--- Begin Message ---
- To: Tzafrir Cohen <tzafrir@cohens.org.il>, 949863@bugs.debian.org
- Cc: Ben Hutchings <benh@debian.org>, krishna.ku@flipkart.com, ashruth@mellanox.com
- Subject: Re: Bug#949863: Info received (#949863: please enable CONFIG_NET_SWITCHDEV)
- From: Vasudev Kamath <vasudev@debian.org>
- Date: Tue, 30 Jun 2020 21:28:59 +0530
- Message-id: <874kqszgsc.fsf@bhrigu>
- In-reply-to: <0e6178f1-9034-b69c-0acc-145b7016bce5@cohens.org.il>
- References: <6fa792fa-f197-4dc6-0650-46d75fe00dd4@cohens.org.il> <handler.949863.B949863.158039162828357.ackinfo@bugs.debian.org> <be3af0e2-0b9f-4829-c306-44d964f6c253@cohens.org.il> <158002988251.3536327.16810236574772890372.reportbug@malaclypse.mth.mellanox.com> <3c63d9c061960c8ca4e17e3570a232faa15b23b8.camel@decadent.org.uk> <87tv22jvph.fsf@bhrigu> <20200626152537.GA41800@malaclypse.cohens.org.il> <d9cb8901c5c3ed6eb3b7e095f5b68ad6f78c5bd4.camel@decadent.org.uk> <0e6178f1-9034-b69c-0acc-145b7016bce5@cohens.org.il>
[Sending reply to bug so we can have a record of discussion]
[Please keep Ashruth and my colleague in CC while replying]
Tzafrir Cohen <tzafrir@cohens.org.il> writes:
[snip]
>>
>> I don't think we should make a config change that hurts users of stock
>> Debian, for the benefit of those using out-of-tree drivers. And in a
>> stable point release, a known regression like that is unacceptable.
>
> Could you please clarify: is that a regression? That is: in the current
> Buster kernel the problem does not exist?
Yes it looks like regression. Had a discussion with Mellanox (Ashruth
Cced) and looks like its fixed upstream [1]. Currently we found a work
around for this that is to make bonding enabled first and then enable
SR-IOV. This bug happens only if we have enabled SR-IOV on slave
interface before enabling bond.
Ashruth is [1] the right patch?. If not can you point to right patch?.
Ben if a patch is available in main-stream can that be cherry picked to
stable point release to enable this config option?.
[1] https://github.com/torvalds/linux/commit/7ff40a46dd188b83311203e72cedf42eb264fdf1
Cheers,
Vasudev
--- End Message ---