[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CALOAHbB0Ura366LyxH2Buy=11bezbqB+4DZVnZRRo-=q=tMgpg@mail.gmail.com>
Date: Fri, 8 Nov 2024 16:39:58 +0800
From: Yafang Shao <laoar.shao@...il.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: Gal Pressman <gal@...dia.com>, Tariq Toukan <ttoukan.linux@...il.com>, saeedm@...dia.com,
tariqt@...dia.com, leon@...nel.org, netdev@...r.kernel.org,
linux-rdma@...r.kernel.org
Subject: Re: [PATCH] net/mlx5e: Report rx_discards_phy via rx_missed_errors
On Thu, Nov 7, 2024 at 9:17 AM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Wed, 6 Nov 2024 21:23:47 +0200 Gal Pressman wrote:
> > > It appears that rx_fifo_errors is a more appropriate counter for this purpose.
> > > I will submit a v2. Thanks for your suggestion.
> >
> > Probably not a good idea:
> > * This statistics was used interchangeably with @rx_over_errors.
> > * Not recommended for use in drivers for high speed interfaces.
>
> FWIW we can change the definition. Let me copy paste below the commit
> which added the docs because it has the background.
>
> tl;dr is that I was trying to push drivers towards a single stat to
> keep things simple. If we have a clear definition of how rx_fifo_errors
> would differ - we can reuse it and update the doc. For example if
> rx_discards_phy usually means that the adapter itself is overwhelmed
> (too many rules etc) that would be a pretty clear, since rx_missed is
> supposed to primarily indicate that the host rings are full or perhaps
> the PCIe interface of the NIC is struggling. But not the packet
> processing.
Thanks for providing the background.
What do you suggest—should we report rx_discards_phy via
rx_fifo_errors and update the documentation accordingly?
--
Regards
Yafang
Powered by blists - more mailing lists