ipmi: Make the message handler easier to use for SMI interfaces

The message handler expected the SMI interface to keep a queue of
messages, but that was kind of silly, the queue would be easier to
manage in the message handler itself.  As part of that, fix the
message cleanup to make sure no messages are outstanding when an
SMI interface is unregistered.  This makes it easier for an SMI
interface to unregister, it just has to call ipmi_unregister_smi()
first and all processing from the message handler will be cleaned
up.
Signed-off-by: Corey Minyard <cminyard@mvista.com>

drivers/char/ipmi/ipmi_msghandler.c

@@ -56,6 +56,8 @@ static int ipmi_init_msghandler(void);
 static void smi_recv_tasklet(unsigned long);
 static void handle_new_recv_msgs(ipmi_smi_t intf);
 static void need_waiter(ipmi_smi_t intf);
+static int handle_one_recv_msg(ipmi_smi_t          intf,
+			       struct ipmi_smi_msg *msg);
 
 static int initialized;
 
@@ -324,6 +326,9 @@ struct ipmi_smi {
 
 	struct kref refcount;
 
+	/* Set when the interface is being unregistered. */
+	bool in_shutdown;
+
 	/* Used for a list of interfaces. */
 	struct list_head link;
 
@@ -382,6 +387,11 @@ struct ipmi_smi {
 	atomic_t	 watchdog_pretimeouts_to_deliver;
 	struct tasklet_struct recv_tasklet;
 
+	spinlock_t             xmit_msgs_lock;
+	struct list_head       xmit_msgs;
+	struct ipmi_smi_msg    *curr_msg;
+	struct list_head       hp_xmit_msgs;
+
 	/*
 	 * The list of command receivers that are registered for commands
 	 * on this interface.
@@ -1488,6 +1498,24 @@ static inline void format_lan_msg(struct ipmi_smi_msg   *smi_msg,
 static void smi_send(ipmi_smi_t intf, struct ipmi_smi_handlers *handlers,
 		     struct ipmi_smi_msg *smi_msg, int priority)
 {
+	int run_to_completion = intf->run_to_completion;
+	unsigned long flags;
+
+	if (!run_to_completion)
+		spin_lock_irqsave(&intf->xmit_msgs_lock, flags);
+	if (intf->curr_msg) {
+		if (priority > 0)
+			list_add_tail(&smi_msg->link, &intf->hp_xmit_msgs);
+		else
+			list_add_tail(&smi_msg->link, &intf->xmit_msgs);
+		smi_msg = NULL;
+	} else {
+		intf->curr_msg = smi_msg;
+	}
+	if (!run_to_completion)
+		spin_unlock_irqrestore(&intf->xmit_msgs_lock, flags);
+
+	if (smi_msg)
 		handlers->sender(intf->send_info, smi_msg, 0);
 }
 
@@ -1515,7 +1543,6 @@ static int i_ipmi_request(ipmi_user_t          user,
 	struct ipmi_smi_msg      *smi_msg;
 	struct ipmi_recv_msg     *recv_msg;
 	unsigned long            flags;
-	struct ipmi_smi_handlers *handlers;
 
 
 	if (supplied_recv)
@@ -1538,8 +1565,7 @@ static int i_ipmi_request(ipmi_user_t          user,
 	}
 
 	rcu_read_lock();
-	handlers = intf->handlers;
-	if (!handlers) {
+	if (intf->in_shutdown) {
 		rv = -ENODEV;
 		goto out_err;
 	}
@@ -1874,7 +1900,7 @@ static int i_ipmi_request(ipmi_user_t          user,
 	}
 #endif
 
