March 2013 - commits - mailman3.cels.anl.gov

[mpich] MPICH primary repository annotated tag, v3.0.3, created. v3.0.3
by noreply＠mpich.org 28 Mar '13

28 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The annotated tag, v3.0.3 has been created at b6ef7ec081ed86814fed393fee1f03bae8fadce8 (tag) tagging 21750800ed2475c43e291dd78c0a285ea97babec (commit) replaces v3.0.2 tagged by Dave Goodell on Thu Mar 28 15:59:10 2013 -0500 - Log ----------------------------------------------------------------- tagging 'v3.0.3' Antonio J. Pena (1): Fixed MPI_Get_elements returning wrong values Dave Goodell (23): clmake: support more patterns tt#1754: fix warnings in ROMIO external32 code explain external32 limitations in RELEASE_NOTES clmake: more patterns and a regex fix fix Fortran MPI_WEIGHTS_EMPTY initialization hydra: squash loop warning from modern clang ROMIO: fix buffer/datatype mismatch in Isend/Irecv clmake: use `\W` instead of `\b` clmake: filter "install-sh" too .gitignore: fixup for [006dd97e] move `_NULL` handle definitions earlier in mpi.h fix compile error when `long double` is disabled configure: add `--disable-long-double` `long double _Complex` depends on `long double` clang annotations: fix `long double _Complex' mpi.h: guard `long double`-derived MPI types Merge branch 'fix-long-double' tests: gate `long double _Complex` fix `long double _Complex` logic from [a9305613] fix `complex<long double>` bug from [3f1fcead] `--enable-g=all` now means _all_, add `=most` update CHANGES for v3.0.3 release bump version numbers for v3.0.3 release James Dinan (48): Fix type incompat. for size_t in BSend degug msgs BUGFIX: Unlock piggybacking for Get_accumulate Eliminate enqueueing of lock op in RMA ops list PERF: Flush should ignore inactive targets PERF: Overlap req/wait for lock-all on shr men win Fix incorrect timing reporting in llist bench Minor cleanups to MCS mutex test Added missing thread safety to Comm_compare Fixed several warnings in lockcontention3 test Cleaned up test compiler warnings Temporarily reverted is_gacc_op bugfix Consolidated RMA op finalization code Converted packet type field to 16-bits Added flags field to RMA op packet headers Integrated flags into RMA op processing Added flags to MPID_Request RMA sync. piggybacking from origin->target Simplify get completion Cleanup of FOP packet header Removed unused single_op_opt field from MPID_Request Flush piggybacking Implemented lock op piggybacking for MODE_NOCHECK Add a MODE_NOCHECK version of the llist bench test Cleaned up error names for CH3 RMA implementation Finished adding memory fences for shr mem win Remove unnecessary flush for empty epochs Added missing CPPFLAGS in rma tests makefile Merge branch 'packet-header-flags' Added missing descriptions for new RMA err classes Added missing function boilerplate decls Moved comm_split_type override from CH3 to Nemesis CH3 default shared memory window implementation Moved window flavor error checking to public API Merge branch 'shr-mem-win-ch3' Added missing CS ENTER/EXIT guards to type_commit Updated changes file with RMA changes for 3.0.3 Updated req-based ops to handle PROC_NULL Cleaned up handling of PROC_NULL in RMA ops Added new operations to RMA PROC_NULL test Fix handling of 0-byte xfers in req-gen ops Test handling of 0-byte RMA xfers Cleanup comment in rmazero test Fix req. op. completion outside of PT epoch Add additional wait/test cases to Req. ops. test Fix Irecv race in wait/test_all/any/some err tests Replace calls to sleep with MTestSleep in tests Fix coll6 test hanging with more than 10 processes Updated CHANGES file Pavan Balaji (6): Bug-fix: Allow configure to check for cobalt as well. Added support for Cobalt to the CHANGES file. Upgrade to hwloc-1.6.1rc2. This version uses libpciaccess instead of Merge hydra-specific hwloc patches from Updated the CHANGES file with the hwloc-1.6.2rc1 information. Bug-fix: correct MPIR_proctable setup when we use wraparound ranks. Ralf Gunter (1): README: fix pamid configuration example Rob Latham (1): feature-encode NTFS quirk William Gropp (5): Added support for passing the length of a Fortran CHARACTER as size_t instead of int. Added a configure option to set; note that despite documentation to the contrary, it looks like some compilers still pass the length as an int (the common practice for decades), so changing to size_t must be made with extreme caution Added needed tests based on coverage analysis Added tests for user-defined error handlers in Fortran Files that were modified to supprt the tests of user-defined error handlers from Fortran More files for user error handler tests ----------------------------------------------------------------------- hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-85-g2175080
by noreply＠mpich.org 28 Mar '13

28 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via 21750800ed2475c43e291dd78c0a285ea97babec (commit) from 9ee964cc057a58a5d484ccd92c98ad8f69cce9e0 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/21750800ed2475c43e291dd78c0a285ea… commit 21750800ed2475c43e291dd78c0a285ea97babec Author: Dave Goodell <goodell(a)mcs.anl.gov> Date: Thu Mar 28 15:57:03 2013 -0500 bump version numbers for v3.0.3 release AFAICS, no changes of substance were made to mpi.h or any of the buildiface scripts, so the ABI string only has its "revision" field incremented by 1 to 10:3:0. No reviewer. diff --git a/maint/version.m4 b/maint/version.m4 index 03dbaab..0660b45 100644 --- a/maint/version.m4 +++ b/maint/version.m4 @@ -14,7 +14,7 @@ # changing this by playing with diversions, but then we would probably be # playing with autotools-fire. -m4_define([MPICH_VERSION_m4],[3.0.2])dnl +m4_define([MPICH_VERSION_m4],[3.0.3])dnl m4_define([MPICH_RELEASE_DATE_m4],[unreleased development copy])dnl # For libtool ABI versioning rules see: @@ -31,6 +31,6 @@ m4_define([MPICH_RELEASE_DATE_m4],[unreleased development copy])dnl # # 4. If any interfaces have been removed since the last public # release, then set age to 0. -m4_define([libmpich_so_version_m4],[10:2:0])dnl +m4_define([libmpich_so_version_m4],[10:3:0])dnl [#] end of __file__ ----------------------------------------------------------------------- Summary of changes: maint/version.m4 | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-84-g9ee964c
by noreply＠mpich.org 28 Mar '13

