[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAL+tcoCZ_pfv5g1P+x+gTeAseifJ=W6y+0GGgS31ih7BqGCrTw@mail.gmail.com>
Date: Tue, 11 Jun 2024 10:56:09 +0800
From: Jason Xing <kerneljasonxing@...il.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: Eric Dumazet <edumazet@...gle.com>, pabeni@...hat.com, davem@...emloft.net,
dsahern@...nel.org, mst@...hat.com, jasowang@...hat.com,
xuanzhuo@...ux.alibaba.com, eperezma@...hat.com, leitao@...ian.org,
netdev@...r.kernel.org, Jason Xing <kernelxing@...cent.com>
Subject: Re: [PATCH net-next] net: dqs: introduce NETIF_F_NO_BQL device feature
On Tue, Jun 11, 2024 at 9:45 AM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Mon, 10 Jun 2024 07:55:55 +0800 Jason Xing wrote:
> > > (I think Vladimir was trying to make some room, this was a discussion
> > > we had last year)
>
> s/Vladimir/Olek/ ?
>
> > Thanks for your reminder. When I was trying to introduce one new bit,
> > I noticed an overflow warning when compiling.
> >
> > > I do not see the reason to report to ethtool the 'nobql bit' :
> > > If a driver opts-out, then the bql sysfs files will not be there, user
> > > space can see the absence of the files.
> >
> > The reason is that I just followed the comment to force myself to
> > report to ethtool. Now I see.
> >
> > It seems not that easy to consider all the non-BQL drivers. Let me
> > think more about it.
>
> All Eric was saying, AFAIU, is that you can for example add a bit
> in somewhere towards the end of struct nedevice, no need to pack
> this info into feature bits.
Oh, thanks for pointing this out.
I would like to add a new bit field in the enum netdev_priv_flags
because it is better that it's grouped into an existing enum for
future compatibility.
>
> BTW the Fixes tag is a bit of an exaggeration here. The heuristic in
> netdev_uses_bql() is best effort, its fine to miss some devices.
I see.
Thanks,
Jason
Powered by blists - more mailing lists