem send their data to
   // (rank+1). These even-numbered processes no longer
   // participate in the algorithm until the very end. The
-  // remaining processes form a nice power-of-two. 
+  // remaining processes form a nice power-of-two.
 
   if (rank < 2 * rem) {
-    // even       
+    // even
     if (rank % 2 == 0) {
 
       Request::send(rbuff, count, dtype, rank + 1, tag, comm);
@@ -67,12 +67,12 @@ int Coll_allreduce_rdb::allreduce(void *sbuff, void *rbuff, int count,
       // the operation is commutative or not.
       if(op!=MPI_OP_NULL) op->apply( tmp_buf, rbuff, &count, dtype);
 
-      // change the rank 
+      // change the rank
       newrank = rank / 2;
     }
   }
 
-  else                          // rank >= 2 * rem 
+  else                          // rank >= 2 * rem
     newrank = rank - rem;
 
   // If op is user-defined or count is less than pof2, use
@@ -82,22 +82,22 @@ int Coll_allreduce_rdb::allreduce(void *sbuff, void *rbuff, int count,
   // datatypes on one process and derived on another as long as
   // the type maps are the same. Breaking up derived
   // datatypes to do the reduce-scatter is tricky, therefore
-  // using recursive doubling in that case.) 
+  // using recursive doubling in that case.)
 
   if (newrank != -1) {
     mask = 0x1;
     while (mask < pof2) {
       newdst = newrank ^ mask;
-      // find real rank of dest 
+      // find real rank of dest
       dst = (newdst < rem) ? newdst * 2 + 1 : newdst + rem;
 
       // Send the most current data, which is in recvbuf. Recv
-      // into tmp_buf 
+      // into tmp_buf
       Request::sendrecv(rbuff, count, dtype, dst, tag, tmp_buf, count, dtype,
                    dst, tag, comm, &status);
 
       // tmp_buf contains data received in this step.
-      // recvbuf contains data accumulated so far 
+      // recvbuf contains data accumulated so far
 
       // op is commutative OR the order is already right
       // we assume it is commuttive op
@@ -120,9 +120,9 @@ int Coll_allreduce_rdb::allreduce(void *sbuff, void *rbuff, int count,
   // participate above.
 
   if (rank < 2 * rem) {
-    if (rank % 2)               // odd 
+    if (rank % 2)               // odd
       Request::send(rbuff, count, dtype, rank - 1, tag, comm);
-    else                        // even 
+    else                        // even
       Request::recv(rbuff, count, dtype, rank + 1, tag, comm, &status);
   }
 
diff --git a/src/smpi/colls/allreduce/allreduce-smp-binomial-pipeline.cpp b/src/smpi/colls/allreduce/allreduce-smp-binomial-pipeline.cpp
index 4750e238f4..e372373093 100644
--- a/src/smpi/colls/allreduce/allreduce-smp-binomial-pipeline.cpp
+++ b/src/smpi/colls/allreduce/allreduce-smp-binomial-pipeline.cpp
@@ -5,16 +5,16 @@
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
 #include "../colls_private.h"
-/* IMPLEMENTED BY PITCH PATARASUK 
-   Non-topoloty-specific (however, number of cores/node need to be changed) 
+/* IMPLEMENTED BY PITCH PATARASUK
+   Non-topoloty-specific (however, number of cores/node need to be changed)
    all-reduce operation designed for smp clusters
-   It uses 2-layer communication: binomial for both intra-communication 
+   It uses 2-layer communication: binomial for both intra-communication
    inter-communication
    The communication are done in a pipeline fashion */
 
 
 
-/* this is a default segment size for pipelining, 
+/* this is a default segment size for pipelining,
    but it is typically passed as a command line argument */
 int allreduce_smp_binomial_pipeline_segment_size = 4096;
 
@@ -113,7 +113,7 @@ int Coll_allreduce_smp_binomial_pipeline::allreduce(void *send_buf,
       }
     }
 
-    /* start binomial reduce inter-communication between each SMP nodes: 
+    /* start binomial reduce inter-communication between each SMP nodes:
        each node only have one process that can communicate to other nodes */
     if ((phase > 0) && (phase < (pipelength + 1))) {
       if (intra_rank == 0) {
@@ -138,7 +138,7 @@ int Coll_allreduce_smp_binomial_pipeline::allreduce(void *send_buf,
       }
     }
 
-    /* start binomial broadcast inter-communication between each SMP nodes: 
+    /* start binomial broadcast inter-communication between each SMP nodes:
        each node only have one process that can communicate to other nodes */
     if ((phase > 1) && (phase < (pipelength + 2))) {
       if (intra_rank == 0) {
diff --git a/src/smpi/colls/allreduce/allreduce-smp-binomial.cpp b/src/smpi/colls/allreduce/allreduce-smp-binomial.cpp
index d3bc4a23ff..ff4e5d0f37 100644
--- a/src/smpi/colls/allreduce/allreduce-smp-binomial.cpp
+++ b/src/smpi/colls/allreduce/allreduce-smp-binomial.cpp
@@ -5,10 +5,10 @@
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
 #include "../colls_private.h"
-/* IMPLEMENTED BY PITCH PATARASUK 
-   Non-topoloty-specific (however, number of cores/node need to be changed) 
+/* IMPLEMENTED BY PITCH PATARASUK
+   Non-topoloty-specific (however, number of cores/node need to be changed)
    all-reduce operation designed for smp clusters
-   It uses 2-layer communication: binomial for both intra-communication 
+   It uses 2-layer communication: binomial for both intra-communication
    inter-communication*/
 
 
@@ -83,7 +83,7 @@ int Coll_allreduce_smp_binomial::allreduce(void *send_buf, void *recv_buf,
     mask <<= 1;
   }
 
-  /* start binomial reduce inter-communication between each SMP nodes: 
+  /* start binomial reduce inter-communication between each SMP nodes:
      each node only have one process that can communicate to other nodes */
   if (intra_rank == 0) {
     mask = 1;
@@ -103,7 +103,7 @@ int Coll_allreduce_smp_binomial::allreduce(void *send_buf, void *recv_buf,
     }
   }
 
-  /* start binomial broadcast inter-communication between each SMP nodes: 
+  /* start binomial broadcast inter-communication between each SMP nodes:
      each node only have one process that can communicate to other nodes */
   if (intra_rank == 0) {
     mask = 1;
diff --git a/src/smpi/colls/allreduce/allreduce-smp-rdb.cpp b/src/smpi/colls/allreduce/allreduce-smp-rdb.cpp
index 35bde00925..52453c263d 100644
--- a/src/smpi/colls/allreduce/allreduce-smp-rdb.cpp
+++ b/src/smpi/colls/allreduce/allreduce-smp-rdb.cpp
@@ -5,10 +5,10 @@
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
 #include "../colls_private.h"
-/* IMPLEMENTED BY PITCH PATARASUK 
-   Non-topoloty-specific (however, number of cores/node need to be changed) 
+/* IMPLEMENTED BY PITCH PATARASUK
+   Non-topoloty-specific (however, number of cores/node need to be changed)
    all-reduce operation designed for smp clusters
-   It uses 2-layer communication: binomial for intra-communication 
+   It uses 2-layer communication: binomial for intra-communication
    and rdb for inter-communication*/
 
 
@@ -91,7 +91,7 @@ int Coll_allreduce_smp_rdb::allreduce(void *send_buf, void *recv_buf, int count,
   }                             /* end binomial reduce intra-communication */
 
 
-  /* start rdb (recursive doubling) all-reduce inter-communication 
+  /* start rdb (recursive doubling) all-reduce inter-communication
      between each SMP nodes : each node only have one process that can communicate
      to other nodes */
   if (intra_rank == 0) {
@@ -124,11 +124,11 @@ int Coll_allreduce_smp_rdb::allreduce(void *send_buf, void *recv_buf, int count,
       newrank = inter_rank - rem;
     }
 
-    /* example inter-communication RDB rank change algorithm 
+    /* example inter-communication RDB rank change algorithm
        0,4,8,12..36 <= true rank (assume 4 core per SMP)
        0123 4567 89 <= inter_rank
        1 3 4567 89 (1,3 got data from 0,2 : 0,2 will be idle until the end)
-       0 1 4567 89 
+       0 1 4567 89
        0 1 2345 67 => newrank
      */
 
@@ -148,8 +148,8 @@ int Coll_allreduce_smp_rdb::allreduce(void *send_buf, void *recv_buf, int count,
       }
     }
 
-    /* non pof2 case 
-       left-over processes (all even ranks: < 2 * rem) get the result    
+    /* non pof2 case
+       left-over processes (all even ranks: < 2 * rem) get the result
      */
     if (inter_rank < 2 * rem) {
       if (inter_rank % 2) {
diff --git a/src/smpi/colls/allreduce/allreduce-smp-rsag-rab.cpp b/src/smpi/colls/allreduce/allreduce-smp-rsag-rab.cpp
index 8ed15d2b7b..c0915a00ba 100644
--- a/src/smpi/colls/allreduce/allreduce-smp-rsag-rab.cpp
+++ b/src/smpi/colls/allreduce/allreduce-smp-rsag-rab.cpp
@@ -4,7 +4,7 @@
 /* This program is free software; you can redistribute it and/or modify it
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
-/* 
+/*
  * implemented by Pitch Patarasuk, 07/01/2007
  */
 #include "../colls_private.h"
diff --git a/src/smpi/colls/alltoall/alltoall-2dmesh.cpp b/src/smpi/colls/alltoall/alltoall-2dmesh.cpp
index 7c476a370b..062d5cbe3c 100644
--- a/src/smpi/colls/alltoall/alltoall-2dmesh.cpp
+++ b/src/smpi/colls/alltoall/alltoall-2dmesh.cpp
@@ -26,7 +26,7 @@
            algorithm. It actually performs allgather operation in x dimension
            then in the y dimension. Each node then extracts the needed data.
            The communication in each dimension follows "simple."
- 
+
  * Auther: Ahmad Faraj
 
 ****************************************************************************/
diff --git a/src/smpi/colls/alltoall/alltoall-basic-linear.cpp b/src/smpi/colls/alltoall/alltoall-basic-linear.cpp
index 723c5b6121..9b1b1a2aad 100644
--- a/src/smpi/colls/alltoall/alltoall-basic-linear.cpp
+++ b/src/smpi/colls/alltoall/alltoall-basic-linear.cpp
@@ -5,7 +5,7 @@
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
  #include "../colls_private.h"
- 
+
 /*Naive and simple basic alltoall implementation. */
 
 
diff --git a/src/smpi/colls/alltoall/alltoall-mvapich-scatter-dest.cpp b/src/smpi/colls/alltoall/alltoall-mvapich-scatter-dest.cpp
index 7dbad24d59..674525263b 100644
--- a/src/smpi/colls/alltoall/alltoall-mvapich-scatter-dest.cpp
+++ b/src/smpi/colls/alltoall/alltoall-mvapich-scatter-dest.cpp
@@ -37,10 +37,10 @@
  * copyright file COPYRIGHT in the top level MVAPICH2 directory.
  *
  */
- 
+
 //correct on stampede
 #define MV2_ALLTOALL_THROTTLE_FACTOR         4
- 
+
 #include "../colls_private.h"
 namespace simgrid{
 namespace smpi{
@@ -59,16 +59,16 @@ int Coll_alltoall_mvapich2_scatter_dest::alltoall(
     int dst, rank;
     MPI_Request *reqarray;
     MPI_Status *starray;
-    
+
     if (recvcount == 0) return MPI_SUCCESS;
-    
+
     comm_size =  comm->size();
     rank = comm->rank();
-    
+
     /* Get extent of send and recv types */
     recvtype_extent = recvtype->get_extent();
     sendtype_extent = sendtype->get_extent();
-    
+
     /* Medium-size message. Use isend/irecv with scattered
      destinations. Use Tony Ladd's modification to post only
      a small number of isends/irecvs at a time. */
@@ -83,20 +83,20 @@ int Coll_alltoall_mvapich2_scatter_dest::alltoall(
      there are only a few isend/irecvs left)
      */
     int ii, ss, bblock;
-        
+
     //Stampede is configured with
     bblock = MV2_ALLTOALL_THROTTLE_FACTOR;//mv2_coll_param.alltoall_throttle_factor;
-    
+
     if (bblock >= comm_size) bblock = comm_size;
     /* If throttle_factor is n, each process posts n pairs of isend/irecv
      in each iteration. */
-    
+
     /* FIXME: This should use the memory macros (there are storage
      leaks here if there is an error, for example) */
     reqarray= (MPI_Request*)xbt_malloc(2*bblock*sizeof(MPI_Request));
-    
+
     starray=(MPI_Status *)xbt_malloc(2*bblock*sizeof(MPI_Status));
- 
+
     for (ii=0; ii<comm_size; ii+=bblock) {
         ss = comm_size-ii < bblock ? comm_size-ii : bblock;
         /* do the communication -- post ss sends and receives: */
@@ -116,11 +116,11 @@ int Coll_alltoall_mvapich2_scatter_dest::alltoall(
                                           COLL_TAG_ALLTOALL, comm);
 
         }
-        
+
         /* ... then wait for them to finish: */
         Request::waitall(2*ss,reqarray,starray);
-        
-       
+
+
         /* --BEGIN ERROR HANDLING-- */
         if (mpi_errno == MPI_ERR_IN_STATUS) {
                 for (j=0; j<2*ss; j++) {
@@ -134,7 +134,7 @@ int Coll_alltoall_mvapich2_scatter_dest::alltoall(
     xbt_free(starray);
     xbt_free(reqarray);
     return (mpi_errno);
-    
+
 }
 }
 }
diff --git a/src/smpi/colls/alltoall/alltoall-rdb.cpp b/src/smpi/colls/alltoall/alltoall-rdb.cpp
index 0a16b23a38..be4860d8d1 100644
--- a/src/smpi/colls/alltoall/alltoall-rdb.cpp
+++ b/src/smpi/colls/alltoall/alltoall-rdb.cpp
@@ -13,7 +13,7 @@
 
  * Return: int
 
- * inputs: 
+ * inputs:
     send_buff: send input buffer
     send_count: number of elements to send
     send_type: data type of elements being sent
@@ -25,7 +25,7 @@
  * Descrp: Function realizes the allgather operation using the recursive
            doubling algorithm.
 
- * Auther: MPICH / slightly modified by Ahmad Faraj.  
+ * Auther: MPICH / slightly modified by Ahmad Faraj.
 
  ****************************************************************************/
 namespace simgrid{
diff --git a/src/smpi/colls/alltoallv/alltoallv-ompi-basic-linear.cpp b/src/smpi/colls/alltoallv/alltoallv-ompi-basic-linear.cpp
index 2d701eb400..8c9bc2902a 100644
--- a/src/smpi/colls/alltoallv/alltoallv-ompi-basic-linear.cpp
+++ b/src/smpi/colls/alltoallv/alltoallv-ompi-basic-linear.cpp
@@ -5,14 +5,14 @@
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
 #include "../colls_private.h"
-/*  
+/*
  * Linear functions are copied from the basic coll module.  For
  * some small number of nodes and/or small data sizes they are just as
  * fast as tuned/tree based segmenting operations and as such may be
  * selected by the decision functions.  These are copied into this module
  * due to the way we select modules in V1. i.e. in V2 we will handle this
- * differently and so will not have to duplicate code.  
- * GEF Oct05 after asking Jeff.  
+ * differently and so will not have to duplicate code.
+ * GEF Oct05 after asking Jeff.
  */
 namespace simgrid{
 namespace smpi{
@@ -67,7 +67,7 @@ Coll_alltoallv_ompi_basic_linear::alltoallv(void *sbuf, int *scounts, int *sdisp
                                       );
         preq++;
         ++nreqs;
-        
+
     }
 
     /* Now post all sends */
diff --git a/src/smpi/colls/alltoallv/alltoallv-ring.cpp b/src/smpi/colls/alltoallv/alltoallv-ring.cpp
index 2290059223..ddc591b507 100644
--- a/src/smpi/colls/alltoallv/alltoallv-ring.cpp
+++ b/src/smpi/colls/alltoallv/alltoallv-ring.cpp
@@ -45,7 +45,7 @@ int Coll_alltoallv_ring::alltoallv(void* send_buff, int* send_counts, int* send_
   recv_chunk = recv_type->get_extent();
   int pof2 = ((num_procs != 0) && ((num_procs & (~num_procs + 1)) == num_procs));
   for (i = 0; i < num_procs; i++) {
-  
+
     if (pof2 == 1) {
       /* use exclusive-or algorithm */
       src = dst = rank ^ i;
diff --git a/src/smpi/colls/barrier/barrier-mvapich2-pair.cpp b/src/smpi/colls/barrier/barrier-mvapich2-pair.cpp
index edf35d0a7e..e5f743d890 100644
--- a/src/smpi/colls/barrier/barrier-mvapich2-pair.cpp
+++ b/src/smpi/colls/barrier/barrier-mvapich2-pair.cpp
@@ -19,7 +19,7 @@
  *
  * Additional copyrights may follow
  */
- 
+
  /* -*- Mode: C; c-basic-offset:4 ; -*- */
 /*
  *
@@ -60,7 +60,7 @@ int Coll_barrier_mvapich2_pair::barrier(MPI_Comm comm)
     /*  N2_prev = greatest power of two < size of Comm  */
     for( N2_prev = 1; N2_prev <= size; N2_prev <<= 1 );
     N2_prev >>= 1;
-    
+
     int surfeit = size - N2_prev;
 
     /* Perform a combine-like operation */
diff --git a/src/smpi/colls/barrier/barrier-ompi.cpp b/src/smpi/colls/barrier/barrier-ompi.cpp
index 94131da717..12c3af3a2c 100644
--- a/src/smpi/colls/barrier/barrier-ompi.cpp
+++ b/src/smpi/colls/barrier/barrier-ompi.cpp
@@ -25,15 +25,15 @@
 
 
 /*
- * Barrier is ment to be a synchronous operation, as some BTLs can mark 
- * a request done before its passed to the NIC and progress might not be made 
- * elsewhere we cannot allow a process to exit the barrier until its last 
+ * Barrier is ment to be a synchronous operation, as some BTLs can mark
+ * a request done before its passed to the NIC and progress might not be made
+ * elsewhere we cannot allow a process to exit the barrier until its last
  * [round of] sends are completed.
  *
- * It is last round of sends rather than 'last' individual send as each pair of 
- * peers can use different channels/devices/btls and the receiver of one of 
+ * It is last round of sends rather than 'last' individual send as each pair of
+ * peers can use different channels/devices/btls and the receiver of one of
  * these sends might be forced to wait as the sender
- * leaves the collective and does not make progress until the next mpi call 
+ * leaves the collective and does not make progress until the next mpi call
  *
  */
 
@@ -60,38 +60,38 @@ int Coll_barrier_ompi_doublering::barrier(MPI_Comm comm)
     right = ((rank+1)%size);
 
     if (rank > 0) { /* receive message from the left */
-        Request::recv((void*)NULL, 0, MPI_BYTE, left, 
+        Request::recv((void*)NULL, 0, MPI_BYTE, left,
                                 COLL_TAG_BARRIER, comm,
                                 MPI_STATUS_IGNORE);
     }
 
     /* Send message to the right */
-    Request::send((void*)NULL, 0, MPI_BYTE, right, 
+    Request::send((void*)NULL, 0, MPI_BYTE, right,
                             COLL_TAG_BARRIER,
                              comm);
 
     /* root needs to receive from the last node */
     if (rank == 0) {
-        Request::recv((void*)NULL, 0, MPI_BYTE, left, 
+        Request::recv((void*)NULL, 0, MPI_BYTE, left,
                                 COLL_TAG_BARRIER, comm,
                                 MPI_STATUS_IGNORE);
     }
 
     /* Allow nodes to exit */
     if (rank > 0) { /* post Receive from left */
-        Request::recv((void*)NULL, 0, MPI_BYTE, left, 
+        Request::recv((void*)NULL, 0, MPI_BYTE, left,
                                 COLL_TAG_BARRIER, comm,
                                 MPI_STATUS_IGNORE);
     }
 
     /* send message to the right one */
-    Request::send((void*)NULL, 0, MPI_BYTE, right, 
+    Request::send((void*)NULL, 0, MPI_BYTE, right,
                             COLL_TAG_BARRIER,
                              comm);
- 
+
     /* rank 0 post receive from the last node */
     if (rank == 0) {
-        Request::recv((void*)NULL, 0, MPI_BYTE, left, 
+        Request::recv((void*)NULL, 0, MPI_BYTE, left,
                                 COLL_TAG_BARRIER, comm,
                                 MPI_STATUS_IGNORE);
     }
@@ -113,7 +113,7 @@ int Coll_barrier_ompi_recursivedoubling::barrier(MPI_Comm comm)
     rank = comm->rank();
     size = comm->size();
     XBT_DEBUG(
-                 "ompi_coll_tuned_barrier_ompi_recursivedoubling rank %d", 
+                 "ompi_coll_tuned_barrier_ompi_recursivedoubling rank %d",
                  rank);
 
     /* do nearest power of 2 less than size calc */
@@ -163,7 +163,7 @@ int Coll_barrier_ompi_recursivedoubling::barrier(MPI_Comm comm)
         if (rank < (size - adjsize)) {
             /* send enter message to higher ranked node */
             remote = rank + adjsize;
-            Request::send((void*)NULL, 0, MPI_BYTE, remote, 
+            Request::send((void*)NULL, 0, MPI_BYTE, remote,
                                     COLL_TAG_BARRIER,
                                      comm);
 
@@ -190,14 +190,14 @@ int Coll_barrier_ompi_bruck::barrier(MPI_Comm comm)
                  "ompi_coll_tuned_barrier_ompi_bruck rank %d", rank);
 
     /* exchange data with rank-2^k and rank+2^k */
-    for (distance = 1; distance < size; distance <<= 1) { 
+    for (distance = 1; distance < size; distance <<= 1) {
         from = (rank + size - distance) % size;
         to   = (rank + distance) % size;
 
         /* send message to lower ranked node */
-        Request::sendrecv(NULL, 0, MPI_BYTE, to, 
+        Request::sendrecv(NULL, 0, MPI_BYTE, to,
                                               COLL_TAG_BARRIER,
-                                              NULL, 0, MPI_BYTE, from, 
+                                              NULL, 0, MPI_BYTE, from,
                                               COLL_TAG_BARRIER,
                                               comm, MPI_STATUS_IGNORE);
     }
@@ -220,9 +220,9 @@ int Coll_barrier_ompi_two_procs::barrier(MPI_Comm comm)
                  "ompi_coll_tuned_barrier_ompi_two_procs rank %d", remote);
     remote = (remote + 1) & 0x1;
 
-    Request::sendrecv(NULL, 0, MPI_BYTE, remote, 
+    Request::sendrecv(NULL, 0, MPI_BYTE, remote,
                                           COLL_TAG_BARRIER,
-                                          NULL, 0, MPI_BYTE, remote, 
+                                          NULL, 0, MPI_BYTE, remote,
                                           COLL_TAG_BARRIER,
                                           comm, MPI_STATUS_IGNORE);
     return (MPI_SUCCESS);
@@ -252,11 +252,11 @@ int Coll_barrier_ompi_basic_linear::barrier(MPI_Comm comm)
     /* All non-root send & receive zero-length message. */
 
     if (rank > 0) {
-        Request::send (NULL, 0, MPI_BYTE, 0, 
+        Request::send (NULL, 0, MPI_BYTE, 0,
                                  COLL_TAG_BARRIER,
                                   comm);
 
-        Request::recv (NULL, 0, MPI_BYTE, 0, 
+        Request::recv (NULL, 0, MPI_BYTE, 0,
                                  COLL_TAG_BARRIER,
                                  comm, MPI_STATUS_IGNORE);
     }
@@ -293,7 +293,7 @@ int Coll_barrier_ompi_basic_linear::barrier(MPI_Comm comm)
 
 /*
  * Another recursive doubling type algorithm, but in this case
- * we go up the tree and back down the tree.  
+ * we go up the tree and back down the tree.
  */
 int Coll_barrier_ompi_tree::barrier(MPI_Comm comm)
 {
@@ -303,7 +303,7 @@ int Coll_barrier_ompi_tree::barrier(MPI_Comm comm)
     rank = comm->rank();
     size = comm->size();
     XBT_DEBUG(
-                 "ompi_coll_tuned_barrier_ompi_tree %d", 
+                 "ompi_coll_tuned_barrier_ompi_tree %d",
                  rank);
 
     /* Find the nearest power of 2 of the communicator size. */
@@ -313,7 +313,7 @@ int Coll_barrier_ompi_tree::barrier(MPI_Comm comm)
         partner = rank ^ jump;
         if (!(partner & (jump-1)) && partner < size) {
             if (partner > rank) {
-                Request::recv (NULL, 0, MPI_BYTE, partner, 
+                Request::recv (NULL, 0, MPI_BYTE, partner,
                                          COLL_TAG_BARRIER, comm,
                                          MPI_STATUS_IGNORE);
             } else if (partner < rank) {
@@ -323,7 +323,7 @@ int Coll_barrier_ompi_tree::barrier(MPI_Comm comm)
             }
         }
     }
-    
+
     depth>>=1;
     for (jump = depth; jump>0; jump>>=1) {
         partner = rank ^ jump;
@@ -333,7 +333,7 @@ int Coll_barrier_ompi_tree::barrier(MPI_Comm comm)
                                          COLL_TAG_BARRIER,
                                           comm);
             } else if (partner < rank) {
-                Request::recv (NULL, 0, MPI_BYTE, partner, 
+                Request::recv (NULL, 0, MPI_BYTE, partner,
                                          COLL_TAG_BARRIER, comm,
                                          MPI_STATUS_IGNORE);
             }
diff --git a/src/smpi/colls/bcast/bcast-NTSB.cpp b/src/smpi/colls/bcast/bcast-NTSB.cpp
index d91d728bad..9e1e974fa6 100644
--- a/src/smpi/colls/bcast/bcast-NTSB.cpp
+++ b/src/smpi/colls/bcast/bcast-NTSB.cpp
@@ -39,14 +39,14 @@ int Coll_bcast_NTSB::bcast(void *buf, int count, MPI_Datatype datatype,
 
   /* segment is segment size in number of elements (not bytes) */
   int segment = bcast_NTSB_segment_size_in_byte / extent;
-  segment =  segment == 0 ? 1 :segment; 
+  segment =  segment == 0 ? 1 :segment;
   /* pipeline length */
   int pipe_length = count / segment;
 
   /* use for buffer offset for sending and receiving data = segment size in byte */
   int increment = segment * extent;
 
-  /* if the input size is not divisible by segment size => 
+  /* if the input size is not divisible by segment size =>
      the small remainder will be done with native implementation */
   int remainder = count % segment;
 
diff --git a/src/smpi/colls/bcast/bcast-NTSL-Isend.cpp b/src/smpi/colls/bcast/bcast-NTSL-Isend.cpp
index 64e97a4fca..df3cf2b891 100644
--- a/src/smpi/colls/bcast/bcast-NTSL-Isend.cpp
+++ b/src/smpi/colls/bcast/bcast-NTSL-Isend.cpp
@@ -8,7 +8,7 @@
 
 static int bcast_NTSL_segment_size_in_byte = 8192;
 
-/* Non-topology-specific pipelined linear-bcast function 
+/* Non-topology-specific pipelined linear-bcast function
    0->1, 1->2 ,2->3, ....., ->last node : in a pipeline fashion
 */
 namespace simgrid{
@@ -37,14 +37,14 @@ int Coll_bcast_NTSL_Isend::bcast(void *buf, int count, MPI_Datatype datatype,
 
   /* segment is segment size in number of elements (not bytes) */
   int segment = bcast_NTSL_segment_size_in_byte / extent;
-  segment =  segment == 0 ? 1 :segment; 
+  segment =  segment == 0 ? 1 :segment;
   /* pipeline length */
   int pipe_length = count / segment;
 
   /* use for buffer offset for sending and receiving data = segment size in byte */
   int increment = segment * extent;
 
-  /* if the input size is not divisible by segment size => 
+  /* if the input size is not divisible by segment size =>
      the small remainder will be done with native implementation */
   int remainder = count % segment;
 
diff --git a/src/smpi/colls/bcast/bcast-NTSL.cpp b/src/smpi/colls/bcast/bcast-NTSL.cpp
index 1f8e4fa711..a9d3b83f50 100644
--- a/src/smpi/colls/bcast/bcast-NTSL.cpp
+++ b/src/smpi/colls/bcast/bcast-NTSL.cpp
@@ -8,7 +8,7 @@
 
 static int bcast_NTSL_segment_size_in_byte = 8192;
 
-/* Non-topology-specific pipelined linear-bcast function 
+/* Non-topology-specific pipelined linear-bcast function
    0->1, 1->2 ,2->3, ....., ->last node : in a pipeline fashion
 */
 namespace simgrid{
@@ -37,14 +37,14 @@ int Coll_bcast_NTSL::bcast(void *buf, int count, MPI_Datatype datatype,
 
   /* segment is segment size in number of elements (not bytes) */
   int segment = extent == 0 ? 1 : (bcast_NTSL_segment_size_in_byte / extent);
-  segment =  segment == 0 ? 1 :segment; 
+  segment =  segment == 0 ? 1 :segment;
   /* pipeline length */
   int pipe_length = count / segment;
 
   /* use for buffer offset for sending and receiving data = segment size in byte */
   int increment = segment * extent;
 
-  /* if the input size is not divisible by segment size => 
+  /* if the input size is not divisible by segment size =>
      the small remainder will be done with native implementation */
   int remainder = count % segment;
 
diff --git a/src/smpi/colls/bcast/bcast-SMP-binary.cpp b/src/smpi/colls/bcast/bcast-SMP-binary.cpp
index 801889e689..f4eb6a216f 100644
--- a/src/smpi/colls/bcast/bcast-SMP-binary.cpp
+++ b/src/smpi/colls/bcast/bcast-SMP-binary.cpp
@@ -62,7 +62,7 @@ int Coll_bcast_SMP_binary::bcast(void *buf, int count,
     else if (rank == 0)
       Request::recv(buf, count, datatype, root, tag, comm, &status);
   }
-  // when a message is smaller than a block size => no pipeline 
+  // when a message is smaller than a block size => no pipeline
   if (count <= segment) {
     // case ROOT-of-each-SMP
     if (rank % host_num_core == 0) {
diff --git a/src/smpi/colls/bcast/bcast-SMP-linear.cpp b/src/smpi/colls/bcast/bcast-SMP-linear.cpp
index 8c241b94f2..691706ecce 100644
--- a/src/smpi/colls/bcast/bcast-SMP-linear.cpp
+++ b/src/smpi/colls/bcast/bcast-SMP-linear.cpp
@@ -38,7 +38,7 @@ int Coll_bcast_SMP_linear::bcast(void *buf, int count,
   }
 
   int segment = bcast_SMP_linear_segment_byte / extent;
-  segment =  segment == 0 ? 1 :segment; 
+  segment =  segment == 0 ? 1 :segment;
   int pipe_length = count / segment;
   int remainder = count % segment;
   int increment = segment * extent;
@@ -55,7 +55,7 @@ int Coll_bcast_SMP_linear::bcast(void *buf, int count,
   if (size <= num_core) {
     XBT_WARN("MPI_bcast_SMP_linear use default MPI_bcast.");
     Coll_bcast_default::bcast(buf, count, datatype, root, comm);
-    return MPI_SUCCESS;            
+    return MPI_SUCCESS;
   }
   // if root is not zero send to rank zero first
   if (root != 0) {
@@ -64,7 +64,7 @@ int Coll_bcast_SMP_linear::bcast(void *buf, int count,
     else if (rank == 0)
       Request::recv(buf, count, datatype, root, tag, comm, &status);
   }
-  // when a message is smaller than a block size => no pipeline 
+  // when a message is smaller than a block size => no pipeline
   if (count <= segment) {
     // case ROOT
     if (rank == 0) {
diff --git a/src/smpi/colls/bcast/bcast-arrival-pattern-aware-wait.cpp b/src/smpi/colls/bcast/bcast-arrival-pattern-aware-wait.cpp
index 66da22569b..3256363a12 100644
--- a/src/smpi/colls/bcast/bcast-arrival-pattern-aware-wait.cpp
+++ b/src/smpi/colls/bcast/bcast-arrival-pattern-aware-wait.cpp
@@ -62,14 +62,14 @@ int Coll_bcast_arrival_pattern_aware_wait::bcast(void *buf, int count,
 
   /* segment is segment size in number of elements (not bytes) */
   int segment = bcast_arrival_pattern_aware_wait_segment_size_in_byte / extent;
-  segment =  segment == 0 ? 1 :segment; 
+  segment =  segment == 0 ? 1 :segment;
   /* pipeline length */
   int pipe_length = count / segment;
 
   /* use for buffer offset for sending and receiving data = segment size in byte */
   int increment = segment * extent;
 
-  /* if the input size is not divisible by segment size => 
+  /* if the input size is not divisible by segment size =>
      the small remainder will be done with native implementation */
   int remainder = count % segment;
 
diff --git a/src/smpi/colls/bcast/bcast-arrival-pattern-aware.cpp b/src/smpi/colls/bcast/bcast-arrival-pattern-aware.cpp
index b4202d34be..102be48d22 100644
--- a/src/smpi/colls/bcast/bcast-arrival-pattern-aware.cpp
+++ b/src/smpi/colls/bcast/bcast-arrival-pattern-aware.cpp
@@ -53,14 +53,14 @@ int Coll_bcast_arrival_pattern_aware::bcast(void *buf, int count,
 
   /* segment is segment size in number of elements (not bytes) */
   int segment = bcast_NTSL_segment_size_in_byte / extent;
-  segment =  segment == 0 ? 1 :segment; 
+  segment =  segment == 0 ? 1 :segment;
   /* pipeline length */
   int pipe_length = count / segment;
 
   /* use for buffer offset for sending and receiving data = segment size in byte */
   int increment = segment * extent;
 
-  /* if the input size is not divisible by segment size => 
+  /* if the input size is not divisible by segment size =>
      the small remainder will be done with native implementation */
   int remainder = count % segment;
 
@@ -347,7 +347,7 @@ int Coll_bcast_arrival_pattern_aware::bcast(void *buf, int count,
       }else{
           Request::waitall(pipe_length, recv_request_array, recv_status_array);
           }
-    
+
     }
 
     free(send_request_array);
diff --git a/src/smpi/colls/bcast/bcast-arrival-scatter.cpp b/src/smpi/colls/bcast/bcast-arrival-scatter.cpp
index ccccf7a05d..3ecba94976 100644
--- a/src/smpi/colls/bcast/bcast-arrival-scatter.cpp
+++ b/src/smpi/colls/bcast/bcast-arrival-scatter.cpp
@@ -61,7 +61,7 @@ int Coll_bcast_arrival_scatter::bcast(void *buf, int count,
   if (count < size) {
     XBT_WARN("MPI_bcast_arrival_scatter use default MPI_bcast.");
     Colls::bcast(buf, count, datatype, root, comm);
-    return MPI_SUCCESS;        
+    return MPI_SUCCESS;
   }
 
 
@@ -193,7 +193,7 @@ int Coll_bcast_arrival_scatter::bcast(void *buf, int count,
              0, tag, comm, &status);
 
     /* at this point all nodes in this set perform all-gather operation */
-    
+
     to = (myordering == (total_nodes - 1)) ? header_buf[0] : header_buf[myordering + 1];
     from = (myordering == 0) ? header_buf[total_nodes - 1] : header_buf[myordering - 1];
 
diff --git a/src/smpi/colls/bcast/bcast-flattree-pipeline.cpp b/src/smpi/colls/bcast/bcast-flattree-pipeline.cpp
index 3b5bf4b71e..f03eb3aaae 100644
--- a/src/smpi/colls/bcast/bcast-flattree-pipeline.cpp
+++ b/src/smpi/colls/bcast/bcast-flattree-pipeline.cpp
@@ -21,7 +21,7 @@ Coll_bcast_flattree_pipeline::bcast(void *buff, int count,
   extent = data_type->get_extent();
 
   int segment = flattree_segment_in_byte / extent;
-  segment =  segment == 0 ? 1 :segment; 
+  segment =  segment == 0 ? 1 :segment;
   int pipe_length = count / segment;
   int increment = segment * extent;
   if (pipe_length==0) {
diff --git a/src/smpi/colls/bcast/bcast-mvapich-smp.cpp b/src/smpi/colls/bcast/bcast-mvapich-smp.cpp
index 8ac87dbe84..b0c0f31a8c 100644
--- a/src/smpi/colls/bcast/bcast-mvapich-smp.cpp
+++ b/src/smpi/colls/bcast/bcast-mvapich-smp.cpp
@@ -42,7 +42,7 @@ extern int (*MV2_Bcast_function) (void *buffer, int count, MPI_Datatype datatype
 
 extern int (*MV2_Bcast_intra_node_function) (void *buffer, int count, MPI_Datatype datatype,
                                       int root, MPI_Comm comm_ptr);
-                                      
+
 extern int zcpy_knomial_factor;
 extern int mv2_pipelined_zcpy_knomial_factor;
 extern int bcast_segment_size;
@@ -95,15 +95,15 @@ int Coll_bcast_mvapich2_inter_node::bcast(void *buffer,
     if (MV2_Bcast_function==NULL){
       MV2_Bcast_function=Coll_bcast_mpich::bcast;
     }
-    
+
     if (MV2_Bcast_intra_node_function==NULL){
       MV2_Bcast_intra_node_function= Coll_bcast_mpich::bcast;
     }
-    
+
     if(comm->get_leaders_comm()==MPI_COMM_NULL){
       comm->init_smp();
     }
-    
+
     shmem_comm = comm->get_intra_comm();
     local_rank = shmem_comm->rank();
     local_size = shmem_comm->size();
@@ -117,8 +117,8 @@ int Coll_bcast_mvapich2_inter_node::bcast(void *buffer,
     int* leaders_map = comm->get_leaders_map();
     leader_of_root = comm->group()->rank(leaders_map[root]);
     leader_root = leader_comm->group()->rank(leaders_map[root]);
-    
-    
+
+
     if (local_size > 1) {
         if ((local_rank == 0) && (root != rank) && (leader_root == global_rank)) {
             Request::recv(buffer, count, datatype, root,
@@ -184,15 +184,15 @@ int Coll_bcast_mvapich2_knomial_intra_node::bcast(void *buffer,
     if (MV2_Bcast_function==NULL){
       MV2_Bcast_function=Coll_bcast_mpich::bcast;
     }
-    
+
     if (MV2_Bcast_intra_node_function==NULL){
       MV2_Bcast_intra_node_function= Coll_bcast_mpich::bcast;
     }
-    
+
     if(comm->get_leaders_comm()==MPI_COMM_NULL){
       comm->init_smp();
     }
-    
+
     local_size = comm->size();
     rank = comm->rank();
 
@@ -254,7 +254,7 @@ int Coll_bcast_mvapich2_intra_node::bcast(void *buffer,
     int mpi_errno = MPI_SUCCESS;
     int comm_size;
     int two_level_bcast = 1;
-    size_t nbytes = 0; 
+    size_t nbytes = 0;
     int is_homogeneous, is_contig;
     MPI_Aint type_size;
     void *tmp_buf = NULL;
@@ -265,15 +265,15 @@ int Coll_bcast_mvapich2_intra_node::bcast(void *buffer,
     if (MV2_Bcast_function==NULL){
       MV2_Bcast_function=Coll_bcast_mpich::bcast;
     }
-    
+
     if (MV2_Bcast_intra_node_function==NULL){
       MV2_Bcast_intra_node_function= Coll_bcast_mpich::bcast;
     }
-    
+
     if(comm->get_leaders_comm()==MPI_COMM_NULL){
       comm->init_smp();
     }
-    
+
     comm_size = comm->size();
    // rank = comm->rank();
 /*
diff --git a/src/smpi/colls/bcast/bcast-ompi-pipeline.cpp b/src/smpi/colls/bcast/bcast-ompi-pipeline.cpp
index 0e328aba0a..425fcab111 100644
--- a/src/smpi/colls/bcast/bcast-ompi-pipeline.cpp
+++ b/src/smpi/colls/bcast/bcast-ompi-pipeline.cpp
@@ -12,8 +12,8 @@
 namespace simgrid{
 namespace smpi{
 int Coll_bcast_ompi_pipeline::bcast( void* buffer,
-                                      int original_count, 
-                                      MPI_Datatype datatype, 
+                                      int original_count,
+                                      MPI_Datatype datatype,
                                       int root,
                                       MPI_Comm comm)
 {
@@ -22,7 +22,7 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
     size_t segsize =1024  << 7;
     //mca_coll_tuned_module_t *tuned_module = (mca_coll_tuned_module_t*) module;
     //mca_coll_tuned_comm_t *data = tuned_module->tuned_data;
-    
+
 //    return ompi_coll_tuned_bcast_intra_generic( buffer, count, datatype, root, comm, module,
 //                                                count_by_segment, data->cached_pipeline );
     ompi_coll_tree_t * tree = ompi_coll_tuned_topo_build_chain( 1, comm, root );
@@ -30,14 +30,14 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
     int rank, size;
     int segindex;
     int num_segments; /* Number of segments */
-    int sendcount;    /* number of elements sent in this segment */ 
+    int sendcount;    /* number of elements sent in this segment */
     size_t realsegsize;
     char *tmpbuf;
     ptrdiff_t extent;
     MPI_Request recv_reqs[2] = {MPI_REQUEST_NULL, MPI_REQUEST_NULL};
     MPI_Request *send_reqs = NULL;
     int req_index;
-    
+
     /**
      * Determine number of elements sent per operation.
      */
@@ -49,9 +49,9 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
 
 
     const double a_p16  = 3.2118e-6; /* [1 / byte] */
-    const double b_p16  = 8.7936;   
+    const double b_p16  = 8.7936;
     const double a_p64  = 2.3679e-6; /* [1 / byte] */
-    const double b_p64  = 1.1787;     
+    const double b_p64  = 1.1787;
     const double a_p128 = 1.6134e-6; /* [1 / byte] */
     const double b_p128 = 2.1102;
     size_t message_size;
@@ -60,13 +60,13 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
     message_size = type_size * (unsigned long)original_count;   /* needed for decision */
 
     if (size < (a_p128 * message_size + b_p128)) {
-            //Pipeline with 128KB segments 
+            //Pipeline with 128KB segments
             segsize = 1024  << 7;
     }else if (size < (a_p64 * message_size + b_p64)) {
-            // Pipeline with 64KB segments 
+            // Pipeline with 64KB segments
             segsize = 1024 << 6;
     }else if (size < (a_p16 * message_size + b_p16)) {
-            //Pipeline with 16KB segments 
+            //Pipeline with 16KB segments
             segsize = 1024 << 4;
     }
 
@@ -80,7 +80,7 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
     extent = datatype->get_extent();
     num_segments = (original_count + count_by_segment - 1) / count_by_segment;
     realsegsize = count_by_segment * extent;
-    
+
     /* Set the buffer pointers */
     tmpbuf = (char *) buffer;
 
@@ -90,7 +90,7 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
 
     /* Root code */
     if( rank == root ) {
-        /* 
+        /*
            For each segment:
            - send segment to all children.
              The last segment may have less elements than other segments.
@@ -100,26 +100,26 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
             if( segindex == (num_segments - 1) ) {
                 sendcount = original_count - segindex * count_by_segment;
             }
-            for( i = 0; i < tree->tree_nextsize; i++ ) { 
+            for( i = 0; i < tree->tree_nextsize; i++ ) {
                 send_reqs[i] = Request::isend(tmpbuf, sendcount, datatype,
-                                         tree->tree_next[i], 
+                                         tree->tree_next[i],
                                          COLL_TAG_BCAST, comm);
-           } 
+           }
 
             /* complete the sends before starting the next sends */
-            Request::waitall( tree->tree_nextsize, send_reqs, 
+            Request::waitall( tree->tree_nextsize, send_reqs,
                                          MPI_STATUSES_IGNORE );
 
             /* update tmp buffer */
             tmpbuf += realsegsize;
 
         }
-    } 
-    
+    }
+
     /* Intermediate nodes code */
-    else if( tree->tree_nextsize > 0 ) { 
-        /* 
-           Create the pipeline. 
+    else if( tree->tree_nextsize > 0 ) {
+        /*
+           Create the pipeline.
            1) Post the first receive
            2) For segments 1 .. num_segments
               - post new receive
@@ -133,31 +133,31 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
         recv_reqs[req_index]=Request::irecv(tmpbuf, count_by_segment, datatype,
                            tree->tree_prev, COLL_TAG_BCAST,
                            comm);
-        
+
         for( segindex = 1; segindex < num_segments; segindex++ ) {
-            
+
             req_index = req_index ^ 0x1;
-            
+
             /* post new irecv */
             recv_reqs[req_index]= Request::irecv( tmpbuf + realsegsize, count_by_segment,
-                                datatype, tree->tree_prev, 
+                                datatype, tree->tree_prev,
                                 COLL_TAG_BCAST,
                                 comm);
-            
+
             /* wait for and forward the previous segment to children */
-            Request::wait( &recv_reqs[req_index ^ 0x1], 
+            Request::wait( &recv_reqs[req_index ^ 0x1],
                                      MPI_STATUSES_IGNORE );
-            
-            for( i = 0; i < tree->tree_nextsize; i++ ) { 
+
+            for( i = 0; i < tree->tree_nextsize; i++ ) {
                 send_reqs[i]=Request::isend(tmpbuf, count_by_segment, datatype,
-                                         tree->tree_next[i], 
+                                         tree->tree_next[i],
                                          COLL_TAG_BCAST, comm );
-            } 
-            
+            }
+
             /* complete the sends before starting the next iteration */
-            Request::waitall( tree->tree_nextsize, send_reqs, 
+            Request::waitall( tree->tree_nextsize, send_reqs,
                                          MPI_STATUSES_IGNORE );
-            
+
             /* Update the receive buffer */
             tmpbuf += realsegsize;
         }
@@ -167,17 +167,17 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
         sendcount = original_count - (num_segments - 1) * count_by_segment;
         for( i = 0; i < tree->tree_nextsize; i++ ) {
             send_reqs[i] = Request::isend(tmpbuf, sendcount, datatype,
-                                     tree->tree_next[i], 
+                                     tree->tree_next[i],
                                      COLL_TAG_BCAST, comm);
         }
-        
-        Request::waitall( tree->tree_nextsize, send_reqs, 
+
+        Request::waitall( tree->tree_nextsize, send_reqs,
                                      MPI_STATUSES_IGNORE );
     }
-  
+
     /* Leaf nodes */
     else {
-        /* 
+        /*
            Receive all segments from parent in a loop:
            1) post irecv for the first segment
            2) for segments 1 .. num_segments
@@ -194,11 +194,11 @@ int Coll_bcast_ompi_pipeline::bcast( void* buffer,
             req_index = req_index ^ 0x1;
             tmpbuf += realsegsize;
             /* post receive for the next segment */
-            recv_reqs[req_index] = Request::irecv(tmpbuf, count_by_segment, datatype, 
+            recv_reqs[req_index] = Request::irecv(tmpbuf, count_by_segment, datatype,
                                      tree->tree_prev, COLL_TAG_BCAST,
                                      comm);
             /* wait on the previous segment */
-            Request::wait( &recv_reqs[req_index ^ 0x1], 
+            Request::wait( &recv_reqs[req_index ^ 0x1],
                                      MPI_STATUS_IGNORE );
         }
 
diff --git a/src/smpi/colls/bcast/bcast-ompi-split-bintree.cpp b/src/smpi/colls/bcast/bcast-ompi-split-bintree.cpp
index cd8ea1063b..e2a095851c 100644
--- a/src/smpi/colls/bcast/bcast-ompi-split-bintree.cpp
+++ b/src/smpi/colls/bcast/bcast-ompi-split-bintree.cpp
@@ -60,11 +60,11 @@
   #define MAXTREEFANOUT 32
 namespace simgrid{
 namespace smpi{
- 
+
 int
 Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
-                                            int count, 
-                                            MPI_Datatype datatype, 
+                                            int count,
+                                            MPI_Datatype datatype,
                                             int root,
                                             MPI_Comm comm)
 {
@@ -74,13 +74,13 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
     int segcount[2];       /* Number ompi_request_wait_allof elements sent with each segment */
     uint32_t counts[2];
     int num_segments[2];   /* Number of segmenets */
-    int sendcount[2];      /* the same like segcount, except for the last segment */ 
+    int sendcount[2];      /* the same like segcount, except for the last segment */
     size_t realsegsize[2];
     char *tmpbuf[2];
     size_t type_size;
     ptrdiff_t type_extent;
-    
-    
+
+
     MPI_Request base_req, new_req;
     ompi_coll_tree_t *tree;
 //    mca_coll_tuned_module_t *tuned_module = (mca_coll_tuned_module_t*) module;
@@ -93,11 +93,11 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
     //compute again segsize
     const size_t intermediate_message_size = 370728;
     size_t message_size = datatype->size() * (unsigned long)count;
-    if(message_size < intermediate_message_size) 
+    if(message_size < intermediate_message_size)
       segsize = 1024 ;
     else
       segsize = 1024 << 3;
-      
+
     XBT_DEBUG("ompi_coll_tuned_bcast_intra_split_bintree rank %d root %d ss %5d", rank, root, segsize);
 
     if (size == 1) {
@@ -120,7 +120,7 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
         if (segsize < ((uint32_t)type_size)) {
           segsize = type_size; /* push segsize up to hold one type */
         }
-        segcount[0] = segcount[1] = segsize / type_size; 
+        segcount[0] = segcount[1] = segsize / type_size;
         num_segments[0] = counts[0]/segcount[0];
         if ((counts[0] % segcount[0]) != 0) num_segments[0]++;
         num_segments[1] = counts[1]/segcount[1];
@@ -136,16 +136,16 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
         (segsize > counts[0] * type_size) ||
         (segsize > counts[1] * type_size) ) {
         /* call linear version here ! */
-        return (Coll_bcast_SMP_linear::bcast ( buffer, count, datatype, 
+        return (Coll_bcast_SMP_linear::bcast ( buffer, count, datatype,
                                                     root, comm));
     }
     type_extent = datatype->get_extent();
 
-    
+
     /* Determine real segment size */
     realsegsize[0] = segcount[0] * type_extent;
     realsegsize[1] = segcount[1] * type_extent;
-  
+
     /* set the buffer pointers */
     tmpbuf[0] = (char *) buffer;
     tmpbuf[1] = (char *) buffer+counts[0] * type_extent;
@@ -158,11 +158,11 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
 
     /* determine if I am left (0) or right (1), (root is right) */
     lr = ((rank + size - root)%size + 1)%2;
-  
+
     /* root code */
     if( rank == root ) {
         /* determine segment count */
-        sendcount[0] = segcount[0]; 
+        sendcount[0] = segcount[0];
         sendcount[1] = segcount[1];
         /* for each segment */
         for (segindex = 0; segindex < num_segments[0]; segindex++) {
@@ -172,7 +172,7 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
                     continue;
                 }
                 /* determine how many elements are being sent in this round */
-                if(segindex == (num_segments[i] - 1)) 
+                if(segindex == (num_segments[i] - 1))
                     sendcount[i] = counts[i] - segindex*segcount[i];
                 /* send data */
                 Request::send(tmpbuf[i], sendcount[i], datatype,
@@ -181,19 +181,19 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
                 tmpbuf[i] += realsegsize[i];
             }
         }
-    } 
-    
+    }
+
     /* intermediate nodes code */
-    else if( tree->tree_nextsize > 0 ) { 
+    else if( tree->tree_nextsize > 0 ) {
         /* Intermediate nodes:
          * It will receive segments only from one half of the data.
-         * Which one is determined by whether the node belongs to the "left" or "right" 
+         * Which one is determined by whether the node belongs to the "left" or "right"
          * subtree. Topoloby building function builds binary tree such that
          * odd "shifted ranks" ((rank + size - root)%size) are on the left subtree,
          * and even on the right subtree.
          *
          * Create the pipeline. We first post the first receive, then in the loop we
-         * post the next receive and after that wait for the previous receive to complete 
+         * post the next receive and after that wait for the previous receive to complete
          * and we disseminating the data to all children.
          */
         sendcount[lr] = segcount[lr];
@@ -203,7 +203,7 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
 
         for( segindex = 1; segindex < num_segments[lr]; segindex++ ) {
             /* determine how many elements to expect in this round */
-            if( segindex == (num_segments[lr] - 1)) 
+            if( segindex == (num_segments[lr] - 1))
                 sendcount[lr] = counts[lr] - segindex*segcount[lr];
             /* post new irecv */
             new_req = Request::irecv( tmpbuf[lr] + realsegsize[lr], sendcount[lr],
@@ -219,7 +219,7 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
             } /* end of for each child */
 
             /* upate the base request */
-            base_req = new_req;     
+            base_req = new_req;
             /* go to the next buffer (ie. the one corresponding to the next recv) */
             tmpbuf[lr] += realsegsize[lr];
         } /* end of for segindex */
@@ -230,10 +230,10 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
             Request::send(tmpbuf[lr], sendcount[lr], datatype,
                               tree->tree_next[i], COLL_TAG_BCAST, comm);
         } /* end of for each child */
-    } 
-  
+    }
+
     /* leaf nodes */
-    else { 
+    else {
         /* Just consume segments as fast as possible */
         sendcount[lr] = segcount[lr];
         for (segindex = 0; segindex < num_segments[lr]; segindex++) {
@@ -253,9 +253,9 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
     tmpbuf[1] = (char *) buffer+counts[0] * type_extent;
 
     /* Step 2:
-       Find your immediate pair (identical node in opposite subtree) and SendRecv 
+       Find your immediate pair (identical node in opposite subtree) and SendRecv
        data buffer with them.
-       The tree building function ensures that 
+       The tree building function ensures that
        if (we are not root)
        if we are in the left subtree (lr == 0) our pair is (rank+1)%size.
        if we are in the right subtree (lr == 1) our pair is (rank-1)%size
@@ -267,7 +267,7 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
         pair = (rank+size-1)%size;
     }
 
-    if ( (size%2) != 0 && rank != root) { 
+    if ( (size%2) != 0 && rank != root) {
 
         Request::sendrecv( tmpbuf[lr], counts[lr], datatype,
                                         pair, COLL_TAG_BCAST,
@@ -280,25 +280,25 @@ Coll_bcast_ompi_split_bintree::bcast ( void* buffer,
             Request::send(tmpbuf[1], counts[1], datatype,
                               (root+size-1)%size, COLL_TAG_BCAST, comm);
 
-        } 
+        }
         /* last node receives right buffer from the root */
         else if (rank == (root+size-1)%size) {
             Request::recv(tmpbuf[1], counts[1], datatype,
                               root, COLL_TAG_BCAST,
                               comm, MPI_STATUS_IGNORE);
-        } 
+        }
         /* everyone else exchanges buffers */
         else {
             Request::sendrecv( tmpbuf[lr], counts[lr], datatype,
                                             pair, COLL_TAG_BCAST,
                                             tmpbuf[(lr+1)%2], counts[(lr+1)%2], datatype,
                                             pair, COLL_TAG_BCAST,
-                                            comm, MPI_STATUS_IGNORE); 
+                                            comm, MPI_STATUS_IGNORE);
         }
     }
     xbt_free(tree);
     return (MPI_SUCCESS);
-  
+
 
 }
 
diff --git a/src/smpi/colls/bcast/bcast-scatter-LR-allgather.cpp b/src/smpi/colls/bcast/bcast-scatter-LR-allgather.cpp
index 66ac6bb0b2..3466fdf996 100644
--- a/src/smpi/colls/bcast/bcast-scatter-LR-allgather.cpp
+++ b/src/smpi/colls/bcast/bcast-scatter-LR-allgather.cpp
@@ -87,7 +87,7 @@ Coll_bcast_scatter_LR_allgather::bcast(void *buff, int count,
 
 
   nbytes = extent * count;
-  scatter_size = (nbytes + num_procs - 1) / num_procs;  // ceiling division 
+  scatter_size = (nbytes + num_procs - 1) / num_procs;  // ceiling division
   curr_size = (rank == root) ? nbytes : 0;      // root starts with all the data
   relative_rank = (rank >= root) ? rank - root : rank - root + num_procs;
 
@@ -103,7 +103,7 @@ Coll_bcast_scatter_LR_allgather::bcast(void *buff, int count,
       //  allows you to post a larger recv.
       if (recv_size <= 0)
         curr_size = 0;          // this process doesn't receive any data
-      // because of uneven division 
+      // because of uneven division
       else {
         Request::recv((char *) buff + relative_rank * scatter_size, recv_size,
                  MPI_BYTE, src, tag, comm, &status);
@@ -123,7 +123,7 @@ Coll_bcast_scatter_LR_allgather::bcast(void *buff, int count,
   while (mask > 0) {
     if (relative_rank + mask < num_procs) {
       send_size = curr_size - scatter_size * mask;
-      // mask is also the size of this process's subtree 
+      // mask is also the size of this process's subtree
 
       if (send_size > 0) {
         dst = rank + mask;
diff --git a/src/smpi/colls/bcast/bcast-scatter-rdb-allgather.cpp b/src/smpi/colls/bcast/bcast-scatter-rdb-allgather.cpp
index 775b82ea73..f81694d4d1 100644
--- a/src/smpi/colls/bcast/bcast-scatter-rdb-allgather.cpp
+++ b/src/smpi/colls/bcast/bcast-scatter-rdb-allgather.cpp
@@ -35,7 +35,7 @@ static int scatter_for_bcast(
        processes may not get any data. For example if bufsize = 97 and
        nprocs = 16, ranks 15 and 16 will get 0 data. On each process, the
        scattered data is stored at the same offset in the buffer as it is
-       on the root process. */ 
+       on the root process. */
 
     scatter_size = (nbytes + comm_size - 1)/comm_size; /* ceiling division */
     curr_size = (rank == root) ? nbytes : 0; /* root starts with all the
@@ -46,12 +46,12 @@ static int scatter_for_bcast(
     {
         if (relative_rank & mask)
         {
-            src = rank - mask; 
+            src = rank - mask;
             if (src < 0) src += comm_size;
             recv_size = nbytes - relative_rank*scatter_size;
             /* recv_size is larger than what might actually be sent by the
                sender. We don't need compute the exact value because MPI
-               allows you to post a larger recv.*/ 
+               allows you to post a larger recv.*/
             if (recv_size <= 0)
             {
                 curr_size = 0; /* this process doesn't receive any data
@@ -81,7 +81,7 @@ static int scatter_for_bcast(
     {
         if (relative_rank + mask < comm_size)
         {
-            send_size = curr_size - scatter_size * mask; 
+            send_size = curr_size - scatter_size * mask;
             /* mask is also the size of this process's subtree */
 
             if (send_size > 0)
@@ -104,10 +104,10 @@ static int scatter_for_bcast(
 
 int
 Coll_bcast_scatter_rdb_allgather::bcast (
-    void *buffer, 
-    int count, 
-    MPI_Datatype datatype, 
-    int root, 
+    void *buffer,
+    int count,
+    MPI_Datatype datatype,
+    int root,
     MPI_Comm comm)
 {
     MPI_Status status;
@@ -196,13 +196,13 @@ Coll_bcast_scatter_rdb_allgather::bcast (
     {
         relative_dst = relative_rank ^ mask;
 
-        dst = (relative_dst + root) % comm_size; 
+        dst = (relative_dst + root) % comm_size;
 
         /* find offset into send and recv buffers.
            zero out the least significant "i" bits of relative_rank and
            relative_dst to find root of src and dst
            subtrees. Use ranks of roots as index to send from
-           and recv into  buffer */ 
+           and recv into  buffer */
 
         dst_tree_root = relative_dst >> i;
         dst_tree_root <<= i;
@@ -216,9 +216,9 @@ Coll_bcast_scatter_rdb_allgather::bcast (
         if (relative_dst < comm_size)
         {
             Request::sendrecv(((char *)tmp_buf + send_offset),
-                                         curr_size, MPI_BYTE, dst, COLL_TAG_BCAST, 
+                                         curr_size, MPI_BYTE, dst, COLL_TAG_BCAST,
                                          ((char *)tmp_buf + recv_offset),
-                                         (nbytes-recv_offset < 0 ? 0 : nbytes-recv_offset), 
+                                         (nbytes-recv_offset < 0 ? 0 : nbytes-recv_offset),
                                          MPI_BYTE, dst, COLL_TAG_BCAST, comm, &status);
             recv_size=Status::get_count(&status, MPI_BYTE);
             curr_size += recv_size;
@@ -235,7 +235,7 @@ Coll_bcast_scatter_rdb_allgather::bcast (
         /* This part of the code will not currently be
            executed because we are not using recursive
            doubling for non power of two. Mark it as experimental
-           so that it doesn't show up as red in the coverage tests. */  
+           so that it doesn't show up as red in the coverage tests. */
 
         /* --BEGIN EXPERIMENTAL-- */
         if (dst_tree_root + mask > comm_size)
@@ -246,7 +246,7 @@ Coll_bcast_scatter_rdb_allgather::bcast (
                in a tree fashion. First find root of current tree
                that is being divided into two. k is the number of
                least-significant bits in this process's rank that
-               must be zeroed out to find the rank of the root */ 
+               must be zeroed out to find the rank of the root */
             j = mask;
             k = 0;
             while (j)
@@ -262,7 +262,7 @@ Coll_bcast_scatter_rdb_allgather::bcast (
             while (tmp_mask)
             {
                 relative_dst = relative_rank ^ tmp_mask;
-                dst = (relative_dst + root) % comm_size; 
+                dst = (relative_dst + root) % comm_size;
 
                 tree_root = relative_rank >> k;
                 tree_root <<= k;
@@ -270,12 +270,12 @@ Coll_bcast_scatter_rdb_allgather::bcast (
                 /* send only if this proc has data and destination
                    doesn't have data. */
 
-                /* if (rank == 3) { 
+                /* if (rank == 3) {
                    printf("rank %d, dst %d, root %d, nprocs_completed %d\n", relative_rank, relative_dst, tree_root, nprocs_completed);
                    fflush(stdout);
                    }*/
 
-                if ((relative_dst > relative_rank) && 
+                if ((relative_dst > relative_rank) &&
                     (relative_rank < tree_root + nprocs_completed)
                     && (relative_dst >= tree_root + nprocs_completed))
                 {
@@ -291,14 +291,14 @@ Coll_bcast_scatter_rdb_allgather::bcast (
                 }
                 /* recv only if this proc. doesn't have data and sender
                    has data */
-                else if ((relative_dst < relative_rank) && 
+                else if ((relative_dst < relative_rank) &&
                          (relative_dst < tree_root + nprocs_completed) &&
                          (relative_rank >= tree_root + nprocs_completed))
                 {
                     /* printf("Rank %d waiting to recv from rank %d\n",
                        relative_rank, dst); */
                     Request::recv(((char *)tmp_buf + offset),
-                                             nbytes - offset, 
+                                             nbytes - offset,
                                              MPI_BYTE, dst, COLL_TAG_BCAST,
                                              comm, &status);
                     /* nprocs_completed is also equal to the no. of processes
diff --git a/src/smpi/colls/coll_tuned_topo.cpp b/src/smpi/colls/coll_tuned_topo.cpp
index f90e726f07..aa797a9884 100644
--- a/src/smpi/colls/coll_tuned_topo.cpp
+++ b/src/smpi/colls/coll_tuned_topo.cpp
@@ -11,11 +11,11 @@
  * Copyright (c) 2004-2005 The University of Tennessee and The University
  *                         of Tennessee Research Foundation.  All rights
  *                         reserved.
- * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
  *                         University of Stuttgart.  All rights reserved.
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
- * 
+ *
  * Additional copyrights may follow
  */
 
@@ -76,7 +76,7 @@ ompi_coll_tuned_topo_build_tree( int fanout,
     int schild, sparent;
     int level; /* location of my rank in the tree structure of size */
     int delta; /* number of nodes on my level */
-    int slimit; /* total number of nodes on levels above me */ 
+    int slimit; /* total number of nodes on levels above me */
     int shiftedrank;
     int i;
     ompi_coll_tree_t* tree;
@@ -92,8 +92,8 @@ ompi_coll_tuned_topo_build_tree( int fanout,
         return NULL;
     }
 
-    /* 
-     * Get size and rank of the process in this communicator 
+    /*
+     * Get size and rank of the process in this communicator
      */
     size = comm->size();
     rank = comm->rank();
@@ -111,8 +111,8 @@ ompi_coll_tuned_topo_build_tree( int fanout,
      * Set root
      */
     tree->tree_root = root;
-  
-    /* 
+
+    /*
      * Initialize tree
      */
     tree->tree_fanout   = fanout;
@@ -128,11 +128,11 @@ ompi_coll_tuned_topo_build_tree( int fanout,
     if( size < 2 ) {
         return tree;
     }
-  
+
     /*
-     * Shift all ranks by root, so that the algorithm can be 
+     * Shift all ranks by root, so that the algorithm can be
      * designed as if root would be always 0
-     * shiftedrank should be used in calculating distances 
+     * shiftedrank should be used in calculating distances
      * and position in tree
      */
     shiftedrank = rank - root;
@@ -154,7 +154,7 @@ ompi_coll_tuned_topo_build_tree( int fanout,
             break;
         }
     }
-    
+
     /* find my parent */
     slimit = calculate_num_nodes_up_to_level( fanout, level );
     sparent = shiftedrank;
@@ -166,12 +166,12 @@ ompi_coll_tuned_topo_build_tree( int fanout,
         }
     }
     tree->tree_prev = (sparent+root)%size;
-  
+
     return tree;
 }
 
 /*
- * Constructs in-order binary tree which can be used for non-commutative reduce 
+ * Constructs in-order binary tree which can be used for non-commutative reduce
  * operations.
  * Root of this tree is always rank (size-1) and fanout is 2.
  * Here are some of the examples of this tree:
@@ -192,8 +192,8 @@ ompi_coll_tuned_topo_build_in_order_bintree( MPI_Comm comm )
     int parent, lchild, rchild;
     ompi_coll_tree_t* tree;
 
-    /* 
-     * Get size and rank of the process in this communicator 
+    /*
+     * Get size and rank of the process in this communicator
      */
     size = comm->size();
     rank = comm->rank();
@@ -207,7 +207,7 @@ ompi_coll_tuned_topo_build_in_order_bintree( MPI_Comm comm )
     tree->tree_root     = MPI_UNDEFINED;
     tree->tree_nextsize = MPI_UNDEFINED;
 
-    /* 
+    /*
      * Initialize tree
      */
     tree->tree_fanout   = 2;
@@ -218,10 +218,10 @@ ompi_coll_tuned_topo_build_in_order_bintree( MPI_Comm comm )
     tree->tree_next[0]  = -1;
     tree->tree_next[1]  = -1;
     XBT_DEBUG(
-                 "coll:tuned:topo_build_in_order_tree Building fo %d rt %d", 
+                 "coll:tuned:topo_build_in_order_tree Building fo %d rt %d",
                  tree->tree_fanout, tree->tree_root);
 
-    /* 
+    /*
      * Build the tree
      */
     myrank = rank;
@@ -237,18 +237,18 @@ ompi_coll_tuned_topo_build_in_order_bintree( MPI_Comm comm )
         rchild = -1;
         if (size - 1 > 0) {
             lchild = parent - 1;
-            if (lchild > 0) { 
+            if (lchild > 0) {
                 rchild = rightsize - 1;
             }
         }
-       
-        /* The following cases are possible: myrank can be 
+
+        /* The following cases are possible: myrank can be
            - a parent,
            - belong to the left subtree, or
            - belong to the right subtee
            Each of the cases need to be handled differently.
         */
-          
+
         if (myrank == parent) {
             /* I am the parent:
                - compute real ranks of my children, and exit the loop. */
@@ -259,7 +259,7 @@ ompi_coll_tuned_topo_build_in_order_bintree( MPI_Comm comm )
         if (myrank > rchild) {
             /* I belong to the left subtree:
                - If I am the left child, compute real rank of my parent
-               - Iterate down through tree: 
+               - Iterate down through tree:
                compute new size, shift ranks down, and update delta.
             */
             if (myrank == lchild) {
@@ -273,8 +273,8 @@ ompi_coll_tuned_topo_build_in_order_bintree( MPI_Comm comm )
         } else {
             /* I belong to the right subtree:
                - If I am the right child, compute real rank of my parent
-               - Iterate down through tree:  
-               compute new size and parent, 
+               - Iterate down through tree:
+               compute new size and parent,
                but the delta and rank do not need to change.
             */
             if (myrank == rchild) {
@@ -284,7 +284,7 @@ ompi_coll_tuned_topo_build_in_order_bintree( MPI_Comm comm )
             parent = rchild;
         }
     }
-    
+
     if (tree->tree_next[0] >= 0) { tree->tree_nextsize = 1; }
     if (tree->tree_next[1] >= 0) { tree->tree_nextsize += 1; }
 
@@ -308,7 +308,7 @@ int ompi_coll_tuned_topo_destroy_tree( ompi_coll_tree_t** tree )
 }
 
 /*
- * 
+ *
  * Here are some of the examples of this tree:
  * size == 2                   size = 4                 size = 8
  *      0                           0                        0
@@ -334,8 +334,8 @@ ompi_coll_tuned_topo_build_bmtree( MPI_Comm comm,
 
     XBT_DEBUG("coll:tuned:topo:build_bmtree rt %d", root);
 
-    /* 
-     * Get size and rank of the process in this communicator 
+    /*
+     * Get size and rank of the process in this communicator
      */
     size = comm->size();
     rank = comm->rank();
@@ -390,7 +390,7 @@ ompi_coll_tuned_topo_build_bmtree( MPI_Comm comm,
 /*
  * Constructs in-order binomial tree which can be used for gather/scatter
  * operations.
- * 
+ *
  * Here are some of the examples of this tree:
  * size == 2                   size = 4                 size = 8
  *      0                           0                        0
@@ -413,8 +413,8 @@ ompi_coll_tree_t* ompi_coll_tuned_topo_build_in_order_bmtree(MPI_Comm comm, int
 
     XBT_DEBUG("coll:tuned:topo:build_in_order_bmtree rt %d", root);
 
-    /* 
-     * Get size and rank of the process in this communicator 
+    /*
+     * Get size and rank of the process in this communicator
      */
     size = comm->size();
     rank = comm->rank();
@@ -473,8 +473,8 @@ ompi_coll_tuned_topo_build_chain( int fanout,
 
     XBT_DEBUG("coll:tuned:topo:build_chain fo %d rt %d", fanout, root);
 
-    /* 
-     * Get size and rank of the process in this communicator 
+    /*
+     * Get size and rank of the process in this communicator
      */
     size = comm->size();
     rank = comm->rank();
@@ -489,7 +489,7 @@ ompi_coll_tuned_topo_build_chain( int fanout,
     }
 
     /*
-     * Allocate space for topology arrays if needed 
+     * Allocate space for topology arrays if needed
      */
     chain = (ompi_coll_tree_t*)malloc( sizeof(ompi_coll_tree_t) );
     if (not chain) {
@@ -501,17 +501,17 @@ ompi_coll_tuned_topo_build_chain( int fanout,
     chain->tree_nextsize = -1;
     for(i=0;i<fanout;i++) chain->tree_next[i] = -1;
 
