All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 1 0/4] Move post-send page release into svc_sendto
@ 2021-03-23 20:37 Chuck Lever
  2021-03-23 20:37 ` [PATCH 1 1/4] svcrdma: Add a "deferred close" helper Chuck Lever
                   ` (3 more replies)
  0 siblings, 4 replies; 5+ messages in thread
From: Chuck Lever @ 2021-03-23 20:37 UTC (permalink / raw)
  To: linux-nfs, linux-rdma

I'm trying to move work out of the completion handlers, because these
are a convenient but limited resource due to their single-threaded
nature.

As a first step I've moved the code that DMA unmaps each
svc_rdma_send_ctxt and returns it to the free list to the tail of
svc_rdma_sendto().

---

Chuck Lever (4):
      svcrdma: Add a "deferred close" helper
      svcrdma: Normalize Send page handling
      svcrdma: Remove unused sc_pages field
      svcrdma: Retain the page backing rq_res.head[0].iov_base


 include/linux/sunrpc/svc_rdma.h            |  4 +-
 net/sunrpc/xprtrdma/svc_rdma_backchannel.c |  8 ++-
 net/sunrpc/xprtrdma/svc_rdma_sendto.c      | 57 ++++++++--------------
 3 files changed, 29 insertions(+), 40 deletions(-)

--
Chuck Lever


^ permalink raw reply	[flat|nested] 5+ messages in thread
* [PATCH 1 1/4] svcrdma: Add a "deferred close" helper
  2021-03-23 20:37 [PATCH 1 0/4] Move post-send page release into svc_sendto Chuck Lever
@ 2021-03-23 20:37 ` Chuck Lever
  2021-03-23 20:37 ` [PATCH 1 2/4] svcrdma: Normalize Send page handling Chuck Lever
                   ` (2 subsequent siblings)
  3 siblings, 0 replies; 5+ messages in thread
From: Chuck Lever @ 2021-03-23 20:37 UTC (permalink / raw)
  To: linux-nfs, linux-rdma

Refactor a bit of commonly used logic so that every site that wants
a close deferred to an nfsd thread does all the right things
(set_bit(XPT_CLOSE) then enqueue).

Also, once XPT_CLOSE is set on a transport, it is never cleared. If
XPT_CLOSE is already set, then the close is already being handled
and the enqueue can be skipped.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 include/linux/sunrpc/svc_xprt.h          |    1 +
 net/sunrpc/svc_xprt.c                    |   14 ++++++++++++++
 net/sunrpc/svcsock.c                     |   15 ++++++---------
 net/sunrpc/xprtrdma/svc_rdma_recvfrom.c  |    3 +--
 net/sunrpc/xprtrdma/svc_rdma_rw.c        |    5 ++---
 net/sunrpc/xprtrdma/svc_rdma_sendto.c    |   10 ++++------
 net/sunrpc/xprtrdma/svc_rdma_transport.c |    6 ++----
 7 files changed, 30 insertions(+), 24 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 92455e0d5244..34dacadfe517 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -143,6 +143,7 @@ struct	svc_xprt *svc_find_xprt(struct svc_serv *serv, const char *xcl_name,
 int	svc_xprt_names(struct svc_serv *serv, char *buf, const int buflen);
 void	svc_add_new_perm_xprt(struct svc_serv *serv, struct svc_xprt *xprt);
 void	svc_age_temp_xprts_now(struct svc_serv *, struct sockaddr *);
+void	svc_xprt_deferred_close(struct svc_xprt *xprt);
 
 static inline void svc_xprt_get(struct svc_xprt *xprt)
 {
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 3cdd71a8df1e..b134fc5f3b8d 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -139,6 +139,20 @@ int svc_print_xprts(char *buf, int maxlen)
 	return len;
 }
 