28 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via 9ee964cc057a58a5d484ccd92c98ad8f69cce9e0 (commit) from a9d4db5f4bd74cfa2e35e63c17bd622711bb44a9 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/9ee964cc057a58a5d484ccd92c98ad8f6… commit 9ee964cc057a58a5d484ccd92c98ad8f69cce9e0 Author: James Dinan <dinan(a)mcs.anl.gov> Date: Thu Mar 28 14:32:45 2013 -0500 Updated CHANGES file diff --git a/CHANGES b/CHANGES index 536973f..10f35dd 100644 --- a/CHANGES +++ b/CHANGES @@ -13,6 +13,9 @@ # RMA: Added a default implementation of shared memory windows to CH3. This adds support for this MPI 3.0 feature to the ch3:sock device. + # RMA: Fix a bug that resulted in an error when RMA operation request handles + where completed outside of a synchronization epoch. + # PM/PMI: Upgraded to hwloc-1.6.2rc1. This version uses libpciaccess instead of libpci, to workaround the GPL license used by libpci. ----------------------------------------------------------------------- Summary of changes: CHANGES | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-83-ga9d4db5
by noreply＠mpich.org 28 Mar '13

28 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via a9d4db5f4bd74cfa2e35e63c17bd622711bb44a9 (commit) from 7b7b759223da7347b24313e89a76193d267d7afc (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/a9d4db5f4bd74cfa2e35e63c17bd62271… commit a9d4db5f4bd74cfa2e35e63c17bd622711bb44a9 Author: Dave Goodell <goodell(a)mcs.anl.gov> Date: Thu Mar 28 14:00:49 2013 -0500 update CHANGES for v3.0.3 release diff --git a/CHANGES b/CHANGES index 20f4067..536973f 100644 --- a/CHANGES +++ b/CHANGES @@ -18,6 +18,13 @@ # PM/PMI: Added support for the Cobalt process manager. + # BUILD SYSTEM: allow MPI_LONG_DOUBLE_SUPPORT to be disabled with a configure + option. + + # FORTRAN: fix MPI_WEIGHTS_EMPTY in the Fortran bindings + + # MISC: fix a bug in MPI_Get_elements where it could return incorrect values + # Several other minor bug fixes, memory leak fixes, and code cleanup. A full list of changes is available at the following link: ----------------------------------------------------------------------- Summary of changes: CHANGES | 7 +++++++ 1 files changed, 7 insertions(+), 0 deletions(-) hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-82-g7b7b759
by noreply＠mpich.org 28 Mar '13

28 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via 7b7b759223da7347b24313e89a76193d267d7afc (commit) from 317b2f3e7ab4efb58b20cf51e35871d4e88f630e (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/7b7b759223da7347b24313e89a76193d2… commit 7b7b759223da7347b24313e89a76193d267d7afc Author: Ralf Gunter <ralfgunter(a)gmail.com> Date: Wed Mar 27 16:38:25 2013 -0500 README: fix pamid configuration example An old driver version (`V1R1M2`) was hard-coded into the compiler paths. Reviewed-by: goodell diff --git a/README.vin b/README.vin index c7f7d3e..6910baa 100644 --- a/README.vin +++ b/README.vin @@ -502,10 +502,10 @@ configure options can be used: --enable-error-messages --disable-debuginfo --enable-thread-cs=per-object --enable-handle-allocation=tls --enable-refcount=lock-free --disable-predefined-refcount - CC=/bgsys/drivers/V1R1M2/ppc64/gnu-linux/bin/powerpc64-bgq-linux-gcc - CXX=/bgsys/drivers/V1R1M2/ppc64/gnu-linux/bin/powerpc64-bgq-linux-g++ - F77=/bgsys/drivers/V1R1M2/ppc64/gnu-linux/bin/powerpc64-bgq-linux-gfortran - FC=/bgsys/drivers/V1R1M2/ppc64/gnu-linux/bin/powerpc64-bgq-linux-gfortran + CC=powerpc64-bgq-linux-gcc + CXX=powerpc64-bgq-linux-g++ + F77=powerpc64-bgq-linux-gfortran + FC=powerpc64-bgq-linux-gfortran ------------------------------------------------------------------------- ----------------------------------------------------------------------- Summary of changes: README.vin | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-) hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-81-g317b2f3
by noreply＠mpich.org 21 Mar '13

