Fixes #1868.

Bill found some misuse of MPIR_ERR_RECOVERABLE and few other error
messages and wrote a test in autogen.sh to discover/report them. This
fixes the errors raised by that test and re-enables it at autogen time.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>

maint/extracterrmsgs

@@ -24,7 +24,7 @@ $quiet = 0;
 $build_test_pgm = 1;
 # FIXME: checkErrClass should be set to 1; currently set to zero
 # to permit autogen.sh to complete
-$checkErrClass = 0;
+$checkErrClass = 1;
 
 # Strict is used to control checking of error message strings.
 $gStrict = 0;

src/include/mpimem.h

@@ -426,7 +426,7 @@ if (pointer_) { \
     void *realloc_tmp_ = MPIU_Realloc((ptr_), (size_)); \
     if ((size_) && !realloc_tmp_) { \
         MPIU_Free(ptr_); \
-        MPIU_ERR_SETANDJUMP2(rc_,MPIU_CHKMEM_ISFATAL,"**nomem2","**nomem2 %d %s",(size_),MPIU_QUOTE(ptr_)); \
+        MPIU_ERR_SETANDJUMP2(rc_,MPI_ERR_OTHER,"**nomem2","**nomem2 %d %s",(size_),MPIU_QUOTE(ptr_)); \
     } \
     (ptr_) = realloc_tmp_; \
 } while (0)
@@ -434,7 +434,7 @@ if (pointer_) { \
 #define MPIU_REALLOC_ORJUMP(ptr_,size_,rc_) do { \
     void *realloc_tmp_ = MPIU_Realloc((ptr_), (size_)); \
     if (size_) \
-        MPIU_ERR_CHKANDJUMP2(!realloc_tmp_,rc_,MPIU_CHKMEM_ISFATAL,"**nomem2","**nomem2 %d %s",(size_),MPIU_QUOTE(ptr_)); \
+        MPIU_ERR_CHKANDJUMP2(!realloc_tmp_,rc_,MPI_ERR_OTHER,"**nomem2","**nomem2 %d %s",(size_),MPIU_QUOTE(ptr_)); \
     (ptr_) = realloc_tmp_; \
 } while (0)
 

src/mpi/coll/alltoallv.c

@@ -475,7 +475,7 @@ int MPI_Alltoallv(const void *sendbuf, const int *sendcounts,
                 comm_size = comm_ptr->remote_size;
 
             if (comm_ptr->comm_kind == MPID_INTERCOMM && sendbuf == MPI_IN_PLACE) {
-                MPIU_ERR_SETANDJUMP(mpi_errno, MPIR_ERR_RECOVERABLE, "**sendbuf_inplace");
+                MPIU_ERR_SETANDJUMP(mpi_errno, MPI_ERR_OTHER, "**sendbuf_inplace");
             }
 
             for (i=0; i<comm_size; i++) {

src/mpi/coll/alltoallw.c

@@ -468,7 +468,7 @@ int MPI_Alltoallw(const void *sendbuf, const int sendcounts[],
             check_send = (comm_ptr->comm_kind == MPID_INTRACOMM && sendbuf != MPI_IN_PLACE);
 
             if (comm_ptr->comm_kind == MPID_INTERCOMM && sendbuf == MPI_IN_PLACE) {
-                MPIU_ERR_SETANDJUMP(mpi_errno, MPIR_ERR_RECOVERABLE, "**sendbuf_inplace");
+                MPIU_ERR_SETANDJUMP(mpi_errno, MPI_ERR_OTHER, "**sendbuf_inplace");
             }
 
             if (comm_ptr->comm_kind == MPID_INTRACOMM)

src/mpi/coll/ialltoallw.c

@@ -364,7 +364,7 @@ int MPI_Ialltoallw(const void *sendbuf, const int sendcounts[], const int sdispl
             MPIR_ERRTEST_ARGNULL(rdispls,"rdispls", mpi_errno);
             MPIR_ERRTEST_ARGNULL(recvtypes,"recvtypes", mpi_errno);
             if (comm_ptr->comm_kind == MPID_INTERCOMM && sendbuf == MPI_IN_PLACE) {
-                MPIU_ERR_SETANDJUMP(mpi_errno, MPIR_ERR_RECOVERABLE, "**sendbuf_inplace");
+                MPIU_ERR_SETANDJUMP(mpi_errno, MPI_ERR_OTHER, "**sendbuf_inplace");
             }
             MPIR_ERRTEST_ARGNULL(request,"request", mpi_errno);
             /* TODO more checks may be appropriate (counts, in_place, buffer aliasing, etc) */

src/mpi/comm/commutil.c

@@ -863,7 +863,7 @@ int MPIR_Get_contextid_sparse_group(MPID_Comm *comm_ptr, MPID_Group *group_ptr,
             int nfree = -1;
             int ntotal = -1;
             MPIR_ContextMaskStats(&nfree, &ntotal);
-            MPIU_ERR_SETANDJUMP3(mpi_errno, MPIR_ERR_RECOVERABLE,
+            MPIU_ERR_SETANDJUMP3(mpi_errno, MPI_ERR_OTHER,
                                  "**toomanycomm", "**toomanycomm %d %d %d",
                                  nfree, ntotal, ignore_id);
         }
@@ -874,7 +874,7 @@ int MPIR_Get_contextid_sparse_group(MPID_Comm *comm_ptr, MPID_Group *group_ptr,
             int nfree = -1;
             int ntotal = -1;
             MPIR_ContextMaskStats(&nfree, &ntotal);
-            MPIU_ERR_SETANDJUMP3(mpi_errno, MPIR_ERR_RECOVERABLE,
+            MPIU_ERR_SETANDJUMP3(mpi_errno, MPI_ERR_OTHER,
                                  "**toomanycomm", "**toomanycomm %d %d %d",
                                  nfree, ntotal, ignore_id);
         }
@@ -1141,7 +1141,7 @@ int MPIR_Get_contextid_sparse_group(MPID_Comm *comm_ptr, MPID_Group *group_ptr,
             }
 
