[SMPI] Replay: Make error message more verbose

[simgrid.git] / src / smpi / internals / smpi_replay.cpp

diff --git a/src/smpi/internals/smpi_replay.cpp b/src/smpi/internals/smpi_replay.cpp
index 278f0a5..2d03b91 100644 (file)
--- a/src/smpi/internals/smpi_replay.cpp
+++ b/src/smpi/internals/smpi_replay.cpp
@@ -16,25 +16,31 @@
 #include <memory>
 #include <numeric>
 #include <unordered_map>
+#include <sstream>
 #include <vector>
 
 using simgrid::s4u::Actor;
 
 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(smpi_replay,smpi,"Trace Replay with SMPI");
 
-static int active_processes  = 0;
 static std::unordered_map<int, std::vector<MPI_Request>*> reqq;
 
 static MPI_Datatype MPI_DEFAULT_TYPE;
 
 #define CHECK_ACTION_PARAMS(action, mandatory, optional)                                                               \
   {                                                                                                                    \
-    if (action.size() < static_cast<unsigned long>(mandatory + 2))                                                     \
+    if (action.size() < static_cast<unsigned long>(mandatory + 2)) {                                                     \
+      std::stringstream ss; \
+      for (const auto& elem : action) { \
+        ss << elem << " "; \
+      } \
       THROWF(arg_error, 0, "%s replay failed.\n"                                                                       \
                            "%zu items were given on the line. First two should be process_id and action.  "            \
                            "This action needs after them %lu mandatory arguments, and accepts %lu optional ones. \n"   \
+                           "The full line that was given is:\n   %s\n" \
                            "Please contact the Simgrid team if support is needed",                                     \
-             __func__, action.size(), static_cast<unsigned long>(mandatory), static_cast<unsigned long>(optional));    \
+             __func__, action.size(), static_cast<unsigned long>(mandatory), static_cast<unsigned long>(optional), ss.str().c_str());    \
+    }\
   }
 
 static void log_timed_action(simgrid::xbt::ReplayAction& action, double clock)
@@ -394,10 +400,9 @@ public:
 template <class T> class ReplayAction {
 protected:
   const std::string name;
+  const int my_proc_id;
   T args;
 
-  int my_proc_id;
-
 public:
   explicit ReplayAction(std::string name) : name(name), my_proc_id(simgrid::s4u::this_actor::getPid()) {}
   virtual ~ReplayAction() = default;
@@ -466,7 +471,7 @@ public:
   explicit SendAction(std::string name) : ReplayAction(name) {}
   void kernel(simgrid::xbt::ReplayAction& action) override
   {
-    int dst_traced = MPI_COMM_WORLD->group()->actor(args.partner)->getPid();
+    int dst_traced = MPI_COMM_WORLD->group()->actor(args.partner)->get_pid();
 
     TRACE_smpi_comm_in(my_proc_id, __func__, new simgrid::instr::Pt2PtTIData(name, args.partner, args.size,
                                                                              Datatype::encode(args.datatype1)));
@@ -492,7 +497,7 @@ public:
   explicit RecvAction(std::string name) : ReplayAction(name) {}
   void kernel(simgrid::xbt::ReplayAction& action) override
   {
-    int src_traced = MPI_COMM_WORLD->group()->actor(args.partner)->getPid();
+    int src_traced = MPI_COMM_WORLD->group()->actor(args.partner)->get_pid();
 
     TRACE_smpi_comm_in(my_proc_id, __func__, new simgrid::instr::Pt2PtTIData(name, args.partner, args.size,
                                                                              Datatype::encode(args.datatype1)));
@@ -567,9 +572,6 @@ public:
 
     /* start a simulated timer */
     smpi_process()->simulated_start();
-    /*initialize the number of active processes */
-    active_processes = smpi_process_count();
-
     set_reqq_self(new std::vector<MPI_Request>);
   }
 };
@@ -623,7 +625,7 @@ public:
   void kernel(simgrid::xbt::ReplayAction& action) override
   {
     TRACE_smpi_comm_in(my_proc_id, "action_bcast",
-                       new simgrid::instr::CollTIData("bcast", MPI_COMM_WORLD->group()->actor(args.root)->getPid(),
+                       new simgrid::instr::CollTIData("bcast", MPI_COMM_WORLD->group()->actor(args.root)->get_pid(),
                                                       -1.0, args.size, -1, Datatype::encode(args.datatype1), ""));
 
     Colls::bcast(send_buffer(args.size * args.datatype1->size()), args.size, args.datatype1, args.root, MPI_COMM_WORLD);
@@ -638,8 +640,9 @@ public:
   void kernel(simgrid::xbt::ReplayAction& action) override
   {
     TRACE_smpi_comm_in(my_proc_id, "action_reduce",
-                       new simgrid::instr::CollTIData("reduce", MPI_COMM_WORLD->group()->actor(args.root)->getPid(), args.comp_size,
-                                                      args.comm_size, -1, Datatype::encode(args.datatype1), ""));
+                       new simgrid::instr::CollTIData("reduce", MPI_COMM_WORLD->group()->actor(args.root)->get_pid(),
+                                                      args.comp_size, args.comm_size, -1,
+                                                      Datatype::encode(args.datatype1), ""));
 