21 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via 317b2f3e7ab4efb58b20cf51e35871d4e88f630e (commit) via a0e9eb583b383f234bedd58a7af9d132dcdd45bf (commit) via 803fda10de941c79e2dc1bfbcfe65d0cb68e328f (commit) from 5ac51edf49d664d33b4a88fc6a5cebeb51950149 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/317b2f3e7ab4efb58b20cf51e35871d4e… commit 317b2f3e7ab4efb58b20cf51e35871d4e88f630e Author: James Dinan <dinan(a)mcs.anl.gov> Date: Thu Mar 21 11:35:27 2013 -0500 Fix coll6 test hanging with more than 10 processes This closes ticket #1798. Patch provided by Jeff Hammond. Reviewer: dinan diff --git a/test/mpi/coll/coll6.c b/test/mpi/coll/coll6.c index 5691357..6e8ea36 100644 --- a/test/mpi/coll/coll6.c +++ b/test/mpi/coll/coll6.c @@ -17,21 +17,23 @@ int main( int argc, char **argv ) int participants; int displs[MAX_PROCESSES]; int recv_counts[MAX_PROCESSES]; + MPI_Comm test_comm; MTest_Init( &argc, &argv ); MPI_Comm_rank( MPI_COMM_WORLD, &rank ); MPI_Comm_size( MPI_COMM_WORLD, &size ); /* A maximum of MAX_PROCESSES processes can participate */ - if ( size > MAX_PROCESSES ) participants = MAX_PROCESSES; - else participants = size; - /* while (MAX_PROCESSES % participants) participants--; */ + participants = ( size > MAX_PROCESSES ) ? MAX_PROCESSES : size; + if (MAX_PROCESSES % participants) { fprintf( stderr, "Number of processors must divide %d\n", MAX_PROCESSES ); MPI_Abort( MPI_COMM_WORLD, 1 ); } - if ( (rank < participants) ) { + MPI_Comm_split(MPI_COMM_WORLD, rank<participants, rank, &test_comm); + + if ( rank < participants ) { /* Determine what rows are my responsibility */ int block_size = MAX_PROCESSES / participants; @@ -53,7 +55,7 @@ int main( int argc, char **argv ) /* Everybody gets the gathered data */ MPI_Allgatherv(&table[begin_row][0], send_count, MPI_INT, &table[0][0], recv_counts, displs, - MPI_INT, MPI_COMM_WORLD); + MPI_INT, test_comm); /* Everybody should have the same table now. @@ -80,6 +82,8 @@ int main( int argc, char **argv ) } MTest_Finalize( errors ); + + MPI_Comm_free(&test_comm); MPI_Finalize(); return MTestReturnValue( errors ); } http://git.mpich.org/mpich.git/commitdiff/a0e9eb583b383f234bedd58a7af9d132d… commit a0e9eb583b383f234bedd58a7af9d132dcdd45bf Author: James Dinan <dinan(a)mcs.anl.gov> Date: Thu Mar 21 09:23:43 2013 -0500 Replace calls to sleep with MTestSleep in tests This replaces calls to non-portable (to e.g. Windows) sleep with calls to MTestSleep. Submitted by Jeff Hammond. Where possible, dependencies on unistd.h have been removed. Reviewer: dinan diff --git a/test/mpi/basic/wtime.c b/test/mpi/basic/wtime.c index bcad008..ec09cb3 100644 --- a/test/mpi/basic/wtime.c +++ b/test/mpi/basic/wtime.c @@ -5,7 +5,7 @@ */ #include "mpi.h" #include <stdio.h> -#include <unistd.h> +#include "mpitest.h" int main(int argc, char *argv[]) { @@ -16,7 +16,7 @@ int main(int argc, char *argv[]) MPI_Comm_rank(MPI_COMM_WORLD, &i); dStart = MPI_Wtime(); - sleep(1); + MTestSleep(1); dFinish = MPI_Wtime(); dDuration = dFinish - dStart; diff --git a/test/mpi/errors/faults/pt2ptf2.c b/test/mpi/errors/faults/pt2ptf2.c index 248eb28..0808ac2 100644 --- a/test/mpi/errors/faults/pt2ptf2.c +++ b/test/mpi/errors/faults/pt2ptf2.c @@ -7,7 +7,6 @@ #include "mpi.h" #include <stdio.h> #include <stdlib.h> -#include <unistd.h> #include "mpitest.h" /* @@ -49,7 +48,7 @@ int main( int argc, char *argv[] ) else { /* To improve the chance that the "faulted" processes will have exited, wait for 1 second */ - sleep( 1 ); + MTestSleep( 1 ); } /* Can we still use newcomm? */ diff --git a/test/mpi/group/gtranksperf.c b/test/mpi/group/gtranksperf.c index 9dfb52e..98b79d2 100644 --- a/test/mpi/group/gtranksperf.c +++ b/test/mpi/group/gtranksperf.c @@ -9,9 +9,6 @@ #include "mpitest.h" #include <math.h> /* for fabs(3) */ -#ifdef HAVE_UNISTD_H -#include <unistd.h> /* for sleep(3) */ -#endif /* Measure and compare the relative performance of MPI_Group_translate_ranks * with small and large group2 sizes but a constant number of ranks. This @@ -89,10 +86,10 @@ int main( int argc, char *argv[] ) MPI_Barrier(comm); if (rank != 0) { - sleep(10); + MTestSleep(10); } else /* rank==0 */ { - sleep(1); /* try to avoid timing while everyone else is making syscalls */ + MTestSleep(1); /* try to avoid timing while everyone else is making syscalls */ MPI_Group_translate_ranks(grev, size, ranks, gworld, ranksout); /*throwaway iter*/ start = MPI_Wtime(); diff --git a/test/mpi/manual/mpi_t/nem_fbox_fallback_to_queue_count.c b/test/mpi/manual/mpi_t/nem_fbox_fallback_to_queue_count.c index c6dec0b..04c1cba 100644 --- a/test/mpi/manual/mpi_t/nem_fbox_fallback_to_queue_count.c +++ b/test/mpi/manual/mpi_t/nem_fbox_fallback_to_queue_count.c @@ -19,10 +19,6 @@ #include <stdio.h> #include "mpitest.h" -#ifdef HAVE_UNISTD_H -#include <unistd.h> -#endif - #define TRY(func) \ do { \ err = (func); \ @@ -41,7 +37,6 @@ MPI_T_pvar_handle fbox_handle; /* Check that we can successfuly write to the variable. */ void blank_test() { - int i; uint64_t temp[2] = { -1 }; temp[0] = 0x1234; temp[1] = 0xABCD; @@ -92,13 +87,13 @@ void send_first_test() * FIXME: Ideally this should use a barrier, but that uses messages * internally and hence will sometimes screw up the asserts above. */ - sleep(1); + MTestSleep(1); } else if (rank == 1) { char recv_buf[BUF_COUNT]; MPI_Status status; - sleep(1); /* see above */ + MTestSleep(1); /* see above */ MPI_Recv(recv_buf, BUF_COUNT, MPI_CHAR, 0, 0, MPI_COMM_WORLD, &status); MPI_Recv(recv_buf, BUF_COUNT, MPI_CHAR, 0, 0, MPI_COMM_WORLD, &status); diff --git a/test/mpi/manual/spawntest_master.c b/test/mpi/manual/spawntest_master.c index 62e87fd..b576649 100644 --- a/test/mpi/manual/spawntest_master.c +++ b/test/mpi/manual/spawntest_master.c @@ -11,7 +11,7 @@ #include "mpi.h" #include <stdio.h> #include <stdlib.h> -#include <unistd.h> +#include "mpitest.h" #define NUM_SPAWNS 4 @@ -48,11 +48,11 @@ int main( int argc, char *argv[] ) { MPI_Comm_free( &allcomm ); MPI_Comm_disconnect( &intercomm ); - fprintf(stdout, "%s:%d: Sleep starting; children should exit\n", + fprintf(stdout, "%s:%d: MTestSleep starting; children should exit\n", __FILE__, __LINE__ );fflush(stdout); - sleep(30); + MTestSleep(30); fprintf(stdout, - "%s:%d: Sleep done; all children should have already exited\n", + "%s:%d: MTestSleep done; all children should have already exited\n", __FILE__, __LINE__ );fflush(stdout); MPI_Finalize(); diff --git a/test/mpi/manual/testconnectserial.c b/test/mpi/manual/testconnectserial.c index 88d3bf4..8e3fc69 100644 --- a/test/mpi/manual/testconnectserial.c +++ b/test/mpi/manual/testconnectserial.c @@ -11,6 +11,7 @@ #include <string.h> #include <time.h> #include <unistd.h> +#include "mpitest.h" #include "connectstuff.h" @@ -119,7 +120,7 @@ int main( int argc, char ** argv ) { if( expectedRank == 0 ) { /* Cleanup on rank zero - delete some files */ - sleep( 4 ); + MTestSleep( 4 ); unlink( actualFname ); free( actualFname ); unlink( globalFname ); diff --git a/test/mpi/threads/comm/comm_dup_deadlock.c b/test/mpi/threads/comm/comm_dup_deadlock.c index c44b8fd..1bfc29a 100644 --- a/test/mpi/threads/comm/comm_dup_deadlock.c +++ b/test/mpi/threads/comm/comm_dup_deadlock.c @@ -4,7 +4,6 @@ * See COPYRIGHT in top-level directory. */ -#include <unistd.h> #include <stdio.h> #include <mpi.h> #include "mpitest.h" @@ -21,10 +20,6 @@ } \ } while (0) -#ifdef HAVE_WINDOWS_H -#define sleep(a) Sleep(a*1000) -#endif - MPI_Comm comms[NUM_THREADS]; MTEST_THREAD_LOCK_TYPE comm_lock; int rank, size; @@ -41,7 +36,7 @@ MTEST_THREAD_RETURN_TYPE test_comm_dup(void *arg) MPI_Comm comm, self_dup; if (*(int*)arg == rank) { - sleep(1); + MTestSleep(1); } MTest_thread_lock(&comm_lock); @@ -66,7 +61,7 @@ MTEST_THREAD_RETURN_TYPE test_comm_dup(void *arg) int main(int argc, char **argv) { int thread_args[NUM_THREADS]; - int i, err, provided; + int i, provided; MPI_Init_thread(&argc, &argv, MPI_THREAD_MULTIPLE, &provided); diff --git a/test/mpi/threads/comm/ctxdup.c b/test/mpi/threads/comm/ctxdup.c index 3f3a9b3..6d3aaa8 100644 --- a/test/mpi/threads/comm/ctxdup.c +++ b/test/mpi/threads/comm/ctxdup.c @@ -7,7 +7,6 @@ #include "mpi.h" #include <stdio.h> #include <stdlib.h> -#include <unistd.h> #include "mpitest.h" #include "mpithreadtest.h" @@ -15,10 +14,6 @@ static char MTEST_Descrip[] = "Creating communications concurrently in different threads"; */ -#ifdef HAVE_WINDOWS_H -#define sleep(a) Sleep(a*1000) -#endif - MTEST_THREAD_RETURN_TYPE dup_thread(void *); MTEST_THREAD_RETURN_TYPE dup_thread(void *p) @@ -88,7 +83,7 @@ int main( int argc, char *argv[] ) /* If we're odd, send to our new thread and then delay */ if (rank & 0x1) { MPI_Ssend( buffer, 0, MPI_INT, rank, 0, MPI_COMM_WORLD ); - sleep(1); + MTestSleep(1); } MPI_Comm_dup( comm1, &comm4 ); http://git.mpich.org/mpich.git/commitdiff/803fda10de941c79e2dc1bfbcfe65d0cb… commit 803fda10de941c79e2dc1bfbcfe65d0cb68e328f Author: James Dinan <dinan(a)mcs.anl.gov> Date: Tue Mar 19 20:31:46 2013 -0500 Fix Irecv race in wait/test_all/any/some err tests This fixes a race in the ERR_IN_STATUS tests for wait/test_all/any/some tests. In these tests, there was a race that could cause the tested error to occur within the Irecv call, rather than the wait/test call (expected behavior). This race was happening when MPICH_ASYNC_PROGRESS was enabled. It should be noted that tests tests still rely on non-standard behavior, in particular some/any calls are expected to complete all operations. This closes ticket #1802. Reviewer: balaji diff --git a/test/mpi/errors/pt2pt/errinstatta.c b/test/mpi/errors/pt2pt/errinstatta.c index 861507f..49929bd 100644 --- a/test/mpi/errors/pt2pt/errinstatta.c +++ b/test/mpi/errors/pt2pt/errinstatta.c @@ -47,7 +47,19 @@ int main( int argc, char *argv[] ) printf( "Error returned from Irecv\n" ); } - /* synchronize */ + /* Wait for Irecvs to be posted before the sender calls send. This + * prevents the operation from completing and returning an error in the + * Irecv. */ + errval = MPI_Recv(NULL, 0, MPI_INT, src, 100, comm, MPI_STATUS_IGNORE); + if (errval) { + errs++; + MTestPrintError( errval ); + printf( "Error returned from Recv\n" ); + } + + /* Wait for sends to complete at the sender before proceeding */ + /* WARNING: This does not guarantee that the sends are ready to + * complete at the receiver. */ errval = MPI_Recv(NULL, 0, MPI_INT, src, 10, comm, MPI_STATUS_IGNORE); if (errval) { errs++; @@ -57,6 +69,9 @@ int main( int argc, char *argv[] ) for (i=0; i<2; i++) { s[i].MPI_ERROR = -1; } + + /* WARNING: The following assumes that Testall will complete both + * send/irecv pairs. This is *not* guaranteed by the MPI standard. */ errval = MPI_Testall( 2, r, &flag, s ); MPI_Error_class( errval, &errclass ); if (errclass != MPI_ERR_IN_STATUS) { @@ -84,12 +99,15 @@ int main( int argc, char *argv[] ) } else if (rank == src) { + /* Wait for Irecvs to be posted before the sender calls send */ + MPI_Ssend( NULL, 0, MPI_INT, dest, 100, comm ); + /* Send test messages, then send another message so that the test does not start until we are sure that the sends have begun */ MPI_Send( b1, 10, MPI_INT, dest, 0, comm ); MPI_Send( b2, 11, MPI_INT, dest, 10, comm ); - /* synchronize */ + /* Wait for sends to complete before proceeding to the testsome. */ MPI_Ssend( NULL, 0, MPI_INT, dest, 10, comm ); } diff --git a/test/mpi/errors/pt2pt/errinstatts.c b/test/mpi/errors/pt2pt/errinstatts.c index e588373..dd247f2 100644 --- a/test/mpi/errors/pt2pt/errinstatts.c +++ b/test/mpi/errors/pt2pt/errinstatts.c @@ -49,7 +49,19 @@ int main( int argc, char *argv[] ) printf( "Error returned from Irecv\n" ); } - /* synchronize */ + /* Wait for Irecvs to be posted before the sender calls send. This + * prevents the operation from completing and returning an error in the + * Irecv. */ + errval = MPI_Recv(NULL, 0, MPI_INT, src, 100, comm, MPI_STATUS_IGNORE); + if (errval) { + errs++; + MTestPrintError( errval ); + printf( "Error returned from Recv\n" ); + } + + /* Wait for sends to complete at the sender before proceeding */ + /* WARNING: This does not guarantee that the sends are ready to + * complete at the receiver. */ errval = MPI_Recv(NULL, 0, MPI_INT, src, 10, comm, MPI_STATUS_IGNORE); if (errval) { errs++; @@ -59,6 +71,9 @@ int main( int argc, char *argv[] ) for (i=0; i<2; i++) { s[i].MPI_ERROR = -1; } + + /* WARNING: The following assumes that Testsome will complete both + * send/irecv pairs. This is *not* guaranteed by the MPI standard. */ errval = MPI_Testsome( 2, r, &outcount, indices, s ); MPI_Error_class( errval, &errclass ); if (errclass != MPI_ERR_IN_STATUS) { @@ -89,12 +104,15 @@ int main( int argc, char *argv[] ) } else if (rank == src) { + /* Wait for Irecvs to be posted before the sender calls send */ + MPI_Ssend( NULL, 0, MPI_INT, dest, 100, comm ); + /* Send test messages, then send another message so that the test does not start until we are sure that the sends have begun */ MPI_Send( b1, 10, MPI_INT, dest, 0, comm ); MPI_Send( b2, 11, MPI_INT, dest, 10, comm ); - /* synchronize */ + /* Wait for sends to complete before proceeding to the testsome. */ MPI_Ssend( NULL, 0, MPI_INT, dest, 10, comm ); } diff --git a/test/mpi/errors/pt2pt/errinstatwa.c b/test/mpi/errors/pt2pt/errinstatwa.c index 98a70ec..99308e4 100644 --- a/test/mpi/errors/pt2pt/errinstatwa.c +++ b/test/mpi/errors/pt2pt/errinstatwa.c @@ -48,7 +48,16 @@ int main( int argc, char *argv[] ) printf( "Error returned from Irecv\n" ); } - errval = MPI_Barrier(comm); + /* Wait for Irecvs to be posted before the sender calls send. This + * prevents the operation from completing and returning an error in the + * Irecv. */ + errval = MPI_Recv(NULL, 0, MPI_INT, src, 100, comm, MPI_STATUS_IGNORE); + if (errval) { + errs++; + MTestPrintError( errval ); + printf( "Error returned from Recv\n" ); + } + if (errval) { errs++; MTestPrintError( errval ); @@ -89,14 +98,11 @@ int main( int argc, char *argv[] ) } else if (rank == src) { - /* Send messages, then barrier so that the wait does not start - until we are sure that the sends have begun */ + /* Wait for Irecvs to be posted before the sender calls send */ + MPI_Ssend( NULL, 0, MPI_INT, dest, 100, comm ); + MPI_Send( b1, 10, MPI_INT, dest, 0, comm ); MPI_Send( b2, 11, MPI_INT, dest, 10, comm ); - MPI_Barrier(comm); - } - else { - MPI_Barrier(comm); } MTest_Finalize( errs ); diff --git a/test/mpi/errors/pt2pt/errinstatws.c b/test/mpi/errors/pt2pt/errinstatws.c index 21f0243..002d781 100644 --- a/test/mpi/errors/pt2pt/errinstatws.c +++ b/test/mpi/errors/pt2pt/errinstatws.c @@ -49,7 +49,19 @@ int main( int argc, char *argv[] ) printf( "Error returned from Irecv\n" ); } - /* synchronize */ + /* Wait for Irecvs to be posted before the sender calls send. This + * prevents the operation from completing and returning an error in the + * Irecv. */ + errval = MPI_Recv(NULL, 0, MPI_INT, src, 100, comm, MPI_STATUS_IGNORE); + if (errval) { + errs++; + MTestPrintError( errval ); + printf( "Error returned from Recv\n" ); + } + + /* Wait for sends to complete at the sender before proceeding */ + /* WARNING: This does not guarantee that the sends are ready to + * complete at the receiver. */ errval = MPI_Recv(NULL, 0, MPI_INT, src, 10, comm, MPI_STATUS_IGNORE); if (errval) { errs++; @@ -59,6 +71,9 @@ int main( int argc, char *argv[] ) for (i=0; i<2; i++) { s[i].MPI_ERROR = -1; } + + /* WARNING: The following assumes that Waitsome will complete both + * send/irecv pairs. This is *not* guaranteed by the MPI standard. */ errval = MPI_Waitsome( 2, r, &outcount, indices, s ); MPI_Error_class( errval, &errclass ); if (errclass != MPI_ERR_IN_STATUS) { @@ -88,12 +103,15 @@ int main( int argc, char *argv[] ) } else if (rank == src) { + /* Wait for Irecvs to be posted before the sender calls send */ + MPI_Ssend( NULL, 0, MPI_INT, dest, 100, comm ); + /* Send test messages, then send another message so that the test does not start until we are sure that the sends have begun */ MPI_Send( b1, 10, MPI_INT, dest, 0, comm ); MPI_Send( b2, 11, MPI_INT, dest, 10, comm ); - /* synchronize */ + /* Wait for sends to complete before proceeding to the testsome. */ MPI_Ssend( NULL, 0, MPI_INT, dest, 10, comm ); } ----------------------------------------------------------------------- Summary of changes: test/mpi/basic/wtime.c | 4 +- test/mpi/coll/coll6.c | 14 ++++++++---- test/mpi/errors/faults/pt2ptf2.c | 3 +- test/mpi/errors/pt2pt/errinstatta.c | 22 ++++++++++++++++++- test/mpi/errors/pt2pt/errinstatts.c | 22 ++++++++++++++++++- test/mpi/errors/pt2pt/errinstatwa.c | 20 +++++++++++------ test/mpi/errors/pt2pt/errinstatws.c | 22 ++++++++++++++++++- test/mpi/group/gtranksperf.c | 7 +---- .../mpi_t/nem_fbox_fallback_to_queue_count.c | 9 +------ test/mpi/manual/spawntest_master.c | 8 +++--- test/mpi/manual/testconnectserial.c | 3 +- test/mpi/threads/comm/comm_dup_deadlock.c | 9 +------ test/mpi/threads/comm/ctxdup.c | 7 +----- 13 files changed, 98 insertions(+), 52 deletions(-) hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-78-g5ac51ed
by noreply＠mpich.org 15 Mar '13

