Re: [2.4.19-rc4 and 2.4.19-rc4-mm2] super block list corruption following fill_super returns fail

[Ian Kent <raven@themaw.net>]

On Fri, 2006-11-10 at 16:55 -0800, Andrew Morton wrote:
> On Fri, 10 Nov 2006 20:33:42 +0800
> Ian Kent <raven@themaw.net> wrote:
> 
> > I'm seeing an oops after returning a fail status from the autofs and
> > autofs4 fill_super methods. The scenario is a little contrived but does
> > demonstrate the mount fail case.
> > 
> > get_super+0x78 corresponds to:
> > 
> >                         down_read(&sb->s_umount);
> > ---->                   if (sb->s_root)
> >                                 return sb;
> >                         up_read(&sb->s_umount);
> > 
> > So I believe that, following the fill_super call in get_sb_nodev the
> > super block is freed during the call to deactivate_super but not removed
> > from the supers list.
> > 
> > As far as I can tell I've done the appropriate housekeeping in the
> > autofs[4] fill_super function. In particular, sb->s_root is not set upon
> > mount fail.
> > 
> 
> Yup, sget() adds the superblock to super_blocks and deactivate_super()
> doesn't take it off.
> 
> > Does anyone have any suggestions as to what I might not be doing that I
> > should be doing that is preventing this removal? 
> 
> Well afacit the only piece of code which knows how to remove a superblock
> from the global list is generic_shutdown_super().  So perhaps your
> ->fill_super() implementation is supposed to run generic_shutdown_super()
> if it's about to return an error.

I came to the same conclusion.

There are a couple of ways I could do it but I was hoping to get a
recommendation from someone that is familiar with the way it's supposed
to be done.

I'll have a look at some of the other file systems.

Ian