Date: Mon, 17 Dec 2007 23:55:17 +0200 (EET)
From: Pekka J Enberg <penberg@cs.helsinki.fi>
To: John Reiser <jreiser@BitWagon.com>
cc: linux-kernel@vger.kernel.org
Subject: Re: slab quirks in DEBUG, ctor, and initialization
In-Reply-To: <4766E7D7.6050003@BitWagon.com>
Message-ID: <Pine.LNX.4.64.0712172343370.22511@sbz-30.cs.Helsinki.FI>
References: <47669A16.3010604@BitWagon.com> <84144f020712171151j5e2a2934qc8bf2fbc9020a545@mail.gmail.com>
 <4766E7D7.6050003@BitWagon.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1009
Lines: 22

Hi John,

On Mon, 17 Dec 2007, John Reiser wrote:
> idr_pre_get calls kmem_cache_alloc, which constructs 'struct idr_layer'
> via the cachep->ctor() call from cache_alloc_debugcheck_after to
> idr_cache_ctor, and not via cache_init_objs.  So if DEBUG is off,
> then idr_cache_ctor does not get its chance to call memset,
> which could leave the struct idr_layer potentially undefined.

No, init_cache_objs() will call the constructor, if the cache has one when 
DEBUG is not set so the struct idr_layer can never be undefined. However, 
struct idr_layer can contain non-zero elements if someone does a 
kmem_cache_free() on an object that hasn't been zeroed. If that matters 
here, idr_pre_get should call kmem_cache_zalloc() and drop the 
constructor.

				Pekka
--
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/