[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20240911152539.4030764b@kernel.org>
Date: Wed, 11 Sep 2024 15:25:39 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Andrew Lunn <andrew@...n.ch>
Cc: Vadim Fedorenko <vadim.fedorenko@...ux.dev>, David Ahern
<dsahern@...nel.org>, Paolo Abeni <pabeni@...hat.com>, "David S. Miller"
<davem@...emloft.net>, Alexander Duyck <alexanderduyck@...com>,
netdev@...r.kernel.org
Subject: Re: [PATCH net-next 2/5] eth: fbnic: add initial PHC support
On Wed, 11 Sep 2024 22:45:11 +0200 Andrew Lunn wrote:
> > > That's good question. Do we need another set of helpers just because of
> > > names? Obviously, the internals will be the same sequence magic.
> >
> > Good question. To be clear we want a seq lock that goes away on 64b
> > since what it protects is accessed on the fast path (potentially per
> > packet). We could s/u64_stats/u64_seq/ the existing helpers. But that
> > sounds like a lot for a single user. Dunno..
>
> It does sound like a lot of a single user.
>
> And what is the likelihood of this device ever being used on a 32 bit
> system? It is a server class NIC. Are there still 32 bit servers in
> use?
>
> Maybe "depends on 64BIT" with a good commit message why?
Will this not apply all new MMIO devices? They will either be high
enough class to be only used on 64b systems, or built into a 64b SoC.
No strong feelings either way, but I think we'd need a better defined
guidance on when "depends on 64BIT" is acceptable and when developers
have to go thru the pain of using u64_stats_*.
Powered by blists - more mailing lists