On Thu, Apr 14, 2005 at 11:39:17AM -0700, Roland Dreier wrote: > I think I have this figured out: if you unload ib_ipoib and > ib_sa_query in quick succession, ib_ipoib sends MCMember requests to > the SA to leave its multicast groups. Normally, because IPoIB sets a > timeout of 0, no callback is generated and so it's fine that IPoIB > passes a NULL callback. However, if ib_sa_query is unloaded right > afterwards, the send of the request doesn't get a chance to complete > and so a cancel callback is generated. > > If this crash is at all reproducible for you, can you try this patch > and see if it helps?
I haven't reproduced it yet...but I'm going to put a machine in an infinite loop running the unload/load script. Once I know how long it takes to reproduce, I can comfortably tell you if it's fixed or not. thanks, grant _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
