rxrpc: Calls shouldn't hold socket refs

rxrpc calls shouldn't hold refs on the sock struct.  This was done so that
the socket wouldn't go away whilst the call was in progress, such that the
call could reach the socket's queues.

However, we can mark the socket as requiring an RCU release and rely on the
RCU read lock.

To make this work, we do:

 (1) rxrpc_release_call() removes the call's call user ID.  This is now
     only called from socket operations and not from the call processor:

	rxrpc_accept_call() / rxrpc_kernel_accept_call()
	rxrpc_reject_call() / rxrpc_kernel_reject_call()
	rxrpc_kernel_end_call()
	rxrpc_release_calls_on_socket()
	rxrpc_recvmsg()

     Though it is also called in the cleanup path of
     rxrpc_accept_incoming_call() before we assign a user ID.

 (2) Pass the socket pointer into rxrpc_release_call() rather than getting
     it from the call so that we can get rid of uninitialised calls.

 (3) Fix call processor queueing to pass a ref to the work queue and to
     release that ref at the end of the processor function (or to pass it
     back to the work queue if we have to requeue).

 (4) Skip out of the call processor function asap if the call is complete
     and don't requeue it if the call is complete.

 (5) Clean up the call immediately that the refcount reaches 0 rather than
     trying to defer it.  Actual deallocation is deferred to RCU, however.

 (6) Don't hold socket refs for allocated calls.

 (7) Use the RCU read lock when queueing a message on a socket and treat
     the call's socket pointer according to RCU rules and check it for
     NULL.

     We also need to use the RCU read lock when viewing a call through
     procfs.

 (8) Transmit the final ACK/ABORT to a client call in rxrpc_release_call()
     if this hasn't been done yet so that we can then disconnect the call.
     Once the call is disconnected, it won't have any access to the
     connection struct and the UDP socket for the call work processor to be
     able to send the ACK.  Terminal retransmission will be handled by the
     connection processor.

 (9) Release all calls immediately on the closing of a socket rather than
     trying to defer this.  Incomplete calls will be aborted.

The call refcount model is much simplified.  Refs are held on the call by:

 (1) A socket's user ID tree.

 (2) A socket's incoming call secureq and acceptq.

 (3) A kernel service that has a call in progress.

 (4) A queued call work processor.  We have to take care to put any call
     that we failed to queue.

 (5) sk_buffs on a socket's receive queue.  A future patch will get rid of
     this.

Whilst we're at it, we can do:

 (1) Get rid of the RXRPC_CALL_EV_RELEASE event.  Release is now done
     entirely from the socket routines and never from the call's processor.

 (2) Get rid of the RXRPC_CALL_DEAD state.  Calls now end in the
     RXRPC_CALL_COMPLETE state.

 (3) Get rid of the rxrpc_call::destroyer work item.  Calls are now torn
     down when their refcount reaches 0 and then handed over to RCU for
     final cleanup.

 (4) Get rid of the rxrpc_call::deadspan timer.  Calls are cleaned up
     immediately they're finished with and don't hang around.
     Post-completion retransmission is handled by the connection processor
     once the call is disconnected.

 (5) Get rid of the dead call expiry setting as there's no longer a timer
     to set.

 (6) rxrpc_destroy_all_calls() can just check that the call list is empty.

Signed-off-by: David Howells <dhowells@redhat.com>

1 files changed, 88 insertions, 136 deletions

diff --git a/net/rxrpc/call_object.c b/net/rxrpc/call_object.c
index 83019e4..be5733d 100644
--- a/net/rxrpc/call_object.c
+++ b/net/rxrpc/call_object.c
@@ -24,11 +24,6 @@
  */
 unsigned int rxrpc_max_call_lifetime = 60 * HZ;
 
-/*
- * Time till dead call expires after last use (in jiffies).
- */
-unsigned int rxrpc_dead_call_expiry = 2 * HZ;
-
 const char *const rxrpc_call_states[NR__RXRPC_CALL_STATES] = {
 	[RXRPC_CALL_UNINITIALISED]		= "Uninit  ",
 	[RXRPC_CALL_CLIENT_AWAIT_CONN]		= "ClWtConn",
@@ -43,7 +38,6 @@ const char *const rxrpc_call_states[NR__RXRPC_CALL_STATES] = {
 	[RXRPC_CALL_SERVER_SEND_REPLY]		= "SvSndRpl",
 	[RXRPC_CALL_SERVER_AWAIT_ACK]		= "SvAwtACK",
 	[RXRPC_CALL_COMPLETE]			= "Complete",
-	[RXRPC_CALL_DEAD]			= "Dead    ",
 };
 
 const char *const rxrpc_call_completions[NR__RXRPC_CALL_COMPLETIONS] = {
@@ -74,11 +68,10 @@ struct kmem_cache *rxrpc_call_jar;
 LIST_HEAD(rxrpc_calls);
 DEFINE_RWLOCK(rxrpc_call_lock);
 