-    /* 
+    /*
      * Set root & numchain
      */
     chain->tree_root = root;
-    if( (size - 1) < fanout ) { 
+    if( (size - 1) < fanout ) {
         chain->tree_nextsize = size-1;
         fanout = size-1;
     } else {
         chain->tree_nextsize = fanout;
     }
-    
+
     /*
      * Shift ranks
      */
@@ -582,13 +582,13 @@ ompi_coll_tuned_topo_build_chain( int fanout,
                 chain->tree_nextsize = 1;
             } else {
                 chain->tree_next[0] = -1;
-                chain->tree_nextsize = 0;    
+                chain->tree_nextsize = 0;
             }
         }
     }
-    
+
     /*
-     * Unshift values 
+     * Unshift values
      */
     if( rank == root ) {
         chain->tree_prev = -1;
diff --git a/src/smpi/colls/coll_tuned_topo.h b/src/smpi/colls/coll_tuned_topo.h
index 2702723976..907fdecf70 100644
--- a/src/smpi/colls/coll_tuned_topo.h
+++ b/src/smpi/colls/coll_tuned_topo.h
@@ -10,11 +10,11 @@
  * Copyright (c) 2004-2005 The University of Tennessee and The University
  *                         of Tennessee Research Foundation.  All rights
  *                         reserved.
- * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart, 
+ * Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
  *                         University of Stuttgart.  All rights reserved.
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
- * 
+ *
  * Additional copyrights may follow
  */
 
diff --git a/src/smpi/colls/gather/gather-mvapich.cpp b/src/smpi/colls/gather/gather-mvapich.cpp
index 407a3b3544..9cc9a36259 100644
--- a/src/smpi/colls/gather/gather-mvapich.cpp
+++ b/src/smpi/colls/gather/gather-mvapich.cpp
@@ -51,7 +51,7 @@ typedef int (*MV2_Gather_function_ptr) (void *sendbuf,
     int recvcnt,
     MPI_Datatype recvtype,
     int root, MPI_Comm comm);
-    
+
 extern MV2_Gather_function_ptr MV2_Gather_inter_leader_function;
 extern MV2_Gather_function_ptr MV2_Gather_intra_node_function;
 
@@ -80,15 +80,15 @@ namespace smpi{
  *                     (shmem_comm or intra_sock_comm or
  *                     inter-sock_leader_comm)
  * intra_node_fn_ptr - (in) Function ptr to choose the
- *                      intra node gather function  
+ *                      intra node gather function
  * errflag           - (out) to record errors
  */
 static int MPIR_pt_pt_intra_gather( void *sendbuf, int sendcnt, MPI_Datatype sendtype,
                             void *recvbuf, int recvcnt, MPI_Datatype recvtype,
-                            int root, int rank, 
+                            int root, int rank,
                             void *tmp_buf, int nbytes,
                             int is_data_avail,
-                            MPI_Comm comm,  
+                            MPI_Comm comm,
                             MV2_Gather_function_ptr intra_node_fn_ptr)
 {
     int mpi_errno = MPI_SUCCESS;
@@ -105,10 +105,10 @@ static int MPIR_pt_pt_intra_gather( void *sendbuf, int sendcnt, MPI_Datatype sen
         recvtype->extent(&true_lb,
                                        &recvtype_true_extent);
     }
-    
+
     /* Special case, when tmp_buf itself has data */
     if (rank == root && sendbuf == MPI_IN_PLACE && is_data_avail) {
-         
+
          mpi_errno = intra_node_fn_ptr(MPI_IN_PLACE,
                                        sendcnt, sendtype, tmp_buf, nbytes,
                                        MPI_BYTE, 0, comm);
@@ -151,12 +151,12 @@ int Coll_gather_mvapich2_two_level::gather(void *sendbuf,
     MPI_Aint true_lb = 0, sendtype_true_extent = 0, recvtype_true_extent = 0;
     MPI_Comm shmem_comm, leader_comm;
     void* tmp_buf = NULL;
-    
+
 
     //if not set (use of the algo directly, without mvapich2 selector)
     if(MV2_Gather_intra_node_function==NULL)
       MV2_Gather_intra_node_function= Coll_gather_mpich::gather;
-    
+
     if(comm->get_leaders_comm()==MPI_COMM_NULL){
       comm->init_smp();
     }
@@ -186,7 +186,7 @@ int Coll_gather_mvapich2_two_level::gather(void *sendbuf,
     shmem_comm = comm->get_intra_comm();
     local_rank = shmem_comm->rank();
     local_size = shmem_comm->size();
-    
+
     if (local_rank == 0) {
         /* Node leader. Extract the rank, size information for the leader
          * communicator */
@@ -206,23 +206,23 @@ int Coll_gather_mvapich2_two_level::gather(void *sendbuf,
     }
 
 #if defined(_SMP_LIMIC_)
-     if((g_use_limic2_coll) && (shmem_commptr->ch.use_intra_sock_comm == 1) 
+     if((g_use_limic2_coll) && (shmem_commptr->ch.use_intra_sock_comm == 1)
          && (use_limic_gather)
-         &&((num_scheme == USE_GATHER_PT_PT_BINOMIAL) 
+         &&((num_scheme == USE_GATHER_PT_PT_BINOMIAL)
             || (num_scheme == USE_GATHER_PT_PT_DIRECT)
-            ||(num_scheme == USE_GATHER_PT_LINEAR_BINOMIAL) 
+            ||(num_scheme == USE_GATHER_PT_LINEAR_BINOMIAL)
             || (num_scheme == USE_GATHER_PT_LINEAR_DIRECT)
             || (num_scheme == USE_GATHER_LINEAR_PT_BINOMIAL)
             || (num_scheme == USE_GATHER_LINEAR_PT_DIRECT)
             || (num_scheme == USE_GATHER_LINEAR_LINEAR)
             || (num_scheme == USE_GATHER_SINGLE_LEADER))) {
-            
+
             mpi_errno = MV2_Gather_intra_node_function(sendbuf, sendcnt, sendtype,
-                                                    recvbuf, recvcnt,recvtype, 
+                                                    recvbuf, recvcnt,recvtype,
                                                     root, comm);
      } else
 
-#endif/*#if defined(_SMP_LIMIC_)*/    
+#endif/*#if defined(_SMP_LIMIC_)*/
     {
         if (local_rank == 0) {
             /* Node leader, allocate tmp_buffer */
@@ -241,17 +241,17 @@ int Coll_gather_mvapich2_two_level::gather(void *sendbuf,
         }
          /*while testing mpich2 gather test, we see that
          * which basically splits the comm, and we come to
-         * a point, where use_intra_sock_comm == 0, but if the 
+         * a point, where use_intra_sock_comm == 0, but if the
          * intra node function is MPIR_Intra_node_LIMIC_Gather_MV2,
-         * it would use the intra sock comm. In such cases, we 
+         * it would use the intra sock comm. In such cases, we
          * fallback to binomial as a default case.*/
-#if defined(_SMP_LIMIC_)         
+#if defined(_SMP_LIMIC_)
         if(*MV2_Gather_intra_node_function == MPIR_Intra_node_LIMIC_Gather_MV2) {
 
             mpi_errno  = MPIR_pt_pt_intra_gather(sendbuf,sendcnt, sendtype,
                                                  recvbuf, recvcnt, recvtype,
-                                                 root, rank, 
-                                                 tmp_buf, nbytes, 
+                                                 root, rank,
+                                                 tmp_buf, nbytes,
                                                  TEMP_BUF_HAS_NO_DATA,
                                                  shmem_commptr,
                                                  MPIR_Gather_intra);
@@ -263,8 +263,8 @@ int Coll_gather_mvapich2_two_level::gather(void *sendbuf,
              * local data, we pass is_data_avail = TEMP_BUF_HAS_NO_DATA*/
             mpi_errno  = MPIR_pt_pt_intra_gather(sendbuf,sendcnt, sendtype,
                                                  recvbuf, recvcnt, recvtype,
-                                                 root, rank, 
-                                                 tmp_buf, nbytes, 
+                                                 root, rank,
+                                                 tmp_buf, nbytes,
                                                  TEMP_BUF_HAS_NO_DATA,
                                                  shmem_comm,
                                                  MV2_Gather_intra_node_function
@@ -275,8 +275,8 @@ int Coll_gather_mvapich2_two_level::gather(void *sendbuf,
     int* leaders_map = comm->get_leaders_map();
     leader_of_root = comm->group()->rank(leaders_map[root]);
     leader_root = leader_comm->group()->rank(leaders_map[root]);
-    /* leader_root is the rank of the leader of the root in leader_comm. 
-     * leader_root is to be used as the root of the inter-leader gather ops 
+    /* leader_root is the rank of the leader of the root in leader_comm.
+     * leader_root is to be used as the root of the inter-leader gather ops
      */
     if (not comm->is_uniform()) {
       if (local_rank == 0) {
@@ -353,8 +353,8 @@ int Coll_gather_mvapich2_two_level::gather(void *sendbuf,
         }
       }
     } else {
-        /* All nodes have the same number of processes. 
-         * Just do one Gather to get all 
+        /* All nodes have the same number of processes.
+         * Just do one Gather to get all
          * the data at the leader of the root process */
         if (local_rank == 0) {
             if (leader_comm_rank == leader_root && root != leader_of_root) {
@@ -373,7 +373,7 @@ int Coll_gather_mvapich2_two_level::gather(void *sendbuf,
                                                    recvcnt * local_size,
                                                    recvtype, leader_root,
                                                    leader_comm);
-                 
+
             } else {
                 mpi_errno = MPIR_Gather_MV2_Direct(tmp_buf, nbytes * local_size,
                                                    MPI_BYTE, leader_gather_buf,
diff --git a/src/smpi/colls/gather/gather-ompi.cpp b/src/smpi/colls/gather/gather-ompi.cpp
index 0420e514fd..6190e56669 100644
--- a/src/smpi/colls/gather/gather-ompi.cpp
+++ b/src/smpi/colls/gather/gather-ompi.cpp
@@ -39,7 +39,7 @@ int Coll_gather_ompi_binomial::gather(void* sbuf, int scount, MPI_Datatype sdtyp
     int err;
     ompi_coll_tree_t* bmtree;
     MPI_Status status;
-    MPI_Aint sextent, slb, strue_lb, strue_extent; 
+    MPI_Aint sextent, slb, strue_lb, strue_extent;
     MPI_Aint rextent, rlb, rtrue_lb, rtrue_extent;
 
 
@@ -203,7 +203,7 @@ int Coll_gather_ompi_linear_sync::gather(void *sbuf, int scount,
                                          MPI_Datatype sdtype,
                                          void *rbuf, int rcount,
                                          MPI_Datatype rdtype,
-                                         int root, 
+                                         int root,
                                          MPI_Comm comm)
 {
     int i;
@@ -217,7 +217,7 @@ int Coll_gather_ompi_linear_sync::gather(void *sbuf, int scount,
     int first_segment_size=0;
     size = comm->size();
     rank = comm->rank();
-    
+
     size_t dsize, block_size;
     if (rank == root) {
         dsize= rdtype->size();
@@ -226,7 +226,7 @@ int Coll_gather_ompi_linear_sync::gather(void *sbuf, int scount,
         dsize=sdtype->size();
         block_size = dsize * scount;
     }
-    
+
      if (block_size > 92160){
      first_segment_size = 32768;
      }else{
@@ -272,26 +272,26 @@ int Coll_gather_ompi_linear_sync::gather(void *sbuf, int scount,
         ret  = -1;
         line = __LINE__;
         goto error_hndl; }
-        
+
         typelng=rdtype->size();
         rdtype->extent(&lb, &extent);
         first_segment_count = rcount;
-        COLL_TUNED_COMPUTED_SEGCOUNT( (size_t)first_segment_size, typelng, 
+        COLL_TUNED_COMPUTED_SEGCOUNT( (size_t)first_segment_size, typelng,
                                       first_segment_count );
 
         for (i = 0; i < size; ++i) {
-            if (i == rank) {  
+            if (i == rank) {
                 /* skip myself */
-                reqs[i] = MPI_REQUEST_NULL; 
-                continue; 
-            } 
+                reqs[i] = MPI_REQUEST_NULL;
+                continue;
+            }
 
             /* irecv for the first segment from i */
             ptmp = (char*)rbuf + i * rcount * extent;
             first_segment_req = Request::irecv(ptmp, first_segment_count, rdtype, i,
                                      COLL_TAG_GATHER, comm
                                      );
-            
+
             /* send sync message */
             Request::send(rbuf, 0, MPI_BYTE, i,
                                     COLL_TAG_GATHER,
@@ -299,7 +299,7 @@ int Coll_gather_ompi_linear_sync::gather(void *sbuf, int scount,
 
             /* irecv for the second segment */
             ptmp = (char*)rbuf + (i * rcount + first_segment_count) * extent;
-            reqs[i]=Request::irecv(ptmp, (rcount - first_segment_count), 
+            reqs[i]=Request::irecv(ptmp, (rcount - first_segment_count),
                                      rdtype, i, COLL_TAG_GATHER, comm
                                      );
 
@@ -310,11 +310,11 @@ int Coll_gather_ompi_linear_sync::gather(void *sbuf, int scount,
         /* copy local data if necessary */
         if (MPI_IN_PLACE != sbuf) {
             ret = Datatype::copy(sbuf, scount, sdtype,
-                                  (char*)rbuf + rank * rcount * extent, 
+                                  (char*)rbuf + rank * rcount * extent,
                                   rcount, rdtype);
             if (ret != MPI_SUCCESS) { line = __LINE__; goto error_hndl; }
         }
-        
+
         /* wait all second segments to complete */
         ret = Request::waitall(size, reqs, MPI_STATUSES_IGNORE);
         if (ret != MPI_SUCCESS) { line = __LINE__; goto error_hndl; }
@@ -326,8 +326,8 @@ int Coll_gather_ompi_linear_sync::gather(void *sbuf, int scount,
 
     return MPI_SUCCESS;
  error_hndl:
-    XBT_DEBUG( 
-                   "ERROR_HNDL: node %d file %s line %d error %d\n", 
+    XBT_DEBUG(
+                   "ERROR_HNDL: node %d file %s line %d error %d\n",
                    rank, __FILE__, line, ret );
     return ret;
 }
@@ -335,8 +335,8 @@ int Coll_gather_ompi_linear_sync::gather(void *sbuf, int scount,
 /*
  * Linear functions are copied from the BASIC coll module
  * they do not segment the message and are simple implementations
- * but for some small number of nodes and/or small data sizes they 
- * are just as fast as tuned/tree based segmenting operations 
+ * but for some small number of nodes and/or small data sizes they
+ * are just as fast as tuned/tree based segmenting operations
  * and as such may be selected by the decision functions
  * These are copied into this module due to the way we select modules
  * in V1. i.e. in V2 we will handle this differently and so will not
diff --git a/src/smpi/colls/reduce/reduce-NTSL.cpp b/src/smpi/colls/reduce/reduce-NTSL.cpp
index 4032b3deca..b15c4f7056 100644
--- a/src/smpi/colls/reduce/reduce-NTSL.cpp
+++ b/src/smpi/colls/reduce/reduce-NTSL.cpp
@@ -9,7 +9,7 @@
 
 int reduce_NTSL_segment_size_in_byte = 8192;
 
-/* Non-topology-specific pipelined linear-bcast function 
+/* Non-topology-specific pipelined linear-bcast function
    0->1, 1->2 ,2->3, ....., ->last node : in a pipeline fashion
 */
 namespace simgrid{
@@ -45,7 +45,7 @@ int Coll_reduce_NTSL::reduce(void *buf, void *rbuf, int count,
   /* use for buffer offset for sending and receiving data = segment size in byte */
   int increment = segment * extent;
 
-  /* if the input size is not divisible by segment size => 
+  /* if the input size is not divisible by segment size =>
      the small remainder will be done with native implementation */
   int remainder = count % segment;
 
@@ -59,7 +59,7 @@ int Coll_reduce_NTSL::reduce(void *buf, void *rbuf, int count,
      Request::send(buf,count,datatype,0,tag,comm);
      }
      else if (rank == 0) {
-     Request::recv(buf,count,datatype,root,tag,comm,&status);        
+     Request::recv(buf,count,datatype,root,tag,comm,&status);
      }
      }
    */
diff --git a/src/smpi/colls/reduce/reduce-arrival-pattern-aware.cpp b/src/smpi/colls/reduce/reduce-arrival-pattern-aware.cpp
index ed8fb0e573..91a371f256 100644
--- a/src/smpi/colls/reduce/reduce-arrival-pattern-aware.cpp
+++ b/src/smpi/colls/reduce/reduce-arrival-pattern-aware.cpp
@@ -62,7 +62,7 @@ int Coll_reduce_arrival_pattern_aware::reduce(void *buf, void *rbuf,
   /* use for buffer offset for sending and receiving data = segment size in byte */
   int increment = segment * extent;
 
-  /* if the input size is not divisible by segment size => 
+  /* if the input size is not divisible by segment size =>
      the small remainder will be done with native implementation */
   int remainder = count % segment;
 
diff --git a/src/smpi/colls/reduce/reduce-binomial.cpp b/src/smpi/colls/reduce/reduce-binomial.cpp
index e0cdc0c94e..7321767499 100644
--- a/src/smpi/colls/reduce/reduce-binomial.cpp
+++ b/src/smpi/colls/reduce/reduce-binomial.cpp
@@ -31,9 +31,9 @@ int Coll_reduce_binomial::reduce(void *sendbuf, void *recvbuf, int count,
   tmp_buf = (void *) smpi_get_tmp_sendbuffer(count * extent);
   int is_commutative =  (op==MPI_OP_NULL || op->is_commutative());
   mask = 1;
-  
+
   int lroot;
-  if (is_commutative) 
+  if (is_commutative)
         lroot   = root;
   else
         lroot   = 0;
@@ -43,7 +43,7 @@ int Coll_reduce_binomial::reduce(void *sendbuf, void *recvbuf, int count,
 
   /* adjust for potential negative lower bound in datatype */
   tmp_buf = (void *)((char*)tmp_buf - true_lb);
-    
+
   /* If I'm not the root, then my recvbuf may not be valid, therefore
      I have to allocate a temporary one */
   if (rank != root) {
@@ -61,7 +61,7 @@ int Coll_reduce_binomial::reduce(void *sendbuf, void *recvbuf, int count,
       if (source < comm_size) {
         source = (source + lroot) % comm_size;
         Request::recv(tmp_buf, count, datatype, source, tag, comm, &status);
-        
+
         if (is_commutative) {
           if(op!=MPI_OP_NULL) op->apply( tmp_buf, recvbuf, &count, datatype);
         } else {
diff --git a/src/smpi/colls/reduce/reduce-mvapich-knomial.cpp b/src/smpi/colls/reduce/reduce-mvapich-knomial.cpp
index 4e9a35f078..2ab0774f80 100644
--- a/src/smpi/colls/reduce/reduce-mvapich-knomial.cpp
+++ b/src/smpi/colls/reduce/reduce-mvapich-knomial.cpp
@@ -37,7 +37,7 @@
  * copyright file COPYRIGHT in the top level MVAPICH2 directory.
  *
  */
- 
+
 #include "../colls_private.h"
 extern int mv2_reduce_intra_knomial_factor;
 extern int mv2_reduce_inter_knomial_factor;
@@ -45,15 +45,15 @@ extern int mv2_reduce_inter_knomial_factor;
 #define SMPI_DEFAULT_KNOMIAL_FACTOR 4
 
 //        int mv2_reduce_knomial_factor = 2;
-        
-        
-        
-static int MPIR_Reduce_knomial_trace(int root, int reduce_knomial_factor,  
+
+
+
+static int MPIR_Reduce_knomial_trace(int root, int reduce_knomial_factor,
         MPI_Comm comm, int *dst, int *expected_send_count,
         int *expected_recv_count, int **src_array)
 {
     int mask=0x1, k, comm_size, src, rank, relative_rank, lroot=0;
-    int orig_mask=0x1; 
+    int orig_mask=0x1;
     int recv_iter=0, send_iter=0;
     int *knomial_reduce_src_array=NULL;
     comm_size =  comm->size();
@@ -62,7 +62,7 @@ static int MPIR_Reduce_knomial_trace(int root, int reduce_knomial_factor,
     lroot = root;
     relative_rank = (rank - lroot + comm_size) % comm_size;
 
-    /* First compute to whom we need to send data */ 
+    /* First compute to whom we need to send data */
     while (mask < comm_size) {
         if (relative_rank % (reduce_knomial_factor*mask)) {
             *dst = relative_rank/(reduce_knomial_factor*mask)*
@@ -77,8 +77,8 @@ static int MPIR_Reduce_knomial_trace(int root, int reduce_knomial_factor,
     }
     mask /= reduce_knomial_factor;
 
-    /* Now compute how many children we have in the knomial-tree */ 
-    orig_mask = mask; 
+    /* Now compute how many children we have in the knomial-tree */
+    orig_mask = mask;
     while (mask > 0) {
         for(k=1;k<reduce_knomial_factor;k++) {
             if (relative_rank + mask*k < comm_size) {
@@ -88,13 +88,13 @@ static int MPIR_Reduce_knomial_trace(int root, int reduce_knomial_factor,
         mask /= reduce_knomial_factor;
     }
 
-    /* Finally, fill up the src array */ 
-    if(recv_iter > 0) { 
+    /* Finally, fill up the src array */
+    if(recv_iter > 0) {
         knomial_reduce_src_array = static_cast<int*>(smpi_get_tmp_sendbuffer(sizeof(int)*recv_iter));
-    } 
+    }
 
-    mask = orig_mask; 
-    recv_iter=0; 
+    mask = orig_mask;
+    recv_iter=0;
     while (mask > 0) {
         for(k=1;k<reduce_knomial_factor;k++) {
             if (relative_rank + mask*k < comm_size) {
@@ -110,10 +110,10 @@ static int MPIR_Reduce_knomial_trace(int root, int reduce_knomial_factor,
 
     *expected_recv_count = recv_iter;
     *expected_send_count = send_iter;
-    *src_array = knomial_reduce_src_array; 
-    return 0; 
+    *src_array = knomial_reduce_src_array;
+    return 0;
 }
-        
+
 namespace simgrid{
 namespace smpi{
 int Coll_reduce_mvapich2_knomial::reduce (
@@ -131,7 +131,7 @@ int Coll_reduce_mvapich2_knomial::reduce (
     MPI_Request send_request;
     int index=0;
     MPI_Aint true_lb, true_extent, extent;
-    MPI_Status status; 
+    MPI_Status status;
     int recv_iter=0, dst=-1, expected_send_count, expected_recv_count;
     int *src_array=NULL;
     void **tmp_buf=NULL;
@@ -170,7 +170,7 @@ int Coll_reduce_mvapich2_knomial::reduce (
       }
 
 
-    MPIR_Reduce_knomial_trace(root, mv2_reduce_intra_knomial_factor, comm, 
+    MPIR_Reduce_knomial_trace(root, mv2_reduce_intra_knomial_factor, comm,
            &dst, &expected_send_count, &expected_recv_count, &src_array);
 
     if(expected_recv_count > 0 ) {
@@ -208,9 +208,9 @@ int Coll_reduce_mvapich2_knomial::reduce (
         xbt_free(requests);
     }
 
-    if(src_array != NULL) { 
+    if(src_array != NULL) {
         xbt_free(src_array);
-    } 
+    }
 
     if(rank != root) {
         send_request=Request::isend(recvbuf,count, datatype, dst,
diff --git a/src/smpi/colls/reduce/reduce-mvapich-two-level.cpp b/src/smpi/colls/reduce/reduce-mvapich-two-level.cpp
index c60e9f920f..6d031e0363 100644
--- a/src/smpi/colls/reduce/reduce-mvapich-two-level.cpp
+++ b/src/smpi/colls/reduce/reduce-mvapich-two-level.cpp
@@ -89,8 +89,8 @@ int Coll_reduce_mvapich2_two_level::reduce( void *sendbuf,
     void *in_buf = NULL, *out_buf = NULL, *tmp_buf = NULL;
     MPI_Aint true_lb, true_extent, extent;
     int is_commutative = 0, stride = 0;
-    int intra_node_root=0; 
-    
+    int intra_node_root=0;
+
     //if not set (use of the algo directly, without mvapich2 selector)
     if(MV2_Reduce_function==NULL)
       MV2_Reduce_function=Coll_reduce_mpich::reduce;
@@ -100,13 +100,13 @@ int Coll_reduce_mvapich2_two_level::reduce( void *sendbuf,
     if(comm->get_leaders_comm()==MPI_COMM_NULL){
       comm->init_smp();
     }
-  
+
     my_rank = comm->rank();
     total_size = comm->size();
     shmem_comm = comm->get_intra_comm();
     local_rank = shmem_comm->rank();
     local_size = shmem_comm->size();
-    
+
     leader_comm = comm->get_leaders_comm();
     int* leaders_map = comm->get_leaders_map();
     leader_of_root = comm->group()->rank(leaders_map[root]);
@@ -135,18 +135,18 @@ int Coll_reduce_mvapich2_two_level::reduce( void *sendbuf,
                 in_buf = recvbuf;
             }
 
-            if (local_rank == 0) { 
+            if (local_rank == 0) {
                  if( my_rank != root) {
                      out_buf = tmp_buf;
-                 } else { 
-                     out_buf = recvbuf; 
-                     if(in_buf == out_buf) { 
-                        in_buf = MPI_IN_PLACE; 
-                        out_buf = recvbuf; 
-                     } 
-                 } 
+                 } else {
+                     out_buf = recvbuf;
+                     if(in_buf == out_buf) {
+                        in_buf = MPI_IN_PLACE;
+                        out_buf = recvbuf;
+                     }
+                 }
             } else {
-                in_buf  = (void *)sendbuf; 
+                in_buf  = (void *)sendbuf;
                 out_buf = NULL;
             }
 
@@ -166,21 +166,21 @@ int Coll_reduce_mvapich2_two_level::reduce( void *sendbuf,
                                          MPI_STATUS_IGNORE);
             }
         } else {
-            if(mv2_use_knomial_reduce == 1) { 
-                reduce_fn = &MPIR_Reduce_intra_knomial_wrapper_MV2; 
-            } else { 
-                reduce_fn = &MPIR_Reduce_binomial_MV2; 
-            } 
+            if(mv2_use_knomial_reduce == 1) {
+                reduce_fn = &MPIR_Reduce_intra_knomial_wrapper_MV2;
+            } else {
+                reduce_fn = &MPIR_Reduce_binomial_MV2;
+            }
             mpi_errno = reduce_fn(sendbuf, recvbuf, count,
                                   datatype, op,
                                   root, comm);
         }
         /* We are done */
-        if(tmp_buf!=NULL) 
+        if(tmp_buf!=NULL)
           smpi_free_tmp_buffer((void *) ((char *) tmp_buf + true_lb));
         goto fn_exit;
     }
-    
+
 
     if (local_rank == 0) {
         leader_comm = comm->get_leaders_comm();
@@ -205,12 +205,12 @@ int Coll_reduce_mvapich2_two_level::reduce( void *sendbuf,
     }
 
 
-    if(local_size > 1) { 
+    if(local_size > 1) {
         /* Lets do the intra-node reduce operations, if we have more than one
          * process in the node */
 
         /*Fix the input and outbuf buffers for the intra-node reduce.
-         *Node leaders will have the reduced data in tmp_buf after 
+         *Node leaders will have the reduced data in tmp_buf after
          *this step*/
         if (MV2_Reduce_intra_function == & MPIR_Reduce_shmem_MV2)
         {
@@ -227,33 +227,33 @@ int Coll_reduce_mvapich2_two_level::reduce( void *sendbuf,
                                       datatype, op,
                                       intra_node_root, shmem_comm);
         }
-    } else { 
+    } else {
         smpi_free_tmp_buffer((void *) ((char *) tmp_buf + true_lb));
-        tmp_buf = in_buf; 
-    } 
+        tmp_buf = in_buf;
+    }
 
     /* Now work on the inter-leader phase. Data is in tmp_buf */
     if (local_rank == 0 && leader_comm_size > 1) {
-        /*The leader of root will have the global reduced data in tmp_buf 
+        /*The leader of root will have the global reduced data in tmp_buf
            or recv_buf
            at the end of the reduce */
         if (leader_comm_rank == leader_root) {
             if (my_rank == root) {
-                /* I am the root of the leader-comm, and the 
-                 * root of the reduce op. So, I will write the 
+                /* I am the root of the leader-comm, and the
+                 * root of the reduce op. So, I will write the
                  * final result directly into my recvbuf */
-                if(tmp_buf != recvbuf) { 
+                if(tmp_buf != recvbuf) {
                     in_buf = tmp_buf;
                     out_buf = recvbuf;
-                } else { 
+                } else {
 
                      in_buf = (char *)smpi_get_tmp_sendbuffer(count*
                                        datatype->get_extent());
                      Datatype::copy(tmp_buf, count, datatype,
                                         in_buf, count, datatype);
-                    //in_buf = MPI_IN_PLACE; 
-                    out_buf = recvbuf; 
-                } 
+                    //in_buf = MPI_IN_PLACE;
+                    out_buf = recvbuf;
+                }
             } else {
                 in_buf = (char *)smpi_get_tmp_sendbuffer(count*
                                        datatype->get_extent());
@@ -291,7 +291,7 @@ int Coll_reduce_mvapich2_two_level::reduce( void *sendbuf,
       smpi_free_tmp_buffer((void *) ((char *) tmp_buf + true_lb));
 
       if (leader_comm_rank == leader_root) {
-        if (my_rank != root || (my_rank == root && tmp_buf == recvbuf)) { 
+        if (my_rank != root || (my_rank == root && tmp_buf == recvbuf)) {
           smpi_free_tmp_buffer(in_buf);
         }
       }
diff --git a/src/smpi/colls/reduce/reduce-ompi.cpp b/src/smpi/colls/reduce/reduce-ompi.cpp
index b476889ae2..25d1cb85f1 100644
--- a/src/smpi/colls/reduce/reduce-ompi.cpp
+++ b/src/smpi/colls/reduce/reduce-ompi.cpp
@@ -38,7 +38,7 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
  * the number of datatype to the original count (original_count)
  *
  * Note that for non-commutative operations we cannot save memory copy
- * for the first block: thus we must copy sendbuf to accumbuf on intermediate 
+ * for the first block: thus we must copy sendbuf to accumbuf on intermediate
  * to keep the optimized loop happy.
  */
 int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int original_count,
@@ -63,33 +63,33 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
     num_segments = (original_count + count_by_segment - 1) / count_by_segment;
     segment_increment = count_by_segment * extent;
 
-    sendtmpbuf = (char*) sendbuf; 
-    if( sendbuf == MPI_IN_PLACE ) { 
-        sendtmpbuf = (char *)recvbuf; 
+    sendtmpbuf = (char*) sendbuf;
+    if( sendbuf == MPI_IN_PLACE ) {
+        sendtmpbuf = (char *)recvbuf;
     }
 
     XBT_DEBUG( "coll:tuned:reduce_generic count %d, msg size %ld, segsize %ld, max_requests %d", original_count, (unsigned long)(num_segments * segment_increment), (unsigned long)segment_increment, max_outstanding_reqs);
 
     rank = comm->rank();
 
-    /* non-leaf nodes - wait for children to send me data & forward up 
+    /* non-leaf nodes - wait for children to send me data & forward up
        (if needed) */
     if( tree->tree_nextsize > 0 ) {
         ptrdiff_t true_extent, real_segment_size;
         true_extent=datatype->get_extent();
 
-        /* handle non existant recv buffer (i.e. its NULL) and 
+        /* handle non existant recv buffer (i.e. its NULL) and
            protect the recv buffer on non-root nodes */
         accumbuf = (char*)recvbuf;
         if( (NULL == accumbuf) || (root != rank) ) {
             /* Allocate temporary accumulator buffer. */
             accumbuf_free = (char*)smpi_get_tmp_sendbuffer(true_extent +
                                           (original_count - 1) * extent);
-            if (accumbuf_free == NULL) { 
-                line = __LINE__; ret = -1; goto error_hndl; 
+            if (accumbuf_free == NULL) {
+                line = __LINE__; ret = -1; goto error_hndl;
             }
             accumbuf = accumbuf_free - lower_bound;
-        } 
+        }
 
         /* If this is a non-commutative operation we must copy
            sendbuf to the accumbuf, in order to simplfy the loops */
@@ -99,19 +99,19 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
         /* Allocate two buffers for incoming segments */
         real_segment_size = true_extent + (count_by_segment - 1) * extent;
         inbuf_free[0] = (char*) smpi_get_tmp_recvbuffer(real_segment_size);
-        if( inbuf_free[0] == NULL ) { 
-            line = __LINE__; ret = -1; goto error_hndl; 
+        if( inbuf_free[0] == NULL ) {
+            line = __LINE__; ret = -1; goto error_hndl;
         }
         inbuf[0] = inbuf_free[0] - lower_bound;
         /* if there is chance to overlap communication -
            allocate second buffer */
         if( (num_segments > 1) || (tree->tree_nextsize > 1) ) {
             inbuf_free[1] = (char*) smpi_get_tmp_recvbuffer(real_segment_size);
-            if( inbuf_free[1] == NULL ) { 
+            if( inbuf_free[1] == NULL ) {
                 line = __LINE__; ret = -1; goto error_hndl;
             }
             inbuf[1] = inbuf_free[1] - lower_bound;
-        } 
+        }
 
         /* reset input buffer index and receive count */
         inbi = 0;
@@ -134,14 +134,14 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
                 if( segindex < num_segments ) {
                     void* local_recvbuf = inbuf[inbi];
                     if( 0 == i ) {
-                        /* for the first step (1st child per segment) and 
-                         * commutative operations we might be able to irecv 
-                         * directly into the accumulate buffer so that we can 
-                         * reduce(op) this with our sendbuf in one step as 
-                         * ompi_op_reduce only has two buffer pointers, 
+                        /* for the first step (1st child per segment) and
+                         * commutative operations we might be able to irecv
+                         * directly into the accumulate buffer so that we can
+                         * reduce(op) this with our sendbuf in one step as
+                         * ompi_op_reduce only has two buffer pointers,
                          * this avoids an extra memory copy.
                          *
-                         * BUT if the operation is non-commutative or 
+                         * BUT if the operation is non-commutative or
                          * we are root and are USING MPI_IN_PLACE this is wrong!
                          */
                         if(  (op==MPI_OP_NULL || op->is_commutative()) &&
@@ -151,32 +151,32 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
                     }
 
                     reqs[inbi]=Request::irecv(local_recvbuf, recvcount, datatype,
-                                             tree->tree_next[i], 
+                                             tree->tree_next[i],
                                              COLL_TAG_REDUCE, comm
                                              );
                 }
                 /* wait for previous req to complete, if any.
-                   if there are no requests reqs[inbi ^1] will be 
+                   if there are no requests reqs[inbi ^1] will be
                    MPI_REQUEST_NULL. */
                 /* wait on data from last child for previous segment */
-                Request::waitall( 1, &reqs[inbi ^ 1], 
+                Request::waitall( 1, &reqs[inbi ^ 1],
                                              MPI_STATUSES_IGNORE );
                 local_op_buffer = inbuf[inbi ^ 1];
                 if( i > 0 ) {
-                    /* our first operation is to combine our own [sendbuf] data 
-                     * with the data we recvd from down stream (but only 
-                     * the operation is commutative and if we are not root and 
+                    /* our first operation is to combine our own [sendbuf] data
+                     * with the data we recvd from down stream (but only
+                     * the operation is commutative and if we are not root and
                      * not using MPI_IN_PLACE)
                      */
                     if( 1 == i ) {
-                        if( (op==MPI_OP_NULL || op->is_commutative())&& 
+                        if( (op==MPI_OP_NULL || op->is_commutative())&&
                             !((MPI_IN_PLACE == sendbuf) && (rank == tree->tree_root)) ) {
                             local_op_buffer = sendtmpbuf + segindex * segment_increment;
                         }
                     }
                     /* apply operation */
-                    if(op!=MPI_OP_NULL) op->apply( local_op_buffer, 
-                                   accumbuf + segindex * segment_increment, 
+                    if(op!=MPI_OP_NULL) op->apply( local_op_buffer,
+                                   accumbuf + segindex * segment_increment,
                                    &recvcount, datatype );
                 } else if ( segindex > 0 ) {
                     void* accumulator = accumbuf + (segindex-1) * segment_increment;
@@ -186,21 +186,21 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
                             local_op_buffer = sendtmpbuf + (segindex-1) * segment_increment;
                         }
                     }
-                    if(op!=MPI_OP_NULL) op->apply( local_op_buffer, accumulator, &prevcount, 
+                    if(op!=MPI_OP_NULL) op->apply( local_op_buffer, accumulator, &prevcount,
                                    datatype );
 
-                    /* all reduced on available data this step (i) complete, 
+                    /* all reduced on available data this step (i) complete,
                      * pass to the next process unless you are the root.
                      */
                     if (rank != tree->tree_root) {
                         /* send combined/accumulated data to parent */
-                        Request::send( accumulator, prevcount, 
-                                                  datatype, tree->tree_prev, 
+                        Request::send( accumulator, prevcount,
+                                                  datatype, tree->tree_prev,
                                                   COLL_TAG_REDUCE,
                                                   comm);
                     }
 
-                    /* we stop when segindex = number of segments 
+                    /* we stop when segindex = number of segments
                        (i.e. we do num_segment+1 steps for pipelining */
                     if (segindex == num_segments) break;
                 }
@@ -216,33 +216,33 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
         smpi_free_tmp_buffer(accumbuf_free);
     }
 
-    /* leaf nodes 
-       Depending on the value of max_outstanding_reqs and 
+    /* leaf nodes
+       Depending on the value of max_outstanding_reqs and
        the number of segments we have two options:
        - send all segments using blocking send to the parent, or
-       - avoid overflooding the parent nodes by limiting the number of 
+       - avoid overflooding the parent nodes by limiting the number of
        outstanding requests to max_oustanding_reqs.
-       TODO/POSSIBLE IMPROVEMENT: If there is a way to determine the eager size 
-       for the current communication, synchronization should be used only 
+       TODO/POSSIBLE IMPROVEMENT: If there is a way to determine the eager size
+       for the current communication, synchronization should be used only
        when the message/segment size is smaller than the eager size.
     */
     else {
 
         /* If the number of segments is less than a maximum number of oustanding
-           requests or there is no limit on the maximum number of outstanding 
+           requests or there is no limit on the maximum number of outstanding
            requests, we send data to the parent using blocking send */
-        if ((0 == max_outstanding_reqs) || 
+        if ((0 == max_outstanding_reqs) ||
             (num_segments <= max_outstanding_reqs)) {
-            
+
             segindex = 0;
             while ( original_count > 0) {
                 if (original_count < count_by_segment) {
                     count_by_segment = original_count;
                 }
-                Request::send((char*)sendbuf + 
+                Request::send((char*)sendbuf +
                                          segindex * segment_increment,
                                          count_by_segment, datatype,
-                                         tree->tree_prev, 
+                                         tree->tree_prev,
                                          COLL_TAG_REDUCE,
                                          comm) ;
                 segindex++;
@@ -270,7 +270,7 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
                 sreq[segindex]=Request::isend((char*)sendbuf +
                                           segindex * segment_increment,
                                           count_by_segment, datatype,
-                                          tree->tree_prev, 
+                                          tree->tree_prev,
                                           COLL_TAG_REDUCE,
                                           comm);
                 original_count -= count_by_segment;
@@ -285,10 +285,10 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
                 if( original_count < count_by_segment ) {
                     count_by_segment = original_count;
                 }
-                sreq[creq]=Request::isend((char*)sendbuf + 
-                                          segindex * segment_increment, 
-                                          count_by_segment, datatype, 
-                                          tree->tree_prev, 
+                sreq[creq]=Request::isend((char*)sendbuf +
+                                          segindex * segment_increment,
+                                          count_by_segment, datatype,
+                                          tree->tree_prev,
                                           COLL_TAG_REDUCE,
                                           comm );
                 creq = (creq + 1) % max_outstanding_reqs;
@@ -297,7 +297,7 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
             }
 
             /* Wait on the remaining request to complete */
-            Request::waitall( max_outstanding_reqs, sreq, 
+            Request::waitall( max_outstanding_reqs, sreq,
                                          MPI_STATUSES_IGNORE );
 
             /* free requests */
@@ -308,7 +308,7 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
     return MPI_SUCCESS;
 
  error_hndl:  /* error handler */
-    XBT_DEBUG("ERROR_HNDL: node %d file %s line %d error %d\n", 
+    XBT_DEBUG("ERROR_HNDL: node %d file %s line %d error %d\n",
                    rank, __FILE__, line, ret );
     if( inbuf_free[0] != NULL ) free(inbuf_free[0]);
     if( inbuf_free[1] != NULL ) free(inbuf_free[1]);
@@ -325,8 +325,8 @@ int smpi_coll_tuned_ompi_reduce_generic( void* sendbuf, void* recvbuf, int origi
 
 
 int Coll_reduce_ompi_chain::reduce( void *sendbuf, void *recvbuf, int count,
-                                        MPI_Datatype datatype, 
-                                        MPI_Op  op, int root, 
+                                        MPI_Datatype datatype,
+                                        MPI_Op  op, int root,
                                         MPI_Comm  comm
                                         )
 {
@@ -342,12 +342,12 @@ int Coll_reduce_ompi_chain::reduce( void *sendbuf, void *recvbuf, int count,
      * sent per operation
      */
     typelng = datatype->size();
-    
+
     COLL_TUNED_COMPUTED_SEGCOUNT( segsize, typelng, segcount );
 
-    return smpi_coll_tuned_ompi_reduce_generic( sendbuf, recvbuf, count, datatype, 
+    return smpi_coll_tuned_ompi_reduce_generic( sendbuf, recvbuf, count, datatype,
                                            op, root, comm,
-                                           ompi_coll_tuned_topo_build_chain(fanout, comm, root), 
+                                           ompi_coll_tuned_topo_build_chain(fanout, comm, root),
                                            segcount, 0 );
 }
 
@@ -373,16 +373,16 @@ int Coll_reduce_ompi_pipeline::reduce( void *sendbuf, void *recvbuf,
     const double b4 =  1.6761;
     typelng= datatype->size();
     int communicator_size = comm->size();
-    size_t message_size = typelng * count; 
+    size_t message_size = typelng * count;
 
     if (communicator_size > (a2 * message_size + b2)) {
-        // Pipeline_1K 
+        // Pipeline_1K
         segsize = 1024;
     }else if (communicator_size > (a4 * message_size + b4)) {
-        // Pipeline_32K 
+        // Pipeline_32K
         segsize = 32*1024;
     } else {
-        // Pipeline_64K 
+        // Pipeline_64K
         segsize = 64*1024;
     }
 
@@ -391,9 +391,9 @@ int Coll_reduce_ompi_pipeline::reduce( void *sendbuf, void *recvbuf,
 
     COLL_TUNED_COMPUTED_SEGCOUNT( segsize, typelng, segcount );
 
-    return smpi_coll_tuned_ompi_reduce_generic( sendbuf, recvbuf, count, datatype, 
+    return smpi_coll_tuned_ompi_reduce_generic( sendbuf, recvbuf, count, datatype,
                                            op, root, comm,
-                                           ompi_coll_tuned_topo_build_chain( 1, comm, root), 
+                                           ompi_coll_tuned_topo_build_chain( 1, comm, root),
                                            segcount, 0);
 }
 
@@ -414,7 +414,7 @@ int Coll_reduce_ompi_binary::reduce( void *sendbuf, void *recvbuf,
      */
     typelng=datatype->size();
 
-        // Binary_32K 
+        // Binary_32K
     segsize = 32*1024;
 
     XBT_DEBUG("coll:tuned:reduce_intra_binary rank %d ss %5d",
@@ -422,9 +422,9 @@ int Coll_reduce_ompi_binary::reduce( void *sendbuf, void *recvbuf,
 
     COLL_TUNED_COMPUTED_SEGCOUNT( segsize, typelng, segcount );
 
-    return smpi_coll_tuned_ompi_reduce_generic( sendbuf, recvbuf, count, datatype, 
-                                           op, root, comm, 
-                                           ompi_coll_tuned_topo_build_tree(2, comm, root), 
+    return smpi_coll_tuned_ompi_reduce_generic( sendbuf, recvbuf, count, datatype,
+                                           op, root, comm,
+                                           ompi_coll_tuned_topo_build_tree(2, comm, root),
                                            segcount, 0);
 }
 
@@ -449,13 +449,13 @@ int Coll_reduce_ompi_binomial::reduce( void *sendbuf, void *recvbuf,
      */
     typelng= datatype->size();
     int communicator_size = comm->size();
-    size_t message_size = typelng * count; 
+    size_t message_size = typelng * count;
     if (((communicator_size < 8) && (message_size < 20480)) ||
                (message_size < 2048) || (count <= 1)) {
         /* Binomial_0K */
         segsize = 0;
     } else if (communicator_size > (a1 * message_size + b1)) {
-        // Binomial_1K 
+        // Binomial_1K
         segsize = 1024;
     }
 
@@ -463,21 +463,21 @@ int Coll_reduce_ompi_binomial::reduce( void *sendbuf, void *recvbuf,
                  comm->rank(), segsize);
     COLL_TUNED_COMPUTED_SEGCOUNT( segsize, typelng, segcount );
 
-    return smpi_coll_tuned_ompi_reduce_generic( sendbuf, recvbuf, count, datatype, 
-                                           op, root, comm, 
-                                           ompi_coll_tuned_topo_build_in_order_bmtree(comm, root), 
+    return smpi_coll_tuned_ompi_reduce_generic( sendbuf, recvbuf, count, datatype,
+                                           op, root, comm,
+                                           ompi_coll_tuned_topo_build_in_order_bmtree(comm, root),
                                            segcount, 0);
 }
 
 /*
- * reduce_intra_in_order_binary 
- * 
+ * reduce_intra_in_order_binary
+ *
  * Function:      Logarithmic reduce operation for non-commutative operations.
  * Acecpts:       same as MPI_Reduce()
  * Returns:       MPI_SUCCESS or error code
  */
 int Coll_reduce_ompi_in_order_binary::reduce( void *sendbuf, void *recvbuf,
-                                                  int count, 
+                                                  int count,
                                                   MPI_Datatype datatype,
                                                   MPI_Op  op, int root,
                                                   MPI_Comm  comm)
@@ -503,9 +503,9 @@ int Coll_reduce_ompi_in_order_binary::reduce( void *sendbuf, void *recvbuf,
 
     /* An in-order binary tree must use root (size-1) to preserve the order of
        operations.  Thus, if root is not rank (size - 1), then we must handle
-       1. MPI_IN_PLACE option on real root, and 
+       1. MPI_IN_PLACE option on real root, and
        2. we must allocate temporary recvbuf on rank (size - 1).
-       Note that generic function must be careful not to switch order of 
+       Note that generic function must be careful not to switch order of
        operations for non-commutative ops.
     */
     io_root = size - 1;
@@ -514,7 +514,7 @@ int Coll_reduce_ompi_in_order_binary::reduce( void *sendbuf, void *recvbuf,
     if (io_root != root) {
         ptrdiff_t text, ext;
         char *tmpbuf = NULL;
-    
+
         ext=datatype->get_extent();
         text=datatype->get_extent();
 
@@ -538,8 +538,8 @@ int Coll_reduce_ompi_in_order_binary::reduce( void *sendbuf, void *recvbuf,
 
     /* Use generic reduce with in-order binary tree topology and io_root */
     ret = smpi_coll_tuned_ompi_reduce_generic( use_this_sendbuf, use_this_recvbuf, count, datatype,
-                                          op, io_root, comm, 
-                                          ompi_coll_tuned_topo_build_in_order_bintree(comm), 
+                                          op, io_root, comm,
+                                          ompi_coll_tuned_topo_build_in_order_bintree(comm),
                                           segcount, 0 );
     if (MPI_SUCCESS != ret) { return ret; }
 
@@ -553,7 +553,7 @@ int Coll_reduce_ompi_in_order_binary::reduce( void *sendbuf, void *recvbuf,
             if (MPI_IN_PLACE == sendbuf) {
               smpi_free_tmp_buffer(use_this_sendbuf);
             }
-          
+
         } else if (io_root == rank) {
             /* Send result from use_this_recvbuf to root */
             Request::send(use_this_recvbuf, count, datatype, root,
@@ -569,8 +569,8 @@ int Coll_reduce_ompi_in_order_binary::reduce( void *sendbuf, void *recvbuf,
 /*
  * Linear functions are copied from the BASIC coll module
  * they do not segment the message and are simple implementations
- * but for some small number of nodes and/or small data sizes they 
- * are just as fast as tuned/tree based segmenting operations 
+ * but for some small number of nodes and/or small data sizes they
+ * are just as fast as tuned/tree based segmenting operations
  * and as such may be selected by the decision functions
  * These are copied into this module due to the way we select modules
  * in V1. i.e. in V2 we will handle this differently and so will not
@@ -618,7 +618,7 @@ Coll_reduce_ompi_basic_linear::reduce(void *sbuf, void *rbuf, int count,
         return MPI_SUCCESS;
     }
 
-    /* see discussion in ompi_coll_basic_reduce_lin_intra about 
+    /* see discussion in ompi_coll_basic_reduce_lin_intra about
        extent and true extent */
     /* for reducing buffer allocation lengths.... */
 
diff --git a/src/smpi/colls/reduce/reduce-scatter-gather.cpp b/src/smpi/colls/reduce/reduce-scatter-gather.cpp
index cc0eb118af..4b987aa2f9 100644
--- a/src/smpi/colls/reduce/reduce-scatter-gather.cpp
+++ b/src/smpi/colls/reduce/reduce-scatter-gather.cpp
@@ -34,7 +34,7 @@ int Coll_reduce_scatter_gather::reduce(void *sendbuf, void *recvbuf,
     return 0;
   rank = comm->rank();
   comm_size = comm->size();
-  
+
 
 
   extent = datatype->get_extent();
diff --git a/src/smpi/colls/reduce_scatter/reduce_scatter-mpich.cpp b/src/smpi/colls/reduce_scatter/reduce_scatter-mpich.cpp
index 60f03bf7a1..39128a2e9f 100644
--- a/src/smpi/colls/reduce_scatter/reduce_scatter-mpich.cpp
+++ b/src/smpi/colls/reduce_scatter/reduce_scatter-mpich.cpp
@@ -27,7 +27,7 @@ int Coll_reduce_scatter_mpich_pair::reduce_scatter(void *sendbuf, void *recvbuf,
                               MPI_Datatype datatype, MPI_Op op, MPI_Comm comm)
 {
     int   rank, comm_size, i;
-    MPI_Aint extent, true_extent, true_lb; 
+    MPI_Aint extent, true_extent, true_lb;
     int  *disps;
     void *tmp_recvbuf;
     int mpi_errno = MPI_SUCCESS;
@@ -38,7 +38,7 @@ int Coll_reduce_scatter_mpich_pair::reduce_scatter(void *sendbuf, void *recvbuf,
 
     extent =datatype->get_extent();
     datatype->extent(&true_lb, &true_extent);
-    
+
     if (op->is_commutative()) {
         is_commutative = 1;
     }
@@ -50,7 +50,7 @@ int Coll_reduce_scatter_mpich_pair::reduce_scatter(void *sendbuf, void *recvbuf,
         disps[i] = total_count;
         total_count += recvcounts[i];
     }
-    
+
     if (total_count == 0) {
         xbt_free(disps);
         return MPI_ERR_COUNT;
@@ -62,33 +62,33 @@ int Coll_reduce_scatter_mpich_pair::reduce_scatter(void *sendbuf, void *recvbuf,
                                        recvcounts[rank], datatype, recvbuf,
                                        recvcounts[rank], datatype);
         }
-        
+
         /* allocate temporary buffer to store incoming data */
         tmp_recvbuf = (void*)smpi_get_tmp_recvbuffer(recvcounts[rank]*(MAX(true_extent,extent))+1);
         /* adjust for potential negative lower bound in datatype */
         tmp_recvbuf = (void *)((char*)tmp_recvbuf - true_lb);
-        
+
         for (i=1; i<comm_size; i++) {
             src = (rank - i + comm_size) % comm_size;
             dst = (rank + i) % comm_size;
-            
+
             /* send the data that dst needs. recv data that this process
                needs from src into tmp_recvbuf */
-            if (sendbuf != MPI_IN_PLACE) 
-                Request::sendrecv(((char *)sendbuf+disps[dst]*extent), 
+            if (sendbuf != MPI_IN_PLACE)
+                Request::sendrecv(((char *)sendbuf+disps[dst]*extent),
                                              recvcounts[dst], datatype, dst,
                                              COLL_TAG_SCATTER, tmp_recvbuf,
                                              recvcounts[rank], datatype, src,
                                              COLL_TAG_SCATTER, comm,
                                              MPI_STATUS_IGNORE);
             else
-                Request::sendrecv(((char *)recvbuf+disps[dst]*extent), 
+                Request::sendrecv(((char *)recvbuf+disps[dst]*extent),
                                              recvcounts[dst], datatype, dst,
                                              COLL_TAG_SCATTER, tmp_recvbuf,
                                              recvcounts[rank], datatype, src,
                                              COLL_TAG_SCATTER, comm,
                                              MPI_STATUS_IGNORE);
-            
+
             if (is_commutative || (src < rank)) {
                 if (sendbuf != MPI_IN_PLACE) {
                   if (op != MPI_OP_NULL)
@@ -125,24 +125,24 @@ int Coll_reduce_scatter_mpich_pair::reduce_scatter(void *sendbuf, void *recvbuf,
                 }
             }
         }
-        
+
         /* if MPI_IN_PLACE, move output data to the beginning of
            recvbuf. already done for rank 0. */
         if ((sendbuf == MPI_IN_PLACE) && (rank != 0)) {
             mpi_errno = Datatype::copy(((char *)recvbuf +
-                                        disps[rank]*extent),  
+                                        disps[rank]*extent),
                                        recvcounts[rank], datatype,
-                                       recvbuf, 
+                                       recvbuf,
                                        recvcounts[rank], datatype );
             if (mpi_errno) return(mpi_errno);
         }
-    
+
         xbt_free(disps);
         smpi_free_tmp_buffer(tmp_recvbuf);
 
         return MPI_SUCCESS;
 }
-    
+
 
 int Coll_reduce_scatter_mpich_noncomm::reduce_scatter(void *sendbuf, void *recvbuf, int recvcounts[],
                               MPI_Datatype datatype, MPI_Op op, MPI_Comm comm)
@@ -228,7 +228,7 @@ int Coll_reduce_scatter_mpich_noncomm::reduce_scatter(void *sendbuf, void *recvb
            is now our peer's responsibility */
         if (rank > peer) {
             /* higher ranked value so need to call op(received_data, my_data) */
-            if(op!=MPI_OP_NULL) op->apply( 
+            if(op!=MPI_OP_NULL) op->apply(
                    incoming_data + recv_offset*true_extent,
                      outgoing_data + recv_offset*true_extent,
                      &size, datatype );
@@ -265,7 +265,7 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
                               MPI_Datatype datatype, MPI_Op op, MPI_Comm comm)
 {
     int   rank, comm_size, i;
-    MPI_Aint extent, true_extent, true_lb; 
+    MPI_Aint extent, true_extent, true_lb;
     int  *disps;
     void *tmp_recvbuf, *tmp_results;
     int mpi_errno = MPI_SUCCESS;
@@ -279,7 +279,7 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
 
     extent =datatype->get_extent();
     datatype->extent(&true_lb, &true_extent);
-    
+
     if ((op==MPI_OP_NULL) || op->is_commutative()) {
         is_commutative = 1;
     }
@@ -291,7 +291,7 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
         disps[i] = total_count;
         total_count += recvcounts[i];
     }
-    
+
             /* noncommutative and (non-pof2 or block irregular), use recursive doubling. */
 
             /* need to allocate temporary buffer to receive incoming data*/
@@ -349,7 +349,7 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
 
                 mpi_errno = Datatype::create_indexed(2, blklens, dis, datatype, &sendtype);
                 if (mpi_errno) return(mpi_errno);
-                
+
                 sendtype->commit();
 
                 /* calculate recvtype */
@@ -366,14 +366,14 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
 
                 mpi_errno = Datatype::create_indexed(2, blklens, dis, datatype, &recvtype);
                 if (mpi_errno) return(mpi_errno);
-                
+
                 recvtype->commit();
 
                 received = 0;
                 if (dst < comm_size) {
                     /* tmp_results contains data to be sent in each step. Data is
                        received in tmp_recvbuf and then accumulated into
-                       tmp_results. accumulation is done later below.   */ 
+                       tmp_results. accumulation is done later below.   */
 
                     Request::sendrecv(tmp_results, 1, sendtype, dst,
                                                  COLL_TAG_SCATTER,
@@ -396,7 +396,7 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
                        in a tree fashion. First find root of current tree
                        that is being divided into two. k is the number of
                        least-significant bits in this process's rank that
-                       must be zeroed out to find the rank of the root */ 
+                       must be zeroed out to find the rank of the root */
                     j = mask;
                     k = 0;
                     while (j) {
@@ -415,7 +415,7 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
                         /* send only if this proc has data and destination
                            doesn't have data. at any step, multiple processes
                            can send if they have the data */
-                        if ((dst > rank) && 
+                        if ((dst > rank) &&
                             (rank < tree_root + nprocs_completed)
                             && (dst >= tree_root + nprocs_completed)) {
                             /* send the current result */
@@ -425,12 +425,12 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
                         }
                         /* recv only if this proc. doesn't have data and sender
                            has data */
-                        else if ((dst < rank) && 
+                        else if ((dst < rank) &&
                                  (dst < tree_root + nprocs_completed) &&
                                  (rank >= tree_root + nprocs_completed)) {
                             Request::recv(tmp_recvbuf, 1, recvtype, dst,
                                                      COLL_TAG_SCATTER,
-                                                     comm, MPI_STATUS_IGNORE); 
+                                                     comm, MPI_STATUS_IGNORE);
                             received = 1;
                         }
                         tmp_mask >>= 1;
@@ -438,9 +438,9 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
                     }
                 }
 
-                /* The following reduction is done here instead of after 
+                /* The following reduction is done here instead of after
                    the MPIC_Sendrecv_ft or MPIC_Recv_ft above. This is
-                   because to do it above, in the noncommutative 
+                   because to do it above, in the noncommutative
                    case, we would need an extra temp buffer so as not to
                    overwrite temp_recvbuf, because temp_recvbuf may have
                    to be communicated to other processes in the
@@ -465,7 +465,7 @@ int Coll_reduce_scatter_mpich_rdb::reduce_scatter(void *sendbuf, void *recvbuf,
                                       &blklens[1], datatype);
                         }
                         /* copy result back into tmp_results */
-                        mpi_errno = Datatype::copy(tmp_recvbuf, 1, recvtype, 
+                        mpi_errno = Datatype::copy(tmp_recvbuf, 1, recvtype,
                                                    tmp_results, 1, recvtype);
                         if (mpi_errno) return(mpi_errno);
                     }
diff --git a/src/smpi/colls/reduce_scatter/reduce_scatter-ompi.cpp b/src/smpi/colls/reduce_scatter/reduce_scatter-ompi.cpp
index 84fd73740f..8a28d8854a 100644
--- a/src/smpi/colls/reduce_scatter/reduce_scatter-ompi.cpp
+++ b/src/smpi/colls/reduce_scatter/reduce_scatter-ompi.cpp
@@ -26,7 +26,7 @@
 
 /*
  * Recursive-halving function is (*mostly*) copied from the BASIC coll module.
- * I have removed the part which handles "large" message sizes 
+ * I have removed the part which handles "large" message sizes
  * (non-overlapping version of reduce_Scatter).
  */
 
@@ -35,7 +35,7 @@
 /*
  *  reduce_scatter_ompi_basic_recursivehalving
  *
- *  Function:   - reduce scatter implementation using recursive-halving 
+ *  Function:   - reduce scatter implementation using recursive-halving
  *                algorithm
  *  Accepts:    - same as MPI_Reduce_scatter()
  *  Returns:    - MPI_SUCCESS or error code
@@ -44,8 +44,8 @@
 namespace simgrid{
 namespace smpi{
 int
-Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf, 
-                                                            void *rbuf, 
+Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
+                                                            void *rbuf,
                                                             int *rcounts,
                                                             MPI_Datatype dtype,
                                                             MPI_Op op,
@@ -57,11 +57,11 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
     ptrdiff_t true_lb, true_extent, lb, extent, buf_size;
     char *recv_buf = NULL, *recv_buf_free = NULL;
     char *result_buf = NULL, *result_buf_free = NULL;
-   
+
     /* Initialize */
     rank = comm->rank();
     size = comm->size();
-   
+
     XBT_DEBUG("coll:tuned:reduce_scatter_ompi_basic_recursivehalving, rank %d", rank);
     if ((op != MPI_OP_NULL && not op->is_commutative()))
       THROWF(arg_error,0, " reduce_scatter ompi_basic_recursivehalving can only be used for commutative operations! ");
@@ -100,29 +100,29 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
         err = MPI_ERR_OTHER;
         goto cleanup;
     }
-   
+
     /* allocate temporary buffer for results */
     result_buf_free = (char*) smpi_get_tmp_sendbuffer(buf_size);
 
     result_buf = result_buf_free - lb;
-   
+
     /* copy local buffer into the temporary results */
     err =Datatype::copy(sbuf, count, dtype, result_buf, count, dtype);
     if (MPI_SUCCESS != err) goto cleanup;
-   
+
     /* figure out power of two mapping: grow until larger than
        comm size, then go back one, to get the largest power of
        two less than comm size */
     while (tmp_size <= size) tmp_size <<= 1;
     tmp_size >>= 1;
     remain = size - tmp_size;
-   
+
     /* If comm size is not a power of two, have the first "remain"
        procs with an even rank send to rank + 1, leaving a power of
        two procs to do the rest of the algorithm */
     if (rank < 2 * remain) {
         if ((rank & 1) == 0) {
-            Request::send(result_buf, count, dtype, rank + 1, 
+            Request::send(result_buf, count, dtype, rank + 1,
                                     COLL_TAG_REDUCE_SCATTER,
                                     comm);
             /* we don't participate from here on out */
@@ -131,10 +131,10 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
             Request::recv(recv_buf, count, dtype, rank - 1,
                                     COLL_TAG_REDUCE_SCATTER,
                                     comm, MPI_STATUS_IGNORE);
-         
+
             /* integrate their results into our temp results */
             if(op!=MPI_OP_NULL) op->apply( recv_buf, result_buf, &count, dtype);
-         
+
             /* adjust rank to be the bottom "remain" ranks */
             tmp_rank = rank / 2;
         }
@@ -143,13 +143,13 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
            remain" ranks dropped out */
         tmp_rank = rank - remain;
     }
-   
+
     /* For ranks not kicked out by the above code, perform the
        recursive halving */
     if (tmp_rank >= 0) {
         int *tmp_disps = NULL, *tmp_rcounts = NULL;
         int mask, send_index, recv_index, last_index;
-      
+
         /* recalculate disps and rcounts to account for the
            special "remainder" processes that are no longer doing
            anything */
@@ -224,18 +224,18 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
                     xbt_free(tmp_rcounts);
                     xbt_free(tmp_disps);
                     goto cleanup;
-                }                                             
+                }
             }
             if (recv_count > 0 && send_count != 0) {
                 Request::send(result_buf + (ptrdiff_t)tmp_disps[send_index] * extent,
-                                        send_count, dtype, peer, 
+                                        send_count, dtype, peer,
                                         COLL_TAG_REDUCE_SCATTER,
                                         comm);
                 if (MPI_SUCCESS != err) {
                     xbt_free(tmp_rcounts);
                     xbt_free(tmp_disps);
                     goto cleanup;
-                }                                             
+                }
             }
             if (send_count > 0 && recv_count != 0) {
                 Request::wait(&request, MPI_STATUS_IGNORE);
@@ -244,8 +244,8 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
             /* if we received something on this step, push it into
                the results buffer */
             if (recv_count > 0) {
-                if(op!=MPI_OP_NULL) op->apply( 
-                               recv_buf + (ptrdiff_t)tmp_disps[recv_index] * extent, 
+                if(op!=MPI_OP_NULL) op->apply(
+                               recv_buf + (ptrdiff_t)tmp_disps[recv_index] * extent,
                                result_buf + (ptrdiff_t)tmp_disps[recv_index] * extent,
                                &recv_count, dtype);
             }
@@ -259,13 +259,13 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
         /* copy local results from results buffer into real receive buffer */
         if (0 != rcounts[rank]) {
             err = Datatype::copy(result_buf + disps[rank] * extent,
-                                       rcounts[rank], dtype, 
+                                       rcounts[rank], dtype,
                                        rbuf, rcounts[rank], dtype);
             if (MPI_SUCCESS != err) {
                 xbt_free(tmp_rcounts);
                 xbt_free(tmp_disps);
                 goto cleanup;
-            }                                             
+            }
         }
 
         xbt_free(tmp_rcounts);
@@ -288,7 +288,7 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
                                         COLL_TAG_REDUCE_SCATTER,
                                         comm);
             }
-        }            
+        }
     }
 
  cleanup:
@@ -309,12 +309,12 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
  *   Accepts:        Same as MPI_Reduce_scatter()
  *   Returns:        MPI_SUCCESS or error code
  *
- *   Description:    Implements ring algorithm for reduce_scatter: 
- *                   the block sizes defined in rcounts are exchanged and 
+ *   Description:    Implements ring algorithm for reduce_scatter:
+ *                   the block sizes defined in rcounts are exchanged and
  8                    updated until they reach proper destination.
  *                   Algorithm requires 2 * max(rcounts) extra buffering
  *
- *   Limitations:    The algorithm DOES NOT preserve order of operations so it 
+ *   Limitations:    The algorithm DOES NOT preserve order of operations so it
  *                   can be used only for commutative operations.
  *         Example on 5 nodes:
  *         Initial state
@@ -326,7 +326,7 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
  *        [04]  ->       [14]          [24]           [34]           [44]
  *
  *        COMPUTATION PHASE
- *         Step 0: rank r sends block (r-1) to rank (r+1) and 
+ *         Step 0: rank r sends block (r-1) to rank (r+1) and
  *                 receives block (r+1) from rank (r-1) [with wraparound].
  *   #      0              1             2              3             4
  *        [00]           [10]        [10+20]   ->     [30]           [40]
@@ -334,12 +334,12 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
  *    ->  [02]           [12]          [22]           [32]         [32+42] -->..
  *      [43+03] ->       [13]          [23]           [33]           [43]
  *        [04]         [04+14]  ->     [24]           [34]           [44]
- *         
+ *
  *         Step 1:
  *   #      0              1             2              3             4
  *        [00]           [10]        [10+20]       [10+20+30] ->     [40]
  *    ->  [01]           [11]          [21]          [21+31]      [21+31+41] ->
- *     [32+42+02] ->     [12]          [22]           [32]         [32+42] 
+ *     [32+42+02] ->     [12]          [22]           [32]         [32+42]
  *        [03]        [43+03+13] ->    [23]           [33]           [43]
  *        [04]         [04+14]      [04+14+24]  ->    [34]           [44]
  *
@@ -347,7 +347,7 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
  *   #      0              1             2              3             4
  *     -> [00]           [10]        [10+20]       [10+20+30]   [10+20+30+40] ->
  *   [21+31+41+01]->     [11]          [21]          [21+31]      [21+31+41]
- *     [32+42+02]   [32+42+02+12]->    [22]           [32]         [32+42] 
+ *     [32+42+02]   [32+42+02+12]->    [22]           [32]         [32+42]
  *        [03]        [43+03+13]   [43+03+13+23]->    [33]           [43]
  *        [04]         [04+14]      [04+14+24]    [04+14+24+34] ->   [44]
  *
@@ -355,13 +355,13 @@ Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(void *sbuf,
  *   #      0             1              2              3             4
  * [10+20+30+40+00]     [10]         [10+20]       [10+20+30]   [10+20+30+40]
  *  [21+31+41+01] [21+31+41+01+11]     [21]          [21+31]      [21+31+41]
- *    [32+42+02]   [32+42+02+12] [32+42+02+12+22]     [32]         [32+42] 
+ *    [32+42+02]   [32+42+02+12] [32+42+02+12+22]     [32]         [32+42]
  *       [03]        [43+03+13]    [43+03+13+23] [43+03+13+23+33]    [43]
  *       [04]         [04+14]       [04+14+24]    [04+14+24+34] [04+14+24+34+44]
  *    DONE :)
  *
  */
-int 
+int
 Coll_reduce_scatter_ompi_ring::reduce_scatter(void *sbuf, void *rbuf, int *rcounts,
                                           MPI_Datatype dtype,
                                           MPI_Op op,
@@ -378,10 +378,10 @@ Coll_reduce_scatter_ompi_ring::reduce_scatter(void *sbuf, void *rbuf, int *rcoun
     size = comm->size();
     rank = comm->rank();
 
-    XBT_DEBUG(  "coll:tuned:reduce_scatter_ompi_ring rank %d, size %d", 
+    XBT_DEBUG(  "coll:tuned:reduce_scatter_ompi_ring rank %d, size %d",
                  rank, size);
 
-    /* Determine the maximum number of elements per node, 
+    /* Determine the maximum number of elements per node,
        corresponding block size, and displacements array.
     */
     displs = (int*) xbt_malloc(size * sizeof(int));
@@ -389,12 +389,12 @@ Coll_reduce_scatter_ompi_ring::reduce_scatter(void *sbuf, void *rbuf, int *rcoun
     displs[0] = 0;
     total_count = rcounts[0];
     max_block_count = rcounts[0];
-    for (i = 1; i < size; i++) { 
+    for (i = 1; i < size; i++) {
         displs[i] = total_count;
         total_count += rcounts[i];
         if (max_block_count < rcounts[i]) max_block_count = rcounts[i];
     }
-      
+
     /* Special case for size == 1 */
     if (1 == size) {
         if (MPI_IN_PLACE != sbuf) {
@@ -438,7 +438,7 @@ Coll_reduce_scatter_ompi_ring::reduce_scatter(void *sbuf, void *rbuf, int *rcoun
 
     /* Computation loop */
 
-    /* 
+    /*
        For each of the remote nodes:
        - post irecv for block (r-2) from (r-1) with wrap around
        - send block (r-1) to (r+1)
@@ -468,23 +468,23 @@ Coll_reduce_scatter_ompi_ring::reduce_scatter(void *sbuf, void *rbuf, int *rcoun
 
     for (k = 2; k < size; k++) {
         const int prevblock = (rank + size - k) % size;
-      
+
         inbi = inbi ^ 0x1;
 
         /* Post irecv for the current block */
         reqs[inbi]=Request::irecv(inbuf[inbi], max_block_count, dtype, recv_from,
                                  COLL_TAG_REDUCE_SCATTER, comm
                                  );
-      
+
         /* Wait on previous block to arrive */
         Request::wait(&reqs[inbi ^ 0x1], MPI_STATUS_IGNORE);
-      
+
         /* Apply operation on previous block: result goes to rbuf
            rbuf[prevblock] = inbuf[inbi ^ 0x1] (op) rbuf[prevblock]
         */
         tmprecv = accumbuf + (ptrdiff_t)displs[prevblock] * extent;
         if(op!=MPI_OP_NULL) op->apply( inbuf[inbi ^ 0x1], tmprecv, &(rcounts[prevblock]), dtype);
-      
+
         /* send previous block to send_to */
         Request::send(tmprecv, rcounts[prevblock], dtype, send_to,
                                 COLL_TAG_REDUCE_SCATTER,
@@ -498,7 +498,7 @@ Coll_reduce_scatter_ompi_ring::reduce_scatter(void *sbuf, void *rbuf, int *rcoun
        rbuf[rank] = inbuf[inbi] (op) rbuf[rank] */
     tmprecv = accumbuf + (ptrdiff_t)displs[rank] * extent;
     if(op!=MPI_OP_NULL) op->apply( inbuf[inbi], tmprecv, &(rcounts[rank]), dtype);
-   
+
     /* Copy result from tmprecv to rbuf */
     ret = Datatype::copy(tmprecv, rcounts[rank], dtype, (char*)rbuf, rcounts[rank], dtype);
     if (ret < 0) { line = __LINE__; goto error_hndl; }
diff --git a/src/smpi/colls/scatter/scatter-mvapich-two-level.cpp b/src/smpi/colls/scatter/scatter-mvapich-two-level.cpp
index cff366cd24..cd701f5f58 100644
--- a/src/smpi/colls/scatter/scatter-mvapich-two-level.cpp
+++ b/src/smpi/colls/scatter/scatter-mvapich-two-level.cpp
@@ -67,7 +67,7 @@ int Coll_scatter_mvapich2_two_level_direct::scatter(void *sendbuf,
     //if not set (use of the algo directly, without mvapich2 selector)
     if(MV2_Scatter_intra_function==NULL)
       MV2_Scatter_intra_function=Coll_scatter_mpich::scatter;
-    
+
     if(comm->get_leaders_comm()==MPI_COMM_NULL){
       comm->init_smp();
     }
@@ -246,7 +246,7 @@ int Coll_scatter_mvapich2_two_level_binomial::scatter(void *sendbuf,
     //if not set (use of the algo directly, without mvapich2 selector)
     if(MV2_Scatter_intra_function==NULL)
       MV2_Scatter_intra_function=Coll_scatter_mpich::scatter;
-    
+
     if(comm->get_leaders_comm()==MPI_COMM_NULL){
       comm->init_smp();
     }
diff --git a/src/smpi/colls/scatter/scatter-ompi.cpp b/src/smpi/colls/scatter/scatter-ompi.cpp
index 9d14907810..5a2da7fa8f 100644
--- a/src/smpi/colls/scatter/scatter-ompi.cpp
+++ b/src/smpi/colls/scatter/scatter-ompi.cpp
@@ -40,7 +40,7 @@ int Coll_scatter_ompi_binomial::scatter(void* sbuf, int scount, MPI_Datatype sdt
     int err;
     ompi_coll_tree_t* bmtree;
     MPI_Status status;
-    MPI_Aint sextent, slb, strue_lb, strue_extent; 
+    MPI_Aint sextent, slb, strue_lb, strue_extent;
     MPI_Aint rextent, rlb, rtrue_lb, rtrue_extent;
 
     size = comm->size();
@@ -50,7 +50,7 @@ int Coll_scatter_ompi_binomial::scatter(void* sbuf, int scount, MPI_Datatype sdt
                  "Coll_scatter_ompi_binomial::scatter rank %d", rank);
 
