[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240610105717.7fbb1caa@kernel.org>
Date: Mon, 10 Jun 2024 10:57:17 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Jason Xing <kerneljasonxing@...il.com>
Cc: Dan Jurgens <danielj@...dia.com>, "Michael S. Tsirkin" <mst@...hat.com>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>, "jasowang@...hat.com"
<jasowang@...hat.com>, "xuanzhuo@...ux.alibaba.com"
<xuanzhuo@...ux.alibaba.com>, "virtualization@...ts.linux.dev"
<virtualization@...ts.linux.dev>, "davem@...emloft.net"
<davem@...emloft.net>, "edumazet@...gle.com" <edumazet@...gle.com>,
"abeni@...hat.com" <abeni@...hat.com>, Parav Pandit <parav@...dia.com>
Subject: Re: [PATCH net-next] virtio_net: Add TX stop and wake counters
On Sat, 8 Jun 2024 08:41:44 +0800 Jason Xing wrote:
> > > Sorry to revive this thread. I wonder why not use this patch like mlnx driver
> > > does instead of adding statistics into the yaml file? Are we gradually using or
> > > adding more fields into the yaml file to replace the 'ethtool -S' command?
> > >
> >
> > It's trivial to have the stats in ethtool as well. But I noticed
> > the stats series intentionally removed some stats from ethtool. So
> > I didn't put it both places.
>
> Thank you for the reply. I thought there was some particular reason
> :-)
Yes, we don't want duplication. We have a long standing (and
documented) policy against duplicating normal stats in custom stat
APIs, otherwise vendors pile everything into the custom stats.
Powered by blists - more mailing lists