aboutsummaryrefslogtreecommitdiffstats
path: root/net/sunrpc/xprtrdma
diff options
context:
space:
mode:
Diffstat (limited to 'net/sunrpc/xprtrdma')
-rw-r--r--net/sunrpc/xprtrdma/backchannel.c3
-rw-r--r--net/sunrpc/xprtrdma/frwr_ops.c4
-rw-r--r--net/sunrpc/xprtrdma/rpc_rdma.c22
-rw-r--r--net/sunrpc/xprtrdma/svc_rdma_backchannel.c1
-rw-r--r--net/sunrpc/xprtrdma/svc_rdma_recvfrom.c12
-rw-r--r--net/sunrpc/xprtrdma/svc_rdma_rw.c17
-rw-r--r--net/sunrpc/xprtrdma/svc_rdma_sendto.c4
-rw-r--r--net/sunrpc/xprtrdma/svc_rdma_transport.c9
-rw-r--r--net/sunrpc/xprtrdma/transport.c2
-rw-r--r--net/sunrpc/xprtrdma/verbs.c2
-rw-r--r--net/sunrpc/xprtrdma/xprt_rdma.h12
11 files changed, 45 insertions, 43 deletions
diff --git a/net/sunrpc/xprtrdma/backchannel.c b/net/sunrpc/xprtrdma/backchannel.c
index 0de9b3e63770..d79b18c1f4cd 100644
--- a/net/sunrpc/xprtrdma/backchannel.c
+++ b/net/sunrpc/xprtrdma/backchannel.c
@@ -123,7 +123,7 @@ static int rpcrdma_bc_marshal_reply(struct rpc_rqst *rqst)
rpcrdma_set_xdrlen(&req->rl_hdrbuf, 0);
xdr_init_encode(&req->rl_stream, &req->rl_hdrbuf,
- req->rl_rdmabuf->rg_base);
+ req->rl_rdmabuf->rg_base, rqst);
p = xdr_reserve_space(&req->rl_stream, 28);
if (unlikely(!p))
@@ -267,7 +267,6 @@ void rpcrdma_bc_receive_call(struct rpcrdma_xprt *r_xprt,
/* Prepare rqst */
rqst->rq_reply_bytes_recvd = 0;
- rqst->rq_bytes_sent = 0;
rqst->rq_xid = *p;
rqst->rq_private_buf.len = size;
diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c
index 6a561056b538..52cb6c1b0c2b 100644
--- a/net/sunrpc/xprtrdma/frwr_ops.c
+++ b/net/sunrpc/xprtrdma/frwr_ops.c
@@ -391,7 +391,7 @@ frwr_wc_localinv_wake(struct ib_cq *cq, struct ib_wc *wc)
*/
struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt,
struct rpcrdma_mr_seg *seg,
- int nsegs, bool writing, u32 xid,
+ int nsegs, bool writing, __be32 xid,
struct rpcrdma_mr **out)
{
struct rpcrdma_ia *ia = &r_xprt->rx_ia;
@@ -446,7 +446,7 @@ struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt,
goto out_mapmr_err;
ibmr->iova &= 0x00000000ffffffff;
- ibmr->iova |= ((u64)cpu_to_be32(xid)) << 32;
+ ibmr->iova |= ((u64)be32_to_cpu(xid)) << 32;
key = (u8)(ibmr->rkey & 0x000000FF);
ib_update_fast_reg_key(ibmr, ++key);
diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c
index d18614e02b4e..6c1fb270f127 100644
--- a/net/sunrpc/xprtrdma/rpc_rdma.c
+++ b/net/sunrpc/xprtrdma/rpc_rdma.c
@@ -164,6 +164,21 @@ static bool rpcrdma_results_inline(struct rpcrdma_xprt *r_xprt,
return rqst->rq_rcv_buf.buflen <= ia->ri_max_inline_read;
}
+/* The client is required to provide a Reply chunk if the maximum
+ * size of the non-payload part of the RPC Reply is larger than
+ * the inline threshold.