     Colls::reduce(send_buffer(args.comm_size * args.datatype1->size()),
         recv_buffer(args.comm_size * args.datatype1->size()), args.comm_size, args.datatype1, MPI_OP_NULL, args.root, MPI_COMM_WORLD);
@@ -675,9 +678,9 @@ public:
                                                     Datatype::encode(args.datatype1),
                                                     Datatype::encode(args.datatype2)));
 
-    Colls::alltoall(send_buffer(args.send_size*args.comm_size* args.datatype1->size()), 
-      args.send_size, args.datatype1, recv_buffer(args.recv_size * args.comm_size * args.datatype2->size()),
-      args.recv_size, args.datatype2, MPI_COMM_WORLD);
+    Colls::alltoall(send_buffer(args.send_size * args.comm_size * args.datatype1->size()), args.send_size,
+                    args.datatype1, recv_buffer(args.recv_size * args.comm_size * args.datatype2->size()),
+                    args.recv_size, args.datatype2, MPI_COMM_WORLD);
 
     TRACE_smpi_comm_out(my_proc_id);
   }
@@ -716,14 +719,14 @@ public:
                                                Datatype::encode(args.datatype1), Datatype::encode(args.datatype2)));
 
     if (name == "gatherV") {
-      Colls::gatherv(send_buffer(args.send_size * args.datatype1->size()), args.send_size, args.datatype1, 
-                     (rank == args.root) ? recv_buffer(args.recv_size_sum  * args.datatype2->size()) : nullptr, args.recvcounts->data(), args.disps.data(), args.datatype2, args.root,
-                     MPI_COMM_WORLD);
+      Colls::gatherv(send_buffer(args.send_size * args.datatype1->size()), args.send_size, args.datatype1,
+                     (rank == args.root) ? recv_buffer(args.recv_size_sum * args.datatype2->size()) : nullptr,
+                     args.recvcounts->data(), args.disps.data(), args.datatype2, args.root, MPI_COMM_WORLD);
     }
     else {
-      Colls::allgatherv(send_buffer(args.send_size * args.datatype1->size()), args.send_size, args.datatype1, 
-                        recv_buffer(args.recv_size_sum * args.datatype2->size()), args.recvcounts->data(), args.disps.data(), args.datatype2,
-                    MPI_COMM_WORLD);
+      Colls::allgatherv(send_buffer(args.send_size * args.datatype1->size()), args.send_size, args.datatype1,
+                        recv_buffer(args.recv_size_sum * args.datatype2->size()), args.recvcounts->data(),
+                        args.disps.data(), args.datatype2, MPI_COMM_WORLD);
     }
 
     TRACE_smpi_comm_out(my_proc_id);
@@ -758,9 +761,10 @@ public:
           nullptr, Datatype::encode(args.datatype1),
           Datatype::encode(args.datatype2)));
 
-    Colls::scatterv((rank == args.root) ? send_buffer(args.send_size_sum * args.datatype1->size()) : nullptr, args.sendcounts->data(), args.disps.data(), 
-        args.datatype1, recv_buffer(args.recv_size * args.datatype2->size()), args.recv_size, args.datatype2, args.root,
-        MPI_COMM_WORLD);
+    Colls::scatterv((rank == args.root) ? send_buffer(args.send_size_sum * args.datatype1->size()) : nullptr,
+                    args.sendcounts->data(), args.disps.data(), args.datatype1,
+                    recv_buffer(args.recv_size * args.datatype2->size()), args.recv_size, args.datatype2, args.root,
+                    MPI_COMM_WORLD);
 
     TRACE_smpi_comm_out(my_proc_id);
   }
@@ -776,8 +780,9 @@ public:
                                                          std::to_string(args.comp_size), /* ugly hack to print comp_size */
                                                          Datatype::encode(args.datatype1)));
 
-    Colls::reduce_scatter(send_buffer(args.recv_size_sum * args.datatype1->size()), recv_buffer(args.recv_size_sum * args.datatype1->size()), 
-                          args.recvcounts->data(), args.datatype1, MPI_OP_NULL, MPI_COMM_WORLD);
+    Colls::reduce_scatter(send_buffer(args.recv_size_sum * args.datatype1->size()),
+                          recv_buffer(args.recv_size_sum * args.datatype1->size()), args.recvcounts->data(),
+                          args.datatype1, MPI_OP_NULL, MPI_COMM_WORLD);
 
     smpi_execute_flops(args.comp_size);
     TRACE_smpi_comm_out(my_proc_id);
@@ -858,6 +863,8 @@ void smpi_replay_init(int* argc, char*** argv)
 /** @brief actually run the replay after initialization */
 void smpi_replay_main(int* argc, char*** argv)
 {
+  static int active_processes = 0;
+  active_processes++;
   simgrid::xbt::replay_runner(*argc, *argv);
 
   /* and now, finalize everything */