+/**
+ * svc_xprt_deferred_close - Close a transport
+ * @xprt: transport instance
+ *
+ * Used in contexts that need to defer the work of shutting down
+ * the transport to an nfsd thread.
+ */
+void svc_xprt_deferred_close(struct svc_xprt *xprt)
+{
+	if (!test_and_set_bit(XPT_CLOSE, &xprt->xpt_flags))
+		svc_xprt_enqueue(xprt);
+}
+EXPORT_SYMBOL_GPL(svc_xprt_deferred_close);
+
 static void svc_xprt_free(struct kref *kref)
 {
 	struct svc_xprt *xprt =
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 2e2f007dfc9f..22454b2df5be 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -728,10 +728,8 @@ static void svc_tcp_state_change(struct sock *sk)
 		rmb();
 		svsk->sk_ostate(sk);
 		trace_svcsock_tcp_state(&svsk->sk_xprt, svsk->sk_sock);
-		if (sk->sk_state != TCP_ESTABLISHED) {
-			set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
-			svc_xprt_enqueue(&svsk->sk_xprt);
-		}
+		if (sk->sk_state != TCP_ESTABLISHED)
+			svc_xprt_deferred_close(&svsk->sk_xprt);
 	}
 }
 
@@ -901,7 +899,7 @@ static ssize_t svc_tcp_read_marker(struct svc_sock *svsk,
 	net_notice_ratelimited("svc: %s %s RPC fragment too large: %d\n",
 			       __func__, svsk->sk_xprt.xpt_server->sv_name,
 			       svc_sock_reclen(svsk));
-	set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
+	svc_xprt_deferred_close(&svsk->sk_xprt);
 err_short:
 	return -EAGAIN;
 }
@@ -1057,7 +1055,7 @@ static int svc_tcp_recvfrom(struct svc_rqst *rqstp)
 	svsk->sk_datalen = 0;
 err_delete:
 	trace_svcsock_tcp_recv_err(&svsk->sk_xprt, len);
-	set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
+	svc_xprt_deferred_close(&svsk->sk_xprt);
 err_noclose:
 	return 0;	/* record not complete */
 }
@@ -1188,8 +1186,7 @@ static int svc_tcp_sendto(struct svc_rqst *rqstp)
 		  xprt->xpt_server->sv_name,
 		  (err < 0) ? "got error" : "sent",
 		  (err < 0) ? err : sent, xdr->len);
-	set_bit(XPT_CLOSE, &xprt->xpt_flags);
-	svc_xprt_enqueue(xprt);
+	svc_xprt_deferred_close(xprt);
 	atomic_dec(&svsk->sk_sendqlen);
 	mutex_unlock(&xprt->xpt_mutex);
 	return -EAGAIN;
@@ -1268,7 +1265,7 @@ static void svc_tcp_init(struct svc_sock *svsk, struct svc_serv *serv)
 		case TCP_ESTABLISHED:
 			break;
 		default:
-			set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
+			svc_xprt_deferred_close(&svsk->sk_xprt);
 		}
 	}
 }
diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
index 2571188ef7f2..8d93d26e0318 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
@@ -367,8 +367,7 @@ static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
 
 flushed:
 	svc_rdma_recv_ctxt_put(rdma, ctxt);
-	set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
-	svc_xprt_enqueue(&rdma->sc_xprt);
+	svc_xprt_deferred_close(&rdma->sc_xprt);
 }
 
 /**
diff --git a/net/sunrpc/xprtrdma/svc_rdma_rw.c b/net/sunrpc/xprtrdma/svc_rdma_rw.c
index 693d139a8633..d7054e3a8e33 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_rw.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_rw.c
@@ -250,7 +250,7 @@ static void svc_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
 	wake_up(&rdma->sc_send_wait);
 
 	if (unlikely(wc->status != IB_WC_SUCCESS))
-		set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
+		svc_xprt_deferred_close(&rdma->sc_xprt);
 
 	svc_rdma_write_info_free(info);
 }
@@ -334,7 +334,6 @@ static void svc_rdma_wc_read_done(struct ib_cq *cq, struct ib_wc *wc)
 static int svc_rdma_post_chunk_ctxt(struct svc_rdma_chunk_ctxt *cc)
 {
 	struct svcxprt_rdma *rdma = cc->cc_rdma;
-	struct svc_xprt *xprt = &rdma->sc_xprt;
 	struct ib_send_wr *first_wr;
 	const struct ib_send_wr *bad_wr;
 	struct list_head *tmp;
@@ -373,7 +372,7 @@ static int svc_rdma_post_chunk_ctxt(struct svc_rdma_chunk_ctxt *cc)
 	} while (1);
 
 	trace_svcrdma_sq_post_err(rdma, ret);
-	set_bit(XPT_CLOSE, &xprt->xpt_flags);
+	svc_xprt_deferred_close(&rdma->sc_xprt);
 