-static void rxrpc_destroy_call(struct work_struct *work);
 static void rxrpc_call_life_expired(unsigned long _call);
-static void rxrpc_dead_call_expired(unsigned long _call);
 static void rxrpc_ack_time_expired(unsigned long _call);
 static void rxrpc_resend_time_expired(unsigned long _call);
+static void rxrpc_cleanup_call(struct rxrpc_call *call);
 
 /*
  * find an extant server call
@@ -138,13 +131,10 @@ static struct rxrpc_call *rxrpc_alloc_call(gfp_t gfp)
 
 	setup_timer(&call->lifetimer, &rxrpc_call_life_expired,
 		    (unsigned long) call);
-	setup_timer(&call->deadspan, &rxrpc_dead_call_expired,
-		    (unsigned long) call);
 	setup_timer(&call->ack_timer, &rxrpc_ack_time_expired,
 		    (unsigned long) call);
 	setup_timer(&call->resend_timer, &rxrpc_resend_time_expired,
 		    (unsigned long) call);
-	INIT_WORK(&call->destroyer, &rxrpc_destroy_call);
 	INIT_WORK(&call->processor, &rxrpc_process_call);
 	INIT_LIST_HEAD(&call->link);
 	INIT_LIST_HEAD(&call->chan_wait_link);
@@ -185,11 +175,9 @@ static struct rxrpc_call *rxrpc_alloc_client_call(struct rxrpc_sock *rx,
 	if (!call)
 		return ERR_PTR(-ENOMEM);
 	call->state = RXRPC_CALL_CLIENT_AWAIT_CONN;
-
-	sock_hold(&rx->sk);
-	call->socket = rx;
 	call->rx_data_post = 1;
 	call->service_id = srx->srx_service;
+	rcu_assign_pointer(call->socket, rx);
 
 	_leave(" = %p", call);
 	return call;
@@ -244,8 +232,9 @@ struct rxrpc_call *rxrpc_new_client_call(struct rxrpc_sock *rx,
 		return call;
 	}
 
-	trace_rxrpc_call(call, 0, atomic_read(&call->usage), 0, here,
-			 (const void *)user_call_ID);
+	trace_rxrpc_call(call, rxrpc_call_new_client,
+			 atomic_read(&call->usage), 0,
+			 here, (const void *)user_call_ID);
 
 	/* Publish the call, even though it is incompletely set up as yet */
 	call->user_call_ID = user_call_ID;
@@ -295,8 +284,10 @@ error:
 	list_del_init(&call->link);
 	write_unlock_bh(&rxrpc_call_lock);
 
+error_out:
+	__rxrpc_set_call_completion(call, RXRPC_CALL_LOCAL_ERROR,
+				    RX_CALL_DEAD, ret);
 	set_bit(RXRPC_CALL_RELEASED, &call->flags);
-	call->state = RXRPC_CALL_DEAD;
 	rxrpc_put_call(call, rxrpc_call_put);
 	_leave(" = %d", ret);
 	return ERR_PTR(ret);
@@ -308,11 +299,8 @@ error:
 	 */
 found_user_ID_now_present:
 	write_unlock(&rx->call_lock);
-	set_bit(RXRPC_CALL_RELEASED, &call->flags);
-	call->state = RXRPC_CALL_DEAD;
-	rxrpc_put_call(call, rxrpc_call_put);
-	_leave(" = -EEXIST [%p]", call);
-	return ERR_PTR(-EEXIST);
+	ret = -EEXIST;
+	goto error_out;
 }
 
 /*
@@ -340,7 +328,6 @@ struct rxrpc_call *rxrpc_incoming_call(struct rxrpc_sock *rx,
 			 atomic_read(&candidate->usage), 0, here, NULL);
 
 	chan = sp->hdr.cid & RXRPC_CHANNELMASK;
-	candidate->socket	= rx;
 	candidate->conn		= conn;
 	candidate->peer		= conn->params.peer;
 	candidate->cid		= sp->hdr.cid;
@@ -351,6 +338,7 @@ struct rxrpc_call *rxrpc_incoming_call(struct rxrpc_sock *rx,
 	candidate->flags	|= (1 << RXRPC_CALL_IS_SERVICE);
 	if (conn->security_ix > 0)
 		candidate->state = RXRPC_CALL_SERVER_SECURING;
+	rcu_assign_pointer(candidate->socket, rx);
 
 	spin_lock(&conn->channel_lock);
 