15 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via 5ac51edf49d664d33b4a88fc6a5cebeb51950149 (commit) via 9f3b12348c4f94bc059fcc6350a660dabb020c4e (commit) from e04dd4b64ff618f2df58789265b741a8e9fab081 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/5ac51edf49d664d33b4a88fc6a5cebeb5… commit 5ac51edf49d664d33b4a88fc6a5cebeb51950149 Author: James Dinan <dinan(a)mcs.anl.gov> Date: Fri Mar 1 09:12:19 2013 -0600 Add additional wait/test cases to Req. ops. test This adds additional cases to the request-generation RMA operations test case that ensures the implementation correctly handles waiting on the request inside of the same passive target epoch, a different passive target epoch, a different fence epoch, and no epoch. Reviewer: goodell diff --git a/test/mpi/rma/reqops.c b/test/mpi/rma/reqops.c index 7622dd1..ef2636f 100644 --- a/test/mpi/rma/reqops.c +++ b/test/mpi/rma/reqops.c @@ -172,6 +172,106 @@ int main( int argc, char *argv[] ) } MPI_Win_unlock(0, window); + MPI_Barrier(MPI_COMM_WORLD); + + /* Wait inside of an epoch */ + { + MPI_Request pn_req[4]; + int val[4], res; + const int target = 0; + + MPI_Win_lock_all(0, window); + + MPI_Rget_accumulate(&val[0], 1, MPI_INT, &res, 1, MPI_INT, target, 0, 1, MPI_INT, MPI_REPLACE, window, &pn_req[0]); + MPI_Rget(&val[1], 1, MPI_INT, target, 1, 1, MPI_INT, window, &pn_req[1]); + MPI_Rput(&val[2], 1, MPI_INT, target, 2, 1, MPI_INT, window, &pn_req[2]); + MPI_Raccumulate(&val[3], 1, MPI_INT, target, 3, 1, MPI_INT, MPI_REPLACE, window, &pn_req[3]); + + assert(pn_req[0] != MPI_REQUEST_NULL); + assert(pn_req[1] != MPI_REQUEST_NULL); + assert(pn_req[2] != MPI_REQUEST_NULL); + assert(pn_req[3] != MPI_REQUEST_NULL); + + MPI_Waitall(4, pn_req, MPI_STATUSES_IGNORE); + + MPI_Win_unlock_all(window); + } + + MPI_Barrier(MPI_COMM_WORLD); + + /* Wait outside of an epoch */ + { + MPI_Request pn_req[4]; + int val[4], res; + const int target = 0; + + MPI_Win_lock_all(0, window); + + MPI_Rget_accumulate(&val[0], 1, MPI_INT, &res, 1, MPI_INT, target, 0, 1, MPI_INT, MPI_REPLACE, window, &pn_req[0]); + MPI_Rget(&val[1], 1, MPI_INT, target, 1, 1, MPI_INT, window, &pn_req[1]); + MPI_Rput(&val[2], 1, MPI_INT, target, 2, 1, MPI_INT, window, &pn_req[2]); + MPI_Raccumulate(&val[3], 1, MPI_INT, target, 3, 1, MPI_INT, MPI_REPLACE, window, &pn_req[3]); + + assert(pn_req[0] != MPI_REQUEST_NULL); + assert(pn_req[1] != MPI_REQUEST_NULL); + assert(pn_req[2] != MPI_REQUEST_NULL); + assert(pn_req[3] != MPI_REQUEST_NULL); + + MPI_Win_unlock_all(window); + + MPI_Waitall(4, pn_req, MPI_STATUSES_IGNORE); + } + + /* Wait in a different epoch */ + { + MPI_Request pn_req[4]; + int val[4], res; + const int target = 0; + + MPI_Win_lock_all(0, window); + + MPI_Rget_accumulate(&val[0], 1, MPI_INT, &res, 1, MPI_INT, target, 0, 1, MPI_INT, MPI_REPLACE, window, &pn_req[0]); + MPI_Rget(&val[1], 1, MPI_INT, target, 1, 1, MPI_INT, window, &pn_req[1]); + MPI_Rput(&val[2], 1, MPI_INT, target, 2, 1, MPI_INT, window, &pn_req[2]); + MPI_Raccumulate(&val[3], 1, MPI_INT, target, 3, 1, MPI_INT, MPI_REPLACE, window, &pn_req[3]); + + assert(pn_req[0] != MPI_REQUEST_NULL); + assert(pn_req[1] != MPI_REQUEST_NULL); + assert(pn_req[2] != MPI_REQUEST_NULL); + assert(pn_req[3] != MPI_REQUEST_NULL); + + MPI_Win_unlock_all(window); + + MPI_Win_lock_all(0, window); + MPI_Waitall(4, pn_req, MPI_STATUSES_IGNORE); + MPI_Win_unlock_all(window); + } + + /* Wait in a fence epoch */ + { + MPI_Request pn_req[4]; + int val[4], res; + const int target = 0; + + MPI_Win_lock_all(0, window); + + MPI_Rget_accumulate(&val[0], 1, MPI_INT, &res, 1, MPI_INT, target, 0, 1, MPI_INT, MPI_REPLACE, window, &pn_req[0]); + MPI_Rget(&val[1], 1, MPI_INT, target, 1, 1, MPI_INT, window, &pn_req[1]); + MPI_Rput(&val[2], 1, MPI_INT, target, 2, 1, MPI_INT, window, &pn_req[2]); + MPI_Raccumulate(&val[3], 1, MPI_INT, target, 3, 1, MPI_INT, MPI_REPLACE, window, &pn_req[3]); + + assert(pn_req[0] != MPI_REQUEST_NULL); + assert(pn_req[1] != MPI_REQUEST_NULL); + assert(pn_req[2] != MPI_REQUEST_NULL); + assert(pn_req[3] != MPI_REQUEST_NULL); + + MPI_Win_unlock_all(window); + + MPI_Win_fence(0, window); + MPI_Waitall(4, pn_req, MPI_STATUSES_IGNORE); + MPI_Win_fence(0, window); + } + MPI_Win_free(&window); if (buf) MPI_Free_mem(buf); http://git.mpich.org/mpich.git/commitdiff/9f3b12348c4f94bc059fcc6350a660dab… commit 9f3b12348c4f94bc059fcc6350a660dabb020c4e Author: James Dinan <dinan(a)mcs.anl.gov> Date: Fri Mar 15 14:14:44 2013 -0500 Fix req. op. completion outside of PT epoch This is a temporary fix for request-generating operations to allow their requests to be completed after the user has called unlock on the given target. This closes ticket #1801. Ticket #1741 is still active and is keeping track of the fact that an implementation of req. ops that tracks and completes individual operations (rather than the current approach, which just calls flush) is still needed. Reviewer: goodell diff --git a/src/mpid/ch3/src/ch3u_rma_reqops.c b/src/mpid/ch3/src/ch3u_rma_reqops.c index 09cf9d9..acfe1cb 100644 --- a/src/mpid/ch3/src/ch3u_rma_reqops.c +++ b/src/mpid/ch3/src/ch3u_rma_reqops.c @@ -30,10 +30,16 @@ static int MPIDI_CH3I_Rma_req_poll(void *state, MPI_Status *status) MPIU_UNREFERENCED_ARG(status); - /* Call flush to complete the operation */ + /* Call flush to complete the operation. Check that a passive target epoch + * is still active first; the user could complete the request after calling + * unlock. */ /* FIXME: We need per-operation completion to make this more efficient. */ - mpi_errno = req_state->win_ptr->RMAFns.Win_flush(req_state->target_rank, - req_state->win_ptr); + if (req_state->win_ptr->targets[req_state->target_rank].remote_lock_state + != MPIDI_CH3_WIN_LOCK_NONE) + { + mpi_errno = req_state->win_ptr->RMAFns.Win_flush(req_state->target_rank, + req_state->win_ptr); + } if (mpi_errno != MPI_SUCCESS) { MPIU_ERR_POP(mpi_errno); } ----------------------------------------------------------------------- Summary of changes: src/mpid/ch3/src/ch3u_rma_reqops.c | 12 +++- test/mpi/rma/reqops.c | 100 ++++++++++++++++++++++++++++++++++++ 2 files changed, 109 insertions(+), 3 deletions(-) hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-76-ge04dd4b
by noreply＠mpich.org 15 Mar '13

