aboutsummaryrefslogtreecommitdiffstats
path: root/fs/ocfs2
diff options
context:
space:
mode:
authorZhen Wei <zwei@novell.com>2007-01-23 20:19:59 -0500
committerMark Fasheh <mark.fasheh@oracle.com>2007-02-07 15:15:11 -0500
commit925037bcba7691db2403684141a276930ad184f3 (patch)
tree5a928f3d3f8488d1094a4ced8f39228c9d5a8ca9 /fs/ocfs2
parentf71aa8a55a0ae1a0d06c6079265d16502a678e8e (diff)
ocfs2: introduce sc->sc_send_lock to protect outbound outbound messages
When there is a lot of multithreaded I/O usage, two threads can collide while sending out a message to the other nodes. This is due to the lack of locking between threads while sending out the messages. When a connected TCP send(), sendto(), or sendmsg() arrives in the Linux kernel, it eventually comes through tcp_sendmsg(). tcp_sendmsg() protects itself by acquiring a lock at invocation by calling lock_sock(). tcp_sendmsg() then loops over the buffers in the iovec, allocating associated sk_buff's and cache pages for use in the actual send. As it does so, it pushes the data out to tcp for actual transmission. However, if one of those allocation fails (because a large number of large sends is being processed, for example), it must wait for memory to become available. It does so by jumping to wait_for_sndbuf or wait_for_memory, both of which eventually cause a call to sk_stream_wait_memory(). sk_stream_wait_memory() contains a code path that calls sk_wait_event(). Finally, sk_wait_event() contains the call to release_sock(). The following patch adds a lock to the socket container in order to properly serialize outbound requests. From: Zhen Wei <zwei@novell.com> Acked-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Diffstat (limited to 'fs/ocfs2')
-rw-r--r--fs/ocfs2/cluster/tcp.c8
-rw-r--r--fs/ocfs2/cluster/tcp_internal.h2
2 files changed, 10 insertions, 0 deletions
diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
index 2021aec7cbbd..1718215fc018 100644
--- a/fs/ocfs2/cluster/tcp.c
+++ b/fs/ocfs2/cluster/tcp.c
@@ -556,6 +556,8 @@ static void o2net_register_callbacks(struct sock *sk,
556 sk->sk_data_ready = o2net_data_ready; 556 sk->sk_data_ready = o2net_data_ready;
557 sk->sk_state_change = o2net_state_change; 557 sk->sk_state_change = o2net_state_change;
558 558
559 mutex_init(&sc->sc_send_lock);
560
559 write_unlock_bh(&sk->sk_callback_lock); 561 write_unlock_bh(&sk->sk_callback_lock);
560} 562}
561 563
@@ -858,10 +860,12 @@ static void o2net_sendpage(struct o2net_sock_container *sc,
858 ssize_t ret; 860 ssize_t ret;
859 861
860 862
863 mutex_lock(&sc->sc_send_lock);
861 ret = sc->sc_sock->ops->sendpage(sc->sc_sock, 864 ret = sc->sc_sock->ops->sendpage(sc->sc_sock,
862 virt_to_page(kmalloced_virt), 865 virt_to_page(kmalloced_virt),
863 (long)kmalloced_virt & ~PAGE_MASK, 866 (long)kmalloced_virt & ~PAGE_MASK,
864 size, MSG_DONTWAIT); 867 size, MSG_DONTWAIT);
868 mutex_unlock(&sc->sc_send_lock);
865 if (ret != size) { 869 if (ret != size) {
866 mlog(ML_ERROR, "sendpage of size %zu to " SC_NODEF_FMT 870 mlog(ML_ERROR, "sendpage of size %zu to " SC_NODEF_FMT
867 " failed with %zd\n", size, SC_NODEF_ARGS(sc), ret); 871 " failed with %zd\n", size, SC_NODEF_ARGS(sc), ret);
@@ -976,8 +980,10 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *caller_vec,
976 980
977 /* finally, convert the message header to network byte-order 981 /* finally, convert the message header to network byte-order
978 * and send */ 982 * and send */
983 mutex_lock(&sc->sc_send_lock);
979 ret = o2net_send_tcp_msg(sc->sc_sock, vec, veclen, 984 ret = o2net_send_tcp_msg(sc->sc_sock, vec, veclen,
980 sizeof(struct o2net_msg) + caller_bytes); 985 sizeof(struct o2net_msg) + caller_bytes);
986 mutex_unlock(&sc->sc_send_lock);
981 msglog(msg, "sending returned %d\n", ret); 987 msglog(msg, "sending returned %d\n", ret);
982 if (ret < 0) { 988 if (ret < 0) {
983 mlog(0, "error returned from o2net_send_tcp_msg=%d\n", ret); 989 mlog(0, "error returned from o2net_send_tcp_msg=%d\n", ret);
@@ -1109,8 +1115,10 @@ static int o2net_process_message(struct o2net_sock_container *sc,
1109 1115
1110out_respond: 1116out_respond:
1111 /* this destroys the hdr, so don't use it after this */ 1117 /* this destroys the hdr, so don't use it after this */
1118 mutex_lock(&sc->sc_send_lock);
1112 ret = o2net_send_status_magic(sc->sc_sock, hdr, syserr, 1119 ret = o2net_send_status_magic(sc->sc_sock, hdr, syserr,
1113 handler_status); 1120 handler_status);
1121 mutex_unlock(&sc->sc_send_lock);
1114 hdr = NULL; 1122 hdr = NULL;
1115 mlog(0, "sending handler status %d, syserr %d returned %d\n", 1123 mlog(0, "sending handler status %d, syserr %d returned %d\n",
1116 handler_status, syserr, ret); 1124 handler_status, syserr, ret);
diff --git a/fs/ocfs2/cluster/tcp_internal.h b/fs/ocfs2/cluster/tcp_internal.h
index 177927a8f007..4dae5df5e467 100644
--- a/fs/ocfs2/cluster/tcp_internal.h
+++ b/fs/ocfs2/cluster/tcp_internal.h
@@ -155,6 +155,8 @@ struct o2net_sock_container {
155 struct timeval sc_tv_func_stop; 155 struct timeval sc_tv_func_stop;
156 u32 sc_msg_key; 156 u32 sc_msg_key;
157 u16 sc_msg_type; 157 u16 sc_msg_type;
158
159 struct mutex sc_send_lock;
158}; 160};
159 161
160struct o2net_msg_handler { 162struct o2net_msg_handler {