netdev - Re: [RFC PATCH 0/2] net: macb: Disable TX checksum offloading on all Zynq

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 7 Aug 2018 11:50:39 +0300
From:   Claudiu Beznea <Claudiu.Beznea@...rochip.com>
To:     Harini Katakam <harinik@...inx.com>,
        Jennifer Dahm <jennifer.dahm@...com>
CC:     <netdev@...r.kernel.org>, "David S . Miller" <davem@...emloft.net>,
        Nathan Sullivan <nathan.sullivan@...com>,
        Rafal Ozieblo <rafalo@...ence.com>,
        Harini Katakam <harini.katakam@...inx.com>,
        Nicolas Ferre <nicolas.ferre@...rochip.com>
Subject: Re: [RFC PATCH 0/2] net: macb: Disable TX checksum offloading on all
 Zynq

Hi Harini,

On 01.08.2018 15:53, Harini Katakam wrote:
> Hi Jennifer,
> 
> On Tue, Jun 5, 2018 at 10:21 AM, Harini Katakam <harinik@...inx.com> wrote:
>> Hi Jeniffer,
>>
>> On Mon, Jun 4, 2018 at 8:35 PM, Nicolas Ferre
>> <nicolas.ferre@...rochip.com> wrote:
>>> Jennifer,
>>>
>>> On 25/05/2018 at 23:44, Jennifer Dahm wrote:
>>>>
>>>> During testing, I discovered that the Zynq GEM hardware overwrites all
>>>> outgoing UDP packet checksums, which is illegal in packet forwarding
>>>> cases. This happens both with and without the checksum-zeroing
>>>> behavior  introduced  in  007e4ba3ee137f4700f39aa6dbaf01a71047c5f6
>>>> ("net: macb: initialize checksum when using checksum offloading"). The
>>>> only solution to both the small packet bug and the packet forwarding
>>>> bug that I can find is to disable TX checksum offloading entirely.
>>>
>>>
>>
>> Thanks for the extensive testing.
>> I'll try to reproduce and see if it is something to be fixed in the driver.
>>
>>> Are the bugs listed above present in all revisions of the GEM IP, only for
>>> some revisions?
>>> Is there an errata that describe this issue for the Zynq GEM?
>>
>> @Nicolas, AFAIK, there is no errata for this in either Cadence or
>> Zynq documentation.
> 
> I was unable to reproduce this issue on Zynq.
> Although I do not have HW with two GEM ports,
> I tried by routing one GEM via PL and another via on board RGMII.
> Since there was no specific errata related to this, I also tried on
> subsequent ZynqMP versions with multiple GEM ports but dint find any
> checksum issues. I discussed the same with cadence and they
> tried the test with 2 bytes of UDP payload on the Zynq GEM IP
> version in their regressions and did not hit any issue either.
> 
> I tried to reach out earlier to see if you can share your exact
> application. Could you please let me know if you have any
> further updates?

I manage to reproduce the issue and provide a patch for this (see patch 3/3
from [1]).

[1] https://www.spinics.net/lists/netdev/msg513848.html

> 
> Regards,
> Harini
>