-	smi_send(intf, handlers, smi_msg, priority);
+	smi_send(intf, intf->handlers, smi_msg, priority);
 	rcu_read_unlock();
 
 	return 0;
@@ -2810,6 +2836,9 @@ int ipmi_register_smi(struct ipmi_smi_handlers *handlers,
 		     smi_recv_tasklet,
 		     (unsigned long) intf);
 	atomic_set(&intf->watchdog_pretimeouts_to_deliver, 0);
+	spin_lock_init(&intf->xmit_msgs_lock);
+	INIT_LIST_HEAD(&intf->xmit_msgs);
+	INIT_LIST_HEAD(&intf->hp_xmit_msgs);
 	spin_lock_init(&intf->events_lock);
 	atomic_set(&intf->event_waiters, 0);
 	intf->ticks_to_req_ev = IPMI_REQUEST_EV_TIME;
@@ -2909,12 +2938,50 @@ int ipmi_register_smi(struct ipmi_smi_handlers *handlers,
 }
 EXPORT_SYMBOL(ipmi_register_smi);
 
+static void deliver_smi_err_response(ipmi_smi_t intf,
+				     struct ipmi_smi_msg *msg,
+				     unsigned char err)
+{
+	msg->rsp[0] = msg->data[0] | 4;
+	msg->rsp[1] = msg->data[1];
+	msg->rsp[2] = err;
+	msg->rsp_size = 3;
+	/* It's an error, so it will never requeue, no need to check return. */
+	handle_one_recv_msg(intf, msg);
+}
+
 static void cleanup_smi_msgs(ipmi_smi_t intf)
 {
 	int              i;
 	struct seq_table *ent;
+	struct ipmi_smi_msg *msg;
+	struct list_head *entry;
+	struct list_head tmplist;
+
+	/* Clear out our transmit queues and hold the messages. */
+	INIT_LIST_HEAD(&tmplist);
+	list_splice_tail(&intf->hp_xmit_msgs, &tmplist);
+	list_splice_tail(&intf->xmit_msgs, &tmplist);
+
+	/* Current message first, to preserve order */
+	while (intf->curr_msg && !list_empty(&intf->waiting_rcv_msgs)) {
+		/* Wait for the message to clear out. */
+		schedule_timeout(1);
+	}
 
 	/* No need for locks, the interface is down. */
+
+	/*
+	 * Return errors for all pending messages in queue and in the
+	 * tables waiting for remote responses.
+	 */
+	while (!list_empty(&tmplist)) {
+		entry = tmplist.next;
+		list_del(entry);
+		msg = list_entry(entry, struct ipmi_smi_msg, link);
+		deliver_smi_err_response(intf, msg, IPMI_ERR_UNSPECIFIED);
+	}
+
 	for (i = 0; i < IPMI_IPMB_NUM_SEQ; i++) {
 		ent = &(intf->seq_table[i]);
 		if (!ent->inuse)
@@ -2927,19 +2994,32 @@ int ipmi_unregister_smi(ipmi_smi_t intf)
 {
 	struct ipmi_smi_watcher *w;
 	int intf_num = intf->intf_num;
+	ipmi_user_t user;
 
 	ipmi_bmc_unregister(intf);
 
 	mutex_lock(&smi_watchers_mutex);
 	mutex_lock(&ipmi_interfaces_mutex);
 	intf->intf_num = -1;
-	intf->handlers = NULL;
+	intf->in_shutdown = true;
 	list_del_rcu(&intf->link);
 	mutex_unlock(&ipmi_interfaces_mutex);
 	synchronize_rcu();
 
 	cleanup_smi_msgs(intf);
 
+	/* Clean up the effects of users on the lower-level software. */
+	mutex_lock(&ipmi_interfaces_mutex);
+	rcu_read_lock();
+	list_for_each_entry_rcu(user, &intf->users, link) {
+		module_put(intf->handlers->owner);
+		if (intf->handlers->dec_usecount)
+			intf->handlers->dec_usecount(intf->send_info);
+	}
+	rcu_read_unlock();
+	intf->handlers = NULL;
+	mutex_unlock(&ipmi_interfaces_mutex);
+
 	remove_proc_entries(intf);
 