+ */
+static bool
+rpcrdma_nonpayload_inline(const struct rpcrdma_xprt *r_xprt,
+ const struct rpc_rqst *rqst)
+{
+ const struct xdr_buf *buf = &rqst->rq_rcv_buf;
+ const struct rpcrdma_ia *ia = &r_xprt->rx_ia;
+
+ return buf->head[0].iov_len + buf->tail[0].iov_len <
+ ia->ri_max_inline_read;
+}
+
/* Split @vec on page boundaries into SGEs. FMR registers pages, not
* a byte range. Other modes coalesce these SGEs into a single MR
* when they can.
@@ -733,7 +748,7 @@ rpcrdma_marshal_req(struct rpcrdma_xprt *r_xprt, struct rpc_rqst *rqst)
rpcrdma_set_xdrlen(&req->rl_hdrbuf, 0);
xdr_init_encode(xdr, &req->rl_hdrbuf,
- req->rl_rdmabuf->rg_base);
+ req->rl_rdmabuf->rg_base, rqst);
/* Fixed header fields */
ret = -EMSGSIZE;
@@ -762,7 +777,8 @@ rpcrdma_marshal_req(struct rpcrdma_xprt *r_xprt, struct rpc_rqst *rqst)
*/
if (rpcrdma_results_inline(r_xprt, rqst))
wtype = rpcrdma_noch;
- else if (ddp_allowed && rqst->rq_rcv_buf.flags & XDRBUF_READ)
+ else if ((ddp_allowed && rqst->rq_rcv_buf.flags & XDRBUF_READ) &&
+ rpcrdma_nonpayload_inline(r_xprt, rqst))
wtype = rpcrdma_writech;
else
wtype = rpcrdma_replych;
@@ -1313,7 +1329,7 @@ void rpcrdma_reply_handler(struct rpcrdma_rep *rep)
/* Fixed transport header fields */
xdr_init_decode(&rep->rr_stream, &rep->rr_hdrbuf,
- rep->rr_hdrbuf.head[0].iov_base);
+ rep->rr_hdrbuf.head[0].iov_base, NULL);
p = xdr_inline_decode(&rep->rr_stream, 4 * sizeof(*p));
if (unlikely(!p))
goto out_shortreply;
diff --git a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
index b908f2ca08fd..907464c2a9f0 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
@@ -304,7 +304,6 @@ xprt_setup_rdma_bc(struct xprt_create *args)
xprt->idle_timeout = RPCRDMA_IDLE_DISC_TO;
xprt->prot = XPRT_TRANSPORT_BC_RDMA;
- xprt->tsh_size = 0;
xprt->ops = &xprt_rdma_bc_procs;
memcpy(&xprt->addr, args->dstaddr, args->addrlen);
diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
index 828b149eaaef..65e2fb9aac65 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
@@ -272,11 +272,8 @@ bool svc_rdma_post_recvs(struct svcxprt_rdma *rdma)
return false;
ctxt->rc_temp = true;
ret = __svc_rdma_post_recv(rdma, ctxt);
- if (ret) {
- pr_err("svcrdma: failure posting recv buffers: %d\n",
- ret);
+ if (ret)
return false;
- }
}
return true;
}
@@ -314,17 +311,14 @@ static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
spin_lock(&rdma->sc_rq_dto_lock);
list_add_tail(&ctxt->rc_list, &rdma->sc_rq_dto_q);
- spin_unlock(&rdma->sc_rq_dto_lock);
+ /* Note the unlock pairs with the smp_rmb in svc_xprt_ready: */
set_bit(XPT_DATA, &rdma->sc_xprt.xpt_flags);
+ spin_unlock(&rdma->sc_rq_dto_lock);
if (!test_bit(RDMAXPRT_CONN_PENDING, &rdma->sc_flags))
svc_xprt_enqueue(&rdma->sc_xprt);
goto out;
flushed:
- if (wc->status != IB_WC_WR_FLUSH_ERR)
- pr_err("svcrdma: Recv: %s (%u/0x%x)\n",
- ib_wc_status_msg(wc->status),
- wc->status, wc->vendor_err);
post_err:
svc_rdma_recv_ctxt_put(rdma, ctxt);
set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
diff --git a/net/sunrpc/xprtrdma/svc_rdma_rw.c b/net/sunrpc/xprtrdma/svc_rdma_rw.c