     /* create the binomial tree */
-    
+
 //    COLL_TUNED_UPDATE_IN_ORDER_BMTREE( comm, tuned_module, root );
     bmtree =  ompi_coll_tuned_topo_build_in_order_bmtree( comm, root);//ompi_ data->cached_in_order_bmtree;
 
@@ -175,8 +175,8 @@ int Coll_scatter_ompi_binomial::scatter(void* sbuf, int scount, MPI_Datatype sdt
 /*
  * Linear functions are copied from the BASIC coll module
  * they do not segment the message and are simple implementations
- * but for some small number of nodes and/or small data sizes they 
- * are just as fast as tuned/tree based segmenting operations 
+ * but for some small number of nodes and/or small data sizes they
+ * are just as fast as tuned/tree based segmenting operations
  * and as such may be selected by the decision functions
  * These are copied into this module due to the way we select modules
  * in V1. i.e. in V2 we will handle this differently and so will not
diff --git a/src/smpi/colls/smpi_default_selector.cpp b/src/smpi/colls/smpi_default_selector.cpp
index 6c48761749..d674e4ad56 100644
--- a/src/smpi/colls/smpi_default_selector.cpp
+++ b/src/smpi/colls/smpi_default_selector.cpp
@@ -222,7 +222,7 @@ int Coll_reduce_default::reduce(void *sendbuf, void *recvbuf, int count, MPI_Dat
     sendtmpbuf = static_cast<char *>(smpi_get_tmp_sendbuffer(count*datatype->get_extent()));
     Datatype::copy(recvbuf, count, datatype,sendtmpbuf, count, datatype);
   }
-  
+
   if(rank != root) {
     // Send buffer to root
     Request::send(sendtmpbuf, count, datatype, root, system_tag, comm);
@@ -337,7 +337,7 @@ int Coll_alltoallv_default::alltoallv(void *sendbuf, int *sendcounts, int *sendd
     XBT_DEBUG("<%d> wait for %d requests", rank, count);
     Request::waitall(count, requests, MPI_STATUS_IGNORE);
     for(i = 0; i < count; i++) {
-      if(requests[i]!=MPI_REQUEST_NULL) 
+      if(requests[i]!=MPI_REQUEST_NULL)
         Request::unref(&requests[i]);
     }
     xbt_free(requests);
diff --git a/src/smpi/colls/smpi_intel_mpi_selector.cpp b/src/smpi/colls/smpi_intel_mpi_selector.cpp
index fd237fa597..6bec031ed8 100644
--- a/src/smpi/colls/smpi_intel_mpi_selector.cpp
+++ b/src/smpi/colls/smpi_intel_mpi_selector.cpp
@@ -41,7 +41,7 @@ MPI_Allreduce
 4 - Topology aware Reduce + Bcast algorithm
 5 - Binomial gather + scatter algorithm
 6 - Topology aware binominal gather + scatter algorithm
-7 - Shumilin's ring algorithm 
+7 - Shumilin's ring algorithm
 8 - Ring algorithm
 
   as Shumilin's ring algorithm is unknown, default to ring'
@@ -67,7 +67,7 @@ int (*intel_allreduce_functions_table[])(void *sendbuf,
 
 intel_tuning_table_element intel_allreduce_table[] =
 {
-  {1,{ 
+  {1,{
     { 2,9,{
       {6,7},
       {85,1},
@@ -382,13 +382,13 @@ intel_tuning_table_element intel_allreduce_table[] =
 
 
 
-/*I_MPI_ADJUST_ALLTOALL 
+/*I_MPI_ADJUST_ALLTOALL
 
-MPI_Alltoall 
+MPI_Alltoall
 
-1. Bruck's algorithm 
-2. Isend/Irecv + waitall algorithm 
-3. Pair wise exchange algorithm 
+1. Bruck's algorithm
+2. Isend/Irecv + waitall algorithm
+3. Pair wise exchange algorithm
 4. Plum's algorithm
 
 */
@@ -634,10 +634,10 @@ intel_tuning_table_element intel_alltoall_table[] =
   }
   }
 };
-int (*intel_alltoall_functions_table[])(void *sbuf, int scount, 
+int (*intel_alltoall_functions_table[])(void *sbuf, int scount,
                                              MPI_Datatype sdtype,
-                                             void* rbuf, int rcount, 
-                                             MPI_Datatype rdtype, 
+                                             void* rbuf, int rcount,
+                                             MPI_Datatype rdtype,
                                              MPI_Comm comm) ={
       Coll_alltoall_bruck::alltoall,
       Coll_alltoall_mvapich2_scatter_dest::alltoall,
@@ -645,16 +645,16 @@ int (*intel_alltoall_functions_table[])(void *sbuf, int scount,
       Coll_alltoall_mvapich2::alltoall//Plum is proprietary ? (and super efficient)
 };
 
-/*I_MPI_ADJUST_BARRIER 
+/*I_MPI_ADJUST_BARRIER
 
-MPI_Barrier 
+MPI_Barrier
 
-1. Dissemination algorithm 
-2. Recursive doubling algorithm 
-3. Topology aware dissemination algorithm 
-4. Topology aware recursive doubling algorithm 
-5. Binominal gather + scatter algorithm 
-6. Topology aware binominal gather + scatter algorithm 
+1. Dissemination algorithm
+2. Recursive doubling algorithm
+3. Topology aware dissemination algorithm
+4. Topology aware recursive doubling algorithm
+5. Binominal gather + scatter algorithm
+6. Topology aware binominal gather + scatter algorithm
 
 */
 static int intel_barrier_gather_scatter(MPI_Comm comm){
@@ -785,17 +785,17 @@ intel_tuning_table_element intel_barrier_table[] =
 };
 
 
-/*I_MPI_ADJUST_BCAST 
+/*I_MPI_ADJUST_BCAST
 
-MPI_Bcast 
+MPI_Bcast
 
-1. Binomial algorithm 
-2. Recursive doubling algorithm 
-3. Ring algorithm 
-4. Topology aware binomial algorithm 
-5. Topology aware recursive doubling algorithm 
-6. Topology aware ring algorithm 
-7. Shumilin's bcast algorithm 
+1. Binomial algorithm
+2. Recursive doubling algorithm
+3. Ring algorithm
+4. Topology aware binomial algorithm
+5. Topology aware recursive doubling algorithm
+6. Topology aware ring algorithm
+7. Shumilin's bcast algorithm
 */
 
 int (*intel_bcast_functions_table[])(void *buff, int count,
@@ -954,15 +954,15 @@ intel_tuning_table_element intel_bcast_table[] =
 };
 
 
-/*I_MPI_ADJUST_REDUCE 
+/*I_MPI_ADJUST_REDUCE
 
-MPI_Reduce 
+MPI_Reduce
 
-1. Shumilin's algorithm 
-2. Binomial algorithm 
-3. Topology aware Shumilin's algorithm 
-4. Topology aware binomial algorithm 
-5. Rabenseifner's algorithm 
+1. Shumilin's algorithm
+2. Binomial algorithm
+3. Topology aware Shumilin's algorithm
+4. Topology aware binomial algorithm
+5. Rabenseifner's algorithm
 6. Topology aware Rabenseifner's algorithm
 
 */
@@ -1044,15 +1044,15 @@ intel_tuning_table_element intel_reduce_table[] =
   }
 };
 
-/* I_MPI_ADJUST_REDUCE_SCATTER 
+/* I_MPI_ADJUST_REDUCE_SCATTER
 
-MPI_Reduce_scatter 
+MPI_Reduce_scatter
 
-1. Recursive having algorithm 
-2. Pair wise exchange algorithm 
-3. Recursive doubling algorithm 
-4. Reduce + Scatterv algorithm 
-5. Topology aware Reduce + Scatterv algorithm 
+1. Recursive having algorithm
+2. Pair wise exchange algorithm
+3. Recursive doubling algorithm
+4. Reduce + Scatterv algorithm
+5. Topology aware Reduce + Scatterv algorithm
 
 */
 static  int intel_reduce_scatter_reduce_scatterv(void *sbuf, void *rbuf,
@@ -1476,21 +1476,21 @@ intel_tuning_table_element intel_reduce_scatter_table[] =
   }
 };
 
-/* I_MPI_ADJUST_ALLGATHER 
+/* I_MPI_ADJUST_ALLGATHER
 
-MPI_Allgather 
+MPI_Allgather
 
-1. Recursive doubling algorithm 
-2. Bruck's algorithm 
-3. Ring algorithm 
-4. Topology aware Gatherv + Bcast algorithm 
+1. Recursive doubling algorithm
+2. Bruck's algorithm
+3. Ring algorithm
+4. Topology aware Gatherv + Bcast algorithm
 
 */
 
-int (*intel_allgather_functions_table[])(void *sbuf, int scount, 
+int (*intel_allgather_functions_table[])(void *sbuf, int scount,
                                               MPI_Datatype sdtype,
-                                              void* rbuf, int rcount, 
-                                              MPI_Datatype rdtype, 
+                                              void* rbuf, int rcount,
+                                              MPI_Datatype rdtype,
                                               MPI_Comm  comm
                                                     ) ={
       Coll_allgather_rdb::allgather,
@@ -1645,22 +1645,22 @@ intel_tuning_table_element intel_allgather_table[] =
   }
 };
 
-/* I_MPI_ADJUST_ALLGATHERV 
+/* I_MPI_ADJUST_ALLGATHERV
 
-MPI_Allgatherv 
+MPI_Allgatherv
 
-1. Recursive doubling algorithm 
-2. Bruck's algorithm 
-3. Ring algorithm 
-4. Topology aware Gatherv + Bcast algorithm 
+1. Recursive doubling algorithm
+2. Bruck's algorithm
+3. Ring algorithm
+4. Topology aware Gatherv + Bcast algorithm
 
 */
 
-int (*intel_allgatherv_functions_table[])(void *sbuf, int scount, 
+int (*intel_allgatherv_functions_table[])(void *sbuf, int scount,
                                                MPI_Datatype sdtype,
-                                               void* rbuf, int *rcounts, 
+                                               void* rbuf, int *rcounts,
                                                int *rdispls,
-                                               MPI_Datatype rdtype, 
+                                               MPI_Datatype rdtype,
                                                MPI_Comm  comm
                                                     ) ={
       Coll_allgatherv_mpich_rdb::allgatherv,
@@ -1861,16 +1861,16 @@ intel_tuning_table_element intel_allgatherv_table[] =
 
 MPI_Gather
 
-1. Binomial algorithm 
-2. Topology aware binomial algorithm 
+1. Binomial algorithm
+2. Topology aware binomial algorithm
 3. Shumilin's algorithm
 
 */
 
-int (*intel_gather_functions_table[])(void *sbuf, int scount, 
+int (*intel_gather_functions_table[])(void *sbuf, int scount,
                                            MPI_Datatype sdtype,
-                                           void* rbuf, int rcount, 
-                                           MPI_Datatype rdtype, 
+                                           void* rbuf, int rcount,
+                                           MPI_Datatype rdtype,
                                            int root,
                                            MPI_Comm  comm
                                                     ) ={
@@ -1961,20 +1961,20 @@ intel_tuning_table_element intel_gather_table[] =
 };
 
 
-/* I_MPI_ADJUST_SCATTER 
+/* I_MPI_ADJUST_SCATTER
 
-MPI_Scatter 
+MPI_Scatter
 
-1. Binomial algorithm 
-2. Topology aware binomial algorithm 
-3. Shumilin's algorithm 
+1. Binomial algorithm
+2. Topology aware binomial algorithm
+3. Shumilin's algorithm
 
 */
 
-int (*intel_scatter_functions_table[])(void *sbuf, int scount, 
+int (*intel_scatter_functions_table[])(void *sbuf, int scount,
                                             MPI_Datatype sdtype,
-                                            void* rbuf, int rcount, 
-                                            MPI_Datatype rdtype, 
+                                            void* rbuf, int rcount,
+                                            MPI_Datatype rdtype,
                                             int root, MPI_Comm  comm
                                                     ) ={
       Coll_scatter_ompi_binomial::scatter,
@@ -2136,12 +2136,12 @@ intel_tuning_table_element intel_scatter_table[] =
 
 
 
-/* I_MPI_ADJUST_ALLTOALLV 
+/* I_MPI_ADJUST_ALLTOALLV
 
-MPI_Alltoallv 
+MPI_Alltoallv
 
-1. Isend/Irecv + waitall algorithm 
-2. Plum's algorithm 
+1. Isend/Irecv + waitall algorithm
+2. Plum's algorithm
 
 */
 
@@ -2179,7 +2179,7 @@ intel_tuning_table_element intel_alltoallv_table[] =
     }
     },
     { 2147483647,2,{
-      {4,1},//0 again 
+      {4,1},//0 again
       {2147483647,2}
     }
     }
@@ -2212,19 +2212,19 @@ intel_tuning_table_element intel_alltoallv_table[] =
 };
 
 
-//These are collected from table 3.5-2 of the Intel MPI Reference Manual 
+//These are collected from table 3.5-2 of the Intel MPI Reference Manual
+
 
-    
 #define SIZECOMP_reduce_scatter\
     int total_message_size = 0;\
     for (i = 0; i < comm_size; i++) { \
         total_message_size += rcounts[i];\
     }\
     size_t block_dsize = total_message_size*dtype->size();\
-    
+
 #define SIZECOMP_allreduce\
   size_t block_dsize =rcount * dtype->size();
-  
+
 #define SIZECOMP_alltoall\
   size_t block_dsize =send_count * send_type->size();
 
@@ -2246,7 +2246,7 @@ intel_tuning_table_element intel_alltoallv_table[] =
         total_message_size += recv_count[i];\
     }\
     size_t block_dsize = total_message_size*recv_type->size();
-    
+
 #define SIZECOMP_gather\
   int rank = comm->rank();\
   size_t block_dsize = (send_buff == MPI_IN_PLACE || rank ==root) ?\
@@ -2261,7 +2261,7 @@ intel_tuning_table_element intel_alltoallv_table[] =
 
 #define SIZECOMP_alltoallv\
   size_t block_dsize = 1;
-  
+
 #define IMPI_COLL_SELECT(cat, ret, args, args2)\
 ret Coll_ ## cat ## _impi:: cat (COLL_UNPAREN args)\
 {\
diff --git a/src/smpi/colls/smpi_mpich_selector.cpp b/src/smpi/colls/smpi_mpich_selector.cpp
index 7fd3910018..b5a0858c6e 100644
--- a/src/smpi/colls/smpi_mpich_selector.cpp
+++ b/src/smpi/colls/smpi_mpich_selector.cpp
@@ -9,7 +9,7 @@
 #include "colls_private.h"
 
 /* This is the default implementation of allreduce. The algorithm is:
-   
+
    Algorithm: MPI_Allreduce
 
    For the heterogeneous case, we call MPI_Reduce followed by MPI_Bcast
@@ -19,12 +19,12 @@
 
    For long messages and for builtin ops and if count >= pof2 (where
    pof2 is the nearest power-of-two less than or equal to the number
-   of processes), we use Rabenseifner's algorithm (see 
+   of processes), we use Rabenseifner's algorithm (see
    http://www.hlrs.de/mpi/myreduce.html).
    This algorithm implements the allreduce in two steps: first a
    reduce-scatter, followed by an allgather. A recursive-halving
    algorithm (beginning with processes that are distance 1 apart) is
-   used for the reduce-scatter, and a recursive doubling 
+   used for the reduce-scatter, and a recursive doubling
    algorithm is used for the allgather. The non-power-of-two case is
    handled by dropping to the nearest lower power-of-two: the first
    few even-numbered processes send their data to their right neighbors
@@ -32,27 +32,27 @@
    power-of-two processes. At the end, the first few even-numbered
    processes get the result from their right neighbors.
 
-   For the power-of-two case, the cost for the reduce-scatter is 
+   For the power-of-two case, the cost for the reduce-scatter is
    lgp.alpha + n.((p-1)/p).beta + n.((p-1)/p).gamma. The cost for the
    allgather lgp.alpha + n.((p-1)/p).beta. Therefore, the
    total cost is:
    Cost = 2.lgp.alpha + 2.n.((p-1)/p).beta + n.((p-1)/p).gamma
 
-   For the non-power-of-two case, 
+   For the non-power-of-two case,
    Cost = (2.floor(lgp)+2).alpha + (2.((p-1)/p) + 2).n.beta + n.(1+(p-1)/p).gamma
 
-   
-   For short messages, for user-defined ops, and for count < pof2 
+
+   For short messages, for user-defined ops, and for count < pof2
    we use a recursive doubling algorithm (similar to the one in
    MPI_Allgather). We use this algorithm in the case of user-defined ops
    because in this case derived datatypes are allowed, and the user
    could pass basic datatypes on one process and derived on another as
    long as the type maps are the same. Breaking up derived datatypes
-   to do the reduce-scatter is tricky. 
+   to do the reduce-scatter is tricky.
 
    Cost = lgp.alpha + n.lgp.beta + n.lgp.gamma
 
-   Possible improvements: 
+   Possible improvements:
 
    End Algorithm: MPI_Allreduce
 */
@@ -76,12 +76,12 @@ int Coll_allreduce_mpich::allreduce(void *sbuf, void *rbuf, int count,
 
     if (block_dsize > large_message && count >= pof2 && (op==MPI_OP_NULL || op->is_commutative())) {
       //for long messages
-       return (Coll_allreduce_rab_rdb::allreduce (sbuf, rbuf, 
+       return (Coll_allreduce_rab_rdb::allreduce (sbuf, rbuf,
                                                                    count, dtype,
                                                                    op, comm));
     }else {
       //for short ones and count < pof2
-      return (Coll_allreduce_rdb::allreduce (sbuf, rbuf, 
+      return (Coll_allreduce_rdb::allreduce (sbuf, rbuf,
                                                                    count, dtype,
                                                                    op, comm));
     }
@@ -89,7 +89,7 @@ int Coll_allreduce_mpich::allreduce(void *sbuf, void *rbuf, int count,
 
 
 /* This is the default implementation of alltoall. The algorithm is:
-   
+
    Algorithm: MPI_Alltoall
 
    We use four algorithms for alltoall. For short messages and
@@ -109,9 +109,9 @@ int Coll_allreduce_mpich::allreduce(void *sbuf, void *rbuf, int count,
    processes, so that all processes don't try to send/recv to/from the
    same process at the same time.
 
-   *** Modification: We post only a small number of isends and irecvs 
+   *** Modification: We post only a small number of isends and irecvs
    at a time and wait on them as suggested by Tony Ladd. ***
-   *** See comments below about an additional modification that 
+   *** See comments below about an additional modification that
    we may want to consider ***
 
    For long messages and power-of-two number of processes, we use a
@@ -122,22 +122,22 @@ int Coll_allreduce_mpich::allreduce(void *sbuf, void *rbuf, int count,
    This algorithm doesn't work if the number of processes is not a power of
    two. For a non-power-of-two number of processes, we use an
    algorithm in which, in step i, each process  receives from (rank-i)
-   and sends to (rank+i). 
+   and sends to (rank+i).
 
    Cost = (p-1).alpha + n.beta
 
    where n is the total amount of data a process needs to send to all
    other processes.
 
-   Possible improvements: 
+   Possible improvements:
 
    End Algorithm: MPI_Alltoall
 */
 
-int Coll_alltoall_mpich::alltoall( void *sbuf, int scount, 
+int Coll_alltoall_mpich::alltoall( void *sbuf, int scount,
                                              MPI_Datatype sdtype,
-                                             void* rbuf, int rcount, 
-                                             MPI_Datatype rdtype, 
+                                             void* rbuf, int rcount,
+                                             MPI_Datatype rdtype,
                                              MPI_Comm comm)
 {
     int communicator_size;
@@ -147,34 +147,34 @@ int Coll_alltoall_mpich::alltoall( void *sbuf, int scount,
     unsigned int short_size=256;
     unsigned int medium_size=32768;
     //short size and comm_size >=8   -> bruck
-    
+
 //     medium size messages and (short messages for comm_size < 8), we
 //     use an algorithm that posts all irecvs and isends and then does a
-//     waitall. 
-    
+//     waitall.
+
 //    For long messages and power-of-two number of processes, we use a
 //   pairwise exchange algorithm
 
 //   For a non-power-of-two number of processes, we use an
 //   algorithm in which, in step i, each process  receives from (rank-i)
-//   and sends to (rank+i). 
+//   and sends to (rank+i).
 
 
     dsize = sdtype->size();
     block_dsize = dsize * scount;
 
     if ((block_dsize < short_size) && (communicator_size >= 8)) {
-        return Coll_alltoall_bruck::alltoall(sbuf, scount, sdtype, 
+        return Coll_alltoall_bruck::alltoall(sbuf, scount, sdtype,
                                                     rbuf, rcount, rdtype,
                                                     comm);
 
     } else if (block_dsize < medium_size) {
-        return Coll_alltoall_basic_linear::alltoall(sbuf, scount, sdtype, 
-                                                           rbuf, rcount, rdtype, 
+        return Coll_alltoall_basic_linear::alltoall(sbuf, scount, sdtype,
+                                                           rbuf, rcount, rdtype,
                                                            comm);
     }else if (communicator_size%2){
-        return Coll_alltoall_ring::alltoall(sbuf, scount, sdtype, 
-                                                           rbuf, rcount, rdtype, 
+        return Coll_alltoall_ring::alltoall(sbuf, scount, sdtype,
+                                                           rbuf, rcount, rdtype,
                                                            comm);
     }
 
@@ -191,25 +191,25 @@ int Coll_alltoallv_mpich::alltoallv(void *sbuf, int *scounts, int *sdisps,
                                               )
 {
     /* For starters, just keep the original algorithm. */
-    return Coll_alltoallv_bruck::alltoallv(sbuf, scounts, sdisps, sdtype, 
+    return Coll_alltoallv_bruck::alltoallv(sbuf, scounts, sdisps, sdtype,
                                                         rbuf, rcounts, rdisps,rdtype,
                                                         comm);
 }
 
 
 int Coll_barrier_mpich::barrier(MPI_Comm  comm)
-{   
+{
     return Coll_barrier_ompi_bruck::barrier(comm);
 }
 
 /* This is the default implementation of broadcast. The algorithm is:
-   
+
    Algorithm: MPI_Bcast
 
-   For short messages, we use a binomial tree algorithm. 
+   For short messages, we use a binomial tree algorithm.
    Cost = lgp.alpha + n.lgp.beta
 
-   For long messages, we do a scatter followed by an allgather. 
+   For long messages, we do a scatter followed by an allgather.
    We first scatter the buffer using a binomial tree algorithm. This costs
    lgp.alpha + n.((p-1)/p).beta
    If the datatype is contiguous and the communicator is homogeneous,
@@ -218,7 +218,7 @@ int Coll_barrier_mpich::barrier(MPI_Comm  comm)
    cases, we first pack the data into a temporary buffer by using
    MPI_Pack, scatter it as bytes, and unpack it after the allgather.
 
-   For the allgather, we use a recursive doubling algorithm for 
+   For the allgather, we use a recursive doubling algorithm for
    medium-size messages and power-of-two number of processes. This
    takes lgp steps. In each step pairs of processes exchange all the
    data they have (we take care of non-power-of-two situations). This
@@ -232,12 +232,12 @@ int Coll_barrier_mpich::barrier(MPI_Comm  comm)
    versus n.lgp.beta. Therefore, for long messages and when lgp > 2,
    this algorithm will perform better.
 
-   For long messages and for medium-size messages and non-power-of-two 
-   processes, we use a ring algorithm for the allgather, which 
+   For long messages and for medium-size messages and non-power-of-two
+   processes, we use a ring algorithm for the allgather, which
    takes p-1 steps, because it performs better than recursive doubling.
    Total Cost = (lgp+p-1).alpha + 2.n.((p-1)/p).beta
 
-   Possible improvements: 
+   Possible improvements:
    For clusters of SMPs, we may want to do something differently to
    take advantage of shared memory on each node.
 
@@ -250,7 +250,7 @@ int Coll_bcast_mpich::bcast(void *buff, int count,
                                           MPI_Comm  comm
                                           )
 {
-    /* Decision function based on MX results for 
+    /* Decision function based on MX results for
        messages up to 36MB and communicator sizes up to 64 nodes */
     const size_t small_message_size = 12288;
     const size_t intermediate_message_size = 524288;
@@ -265,34 +265,34 @@ int Coll_bcast_mpich::bcast(void *buff, int count,
     dsize = datatype->size();
     message_size = dsize * (unsigned long)count;   /* needed for decision */
 
-    /* Handle messages of small and intermediate size, and 
+    /* Handle messages of small and intermediate size, and
        single-element broadcasts */
     if ((message_size < small_message_size) || (communicator_size <= 8)) {
         /* Binomial without segmentation */
-        return  Coll_bcast_binomial_tree::bcast (buff, count, datatype, 
+        return  Coll_bcast_binomial_tree::bcast (buff, count, datatype,
                                                       root, comm);
 
     } else if (message_size < intermediate_message_size && !(communicator_size%2)) {
         // SplittedBinary with 1KB segments
-        return Coll_bcast_scatter_rdb_allgather::bcast(buff, count, datatype, 
+        return Coll_bcast_scatter_rdb_allgather::bcast(buff, count, datatype,
                                                          root, comm);
 
     }
-     //Handle large message sizes 
-     return Coll_bcast_scatter_LR_allgather::bcast (buff, count, datatype, 
+     //Handle large message sizes
+     return Coll_bcast_scatter_LR_allgather::bcast (buff, count, datatype,
                                                      root, comm);
-                                                         
+
 }
 
 
 
 /* This is the default implementation of reduce. The algorithm is:
-   
+
    Algorithm: MPI_Reduce
 
    For long messages and for builtin ops and if count >= pof2 (where
    pof2 is the nearest power-of-two less than or equal to the number
-   of processes), we use Rabenseifner's algorithm (see 
+   of processes), we use Rabenseifner's algorithm (see
    http://www.hlrs.de/organization/par/services/models/mpi/myreduce.html ).
    This algorithm implements the reduce in two steps: first a
    reduce-scatter, followed by a gather to the root. A
@@ -307,7 +307,7 @@ int Coll_bcast_mpich::bcast(void *buff, int count,
    the root and exits; the root now acts as rank 0 in the binomial tree
    algorithm for gather.
 
-   For the power-of-two case, the cost for the reduce-scatter is 
+   For the power-of-two case, the cost for the reduce-scatter is
    lgp.alpha + n.((p-1)/p).beta + n.((p-1)/p).gamma. The cost for the
    gather to root is lgp.alpha + n.((p-1)/p).beta. Therefore, the
    total cost is:
@@ -315,12 +315,12 @@ int Coll_bcast_mpich::bcast(void *buff, int count,
 
    For the non-power-of-two case, assuming the root is not one of the
    odd-numbered processes that get excluded in the reduce-scatter,
-   Cost = (2.floor(lgp)+1).alpha + (2.((p-1)/p) + 1).n.beta + 
+   Cost = (2.floor(lgp)+1).alpha + (2.((p-1)/p) + 1).n.beta +
            n.(1+(p-1)/p).gamma
 
 
    For short messages, user-defined ops, and count < pof2, we use a
-   binomial tree algorithm for both short and long messages. 
+   binomial tree algorithm for both short and long messages.
 
    Cost = lgp.alpha + n.lgp.beta + n.lgp.gamma
 
@@ -335,7 +335,7 @@ int Coll_bcast_mpich::bcast(void *buff, int count,
    should be able to use the reduce-scatter/gather approach as long as
    count >= pof2.  [goodell@ 2009-01-21]
 
-   Possible improvements: 
+   Possible improvements:
 
    End Algorithm: MPI_Reduce
 */
@@ -361,7 +361,7 @@ int Coll_reduce_mpich::reduce( void *sendbuf, void *recvbuf,
     pof2 >>= 1;
 
     if ((count < pof2) || (message_size < 2048) || (op != MPI_OP_NULL && not op->is_commutative())) {
-      return Coll_reduce_binomial::reduce(sendbuf, recvbuf, count, datatype, op, root, comm); 
+      return Coll_reduce_binomial::reduce(sendbuf, recvbuf, count, datatype, op, root, comm);
     }
         return Coll_reduce_scatter_gather::reduce(sendbuf, recvbuf, count, datatype, op, root, comm/*, module,
                                                      segsize, max_requests*/);
@@ -430,17 +430,17 @@ int Coll_reduce_scatter_mpich::reduce_scatter( void *sbuf, void *rbuf,
     if(sbuf==rbuf)sbuf=MPI_IN_PLACE; //restore MPI_IN_PLACE as these algorithms handle it
 
     XBT_DEBUG("Coll_reduce_scatter_mpich::reduce");
-    
+
     comm_size = comm->size();
-    // We need data size for decision function 
+    // We need data size for decision function
     total_message_size = 0;
-    for (i = 0; i < comm_size; i++) { 
+    for (i = 0; i < comm_size; i++) {
         total_message_size += rcounts[i];
     }
 
-    if( (op==MPI_OP_NULL || op->is_commutative()) &&  total_message_size > 524288) { 
-        return Coll_reduce_scatter_mpich_pair::reduce_scatter (sbuf, rbuf, rcounts, 
-                                                                    dtype, op, 
+    if( (op==MPI_OP_NULL || op->is_commutative()) &&  total_message_size > 524288) {
+        return Coll_reduce_scatter_mpich_pair::reduce_scatter (sbuf, rbuf, rcounts,
+                                                                    dtype, op,
                                                                     comm);
     } else if ((op != MPI_OP_NULL && not op->is_commutative())) {
       int is_block_regular = 1;
@@ -462,14 +462,14 @@ int Coll_reduce_scatter_mpich::reduce_scatter( void *sbuf, void *rbuf,
       }
 
       return Coll_reduce_scatter_mpich_rdb::reduce_scatter(sbuf, rbuf, rcounts, dtype, op, comm);
-    }else{      
+    }else{
        return Coll_reduce_scatter_mpich_rdb::reduce_scatter(sbuf, rbuf, rcounts, dtype, op, comm);
     }
 }
 
 
 /* This is the default implementation of allgather. The algorithm is:
-   
+
    Algorithm: MPI_Allgather
 
    For short messages and non-power-of-two no. of processes, we use
@@ -508,15 +508,15 @@ int Coll_reduce_scatter_mpich::reduce_scatter( void *sbuf, void *rbuf,
    neighbor) performs twice as fast as recursive doubling for long
    messages (on Myrinet and IBM SP).
 
-   Possible improvements: 
+   Possible improvements:
 
    End Algorithm: MPI_Allgather
 */
 
-int Coll_allgather_mpich::allgather(void *sbuf, int scount, 
+int Coll_allgather_mpich::allgather(void *sbuf, int scount,
                                               MPI_Datatype sdtype,
-                                              void* rbuf, int rcount, 
-                                              MPI_Datatype rdtype, 
+                                              void* rbuf, int rcount,
+                                              MPI_Datatype rdtype,
                                               MPI_Comm  comm
                                               )
 {
@@ -527,36 +527,36 @@ int Coll_allgather_mpich::allgather(void *sbuf, int scount,
 
     /* Determine complete data size */
     dsize=sdtype->size();
-    total_dsize = dsize * scount * communicator_size;   
-   
-    for (pow2_size  = 1; pow2_size < communicator_size; pow2_size <<=1); 
+    total_dsize = dsize * scount * communicator_size;
 
-    /* Decision as in MPICH-2 
-       presented in Thakur et.al. "Optimization of Collective Communication 
-       Operations in MPICH", International Journal of High Performance Computing 
+    for (pow2_size  = 1; pow2_size < communicator_size; pow2_size <<=1);
+
+    /* Decision as in MPICH-2
+       presented in Thakur et.al. "Optimization of Collective Communication
+       Operations in MPICH", International Journal of High Performance Computing
        Applications, Vol. 19, No. 1, 49-66 (2005)
-       - for power-of-two processes and small and medium size messages 
+       - for power-of-two processes and small and medium size messages
        (up to 512KB) use recursive doubling
        - for non-power-of-two processes and small messages (80KB) use bruck,
        - for everything else use ring.
     */
     if ((pow2_size == communicator_size) && (total_dsize < 524288)) {
-        return Coll_allgather_rdb::allgather(sbuf, scount, sdtype, 
-                                                                 rbuf, rcount, rdtype, 
+        return Coll_allgather_rdb::allgather(sbuf, scount, sdtype,
+                                                                 rbuf, rcount, rdtype,
                                                                  comm);
-    } else if (total_dsize <= 81920) { 
-        return Coll_allgather_bruck::allgather(sbuf, scount, sdtype, 
+    } else if (total_dsize <= 81920) {
+        return Coll_allgather_bruck::allgather(sbuf, scount, sdtype,
                                                      rbuf, rcount, rdtype,
                                                      comm);
-    } 
-    return Coll_allgather_ring::allgather(sbuf, scount, sdtype, 
+    }
+    return Coll_allgather_ring::allgather(sbuf, scount, sdtype,
                                                 rbuf, rcount, rdtype,
                                                 comm);
 }
 
 
 /* This is the default implementation of allgatherv. The algorithm is:
-   
+
    Algorithm: MPI_Allgatherv
 
    For short messages and non-power-of-two no. of processes, we use
@@ -587,15 +587,15 @@ int Coll_allgather_mpich::allgather(void *sbuf, int scount,
 
    Cost = (p-1).alpha + n.((p-1)/p).beta
 
-   Possible improvements: 
+   Possible improvements:
 
    End Algorithm: MPI_Allgatherv
 */
-int Coll_allgatherv_mpich::allgatherv(void *sbuf, int scount, 
+int Coll_allgatherv_mpich::allgatherv(void *sbuf, int scount,
                                                MPI_Datatype sdtype,
-                                               void* rbuf, int *rcounts, 
+                                               void* rbuf, int *rcounts,
                                                int *rdispls,
-                                               MPI_Datatype rdtype, 
+                                               MPI_Datatype rdtype,
                                                MPI_Comm  comm
                                                )
 {
@@ -610,25 +610,25 @@ int Coll_allgatherv_mpich::allgatherv(void *sbuf, int scount,
         total_dsize += rcounts[i];
     if (total_dsize == 0)
       return MPI_SUCCESS;
-    
-    for (pow2_size  = 1; pow2_size < communicator_size; pow2_size <<=1); 
+
+    for (pow2_size  = 1; pow2_size < communicator_size; pow2_size <<=1);
 
     if ((pow2_size == communicator_size) && (total_dsize < 524288)) {
-        return Coll_allgatherv_mpich_rdb::allgatherv(sbuf, scount, sdtype, 
-                                                                 rbuf, rcounts, rdispls, rdtype, 
+        return Coll_allgatherv_mpich_rdb::allgatherv(sbuf, scount, sdtype,
+                                                                 rbuf, rcounts, rdispls, rdtype,
                                                                  comm);
-    } else if (total_dsize <= 81920) { 
-        return Coll_allgatherv_ompi_bruck::allgatherv(sbuf, scount, sdtype, 
+    } else if (total_dsize <= 81920) {
+        return Coll_allgatherv_ompi_bruck::allgatherv(sbuf, scount, sdtype,
                                                      rbuf, rcounts, rdispls, rdtype,
                                                      comm);
-    } 
+    }
     return Coll_allgatherv_mpich_ring::allgatherv(sbuf, scount, sdtype,
                                                 rbuf, rcounts, rdispls, rdtype,
                                                 comm);
 }
 
 /* This is the default implementation of gather. The algorithm is:
-   
+
    Algorithm: MPI_Gather
 
    We use a binomial tree algorithm for both short and long
@@ -644,50 +644,50 @@ int Coll_allgatherv_mpich::allgatherv(void *sbuf, int scount,
    Cost = lgp.alpha + n.((p-1)/p).beta
    where n is the total size of the data gathered at the root.
 
-   Possible improvements: 
+   Possible improvements:
 
    End Algorithm: MPI_Gather
 */
 
-int Coll_gather_mpich::gather(void *sbuf, int scount, 
+int Coll_gather_mpich::gather(void *sbuf, int scount,
                                            MPI_Datatype sdtype,
-                                           void* rbuf, int rcount, 
-                                           MPI_Datatype rdtype, 
+                                           void* rbuf, int rcount,
+                                           MPI_Datatype rdtype,
                                            int root,
                                            MPI_Comm  comm
                                            )
 {
-        return Coll_gather_ompi_binomial::gather (sbuf, scount, sdtype, 
-                                                      rbuf, rcount, rdtype, 
+        return Coll_gather_ompi_binomial::gather (sbuf, scount, sdtype,
+                                                      rbuf, rcount, rdtype,
                                                       root, comm);
 }
 
 /* This is the default implementation of scatter. The algorithm is:
-   
+
    Algorithm: MPI_Scatter
 
    We use a binomial tree algorithm for both short and
    long messages. At nodes other than leaf nodes we need to allocate
    a temporary buffer to store the incoming message. If the root is
-   not rank 0, we reorder the sendbuf in order of relative ranks by 
+   not rank 0, we reorder the sendbuf in order of relative ranks by
    copying it into a temporary buffer, so that all the sends from the
    root are contiguous and in the right order. In the heterogeneous
    case, we first pack the buffer by using MPI_Pack and then do the
-   scatter. 
+   scatter.
 
    Cost = lgp.alpha + n.((p-1)/p).beta
    where n is the total size of the data to be scattered from the root.
 
-   Possible improvements: 
+   Possible improvements:
 
    End Algorithm: MPI_Scatter
 */
 
 
-int Coll_scatter_mpich::scatter(void *sbuf, int scount, 
+int Coll_scatter_mpich::scatter(void *sbuf, int scount,
                                             MPI_Datatype sdtype,
-                                            void* rbuf, int rcount, 
-                                            MPI_Datatype rdtype, 
+                                            void* rbuf, int rcount,
+                                            MPI_Datatype rdtype,
                                             int root, MPI_Comm  comm
                                             )
 {
@@ -697,7 +697,7 @@ int Coll_scatter_mpich::scatter(void *sbuf, int scount,
       sdtype=rdtype;
   }
   int ret= Coll_scatter_ompi_binomial::scatter (sbuf, scount, sdtype,
-                                                       rbuf, rcount, rdtype, 
+                                                       rbuf, rcount, rdtype,
                                                        root, comm);
   if(comm->rank()!=root){
       xbt_free(sbuf);
diff --git a/src/smpi/colls/smpi_mvapich2_selector.cpp b/src/smpi/colls/smpi_mvapich2_selector.cpp
index 99290fb14a..837584f694 100644
--- a/src/smpi/colls/smpi_mvapich2_selector.cpp
+++ b/src/smpi/colls/smpi_mvapich2_selector.cpp
@@ -14,7 +14,7 @@ namespace simgrid{
 namespace smpi{
 
 
-int Coll_alltoall_mvapich2::alltoall( void *sendbuf, int sendcount, 
+int Coll_alltoall_mvapich2::alltoall( void *sendbuf, int sendcount,
     MPI_Datatype sendtype,
     void* recvbuf, int recvcount,
     MPI_Datatype recvtype,
@@ -106,7 +106,7 @@ int Coll_allgather_mvapich2::allgather(void *sendbuf, int sendcount, MPI_Datatyp
 
   if(mv2_allgather_table_ppn_conf==NULL)
     init_mv2_allgather_tables_stampede();
-    
+
   if(comm->get_leaders_comm()==MPI_COMM_NULL){
     comm->init_smp();
   }
@@ -134,7 +134,7 @@ int Coll_allgather_mvapich2::allgather(void *sendbuf, int sendcount, MPI_Datatyp
   if (partial_sub_ok != 1) {
     conf_index = 0;
   }
-  
+
   /* Search for the corresponding system size inside the tuning table */
   while ((range < (mv2_size_allgather_tuning_table[conf_index] - 1)) &&
       (comm_size >
@@ -239,17 +239,17 @@ int Coll_gather_mvapich2::gather(void *sendbuf,
       -1)) {
       range_intra_threshold++;
   }
-  
+
     if (comm->is_blocked() ) {
-        // Set intra-node function pt for gather_two_level 
-        MV2_Gather_intra_node_function = 
+        // Set intra-node function pt for gather_two_level
+        MV2_Gather_intra_node_function =
                               mv2_gather_thresholds_table[range].intra_node[range_intra_threshold].
                               MV2_pt_Gather_function;
-        //Set inter-leader pt 
+        //Set inter-leader pt
         MV2_Gather_inter_leader_function =
                               mv2_gather_thresholds_table[range].inter_leader[range_threshold].
                               MV2_pt_Gather_function;
-        // We call Gather function 
+        // We call Gather function
         mpi_errno =
             MV2_Gather_inter_leader_function(sendbuf, sendcnt, sendtype, recvbuf, recvcnt,
                                              recvtype, root, comm);
@@ -465,7 +465,7 @@ int Coll_alltoallv_mvapich2::alltoallv(void *sbuf, int *scounts, int *sdisps,
 
 
 int Coll_barrier_mvapich2::barrier(MPI_Comm  comm)
-{   
+{
   return Coll_barrier_mvapich2_pair::barrier(comm);
 }
 
@@ -480,7 +480,7 @@ int Coll_bcast_mvapich2::bcast(void *buffer,
     int mpi_errno = MPI_SUCCESS;
     int comm_size/*, rank*/;
     int two_level_bcast = 1;
-    long nbytes = 0; 
+    long nbytes = 0;
     int range = 0;
     int range_threshold = 0;
     int range_threshold_intra = 0;
@@ -564,7 +564,7 @@ int Coll_bcast_mvapich2::bcast(void *buffer,
 
     if (mv2_bcast_thresholds_table[range].inter_leader[range_threshold].
         zcpy_pipelined_knomial_factor != -1) {
-        zcpy_knomial_factor = 
+        zcpy_knomial_factor =
             mv2_bcast_thresholds_table[range].inter_leader[range_threshold].
             zcpy_pipelined_knomial_factor;
     }
@@ -581,7 +581,7 @@ int Coll_bcast_mvapich2::bcast(void *buffer,
 
     /* Set value of pipeline segment size */
     bcast_segment_size = mv2_bcast_thresholds_table[range].bcast_segment_size;
-    
+
     /* Set value of inter node knomial factor */
     mv2_inter_node_knomial_factor = mv2_bcast_thresholds_table[range].inter_node_knomial_factor;
 
@@ -591,7 +591,7 @@ int Coll_bcast_mvapich2::bcast(void *buffer,
     /* Check if we will use a two level algorithm or not */
     two_level_bcast =
 #if defined(_MCST_SUPPORT_)
-        mv2_bcast_thresholds_table[range].is_two_level_bcast[range_threshold] 
+        mv2_bcast_thresholds_table[range].is_two_level_bcast[range_threshold]
         || comm->ch.is_mcast_ok;
 #else
         mv2_bcast_thresholds_table[range].is_two_level_bcast[range_threshold];
@@ -613,13 +613,13 @@ int Coll_bcast_mvapich2::bcast(void *buffer,
               (&MPIR_Pipelined_Bcast_Zcpy_MV2 == MV2_Bcast_function)) {
           if (not is_contig || not is_homogeneous) {
             mpi_errno = MPIR_Pipelined_Bcast_Zcpy_MV2(tmp_buf, nbytes, MPI_BYTE, root, comm);
-            } else { 
+            } else {
                 mpi_errno = MPIR_Pipelined_Bcast_Zcpy_MV2(buffer, count, datatype,
                                                  root, comm);
-            } 
-        } else 
+            }
+        } else
 #endif /* defined(CHANNEL_MRAIL_GEN2) */
-        { 
+        {
             shmem_comm = comm->get_intra_comm();
             if (not is_contig || not is_homogeneous) {
               mpi_errno = MPIR_Bcast_tune_inter_node_helper_MV2(tmp_buf, nbytes, MPI_BYTE, root, comm);
@@ -752,7 +752,7 @@ int Coll_reduce_mvapich2::reduce( void *sendbuf,
          if(comm->get_leaders_comm()==MPI_COMM_NULL){
            comm->init_smp();
          }
-         mpi_errno = MPIR_Reduce_two_level_helper_MV2(sendbuf, recvbuf, count, 
+         mpi_errno = MPIR_Reduce_two_level_helper_MV2(sendbuf, recvbuf, count,
                                            datatype, op, root, comm);
         } else {
       mpi_errno = MPIR_Reduce_binomial_MV2(sendbuf, recvbuf, count,
@@ -761,7 +761,7 @@ int Coll_reduce_mvapich2::reduce( void *sendbuf,
     } else if(MV2_Reduce_function == &MPIR_Reduce_inter_knomial_wrapper_MV2 ){
         if(is_commutative ==1)
           {
-            mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count, 
+            mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count,
                 datatype, op, root, comm);
           } else {
               mpi_errno = MPIR_Reduce_binomial_MV2(sendbuf, recvbuf, count,
@@ -770,14 +770,14 @@ int Coll_reduce_mvapich2::reduce( void *sendbuf,
     } else if(MV2_Reduce_function == &MPIR_Reduce_redscat_gather_MV2){
         if (/*(HANDLE_GET_KIND(op) == HANDLE_KIND_BUILTIN) &&*/ (count >= pof2))
           {
-            mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count, 
+            mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count,
                 datatype, op, root, comm);
           } else {
               mpi_errno = MPIR_Reduce_binomial_MV2(sendbuf, recvbuf, count,
                   datatype, op, root, comm);
           }
     } else {
-        mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count, 
+        mpi_errno = MV2_Reduce_function(sendbuf, recvbuf, count,
             datatype, op, root, comm);
     }
 
@@ -887,7 +887,7 @@ int Coll_scatter_mvapich2::scatter(void *sendbuf,
   if(comm->get_leaders_comm()==MPI_COMM_NULL){
     comm->init_smp();
   }
-  
+
   comm_size = comm->size();
 
   rank = comm->rank();
@@ -899,15 +899,15 @@ int Coll_scatter_mvapich2::scatter(void *sendbuf,
       recvtype_size=recvtype->size();
       nbytes = recvcnt * recvtype_size;
   }
-  
-    // check if safe to use partial subscription mode 
+
+    // check if safe to use partial subscription mode
     if (comm->is_uniform()) {
 
         shmem_comm = comm->get_intra_comm();
         local_size = shmem_comm->size();
         i = 0;
         if (mv2_scatter_table_ppn_conf[0] == -1) {
-            // Indicating user defined tuning 
+            // Indicating user defined tuning
             conf_index = 0;
         }else{
             do {
@@ -920,7 +920,7 @@ int Coll_scatter_mvapich2::scatter(void *sendbuf,
             } while(i < mv2_scatter_num_ppn_conf);
         }
     }
-   
+
   if (partial_sub_ok != 1) {
       conf_index = 0;
   }
@@ -968,7 +968,7 @@ int Coll_scatter_mvapich2::scatter(void *sendbuf,
               /* Fallback! */
               MV2_Scatter_function = &MPIR_Scatter_MV2_Binomial;
           }
-        } 
+        }
   }
 
   if( (MV2_Scatter_function == &MPIR_Scatter_MV2_two_level_Direct) ||
diff --git a/src/smpi/colls/smpi_mvapich2_selector_stampede.h b/src/smpi/colls/smpi_mvapich2_selector_stampede.h
index e8a89b5185..ca81613bf4 100644
--- a/src/smpi/colls/smpi_mvapich2_selector_stampede.h
+++ b/src/smpi/colls/smpi_mvapich2_selector_stampede.h
@@ -42,7 +42,7 @@ mv2_alltoall_tuning_table **mv2_alltoall_thresholds_table = NULL;
 #define MPIR_Alltoall_RD_MV2 simgrid::smpi::Coll_alltoall_rdb::alltoall
 #define MPIR_Alltoall_Scatter_dest_MV2 simgrid::smpi::Coll_alltoall_mvapich2_scatter_dest::alltoall
 #define MPIR_Alltoall_pairwise_MV2 simgrid::smpi::Coll_alltoall_pair::alltoall
-#define MPIR_Alltoall_inplace_MV2 simgrid::smpi::Coll_alltoall_ring::alltoall 
+#define MPIR_Alltoall_inplace_MV2 simgrid::smpi::Coll_alltoall_ring::alltoall
 
 
 static void init_mv2_alltoall_tables_stampede(){
@@ -578,7 +578,7 @@ typedef struct {
 } mv2_gather_tuning_table;
 
 int mv2_size_gather_tuning_table=7;
-mv2_gather_tuning_table * mv2_gather_thresholds_table=NULL; 
+mv2_gather_tuning_table * mv2_gather_thresholds_table=NULL;
 
 typedef int (*MV2_Gather_function_ptr) (void *sendbuf,
     int sendcnt,
@@ -796,7 +796,7 @@ static int MPIR_Allreduce_mcst_reduce_two_level_helper_MV2( void *sendbuf,
     int count,
     MPI_Datatype datatype,
     MPI_Op op, MPI_Comm comm)
-{ 
+{
   return 0;
 }
 
@@ -1982,6 +1982,6 @@ static void init_mv2_scatter_tables_stampede(){
                 * mv2_size_scatter_tuning_table[i]));
     }
     xbt_free(table_ptrs);
-  
+
 }
 
diff --git a/src/smpi/colls/smpi_openmpi_selector.cpp b/src/smpi/colls/smpi_openmpi_selector.cpp
index 509f00701c..9770eb0b88 100644
--- a/src/smpi/colls/smpi_openmpi_selector.cpp
+++ b/src/smpi/colls/smpi_openmpi_selector.cpp
@@ -20,8 +20,8 @@ int Coll_allreduce_ompi::allreduce(void *sbuf, void *rbuf, int count,
 
     /**
      * Decision function based on MX results from the Grig cluster at UTK.
-     * 
-     * Currently, linear, recursive doubling, and nonoverlapping algorithms 
+     *
+     * Currently, linear, recursive doubling, and nonoverlapping algorithms
      * can handle both commutative and non-commutative operations.
      * Ring algorithm does not support non-commutative operations.
      */
@@ -29,10 +29,10 @@ int Coll_allreduce_ompi::allreduce(void *sbuf, void *rbuf, int count,
     block_dsize = dsize * count;
 
     if (block_dsize < intermediate_message) {
-        return (Coll_allreduce_rdb::allreduce (sbuf, rbuf, 
+        return (Coll_allreduce_rdb::allreduce (sbuf, rbuf,
                                                                    count, dtype,
                                                                    op, comm));
-    } 
+    }
 
     if( ((op==MPI_OP_NULL) || op->is_commutative()) && (count > comm_size) ) {
         const size_t segment_size = 1 << 20; /* 1 MB */
@@ -43,29 +43,29 @@ int Coll_allreduce_ompi::allreduce(void *sbuf, void *rbuf, int count,
                                               op, comm);
         } else {
            return (Coll_allreduce_ompi_ring_segmented::allreduce (sbuf, rbuf,
-                                                                    count, dtype, 
-                                                                    op, comm 
+                                                                    count, dtype,
+                                                                    op, comm
                                                                     /*segment_size*/));
         }
     }
 
-    return (Coll_allreduce_redbcast::allreduce(sbuf, rbuf, count, 
+    return (Coll_allreduce_redbcast::allreduce(sbuf, rbuf, count,
                                                             dtype, op, comm));
 }
 
 
 
-int Coll_alltoall_ompi::alltoall( void *sbuf, int scount, 
+int Coll_alltoall_ompi::alltoall( void *sbuf, int scount,
                                              MPI_Datatype sdtype,
-                                             void* rbuf, int rcount, 
-                                             MPI_Datatype rdtype, 
+                                             void* rbuf, int rcount,
+                                             MPI_Datatype rdtype,
                                              MPI_Comm comm)
 {
     int communicator_size;
     size_t dsize, block_dsize;
     communicator_size = comm->size();
 
-    /* Decision function based on measurement on Grig cluster at 
+    /* Decision function based on measurement on Grig cluster at
        the University of Tennessee (2GB MX) up to 64 nodes.
        Has better performance for messages of intermediate sizes than the old one */
     /* determine block size */
@@ -73,17 +73,17 @@ int Coll_alltoall_ompi::alltoall( void *sbuf, int scount,
     block_dsize = dsize * scount;
 
     if ((block_dsize < 200) && (communicator_size > 12)) {
-        return Coll_alltoall_bruck::alltoall(sbuf, scount, sdtype, 
+        return Coll_alltoall_bruck::alltoall(sbuf, scount, sdtype,
                                                     rbuf, rcount, rdtype,
                                                     comm);
 
     } else if (block_dsize < 3000) {
-        return Coll_alltoall_basic_linear::alltoall(sbuf, scount, sdtype, 
-                                                           rbuf, rcount, rdtype, 
+        return Coll_alltoall_basic_linear::alltoall(sbuf, scount, sdtype,
+                                                           rbuf, rcount, rdtype,
                                                            comm);
     }
 
-    return Coll_alltoall_ring::alltoall (sbuf, scount, sdtype, 
+    return Coll_alltoall_ring::alltoall (sbuf, scount, sdtype,
                                                     rbuf, rcount, rdtype,
                                                     comm);
 }
@@ -96,7 +96,7 @@ int Coll_alltoallv_ompi::alltoallv(void *sbuf, int *scounts, int *sdisps,
                                               )
 {
     /* For starters, just keep the original algorithm. */
-    return Coll_alltoallv_ompi_basic_linear::alltoallv(sbuf, scounts, sdisps, sdtype, 
+    return Coll_alltoallv_ompi_basic_linear::alltoallv(sbuf, scounts, sdisps, sdtype,
                                                         rbuf, rcounts, rdisps,rdtype,
                                                         comm);
 }
@@ -128,14 +128,14 @@ int Coll_bcast_ompi::bcast(void *buff, int count,
                                           MPI_Comm  comm
                                           )
 {
-    /* Decision function based on MX results for 
+    /* Decision function based on MX results for
        messages up to 36MB and communicator sizes up to 64 nodes */
     const size_t small_message_size = 2048;
     const size_t intermediate_message_size = 370728;
     const double a_p16  = 3.2118e-6; /* [1 / byte] */
-    const double b_p16  = 8.7936;   
+    const double b_p16  = 8.7936;
     const double a_p64  = 2.3679e-6; /* [1 / byte] */
-    const double b_p64  = 1.1787;     
+    const double b_p64  = 1.1787;
     const double a_p128 = 1.6134e-6; /* [1 / byte] */
     const double b_p128 = 2.1102;
 
@@ -149,50 +149,50 @@ int Coll_bcast_ompi::bcast(void *buff, int count,
     dsize = datatype->size();
     message_size = dsize * (unsigned long)count;   /* needed for decision */
 
-    /* Handle messages of small and intermediate size, and 
+    /* Handle messages of small and intermediate size, and
        single-element broadcasts */
     if ((message_size < small_message_size) || (count <= 1)) {
         /* Binomial without segmentation */
-        return  Coll_bcast_binomial_tree::bcast (buff, count, datatype, 
+        return  Coll_bcast_binomial_tree::bcast (buff, count, datatype,
                                                       root, comm);
 
     } else if (message_size < intermediate_message_size) {
         // SplittedBinary with 1KB segments
-        return Coll_bcast_ompi_split_bintree::bcast(buff, count, datatype, 
+        return Coll_bcast_ompi_split_bintree::bcast(buff, count, datatype,
                                                          root, comm);
 
     }
-     //Handle large message sizes 
+     //Handle large message sizes
     else if (communicator_size < (a_p128 * message_size + b_p128)) {
-        //Pipeline with 128KB segments 
+        //Pipeline with 128KB segments
         //segsize = 1024  << 7;
-        return Coll_bcast_ompi_pipeline::bcast (buff, count, datatype, 
+        return Coll_bcast_ompi_pipeline::bcast (buff, count, datatype,
                                                      root, comm);
-                                                     
+
 
     } else if (communicator_size < 13) {
-        // Split Binary with 8KB segments 
-        return Coll_bcast_ompi_split_bintree::bcast(buff, count, datatype, 
+        // Split Binary with 8KB segments
+        return Coll_bcast_ompi_split_bintree::bcast(buff, count, datatype,
                                                          root, comm);
-       
+
     } else if (communicator_size < (a_p64 * message_size + b_p64)) {
-        // Pipeline with 64KB segments 
+        // Pipeline with 64KB segments
         //segsize = 1024 << 6;
-        return Coll_bcast_ompi_pipeline::bcast (buff, count, datatype, 
+        return Coll_bcast_ompi_pipeline::bcast (buff, count, datatype,
                                                      root, comm);
-                                                     
+
 
     } else if (communicator_size < (a_p16 * message_size + b_p16)) {
-        //Pipeline with 16KB segments 
+        //Pipeline with 16KB segments
         //segsize = 1024 << 4;
-        return Coll_bcast_ompi_pipeline::bcast (buff, count, datatype, 
+        return Coll_bcast_ompi_pipeline::bcast (buff, count, datatype,
                                                      root, comm);
-                                                     
+
 
     }
     /* Pipeline with 8KB segments */
     //segsize = 1024 << 3;
-    return Coll_bcast_flattree_pipeline::bcast (buff, count, datatype, 
+    return Coll_bcast_flattree_pipeline::bcast (buff, count, datatype,
                                                  root, comm
                                                  /*segsize*/);
 #if 0
@@ -252,7 +252,7 @@ int Coll_reduce_ompi::reduce( void *sendbuf, void *recvbuf,
     message_size = dsize * count;   /* needed for decision */
 
     /**
-     * If the operation is non commutative we currently have choice of linear 
+     * If the operation is non commutative we currently have choice of linear
      * or in-order binary tree algorithm.
      */
     if ((op != MPI_OP_NULL) && not op->is_commutative()) {
@@ -260,12 +260,12 @@ int Coll_reduce_ompi::reduce( void *sendbuf, void *recvbuf,
         return Coll_reduce_ompi_basic_linear::reduce(sendbuf, recvbuf, count, datatype, op, root, comm /*, module*/);
       }
       return Coll_reduce_ompi_in_order_binary::reduce(sendbuf, recvbuf, count, datatype, op, root, comm /*, module,
-                                                             0, max_requests*/); 
+                                                             0, max_requests*/);
     }
 
     if ((communicator_size < 8) && (message_size < 512)){
         /* Linear_0K */
-        return Coll_reduce_ompi_basic_linear::reduce (sendbuf, recvbuf, count, datatype, op, root, comm); 
+        return Coll_reduce_ompi_basic_linear::reduce (sendbuf, recvbuf, count, datatype, op, root, comm);
     } else if (((communicator_size < 8) && (message_size < 20480)) ||
                (message_size < 2048) || (count <= 1)) {
         /* Binomial_0K */
@@ -273,29 +273,29 @@ int Coll_reduce_ompi::reduce( void *sendbuf, void *recvbuf,
         return Coll_reduce_ompi_binomial::reduce(sendbuf, recvbuf, count, datatype, op, root, comm/*, module,
                                                      segsize, max_requests*/);
     } else if (communicator_size > (a1 * message_size + b1)) {
-        // Binomial_1K 
+        // Binomial_1K
         //segsize = 1024;
         return Coll_reduce_ompi_binomial::reduce(sendbuf, recvbuf, count, datatype, op, root, comm/*, module,
                                                      segsize, max_requests*/);
     } else if (communicator_size > (a2 * message_size + b2)) {
-        // Pipeline_1K 
+        // Pipeline_1K
         //segsize = 1024;
-        return Coll_reduce_ompi_pipeline::reduce (sendbuf, recvbuf, count, datatype, op, root, comm/*, module, 
+        return Coll_reduce_ompi_pipeline::reduce (sendbuf, recvbuf, count, datatype, op, root, comm/*, module,
                                                       segsize, max_requests*/);
     } else if (communicator_size > (a3 * message_size + b3)) {
-        // Binary_32K 
+        // Binary_32K
         //segsize = 32*1024;
         return Coll_reduce_ompi_binary::reduce( sendbuf, recvbuf, count, datatype, op, root,
                                                     comm/*, module, segsize, max_requests*/);
     }
 //    if (communicator_size > (a4 * message_size + b4)) {
-        // Pipeline_32K 
+        // Pipeline_32K
 //        segsize = 32*1024;
 //    } else {
-        // Pipeline_64K 
+        // Pipeline_64K
 //        segsize = 64*1024;
 //    }
-    return Coll_reduce_ompi_pipeline::reduce (sendbuf, recvbuf, count, datatype, op, root, comm/*, module, 
+    return Coll_reduce_ompi_pipeline::reduce (sendbuf, recvbuf, count, datatype, op, root, comm/*, module,
                                                   segsize, max_requests*/);
 
 #if 0
@@ -305,7 +305,7 @@ int Coll_reduce_ompi::reduce( void *sendbuf, void *recvbuf,
         fanout = communicator_size - 1;
         /* when linear implemented or taken from basic put here, right now using chain as a linear system */
         /* it is implemented and I shouldn't be calling a chain with a fanout bigger than MAXTREEFANOUT from topo.h! */
-        return Coll_reduce_intra_basic_linear::reduce (sendbuf, recvbuf, count, datatype, op, root, comm, module); 
+        return Coll_reduce_intra_basic_linear::reduce (sendbuf, recvbuf, count, datatype, op, root, comm, module);
         /*        return Coll_reduce_intra_chain::reduce (sendbuf, recvbuf, count, datatype, op, root, comm, segsize, fanout); */
     }
     if (message_size < 524288) {
@@ -343,12 +343,12 @@ int Coll_reduce_scatter_ompi::reduce_scatter( void *sbuf, void *rbuf,
     int zerocounts = 0;
 
     XBT_DEBUG("Coll_reduce_scatter_ompi::reduce_scatter");
-    
+
     comm_size = comm->size();
-    // We need data size for decision function 
+    // We need data size for decision function
     dsize=dtype->size();
     total_message_size = 0;
-    for (i = 0; i < comm_size; i++) { 
+    for (i = 0; i < comm_size; i++) {
         total_message_size += rcounts[i];
         if (0 == rcounts[i]) {
             zerocounts = 1;
@@ -359,20 +359,20 @@ int Coll_reduce_scatter_ompi::reduce_scatter( void *sbuf, void *rbuf,
       Coll_reduce_scatter_default::reduce_scatter(sbuf, rbuf, rcounts, dtype, op, comm);
       return MPI_SUCCESS;
     }
-   
+
     total_message_size *= dsize;
 
-    // compute the nearest power of 2 
+    // compute the nearest power of 2
     for (pow2 = 1; pow2 < comm_size; pow2 <<= 1);
 
     if ((total_message_size <= small_message_size) ||
         ((total_message_size <= large_message_size) && (pow2 == comm_size)) ||
         (comm_size >= a * total_message_size + b)) {
-        return 
+        return
             Coll_reduce_scatter_ompi_basic_recursivehalving::reduce_scatter(sbuf, rbuf, rcounts,
                                                                         dtype, op,
                                                                         comm);
-    } 
+    }
     return Coll_reduce_scatter_ompi_ring::reduce_scatter(sbuf, rbuf, rcounts,
                                                      dtype, op,
                                                      comm);
@@ -381,10 +381,10 @@ int Coll_reduce_scatter_ompi::reduce_scatter( void *sbuf, void *rbuf,
 
 }
 
-int Coll_allgather_ompi::allgather(void *sbuf, int scount, 
+int Coll_allgather_ompi::allgather(void *sbuf, int scount,
                                               MPI_Datatype sdtype,
-                                              void* rbuf, int rcount, 
-                                              MPI_Datatype rdtype, 
+                                              void* rbuf, int rcount,
+                                              MPI_Datatype rdtype,
                                               MPI_Comm  comm
                                               )
 {
@@ -395,39 +395,39 @@ int Coll_allgather_ompi::allgather(void *sbuf, int scount,
 
     /* Special case for 2 processes */
     if (communicator_size == 2) {
-        return Coll_allgather_pair::allgather (sbuf, scount, sdtype, 
-                                                          rbuf, rcount, rdtype, 
+        return Coll_allgather_pair::allgather (sbuf, scount, sdtype,
+                                                          rbuf, rcount, rdtype,
                                                           comm/*, module*/);
     }
 
     /* Determine complete data size */
     dsize=sdtype->size();
-    total_dsize = dsize * scount * communicator_size;   
-   
-    for (pow2_size  = 1; pow2_size < communicator_size; pow2_size <<=1); 
-
-    /* Decision based on MX 2Gb results from Grig cluster at 
-       The University of Tennesse, Knoxville 
-       - if total message size is less than 50KB use either bruck or 
-       recursive doubling for non-power of two and power of two nodes, 
+    total_dsize = dsize * scount * communicator_size;
+
+    for (pow2_size  = 1; pow2_size < communicator_size; pow2_size <<=1);
+
+    /* Decision based on MX 2Gb results from Grig cluster at
+       The University of Tennesse, Knoxville
+       - if total message size is less than 50KB use either bruck or
+       recursive doubling for non-power of two and power of two nodes,
        respectively.
-       - else use ring and neighbor exchange algorithms for odd and even 
+       - else use ring and neighbor exchange algorithms for odd and even
        number of nodes, respectively.
     */
     if (total_dsize < 50000) {
         if (pow2_size == communicator_size) {
-            return Coll_allgather_rdb::allgather(sbuf, scount, sdtype, 
+            return Coll_allgather_rdb::allgather(sbuf, scount, sdtype,
                                                                      rbuf, rcount, rdtype,
                                                                      comm);
         } else {
-            return Coll_allgather_bruck::allgather(sbuf, scount, sdtype, 
-                                                         rbuf, rcount, rdtype, 
+            return Coll_allgather_bruck::allgather(sbuf, scount, sdtype,
+                                                         rbuf, rcount, rdtype,
                                                          comm);
         }
     } else {
         if (communicator_size % 2) {
-            return Coll_allgather_ring::allgather(sbuf, scount, sdtype, 
-                                                        rbuf, rcount, rdtype, 
+            return Coll_allgather_ring::allgather(sbuf, scount, sdtype,
+                                                        rbuf, rcount, rdtype,
                                                         comm);
         } else {
             return  Coll_allgather_ompi_neighborexchange::allgather(sbuf, scount, sdtype,
@@ -435,86 +435,86 @@ int Coll_allgather_ompi::allgather(void *sbuf, int scount,
                                                                      comm);
         }
     }
-   
+
 #if defined(USE_MPICH2_DECISION)
-    /* Decision as in MPICH-2 
-       presented in Thakur et.al. "Optimization of Collective Communication 
-       Operations in MPICH", International Journal of High Performance Computing 
+    /* Decision as in MPICH-2
+       presented in Thakur et.al. "Optimization of Collective Communication
+       Operations in MPICH", International Journal of High Performance Computing
        Applications, Vol. 19, No. 1, 49-66 (2005)
-       - for power-of-two processes and small and medium size messages 
+       - for power-of-two processes and small and medium size messages
        (up to 512KB) use recursive doubling
        - for non-power-of-two processes and small messages (80KB) use bruck,
        - for everything else use ring.
     */
     if ((pow2_size == communicator_size) && (total_dsize < 524288)) {
-        return Coll_allgather_rdb::allgather(sbuf, scount, sdtype, 
-                                                                 rbuf, rcount, rdtype, 
+        return Coll_allgather_rdb::allgather(sbuf, scount, sdtype,
+                                                                 rbuf, rcount, rdtype,
                                                                  comm);
-    } else if (total_dsize <= 81920) { 
-        return Coll_allgather_bruck::allgather(sbuf, scount, sdtype, 
+    } else if (total_dsize <= 81920) {
+        return Coll_allgather_bruck::allgather(sbuf, scount, sdtype,
                                                      rbuf, rcount, rdtype,
                                                      comm);
-    } 
-    return Coll_allgather_ring::allgather(sbuf, scount, sdtype, 
+    }
+    return Coll_allgather_ring::allgather(sbuf, scount, sdtype,
                                                 rbuf, rcount, rdtype,
                                                 comm);
 #endif  /* defined(USE_MPICH2_DECISION) */
 }
 
-int Coll_allgatherv_ompi::allgatherv(void *sbuf, int scount, 
+int Coll_allgatherv_ompi::allgatherv(void *sbuf, int scount,
                                                MPI_Datatype sdtype,
-                                               void* rbuf, int *rcounts, 
+                                               void* rbuf, int *rcounts,
                                                int *rdispls,
-                                               MPI_Datatype rdtype, 
+                                               MPI_Datatype rdtype,
                                                MPI_Comm  comm
                                                )
 {
     int i;
     int communicator_size;
     size_t dsize, total_dsize;
-    
+
     communicator_size = comm->size();
-    
+
     /* Special case for 2 processes */
     if (communicator_size == 2) {
         return Coll_allgatherv_pair::allgatherv(sbuf, scount, sdtype,
-                                                           rbuf, rcounts, rdispls, rdtype, 
+                                                           rbuf, rcounts, rdispls, rdtype,
                                                            comm);
     }
-    
+
     /* Determine complete data size */
     dsize=sdtype->size();
     total_dsize = 0;
     for (i = 0; i < communicator_size; i++) {
         total_dsize += dsize * rcounts[i];
     }
-    
+
     /* Decision based on allgather decision.   */
     if (total_dsize < 50000) {
-/*        return Coll_allgatherv_intra_bruck::allgatherv(sbuf, scount, sdtype, 
-                                                      rbuf, rcounts, rdispls, rdtype, 
+/*        return Coll_allgatherv_intra_bruck::allgatherv(sbuf, scount, sdtype,
+                                                      rbuf, rcounts, rdispls, rdtype,
                                                       comm, module);*/
-    return Coll_allgatherv_ring::allgatherv(sbuf, scount, sdtype, 
-                                                      rbuf, rcounts, rdispls, rdtype, 
+    return Coll_allgatherv_ring::allgatherv(sbuf, scount, sdtype,
+                                                      rbuf, rcounts, rdispls, rdtype,
                                                       comm);
 
     } else {
         if (communicator_size % 2) {
-            return Coll_allgatherv_ring::allgatherv(sbuf, scount, sdtype, 
-                                                         rbuf, rcounts, rdispls, rdtype, 
+            return Coll_allgatherv_ring::allgatherv(sbuf, scount, sdtype,
+                                                         rbuf, rcounts, rdispls, rdtype,
                                                          comm);
         } else {
             return  Coll_allgatherv_ompi_neighborexchange::allgatherv(sbuf, scount, sdtype,
-                                                                      rbuf, rcounts, rdispls, rdtype, 
+                                                                      rbuf, rcounts, rdispls, rdtype,
                                                                       comm);
         }
     }
 }
 
