[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CACKFLik+AOKy3dy6Dv-rLOhsO4m+Q54KZRaNv5W7a6OFbMWorA@mail.gmail.com>
Date: Sun, 22 Jun 2025 21:20:32 -0700
From: Michael Chan <michael.chan@...adcom.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: pavan.chebbi@...adcom.com, willemdebruijn.kernel@...il.com,
netdev@...r.kernel.org, davem@...emloft.net, edumazet@...gle.com,
pabeni@...hat.com, andrew+netdev@...n.ch, horms@...nel.org
Subject: Re: [RFC net-next 4/6] eth: bnxt: support RSS on IPv6 Flow Label
On Thu, Jun 12, 2025 at 6:04 PM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Thu, 12 Jun 2025 17:11:18 -0700 Michael Chan wrote:
> > Just a quick update on this. The FW call fails when this flag is set.
> > I am looking into it.
>
> Thanks for the update! FWIW I'm doing the refactoring to add netlink
> support as requested by Andrew:
> https://lore.kernel.org/all/20250611145949.2674086-1-kuba@kernel.org/
> There's 40 drivers to convert so even if it goes smoothly I wouldn't
> be able to repost for another week.
I've looked into this. The way flow label works on our chip is that
the flow label can be hashed with the IPV6 source and destination
addresses only, effectively becoming a 3-tuple hash. For example,
this would be valid:
ethtool -N eth0 rx-flow-hash tcp6 sdl
4-tuple hash cannot include the flow label. 2-tuple and 3-tuple are
mutually exclusive for IPV6. For example, once tcp6 is configured for
3-tuple, udp6 cannot be configured for 2-tuple.
I can post the patch that I am testing after some more testing. Thanks.
Download attachment "smime.p7s" of type "application/pkcs7-signature" (4196 bytes)
Powered by blists - more mailing lists