160 lines
5.6 KiB
Plaintext
160 lines
5.6 KiB
Plaintext
2016-09-21 Aravind Gopalakrishnan <Aravind.Gopalakrishnan@intel.com>
|
|
|
|
Sync with wfr-psm 10.2 release codebase
|
|
This code surface corresponds to the 10.2 release
|
|
of PSM2. Changes in this patch include few bew
|
|
fixes and patches to improve performance for certain
|
|
workloads bt interleaving the RTS handling.
|
|
|
|
While at it, removing some unnecessary files whose
|
|
functionality have been moved.
|
|
|
|
2016-07-12 Russell McGuire (rwmcguir) <russell.w.mcguire@intel.com>
|
|
|
|
Update psm2_ep_close timeout handling
|
|
This change includes many patches for psm2_ep_close timeouts
|
|
* Fixed timeout logic for IPS layer disconnects
|
|
* Replaced usage of rand/srand/random with drand48_r
|
|
* Cleanup of internal connection symbols names
|
|
* Fix for amsh layer dropping disconnect packets
|
|
* Fixed issue with spec file not obsoleting older rpm name
|
|
* Update README for newer RPM names and usages
|
|
* Reduce timeout scaling for dropped packets during psm2_ep_close
|
|
for large scale jobs.
|
|
|
|
2016-06-29 Russell McGuire (rwmcguir) <russell.w.mcguire@intel.com>
|
|
|
|
Remove deprecated hfi1-spec* files
|
|
These files were left over after the rpm
|
|
rename from hfi1-psm to libpsm2.
|
|
|
|
2016-06-30 Aravind Gopalakrishnan <Aravind.Gopalakrishnan@intel.com>
|
|
|
|
Upstreaming /dev/hfi_N selection logic and some bug fixes
|
|
This change primarily includes the following patches to PSM2-
|
|
* Remove usage of /dev/hfi
|
|
* Logic to select correct /dev/hfi_N
|
|
* Initialize frag_size in tidrecv array tidflows which
|
|
fixes assertion hit on debug builds
|
|
* Clear send_msgoff in req struct which fixes the hang
|
|
while using OPCODE_SHORT messages repeatedly
|
|
|
|
2016-06-27 Russell McGuire (rwmcguir) <russell.w.mcguire@intel.com>
|
|
|
|
Update Makefile to use RELEASE_VER for rpm version
|
|
This change is to allow this git repo, to honor a
|
|
RELEASE_VER file for RPM naming convetion. This will be
|
|
used over counting the number of commits since the last
|
|
v<major>.<minor> tag in the repo.
|
|
|
|
Mostly this is added to allow internal and external
|
|
rpm naming syncronization with Intel IFS releases for OPA.
|
|
|
|
2016-06-28 Aravind Gopalakrishnan <Aravind.Gopalakrishnan@intel.com>
|
|
|
|
Sync with wfr-psm 10.1 branch
|
|
This corresponds to the 10.1 release surface
|
|
|
|
2016-06-28 Paul Reger <paul.j.reger@intel.com>
|
|
|
|
Adds many bug fixes, and heap debug code.
|
|
This surface corresponds to the code being upstreamed to Fedora.
|
|
|
|
2016-06-23 pjreger <paul.j.reger@intel.com>
|
|
|
|
Merge pull request #5 from LeviDettwyler/10.2
|
|
Fix for assertion failure seen in PSM_DEBUG builds
|
|
|
|
2016-06-23 Levi Dettwyler <levi.dettwyler@me.com>
|
|
|
|
Fix for assertion failure seen in PSM_DEBUG builds
|
|
Merging fix from internal repos to public github to correct tidflow frag_size assertion issue when running workloads with PSM_DEBUG builds.
|
|
|
|
2016-05-06 Paul Reger <paul.j.reger@intel.com>
|
|
|
|
First commit for v10.2 branch, and first commit for Fedora 25 support.
|
|
|
|
2016-04-06 Paul Reger <paul.j.reger@intel.com>
|
|
|
|
Adds many bug fixes, and heap debug code. This surface corresponds to the code being upstreamed to Fedora.
|
|
|
|
Adding two bug fixes. This surface corresonds to the 10.0.1 release.
|
|
|
|
2016-03-03 Paul Reger <paul.j.reger@intel.com>
|
|
|
|
Limit rpm builds of the psm library to the x86_64 architecture only.
|
|
|
|
2016-03-02 Paul Reger <paul.j.reger@intel.com>
|
|
|
|
Moved two unnecessary macros under PSM_DEBUG to fix build error on Fedora 24.
|
|
|
|
2016-02-19 Paul Reger <paul.j.reger@intel.com>
|
|
|
|
Update gitub to RELEASE-10.0 version includes:
|
|
- Primarily bug fixes
|
|
- One feature was added that includes a new hashing scheme for MQ's
|
|
- Performance enhancements
|
|
|
|
2015-12-07 Paul Reger <paul.j.reger@intel.com>
|
|
|
|
Update gitub to beta0, includes:
|
|
- Symbol renames from psm* to psm2* and PSM* to PSM2*,
|
|
- Compat library support added,
|
|
- Tid caching requires building and running with a driver
|
|
that includes tid caching support.
|
|
|
|
2015-09-23 nbwhite <nathan.b.white@intel.com>
|
|
|
|
Merge pull request #2 from michich/no-exec-stack
|
|
prevent having executable stack
|
|
|
|
2015-09-04 nbwhite <nathan.b.white@intel.com>
|
|
|
|
Merge pull request #1 from michich/master
|
|
fix issues found during packaging for RHEL
|
|
|
|
2015-09-01 Michal Schmidt <mschmidt@redhat.com>
|
|
|
|
prevent having executable stack
|
|
Add .note.GNU-stack sections to objects compiled from assembly.
|
|
This allows libpsm2.so.2.0 to have non-executable stack.
|
|
|
|
2015-08-25 Michal Schmidt <mschmidt@redhat.com>
|
|
|
|
use pkg-config to find udev rules dir
|
|
and default to /lib/udev (for system rules) rather than /etc/udev (which
|
|
is for the local administrator).
|
|
|
|
2015-08-12 Michal Schmidt <mschmidt@redhat.com>
|
|
|
|
avoid calling memset with zero length
|
|
This avoids the following build error:
|
|
|
|
In file included from /usr/include/string.h:638:0,
|
|
from ./include/opa_user.h:77,
|
|
from psm_user.h:67,
|
|
from psm_diags.c:56:
|
|
In function 'memset',
|
|
inlined from 'memcpy_check_one.constprop.4' at psm_diags.c:278:8,
|
|
inlined from 'memcpy_check_size.constprop.2' at psm_diags.c:335:30,
|
|
inlined from 'psmi_test_memcpy.constprop.1' at psm_diags.c:230:6,
|
|
inlined from 'psmi_diags' at psm_diags.c:84:25:
|
|
/usr/include/bits/string3.h:81:30: error: call to
|
|
'__warn_memset_zero_len' declared with attribute warning: memset used
|
|
with constant zero length parameter; this could be due to transposed
|
|
parameters [-Werror]
|
|
__warn_memset_zero_len ();
|
|
^
|
|
|
|
2015-08-12 Michal Schmidt <mschmidt@redhat.com>
|
|
|
|
avoid executable flag on installed header files
|
|
|
|
2015-07-23 Paul Reger <paul.j.reger@intel.com>
|
|
|
|
Sync github to internal git repo.
|
|
|
|
2015-07-15 nbwhite <nathan.b.white@intel.com>
|
|
|
|
Initial commit
|