[PATCH] SUNRPC: Ensure the bvecs are reset when we re-encode the RPC request

[PATCH] SUNRPC: Ensure the bvecs are reset when we re-encode the RPC request

[Trond Myklebust]

The bvec tracks the list of pages, so if the number of pages changes
due to a re-encode, we need to reset the bvec as well.

Fixes: 277e4ab7d530 ("SUNRPC: Simplify TCP receive code by switching...")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Cc: stable@vger.kernel.org # v4.20+
---
 net/sunrpc/clnt.c     | 3 +--
 net/sunrpc/xprt.c     | 2 ++
 net/sunrpc/xprtsock.c | 1 +
 3 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index 79c849391cb9..d8679b6027e9 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -1865,6 +1865,7 @@ rpc_xdr_encode(struct rpc_task *task)
 	req->rq_snd_buf.head[0].iov_len = 0;
 	xdr_init_encode(&xdr, &req->rq_snd_buf,
 			req->rq_snd_buf.head[0].iov_base, req);
+	xdr_free_bvec(&req->rq_snd_buf);
 	if (rpc_encode_header(task, &xdr))
 		return;
 
@@ -1904,8 +1905,6 @@ call_encode(struct rpc_task *task)
 			rpc_call_rpcerror(task, task->tk_status);
 		}
 		return;
-	} else {
-		xprt_request_prepare(task->tk_rqstp);
 	}
 
 	/* Add task to reply queue before transmission to avoid races */
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index 70a704c44c6d..783748dc5e6f 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -1041,6 +1041,8 @@ xprt_request_enqueue_receive(struct rpc_task *task)
 
 	if (!xprt_request_need_enqueue_receive(task, req))
 		return;
+
+	xprt_request_prepare(task->tk_rqstp);
 	spin_lock(&xprt->queue_lock);
 
 	/* Update the softirq receive buffer */
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 6b1fca51028a..e2176c167a57 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -909,6 +909,7 @@ static int xs_nospace(struct rpc_rqst *req)
 static void
 xs_stream_prepare_request(struct rpc_rqst *req)
 {
+	xdr_free_bvec(&req->rq_rcv_buf);
 	req->rq_task->tk_status = xdr_alloc_bvec(&req->rq_rcv_buf, GFP_KERNEL);
 }
 
-- 
2.21.0

[PATCH] SUNRPC: Fix initialisation of struct rpc_xprt_switch

[Trond Myklebust]

Ensure that we do initialise the fields xps_nactive, xps_queuelen
and xps_net.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 net/sunrpc/xprtmultipath.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/net/sunrpc/xprtmultipath.c b/net/sunrpc/xprtmultipath.c
index c12778e1235e..f3ecd0bee484 100644
--- a/net/sunrpc/xprtmultipath.c
+++ b/net/sunrpc/xprtmultipath.c
@@ -103,7 +103,9 @@ struct rpc_xprt_switch *xprt_switch_alloc(struct rpc_xprt *xprt,
 	if (xps != NULL) {
 		spin_lock_init(&xps->xps_lock);
 		kref_init(&xps->xps_kref);
-		xps->xps_nxprts = 0;
+		xps->xps_nxprts = xps->xps_nactive = 0;
+		atomic_long_set(&xps->xps_queuelen, 0);
+		xps->xps_net = NULL;
 		INIT_LIST_HEAD(&xps->xps_xprt_list);
 		xps->xps_iter_ops = &rpc_xprt_iter_singular;
 		xprt_switch_add_xprt_locked(xps, xprt);
-- 
2.21.0

[PATCH] SUNRPC: Optimise transport balancing code

[Trond Myklebust]

Moves the balancing code to avoid doing cursor changes on every search
iteration.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 net/sunrpc/xprtmultipath.c | 67 +++++++++++++++++++++-----------------
 1 file changed, 38 insertions(+), 29 deletions(-)

diff --git a/net/sunrpc/xprtmultipath.c b/net/sunrpc/xprtmultipath.c
index f3ecd0bee484..78c075a68c04 100644
--- a/net/sunrpc/xprtmultipath.c
+++ b/net/sunrpc/xprtmultipath.c
@@ -19,7 +19,7 @@
 #include <linux/sunrpc/addr.h>
 #include <linux/sunrpc/xprtmultipath.h>
 
