X-Git-Url: http://info.iut-bm.univ-fcomte.fr/pub/gitweb/simgrid.git/blobdiff_plain/4d9809aa64447811ca65a1242d4f9e07972c6472..897c14185f8a5e6506c9037d4342329a460a7a95:/src/smpi/internals/smpi_global.cpp diff --git a/src/smpi/internals/smpi_global.cpp b/src/smpi/internals/smpi_global.cpp index dcc7d0fed2..cfec9be2af 100644 --- a/src/smpi/internals/smpi_global.cpp +++ b/src/smpi/internals/smpi_global.cpp @@ -1,10 +1,11 @@ -/* Copyright (c) 2007-2018. The SimGrid Team. All rights reserved. */ +/* Copyright (c) 2007-2019. 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 "mc/mc.h" #include "simgrid/s4u/Engine.hpp" +#include "simgrid/plugins/file_system.h" #include "smpi_coll.hpp" #include "smpi_f2c.hpp" #include "smpi_host.hpp" @@ -14,14 +15,25 @@ #include "xbt/config.hpp" #include +#include /* trim_right / trim_left */ +#include #include /* DBL_MAX */ #include #include /* intmax_t */ #include #include #include +#include + +#if SG_HAVE_SENDFILE +#include +#endif + +#if HAVE_PAPI +#include "papi.h" +#endif -#if not defined(__APPLE__) +#if not defined(__APPLE__) && not defined(__HAIKU__) #include #endif @@ -30,20 +42,14 @@ # ifndef MAC_OS_X_VERSION_10_12 # define MAC_OS_X_VERSION_10_12 101200 # endif -# define HAVE_WORKING_MMAP (MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_12) -#elif defined(__FreeBSD__) || defined(__FreeBSD_kernel__) -# define HAVE_WORKING_MMAP 0 +constexpr bool HAVE_WORKING_MMAP = (MAC_OS_X_VERSION_MIN_REQUIRED >= MAC_OS_X_VERSION_10_12); +#elif defined(__FreeBSD__) || defined(__FreeBSD_kernel__) || defined(__sun) || defined(__HAIKU__) +constexpr bool HAVE_WORKING_MMAP = false; #else -# define HAVE_WORKING_MMAP 1 -#endif - -#if HAVE_SENDFILE -#include +constexpr bool HAVE_WORKING_MMAP = true; #endif XBT_LOG_NEW_DEFAULT_SUBCATEGORY(smpi_kernel, smpi, "Logging specific to SMPI (kernel)"); -#include -#include /* trim_right / trim_left */ #if SMPI_IFORT extern "C" void for_rtl_init_ (int *, char **); @@ -59,23 +65,20 @@ XBT_LOG_NEW_DEFAULT_SUBCATEGORY(smpi_kernel, smpi, "Logging specific to SMPI (ke * See https://www.akkadia.org/drepper/dsohowto.pdf * and https://lists.freebsd.org/pipermail/freebsd-current/2016-March/060284.html */ -#if !defined(RTLD_DEEPBIND) || HAVE_SANITIZER_ADDRESS || HAVE_SANITIZER_THREAD +#if !RTLD_DEEPBIND || HAVE_SANITIZER_ADDRESS || HAVE_SANITIZER_THREAD #define WANT_RTLD_DEEPBIND 0 #else #define WANT_RTLD_DEEPBIND RTLD_DEEPBIND #endif #if HAVE_PAPI -#include "papi.h" std::string papi_default_config_name = "default"; std::map units2papi_setup; #endif -using simgrid::s4u::Actor; -using simgrid::s4u::ActorPtr; std::unordered_map location2speedup; -static std::map process_data; +static std::map process_data; int process_count = 0; static int smpi_exit_status = 0; int smpi_universe_size = 0; @@ -102,7 +105,8 @@ static const std::string smpi_default_instance_name("smpirun"); static simgrid::config::Flag smpi_init_sleep( "smpi/init", "Time to inject inside a call to MPI_Init", 0.0); -void (*smpi_comm_copy_data_callback) (smx_activity_t, void*, size_t) = &smpi_comm_copy_buffer_callback; +void (*smpi_comm_copy_data_callback)(simgrid::kernel::activity::CommImpl*, void*, + size_t) = &smpi_comm_copy_buffer_callback; int smpi_process_count() { @@ -111,47 +115,47 @@ int