linux-kernel - Re: [PATCH 5/6] IMA: use rbtree instead of radix tree for inode information cache

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 21 Oct 2010 09:05:57 +1100
From:	Dave Chinner <david@...morbit.com>
To:	Peter Zijlstra <peterz@...radead.org>
Cc:	Eric Paris <eparis@...hat.com>, linux-kernel@...r.kernel.org,
	linux-security-module@...r.kernel.org,
	linux-fsdevel@...r.kernel.org, hch@...radead.org, zohar@...ibm.com,
	warthog9@...nel.org, jmorris@...ei.org, kyle@...artin.ca,
	hpa@...or.com, akpm@...ux-foundation.org,
	torvalds@...ux-foundation.org, mingo@...e.hu,
	viro@...iv.linux.org.uk
Subject: Re: [PATCH 5/6] IMA: use rbtree instead of radix tree for inode
 information cache

On Wed, Oct 20, 2010 at 01:31:09PM +0200, Peter Zijlstra wrote:
> On Wed, 2010-10-20 at 10:17 +1100, Dave Chinner wrote:
> > On Tue, Oct 19, 2010 at 06:58:39PM -0400, Eric Paris wrote:
> > > @@ -36,12 +63,11 @@ struct ima_iint_cache *ima_iint_find_get(struct inode *inode)
> > >  	struct ima_iint_cache *iint;
> > >  
> > >  	rcu_read_lock();
> > > -	iint = radix_tree_lookup(&ima_iint_store, (unsigned long)inode);
> > > -	if (!iint)
> > > -		goto out;
> > > -	kref_get(&iint->refcount);
> > > -out:
> > > +	iint = __ima_iint_find(inode);
> > > +	if (iint)
> > > +		kref_get(&iint->refcount);
> > >  	rcu_read_unlock();
> > > +
> > 
> > This is wrong - the rbtree is protected only by the ima_iint_lock(),
> > not RCU. Hence you can't do lockless lookups on an rbtree in this
> > manner as they will race with inserts and deletes.
> 
> Correct, what can be made to work is combine RCU with a seqlock. Retry
> the lookup using read_seqretry(), RCU here helps to ensure you're not
> stepping on already freed memory.
> 
> 
> So, tree modification does:
> 
>   write_seqlock();
>   /* frob RB-tree, using call_rcu() for frees where needed */
>   write_sequnlock();
> 
> Lookup does:
> 
>   unsigned seq;
> 
>   rcu_read_lock()
> again;
>   seq = read_seqbegin();
> 
>   /* RB-tree lookup */
> 
>   if (read_seqretry(seq))
>     goto again;
> 
>   rcu_read_unlock();
> 
>   return obj;

Nice trick, Peter. Thanks for sharing that - I'll definitely find
that useful. :)

/me wanders off to look at converting the xfs buffer cache rbtrees
to RCU....

Cheers,

Dave.
-- 
Dave Chinner
david@...morbit.com
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/