Re: [2.6.36-rc1] List corruption when using initrd.

[Hugh Dickins <hughd@google.com>]

On Wed, 18 Aug 2010, Tetsuo Handa wrote:
> Hugh Dickins wrote:
> > Thanks for reporting and bisecting.  Certainly there's a bug in shmem
> > (that we ought to have caught long before it reached 36-rc1: sorry),
> > and it is probably the cause of your crashes; but it's possible there's
> > a similar bug elsewhere too, something else messing up the percpu_counters
> > list - so please check if the patch below really does fix it for you.
> 
> This patch solved the list_add() corruption messages and

Great.

> "[2.6.36-rc1] Bugs in __free_pipe_info()".

That's good, but puzzling.  I'll mention it in the patch comment, since
that's what you've found: but I've given up trying to understand how.

> 
> > I notice your dmesg also showed fs/inode.c:692 unlock_new_inode()
> > warnings from get_cramfs_inode(): those would be unrelated, and
> > probably from Al Viro's recent changes in cramfs/inode.c - Cc'ed.
> unlock_new_inode() warnings remain after applying this patch.
> 
> Thank you.

Bigger thanks to you.  Patch to Linus follows.

Hugh