-int Coll_gather_ompi::gather(void *sbuf, int scount, 
+int Coll_gather_ompi::gather(void *sbuf, int scount,
                                            MPI_Datatype sdtype,
-                                           void* rbuf, int rcount, 
-                                           MPI_Datatype rdtype, 
+                                           void* rbuf, int rcount,
+                                           MPI_Datatype rdtype,
                                            int root,
                                            MPI_Comm  comm
                                            )
@@ -537,7 +537,7 @@ int Coll_gather_ompi::gather(void *sbuf, int scount,
     communicator_size = comm->size();
     rank = comm->rank();
 
-    // Determine block size 
+    // Determine block size
     if (rank == root) {
         dsize = rdtype->size();
         block_size = dsize * rcount;
@@ -552,29 +552,29 @@ int Coll_gather_ompi::gather(void *sbuf, int scount,
 /*                                                         root, comm);*/
 
 /*    } else*/ if (block_size > intermediate_block_size) {
-        return Coll_gather_ompi_linear_sync::gather (sbuf, scount, sdtype, 
-                                                         rbuf, rcount, rdtype, 
+        return Coll_gather_ompi_linear_sync::gather (sbuf, scount, sdtype,
+                                                         rbuf, rcount, rdtype,
                                                          root, comm);
 
     } else if ((communicator_size > large_communicator_size) ||
                ((communicator_size > small_communicator_size) &&
                 (block_size < small_block_size))) {
-        return Coll_gather_ompi_binomial::gather (sbuf, scount, sdtype, 
-                                                      rbuf, rcount, rdtype, 
+        return Coll_gather_ompi_binomial::gather (sbuf, scount, sdtype,
+                                                      rbuf, rcount, rdtype,
                                                       root, comm);
 
     }
-    // Otherwise, use basic linear 
-    return Coll_gather_ompi_basic_linear::gather (sbuf, scount, sdtype, 
-                                                      rbuf, rcount, rdtype, 
+    // Otherwise, use basic linear
+    return Coll_gather_ompi_basic_linear::gather (sbuf, scount, sdtype,
+                                                      rbuf, rcount, rdtype,
                                                       root, comm);
 }
 
 
-int Coll_scatter_ompi::scatter(void *sbuf, int scount, 
+int Coll_scatter_ompi::scatter(void *sbuf, int scount,
                                             MPI_Datatype sdtype,
-                                            void* rbuf, int rcount, 
-                                            MPI_Datatype rdtype, 
+                                            void* rbuf, int rcount,
+                                            MPI_Datatype rdtype,
                                             int root, MPI_Comm  comm
                                             )
 {
@@ -587,14 +587,14 @@ int Coll_scatter_ompi::scatter(void *sbuf, int scount,
 
     communicator_size = comm->size();
     rank = comm->rank();
-    // Determine block size 
+    // Determine block size
     if (root == rank) {
         dsize=sdtype->size();
         block_size = dsize * scount;
     } else {
         dsize=rdtype->size();
         block_size = dsize * rcount;
-    } 
+    }
 
     if ((communicator_size > small_comm_size) &&
         (block_size < small_block_size)) {
@@ -611,8 +611,8 @@ int Coll_scatter_ompi::scatter(void *sbuf, int scount,
         }
         return ret;
     }
-    return Coll_scatter_ompi_basic_linear::scatter (sbuf, scount, sdtype, 
-                                                       rbuf, rcount, rdtype, 
+    return Coll_scatter_ompi_basic_linear::scatter (sbuf, scount, sdtype,
+                                                       rbuf, rcount, rdtype,
                                                        root, comm);
 }
 
diff --git a/src/smpi/instr_smpi.cpp b/src/smpi/instr_smpi.cpp
index 3a7525eb2c..d0f4c4dd8f 100644
--- a/src/smpi/instr_smpi.cpp
+++ b/src/smpi/instr_smpi.cpp
@@ -77,8 +77,8 @@ static const char *instr_find_color (const char *state)
   while (current != nullptr) {
     if (strcmp (state, current) == 0 //exact match
         || strstr(target, current) != 0 ){//as substring
-         ret = smpi_colors[i+1]; 
-         break; 
+         ret = smpi_colors[i+1];
+         break;
     }
     i+=2;
     current = smpi_colors[i];
diff --git a/src/smpi/private.h b/src/smpi/private.h
index fd8553ab2d..649f42ef9a 100644
--- a/src/smpi/private.h
+++ b/src/smpi/private.h
@@ -61,7 +61,7 @@ XBT_PRIVATE void smpi_deployment_register_process(const char* instance_id, int r
 XBT_PRIVATE MPI_Comm* smpi_deployment_comm_world(const char* instance_id);
 XBT_PRIVATE msg_bar_t smpi_deployment_finalization_barrier(const char* instance_id);
 XBT_PRIVATE void smpi_deployment_cleanup_instances();
- 
+
 XBT_PRIVATE void smpi_comm_copy_buffer_callback(smx_activity_t comm, void *buff, size_t buff_size);
 
 XBT_PRIVATE void smpi_comm_null_copy_buffer_callback(smx_activity_t comm, void *buff, size_t buff_size);
diff --git a/src/smpi/private.hpp b/src/smpi/private.hpp
index 42792da047..87d51f8ec6 100644
--- a/src/smpi/private.hpp
+++ b/src/smpi/private.hpp
@@ -20,7 +20,7 @@
 #define TOPAGE(addr) (void *)(((unsigned long)(addr) / xbt_pagesize) * xbt_pagesize)
 
 #if HAVE_PAPI
-typedef 
+typedef
     std::vector<std::pair</* counter name */std::string, /* counter value */long long>> papi_counter_t;
 XBT_PRIVATE papi_counter_t& smpi_process_papi_counters();
 XBT_PRIVATE int smpi_process_papi_event_set();
diff --git a/src/smpi/smpi_bench.cpp b/src/smpi/smpi_bench.cpp
index 52736c8e06..f902f4e3d6 100644
--- a/src/smpi/smpi_bench.cpp
+++ b/src/smpi/smpi_bench.cpp
@@ -428,7 +428,7 @@ extern "C" { /** These functions will be called from the user code **/
     smpi_trace_set_call_location(file, *line);
   }
 
-  /** 
+  /**
    * Required for Fortran if -fsecond-underscore is activated
    */
   void smpi_trace_set_call_location__(const char* file, int* line) {
diff --git a/src/smpi/smpi_coll.cpp b/src/smpi/smpi_coll.cpp
index bb7d6069fd..a26b1f7ed0 100644
--- a/src/smpi/smpi_coll.cpp
+++ b/src/smpi/smpi_coll.cpp
@@ -320,7 +320,7 @@ int Colls::exscan(void *sendbuf, void *recvbuf, int count, MPI_Datatype datatype
           Datatype::copy(tmpbufs[other], count, datatype, recvbuf, count, datatype);
           recvbuf_is_empty = 0;
         } else
-          if(op!=MPI_OP_NULL) 
+          if(op!=MPI_OP_NULL)
             op->apply( tmpbufs[other], recvbuf, &count, datatype);
       }
     }
diff --git a/src/smpi/smpi_comm.cpp b/src/smpi/smpi_comm.cpp
index 39e51329dc..d1e2ab014f 100644
--- a/src/smpi/smpi_comm.cpp
+++ b/src/smpi/smpi_comm.cpp
@@ -171,7 +171,7 @@ MPI_Comm Comm::get_leaders_comm(){
 }
 
 MPI_Comm Comm::get_intra_comm(){
-  if (this == MPI_COMM_UNINITIALIZED || this==MPI_COMM_WORLD) 
+  if (this == MPI_COMM_UNINITIALIZED || this==MPI_COMM_WORLD)
     return smpi_process()->comm_intra();
   else return intra_comm_;
 }
@@ -251,7 +251,7 @@ MPI_Comm Comm::split(int color, int key)
         }
         if(i != 0 && group_out != MPI_COMM_WORLD->group() && group_out != MPI_GROUP_EMPTY)
           Group::unref(group_out);
-        
+
         Request::waitall(reqs, requests, MPI_STATUS_IGNORE);
         xbt_free(requests);
       }
@@ -318,8 +318,8 @@ void Comm::init_smp(){
     smpi_process()->comm_world()->init_smp();
 
   int comm_size = this->size();
-  
-  // If we are in replay - perform an ugly hack  
+
+  // If we are in replay - perform an ugly hack
   // tell SimGrid we are not in replay for a while, because we need the buffers to be copied for the following calls
   bool replaying = false; //cache data to set it back again after
   if(smpi_process()->replaying()){
@@ -469,9 +469,9 @@ void Comm::init_smp(){
     is_blocked_=global_blocked;
   }
   xbt_free(leader_list);
-  
+
   if(replaying)
-    smpi_process()->set_replaying(true); 
+    smpi_process()->set_replaying(true);
 }
 
 MPI_Comm Comm::f2c(int id) {
diff --git a/src/smpi/smpi_comm.hpp b/src/smpi/smpi_comm.hpp
index 992c9229e0..3e997763cf 100644
--- a/src/smpi/smpi_comm.hpp
+++ b/src/smpi/smpi_comm.hpp
@@ -18,7 +18,7 @@ class Comm : public F2C, public Keyval{
 
   private:
     MPI_Group group_;
-    MPIR_Topo_type topoType_; 
+    MPIR_Topo_type topoType_;
     MPI_Topology topo_; // to be replaced by an union
     int refcount_;
     MPI_Comm leaders_comm_;//inter-node communicator
diff --git a/src/smpi/smpi_datatype.cpp b/src/smpi/smpi_datatype.cpp
index 4faee97a5a..d93fc2da8b 100644
--- a/src/smpi/smpi_datatype.cpp
+++ b/src/smpi/smpi_datatype.cpp
@@ -127,7 +127,7 @@ Datatype::Datatype(Datatype *datatype, int* ret) : name_(nullptr), lb_(datatype-
     void* value_out;
     for(auto it = datatype->attributes()->begin(); it != datatype->attributes()->end(); it++){
       smpi_key_elem elem = keyvals_.at((*it).first);
-      
+
       if (elem != nullptr && elem->copy_fn.type_copy_fn != MPI_NULL_COPY_FN) {
         *ret = elem->copy_fn.type_copy_fn(datatype, (*it).first, nullptr, (*it).second, &value_out, &flag);
         if (*ret != MPI_SUCCESS) {
@@ -328,7 +328,7 @@ int Datatype::create_contiguous(int count, MPI_Datatype old_type, MPI_Aint lb, M
 int Datatype::create_vector(int count, int block_length, int stride, MPI_Datatype old_type, MPI_Datatype* new_type)
 {
   int retval;
-  if (block_length<0) 
+  if (block_length<0)
     return MPI_ERR_ARG;
   MPI_Aint lb = 0;
   MPI_Aint ub = 0;
@@ -353,7 +353,7 @@ int Datatype::create_vector(int count, int block_length, int stride, MPI_Datatyp
 int Datatype::create_hvector(int count, int block_length, MPI_Aint stride, MPI_Datatype old_type, MPI_Datatype* new_type)
 {
   int retval;
-  if (block_length<0) 
+  if (block_length<0)
     return MPI_ERR_ARG;
   MPI_Aint lb = 0;
   MPI_Aint ub = 0;
@@ -422,9 +422,9 @@ int Datatype::create_hindexed(int count, int* block_lengths, MPI_Aint* indices,
       return MPI_ERR_ARG;
     size += block_lengths[i];
 
-    if(indices[i]+old_type->lb()<lb) 
+    if(indices[i]+old_type->lb()<lb)
       lb = indices[i]+old_type->lb();
-    if(indices[i]+block_lengths[i]*old_type->ub()>ub) 
+    if(indices[i]+block_lengths[i]*old_type->ub()>ub)
       ub = indices[i]+block_lengths[i]*old_type->ub();
 
     if ( (i< count -1) && (indices[i]+block_lengths[i]*(static_cast<int>(old_type->size())) != indices[i+1]) )
@@ -479,7 +479,7 @@ int Datatype::create_struct(int count, int* block_lengths, MPI_Aint* indices, MP
       contiguous=false;
   }
   if (not contiguous) {
-    *new_type = new Type_Struct(size, lb,ub, DT_FLAG_DERIVED|DT_FLAG_DATA, 
+    *new_type = new Type_Struct(size, lb,ub, DT_FLAG_DERIVED|DT_FLAG_DATA,
                                 count, block_lengths, indices, old_types);
   }else{
     Datatype::create_contiguous(size, MPI_CHAR, lb, new_type);
diff --git a/src/smpi/smpi_datatype.hpp b/src/smpi/smpi_datatype.hpp
index 220a424ca5..51978b5c71 100644
--- a/src/smpi/smpi_datatype.hpp
+++ b/src/smpi/smpi_datatype.hpp
@@ -111,9 +111,9 @@ class Datatype : public F2C, public Keyval{
     void set_name(char* name);
     static int copy(void *sendbuf, int sendcount, MPI_Datatype sendtype,
                     void *recvbuf, int recvcount, MPI_Datatype recvtype);
-    virtual void serialize( void* noncontiguous, void *contiguous, 
+    virtual void serialize( void* noncontiguous, void *contiguous,
                             int count);
-    virtual void unserialize( void* contiguous, void *noncontiguous, 
+    virtual void unserialize( void* contiguous, void *noncontiguous,
                               int count, MPI_Op op);
     static int keyval_create(MPI_Type_copy_attr_function* copy_fn, MPI_Type_delete_attr_function* delete_fn, int* keyval, void* extra_state);
     static int keyval_free(int* keyval);
diff --git a/src/smpi/smpi_datatype_derived.cpp b/src/smpi/smpi_datatype_derived.cpp
index afc16cae5f..4551048a74 100644
--- a/src/smpi/smpi_datatype_derived.cpp
+++ b/src/smpi/smpi_datatype_derived.cpp
@@ -22,13 +22,13 @@ Type_Contiguous::~Type_Contiguous(){
 }
 
 
-void Type_Contiguous::serialize( void* noncontiguous_buf, void *contiguous_buf, 
+void Type_Contiguous::serialize( void* noncontiguous_buf, void *contiguous_buf,
                             int count){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf)+lb();
   memcpy(contiguous_buf_char, noncontiguous_buf_char, count * block_count_ * old_type_->size());
 }
-void Type_Contiguous::unserialize( void* contiguous_buf, void *noncontiguous_buf, 
+void Type_Contiguous::unserialize( void* contiguous_buf, void *noncontiguous_buf,
                               int count, MPI_Op op){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf)+lb();
@@ -47,7 +47,7 @@ Type_Vector::~Type_Vector(){
 }
 
 
-void Type_Vector::serialize( void* noncontiguous_buf, void *contiguous_buf, 
+void Type_Vector::serialize( void* noncontiguous_buf, void *contiguous_buf,
                             int count){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf);
@@ -66,7 +66,7 @@ void Type_Vector::serialize( void* noncontiguous_buf, void *contiguous_buf,
   }
 }
 
-void Type_Vector::unserialize( void* contiguous_buf, void *noncontiguous_buf, 
+void Type_Vector::unserialize( void* contiguous_buf, void *noncontiguous_buf,
                               int count, MPI_Op op){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf);
@@ -94,7 +94,7 @@ Type_Hvector::~Type_Hvector(){
   Datatype::unref(old_type_);
 }
 
-void Type_Hvector::serialize( void* noncontiguous_buf, void *contiguous_buf, 
+void Type_Hvector::serialize( void* noncontiguous_buf, void *contiguous_buf,
                     int count){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf);
@@ -114,14 +114,14 @@ void Type_Hvector::serialize( void* noncontiguous_buf, void *contiguous_buf,
 }
 
 
-void Type_Hvector::unserialize( void* contiguous_buf, void *noncontiguous_buf, 
+void Type_Hvector::unserialize( void* contiguous_buf, void *noncontiguous_buf,
                               int count, MPI_Op op){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf);
 
   for (int i = 0; i < block_count_ * count; i++) {
     if (not(old_type_->flags() & DT_FLAG_DERIVED)) {
-      if(op!=MPI_OP_NULL) 
+      if(op!=MPI_OP_NULL)
         op->apply( contiguous_buf_char, noncontiguous_buf_char, &block_length_, old_type_);
     }else
       old_type_->unserialize( contiguous_buf_char, noncontiguous_buf_char, block_length_, op);
@@ -152,7 +152,7 @@ Type_Indexed::~Type_Indexed(){
 }
 
 
-void Type_Indexed::serialize( void* noncontiguous_buf, void *contiguous_buf, 
+void Type_Indexed::serialize( void* noncontiguous_buf, void *contiguous_buf,
                         int count){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf)+block_indices_[0] * old_type_->size();
@@ -175,7 +175,7 @@ void Type_Indexed::serialize( void* noncontiguous_buf, void *contiguous_buf,
 }
 
 
-void Type_Indexed::unserialize( void* contiguous_buf, void *noncontiguous_buf, 
+void Type_Indexed::unserialize( void* contiguous_buf, void *noncontiguous_buf,
                       int count, MPI_Op op){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char =
@@ -183,7 +183,7 @@ void Type_Indexed::unserialize( void* contiguous_buf, void *noncontiguous_buf,
   for (int j = 0; j < count; j++) {
     for (int i = 0; i < block_count_; i++) {
       if (not(old_type_->flags() & DT_FLAG_DERIVED)) {
-        if(op!=MPI_OP_NULL) 
+        if(op!=MPI_OP_NULL)
           op->apply( contiguous_buf_char, noncontiguous_buf_char, &block_lengths_[i],
                     old_type_);
       }else
@@ -220,7 +220,7 @@ Type_Hindexed::Type_Hindexed(int size,MPI_Aint lb, MPI_Aint ub, int flags, int c
   }
 }
 
-void Type_Hindexed::serialize( void* noncontiguous_buf, void *contiguous_buf, 
+void Type_Hindexed::serialize( void* noncontiguous_buf, void *contiguous_buf,
                 int count){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf)+ block_indices_[0];
@@ -241,14 +241,14 @@ void Type_Hindexed::serialize( void* noncontiguous_buf, void *contiguous_buf,
   }
 }
 
-void Type_Hindexed::unserialize( void* contiguous_buf, void *noncontiguous_buf, 
+void Type_Hindexed::unserialize( void* contiguous_buf, void *noncontiguous_buf,
                           int count, MPI_Op op){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf)+ block_indices_[0];
   for (int j = 0; j < count; j++) {
     for (int i = 0; i < block_count_; i++) {
       if (not(old_type_->flags() & DT_FLAG_DERIVED)) {
-        if(op!=MPI_OP_NULL) 
+        if(op!=MPI_OP_NULL)
           op->apply( contiguous_buf_char, noncontiguous_buf_char, &block_lengths_[i],
                             old_type_);
       }else
@@ -288,7 +288,7 @@ Type_Struct::~Type_Struct(){
 }
 
 
-void Type_Struct::serialize( void* noncontiguous_buf, void *contiguous_buf, 
+void Type_Struct::serialize( void* noncontiguous_buf, void *contiguous_buf,
                         int count){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf)+ block_indices_[0];
@@ -310,14 +310,14 @@ void Type_Struct::serialize( void* noncontiguous_buf, void *contiguous_buf,
   }
 }
 
-void Type_Struct::unserialize( void* contiguous_buf, void *noncontiguous_buf, 
+void Type_Struct::unserialize( void* contiguous_buf, void *noncontiguous_buf,
                               int count, MPI_Op op){
   char* contiguous_buf_char = static_cast<char*>(contiguous_buf);
   char* noncontiguous_buf_char = static_cast<char*>(noncontiguous_buf)+ block_indices_[0];
   for (int j = 0; j < count; j++) {
     for (int i = 0; i < block_count_; i++) {
       if (not(old_types_[i]->flags() & DT_FLAG_DERIVED)) {
-        if(op!=MPI_OP_NULL) 
+        if(op!=MPI_OP_NULL)
           op->apply( contiguous_buf_char, noncontiguous_buf_char, &block_lengths_[i], old_types_[i]);
       }else
         old_types_[i]->unserialize( contiguous_buf_char, noncontiguous_buf_char,block_lengths_[i], op);
diff --git a/src/smpi/smpi_datatype_derived.hpp b/src/smpi/smpi_datatype_derived.hpp
index 03ac75689c..4ed9f738f5 100644
--- a/src/smpi/smpi_datatype_derived.hpp
+++ b/src/smpi/smpi_datatype_derived.hpp
@@ -19,9 +19,9 @@ class Type_Contiguous: public Datatype {
   public:
     Type_Contiguous(int size, MPI_Aint lb, MPI_Aint ub, int flags, int block_count, MPI_Datatype old_type);
     ~Type_Contiguous();
-    void serialize( void* noncontiguous, void *contiguous, 
+    void serialize( void* noncontiguous, void *contiguous,
                             int count);
-    void unserialize( void* contiguous_vector, void *noncontiguous_vector, 
+    void unserialize( void* contiguous_vector, void *noncontiguous_vector,
                               int count, MPI_Op op);
 };
 
@@ -34,9 +34,9 @@ class Type_Vector: public Datatype{
   public:
     Type_Vector(int size,MPI_Aint lb, MPI_Aint ub, int flags, int count, int blocklen, int stride, MPI_Datatype old_type);
     ~Type_Vector();
-    void serialize( void* noncontiguous, void *contiguous, 
+    void serialize( void* noncontiguous, void *contiguous,
                             int count);
-    void unserialize( void* contiguous_vector, void *noncontiguous_vector, 
+    void unserialize( void* contiguous_vector, void *noncontiguous_vector,
                               int count, MPI_Op op);
 };
 
@@ -49,9 +49,9 @@ class Type_Hvector: public Datatype{
   public:
     Type_Hvector(int size,MPI_Aint lb, MPI_Aint ub, int flags, int block_count, int block_length, MPI_Aint block_stride, MPI_Datatype old_type);
     ~Type_Hvector();
-    void serialize( void* noncontiguous, void *contiguous, 
+    void serialize( void* noncontiguous, void *contiguous,
                             int count);
-    void unserialize( void* contiguous_vector, void *noncontiguous_vector, 
+    void unserialize( void* contiguous_vector, void *noncontiguous_vector,
                               int count, MPI_Op op);
 };
 
@@ -64,9 +64,9 @@ class Type_Indexed: public Datatype{
   public:
     Type_Indexed(int size,MPI_Aint lb, MPI_Aint ub, int flags, int block_count, int* block_lengths, int* block_indices, MPI_Datatype old_type);
     ~Type_Indexed();
-    void serialize( void* noncontiguous, void *contiguous, 
+    void serialize( void* noncontiguous, void *contiguous,
                             int count);
-    void unserialize( void* contiguous_vector, void *noncontiguous_vector, 
+    void unserialize( void* contiguous_vector, void *noncontiguous_vector,
                               int count, MPI_Op op);
 };
 
@@ -79,9 +79,9 @@ class Type_Hindexed: public Datatype{
   public:
     Type_Hindexed(int size,MPI_Aint lb, MPI_Aint ub, int flags, int block_count, int* block_lengths, MPI_Aint* block_indices, MPI_Datatype old_type);
     ~Type_Hindexed();
-    void serialize( void* noncontiguous, void *contiguous, 
+    void serialize( void* noncontiguous, void *contiguous,
                             int count);
-    void unserialize( void* contiguous_vector, void *noncontiguous_vector, 
+    void unserialize( void* contiguous_vector, void *noncontiguous_vector,
                               int count, MPI_Op op);
 };
 
@@ -94,9 +94,9 @@ class Type_Struct: public Datatype{
   public:
     Type_Struct(int size,MPI_Aint lb, MPI_Aint ub, int flags, int block_count, int* block_lengths, MPI_Aint* block_indices, MPI_Datatype* old_types);
     ~Type_Struct();
-    void serialize( void* noncontiguous, void *contiguous, 
+    void serialize( void* noncontiguous, void *contiguous,
                             int count);
-    void unserialize( void* contiguous_vector, void *noncontiguous_vector, 
+    void unserialize( void* contiguous_vector, void *noncontiguous_vector,
                               int count, MPI_Op op);
 };
 
diff --git a/src/smpi/smpi_f2c.hpp b/src/smpi/smpi_f2c.hpp
index bfe726b640..5b77cab24b 100644
--- a/src/smpi/smpi_f2c.hpp
+++ b/src/smpi/smpi_f2c.hpp
@@ -18,7 +18,7 @@ namespace smpi{
 
 class F2C {
   private:
-    // We use a single lookup table for every type. 
+    // We use a single lookup table for every type.
     // Beware of collisions if id in mpif.h is not unique
     static xbt_dict_t f2c_lookup_;
     static int f2c_id_;
diff --git a/src/smpi/smpi_f77.cpp b/src/smpi/smpi_f77.cpp
index 73d502c048..ffaad7c8f1 100644
--- a/src/smpi/smpi_f77.cpp
+++ b/src/smpi/smpi_f77.cpp
@@ -276,7 +276,7 @@ void mpi_startall_(int* count, int* requests, int* ierr) {
 
 void mpi_wait_(int* request, MPI_Status* status, int* ierr) {
    MPI_Request req = simgrid::smpi::Request::f2c(*request);
-   
+
    *ierr = MPI_Wait(&req, FORT_STATUS_IGNORE(status));
    if(req==MPI_REQUEST_NULL){
      simgrid::smpi::Request::free_f(*request);
@@ -564,7 +564,7 @@ void mpi_win_set_name_ (int*  win, char * name, int* ierr, int size){
 
 void mpi_win_get_name_ (int*  win, char * name, int* len, int* ierr){
   *ierr = MPI_Win_get_name(simgrid::smpi::Win::f2c(*win),name,len);
-  if(*len>0) 
+  if(*len>0)
     name[*len]=' ';//blank padding, not \0
 }
 
@@ -586,7 +586,7 @@ void mpi_info_set_( int *info, char *key, char *value, int* ierr, unsigned int k
   }
   char* tkey = xbt_new(char,keylen+1);
   strncpy(tkey, key, keylen);
-  tkey[keylen]='\0';  
+  tkey[keylen]='\0';
 
   while(value[valuelen-1]==' ')
     valuelen--;
@@ -596,7 +596,7 @@ void mpi_info_set_( int *info, char *key, char *value, int* ierr, unsigned int k
   }
   char* tvalue = xbt_new(char,valuelen+1);
   strncpy(tvalue, value, valuelen);
-  tvalue[valuelen]='\0'; 
+  tvalue[valuelen]='\0';
 
   *ierr =  MPI_Info_set( simgrid::smpi::Info::f2c(*info), tkey, tvalue);
   xbt_free(tkey);
@@ -609,7 +609,7 @@ void mpi_info_get_ (int* info,char *key,int* valuelen, char *value, int *flag, i
   while(*key==' '){//handle leading blanks
     keylen--;
     key++;
-  }  
+  }
   char* tkey = xbt_new(char,keylen+1);
   strncpy(tkey, key, keylen);
   tkey[keylen]='\0';
@@ -624,7 +624,7 @@ void mpi_info_get_ (int* info,char *key,int* valuelen, char *value, int *flag, i
       if(replace)
         value[i]=' ';
     }
-  } 
+  }
 }
 
 void mpi_info_free_(int* info, int* ierr){
@@ -703,7 +703,7 @@ void mpi_type_set_name_ (int*  datatype, char * name, int* ierr, int size){
 
 void mpi_type_get_name_ (int*  datatype, char * name, int* len, int* ierr){
  *ierr = MPI_Type_get_name(simgrid::smpi::Datatype::f2c(*datatype),name,len);
-  if(*len>0) 
+  if(*len>0)
     name[*len]=' ';
 }
 
@@ -869,7 +869,7 @@ void mpi_comm_free_keyval_ (int* keyval, int* ierr) {
 
 void mpi_comm_get_name_ (int* comm, char* name, int* len, int* ierr){
  *ierr = MPI_Comm_get_name(simgrid::smpi::Comm::f2c(*comm), name, len);
-  if(*len>0) 
+  if(*len>0)
     name[*len]=' ';
 }
 
diff --git a/src/smpi/smpi_global.cpp b/src/smpi/smpi_global.cpp
index 687e97257e..66a8f26d7a 100644
--- a/src/smpi/smpi_global.cpp
+++ b/src/smpi/smpi_global.cpp
@@ -144,7 +144,8 @@ static void check_blocks(std::vector<std::pair<size_t, size_t>> &private_blocks,
 
 void smpi_comm_copy_buffer_callback(smx_activity_t synchro, void *buff, size_t buff_size)
 {
-  simgrid::kernel::activity::CommImpl* comm = dynamic_cast<simgrid::kernel::activity::CommImpl*>(synchro);
+  simgrid::kernel::activity::CommImplPtr comm =
+      boost::dynamic_pointer_cast<simgrid::kernel::activity::CommImpl>(synchro);
   int src_shared                        = 0;
   int dst_shared                        = 0;
   size_t src_offset                     = 0;
@@ -242,7 +243,7 @@ void smpi_global_init()
     xbt_os_walltimer_start(global_timer);
   }
 
-  if (xbt_cfg_get_string("smpi/comp-adjustment-file")[0] != '\0') { 
+  if (xbt_cfg_get_string("smpi/comp-adjustment-file")[0] != '\0') {
     std::string filename {xbt_cfg_get_string("smpi/comp-adjustment-file")};
     std::ifstream fstream(filename);
     if (not fstream.is_open()) {
@@ -650,7 +651,7 @@ int smpi_main(const char* executable, int argc, char *argv[])
   if (MC_is_active()) {
     MC_run();
   } else {
-  
+
     SIMIX_run();
 
     xbt_os_walltimer_stop(global_timer);
@@ -660,7 +661,7 @@ int smpi_main(const char* executable, int argc, char *argv[])
           "The simulation took %g seconds (after parsing and platform setup)\n"
           "%g seconds were actual computation of the application",
           SIMIX_get_clock(), global_time , smpi_total_benched_time);
-          
+
       if (smpi_total_benched_time/global_time>=0.75)
       XBT_INFO("More than 75%% of the time was spent inside the application code.\n"
       "You may want to use sampling functions or trace replay to reduce this.");
@@ -698,7 +699,7 @@ void SMPI_finalize(){
 }
 
 void smpi_mpi_init() {
-  if(smpi_init_sleep > 0) 
+  if(smpi_init_sleep > 0)
     simcall_process_sleep(smpi_init_sleep);
 }
 
@@ -713,7 +714,7 @@ double smpi_mpi_wtime(){
     //     }
     // because the time will not normally advance when only calls to MPI_Wtime
     // are made -> deadlock (MPI_Wtime never reaches the time limit)
-    if(smpi_wtime_sleep > 0) 
+    if(smpi_wtime_sleep > 0)
       simcall_process_sleep(smpi_wtime_sleep);
     smpi_bench_begin();
   } else {
diff --git a/src/smpi/smpi_keyvals.cpp b/src/smpi/smpi_keyvals.cpp
index 53cf51f8ed..25aafe45d3 100644
--- a/src/smpi/smpi_keyvals.cpp
+++ b/src/smpi/smpi_keyvals.cpp
@@ -18,7 +18,7 @@ std::unordered_map<int, void*>* Keyval::attributes(){
 template <> int Keyval::call_deleter<Comm>(Comm* obj, smpi_key_elem elem, int keyval, void * value, int* flag){
   if(elem->delete_fn.comm_delete_fn!=MPI_NULL_DELETE_FN){
     int ret = elem->delete_fn.comm_delete_fn(obj, keyval, value, flag);
-    if(ret!=MPI_SUCCESS) 
+    if(ret!=MPI_SUCCESS)
       return ret;
   }
   return MPI_SUCCESS;
@@ -27,7 +27,7 @@ template <> int Keyval::call_deleter<Comm>(Comm* obj, smpi_key_elem elem, int ke
 template <> int Keyval::call_deleter<Win>(Win* obj, smpi_key_elem elem, int keyval, void * value, int* flag){
   if(elem->delete_fn.win_delete_fn!=MPI_NULL_DELETE_FN){
     int ret = elem->delete_fn.win_delete_fn(obj, keyval, value, flag);
-    if(ret!=MPI_SUCCESS) 
+    if(ret!=MPI_SUCCESS)
       return ret;
   }
   return MPI_SUCCESS;
@@ -36,7 +36,7 @@ template <> int Keyval::call_deleter<Win>(Win* obj, smpi_key_elem elem, int keyv
 template <> int Keyval::call_deleter<Datatype>(Datatype* obj, smpi_key_elem elem, int keyval, void * value, int* flag){
   if(elem->delete_fn.type_delete_fn!=MPI_NULL_DELETE_FN){
     int ret = elem->delete_fn.type_delete_fn(obj, keyval, value, flag);
-    if(ret!=MPI_SUCCESS) 
+    if(ret!=MPI_SUCCESS)
       return ret;
   }
   return MPI_SUCCESS;
diff --git a/src/smpi/smpi_keyvals.hpp b/src/smpi/smpi_keyvals.hpp
index 52ad77ee45..bc3e2af943 100644
--- a/src/smpi/smpi_keyvals.hpp
+++ b/src/smpi/smpi_keyvals.hpp
@@ -27,7 +27,7 @@ typedef struct s_smpi_key_elem {
   smpi_copy_fn copy_fn;
   smpi_delete_fn delete_fn;
   int refcount;
-} s_smpi_mpi_key_elem_t; 
+} s_smpi_mpi_key_elem_t;
 
 typedef struct s_smpi_key_elem *smpi_key_elem;
 
@@ -40,7 +40,7 @@ class Keyval{
   protected:
     std::unordered_map<int, void*>* attributes();
   public:
-// Each subclass should have two members, as we want to separate the ones for Win, Comm, and Datatypes :  
+// Each subclass should have two members, as we want to separate the ones for Win, Comm, and Datatypes :
 //    static std::unordered_map<int, smpi_key_elem> keyvals_;
 //    static int keyval_id_;
     template <typename T> static int keyval_create(smpi_copy_fn copy_fn, smpi_delete_fn delete_fn, int* keyval, void* extra_statee);
diff --git a/src/smpi/smpi_mpi.cpp b/src/smpi/smpi_mpi.cpp
index aa987f6979..9f344ee5fa 100644
--- a/src/smpi/smpi_mpi.cpp
+++ b/src/smpi/smpi_mpi.cpp
@@ -238,7 +238,7 @@ WRAPPED_PMPI_CALL(MPI_Win, MPI_Win_f2c,(MPI_Fint win),(win))
 /*
   Unimplemented Calls - both PMPI and MPI calls are generated.
   When implementing, please move ahead, swap UNIMPLEMENTED_WRAPPED_PMPI_CALL for WRAPPED_PMPI_CALL,
-  and implement PMPI version of the function in smpi_pmpi.cpp file 
+  and implement PMPI version of the function in smpi_pmpi.cpp file
 */
 
 
diff --git a/src/smpi/smpi_pmpi.cpp b/src/smpi/smpi_pmpi.cpp
index 7e9d901fbf..528bcaac99 100644
--- a/src/smpi/smpi_pmpi.cpp
+++ b/src/smpi/smpi_pmpi.cpp
@@ -1015,10 +1015,10 @@ int PMPI_Ssend(void* buf, int count, MPI_Datatype datatype, int dst, int tag, MP
     extra->send_size = count*dt_size_send;
     TRACE_smpi_ptp_in(rank, rank, dst_traced, __FUNCTION__, extra);
     TRACE_smpi_send(rank, rank, dst_traced, tag,count*datatype->size());
-  
+
     simgrid::smpi::Request::ssend(buf, count, datatype, dst, tag, comm);
     retval = MPI_SUCCESS;
-  
+
     TRACE_smpi_ptp_out(rank, rank, dst_traced, __FUNCTION__);
   }
 
@@ -1044,7 +1044,7 @@ int PMPI_Sendrecv(void *sendbuf, int sendcount, MPI_Datatype sendtype, int dst,
   }else if (dst >= comm->group()->size() || dst <0 ||
       (src!=MPI_ANY_SOURCE && (src >= comm->group()->size() || src <0))){
     retval = MPI_ERR_RANK;
-  } else if ((sendcount < 0 || recvcount<0) || 
+  } else if ((sendcount < 0 || recvcount<0) ||
       (sendbuf==nullptr && sendcount > 0) || (recvbuf==nullptr && recvcount>0)) {
     retval = MPI_ERR_COUNT;
   } else if((sendtag<0 && sendtag !=  MPI_ANY_TAG)||(recvtag<0 && recvtag != MPI_ANY_TAG)){
@@ -2620,7 +2620,7 @@ int PMPI_Get( void *origin_addr, int origin_count, MPI_Datatype origin_datatype,
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if ((origin_count < 0 || target_count < 0) ||
@@ -2655,7 +2655,7 @@ int PMPI_Rget( void *origin_addr, int origin_count, MPI_Datatype origin_datatype
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if ((origin_count < 0 || target_count < 0) ||
@@ -2691,7 +2691,7 @@ int PMPI_Put( void *origin_addr, int origin_count, MPI_Datatype origin_datatype,
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if ((origin_count < 0 || target_count < 0) ||
@@ -2727,7 +2727,7 @@ int PMPI_Rput( void *origin_addr, int origin_count, MPI_Datatype origin_datatype
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if ((origin_count < 0 || target_count < 0) ||
@@ -2764,7 +2764,7 @@ int PMPI_Accumulate( void *origin_addr, int origin_count, MPI_Datatype origin_da
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if ((origin_count < 0 || target_count < 0) ||
@@ -2801,7 +2801,7 @@ int PMPI_Raccumulate( void *origin_addr, int origin_count, MPI_Datatype origin_d
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if ((origin_count < 0 || target_count < 0) ||
@@ -2829,8 +2829,8 @@ int PMPI_Raccumulate( void *origin_addr, int origin_count, MPI_Datatype origin_d
   return retval;
 }
 
-int PMPI_Get_accumulate(void *origin_addr, int origin_count, MPI_Datatype origin_datatype, void *result_addr, 
-int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, int target_count, 
+int PMPI_Get_accumulate(void *origin_addr, int origin_count, MPI_Datatype origin_datatype, void *result_addr,
+int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, int target_count,
 MPI_Datatype target_datatype, MPI_Op op, MPI_Win win){
   int retval = 0;
   smpi_bench_end();
@@ -2840,7 +2840,7 @@ MPI_Datatype target_datatype, MPI_Op op, MPI_Win win){
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if ((origin_count < 0 || target_count < 0 || result_count <0) ||
@@ -2859,8 +2859,8 @@ MPI_Datatype target_datatype, MPI_Op op, MPI_Win win){
     int src_traced = group->index(target_rank);
     TRACE_smpi_ptp_in(rank, src_traced, rank, __FUNCTION__, nullptr);
 
-    retval = win->get_accumulate( origin_addr, origin_count, origin_datatype, result_addr, 
-                                  result_count, result_datatype, target_rank, target_disp, 
+    retval = win->get_accumulate( origin_addr, origin_count, origin_datatype, result_addr,
+                                  result_count, result_datatype, target_rank, target_disp,
                                   target_count, target_datatype, op);
 
     TRACE_smpi_ptp_out(rank, src_traced, rank, __FUNCTION__);
@@ -2870,8 +2870,8 @@ MPI_Datatype target_datatype, MPI_Op op, MPI_Win win){
 }
 
 
-int PMPI_Rget_accumulate(void *origin_addr, int origin_count, MPI_Datatype origin_datatype, void *result_addr, 
-int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, int target_count, 
+int PMPI_Rget_accumulate(void *origin_addr, int origin_count, MPI_Datatype origin_datatype, void *result_addr,
+int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, int target_count,
 MPI_Datatype target_datatype, MPI_Op op, MPI_Win win, MPI_Request* request){
   int retval = 0;
   smpi_bench_end();
@@ -2882,7 +2882,7 @@ MPI_Datatype target_datatype, MPI_Op op, MPI_Win win, MPI_Request* request){
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if ((origin_count < 0 || target_count < 0 || result_count <0) ||
@@ -2903,8 +2903,8 @@ MPI_Datatype target_datatype, MPI_Op op, MPI_Win win, MPI_Request* request){
     int src_traced = group->index(target_rank);
     TRACE_smpi_ptp_in(rank, src_traced, rank, __FUNCTION__, nullptr);
 
-    retval = win->get_accumulate( origin_addr, origin_count, origin_datatype, result_addr, 
-                                  result_count, result_datatype, target_rank, target_disp, 
+    retval = win->get_accumulate( origin_addr, origin_count, origin_datatype, result_addr,
+                                  result_count, result_datatype, target_rank, target_disp,
                                   target_count, target_datatype, op, request);
 
     TRACE_smpi_ptp_out(rank, src_traced, rank, __FUNCTION__);
@@ -2928,7 +2928,7 @@ int PMPI_Compare_and_swap(void *origin_addr, void *compare_addr,
     retval = MPI_SUCCESS;
   } else if (target_rank <0){
     retval = MPI_ERR_RANK;
-  } else if (win->dynamic()==0 && target_disp <0){ 
+  } else if (win->dynamic()==0 && target_disp <0){
     //in case of dynamic window, target_disp can be mistakenly seen as negative, as it is an address
     retval = MPI_ERR_ARG;
   } else if (origin_addr==nullptr || result_addr==nullptr || compare_addr==nullptr){
@@ -2942,7 +2942,7 @@ int PMPI_Compare_and_swap(void *origin_addr, void *compare_addr,
     int src_traced = group->index(target_rank);
     TRACE_smpi_ptp_in(rank, src_traced, rank, __FUNCTION__, nullptr);
 
-    retval = win->compare_and_swap( origin_addr, compare_addr, result_addr, datatype, 
+    retval = win->compare_and_swap( origin_addr, compare_addr, result_addr, datatype,
                                   target_rank, target_disp);
 
     TRACE_smpi_ptp_out(rank, src_traced, rank, __FUNCTION__);
@@ -3024,10 +3024,10 @@ int PMPI_Win_lock(int lock_type, int rank, int assert, MPI_Win win){
   smpi_bench_end();
   if (win == MPI_WIN_NULL) {
     retval = MPI_ERR_WIN;
-  } else if (lock_type != MPI_LOCK_EXCLUSIVE && 
+  } else if (lock_type != MPI_LOCK_EXCLUSIVE &&
              lock_type != MPI_LOCK_SHARED) {
     retval = MPI_ERR_LOCKTYPE;
-  } else if (rank == MPI_PROC_NULL){ 
+  } else if (rank == MPI_PROC_NULL){
     retval = MPI_SUCCESS;
   } else {
     int myrank = smpi_process()->index();
@@ -3044,7 +3044,7 @@ int PMPI_Win_unlock(int rank, MPI_Win win){
   smpi_bench_end();
   if (win == MPI_WIN_NULL) {
     retval = MPI_ERR_WIN;
-  } else if (rank == MPI_PROC_NULL){ 
+  } else if (rank == MPI_PROC_NULL){
     retval = MPI_SUCCESS;
   } else {
     int myrank = smpi_process()->index();
@@ -3091,7 +3091,7 @@ int PMPI_Win_flush(int rank, MPI_Win win){
   smpi_bench_end();
   if (win == MPI_WIN_NULL) {
     retval = MPI_ERR_WIN;
-  } else if (rank == MPI_PROC_NULL){ 
+  } else if (rank == MPI_PROC_NULL){
     retval = MPI_SUCCESS;
   } else {
     int myrank = smpi_process()->index();
@@ -3108,7 +3108,7 @@ int PMPI_Win_flush_local(int rank, MPI_Win win){
   smpi_bench_end();
   if (win == MPI_WIN_NULL) {
     retval = MPI_ERR_WIN;
-  } else if (rank == MPI_PROC_NULL){ 
+  } else if (rank == MPI_PROC_NULL){
     retval = MPI_SUCCESS;
   } else {
     int myrank = smpi_process()->index();
diff --git a/src/smpi/smpi_replay.cpp b/src/smpi/smpi_replay.cpp
index c6267613a2..9ba7295206 100644
--- a/src/smpi/smpi_replay.cpp
+++ b/src/smpi/smpi_replay.cpp
@@ -164,8 +164,8 @@ static void action_init(const char *const *action)
 {
   XBT_DEBUG("Initialize the counters");
   CHECK_ACTION_PARAMS(action, 0, 1)
-  if(action[2]) 
-    MPI_DEFAULT_TYPE=MPI_DOUBLE; // default MPE dataype 
+  if(action[2])
+    MPI_DEFAULT_TYPE=MPI_DOUBLE; // default MPE dataype
   else MPI_DEFAULT_TYPE= MPI_BYTE; // default TAU datatype
 
   /* start a simulated timer */
@@ -253,9 +253,9 @@ static void action_Isend(const char *const *action)
   double size=parse_double(action[3]);
   double clock = smpi_process()->simulated_elapsed();
 
-  if(action[4]) 
+  if(action[4])
     MPI_CURRENT_TYPE=decode_datatype(action[4]);
-  else 
+  else
     MPI_CURRENT_TYPE= MPI_DEFAULT_TYPE;
 
   int rank = smpi_process()->index();
@@ -286,9 +286,9 @@ static void action_recv(const char *const *action) {
   double clock = smpi_process()->simulated_elapsed();
   MPI_Status status;
 
-  if(action[4]) 
+  if(action[4])
     MPI_CURRENT_TYPE=decode_datatype(action[4]);
-  else 
+  else
     MPI_CURRENT_TYPE= MPI_DEFAULT_TYPE;
 
   int rank = smpi_process()->index();
@@ -325,9 +325,9 @@ static void action_Irecv(const char *const *action)
   double size=parse_double(action[3]);
   double clock = smpi_process()->simulated_elapsed();
 
-  if(action[4]) 
+  if(action[4])
     MPI_CURRENT_TYPE=decode_datatype(action[4]);
-  else 
+  else
     MPI_CURRENT_TYPE= MPI_DEFAULT_TYPE;
 
   int rank = smpi_process()->index();
@@ -361,8 +361,8 @@ static void action_test(const char *const *action){
 
   MPI_Request request = get_reqq_self()->back();
   get_reqq_self()->pop_back();
-  //if request is null here, this may mean that a previous test has succeeded 
-  //Different times in traced application and replayed version may lead to this 
+  //if request is null here, this may mean that a previous test has succeeded
+  //Different times in traced application and replayed version may lead to this
   //In this case, ignore the extra calls.
   if(request!=nullptr){
     int rank = smpi_process()->index();
@@ -469,12 +469,12 @@ static void action_bcast(const char *const *action)
   double clock = smpi_process()->simulated_elapsed();
   int root=0;
   /* Initialize MPI_CURRENT_TYPE in order to decrease the number of the checks */
-  MPI_CURRENT_TYPE= MPI_DEFAULT_TYPE;  
+  MPI_CURRENT_TYPE= MPI_DEFAULT_TYPE;
 
   if(action[3]) {
     root= atoi(action[3]);
     if(action[4])
-      MPI_CURRENT_TYPE=decode_datatype(action[4]);   
+      MPI_CURRENT_TYPE=decode_datatype(action[4]);
   }
 
   int rank = smpi_process()->index();
@@ -755,8 +755,8 @@ static void action_allgather(const char *const *action) {
   double clock = smpi_process()->simulated_elapsed();
 
   CHECK_ACTION_PARAMS(action, 2, 2)
-  int sendcount=atoi(action[2]); 
-  int recvcount=atoi(action[3]); 
+  int sendcount=atoi(action[2]);
+  int recvcount=atoi(action[3]);
 
   MPI_Datatype MPI_CURRENT_TYPE2 = MPI_DEFAULT_TYPE;
 
diff --git a/src/smpi/smpi_request.cpp b/src/smpi/smpi_request.cpp
index e839534a11..ebfc570ab5 100644
--- a/src/smpi/smpi_request.cpp
+++ b/src/smpi/smpi_request.cpp
@@ -31,7 +31,7 @@ extern void (*smpi_comm_copy_data_callback) (smx_activity_t, void*, size_t);
 namespace simgrid{
 namespace smpi{
 
-Request::Request(void *buf, int count, MPI_Datatype datatype, int src, int dst, int tag, MPI_Comm comm, unsigned flags) : buf_(buf), old_type_(datatype), src_(src), dst_(dst), tag_(tag), comm_(comm), flags_(flags) 
+Request::Request(void *buf, int count, MPI_Datatype datatype, int src, int dst, int tag, MPI_Comm comm, unsigned flags) : buf_(buf), old_type_(datatype), src_(src), dst_(dst), tag_(tag), comm_(comm), flags_(flags)
 {
   void *old_buf = nullptr;
 // FIXME Handle the case of a partial shared malloc.
@@ -131,7 +131,7 @@ int Request::match_recv(void* a, void* b, smx_activity_t ignored) {
       ref->real_src_ = req->src_;
     if(ref->tag_ == MPI_ANY_TAG)
       ref->real_tag_ = req->tag_;
-    if(ref->real_size_ < req->real_size_) 
+    if(ref->real_size_ < req->real_size_)
       ref->truncated_ = 1;
     if(req->detached_==1)
       ref->detached_sender_=req; //tie the sender to the receiver, as it is detached and has to be freed in the receiver
@@ -339,7 +339,7 @@ void Request::start()
 
     if (async_small_thresh == 0 && (flags_ & RMA) == 0 ) {
       mailbox = process->mailbox();
-    } 
+    }
     else if (((flags_ & RMA) != 0) || static_cast<int>(size_) < async_small_thresh) {
       //We have to check both mailboxes (because SSEND messages are sent to the large mbox).
       //begin with the more appropriate one : the small one.
@@ -481,7 +481,7 @@ void Request::start()
 
 void Request::startall(int count, MPI_Request * requests)
 {
-  if(requests== nullptr) 
+  if(requests== nullptr)
     return;
 
   for(int i = 0; i < count; i++) {
@@ -496,7 +496,7 @@ int Request::test(MPI_Request * request, MPI_Status * status) {
   // because the time will not normally advance when only calls to MPI_Test are made -> deadlock
   // multiplier to the sleeptime, to increase speed of execution, each failed test will increase it
   static int nsleeps = 1;
-  if(smpi_test_sleep > 0)  
+  if(smpi_test_sleep > 0)
     simcall_process_sleep(nsleeps*smpi_test_sleep);
 
   Status::empty(status);
@@ -530,7 +530,7 @@ int Request::testsome(int incount, MPI_Request requests[], int *indices, MPI_Sta
         count++;
         if (status != MPI_STATUSES_IGNORE)
           status[i] = *pstat;
-        if ((requests[i] != MPI_REQUEST_NULL) && requests[i]->flags_ & NON_PERSISTENT)
+        if ((requests[i] != MPI_REQUEST_NULL) && (requests[i]->flags_ & NON_PERSISTENT))
           requests[i] = MPI_REQUEST_NULL;
       }
     } else {
@@ -544,7 +544,7 @@ int Request::testsome(int incount, MPI_Request requests[], int *indices, MPI_Sta
 
 int Request::testany(int count, MPI_Request requests[], int *index, MPI_Status * status)
 {
-  std::vector<simgrid::kernel::activity::ActivityImpl*> comms;
+  std::vector<simgrid::kernel::activity::ActivityImplPtr> comms;
   comms.reserve(count);
 
   int i;
@@ -562,12 +562,12 @@ int Request::testany(int count, MPI_Request requests[], int *index, MPI_Status *
   if (not map.empty()) {
     //multiplier to the sleeptime, to increase speed of execution, each failed testany will increase it
     static int nsleeps = 1;
-    if(smpi_test_sleep > 0) 
+    if(smpi_test_sleep > 0)
       simcall_process_sleep(nsleeps*smpi_test_sleep);
 
     i = simcall_comm_testany(comms.data(), comms.size()); // The i-th element in comms matches!
     if (i != -1) { // -1 is not MPI_UNDEFINED but a SIMIX return code. (nothing matches)
-      *index = map[i]; 
+      *index = map[i];
       finish_wait(&requests[*index],status);
       flag             = 1;
       nsleeps          = 1;
@@ -652,8 +652,8 @@ void Request::iprobe(int source, int tag, MPI_Comm comm, int* flag, MPI_Status*
   }
 
   if (request->action_ != nullptr){
-    simgrid::kernel::activity::CommImpl* sync_comm =
-        static_cast<simgrid::kernel::activity::CommImpl*>(request->action_);
+    simgrid::kernel::activity::CommImplPtr sync_comm =
+        boost::static_pointer_cast<simgrid::kernel::activity::CommImpl>(request->action_);
     MPI_Request req                            = static_cast<MPI_Request>(sync_comm->src_data);
     *flag = 1;
     if(status != MPI_STATUS_IGNORE && (req->flags_ & PREPARED) == 0) {
@@ -757,22 +757,23 @@ void Request::wait(MPI_Request * request, MPI_Status * status)
 int Request::waitany(int count, MPI_Request requests[], MPI_Status * status)
 {
   s_xbt_dynar_t comms; // Keep it on stack to save some extra mallocs
-  int i;
   int size = 0;
   int index = MPI_UNDEFINED;
-  int *map;
 
   if(count > 0) {
     // Wait for a request to complete
-    xbt_dynar_init(&comms, sizeof(smx_activity_t), nullptr);
-    map = xbt_new(int, count);
+    xbt_dynar_init(&comms, sizeof(smx_activity_t), [](void*ptr){
+      intrusive_ptr_release(*(simgrid::kernel::activity::ActivityImpl**)ptr);
+    });
+    int *map = xbt_new(int, count);
     XBT_DEBUG("Wait for one of %d", count);
-    for(i = 0; i < count; i++) {
+    for(int i = 0; i < count; i++) {
       if (requests[i] != MPI_REQUEST_NULL && not(requests[i]->flags_ & PREPARED) &&
           not(requests[i]->flags_ & FINISHED)) {
         if (requests[i]->action_ != nullptr) {
           XBT_DEBUG("Waiting any %p ", requests[i]);
-          xbt_dynar_push(&comms, &requests[i]->action_);
+          intrusive_ptr_add_ref(requests[i]->action_.get());
+          xbt_dynar_push_as(&comms, simgrid::kernel::activity::ActivityImpl*, requests[i]->action_.get());
           map[size] = i;
           size++;
         } else {
@@ -786,8 +787,9 @@ int Request::waitany(int count, MPI_Request requests[], MPI_Status * status)
         }
       }
     }
-    if(size > 0) {
-      i = simcall_comm_waitany(&comms, -1);
+    if (size > 0) {
+      XBT_DEBUG("Enter waitany for %lu comms", xbt_dynar_length(&comms));
+      int i = simcall_comm_waitany(&comms, -1);
 
       // not MPI_UNDEFINED, as this is a simix return code
       if (i != -1) {
@@ -822,7 +824,7 @@ int Request::waitall(int count, MPI_Request requests[], MPI_Status status[])
   std::vector<MPI_Request> accumulates;
   int index;
   MPI_Status stat;
-  MPI_Status *pstat = status == MPI_STATUSES_IGNORE ? MPI_STATUS_IGNORE : &stat;
+  MPI_Status *pstat = (status == MPI_STATUSES_IGNORE ? MPI_STATUS_IGNORE : &stat);
   int retvalue = MPI_SUCCESS;
   //tag invalid requests in the set
   if (status != MPI_STATUSES_IGNORE) {
@@ -840,7 +842,7 @@ int Request::waitall(int count, MPI_Request requests[], MPI_Status status[])
       wait(&requests[c],pstat);
       index = c;
     } else {
-      index = waitany(count, requests, pstat);
+      index = waitany(count, (MPI_Request*)requests, pstat);
       if (index == MPI_UNDEFINED)
         break;
 
diff --git a/src/smpi/smpi_topo.cpp b/src/smpi/smpi_topo.cpp
index e23a62332b..432174ed2b 100644
--- a/src/smpi/smpi_topo.cpp
+++ b/src/smpi/smpi_topo.cpp
@@ -19,14 +19,13 @@ static int getfactors(int num, int *nfators, int **factors);
 namespace simgrid{
 namespace smpi{
 
-
-Topo_Graph::~Topo_Graph() 
+Topo_Graph::~Topo_Graph()
 {
   delete[] index_;
   delete[] edges_;
 }
 
-Topo_Dist_Graph::~Topo_Dist_Graph() 
+Topo_Dist_Graph::~Topo_Dist_Graph()
 {
   delete[] in_;
   delete[] in_weights_;
@@ -37,7 +36,7 @@ Topo_Dist_Graph::~Topo_Dist_Graph()
 /*******************************************************************************
  * Cartesian topologies
  ******************************************************************************/
-Topo_Cart::~Topo_Cart() 
+Topo_Cart::~Topo_Cart()
 {
   delete[] dims_;
   delete[] periodic_;
@@ -56,12 +55,11 @@ Topo_Cart::Topo_Cart(int ndims) : ndims_(ndims)
 Topo_Cart::Topo_Cart(MPI_Comm comm_old, int ndims, int dims[], int periods[], int reorder, MPI_Comm *comm_cart) : Topo_Cart(ndims) {
   MPI_Group newGroup;
   MPI_Group oldGroup;
-  int nranks;
 
   int rank = comm_old->rank();
 
-  int newSize = 1;
   if(ndims != 0) {
+    int newSize = 1;
     for (int i = 0 ; i < ndims ; i++) {
       newSize *= dims[i];
     }
@@ -78,7 +76,7 @@ Topo_Cart::Topo_Cart(MPI_Comm comm_old, int ndims, int dims[], int periods[], in
     nnodes_ = newSize;
 
     //  FIXME : code duplication... See coords
-    nranks = newSize;
+    int nranks = newSize;
     for (int i=0; i<ndims; i++) {
       dims_[i] = dims[i];
      periodic_[i] = periods[i];
@@ -101,7 +99,6 @@ Topo_Cart::Topo_Cart(MPI_Comm comm_old, int ndims, int dims[], int periods[], in
 
 Topo_Cart* Topo_Cart::sub(const int remain_dims[], MPI_Comm *newcomm) {
   int oldNDims = ndims_;
-  int j = 0;
   int *newDims = nullptr;
   int *newPeriodic = nullptr;
 
@@ -119,6 +116,7 @@ Topo_Cart* Topo_Cart::sub(const int remain_dims[], MPI_Comm *newcomm) {
     newPeriodic = xbt_new(int, newNDims);
 
     // that should not segfault
+    int j = 0;
     for (int i = 0 ; j < newNDims ; i++) {
       if(remain_dims[i]) {
         newDims[j] =dims_[i];
@@ -152,12 +150,11 @@ int Topo_Cart::get(int maxdims, int* dims, int* periods, int* coords) {
 
 int Topo_Cart::rank(int* coords, int* rank) {
   int ndims =ndims_;
-  int coord;
   *rank = 0;
   int multiplier = 1;
 
   for (int i=ndims-1; i >=0; i-- ) {
-    coord = coords[i];
+    int coord = coords[i];
 
     /* The user can give us whatever coordinates he wants. If one of them is out of range, either this dimension is
      * periodic, and we consider the equivalent coordinate inside the bounds, or it's not and then it's an error
@@ -242,12 +239,12 @@ int Topo_Cart::dim_get(int *ndims) {
  * Copyright (c) 2004-2005 The University of Tennessee and The University
  *                         of Tennessee Research Foundation.  All rights
  *                         reserved.
- * Copyright (c) 2004-2014 High Performance Computing Center Stuttgart, 
+ * Copyright (c) 2004-2014 High Performance Computing Center Stuttgart,
  *                         University of Stuttgart.  All rights reserved.
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
  * Copyright (c) 2012      Los Alamos National Security, LLC.  All rights
- *                         reserved. 
+ *                         reserved.
  * Copyright (c) 2014      Intel, Inc. All rights reserved
  * $COPYRIGHT$
  *
@@ -256,7 +253,6 @@ int Topo_Cart::dim_get(int *ndims) {
  * $HEADER$
  */
 
-
 /*
  * This is a utility function, no need to have anything in the lower layer for this at all
  */
diff --git a/src/smpi/smpi_topo.hpp b/src/smpi/smpi_topo.hpp
index 9e419cf1dc..dc04e4451d 100644
--- a/src/smpi/smpi_topo.hpp
+++ b/src/smpi/smpi_topo.hpp
@@ -37,7 +37,7 @@ class Topo_Cart: public Topo {
     int *periodic_;
     int *position_;
   public:
-    Topo_Cart(int ndims);
+    explicit Topo_Cart(int ndims);
     ~Topo_Cart();
     Topo_Cart(MPI_Comm comm_old, int ndims, int dims[], int periods[], int reorder, MPI_Comm *comm_cart);
     Topo_Cart* sub(const int remain_dims[], MPI_Comm *newcomm) ;
diff --git a/src/smpi/smpi_win.cpp b/src/smpi/smpi_win.cpp
index 6ddaeb52bf..5d83d8cd41 100644
--- a/src/smpi/smpi_win.cpp
+++ b/src/smpi/smpi_win.cpp
@@ -276,7 +276,7 @@ int Win::get( void *origin_addr, int origin_count, MPI_Datatype origin_datatype,
         comm_->group()->index(target_rank), smpi_process()->index(), SMPI_RMA_TAG+2, comm_,
         MPI_OP_NULL);
 
-    //start the send, with another process than us as sender. 
+    //start the send, with another process than us as sender.
     sreq->start();
     //push request to receiver's win
     xbt_mutex_acquire(send_win->mut_);
@@ -320,7 +320,7 @@ int Win::accumulate( void *origin_addr, int origin_count, MPI_Datatype origin_da
     if(locked != 1)
       return MPI_ERR_WIN;
   }
-  //FIXME: local version 
+  //FIXME: local version
 
   if(target_count*target_datatype->get_extent()>recv_win->size_)
     return MPI_ERR_ARG;
@@ -358,8 +358,8 @@ int Win::accumulate( void *origin_addr, int origin_count, MPI_Datatype origin_da
   return MPI_SUCCESS;
 }
 
-int Win::get_accumulate( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, void *result_addr, 
-              int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, int target_count, 
+int Win::get_accumulate( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, void *result_addr,
+              int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, int target_count,
               MPI_Datatype target_datatype, MPI_Op op, MPI_Request* request){
 
   //get sender pointer
diff --git a/src/smpi/smpi_win.hpp b/src/smpi/smpi_win.hpp
index dd837b6718..e219be5588 100644
--- a/src/smpi/smpi_win.hpp
+++ b/src/smpi/smpi_win.hpp
@@ -70,8 +70,8 @@ public:
               MPI_Aint target_disp, int target_count, MPI_Datatype target_datatype, MPI_Request* request=nullptr);
   int accumulate( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, int target_rank,
               MPI_Aint target_disp, int target_count, MPI_Datatype target_datatype, MPI_Op op, MPI_Request* request=nullptr);
-  int get_accumulate( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, void *result_addr, 
-              int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, int target_count, 
+  int get_accumulate( void *origin_addr, int origin_count, MPI_Datatype origin_datatype, void *result_addr,
+              int result_count, MPI_Datatype result_datatype, int target_rank, MPI_Aint target_disp, int target_count,
               MPI_Datatype target_datatype, MPI_Op op, MPI_Request* request=nullptr);
   int compare_and_swap(void *origin_addr, void *compare_addr,
         void *result_addr, MPI_Datatype datatype, int target_rank,
diff --git a/src/surf/HostImpl.cpp b/src/surf/HostImpl.cpp
index 0833a45c1d..72f0f72f68 100644
--- a/src/surf/HostImpl.cpp
+++ b/src/surf/HostImpl.cpp
@@ -128,19 +128,9 @@ xbt_dict_t HostImpl::getMountedStorageList()
 
 void HostImpl::getAttachedStorageList(std::vector<const char*>* storages)
 {
-  xbt_lib_cursor_t cursor;
-  char* key;
-  void** data;
-  xbt_lib_foreach(storage_lib, cursor, key, data)
-  {
-    if (xbt_lib_get_level(xbt_lib_get_elm_or_null(storage_lib, key), SURF_STORAGE_LEVEL) != nullptr) {
-      simgrid::surf::StorageImpl* storage = static_cast<simgrid::surf::StorageImpl*>(
-          xbt_lib_get_level(xbt_lib_get_elm_or_null(storage_lib, key), SURF_STORAGE_LEVEL));
-      if (not strcmp(static_cast<const char*>(storage->attach_), piface_->cname())) {
-        storages->push_back(storage->cname());
-      }
-    }
-  }
+  for (auto s : storage_)
+    if (not strcmp(static_cast<const char*>(s.second->attach_), piface_->cname()))
+      storages->push_back(s.second->piface_.name());
 }
 
 Action* HostImpl::open(const char* fullpath)
diff --git a/src/surf/StorageImpl.cpp b/src/surf/StorageImpl.cpp
index 3eb9910a3d..aea1e2602a 100644
--- a/src/surf/StorageImpl.cpp
+++ b/src/surf/StorageImpl.cpp
@@ -14,9 +14,6 @@
 
 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(surf_storage, surf, "Logging specific to the SURF storage module");
 
-xbt_lib_t storage_lib;
-int MSG_STORAGE_LEVEL                           = -1; // Msg storage level
-int SURF_STORAGE_LEVEL                          = -1;
 simgrid::surf::StorageModel* surf_storage_model = nullptr;
 
 namespace simgrid {
@@ -31,6 +28,17 @@ simgrid::xbt::signal<void(StorageImpl*)> storageDestructedCallbacks;
 simgrid::xbt::signal<void(StorageImpl*, int, int)> storageStateChangedCallbacks; // signature: wasOn, isOn
 simgrid::xbt::signal<void(StorageAction*, Action::State, Action::State)> storageActionStateChangedCallbacks;
 
+/* List of storages */
+std::unordered_map<std::string, StorageImpl*>* StorageImpl::storages =
+    new std::unordered_map<std::string, StorageImpl*>();
+
+StorageImpl* StorageImpl::byName(const char* name)
+{
+  if (storages->find(name) == storages->end())
+    return nullptr;
+  return storages->at(name);
+}
+
 /*********
  * Model *
  *********/
@@ -53,6 +61,7 @@ StorageModel::~StorageModel()
 StorageImpl::StorageImpl(Model* model, const char* name, lmm_system_t maxminSystem, double bread, double bwrite,
                          const char* type_id, const char* content_name, sg_size_t size, const char* attach)
     : Resource(model, name, lmm_constraint_new(maxminSystem, this, MAX(bread, bwrite)))
+    , piface_(this)
     , size_(size)
     , usedSize_(0)
     , typeId_(xbt_strdup(type_id))
@@ -64,6 +73,7 @@ StorageImpl::StorageImpl(Model* model, const char* name, lmm_system_t maxminSyst
   XBT_DEBUG("Create resource with Bread '%f' Bwrite '%f' and Size '%llu'", bread, bwrite, size);
   constraintRead_  = lmm_constraint_new(maxminSystem, this, bread);
   constraintWrite_ = lmm_constraint_new(maxminSystem, this, bwrite);
+  storages->insert({name, this});
 }
 
 StorageImpl::~StorageImpl()
diff --git a/src/surf/StorageImpl.hpp b/src/surf/StorageImpl.hpp
index b02503e94e..9b1af7f869 100644
--- a/src/surf/StorageImpl.hpp
+++ b/src/surf/StorageImpl.hpp
@@ -7,6 +7,7 @@
 #include <xbt/base.h>
 #include <xbt/signal.hpp>
 
+#include "simgrid/s4u/Storage.hpp"
 #include "src/surf/PropertyHolder.hpp"
 #include "surf_interface.hpp"
 #include <map>
@@ -86,7 +87,12 @@ public:
   StorageImpl(Model* model, const char* name, lmm_system_t maxminSystem, double bread, double bwrite,
               const char* type_id, const char* content_name, sg_size_t size, const char* attach);
 
-  ~StorageImpl();
+  ~StorageImpl() override;
+
+public:
+  /** @brief Public interface */
+  s4u::Storage piface_;
+  static StorageImpl* byName(const char* name);
 
   /** @brief Check if the Storage is used (if an action currently uses its resources) */
   bool isUsed() override;
@@ -160,7 +166,8 @@ public:
   virtual sg_size_t getUsedSize();
 
   std::map<std::string, sg_size_t*>* parseContent(const char* filename);
-
+  static std::unordered_map<std::string, StorageImpl*>* storages;
+  static std::unordered_map<std::string, StorageImpl*>* storagesMap() { return StorageImpl::storages; }
   std::vector<StorageAction*> writeActions_;
 
   lmm_constraint_t constraintWrite_; /* Constraint for maximum write bandwidth*/
diff --git a/src/surf/cpu_cas01.cpp b/src/surf/cpu_cas01.cpp
index 7d4e26ea27..6edd4e9f2d 100644
--- a/src/surf/cpu_cas01.cpp
+++ b/src/surf/cpu_cas01.cpp
@@ -164,7 +164,7 @@ void CpuCas01::apply_event(tmgr_trace_event_t event, double value)
   }
 }
 
-/** @brief Start a new execution on this CPU lasting @size flops and using one core */
+/** @brief Start a new execution on this CPU lasting @param size flops and using one core */
 CpuAction *CpuCas01::execution_start(double size)
 {
   return new CpuCas01Action(model(), size, isOff(), speed_.scale * speed_.peak, constraint());
diff --git a/src/surf/fair_bottleneck.cpp b/src/surf/fair_bottleneck.cpp
index 1b2582bdcc..30b45e290d 100644
--- a/src/surf/fair_bottleneck.cpp
+++ b/src/surf/fair_bottleneck.cpp
@@ -78,7 +78,7 @@ void bottleneck_solve(lmm_system_t sys)
 
   XBT_DEBUG("Fair bottleneck Initialized");
 
-  /* 
+  /*
    * Compute Usage and store the variables that reach the maximum.
    */
   do {
diff --git a/src/surf/instr_routing.cpp b/src/surf/instr_routing.cpp
index 2baf7b9d96..ed350fe675 100644
--- a/src/surf/instr_routing.cpp
+++ b/src/surf/instr_routing.cpp
@@ -162,9 +162,9 @@ static void recursiveGraphExtraction(simgrid::s4u::NetZone* netzone, container_t
 /*
  * Callbacks
  */
-void sg_instr_AS_begin(sg_platf_AS_cbarg_t AS)
+static void sg_instr_AS_begin(simgrid::s4u::NetZone& netzone)
 {
-  const char*id = AS->id;
+  const char* id = netzone.name();
 
   if (PJ_container_get_root() == nullptr){
     PJ_container_alloc ();
@@ -195,7 +195,7 @@ void sg_instr_AS_begin(sg_platf_AS_cbarg_t AS)
   }
 }
 
-void sg_instr_AS_end()
+static void sg_instr_AS_end(simgrid::s4u::NetZone& /*netzone*/)
 {
   if (TRACE_needs_platform()){
     currentContainer.pop_back();
@@ -318,11 +318,15 @@ void instr_routing_define_callbacks ()
 {
   //always need the call backs to ASes (we need only the root AS),
   //to create the rootContainer and the rootType properly
-  if (not TRACE_is_enabled() || not TRACE_needs_platform())
+  if (not TRACE_is_enabled())
     return;
-  simgrid::s4u::Link::onCreation.connect(instr_routing_parse_start_link);
-  simgrid::s4u::onPlatformCreated.connect(instr_routing_parse_end_platform);
-  simgrid::s4u::Host::onCreation.connect(sg_instr_new_host);
+  if (TRACE_needs_platform()) {
+    simgrid::s4u::Link::onCreation.connect(instr_routing_parse_start_link);
+    simgrid::s4u::onPlatformCreated.connect(instr_routing_parse_end_platform);
+    simgrid::s4u::Host::onCreation.connect(sg_instr_new_host);
+  }
+  simgrid::s4u::NetZone::onCreation.connect(sg_instr_AS_begin);
+  simgrid::s4u::NetZone::onSeal.connect(sg_instr_AS_end);
 }
 