@@ -411,7 +399,6 @@ struct rxrpc_call *rxrpc_incoming_call(struct rxrpc_sock *rx,
 	candidate = NULL;
 	conn->channels[chan].call_counter = call_id;
 	rcu_assign_pointer(conn->channels[chan].call, call);
-	sock_hold(&rx->sk);
 	rxrpc_get_connection(conn);
 	rxrpc_get_peer(call->peer);
 	spin_unlock(&conn->channel_lock);
@@ -453,6 +440,39 @@ old_call:
 }
 
 /*
+ * Queue a call's work processor, getting a ref to pass to the work queue.
+ */
+bool rxrpc_queue_call(struct rxrpc_call *call)
+{
+	const void *here = __builtin_return_address(0);
+	int n = __atomic_add_unless(&call->usage, 1, 0);
+	int m = atomic_read(&call->skb_count);
+	if (n == 0)
+		return false;
+	if (rxrpc_queue_work(&call->processor))
+		trace_rxrpc_call(call, rxrpc_call_queued, n + 1, m, here, NULL);
+	else
+		rxrpc_put_call(call, rxrpc_call_put_noqueue);
+	return true;
+}
+
+/*
+ * Queue a call's work processor, passing the callers ref to the work queue.
+ */
+bool __rxrpc_queue_call(struct rxrpc_call *call)
+{
+	const void *here = __builtin_return_address(0);
+	int n = atomic_read(&call->usage);
+	int m = atomic_read(&call->skb_count);
+	ASSERTCMP(n, >=, 1);
+	if (rxrpc_queue_work(&call->processor))
+		trace_rxrpc_call(call, rxrpc_call_queued_ref, n, m, here, NULL);
+	else
+		rxrpc_put_call(call, rxrpc_call_put_noqueue);
+	return true;
+}
+
+/*
  * Note the re-emergence of a call.
  */
 void rxrpc_see_call(struct rxrpc_call *call)
@@ -493,11 +513,8 @@ void rxrpc_get_call_for_skb(struct rxrpc_call *call, struct sk_buff *skb)
 /*
  * detach a call from a socket and set up for release
  */
-void rxrpc_release_call(struct rxrpc_call *call)
+void rxrpc_release_call(struct rxrpc_sock *rx, struct rxrpc_call *call)
 {
-	struct rxrpc_connection *conn = call->conn;
-	struct rxrpc_sock *rx = call->socket;
-
 	_enter("{%d,%d,%d,%d}",
 	       call->debug_id, atomic_read(&call->usage),
 	       atomic_read(&call->ackr_not_idle),
@@ -513,7 +530,7 @@ void rxrpc_release_call(struct rxrpc_call *call)
 	/* dissociate from the socket
 	 * - the socket's ref on the call is passed to the death timer
 	 */
-	_debug("RELEASE CALL %p (%d CONN %p)", call, call->debug_id, conn);
+	_debug("RELEASE CALL %p (%d)", call, call->debug_id);
 
 	if (call->peer) {
 		spin_lock(&call->peer->lock);
@@ -532,20 +549,30 @@ void rxrpc_release_call(struct rxrpc_call *call)
 		rb_erase(&call->sock_node, &rx->calls);
 		memset(&call->sock_node, 0xdd, sizeof(call->sock_node));
 		clear_bit(RXRPC_CALL_HAS_USERID, &call->flags);
+		rxrpc_put_call(call, rxrpc_call_put_userid);
 	}
 	write_unlock_bh(&rx->call_lock);
 
 	/* free up the channel for reuse */
-	write_lock_bh(&call->state_lock);
+	if (call->state == RXRPC_CALL_CLIENT_FINAL_ACK) {
+		clear_bit(RXRPC_CALL_EV_ACK_FINAL, &call->events);
+		rxrpc_send_call_packet(call, RXRPC_PACKET_TYPE_ACK);
+		rxrpc_call_completed(call);
+	} else {
+		write_lock_bh(&call->state_lock);
+
+		if (call->state < RXRPC_CALL_COMPLETE) {
+			_debug("+++ ABORTING STATE %d +++\n", call->state);
+			__rxrpc_abort_call(call, RX_CALL_DEAD, ECONNRESET);
+			clear_bit(RXRPC_CALL_EV_ACK_FINAL, &call->events);
+			rxrpc_send_call_packet(call, RXRPC_PACKET_TYPE_ABORT);
+		}
 
