IB/qib: Optimize RC/UC code by IB operation

The memset for zeroing work completions had been unconditional.

This patch removes the memset and moves the zeroing into the work
completion with a more explicit field by field set.  With this patch,
non-ONLY/non-LAST packets will avoid the overhead since they will not
generate a completion.

Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>

2 files changed, 24 insertions, 9 deletions

diff --git a/drivers/infiniband/hw/qib/qib_rc.c b/drivers/infiniband/hw/qib/qib_rc.c
index eca0c41f1226..5c1c49b299f1 100644
--- a/drivers/infiniband/hw/qib/qib_rc.c
+++ b/drivers/infiniband/hw/qib/qib_rc.c
@@ -1955,8 +1955,6 @@ void qib_rc_rcv(struct qib_ctxtdata *rcd, struct qib_ib_header *hdr,
                 break;
         }
 
-        memset(&wc, 0, sizeof wc);
-
         if (qp->state == IB_QPS_RTR && !(qp->r_flags & QIB_R_COMM_EST)) {
                 qp->r_flags |= QIB_R_COMM_EST;
                 if (qp->ibqp.event_handler) {
@@ -2009,16 +2007,19 @@ send_middle:
                         goto rnr_nak;
                 qp->r_rcv_len = 0;
                 if (opcode == OP(SEND_ONLY))
-                        goto send_last;
-                /* FALLTHROUGH */
+                        goto no_immediate_data;
+                /* FALLTHROUGH for SEND_ONLY_WITH_IMMEDIATE */
         case OP(SEND_LAST_WITH_IMMEDIATE):
 send_last_imm:
                 wc.ex.imm_data = ohdr->u.imm_data;
                 hdrsize += 4;
                 wc.wc_flags = IB_WC_WITH_IMM;
-                /* FALLTHROUGH */
+                goto send_last;
         case OP(SEND_LAST):
         case OP(RDMA_WRITE_LAST):
+no_immediate_data:
+                wc.wc_flags = 0;
+                wc.ex.imm_data = 0;
 send_last:
                 /* Get the number of bytes the message was padded by. */
                 pad = (be32_to_cpu(ohdr->bth[0]) >> 20) & 3;
@@ -2051,6 +2052,12 @@ send_last:
                 wc.src_qp = qp->remote_qpn;
                 wc.slid = qp->remote_ah_attr.dlid;
                 wc.sl = qp->remote_ah_attr.sl;
+                /* zero fields that are N/A */
+                wc.vendor_err = 0;
+                wc.pkey_index = 0;
+                wc.dlid_path_bits = 0;
+                wc.port_num = 0;
+                wc.csum_ok = 0;
                 /* Signal completion event if the solicited bit is set. */
                 qib_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
                              (ohdr->bth[0] &
@@ -2089,7 +2096,7 @@ send_last:
                 if (opcode == OP(RDMA_WRITE_FIRST))
                         goto send_middle;
                 else if (opcode == OP(RDMA_WRITE_ONLY))
-                        goto send_last;
+                        goto no_immediate_data;
                 ret = qib_get_rwqe(qp, 1);
                 if (ret < 0)
                         goto nack_op_err;
diff --git a/drivers/infiniband/hw/qib/qib_uc.c b/drivers/infiniband/hw/qib/qib_uc.c
index 32ccf3c824ca..d31f33a99887 100644
--- a/drivers/infiniband/hw/qib/qib_uc.c
+++ b/drivers/infiniband/hw/qib/qib_uc.c
@@ -270,7 +270,6 @@ void qib_uc_rcv(struct qib_ibport *ibp, struct qib_ib_header *hdr,
 
         psn = be32_to_cpu(ohdr->bth[2]);
         opcode >>= 24;
-        memset(&wc, 0, sizeof wc);
 
         /* Compare the PSN verses the expected PSN. */
         if (unlikely(qib_cmp24(psn, qp->r_psn) != 0)) {
@@ -370,7 +369,7 @@ send_first:
                 }
                 qp->r_rcv_len = 0;
                 if (opcode == OP(SEND_ONLY))
-                        goto send_last;
+                        goto no_immediate_data;
                 else if (opcode == OP(SEND_ONLY_WITH_IMMEDIATE))
                         goto send_last_imm;
                 /* FALLTHROUGH */
@@ -389,8 +388,11 @@ send_last_imm:
                 wc.ex.imm_data = ohdr->u.imm_data;
                 hdrsize += 4;
                 wc.wc_flags = IB_WC_WITH_IMM;
-                /* FALLTHROUGH */
+                goto send_last;
         case OP(SEND_LAST):
+no_immediate_data:
+                wc.ex.imm_data = 0;
+                wc.wc_flags = 0;
 send_last:
                 /* Get the number of bytes the message was padded by. */
                 pad = (be32_to_cpu(ohdr->bth[0]) >> 20) & 3;
@@ -418,6 +420,12 @@ last_imm:
                 wc.src_qp = qp->remote_qpn;
                 wc.slid = qp->remote_ah_attr.dlid;
                 wc.sl = qp->remote_ah_attr.sl;
+                /* zero fields that are N/A */
+                wc.vendor_err = 0;
+                wc.pkey_index = 0;
+                wc.dlid_path_bits = 0;
+                wc.port_num = 0;
+                wc.csum_ok = 0;
                 /* Signal completion event if the solicited bit is set. */
                 qib_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
                              (ohdr->bth[0] &