Re: tty OOPS (Re: 2.6.21-rc5-mm2)

[Alexey Dobriyan <adobriyan@gmail.com>]

On Wed, Mar 28, 2007 at 10:38:14PM +0400, Alexey Dobriyan wrote:
> On Wed, Mar 28, 2007 at 08:04:46PM +0200, Andreas Mohr wrote:
> > [unrelated maintainers removed, Alexey added]
> >
> > On Wed, Mar 28, 2007 at 07:45:24PM +0200, Andreas Mohr wrote:
> > > Hi,
> > >
> > > just wanted to add that when analyzing the backtrace I found the comment
> > > at drivers/char/vt.c/con_close() to be VERY suspicious...
> > > (need to take tty_mutex to prevent concurrent thread tty access).
> > > This might just be what happened here despite trying to protect against it.
> >
> > OK, can we assume that one of
> >
> > +protect-tty-drivers-list-with-tty_mutex.patch
> > +tty-minor-merge-correction.patch
> > +tty-in-tiocsctty-when-we-steal-a-tty-hang-it-up-fix.patch
> >
> > is responsible / not implemented fully?
>
> #2 is just comment removal.
>
> I may state the obvious, but __iget() in sysfs_drop_dentry() gets NULL
> inode and you aren't failing on spin_lock one line above because of UP
> without spinlock debugging.

The only suspicious new patch in -rc5-mm1 to me is
fix-sysfs-reclaim-crash.patch which removes "sd->s_dentry = NULL;". Note
that whole sysfs_drop_dentry() is NOP if ->s_dentry is NULL.

Could you try to revert it?

	Alexey, who knows very little about sysfs internals