15 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via e04dd4b64ff618f2df58789265b741a8e9fab081 (commit) from 4b6df5cd56f9044a03f8ed421b7e8e293ffac286 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/e04dd4b64ff618f2df58789265b741a8e… commit e04dd4b64ff618f2df58789265b741a8e9fab081 Author: Pavan Balaji <balaji(a)mcs.anl.gov> Date: Tue Mar 12 11:02:07 2013 -0500 Bug-fix: correct MPIR_proctable setup when we use wraparound ranks. As reported by John DelSignore @ Roguewave, when we wraparound ranks among nodes, we were not allocating the MPIR_proctable structure in rank order. We were always grouping them per node. Reviewed by dinan. diff --git a/src/pm/hydra/tools/debugger/debugger.c b/src/pm/hydra/tools/debugger/debugger.c index 3637343..b5449dd 100644 --- a/src/pm/hydra/tools/debugger/debugger.c +++ b/src/pm/hydra/tools/debugger/debugger.c @@ -28,7 +28,7 @@ HYD_status HYDT_dbg_setup_procdesc(struct HYD_pg * pg) { struct HYD_proxy *proxy; struct HYD_exec *exec; - int i, j, np; + int i, j, k, np, round; HYD_status status = HYD_SUCCESS; HYDU_FUNC_ENTER(); @@ -36,10 +36,27 @@ HYD_status HYDT_dbg_setup_procdesc(struct HYD_pg * pg) HYDU_MALLOC(MPIR_proctable, struct MPIR_PROCDESC *, pg->pg_process_count * sizeof(struct MPIR_PROCDESC), status); - for (proxy = pg->proxy_list, i = 0; proxy; proxy = proxy->next) { + round = 0; + /* We need to allocate the MPIR_proctable in COMM_WORLD rank + * order. We do this in multiple rounds. In each round, we + * allocate the proctable entries for the executables on the proxy + * that form a contiguous rank list. Then we move to the next + * proxy. When we run out of proxies, we go back to the first + * proxy and find the next set of contiguous ranks on that + * proxy. */ + for (proxy = pg->proxy_list, i = 0;; proxy = proxy->next) { j = 0; + k = 0; + if (!proxy) { + proxy = pg->proxy_list; + round++; + } for (exec = proxy->exec_list; exec; exec = exec->next) { for (np = 0; np < exec->proc_count; np++) { + if (k + np >= ((round + 1) * proxy->node->core_count)) + break; + if (k + np < round * proxy->node->core_count) + continue; MPIR_proctable[i].host_name = HYDU_strdup(proxy->node->hostname); MPIR_proctable[i].pid = proxy->pid[j++]; if (exec->exec[0]) @@ -48,7 +65,11 @@ HYD_status HYDT_dbg_setup_procdesc(struct HYD_pg * pg) MPIR_proctable[i].executable_name = NULL; i++; } + k += exec->proc_count; } + + if (i >= pg->pg_process_count) + break; } MPIR_proctable_size = pg->pg_process_count; ----------------------------------------------------------------------- Summary of changes: src/pm/hydra/tools/debugger/debugger.c | 25 +++++++++++++++++++++++-- 1 files changed, 23 insertions(+), 2 deletions(-) hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-75-g4b6df5c
by noreply＠mpich.org 12 Mar '13

