Talpey, Thomas wrote:
At 03:59 PM 5/27/2008, Steve Wise wrote:
Talpey, Thomas wrote:
What happens if the upper layer gives up and invalidates the stag itself,
and the peer's RDMA Read response arrives later? Nothing bad, I assume,
and the peer's response is denied?
It behaves just like any other tagged message arriving and the target
stag is invalid. The connection is torn down via an RDMAP TERMINATE...
I was wondering more about the dangling stag reference that the original work
request carried. Normally, it would reference the still-valid stag, but if that
stag was torn down (causing the invalidation to point to nothing), or worse,
re-bound (causing it to point at something else!), then it's a possible issue?
Sorry to seem paranoid here. Storage is pretty sensitive to silent data
corruption avenues. Because they always find a way to happen.
These WRs allow the user to shoot himself in the foot. You're describe
one such case...
Steve.
|
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
