Commits · e072e5ca3d2cca0a9028958b8f096bd436681d19 · Rob Latham / MPICH-BlueGene

26 Oct, 2013 19 commits
- Rename parameters MPIR_PARAM_* to MPIR_CVAR_* · e072e5ca
  Junchao Zhang authored Oct 25, 2013 and Pavan Balaji committed Oct 26, 2013
```
To adapt to naming for control variables in MPI_T.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  e072e5ca
- Add a test case to test multithreading MPI_T · fb570792
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
The case is adapted from test/mpi/mpi_t/mpit_vars.c
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  fb570792
- Delete xfail=ticket1898 of mpi_t_str · de70f872
  Junchao Zhang authored Oct 17, 2013 and Pavan Balaji committed Oct 26, 2013
```
This bug was fixed with the new MPI_T implementation.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  de70f872
- Revise two tests under test/mpi/manual/mpi_t/ · 9fcd5712
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  9fcd5712
- Revise the test case mpit_vars.c · d5ca1acd
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Now we can control printing using env var MPITEST_VERBOSE.
The old code used a local variable 'verbose'. It shades the global
variable 'verbose'. So we renamed the local one to 'verb' instead.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  d5ca1acd
- Rename mpit_init_thread.c to mpit_initthread.c · 83f6c94a
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Since initthread.c is used for MPI_Init_thread(), we'd better use
mpit_initthread.c for MPI_T_init_thread().
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  83f6c94a
- Rename cat_get_pv.c to cat_get_pvars.c · 601ac99d
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
The old name doesn't confrom to naming convention of other files under mpi_t
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  601ac99d
- Rename mpich_param_vals.h, param_vals.c · fc51e3ad
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
These two files are renamed to mpich_params.{c, h} since the old
ones are not uniform.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  fc51e3ad
- Revise genparams to gen files needed by new MPI_T · 255d0fb7
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  255d0fb7
- Add missing info of parameters (cvars) · 127ee1d2
  Junchao Zhang authored Oct 24, 2013 and Pavan Balaji committed Oct 26, 2013
```
This makes parameters (cvars) conform to MPI_T standard.
However, probably the info is too conservative.
Need further tuning the info.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  127ee1d2
- Capitalize parameter category names · 4c5f24c8
  Junchao Zhang authored Oct 24, 2013 and Pavan Balaji committed Oct 26, 2013
```
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  4c5f24c8
- Adapt old ch3 pvars to new MPI_T impl · e39e4d2e
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  e39e4d2e
- Adapt old nemesis pvars to new MPI_T impl · 068ac9cd
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  068ac9cd
- Change wording for option --enable-mpit-pvars · b351a97e
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  b351a97e
- Change var names in enable_pvar_* style · 15357039
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  15357039
- Delete the old interfaces on pvars · bfca1382
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
New code should use interfaces provided by new MPI_T impl.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  bfca1382
- In MPI init/finalize, use updated MPI_T routines · 21d84da9
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  21d84da9
- Replace mpi_t_util.c with mpit.c · 72bf15fb
  Junchao Zhang authored Oct 18, 2013 and Pavan Balaji committed Oct 26, 2013
```
mpit.c also includes non-util routines
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  72bf15fb
- Primary implementation of MPI_T specified in MPI3 · 39e5fa15
  Junchao Zhang authored Sep 03, 2013 and Pavan Balaji committed Oct 26, 2013
```
Implemented the framework for MPI_T enums, categories, performance variables.
and control variables specified in MPI3. Fix #1898, #1930.

The code should make MPICH MPI_T compliant, though not all MPICH pvars/cvars
are currently exposed through MPI_T.

