Re: [PATCH] Fix open/close race in saa7134

[Marcin Slusarz <marcin.slusarz@gmail.com>]

On Mon, Jun 23, 2008 at 12:53:48PM -0700, Arjan van de Ven wrote:
> On Mon, 23 Jun 2008 21:22:03 +0200
> Marcin Slusarz <marcin.slusarz@gmail.com> wrote:
> 
> > 
> > If dev->empress_users could be > 1 then ok - it could break, but it
> > can only be 1 or 0. If it's 1 you won't open the device. If it's 0
> > you won't reach ts_close.
> > 
> > If you still see the race, please show me the sequence, because I
> > don't (of course when decrementing is the last operation of ts_close).
> 
> a decrement in C, without locking, is NOT atomic.

I know about it. But in this code, 2 threads cannot modify empress_users!
This variable should be named empress_used_now or something like that.

Look:

static int ts_open(struct inode *inode, struct file *file)
{
	(...)
	err = -EBUSY;
	if (!mutex_trylock(&dev->empress_tsq.vb_lock))
		goto done;
	if (dev->empress_users)                 <-------------
		goto done_up;

	/* Unmute audio */
	saa_writeb(SAA7134_AUDIO_MUTE_CTRL,
		saa_readb(SAA7134_AUDIO_MUTE_CTRL) & ~(1 << 6));

	dev->empress_users++;    <------ this should be "dev->empress_users = 1;"
	file->private_data = dev;
	err = 0;

done_up:
	mutex_unlock(&dev->empress_tsq.vb_lock);
done:
	return err;
}

static int ts_release(struct inode *inode, struct file *file)
{
	(...)
	dev->empress_users--;   <------------ this should be "dev->empress_users = 0;"
	return 0;
}