3.2.100-rc1 review patch. If anyone has any objections, please let me know.
------------------
From: Steve Wise swise@opengridcomputing.com
commit f55688c45442bc863f40ad678c638785b26cdce6 upstream.
If the RECV CQE is in error, ignore the MSN check. This was causing recvs that were flushed into the sw cq to be completed with the wrong status (BAD_MSN instead of FLUSHED).
Signed-off-by: Steve Wise swise@opengridcomputing.com Signed-off-by: Jason Gunthorpe jgg@mellanox.com [bwh: Backported to 3.2: adjust context] Signed-off-by: Ben Hutchings ben@decadent.org.uk --- drivers/infiniband/hw/cxgb4/cq.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
--- a/drivers/infiniband/hw/cxgb4/cq.c +++ b/drivers/infiniband/hw/cxgb4/cq.c @@ -484,10 +484,10 @@ static int poll_cq(struct t4_wq *wq, str ret = -EAGAIN; goto skip_cqe; } - if (unlikely((CQE_WRID_MSN(hw_cqe) != (wq->rq.msn)))) { + if (unlikely(!CQE_STATUS(hw_cqe) && + CQE_WRID_MSN(hw_cqe) != wq->rq.msn)) { t4_set_wq_in_error(wq); - hw_cqe->header |= htonl(V_CQE_STATUS(T4_ERR_MSN)); - goto proc_cqe; + hw_cqe->header |= cpu_to_be32(V_CQE_STATUS(T4_ERR_MSN)); } goto proc_cqe; }