linux-kernel - Re: [PATCH] drm/exynos: Fix memory leak and release IOMMU mapping structures

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, 5 Mar 2020 10:05:03 +0000
From:   Lukasz Luba <lukasz.luba@....com>
To:     Marek Szyprowski <m.szyprowski@...sung.com>,
        linux-kernel@...r.kernel.org, dri-devel@...ts.freedesktop.org,
        linux-arm-kernel@...ts.infradead.org,
        linux-samsung-soc@...r.kernel.org
Cc:     inki.dae@...sung.com, jy0922.shim@...sung.com,
        sw0312.kim@...sung.com, kyungmin.park@...sung.com,
        airlied@...ux.ie, daniel@...ll.ch, kgene@...nel.org,
        krzk@...nel.org, b.zolnierkie@...sung.com, a.hajda@...sung.com,
        Dietmar.Eggemann@....com
Subject: Re: [PATCH] drm/exynos: Fix memory leak and release IOMMU mapping
 structures

Hi Marek,

On 3/5/20 7:07 AM, Marek Szyprowski wrote:
> Hi Lukasz,
> 
> On 04.03.2020 23:00, Lukasz Luba wrote:
>> There is a memory leak which left some objects not freed. The reference
>> counter of mapping: 'mapping->kref' was 2 when calling
>> arm_iommu_detach_device(), so the release_iommu_mapping() won't be called.
>> Since the old mapping structure is not going to be used any more (because
>> it is detached and new one attached), call arm_iommu_release_mapping()
>> to trigger cleanup.
> 
> This will break IOMMU support in Exynos DRM if deferred probe happens.
> Here is a proper fix:

I forgot about the deferred probe.

> 
> https://patchwork.kernel.org/patch/11415715/
> 
> The mapping initially created by DMA-mapping framework should be
> attached back when Exynos DRM releases the subdev device.
> 

Indeed, as you responded in that thread with the example, there is
more dependencies and attaching back the old mapping will work.

I am going add my reviewed-by to your patch.

Regards,
Lukasz