pml-cm bug fixes #637

jithinjosepkl · 2015-06-11T00:38:24Z

Set convertor pDesc and count in OPAL_CONVERTOR_PREPARE (including cases where count = 0)
Do opal_convertor_copy_and_prepare_for_send for buffered send mode as MCA_PML_CM_HVY_SEND_REQUEST_BSEND_ALLOC calls opal_convertor_pack directly

@bosilca - Please review.

where count = 0) Signed-off-by: Jithin Jose <jithin.jose@intel.com>

mellanox-github · 2015-06-11T00:51:51Z

Refer to this link for build results (access rights to CI server needed):
http://bgate.mellanox.com/job/gh-ompi-master-pr/621/

Build Log
last 50 lines

[...truncated 13714 lines...]
pml_cm.h:227: error: (Each undeclared identifier is reported only once
pml_cm.h:227: error: for each function it appears in.)
In file included from pml_cm_request.c:21:
pml_cm.h: In function 'mca_pml_cm_isend_init':
pml_cm.h:227: error: implicit declaration of function 'MCA_PML_CM_HVY_SEND_REQUEST_INIT'
pml_cm.h:227: error: 'ompi_proc' undeclared (first use in this function)
pml_cm.h:227: error: (Each undeclared identifier is reported only once
pml_cm.h:227: error: for each function it appears in.)
pml_cm.h: In function 'mca_pml_cm_isend':
pml_cm.h:256: error: 'ompi_proc' undeclared (first use in this function)
pml_cm.h: In function 'mca_pml_cm_isend':
pml_cm.h:256: error: 'ompi_proc' undeclared (first use in this function)
In file included from pml_cm_sendreq.c:21:
pml_cm.h: In function 'mca_pml_cm_isend_init':
pml_cm.h:227: error: implicit declaration of function 'MCA_PML_CM_HVY_SEND_REQUEST_INIT'
pml_cm.h:227: error: 'ompi_proc' undeclared (first use in this function)
pml_cm.h:227: error: (Each undeclared identifier is reported only once
pml_cm.h:227: error: for each function it appears in.)
pml_cm.h: In function 'mca_pml_cm_isend':
pml_cm.h:256: error: 'ompi_proc' undeclared (first use in this function)
In file included from pml_cm_start.c:26:
pml_cm.h: In function 'mca_pml_cm_isend_init':
pml_cm.h:227: error: implicit declaration of function 'MCA_PML_CM_HVY_SEND_REQUEST_INIT'
pml_cm.h:227: error: 'ompi_proc' undeclared (first use in this function)
pml_cm.h:227: error: (Each undeclared identifier is reported only once
pml_cm.h:227: error: for each function it appears in.)
pml_cm.h: In function 'mca_pml_cm_isend':
pml_cm.h:256: error: 'ompi_proc' undeclared (first use in this function)
make[2]: *** [pml_cm_cancel.lo] Error 1
make[2]: *** Waiting for unfinished jobs....
make[2]: *** [pml_cm_component.lo] Error 1
make[2]: *** [pml_cm_request.lo] Error 1
make[2]: *** [pml_cm.lo] Error 1
make[2]: *** [pml_cm_start.lo] Error 1
make[2]: *** [pml_cm_recvreq.lo] Error 1
make[2]: *** [pml_cm_sendreq.lo] Error 1
make[2]: Leaving directory `/scrap/jenkins/jenkins/jobs/gh-ompi-master-pr/workspace/ompi/mca/pml/cm'
make[1]: *** [install-recursive] Error 1
make[1]: Leaving directory `/scrap/jenkins/jenkins/jobs/gh-ompi-master-pr/workspace/ompi'
make: *** [install-recursive] Error 1
Build step 'Execute shell' marked build as failure
[htmlpublisher] Archiving HTML reports...
[htmlpublisher] Archiving at BUILD level /var/lib/jenkins/jobs/gh-ompi-master-pr/workspace/cov_build to /var/lib/jenkins/jobs/gh-ompi-master-pr/builds/621/htmlreports/Coverity_Report
Setting commit status on GitHub for https://github.com/open-mpi/ompi/commit/94d15be88feb99a5966f0db9ef8544d8a5f52357
[BFA] Scanning build for known causes...
[BFA] No failure causes found
[BFA] Done. 0s
Setting status of 8876976f83efc76cb79977b336ed0cecae4b5895 to FAILURE with url http://bgate.mellanox.com:8888/jenkins/job/gh-ompi-master-pr/621/ and message: 'Build finished.'
Using conext: Mellanox

lanl-ompi · 2015-06-11T00:53:04Z

Refer to this link for build results (access rights to CI server needed):
http://jenkins.open-mpi.org/job/ompi_master_pr_cle5.2up02/131/

Build Log
last 20 lines

[...truncated 9506 lines...]
pml_cm_sendreq.h:273:46: note: in definition of macro 'MCA_PML_CM_HVY_SEND_REQUEST_INIT'
                                              ompi_proc,                 \
                                              ^
pml_cm.h: In function 'mca_pml_cm_isend':
pml_cm.h:257:42: error: 'ompi_proc' undeclared (first use in this function)
                                          ompi_proc, 
                                          ^
pml_cm_sendreq.h:273:46: note: in definition of macro 'MCA_PML_CM_HVY_SEND_REQUEST_INIT'
                                              ompi_proc,                 \
                                              ^
make[2]: *** [pml_cm_component.lo] Error 1
cc1: some warnings being treated as errors
make[2]: *** [pml_cm_recvreq.lo] Error 1
make[2]: Leaving directory `/global/u2/h/hpp/jenkins_edison/workspace/ompi_master_pr_cle5.2up02/ompi/mca/pml/cm'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/global/u2/h/hpp/jenkins_edison/workspace/ompi_master_pr_cle5.2up02/ompi'
make: *** [all-recursive] Error 1
Build step 'Execute shell' marked build as failure
Setting status of 8876976f83efc76cb79977b336ed0cecae4b5895 to FAILURE with url http://jenkins.open-mpi.org/job/ompi_master_pr_cle5.2up02/131/ and message: Build finished.

