V4L/DVB: IR/lirc_dev: fix locking in lirc_dev_fop_read
authorJarod Wilson <jarod@redhat.com>
Wed, 7 Jul 2010 14:29:44 +0000 (11:29 -0300)
committerMauro Carvalho Chehab <mchehab@redhat.com>
Mon, 2 Aug 2010 18:36:23 +0000 (15:36 -0300)
On Wed, Jul 07, 2010 at 02:52:58PM +0200, Jiri Slaby wrote:
> Hi,
>
> stanse found a locking error in lirc_dev_fop_read:
> if (mutex_lock_interruptible(&ir->irctl_lock))
>   return -ERESTARTSYS;
> ...
> while (written < length && ret == 0) {
>   if (mutex_lock_interruptible(&ir->irctl_lock)) {    #1
>     ret = -ERESTARTSYS;
>     break;
>   }
>   ...
> }
>
> remove_wait_queue(&ir->buf->wait_poll, &wait);
> set_current_state(TASK_RUNNING);
> mutex_unlock(&ir->irctl_lock);                        #2
>
> If lock at #1 fails, it beaks out of the loop, with the lock unlocked,
> but there is another "unlock" at #2.

This should do the trick. Completely untested beyond compiling, but its
not exactly a complicated fix, and in practice, I'm not aware of anyone
ever actually tripping that locking bug, so there's zero functional change
in typical use here.

Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
drivers/media/IR/lirc_dev.c

index 9e141d5..64170fa 100644 (file)
@@ -657,7 +657,9 @@ ssize_t lirc_dev_fop_read(struct file *file,
 
                        if (mutex_lock_interruptible(&ir->irctl_lock)) {
                                ret = -ERESTARTSYS;
-                               break;
+                               remove_wait_queue(&ir->buf->wait_poll, &wait);
+                               set_current_state(TASK_RUNNING);
+                               goto out_unlocked;
                        }
 
                        if (!ir->attached) {
@@ -676,6 +678,7 @@ ssize_t lirc_dev_fop_read(struct file *file,
        set_current_state(TASK_RUNNING);
        mutex_unlock(&ir->irctl_lock);
 
+out_unlocked:
        dev_dbg(ir->d.dev, LOGHEAD "read result = %s (%d)\n",
                ir->d.name, ir->d.minor, ret ? "-EFAULT" : "OK", ret);