-typedef struct rpc_xprt *(*xprt_switch_find_xprt_t)(struct list_head *head,
+typedef struct rpc_xprt *(*xprt_switch_find_xprt_t)(struct rpc_xprt_switch *xps,
 		const struct rpc_xprt *cur);
 
 static const struct rpc_xprt_iter_ops rpc_xprt_iter_singular;
@@ -291,22 +291,15 @@ struct rpc_xprt *xprt_switch_find_next_entry(struct list_head *head,
 }
 
 static
-struct rpc_xprt *xprt_switch_set_next_cursor(struct list_head *head,
+struct rpc_xprt *xprt_switch_set_next_cursor(struct rpc_xprt_switch *xps,
 		struct rpc_xprt **cursor,
 		xprt_switch_find_xprt_t find_next)
 {
-	struct rpc_xprt *cur, *pos, *old;
+	struct rpc_xprt *pos, *old;
 
-	cur = READ_ONCE(*cursor);
-	for (;;) {
-		old = cur;
-		pos = find_next(head, old);
-		if (pos == NULL)
-			break;
-		cur = cmpxchg_relaxed(cursor, old, pos);
-		if (cur == old)
-			break;
-	}
+	old = smp_load_acquire(cursor);
+	pos = find_next(xps, old);
+	smp_store_release(cursor, pos);
 	return pos;
 }
 
@@ -318,13 +311,11 @@ struct rpc_xprt *xprt_iter_next_entry_multiple(struct rpc_xprt_iter *xpi,
 
 	if (xps == NULL)
 		return NULL;
-	return xprt_switch_set_next_cursor(&xps->xps_xprt_list,
-			&xpi->xpi_cursor,
-			find_next);
+	return xprt_switch_set_next_cursor(xps, &xpi->xpi_cursor, find_next);
 }
 
 static
-struct rpc_xprt *xprt_switch_find_next_entry_roundrobin(struct list_head *head,
+struct rpc_xprt *__xprt_switch_find_next_entry_roundrobin(struct list_head *head,
 		const struct rpc_xprt *cur)
 {
 	struct rpc_xprt *ret;
@@ -336,31 +327,49 @@ struct rpc_xprt *xprt_switch_find_next_entry_roundrobin(struct list_head *head,
 }
 
 static
-struct rpc_xprt *xprt_iter_next_entry_roundrobin(struct rpc_xprt_iter *xpi)
+struct rpc_xprt *xprt_switch_find_next_entry_roundrobin(struct rpc_xprt_switch *xps,
+		const struct rpc_xprt *cur)
 {
-	struct rpc_xprt_switch *xps = rcu_dereference(xpi->xpi_xpswitch);
+	struct list_head *head = &xps->xps_xprt_list;
 	struct rpc_xprt *xprt;
-	unsigned long xprt_queuelen;
-	unsigned long xps_queuelen;
+	unsigned int nactive;
 
-	do {
-		xprt = xprt_iter_next_entry_multiple(xpi,
-			xprt_switch_find_next_entry_roundrobin);
-		if (xprt == NULL)
+	for (;;) {
+		unsigned long xprt_queuelen, xps_queuelen;
+
+		xprt = __xprt_switch_find_next_entry_roundrobin(head, cur);
+		if (!xprt)
 			break;
 		xprt_queuelen = atomic_long_read(&xprt->queuelen);
-		if (xprt_queuelen <= 2)
-			break;
 		xps_queuelen = atomic_long_read(&xps->xps_queuelen);
+		nactive = READ_ONCE(xps->xps_nactive);
 		/* Exit loop if xprt_queuelen <= average queue length */
-	} while (xprt_queuelen * READ_ONCE(xps->xps_nactive) > xps_queuelen);
+		if (xprt_queuelen * nactive <= xps_queuelen)
+			break;
+		cur = xprt;
+	}
 	return xprt;
 }
 
+static
+struct rpc_xprt *xprt_iter_next_entry_roundrobin(struct rpc_xprt_iter *xpi)
+{
+	return xprt_iter_next_entry_multiple(xpi,
+			xprt_switch_find_next_entry_roundrobin);
+}
+
+static
+struct rpc_xprt *xprt_switch_find_next_entry_all(struct rpc_xprt_switch *xps,
+		const struct rpc_xprt *cur)
+{
+	return xprt_switch_find_next_entry(&xps->xps_xprt_list, cur);
+}
+
 static
 struct rpc_xprt *xprt_iter_next_entry_all(struct rpc_xprt_iter *xpi)
 {
-	return xprt_iter_next_entry_multiple(xpi, xprt_switch_find_next_entry);
+	return xprt_iter_next_entry_multiple(xpi,
+			xprt_switch_find_next_entry_all);
 }
 
 /*
-- 
2.21.0