             MPIR_ContextMaskStats(&nfree, &ntotal);
-            MPIU_ERR_SETANDJUMP3(mpi_errno, MPIR_ERR_RECOVERABLE,
+            MPIU_ERR_SETANDJUMP3(mpi_errno, MPI_ERR_OTHER,
                                  "**toomanycommfrag", "**toomanycommfrag %d %d %d",
                                  nfree, ntotal, ignore_id);
             /* --END ERROR HANDLING-- */
@@ -1192,7 +1192,7 @@ static int gcn_helper(MPID_Comm *comm, int tag, void *state)
         int nfree = -1;
         int ntotal = -1;
         MPIR_ContextMaskStats(&nfree, &ntotal);
-        MPIU_ERR_SETANDJUMP3(mpi_errno, MPIR_ERR_RECOVERABLE,
+        MPIU_ERR_SETANDJUMP3(mpi_errno, MPI_ERR_OTHER,
                              "**toomanycomm", "**toomanycomm %d %d %d",
                              nfree, ntotal, /*ignore_id=*/0);
     }
@@ -1557,7 +1557,7 @@ int MPIR_Comm_copy( MPID_Comm *comm_ptr, int size, MPID_Comm **outcomm_ptr )
         int nfree = -1;
         int ntotal = -1;
         MPIR_ContextMaskStats(&nfree, &ntotal);
-        MPIU_ERR_SETANDJUMP3(mpi_errno, MPIR_ERR_RECOVERABLE,
+        MPIU_ERR_SETANDJUMP3(mpi_errno, MPI_ERR_OTHER,
                              "**toomanycomm", "**toomanycomm %d %d %d",
                              nfree, ntotal, /*ignore_id=*/0);
     }

src/mpi/topo/dist_gr_create.c

@@ -316,7 +316,7 @@ int MPI_Dist_graph_create(MPI_Comm comm_old, int n, const int sources[],
         /* FIXME: Why not - there is only one allocated at a time. Is it only 
            that there is no defined macro to pop and free an item? */
         buf = MPIU_Malloc(count*sizeof(int));
-        MPIU_ERR_CHKANDJUMP(!buf, mpi_errno, MPIR_ERR_RECOVERABLE, "**nomem");
+        MPIU_ERR_CHKANDJUMP(!buf, mpi_errno, MPI_ERR_OTHER, "**nomem");
 
         mpi_errno = MPIC_Recv(buf, count, MPI_INT, MPI_ANY_SOURCE, MPIR_TOPO_A_TAG, comm_old, MPI_STATUS_IGNORE);
         /* FIXME: buf is never freed on error! */
@@ -350,7 +350,7 @@ int MPI_Dist_graph_create(MPI_Comm comm_old, int n, const int sources[],
         /* can't use CHKLMEM macros b/c we are in a loop */
         /* Why not? */
         buf = MPIU_Malloc(count*sizeof(int));
-        MPIU_ERR_CHKANDJUMP(!buf, mpi_errno, MPIR_ERR_RECOVERABLE, "**nomem");
+        MPIU_ERR_CHKANDJUMP(!buf, mpi_errno, MPI_ERR_OTHER, "**nomem");
 
         mpi_errno = MPIC_Recv(buf, count, MPI_INT, MPI_ANY_SOURCE, MPIR_TOPO_B_TAG, comm_old, MPI_STATUS_IGNORE);
         /* FIXME: buf is never freed on error! */

src/mpi/topo/dist_gr_create_adj.c

@@ -116,7 +116,7 @@ int MPI_Dist_graph_create_adjacent(MPI_Comm comm_old,
             if (indegree > 0) {
                 MPIR_ERRTEST_ARGNULL(sources, "sources", mpi_errno);
                 if (sourceweights == MPI_UNWEIGHTED && destweights != MPI_UNWEIGHTED) {
-                    MPIU_ERR_SET(mpi_errno, MPIR_ERR_RECOVERABLE, "**unweightedboth");
+                    MPIU_ERR_SET(mpi_errno, MPI_ERR_TOPOLOGY, "**unweightedboth");
                     goto fn_fail;
                 }
                 /* TODO check ranges for array elements too (**argarrayneg / **rankarray)*/
@@ -124,7 +124,7 @@ int MPI_Dist_graph_create_adjacent(MPI_Comm comm_old,
             if (outdegree > 0) {
                 MPIR_ERRTEST_ARGNULL(destinations, "destinations", mpi_errno);
                 if (destweights == MPI_UNWEIGHTED && sourceweights != MPI_UNWEIGHTED) {
-                    MPIU_ERR_SET(mpi_errno, MPIR_ERR_RECOVERABLE, "**unweightedboth");
+                    MPIU_ERR_SET(mpi_errno, MPI_ERR_TOPOLOGY, "**unweightedboth");
                     goto fn_fail;
                 }
             }