12 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via 4b6df5cd56f9044a03f8ed421b7e8e293ffac286 (commit) from 0f6085639368dc008cf18ca4c85cdc00b880d02b (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/4b6df5cd56f9044a03f8ed421b7e8e293… commit 4b6df5cd56f9044a03f8ed421b7e8e293ffac286 Author: Dave Goodell <goodell(a)mcs.anl.gov> Date: Tue Mar 12 19:28:55 2013 -0500 `--enable-g=all` now means _all_, add `=most` Before this commit `--enable-g` meant `--enable-g=all`, which actually turned on all debugging options *except* `memarena` and `handle`. As of this commit: * `most` is a new valid debug option having the old effect of `all,handle` * `all` really means everything now, including `memarena` and `handle` * `--enable-g` with no `=BLAH` means `most` Reviewed by balaji@. diff --git a/configure.ac b/configure.ac index e3a2b43..d135372 100644 --- a/configure.ac +++ b/configure.ac @@ -352,8 +352,8 @@ AC_ARG_ENABLE(timing, AC_ARG_ENABLE(g, [ --enable-g=option - Control the level of debugging support in the - MPICH implementation. option is a list of comma - separated names including + MPICH implementation. "option" is a list of comma + separated names including. Default is "most". none - No debugging handle - Trace handle operations handlealloc - Trace handle allocations @@ -369,6 +369,9 @@ AC_ARG_ENABLE(g, memarena - Check for overwrite errors in memory allocation arena mutex - Enable error checking on pthread mutexes mutexnesting - Check for non-nesting of mutexes + most - Most of the above options, excluding some with severe + performance impacts. Recommended for typical development. + yes - synonym for "most" (*not* "all") all - All of the above choices ],,enable_g=none) @@ -1372,7 +1375,7 @@ for option in $enable_g ; do perform_handlealloc=yes ;; handle) - AC_DEFINE(MPICH_DEBUG_HANDLES,1,[Define to enable handle checking]) + perform_handle=yes ;; instr) perform_instr=yes @@ -1396,7 +1399,19 @@ for option in $enable_g ; do mutexnesting) perform_mutexnesting=yes ;; - all|yes) + most|yes) + perform_memtracing=yes + perform_dbglog=yes + enable_append_g=yes + perform_meminit=yes + perform_instr=yes + perform_dbgmutex=yes + perform_mutexnesting=yes + perform_handlealloc=yes + perform_handle=yes + ;; + all) + perform_memarena=yes perform_memtracing=yes perform_dbglog=yes enable_append_g=yes @@ -1405,6 +1420,7 @@ for option in $enable_g ; do perform_dbgmutex=yes perform_mutexnesting=yes perform_handlealloc=yes + perform_handle=yes ;; *) IFS=$save_IFS @@ -1437,6 +1453,8 @@ fi if test "$perform_instr" = yes ; then AC_DEFINE(USE_MPIU_INSTR,1,[Define this to enable internal instrumentation] ) fi +AS_IF([test "X$perform_handle" = "Xyes"], + [AC_DEFINE(MPICH_DEBUG_HANDLES,1,[Define to enable handle checking])]) if test -n "$perform_memtracing" ; then enable_g_mem=yes ----------------------------------------------------------------------- Summary of changes: configure.ac | 26 ++++++++++++++++++++++---- 1 files changed, 22 insertions(+), 4 deletions(-) hooks/post-receive -- MPICH primary repository

