On Mon, Mar 12, 2018 at 06:08:04PM +0200, Leon Romanovsky wrote: > On Mon, Mar 12, 2018 at 09:39:14AM -0600, Jason Gunthorpe wrote: > > On Sun, Mar 11, 2018 at 01:28:23PM +0200, Leon Romanovsky wrote: > > > From: Shalom Lagziel <shaloml@xxxxxxxxxxxx> > > > > > > In IPoIB connected mode the WQE opcode can be IB_WR_SEND only (stateless > > > offloads are supported only in datagram mode), but it wasn't initialized. > > > It leads to unpredictable CQE errors which were triggered by > > > ipoib_cm_send() flow. > > > > This description is confusing. I think it is trying to say 'when > > switching from UD to RC mode the wr.opcode can be left at IB_WR_LSO > > ..' > > > > Sounds like -rc and -stable material, no? > > I'm not so sure, I tried hard and failed to find fixes line, so decided > to skip that -rc/-stable part. The race has always existed I think. Lets see what the answers to the other questions in my mail are. Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
