Re: [PATCH] real_lookup fix

[Maneesh Soni <maneesh@in.ibm.com>]

On Fri, Mar 28, 2003 at 02:51:31AM -0800, Andrew Morton wrote:
> Maneesh Soni <maneesh@in.ibm.com> wrote:
> >
> > Hi Andrew,
> > 
> > Here is a patch to use seqlock for real_lookup race with d_lookup as suggested
> > by Linus. The race condition can result in duplicate dentry when d_lookup
> > fails due concurrent d_move in some unrelated directory. 
> 
> I was not aware of this race.  Could you please explain it in more detail?
> 

Sometime back, Linus has pointed a race regading d_lookup and concurrent
d_move (rename). If lookup moves to a different bucket due to d_move, it may
fail the lookup. rename in the same directory is protected by parent's i_sem
but rename in some unrelated directory on the same hash chain can have
this problem. This can result in real_lookup allocating a new dentry for an
existing one.

Now, similar problem is there with lookup_hash()->cached_lookup(), where
lookup_hash() ends up in allocating a duplicate dentry.

Linus, actually fixed the race in real_lookup using dcache_lock around the
d_lookup call. The patch I posted replaces this with seqlock and also fixes
the cached_lookup() case.

Regards,
Maneesh

-- 
Maneesh Soni
IBM Linux Technology Center, 
IBM India Software Lab, Bangalore.
Phone: +91-80-5044999 email: maneesh@in.ibm.com
http://lse.sourceforge.net/