 /*
diff --git a/src/surf/lagrange.cpp b/src/surf/lagrange.cpp
index 20f16b5951..e02b6fc27c 100644
--- a/src/surf/lagrange.cpp
+++ b/src/surf/lagrange.cpp
@@ -33,7 +33,7 @@ double (*func_fpi_def) (lmm_variable_t, double);
 void lagrange_solve(lmm_system_t sys);
 //computes the value of the dichotomy using a initial values, init, with a specific variable or constraint
 static double dichotomy(double init, double diff(double, void *), void *var_cnst, double min_error);
-//computes the value of the differential of constraint param_cnst applied to lambda  
+//computes the value of the differential of constraint param_cnst applied to lambda
 static double partial_diff_lambda(double lambda, void *param_cnst);
 
 static int __check_feasible(xbt_swag_t cnst_list, xbt_swag_t var_list, int warn)
@@ -194,8 +194,8 @@ void lagrange_solve(lmm_system_t sys)
     XBT_DEBUG("#### cnst(%p)->lambda :  %e", cnst, cnst->lambda);
   }
 
-  /* 
-   * Initialize the var list variable with only the active variables. 
+  /*
+   * Initialize the var list variable with only the active variables.
    * Associate an index in the swag variables. Initialize mu.
    */
   var_list = &(sys->variable_set);
@@ -320,7 +320,7 @@ void lagrange_solve(lmm_system_t sys)
  *
  * @param init initial value for \mu or \lambda
  * @param diff a function that computes the differential of with respect a \mu or \lambda
- * @param var_cnst a pointer to a variable or constraint 
+ * @param var_cnst a pointer to a variable or constraint
  * @param min_erro a minimum error tolerated
  *
  * @return a double corresponding to the result of the dichotomy process
@@ -453,7 +453,7 @@ static double partial_diff_lambda(double lambda, void *param_cnst)
     // Initialize the summation variable
     sigma_i = 0.0;
 
-    // Compute sigma_i 
+    // Compute sigma_i
     for (j = 0; j < var->cnsts_number; j++) {
       sigma_i += (var->cnsts[j].constraint)->lambda;
     }
@@ -476,9 +476,9 @@ static double partial_diff_lambda(double lambda, void *param_cnst)
 }
 
 /** \brief Attribute the value bound to var->bound.
- * 
+ *
  *  \param func_fpi  inverse of the partial differential of f (f prime inverse, (f')^{-1})
- * 
+ *
  *  Set default functions to the ones passed as parameters. This is a polymorphism in C pure, enjoy the roots of
  *  programming.
  *
diff --git a/src/surf/maxmin.cpp b/src/surf/maxmin.cpp
index ffa7f5936c..3d9da37aec 100644
--- a/src/surf/maxmin.cpp
+++ b/src/surf/maxmin.cpp
@@ -247,9 +247,9 @@ int lmm_constraint_sharing_policy(lmm_constraint_t cnst)
   return (cnst->sharing_policy);
 }
 
-/* @brief Remove a constraint 
+/* @brief Remove a constraint
  * Currently this is dead code, but it is exposed in maxmin.h
- * Apparently, this call was designed assuming that constraint would no more have elements in it. 
+ * Apparently, this call was designed assuming that constraint would no more have elements in it.
  * If not the case, assertion will fail, and you need to add calls e.g. to lmm_shrink before effectively removing it.
  */
 inline void lmm_constraint_free(lmm_system_t sys,lmm_constraint_t cnst)
@@ -394,7 +394,7 @@ void lmm_shrink(lmm_system_t sys, lmm_constraint_t cnst, lmm_variable_t var)
     make_constraint_inactive(sys, cnst);
   else {
     //Check maxconcurrency to see if we can enable new variables
-    lmm_on_disabled_var(sys,elem->constraint);       
+    lmm_on_disabled_var(sys,elem->constraint);
   }
 
   lmm_check_concurrency(sys);
@@ -418,7 +418,7 @@ void lmm_expand(lmm_system_t sys, lmm_constraint_t cnst, lmm_variable_t var, dou
     }
   }
 
-  //Check if we need to disable the variable 
+  //Check if we need to disable the variable
   if(var->weight>0 && var->concurrency_share-current_share>lmm_concurrency_slack(cnst)) {
     double weight = var->weight;
     lmm_disable_var(sys,var);
@@ -534,7 +534,7 @@ lmm_variable_t lmm_get_var_from_cnst(lmm_system_t /*sys*/, lmm_constraint_t cnst
       if (*elem == nullptr)
         *elem = (lmm_element_t) xbt_swag_getFirst(&(cnst->disabled_element_set));
     } else {
-      *elem = (lmm_element_t) xbt_swag_getNext(*elem, cnst->disabled_element_set.offset);      
+      *elem = (lmm_element_t) xbt_swag_getNext(*elem, cnst->disabled_element_set.offset);
     }
   }
   if (*elem)
@@ -568,7 +568,7 @@ lmm_variable_t lmm_get_var_from_cnst_safe(lmm_system_t /*sys*/, lmm_constraint_t
       if (*nextelem == nullptr)
         *nextelem = (lmm_element_t) xbt_swag_getFirst(&(cnst->disabled_element_set));
     } else {
-      *nextelem = (lmm_element_t) xbt_swag_getNext(*elem, cnst->disabled_element_set.offset);      
+      *nextelem = (lmm_element_t) xbt_swag_getNext(*elem, cnst->disabled_element_set.offset);
     }
     return (*elem)->variable;
   }else
@@ -609,7 +609,7 @@ static inline void saturated_constraint_set_update(double usage, int cnst_light_
 static inline void saturated_variable_set_update(s_lmm_constraint_light_t *cnst_light_tab,
                                                  dyn_light_t saturated_constraint_set, lmm_system_t sys)
 {
-  /* Add active variables (i.e. variables that need to be set) from the set of constraints to saturate (cnst_light_tab)*/ 
+  /* Add active variables (i.e. variables that need to be set) from the set of constraints to saturate (cnst_light_tab)*/
   lmm_constraint_light_t cnst = nullptr;
   void *_elem;
   lmm_element_t elem = nullptr;
@@ -667,7 +667,7 @@ void lmm_print(lmm_system_t sys)
             std::to_string(elem->variable->value) + ")" + ((cnst->sharing_policy) ? " + " : " , ");
       if(cnst->sharing_policy)
         sum += elem->value * elem->variable->value;
-      else 
+      else
         sum = MAX(sum,elem->value * elem->variable->value);
     }
     //TODO: Adding disabled elements only for test compatibility, but do we really want them to be printed?
@@ -678,7 +678,7 @@ void lmm_print(lmm_system_t sys)
             std::to_string(elem->variable->value) + ")" + ((cnst->sharing_policy) ? " + " : " , ");
       if(cnst->sharing_policy)
         sum += elem->value * elem->variable->value;
-      else 
+      else
         sum = MAX(sum,elem->value * elem->variable->value);
     }
 
@@ -936,11 +936,11 @@ void lmm_solve(lmm_system_t sys)
 }
 
 /** \brief Attribute the value bound to var->bound.
- * 
+ *
  *  \param sys the lmm_system_t
  *  \param var the lmm_variable_t
  *  \param bound the new bound to associate with var
- * 
+ *
  *  Makes var->bound equal to bound. Whenever this function is called a change is  signed in the system. To
  *  avoid false system changing detection it is a good idea to test (bound != 0) before calling it.
  */
@@ -958,7 +958,7 @@ int lmm_concurrency_slack(lmm_constraint_t cnstr){
   if(cnstr->concurrency_limit<0)
     return 666;
 
-  return  cnstr->concurrency_limit - cnstr->concurrency_current;  
+  return  cnstr->concurrency_limit - cnstr->concurrency_current;
 }
 
 /** \brief Measure the minimum concurrency slack across all constraints where the given var is involved */
@@ -981,7 +981,7 @@ int lmm_cnstrs_min_concurrency_slack(lmm_variable_t var){
 
 /* /Check if a variable can be enabled
  *
- * Make sure to set staged_weight before, if your intent is only to check concurrency 
+ * Make sure to set staged_weight before, if your intent is only to check concurrency
  */
 int lmm_can_enable_var(lmm_variable_t var){
   return var->staged_weight>0 && lmm_cnstrs_min_concurrency_slack(var)>=var->concurrency_share;
@@ -1016,7 +1016,7 @@ void lmm_enable_var(lmm_system_t sys, lmm_variable_t var){
 
   //When used within lmm_on_disabled_var, we would get an assertion fail, because transiently there can be variables
   // that are staged and could be activated.
-  //Anyway, caller functions all call lmm_check_concurrency() in the end. 
+  //Anyway, caller functions all call lmm_check_concurrency() in the end.
   //  lmm_check_concurrency(sys);
 }
 
@@ -1046,9 +1046,9 @@ void lmm_disable_var(lmm_system_t sys, lmm_variable_t var){
   var->value = 0.0;
   lmm_check_concurrency(sys);
 }
- 
+
 /* /brief Find variables that can be enabled and enable them.
- * 
+ *
  * Assuming that the variable has already been removed from non-zero weights
  * Can we find a staged variable to add?
  * If yes, check that none of the constraints that this variable is involved in is at the limit of its concurrency
@@ -1088,7 +1088,7 @@ void lmm_on_disabled_var(lmm_system_t sys, lmm_constraint_t cnstr){
 
   //We could get an assertion fail, because transiently there can be variables that are staged and could be activated.
   //And we need to go through all constraints of the disabled var before getting back a coherent state.
-  //Anyway, caller functions all call lmm_check_concurrency() in the end. 
+  //Anyway, caller functions all call lmm_check_concurrency() in the end.
   //  lmm_check_concurrency(sys);
 }
 
@@ -1184,7 +1184,7 @@ static void lmm_update_modified_set_rec(lmm_system_t sys, lmm_constraint_t cnst)
         lmm_update_modified_set_rec(sys, cnsts[i].constraint);
       }
     }
-    //var will be ignored in later visits as long as sys->visited_counter does not move 
+    //var will be ignored in later visits as long as sys->visited_counter does not move
     var->visited = sys->visited_counter;
   }
 }
@@ -1214,7 +1214,7 @@ static void lmm_remove_all_modified_set(lmm_system_t sys)
   void *_var;
     xbt_swag_foreach(_var, &sys->variable_set)
       ((lmm_variable_t)_var)->visited = 0;
-  } 
+  }
   xbt_swag_reset(&sys->modified_constraint_set);
 }
 
diff --git a/src/surf/maxmin_private.hpp b/src/surf/maxmin_private.hpp
index 9b4ecae5b4..f0c93fb14d 100644
--- a/src/surf/maxmin_private.hpp
+++ b/src/surf/maxmin_private.hpp
@@ -15,7 +15,7 @@
  * @brief LMM element
  * Elements can be seen as glue between constraint objects and variable objects.
  * Basically, each variable will have a set of elements, one for each constraint where it is involved.
- * Then, it is used to list all variables involved in constraint through constraint's xxx_element_set lists, or vice-versa list all constraints for a given variable.  
+ * Then, it is used to list all variables involved in constraint through constraint's xxx_element_set lists, or vice-versa list all constraints for a given variable.
  */
 typedef struct lmm_element {
   /* hookup to constraint */
@@ -37,11 +37,11 @@ typedef struct lmm_constraint_light {
 
 /** @ingroup SURF_lmm
  * @brief LMM constraint
- * Each constraint contains several partially overlapping logical sets of elements: 
+ * Each constraint contains several partially overlapping logical sets of elements:
  * \li Disabled elements which variable's weight is zero. This variables are not at all processed by LMM, but eventually the corresponding action will enable it (at least this is the idea).
  * \li Enabled elements which variable's weight is non-zero. They are utilized in some LMM functions.
- * \li Active elements which variable's weight is non-zero (i.e. it is enabled) AND its element value is non-zero. LMM_solve iterates over active elements during resolution, dynamically making them active or unactive. 
- * 
+ * \li Active elements which variable's weight is non-zero (i.e. it is enabled) AND its element value is non-zero. LMM_solve iterates over active elements during resolution, dynamically making them active or unactive.
+ *
  */
 typedef struct lmm_constraint {
   /* hookup to system */
@@ -57,10 +57,10 @@ typedef struct lmm_constraint {
   double usage;
   double bound;
   int concurrency_limit; /* The maximum number of variables that may be enabled at any time (stage variables if necessary) */
-  //TODO MARTIN Check maximum value across resources at the end of simulation and give a warning is more than e.g. 500 
+  //TODO MARTIN Check maximum value across resources at the end of simulation and give a warning is more than e.g. 500
   int concurrency_current; /* The current concurrency */
   int concurrency_maximum; /* The maximum number of (enabled and disabled) variables associated to the constraint at any given time (essentially for tracing)*/
-  
+
   int sharing_policy; /* see @e_surf_link_sharing_policy_t (0: FATPIPE, 1: SHARED, 2: FULLDUPLEX) */
   void *id;
   int id_int;
@@ -71,7 +71,7 @@ typedef struct lmm_constraint {
 
 /** @ingroup SURF_lmm
  * @brief LMM variable
- * 
+ *
  * When something prevents us from enabling a variable, we "stage" the weight that we would have like to set, so that as soon as possible we enable the variable with desired weight
  */
 typedef struct lmm_variable {
@@ -136,7 +136,7 @@ typedef struct lmm_system {
 
 /** @ingroup SURF_lmm
  * @brief Print information about a lmm system
- * 
+ *
  * @param sys A lmm system
  */
 //XBT_PRIVATE void lmm_print(lmm_system_t sys);
diff --git a/src/surf/network_cm02.hpp b/src/surf/network_cm02.hpp
index 04f6d98786..a04f21518f 100644
--- a/src/surf/network_cm02.hpp
+++ b/src/surf/network_cm02.hpp
@@ -31,56 +31,55 @@ namespace simgrid {
  *********/
 
 namespace simgrid {
-  namespace surf {
-
-    class NetworkCm02Model : public NetworkModel {
-    public:
-      NetworkCm02Model();
-      explicit NetworkCm02Model(void (*solve_fun)(lmm_system_t self));
-      virtual ~NetworkCm02Model() = default;
-      LinkImpl* createLink(const char* name, double bandwidth, double latency,
-                           e_surf_link_sharing_policy_t policy) override;
-      void updateActionsStateLazy(double now, double delta) override;
-      void updateActionsStateFull(double now, double delta) override;
-      Action* communicate(s4u::Host* src, s4u::Host* dst, double size, double rate) override;
-      virtual void gapAppend(double size, const LinkImpl* link, NetworkAction* action);
-
-    protected:
-      bool haveGap_ = false;
-    };
-
-    /************
-     * Resource *
-     ************/
-
-    class NetworkCm02Link : public LinkImpl {
-    public:
-      NetworkCm02Link(NetworkCm02Model* model, const char* name, double bandwidth, double latency,
-                      e_surf_link_sharing_policy_t policy, lmm_system_t system);
-      virtual ~NetworkCm02Link() = default;
-      void apply_event(tmgr_trace_event_t event, double value) override;
-      void setBandwidth(double value) override;
-      void setLatency(double value) override;
-      virtual void gapAppend(double size, const LinkImpl* link, NetworkAction* action);
-    };
-
-
-    /**********
-     * Action *
-     **********/
-    class NetworkCm02Action : public NetworkAction {
-      friend Action* NetworkCm02Model::communicate(s4u::Host* src, s4u::Host* dst, double size, double rate);
-      friend NetworkSmpiModel;
-    public:
-      NetworkCm02Action(Model *model, double cost, bool failed)
-      : NetworkAction(model, cost, failed) {};
-      virtual ~NetworkCm02Action() = default;
-      void updateRemainingLazy(double now) override;
-    protected:
-      double senderGap_;
-    };
-
-  }
+namespace surf {
+
+class NetworkCm02Model : public NetworkModel {
+public:
+  NetworkCm02Model();
+  explicit NetworkCm02Model(void (*solve_fun)(lmm_system_t self));
+  virtual ~NetworkCm02Model() = default;
+  LinkImpl* createLink(const char* name, double bandwidth, double latency,
+                       e_surf_link_sharing_policy_t policy) override;
+  void updateActionsStateLazy(double now, double delta) override;
+  void updateActionsStateFull(double now, double delta) override;
+  Action* communicate(s4u::Host* src, s4u::Host* dst, double size, double rate) override;
+  virtual void gapAppend(double size, const LinkImpl* link, NetworkAction* action);
+
+protected:
+  bool haveGap_ = false;
+};
+
+/************
+ * Resource *
+ ************/
+
+class NetworkCm02Link : public LinkImpl {
+public:
+  NetworkCm02Link(NetworkCm02Model* model, const char* name, double bandwidth, double latency,
+                  e_surf_link_sharing_policy_t policy, lmm_system_t system);
+  virtual ~NetworkCm02Link() = default;
+  void apply_event(tmgr_trace_event_t event, double value) override;
+  void setBandwidth(double value) override;
+  void setLatency(double value) override;
+  virtual void gapAppend(double size, const LinkImpl* link, NetworkAction* action);
+};
+
+/**********
+ * Action *
+ **********/
+class NetworkCm02Action : public NetworkAction {
+  friend Action* NetworkCm02Model::communicate(s4u::Host* src, s4u::Host* dst, double size, double rate);
+  friend NetworkSmpiModel;
+
+public:
+  NetworkCm02Action(Model* model, double cost, bool failed) : NetworkAction(model, cost, failed){};
+  virtual ~NetworkCm02Action() = default;
+  void updateRemainingLazy(double now) override;
+
+protected:
+  double senderGap_;
+};
+}
 }
 
 #endif /* SURF_NETWORK_CM02_HPP_ */
diff --git a/src/surf/network_ns3.cpp b/src/surf/network_ns3.cpp
index 2cf77d2ed5..9404d6bc43 100644
--- a/src/surf/network_ns3.cpp
+++ b/src/surf/network_ns3.cpp
@@ -57,7 +57,7 @@ NetPointNs3::NetPointNs3()
 static void clusterCreation_cb(sg_platf_cluster_cbarg_t cluster)
 {
   char* lat = bprintf("%fs", cluster->lat);
-  char* bw  = bprintf("%fBps", cluster->bw);
+  char* bw  = bprintf("%fbps", cluster->bw * 8);
 
   for (int i : *cluster->radicals) {
     // Routers don't create a router on the other end of the private link by themselves.
@@ -96,7 +96,7 @@ static void routeCreation_cb(bool symmetrical, simgrid::kernel::routing::NetPoin
 
     XBT_DEBUG("Route from '%s' to '%s' with link '%s' %s", src->cname(), dst->cname(), link->cname(),
               (symmetrical ? "(symmetrical)" : "(not symmetrical)"));
-    char* link_bdw = bprintf("%fBps", link->bandwidth());
+    char* link_bdw = bprintf("%fbps", link->bandwidth() * 8);
     char* link_lat = bprintf("%fs", link->latency());
 
     //   XBT_DEBUG("src (%s), dst (%s), src_id = %d, dst_id = %d",src,dst, src_id, dst_id);
diff --git a/src/surf/plugins/host_energy.cpp b/src/surf/plugins/host_energy.cpp
index 9a441b93ac..f7878d7b11 100644
--- a/src/surf/plugins/host_energy.cpp
+++ b/src/surf/plugins/host_energy.cpp
@@ -113,45 +113,54 @@ void HostEnergy::update()
   double finish_time = surf_get_clock();
   double cpu_load;
   double current_speed = host->speed();
-  if (current_speed <= 0)
-    // Some users declare a pstate of speed 0 flops (e.g., to model boot time).
-    // We consider that the machine is then fully loaded. That's arbitrary but it avoids a NaN
-    cpu_load = 1;
-  else
-    cpu_load = lmm_constraint_get_usage(host->pimpl_cpu->constraint()) / current_speed;
-
-  /** Divide by the number of cores here **/
-  cpu_load /= host->pimpl_cpu->coreCount();
-
-  if (cpu_load > 1) // A machine with a load > 1 consumes as much as a fully loaded machine, not more
-    cpu_load = 1;
-
-  /* The problem with this model is that the load is always 0 or 1, never something less.
-   * Another possibility could be to model the total energy as
-   *
-   *   X/(X+Y)*W_idle + Y/(X+Y)*W_burn
-   *
-   * where X is the amount of idling cores, and Y the amount of computing cores.
-   */
 
-  double previous_energy = this->total_energy;
+  if (start_time < finish_time) {
+    // We may have start == finish if the past consumption was updated since the simcall was started
+    // for example if 2 actors requested to update the same host's consumption in a given scheduling round.
+    //
+    // Even in this case, we need to save the pstate for the next call (after this big if),
+    // which may have changed since that recent update.
+
+    if (current_speed <= 0)
+      // Some users declare a pstate of speed 0 flops (e.g., to model boot time).
+      // We consider that the machine is then fully loaded. That's arbitrary but it avoids a NaN
+      cpu_load = 1;
+    else
+      cpu_load = lmm_constraint_get_usage(host->pimpl_cpu->constraint()) / current_speed;
+
+    /** Divide by the number of cores here **/
+    cpu_load /= host->pimpl_cpu->coreCount();
 
-  double instantaneous_consumption;
-  if (this->pstate == -1) // The host was off at the beginning of this time interval
-    instantaneous_consumption = this->watts_off;
-  else
-    instantaneous_consumption = this->getCurrentWattsValue(cpu_load);
+    if (cpu_load > 1) // A machine with a load > 1 consumes as much as a fully loaded machine, not more
+      cpu_load = 1;
 
-  double energy_this_step = instantaneous_consumption * (finish_time - start_time);
+    /* The problem with this model is that the load is always 0 or 1, never something less.
+     * Another possibility could be to model the total energy as
+     *
+     *   X/(X+Y)*W_idle + Y/(X+Y)*W_burn
+     *
+     * where X is the amount of idling cores, and Y the amount of computing cores.
+     */
+
+    double previous_energy = this->total_energy;
+
+    double instantaneous_consumption;
+    if (this->pstate == -1) // The host was off at the beginning of this time interval
+      instantaneous_consumption = this->watts_off;
+    else
+      instantaneous_consumption = this->getCurrentWattsValue(cpu_load);
 
-  //TODO Trace: Trace energy_this_step from start_time to finish_time in host->name()
+    double energy_this_step = instantaneous_consumption * (finish_time - start_time);
 
-  this->total_energy = previous_energy + energy_this_step;
-  this->last_updated = finish_time;
+    // TODO Trace: Trace energy_this_step from start_time to finish_time in host->name()
 
-  XBT_DEBUG(
-      "[update_energy of %s] period=[%.2f-%.2f]; current power peak=%.0E flop/s; consumption change: %.2f J -> %.2f J",
-      host->cname(), start_time, finish_time, host->pimpl_cpu->speed_.peak, previous_energy, energy_this_step);
+    this->total_energy = previous_energy + energy_this_step;
+    this->last_updated = finish_time;
+
+    XBT_DEBUG("[update_energy of %s] period=[%.2f-%.2f]; current power peak=%.0E flop/s; consumption change: %.2f J -> "
+              "%.2f J",
+              host->cname(), start_time, finish_time, host->pimpl_cpu->speed_.peak, previous_energy, energy_this_step);
+  }
 
   /* Save data for the upcoming time interval: whether it's on/off and the pstate if it's on */
   this->pstate = host->isOn() ? host->pstate() : -1;
@@ -372,7 +381,26 @@ void sg_host_energy_plugin_init()
   simgrid::surf::CpuAction::onStateChange.connect(&onActionStateChange);
 }
 
+/** @brief updates the consumption of all hosts
+ *
+ * After this call, sg_host_get_consumed_energy() will not interrupt your process
+ * (until after the next clock update).
+ */
+void sg_host_energy_update_all()
+{
+  simgrid::simix::kernelImmediate([]() {
+    std::vector<simgrid::s4u::Host*> list;
+    simgrid::s4u::Engine::instance()->hostList(&list);
+    for (auto host : list)
+      host->extension<HostEnergy>()->update();
+  });
+}
+
 /** @brief Returns the total energy consumed by the host so far (in Joules)
+ *
+ *  Please note that since the consumption is lazily updated, it may require a simcall to update it.
+ *  The result is that the actor requesting this value will be interrupted,
+ *  the value will be updated in kernel mode before returning the control to the requesting actor.
  *
  *  See also @ref SURF_plugin_energy.
  */
diff --git a/src/surf/ptask_L07.cpp b/src/surf/ptask_L07.cpp
index c960134bbf..24129d9f56 100644
--- a/src/surf/ptask_L07.cpp
+++ b/src/surf/ptask_L07.cpp
@@ -12,7 +12,6 @@
 #include "ptask_L07.hpp"
 
 #include "cpu_interface.hpp"
-#include "xbt/lib.h"
 
 XBT_LOG_EXTERNAL_DEFAULT_CATEGORY(surf_host);
 XBT_LOG_EXTERNAL_CATEGORY(xbt_cfg);
@@ -41,7 +40,7 @@ HostL07Model::HostL07Model() : HostModel() {
   surf_cpu_model_pm = new CpuL07Model(this,maxminSystem_);
 }
 
-HostL07Model::~HostL07Model() 
+HostL07Model::~HostL07Model()
 {
   lmm_system_free(maxminSystem_);
   maxminSystem_ = nullptr;
diff --git a/src/surf/sg_platf.cpp b/src/surf/sg_platf.cpp
index 2e10ec0209..c9e3c5dd52 100644
--- a/src/surf/sg_platf.cpp
+++ b/src/surf/sg_platf.cpp
@@ -4,6 +4,7 @@
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
 #include "simgrid/s4u/Engine.hpp"
+#include "simgrid/s4u/Storage.hpp"
 
 #include "src/kernel/EngineImpl.hpp"
 #include "src/simix/smx_private.h"
@@ -99,7 +100,7 @@ void sg_platf_new_host(sg_platf_host_cbarg_t args)
     host->pimpl_cpu->setPState(args->pstate);
   if (args->coord && strcmp(args->coord, ""))
     new simgrid::kernel::routing::vivaldi::Coords(host->pimpl_netpoint, args->coord);
-    
+
 }
 
 /** @brief Add a "router" to the network element list */
@@ -424,8 +425,7 @@ void sg_platf_new_mount(sg_platf_mount_cbarg_t mount){
 
   if (mount_list.empty())
     XBT_DEBUG("Create a Mount list for %s", A_surfxml_host_id);
-  mount_list.insert(
-      {std::string(mount->name), surf_storage_resource_priv(surf_storage_resource_by_name(mount->storageId))});
+  mount_list.insert({std::string(mount->name), simgrid::surf::StorageImpl::byName(mount->storageId)});
 }
 
 void sg_platf_new_route(sg_platf_route_cbarg_t route)
@@ -662,8 +662,7 @@ simgrid::s4u::NetZone* sg_platf_new_AS_begin(sg_platf_AS_cbarg_t AS)
   /* set the new current component of the tree */
   current_routing = new_as;
 
-  if (TRACE_is_enabled())
-    sg_instr_AS_begin(AS);
+  simgrid::s4u::NetZone::onCreation(*new_as); // notify the signal
 
   return new_as;
 }
@@ -678,10 +677,8 @@ void sg_platf_new_AS_seal()
 {
   xbt_assert(current_routing, "Cannot seal the current AS: none under construction");
   current_routing->seal();
+  simgrid::s4u::NetZone::onSeal(*current_routing);
   current_routing = static_cast<simgrid::kernel::routing::NetZoneImpl*>(current_routing->father());
-
-  if (TRACE_is_enabled())
-    sg_instr_AS_end();
 }
 
 /** @brief Add a link connecting an host to the rest of its AS (which must be cluster or vivaldi) */
diff --git a/src/surf/storage_n11.cpp b/src/surf/storage_n11.cpp
index 5cffd796e2..2cedbb9eea 100644
--- a/src/surf/storage_n11.cpp
+++ b/src/surf/storage_n11.cpp
@@ -17,17 +17,10 @@ extern std::map<std::string, storage_type_t> storage_types;
 
 static void check_disk_attachment()
 {
-  xbt_lib_cursor_t cursor;
-  char* key;
-  void** data;
-  xbt_lib_foreach(storage_lib, cursor, key, data) {
-    if (xbt_lib_get_level(xbt_lib_get_elm_or_null(storage_lib, key), SURF_STORAGE_LEVEL) != nullptr) {
-      simgrid::surf::StorageImpl* storage =
-          static_cast<simgrid::surf::StorageImpl*>(xbt_lib_get_or_null(storage_lib, key, SURF_STORAGE_LEVEL));
-      simgrid::kernel::routing::NetPoint* host_elm = sg_netpoint_by_name_or_null(storage->attach_);
-      if (not host_elm)
-        surf_parse_error("Unable to attach storage %s: host %s does not exist.", storage->cname(), storage->attach_);
-    }
+  for (auto s : *simgrid::surf::StorageImpl::storagesMap()) {
+    simgrid::kernel::routing::NetPoint* host_elm = sg_netpoint_by_name_or_null(s.second->attach_);
+    if (not host_elm)
+      surf_parse_error("Unable to attach storage %s: host %s does not exist.", s.second->cname(), s.second->attach_);
   }
 }
 
@@ -35,9 +28,6 @@ void storage_register_callbacks()
 {
   simgrid::s4u::onPlatformCreated.connect(check_disk_attachment);
   instr_routing_define_callbacks();
-
-  SURF_STORAGE_LEVEL =
-      xbt_lib_add_level(storage_lib, [](void* self) { delete static_cast<simgrid::surf::StorageImpl*>(self); });
 }
 
 /*********
@@ -65,7 +55,6 @@ StorageImpl* StorageN11Model::createStorage(const char* id, const char* type_id,
 
   StorageImpl* storage = new StorageN11(this, id, maxminSystem_, Bread, Bwrite, type_id, (char*)content_name,
                                         storage_type->size, (char*)attach);
-  xbt_lib_set(storage_lib, id, SURF_STORAGE_LEVEL, storage);
   storageCreatedCallbacks(storage);
 
   XBT_DEBUG("SURF storage create resource\n\t\tid '%s'\n\t\ttype '%s'\n\t\tBread '%f'\n", id, type_id, Bread);
@@ -158,6 +147,7 @@ StorageN11::StorageN11(StorageModel* model, const char* name, lmm_system_t maxmi
     : StorageImpl(model, name, maxminSystem, bread, bwrite, type_id, content_name, size, attach)
 {
   XBT_DEBUG("Create resource with Bread '%f' Bwrite '%f' and Size '%llu'", bread, bwrite, size);
+  simgrid::s4u::Storage::onCreation(this->piface_);
 }
 
 StorageAction *StorageN11::open(const char* mount, const char* path)
diff --git a/src/surf/storage_n11.hpp b/src/surf/storage_n11.hpp
index 00bd08d9ab..72703a1093 100644
--- a/src/surf/storage_n11.hpp
+++ b/src/surf/storage_n11.hpp
@@ -42,6 +42,7 @@ class StorageN11 : public StorageImpl {
 public:
   StorageN11(StorageModel* model, const char* name, lmm_system_t maxminSystem, double bread, double bwrite,
              const char* type_id, char* content_name, sg_size_t size, char* attach);
+  virtual ~StorageN11() = default;
   StorageAction *open(const char* mount, const char* path);
   StorageAction *close(surf_file_t fd);
   StorageAction *ls(const char *path);
diff --git a/src/surf/surf_c_bindings.cpp b/src/surf/surf_c_bindings.cpp
index 23cc5cc12a..0f0061812d 100644
--- a/src/surf/surf_c_bindings.cpp
+++ b/src/surf/surf_c_bindings.cpp
@@ -206,28 +206,34 @@ int surf_host_file_move(sg_host_t host, surf_file_t fd, const char* fullpath){
   return host->pimpl_->fileMove(fd, fullpath);
 }
 
-sg_size_t surf_storage_get_size(surf_resource_t resource){
-  return static_cast<simgrid::surf::StorageImpl*>(surf_storage_resource_priv(resource))->size_;
+sg_size_t surf_storage_get_size(surf_storage_t resource)
+{
+  return static_cast<simgrid::surf::StorageImpl*>(resource)->size_;
 }
 
-sg_size_t surf_storage_get_free_size(surf_resource_t resource){
-  return static_cast<simgrid::surf::StorageImpl*>(surf_storage_resource_priv(resource))->getFreeSize();
+sg_size_t surf_storage_get_free_size(surf_storage_t resource)
+{
+  return static_cast<simgrid::surf::StorageImpl*>(resource)->getFreeSize();
 }
 
-sg_size_t surf_storage_get_used_size(surf_resource_t resource){
-  return static_cast<simgrid::surf::StorageImpl*>(surf_storage_resource_priv(resource))->getUsedSize();
+sg_size_t surf_storage_get_used_size(surf_storage_t resource)
+{
+  return static_cast<simgrid::surf::StorageImpl*>(resource)->getUsedSize();
 }
 
-xbt_dict_t surf_storage_get_properties(surf_resource_t resource){
-  return static_cast<simgrid::surf::StorageImpl*>(surf_storage_resource_priv(resource))->getProperties();
+xbt_dict_t surf_storage_get_properties(surf_storage_t resource)
+{
+  return static_cast<simgrid::surf::StorageImpl*>(resource)->getProperties();
 }
 
-const char* surf_storage_get_host(surf_resource_t resource){
-  return static_cast<simgrid::surf::StorageImpl*>(surf_storage_resource_priv(resource))->attach_;
+const char* surf_storage_get_host(surf_storage_t resource)
+{
+  return static_cast<simgrid::surf::StorageImpl*>(resource)->attach_;
 }
 
-const char* surf_storage_get_name(surf_resource_t resource){
-  return static_cast<simgrid::surf::StorageImpl*>(surf_storage_resource_priv(resource))->cname();
+const char* surf_storage_get_name(surf_storage_t resource)
+{
+  return static_cast<simgrid::surf::StorageImpl*>(resource)->cname();
 }
 
 void surf_cpu_action_set_bound(surf_action_t action, double bound) {
diff --git a/src/surf/surf_interface.cpp b/src/surf/surf_interface.cpp
index f318471add..0bfd73d145 100644
--- a/src/surf/surf_interface.cpp
+++ b/src/surf/surf_interface.cpp
@@ -256,11 +256,6 @@ int find_model_description(s_surf_model_description_t * table,
   return -1;
 }
 
-static inline void surf_storage_free(void *r)
-{
-  delete static_cast<simgrid::surf::StorageImpl*>(r);
-}
-
 void sg_version_check(int lib_version_major, int lib_version_minor, int lib_version_patch)
 {
   if ((lib_version_major != SIMGRID_VERSION_MAJOR) || (lib_version_minor != SIMGRID_VERSION_MINOR)) {
@@ -345,12 +340,8 @@ void surf_init(int *argc, char **argv)
   XBT_DEBUG("Create all Libs");
   USER_HOST_LEVEL = simgrid::s4u::Host::extension_create(nullptr);
 
-  storage_lib = xbt_lib_new();
   watched_hosts_lib = xbt_dict_new_homogeneous(nullptr);
 
-  XBT_DEBUG("Add SURF levels");
-  SURF_STORAGE_LEVEL = xbt_lib_add_level(storage_lib,surf_storage_free);
-
   xbt_init(argc, argv);
   if (not all_existing_models)
     all_existing_models = new std::vector<simgrid::surf::Model*>();
@@ -371,7 +362,6 @@ void surf_exit()
   TRACE_end();                  /* Just in case it was not called by the upper layer (or there is no upper layer) */
 
   sg_host_exit();
-  xbt_lib_free(&storage_lib);
   sg_link_exit();
   xbt_dict_free(&watched_hosts_lib);
   for (auto e : storage_types) {
@@ -383,6 +373,7 @@ void surf_exit()
     delete stype->model_properties;
     free(stype);
   }
+  delete simgrid::surf::StorageImpl::storages;
 
   for (auto model : *all_existing_models)
     delete model;
diff --git a/src/surf/xml/platf_private.hpp b/src/surf/xml/platf_private.hpp
index 2bc61f3f77..931f78a7bc 100644
--- a/src/surf/xml/platf_private.hpp
+++ b/src/surf/xml/platf_private.hpp
@@ -238,16 +238,9 @@ XBT_PUBLIC(int) surf_parse_get_debug();
 XBT_PUBLIC(void) surf_parse_set_debug(int bdebug);
 XBT_PUBLIC(int) surf_parse_lex_destroy();
 
-/* To include files (?) */
-XBT_PRIVATE void surfxml_bufferstack_push(int _new);
-XBT_PRIVATE void surfxml_bufferstack_pop(int _new);
-XBT_PUBLIC_DATA(int) surfxml_bufferstack_size;
-
 XBT_PUBLIC(void) routing_route_free(sg_platf_route_cbarg_t route);
 /********** Instr. **********/
-XBT_PRIVATE void sg_instr_AS_begin(sg_platf_AS_cbarg_t AS);
 XBT_PRIVATE void sg_instr_new_router(const char* name);
-XBT_PRIVATE void sg_instr_AS_end();
 
 SG_END_DECL()
 
diff --git a/src/surf/xml/simgrid_dtd.c b/src/surf/xml/simgrid_dtd.c
index 834cf3fbdb..18e4d7338a 100644
--- a/src/surf/xml/simgrid_dtd.c
+++ b/src/surf/xml/simgrid_dtd.c
@@ -50,7 +50,7 @@
 #if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L
 
 /* C99 says to define __STDC_LIMIT_MACROS before including stdint.h,
- * if you want the limit (max/min) macros for int types. 
+ * if you want the limit (max/min) macros for int types.
  */
 #ifndef __STDC_LIMIT_MACROS
 #define __STDC_LIMIT_MACROS 1
@@ -67,7 +67,7 @@ typedef uint32_t flex_uint32_t;
 typedef signed char flex_int8_t;
 typedef short int flex_int16_t;
 typedef int flex_int32_t;
-typedef unsigned char flex_uint8_t; 
+typedef unsigned char flex_uint8_t;
 typedef unsigned short int flex_uint16_t;
 typedef unsigned int flex_uint32_t;
 
@@ -200,7 +200,7 @@ extern FILE *surf_parse_in, *surf_parse_out;
                     if ( *p == '\n' )\
                         --surf_parse_lineno;\
             }while(0)
-    
+
 /* Return all but the first "n" matched characters back to the input stream. */
 #define yyless(n) \
   do \
@@ -4940,38 +4940,38 @@ static yyconst flex_int16_t yy_chk[13918] =
 /* Table of booleans, true if rule could match eol. */
 static yyconst flex_int32_t yy_rule_can_match_eol[648] =
     {   0,
-0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 
-    1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 
-    0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 
-    1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 
-    0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 
-    1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 
-    1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 
-    1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 
-    0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 
-    1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 
-    1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 
-    1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 
-    1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 
-    1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 
-    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 
-    0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 
-    0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 
+0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0,
+    1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0,
+    0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,
+    1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0,
+    0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,
+    1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1,
+    1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1,
+    1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0,
+    0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1,
+    1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1,
+    1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,
+    1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0,
+    1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1,
+    1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1,
+    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0,
+    0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0,
+    0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0,
     0, 0, 1, 0, 0, 0, 1, 0,     };
 
 static yy_state_type yy_last_accepting_state;
@@ -4994,13 +4994,13 @@ char *surf_parse_text;
  * FleXML is Copyright (C) 1999-2005 Kristoffer Rose.  All rights reserved.
  * FleXML is Copyright (C) 2003-2013 Martin Quinson.  All rights reserved.
  * (1.9.6).
- * 
+ *
  * There are two, intertwined parts to this program, part A and part B.
  *
  * Part A
  * ------
- * 
- * Some parts, here collectively called "Part A", are found in the 
+ *
+ * Some parts, here collectively called "Part A", are found in the
  * FleXML package.  They are Copyright (C) 1999-2005 Kristoffer Rose
  * and Copyright (C) 2003-2013 Martin Quinson. All rights reserved.
  *
@@ -5018,20 +5018,20 @@ char *surf_parse_text;
  * Notice that these are explicit rights granted to you for files
  * generated by the FleXML system.  For your rights in connection with
  * the FleXML system itself please consult the GNU General Public License.
- * 
+ *
  * Part B
  * ------
- * 
- * The other parts, here collectively called "Part B", and which came 
- * from the DTD used by FleXML to generate this program, can be 
+ *
+ * The other parts, here collectively called "Part B", and which came
+ * from the DTD used by FleXML to generate this program, can be
  * distributed (or not, as the case may be) under the terms of whoever
- * wrote them, provided these terms respect and obey the two conditions 
+ * wrote them, provided these terms respect and obey the two conditions
  * above under the heading "Part A".
  *
  * The author of and contributors to FleXML specifically disclaim
- * any copyright interest in "Part B", unless "Part B" was written 
+ * any copyright interest in "Part B", unless "Part B" was written
  * by the author of or contributors to FleXML.
- * 
+ *
  */
 
 /* Version strings. */
@@ -5044,7 +5044,7 @@ const char surfxml_flexml_version[] = "1.9.6";
 #include <assert.h>
 #include <stdarg.h>
 #include <ctype.h>
-     
+
 #if defined(_WIN32) || defined(__WIN32__) || defined(WIN32) || defined(__TOS_WIN__)
 # ifndef __STRICT_ANSI__
 #  include <io.h>
@@ -5053,7 +5053,7 @@ const char surfxml_flexml_version[] = "1.9.6";
 #else
 # include <unistd.h>
 #endif
-     
+
 #ifndef FLEXML_INDEXSTACKSIZE
 #define FLEXML_INDEXSTACKSIZE 1000
 #endif
@@ -5556,7 +5556,7 @@ static void pushbuffer(int p)
 {
     ck_ilimit();
     indexstack[inext++] = p;
-    indexstack[inext++] = bnext;    
+    indexstack[inext++] = bnext;
 }
 
 static int popbuffer(void)
@@ -5806,7 +5806,7 @@ extern int surf_parse_wrap (void );
 #endif
 
 #ifndef YY_NO_UNPUT
-    
+
 #endif
 
 #ifndef yytext_ptr
@@ -5830,11 +5830,11 @@ static int input (void );
         static int yy_start_stack_ptr = 0;
         static int yy_start_stack_depth = 0;
         static int *yy_start_stack = NULL;
-    
+
     static void yy_push_state (int _new_state );
-    
+
     static void yy_pop_state (void );
-    
+
 /* Amount of stuff to slurp up with each read. */
 #ifndef YY_READ_BUF_SIZE
 #ifdef __ia64__
@@ -5942,7 +5942,7 @@ YY_DECL
   yy_state_type yy_current_state;
   char *yy_cp, *yy_bp;
   int yy_act;
-    
+
   if ( !(yy_init) )
     {
     (yy_init) = 1;
@@ -6199,7 +6199,7 @@ yy_find_action:
       unsigned int yyl;
       for ( yyl = 0; yyl < surf_parse_leng; ++yyl )
         if ( surf_parse_text[yyl] == '\n' )
-          
+
     surf_parse_lineno++;
 ;
       }
@@ -6264,7 +6264,7 @@ SKIP;
 case 11:
 /* rule 11 can match eol */
 YY_RULE_SETUP
-SET(DOCTYPE); 
+SET(DOCTYPE);
   YY_BREAK
 case 12:
 /* rule 12 can match eol */
@@ -8855,7 +8855,7 @@ case YY_STATE_EOF(E_surfxml_peer):
 if(!ETag_surfxml_include_state()) FAIL("Premature EOF: `</peer>' expected.");
   YY_BREAK
 
-/* <!-- 
+/* <!--
   *              DTD of SimGrid platform and deployment files.
   * More info: http://simgrid.gforge.inria.fr/simgrid/latest/doc/platform.html
   * To upgrade your files, use the tool simgrid_update_xml
@@ -8868,15 +8868,15 @@ if(!ETag_surfxml_include_state()) FAIL("Premature EOF: `</peer>' expected.");
   *    - Rename (power->speed) the attributes describing the amount of flop
   *      that a <host>, <peer>, <cluster> or <cabinet> can deliver per second.
   *    - In <trace_connect>, attribute kind="POWER" is now kind="SPEED".
-  *    
+  *
   *    - In <host> and <link>, attributes availability and state are gone.
   *      It was redundent with state and availability traces, and with peak values.
-  *      
-  *    - In <cluster>, cannot set the availability nor state traces. 
+  *
+  *    - In <cluster>, cannot set the availability nor state traces.
   *      This was too complex and unused.
   *    - The DOCTYPE points to the right URL (this file):
   *      http://simgrid.gforge.inria.fr/simgrid/simgrid.dtd
-  *      
+  *
   *    - Kill <gpu>. Was not doing anything.
   *    - A warning is emitted for unit-less values (they are still accepted).
   *      - speed. Default: 'f' or 'flops'. Also defined:
@@ -8905,7 +8905,7 @@ if(!ETag_surfxml_include_state()) FAIL("Premature EOF: `</peer>' expected.");
   *      - PLATFORM_DESCRIPTION -> PLATFORM
   * * New in DTD version 1 (in SimGrid 3.3):
   *    - DTD is now versionned with the version attribute of platform
-  *    - Unit change: 
+  *    - Unit change:
   *      - Link bandwidth: from Mb/s to b/s
   *      - CPU speed: from MFlop/s to Flop/s
   * -->  */
@@ -10855,7 +10855,7 @@ static int yy_get_next_buffer (void)
 {
   yy_state_type yy_current_state;
   char *yy_cp;
-    
+
   yy_current_state = (yy_start);
 
   for ( yy_cp = (yytext_ptr) + YY_MORE_ADJ; yy_cp < (yy_c_buf_p); ++yy_cp )
@@ -10919,7 +10919,7 @@ static int yy_get_next_buffer (void)
 
 {
   int c;
-    
+
   *(yy_c_buf_p) = (yy_hold_char);
 
   if ( *(yy_c_buf_p) == YY_END_OF_BUFFER_CHAR )
@@ -10981,7 +10981,7 @@ static int yy_get_next_buffer (void)
   (yy_hold_char) = *++(yy_c_buf_p);
 
   if ( c == '\n' )
-    
+
     surf_parse_lineno++;
 ;
 
@@ -10991,12 +10991,12 @@ static int yy_get_next_buffer (void)
 
 /** Immediately switch to a different input stream.
  * @param input_file A readable stream.
- * 
+ *
  * @note This function does not reset the start condition to @c INITIAL .
  */
     void surf_parse_restart  (FILE * input_file )
 {
-    
+
   if ( ! YY_CURRENT_BUFFER ){
         surf_parse_ensure_buffer_stack ();
     YY_CURRENT_BUFFER_LVALUE =
@@ -11009,11 +11009,11 @@ static int yy_get_next_buffer (void)
 
 /** Switch to a different input buffer.
  * @param new_buffer The new input buffer.
- * 
+ *
  */
     void surf_parse__switch_to_buffer  (YY_BUFFER_STATE  new_buffer )
 {
-    
+
   /* TODO. We should be able to replace this entire function body
    * with
    *    surf_parse_pop_buffer_state();
@@ -11053,13 +11053,13 @@ static void surf_parse__load_buffer_state  (void)
 /** Allocate and initialize an input buffer state.
  * @param file A readable stream.
  * @param size The character buffer size in bytes. When in doubt, use @c YY_BUF_SIZE.
- * 
+ *
  * @return the allocated buffer state.
  */
     YY_BUFFER_STATE surf_parse__create_buffer  (FILE * file, int  size )
 {
   YY_BUFFER_STATE b;
-    
+
   b = (YY_BUFFER_STATE) surf_parse_alloc(sizeof( struct yy_buffer_state )  );
   if ( ! b )
     YY_FATAL_ERROR( "out of dynamic memory in surf_parse__create_buffer()" );
@@ -11082,11 +11082,11 @@ static void surf_parse__load_buffer_state  (void)
 
 /** Destroy the buffer.
  * @param b a buffer created with surf_parse__create_buffer()
- * 
+ *
  */
     void surf_parse__delete_buffer (YY_BUFFER_STATE  b )
 {
-    
+
   if ( ! b )
     return;
 
@@ -11107,7 +11107,7 @@ static void surf_parse__load_buffer_state  (void)
 
 {
   int oerrno = errno;
-    
+
   surf_parse__flush_buffer(b );
 
   b->yy_input_file = file;
@@ -11123,13 +11123,13 @@ static void surf_parse__load_buffer_state  (void)
     }
 
         b->yy_is_interactive = file ? (isatty( fileno(file) ) > 0) : 0;
-    
+
   errno = oerrno;
 }
 
 /** Discard all buffered characters. On the next scan, YY_INPUT will be called.
  * @param b the buffer state to be flushed, usually @c YY_CURRENT_BUFFER.
- * 
+ *
  */
     void surf_parse__flush_buffer (YY_BUFFER_STATE  b )
 {
@@ -11158,7 +11158,7 @@ static void surf_parse__load_buffer_state  (void)
  *  the current state. This function will allocate the stack
  *  if necessary.
  *  @param new_buffer The new state.
- *  
+ *
  */
 void surf_parse_push_buffer_state (YY_BUFFER_STATE new_buffer )
 {
@@ -11188,7 +11188,7 @@ void surf_parse_push_buffer_state (YY_BUFFER_STATE new_buffer )
 
 /** Removes and deletes the top of the stack, if present.
  *  The next element becomes the new top.
- *  
+ *
  */
 void surf_parse_pop_buffer_state (void)
 {
@@ -11212,7 +11212,7 @@ void surf_parse_pop_buffer_state (void)
 static void surf_parse_ensure_buffer_stack (void)
 {
   int num_to_alloc;
-    
+
   if (!(yy_buffer_stack)) {
 
     /* First allocation is just for 2 elements, since we don't know if this
@@ -11255,13 +11255,13 @@ static void surf_parse_ensure_buffer_stack (void)
 /** Setup the input buffer state to scan directly from a user-specified character buffer.
  * @param base the character buffer
  * @param size the size in bytes of the character buffer
- * 
+ *
  * @return the newly allocated buffer state object.
  */
 YY_BUFFER_STATE surf_parse__scan_buffer  (char * base, yy_size_t  size )
 {
   YY_BUFFER_STATE b;
-    
+
   if ( size < 2 ||
        base[size-2] != YY_END_OF_BUFFER_CHAR ||
        base[size-1] != YY_END_OF_BUFFER_CHAR )
@@ -11290,14 +11290,14 @@ YY_BUFFER_STATE surf_parse__scan_buffer  (char * base, yy_size_t  size )
 /** Setup the input buffer state to scan a string. The next call to surf_parse_lex() will
  * scan from a @e copy of @a str.
  * @param yystr a NUL-terminated string to scan
- * 
+ *
  * @return the newly allocated buffer state object.
  * @note If you want to scan bytes that may contain NUL values, then use
  *       surf_parse__scan_bytes() instead.
  */
 YY_BUFFER_STATE surf_parse__scan_string (yyconst char * yystr )
 {
-    
+
   return surf_parse__scan_bytes(yystr,(int) strlen(yystr) );
 }
 
@@ -11305,7 +11305,7 @@ YY_BUFFER_STATE surf_parse__scan_string (yyconst char * yystr )
  * scan from a @e copy of @a bytes.
  * @param yybytes the byte buffer to scan
  * @param _yybytes_len the number of bytes in the buffer pointed to by @a bytes.
- * 
+ *
  * @return the newly allocated buffer state object.
  */
 YY_BUFFER_STATE surf_parse__scan_bytes  (yyconst char * yybytes, int  _yybytes_len )
@@ -11314,7 +11314,7 @@ YY_BUFFER_STATE surf_parse__scan_bytes  (yyconst char * yybytes, int  _yybytes_l
   char *buf;
   yy_size_t n;
   int i;
-    
+
   /* Get memory for full buffer, including space for trailing EOB's. */
   n = (yy_size_t) (_yybytes_len + 2);
   buf = (char *) surf_parse_alloc(n  );
@@ -11400,16 +11400,16 @@ static void yynoreturn yy_fatal_error (yyconst char* msg )
 /* Accessor  methods (get/set functions) to struct members. */
 
 /** Get the current line number.
- * 
+ *
  */
 int surf_parse_get_lineno  (void)
 {
-    
+
     return surf_parse_lineno;
 }
 
 /** Get the input stream.
- * 
+ *
  */
 FILE *surf_parse_get_in  (void)
 {
@@ -11417,7 +11417,7 @@ FILE *surf_parse_get_in  (void)
 }
 
 /** Get the output stream.
- * 
+ *
  */
 FILE *surf_parse_get_out  (void)
 {
@@ -11425,7 +11425,7 @@ FILE *surf_parse_get_out  (void)
 }
 
 /** Get the length of the current token.
- * 
+ *
  */
 int surf_parse_get_leng  (void)
 {
@@ -11433,7 +11433,7 @@ int surf_parse_get_leng  (void)
 }
 
 /** Get the current token.
- * 
+ *
  */
 
 char *surf_parse_get_text  (void)
@@ -11443,18 +11443,18 @@ char *surf_parse_get_text  (void)
 
 /** Set the current line number.
  * @param _line_number line number
- * 
+ *
  */
 void surf_parse_set_lineno (int  _line_number )
 {
-    
+
     surf_parse_lineno = _line_number;
 }
 
 /** Set the input stream. This does not discard the current
  * input buffer.
  * @param _in_str A readable stream.
- * 
+ *
  * @see surf_parse__switch_to_buffer
  */
 void surf_parse_set_in (FILE *  _in_str )
@@ -11485,7 +11485,7 @@ static int yy_init_globals (void)
 
     /* We do not touch surf_parse_lineno unless the option is enabled. */
     surf_parse_lineno =  1;
-    
+
     (yy_buffer_stack) = NULL;
     (yy_buffer_stack_top) = 0;
     (yy_buffer_stack_max) = 0;
@@ -11515,7 +11515,7 @@ static int yy_init_globals (void)
 /* surf_parse_lex_destroy is for both reentrant and non-reentrant scanners. */
 int surf_parse_lex_destroy  (void)
 {
-    
+
     /* Pop the buffer stack, destroying each element. */
   while(YY_CURRENT_BUFFER){
     surf_parse__delete_buffer(YY_CURRENT_BUFFER  );
@@ -11545,7 +11545,7 @@ int surf_parse_lex_destroy  (void)
 #ifndef yytext_ptr
 static void yy_flex_strncpy (char* s1, yyconst char * s2, int n )
 {
-    
+
   int i;
   for ( i = 0; i < n; ++i )
     s1[i] = s2[i];
@@ -11570,7 +11570,7 @@ void *surf_parse_alloc (yy_size_t  size )
 
 void *surf_parse_realloc  (void * ptr, yy_size_t  size )
 {
-    
+
   /* The cast to (char *) in the following accommodates both
    * implementations that use char* generic pointers, and those
    * that use void* generic pointers.  It works with the latter
diff --git a/src/surf/xml/simgrid_dtd.h b/src/surf/xml/simgrid_dtd.h
index fe1abe2cb6..60cd6ffd2e 100644
--- a/src/surf/xml/simgrid_dtd.h
+++ b/src/surf/xml/simgrid_dtd.h
@@ -4,13 +4,13 @@
  * FleXML is Copyright (C) 1999-2005 Kristoffer Rose.  All rights reserved.
  * FleXML is Copyright (C) 2003-2013 Martin Quinson.  All rights reserved.
  * (1.9.6).
- * 
+ *
  * There are two, intertwined parts to this program, part A and part B.
  *
  * Part A
  * ------
- * 
- * Some parts, here collectively called "Part A", are found in the 
+ *
+ * Some parts, here collectively called "Part A", are found in the
  * FleXML package.  They are Copyright (C) 1999-2005 Kristoffer Rose
  * and Copyright (C) 2003-2013 Martin Quinson. All rights reserved.
  *
@@ -28,20 +28,20 @@
  * Notice that these are explicit rights granted to you for files
  * generated by the FleXML system.  For your rights in connection with
  * the FleXML system itself please consult the GNU General Public License.
- * 
+ *
  * Part B
  * ------
- * 
- * The other parts, here collectively called "Part B", and which came 
- * from the DTD used by FleXML to generate this program, can be 
+ *
+ * The other parts, here collectively called "Part B", and which came
+ * from the DTD used by FleXML to generate this program, can be
  * distributed (or not, as the case may be) under the terms of whoever
- * wrote them, provided these terms respect and obey the two conditions 
+ * wrote them, provided these terms respect and obey the two conditions
  * above under the heading "Part A".
  *
  * The author of and contributors to FleXML specifically disclaim
- * any copyright interest in "Part B", unless "Part B" was written 
+ * any copyright interest in "Part B", unless "Part B" was written
  * by the author of or contributors to FleXML.
- * 
+ *
  */
 
 #ifndef _FLEXML_simgrid_H
diff --git a/src/surf/xml/surfxml_parseplatf.cpp b/src/surf/xml/surfxml_parseplatf.cpp
index 7e3c02e28f..5f02c5230a 100644
--- a/src/surf/xml/surfxml_parseplatf.cpp
+++ b/src/surf/xml/surfxml_parseplatf.cpp
@@ -29,93 +29,43 @@ XBT_LOG_EXTERNAL_DEFAULT_CATEGORY(surf_parse);
 
 SG_BEGIN_DECL()
 
-/*
- *  Allow the cluster tag to mess with the parsing buffer.
- * (this will probably become obsolete once the cluster tag do not mess with the parsing callbacks directly)
- */
-
-/* This buffer is used to store the original buffer before substituting it by out own buffer. Useful for the cluster tag */
-static std::vector<char*> surfxml_bufferstack_stack;
-int surfxml_bufferstack_size = 2048;
-
-static char *old_buff = nullptr;
-
-XBT_IMPORT_NO_EXPORT(unsigned int) surfxml_buffer_stack_stack_ptr;
-XBT_IMPORT_NO_EXPORT(unsigned int) surfxml_buffer_stack_stack[1024];
-
-void surfxml_bufferstack_push(int new_one)
-{
-  if (not new_one)
-    old_buff = surfxml_bufferstack;
-  else {
-    surfxml_bufferstack_stack.push_back(surfxml_bufferstack);
-    surfxml_bufferstack = xbt_new0(char, surfxml_bufferstack_size);
-  }
-}
-
-void surfxml_bufferstack_pop(int new_one)
-{
-  if (not new_one)
-    surfxml_bufferstack = old_buff;
-  else {
-    free(surfxml_bufferstack);
-    surfxml_bufferstack = surfxml_bufferstack_stack.back();
-    surfxml_bufferstack_stack.pop_back();
-  }
-}
-
-/*
- * Trace related stuff
- */
+/* Trace related stuff */
 
 xbt_dict_t traces_set_list = nullptr;
-XBT_PRIVATE xbt_dict_t trace_connect_list_host_avail = nullptr;
-XBT_PRIVATE xbt_dict_t trace_connect_list_host_speed = nullptr;
-XBT_PRIVATE xbt_dict_t trace_connect_list_link_avail = nullptr;
-XBT_PRIVATE xbt_dict_t trace_connect_list_link_bw = nullptr;
-XBT_PRIVATE xbt_dict_t trace_connect_list_link_lat = nullptr;
+XBT_PRIVATE std::unordered_map<std::string, std::string> trace_connect_list_host_avail;
+XBT_PRIVATE std::unordered_map<std::string, std::string> trace_connect_list_host_speed;
+XBT_PRIVATE std::unordered_map<std::string, std::string> trace_connect_list_link_avail;
+XBT_PRIVATE std::unordered_map<std::string, std::string> trace_connect_list_link_bw;
+XBT_PRIVATE std::unordered_map<std::string, std::string> trace_connect_list_link_lat;
 
 void sg_platf_trace_connect(sg_platf_trace_connect_cbarg_t trace_connect)
 {
   xbt_assert(xbt_dict_get_or_null(traces_set_list, trace_connect->trace),
-              "Cannot connect trace %s to %s: trace unknown",
-              trace_connect->trace,
-              trace_connect->element);
+             "Cannot connect trace %s to %s: trace unknown", trace_connect->trace, trace_connect->element);
 
   switch (trace_connect->kind) {
   case SURF_TRACE_CONNECT_KIND_HOST_AVAIL:
-    xbt_dict_set(trace_connect_list_host_avail,
-        trace_connect->trace,
-        xbt_strdup(trace_connect->element), nullptr);
+    trace_connect_list_host_avail.insert({trace_connect->trace, trace_connect->element});
     break;
   case SURF_TRACE_CONNECT_KIND_SPEED:
-    xbt_dict_set(trace_connect_list_host_speed, trace_connect->trace,
-        xbt_strdup(trace_connect->element), nullptr);
+    trace_connect_list_host_speed.insert({trace_connect->trace, trace_connect->element});
     break;
   case SURF_TRACE_CONNECT_KIND_LINK_AVAIL:
-    xbt_dict_set(trace_connect_list_link_avail,
-        trace_connect->trace,
-        xbt_strdup(trace_connect->element), nullptr);
+    trace_connect_list_link_avail.insert({trace_connect->trace, trace_connect->element});
     break;
   case SURF_TRACE_CONNECT_KIND_BANDWIDTH:
-    xbt_dict_set(trace_connect_list_link_bw,
-        trace_connect->trace,
-        xbt_strdup(trace_connect->element), nullptr);
+    trace_connect_list_link_bw.insert({trace_connect->trace, trace_connect->element});
     break;
   case SURF_TRACE_CONNECT_KIND_LATENCY:
-    xbt_dict_set(trace_connect_list_link_lat, trace_connect->trace,
-        xbt_strdup(trace_connect->element), nullptr);
+    trace_connect_list_link_lat.insert({trace_connect->trace, trace_connect->element});
     break;
   default:
-  surf_parse_error("Cannot connect trace %s to %s: kind of trace unknown",
-        trace_connect->trace, trace_connect->element);
+    surf_parse_error("Cannot connect trace %s to %s: kind of trace unknown", trace_connect->trace,
+                     trace_connect->element);
     break;
   }
 }
 
-
-/* ***************************************** */
-
 static int after_config_done;
 void parse_after_config() {
   if (not after_config_done) {
@@ -163,76 +113,63 @@ void parse_platform_file(const char *file)
     int parse_status;
 
     /* init the flex parser */
-    surfxml_buffer_stack_stack_ptr = 1;
-    surfxml_buffer_stack_stack[0] = 0;
     after_config_done = 0;
     surf_parse_open(file);
 
     traces_set_list = xbt_dict_new_homogeneous(nullptr);
-    trace_connect_list_host_avail = xbt_dict_new_homogeneous(free);
-    trace_connect_list_host_speed = xbt_dict_new_homogeneous(free);
-    trace_connect_list_link_avail = xbt_dict_new_homogeneous(free);
-    trace_connect_list_link_bw = xbt_dict_new_homogeneous(free);
-    trace_connect_list_link_lat = xbt_dict_new_homogeneous(free);
 
     /* Do the actual parsing */
     parse_status = surf_parse();
 
     /* connect all traces relative to hosts */
-    xbt_dict_cursor_t cursor = nullptr;
-    char* trace_name;
-    char* elm;
+    for (auto elm : trace_connect_list_host_avail) {
+      tmgr_trace_t trace = (tmgr_trace_t)xbt_dict_get_or_null(traces_set_list, elm.first.c_str());
+      xbt_assert(trace, "Trace %s undefined", elm.first.c_str());
 
-    xbt_dict_foreach(trace_connect_list_host_avail, cursor, trace_name, elm) {
-      tmgr_trace_t trace = (tmgr_trace_t) xbt_dict_get_or_null(traces_set_list, trace_name);
-      xbt_assert(trace, "Trace %s undefined", trace_name);
-
-      simgrid::s4u::Host *host = sg_host_by_name(elm);
-      xbt_assert(host, "Host %s undefined", elm);
+      simgrid::s4u::Host* host = sg_host_by_name(elm.second.c_str());
+      xbt_assert(host, "Host %s undefined", elm.second.c_str());
       simgrid::surf::Cpu *cpu = host->pimpl_cpu;
 
       cpu->setStateTrace(trace);
     }
-    xbt_dict_foreach(trace_connect_list_host_speed, cursor, trace_name, elm) {
-      tmgr_trace_t trace = (tmgr_trace_t) xbt_dict_get_or_null(traces_set_list, trace_name);
-      xbt_assert(trace, "Trace %s undefined", trace_name);
 
-      simgrid::s4u::Host *host = sg_host_by_name(elm);
-      xbt_assert(host, "Host %s undefined", elm);
+    for (auto elm : trace_connect_list_host_speed) {
+      tmgr_trace_t trace = (tmgr_trace_t)xbt_dict_get_or_null(traces_set_list, elm.first.c_str());
+      xbt_assert(trace, "Trace %s undefined", elm.first.c_str());
+
+      simgrid::s4u::Host* host = sg_host_by_name(elm.second.c_str());
+      xbt_assert(host, "Host %s undefined", elm.second.c_str());
       simgrid::surf::Cpu *cpu = host->pimpl_cpu;
 
       cpu->setSpeedTrace(trace);
     }
-    xbt_dict_foreach(trace_connect_list_link_avail, cursor, trace_name, elm) {
-      tmgr_trace_t trace = (tmgr_trace_t) xbt_dict_get_or_null(traces_set_list, trace_name);
-      xbt_assert(trace, "Trace %s undefined", trace_name);
-      sg_link_t link = simgrid::s4u::Link::byName(elm);
-      xbt_assert(link, "Link %s undefined", elm);
+
+    for (auto elm : trace_connect_list_link_avail) {
+      tmgr_trace_t trace = (tmgr_trace_t)xbt_dict_get_or_null(traces_set_list, elm.first.c_str());
+      xbt_assert(trace, "Trace %s undefined", elm.first.c_str());
+
+      sg_link_t link = simgrid::s4u::Link::byName(elm.second.c_str());
+      xbt_assert(link, "Link %s undefined", elm.second.c_str());
       link->setStateTrace(trace);
     }
 
-    xbt_dict_foreach(trace_connect_list_link_bw, cursor, trace_name, elm) {
-      tmgr_trace_t trace = (tmgr_trace_t) xbt_dict_get_or_null(traces_set_list, trace_name);
-      xbt_assert(trace, "Trace %s undefined", trace_name);
-      sg_link_t link = simgrid::s4u::Link::byName(elm);
-      xbt_assert(link, "Link %s undefined", elm);
+    for (auto elm : trace_connect_list_link_bw) {
+      tmgr_trace_t trace = (tmgr_trace_t)xbt_dict_get_or_null(traces_set_list, elm.first.c_str());
+      xbt_assert(trace, "Trace %s undefined", elm.first.c_str());
+      sg_link_t link = simgrid::s4u::Link::byName(elm.second.c_str());
+      xbt_assert(link, "Link %s undefined", elm.second.c_str());
       link->setBandwidthTrace(trace);
     }
 
-    xbt_dict_foreach(trace_connect_list_link_lat, cursor, trace_name, elm) {
-      tmgr_trace_t trace = (tmgr_trace_t) xbt_dict_get_or_null(traces_set_list, trace_name);
-      xbt_assert(trace, "Trace %s undefined", trace_name);
-      sg_link_t link = simgrid::s4u::Link::byName(elm);
-      xbt_assert(link, "Link %s undefined", elm);
+    for (auto elm : trace_connect_list_link_lat) {
+      tmgr_trace_t trace = (tmgr_trace_t)xbt_dict_get_or_null(traces_set_list, elm.first.c_str());
+      xbt_assert(trace, "Trace %s undefined", elm.first.c_str());
+      sg_link_t link = simgrid::s4u::Link::byName(elm.second.c_str());
+      xbt_assert(link, "Link %s undefined", elm.second.c_str());
       link->setLatencyTrace(trace);
     }
 
     /* Free my data */
-    xbt_dict_free(&trace_connect_list_host_avail);
-    xbt_dict_free(&trace_connect_list_host_speed);
-    xbt_dict_free(&trace_connect_list_link_avail);
-    xbt_dict_free(&trace_connect_list_link_bw);
-    xbt_dict_free(&trace_connect_list_link_lat);
     xbt_dict_free(&traces_set_list);
 
     surf_parse_close();
diff --git a/src/xbt/automaton/automaton.c b/src/xbt/automaton/automaton.c
index 2fb1c6adad..008ead8886 100644
--- a/src/xbt/automaton/automaton.c
+++ b/src/xbt/automaton/automaton.c
@@ -35,7 +35,7 @@ xbt_automaton_state_t xbt_automaton_state_new(xbt_automaton_t a, int type, char*
   state->type = type;
   state->id = xbt_strdup(id);
   state->in = xbt_dynar_new(sizeof(xbt_automaton_transition_t), xbt_automaton_transition_free_voidp);
-  state->out = xbt_dynar_new(sizeof(xbt_automaton_transition_t), xbt_automaton_transition_free_voidp); 
+  state->out = xbt_dynar_new(sizeof(xbt_automaton_transition_t), xbt_automaton_transition_free_voidp);
   xbt_dynar_push(a->states, &state);
   return state;
 }
@@ -283,7 +283,7 @@ int xbt_automaton_transition_compare(const void *t1, const void *t2){
 
   if(xbt_automaton_state_compare(((xbt_automaton_transition_t)t1)->src, ((xbt_automaton_transition_t)t2)->src))
     return 1;
-  
+
   if(xbt_automaton_state_compare(((xbt_automaton_transition_t)t1)->dst, ((xbt_automaton_transition_t)t2)->dst))
     return 1;
 
@@ -291,7 +291,7 @@ int xbt_automaton_transition_compare(const void *t1, const void *t2){
     return 1;
 
   return 0;
-  
+
 }
 
 int xbt_automaton_exp_label_compare(xbt_automaton_exp_label_t l1, xbt_automaton_exp_label_t l2){
@@ -300,7 +300,7 @@ int xbt_automaton_exp_label_compare(xbt_automaton_exp_label_t l1, xbt_automaton_
     return 1;
 
   switch(l1->type){
-  case 0 : // OR 
+  case 0 : // OR
   case 1 : // AND
     if(xbt_automaton_exp_label_compare(l1->u.or_and.left_exp, l2->u.or_and.left_exp))
       return 1;
@@ -330,7 +330,7 @@ int xbt_automaton_propositional_symbols_compare_value(xbt_dynar_t s1, xbt_dynar_
     int* iptr2 = xbt_dynar_get_ptr(s2, cursor);
     if(*iptr1 != *iptr2)
       return 1;
-  } 
+  }
 
   return 0;
 }
diff --git a/src/xbt/automaton/automaton_lexer.yy.c b/src/xbt/automaton/automaton_lexer.yy.c
index d451e561ec..8715148d49 100644
--- a/src/xbt/automaton/automaton_lexer.yy.c
+++ b/src/xbt/automaton/automaton_lexer.yy.c
@@ -53,7 +53,7 @@
 #if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L
 
 /* C99 says to define __STDC_LIMIT_MACROS before including stdint.h,
- * if you want the limit (max/min) macros for int types. 
+ * if you want the limit (max/min) macros for int types.
  */
 #ifndef __STDC_LIMIT_MACROS
 #define __STDC_LIMIT_MACROS 1
@@ -70,7 +70,7 @@ typedef uint32_t flex_uint32_t;
 typedef signed char flex_int8_t;
 typedef short int flex_int16_t;
 typedef int flex_int32_t;
-typedef unsigned char flex_uint8_t; 
+typedef unsigned char flex_uint8_t;
 typedef unsigned short int flex_uint16_t;
 typedef unsigned int flex_uint32_t;
 
@@ -196,7 +196,7 @@ extern FILE *xbt_automaton_parser_in, *xbt_automaton_parser_out;
 
     #define YY_LESS_LINENO(n)
     #define YY_LINENO_REWIND_TO(ptr)
-    
+
 /* Return all but the first "n" matched characters back to the input stream. */
 #define yyless(n) \
   do \
@@ -253,7 +253,7 @@ struct yy_buffer_state
 
     int yy_bs_lineno; /**< The line count. */
     int yy_bs_column; /**< The column count. */
-    
+
   /* Whether to try to fill the input buffer when we reach the
    * end of it.
    */
@@ -541,9 +541,9 @@ char *xbt_automaton_parser_text;
 
 #include <stdio.h>
 #include "parserPromela.tab.hacc"
-  
+
   extern YYSTYPE yylval;
- 
+
 #line 548 "automaton_lexer.yy.c"
 
 #define INITIAL 0
@@ -604,9 +604,9 @@ extern int xbt_automaton_parser_wrap (void );
 #endif
 
 #ifndef YY_NO_UNPUT
-    
+
     static void yyunput (int c,char *buf_ptr  );
-    
+
 #endif
 
