Date: Sat, 6 Jun 2009 23:29:45 +0100 (BST)
From: Hugh Dickins <hugh.dickins@tiscali.co.uk>
To: Linus Torvalds <torvalds@linux-foundation.org>
cc: Mimi Zohar <zohar@us.ibm.com>, Andrew Morton <akpm@linux-foundation.org>,
       Mimi Zohar <zohar@linux.vnet.ibm.com>, Serge Hallyn <serue@us.ibm.com>,
       James Morris <jmorris@namei.org>, Al Viro <viro@zeniv.linux.org.uk>,
       linux-kernel@vger.kernel.org
Subject: Re: [PATCH] integrity: fix IMA inode leak
In-Reply-To: <alpine.LFD.2.01.0906061433490.6847@localhost.localdomain>
Message-ID: <Pine.LNX.4.64.0906062312310.12529@sister.anvils>
References: <Pine.LNX.4.64.0906062109580.8365@sister.anvils>
 <alpine.LFD.2.01.0906061416170.6847@localhost.localdomain>
 <alpine.LFD.2.01.0906061433490.6847@localhost.localdomain>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1898
Lines: 45

On Sat, 6 Jun 2009, Linus Torvalds wrote:
> On Sat, 6 Jun 2009, Linus Torvalds wrote:
> > On Sat, 6 Jun 2009, Hugh Dickins wrote:
> > >
> > > CONFIG_IMA=y inode activity leaks iint_cache and radix_tree_node objects
> > > until the system runs out of memory.  Nowhere is calling ima_inode_free()
> > > a.k.a. ima_iint_delete().  Fix that by calling it from destroy_inode().
> > 
> > Shouldn't we call it from "security_inode_free()" instead? And shouldn't 
> > it be allocated in "security_inode_alloc()"? That sounds like the correct 
> > nesting here, since the whole integrity thing is under the security 
> > module.
> > 
> > Hmm?

I wondered the same: quite possibly, but I tend to assume there was reason
to do it like that; and thought it best to mirror the security_inode_alloc,
ima_inode_alloc with ima_inode_free, security_inode_free for now.

(I also disliked the obfuscescent #define ima_iint_delete ima_inode_free
in ima_iint.c!)

> 
> Oh well. I applied the patch as-is, since it seems to fix a real issue. 

Thanks, yes.  There is a possibility that it will reveal some warnings
from iint_free which were hidden before; but I've not seen them in normal
working, and I'd anyway prefer a few warnings to my boxes OOMing.  Though
it was only by mistake that I had CONFIG_IMA=y in the first place.

> 
> But I do think fs/inode.c shouldn't care about things like that, and have 
> it internal to security_inode_alloc/free(). But I guess that's a separate 
> cleanup.

The IMA stuff all looks rather bolted in on top; I did fail to persuade
Mimi to move the shmem and shm hooks down a level, so for now at least
they'll stay as they are.

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