From 146d8fef9da1ba854de7fa28b9cb9eb147535f88 Mon Sep 17 00:00:00 2001 From: David Howells Date: Sat, 4 Mar 2017 00:01:41 +0000 Subject: rxrpc: Call state should be read with READ_ONCE() under some circumstances The call state may be changed at any time by the data-ready routine in response to received packets, so if the call state is to be read and acted upon several times in a function, READ_ONCE() must be used unless the call state lock is held. Signed-off-by: David Howells Signed-off-by: David S. Miller --- net/rxrpc/input.c | 12 +++++++----- net/rxrpc/recvmsg.c | 4 ++-- net/rxrpc/sendmsg.c | 48 ++++++++++++++++++++++++++++++------------------ 3 files changed, 39 insertions(+), 25 deletions(-) (limited to 'net/rxrpc') diff --git a/net/rxrpc/input.c b/net/rxrpc/input.c index 9f4cfa25af7c..d74921c4969b 100644 --- a/net/rxrpc/input.c +++ b/net/rxrpc/input.c @@ -420,6 +420,7 @@ static void rxrpc_input_data(struct rxrpc_call *call, struct sk_buff *skb, u16 skew) { struct rxrpc_skb_priv *sp = rxrpc_skb(skb); + enum rxrpc_call_state state; unsigned int offset = sizeof(struct rxrpc_wire_header); unsigned int ix; rxrpc_serial_t serial = sp->hdr.serial, ack_serial = 0; @@ -434,14 +435,15 @@ static void rxrpc_input_data(struct rxrpc_call *call, struct sk_buff *skb, _proto("Rx DATA %%%u { #%u f=%02x }", sp->hdr.serial, seq, sp->hdr.flags); - if (call->state >= RXRPC_CALL_COMPLETE) + state = READ_ONCE(call->state); + if (state >= RXRPC_CALL_COMPLETE) return; /* Received data implicitly ACKs all of the request packets we sent * when we're acting as a client. */ - if ((call->state == RXRPC_CALL_CLIENT_SEND_REQUEST || - call->state == RXRPC_CALL_CLIENT_AWAIT_REPLY) && + if ((state == RXRPC_CALL_CLIENT_SEND_REQUEST || + state == RXRPC_CALL_CLIENT_AWAIT_REPLY) && !rxrpc_receiving_reply(call)) return; @@ -799,7 +801,7 @@ static void rxrpc_input_ack(struct rxrpc_call *call, struct sk_buff *skb, return rxrpc_proto_abort("AK0", call, 0); /* Ignore ACKs unless we are or have just been transmitting. */ - switch (call->state) { + switch (READ_ONCE(call->state)) { case RXRPC_CALL_CLIENT_SEND_REQUEST: case RXRPC_CALL_CLIENT_AWAIT_REPLY: case RXRPC_CALL_SERVER_SEND_REPLY: @@ -940,7 +942,7 @@ static void rxrpc_input_call_packet(struct rxrpc_call *call, static void rxrpc_input_implicit_end_call(struct rxrpc_connection *conn, struct rxrpc_call *call) { - switch (call->state) { + switch (READ_ONCE(call->state)) { case RXRPC_CALL_SERVER_AWAIT_ACK: rxrpc_call_completed(call); break; diff --git a/net/rxrpc/recvmsg.c b/net/rxrpc/recvmsg.c index 6491ca46a03f..3e2f1a8e9c5b 100644 --- a/net/rxrpc/recvmsg.c +++ b/net/rxrpc/recvmsg.c @@ -527,7 +527,7 @@ try_again: msg->msg_namelen = len; } - switch (call->state) { + switch (READ_ONCE(call->state)) { case RXRPC_CALL_SERVER_ACCEPTING: ret = rxrpc_recvmsg_new_call(rx, call, msg, flags); break; @@ -640,7 +640,7 @@ int rxrpc_kernel_recv_data(struct socket *sock, struct rxrpc_call *call, mutex_lock(&call->user_mutex); - switch (call->state) { + switch (READ_ONCE(call->state)) { case RXRPC_CALL_CLIENT_RECV_REPLY: case RXRPC_CALL_SERVER_RECV_REQUEST: case RXRPC_CALL_SERVER_ACK_REQUEST: diff --git a/net/rxrpc/sendmsg.c b/net/rxrpc/sendmsg.c index bc2d3dcff9de..47ccfeacc1c6 100644 --- a/net/rxrpc/sendmsg.c +++ b/net/rxrpc/sendmsg.c @@ -488,6 +488,7 @@ rxrpc_new_client_call_for_sendmsg(struct rxrpc_sock *rx, struct msghdr *msg, int rxrpc_do_sendmsg(struct rxrpc_sock *rx, struct msghdr *msg, size_t len) __releases(&rx->sk.sk_lock.slock) { + enum rxrpc_call_state state; enum rxrpc_command cmd; struct rxrpc_call *call; unsigned long user_call_ID = 0; @@ -526,13 +527,17 @@ int rxrpc_do_sendmsg(struct rxrpc_sock *rx, struct msghdr *msg, size_t len) return PTR_ERR(call); /* ... and we have the call lock. */ } else { - ret = -EBUSY; - if (call->state == RXRPC_CALL_UNINITIALISED || - call->state == RXRPC_CALL_CLIENT_AWAIT_CONN || - call->state == RXRPC_CALL_SERVER_PREALLOC || - call->state == RXRPC_CALL_SERVER_SECURING || - call->state == RXRPC_CALL_SERVER_ACCEPTING) + switch (READ_ONCE(call->state)) { + case RXRPC_CALL_UNINITIALISED: + case RXRPC_CALL_CLIENT_AWAIT_CONN: + case RXRPC_CALL_SERVER_PREALLOC: + case RXRPC_CALL_SERVER_SECURING: + case RXRPC_CALL_SERVER_ACCEPTING: + ret = -EBUSY; goto error_release_sock; + default: + break; + } ret = mutex_lock_interruptible(&call->user_mutex); release_sock(&rx->sk); @@ -542,10 +547,11 @@ int rxrpc_do_sendmsg(struct rxrpc_sock *rx, struct msghdr *msg, size_t len) } } + state = READ_ONCE(call->state); _debug("CALL %d USR %lx ST %d on CONN %p", - call->debug_id, call->user_call_ID, call->state, call->conn); + call->debug_id, call->user_call_ID, state, call->conn); - if (call->state >= RXRPC_CALL_COMPLETE) { + if (state >= RXRPC_CALL_COMPLETE) { /* it's too late for this call */ ret = -ESHUTDOWN; } else if (cmd == RXRPC_CMD_SEND_ABORT) { @@ -555,12 +561,12 @@ int rxrpc_do_sendmsg(struct rxrpc_sock *rx, struct msghdr *msg, size_t len) } else if (cmd != RXRPC_CMD_SEND_DATA) { ret = -EINVAL; } else if (rxrpc_is_client_call(call) && - call->state != RXRPC_CALL_CLIENT_SEND_REQUEST) { + state != RXRPC_CALL_CLIENT_SEND_REQUEST) { /* request phase complete for this client call */ ret = -EPROTO; } else if (rxrpc_is_service_call(call) && - call->state != RXRPC_CALL_SERVER_ACK_REQUEST && - call->state != RXRPC_CALL_SERVER_SEND_REPLY) { + state != RXRPC_CALL_SERVER_ACK_REQUEST && + state != RXRPC_CALL_SERVER_SEND_REPLY) { /* Reply phase not begun or not complete for service call. */ ret = -EPROTO; } else { @@ -605,14 +611,20 @@ int rxrpc_kernel_send_data(struct socket *sock, struct rxrpc_call *call, _debug("CALL %d USR %lx ST %d on CONN %p", call->debug_id, call->user_call_ID, call->state, call->conn); - if (call->state >= RXRPC_CALL_COMPLETE) { - ret = -ESHUTDOWN; /* it's too late for this call */ - } else if (call->state != RXRPC_CALL_CLIENT_SEND_REQUEST && - call->state != RXRPC_CALL_SERVER_ACK_REQUEST && - call->state != RXRPC_CALL_SERVER_SEND_REPLY) { - ret = -EPROTO; /* request phase complete for this client call */ - } else { + switch (READ_ONCE(call->state)) { + case RXRPC_CALL_CLIENT_SEND_REQUEST: + case RXRPC_CALL_SERVER_ACK_REQUEST: + case RXRPC_CALL_SERVER_SEND_REPLY: ret = rxrpc_send_data(rxrpc_sk(sock->sk), call, msg, len); + break; + case RXRPC_CALL_COMPLETE: + /* It's too late for this call */ + ret = -ESHUTDOWN; + break; + default: + /* Request phase complete for this client call */ + ret = -EPROTO; + break; } mutex_unlock(&call->user_mutex); -- cgit v1.2.3 From 6fc166d62c6f9f6eda5ea300f671d34e89bdd3c8 Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 9 Mar 2017 08:10:32 +0000 Subject: rxrpc: rxrpc_kernel_send_data() needs to handle failed call better If rxrpc_kernel_send_data() is asked to send data through a call that has already failed (due to a remote abort, received protocol error or network error), then return the associated error code saved in the call rather than ESHUTDOWN. This allows the caller to work out whether to ask for the abort code or not based on this. Signed-off-by: David Howells Signed-off-by: David S. Miller --- net/rxrpc/sendmsg.