 #ifndef yytext_ptr
@@ -734,7 +734,7 @@ YY_DECL
   yy_state_type yy_current_state;
   char *yy_cp, *yy_bp;
   int yy_act;
-    
+
   if ( !(yy_init) )
     {
     (yy_init) = 1;
@@ -916,13 +916,13 @@ YY_RULE_SETUP
 case 19:
 YY_RULE_SETUP
 #line 63 "parserPromela.lex"
-{ sscanf(xbt_automaton_parser_text,"%lf",&yylval.real); 
+{ sscanf(xbt_automaton_parser_text,"%lf",&yylval.real);
                             return (LITT_REEL); }
   YY_BREAK
 case 20:
 YY_RULE_SETUP
 #line 66 "parserPromela.lex"
-{ sscanf(xbt_automaton_parser_text,"%d",&yylval.integer); 
+{ sscanf(xbt_automaton_parser_text,"%d",&yylval.integer);
                             return (LITT_ENT); }
   YY_BREAK
 case 21:
@@ -930,7 +930,7 @@ case 21:
 YY_RULE_SETUP
 #line 69 "parserPromela.lex"
 { yylval.string=(char *)malloc(strlen(xbt_automaton_parser_text)+1);
-                            sscanf(xbt_automaton_parser_text,"%s",yylval.string); 
+                            sscanf(xbt_automaton_parser_text,"%s",yylval.string);
                             return (LITT_CHAINE); }
   YY_BREAK
 case 22:
@@ -1235,7 +1235,7 @@ static int yy_get_next_buffer (void)
 {
   yy_state_type yy_current_state;
   char *yy_cp;
-    
+
   yy_current_state = (yy_start);
 
   for ( yy_cp = (yytext_ptr) + YY_MORE_ADJ; yy_cp < (yy_c_buf_p); ++yy_cp )
@@ -1291,7 +1291,7 @@ static int yy_get_next_buffer (void)
     static void yyunput (int c, char * yy_bp )
 {
   char *yy_cp;
-    
+
     yy_cp = (yy_c_buf_p);
 
   /* undo effects of setting up xbt_automaton_parser_text */
@@ -1336,7 +1336,7 @@ static int yy_get_next_buffer (void)
 
 {
   int c;
-    
+
   *(yy_c_buf_p) = (yy_hold_char);
 
   if ( *(yy_c_buf_p) == YY_END_OF_BUFFER_CHAR )
@@ -1403,12 +1403,12 @@ static int yy_get_next_buffer (void)
 
 /** Immediately switch to a different input stream.
  * @param input_file A readable stream.
- * 
+ *
  * @note This function does not reset the start condition to @c INITIAL .
  */
     void xbt_automaton_parser_restart  (FILE * input_file )
 {
-    
+
   if ( ! YY_CURRENT_BUFFER ){
         xbt_automaton_parser_ensure_buffer_stack ();
     YY_CURRENT_BUFFER_LVALUE =
@@ -1421,11 +1421,11 @@ static int yy_get_next_buffer (void)
 
 /** Switch to a different input buffer.
  * @param new_buffer The new input buffer.
- * 
+ *
  */
     void xbt_automaton_parser__switch_to_buffer  (YY_BUFFER_STATE  new_buffer )
 {
-    
+
   /* TODO. We should be able to replace this entire function body
    * with
    *    xbt_automaton_parser_pop_buffer_state();
@@ -1465,13 +1465,13 @@ static void xbt_automaton_parser__load_buffer_state  (void)
 /** Allocate and initialize an input buffer state.
  * @param file A readable stream.
  * @param size The character buffer size in bytes. When in doubt, use @c YY_BUF_SIZE.
- * 
+ *
  * @return the allocated buffer state.
  */
     YY_BUFFER_STATE xbt_automaton_parser__create_buffer  (FILE * file, int  size )
 {
   YY_BUFFER_STATE b;
-    
+
   b = (YY_BUFFER_STATE) xbt_automaton_parser_alloc(sizeof( struct yy_buffer_state )  );
   if ( ! b )
     YY_FATAL_ERROR( "out of dynamic memory in xbt_automaton_parser__create_buffer()" );
@@ -1494,11 +1494,11 @@ static void xbt_automaton_parser__load_buffer_state  (void)
 
 /** Destroy the buffer.
  * @param b a buffer created with xbt_automaton_parser__create_buffer()
- * 
+ *
  */
     void xbt_automaton_parser__delete_buffer (YY_BUFFER_STATE  b )
 {
-    
+
   if ( ! b )
     return;
 
@@ -1519,7 +1519,7 @@ static void xbt_automaton_parser__load_buffer_state  (void)
 
 {
   int oerrno = errno;
-    
+
   xbt_automaton_parser__flush_buffer(b );
 
   b->yy_input_file = file;
@@ -1535,13 +1535,13 @@ static void xbt_automaton_parser__load_buffer_state  (void)
     }
 
         b->yy_is_interactive = file ? (isatty( fileno(file) ) > 0) : 0;
-    
+
   errno = oerrno;
 }
 
 /** Discard all buffered characters. On the next scan, YY_INPUT will be called.
  * @param b the buffer state to be flushed, usually @c YY_CURRENT_BUFFER.
- * 
+ *
  */
     void xbt_automaton_parser__flush_buffer (YY_BUFFER_STATE  b )
 {
@@ -1570,7 +1570,7 @@ static void xbt_automaton_parser__load_buffer_state  (void)
  *  the current state. This function will allocate the stack
  *  if necessary.
  *  @param new_buffer The new state.
- *  
+ *
  */
 void xbt_automaton_parser_push_buffer_state (YY_BUFFER_STATE new_buffer )
 {
@@ -1600,7 +1600,7 @@ void xbt_automaton_parser_push_buffer_state (YY_BUFFER_STATE new_buffer )
 
 /** Removes and deletes the top of the stack, if present.
  *  The next element becomes the new top.
- *  
+ *
  */
 void xbt_automaton_parser_pop_buffer_state (void)
 {
@@ -1624,7 +1624,7 @@ void xbt_automaton_parser_pop_buffer_state (void)
 static void xbt_automaton_parser_ensure_buffer_stack (void)
 {
   yy_size_t num_to_alloc;
-    
+
   if (!(yy_buffer_stack)) {
 
     /* First allocation is just for 2 elements, since we don't know if this
@@ -1637,9 +1637,9 @@ static void xbt_automaton_parser_ensure_buffer_stack (void)
                 );
     if ( ! (yy_buffer_stack) )
       YY_FATAL_ERROR( "out of dynamic memory in xbt_automaton_parser_ensure_buffer_stack()" );
-                  
+
     memset((yy_buffer_stack), 0, num_to_alloc * sizeof(struct yy_buffer_state*));
-        
+
     (yy_buffer_stack_max) = num_to_alloc;
     (yy_buffer_stack_top) = 0;
     return;
@@ -1667,13 +1667,13 @@ static void xbt_automaton_parser_ensure_buffer_stack (void)
 /** Setup the input buffer state to scan directly from a user-specified character buffer.
  * @param base the character buffer
  * @param size the size in bytes of the character buffer
- * 
- * @return the newly allocated buffer state object. 
+ *
+ * @return the newly allocated buffer state object.
  */
 YY_BUFFER_STATE xbt_automaton_parser__scan_buffer  (char * base, yy_size_t  size )
 {
   YY_BUFFER_STATE b;
-    
+
   if ( size < 2 ||
        base[size-2] != YY_END_OF_BUFFER_CHAR ||
        base[size-1] != YY_END_OF_BUFFER_CHAR )
@@ -1702,14 +1702,14 @@ YY_BUFFER_STATE xbt_automaton_parser__scan_buffer  (char * base, yy_size_t  size
 /** Setup the input buffer state to scan a string. The next call to xbt_automaton_parser_lex() will
  * scan from a @e copy of @a str.
  * @param yystr a NUL-terminated string to scan
- * 
+ *
  * @return the newly allocated buffer state object.
  * @note If you want to scan bytes that may contain NUL values, then use
  *       xbt_automaton_parser__scan_bytes() instead.
  */
 YY_BUFFER_STATE xbt_automaton_parser__scan_string (yyconst char * yystr )
 {
-    
+
   return xbt_automaton_parser__scan_bytes(yystr,strlen(yystr) );
 }
 
@@ -1717,7 +1717,7 @@ YY_BUFFER_STATE xbt_automaton_parser__scan_string (yyconst char * yystr )
  * scan from a @e copy of @a bytes.
  * @param yybytes the byte buffer to scan
  * @param _yybytes_len the number of bytes in the buffer pointed to by @a bytes.
- * 
+ *
  * @return the newly allocated buffer state object.
  */
 YY_BUFFER_STATE xbt_automaton_parser__scan_bytes  (yyconst char * yybytes, yy_size_t  _yybytes_len )
@@ -1726,7 +1726,7 @@ YY_BUFFER_STATE xbt_automaton_parser__scan_bytes  (yyconst char * yybytes, yy_si
   char *buf;
   yy_size_t n;
   yy_size_t i;
-    
+
   /* Get memory for full buffer, including space for trailing EOB's. */
   n = _yybytes_len + 2;
   buf = (char *) xbt_automaton_parser_alloc(n  );
@@ -1780,16 +1780,16 @@ static void yy_fatal_error (yyconst char* msg )
 /* Accessor  methods (get/set functions) to struct members. */
 
 /** Get the current line number.
- * 
+ *
  */
 int xbt_automaton_parser_get_lineno  (void)
 {
-        
+
     return xbt_automaton_parser_lineno;
 }
 
 /** Get the input stream.
- * 
+ *
  */
 FILE *xbt_automaton_parser_get_in  (void)
 {
@@ -1797,7 +1797,7 @@ FILE *xbt_automaton_parser_get_in  (void)
 }
 
 /** Get the output stream.
- * 
+ *
  */
 FILE *xbt_automaton_parser_get_out  (void)
 {
@@ -1805,7 +1805,7 @@ FILE *xbt_automaton_parser_get_out  (void)
 }
 
 /** Get the length of the current token.
- * 
+ *
  */
 yy_size_t xbt_automaton_parser_get_leng  (void)
 {
@@ -1813,7 +1813,7 @@ yy_size_t xbt_automaton_parser_get_leng  (void)
 }
 
 /** Get the current token.
- * 
+ *
  */
 
 char *xbt_automaton_parser_get_text  (void)
@@ -1823,18 +1823,18 @@ char *xbt_automaton_parser_get_text  (void)
 
 /** Set the current line number.
  * @param _line_number line number
- * 
+ *
  */
 void xbt_automaton_parser_set_lineno (int  _line_number )
 {
-    
+
     xbt_automaton_parser_lineno = _line_number;
 }
 
 /** Set the input stream. This does not discard the current
  * input buffer.
  * @param _in_str A readable stream.
- * 
+ *
  * @see xbt_automaton_parser__switch_to_buffer
  */
 void xbt_automaton_parser_set_in (FILE *  _in_str )
@@ -1888,7 +1888,7 @@ static int yy_init_globals (void)
 /* xbt_automaton_parser_lex_destroy is for both reentrant and non-reentrant scanners. */
 int xbt_automaton_parser_lex_destroy  (void)
 {
-    
+
     /* Pop the buffer stack, destroying each element. */
   while(YY_CURRENT_BUFFER){
     xbt_automaton_parser__delete_buffer(YY_CURRENT_BUFFER  );
@@ -1914,7 +1914,7 @@ int xbt_automaton_parser_lex_destroy  (void)
 #ifndef yytext_ptr
 static void yy_flex_strncpy (char* s1, yyconst char * s2, int n )
 {
-    
+
   int i;
   for ( i = 0; i < n; ++i )
     s1[i] = s2[i];
@@ -1939,7 +1939,7 @@ void *xbt_automaton_parser_alloc (yy_size_t  size )
 
 void *xbt_automaton_parser_realloc  (void * ptr, yy_size_t  size )
 {
-    
+
   /* The cast to (char *) in the following accommodates both
    * implementations that use char* generic pointers, and those
    * that use void* generic pointers.  It works with the latter
diff --git a/src/xbt/automaton/automatonparse_promela.c b/src/xbt/automaton/automatonparse_promela.c
index 6364fe32e3..047079c1ce 100644
--- a/src/xbt/automaton/automatonparse_promela.c
+++ b/src/xbt/automaton/automatonparse_promela.c
@@ -54,7 +54,7 @@ static void new_transition(char* id, xbt_automaton_exp_label_t label)
   new_state(id, 0);
   xbt_automaton_state_t state_dst = xbt_automaton_state_exists(parsed_automaton, id);
   xbt_automaton_state_t state_src = xbt_automaton_state_exists(parsed_automaton, state_id_src);
-  
+
   //xbt_transition_t trans = NULL;
   xbt_automaton_transition_new(parsed_automaton, state_src, state_dst, label);
 
diff --git a/src/xbt/backtrace.cpp b/src/xbt/backtrace.cpp
index fa82b4d617..022ae52a55 100644
--- a/src/xbt/backtrace.cpp
+++ b/src/xbt/backtrace.cpp
@@ -1,10 +1,11 @@
-/* Copyright (c) 2005-2016. The SimGrid Team. All rights reserved.          */
+/* Copyright (c) 2005-2017. The SimGrid Team. All rights reserved.          */
 
 /* This program is free software; you can redistribute it and/or modify it
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
 #include <cstddef>
 #include <cstdlib>
+#include <cstring>
 
 #include <vector>
 
@@ -27,6 +28,11 @@ XBT_LOG_NEW_DEFAULT_SUBCATEGORY(xbt_backtrace, xbt, "Backtrace");
 
 }
 
+static bool startWith(std::string str, const char* prefix)
+{
+  return strncmp(str.c_str(), prefix, strlen(prefix)) == 0;
+}
+
 void xbt_backtrace_display(xbt_backtrace_location_t* loc, std::size_t count)
 {
 #ifdef HAVE_BACKTRACE
@@ -36,8 +42,15 @@ void xbt_backtrace_display(xbt_backtrace_location_t* loc, std::size_t count)
     return;
   }
   fprintf(stderr, "Backtrace (displayed in process %s):\n", SIMIX_process_self_get_name());
-  for (std::string const& s : backtrace)
-    fprintf(stderr, "---> %s\n", s.c_str());
+  for (std::string const& s : backtrace) {
+    if (startWith(s, "xbt_backtrace_display_current"))
+      continue;
+
+    std::fprintf(stderr, "---> '%s'\n", s.c_str());
+    if (startWith(s, "SIMIX_simcall_handle") ||
+        startWith(s, "simgrid::xbt::MainFunction") /* main used with thread factory */)
+      break;
+  }
 #else
   XBT_ERROR("Cannot display backtrace when compiled without libunwind.");
 #endif
diff --git a/src/xbt/dynar.cpp b/src/xbt/dynar.cpp
index 47300c393d..b1a20fd41f 100644
--- a/src/xbt/dynar.cpp
+++ b/src/xbt/dynar.cpp
@@ -454,7 +454,7 @@ extern "C" unsigned int xbt_dynar_search(xbt_dynar_t const dynar, void* const el
  *
  * Beware that if your dynar contains pointed values (such as strings) instead of scalar, this function is probably not
  * what you want. Check the documentation of xbt_dynar_search() for more info.
- * 
+ *
  * Note that usually, the dynar indices are unsigned integers. If you have more than 2 million elements in your dynar,
  * this very function will not work (but the other will).
  */
@@ -470,7 +470,7 @@ extern "C" signed int xbt_dynar_search_or_negative(xbt_dynar_t const dynar, void
   return -1;
 }
 
-/** @brief Returns a boolean indicating whether the element is part of the dynar 
+/** @brief Returns a boolean indicating whether the element is part of the dynar
  *
  * Beware that if your dynar contains pointed values (such as strings) instead of scalar, this function is probably not
  * what you want. Check the documentation of xbt_dynar_search() for more info.
@@ -664,7 +664,7 @@ extern "C" void xbt_dynar_three_way_partition(xbt_dynar_t const dynar, int_f_pvo
   }
 }
 
-/** @brief Transform a dynar into a nullptr terminated array. 
+/** @brief Transform a dynar into a nullptr terminated array.
  *
  *  \param dynar the dynar to transform
  *  \return pointer to the first element of the array
diff --git a/src/xbt/lib.c b/src/xbt/lib.c
deleted file mode 100644
index 068652e41a..0000000000
--- a/src/xbt/lib.c
+++ /dev/null
@@ -1,125 +0,0 @@
-/* lib - a generic library, variation over dictionary                    */
-
-/* Copyright (c) 2011, 2013-2015. The SimGrid Team.
- * All rights reserved.                                                     */
-
-/* This program is free software; you can redistribute it and/or modify it
- * under the terms of the license (GNU LGPL) which comes with this package. */
-
-#include <xbt/asserts.h>
-#include <xbt/sysdep.h>
-#include <xbt/log.h>
-#include <xbt/lib.h>
-
-XBT_LOG_NEW_DEFAULT_SUBCATEGORY(xbt_lib, xbt, "A dict with keys of type (name, level)");
-
-xbt_lib_t xbt_lib_new(void)
-{
-  xbt_lib_t lib  = xbt_new(s_xbt_lib_t, 1);
-  lib->dict = xbt_dict_new_homogeneous(xbt_free_f);
-  lib->levels = 0;
-  lib->free_f = NULL;
-  return lib;
-}
-
-void xbt_lib_free(xbt_lib_t *plib)
-{
-  xbt_lib_t lib = *plib;
-  if (lib) {
-    xbt_dict_cursor_t cursor;
-    char *key;
-    void **elts;
-    xbt_dict_foreach (lib->dict, cursor, key, elts) {
-      int i;
-      for (i = 0 ; i < lib->levels ; i++)
-        if (elts[i] && lib->free_f[i])
-          lib->free_f[i](elts[i]);
-    }
-    xbt_dict_free(&lib->dict);
-    xbt_free(lib->free_f);
-    xbt_free(lib);
-    *plib = NULL;
-  }
-}
-
-int xbt_lib_add_level(xbt_lib_t lib, void_f_pvoid_t free_f)
-{
-  XBT_DEBUG("xbt_lib_add_level");
-  xbt_assert(xbt_dict_is_empty(lib->dict), "Lib is not empty (size: %u), cannot add a level", xbt_dict_size(lib->dict));
-  lib->free_f = xbt_realloc(lib->free_f, sizeof(void_f_pvoid_t) * (lib->levels + 1));
-  lib->free_f[lib->levels] = free_f;
-  return lib->levels++;
-}
-
-void xbt_lib_set(xbt_lib_t lib, const char *key, int level, void *obj)
-{
-  XBT_DEBUG("xbt_lib_set key '%s:%d' with object %p", key, level, obj);
-  xbt_assert(level >= 0);
-  void **elts = xbt_dict_get_or_null(lib->dict, key);
-  if (!elts) {
-    elts = xbt_new0(void *, lib->levels);
-    xbt_dict_set(lib->dict, key, elts, NULL);
-  }
-  if (elts[level]) {
-    XBT_DEBUG("Replace %p by %p element under key '%s:%d'", elts[level], obj, key, level);
-    if (lib->free_f[level])
-      lib->free_f[level](elts[level]);
-  }
-  elts[level] = obj;
-}
-
-void xbt_lib_unset(xbt_lib_t lib, const char *key, int level, int invoke_callback)
-{
-  xbt_assert(level >= 0);
-  void **elts = xbt_dict_get_or_null(lib->dict, key);
-  if (!elts) {
-     XBT_WARN("no key %s", key);
-     return;
-  }
-
-  void *obj = elts[level];
-  if (!obj) {
-     XBT_WARN("no key %s at level %d", key, level);
-     return;
-  }
-
-  XBT_DEBUG("Remove %p of key %s at level %d", obj, key, level);
-  elts[level] = NULL;
-
-  /* check if there still remains any elements of this key */
-  int empty = 1;
-  for (int i = 0; i < lib->levels && empty; i++) {
-     if (elts[i] != NULL)
-       empty = 0;
-  }
-  if (empty) {
-    /* there is no element at any level, so delete the key */
-    xbt_dict_remove(lib->dict, key);
-  }
-
-  if (invoke_callback && lib->free_f[level])
-    lib->free_f[level](obj);
-}
-
-void *xbt_lib_get_or_null(xbt_lib_t lib, const char *key, int level)
-{
-  xbt_assert(level >= 0);
-  void **elts = xbt_dict_get_or_null(lib->dict, key);
-  return elts ? elts[level] : NULL;
-}
-
-xbt_dictelm_t xbt_lib_get_elm_or_null(xbt_lib_t lib, const char *key)
-{
-  return xbt_dict_get_elm_or_null(lib->dict, key);
-}
-
-void *xbt_lib_get_level(xbt_dictelm_t elm, int level)
-{
-  xbt_assert(level >= 0);
-  void **elts = elm->content;
-  return elts ? elts[level] : NULL;
-}
-
-void xbt_lib_remove(xbt_lib_t lib, const char *key){
-  xbt_dict_remove(lib->dict, key);
-}
diff --git a/src/xbt/log.c b/src/xbt/log.c
index a7379c56cd..a2362dc736 100644
--- a/src/xbt/log.c
+++ b/src/xbt/log.c
@@ -112,7 +112,6 @@ static void xbt_log_connect_categories(void)
   XBT_LOG_CONNECT(xbt_exception);
   XBT_LOG_CONNECT(xbt_graph);
   XBT_LOG_CONNECT(xbt_heap);
-  XBT_LOG_CONNECT(xbt_lib);
   XBT_LOG_CONNECT(xbt_mallocator);
   XBT_LOG_CONNECT(xbt_memory_map);
   XBT_LOG_CONNECT(xbt_parmap);
@@ -200,7 +199,7 @@ static void xbt_log_connect_categories(void)
   XBT_LOG_CONNECT(s4u_channel);
   XBT_LOG_CONNECT(s4u_comm);
   XBT_LOG_CONNECT(s4u_file);
-   
+
   /* sg */
   XBT_LOG_CONNECT(sg_host);
 
@@ -260,7 +259,7 @@ static void xbt_log_connect_categories(void)
   XBT_LOG_CONNECT(surf_trace);
   XBT_LOG_CONNECT(surf_vm);
   XBT_LOG_CONNECT(surf_host);
-   
+
 #endif /* simgrid_EXPORTS */
 }
 
diff --git a/src/xbt/mmalloc/mm_module.c b/src/xbt/mmalloc/mm_module.c
index 096941c76a..eac80c81e8 100644
--- a/src/xbt/mmalloc/mm_module.c
+++ b/src/xbt/mmalloc/mm_module.c
@@ -48,7 +48,7 @@
    If FD is a valid file descriptor for an open file then data for the
    mmalloc managed region is mapped to that file, otherwise an anonymous
    map is used if supported by the underlying OS. In case of running in
-   an OS without support of anonymous mappings then "/dev/zero" is used 
+   an OS without support of anonymous mappings then "/dev/zero" is used
    and in both cases the data will not exist in any filesystem object.
 
    If the open file corresponding to FD is from a previous use of
@@ -58,8 +58,8 @@
    the current process as the process that created the file (ignoring
    the BASEADDR parameter).
 
-   For non valid FDs or empty files ones the mapping is established 
-   starting at the specified address BASEADDR in the process address 
+   For non valid FDs or empty files ones the mapping is established
+   starting at the specified address BASEADDR in the process address
    space.
 
    The provided BASEADDR should be choosed carefully in order to avoid
@@ -176,7 +176,7 @@ xbt_mheap_t xbt_mheap_new_options(int fd, void *baseaddr, int options)
   mdp->next_mdesc = NULL;
   mdp->refcount = 1;
   mdp->options = options;
-  
+
   /* If we have not been passed a valid open file descriptor for the file
      to map to, then we go for an anonymous map */
 
@@ -199,7 +199,7 @@ xbt_mheap_t xbt_mheap_new_options(int fd, void *baseaddr, int options)
     abort();
   }
 
-  /* Add the new heap to the linked list of heaps attached by mmalloc */  
+  /* Add the new heap to the linked list of heaps attached by mmalloc */
   if(__mmalloc_default_mdp){
     mdp = __mmalloc_default_mdp;
     while(mdp->next_mdesc)
@@ -388,7 +388,7 @@ ssize_t mmalloc_get_busy_size(xbt_mheap_t heap, void *ptr){
     ssize_t frag = ((uintptr_t) (ADDR2UINT (ptr) % (BLOCKSIZE))) >> heap->heapinfo[block].type;
     return heap->heapinfo[block].busy_frag.frag_size[frag];
   }
-    
+
 }
 
 void mmcheck(xbt_mheap_t heap) {return;
diff --git a/src/xbt/mmalloc/mmprivate.h b/src/xbt/mmalloc/mmprivate.h
index ce97df49df..93f925ab47 100644
--- a/src/xbt/mmalloc/mmprivate.h
+++ b/src/xbt/mmalloc/mmprivate.h
@@ -148,7 +148,7 @@ typedef struct {
   int type; /*  0: busy large block
                 >0: busy fragmented (fragments of size 2^type bytes)
                 <0: free block */
-  
+
   union {
     /* Heap information for a busy block.  */
     struct {
diff --git a/src/xbt/parmap.cpp b/src/xbt/parmap.cpp
index a0689f0c86..4afc9f73d4 100644
--- a/src/xbt/parmap.cpp
+++ b/src/xbt/parmap.cpp
@@ -117,7 +117,7 @@ xbt_parmap_t xbt_parmap_new(unsigned int num_workers, e_xbt_parmap_mode_t mode)
   parmap->workers[0] = nullptr;
 #if HAVE_PTHREAD_SETAFFINITY
   int core_bind = 0;
-#endif  
+#endif
   for (unsigned int i = 1; i < num_workers; i++) {
     xbt_parmap_thread_data_t data = xbt_new0(s_xbt_parmap_thread_data_t, 1);
     data->parmap = parmap;
@@ -128,7 +128,7 @@ xbt_parmap_t xbt_parmap_new(unsigned int num_workers, e_xbt_parmap_mode_t mode)
     if (core_bind != xbt_os_get_numcores() - 1)
       core_bind++;
     else
-      core_bind = 0; 
+      core_bind = 0;
 #endif
   }
   return parmap;
diff --git a/src/xbt/xbt_log_appender_file.c b/src/xbt/xbt_log_appender_file.c
index 93887340cf..74232a6351 100644
--- a/src/xbt/xbt_log_appender_file.c
+++ b/src/xbt/xbt_log_appender_file.c
@@ -130,7 +130,7 @@ xbt_log_appender_t xbt_log_appender2_file_new(char *arg,int roll) {
     data->count=-1;
   else
     data->count=0;
-  open_append2_file(data);  
+  open_append2_file(data);
   res->data = data;
   return res;
 }
diff --git a/src/xbt/xbt_main.cpp b/src/xbt/xbt_main.cpp
index d53d973cf2..f4ba9ab2c8 100644
--- a/src/xbt/xbt_main.cpp
+++ b/src/xbt/xbt_main.cpp
@@ -135,7 +135,7 @@ void xbt_init(int *argc, char **argv)
   xbt_cmdline     = xbt_dynar_new(sizeof(char*), NULL);
   for (int i = 0; i < *argc; i++)
     xbt_dynar_push(xbt_cmdline,&(argv[i]));
-  
+
   xbt_log_init(argc, argv);
 }
 
diff --git a/src/xbt/xbt_os_thread.c b/src/xbt/xbt_os_thread.c
index b83e310022..af1dc12778 100644
--- a/src/xbt/xbt_os_thread.c
+++ b/src/xbt/xbt_os_thread.c
@@ -92,7 +92,7 @@ void xbt_os_thread_mod_preinit(void)
 
   int errcode = pthread_key_create(&xbt_self_thread_key, NULL);
   xbt_assert(errcode == 0, "pthread_key_create failed for xbt_self_thread_key");
-  
+
   main_thread = xbt_new(s_xbt_os_thread_t, 1);
   main_thread->name = NULL;
   main_thread->detached = 0;
@@ -165,7 +165,7 @@ xbt_os_thread_t xbt_os_thread_create(const char *name,  pvoid_f_pvoid_t start_ro
   res_thread->start_routine = start_routine;
   res_thread->param = param;
   res_thread->extra_data = extra_data;
-  
+
   int errcode = pthread_create(&(res_thread->t), &thread_attr, wrapper_start_routine, res_thread);
   xbt_assert(errcode == 0, "pthread_create failed: %s", strerror(errcode));
 
diff --git a/teshsuite/msg/task_destroy_cancel/task_destroy_cancel.cpp b/teshsuite/msg/task_destroy_cancel/task_destroy_cancel.cpp
index 2c686cc4b4..5c22018489 100644
--- a/teshsuite/msg/task_destroy_cancel/task_destroy_cancel.cpp
+++ b/teshsuite/msg/task_destroy_cancel/task_destroy_cancel.cpp
@@ -9,7 +9,7 @@
 
 XBT_LOG_NEW_DEFAULT_CATEGORY(msg_test, "Messages specific for this msg example");
 
-static int master(int argc, char *argv[])
+static int master(int /*argc*/, char* /*argv*/ [])
 {
   double task_comp_size = 5E7;
   double task_comm_size = 1E6;
@@ -61,7 +61,7 @@ static int master(int argc, char *argv[])
   return 0;
 }
 
-static int worker_main(int argc, char *argv[])
+static int worker_main(int /*argc*/, char* /*argv*/ [])
 {
   msg_task_t task = (msg_task_t) MSG_process_get_data(MSG_process_self());
   msg_error_t res;
@@ -72,7 +72,7 @@ static int worker_main(int argc, char *argv[])
   return 0;
 }
 
-static int worker(int argc, char *argv[])
+static int worker(int /*argc*/, char* /*argv*/ [])
 {
   while (1) {
     msg_task_t task = NULL;
diff --git a/teshsuite/s4u/CMakeLists.txt b/teshsuite/s4u/CMakeLists.txt
index b3ed7dad87..27538f6e24 100644
--- a/teshsuite/s4u/CMakeLists.txt
+++ b/teshsuite/s4u/CMakeLists.txt
@@ -1,16 +1,29 @@
-foreach(x actor concurrent_rw host_on_off_wait listen_async pid storage_client_server)
+foreach(x actor comm-pt2pt comm-waitany concurrent_rw host_on_off_wait listen_async pid storage_client_server)
   add_executable       (${x}  ${x}/${x}.cpp)
   target_link_libraries(${x}  simgrid)
   set_target_properties(${x}  PROPERTIES RUNTIME_OUTPUT_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/${x})
 
-  set(tesh_files    ${tesh_files}    ${CMAKE_CURRENT_SOURCE_DIR}/${x}/${x}.tesh)
   set(teshsuite_src ${teshsuite_src} ${CMAKE_CURRENT_SOURCE_DIR}/${x}/${x}.cpp)
 endforeach()
 
+## Add the tests.
+## Some need to be run with all factories, some need not tesh to run
+foreach(x actor concurrent_rw)
+  set(tesh_files    ${tesh_files}    ${CMAKE_CURRENT_SOURCE_DIR}/${x}/${x}.tesh)
+  ADD_TESH_FACTORIES(tesh-s4u-${x} "thread;boost;ucontext;raw" --setenv srcdir=${CMAKE_HOME_DIRECTORY}/teshsuite/s4u/${x} --cd ${CMAKE_BINARY_DIR}/teshsuite/s4u/${x} ${CMAKE_HOME_DIRECTORY}/teshsuite/s4u/${x}/${x}.tesh)
+endforeach()
+
+foreach(x host_on_off_wait listen_async pid storage_client_server)
+  set(tesh_files    ${tesh_files}    ${CMAKE_CURRENT_SOURCE_DIR}/${x}/${x}.tesh)
+  ADD_TESH(tesh-s4u-${x} --setenv srcdir=${CMAKE_HOME_DIRECTORY}/teshsuite/s4u/${x} --cd ${CMAKE_BINARY_DIR}/teshsuite/s4u/${x} ${CMAKE_HOME_DIRECTORY}/teshsuite/s4u/${x}/${x}.tesh)
+endforeach()
+
+# The output is not relevant
+ADD_TEST(tesh-s4u-comm-pt2pt   ${CMAKE_BINARY_DIR}/teshsuite/s4u/comm-pt2pt/comm-pt2pt     ${CMAKE_HOME_DIRECTORY}/examples/platforms/cluster.xml)
+ADD_TEST(tesh-s4u-comm-waitany ${CMAKE_BINARY_DIR}/teshsuite/s4u/comm-waitany/comm-waitany ${CMAKE_HOME_DIRECTORY}/examples/platforms/two_hosts.xml)
+
+
+
 set(teshsuite_src ${teshsuite_src}  PARENT_SCOPE)
 set(tesh_files    ${tesh_files}     PARENT_SCOPE)
 set(xml_files     ${xml_files}      PARENT_SCOPE)
-
-foreach(x actor concurrent_rw host_on_off_wait listen_async pid storage_client_server)
-  ADD_TESH_FACTORIES(tesh-s4u-${x} "thread;boost;ucontext;raw" --setenv srcdir=${CMAKE_HOME_DIRECTORY}/teshsuite/s4u/${x} --cd ${CMAKE_BINARY_DIR}/teshsuite/s4u/${x} ${CMAKE_HOME_DIRECTORY}/teshsuite/s4u/${x}/${x}.tesh)
-endforeach()
diff --git a/teshsuite/s4u/comm-pt2pt/comm-pt2pt.cpp b/teshsuite/s4u/comm-pt2pt/comm-pt2pt.cpp
new file mode 100644
index 0000000000..0eebb04af4
--- /dev/null
+++ b/teshsuite/s4u/comm-pt2pt/comm-pt2pt.cpp
@@ -0,0 +1,195 @@
+/* Copyright (c) 2010-2017. The SimGrid Team. All rights reserved.          */
+
+/* This program is free software; you can redistribute it and/or modify it
+ * under the terms of the license (GNU LGPL) which comes with this package. */
+
+/* We want this test to be exhaustive in term of:
+ *   - communication involved (regular, asynchronous, detached, with a permanent receiver declared)
+ *   - whether the send or the receive was posted first
+ *
+ * FIXME: Missing elements: timeouts, host/actor failures, link failures
+ */
+
+#include "simgrid/s4u.hpp"
+
+#include <cstring>
+
+XBT_LOG_NEW_DEFAULT_CATEGORY(s4u_test, "Messages specific for this s4u example");
+
+using namespace simgrid::s4u;
+
+static void usage(const char* binaryName, const char* defaultSend, const char* defaultRecv)
+{
+  std::fprintf(stderr, "Usage: %s examples/platforms/cluster.xml <send_spec> <recv_spec>\n"
+                       "where spec is a list of letters giving the kind of tests you want to see.\n"
+                       "Existing sender spec:\n"
+                       " r regular send\n"
+                       " R regular send (after a little delay)\n"
+                       " i asynchronous isend\n"
+                       " I asynchronous isend (after a little delay)\n"
+                       " d detached send\n"
+                       " D detached send (after a little delay)\n"
+                       "Existing receiver spec:\n"
+                       " r regular receive\n"
+                       " R regular receive (after a little delay)\n"
+                       " i asynchronous irecv\n"
+                       " I asynchronous irecv (after a little delay)\n"
+                       " p regular receive on permanent mailbox (after a little delay)\n"
+                       " P regular receive on permanent mailbox (after a little delay)\n"
+                       " j irecv on permanent mailbox (after a little delay)\n"
+                       " J irecv on permanent mailbox (after a little delay)\n"
+                       "\n"
+                       "Example 1: %s examples/platforms/cluster.xml rRiIdD rrrrrr # testing all send functions\n"
+                       "Default specs: %s %s (all possible pairs)\n",
+               binaryName, binaryName, defaultSend, defaultRecv);
+  exit(1);
+}
+
+static void sender(std::vector<std::string> args)
+{
+  XBT_INFO("Sender spec: %s", args[0].c_str());
+  for (unsigned int test = 1; test <= args[0].size(); test++) {
+    this_actor::sleep_until(test * 5 - 5);
+    char* mboxName                = bprintf("Test #%u", test);
+    simgrid::s4u::MailboxPtr mbox = simgrid::s4u::Mailbox::byName(mboxName);
+
+    switch (args[0][test - 1]) {
+      case 'r':
+        XBT_INFO("Test %d: r (regular send)", test);
+        simgrid::s4u::this_actor::send(mbox, (void*)mboxName, 42.0);
+        break;
+      case 'R':
+        XBT_INFO("Test %d: R (sleep + regular send)", test);
+        simgrid::s4u::this_actor::sleep_for(0.5);
+        simgrid::s4u::this_actor::send(mbox, (void*)mboxName, 42.0);
+        break;
+
+      case 'i':
+        XBT_INFO("Test %d: i (asynchronous isend)", test);
+        simgrid::s4u::this_actor::isend(mbox, (void*)mboxName, 42.0)->wait();
+        break;
+      case 'I':
+        XBT_INFO("Test %d: I (sleep + isend)", test);
+        simgrid::s4u::this_actor::sleep_for(0.5);
+        simgrid::s4u::this_actor::isend(mbox, (void*)mboxName, 42.0)->wait();
+        break;
+
+      case 'd':
+        XBT_INFO("Test %d: d (detached send)", test);
+        simgrid::s4u::this_actor::dsend(mbox, (void*)mboxName, 42.0);
+        break;
+      case 'D':
+        XBT_INFO("Test %d: D (sleep + detached send)", test);
+        simgrid::s4u::this_actor::sleep_for(0.5);
+        simgrid::s4u::this_actor::dsend(mbox, (void*)mboxName, 42.0);
+        break;
+      default:
+        xbt_die("Unknown sender spec for test %d: '%c'", test, args[0][test - 1]);
+    }
+    XBT_INFO("Test %d OK", test);
+  }
+  simgrid::s4u::this_actor::sleep_for(0.5);
+  // FIXME: we should test what happens when the process ends before the end of remote comm instead of hiding it
+}
+
+static void receiver(std::vector<std::string> args)
+{
+  XBT_INFO("Receiver spec: %s", args[0].c_str());
+  for (unsigned int test = 1; test <= args[0].size(); test++) {
+    this_actor::sleep_until(test * 5 - 5);
+    char* mboxName                = bprintf("Test #%u", test);
+    simgrid::s4u::MailboxPtr mbox = simgrid::s4u::Mailbox::byName(mboxName);
+    void* received                = nullptr;
+
+    switch (args[0][test - 1]) {
+      case 'r':
+        XBT_INFO("Test %d: r (regular receive)", test);
+        received = simgrid::s4u::this_actor::recv(mbox);
+        break;
+      case 'R':
+        XBT_INFO("Test %d: R (sleep + regular receive)", test);
+        simgrid::s4u::this_actor::sleep_for(0.5);
+        received = simgrid::s4u::this_actor::recv(mbox);
+        break;
+
+      case 'i':
+        XBT_INFO("Test %d: i (asynchronous irecv)", test);
+        simgrid::s4u::this_actor::irecv(mbox, &received)->wait();
+        break;
+      case 'I':
+        XBT_INFO("Test %d: I (sleep + asynchronous irecv)", test);
+        simgrid::s4u::this_actor::sleep_for(0.5);
+        simgrid::s4u::this_actor::irecv(mbox, &received)->wait();
+        break;
+      case 'p':
+        XBT_INFO("Test %d: p (regular receive on permanent mailbox)", test);
+        mbox->setReceiver(Actor::self());
+        received = simgrid::s4u::this_actor::recv(mbox);
+        break;
+      case 'P':
+        XBT_INFO("Test %d: P (sleep + regular receive on permanent mailbox)", test);
+        simgrid::s4u::this_actor::sleep_for(0.5);
+        mbox->setReceiver(Actor::self());
+        received = simgrid::s4u::this_actor::recv(mbox);
+        break;
+      case 'j':
+        XBT_INFO("Test %d: j (irecv on permanent mailbox)", test);
+        mbox->setReceiver(Actor::self());
+        simgrid::s4u::this_actor::irecv(mbox, &received)->wait();
+        break;
+      case 'J':
+        XBT_INFO("Test %d: J (sleep + irecv on permanent mailbox)", test);
+        simgrid::s4u::this_actor::sleep_for(0.5);
+        mbox->setReceiver(Actor::self());
+        simgrid::s4u::this_actor::irecv(mbox, &received)->wait();
+        break;
+      default:
+        xbt_die("Unknown receiver spec for test %d: '%c'", test, args[0][test - 1]);
+    }
+
+    xbt_assert(strcmp(static_cast<char*>(received), mboxName) == 0);
+    xbt_free(received);
+    xbt_free(mboxName);
+    XBT_INFO("Test %d OK", test);
+  }
+  simgrid::s4u::this_actor::sleep_for(0.5);
+}
+
+int main(int argc, char* argv[])
+{
+  std::string specSend;
+  std::string specRecv;
+  for (char s : {'r', 'R', 'i', 'I', 'd', 'D'})
+    for (char r : {'r', 'R', 'i', 'I', 'p', 'P', 'j', 'J'}) {
+      specSend += s;
+      specRecv += r;
+    }
+  std::vector<std::string> argSend{specSend.c_str()};
+  std::vector<std::string> argRecv{specRecv.c_str()};
+
+  simgrid::s4u::Engine* e = new simgrid::s4u::Engine(&argc, argv);
+  if (argc < 2)
+    usage(argv[0], specSend.c_str(), specRecv.c_str());
+
+  e->loadPlatform(argv[1]);
+
+  if (argc >= 3) {
+    argSend.clear();
+    argSend.push_back(argv[2]);
+  }
+  if (argc >= 4) {
+    argRecv.clear();
+    argRecv.push_back(argv[3]);
+  }
+  xbt_assert(argSend.front().size() == argRecv.front().size(), "Sender and receiver spec must be of the same size");
+
+  simgrid::s4u::Host** hosts = sg_host_list();
+  simgrid::s4u::Actor::createActor("sender", hosts[0], sender, argSend);
+  simgrid::s4u::Actor::createActor("recver", hosts[1], receiver, argRecv);
+  xbt_free(hosts);
+
+  e->run();
+  XBT_INFO("Simulation time %g", e->getClock());
+
+  return 0;
+}
diff --git a/teshsuite/s4u/comm-waitany/comm-waitany.cpp b/teshsuite/s4u/comm-waitany/comm-waitany.cpp
new file mode 100644
index 0000000000..24420571c1
--- /dev/null
+++ b/teshsuite/s4u/comm-waitany/comm-waitany.cpp
@@ -0,0 +1,75 @@
+#include <iostream>
+#include <simgrid/s4u.hpp>
+#include <stdlib.h>
+#include <vector>
+
+#define NUM_COMMS 1
+
+XBT_LOG_NEW_DEFAULT_CATEGORY(mwe, "Minimum Working Example");
+
+static void receiver()
+{
+  simgrid::s4u::MailboxPtr mymailbox    = simgrid::s4u::Mailbox::byName("receiver_mailbox");
+  simgrid::s4u::MailboxPtr theirmailbox = simgrid::s4u::Mailbox::byName("sender_mailbox");
+
+  std::vector<simgrid::s4u::CommPtr> pending_comms;
+
+  XBT_INFO("Placing %d asynchronous recv requests", NUM_COMMS);
+  void* data;
+  for (int i = 0; i < NUM_COMMS; i++) {
+    simgrid::s4u::CommPtr comm = simgrid::s4u::Comm::recv_async(mymailbox, &data);
+    pending_comms.push_back(comm);
+  }
+
+  for (int i = 0; i < NUM_COMMS; i++) {
+    XBT_INFO("Sleeping for 3 seconds (for the %dth time)...", i + 1);
+    simgrid::s4u::this_actor::sleep_for(3.0);
+    XBT_INFO("Calling wait_any() for %zu pending comms", pending_comms.size());
+    std::vector<simgrid::s4u::CommPtr>::iterator ret_it =
+        simgrid::s4u::Comm::wait_any(pending_comms.begin(), pending_comms.end());
+    XBT_INFO("Counting the number of completed comms...");
+
+    int count = 0;
+    for (; ret_it != pending_comms.end(); count++, ret_it++)
+      ;
+
+    XBT_INFO("wait_any() replied that %d comms have completed", count);
+    // xbt_assert(count == 1, "wait_any() replied that %d comms have completed, which is broken!", count);
+  }
+}
+
+static void sender()
+{
+  simgrid::s4u::MailboxPtr mymailbox    = simgrid::s4u::Mailbox::byName("sender_mailbox");
+  simgrid::s4u::MailboxPtr theirmailbox = simgrid::s4u::Mailbox::byName("receiver_mailbox");
+
+  void* data = (void*)"data";
+
+  for (int i = 0; i < NUM_COMMS; i++) {
+    XBT_INFO("Sending a message to the receiver");
+    simgrid::s4u::this_actor::send(theirmailbox, &data, 4);
+    XBT_INFO("Sleeping for 1000 seconds");
+    simgrid::s4u::this_actor::sleep_for(1000.0);
+  }
+}
+
+int main(int argc, char** argv)
+{
+
+  simgrid::s4u::Engine* engine = new simgrid::s4u::Engine(&argc, argv);
+
+  if (argc < 2) {
+    std::cerr << "Usage: " << argv[0] << " <xml platform file>" << std::endl;
+    exit(1);
+  }
+
+  engine->loadPlatform(argv[1]);
+  simgrid::s4u::Host* host = simgrid::s4u::Host::by_name("Tremblay");
+
+  simgrid::s4u::Actor::createActor("Receiver", host, receiver);
+  simgrid::s4u::Actor::createActor("Sender", host, sender);
+
+  simgrid::s4u::Engine::instance()->run();
+
+  return 0;
+}
diff --git a/teshsuite/s4u/storage_client_server/storage_client_server.cpp b/teshsuite/s4u/storage_client_server/storage_client_server.cpp
index 735879fe9f..b2fdc637b2 100644
--- a/teshsuite/s4u/storage_client_server/storage_client_server.cpp
+++ b/teshsuite/s4u/storage_client_server/storage_client_server.cpp
@@ -76,34 +76,32 @@ static void display_storage_content(simgrid::s4u::Storage* storage)
 static void dump_storage_by_name(char* name)
 {
   XBT_INFO("*** Dump a storage element ***");
-  simgrid::s4u::Storage& storage = simgrid::s4u::Storage::byName(name);
-  display_storage_content(&storage);
+  simgrid::s4u::Storage* storage = simgrid::s4u::Storage::byName(name);
+  display_storage_content(storage);
 }
 
 static void get_set_storage_data(const char* storage_name)
 {
   XBT_INFO("*** GET/SET DATA for storage element: %s ***", storage_name);
-  simgrid::s4u::Storage& storage = simgrid::s4u::Storage::byName(storage_name);
+  simgrid::s4u::Storage* storage = simgrid::s4u::Storage::byName(storage_name);
 
-  char* data = static_cast<char*>(storage.userdata());
+  char* data = static_cast<char*>(storage->userdata());
   XBT_INFO("Get data: '%s'", data);
-  storage.setUserdata(xbt_strdup("Some data"));
-  data = static_cast<char*>(storage.userdata());
+  storage->setUserdata(xbt_strdup("Some data"));
+  data = static_cast<char*>(storage->userdata());
   XBT_INFO("\tSet and get data: '%s'", data);
   xbt_free(data);
 }
 
 static void dump_platform_storages()
 {
-  std::unordered_map<std::string, simgrid::s4u::Storage*>* storages = simgrid::s4u::Storage().allStorages();
+  std::map<std::string, simgrid::s4u::Storage*>* storages = simgrid::s4u::allStorages();
 
   for (auto storage : *storages) {
     XBT_INFO("Storage %s is attached to %s", storage.first.c_str(), storage.second->host());
     storage.second->setProperty("other usage", xbt_strdup("gpfs"));
   }
-  // Expected output in tesh file that's missing for now
-  //> [  1.207952] (server@alice) Storage Disk3 is attached to carl
-  //> [  1.207952] (server@alice) Storage Disk4 is attached to denise
+  delete storages;
 }
 
 static void storage_info(simgrid::s4u::Host* host)
@@ -116,15 +114,15 @@ static void storage_info(simgrid::s4u::Host* host)
   xbt_dict_t storage_list = host->mountedStoragesAsDict();
   xbt_dict_foreach (storage_list, cursor, mount_name, storage_name) {
     XBT_INFO("\tStorage name: %s, mount name: %s", storage_name, mount_name);
-    simgrid::s4u::Storage& storage = simgrid::s4u::Storage::byName(storage_name);
+    simgrid::s4u::Storage* storage = simgrid::s4u::Storage::byName(storage_name);
 
-    sg_size_t free_size = storage.sizeFree();
-    sg_size_t used_size = storage.sizeUsed();
+    sg_size_t free_size = storage->sizeFree();
+    sg_size_t used_size = storage->sizeUsed();
 
     XBT_INFO("\t\tFree size: %llu bytes", free_size);
     XBT_INFO("\t\tUsed size: %llu bytes", used_size);
 
-    display_storage_properties(&storage);
+    display_storage_properties(storage);
     dump_storage_by_name(storage_name);
   }
   xbt_dict_free(&storage_list);
diff --git a/teshsuite/s4u/storage_client_server/storage_client_server.tesh b/teshsuite/s4u/storage_client_server/storage_client_server.tesh
index ee942ed592..d334f6b9f4 100644
--- a/teshsuite/s4u/storage_client_server/storage_client_server.tesh
+++ b/teshsuite/s4u/storage_client_server/storage_client_server.tesh
@@ -103,4 +103,6 @@ $ ./storage_client_server$EXEEXT ${srcdir:=.}/../../../examples/platforms/storag
 > [  1.207952] (server@alice) 	\Windows\write.exe size: 10752 bytes
 > [  1.207952] (server@alice) Storage Disk1 is attached to bob
 > [  1.207952] (server@alice) Storage Disk2 is attached to alice
+> [  1.207952] (server@alice) Storage Disk3 is attached to carl
+> [  1.207952] (server@alice) Storage Disk4 is attached to denise
 > [  1.207952] (maestro@) Simulated time: 1.20795
diff --git a/teshsuite/simdag/basic-link-test/basic-link-test.c b/teshsuite/simdag/basic-link-test/basic-link-test.c
index e3a01ade8a..5cf1254a31 100644
--- a/teshsuite/simdag/basic-link-test/basic-link-test.c
+++ b/teshsuite/simdag/basic-link-test/basic-link-test.c
@@ -26,7 +26,7 @@ int main(int argc, char **argv)
   int count = sg_link_count();
   XBT_INFO("Link count: %d", count);
   qsort((void *)links, count, sizeof(SD_link_t), cmp_link);
-   
+
   for (int i=0; i < count; i++){
     XBT_INFO("%s: latency = %.5f, bandwidth = %f", sg_link_name(links[i]),
              sg_link_latency(links[i]), sg_link_bandwidth(links[i]));
diff --git a/teshsuite/simdag/comm-mxn-all2all/comm-mxn-all2all.c b/teshsuite/simdag/comm-mxn-all2all/comm-mxn-all2all.c
index 45e8e62849..285f0bdecb 100644
--- a/teshsuite/simdag/comm-mxn-all2all/comm-mxn-all2all.c
+++ b/teshsuite/simdag/comm-mxn-all2all/comm-mxn-all2all.c
@@ -12,7 +12,7 @@ XBT_LOG_NEW_DEFAULT_SUBCATEGORY(comm_mxn_all2all, sd, "SimDag test All2All");
 /*
  * intra communication test
  * All2All
- * 
+ *
  * send 1 byte from all to all
  * + 2 secs latency
  * should be 8  (platform_4p_1switch.xml)
diff --git a/teshsuite/simdag/comm-mxn-independent/comm-mxn-independent.c b/teshsuite/simdag/comm-mxn-independent/comm-mxn-independent.c
index 7e9e61921d..789906a2c5 100644
--- a/teshsuite/simdag/comm-mxn-independent/comm-mxn-independent.c
+++ b/teshsuite/simdag/comm-mxn-independent/comm-mxn-independent.c
@@ -12,8 +12,8 @@ XBT_LOG_NEW_DEFAULT_SUBCATEGORY(comm_mxn_independent, sd, "SimDag test independe
 /*
  * intra communication test
  * independent communication
- * 
- * 0 -> 1 
+ *
+ * 0 -> 1
  * 2 -> 3
  * shared is only switch which is fat pipe
  * should be 1 + 2 latency = 3
diff --git a/teshsuite/simdag/comm-mxn-scatter/comm-mxn-scatter.c b/teshsuite/simdag/comm-mxn-scatter/comm-mxn-scatter.c
index 6116df43e7..5b77c2c486 100644
--- a/teshsuite/simdag/comm-mxn-scatter/comm-mxn-scatter.c
+++ b/teshsuite/simdag/comm-mxn-scatter/comm-mxn-scatter.c
@@ -12,7 +12,7 @@ XBT_LOG_NEW_DEFAULT_SUBCATEGORY(comm_mxn_scatter, sd, "SimDag test scatter");
 /*
  * intra communication test 1
  * scatter
- * 
+ *
  * start: 1 2 3 (each having 1/3 of the bandwidth)
  * after 3 sec: 0 1 2 (having 1/2 of the bandwidth)
  * after another 2 sec: 0 0 1 (having all the bandwidth)
diff --git a/teshsuite/simdag/comm-p2p-latency-1/comm-p2p-latency-1.c b/teshsuite/simdag/comm-p2p-latency-1/comm-p2p-latency-1.c
index e4bce55f33..1445df4f8f 100644
--- a/teshsuite/simdag/comm-p2p-latency-1/comm-p2p-latency-1.c
+++ b/teshsuite/simdag/comm-p2p-latency-1/comm-p2p-latency-1.c
@@ -11,8 +11,8 @@
 
 /*
  * simple latency test
- * send one byte from 0 to 1 
- * 
+ * send one byte from 0 to 1
+ *
  * this is a test for multiple platforms
  * see tesh file for expected output
  */
diff --git a/teshsuite/simdag/comm-p2p-latency-2/comm-p2p-latency-2.c b/teshsuite/simdag/comm-p2p-latency-2/comm-p2p-latency-2.c
index 95f3491074..b8fa568234 100644
--- a/teshsuite/simdag/comm-p2p-latency-2/comm-p2p-latency-2.c
+++ b/teshsuite/simdag/comm-p2p-latency-2/comm-p2p-latency-2.c
@@ -11,7 +11,7 @@
 
 /*
  * bw and latency test 2
- * send 2 x 1 byte from 2 task in same direction 0 -> 1 
+ * send 2 x 1 byte from 2 task in same direction 0 -> 1
  */
 
 int main(int argc, char **argv)
diff --git a/teshsuite/simdag/comm-p2p-latency-3/comm-p2p-latency-3.c b/teshsuite/simdag/comm-p2p-latency-3/comm-p2p-latency-3.c
index abdfb6e15f..c95a637fd6 100644
--- a/teshsuite/simdag/comm-p2p-latency-3/comm-p2p-latency-3.c
+++ b/teshsuite/simdag/comm-p2p-latency-3/comm-p2p-latency-3.c
@@ -12,7 +12,7 @@
 /**
  * bw and latency test 3
  * same intention as test 2
- * sending 2 x 1 bytes at the same time 
+ * sending 2 x 1 bytes at the same time
  * this time in opposite direction
  */
 
diff --git a/teshsuite/simdag/comm-p2p-latency-bound/comm-p2p-latency-bound.c b/teshsuite/simdag/comm-p2p-latency-bound/comm-p2p-latency-bound.c
index db1fa4ff49..407b0f3a0a 100644
--- a/teshsuite/simdag/comm-p2p-latency-bound/comm-p2p-latency-bound.c
+++ b/teshsuite/simdag/comm-p2p-latency-bound/comm-p2p-latency-bound.c
@@ -12,7 +12,7 @@
 #define TASK_NUM 3
 
 /**
- * 3 tasks send 1 byte in parallel 
+ * 3 tasks send 1 byte in parallel
  * 3 flows exceed bandwidth
  * should be 10001.5
  * because the max tcp win size is 20000
diff --git a/teshsuite/smpi/bug-17132/bug-17132.c b/teshsuite/smpi/bug-17132/bug-17132.c
index d260080644..432d44a278 100644
--- a/teshsuite/smpi/bug-17132/bug-17132.c
+++ b/teshsuite/smpi/bug-17132/bug-17132.c
@@ -5,7 +5,7 @@
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
 /* Bug report: https://gforge.inria.fr/tracker/index.php?func=detail&aid=17132&group_id=12&atid=165 */
- 
+
 #include "xbt/log.h"
 #include <stdio.h>
 #include <mpi.h>
diff --git a/teshsuite/smpi/coll-allgather/coll-allgather.c b/teshsuite/smpi/coll-allgather/coll-allgather.c
index 9b486d2441..68cfc76500 100644
--- a/teshsuite/smpi/coll-allgather/coll-allgather.c
+++ b/teshsuite/smpi/coll-allgather/coll-allgather.c
@@ -28,7 +28,7 @@ int main(int argc, char *argv[])
   int count = 2;
   int* sb = (int *) xbt_malloc(count * sizeof(int));
   int* rb = (int *) xbt_malloc(count * size * sizeof(int));
-  
+
   for (int i = 0; i < count; ++i)
     sb[i] = rank * count + i;
   for (int i = 0; i < count * size; ++i)
diff --git a/teshsuite/smpi/coll-allgatherv/coll-allgatherv.c b/teshsuite/smpi/coll-allgatherv/coll-allgatherv.c
index 664b3c2912..1cbda3cc01 100644
--- a/teshsuite/smpi/coll-allgatherv/coll-allgatherv.c
+++ b/teshsuite/smpi/coll-allgatherv/coll-allgatherv.c
@@ -41,7 +41,7 @@ int main(int argc, char *argv[])
 
   for (i = 0; i < recv_counts[rank]; ++i)
     sb[i] = recv_disps[rank] + i;
-  for (i = 0; i < recv_sb_size; ++i)  
+  for (i = 0; i < recv_sb_size; ++i)
     rb[i] = -1;
 
   printf("[%d] sndbuf=[", rank);
diff --git a/teshsuite/smpi/coll-allreduce/coll-allreduce.c b/teshsuite/smpi/coll-allreduce/coll-allreduce.c
index 7eef33992a..45eeb9c8e5 100644
--- a/teshsuite/smpi/coll-allreduce/coll-allreduce.c
+++ b/teshsuite/smpi/coll-allreduce/coll-allreduce.c
@@ -32,7 +32,7 @@ int main(int argc, char *argv[])
     mult = size;
   int* sb = (int *) xbt_malloc(size *maxlen * sizeof(int));
   int* rb = (int *) xbt_malloc(size *maxlen * sizeof(int));
-  
+
   for (i = 0; i < size *maxlen; ++i) {
     sb[i] = rank*size + i;
     rb[i] = 0;
diff --git a/teshsuite/smpi/coll-alltoall/coll-alltoall.c b/teshsuite/smpi/coll-alltoall/coll-alltoall.c
index 4ad259883b..1bbcf69d17 100644
--- a/teshsuite/smpi/coll-alltoall/coll-alltoall.c
+++ b/teshsuite/smpi/coll-alltoall/coll-alltoall.c
@@ -28,7 +28,7 @@ int main(int argc, char *argv[])
 
   int* sb = (int *) xbt_malloc(size * sizeof(int) * 2);
   int* rb = (int *) xbt_malloc(size * sizeof(int) * 2);
-  
+
   for (i = 0; i < size; ++i) {
     sb[i] = rank*size + i;
     rb[i] = 0;
diff --git a/teshsuite/smpi/coll-alltoallv/coll-alltoallv.c b/teshsuite/smpi/coll-alltoallv/coll-alltoallv.c
index d6bf366320..780ee6354f 100644
--- a/teshsuite/smpi/coll-alltoallv/coll-alltoallv.c
+++ b/teshsuite/smpi/coll-alltoallv/coll-alltoallv.c
@@ -101,7 +101,7 @@ int main(int argc, char **argv)
 
   print_buffer_int( rbuf, size*size, strdup("rbuf:"),rank);
 
-  MPI_Barrier(MPI_COMM_WORLD);  
+  MPI_Barrier(MPI_COMM_WORLD);
   if (0 == rank) {
     printf("Alltoallv TEST COMPLETE.\n");
   }
diff --git a/teshsuite/smpi/coll-bcast/coll-bcast.c b/teshsuite/smpi/coll-bcast/coll-bcast.c
index 2f40572ebd..e8aeaecc0a 100644
--- a/teshsuite/smpi/coll-bcast/coll-bcast.c
+++ b/teshsuite/smpi/coll-bcast/coll-bcast.c
@@ -34,7 +34,7 @@ int main(int argc, char **argv)
   xbt_free(values);
 
   count = 4096;
-  values = (int *) xbt_malloc(count * sizeof(int));  
+  values = (int *) xbt_malloc(count * sizeof(int));
 
   for (i = 0; i < count; i++)
     values[i] = (size -1 == rank) ? 17 : 3;
diff --git a/teshsuite/smpi/coll-gather/coll-gather.c b/teshsuite/smpi/coll-gather/coll-gather.c
index a8cd2c4389..6c593da19c 100644
--- a/teshsuite/smpi/coll-gather/coll-gather.c
+++ b/teshsuite/smpi/coll-gather/coll-gather.c
@@ -30,7 +30,7 @@ int main(int argc, char *argv[])
   int count = 2;
   int* sb = (int *) xbt_malloc(count * sizeof(int));
   int* rb = (int *) xbt_malloc(count * size * sizeof(int));
-  
+
   for (int i = 0; i < count; ++i)
     sb[i] = rank * count + i;
   for (int i = 0; i < count * size; ++i)
diff --git a/teshsuite/smpi/coll-reduce-scatter/coll-reduce-scatter.c b/teshsuite/smpi/coll-reduce-scatter/coll-reduce-scatter.c
index 6299336122..6a866f042c 100644
--- a/teshsuite/smpi/coll-reduce-scatter/coll-reduce-scatter.c
+++ b/teshsuite/smpi/coll-reduce-scatter/coll-reduce-scatter.c
@@ -4,7 +4,7 @@
 /* This program is free software; you can redistribute it and/or modify it
  * under the terms of the license (GNU LGPL) which comes with this package. */
 
-/* 
+/*
  * Test of reduce scatter.
  * Each processor contributes its rank + the index to the reduction,  then receives the ith sum
  * Can be called with any number of processors.
@@ -29,11 +29,11 @@ int main( int argc, char **argv )
     MPI_Comm_size( comm, &size );
     MPI_Comm_rank( comm, &rank );
     int* sendbuf = (int *) malloc( size * sizeof(int) );
-    for (i=0; i<size; i++) 
+    for (i=0; i<size; i++)
       sendbuf[i] = rank + i;
     int* recvcounts = (int*) malloc (size * sizeof(int));
     int* recvbuf    = (int*) malloc (size * sizeof(int));
-    for (i=0; i<size; i++) 
+    for (i=0; i<size; i++)
       recvcounts[i] = 1;
     MPI_Reduce_scatter( sendbuf, recvbuf, recvcounts, MPI_INT, MPI_SUM, comm );
     int sumval = size * rank + ((size - 1) * size)/2;
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock.c
index 9fe72c80ee..8f14987a63 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock.c
@@ -36,12 +36,12 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf0, buf_size, MPI_INT, 
+      MPI_Recv (buf0, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
     }
   else if (rank == 1)
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock10.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock10.c
index 65e3a6e08e..1d0905954f 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock10.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock10.c
@@ -40,14 +40,14 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock10_mod.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock10_mod.c
index 25d2b5ab6c..67cef7dc5b 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock10_mod.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock10_mod.c
@@ -40,14 +40,14 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock11.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock11.c
index 48717b8992..63058b185f 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock11.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock11.c
@@ -40,14 +40,14 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock2.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock2.c
index 6559eb23b6..562af09d34 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock2.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock2.c
@@ -36,12 +36,12 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf0, buf_size, MPI_INT, 
+      MPI_Recv (buf0, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
     }
   else if (rank == 1)
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock3.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock3.c
index 53015d003e..889cf38c5a 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock3.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock3.c
@@ -36,12 +36,12 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf0, buf_size, MPI_INT, 
+      MPI_Recv (buf0, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
     }
   else if (rank == 1)
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock4.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock4.c
index 94dd335781..a68a1589fe 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock4.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock4.c
@@ -37,14 +37,14 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock4_mod.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock4_mod.c
index 51cf1cedf2..0d3adfc8fb 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock4_mod.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock4_mod.c
@@ -37,14 +37,14 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock5.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock5.c
index 050d5577b6..1f4c0cd7e6 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock5.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock5.c
@@ -41,14 +41,14 @@ main (int argc, char **argv)
     {
       //sleep (60);
 
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock5_mod.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock5_mod.c
index d06d03105c..b43b516e64 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock5_mod.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock5_mod.c
@@ -41,14 +41,14 @@ main (int argc, char **argv)
     {
       //sleep (60);
 
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock6.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock6.c
index 4ca42bc1dc..bc9bef9fb4 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock6.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock6.c
@@ -41,14 +41,14 @@ main (int argc, char **argv)
     {
 //      sleep (60);
 
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock6_mod.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock6_mod.c
index 6e80869ee1..3dc1c190e3 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock6_mod.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock6_mod.c
@@ -41,14 +41,14 @@ main (int argc, char **argv)
     {
 //      sleep (60);
 
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock7.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock7.c
index b265d21ae8..2cf5d54a9b 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock7.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock7.c
@@ -39,14 +39,14 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock8.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock8.c
index d789bebdc8..0d2280845d 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock8.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock8.c
@@ -42,14 +42,14 @@ main (int argc, char **argv)
     {
 //      sleep (60);
 
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-can-deadlock9.c b/teshsuite/smpi/isp/umpire/any_src-can-deadlock9.c
index 3778ffe7bd..b4849392b4 100644
--- a/teshsuite/smpi/isp/umpire/any_src-can-deadlock9.c
+++ b/teshsuite/smpi/isp/umpire/any_src-can-deadlock9.c
@@ -42,14 +42,14 @@ main (int argc, char **argv)
     {
       //sleep (60);
 
-     MPI_Irecv (buf0, buf_size, MPI_INT, 
+     MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD, &status);
 
       MPI_Send (buf1, buf_size, MPI_INT, 2, 0, MPI_COMM_WORLD);
 
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-deadlock.c b/teshsuite/smpi/isp/umpire/any_src-deadlock.c
index e8beec3ce9..e6b990ff2b 100644
--- a/teshsuite/smpi/isp/umpire/any_src-deadlock.c
+++ b/teshsuite/smpi/isp/umpire/any_src-deadlock.c
@@ -34,7 +34,7 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Recv (buf1, buf_size, MPI_INT, 
+      MPI_Recv (buf1, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 
       MPI_Recv (buf0, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-wait-deadlock.c b/teshsuite/smpi/isp/umpire/any_src-wait-deadlock.c
index 130c972068..1b91f58abb 100644
--- a/teshsuite/smpi/isp/umpire/any_src-wait-deadlock.c
+++ b/teshsuite/smpi/isp/umpire/any_src-wait-deadlock.c
@@ -37,7 +37,7 @@ main (int argc, char **argv)
     {
       memset (buf0, 0, buf_size);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Wait (&req, &status);
@@ -48,7 +48,7 @@ main (int argc, char **argv)
     {
       memset (buf1, 1, buf_size);
 
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-wait-deadlock2.c b/teshsuite/smpi/isp/umpire/any_src-wait-deadlock2.c
index 6112e1235f..bb3eda69c9 100644
--- a/teshsuite/smpi/isp/umpire/any_src-wait-deadlock2.c
+++ b/teshsuite/smpi/isp/umpire/any_src-wait-deadlock2.c
@@ -37,10 +37,10 @@ main (int argc, char **argv)
     {
       memset (buf0, 0, buf_size);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
       printf("Proc 0:  Request number - %p\n",req);
-      
+
       MPI_Send (buf0, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
 	
       MPI_Recv (buf0, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &status);
@@ -54,7 +54,7 @@ main (int argc, char **argv)
     {
       memset (buf1, 1, buf_size);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
       printf("Proc 1:  Request number - %p\n",req);
 
diff --git a/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock.c b/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock.c
index 246863705c..5e2dc2981d 100644
--- a/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock.c
+++ b/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock.c
@@ -37,10 +37,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
diff --git a/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock2.c b/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock2.c
index fb2fdefea1..461953212b 100644
--- a/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock2.c
+++ b/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock2.c
@@ -37,10 +37,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
diff --git a/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock3.c b/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock3.c
index 64591d66e4..7fcd71dc0f 100644
--- a/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock3.c
+++ b/teshsuite/smpi/isp/umpire/any_src-waitall-deadlock3.c
@@ -37,10 +37,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
diff --git a/teshsuite/smpi/isp/umpire/any_src-waitany-deadlock.c b/teshsuite/smpi/isp/umpire/any_src-waitany-deadlock.c
index 9bf356e9f0..008c7400ed 100644
--- a/teshsuite/smpi/isp/umpire/any_src-waitany-deadlock.c
+++ b/teshsuite/smpi/isp/umpire/any_src-waitany-deadlock.c
@@ -38,10 +38,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitany (2, reqs, &done, &status);
diff --git a/teshsuite/smpi/isp/umpire/any_src-waitany-deadlock2.c b/teshsuite/smpi/isp/umpire/any_src-waitany-deadlock2.c
index 5db599293c..8b345592ac 100644
--- a/teshsuite/smpi/isp/umpire/any_src-waitany-deadlock2.c
+++ b/teshsuite/smpi/isp/umpire/any_src-waitany-deadlock2.c
@@ -38,10 +38,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
@@ -66,7 +66,7 @@ main (int argc, char **argv)
     }
   else if (rank == 2)
     {
-    //  sleep (60);      
+    //  sleep (60);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 1, 1, MPI_COMM_WORLD, statuses);
     }
diff --git a/teshsuite/smpi/isp/umpire/basic-deadlock-comm_dup.c b/teshsuite/smpi/isp/umpire/basic-deadlock-comm_dup.c
index d71c51023d..c115c5aa27 100644
--- a/teshsuite/smpi/isp/umpire/basic-deadlock-comm_dup.c
+++ b/teshsuite/smpi/isp/umpire/basic-deadlock-comm_dup.c
@@ -34,12 +34,12 @@ main (int argc, char **argv)
   }
   else {
     MPI_Comm_dup (MPI_COMM_WORLD, &comm);
-    
+
     if (rank == 0) {
       memset (buf0, 0, buf_size);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 1, 0, comm, &status);
-      
+
       MPI_Send (buf0, buf_size, MPI_INT, 1, 0, comm);
     }
     else if (rank == 1) {
diff --git a/teshsuite/smpi/isp/umpire/basic-deadlock-comm_split.c b/teshsuite/smpi/isp/umpire/basic-deadlock-comm_split.c
index 89828d4a8e..17ce7e5c1c 100644
--- a/teshsuite/smpi/isp/umpire/basic-deadlock-comm_split.c
+++ b/teshsuite/smpi/isp/umpire/basic-deadlock-comm_split.c
@@ -35,7 +35,7 @@ main (int argc, char **argv)
   }
   else {
     MPI_Comm_split (MPI_COMM_WORLD, rank % 2, nprocs - rank, &comm);
-    
+
     if (comm != MPI_COMM_NULL) {
       MPI_Comm_size (comm, &dnprocs);
       MPI_Comm_rank (comm, &drank);
diff --git a/teshsuite/smpi/isp/umpire/basic-deadlock-graph_create.c b/teshsuite/smpi/isp/umpire/basic-deadlock-graph_create.c
index 2017b4c48f..c800ba65db 100644
--- a/teshsuite/smpi/isp/umpire/basic-deadlock-graph_create.c
+++ b/teshsuite/smpi/isp/umpire/basic-deadlock-graph_create.c
@@ -39,7 +39,7 @@ main (int argc, char **argv)
   }
   else {
     /* create the graph on p.268 MPI: The Complete Reference... */
-    MPI_Graph_create (MPI_COMM_WORLD, GRAPH_SZ, 
+    MPI_Graph_create (MPI_COMM_WORLD, GRAPH_SZ,
 		      graph_index, graph_edges, 1, &comm);
 
     if (comm != MPI_COMM_NULL) {
diff --git a/teshsuite/smpi/isp/umpire/basic-deadlock-intercomm_create.c b/teshsuite/smpi/isp/umpire/basic-deadlock-intercomm_create.c
index 78beb4fb05..f3a345e2f2 100644
--- a/teshsuite/smpi/isp/umpire/basic-deadlock-intercomm_create.c
+++ b/teshsuite/smpi/isp/umpire/basic-deadlock-intercomm_create.c
@@ -46,12 +46,12 @@ main (int argc, char **argv)
 			    INTERCOMM_CREATE_TAG, &intercomm);
 