The code supports multithreading up to MPI_THREAD_MULTIPLE.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>
```
  39e5fa15
24 Oct, 2013 3 commits

Rob Latham authored Oct 24, 2013

Jeff Hammond's test to send/recieve a message described by a large datatype.

7ff6a970

Fix for uncommitted datatypes · 10073a33

Lisandro Dalcin authored Oct 23, 2013 and

Kenneth Raffenetti committed Oct 24, 2013



MPI_Type_get_extent_x and MPI_Type_get_true_extent_x were throwing
errors when passed an uncommitted user-defined datatype. This was
incorrect. Closes #1774
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

10073a33

Allow MPI_ERR_IO as error class in test · a5d05b36

William Gropp authored Oct 24, 2013 and

Rob Latham committed Oct 24, 2013



The test was too prescriptive about the allowed error classes.
ERR_NO_SUCH_FILE is a better choice, but ERR_IO is valid.
Signed-off-by: Rob Latham <robl@mcs.anl.gov>

a5d05b36

21 Oct, 2013 2 commits

Add way to test implementation specific features · 9f7f1125

William Gropp authored Oct 17, 2013



Adds to the test suite a way to include tests that are specific to
a particular implementation.  Currently includes only one for MPICH,
testing the basic features of mpi_t.  This also serves as a test of the
implementation tests themselves.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>

9f7f1125

Ensure MPIR_COUNT_ALIGNMENT set · c580679c

William Gropp authored Oct 20, 2013

In some cases, MPIR_COUNT_ALIGNMENT was not being defined.  This
fixes that case, which came up when trying the -feature=fint=on option
for the Fortran 77 code.

c580679c

20 Oct, 2013 1 commit

Multi-threaded RMA test. · 011cd3df

Xin Zhao authored Oct 20, 2013 and

Pavan Balaji committed Oct 20, 2013



Multiple threads do RMA operations and flush on the same target.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>

011cd3df

19 Oct, 2013 2 commits

Add a test for allocating 1GB SHM window memory per process. · 269fc330

Xin Zhao authored Oct 18, 2013 and

Pavan Balaji committed Oct 19, 2013



Test MPI_WIN_ALLOCATE and MPI_WIN_ALLOCATE_SHARED when allocating
large SHM window memory (1GB) per process.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>

269fc330

make MMAP SHM handle as MPI_Aint type. · 2b572f0c

Xin Zhao authored Oct 18, 2013 and

Pavan Balaji committed Oct 19, 2013

make SHM handle type in MMAP SHM allocation defined as MPI_Aint instead of
integer to accomodate allocation of SHM memory with size larger than 2^32.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>

2b572f0c

18 Oct, 2013 4 commits

Update libmpich ABI string for the 3.1b2 release. · beac0a93

Pavan Balaji authored Oct 03, 2013



The ABI is broken from the changes contributed by Intel.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

beac0a93

Fortran ABI changes from Intel. · 3f1080e7

Pavan Balaji authored Sep 22, 2013



This is a part of the contribution from Intel to update mpich's ABI to
match that of Intel MPI's previous versions.
Signed-off-by: William Gropp <wgropp@illinois.edu>

3f1080e7

Renumber MPI_COMBINER constants to maintain ABI. · a4041b98

Pavan Balaji authored Sep 22, 2013



The numbering of MPI_COMBINER constants was changed when we added
MPI-2.2 support.  This patch reverts that numbering and adds the
MPI-2.2 specific combiners to the end, to maintain ABI compatibility
with older mpich releases.

An initial version of this patch was contributed by Intel.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

a4041b98

Change the MPI_MAX_ERROR_STRING back to 512. · 8927c1b1

Pavan Balaji authored Sep 21, 2013



This was changed to 1024 to give space for larger error strings.  But
this caused an ABI breakage for a number of MPICH partners.  This
patch reverts the size back the size to what we had in older mpich
releases.

An initial version of this patch was contributed by Intel.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

8927c1b1

17 Oct, 2013 1 commit
- update discuss email address · 399b6b86
  Jeff Hammond authored Oct 16, 2013
```
Signed-off-by: Wesley Bland <wbland@mcs.anl.gov>
```
  399b6b86
15 Oct, 2013 1 commit
- fix check for duplicate KVS entries · f1dcc582
  Jan Bierbaum authored Oct 14, 2013 and Kenneth Raffenetti committed Oct 15, 2013
```
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>
```
  f1dcc582
13 Oct, 2013 1 commit

Disable xfail test from running by default · da5d5541

Kenneth Raffenetti authored Oct 09, 2013



xfails are primarily for our consumption, so we disable them by
default. This will result in cleaner test output for the average
user.

Also note that if the testsuite is configured in strict MPI mode,
all strict tests, regardless of xfail, will run.
Signed-off-by: Pavan Balaji <balaji@mcs.anl.gov>

da5d5541

12 Oct, 2013 1 commit

teach type-dumping code about subarray · 307e5fc6

Rob Latham authored Oct 12, 2013

MPIDU_Datatype_debug is really helpful when trying to figure out what a
library or someone else's datatype code is feeding to MPICH. For some
reason it didn't know how to dump subarray types. now it does.

307e5fc6

11 Oct, 2013 2 commits

Warning squash. · 0706e96c
Pavan Balaji authored Oct 09, 2013
```
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>
```
0706e96c

clang compilation fixes. · d18633ac

Pavan Balaji authored Oct 09, 2013



clang seems to be unhappy when we use "0" instead of "NULL" for buffer
addresses.
Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>

d18633ac

09 Oct, 2013 2 commits
- clang finds an incorrect header guard · a1b3657e
  Rob Latham authored Oct 09, 2013
```
bglockless feature file should not define anything pvfs2 related.
```
  a1b3657e
- removed unused variable · 7d94f24c
  Rob Latham authored Oct 09, 2013
```
this guy no longer gets used now that there's common hint processing
code
```
  7d94f24c
04 Oct, 2013 1 commit
- compile error · 4a13e986
  Rob Latham authored Oct 04, 2013
  
  4a13e986