Re: [PATCH 1/2] TTY: ldisc, fix open flag handling

[Greg KH <greg@kroah.com>]

On Mon, Nov 29, 2010 at 11:41:45PM +0100, Jiri Slaby wrote:
> On 11/29/2010 10:50 PM, Greg KH wrote:
> > On Thu, Nov 25, 2010 at 12:27:54AM +0100, Jiri Slaby wrote:
> >> When a concrete ldisc open fails in tty_ldisc_open, we forget to clear
> >> TTY_LDISC_OPEN. This causes a false warning on the next ldisc open:
> >> WARNING: at drivers/char/tty_ldisc.c:445 tty_ldisc_open+0x26/0x38()
> >> Hardware name: System Product Name
> >> Modules linked in: ...
> >> Pid: 5251, comm: a.out Tainted: G        W  2.6.32-5-686 #1
> >> Call Trace:
> >>  [<c1030321>] ? warn_slowpath_common+0x5e/0x8a
> >>  [<c1030357>] ? warn_slowpath_null+0xa/0xc
> >>  [<c119311c>] ? tty_ldisc_open+0x26/0x38
> >>  [<c11936c5>] ? tty_set_ldisc+0x218/0x304
> >> ...
> >>
> >> So clear the bit when failing...
> >>
> >> Introduced in c65c9bc3efa (tty: rewrite the ldisc locking) back in
> >> 2.6.31-rc1.
> >>
> >> Signed-off-by: Jiri Slaby <jslaby@suse.cz>
> >> Cc: Alan Cox <alan@linux.intel.com>
> >> Reported-by: Sergey Lapin <slapin@ossfans.org>
> >> Tested-by: Sergey Lapin <slapin@ossfans.org>
> > 
> > Is this still needed, or can I just use your:
> > 	[PATCH v2 1/2] TTY: don't allow reopen when ldisc is changing
> > patch instead?
> 
> This patch is still needed, it fixes a fail path.

Ok, thanks.

> Other than that there are 3 races in 2.6.36, one of them is introduced
> in 2.6.36, the rest in 2.6.32. For each bug there is a single patch I sent:
> * TTY: open/hangup race fixup
>   - introduced in 2.6.36
>   - open vs hangup race
> * TTY: don't allow reopen when ldisc is changing
>   - tiocsetd vs open race
> * TTY: ldisc, fix open flag handling
>   - this you are asking about
>   - when ldisc->open fails, we blow up

Good, I have these three ready to go to Linus now.

> * Char: TTY, restore tty_ldisc_wait_idle
>   - this is in 37-rc2 already
>   - multiple opens followed by tiocsetd blows the machine up
> 
> All of them are stable candidates (but I would give them some time in
> HEAD to see if something breaks, since I opened a can of worms). The
> first one is applicable only to 2.6.36 indeed.

Thanks for the details, much appreciated, as are your patches :)

greg k-h