Input: serio/gameport - use 'long' system workqueue
authorDmitry Torokhov <dmitry.torokhov@gmail.com>
Wed, 23 Feb 2011 16:51:28 +0000 (08:51 -0800)
committerDmitry Torokhov <dmitry.torokhov@gmail.com>
Wed, 23 Feb 2011 16:53:07 +0000 (08:53 -0800)
Commit 8ee294cd9def0004887da7f44b80563493b0a097 converted serio
subsystem event handling from using a dedicated thread to using
common workqueue. Unfortunately, this regressed our boot times,
due to the fact that serio jobs take long time to execute. While
the new concurrency managed workqueue code manages long-playing
works just fine and schedules additional workers as needed, such
works wreck havoc among remaining users of flush_scheduled_work().

To solve this problem let's move serio/gameport works from system_wq
to system_long_wq which nobody tries to flush.

Reported-and-tested-by: Hernando Torque <pantherchen@versanet.de>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
drivers/input/gameport/gameport.c
drivers/input/serio/serio.c

index dbf741c95835f953c8d6d554f7f087286f3c9381..ce57ede6e981e85aeaeb2c84b1a2cb0b1a0ddf96 100644 (file)
@@ -360,7 +360,7 @@ static int gameport_queue_event(void *object, struct module *owner,
        event->owner = owner;
 
        list_add_tail(&event->node, &gameport_event_list);
-       schedule_work(&gameport_event_work);
+       queue_work(system_long_wq, &gameport_event_work);
 
 out:
        spin_unlock_irqrestore(&gameport_event_lock, flags);
index 7c38d1fbabf2339afeedd1ab2d21ec909de61621..ba70058e2be3ae2bc6a3289c08d541f336a4aad8 100644 (file)
@@ -299,7 +299,7 @@ static int serio_queue_event(void *object, struct module *owner,
        event->owner = owner;
 
        list_add_tail(&event->node, &serio_event_list);
-       schedule_work(&serio_event_work);
+       queue_work(system_long_wq, &serio_event_work);
 
 out:
        spin_unlock_irqrestore(&serio_event_lock, flags);