Test FAILed.

lanl-ompi · 2015-06-11T01:40:08Z

Refer to this link for build results (access rights to CI server needed):
http://jenkins.open-mpi.org/job/ompi_master_pr_cle5.2up02/132/
Test PASSed.

mellanox-github · 2015-06-11T01:45:56Z

Coverity found 812 errors for all
Coverity found 5 errors for oshmem
Coverity found 2 errors for fca
Coverity report at http://bgate.mellanox.com:8888/jenkins/job/gh-ompi-master-pr//622/Coverity_Report

Sent from GH.

mellanox-github · 2015-06-11T02:00:51Z

Refer to this link for build results (access rights to CI server needed):
http://bgate.mellanox.com/job/gh-ompi-master-pr/622/

hppritcha · 2015-06-11T03:46:48Z

bot:retest

lanl-ompi · 2015-06-11T04:02:05Z

Refer to this link for build results (access rights to CI server needed):
http://jenkins.open-mpi.org/job/ompi_master_pr_distcheck/41/
Test PASSed.

lanl-ompi · 2015-06-11T04:02:15Z

Refer to this link for build results (access rights to CI server needed):
http://jenkins.open-mpi.org/job/ompi_master_pr_cle5.2up02/134/
Test PASSed.

jithinjosepkl · 2015-06-11T23:00:59Z

@bosilca - the build error was due to a last minute macro name change. Looks sanity tests are fine now.
Can you please take a look?

bosilca · 2015-06-11T23:05:04Z

Overall it looks OK. However, I wonder why you need the changes in the convertor, as if the local_size of zero the convertor should not be used in any case.

jithinjosepkl · 2015-06-11T23:12:26Z

@bosilca - Thanks.

In ompi_mtl_datatype_pack, we check for these two fields (datatype and count) while invoking opal_datatype_is_contiguous_memory_layout. Earlier, these fields were not set in OPAL_CONVERTOR_PREPARE macro if (0 == count) || (0 == datatype->size).

hppritcha · 2015-06-15T12:47:48Z

If you want this merged in to master before fork it should be checked in today or latest tomorrow.

jithinjosepkl · 2015-06-15T16:30:16Z

@hppritcha - Thanks for the reminder. Yes I would like to.
@bosilca - Any further comments?

bosilca · 2015-06-15T20:06:03Z

A convertor without anything to do is marked with the special flag COMPLETED. You might want to use that as a marker for the fact that the MTL has nothing to do. Otherwise I have no other comments.

MCA_PML_CM_HVY_SEND_REQUEST_BSEND_ALLOC calls opal_convertor_pack directly. Signed-off-by: Jithin Jose <jithin.jose@intel.com>

jithinjosepkl · 2015-06-16T00:53:59Z

bot:retest

rhc54 · 2015-06-16T01:13:19Z

bot:retest

pml-cm bug fixes

Fix typo that caused compile error

Set convertor pDesc and count in OPAL_CONVERTOR_PREPARE (including cases

5922645

where count = 0) Signed-off-by: Jithin Jose <jithin.jose@intel.com>

jithinjosepkl force-pushed the pr/pml-cm-opt branch from 8876976 to e9c02bd Compare June 11, 2015 01:24

Do opal_convertor_copy_and_prepare_for_send for buffered send mode as

7ccde09

MCA_PML_CM_HVY_SEND_REQUEST_BSEND_ALLOC calls opal_convertor_pack directly. Signed-off-by: Jithin Jose <jithin.jose@intel.com>

jithinjosepkl force-pushed the pr/pml-cm-opt branch from e9c02bd to 7ccde09 Compare June 16, 2015 00:26

rhc54 pushed a commit that referenced this pull request Jun 16, 2015

Merge pull request #637 from jithinjosepkl/pr/pml-cm-opt

9a8bda0

pml-cm bug fixes

rhc54 merged commit 9a8bda0 into open-mpi:master Jun 16, 2015

jsquyres added a commit to jsquyres/ompi that referenced this pull request Nov 10, 2015

Merge pull request open-mpi#637 from rolfv/pr/fix-cuda-compile-error-2.x

b2563fb

Fix typo that caused compile error

pml-cm bug fixes #637

pml-cm bug fixes #637

Uh oh!

Conversation

jithinjosepkl commented Jun 11, 2015

Uh oh!

mellanox-github commented Jun 11, 2015

Uh oh!

lanl-ompi commented Jun 11, 2015

Uh oh!

lanl-ompi commented Jun 11, 2015

Uh oh!

mellanox-github commented Jun 11, 2015

Uh oh!

mellanox-github commented Jun 11, 2015

Uh oh!

hppritcha commented Jun 11, 2015

Uh oh!

lanl-ompi commented Jun 11, 2015

Uh oh!

lanl-ompi commented Jun 11, 2015

Uh oh!

jithinjosepkl commented Jun 11, 2015

Uh oh!

bosilca commented Jun 11, 2015

Uh oh!

jithinjosepkl commented Jun 11, 2015

Uh oh!

hppritcha commented Jun 15, 2015

Uh oh!

jithinjosepkl commented Jun 15, 2015

Uh oh!

bosilca commented Jun 15, 2015

Uh oh!

jithinjosepkl commented Jun 16, 2015

Uh oh!

rhc54 commented Jun 16, 2015

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants