[mpich-discuss] crash in MPICH2 1.3.1

[Dave Goodell]

On Jan 6, 2011, at 4:01 PM CST, Blair, David wrote:

> Sounds like taking the lock is the responsibility of the caller of
> the progress engine.

Mostly.  That's good enough for what we're talking about here, but the real rules for locking in MPICH2 are far more complicated than that, unfortunately.

> So the right fix is to add
> MPIU_THREAD_CS_ENTER/EXIT(ALLFUNC,) in MPI_Request_get_status?

Yes, as in the (untested) patch that I've attached to this mail.  It's based on the current trunk at 7673, but it will probably apply cleanly to the past year or two of releases.

Let us know if it works for you and I'll commit it.  Thanks for telling us about the bug.

-Dave

-------------- next part --------------
A non-text attachment was scrubbed...
Name: req_get_status.patch
Type: application/octet-stream
Size: 649 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20110106/47e7d5cd/attachment.obj>