IB/mthca: Fix posting lists of 256 receive requests to SRQ for Tavor
If we post a list of length exactly a multiple of 256, nreq in
doorbell gets set to 256 which is wrong: it should be encoded by 0.
This is because we only zero it out on the next WR, which may not be
there. The solution is to ring the doorbell after posting a WQE, not
before posting the next one.
This is the same bug that we just fixed for QPs with non-shared RQ.
Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
diff --git a/drivers/infiniband/hw/mthca/mthca_srq.c b/drivers/infiniband/hw/mthca/mthca_srq.c
index 1ea4332..b292fef 100644
--- a/drivers/infiniband/hw/mthca/mthca_srq.c
+++ b/drivers/infiniband/hw/mthca/mthca_srq.c
@@ -490,26 +490,7 @@
first_ind = srq->first_free;
- for (nreq = 0; wr; ++nreq, wr = wr->next) {
- if (unlikely(nreq == MTHCA_TAVOR_MAX_WQES_PER_RECV_DB)) {
- nreq = 0;
-
- doorbell[0] = cpu_to_be32(first_ind << srq->wqe_shift);
- doorbell[1] = cpu_to_be32(srq->srqn << 8);
-
- /*
- * Make sure that descriptors are written
- * before doorbell is rung.
- */
- wmb();
-
- mthca_write64(doorbell,
- dev->kar + MTHCA_RECEIVE_DOORBELL,
- MTHCA_GET_DOORBELL_LOCK(&dev->doorbell_lock));
-
- first_ind = srq->first_free;
- }
-
+ for (nreq = 0; wr; wr = wr->next) {
ind = srq->first_free;
if (ind < 0) {
@@ -569,6 +550,26 @@
srq->wrid[ind] = wr->wr_id;
srq->first_free = next_ind;
+
+ ++nreq;
+ if (unlikely(nreq == MTHCA_TAVOR_MAX_WQES_PER_RECV_DB)) {
+ nreq = 0;
+
+ doorbell[0] = cpu_to_be32(first_ind << srq->wqe_shift);
+ doorbell[1] = cpu_to_be32(srq->srqn << 8);
+
+ /*
+ * Make sure that descriptors are written
+ * before doorbell is rung.
+ */
+ wmb();
+
+ mthca_write64(doorbell,
+ dev->kar + MTHCA_RECEIVE_DOORBELL,
+ MTHCA_GET_DOORBELL_LOCK(&dev->doorbell_lock));
+
+ first_ind = srq->first_free;
+ }
}
if (likely(nreq)) {