[openib-general] RE: [PATCH] [ucm] fix for potential deadlock

Arlin Davis arlin.r.davis at intel.com
Wed Aug 10 11:31:16 PDT 2005



>-----Original Message-----
>From: Hefty, Sean
>Sent: Tuesday, August 09, 2005 4:24 PM
>To: openib-general at openib.org; Davis, Arlin R
>Subject: [PATCH] [ucm] fix for potential deadlock
>
>The following patch fixes a potential deadlock condition in the kernel
>ucm code resulting from trying to destroy a cm_id while in the context
>of a CM callback thread.  The synchronization around the ucm context
>structure was simplified as a result, and some simple code cleanup is
>included.  (I tried keeping the code cleanup separate, but it was
>turning out to be more work.)
>
>Arlin, can you please test with this and see if your problems (well the
>ones related to IB anyway ;) go away.

Yes, the issues with dapltest regress.sh (client 6) are resolved with this change.

Thanks!

-arlin

>
>Signed-off-by: Sean Hefty <sean.hefty at intel.com>
>
>





More information about the general mailing list