[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250324073509.6571ade3@kernel.org>
Date: Mon, 24 Mar 2025 07:35:09 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Gal Pressman <gal@...dia.com>
Cc: Paolo Abeni <pabeni@...hat.com>, "David S. Miller"
<davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, Andrew Lunn
<andrew+netdev@...n.ch>, netdev@...r.kernel.org, Simon Horman
<horms@...nel.org>, Andrew Lunn <andrew@...n.ch>, Tariq Toukan
<tariqt@...dia.com>
Subject: Re: [PATCH net-next] ethtool: Block setting of symmetric RSS when
non-symmetric rx-flow-hash is requested
On Tue, 18 Mar 2025 11:00:10 +0200 Gal Pressman wrote:
> > I guess something alike the following could be quite common or at least
> > possible:
> >
> > #ifdef AH_V4_FLOW
> >
> > // kernel support AH flow, implement user-space side
> >
> > #else
> >
> > // fail on any AH-flow related code path
> >
> > #endif
>
> Right, thanks Paolo!
I don't see a v2, so commenting here.
I believe that we have had this conversation in BPF context in the past.
BPF likes to have things in enums because enums are preserved in debug
info / BTF and defines (obviously) aren't. So we converted a bunch of
things in uAPI to enums in the past for BPF's benefit. While Paolo's
concern is correct (and I believe I voiced similar concerns myself),
in practice we have never encountered any issues.
No strong preference from my side, but FWIW I think there's significant
precedent for such conversions.
One nit if you decide to keep the enum, Gal, please keep the comments
aligned.
Powered by blists - more mailing lists