Re: proper tty-ldisc refcounting (was Re: WARNING at:drivers/char/tty_ldisc.c)

[Sergey Senozhatsky]

On (08/04/09 00:23), Greg KH wrote:
> On Mon, Aug 03, 2009 at 11:19:53PM -0700, Linus Torvalds wrote:
> > 
> > 
> > On Mon, 3 Aug 2009, Greg KH wrote:
> > > 
> > > Oh wait, the original problem, single user mode.  Hm, we need all of
> > > these to fix that problem?  Or just the first one?
> > 
> > Patches 1-3 should fix that one. 4-5 are just cleanups with no semantic 
> > changes.
> 
> Ok, but due to the lateness of the release cycle, is it worth it to add
> those 3 right now?  Or do we just take the BUG_ON() out as it's pretty
> harmless while shutting down in single user mode?
> 
> What do you think?
> 

I don't think that take the BUG_ON out is the best we can do. 

If I understand correctly, this one (tty_ldisc.c:209):
static void tty_ldisc_put(struct tty_ldisc *ld)
{
	[...]
	spin_unlock_irqrestore(&tty_ldisc_lock, flags);
-	WARN_ON(ld->refcount);
	kfree(ld);
}

It's better to leave everything as is in that case and wait 32 (to my mind).

"there may certainly be more tty breakage there too", Linus
and WARN_ON may be quite usefull.

	Sergey

Attachment: signature.asc
Description: Digital signature