 	/* If even one was posted, there will be a completion. */
 	if (bad_wr != first_wr)
diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index e6fab5dd20d0..4471a0fcd3a3 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -285,10 +285,8 @@ static void svc_rdma_wc_send(struct ib_cq *cq, struct ib_wc *wc)
 
 	svc_rdma_send_ctxt_put(rdma, ctxt);
 
-	if (unlikely(wc->status != IB_WC_SUCCESS)) {
-		set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
-		svc_xprt_enqueue(&rdma->sc_xprt);
-	}
+	if (unlikely(wc->status != IB_WC_SUCCESS))
+		svc_xprt_deferred_close(&rdma->sc_xprt);
 }
 
 /**
@@ -334,7 +332,7 @@ int svc_rdma_send(struct svcxprt_rdma *rdma, struct svc_rdma_send_ctxt *ctxt)
 	}
 
 	trace_svcrdma_sq_post_err(rdma, ret);
-	set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
+	svc_xprt_deferred_close(&rdma->sc_xprt);
 	wake_up(&rdma->sc_send_wait);
 	return ret;
 }
@@ -994,7 +992,7 @@ int svc_rdma_sendto(struct svc_rqst *rqstp)
 	svc_rdma_send_ctxt_put(rdma, sctxt);
  err0:
 	trace_svcrdma_send_err(rqstp, ret);
-	set_bit(XPT_CLOSE, &xprt->xpt_flags);
+	svc_xprt_deferred_close(&rdma->sc_xprt);
 	return -ENOTCONN;
 }
 
diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
index e629eacfedfc..3646216211c5 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
@@ -119,8 +119,7 @@ static void qp_event_handler(struct ib_event *event, void *context)
 	case IB_EVENT_QP_ACCESS_ERR:
 	case IB_EVENT_DEVICE_FATAL:
 	default:
-		set_bit(XPT_CLOSE, &xprt->xpt_flags);
-		svc_xprt_enqueue(xprt);
+		svc_xprt_deferred_close(xprt);
 		break;
 	}
 }
@@ -286,8 +285,7 @@ static int svc_rdma_cma_handler(struct rdma_cm_id *cma_id,
 		break;
 	case RDMA_CM_EVENT_DISCONNECTED:
 	case RDMA_CM_EVENT_DEVICE_REMOVAL:
-		set_bit(XPT_CLOSE, &xprt->xpt_flags);
-		svc_xprt_enqueue(xprt);
+		svc_xprt_deferred_close(xprt);
 		break;
 	default:
 		break;



^ permalink raw reply related	[flat|nested] 5+ messages in thread
* [PATCH 1 2/4] svcrdma: Normalize Send page handling
  2021-03-23 20:37 [PATCH 1 0/4] Move post-send page release into svc_sendto Chuck Lever
  2021-03-23 20:37 ` [PATCH 1 1/4] svcrdma: Add a "deferred close" helper Chuck Lever
@ 2021-03-23 20:37 ` Chuck Lever
  2021-03-23 20:37 ` [PATCH 1 3/4] svcrdma: Remove unused sc_pages field Chuck Lever
  2021-03-23 20:37 ` [PATCH 1 4/4] svcrdma: Retain the page backing rq_res.head[0].iov_base Chuck Lever
  3 siblings, 0 replies; 5+ messages in thread
From: Chuck Lever @ 2021-03-23 20:37 UTC (permalink / raw)
  To: linux-nfs, linux-rdma

Currently svc_rdma_sendto() migrates xdr_buf pages into a separate
page list and NULLs out a bunch of entries in rq_pages while the
pages are under I/O. The Send completion handler then frees those
pages later.

Instead, let's wait for the Send completion, then handle page
releasing in the nfsd thread. I'd like to avoid the cost of 250+
put_page() calls in the Send completion handler, which is single-
threaded.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 include/linux/sunrpc/svc_rdma.h            |    1 +
 net/sunrpc/xprtrdma/svc_rdma_backchannel.c |    8 +++++++-
 net/sunrpc/xprtrdma/svc_rdma_sendto.c      |   29 +++++++++++++++-------------
 3 files changed, 24 insertions(+), 14 deletions(-)

diff --git a/include/linux/sunrpc/svc_rdma.h b/include/linux/sunrpc/svc_rdma.h
index 722fc7c48725..5841978550c6 100644
--- a/include/linux/sunrpc/svc_rdma.h
+++ b/include/linux/sunrpc/svc_rdma.h
@@ -160,6 +160,7 @@ struct svc_rdma_send_ctxt {
 