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'net/rxrpc') diff --git a/net/rxrpc/sendmsg.c b/net/rxrpc/sendmsg.c index 47ccfeacc1c6..97ab214ca411 100644 --- a/net/rxrpc/sendmsg.c +++ b/net/rxrpc/sendmsg.c @@ -618,8 +618,9 @@ int rxrpc_kernel_send_data(struct socket *sock, struct rxrpc_call *call, ret = rxrpc_send_data(rxrpc_sk(sock->sk), call, msg, len); break; case RXRPC_CALL_COMPLETE: - /* It's too late for this call */ - ret = -ESHUTDOWN; + read_lock_bh(&call->state_lock); + ret = -call->error; + read_unlock_bh(&call->state_lock); break; default: /* Request phase complete for this client call */ -- cgit v1.2.3 From 702f2ac87a9a8da23bf8506466bc70175fc970b2 Mon Sep 17 00:00:00 2001 From: David Howells Date: Fri, 10 Mar 2017 07:48:49 +0000 Subject: rxrpc: Wake up the transmitter if Rx window size increases on the peer The RxRPC ACK packet may contain an extension that includes the peer's current Rx window size for this call. We adjust the local Tx window size to match. However, the transmitter can stall if the receive window is reduced to 0 by the peer and then reopened. This is because the normal way that the transmitter is re-energised is by dropping something out of our Tx queue and thus making space. When a single gap is made, the transmitter is woken up. However, because there's nothing in the Tx queue at this point, this doesn't happen. To fix this, perform a wake_up() any time we see the peer's Rx window size increasing. The observable symptom is that calls start failing on ETIMEDOUT and the following: kAFS: SERVER DEAD state=-62 appears in dmesg. Signed-off-by: David Howells Signed-off-by: David S. Miller --- net/rxrpc/input.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) (limited to 'net/rxrpc') diff --git a/net/rxrpc/input.c b/net/rxrpc/input.c index d74921c4969b..18b2ad8be8e2 100644 --- a/net/rxrpc/input.c +++ b/net/rxrpc/input.c @@ -652,6 +652,7 @@ static void rxrpc_input_ackinfo(struct rxrpc_call *call, struct sk_buff *skb, struct rxrpc_skb_priv *sp = rxrpc_skb(skb); struct rxrpc_peer *peer; unsigned int mtu; + bool wake = false; u32 rwind = ntohl(ackinfo->rwind); _proto("Rx ACK %%%u Info { rx=%u max=%u rwin=%u jm=%u }", @@ -659,9 +660,14 @@ static void rxrpc_input_ackinfo(struct rxrpc_call *call, struct sk_buff *skb, ntohl(ackinfo->rxMTU), ntohl(ackinfo->maxMTU), rwind, ntohl(ackinfo->jumbo_max)); - if (rwind > RXRPC_RXTX_BUFF_SIZE - 1) - rwind = RXRPC_RXTX_BUFF_SIZE - 1; - call->tx_winsize = rwind; + if (call->tx_winsize != rwind) { + if (rwind > RXRPC_RXTX_BUFF_SIZE - 1) + rwind = RXRPC_RXTX_BUFF_SIZE - 1; + if (rwind > call->tx_winsize) + wake = true; + call->tx_winsize = rwind; + } + if (call->cong_ssthresh > rwind) call->cong_ssthresh = rwind; @@ -675,6 +681,9 @@ static void rxrpc_input_ackinfo(struct rxrpc_call *call, struct sk_buff *skb, spin_unlock_bh(&peer->lock); _net("Net MTU %u (maxdata %u)", peer->mtu, peer->maxdata); } + + if (wake) + wake_up(&call->waitq); } /* -- cgit v1.2.3