       MPI_Comm_free (&temp);
-      
+
       if (intercomm != MPI_COMM_NULL) {
 	MPI_Comm_size (intercomm, &dnprocs);
 	MPI_Comm_rank (intercomm, &drank);
  	MPI_Comm_remote_size (intercomm, &rnprocs);
-     
+
 	if (rnprocs > drank) {
 	  if (rank % 2) {
 	    memset (buf1, 1, buf_size);
@@ -62,7 +62,7 @@ main (int argc, char **argv)
 	  }
 	  else {
 	    memset (buf0, 0, buf_size);
-	    
+	
 	    MPI_Recv (buf1, buf_size, MPI_INT, drank, 0, intercomm, &status);
 	
 	    MPI_Send (buf0, buf_size, MPI_INT, drank, 0, intercomm);
diff --git a/teshsuite/smpi/isp/umpire/basic-deadlock-intercomm_merge.c b/teshsuite/smpi/isp/umpire/basic-deadlock-intercomm_merge.c
index e9361205b6..587569c5ce 100644
--- a/teshsuite/smpi/isp/umpire/basic-deadlock-intercomm_merge.c
+++ b/teshsuite/smpi/isp/umpire/basic-deadlock-intercomm_merge.c
@@ -46,7 +46,7 @@ main (int argc, char **argv)
 			    INTERCOMM_CREATE_TAG, &intercomm);
 
       MPI_Comm_free (&temp);
-      
+
       if (intercomm == MPI_COMM_NULL) {
 	printf ("(%d) MPI_Intercomm_Create returned MPI_COMM_NULL\n", rank);
 	printf ("(%d) Aborting...\n", rank);
@@ -61,11 +61,11 @@ main (int argc, char **argv)
       if (comm != MPI_COMM_NULL) {
 	MPI_Comm_size (comm, &dnprocs);
 	MPI_Comm_rank (comm, &drank);
-      
+
 	if (dnprocs > 1) {
 	  if (drank == 0) {
 	    memset (buf0, 0, buf_size);
-	    
+	
 	    MPI_Recv (buf1, buf_size, MPI_INT, 1, 0, comm, &status);
 	
 	    MPI_Send (buf0, buf_size, MPI_INT, 1, 0, comm);
diff --git a/teshsuite/smpi/isp/umpire/change-send-buffer-exhaustive.c b/teshsuite/smpi/isp/umpire/change-send-buffer-exhaustive.c
index 661ba0aa6d..aec87df2de 100644
--- a/teshsuite/smpi/isp/umpire/change-send-buffer-exhaustive.c
+++ b/teshsuite/smpi/isp/umpire/change-send-buffer-exhaustive.c
@@ -42,42 +42,42 @@ main (int argc, char **argv)
   printf ("(%d) is alive on %s\n", rank, processor_name);
   fflush (stdout);
 
-  MPI_Buffer_attach (bbuf, sizeof(int) * 
+  MPI_Buffer_attach (bbuf, sizeof(int) *
 		     (BUF_SIZE + MPI_BSEND_OVERHEAD) * 2 * NUM_BSEND_TYPES);
 
   if (rank == 0) {
     /* set up persistent sends... */
     send_t_number = NUM_SEND_TYPES - NUM_PERSISTENT_SEND_TYPES;
 
-    MPI_Send_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Send_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Send_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Send_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
     send_t_number++;
 
-    MPI_Bsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Bsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Bsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Bsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
 
     send_t_number++;
 
-    MPI_Rsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Rsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Rsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Rsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
     send_t_number++;
 
-    MPI_Ssend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Ssend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Ssend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Ssend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
   }
 
@@ -101,19 +101,19 @@ main (int argc, char **argv)
     if (rank == 0) {
       /* set up transient sends... */
       send_t_number = 0;
-    
+
       MPI_Isend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		 1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
       MPI_Isend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		 BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		 BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		 comm, &aReq[send_t_number * 2 + 1]);
 
       send_t_number++;
-      
+
       MPI_Ibsend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		  1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
       MPI_Ibsend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		  BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		  BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		  comm, &aReq[send_t_number * 2 + 1]);
 
       send_t_number++;
@@ -124,7 +124,7 @@ main (int argc, char **argv)
       MPI_Irsend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		  1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
       MPI_Irsend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		  BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		  BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		  comm, &aReq[send_t_number * 2 + 1]);
 
       send_t_number++;
@@ -132,7 +132,7 @@ main (int argc, char **argv)
       MPI_Issend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		  1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
       MPI_Issend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		  BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		  BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		  comm, &aReq[send_t_number * 2 + 1]);
 
       /* just to be paranoid */
@@ -148,7 +148,7 @@ main (int argc, char **argv)
 	  MPI_Start (&aReq[2 * send_t_number + j]);
 	}
       }
-    
+
       /* NOTE: Changing the send buffer of a Bsend is NOT an error... */
       for (j = 0; j < NUM_SEND_TYPES; j++) {
 	/* muck the buffers */
@@ -229,7 +229,7 @@ main (int argc, char **argv)
 	  outcount = 0;
 
 	  while (!outcount) {
-	    MPI_Testsome (NUM_SEND_TYPES * 2, aReq, 
+	    MPI_Testsome (NUM_SEND_TYPES * 2, aReq,
 			  &outcount, indices, aStatus);
 	  }
 
@@ -246,7 +246,7 @@ main (int argc, char **argv)
     else if (rank == 1) {
       /* set up receives for all of the sends */
       for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
-	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE, 
+	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE,
 		   MPI_INT, 0, j, comm, &aReq[j]);
       }
 
@@ -274,7 +274,7 @@ main (int argc, char **argv)
 
   MPI_Buffer_detach (bbuf, &at_size);
 
-  assert (at_size == 
+  assert (at_size ==
 	  sizeof(int) * (BUF_SIZE + MPI_BSEND_OVERHEAD) * 2 * NUM_BSEND_TYPES);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/change-send-buffer-type-exhaustive.c b/teshsuite/smpi/isp/umpire/change-send-buffer-type-exhaustive.c
index f95139add9..c770a4cb0a 100644
--- a/teshsuite/smpi/isp/umpire/change-send-buffer-type-exhaustive.c
+++ b/teshsuite/smpi/isp/umpire/change-send-buffer-type-exhaustive.c
@@ -57,8 +57,8 @@ main (int argc, char **argv)
   int namelen = 128;
   int i, j, k, basic_extent;
   int blocklens[4], displs[4];
-  MPI_Datatype structtypes[4]; 
-  MPI_Datatype newtype[TYPE_CONSTRUCTOR_COUNT]; 
+  MPI_Datatype structtypes[4];
+  MPI_Datatype newtype[TYPE_CONSTRUCTOR_COUNT];
   MPI_Request aReq[TYPE_CONSTRUCTOR_COUNT];
   MPI_Status aStatus[TYPE_CONSTRUCTOR_COUNT];
 #ifdef RUN_TYPE_STRUCT
@@ -102,7 +102,7 @@ main (int argc, char **argv)
 
   /* create the types */
   MPI_Type_struct (2, blocklens, displs, structtypes, &newtype[0]);
-  
+
   MPI_Type_extent (newtype[0], &basic_extent);
   if (basic_extent != sizeof (test_basic_struct_t)) {
     fprintf (stderr, "(%d): Unexpect extent for struct\n");
@@ -110,7 +110,7 @@ main (int argc, char **argv)
   }
 
   MPI_Type_vector (2, 3, 4, newtype[0], &newtype[1]);
-  MPI_Type_hvector (3, 2, 15 * sizeof (test_basic_struct_t), 
+  MPI_Type_hvector (3, 2, 15 * sizeof (test_basic_struct_t),
 		    newtype[1], &newtype[2]);
   displs[1] = 2;
   MPI_Type_indexed (2, blocklens, displs, newtype[2], &newtype[3]);
@@ -130,7 +130,7 @@ main (int argc, char **argv)
 
   MPI_Type_struct (4, blocklens, displs, structtypes, &newtype[6]);
 
-#ifdef RUN_TYPE_STRUCT 
+#ifdef RUN_TYPE_STRUCT
   MPI_Type_commit (&newtype[0]);
 #endif
 
@@ -154,7 +154,7 @@ main (int argc, char **argv)
   MPI_Type_commit (&newtype[5]);
 #endif
 
-#ifdef RUN_TYPE_STRUCT_LB_UB 
+#ifdef RUN_TYPE_STRUCT_LB_UB
   MPI_Type_commit (&newtype[6]);
 #endif
 
@@ -248,7 +248,7 @@ main (int argc, char **argv)
 #endif
 
 #ifdef RUN_TYPE_STRUCT
-    MPI_Isend (&(struct_lb_ub_send_buf[0].the_double_to_send), 
+    MPI_Isend (&(struct_lb_ub_send_buf[0].the_double_to_send),
 	       MSG_COUNT, newtype[6], 1, 6, comm, &aReq[6]);
 #else
     aReq[6] = MPI_REQUEST_NULL;
@@ -341,7 +341,7 @@ main (int argc, char **argv)
 #endif
 
 #ifdef RUN_TYPE_STRUCT_LB_UB
-    MPI_Irecv (struct_lb_ub_recv_buf, 
+    MPI_Irecv (struct_lb_ub_recv_buf,
 	       MSG_COUNT, newtype[0], 0, 6, comm, &aReq[6]);
 #else
     aReq[6] = MPI_REQUEST_NULL;
@@ -416,7 +416,7 @@ main (int argc, char **argv)
 #ifdef RUN_TYPE_HVECTOR
       /* eight holes in hvector_buf... */
       /* hole in first vector, first block... */
-      assert ((hvector_buf[i*44 + 3].the_double == 2.0) && 
+      assert ((hvector_buf[i*44 + 3].the_double == 2.0) &&
 	      (hvector_buf[i*44 + 3].the_char == 'b'));
       /* hole in second vector, first block... */
       assert ((hvector_buf[i*44 + 10].the_double == 2.0) &&
diff --git a/teshsuite/smpi/isp/umpire/collective-exhaustive-byte-int-mismatch.c b/teshsuite/smpi/isp/umpire/collective-exhaustive-byte-int-mismatch.c
index 42553826d0..a32b1f4e0c 100644
--- a/teshsuite/smpi/isp/umpire/collective-exhaustive-byte-int-mismatch.c
+++ b/teshsuite/smpi/isp/umpire/collective-exhaustive-byte-int-mismatch.c
@@ -63,14 +63,14 @@ complex_prod (void *inp, void *inoutp, int *len, MPI_Datatype *dptr)
 
   if (*dptr == MPI_BYTE)
     stop = (*len)/(2 * sizeof(double));
-  else 
+  else
     stop = *len;
 
   for (i = 0; i < stop; i++) {
     c.real = inout->real * in->real - inout->imag * in->imag;
     c.imag = inout->real * in->imag + inout->imag * in->real;
     *inout = c;
-    in++; 
+    in++;
     inout++;
   }
 
@@ -104,7 +104,7 @@ main (int argc, char **argv)
 
   buf0 = (int *) malloc (buf_size * nprocs * sizeof(int));
   assert (buf0);
-  for (i = 0; i < buf_size * nprocs; i++) 
+  for (i = 0; i < buf_size * nprocs; i++)
     buf0[i] = rank;
 
 #ifdef RUN_ALLTOALLV
@@ -128,7 +128,7 @@ main (int argc, char **argv)
   if (rank == 0) {
     buf1 = (int *) malloc (buf_size * nprocs * sizeof(int));
     assert (buf1);
-    for (i = 0; i < buf_size * nprocs; i++) 
+    for (i = 0; i < buf_size * nprocs; i++)
       buf1[i] = i;
 
     displs = (int *) malloc (nprocs * sizeof(int));
@@ -152,286 +152,286 @@ main (int argc, char **argv)
 #endif
 
 #ifdef RUN_BARRIER
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Barrier (comm);
 #endif
 
 #ifdef RUN_BCAST
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Bcast (buf0, buf_size, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_GATHER
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Gather (&buf0[rank*buf_size], buf_size, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Gather (&buf0[rank*buf_size], buf_size,
 		  MPI_INT, buf1, buf_size, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_SCATTER
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scatter (buf1, buf_size, MPI_INT, buf0, buf_size, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_GATHERV
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Gatherv (&buf0[rank*buf_size], 
-		   (rank < buf_size) ? rank : buf_size, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Gatherv (&buf0[rank*buf_size],
+		   (rank < buf_size) ? rank : buf_size,
 		   MPI_INT, buf1, rcounts, displs, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_SCATTERV
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Scatterv (buf1, counts, displs, MPI_INT, buf0, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Scatterv (buf1, counts, displs, MPI_INT, buf0,
 		    (rank < buf_size) ? rank : buf_size, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_REDUCE
 #ifdef RUN_MAX
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_MAX, 0, comm);
 #endif
 
 #ifdef RUN_MIN
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_MIN, 0, comm);
 #endif
 
 #ifdef RUN_SUM
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_SUM, 0, comm);
 #endif
 
 #ifdef RUN_PROD
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_PROD, 0, comm);
 #endif
 
 #ifdef RUN_LAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_LAND, 0, comm);
 #endif
 
 #ifdef RUN_BAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_BAND, 0, comm);
 #endif
 
 #ifdef RUN_LOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_LOR, 0, comm);
 #endif
 
 #ifdef RUN_BOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_BOR, 0, comm);
 #endif
 
 #ifdef RUN_LXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_LXOR, 0, comm);
 #endif
 
 #ifdef RUN_BXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_BXOR, 0, comm);
 #endif
 
 #ifdef RUN_USEROP
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (a, answer, buf_size, ctype, user_op, 0, comm);
 #endif
 #endif
 
 #ifdef RUN_ALLGATHER
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allgather (buf0, buf_size, MPI_INT, buf1, buf_size, MPI_INT, comm);
 #endif
 
 #ifdef RUN_ALLTOALL
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Alltoall (buf1, buf_size, MPI_INT, buf0, buf_size, MPI_INT, comm);
 #endif
 
 #ifdef RUN_ALLGATHERV
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Allgatherv (buf0, 
-		      (rank < buf_size) ? rank : buf_size, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Allgatherv (buf0,
+		      (rank < buf_size) ? rank : buf_size,
 		      MPI_INT, buf1, rcounts, displs, MPI_INT, comm);
 #endif
 
 #ifdef RUN_ALLTOALLV
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Alltoallv (buf1, alltoallvcounts, displs, MPI_INT, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Alltoallv (buf1, alltoallvcounts, displs, MPI_INT,
 		     buf0, alltoallvcounts, displs, MPI_INT, comm);
 #endif
 
 #ifdef RUN_ALLREDUCE
 #ifdef RUN_MAX
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_USEROP
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (a, answer, buf_size, ctype, user_op, comm);
 #endif
 #endif
 
 #ifdef RUN_REDUCE_SCATTER
 #ifdef RUN_MAX
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_USEROP
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (a, answer, rcounts, ctype, user_op, comm);
 #endif
 #endif
 
 #ifdef RUN_SCAN
 #ifdef RUN_MAX
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_USEROP
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (a, answer, buf_size, ctype, user_op, comm);
 #endif
 #endif
@@ -440,105 +440,105 @@ main (int argc, char **argv)
     int *ricounts, *rdcounts;
 
 #ifdef RUN_BARRIER
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Barrier (comm);
 #endif
 
 #ifdef RUN_BCAST
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Bcast (buf0, buf_size * sizeof(int), MPI_BYTE, 0, comm);
 #endif
 
 #ifdef RUN_GATHER
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Gather (&buf0[rank*buf_size], buf_size * sizeof(int), 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Gather (&buf0[rank*buf_size], buf_size * sizeof(int),
 		  MPI_BYTE, buf1, buf_size * sizeof(int), MPI_BYTE, 0, comm);
 #endif
 
 #ifdef RUN_SCATTER
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Scatter (buf1, buf_size * sizeof(int), MPI_BYTE, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Scatter (buf1, buf_size * sizeof(int), MPI_BYTE,
 		   buf0, buf_size * sizeof(int), MPI_BYTE, 0, comm);
 #endif
 
 #ifdef RUN_GATHERV
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Gatherv (&buf0[rank*buf_size], 
-		   ((rank < buf_size) ? rank : buf_size) * sizeof(int), 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Gatherv (&buf0[rank*buf_size],
+		   ((rank < buf_size) ? rank : buf_size) * sizeof(int),
 		   MPI_BYTE, buf1, rcounts, displs, MPI_BYTE, 0, comm);
 #endif
 
 #ifdef RUN_SCATTERV
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Scatterv (buf1, counts, displs, MPI_BYTE, buf0, 
-		    ((rank < buf_size) ? rank : buf_size) * sizeof(int), 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Scatterv (buf1, counts, displs, MPI_BYTE, buf0,
+		    ((rank < buf_size) ? rank : buf_size) * sizeof(int),
 		    MPI_BYTE, 0, comm);
 #endif
 
 #ifdef RUN_REDUCE
 #ifdef RUN_MAX
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_MAX, 0, comm);
 #endif
 
 #ifdef RUN_MIN
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_MIN, 0, comm);
 #endif
 
 #ifdef RUN_SUM
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_SUM, 0, comm);
 #endif
 
 #ifdef RUN_PROD
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_PROD, 0, comm);
 #endif
 
 #ifdef RUN_LAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LAND, 0, comm);
 #endif
 
 #ifdef RUN_BAND
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Reduce (buf0, buf1, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Reduce (buf0, buf1,
 		  buf_size * sizeof(int), MPI_BYTE, MPI_BAND, 0, comm);
 #endif
 
 #ifdef RUN_LOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LOR, 0, comm);
 #endif
 
 #ifdef RUN_BOR
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Reduce (buf0, buf1, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Reduce (buf0, buf1,
 		  buf_size * sizeof(int), MPI_BYTE, MPI_BOR, 0, comm);
 #endif
 
 #ifdef RUN_LXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LXOR, 0, comm);
 #endif
 
 #ifdef RUN_BXOR
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Reduce (buf0, buf1, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Reduce (buf0, buf1,
 		  buf_size * sizeof(int), MPI_BYTE, MPI_BXOR, 0, comm);
 #endif
 
 #ifdef RUN_USEROP
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Reduce (a, answer, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Reduce (a, answer,
 		  buf_size * 2 * sizeof(double), MPI_BYTE, user_op, 0, comm);
 #endif
 #endif
 
     buf1 = (int *) malloc (buf_size * nprocs * sizeof(int));
     assert (buf1);
-    for (i = 0; i < buf_size * nprocs; i++) 
+    for (i = 0; i < buf_size * nprocs; i++)
       buf1[i] = i;
 
     displs = (int *) malloc (nprocs * sizeof(int));
@@ -570,202 +570,202 @@ main (int argc, char **argv)
 #endif
 
 #ifdef RUN_ALLGATHER
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Allgather (buf0, buf_size * sizeof(int), MPI_BYTE, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Allgather (buf0, buf_size * sizeof(int), MPI_BYTE,
 		     buf1, buf_size * sizeof(int), MPI_BYTE, comm);
 #endif
 
 #ifdef RUN_ALLTOALL
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Alltoall (buf1, buf_size * sizeof(int), MPI_BYTE, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Alltoall (buf1, buf_size * sizeof(int), MPI_BYTE,
 		    buf0, buf_size * sizeof(int), MPI_BYTE, comm);
 #endif
 
 #ifdef RUN_ALLGATHERV
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Allgatherv (buf0, 
-		      ((rank < buf_size) ? rank : buf_size) * sizeof(int), 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Allgatherv (buf0,
+		      ((rank < buf_size) ? rank : buf_size) * sizeof(int),
 		      MPI_BYTE, buf1, rcounts, displs, MPI_BYTE, comm);
 #endif
 
 #ifdef RUN_ALLTOALLV
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Alltoallv (buf1, alltoallvcounts, displs, MPI_BYTE, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Alltoallv (buf1, alltoallvcounts, displs, MPI_BYTE,
 		     buf0, alltoallvcounts, displs, MPI_BYTE, comm);
 #endif
 
 #ifdef RUN_ALLREDUCE
 #ifdef RUN_MAX
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Allreduce (buf0, buf1, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Allreduce (buf0, buf1,
 		     buf_size * sizeof (int), MPI_BYTE, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Allreduce (buf0, buf1, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Allreduce (buf0, buf1,
 		     buf_size * sizeof (int), MPI_BYTE, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Allreduce (buf0, buf1, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Allreduce (buf0, buf1,
 		     buf_size * sizeof (int), MPI_BYTE, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_USEROP
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Allreduce (a, answer, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Allreduce (a, answer,
 		     buf_size * 2 * sizeof (double), MPI_BYTE, user_op, comm);
 #endif
 #endif
 
 #ifdef RUN_REDUCE_SCATTER
 #ifdef RUN_MAX
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, ricounts, MPI_UNSIGNED, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, ricounts, MPI_UNSIGNED, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, ricounts, MPI_UNSIGNED, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, ricounts, MPI_UNSIGNED, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, ricounts, MPI_UNSIGNED, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_BYTE, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, ricounts, MPI_UNSIGNED, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_BYTE, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, ricounts, MPI_UNSIGNED, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_BYTE, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_USEROP
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (a, answer, rdcounts, MPI_BYTE, user_op, comm);
 #endif
 #endif
 
 #ifdef RUN_SCAN
 #ifdef RUN_MAX
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size * sizeof(int), MPI_BYTE, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size * sizeof(int), MPI_BYTE, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_UNSIGNED, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size * sizeof(int), MPI_BYTE, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_USEROP
-    for (i = 0; i < OP_COUNT; i++) 
-      MPI_Scan (a, answer, 
+    for (i = 0; i < OP_COUNT; i++)
+      MPI_Scan (a, answer,
 		buf_size * 2 * sizeof(double), MPI_BYTE, user_op, comm);
 #endif
 #endif
diff --git a/teshsuite/smpi/isp/umpire/collective-exhaustive-no-error.c b/teshsuite/smpi/isp/umpire/collective-exhaustive-no-error.c
index 1e53e147c4..98fcd9fcbe 100644
--- a/teshsuite/smpi/isp/umpire/collective-exhaustive-no-error.c
+++ b/teshsuite/smpi/isp/umpire/collective-exhaustive-no-error.c
@@ -65,7 +65,7 @@ complex_prod (void *inp, void *inoutp, int *len, MPI_Datatype *dptr)
     c.real = inout->real * in->real - inout->imag * in->imag;
     c.imag = inout->real * in->imag + inout->imag * in->real;
     *inout = c;
-    in++; 
+    in++;
     inout++;
   }
 
@@ -99,7 +99,7 @@ main (int argc, char **argv)
 
   buf0 = (int *) malloc (buf_size * nprocs * sizeof(int));
   assert (buf0);
-  for (i = 0; i < buf_size * nprocs; i++) 
+  for (i = 0; i < buf_size * nprocs; i++)
     buf0[i] = rank;
 
 #ifdef RUN_ALLTOALLV
@@ -128,7 +128,7 @@ main (int argc, char **argv)
   if (rank == 0) {
     buf1 = (int *) malloc (buf_size * nprocs * sizeof(int));
     assert (buf1);
-    for (i = 0; i < buf_size * nprocs; i++) 
+    for (i = 0; i < buf_size * nprocs; i++)
       buf1[i] = i;
 
     displs = (int *) malloc (nprocs * sizeof(int));
@@ -149,93 +149,93 @@ main (int argc, char **argv)
   }
 
 #ifdef RUN_BARRIER
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Barrier (comm);
 #endif
 
 #ifdef RUN_BCAST
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Bcast (buf0, buf_size, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_GATHER
-  for (i = 0; i < OP_COUNT; i++) 
-    MPI_Gather (&buf0[rank*buf_size], buf_size, 
+  for (i = 0; i < OP_COUNT; i++)
+    MPI_Gather (&buf0[rank*buf_size], buf_size,
 		MPI_INT, buf1, buf_size, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_SCATTER
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scatter (buf1, buf_size, MPI_INT, buf0, buf_size, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_GATHERV
-  for (i = 0; i < OP_COUNT; i++) 
-    MPI_Gatherv (&buf0[rank*buf_size], 
-		 (rank < buf_size) ? rank : buf_size, 
+  for (i = 0; i < OP_COUNT; i++)
+    MPI_Gatherv (&buf0[rank*buf_size],
+		 (rank < buf_size) ? rank : buf_size,
 		 MPI_INT, buf1, rcounts, displs, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_SCATTERV
-  for (i = 0; i < OP_COUNT; i++) 
-    MPI_Scatterv (buf1, counts, displs, MPI_INT, buf0, 
+  for (i = 0; i < OP_COUNT; i++)
+    MPI_Scatterv (buf1, counts, displs, MPI_INT, buf0,
 		  (rank < buf_size) ? rank : buf_size, MPI_INT, 0, comm);
 #endif
 
 #ifdef RUN_REDUCE
 #ifdef RUN_MAX
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_MAX, 0, comm);
 #endif
 
 #ifdef RUN_MIN
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_MIN, 0, comm);
 #endif
 
 #ifdef RUN_SUM
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_SUM, 0, comm);
 #endif
 
 #ifdef RUN_PROD
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_PROD, 0, comm);
 #endif
 
 #ifdef RUN_LAND
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_LAND, 0, comm);
 #endif
 
 #ifdef RUN_BAND
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_BAND, 0, comm);
 #endif
 
 #ifdef RUN_LOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_LOR, 0, comm);
 #endif
 
 #ifdef RUN_BOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_BOR, 0, comm);
 #endif
 
 #ifdef RUN_LXOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_LXOR, 0, comm);
 #endif
 
 #ifdef RUN_BXOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (buf0, buf1, buf_size, MPI_INT, MPI_BXOR, 0, comm);
 #endif
 
 #ifdef RUN_MAXLOC
   if (nprocs > 1)
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_2INT, MPI_MAXLOC, 0, comm);
   else
     fprintf (stderr, "Not enough tasks for MAXLOC test\n");
@@ -243,14 +243,14 @@ main (int argc, char **argv)
 
 #ifdef RUN_MINLOC
   if (nprocs > 1)
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce (buf0, buf1, buf_size, MPI_2INT, MPI_MINLOC, 0, comm);
   else
     fprintf (stderr, "Not enough tasks for MINLOC test\n");
 #endif
 
 #ifdef RUN_USEROP
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce (a, answer, buf_size, ctype, user_op, 0, comm);
 #endif
 #endif
@@ -258,7 +258,7 @@ main (int argc, char **argv)
   if (rank != 0) {
     buf1 = (int *) malloc (buf_size * nprocs * sizeof(int));
     assert (buf1);
-    for (i = 0; i < buf_size * nprocs; i++) 
+    for (i = 0; i < buf_size * nprocs; i++)
       buf1[i] = i;
 
     displs = (int *) malloc (nprocs * sizeof(int));
@@ -279,82 +279,82 @@ main (int argc, char **argv)
   }
 
 #ifdef RUN_ALLGATHER
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allgather (buf0, buf_size, MPI_INT, buf1, buf_size, MPI_INT, comm);
 #endif
 
 #ifdef RUN_ALLTOALL
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Alltoall (buf1, buf_size, MPI_INT, buf0, buf_size, MPI_INT, comm);
 #endif
 
 #ifdef RUN_ALLGATHERV
-  for (i = 0; i < OP_COUNT; i++) 
-    MPI_Allgatherv (buf0, 
-		    (rank < buf_size) ? rank : buf_size, 
+  for (i = 0; i < OP_COUNT; i++)
+    MPI_Allgatherv (buf0,
+		    (rank < buf_size) ? rank : buf_size,
 		    MPI_INT, buf1, rcounts, displs, MPI_INT, comm);
 #endif
 
 #ifdef RUN_ALLTOALLV
-  for (i = 0; i < OP_COUNT; i++) 
-    MPI_Alltoallv (buf1, alltoallvcounts, displs, MPI_INT, 
+  for (i = 0; i < OP_COUNT; i++)
+    MPI_Alltoallv (buf1, alltoallvcounts, displs, MPI_INT,
 		   buf0, alltoallvcounts, displs, MPI_INT, comm);
 #endif
 
 #ifdef RUN_ALLREDUCE
 #ifdef RUN_MAX
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (buf0, buf1, buf_size, MPI_INT, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_MAXLOC
   if (nprocs > 1)
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_2INT, MPI_MAXLOC, comm);
   else
     fprintf (stderr, "Not enough tasks for MAXLOC test\n");
@@ -362,72 +362,72 @@ main (int argc, char **argv)
 
 #ifdef RUN_MINLOC
   if (nprocs > 1)
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Allreduce (buf0, buf1, buf_size, MPI_2INT, MPI_MINLOC, comm);
   else
     fprintf (stderr, "Not enough tasks for MINLOC test\n");
 #endif
 
 #ifdef RUN_USEROP
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Allreduce (a, answer, buf_size, ctype, user_op, comm);
 #endif
 #endif
 
 #ifdef RUN_REDUCE_SCATTER
 #ifdef RUN_MAX
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_INT, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_MAXLOC
   if (nprocs > 1)
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_2INT, MPI_MAXLOC, comm);
   else
     fprintf (stderr, "Not enough tasks for MAXLOC test\n");
@@ -435,72 +435,72 @@ main (int argc, char **argv)
 
 #ifdef RUN_MINLOC
   if (nprocs > 1)
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Reduce_scatter (buf0, buf1, rcounts, MPI_2INT, MPI_MINLOC, comm);
   else
     fprintf (stderr, "Not enough tasks for MINLOC test\n");
 #endif
 
 #ifdef RUN_USEROP
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Reduce_scatter (a, answer, rcounts, ctype, user_op, comm);
 #endif
 #endif
 
 #ifdef RUN_SCAN
 #ifdef RUN_MAX
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_MAX, comm);
 #endif
 
 #ifdef RUN_MIN
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_MIN, comm);
 #endif
 
 #ifdef RUN_SUM
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_SUM, comm);
 #endif
 
 #ifdef RUN_PROD
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_PROD, comm);
 #endif
 
 #ifdef RUN_LAND
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_LAND, comm);
 #endif
 
 #ifdef RUN_BAND
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_BAND, comm);
 #endif
 
 #ifdef RUN_LOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_LOR, comm);
 #endif
 
 #ifdef RUN_BOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_BOR, comm);
 #endif
 
 #ifdef RUN_LXOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_LXOR, comm);
 #endif
 
 #ifdef RUN_BXOR
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (buf0, buf1, buf_size, MPI_INT, MPI_BXOR, comm);
 #endif
 
 #ifdef RUN_MAXLOC
   if (nprocs > 1)
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_2INT, MPI_MAXLOC, comm);
   else
     fprintf (stderr, "Not enough tasks for MAXLOC test\n");
@@ -508,14 +508,14 @@ main (int argc, char **argv)
 
 #ifdef RUN_MINLOC
   if (nprocs > 1)
-    for (i = 0; i < OP_COUNT; i++) 
+    for (i = 0; i < OP_COUNT; i++)
       MPI_Scan (buf0, buf1, buf_size, MPI_2INT, MPI_MINLOC, comm);
   else
     fprintf (stderr, "Not enough tasks for MINLOC test\n");
 #endif
 
 #ifdef RUN_USEROP
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Scan (a, answer, buf_size, ctype, user_op, comm);
 #endif
 #endif
diff --git a/teshsuite/smpi/isp/umpire/errhandler-no-error.c b/teshsuite/smpi/isp/umpire/errhandler-no-error.c
index 10239d4cb9..5cc6783fa7 100644
--- a/teshsuite/smpi/isp/umpire/errhandler-no-error.c
+++ b/teshsuite/smpi/isp/umpire/errhandler-no-error.c
@@ -58,17 +58,17 @@ main (int argc, char **argv)
 
   MPI_Barrier (comm);
 
-  for (i = 0; i < ERRHANDLER_COUNT; i++) 
+  for (i = 0; i < ERRHANDLER_COUNT; i++)
     MPI_Errhandler_create (myErrhandler, &newerrhandler[i]);
 
-  for (i = 0; i < ERRHANDLER_COUNT; i++) 
+  for (i = 0; i < ERRHANDLER_COUNT; i++)
     MPI_Errhandler_free (&newerrhandler[i]);
 
   MPI_Barrier (comm);
 
   /* now with an alias... */
 
-  for (i = 0; i < ERRHANDLER_COUNT; i++) 
+  for (i = 0; i < ERRHANDLER_COUNT; i++)
     MPI_Errhandler_create (myErrhandler, &newerrhandler[i]);
 
   for (i = 0; i < ERRHANDLER_COUNT; i++) {
diff --git a/teshsuite/smpi/isp/umpire/errhandler-no-free.c b/teshsuite/smpi/isp/umpire/errhandler-no-free.c
index 2d55368276..488a1df238 100644
--- a/teshsuite/smpi/isp/umpire/errhandler-no-free.c
+++ b/teshsuite/smpi/isp/umpire/errhandler-no-free.c
@@ -57,7 +57,7 @@ main (int argc, char **argv)
 
   MPI_Barrier (comm);
 
-  for (i = 0; i < ERRHANDLER_COUNT; i++) 
+  for (i = 0; i < ERRHANDLER_COUNT; i++)
     MPI_Errhandler_create (myErrhandler, &newerrhandler[i]);
 
   MPI_Barrier (comm);
diff --git a/teshsuite/smpi/isp/umpire/group-no-error-exhaustive.c b/teshsuite/smpi/isp/umpire/group-no-error-exhaustive.c
index 01e7b75c63..fd07b400d0 100644
--- a/teshsuite/smpi/isp/umpire/group-no-error-exhaustive.c
+++ b/teshsuite/smpi/isp/umpire/group-no-error-exhaustive.c
@@ -27,8 +27,8 @@ main (int argc, char **argv)
   int namelen = 128;
   int i;
   int ranks[2], ranges[1][3];
-  MPI_Group newgroup[GROUP_CONSTRUCTOR_COUNT]; 
-  MPI_Group newgroup2[GROUP_CONSTRUCTOR_COUNT]; 
+  MPI_Group newgroup[GROUP_CONSTRUCTOR_COUNT];
+  MPI_Group newgroup2[GROUP_CONSTRUCTOR_COUNT];
   MPI_Comm temp;
   MPI_Comm intercomm = MPI_COMM_NULL;
 
@@ -58,16 +58,16 @@ main (int argc, char **argv)
       MPI_Comm_group (MPI_COMM_WORLD, &newgroup[0]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 1)
-      MPI_Group_incl (newgroup[0], 2, ranks, &newgroup[1]);    
+      MPI_Group_incl (newgroup[0], 2, ranks, &newgroup[1]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 2)
       MPI_Group_excl (newgroup[0], 2, ranks, &newgroup[2]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 3)
-      MPI_Group_range_incl (newgroup[0], 1, ranges, &newgroup[3]);    
+      MPI_Group_range_incl (newgroup[0], 1, ranges, &newgroup[3]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 4)
-      MPI_Group_range_excl (newgroup[0], 1, ranges, &newgroup[4]);    
+      MPI_Group_range_excl (newgroup[0], 1, ranges, &newgroup[4]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 5)
       MPI_Group_union (newgroup[1], newgroup[3], &newgroup[5]);
@@ -83,7 +83,7 @@ main (int argc, char **argv)
       MPI_Comm_split (MPI_COMM_WORLD, rank % 3, nprocs - rank, &temp);
 
       if (rank % 3) {
-	MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, 
+	MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD,
 			      (((nprocs % 3) == 2) && ((rank % 3) == 2)) ?
 			      nprocs - 1 : nprocs - (rank % 3) - (nprocs % 3),
 			      INTERCOMM_CREATE_TAG, &intercomm);
@@ -98,7 +98,7 @@ main (int argc, char **argv)
 
       MPI_Comm_free (&temp);
     }
-      
+
     for (i = 0; i < GROUP_CONSTRUCTOR_COUNT; i++)
       MPI_Group_free (&newgroup[i]);
 
@@ -109,16 +109,16 @@ main (int argc, char **argv)
       MPI_Comm_group (MPI_COMM_WORLD, &newgroup[0]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 1)
-      MPI_Group_incl (newgroup[0], 2, ranks, &newgroup[1]);    
+      MPI_Group_incl (newgroup[0], 2, ranks, &newgroup[1]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 2)
       MPI_Group_excl (newgroup[0], 2, ranks, &newgroup[2]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 3)
-      MPI_Group_range_incl (newgroup[0], 1, ranges, &newgroup[3]);    
+      MPI_Group_range_incl (newgroup[0], 1, ranges, &newgroup[3]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 4)
-      MPI_Group_range_excl (newgroup[0], 1, ranges, &newgroup[4]);    
+      MPI_Group_range_excl (newgroup[0], 1, ranges, &newgroup[4]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 5)
       MPI_Group_union (newgroup[1], newgroup[3], &newgroup[5]);
@@ -134,7 +134,7 @@ main (int argc, char **argv)
       MPI_Comm_split (MPI_COMM_WORLD, rank % 3, nprocs - rank, &temp);
 
       if (rank % 3) {
-	MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, 
+	MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD,
 			      (((nprocs % 3) == 2) && ((rank % 3) == 2)) ?
 			      nprocs - 1 : nprocs - (rank % 3) - (nprocs % 3),
 			      INTERCOMM_CREATE_TAG, &intercomm);
@@ -149,7 +149,7 @@ main (int argc, char **argv)
 
       MPI_Comm_free (&temp);
     }
-      
+
     for (i = 0; i < GROUP_CONSTRUCTOR_COUNT; i++) {
       newgroup2[i] = newgroup[i];
       MPI_Group_free (&newgroup2[i]);
diff --git a/teshsuite/smpi/isp/umpire/group-no-free-exhaustive.c b/teshsuite/smpi/isp/umpire/group-no-free-exhaustive.c
index d2c9a7442a..9956ad174b 100644
--- a/teshsuite/smpi/isp/umpire/group-no-free-exhaustive.c
+++ b/teshsuite/smpi/isp/umpire/group-no-free-exhaustive.c
@@ -27,8 +27,8 @@ main (int argc, char **argv)
   int namelen = 128;
   int i;
   int ranks[2], ranges[1][3];
-  MPI_Group newgroup[GROUP_CONSTRUCTOR_COUNT]; 
-  MPI_Group newgroup2[GROUP_CONSTRUCTOR_COUNT]; 
+  MPI_Group newgroup[GROUP_CONSTRUCTOR_COUNT];
+  MPI_Group newgroup2[GROUP_CONSTRUCTOR_COUNT];
   MPI_Comm temp;
   MPI_Comm intercomm = MPI_COMM_NULL;
 
@@ -58,16 +58,16 @@ main (int argc, char **argv)
       MPI_Comm_group (MPI_COMM_WORLD, &newgroup[0]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 1)
-      MPI_Group_incl (newgroup[0], 2, ranks, &newgroup[1]);    
+      MPI_Group_incl (newgroup[0], 2, ranks, &newgroup[1]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 2)
       MPI_Group_excl (newgroup[0], 2, ranks, &newgroup[2]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 3)
-      MPI_Group_range_incl (newgroup[0], 1, ranges, &newgroup[3]);    
+      MPI_Group_range_incl (newgroup[0], 1, ranges, &newgroup[3]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 4)
-      MPI_Group_range_excl (newgroup[0], 1, ranges, &newgroup[4]);    
+      MPI_Group_range_excl (newgroup[0], 1, ranges, &newgroup[4]);
 
     if (GROUP_CONSTRUCTOR_COUNT > 5)
       MPI_Group_union (newgroup[1], newgroup[3], &newgroup[5]);
@@ -83,7 +83,7 @@ main (int argc, char **argv)
       MPI_Comm_split (MPI_COMM_WORLD, rank % 3, nprocs - rank, &temp);
 
       if (rank % 3) {
-	MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, 
+	MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD,
 			      (((nprocs % 3) == 2) && ((rank % 3) == 2)) ?
 			      nprocs - 1 : nprocs - (rank % 3) - (nprocs % 3),
 			      INTERCOMM_CREATE_TAG, &intercomm);
@@ -98,7 +98,7 @@ main (int argc, char **argv)
 
       MPI_Comm_free (&temp);
     }
-  }      
+  }
 
   MPI_Barrier (comm);
 
diff --git a/teshsuite/smpi/isp/umpire/hello.c b/teshsuite/smpi/isp/umpire/hello.c
index 20832d3eb2..37257e95f0 100644
--- a/teshsuite/smpi/isp/umpire/hello.c
+++ b/teshsuite/smpi/isp/umpire/hello.c
@@ -1,8 +1,8 @@
-/* -*- C -*- 
+/* -*- C -*-
 
    @PROLOGUE@
 
-   ----- 
+   -----
 
    Jeffrey Vetter vetter@llnl.gov
    Center for Applied Scientific Computing, LLNL
diff --git a/teshsuite/smpi/isp/umpire/intercomm_create-deadlock.c b/teshsuite/smpi/isp/umpire/intercomm_create-deadlock.c
index afe7232184..5f05b58005 100644
--- a/teshsuite/smpi/isp/umpire/intercomm_create-deadlock.c
+++ b/teshsuite/smpi/isp/umpire/intercomm_create-deadlock.c
@@ -66,12 +66,12 @@ main (int argc, char **argv)
       }
 
       MPI_Comm_free (&temp);
-      
+
       if (intercomm != MPI_COMM_NULL) {
 	MPI_Comm_size (intercomm, &dnprocs);
 	MPI_Comm_rank (intercomm, &drank);
  	MPI_Comm_remote_size (intercomm, &rnprocs);
-     
+
 	if (rnprocs > drank) {
 	  if (rank % 2) {
 	    memset (buf1, 1, buf_size);
@@ -84,7 +84,7 @@ main (int argc, char **argv)
 	    memset (buf0, 0, buf_size);
 	
 	    MPI_Send (buf0, buf_size, MPI_INT, drank, 0, intercomm);
-	    
+	
 	    MPI_Recv (buf1, buf_size, MPI_INT, drank, 0, intercomm, &status);
 	  }
 	}
diff --git a/teshsuite/smpi/isp/umpire/intercomm_create-deadlock2.c b/teshsuite/smpi/isp/umpire/intercomm_create-deadlock2.c
index 332dce04b7..3dbfc76be1 100644
--- a/teshsuite/smpi/isp/umpire/intercomm_create-deadlock2.c
+++ b/teshsuite/smpi/isp/umpire/intercomm_create-deadlock2.c
@@ -48,7 +48,7 @@ main (int argc, char **argv)
       rleader = ((rank + nprocs) % 2) ?  nprocs - 2 : nprocs - 1;
 
       if ((trank == 0) && (rank % 2)) {
-	MPI_Recv (buf0, buf_size, MPI_INT, 
+	MPI_Recv (buf0, buf_size, MPI_INT,
 		  rleader, 0, MPI_COMM_WORLD, &status);
       }
 
@@ -65,12 +65,12 @@ main (int argc, char **argv)
       }
 
       MPI_Comm_free (&temp);
-      
+
       if (intercomm != MPI_COMM_NULL) {
 	MPI_Comm_size (intercomm, &dnprocs);
 	MPI_Comm_rank (intercomm, &drank);
  	MPI_Comm_remote_size (intercomm, &rnprocs);
-     
+
 	if (rnprocs > drank) {
 	  if (rank % 2) {
 	    memset (buf1, 1, buf_size);
@@ -83,7 +83,7 @@ main (int argc, char **argv)
 	    memset (buf0, 0, buf_size);
 	
 	    MPI_Send (buf0, buf_size, MPI_INT, drank, 0, intercomm);
-	    
+	
 	    MPI_Recv (buf1, buf_size, MPI_INT, drank, 0, intercomm, &status);
 	  }
 	}
diff --git a/teshsuite/smpi/isp/umpire/intercomm_create-deadlock3.c b/teshsuite/smpi/isp/umpire/intercomm_create-deadlock3.c
index d546995d5e..908b8873f4 100644
--- a/teshsuite/smpi/isp/umpire/intercomm_create-deadlock3.c
+++ b/teshsuite/smpi/isp/umpire/intercomm_create-deadlock3.c
@@ -61,12 +61,12 @@ main (int argc, char **argv)
       }
 
       MPI_Comm_free (&temp);
-      
+
       if (intercomm != MPI_COMM_NULL) {
 	MPI_Comm_size (intercomm, &dnprocs);
 	MPI_Comm_rank (intercomm, &drank);
  	MPI_Comm_remote_size (intercomm, &rnprocs);
-     
+
 	if (rnprocs > drank) {
 	  if (rank % 2) {
 	    memset (buf1, 1, buf_size);
@@ -79,7 +79,7 @@ main (int argc, char **argv)
 	    memset (buf0, 0, buf_size);
 	
 	    MPI_Send (buf0, buf_size, MPI_INT, drank, 0, intercomm);
-	    
+	
 	    MPI_Recv (buf1, buf_size, MPI_INT, drank, 0, intercomm, &status);
 	  }
 	}
diff --git a/teshsuite/smpi/isp/umpire/intercomm_create-deadlock4.c b/teshsuite/smpi/isp/umpire/intercomm_create-deadlock4.c
index 8c45d54baa..c84e94fb82 100644
--- a/teshsuite/smpi/isp/umpire/intercomm_create-deadlock4.c
+++ b/teshsuite/smpi/isp/umpire/intercomm_create-deadlock4.c
@@ -69,12 +69,12 @@ main (int argc, char **argv)
       }
 
       MPI_Comm_free (&temp);
-      
+
       if (intercomm != MPI_COMM_NULL) {
 	MPI_Comm_size (intercomm, &dnprocs);
 	MPI_Comm_rank (intercomm, &drank);
  	MPI_Comm_remote_size (intercomm, &rnprocs);
-     
+
 	if (rnprocs > drank) {
 	  if (rank % 2) {
 	    memset (buf1, 1, buf_size);
@@ -87,7 +87,7 @@ main (int argc, char **argv)
 	    memset (buf0, 0, buf_size);
 	
 	    MPI_Send (buf0, buf_size, MPI_INT, drank, 0, intercomm);
-	    
+	
 	    MPI_Recv (buf1, buf_size, MPI_INT, drank, 0, intercomm, &status);
 	  }
 	}
diff --git a/teshsuite/smpi/isp/umpire/intercomm_create-no-error.c b/teshsuite/smpi/isp/umpire/intercomm_create-no-error.c
index 700edce621..e85dfcecde 100644
--- a/teshsuite/smpi/isp/umpire/intercomm_create-no-error.c
+++ b/teshsuite/smpi/isp/umpire/intercomm_create-no-error.c
@@ -43,11 +43,11 @@ main (int argc, char **argv)
     /* need lots of stuff for this constructor... */
     MPI_Comm_split (MPI_COMM_WORLD, rank % 2, nprocs - rank, &temp);
 
-    MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, 
-			  nprocs - 
+    MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD,
+			  nprocs -
 			  ((rank % 2) ? 2 - (nprocs % 2) : 1 + (nprocs % 2)),
 			  INTERCOMM_CREATE_TAG, &intercomm);
-      
+
     MPI_Comm_free (&intercomm);
 
     MPI_Comm_free (&temp);
diff --git a/teshsuite/smpi/isp/umpire/intercomm_merge-deadlock.c b/teshsuite/smpi/isp/umpire/intercomm_merge-deadlock.c
index 6d35a0aa01..f6dfc567fb 100644
--- a/teshsuite/smpi/isp/umpire/intercomm_merge-deadlock.c
+++ b/teshsuite/smpi/isp/umpire/intercomm_merge-deadlock.c
@@ -46,10 +46,10 @@ main (int argc, char **argv)
       MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, rleader,
 			    INTERCOMM_CREATE_TAG, &intercomm);
       MPI_Comm_free (&temp);
-      
+
       if (intercomm != MPI_COMM_NULL) {
 	/* need to make a different split communicator temporarily... */
-	MPI_Comm_split (MPI_COMM_WORLD, 
+	MPI_Comm_split (MPI_COMM_WORLD,
 			rank < nprocs/2, nprocs - rank, &temp);
 
 	if (temp != MPI_COMM_NULL) {
@@ -58,7 +58,7 @@ main (int argc, char **argv)
 	  MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, rleader,
 				INTERCOMM_CREATE_TAG, &intercomm2);
 	  MPI_Comm_free (&temp);
-      
+
 	  if (intercomm2 != MPI_COMM_NULL) {
 	    if (rank < nprocs/2) {
 	      MPI_Intercomm_merge (intercomm2, rank < nprocs/2, &comm2);
@@ -89,7 +89,7 @@ main (int argc, char **argv)
 		  memset (buf0, 0, buf_size);
 	
 		  MPI_Send (buf0, buf_size, MPI_INT, 1, 0, comm);
-	    
+	
 		  MPI_Recv (buf1, buf_size, MPI_INT, 1, 0, comm, &status);
 		}
 	      }
@@ -112,7 +112,7 @@ main (int argc, char **argv)
 		  memset (buf0, 0, buf_size);
 	
 		  MPI_Send (buf0, buf_size, MPI_INT, 1, 0, comm2);
-	    
+	
 		  MPI_Recv (buf1, buf_size, MPI_INT, 1, 0, comm2, &status);
 		}
 	      }
diff --git a/teshsuite/smpi/isp/umpire/irecv-isend-ok2.c b/teshsuite/smpi/isp/umpire/irecv-isend-ok2.c
index 80a63886c0..b5f60b2c6e 100644
--- a/teshsuite/smpi/isp/umpire/irecv-isend-ok2.c
+++ b/teshsuite/smpi/isp/umpire/irecv-isend-ok2.c
@@ -42,7 +42,7 @@ main (int argc, char **argv)
       int src = (rank == 0) ? (nprocs - 1) : (rank - 1);
       memset (buf0, rank, buf_size);
       memset (buf1, rank, buf_size);
-      
+
       for (i = 0; i < req_count; i++) {
 	MPI_Irecv (buf0, buf_size, MPI_INT, src, 0, MPI_COMM_WORLD, &req1);
 	MPI_Isend (buf1, buf_size, MPI_INT, dest, 0, MPI_COMM_WORLD, &req2);
diff --git a/teshsuite/smpi/isp/umpire/lost-request2.c b/teshsuite/smpi/isp/umpire/lost-request2.c
index 0953cb58f5..614566448f 100644
--- a/teshsuite/smpi/isp/umpire/lost-request2.c
+++ b/teshsuite/smpi/isp/umpire/lost-request2.c
@@ -34,7 +34,7 @@ int main
   int done;
 
   rc = MPI_Init(&argc,&argv);
-  if (rc != MPI_SUCCESS) 
+  if (rc != MPI_SUCCESS)
     {
       printf ("Error starting MPI program. Terminating.\n");
       MPI_Abort(MPI_COMM_WORLD, rc);
@@ -45,7 +45,7 @@ int main
   MPI_Get_processor_name (processor_name, &namelen);
   printf ("(%d) is alive on %s\n", rank, processor_name);
   fflush (stdout);
-  
+
   /* Must be multiple of two for this test */
   if ((numtasks & 1) != 0)
     {
@@ -86,7 +86,7 @@ int main
 	}
       else if (split_rank == 1)
 	{
-	  rc = MPI_Irecv (&recv_int, 1, MPI_INT, 0, 0, split_comm, 
+	  rc = MPI_Irecv (&recv_int, 1, MPI_INT, 0, 0, split_comm,
 			  &request);
 	  if (rc != MPI_SUCCESS)
 	    {
diff --git a/teshsuite/smpi/isp/umpire/lost-request3.c b/teshsuite/smpi/isp/umpire/lost-request3.c
index 7d4875c16d..3be717c059 100644
--- a/teshsuite/smpi/isp/umpire/lost-request3.c
+++ b/teshsuite/smpi/isp/umpire/lost-request3.c
@@ -54,7 +54,7 @@ main (int argc, char **argv)
 	memset (buf, 1, buf_size * REQS_PER_ITERATION);
 
 	for (j = 0; j < REQS_PER_ITERATION; j++) {
-	  MPI_Isend (&buf[j*buf_size], buf_size, MPI_INT, 
+	  MPI_Isend (&buf[j*buf_size], buf_size, MPI_INT,
 		     1, j, MPI_COMM_WORLD, &req[j]);
 	}
 
@@ -68,7 +68,7 @@ main (int argc, char **argv)
 	memset (buf, 2, buf_size * REQS_PER_ITERATION);
 
 	for (j = 0; j < REQS_PER_ITERATION; j++) {
-	  MPI_Irecv (&buf[j*buf_size], buf_size, MPI_INT, 
+	  MPI_Irecv (&buf[j*buf_size], buf_size, MPI_INT,
 		     0, j, MPI_COMM_WORLD, &req[j]);
 	}
 
diff --git a/teshsuite/smpi/isp/umpire/no-error-any_src.c b/teshsuite/smpi/isp/umpire/no-error-any_src.c
index 48aa54e58f..3d408f672e 100644
--- a/teshsuite/smpi/isp/umpire/no-error-any_src.c
+++ b/teshsuite/smpi/isp/umpire/no-error-any_src.c
@@ -34,7 +34,7 @@ main (int argc, char **argv)
   else {
     if (rank == 0) {
       for (i = 1; i < nprocs; i++) {
-	MPI_Recv (buf, buf_size, MPI_INT, 
+	MPI_Recv (buf, buf_size, MPI_INT,
 		  MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
       }
     }
diff --git a/teshsuite/smpi/isp/umpire/no-error-derived-comms.c b/teshsuite/smpi/isp/umpire/no-error-derived-comms.c
index 4f5a479eb1..4f64a2776d 100644
--- a/teshsuite/smpi/isp/umpire/no-error-derived-comms.c
+++ b/teshsuite/smpi/isp/umpire/no-error-derived-comms.c
@@ -19,7 +19,7 @@
 #define DCOMM_CALL_COUNT  7 /* MPI_Cart_create; MPI_Cart_sub;
 			       MPI_Comm_create; MPI_Comm_dup;
 			       MPI_Comm_split; MPI_Graph_create;
-			       and MPI_Intercomm_merge; store 
+			       and MPI_Intercomm_merge; store
 			       MPI_Intercomm_create separately... */
 #define TWOD     2
 #define GRAPH_SZ 4
@@ -105,7 +105,7 @@ main (int argc, char **argv)
       temp = dcomms[2];
     }
     if (rank % 3) {
-      MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, 
+      MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD,
 			    (((nprocs % 3) == 2) && ((rank % 3) == 2)) ?
 			    nprocs - 1 : nprocs - (rank % 3) - (nprocs % 3),
 			    INTERCOMM_CREATE_TAG, &intercomm);
@@ -130,7 +130,7 @@ main (int argc, char **argv)
     if (DCOMM_CALL_COUNT > 4) {
 #ifdef RUN_GRAPH_CREATE
       /* create the graph on p.268 MPI: The Complete Reference... */
-      MPI_Graph_create (MPI_COMM_WORLD, GRAPH_SZ, 
+      MPI_Graph_create (MPI_COMM_WORLD, GRAPH_SZ,
 			graph_index, graph_edges, 1, &dcomms[4]);
 #else
       dcomms[4] = MPI_COMM_NULL;
@@ -177,7 +177,7 @@ main (int argc, char **argv)
       /* create an intercommunicator and merge it... */
       if (rank % 3) {
 #ifndef RUN_INTERCOMM_CREATE
-	MPI_Intercomm_create (dcomms[2], 0, MPI_COMM_WORLD, 
+	MPI_Intercomm_create (dcomms[2], 0, MPI_COMM_WORLD,
 			      (((nprocs % 3) == 2) && ((rank % 3) == 2)) ?
 			      nprocs - 1 : nprocs - (rank % 3) - (nprocs % 3),
 			      INTERCOMM_CREATE_TAG, &intercomm);
@@ -192,7 +192,7 @@ main (int argc, char **argv)
       }
       else {
 	dcomms[6] = MPI_COMM_NULL;
-      }    
+      }
 #ifndef RUN_INTERCOMM_CREATE
 #ifndef RUN_COMM_SPLIT
       if (dcomms[2] != MPI_COMM_NULL)
@@ -268,7 +268,7 @@ main (int argc, char **argv)
       if (dnprocs[i] > 1) {
 	if (drank[i] == 0) {
 	  for (j = 1; j < dnprocs[i]; j++) {
-	    MPI_Recv (buf, buf_size, MPI_INT, 
+	    MPI_Recv (buf, buf_size, MPI_INT,
 		      MPI_ANY_SOURCE, 0, dcomms[i], &status);
 	  }
 	}
@@ -284,7 +284,7 @@ main (int argc, char **argv)
     /* do any source receives on the intercomm... */
     if ((rank % 3) == 1) {
       for (j = 0; j < intersize; j++) {
-	MPI_Recv (buf, buf_size, MPI_INT, 
+	MPI_Recv (buf, buf_size, MPI_INT,
 		  MPI_ANY_SOURCE, 0, intercomm, &status);
       }
     }
diff --git a/teshsuite/smpi/isp/umpire/no-error-interleaved-isend.c b/teshsuite/smpi/isp/umpire/no-error-interleaved-isend.c
index e40a9c4d72..4b534685f1 100644
--- a/teshsuite/smpi/isp/umpire/no-error-interleaved-isend.c
+++ b/teshsuite/smpi/isp/umpire/no-error-interleaved-isend.c
@@ -46,10 +46,10 @@ main (int argc, char **argv)
       for (i = rank; i < buf_size; i = i + 2)
 	buf[i] = i;
 
-      MPI_Isend (&buf[rank], 1, 
+      MPI_Isend (&buf[rank], 1,
 		 strided_type, (rank + 1) % 2, 0, comm, &req);
 
-      MPI_Recv (&buf[(rank + 1) % 2], 1, 
+      MPI_Recv (&buf[(rank + 1) % 2], 1,
 		strided_type, (rank + 1) % 2, 0, comm, &status);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/no-error-persistent-all-completions.c b/teshsuite/smpi/isp/umpire/no-error-persistent-all-completions.c
index eb77c9cdae..6d9648fed9 100644
--- a/teshsuite/smpi/isp/umpire/no-error-persistent-all-completions.c
+++ b/teshsuite/smpi/isp/umpire/no-error-persistent-all-completions.c
@@ -67,7 +67,7 @@ main (int argc, char **argv)
 	  MPI_Start (&aReq[j]);
 	}
       }
-    
+
       /* complete the sends */
       switch (k/2) {
       case 0:
@@ -146,7 +146,7 @@ main (int argc, char **argv)
     else if (rank == 1) {
       /* set up receives for all of the sends */
       for (j = 0; j < 2; j++) {
-	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE, 
+	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE,
 		   MPI_INT, 0, j, comm, &aReq[j]);
       }
       /* complete all of the receives... */
diff --git a/teshsuite/smpi/isp/umpire/no-error-persistent-test.c b/teshsuite/smpi/isp/umpire/no-error-persistent-test.c
index 930178c5ac..c8ac0d924f 100644
--- a/teshsuite/smpi/isp/umpire/no-error-persistent-test.c
+++ b/teshsuite/smpi/isp/umpire/no-error-persistent-test.c
@@ -66,7 +66,7 @@ main (int argc, char **argv)
 	  MPI_Start (&aReq[j]);
 	}
       }
-    
+
       /* complete the sends */
       if (k < 2) {
 	/* use MPI_Test */
@@ -88,7 +88,7 @@ main (int argc, char **argv)
     else if (rank == 1) {
       /* set up receives for all of the sends */
       for (j = 0; j < 2; j++) {
-	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE, 
+	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE,
 		   MPI_INT, 0, j, comm, &aReq[j]);
       }
       /* complete all of the receives... */
diff --git a/teshsuite/smpi/isp/umpire/no-error-persistent-testpartial.c b/teshsuite/smpi/isp/umpire/no-error-persistent-testpartial.c
index fd2a82ef49..7d5fd62138 100644
--- a/teshsuite/smpi/isp/umpire/no-error-persistent-testpartial.c
+++ b/teshsuite/smpi/isp/umpire/no-error-persistent-testpartial.c
@@ -66,7 +66,7 @@ main (int argc, char **argv)
 	  MPI_Start (&aReq[j]);
 	}
       }
-    
+
       /* complete the sends */
       if (k < 2) {
 	/* use MPI_Testany */
@@ -92,7 +92,7 @@ main (int argc, char **argv)
     else if (rank == 1) {
       /* set up receives for all of the sends */
       for (j = 0; j < 2; j++) {
-	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE, 
+	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE,
 		   MPI_INT, 0, j, comm, &aReq[j]);
       }
       /* complete all of the receives... */
diff --git a/teshsuite/smpi/isp/umpire/no-error-persistent-waitpartial.c b/teshsuite/smpi/isp/umpire/no-error-persistent-waitpartial.c
index 1ff4276fe4..f8abc6b637 100644
--- a/teshsuite/smpi/isp/umpire/no-error-persistent-waitpartial.c
+++ b/teshsuite/smpi/isp/umpire/no-error-persistent-waitpartial.c
@@ -66,7 +66,7 @@ main (int argc, char **argv)
 	  MPI_Start (&aReq[j]);
 	}
       }
-    
+
       /* complete the sends */
       if (k < 2) {
 	/* use MPI_Waitany */
@@ -85,7 +85,7 @@ main (int argc, char **argv)
     else if (rank == 1) {
       /* set up receives for all of the sends */
       for (j = 0; j < 2; j++) {
-	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE, 
+	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE,
 		   MPI_INT, 0, j, comm, &aReq[j]);
       }
       /* complete all of the receives... */
diff --git a/teshsuite/smpi/isp/umpire/no-error-persistent.c b/teshsuite/smpi/isp/umpire/no-error-persistent.c
index 8b157b30d2..e9fe2a6d57 100644
--- a/teshsuite/smpi/isp/umpire/no-error-persistent.c
+++ b/teshsuite/smpi/isp/umpire/no-error-persistent.c
@@ -66,20 +66,20 @@ main (int argc, char **argv)
 	  MPI_Start (&aReq[j]);
 	}
       }
-    
+
       /* complete the sends */
       if (k < 2)
 	/* use MPI_Wait */
 	for (j = 0; j < 2; j++)
 	  MPI_Wait (&aReq[j], &aStatus[j]);
-      else 
+      else
 	/* use MPI_Waitall */
 	MPI_Waitall (2, aReq, aStatus);
     }
     else if (rank == 1) {
       /* set up receives for all of the sends */
       for (j = 0; j < 2; j++) {
-	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE, 
+	MPI_Irecv (&buf[j * BUF_SIZE], BUF_SIZE,
 		   MPI_INT, 0, j, comm, &aReq[j]);
       }
       /* complete all of the receives... */
diff --git a/teshsuite/smpi/isp/umpire/no-error-probe-any_src.c b/teshsuite/smpi/isp/umpire/no-error-probe-any_src.c
index 4ea1ef1cb0..a40c21c463 100644
--- a/teshsuite/smpi/isp/umpire/no-error-probe-any_src.c
+++ b/teshsuite/smpi/isp/umpire/no-error-probe-any_src.c
@@ -50,14 +50,14 @@ main (int argc, char **argv)
     {
       for (j = 0; j < 2; j++) {
 	MPI_Probe (MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
-      
+
 	if (status.MPI_SOURCE == 0)
 	  MPI_Recv (&i, 1, MPI_INT, 0, 0, MPI_COMM_WORLD, &status);
-	else 
+	else
 	  MPI_Recv (&x, 1, MPI_DOUBLE, 1, 0, MPI_COMM_WORLD, &status);
       }
     }
-      
+
   MPI_Barrier (MPI_COMM_WORLD);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/no-error-probe-any_tag.c b/teshsuite/smpi/isp/umpire/no-error-probe-any_tag.c
index 9be706c8c3..4bfc777a3a 100644
--- a/teshsuite/smpi/isp/umpire/no-error-probe-any_tag.c
+++ b/teshsuite/smpi/isp/umpire/no-error-probe-any_tag.c
@@ -47,18 +47,18 @@ main (int argc, char **argv)
 
       MPI_Wait (&req0, &status);
     }
-  else if (rank == 1) 
+  else if (rank == 1)
     {
       for (j = 0; j < 2; j++) {
 	MPI_Probe (0, MPI_ANY_TAG, MPI_COMM_WORLD, &status);
-      
+
 	if (status.MPI_TAG == 0)
 	  MPI_Recv (&i, 1, MPI_INT, 0, 0, MPI_COMM_WORLD, &status);
 	else
 	  MPI_Recv (&x, 1, MPI_DOUBLE, 0, 1, MPI_COMM_WORLD, &status);
       }
     }
-      
+
   MPI_Barrier (MPI_COMM_WORLD);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/no-error-probe.c b/teshsuite/smpi/isp/umpire/no-error-probe.c
index 159bfc33e4..ee5d4ec9f7 100644
--- a/teshsuite/smpi/isp/umpire/no-error-probe.c
+++ b/teshsuite/smpi/isp/umpire/no-error-probe.c
@@ -39,7 +39,7 @@ main (int argc, char **argv)
       MPI_Send (&i, 1, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
       MPI_Probe (1, 0, MPI_COMM_WORLD, &status);
-      
+
       MPI_Recv (&x, 1, MPI_DOUBLE, 1, 0, MPI_COMM_WORLD, &status);
     }
   else if (rank == 1)
@@ -52,7 +52,7 @@ main (int argc, char **argv)
 
       MPI_Send (&x, 1, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
     }
-      
+
   MPI_Barrier (MPI_COMM_WORLD);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/no-error-testall.c b/teshsuite/smpi/isp/umpire/no-error-testall.c
index da5cccae69..3f497c511b 100644
--- a/teshsuite/smpi/isp/umpire/no-error-testall.c
+++ b/teshsuite/smpi/isp/umpire/no-error-testall.c
@@ -40,7 +40,7 @@ main (int argc, char **argv)
 
       MPI_Irecv (buf1, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &reqs[1]);
 
-      while (!flag) 
+      while (!flag)
 	MPI_Testall (2, reqs, &flag, statuses);
 
       MPI_Send (buf1, buf_size, MPI_INT, 1, 1, MPI_COMM_WORLD);
@@ -53,7 +53,7 @@ main (int argc, char **argv)
 
       MPI_Isend (buf0, buf_size, MPI_INT, 2, 1, MPI_COMM_WORLD, &reqs[1]);
 
-      while (!flag) 
+      while (!flag)
 	MPI_Testall (2, reqs, &flag, statuses);
 
       MPI_Send (buf0, buf_size, MPI_INT, 0, 0, MPI_COMM_WORLD);
diff --git a/teshsuite/smpi/isp/umpire/no-error-vector-isend.c b/teshsuite/smpi/isp/umpire/no-error-vector-isend.c
index 0c1fc0950c..a3a86216b4 100644
--- a/teshsuite/smpi/isp/umpire/no-error-vector-isend.c
+++ b/teshsuite/smpi/isp/umpire/no-error-vector-isend.c
@@ -59,7 +59,7 @@ main (int argc, char **argv)
     for (i = 1; i < buf_size; i = i + 2)
       buf[i] = i;
 
-    MPI_Recv (&buf[(rank + 1) % 2], 1, 
+    MPI_Recv (&buf[(rank + 1) % 2], 1,
 	      strided_type, (rank + 1) % 2, 0, comm, &status);
 
     for (i = 0; i < buf_size; i++)
diff --git a/teshsuite/smpi/isp/umpire/no-error-wait-any_src.c b/teshsuite/smpi/isp/umpire/no-error-wait-any_src.c
index e33c2c9041..0ab83aa9ae 100644
--- a/teshsuite/smpi/isp/umpire/no-error-wait-any_src.c
+++ b/teshsuite/smpi/isp/umpire/no-error-wait-any_src.c
@@ -36,7 +36,7 @@ main (int argc, char **argv)
     {
       memset (buf0, 0, buf_size);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Send (buf0, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
@@ -47,7 +47,7 @@ main (int argc, char **argv)
     {
       memset (buf1, 1, buf_size);
 
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Wait (&req, &status);
diff --git a/teshsuite/smpi/isp/umpire/no-error-wait-any_src2.c b/teshsuite/smpi/isp/umpire/no-error-wait-any_src2.c
index 364993922d..59ed5d0036 100644
--- a/teshsuite/smpi/isp/umpire/no-error-wait-any_src2.c
+++ b/teshsuite/smpi/isp/umpire/no-error-wait-any_src2.c
@@ -34,7 +34,7 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 1, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf0, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &status);
diff --git a/teshsuite/smpi/isp/umpire/no-error-wait-any_src3.c b/teshsuite/smpi/isp/umpire/no-error-wait-any_src3.c
index db9980b172..7d17a6cc21 100644
--- a/teshsuite/smpi/isp/umpire/no-error-wait-any_src3.c
+++ b/teshsuite/smpi/isp/umpire/no-error-wait-any_src3.c
@@ -43,7 +43,7 @@ main (int argc, char **argv)
     {
       memset (buf0, 0, buf_size);
 
-       MPI_Irecv (buf1, buf_size, MPI_INT, 
+       MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Send (buf0, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD);
@@ -58,7 +58,7 @@ main (int argc, char **argv)
     }
   else if (rank == 1)
     {
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &req);
 
       MPI_Recv (buf0, buf_size, MPI_INT, 0, 0, MPI_COMM_WORLD, &status);
diff --git a/teshsuite/smpi/isp/umpire/no-error-wait-any_src4.c b/teshsuite/smpi/isp/umpire/no-error-wait-any_src4.c
index fd6b2a1cc9..d76749fe07 100644
--- a/teshsuite/smpi/isp/umpire/no-error-wait-any_src4.c
+++ b/teshsuite/smpi/isp/umpire/no-error-wait-any_src4.c
@@ -37,10 +37,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
@@ -65,7 +65,7 @@ main (int argc, char **argv)
     }
   else if (rank == 2)
     {
-      //sleep (60);      
+      //sleep (60);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 1, 1, MPI_COMM_WORLD, statuses);
     }
diff --git a/teshsuite/smpi/isp/umpire/no-error-waitall-any_src.c b/teshsuite/smpi/isp/umpire/no-error-waitall-any_src.c
index 2fe9b32212..2c7f144586 100644
--- a/teshsuite/smpi/isp/umpire/no-error-waitall-any_src.c
+++ b/teshsuite/smpi/isp/umpire/no-error-waitall-any_src.c
@@ -43,10 +43,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
diff --git a/teshsuite/smpi/isp/umpire/no-error-waitall-any_src2.c b/teshsuite/smpi/isp/umpire/no-error-waitall-any_src2.c
index f6d3e36de8..56c7e6cfd1 100644
--- a/teshsuite/smpi/isp/umpire/no-error-waitall-any_src2.c
+++ b/teshsuite/smpi/isp/umpire/no-error-waitall-any_src2.c
@@ -35,10 +35,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
diff --git a/teshsuite/smpi/isp/umpire/no-error-waitall-any_src3.c b/teshsuite/smpi/isp/umpire/no-error-waitall-any_src3.c
index bed4b5e6e2..835d2a560d 100644
--- a/teshsuite/smpi/isp/umpire/no-error-waitall-any_src3.c
+++ b/teshsuite/smpi/isp/umpire/no-error-waitall-any_src3.c
@@ -37,10 +37,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
diff --git a/teshsuite/smpi/isp/umpire/no-error-waitany-any_src.c b/teshsuite/smpi/isp/umpire/no-error-waitany-any_src.c
index 3aaa049ba2..d8a7e89560 100644
--- a/teshsuite/smpi/isp/umpire/no-error-waitany-any_src.c
+++ b/teshsuite/smpi/isp/umpire/no-error-waitany-any_src.c
@@ -39,11 +39,11 @@ main (int argc, char **argv)
     {
       MPI_Request reqs[3];
 
