[rds-devel] [git pull] rds-stress -I fix
Zach Brown
zach.brown at oracle.com
Wed Aug 4 11:22:10 PDT 2010
Hey Andy,
I think I finally got to the bottom of the errors we were seeing with -I in rds-tools.
The great news is that it doesn't look like the kernel did anything wrong. It looks to my eye that the mapping call in rds-stress wasn't taking the -I option into account. It would map a single -D sized buffer. Then the remote peer would try and issue rdma ops against that buffer with multiple -D sized buffers. The ops would generate an access violation and tear down the QP.
The patch certainly stops the problem over here. I might have misunderstood the rds-stress code, though. Can you eyeball it before pulling?
- z
The following changes since commit c6992212e442a301d40d9a0265db12f65eab1b81:
Andy Grover (1):
update rpm/tgz build process
are available in the git repository at:
git://oss.oracle.com/git/zab/rds-tools.git vectored-map-fix
Zach Brown (1):
account for vectored rdma size when mapping
rds-stress.c | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
More information about the rds-devel
mailing list