index dc1951759a8e..2121c9b4d275 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_rw.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_rw.c
@@ -64,8 +64,7 @@ svc_rdma_get_rw_ctxt(struct svcxprt_rdma *rdma, unsigned int sges)
spin_unlock(&rdma->sc_rw_ctxt_lock);
} else {
spin_unlock(&rdma->sc_rw_ctxt_lock);
- ctxt = kmalloc(sizeof(*ctxt) +
- SG_CHUNK_SIZE * sizeof(struct scatterlist),
+ ctxt = kmalloc(struct_size(ctxt, rw_first_sgl, SG_CHUNK_SIZE),
GFP_KERNEL);
if (!ctxt)
goto out;
@@ -213,13 +212,8 @@ static void svc_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
atomic_add(cc->cc_sqecount, &rdma->sc_sq_avail);
wake_up(&rdma->sc_send_wait);
- if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ if (unlikely(wc->status != IB_WC_SUCCESS))
set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
- if (wc->status != IB_WC_WR_FLUSH_ERR)
- pr_err("svcrdma: write ctx: %s (%u/0x%x)\n",
- ib_wc_status_msg(wc->status),
- wc->status, wc->vendor_err);
- }
svc_rdma_write_info_free(info);
}
@@ -278,18 +272,15 @@ static void svc_rdma_wc_read_done(struct ib_cq *cq, struct ib_wc *wc)
if (unlikely(wc->status != IB_WC_SUCCESS)) {
set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
- if (wc->status != IB_WC_WR_FLUSH_ERR)
- pr_err("svcrdma: read ctx: %s (%u/0x%x)\n",
- ib_wc_status_msg(wc->status),
- wc->status, wc->vendor_err);
svc_rdma_recv_ctxt_put(rdma, info->ri_readctxt);
} else {
spin_lock(&rdma->sc_rq_dto_lock);
list_add_tail(&info->ri_readctxt->rc_list,
&rdma->sc_read_complete_q);
+ /* Note the unlock pairs with the smp_rmb in svc_xprt_ready: */
+ set_bit(XPT_DATA, &rdma->sc_xprt.xpt_flags);
spin_unlock(&rdma->sc_rq_dto_lock);
- set_bit(XPT_DATA, &rdma->sc_xprt.xpt_flags);
svc_xprt_enqueue(&rdma->sc_xprt);
}
diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index 1f200119268c..6fdba72f89f4 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -272,10 +272,6 @@ static void svc_rdma_wc_send(struct ib_cq *cq, struct ib_wc *wc)
if (unlikely(wc->status != IB_WC_SUCCESS)) {
set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
svc_xprt_enqueue(&rdma->sc_xprt);
- if (wc->status != IB_WC_WR_FLUSH_ERR)
- pr_err("svcrdma: Send: %s (%u/0x%x)\n",
- ib_wc_status_msg(wc->status),
- wc->status, wc->vendor_err);
}
svc_xprt_put(&rdma->sc_xprt);
diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
index 57f86c63a463..027a3b07d329 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
@@ -390,8 +390,8 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
struct ib_qp_init_attr qp_attr;
unsigned int ctxts, rq_depth;
struct ib_device *dev;
- struct sockaddr *sap;
int ret = 0;
+ RPC_IFDEBUG(struct sockaddr *sap);
listen_rdma = container_of(xprt, struct svcxprt_rdma, sc_xprt);
clear_bit(XPT_CONN, &xprt->xpt_flags);
@@ -525,6 +525,7 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
if (ret)
goto errout;
+#if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
dprintk("svcrdma: new connection %p accepted:\n", newxprt);