smpi_process_count() simgrid::smpi::ActorExt* smpi_process() { - ActorPtr me = Actor::self(); + simgrid::s4u::ActorPtr me = simgrid::s4u::Actor::self(); if (me == nullptr) // This happens sometimes (eg, when linking against NS3 because it pulls openMPI...) return nullptr; - return process_data.at(me); + return process_data.at(me.get()); } -simgrid::smpi::ActorExt* smpi_process_remote(ActorPtr actor) +simgrid::smpi::ActorExt* smpi_process_remote(simgrid::s4u::ActorPtr actor) { - return process_data.at(actor); + return process_data.at(actor.get()); } MPI_Comm smpi_process_comm_self(){ return smpi_process()->comm_self(); } -void smpi_process_init(int *argc, char ***argv){ +MPI_Info smpi_process_info_env(){ + return smpi_process()->info_env(); +} + +void smpi_process_init(int*, char***) +{ simgrid::smpi::ActorExt::init(); } void * smpi_process_get_user_data(){ - return Actor::self()->get_impl()->get_user_data(); + return simgrid::s4u::Actor::self()->get_impl()->get_user_data(); } void smpi_process_set_user_data(void *data){ - Actor::self()->get_impl()->set_user_data(data); -} - - -int smpi_global_size() -{ - char *value = getenv("SMPI_GLOBAL_SIZE"); - xbt_assert(value,"Please set env var SMPI_GLOBAL_SIZE to the expected number of processes."); - - return xbt_str_parse_int(value, "SMPI_GLOBAL_SIZE contains a non-numerical value: %s"); + simgrid::s4u::Actor::self()->get_impl()->set_user_data(data); } void smpi_comm_set_copy_data_callback(void (*callback) (smx_activity_t, void*, size_t)) { - smpi_comm_copy_data_callback = callback; + static void (*saved_callback)(smx_activity_t, void*, size_t); + saved_callback = callback; + smpi_comm_copy_data_callback = [](simgrid::kernel::activity::CommImpl* comm, void* buff, size_t size) { + saved_callback(smx_activity_t(comm), buff, size); + }; } static void memcpy_private(void* dest, const void* src, std::vector>& private_blocks) @@ -165,18 +169,14 @@ static void check_blocks(std::vector> &private_blocks, xbt_assert(block.first <= block.second && block.second <= buff_size, "Oops, bug in shared malloc."); } -void smpi_comm_copy_buffer_callback(smx_activity_t synchro, void *buff, size_t buff_size) +void smpi_comm_copy_buffer_callback(simgrid::kernel::activity::CommImpl* comm, void* buff, size_t buff_size) { - simgrid::kernel::activity::CommImplPtr comm = - boost::dynamic_pointer_cast(synchro); - int src_shared = 0; - int dst_shared = 0; size_t src_offset = 0; size_t dst_offset = 0; std::vector> src_private_blocks; std::vector> dst_private_blocks; XBT_DEBUG("Copy the data over"); - if((src_shared=smpi_is_shared(buff, src_private_blocks, &src_offset))) { + if(smpi_is_shared(buff, src_private_blocks, &src_offset)) { XBT_DEBUG("Sender %p is shared. Let's ignore it.", buff); src_private_blocks = shift_and_frame_private_blocks(src_private_blocks, src_offset, buff_size); } @@ -184,8 +184,8 @@ void smpi_comm_copy_buffer_callback(smx_activity_t synchro, void *buff, size_t b src_private_blocks.clear(); src_private_blocks.push_back(std::make_pair(0, buff_size)); } - if((dst_shared=smpi_is_shared((char*)comm->dst_buff, dst_private_blocks, &dst_offset))) { - XBT_DEBUG("Receiver %p is shared. Let's ignore it.", (char*)comm->dst_buff); + if (smpi_is_shared((char*)comm->dst_buff_, dst_private_blocks, &dst_offset)) { + XBT_DEBUG("Receiver %p is shared. Let's ignore it.", (char*)comm->dst_buff_); dst_private_blocks = shift_and_frame_private_blocks(dst_private_blocks, dst_offset, buff_size); } else { @@ -201,32 +201,33 @@ void smpi_comm_copy_buffer_callback(smx_activity_t synchro, void *buff, size_t b (static_cast(buff) >= smpi_data_exe_start) && (static_cast(buff) < smpi_data_exe_start + smpi_data_exe_size)) { XBT_DEBUG("Privatization : We are copying from a zone inside global memory... Saving data to temp buffer !"); - smpi_switch_data_segment(comm->src_proc->iface()); + smpi_switch_data_segment(comm->src_actor_->iface()); tmpbuff = static_cast(xbt_malloc(buff_size)); memcpy_private(tmpbuff, buff, private_blocks); } - if ((smpi_privatize_global_variables == SmpiPrivStrategies::MMAP) && ((char*)comm->dst_buff >= smpi_data_exe_start) && - ((char*)comm->dst_buff < smpi_data_exe_start + smpi_data_exe_size)) { + if ((smpi_privatize_global_variables == SmpiPrivStrategies::MMAP) && + ((char*)comm->dst_buff_ >= smpi_data_exe_start) && + ((char*)comm->dst_buff_ < smpi_data_exe_start + smpi_data_exe_size)) { XBT_DEBUG("Privatization : We are copying to a zone inside global memory - Switch data segment"); - smpi_switch_data_segment(comm->dst_proc->iface()); + smpi_switch_data_segment(comm->dst_actor_->iface()); } - XBT_DEBUG("Copying %zu bytes from %p to %p", buff_size, tmpbuff,comm->dst_buff); - memcpy_private(comm->dst_buff, tmpbuff, private_blocks); + XBT_DEBUG("Copying %zu bytes from %p to %p", buff_size, tmpbuff, comm->dst_buff_); + memcpy_private(comm->dst_buff_, tmpbuff, private_blocks); - if (comm->detached) { + if (comm->detached()) { // if this is a detached send, the source buffer was duplicated by SMPI // sender to make the original buffer available to the application ASAP xbt_free(buff); //It seems that the request is used after the call there this should be free somewhere else but where??? //xbt_free(comm->comm.src_data);// inside SMPI the request is kept inside the user data and should be free - comm->src_buff = nullptr; + comm->src_buff_ = nullptr; } if (tmpbuff != buff) xbt_free(tmpbuff); } -void smpi_comm_null_copy_buffer_callback(smx_activity_t comm, void *buff, size_t buff_size) +void smpi_comm_null_copy_buffer_callback(simgrid::kernel::activity::CommImpl*, void*, size_t) { /* nothing done in this version */ } @@ -238,11 +239,13 @@ static void smpi_check_options() xbt_assert(simgrid::config::get_value("smpi/async-small-thresh") <= simgrid::config::get_value("smpi/send-is-detached-thresh")); - if (simgrid::config::is_default("smpi/host-speed")) { + if (simgrid::config::is_default("smpi/host-speed") && not MC_is_active()) { XBT_INFO("You did not set the power of the host running the simulation. " "The timings will certainly not be accurate. " - "Use the option \"--cfg=smpi/host-speed:\" to set its value." - "Check http://simgrid.org/simgrid/latest/doc/options.html#options_smpi_bench for more information."); + "Use the option \"--cfg=smpi/host-speed:\" to set its value. " + "Check " + "https://simgrid.org/doc/latest/Configuring_SimGrid.html#automatic-benchmarking-of-smpi-code for more " + "information."); } xbt_assert(simgrid::config::get_value("smpi/cpu-threshold") >= 0, @@ -323,17 +326,15 @@ void smpi_global_init() int event_code = PAPI_NULL; char* event_name = const_cast((*events_it).c_str()); - if (PAPI_event_name_to_code(event_name, &event_code) == PAPI_OK) { - if (PAPI_add_event(event_set, event_code) != PAPI_OK) { - XBT_ERROR("Could not add PAPI event '%s'. Skipping.", event_name); - continue; - } else { - XBT_DEBUG("Successfully added PAPI event '%s' to the event set.", event_name); - } - } else { + if (PAPI_event_name_to_code(event_name, &event_code) != PAPI_OK) { XBT_CRITICAL("Could not find PAPI event '%s'. Skipping.", event_name); continue; } + if (PAPI_add_event(event_set, event_code) != PAPI_OK) { + XBT_ERROR("Could not add PAPI event '%s'. Skipping.", event_name); + continue; + } + XBT_DEBUG("Successfully added PAPI event '%s' to the event set.", event_name); counters2values.push_back( // We cannot just pass *events_it, as this is of type const basic_string @@ -366,7 +367,6 @@ void smpi_global_destroy() if (smpi_privatize_global_variables == SmpiPrivStrategies::MMAP) smpi_destroy_global_memory_segments(); - smpi_free_static(); if(simgrid::smpi::F2C::lookup() != nullptr) simgrid::smpi::F2C::delete_lookup(); } @@ -378,8 +378,10 @@ static void smpi_init_options(){ simgrid::smpi::Colls::set_collectives(); simgrid::smpi::Colls::smpi_coll_cleanup_callback = nullptr; smpi_cpu_threshold = simgrid::config::get_value("smpi/cpu-threshold"); - smpi_host_speed = simgrid::config::get_value("smpi/host-speed"); - xbt_assert(smpi_host_speed > 0.0, "You're trying to set the host_speed to a non-positive value (given: %f)", smpi_host_speed); + if (smpi_cpu_threshold < 0) + smpi_cpu_threshold = DBL_MAX; + + smpi_host_speed = simgrid::config::get_value("smpi/host-speed"); std::string smpi_privatize_option = simgrid::config::get_value("smpi/privatization"); if (smpi_privatize_option == "no" || smpi_privatize_option == "0") smpi_privatize_global_variables = SmpiPrivStrategies::NONE; @@ -396,15 +398,10 @@ static void smpi_init_options(){ XBT_DEBUG("Running without smpi_main(); disable smpi/privatization."); smpi_privatize_global_variables = SmpiPrivStrategies::NONE; } -#if !HAVE_WORKING_MMAP - if (smpi_privatize_global_variables == SmpiPrivStrategies::MMAP) { + if (not HAVE_WORKING_MMAP && smpi_privatize_global_variables == SmpiPrivStrategies::MMAP) { XBT_INFO("mmap privatization is broken on this platform, switching to dlopen privatization instead."); smpi_privatize_global_variables = SmpiPrivStrategies::DLOPEN; } -#endif - - if (smpi_cpu_threshold < 0) - smpi_cpu_threshold = DBL_MAX; std::string val = simgrid::config::get_value("smpi/shared-malloc"); if ((val == "yes") || (val == "1") || (val == "on") || (val == "global")) { @@ -423,8 +420,8 @@ typedef std::function smpi_entry_point_type; typedef int (* smpi_c_entry_point_type)(int argc, char **argv); typedef void (*smpi_fortran_entry_point_type)(); -static int smpi_run_entry_point(smpi_entry_point_type entry_point, const std::string& executable_path, - std::vector args) +template +static int smpi_run_entry_point(const F& entry_point, const std::string& executable_path, std::vector args) { // copy C strings, we need them writable std::vector* args4argv = new std::vector(args.size()); @@ -475,7 +472,7 @@ static smpi_entry_point_type smpi_resolve_function(void* handle) { smpi_fortran_entry_point_type entry_point_fortran = (smpi_fortran_entry_point_type)dlsym(handle, "user_main_"); if (entry_point_fortran != nullptr) { - return [entry_point_fortran](int argc, char** argv) { + return [entry_point_fortran](int, char**) { entry_point_fortran(); return 0; }; @@ -489,7 +486,7 @@ static smpi_entry_point_type smpi_resolve_function(void* handle) return smpi_entry_point_type(); } -static void smpi_copy_file(std::string src, std::string target, off_t fdin_size) +static void smpi_copy_file(const std::string& src, const std::string& target, off_t fdin_size) { int fdin = open(src.c_str(), O_RDONLY); xbt_assert(fdin >= 0, "Cannot read from %s. Please make sure that the file exists and is executable.", src.c_str()); @@ -497,40 +494,42 @@ static void smpi_copy_file(std::string src, std::string target, off_t fdin_size) xbt_assert(fdout >= 0, "Cannot write into %s", target.c_str()); XBT_DEBUG("Copy %" PRIdMAX " bytes into %s", static_cast(fdin_size), target.c_str()); - bool slow_copy = true; -#if HAVE_SENDFILE +#if SG_HAVE_SENDFILE ssize_t sent_size = sendfile(fdout, fdin, NULL, fdin_size); - if (sent_size == fdin_size) - slow_copy = false; - else if (sent_size != -1 || errno != ENOSYS) + if (sent_size == fdin_size) { + close(fdin); + close(fdout); + return; + } else if (sent_size != -1 || errno != ENOSYS) { xbt_die("Error while copying %s: only %zd bytes copied instead of %" PRIdMAX " (errno: %d -- %s)", target.c_str(), sent_size, static_cast(fdin_size), errno, strerror(errno)); + } #endif - if (slow_copy) { - const int bufsize = 1024 * 1024 * 4; - char buf[bufsize]; - while (int got = read(fdin, buf, bufsize)) { - if (got == -1) { - xbt_assert(errno == EINTR, "Cannot read from %s", src.c_str()); - } else { - char* p = buf; - int todo = got; - while (int done = write(fdout, p, todo)) { - if (done == -1) { - xbt_assert(errno == EINTR, "Cannot write into %s", target.c_str()); - } else { - p += done; - todo -= done; - } + // If this point is reached, sendfile() actually is not available. Copy file by hand. + const int bufsize = 1024 * 1024 * 4; + char* buf = new char[bufsize]; + while (int got = read(fdin, buf, bufsize)) { + if (got == -1) { + xbt_assert(errno == EINTR, "Cannot read from %s", src.c_str()); + } else { + char* p = buf; + int todo = got; + while (int done = write(fdout, p, todo)) { + if (done == -1) { + xbt_assert(errno == EINTR, "Cannot write into %s", target.c_str()); + } else { + p += done; + todo -= done; } } } } + delete[] buf; close(fdin); close(fdout); } -#if not defined(__APPLE__) +#if not defined(__APPLE__) && not defined(__HAIKU__) static int visit_libs(struct dl_phdr_info* info, size_t, void* data) { char* libname = (char*)(data); @@ -550,7 +549,6 @@ static void smpi_init_privatization_dlopen(const std::string& executable) struct stat fdin_stat; stat(executable.c_str(), &fdin_stat); off_t fdin_size = fdin_stat.st_size; - static std::size_t rank = 0; std::string libnames = simgrid::config::get_value("smpi/privatize-libs"); if (not libnames.empty()) { @@ -563,15 +561,13 @@ static void smpi_init_privatization_dlopen(const std::string& executable) void* libhandle = dlopen(libname.c_str(), RTLD_LAZY); // get library name from path char fullpath[512] = {'\0'}; - strcpy(fullpath, libname.c_str()); -#if not defined(__APPLE__) - int ret = dl_iterate_phdr(visit_libs, fullpath); - if (ret == 0) - xbt_die("Can't find a linked %s - check the setting you gave to smpi/privatize-libs", fullpath); - else - XBT_DEBUG("Extra lib to privatize found : %s", fullpath); + strncpy(fullpath, libname.c_str(), 511); +#if not defined(__APPLE__) && not defined(__HAIKU__) + xbt_assert(0 != dl_iterate_phdr(visit_libs, fullpath), + "Can't find a linked %s - check your settings in smpi/privatize-libs", fullpath); + XBT_DEBUG("Extra lib to privatize '%s' found", fullpath); #else - xbt_die("smpi/privatize-libs is not (yet) compatible with OSX"); + xbt_die("smpi/privatize-libs is not (yet) compatible with OSX nor with Haiku"); #endif privatize_libs_paths.push_back(fullpath); dlclose(libhandle); @@ -580,6 +576,7 @@ static void smpi_init_privatization_dlopen(const std::string& executable) simix_global->default_function = [executable, fdin_size](std::vector args) { return std::function([executable, fdin_size, args] { + static std::size_t rank = 0; // Copy the dynamic library: std::string target_executable = executable + "_" + std::to_string(getpid()) + "_" + std::to_string(rank) + ".so"; @@ -660,8 +657,6 @@ static void smpi_init_privatization_no_dlopen(const std::string& executable) int smpi_main(const char* executable, int argc, char* argv[]) { - srand(SMPI_RAND_SEED); - if (getenv("SMPI_PRETEND_CC") != nullptr) { /* Hack to ensure that smpicc can pretend to be a simple compiler. Particularly handy to pass it to the * configuration tools */ @@ -672,13 +667,7 @@ int smpi_main(const char* executable, int argc, char* argv[]) SIMIX_global_init(&argc, argv); SMPI_switch_data_segment = &smpi_switch_data_segment; - - // TODO This will not be executed in the case where smpi_main is not called, - // e.g., not for smpi_msg_masterslave. This should be moved to another location - // that is always called -- maybe close to Actor::on_creation? - simgrid::s4u::Host::on_creation.connect( - [](simgrid::s4u::Host& host) { host.extension_set(new simgrid::smpi::Host(&host)); }); - + sg_storage_file_system_init(); // parse the platform file: get the host list simgrid::s4u::Engine::get_instance()->load_platform(argv[1]); SIMIX_comm_set_copy_data_callback(smpi_comm_copy_buffer_callback); @@ -727,18 +716,20 @@ int smpi_main(const char* executable, int argc, char* argv[]) // Called either directly from the user code, or from the code called by smpirun void SMPI_init(){ - simgrid::s4u::Actor::on_creation.connect([](simgrid::s4u::ActorPtr actor) { - if (not actor->is_daemon()) { - process_data.insert({actor, new simgrid::smpi::ActorExt(actor, nullptr)}); + simgrid::s4u::Actor::on_creation.connect([](simgrid::s4u::Actor& actor) { + if (not actor.is_daemon()) { + process_data.insert({&actor, new simgrid::smpi::ActorExt(&actor, nullptr)}); } }); - simgrid::s4u::Actor::on_destruction.connect([](simgrid::s4u::ActorPtr actor) { - auto it = process_data.find(actor); + simgrid::s4u::Actor::on_destruction.connect([](simgrid::s4u::Actor const& actor) { + auto it = process_data.find(&actor); if (it != process_data.end()) { delete it->second; process_data.erase(it); } }); + simgrid::s4u::Host::on_creation.connect( + [](simgrid::s4u::Host& host) { host.extension_set(new simgrid::smpi::Host(&host)); }); smpi_init_options(); smpi_global_init();