 	/*
@@ -3029,7 +3109,6 @@ static int handle_ipmb_get_msg_cmd(ipmi_smi_t          intf,
 	ipmi_user_t              user = NULL;
 	struct ipmi_ipmb_addr    *ipmb_addr;
 	struct ipmi_recv_msg     *recv_msg;
-	struct ipmi_smi_handlers *handlers;
 
 	if (msg->rsp_size < 10) {
 		/* Message not big enough, just ignore it. */
@@ -3083,9 +3162,8 @@ static int handle_ipmb_get_msg_cmd(ipmi_smi_t          intf,
 	}
 #endif
 		rcu_read_lock();
-		handlers = intf->handlers;
-		if (handlers) {
-			smi_send(intf, handlers, msg, 0);
+		if (!intf->in_shutdown) {
+			smi_send(intf, intf->handlers, msg, 0);
 			/*
 			 * We used the message, so return the value
 			 * that causes it to not be freed or
@@ -3756,25 +3834,24 @@ static void handle_new_recv_msgs(ipmi_smi_t intf)
 	while (!list_empty(&intf->waiting_rcv_msgs)) {
 		smi_msg = list_entry(intf->waiting_rcv_msgs.next,
 				     struct ipmi_smi_msg, link);
-		list_del(&smi_msg->link);
 		if (!run_to_completion)
 			spin_unlock_irqrestore(&intf->waiting_rcv_msgs_lock,
 					       flags);
 		rv = handle_one_recv_msg(intf, smi_msg);
 		if (!run_to_completion)
 			spin_lock_irqsave(&intf->waiting_rcv_msgs_lock, flags);
-		if (rv == 0) {
-			/* Message handled */
-			ipmi_free_smi_msg(smi_msg);
-		} else if (rv < 0) {
-			/* Fatal error on the message, del but don't free. */
-		} else {
+		if (rv > 0) {
 			/*
 			 * To preserve message order, quit if we
 			 * can't handle a message.
 			 */
-			list_add(&smi_msg->link, &intf->waiting_rcv_msgs);
 			break;
+		} else {
+			list_del(&smi_msg->link);
+			if (rv == 0)
+				/* Message handled */
+				ipmi_free_smi_msg(smi_msg);
+			/* If rv < 0, fatal error, del but don't free. */
 		}
 	}
 	if (!run_to_completion)
@@ -3799,7 +3876,41 @@ static void handle_new_recv_msgs(ipmi_smi_t intf)
 
 static void smi_recv_tasklet(unsigned long val)
 {
-	handle_new_recv_msgs((ipmi_smi_t) val);
+	unsigned long flags = 0; /* keep us warning-free. */
+	ipmi_smi_t intf = (ipmi_smi_t) val;
+	int run_to_completion = intf->run_to_completion;
+	struct ipmi_smi_msg *newmsg = NULL;
+
+	/*
+	 * Start the next message if available.
+	 *
+	 * Do this here, not in the actual receiver, because we may deadlock
+	 * because the lower layer is allowed to hold locks while calling
+	 * message delivery.
+	 */
+	if (!run_to_completion)
+		spin_lock_irqsave(&intf->xmit_msgs_lock, flags);
+	if (intf->curr_msg == NULL && !intf->in_shutdown) {
+		struct list_head *entry = NULL;
+
+		/* Pick the high priority queue first. */
+		if (!list_empty(&intf->hp_xmit_msgs))
+			entry = intf->hp_xmit_msgs.next;
+		else if (!list_empty(&intf->xmit_msgs))
+			entry = intf->xmit_msgs.next;
+
+		if (entry) {
+			list_del(entry);
+			newmsg = list_entry(entry, struct ipmi_smi_msg, link);
+			intf->curr_msg = newmsg;
+		}
+	}
+	if (!run_to_completion)
+		spin_unlock_irqrestore(&intf->xmit_msgs_lock, flags);
+	if (newmsg)
+		intf->handlers->sender(intf->send_info, newmsg, 0);
+
+	handle_new_recv_msgs(intf);
 }
 
