[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CACKFLi=cjSA3kC-9tZTGwKzumC5JAZ-+50=iLyro4pgeQ5SfHA@mail.gmail.com>
Date: Thu, 19 Dec 2024 11:18:42 -0800
From: Michael Chan <michael.chan@...adcom.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: davem@...emloft.net, netdev@...r.kernel.org, edumazet@...gle.com,
pabeni@...hat.com, andrew+netdev@...n.ch, pavan.chebbi@...adcom.com,
andrew.gospodarek@...adcom.com, Ajit Khaparde <ajit.khaparde@...adcom.com>,
Kalesh AP <kalesh-anakkur.purayil@...adcom.com>,
Somnath Kotur <somnath.kotur@...adcom.com>
Subject: Re: [PATCH net-next v2 5/6] bnxt_en: Skip reading PXP registers
during ethtool -d if unsupported
On Thu, Dec 19, 2024 at 9:53 AM Michael Chan <michael.chan@...adcom.com> wrote:
>
> On Thu, Dec 19, 2024 at 6:59 AM Jakub Kicinski <kuba@...nel.org> wrote:
> >
> > On Wed, 18 Dec 2024 22:57:09 -0800 Michael Chan wrote:
> > > The existing bnxt.c in userspace since 2020 always assumes that the
> > > beginning part always contains the PXP register block regardless of
> > > regs->version as long as the register length >= the length of the
> > > register block. I guess we didn't anticipate that this PXP block
> > > would ever be changed or FW would disallow reading it.
> >
> > So if you bumped the version the existing userspace wouldn't care but
> > then you _could_ follow up and update user space to ignore these
> > registers when version is 1?
>
> Sure, will do.
Actually, it's not that easy. We already use version 1 to indicate
that there are PCIe stats following the PXP registers. To make this
work, version will have to be used as flags.
Userspace bnxt.c can detect valid PXP registers in a different way.
If certain non-zero PXP registers are returned as zeros, we know the
PXP register block is not valid and can skip displaying it.
Download attachment "smime.p7s" of type "application/pkcs7-signature" (4209 bytes)
Powered by blists - more mailing lists