1 0

[mpich] MPICH primary repository branch, master, updated. v3.0.2-74-g0f60856
by noreply＠mpich.org 07 Mar '13

07 Mar '13

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MPICH primary repository". The branch, master has been updated via 0f6085639368dc008cf18ca4c85cdc00b880d02b (commit) from 115fc21aff37499f35ebde653833103c0f5a963a (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- http://git.mpich.org/mpich.git/commitdiff/0f6085639368dc008cf18ca4c85cdc00b… commit 0f6085639368dc008cf18ca4c85cdc00b880d02b Author: Dave Goodell <goodell(a)mcs.anl.gov> Date: Thu Mar 7 14:29:15 2013 -0600 fix `complex<long double>` bug from [3f1fcead] If C++ is disabled or the C++ compiler does not support `complex<long double>`, then bad Fortran code would be emitted into `mpif.h`. No reviewer. diff --git a/configure.ac b/configure.ac index 4a80c12..e3a2b43 100644 --- a/configure.ac +++ b/configure.ac @@ -4185,10 +4185,10 @@ MPIR_CXX_BOOL=0x0c000000 MPIR_CXX_COMPLEX=0x0c000000 MPIR_CXX_DOUBLE_COMPLEX=0x0c000000 MPIR_CXX_LONG_DOUBLE_COMPLEX=0x0c000000 -MPI_F77_CXX_BOOL=0x0c000000 -MPI_F77_CXX_FLOAT_COMPLEX=0x0c000000 -MPI_F77_CXX_DOUBLE_COMPLEX=0x0c000000 -MPI_F77_CXX_LONG_DOUBLE_COMPLEX=0x0c000000 +MPI_F77_CXX_BOOL=MPI_DATATYPE_NULL +MPI_F77_CXX_FLOAT_COMPLEX=MPI_DATATYPE_NULL +MPI_F77_CXX_DOUBLE_COMPLEX=MPI_DATATYPE_NULL +MPI_F77_CXX_LONG_DOUBLE_COMPLEX=MPI_DATATYPE_NULL if test "$enable_cxx" = "yes" ; then AC_LANG_CPLUSPLUS AC_CHECK_SIZEOF(bool) ----------------------------------------------------------------------- Summary of changes: configure.ac | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-) hooks/post-receive -- MPICH primary repository

1 0