Date: Wed, 02 Jun 2010 14:34:58 +0300 (EEST)
Message-Id: <20100602.143458.232754971.Hiroshi.DOYU@nokia.com>
To: catalin.marinas@arm.com
Cc: linux-kernel@vger.kernel.org, ext-phil.2.carmody@nokia.com,
       linux-omap@vger.kernel.org
Subject: Re: [PATCH v2 0/3] kmemleak: Fix false positive with special scan
From: Hiroshi DOYU <Hiroshi.DOYU@nokia.com>
In-Reply-To: <1275472884.13743.29.camel@e102109-lin.cambridge.arm.com>
References: <1273821401-26578-2-git-send-email-Hiroshi.DOYU@nokia.com>
	<1275387911-13030-1-git-send-email-Hiroshi.DOYU@nokia.com>
	<1275472884.13743.29.camel@e102109-lin.cambridge.arm.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: 1938
Lines: 44

From: ext Catalin Marinas <catalin.marinas@arm.com>
Subject: Re: [PATCH v2 0/3] kmemleak: Fix false positive with special scan
Date: Wed, 2 Jun 2010 12:01:24 +0200

> Hi,
> 
> Sorry for the delay, I eventually got the time to look at your patches.

Thank you for your review.

> On Tue, 2010-06-01 at 11:25 +0100, Hiroshi DOYU wrote:
>> There is a false positive case that a pointer is calculated by other
>> methods than the usual container_of macro. "kmemleak_ignore" can cover
>> such a false positive, but it would loose the advantage of memory leak
>> detection. This patch allows kmemleak to work with such false
>> positives by introducing a new special memory block with a specified
>> calculation formula. A client module can register its area with a
>> conversion function, with which function kmemleak scan could calculate
>> a correct pointer.
> 
> While something needs to be done to cover these situations, I'm not so
> convinced about the method as it complicates the code requiring such
> conversion by having to insert two kmemleak hooks and a callback
> function.
>
> Can we not add a new prio tree (or just use the existing one) for
> pointer aliases? The advantage is that you only have a single function
> to call, something like kmemleak_add_alias() and you do it at the point
> the value was converted.

Actually I considered the above aliasing a little bit but I gave up
soon.

I was afraid that this method might consume way more memory since this
just adds another member for "struct kmemleak_object", but adding a
single member for all objects. The number of kmemleak_object is
usually numerous.

Do you think that this increase of memory consumption is acceptable?
--
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/