[openib-general] Re: executing the ibv_srq_pingpong with many QPs sometimes causes to a test failure

[Roland Dreier]

Thanks, I think I tracked down the seg fault and pushed a fix out to
the subversion repository.

I still see some problems with the SRQ pingpong test with many QPs.
However I think the problems are in the pingpong test and not the
verbs libraries -- the pingpong tests all make bogus assumptions about
the ordering of send vs receive completions, so I'm sure there are
race conditions.

 - R.