netdev - Re: [PATCH] net/mlx4_en: fix potential use-after-free with dma_unmap

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]

Message-ID: <c386e907-51fa-f9e3-2207-09e137b8bcfa@mellanox.com>
Date:   Sun, 11 Mar 2018 17:15:34 +0200
From:   Tariq Toukan <tariqt@...lanox.com>
To:     Sarah Newman <srn@...mr.com>, Tariq Toukan <tariqt@...lanox.com>,
        Yishai Hadas <yishaih@...lanox.com>
Cc:     netdev@...r.kernel.org
Subject: Re: [PATCH] net/mlx4_en: fix potential use-after-free with
 dma_unmap_page



On 06/03/2018 10:16 PM, Sarah Newman wrote:
> On 03/06/2018 08:13 AM, Tariq Toukan wrote:
> 
>> I have a general question about the process.
>> I don't totally get what branch this patch is targeted to.
>> It touches critical areas in datapath and should go through regression tests before it is accepted to any branch.
>>
> 
> This one is against 4.9 -
> https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/log/?h=linux-4.9.y
> 
> I assume you'd be interested in applying a fix to all currently maintained stable branches between 3.6 to 4.11, per
> https://www.kernel.org/category/releases.html .
> 

Yes.

> I know of at least one other person whose workaround has been to disable GRO, but I don't understand the networking code well enough to guarantee that
> disabling GRO means the problem will never occur under all possible error conditions. If it's guaranteed that disabling GRO will definitely mitigate
> the bug, probably it's better to stop supporting GRO for these versions of the driver instead of trying to implement some other fix.

I don't think we have a guarantee, it just happens to work.
Also, disabling GRO will cause dramatic performance degradation.
We should fix this, not WA.

> 
> --Sarah
>