 	struct ib_send_wr	sc_send_wr;
 	struct ib_cqe		sc_cqe;
+	struct completion	sc_done;
 	struct xdr_buf		sc_hdrbuf;
 	struct xdr_stream	sc_stream;
 	void			*sc_xprt_buf;
diff --git a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
index 9150df35fb6f..16897fcb659c 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
@@ -93,7 +93,13 @@ static int svc_rdma_bc_sendto(struct svcxprt_rdma *rdma,
 	 */
 	get_page(virt_to_page(rqst->rq_buffer));
 	sctxt->sc_send_wr.opcode = IB_WR_SEND;
-	return svc_rdma_send(rdma, sctxt);
+	ret = svc_rdma_send(rdma, sctxt);
+	if (ret < 0)
+		return ret;
+
+	ret = wait_for_completion_killable(&sctxt->sc_done);
+	svc_rdma_send_ctxt_put(rdma, sctxt);
+	return ret;
 }
 
 /* Server-side transport endpoint wants a whole page for its send
diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index 4471a0fcd3a3..62d55850ca54 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -155,6 +155,7 @@ svc_rdma_send_ctxt_alloc(struct svcxprt_rdma *rdma)
 	ctxt->sc_send_wr.wr_cqe = &ctxt->sc_cqe;
 	ctxt->sc_send_wr.sg_list = ctxt->sc_sges;
 	ctxt->sc_send_wr.send_flags = IB_SEND_SIGNALED;
+	init_completion(&ctxt->sc_done);
 	ctxt->sc_cqe.done = svc_rdma_wc_send;
 	ctxt->sc_xprt_buf = buffer;
 	xdr_buf_init(&ctxt->sc_hdrbuf, ctxt->sc_xprt_buf,
@@ -280,11 +281,11 @@ static void svc_rdma_wc_send(struct ib_cq *cq, struct ib_wc *wc)
 
 	trace_svcrdma_wc_send(wc, &ctxt->sc_cid);
 
+	complete(&ctxt->sc_done);
+
 	atomic_inc(&rdma->sc_sq_avail);
 	wake_up(&rdma->sc_send_wait);
 
-	svc_rdma_send_ctxt_put(rdma, ctxt);
-
 	if (unlikely(wc->status != IB_WC_SUCCESS))
 		svc_xprt_deferred_close(&rdma->sc_xprt);
 }
@@ -294,7 +295,7 @@ static void svc_rdma_wc_send(struct ib_cq *cq, struct ib_wc *wc)
  * @rdma: transport on which to post the WR
  * @ctxt: send ctxt with a Send WR ready to post
  *
- * Returns zero the Send WR was posted successfully. Otherwise, a
+ * Returns zero if the Send WR was posted successfully. Otherwise, a
  * negative errno is returned.
  */
 int svc_rdma_send(struct svcxprt_rdma *rdma, struct svc_rdma_send_ctxt *ctxt)
@@ -302,7 +303,7 @@ int svc_rdma_send(struct svcxprt_rdma *rdma, struct svc_rdma_send_ctxt *ctxt)
 	struct ib_send_wr *wr = &ctxt->sc_send_wr;
 	int ret;
 
-	might_sleep();
+	reinit_completion(&ctxt->sc_done);
 
 	/* Sync the transport header buffer */
 	ib_dma_sync_single_for_device(rdma->sc_pd->device,
@@ -795,7 +796,7 @@ int svc_rdma_map_reply_msg(struct svcxprt_rdma *rdma,
  * svc_rdma_sendto returns. Transfer pages under I/O to the ctxt
  * so they are released by the Send completion handler.
  */
-static void svc_rdma_save_io_pages(struct svc_rqst *rqstp,
+static inline void svc_rdma_save_io_pages(struct svc_rqst *rqstp,
 				   struct svc_rdma_send_ctxt *ctxt)
 {
 	int i, pages = rqstp->rq_next_page - rqstp->rq_respages;
@@ -839,15 +840,20 @@ static int svc_rdma_send_reply_msg(struct svcxprt_rdma *rdma,
 	if (ret < 0)
 		return ret;
 