-	if (call->state < RXRPC_CALL_COMPLETE &&
-	    call->state != RXRPC_CALL_CLIENT_FINAL_ACK) {
-		_debug("+++ ABORTING STATE %d +++\n", call->state);
-		__rxrpc_abort_call(call, RX_CALL_DEAD, ECONNRESET);
+		write_unlock_bh(&call->state_lock);
 	}
-	write_unlock_bh(&call->state_lock);
 
-	rxrpc_disconnect_call(call);
+	if (call->conn)
+		rxrpc_disconnect_call(call);
 
 	/* clean up the Rx queue */
 	if (!skb_queue_empty(&call->rx_queue) ||
@@ -569,53 +596,16 @@ void rxrpc_release_call(struct rxrpc_call *call)
 		}
 		spin_unlock_bh(&call->lock);
 	}
+	rxrpc_purge_queue(&call->knlrecv_queue);
 
 	del_timer_sync(&call->resend_timer);
 	del_timer_sync(&call->ack_timer);
 	del_timer_sync(&call->lifetimer);
-	call->deadspan.expires = jiffies + rxrpc_dead_call_expiry;
-	add_timer(&call->deadspan);
 
 	_leave("");
 }
 
 /*
- * handle a dead call being ready for reaping
- */
-static void rxrpc_dead_call_expired(unsigned long _call)
-{
-	struct rxrpc_call *call = (struct rxrpc_call *) _call;
-
-	_enter("{%d}", call->debug_id);
-
-	rxrpc_see_call(call);
-	write_lock_bh(&call->state_lock);
-	call->state = RXRPC_CALL_DEAD;
-	write_unlock_bh(&call->state_lock);
-	rxrpc_put_call(call, rxrpc_call_put);
-}
-
-/*
- * mark a call as to be released, aborting it if it's still in progress
- * - called with softirqs disabled
- */
-static void rxrpc_mark_call_released(struct rxrpc_call *call)
-{
-	bool sched = false;
-
-	rxrpc_see_call(call);
-	write_lock(&call->state_lock);
-	if (call->state < RXRPC_CALL_DEAD) {
-		sched = __rxrpc_abort_call(call, RX_CALL_DEAD, ECONNRESET);
-		if (!test_and_set_bit(RXRPC_CALL_EV_RELEASE, &call->events))
-			sched = true;
-	}
-	write_unlock(&call->state_lock);
-	if (sched)
-		rxrpc_queue_call(call);
-}
-
-/*
  * release all the calls associated with a socket
  */
 void rxrpc_release_calls_on_socket(struct rxrpc_sock *rx)
@@ -629,17 +619,17 @@ void rxrpc_release_calls_on_socket(struct rxrpc_sock *rx)
 
 	/* kill the not-yet-accepted incoming calls */
 	list_for_each_entry(call, &rx->secureq, accept_link) {
-		rxrpc_mark_call_released(call);
+		rxrpc_release_call(rx, call);
 	}
 
 	list_for_each_entry(call, &rx->acceptq, accept_link) {
-		rxrpc_mark_call_released(call);
+		rxrpc_release_call(rx, call);
 	}
 
 	/* mark all the calls as no longer wanting incoming packets */
 	for (p = rb_first(&rx->calls); p; p = rb_next(p)) {
 		call = rb_entry(p, struct rxrpc_call, sock_node);
-		rxrpc_mark_call_released(call);
+		rxrpc_release_call(rx, call);
 	}
 
 	read_unlock_bh(&rx->call_lock);
@@ -663,8 +653,7 @@ void rxrpc_put_call(struct rxrpc_call *call, enum rxrpc_call_trace op)
 	if (n == 0) {
 		_debug("call %d dead", call->debug_id);
 		WARN_ON(m != 0);
-		ASSERTCMP(call->state, ==, RXRPC_CALL_DEAD);
-		rxrpc_queue_work(&call->destroyer);
+		rxrpc_cleanup_call(call);
 	}
 }
 
@@ -683,8 +672,7 @@ void rxrpc_put_call_for_skb(struct rxrpc_call *call, struct sk_buff *skb)
 	if (n == 0) {
 		_debug("call %d dead", call->debug_id);
 		WARN_ON(m != 0);
-		ASSERTCMP(call->state, ==, RXRPC_CALL_DEAD);
-		rxrpc_queue_work(&call->destroyer);
+		rxrpc_cleanup_call(call);
 	}
 }
 
