IB/ib_mthca: Pre-link receive WQEs in Tavor mode
We have recently discovered that Tavor mode requires each WQE in a
posted list of receive WQEs to have a valid NDA field at all times.
This requirement holds true for regular QPs as well as for SRQs. This
patch prelinks the receive queue in a regular QP and keeps the free
list in SRQ always properly linked.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Reviewed-by: Jack Morgenstein <jackm@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 ec63adc..a5ffff6 100644
--- a/drivers/infiniband/hw/mthca/mthca_srq.c
+++ b/drivers/infiniband/hw/mthca/mthca_srq.c
@@ -175,9 +175,17 @@
* scatter list L_Keys to the sentry value of 0x100.
*/
for (i = 0; i < srq->max; ++i) {
- wqe = get_wqe(srq, i);
+ struct mthca_next_seg *next;
- *wqe_to_link(wqe) = i < srq->max - 1 ? i + 1 : -1;
+ next = wqe = get_wqe(srq, i);
+
+ if (i < srq->max - 1) {
+ *wqe_to_link(wqe) = i + 1;
+ next->nda_op = htonl(((i + 1) << srq->wqe_shift) | 1);
+ } else {
+ *wqe_to_link(wqe) = -1;
+ next->nda_op = 0;
+ }
for (scatter = wqe + sizeof (struct mthca_next_seg);
(void *) scatter < wqe + (1 << srq->wqe_shift);
@@ -470,12 +478,15 @@
void mthca_free_srq_wqe(struct mthca_srq *srq, u32 wqe_addr)
{
int ind;
+ struct mthca_next_seg *last_free;
ind = wqe_addr >> srq->wqe_shift;
spin_lock(&srq->lock);
- *wqe_to_link(get_wqe(srq, srq->last_free)) = ind;
+ last_free = get_wqe(srq, srq->last_free);
+ *wqe_to_link(last_free) = ind;
+ last_free->nda_op = htonl((ind << srq->wqe_shift) | 1);
*wqe_to_link(get_wqe(srq, ind)) = -1;
srq->last_free = ind;
@@ -516,7 +527,6 @@
prev_wqe = srq->last;
srq->last = wqe;
- ((struct mthca_next_seg *) wqe)->nda_op = 0;
((struct mthca_next_seg *) wqe)->ee_nds = 0;
/* flags field will always remain 0 */
@@ -537,9 +547,6 @@
if (i < srq->max_gs)
mthca_set_data_seg_inval(wqe);
- ((struct mthca_next_seg *) prev_wqe)->nda_op =
- cpu_to_be32((ind << srq->wqe_shift) | 1);
- wmb();
((struct mthca_next_seg *) prev_wqe)->ee_nds =
cpu_to_be32(MTHCA_NEXT_DBD);
@@ -613,8 +620,6 @@
break;
}
- ((struct mthca_next_seg *) wqe)->nda_op =
- cpu_to_be32((next_ind << srq->wqe_shift) | 1);
((struct mthca_next_seg *) wqe)->ee_nds = 0;
/* flags field will always remain 0 */