-	svc_rdma_save_io_pages(rqstp, sctxt);
-
 	if (rctxt->rc_inv_rkey) {
 		sctxt->sc_send_wr.opcode = IB_WR_SEND_WITH_INV;
 		sctxt->sc_send_wr.ex.invalidate_rkey = rctxt->rc_inv_rkey;
 	} else {
 		sctxt->sc_send_wr.opcode = IB_WR_SEND;
 	}
-	return svc_rdma_send(rdma, sctxt);
+
+	ret = svc_rdma_send(rdma, sctxt);
+	if (ret < 0)
+		return ret;
+
+	ret = wait_for_completion_killable(&sctxt->sc_done);
+	svc_rdma_send_ctxt_put(rdma, sctxt);
+	return ret;
 }
 
 /**
@@ -913,7 +919,8 @@ void svc_rdma_send_error_msg(struct svcxprt_rdma *rdma,
 	sctxt->sc_sges[0].length = sctxt->sc_hdrbuf.len;
 	if (svc_rdma_send(rdma, sctxt))
 		goto put_ctxt;
-	return;
+
+	wait_for_completion_killable(&sctxt->sc_done);
 
 put_ctxt:
 	svc_rdma_send_ctxt_put(rdma, sctxt);
@@ -981,10 +988,6 @@ int svc_rdma_sendto(struct svc_rqst *rqstp)
 	if (ret != -E2BIG && ret != -EINVAL)
 		goto err1;
 
-	/* Send completion releases payload pages that were part
-	 * of previously posted RDMA Writes.
-	 */
-	svc_rdma_save_io_pages(rqstp, sctxt);
 	svc_rdma_send_error_msg(rdma, sctxt, rctxt, ret);
 	return 0;
 



^ permalink raw reply related	[flat|nested] 5+ messages in thread
* [PATCH 1 3/4] svcrdma: Remove unused sc_pages field
  2021-03-23 20:37 [PATCH 1 0/4] Move post-send page release into svc_sendto Chuck Lever
  2021-03-23 20:37 ` [PATCH 1 1/4] svcrdma: Add a "deferred close" helper Chuck Lever
  2021-03-23 20:37 ` [PATCH 1 2/4] svcrdma: Normalize Send page handling Chuck Lever
@ 2021-03-23 20:37 ` Chuck Lever
  2021-03-23 20:37 ` [PATCH 1 4/4] svcrdma: Retain the page backing rq_res.head[0].iov_base Chuck Lever
  3 siblings, 0 replies; 5+ messages in thread
From: Chuck Lever @ 2021-03-23 20:37 UTC (permalink / raw)
  To: linux-nfs, linux-rdma

Clean up. This significantly reduces the size of struct
svc_rdma_send_ctxt.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 include/linux/sunrpc/svc_rdma.h       |    3 +--
 net/sunrpc/xprtrdma/svc_rdma_sendto.c |   25 -------------------------
 2 files changed, 1 insertion(+), 27 deletions(-)

diff --git a/include/linux/sunrpc/svc_rdma.h b/include/linux/sunrpc/svc_rdma.h
index 5841978550c6..6e621e1f56b8 100644
--- a/include/linux/sunrpc/svc_rdma.h
+++ b/include/linux/sunrpc/svc_rdma.h
@@ -164,9 +164,8 @@ struct svc_rdma_send_ctxt {
 	struct xdr_buf		sc_hdrbuf;
 	struct xdr_stream	sc_stream;
 	void			*sc_xprt_buf;
-	int			sc_page_count;
 	int			sc_cur_sge_no;
-	struct page		*sc_pages[RPCSVC_MAXPAGES];
+
 	struct ib_sge		sc_sges[];
 };
 
diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index 62d55850ca54..f093c9b536ff 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -219,7 +219,6 @@ struct svc_rdma_send_ctxt *svc_rdma_send_ctxt_get(struct svcxprt_rdma *rdma)
 
 	ctxt->sc_send_wr.num_sge = 0;
 	ctxt->sc_cur_sge_no = 0;
-	ctxt->sc_page_count = 0;
 	return ctxt;
 
 out_empty:
@@ -234,8 +233,6 @@ struct svc_rdma_send_ctxt *svc_rdma_send_ctxt_get(struct svcxprt_rdma *rdma)
  * svc_rdma_send_ctxt_put - Return send_ctxt to free list
  * @rdma: controlling svcxprt_rdma
  * @ctxt: object to return to the free list
- *
- * Pages left in sc_pages are DMA unmapped and released.
  */
 void svc_rdma_send_ctxt_put(struct svcxprt_rdma *rdma,
 			    struct svc_rdma_send_ctxt *ctxt)
@@ -256,9 +253,6 @@ void svc_rdma_send_ctxt_put(struct svcxprt_rdma *rdma,
 					     ctxt->sc_sges[i].length);
 	}
 
