netdev - Re: crash since commit "net/flow_dissector: switch to siphash"

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:   Mon, 18 Nov 2019 01:11:49 +0100
From:   Hauke Mehrtens <hauke@...ke-m.de>
To:     Eric Dumazet <edumazet@...gle.com>,
        Koen Vandeputte <koen.vandeputte@...ntric.com>
Cc:     netdev <netdev@...r.kernel.org>, Greg KH <greg@...ah.com>
Subject: Re: crash since commit "net/flow_dissector: switch to siphash"

On 11/15/19 6:09 PM, Eric Dumazet wrote:
> On Fri, Nov 15, 2019 at 3:30 AM Koen Vandeputte
> <koen.vandeputte@...ntric.com> wrote:
>>
>>
>> On 13.11.19 15:50, Eric Dumazet wrote:
>>> On Wed, Nov 13, 2019 at 3:52 AM Koen Vandeputte
>>> <koen.vandeputte@...ntric.com> wrote:
>>>> Hi Eric,
>>>>
>>>> I'm currently testing kernel 4.14.153 bump for OpenWrt and noticed splat
>>>> below on my testing boards.
>>>> They all reboot continuously nearly immediately when linked.
>>>>
>>>> It feels like it's tied to a commit of yours [1]
>>> Have you tried current upstream kernels ?
>>
>> No.
>>
>> This board is only supported on OpenWrt currently using 4.14 and it's
>> not natively supported by upstream.
>>
>>>
>>> Is is a backport issue, or a bug in original commit ?
>>
>> No idea .. and I'm not profound enough on that part of the code to judge
>> this,
>>
>> which is why I'm consulting your expert opinion.
>>
>>>
>>> Can you give us gcc version ?
>> 7.4.0
>>>
>>> Can you check what SIPHASH_ALIGNMENT value is at compile time ?
>> 8 (exposed it in dmesg on boot)
> 
> Please ask OpenWrt specialists for support.
> 
> The code is probably mishandled by the compiler.
> 
> siphash() is supposed to handle misaligned data just fine, and
> net/core/flow_dissector.c tries hard to align the keys anyway.

Hi Koen,

This is probably related to backports for the wireless driver which uses
its own fq.h and fq_impl.h implementation and this is now conflicting
with Eric's patch which got backported.
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v4.14.154&id=a9de6f42e945cdb24b59c7ab7ebad1eba6cb5875

I saw this as a compile warning when creating a new backports version
and created this patch:
https://www.spinics.net/lists/backports/msg04930.html
This is for the other way around.

I will try to create a patch for OpenWrt backports tomorrow to fix this
problem and I am also planning to create a new backports release in the
next few days.

Hauke