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 1, MPI_COMM_WORLD, &reqs[0]);
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 2, MPI_COMM_WORLD, &reqs[1]);
-      MPI_Irecv (buf2, buf_size, MPI_INT, 
+      MPI_Irecv (buf2, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 3, MPI_COMM_WORLD, &reqs[2]);
 
       for (i = 3; i > 0; i--) {
@@ -59,7 +59,7 @@ main (int argc, char **argv)
       memset (buf0, 1, buf_size);
 
       for (i = 3; i > 0; i--) {
-	MPI_Recv (&flipbit, 1, MPI_INT, 
+	MPI_Recv (&flipbit, 1, MPI_INT,
 		  MPI_ANY_SOURCE, i, MPI_COMM_WORLD, &status);
 	
 	MPI_Send (buf0, buf_size, MPI_INT, 0, i, MPI_COMM_WORLD);
diff --git a/teshsuite/smpi/isp/umpire/no-error-waitany-any_src2.c b/teshsuite/smpi/isp/umpire/no-error-waitany-any_src2.c
index 93d3582893..4a64b34b2c 100644
--- a/teshsuite/smpi/isp/umpire/no-error-waitany-any_src2.c
+++ b/teshsuite/smpi/isp/umpire/no-error-waitany-any_src2.c
@@ -43,10 +43,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitany (2, reqs, &done, &status);
diff --git a/teshsuite/smpi/isp/umpire/no-error-waitany-any_src3.c b/teshsuite/smpi/isp/umpire/no-error-waitany-any_src3.c
index 2a690bceff..a9e522b0e9 100644
--- a/teshsuite/smpi/isp/umpire/no-error-waitany-any_src3.c
+++ b/teshsuite/smpi/isp/umpire/no-error-waitany-any_src3.c
@@ -38,10 +38,10 @@ main (int argc, char **argv)
     }
   else if (rank == 0)
     {
-      MPI_Irecv (buf0, buf_size, MPI_INT, 
+      MPI_Irecv (buf0, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[0]);
 
-      MPI_Irecv (buf1, buf_size, MPI_INT, 
+      MPI_Irecv (buf1, buf_size, MPI_INT,
 		 MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &reqs[1]);
 
       MPI_Waitall (2, reqs, statuses);
@@ -66,7 +66,7 @@ main (int argc, char **argv)
     }
   else if (rank == 2)
     {
-      sleep (60);      
+      sleep (60);
 
       MPI_Recv (buf1, buf_size, MPI_INT, 1, 1, MPI_COMM_WORLD, statuses);
     }
diff --git a/teshsuite/smpi/isp/umpire/no-error3-any_src.c b/teshsuite/smpi/isp/umpire/no-error3-any_src.c
index c80985f93d..c3ba9bb7e4 100644
--- a/teshsuite/smpi/isp/umpire/no-error3-any_src.c
+++ b/teshsuite/smpi/isp/umpire/no-error3-any_src.c
@@ -42,9 +42,9 @@ main (int argc, char **argv)
     else if (rank == 1) {
       memset (buf1, 1, buf_size);
 
-      MPI_Recv (buf0, buf_size, MPI_INT, 
+      MPI_Recv (buf0, buf_size, MPI_INT,
 		MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
-      
+
       MPI_Send (buf1, buf_size, MPI_INT, 0, 0, MPI_COMM_WORLD);
     }
   }
diff --git a/teshsuite/smpi/isp/umpire/no-error3.c b/teshsuite/smpi/isp/umpire/no-error3.c
index ec91c5d08e..30e6e115dd 100644
--- a/teshsuite/smpi/isp/umpire/no-error3.c
+++ b/teshsuite/smpi/isp/umpire/no-error3.c
@@ -43,7 +43,7 @@ main (int argc, char **argv)
       memset (buf1, 1, buf_size);
 
       MPI_Recv (buf0, buf_size, MPI_INT, 0, 0, MPI_COMM_WORLD, &status);
-      
+
       MPI_Send (buf1, buf_size, MPI_INT, 0, 0, MPI_COMM_WORLD);
     }
   }
diff --git a/teshsuite/smpi/isp/umpire/no-error4-any_src.c b/teshsuite/smpi/isp/umpire/no-error4-any_src.c
index 426149a21e..ecadd932a9 100644
--- a/teshsuite/smpi/isp/umpire/no-error4-any_src.c
+++ b/teshsuite/smpi/isp/umpire/no-error4-any_src.c
@@ -39,7 +39,7 @@ main (int argc, char **argv)
       memset (buf0, 0, buf_size);
 
       MPI_Send_init (buf0, buf_size, MPI_INT, 1, 0, MPI_COMM_WORLD, &aReq[0]);
-      MPI_Recv_init (buf1, buf_size, MPI_INT, 
+      MPI_Recv_init (buf1, buf_size, MPI_INT,
 		     MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &aReq[1]);
 
       MPI_Start (&aReq[0]);
@@ -55,7 +55,7 @@ main (int argc, char **argv)
     else if (rank == 1) {
       memset (buf1, 1, buf_size);
 
-      MPI_Recv_init (buf0, buf_size, MPI_INT, 
+      MPI_Recv_init (buf0, buf_size, MPI_INT,
 		     MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &aReq[0]);
       MPI_Send_init (buf1, buf_size, MPI_INT, 0, 0, MPI_COMM_WORLD, &aReq[1]);
 
diff --git a/teshsuite/smpi/isp/umpire/op-no-error.c b/teshsuite/smpi/isp/umpire/op-no-error.c
index 6f4cf31869..41d5864c04 100644
--- a/teshsuite/smpi/isp/umpire/op-no-error.c
+++ b/teshsuite/smpi/isp/umpire/op-no-error.c
@@ -20,8 +20,8 @@ typedef struct {
   double real, imag;
 } Complex;
 
-void 
-myProd (void *inp, void *inoutp, int *len, MPI_Datatype *dptr) 
+void
+myProd (void *inp, void *inoutp, int *len, MPI_Datatype *dptr)
 {
   int i;
   Complex c;
@@ -61,16 +61,16 @@ main (int argc, char **argv)
 
   MPI_Barrier (comm);
 
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Op_create (myProd, 1, &newop[i]);
 
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Op_free (&newop[i]);
 
   MPI_Barrier (comm);
 
   /* now with an alias... */
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Op_create (myProd, 1, &newop[i]);
 
   for (i = 0; i < OP_COUNT; i++) {
diff --git a/teshsuite/smpi/isp/umpire/op-no-free.c b/teshsuite/smpi/isp/umpire/op-no-free.c
index f6115044b4..40e31e623a 100644
--- a/teshsuite/smpi/isp/umpire/op-no-free.c
+++ b/teshsuite/smpi/isp/umpire/op-no-free.c
@@ -20,8 +20,8 @@ typedef struct {
   double real, imag;
 } Complex;
 
-void 
-myProd (void *inp, void *inoutp, int *len, MPI_Datatype *dptr) 
+void
+myProd (void *inp, void *inoutp, int *len, MPI_Datatype *dptr)
 {
   int i;
   Complex c;
@@ -60,7 +60,7 @@ main (int argc, char **argv)
 
   MPI_Barrier (comm);
 
-  for (i = 0; i < OP_COUNT; i++) 
+  for (i = 0; i < OP_COUNT; i++)
     MPI_Op_create (myProd, 1, &newop[i]);
 
   MPI_Barrier (comm);
diff --git a/teshsuite/smpi/isp/umpire/partial-recv-exhaustive.c b/teshsuite/smpi/isp/umpire/partial-recv-exhaustive.c
index 03bcdddd52..0759c7a054 100644
--- a/teshsuite/smpi/isp/umpire/partial-recv-exhaustive.c
+++ b/teshsuite/smpi/isp/umpire/partial-recv-exhaustive.c
@@ -45,42 +45,42 @@ main (int argc, char **argv)
   printf ("(%d) is alive on %s\n", rank, processor_name);
   fflush (stdout);
 
-  MPI_Buffer_attach (bbuf, sizeof(int) * 
+  MPI_Buffer_attach (bbuf, sizeof(int) *
 		     (BUF_SIZE + MPI_BSEND_OVERHEAD) * 2 * NUM_BSEND_TYPES);
 
   if (rank == 0) {
     /* set up persistent sends... */
     send_t_number = NUM_SEND_TYPES - NUM_PERSISTENT_SEND_TYPES;
 
-    MPI_Send_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Send_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Send_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Send_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
     send_t_number++;
 
-    MPI_Bsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Bsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Bsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Bsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
 
     send_t_number++;
 
-    MPI_Rsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Rsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Rsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Rsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
     send_t_number++;
 
-    MPI_Ssend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Ssend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Ssend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Ssend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
   }
 
@@ -88,8 +88,8 @@ main (int argc, char **argv)
     if ((m == 1) && (rank == 1)) {
       /* set up the persistent receives... */
       for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
-	MPI_Recv_init (&buf[j * BUF_SIZE], 
-		       BUF_SIZE + (j % 2) * SLOP, 
+	MPI_Recv_init (&buf[j * BUF_SIZE],
+		       BUF_SIZE + (j % 2) * SLOP,
 		       MPI_INT, 0, j, comm, &aReq[j]);
       }
     }
@@ -115,11 +115,11 @@ main (int argc, char **argv)
 	if (rank == 0) {
 	  /* set up transient sends... */
 	  send_t_number = 0;
-    
+
 	  MPI_Isend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		     1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
 	  MPI_Isend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		     BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		     BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		     comm, &aReq[send_t_number * 2 + 1]);
 
 	  send_t_number++;
@@ -127,18 +127,18 @@ main (int argc, char **argv)
 	  MPI_Ibsend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		      1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
 	  MPI_Ibsend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		      comm, &aReq[send_t_number * 2 + 1]);
 
 	  send_t_number++;
 
 	  /* Barrier to ensure receives are posted for rsends... */
 	  MPI_Barrier(MPI_COMM_WORLD);
-	  
+	
 	  MPI_Irsend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
 	  MPI_Irsend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		      comm, &aReq[send_t_number * 2 + 1]);
 
 	  send_t_number++;
@@ -146,17 +146,17 @@ main (int argc, char **argv)
 	  MPI_Issend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		      1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
 	  MPI_Issend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		      comm, &aReq[send_t_number * 2 + 1]);
 
 	  /* just to be paranoid */
 	  send_t_number++;
-	  assert (send_t_number == 
+	  assert (send_t_number ==
 		  NUM_SEND_TYPES - NUM_PERSISTENT_SEND_TYPES);
 
 	  /* start the persistent sends... */
 	  if (k % 2) {
-	    MPI_Startall (NUM_PERSISTENT_SEND_TYPES * 2, 
+	    MPI_Startall (NUM_PERSISTENT_SEND_TYPES * 2,
 			  &aReq[2 * send_t_number]);
 	  }
 	  else {
@@ -173,7 +173,7 @@ main (int argc, char **argv)
 	      MPI_Wait (&aReq[j], &aStatus[j]);
 	    }
 	    break;
-	  
+	
 	  case 1:
 	    /* use MPI_Waitall */
 	    MPI_Waitall (NUM_SEND_TYPES * 2, aReq, aStatus);
@@ -184,14 +184,14 @@ main (int argc, char **argv)
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      MPI_Waitany (NUM_SEND_TYPES * 2, aReq, &index, aStatus);
 	    }
-	    
+	
 	    break;
 	
 	  case 3:
 	    /* use MPI_Waitsome */
 	    total = 0;
 	    while (total < NUM_SEND_TYPES * 2) {
-	      MPI_Waitsome (NUM_SEND_TYPES * 2, aReq, 
+	      MPI_Waitsome (NUM_SEND_TYPES * 2, aReq,
 			    &outcount, indices, aStatus);
 
 	      total += outcount;
@@ -203,12 +203,12 @@ main (int argc, char **argv)
 	    /* use MPI_Test */
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      flag = 0;
-	    
+	
 	      while (!flag) {
 		MPI_Test (&aReq[j], &flag, &aStatus[j]);
 	      }
 	    }
-	  
+	
 	    break;
 	
 	  case 5:
@@ -217,7 +217,7 @@ main (int argc, char **argv)
 	    while (!flag) {
 	      MPI_Testall (NUM_SEND_TYPES * 2, aReq, &flag, aStatus);
 	    }
-	  
+	
 	    break;
 	
 	  case 6:
@@ -225,7 +225,7 @@ main (int argc, char **argv)
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      flag = 0;
 	      while (!flag) {
-		MPI_Testany (NUM_SEND_TYPES * 2, aReq, 
+		MPI_Testany (NUM_SEND_TYPES * 2, aReq,
 			     &index, &flag, aStatus);
 	      }
 	    }
@@ -239,7 +239,7 @@ main (int argc, char **argv)
 	      outcount = 0;
 
 	      while (!outcount) {
-		MPI_Testsome (NUM_SEND_TYPES * 2, aReq, 
+		MPI_Testsome (NUM_SEND_TYPES * 2, aReq,
 			      &outcount, indices, aStatus);
 	      }
 
@@ -257,8 +257,8 @@ main (int argc, char **argv)
 	  /* start receives for all of the sends */
 	  if (m == 0) {
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
-	      MPI_Irecv (&buf[j * BUF_SIZE], 
-			 BUF_SIZE + (j % 2) * SLOP, 
+	      MPI_Irecv (&buf[j * BUF_SIZE],
+			 BUF_SIZE + (j % 2) * SLOP,
 			 MPI_INT, 0, j, comm, &aReq[j]);
 	    }
 	  }
@@ -279,13 +279,13 @@ main (int argc, char **argv)
 	
 	  /* complete all of the receives... */
 	  switch (l/2) {
-	  case 0: 
+	  case 0:
 	    /* use MPI_Wait */
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      MPI_Wait (&aReq[j], &aStatus[j]);
 	    }
 	    break;
-	    
+	
 	  case 1:
 	    /* use MPI_Waitall */
 	    MPI_Waitall (NUM_SEND_TYPES * 2, aReq, aStatus);
@@ -296,54 +296,54 @@ main (int argc, char **argv)
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      MPI_Waitany (NUM_SEND_TYPES * 2, aReq, &index, aStatus);
 	    }
-	    
+	
 	    break;
 	
 	  case 3:
 	    /* use MPI_Waitsome */
 	    total = 0;
 	    while (total < NUM_SEND_TYPES * 2) {
-	      MPI_Waitsome (NUM_SEND_TYPES * 2, aReq, 
+	      MPI_Waitsome (NUM_SEND_TYPES * 2, aReq,
 			    &outcount, indices, aStatus);
 
 	      total += outcount;
 	    }
 
 	    break;
-	    
+	
 	  case 4:
 	    /* use MPI_Test */
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      flag = 0;
-	      
+	
 	      while (!flag) {
 		MPI_Test (&aReq[j], &flag, &aStatus[j]);
 	      }
 	    }
-	    
+	
 	    break;
-	    
+	
 	  case 5:
 	    /* use MPI_Testall */
 	    flag = 0;
 	    while (!flag) {
 	      MPI_Testall (NUM_SEND_TYPES * 2, aReq, &flag, aStatus);
 	    }
-	    
+	
 	    break;
-	    
+	
 	  case 6:
 	    /* use MPI_Testany */
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      flag = 0;
 	      while (!flag) {
-		MPI_Testany (NUM_SEND_TYPES * 2, aReq, 
+		MPI_Testany (NUM_SEND_TYPES * 2, aReq,
 			     &index, &flag, aStatus);
 	      }
 	    }
-	    
+	
 	    break;
-	    
+	
 	  case 7:
 	    /* use MPI_Testsome */
 	    total = 0;
@@ -351,15 +351,15 @@ main (int argc, char **argv)
 	      outcount = 0;
 
 	      while (!outcount) {
-		MPI_Testsome (NUM_SEND_TYPES * 2, aReq, 
+		MPI_Testsome (NUM_SEND_TYPES * 2, aReq,
 			      &outcount, indices, aStatus);
 	      }
-	      
+	
 	      total += outcount;
 	    }
-	    
+	
 	    break;
-	    
+	
 	  default:
 	    assert (0);
 	    break;
@@ -392,7 +392,7 @@ main (int argc, char **argv)
 
   MPI_Buffer_detach (bbuf, &at_size);
 
-  assert (at_size == 
+  assert (at_size ==
 	  sizeof(int) * (BUF_SIZE + MPI_BSEND_OVERHEAD) * 2 * NUM_BSEND_TYPES);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/partial-recv-persistent.c b/teshsuite/smpi/isp/umpire/partial-recv-persistent.c
index 81fbd8adea..1ccac54316 100644
--- a/teshsuite/smpi/isp/umpire/partial-recv-persistent.c
+++ b/teshsuite/smpi/isp/umpire/partial-recv-persistent.c
@@ -84,7 +84,7 @@ main (int argc, char **argv)
 	  MPI_Testall (2, aReq, &flag, aStatus);
 	}
       }
-    }  
+    }
     else {
       /* Barrier to ensure receives are posted for rsends... */
       MPI_Barrier(MPI_COMM_WORLD);
diff --git a/teshsuite/smpi/isp/umpire/partial-recv-persistent2.c b/teshsuite/smpi/isp/umpire/partial-recv-persistent2.c
index 336da22219..754749acd3 100644
--- a/teshsuite/smpi/isp/umpire/partial-recv-persistent2.c
+++ b/teshsuite/smpi/isp/umpire/partial-recv-persistent2.c
@@ -88,7 +88,7 @@ main (int argc, char **argv)
 	  }
 	}
       }
-    }  
+    }
     else {
       /* Barrier to ensure receives are posted for rsends... */
       MPI_Barrier(MPI_COMM_WORLD);
diff --git a/teshsuite/smpi/isp/umpire/partial-recv-persistent3.c b/teshsuite/smpi/isp/umpire/partial-recv-persistent3.c
index ef15c1c765..27a4757aaf 100644
--- a/teshsuite/smpi/isp/umpire/partial-recv-persistent3.c
+++ b/teshsuite/smpi/isp/umpire/partial-recv-persistent3.c
@@ -94,7 +94,7 @@ main (int argc, char **argv)
 	  total += outcount;
 	}
       }
-    }  
+    }
     else {
       /* Barrier to ensure receives are posted for rsends... */
       MPI_Barrier(MPI_COMM_WORLD);
diff --git a/teshsuite/smpi/isp/umpire/partial-recv-persistent4.c b/teshsuite/smpi/isp/umpire/partial-recv-persistent4.c
index 1a7f292f18..44295dd9c5 100644
--- a/teshsuite/smpi/isp/umpire/partial-recv-persistent4.c
+++ b/teshsuite/smpi/isp/umpire/partial-recv-persistent4.c
@@ -83,13 +83,13 @@ main (int argc, char **argv)
 	/* use MPI_Test */
 	for (j = 0; j < 2; j++) {
 	  flag = 0;
-	  
+	
 	  while (!flag) {
 	    MPI_Test (&aReq[j], &flag, &aStatus[j]);
 	  }
 	}
       }
-    }  
+    }
     else {
       /* Barrier to ensure receives are posted for rsends... */
       MPI_Barrier(MPI_COMM_WORLD);
diff --git a/teshsuite/smpi/isp/umpire/partial-recv.c b/teshsuite/smpi/isp/umpire/partial-recv.c
index 59835fbf3c..ae43cf8d9d 100644
--- a/teshsuite/smpi/isp/umpire/partial-recv.c
+++ b/teshsuite/smpi/isp/umpire/partial-recv.c
@@ -46,8 +46,8 @@ main (int argc, char **argv)
   MPI_Aint basic_extent;
   int blocklens[3];
   MPI_Aint displs[3];
-  MPI_Datatype structtypes[3]; 
-  MPI_Datatype newtype[2]; 
+  MPI_Datatype structtypes[3];
+  MPI_Datatype newtype[2];
   MPI_Request aReq[2];
   MPI_Status aStatus[2];
   test_small_struct_t small_struct_buf[SMALL_SIZE];
@@ -67,8 +67,8 @@ main (int argc, char **argv)
   blocklens[0] = blocklens[1] = blocklens[2] = 1;
   displs[0] = 0;
   displs[1] = sizeof(double);
-  displs[2] = 
-    ((void *) &(big_struct_buf[0].the_other_double)) - 
+  displs[2] =
+    ((void *) &(big_struct_buf[0].the_other_double)) -
     ((void *) big_struct_buf);
 
   if (displs[2] < 0) displs[2] = -displs[2];
@@ -78,13 +78,13 @@ main (int argc, char **argv)
   /* create the types */
   MPI_Type_struct (2, blocklens, displs, structtypes, &newtype[0]);
   MPI_Type_struct (3, blocklens, displs, structtypes, &newtype[1]);
-  
+
   MPI_Type_extent (newtype[0], &basic_extent);
   if (basic_extent != sizeof (test_small_struct_t)) {
     fprintf (stderr, "(%d): Unexpected extent for small struct\n", rank);
     MPI_Abort (MPI_COMM_WORLD, 666);
   }
-  
+
   MPI_Type_extent (newtype[1], &basic_extent);
   if (basic_extent != sizeof (test_big_struct_t)) {
     fprintf (stderr, "(%d): Unexpected extent for big struct\n", rank);
diff --git a/teshsuite/smpi/isp/umpire/probe-any_src-types-can-mismatch.c b/teshsuite/smpi/isp/umpire/probe-any_src-types-can-mismatch.c
index b4e63bf1fc..3186ae87bf 100644
--- a/teshsuite/smpi/isp/umpire/probe-any_src-types-can-mismatch.c
+++ b/teshsuite/smpi/isp/umpire/probe-any_src-types-can-mismatch.c
@@ -54,22 +54,22 @@ main (int argc, char **argv)
     {
       for (j = 0; j < 2; j++) {
 	MPI_Probe (MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
-      
+
 	if (status.MPI_SOURCE == 0) {
-	  MPI_Recv (&i, 1, MPI_INT, 
+	  MPI_Recv (&i, 1, MPI_INT,
 		    MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 	  if (status.MPI_SOURCE != 0)
 	    printf ("(%d) Type mismatch from matching other message\n", rank);
 	}
 	else {
-	  MPI_Recv (&x, 1, MPI_DOUBLE, 
+	  MPI_Recv (&x, 1, MPI_DOUBLE,
 		    MPI_ANY_SOURCE, 0, MPI_COMM_WORLD, &status);
 	  if (status.MPI_SOURCE == 0)
 	    printf ("(%d) Type mismatch from matching other message\n", rank);
 	}
       }
     }
-      
+
   MPI_Barrier (MPI_COMM_WORLD);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/probe-any_tag-types-can-mismatch.c b/teshsuite/smpi/isp/umpire/probe-any_tag-types-can-mismatch.c
index 1aac3fde74..b5b02db77f 100644
--- a/teshsuite/smpi/isp/umpire/probe-any_tag-types-can-mismatch.c
+++ b/teshsuite/smpi/isp/umpire/probe-any_tag-types-can-mismatch.c
@@ -51,11 +51,11 @@ main (int argc, char **argv)
 
       MPI_Wait (&req0, &status);
     }
-  else if (rank == 1) 
+  else if (rank == 1)
     {
       for (j = 0; j < 2; j++) {
 	MPI_Probe (0, MPI_ANY_TAG, MPI_COMM_WORLD, &status);
-      
+
 	if (status.MPI_TAG == 0) {
 	  MPI_Recv (&i, 1, MPI_INT, 0, MPI_ANY_TAG, MPI_COMM_WORLD, &status);
 
@@ -63,7 +63,7 @@ main (int argc, char **argv)
 	    printf ("(%d) Type mismatch from matching other message\n", rank);
 	}
 	else {
-	  MPI_Recv (&x, 1, MPI_DOUBLE, 0, 
+	  MPI_Recv (&x, 1, MPI_DOUBLE, 0,
 		    MPI_ANY_TAG, MPI_COMM_WORLD, &status);
 
 	  if (status.MPI_TAG == 0)
@@ -71,7 +71,7 @@ main (int argc, char **argv)
 	}
       }
     }
-      
+
   MPI_Barrier (MPI_COMM_WORLD);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/probe-deadlock.c b/teshsuite/smpi/isp/umpire/probe-deadlock.c
index e41ece3273..80ed338f99 100644
--- a/teshsuite/smpi/isp/umpire/probe-deadlock.c
+++ b/teshsuite/smpi/isp/umpire/probe-deadlock.c
@@ -37,7 +37,7 @@ main (int argc, char **argv)
       i = 0;
 
       MPI_Probe (1, 0, MPI_COMM_WORLD, &status);
-      
+
       MPI_Send (&i, 1, MPI_INT, 1, 0, MPI_COMM_WORLD);
 
       MPI_Recv (&x, 1, MPI_DOUBLE, 1, 0, MPI_COMM_WORLD, &status);
@@ -52,7 +52,7 @@ main (int argc, char **argv)
 
       MPI_Send (&x, 1, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
     }
-      
+
   MPI_Barrier (MPI_COMM_WORLD);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/pt2pt-byte-int-mismatch.c b/teshsuite/smpi/isp/umpire/pt2pt-byte-int-mismatch.c
index 13fbd5510b..8ff6b91622 100644
--- a/teshsuite/smpi/isp/umpire/pt2pt-byte-int-mismatch.c
+++ b/teshsuite/smpi/isp/umpire/pt2pt-byte-int-mismatch.c
@@ -44,42 +44,42 @@ main (int argc, char **argv)
   printf ("(%d) is alive on %s\n", rank, processor_name);
   fflush (stdout);
 
-  MPI_Buffer_attach (bbuf, sizeof(int) * 
+  MPI_Buffer_attach (bbuf, sizeof(int) *
 		     (BUF_SIZE + MPI_BSEND_OVERHEAD) * 2 * NUM_BSEND_TYPES);
 
   if (rank == 0) {
     /* set up persistent sends... */
     send_t_number = NUM_SEND_TYPES - NUM_PERSISTENT_SEND_TYPES;
 
-    MPI_Send_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Send_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Send_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Send_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
     send_t_number++;
 
-    MPI_Bsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Bsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Bsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Bsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
 
     send_t_number++;
 
-    MPI_Rsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Rsend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Rsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Rsend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
 
     send_t_number++;
 
-    MPI_Ssend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT, 
+    MPI_Ssend_init (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
-    MPI_Ssend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE], 
-		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+    MPI_Ssend_init (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
+		    BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		    comm, &aReq[send_t_number * 2 + 1]);
   }
 
@@ -87,10 +87,10 @@ main (int argc, char **argv)
     if ((m == 1) && (rank == 1)) {
       /* set up the persistent receives... */
       for (j = 0; j < NUM_SEND_TYPES * 2; j+=2) {
-	MPI_Recv_init (&buf[j * BUF_SIZE], 
+	MPI_Recv_init (&buf[j * BUF_SIZE],
 		       BUF_SIZE, MPI_INT, 0, j, comm, &aReq[j]);
-	MPI_Recv_init (&buf[(j + 1) * BUF_SIZE], 
-		       BUF_SIZE * sizeof(int), 
+	MPI_Recv_init (&buf[(j + 1) * BUF_SIZE],
+		       BUF_SIZE * sizeof(int),
 		       MPI_BYTE, 0, j + 1, comm, &aReq[j + 1]);
       }
     }
@@ -116,11 +116,11 @@ main (int argc, char **argv)
 	if (rank == 0) {
 	  /* set up transient sends... */
 	  send_t_number = 0;
-    
+
 	  MPI_Isend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		     1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
 	  MPI_Isend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		     BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		     BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		     comm, &aReq[send_t_number * 2 + 1]);
 
 	  send_t_number++;
@@ -128,18 +128,18 @@ main (int argc, char **argv)
 	  MPI_Ibsend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		      1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
 	  MPI_Ibsend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		      comm, &aReq[send_t_number * 2 + 1]);
 
 	  send_t_number++;
 
 	  /* Barrier to ensure receives are posted for rsends... */
 	  MPI_Barrier(MPI_COMM_WORLD);
-	  
+	
 	  MPI_Irsend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		    1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
 	  MPI_Irsend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		      comm, &aReq[send_t_number * 2 + 1]);
 
 	  send_t_number++;
@@ -147,17 +147,17 @@ main (int argc, char **argv)
 	  MPI_Issend (&buf[send_t_number * 2 * BUF_SIZE], BUF_SIZE, MPI_INT,
 		      1, send_t_number * 2, comm, &aReq[send_t_number * 2]);
 	  MPI_Issend (&buf[(send_t_number * 2 + 1) * BUF_SIZE],
-		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1, 
+		      BUF_SIZE, MPI_INT, 1, send_t_number * 2 + 1,
 		      comm, &aReq[send_t_number * 2 + 1]);
 
 	  /* just to be paranoid */
 	  send_t_number++;
-	  assert (send_t_number == 
+	  assert (send_t_number ==
 		  NUM_SEND_TYPES - NUM_PERSISTENT_SEND_TYPES);
 
 	  /* start the persistent sends... */
 	  if (k % 2) {
-	    MPI_Startall (NUM_PERSISTENT_SEND_TYPES * 2, 
+	    MPI_Startall (NUM_PERSISTENT_SEND_TYPES * 2,
 			  &aReq[2 * send_t_number]);
 	  }
 	  else {
@@ -174,7 +174,7 @@ main (int argc, char **argv)
 	      MPI_Wait (&aReq[j], &aStatus[j]);
 	    }
 	    break;
-	  
+	
 	  case 1:
 	    /* use MPI_Waitall */
 	    MPI_Waitall (NUM_SEND_TYPES * 2, aReq, aStatus);
@@ -185,14 +185,14 @@ main (int argc, char **argv)
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      MPI_Waitany (NUM_SEND_TYPES * 2, aReq, &index, aStatus);
 	    }
-	    
+	
 	    break;
 	
 	  case 3:
 	    /* use MPI_Waitsome */
 	    total = 0;
 	    while (total < NUM_SEND_TYPES * 2) {
-	      MPI_Waitsome (NUM_SEND_TYPES * 2, aReq, 
+	      MPI_Waitsome (NUM_SEND_TYPES * 2, aReq,
 			    &outcount, indices, aStatus);
 
 	      total += outcount;
@@ -204,12 +204,12 @@ main (int argc, char **argv)
 	    /* use MPI_Test */
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      flag = 0;
-	    
+	
 	      while (!flag) {
 		MPI_Test (&aReq[j], &flag, &aStatus[j]);
 	      }
 	    }
-	  
+	
 	    break;
 	
 	  case 5:
@@ -218,7 +218,7 @@ main (int argc, char **argv)
 	    while (!flag) {
 	      MPI_Testall (NUM_SEND_TYPES * 2, aReq, &flag, aStatus);
 	    }
-	  
+	
 	    break;
 	
 	  case 6:
@@ -226,7 +226,7 @@ main (int argc, char **argv)
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      flag = 0;
 	      while (!flag) {
-		MPI_Testany (NUM_SEND_TYPES * 2, aReq, 
+		MPI_Testany (NUM_SEND_TYPES * 2, aReq,
 			     &index, &flag, aStatus);
 	      }
 	    }
@@ -240,7 +240,7 @@ main (int argc, char **argv)
 	      outcount = 0;
 
 	      while (!outcount) {
-		MPI_Testsome (NUM_SEND_TYPES * 2, aReq, 
+		MPI_Testsome (NUM_SEND_TYPES * 2, aReq,
 			      &outcount, indices, aStatus);
 	      }
 
@@ -258,10 +258,10 @@ main (int argc, char **argv)
 	  /* start receives for all of the sends */
 	  if (m == 0) {
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j+=2) {
-	      MPI_Irecv (&buf[j * BUF_SIZE], 
+	      MPI_Irecv (&buf[j * BUF_SIZE],
 			 BUF_SIZE, MPI_INT, 0, j, comm, &aReq[j]);
-	      MPI_Irecv (&buf[(j + 1) * BUF_SIZE], 
-			 BUF_SIZE * sizeof(int), 
+	      MPI_Irecv (&buf[(j + 1) * BUF_SIZE],
+			 BUF_SIZE * sizeof(int),
 			 MPI_BYTE, 0, j + 1, comm, &aReq[j + 1]);
 	    }
 	  }
@@ -282,13 +282,13 @@ main (int argc, char **argv)
 	
 	  /* complete all of the receives... */
 	  switch (l/2) {
-	  case 0: 
+	  case 0:
 	    /* use MPI_Wait */
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      MPI_Wait (&aReq[j], &aStatus[j]);
 	    }
 	    break;
-	    
+	
 	  case 1:
 	    /* use MPI_Waitall */
 	    MPI_Waitall (NUM_SEND_TYPES * 2, aReq, aStatus);
@@ -299,54 +299,54 @@ main (int argc, char **argv)
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      MPI_Waitany (NUM_SEND_TYPES * 2, aReq, &index, aStatus);
 	    }
-	    
+	
 	    break;
 	
 	  case 3:
 	    /* use MPI_Waitsome */
 	    total = 0;
 	    while (total < NUM_SEND_TYPES * 2) {
-	      MPI_Waitsome (NUM_SEND_TYPES * 2, aReq, 
+	      MPI_Waitsome (NUM_SEND_TYPES * 2, aReq,
 			    &outcount, indices, aStatus);
 
 	      total += outcount;
 	    }
 
 	    break;
-	    
+	
 	  case 4:
 	    /* use MPI_Test */
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      flag = 0;
-	      
+	
 	      while (!flag) {
 		MPI_Test (&aReq[j], &flag, &aStatus[j]);
 	      }
 	    }
-	    
+	
 	    break;
-	    
+	
 	  case 5:
 	    /* use MPI_Testall */
 	    flag = 0;
 	    while (!flag) {
 	      MPI_Testall (NUM_SEND_TYPES * 2, aReq, &flag, aStatus);
 	    }
-	    
+	
 	    break;
-	    
+	
 	  case 6:
 	    /* use MPI_Testany */
 	    for (j = 0; j < NUM_SEND_TYPES * 2; j++) {
 	      flag = 0;
 	      while (!flag) {
-		MPI_Testany (NUM_SEND_TYPES * 2, aReq, 
+		MPI_Testany (NUM_SEND_TYPES * 2, aReq,
 			     &index, &flag, aStatus);
 	      }
 	    }
-	    
+	
 	    break;
-	    
+	
 	  case 7:
 	    /* use MPI_Testsome */
 	    total = 0;
@@ -354,15 +354,15 @@ main (int argc, char **argv)
 	      outcount = 0;
 
 	      while (!outcount) {
-		MPI_Testsome (NUM_SEND_TYPES * 2, aReq, 
+		MPI_Testsome (NUM_SEND_TYPES * 2, aReq,
 			      &outcount, indices, aStatus);
 	      }
-	      
+	
 	      total += outcount;
 	    }
-	    
+	
 	    break;
-	    
+	
 	  default:
 	    assert (0);
 	    break;
@@ -395,7 +395,7 @@ main (int argc, char **argv)
 
   MPI_Buffer_detach (bbuf, &at_size);
 
-  assert (at_size == 
+  assert (at_size ==
 	  sizeof(int) * (BUF_SIZE + MPI_BSEND_OVERHEAD) * 2 * NUM_BSEND_TYPES);
 
   MPI_Finalize ();
diff --git a/teshsuite/smpi/isp/umpire/remote_group-no-error.c b/teshsuite/smpi/isp/umpire/remote_group-no-error.c
index a93eff6084..e62fc50270 100644
--- a/teshsuite/smpi/isp/umpire/remote_group-no-error.c
+++ b/teshsuite/smpi/isp/umpire/remote_group-no-error.c
@@ -25,8 +25,8 @@ main (int argc, char **argv)
   char processor_name[128];
   int namelen = 128;
   int i;
-  MPI_Group newgroup; 
-  MPI_Group newgroup2; 
+  MPI_Group newgroup;
+  MPI_Group newgroup2;
   MPI_Comm temp;
   MPI_Comm intercomm = MPI_COMM_NULL;
 
@@ -49,13 +49,13 @@ main (int argc, char **argv)
     MPI_Comm_split (MPI_COMM_WORLD, rank % 3, nprocs - rank, &temp);
 
     if (rank % 3) {
-      MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, 
+      MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD,
 			    (((nprocs % 3) == 2) && ((rank % 3) == 2)) ?
 			    nprocs - 1 : nprocs - (rank % 3) - (nprocs % 3),
 			    INTERCOMM_CREATE_TAG, &intercomm);
 
       MPI_Comm_remote_group (intercomm, &newgroup);
-      
+
       MPI_Comm_free (&intercomm);
     }
     else {
@@ -63,7 +63,7 @@ main (int argc, char **argv)
     }
 
     MPI_Comm_free (&temp);
-      
+
     MPI_Group_free (&newgroup);
 
     MPI_Barrier (comm);
@@ -73,7 +73,7 @@ main (int argc, char **argv)
     MPI_Comm_split (MPI_COMM_WORLD, rank % 3, nprocs - rank, &temp);
 
     if (rank % 3) {
-      MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD, 
+      MPI_Intercomm_create (temp, 0, MPI_COMM_WORLD,
 			    (((nprocs % 3) == 2) && ((rank % 3) == 2)) ?
 			    nprocs - 1 : nprocs - (rank % 3) - (nprocs % 3),
 			    INTERCOMM_CREATE_TAG, &intercomm);
@@ -87,7 +87,7 @@ main (int argc, char **argv)
     }
 
     MPI_Comm_free (&temp);
-      
+
     newgroup2 = newgroup;
     MPI_Group_free (&newgroup2);
   }
diff --git a/teshsuite/smpi/isp/umpire/type-commit-twice.c b/teshsuite/smpi/isp/umpire/type-commit-twice.c
index 76debc31bf..34c84661f3 100644
--- a/teshsuite/smpi/isp/umpire/type-commit-twice.c
+++ b/teshsuite/smpi/isp/umpire/type-commit-twice.c
@@ -39,7 +39,7 @@ main (int argc, char **argv)
 
   /* COMMITTING TWICE IS NOT AN ERROR - SEE:
      http://www.mpi-forum.org/docs/mpi-20-html/node50.htm#Node50
-     AT MOST, UMPIRE SHOULD PROVIDE A CLEAR WARNING ABOUT MINOR 
+     AT MOST, UMPIRE SHOULD PROVIDE A CLEAR WARNING ABOUT MINOR
      PERFORMANCE CONSEQUENCE (JUST A WASTED FUNCTION CALL)... */
   MPI_Type_commit (&newtype);
 
diff --git a/teshsuite/smpi/isp/umpire/type-no-error-exhaustive-with-isends.c b/teshsuite/smpi/isp/umpire/type-no-error-exhaustive-with-isends.c
index 8826ce2133..e0525a6293 100644
--- a/teshsuite/smpi/isp/umpire/type-no-error-exhaustive-with-isends.c
+++ b/teshsuite/smpi/isp/umpire/type-no-error-exhaustive-with-isends.c
@@ -59,8 +59,8 @@ main (int argc, char **argv)
   int namelen = 128;
   int i, j, k, basic_extent;
   int blocklens[4], displs[4];
-  MPI_Datatype structtypes[4]; 
-  MPI_Datatype newtype[TYPE_CONSTRUCTOR_COUNT]; 
+  MPI_Datatype structtypes[4];
+  MPI_Datatype newtype[TYPE_CONSTRUCTOR_COUNT];
   MPI_Request aReq[TYPE_CONSTRUCTOR_COUNT];
   MPI_Status aStatus[TYPE_CONSTRUCTOR_COUNT];
 #ifdef RUN_TYPE_STRUCT
@@ -104,7 +104,7 @@ main (int argc, char **argv)
 
   /* create the types */
   MPI_Type_struct (2, blocklens, displs, structtypes, &newtype[0]);
-  
+
   MPI_Type_extent (newtype[0], &basic_extent);
   if (basic_extent != sizeof (test_basic_struct_t)) {
     fprintf (stderr, "(%d): Unexpected extent for struct\n");
@@ -112,7 +112,7 @@ main (int argc, char **argv)
   }
 
   MPI_Type_vector (2, 3, 4, newtype[0], &newtype[1]);
-  MPI_Type_hvector (3, 2, 15 * sizeof (test_basic_struct_t), 
+  MPI_Type_hvector (3, 2, 15 * sizeof (test_basic_struct_t),
 		    newtype[1], &newtype[2]);
   displs[1] = 2;
   MPI_Type_indexed (2, blocklens, displs, newtype[2], &newtype[3]);
@@ -132,7 +132,7 @@ main (int argc, char **argv)
 
   MPI_Type_struct (4, blocklens, displs, structtypes, &newtype[6]);
 
-#ifdef RUN_TYPE_STRUCT 
+#ifdef RUN_TYPE_STRUCT
   MPI_Type_commit (&newtype[0]);
 #endif
 
@@ -156,8 +156,8 @@ main (int argc, char **argv)
   MPI_Type_commit (&newtype[5]);
 #endif
 
-#ifdef RUN_TYPE_STRUCT_LB_UB 
-#ifndef RUN_TYPE_STRUCT 
+#ifdef RUN_TYPE_STRUCT_LB_UB
+#ifndef RUN_TYPE_STRUCT
   /* need the struct type for the receive... */
   MPI_Type_commit (&newtype[0]);
 #endif
@@ -254,7 +254,7 @@ main (int argc, char **argv)
 #endif
 
 #ifdef RUN_TYPE_STRUCT_LB_UB
-    MPI_Isend (&(struct_lb_ub_send_buf[0].the_double_to_send), 
+    MPI_Isend (&(struct_lb_ub_send_buf[0].the_double_to_send),
 	       MSG_COUNT, newtype[6], 1, 6, comm, &aReq[6]);
 #else
     aReq[6] = MPI_REQUEST_NULL;
@@ -347,7 +347,7 @@ main (int argc, char **argv)
 #endif
 
 #ifdef RUN_TYPE_STRUCT_LB_UB
-    MPI_Irecv (struct_lb_ub_recv_buf, 
+    MPI_Irecv (struct_lb_ub_recv_buf,
 	       MSG_COUNT, newtype[0], 0, 6, comm, &aReq[6]);
 #else
     aReq[6] = MPI_REQUEST_NULL;
@@ -714,7 +714,7 @@ main (int argc, char **argv)
 #ifdef RUN_TYPE_HVECTOR
       /* eight holes in hvector_buf... */
       /* hole in first vector, first block... */
-      assert ((hvector_buf[i*44 + 3].the_double == 2.0) && 
+      assert ((hvector_buf[i*44 + 3].the_double == 2.0) &&
 	      (hvector_buf[i*44 + 3].the_char == 'b'));
       /* hole in second vector, first block... */
       assert ((hvector_buf[i*44 + 10].the_double == 2.0) &&
diff --git a/teshsuite/smpi/isp/umpire/type-no-error-exhaustive.c b/teshsuite/smpi/isp/umpire/type-no-error-exhaustive.c
index 6515ea223d..1df0319294 100644
--- a/teshsuite/smpi/isp/umpire/type-no-error-exhaustive.c
+++ b/teshsuite/smpi/isp/umpire/type-no-error-exhaustive.c
@@ -25,8 +25,8 @@ main (int argc, char **argv)
   int namelen = 128;
   int i;
   int blocklens[2], displs[2];
-  MPI_Datatype newtype[TYPE_CONSTRUCTOR_COUNT]; 
-  MPI_Datatype newtype2[TYPE_CONSTRUCTOR_COUNT]; 
+  MPI_Datatype newtype[TYPE_CONSTRUCTOR_COUNT];
+  MPI_Datatype newtype2[TYPE_CONSTRUCTOR_COUNT];
 
   /* init */
   MPI_Init (&argc, &argv);
diff --git a/teshsuite/smpi/isp/umpire/type-no-free-exhaustive.c b/teshsuite/smpi/isp/umpire/type-no-free-exhaustive.c
index 7cd7c0cadb..51b38a713c 100644
--- a/teshsuite/smpi/isp/umpire/type-no-free-exhaustive.c
+++ b/teshsuite/smpi/isp/umpire/type-no-free-exhaustive.c
@@ -25,8 +25,8 @@ main (int argc, char **argv)
   int namelen = 128;
   int i;
   int blocklens[2], displs[2];
-  MPI_Datatype newtype[TYPE_CONSTRUCTOR_COUNT]; 
-  MPI_Datatype newtype2[TYPE_CONSTRUCTOR_COUNT]; 
+  MPI_Datatype newtype[TYPE_CONSTRUCTOR_COUNT];
+  MPI_Datatype newtype2[TYPE_CONSTRUCTOR_COUNT];
 
   /* init */
   MPI_Init (&argc, &argv);
diff --git a/teshsuite/smpi/isp/umpire/type-no-free3.c b/teshsuite/smpi/isp/umpire/type-no-free3.c
index 8fc526ba70..572a1b5fb5 100644
--- a/teshsuite/smpi/isp/umpire/type-no-free3.c
+++ b/teshsuite/smpi/isp/umpire/type-no-free3.c
@@ -85,7 +85,7 @@ main (int argc, char **argv)
 	  MPI_Type_free (&newtype[j]);
 	}
       }
-      
+
       MPI_Waitall (TYPES_TO_COMMIT, reqs, statuses);
     }
   }
diff --git a/teshsuite/smpi/macro-shared/macro-shared.c b/teshsuite/smpi/macro-shared/macro-shared.c
index d1d9966e82..1e09ac43de 100644
--- a/teshsuite/smpi/macro-shared/macro-shared.c
+++ b/teshsuite/smpi/macro-shared/macro-shared.c
@@ -47,7 +47,7 @@ int main(int argc, char *argv[])
   //Try SMPI_SHARED_CALL function, which should call hash only once and for all.
   char *str = strdup("onceandforall");
   if(rank==size-1){
-    SMPI_SHARED_CALL(hash,str,str,buf);  
+    SMPI_SHARED_CALL(hash,str,str,buf);
   }
 
   MPI_Barrier(MPI_COMM_WORLD);
diff --git a/teshsuite/smpi/mpich3-test/coll/allgatherv4_manual.c b/teshsuite/smpi/mpich3-test/coll/allgatherv4_manual.c
index 4249a92524..c5cbb95a71 100644
--- a/teshsuite/smpi/mpich3-test/coll/allgatherv4_manual.c
+++ b/teshsuite/smpi/mpich3-test/coll/allgatherv4_manual.c
@@ -177,7 +177,7 @@ void comm_tests(MPI_Comm comm)
     }
 }
 
-double run_test(long long msg_size, MPI_Comm comm, test_t test_type, 
+double run_test(long long msg_size, MPI_Comm comm, test_t test_type,
 		double * max_time)
 {
     int i, j;
@@ -230,7 +230,7 @@ double run_test(long long msg_size, MPI_Comm comm, test_t test_type,
     /* Test that:
        1: sbuf is large enough
        2: rbuf is large enough
-       3: There were no failures (e.g., tmp nowhere > rbuf size 
+       3: There were no failures (e.g., tmp nowhere > rbuf size
     */
     MPI_Barrier(comm);
     start = MPI_Wtime();
diff --git a/teshsuite/smpi/mpich3-test/datatype/typename.c b/teshsuite/smpi/mpich3-test/datatype/typename.c
index 60845c16f4..8ff1fc0f3d 100644
--- a/teshsuite/smpi/mpich3-test/datatype/typename.c
+++ b/teshsuite/smpi/mpich3-test/datatype/typename.c
@@ -85,10 +85,10 @@ mpi_names_t mpi_names[] = {
     { MPI_OFFSET, "MPI_OFFSET" },
 #endif
     /* Size-specific types */
-    /* Do not move MPI_REAL4 - this is used to indicate the very first 
+    /* Do not move MPI_REAL4 - this is used to indicate the very first
        optional type.  In addition, you must not add any required types
        after this type */
-    /* See MPI 2.1, Section 16.2.  These are required, predefined types. 
+    /* See MPI 2.1, Section 16.2.  These are required, predefined types.
        If the type is not available (e.g., *only* because the Fortran
        compiler does not support it), the value may be MPI_DATATYPE_NULL */
     { MPI_REAL4, "MPI_REAL4" },
@@ -110,9 +110,9 @@ mpi_names_t mpi_names[] = {
     /* Semi-optional types - if the compiler doesn't support long double
        or long long, these might be MPI_DATATYPE_NULL */
     { MPI_LONG_DOUBLE, "MPI_LONG_DOUBLE" },
-    { MPI_LONG_LONG_INT, "MPI_LONG_LONG_INT" }, 
+    { MPI_LONG_LONG_INT, "MPI_LONG_LONG_INT" },
     { MPI_LONG_LONG, "MPI_LONG_LONG" },
-    { MPI_UNSIGNED_LONG_LONG, "MPI_UNSIGNED_LONG_LONG" }, 
+    { MPI_UNSIGNED_LONG_LONG, "MPI_UNSIGNED_LONG_LONG" },
     { MPI_LONG_DOUBLE_INT, "MPI_LONG_DOUBLE_INT" },
 #if MTEST_HAVE_MIN_MPI_VERSION(2,2)
     /* added in MPI-2.2 */
@@ -132,7 +132,7 @@ mpi_names_t mpi_names[] = {
     int errs = 0;
 
     MTest_Init( &argc, &argv );
-    
+
     /* Sample some datatypes */
     /* See 8.4, "Naming Objects" in MPI-2.  The default name is the same
        as the datatype name */
@@ -152,15 +152,15 @@ mpi_names_t mpi_names[] = {
     inOptional = 0;
     for (i=0; mpi_names[i].name != 0; i++) {
 	/* Are we in the optional types? */
-	if (strcmp( mpi_names[i].name, "MPI_REAL4" ) == 0) 
+	if (strcmp( mpi_names[i].name, "MPI_REAL4" ) == 0)
 	    inOptional = 1;
 	/* If this optional type is not supported, skip it */
 	if (inOptional && mpi_names[i].dtype == MPI_DATATYPE_NULL) continue;
 	if (mpi_names[i].dtype == MPI_DATATYPE_NULL) {
-	    /* Report an error because all of the standard types 
+	    /* Report an error because all of the standard types
 	       must be supported */
 	    errs++;
-	    fprintf( stderr, "MPI Datatype %s is MPI_DATATYPE_NULL\n", 
+	    fprintf( stderr, "MPI Datatype %s is MPI_DATATYPE_NULL\n",
 		     mpi_names[i].name );
 	    continue;
 	}
@@ -169,7 +169,7 @@ mpi_names_t mpi_names[] = {
 	MPI_Type_get_name( mpi_names[i].dtype, name, &namelen );
 	if (strncmp( name, mpi_names[i].name, namelen )) {
 	    errs++;
-	    fprintf( stderr, "Expected %s but got %s\n", 
+	    fprintf( stderr, "Expected %s but got %s\n",
 		     mpi_names[i].name, name );
 	}
     }
diff --git a/teshsuite/smpi/mpich3-test/f77/attr/attraints.h b/teshsuite/smpi/mpich3-test/f77/attr/attraints.h
index 182b04567a..fb0747f683 100644
--- a/teshsuite/smpi/mpich3-test/f77/attr/attraints.h
+++ b/teshsuite/smpi/mpich3-test/f77/attr/attraints.h
@@ -1,4 +1,4 @@
-C -*- Mode: Fortran; -*- 
+C -*- Mode: Fortran; -*-
 C
 C  (C) 2003 by Argonne National Laboratory.
 C      See COPYRIGHT in top-level directory.
diff --git a/teshsuite/smpi/mpich3-test/f77/ext/add1size.h b/teshsuite/smpi/mpich3-test/f77/ext/add1size.h
index 940a4c315a..85d2b58c1e 100644
--- a/teshsuite/smpi/mpich3-test/f77/ext/add1size.h
+++ b/teshsuite/smpi/mpich3-test/f77/ext/add1size.h
@@ -1,4 +1,4 @@
-C -*- Mode: Fortran; -*- 
+C -*- Mode: Fortran; -*-
 C
 C  (C) 2003 by Argonne National Laboratory.
 C      See COPYRIGHT in top-level directory.
diff --git a/teshsuite/smpi/mpich3-test/f77/ext/c2f2c.c b/teshsuite/smpi/mpich3-test/f77/ext/c2f2c.c
index 4e048b272f..4454e702f3 100644
--- a/teshsuite/smpi/mpich3-test/f77/ext/c2f2c.c
+++ b/teshsuite/smpi/mpich3-test/f77/ext/c2f2c.c
@@ -5,7 +5,7 @@
  *      See COPYRIGHT in top-level directory.
  */
 /*
- * This file contains the C routines used in testing the c2f and f2c 
+ * This file contains the C routines used in testing the c2f and f2c
  * handle conversion functions, except for MPI_File and MPI_Win (to
  * allow working with MPI implementations that do not include those
  * features).
@@ -32,11 +32,11 @@
 #include "../../include/mpitestconf.h"
 #include <string.h>
 
-/* 
+/*
    Name mapping.  All routines are created with names that are lower case
    with a single trailing underscore.  This matches many compilers.
    We use #define to change the name for Fortran compilers that do
-   not use the lowercase/underscore pattern 
+   not use the lowercase/underscore pattern
 */
 
 #ifdef F77_NAME_UPPER
@@ -78,7 +78,7 @@
       defined(F77_NAME_MIXED_USCORE)
 /* Else leave name alone (routines have no underscore, so both
    of these map to a lowercase, single underscore) */
-#else 
+#else
 #error 'Unrecognized Fortran name mapping'
 #endif
 
@@ -176,7 +176,7 @@ MPI_Fint c2frequest_ ( MPI_Fint *request )
     int flag;
     MPI_Test( &req, &flag, &status );
     MPI_Test_cancelled( &status, &flag );
-    if (!flag) { 
+    if (!flag) {
 	fprintf( stderr, "Request: Wrong value for flag\n" );
 	return 1;
     }
@@ -189,7 +189,7 @@ MPI_Fint c2frequest_ ( MPI_Fint *request )
 MPI_Fint c2fop_ ( MPI_Fint *op )
 {
     MPI_Op cOp = MPI_Op_f2c( *op );
-    
+
     if (cOp != MPI_SUM) {
 	fprintf( stderr, "Op: did not get sum\n" );
 	return 1;
@@ -209,7 +209,7 @@ MPI_Fint c2ferrhandler_ ( MPI_Fint *errh )
     return 0;
 }
 
-/* 
+/*
  * The following routines provide handles to the calling Fortran program
  */
 void f2ccomm_( MPI_Fint * comm )
@@ -244,11 +244,11 @@ void f2crequest_( MPI_Fint * req )
 {
     MPI_Request cReq;
 
-    MPI_Irecv( NULL, 0, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG, 
+    MPI_Irecv( NULL, 0, MPI_INT, MPI_ANY_SOURCE, MPI_ANY_TAG,
 	       MPI_COMM_WORLD, &cReq );
     MPI_Cancel( &cReq );
     *req = MPI_Request_c2f( cReq );
-    
+
 }
 
 void f2cop_( MPI_Fint * op )
diff --git a/teshsuite/smpi/mpich3-test/f77/ext/c2fmult.c b/teshsuite/smpi/mpich3-test/f77/ext/c2fmult.c
index 07c21d6e36..4f1d343838 100644
--- a/teshsuite/smpi/mpich3-test/f77/ext/c2fmult.c
+++ b/teshsuite/smpi/mpich3-test/f77/ext/c2fmult.c
@@ -6,8 +6,8 @@
 
 /*
   Check that MPI_xxxx_c2f, applied to the same object several times,
-  yields the same handle.  We do this because when MPI handles in 
-  C are a different length than those in Fortran, care needs to 
+  yields the same handle.  We do this because when MPI handles in
+  C are a different length than those in Fortran, care needs to
   be exercised to ensure that the mapping from one to another is unique.
   (Test added to test a potential problem in ROMIO for handling MPI_File
   on 64-bit systems)
@@ -55,6 +55,6 @@ int main( int argc, char *argv[] )
 
     MTest_Finalize( errs );
     MPI_Finalize();
-    
+
     return 0;
 }
diff --git a/teshsuite/smpi/mpich3-test/f77/ext/ctypesfromc.c b/teshsuite/smpi/mpich3-test/f77/ext/ctypesfromc.c
index 51015da908..e0e5f60d92 100644
--- a/teshsuite/smpi/mpich3-test/f77/ext/ctypesfromc.c
+++ b/teshsuite/smpi/mpich3-test/f77/ext/ctypesfromc.c
@@ -11,8 +11,8 @@
  * The tests follow this pattern:
  *
  *  Fortran main program
- *     calls the c routine f2ctype with each of the C types and the name of 
- *     the type.  That c routine using MPI_Type_f2c to convert the 
+ *     calls the c routine f2ctype with each of the C types and the name of
+ *     the type.  That c routine using MPI_Type_f2c to convert the
  *     Fortran handle to a C handle, and then compares it to the corresponding
  *     C type, which is found by looking up the C handle by name
  *
@@ -60,18 +60,18 @@ static mpi_names_t mpi_names[] = {
     { MPI_SHORT_INT, "MPI_SHORT_INT" },
     { MPI_2INT, "MPI_2INT" },
     { MPI_LONG_DOUBLE, "MPI_LONG_DOUBLE" },
-    { MPI_LONG_LONG_INT, "MPI_LONG_LONG_INT" }, 
+    { MPI_LONG_LONG_INT, "MPI_LONG_LONG_INT" },
     { MPI_LONG_LONG, "MPI_LONG_LONG" },
-    { MPI_UNSIGNED_LONG_LONG, "MPI_UNSIGNED_LONG_LONG" }, 
+    { MPI_UNSIGNED_LONG_LONG, "MPI_UNSIGNED_LONG_LONG" },
     { MPI_LONG_DOUBLE_INT, "MPI_LONG_DOUBLE_INT" },
     { 0, (char *)0 },  /* Sentinal used to indicate the last element */
 };
 
-/* 
+/*
    Name mapping.  All routines are created with names that are lower case
    with a single trailing underscore.  This matches many compilers.
    We use #define to change the name for Fortran compilers that do
-   not use the lowercase/underscore pattern 
+   not use the lowercase/underscore pattern
 */
 
 #ifdef F77_NAME_UPPER
@@ -85,7 +85,7 @@ static mpi_names_t mpi_names[] = {
       defined(F77_NAME_MIXED_USCORE)
 /* Else leave name alone (routines have no underscore, so both
    of these map to a lowercase, single underscore) */
-#else 
+#else
 #error 'Unrecognized Fortran name mapping'
 #endif
 
@@ -103,8 +103,8 @@ int f2ctype_( MPI_Fint *fhandle, MPI_Fint *typeidx )
     if (ctype != mpi_names[*typeidx].dtype) {
 	char mytypename[MPI_MAX_OBJECT_NAME];
 	int mytypenamelen;
-	/* An implementation is not *required* to deliver the 
-	   corresponding C version of the MPI Datatype bit-for-bit.  But 
+	/* An implementation is not *required* to deliver the
+	   corresponding C version of the MPI Datatype bit-for-bit.  But
 	   if *must* act like it - e.g., the datatype name must be the same */
 	MPI_Type_get_name( ctype, mytypename, &mytypenamelen );
 	if (strcmp( mytypename, mpi_names[*typeidx].name ) != 0) {
@@ -113,6 +113,6 @@ int f2ctype_( MPI_Fint *fhandle, MPI_Fint *typeidx )
 		    mpi_names[*typeidx].name, mytypename, *fhandle, MPI_Type_c2f( ctype ) );
 	}
     }
-    
+
     return errs;
 }
diff --git a/teshsuite/smpi/mpich3-test/f77/pt2pt/attr1aints.h b/teshsuite/smpi/mpich3-test/f77/pt2pt/attr1aints.h
index 182b04567a..fb0747f683 100644
--- a/teshsuite/smpi/mpich3-test/f77/pt2pt/attr1aints.h
+++ b/teshsuite/smpi/mpich3-test/f77/pt2pt/attr1aints.h
@@ -1,4 +1,4 @@
-C -*- Mode: Fortran; -*- 
+C -*- Mode: Fortran; -*-
 C
 C  (C) 2003 by Argonne National Laboratory.
 C      See COPYRIGHT in top-level directory.
diff --git a/teshsuite/smpi/mpich3-test/f77/rma/addsize.h b/teshsuite/smpi/mpich3-test/f77/rma/addsize.h
index 300e81f0bd..ed9d7e7e2c 100644
--- a/teshsuite/smpi/mpich3-test/f77/rma/addsize.h
+++ b/teshsuite/smpi/mpich3-test/f77/rma/addsize.h
@@ -1,4 +1,4 @@
-C -*- Mode: Fortran; -*- 
+C -*- Mode: Fortran; -*-
 C
 C  (C) 2003 by Argonne National Laboratory.
 C      See COPYRIGHT in top-level directory.
diff --git a/teshsuite/smpi/mpich3-test/f77/rma/c2f2cwin.c b/teshsuite/smpi/mpich3-test/f77/rma/c2f2cwin.c
index c09933ebd4..bd0c42c4f3 100644
--- a/teshsuite/smpi/mpich3-test/f77/rma/c2f2cwin.c
+++ b/teshsuite/smpi/mpich3-test/f77/rma/c2f2cwin.c
@@ -5,8 +5,8 @@
  *      See COPYRIGHT in top-level directory.
  */
 /*
- * This file contains the C routines used in testing the c2f and f2c 
- * handle conversion functions for MPI_Win 
+ * This file contains the C routines used in testing the c2f and f2c
+ * handle conversion functions for MPI_Win
  *
  * The tests follow this pattern:
  *
@@ -30,11 +30,11 @@
 #include "../../include/mpitestconf.h"
 #include <string.h>
 
-/* 
+/*
    Name mapping.  All routines are created with names that are lower case
    with a single trailing underscore.  This matches many compilers.
    We use #define to change the name for Fortran compilers that do
-   not use the lowercase/underscore pattern 
+   not use the lowercase/underscore pattern
 */
 
 #ifdef F77_NAME_UPPER
@@ -50,7 +50,7 @@
       defined(F77_NAME_MIXED_USCORE)
 /* Else leave name alone (routines have no underscore, so both
    of these map to a lowercase, single underscore) */
-#else 
+#else
 #error 'Unrecognized Fortran name mapping'
 #endif
 
@@ -79,7 +79,7 @@ int c2fwin_( int *win )
     return 0;
 }
 
-/* 
+/*
  * The following routines provide handles to the calling Fortran program
  */
 void f2cwin_( int *win )
diff --git a/teshsuite/smpi/mpich3-test/f90/rma/c2f902cwin.c b/teshsuite/smpi/mpich3-test/f90/rma/c2f902cwin.c
index ce35ccf8f7..fb18e1ba0d 100644
--- a/teshsuite/smpi/mpich3-test/f90/rma/c2f902cwin.c
+++ b/teshsuite/smpi/mpich3-test/f90/rma/c2f902cwin.c
@@ -6,8 +6,8 @@
  *      See COPYRIGHT in top-level directory.
  */
 /*
- * This file contains the C routines used in testing the c2f and f2c 
- * handle conversion functions for MPI_Win 
+ * This file contains the C routines used in testing the c2f and f2c
+ * handle conversion functions for MPI_Win
  *
  * The tests follow this pattern:
  *
@@ -31,11 +31,11 @@
 #include "../../include/mpitestconf.h"
 #include <string.h>
 
-/* 
+/*
    Name mapping.  All routines are created with names that are lower case
    with a single trailing underscore.  This matches many compilers.
    We use #define to change the name for Fortran compilers that do
-   not use the lowercase/underscore pattern 
+   not use the lowercase/underscore pattern
 */
 
 #ifdef F77_NAME_UPPER
@@ -51,7 +51,7 @@
       defined(F77_NAME_MIXED_USCORE)
 /* Else leave name alone (routines have no underscore, so both
    of these map to a lowercase, single underscore) */
-#else 
+#else
 #error 'Unrecognized Fortran name mapping'
 #endif
 
@@ -80,7 +80,7 @@ int c2fwin_( int *win )
     return 0;
 }
 
-/* 
+/*
  * The following routines provide handles to the calling Fortran program
  */
 void f2cwin_( int *win )
diff --git a/teshsuite/smpi/mpich3-test/include/mpitestconf.h b/teshsuite/smpi/mpich3-test/include/mpitestconf.h
index 03c66766d3..6ccd8014d0 100644
--- a/teshsuite/smpi/mpich3-test/include/mpitestconf.h
+++ b/teshsuite/smpi/mpich3-test/include/mpitestconf.h
@@ -2,7 +2,7 @@
 /* include/mpitestconf.h.in.  Generated from configure.ac by autoheader.  */
 
 /* -*- Mode: C; c-basic-offset:4 ; -*- */
-/*  
+/*
  *  (C) 2001 by Argonne National Laboratory.
  *      See COPYRIGHT in top-level directory.
  */
@@ -188,25 +188,25 @@
 #define POINTERINT_t int
 
 /* The size of `int', as computed by sizeof. */
-#define SIZEOF_INT 
+#define SIZEOF_INT
 
 /* The size of `long', as computed by sizeof. */
-#define SIZEOF_LONG 
+#define SIZEOF_LONG
 
 /* The size of `long long', as computed by sizeof. */
-#define SIZEOF_LONG_LONG 
+#define SIZEOF_LONG_LONG
 
 /* The size of `MPI_Offset', as computed by sizeof. */
-#define SIZEOF_MPI_OFFSET 
+#define SIZEOF_MPI_OFFSET
 
 /* The size of `short', as computed by sizeof. */
-#define SIZEOF_SHORT 
+#define SIZEOF_SHORT
 
 /* The size of `void *', as computed by sizeof. */
-#define SIZEOF_VOID_P 
+#define SIZEOF_VOID_P
 
 /* Define calling convention */
-#define STDCALL 
+#define STDCALL
 
 /* Define to 1 if you have the ANSI C header files. */
 #define STDC_HEADERS 1
diff --git a/teshsuite/smpi/type-struct/type-struct.c b/teshsuite/smpi/type-struct/type-struct.c
index 6e24484110..35460ea492 100644
--- a/teshsuite/smpi/type-struct/type-struct.c
+++ b/teshsuite/smpi/type-struct/type-struct.c
@@ -21,7 +21,7 @@ int main(int argc, char **argv)
 
     MPI_Comm_rank( MPI_COMM_WORLD, &rank );
 
-    int tab[2][3]={{1*rank,2*rank,3*rank},{7*rank,8*rank,9*rank}}; 
+    int tab[2][3]={{1*rank,2*rank,3*rank},{7*rank,8*rank,9*rank}};
     MPI_Type_contiguous(3, MPI_INT, &type2);
     MPI_Type_commit(&type2);
 
@@ -50,7 +50,7 @@ int main(int argc, char **argv)
       value.b=8.0;
     }else{
       value.a=10000;
-      value.b=5.0; 
+      value.b=5.0;
     }
 
     MPI_Bcast( &value, 1, mystruct, 0, MPI_COMM_WORLD );
diff --git a/teshsuite/surf/lmm_usage/lmm_usage.cpp b/teshsuite/surf/lmm_usage/lmm_usage.cpp
index fdc551ab54..01a2dfadb4 100644
--- a/teshsuite/surf/lmm_usage/lmm_usage.cpp
+++ b/teshsuite/surf/lmm_usage/lmm_usage.cpp
@@ -259,7 +259,7 @@ static void test3(method_t method)
 
   /* Creates the constraints */
   lmm_constraint_t *tmp_cnst = xbt_new0(lmm_constraint_t, 15);
-  for (int i = 0; i < 15; i++) 
+  for (int i = 0; i < 15; i++)
     tmp_cnst[i] = lmm_constraint_new(Sys, nullptr, B[i]);
 
   /* Creates the variables */
diff --git a/teshsuite/surf/maxmin_bench/maxmin_bench.cpp b/teshsuite/surf/maxmin_bench/maxmin_bench.cpp
index e5a911a07e..c4972b26da 100644
--- a/teshsuite/surf/maxmin_bench/maxmin_bench.cpp
+++ b/teshsuite/surf/maxmin_bench/maxmin_bench.cpp
@@ -48,7 +48,7 @@ static void test(int nb_cnst, int nb_var, int nb_elem, unsigned int pw_base_limi
     cnst[i] = lmm_constraint_new(Sys, NULL, float_random(10.0));
     int l;
     if(rate_no_limit>float_random(1.0))
-      //Look at what happens when there is no concurrency limit 
+      //Look at what happens when there is no concurrency limit
       l=-1;
     else
       //Badly logarithmically random concurrency limit in [2^pw_base_limit+1,2^pw_base_limit+2^pw_max_limit]
@@ -108,12 +108,12 @@ static void test(int nb_cnst, int nb_var, int nb_elem, unsigned int pw_base_limi
 }
 
 unsigned int TestClasses [][4]=
-  //Nbcnst Nbvar Baselimit Maxlimit 
+  //Nbcnst Nbvar Baselimit Maxlimit
   {{  10  ,10    ,1        ,2 }, //small
    {  100 ,100   ,3        ,6 }, //medium
    {  2000,2000  ,5        ,8 }, //big
    { 20000,20000 ,7        ,10}  //huge
-  }; 
+  };
 
 int main(int argc, char **argv)
 {
@@ -179,7 +179,7 @@ int main(int argc, char **argv)
     acc_date2+=date*date;
   }
 
-  float mean_date= acc_date/(float)testcount;  
+  float mean_date= acc_date/(float)testcount;
   float stdev_date= sqrt(acc_date2/(float)testcount-mean_date*mean_date);
 
   fprintf(stderr,
diff --git a/tools/cmake/DefinePackages.cmake b/tools/cmake/DefinePackages.cmake
index 1226d706d8..d07023604b 100644
--- a/tools/cmake/DefinePackages.cmake
+++ b/tools/cmake/DefinePackages.cmake
@@ -263,7 +263,6 @@ set(XBT_SRC
   src/xbt/exception.cpp
   src/xbt/graph.c
   src/xbt/heap.c
-  src/xbt/lib.c
   src/xbt/log.c
   src/xbt/mallocator.c
   src/xbt/memory_map.cpp
@@ -688,7 +687,6 @@ set(headers_to_install
   include/smpi/smpi_extended_traces.h
   include/smpi/smpi_extended_traces_fortran.h
   include/smpi/forward.hpp
-  include/surf/surf_routing.h
   include/xbt.h
   include/xbt/RngStream.h
   include/xbt/asserts.h
@@ -714,7 +712,6 @@ set(headers_to_install
   include/xbt/future.hpp
   include/xbt/graph.h
   include/xbt/heap.h
-  include/xbt/lib.h
   include/xbt/Extendable.hpp
   include/xbt/log.h
   include/xbt/log.hpp
diff --git a/tools/cmake/test_prog/prog_sem_init.c b/tools/cmake/test_prog/prog_sem_init.c
index 43540a0d12..5215f33394 100644
--- a/tools/cmake/test_prog/prog_sem_init.c
+++ b/tools/cmake/test_prog/prog_sem_init.c
@@ -12,7 +12,7 @@
 int main(void)
 {
   sem_t s;
-  if (sem_init(&s, 0, 0) != 0) 
+  if (sem_init(&s, 0, 0) != 0)
      return 1;
   return 0;
 }
diff --git a/tools/cmake/test_prog/prog_sem_open.c b/tools/cmake/test_prog/prog_sem_open.c
index ad0538460c..f983986dfc 100644
--- a/tools/cmake/test_prog/prog_sem_open.c
+++ b/tools/cmake/test_prog/prog_sem_open.c
@@ -25,6 +25,6 @@ int main(void) {
 //     printf("sem_open failed\n");
      return 1;
    }
-//   printf("sem_open succeeded\n");   
+//   printf("sem_open succeeded\n");
    return 0;
 }
diff --git a/tools/cmake/test_prog/prog_stackgrowth.c b/tools/cmake/test_prog/prog_stackgrowth.c
index dabc6b7872..dc7c0df0d8 100644
--- a/tools/cmake/test_prog/prog_stackgrowth.c
+++ b/tools/cmake/test_prog/prog_stackgrowth.c
@@ -15,7 +15,7 @@ static int growsdown(int *x)
   if (--iterate > 0)
     y = growsdown(&y);
 
-  /* The stack sometimes changes at the 0th level. 
+  /* The stack sometimes changes at the 0th level.
    * Original version did fail in this case, but I changed this around SimGrid 3.13 because of https://bugs.debian.org/814272
    * Every arch failed on that day :(
    */