-	for (i = 0; i < ctxt->sc_page_count; ++i)
-		put_page(ctxt->sc_pages[i]);
-
 	spin_lock(&rdma->sc_send_lock);
 	list_add(&ctxt->sc_list, &rdma->sc_send_ctxts);
 	spin_unlock(&rdma->sc_send_lock);
@@ -792,25 +786,6 @@ int svc_rdma_map_reply_msg(struct svcxprt_rdma *rdma,
 				       svc_rdma_xb_dma_map, &args);
 }
 
-/* The svc_rqst and all resources it owns are released as soon as
- * svc_rdma_sendto returns. Transfer pages under I/O to the ctxt
- * so they are released by the Send completion handler.
- */
-static inline void svc_rdma_save_io_pages(struct svc_rqst *rqstp,
-				   struct svc_rdma_send_ctxt *ctxt)
-{
-	int i, pages = rqstp->rq_next_page - rqstp->rq_respages;
-
-	ctxt->sc_page_count += pages;
-	for (i = 0; i < pages; i++) {
-		ctxt->sc_pages[i] = rqstp->rq_respages[i];
-		rqstp->rq_respages[i] = NULL;
-	}
-
-	/* Prevent svc_xprt_release from releasing pages in rq_pages */
-	rqstp->rq_next_page = rqstp->rq_respages;
-}
-
 /* Prepare the portion of the RPC Reply that will be transmitted
  * via RDMA Send. The RPC-over-RDMA transport header is prepared
  * in sc_sges[0], and the RPC xdr_buf is prepared in following sges.



^ permalink raw reply related	[flat|nested] 5+ messages in thread
* [PATCH 1 4/4] svcrdma: Retain the page backing rq_res.head[0].iov_base
  2021-03-23 20:37 [PATCH 1 0/4] Move post-send page release into svc_sendto Chuck Lever
                   ` (2 preceding siblings ...)
  2021-03-23 20:37 ` [PATCH 1 3/4] svcrdma: Remove unused sc_pages field Chuck Lever
@ 2021-03-23 20:37 ` Chuck Lever
  3 siblings, 0 replies; 5+ messages in thread
From: Chuck Lever @ 2021-03-23 20:37 UTC (permalink / raw)
  To: linux-nfs, linux-rdma

svc_rdma_sendto() now waits for the NIC hardware to finish with
the pages backing rq_res. We still have to release the page array
in some cases, but now it's always safe to immediately re-use the
page backing rq_res's head buffer.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 net/sunrpc/xprtrdma/svc_rdma_sendto.c |    5 +++++
 1 file changed, 5 insertions(+)

diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index f093c9b536ff..056452cabc98 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -957,6 +957,11 @@ int svc_rdma_sendto(struct svc_rqst *rqstp)
 	ret = svc_rdma_send_reply_msg(rdma, sctxt, rctxt, rqstp);
 	if (ret < 0)
 		goto err1;
+
+	/* Prevent svc_xprt_release() from releasing the page backing
+	 * rq_res.head[0].iov_base. It's no longer being accessed by
+	 * the I/O device. */
+	rqstp->rq_respages++;
 	return 0;
 
  err2:



^ permalink raw reply related	[flat|nested] 5+ messages in thread
end of thread, other threads:[~2021-03-23 20:38 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-03-23 20:37 [PATCH 1 0/4] Move post-send page release into svc_sendto Chuck Lever
2021-03-23 20:37 ` [PATCH 1 1/4] svcrdma: Add a "deferred close" helper Chuck Lever
2021-03-23 20:37 ` [PATCH 1 2/4] svcrdma: Normalize Send page handling Chuck Lever
2021-03-23 20:37 ` [PATCH 1 3/4] svcrdma: Remove unused sc_pages field Chuck Lever
2021-03-23 20:37 ` [PATCH 1 4/4] svcrdma: Retain the page backing rq_res.head[0].iov_base Chuck Lever

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.