sap = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.src_addr;
dprintk(" local address : %pIS:%u\n", sap, rpc_get_port(sap));
@@ -535,6 +536,7 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
dprintk(" rdma_rw_ctxs : %d\n", ctxts);
dprintk(" max_requests : %d\n", newxprt->sc_max_requests);
dprintk(" ord : %d\n", conn_param.initiator_depth);
+#endif
trace_svcrdma_xprt_accept(&newxprt->sc_xprt);
return &newxprt->sc_xprt;
@@ -588,11 +590,6 @@ static void __svc_rdma_free(struct work_struct *work)
if (rdma->sc_qp && !IS_ERR(rdma->sc_qp))
ib_drain_qp(rdma->sc_qp);
- /* We should only be called from kref_put */
- if (kref_read(&xprt->xpt_ref) != 0)
- pr_err("svcrdma: sc_xprt still in use? (%d)\n",
- kref_read(&xprt->xpt_ref));
-
svc_rdma_flush_recv_queues(rdma);
/* Final put of backchannel client transport */
diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c
index fbc171ebfe91..5d261353bd90 100644
--- a/net/sunrpc/xprtrdma/transport.c
+++ b/net/sunrpc/xprtrdma/transport.c
@@ -332,7 +332,6 @@ xprt_setup_rdma(struct xprt_create *args)
xprt->idle_timeout = RPCRDMA_IDLE_DISC_TO;
xprt->resvport = 0; /* privileged port not needed */
- xprt->tsh_size = 0; /* RPC-RDMA handles framing */
xprt->ops = &xprt_rdma_procs;
/*
@@ -738,7 +737,6 @@ xprt_rdma_send_request(struct rpc_rqst *rqst)
goto drop_connection;
rqst->rq_xmit_bytes_sent += rqst->rq_snd_buf.len;
- rqst->rq_bytes_sent = 0;
/* An RPC with no reply will throw off credit accounting,
* so drop the connection to reset the credit grant.
diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c
index 21113bfd4eca..89a63391d4d4 100644
--- a/net/sunrpc/xprtrdma/verbs.c
+++ b/net/sunrpc/xprtrdma/verbs.c
@@ -1481,6 +1481,8 @@ rpcrdma_post_recvs(struct rpcrdma_xprt *r_xprt, bool temp)
if (ep->rep_receive_count > needed)
goto out;
needed -= ep->rep_receive_count;
+ if (!temp)
+ needed += RPCRDMA_MAX_RECV_BATCH;
count = 0;
wr = NULL;
diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
index 5a18472f2c9c..10f6593e1a6a 100644
--- a/net/sunrpc/xprtrdma/xprt_rdma.h
+++ b/net/sunrpc/xprtrdma/xprt_rdma.h
@@ -205,6 +205,16 @@ struct rpcrdma_rep {
struct ib_recv_wr rr_recv_wr;
};
+/* To reduce the rate at which a transport invokes ib_post_recv
+ * (and thus the hardware doorbell rate), xprtrdma posts Receive
+ * WRs in batches.
+ *
+ * Setting this to zero disables Receive post batching.
+ */
+enum {
+ RPCRDMA_MAX_RECV_BATCH = 7,
+};
+
/* struct rpcrdma_sendctx - DMA mapped SGEs to unmap after Send completes
*/
struct rpcrdma_req;
@@ -577,7 +587,7 @@ void frwr_release_mr(struct rpcrdma_mr *mr);
size_t frwr_maxpages(struct rpcrdma_xprt *r_xprt);
struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt,
struct rpcrdma_mr_seg *seg,
- int nsegs, bool writing, u32 xid,
+ int nsegs, bool writing, __be32 xid,
struct rpcrdma_mr **mr);
int frwr_send(struct rpcrdma_ia *ia, struct rpcrdma_req *req);
void frwr_reminv(struct rpcrdma_rep *rep, struct list_head *mrs);
Copyright © 1996 – 2023 Jason A. Donenfeld. All Rights Reverse Engineered.