@@ -708,23 +696,19 @@ static void rxrpc_cleanup_call(struct rxrpc_call *call)
 {
 	_net("DESTROY CALL %d", call->debug_id);
 
-	ASSERT(call->socket);
+	write_lock_bh(&rxrpc_call_lock);
+	list_del_init(&call->link);
+	write_unlock_bh(&rxrpc_call_lock);
 
 	memset(&call->sock_node, 0xcd, sizeof(call->sock_node));
 
 	del_timer_sync(&call->lifetimer);
-	del_timer_sync(&call->deadspan);
 	del_timer_sync(&call->ack_timer);
 	del_timer_sync(&call->resend_timer);
 
+	ASSERTCMP(call->state, ==, RXRPC_CALL_COMPLETE);
 	ASSERT(test_bit(RXRPC_CALL_RELEASED, &call->flags));
-	ASSERTCMP(call->events, ==, 0);
-	if (work_pending(&call->processor)) {
-		_debug("defer destroy");
-		rxrpc_queue_work(&call->destroyer);
-		return;
-	}
-
+	ASSERT(!work_pending(&call->processor));
 	ASSERTCMP(call->conn, ==, NULL);
 
 	if (call->acks_window) {
@@ -753,40 +737,21 @@ static void rxrpc_cleanup_call(struct rxrpc_call *call)
 	rxrpc_purge_queue(&call->rx_queue);
 	ASSERT(skb_queue_empty(&call->rx_oos_queue));
 	rxrpc_purge_queue(&call->knlrecv_queue);
-	sock_put(&call->socket->sk);
 	call_rcu(&call->rcu, rxrpc_rcu_destroy_call);
 }
 
 /*
- * destroy a call
- */
-static void rxrpc_destroy_call(struct work_struct *work)
-{
-	struct rxrpc_call *call =
-		container_of(work, struct rxrpc_call, destroyer);
-
-	_enter("%p{%d,%x,%p}",
-	       call, atomic_read(&call->usage), call->cid, call->conn);
-
-	ASSERTCMP(call->state, ==, RXRPC_CALL_DEAD);
-
-	write_lock_bh(&rxrpc_call_lock);
-	list_del_init(&call->link);
-	write_unlock_bh(&rxrpc_call_lock);
-
-	rxrpc_cleanup_call(call);
-	_leave("");
-}
-
-/*
- * preemptively destroy all the call records from a transport endpoint rather
- * than waiting for them to time out
+ * Make sure that all calls are gone.
  */
 void __exit rxrpc_destroy_all_calls(void)
 {
 	struct rxrpc_call *call;
 
 	_enter("");
+
+	if (list_empty(&rxrpc_calls))
+		return;
+	
 	write_lock_bh(&rxrpc_call_lock);
 
 	while (!list_empty(&rxrpc_calls)) {
@@ -796,28 +761,15 @@ void __exit rxrpc_destroy_all_calls(void)
 		rxrpc_see_call(call);
 		list_del_init(&call->link);
 
-		switch (atomic_read(&call->usage)) {
-		case 0:
-			ASSERTCMP(call->state, ==, RXRPC_CALL_DEAD);
-			break;
-		case 1:
-			if (del_timer_sync(&call->deadspan) != 0 &&
-			    call->state != RXRPC_CALL_DEAD)
-				rxrpc_dead_call_expired((unsigned long) call);
-			if (call->state != RXRPC_CALL_DEAD)
-				break;
-		default:
-			pr_err("Call %p still in use (%d,%d,%s,%lx,%lx)!\n",
-			       call, atomic_read(&call->usage),
-			       atomic_read(&call->ackr_not_idle),
-			       rxrpc_call_states[call->state],
-			       call->flags, call->events);
-			if (!skb_queue_empty(&call->rx_queue))
-				pr_err("Rx queue occupied\n");
-			if (!skb_queue_empty(&call->rx_oos_queue))
-				pr_err("OOS queue occupied\n");
-			break;
-		}
+		pr_err("Call %p still in use (%d,%d,%s,%lx,%lx)!\n",
+		       call, atomic_read(&call->usage),
+		       atomic_read(&call->ackr_not_idle),
+		       rxrpc_call_states[call->state],
+		       call->flags, call->events);
+		if (!skb_queue_empty(&call->rx_queue))
+			pr_err("Rx queue occupied\n");
+		if (!skb_queue_empty(&call->rx_oos_queue))
+			pr_err("OOS queue occupied\n");
 
 		write_unlock_bh(&rxrpc_call_lock);
 		cond_resched();