 /* Handle a new message from the lower layer. */
@@ -3807,13 +3918,16 @@ void ipmi_smi_msg_received(ipmi_smi_t          intf,
 			   struct ipmi_smi_msg *msg)
 {
 	unsigned long flags = 0; /* keep us warning-free. */
-	int           run_to_completion;
-
+	int run_to_completion = intf->run_to_completion;
 
 	if ((msg->data_size >= 2)
 	    && (msg->data[0] == (IPMI_NETFN_APP_REQUEST << 2))
 	    && (msg->data[1] == IPMI_SEND_MSG_CMD)
 	    && (msg->user_data == NULL)) {
+
+		if (intf->in_shutdown)
+			goto free_msg;
+
 		/*
 		 * This is the local response to a command send, start
 		 * the timer for these.  The user_data will not be
@@ -3849,29 +3963,40 @@ void ipmi_smi_msg_received(ipmi_smi_t          intf,
 			/* The message was sent, start the timer. */
 			intf_start_seq_timer(intf, msg->msgid);
 
+free_msg:
 		ipmi_free_smi_msg(msg);
-		goto out;
-	}
-
+	} else {
 		/*
-	 * To preserve message order, if the list is not empty, we
-	 * tack this message onto the end of the list.
+		 * To preserve message order, we keep a queue and deliver from
+		 * a tasklet.
 		 */
-	run_to_completion = intf->run_to_completion;
 		if (!run_to_completion)
 			spin_lock_irqsave(&intf->waiting_rcv_msgs_lock, flags);
 		list_add_tail(&msg->link, &intf->waiting_rcv_msgs);
 		if (!run_to_completion)
-		spin_unlock_irqrestore(&intf->waiting_rcv_msgs_lock, flags);
+			spin_unlock_irqrestore(&intf->waiting_rcv_msgs_lock,
+					       flags);
+	}
+
+	if (!run_to_completion)
+		spin_lock_irqsave(&intf->xmit_msgs_lock, flags);
+	if (msg == intf->curr_msg)
+		intf->curr_msg = NULL;
+	if (!run_to_completion)
+		spin_unlock_irqrestore(&intf->xmit_msgs_lock, flags);
 
+	if (run_to_completion)
+		smi_recv_tasklet((unsigned long) intf);
+	else
 		tasklet_schedule(&intf->recv_tasklet);
- out:
-	return;
 }
 EXPORT_SYMBOL(ipmi_smi_msg_received);
 
 void ipmi_smi_watchdog_pretimeout(ipmi_smi_t intf)
 {
+	if (intf->in_shutdown)
+		return;
+
 	atomic_set(&intf->watchdog_pretimeouts_to_deliver, 1);
 	tasklet_schedule(&intf->recv_tasklet);
 }
@@ -3913,7 +4038,7 @@ static void check_msg_timeout(ipmi_smi_t intf, struct seq_table *ent,
 	struct ipmi_recv_msg     *msg;
 	struct ipmi_smi_handlers *handlers;
 
-	if (intf->intf_num == -1)
+	if (intf->in_shutdown)
 		return;
 
 	if (!ent->inuse)
@@ -4040,15 +4165,12 @@ static unsigned int ipmi_timeout_handler(ipmi_smi_t intf, long timeout_period)
 
 static void ipmi_request_event(ipmi_smi_t intf)
 {
-	struct ipmi_smi_handlers *handlers;
-
 	/* No event requests when in maintenance mode. */
 	if (intf->maintenance_mode_enable)
 		return;
 
-	handlers = intf->handlers;
-	if (handlers)
-		handlers->request_events(intf->send_info);
+	if (!intf->in_shutdown)
+		intf->handlers->request_events(intf->send_info);
 }
 
 static struct timer_list ipmi_timer;