[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20230920075418.TdZ0jsM_@linutronix.de>
Date: Wed, 20 Sep 2023 09:54:18 +0200
From: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
To: Jesper Dangaard Brouer <hawk@...nel.org>
Cc: netdev@...r.kernel.org, bpf@...r.kernel.org,
"David S. Miller" <davem@...emloft.net>,
Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Eric Dumazet <edumazet@...gle.com>,
Jakub Kicinski <kuba@...nel.org>,
John Fastabend <john.fastabend@...il.com>,
Paolo Abeni <pabeni@...hat.com>,
Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH net v2 0/3] Add missing xdp_do_flush() invocations.
On 2023-09-20 09:04:27 [+0200], Jesper Dangaard Brouer wrote:
> Hi Sebastian,
Hi Jesper,
> On 18/09/2023 17.36, Sebastian Andrzej Siewior wrote:
> > Hi,
> >
> > I've been looking at the drivers/ XDP users and noticed that some
> > XDP_REDIRECT user don't invoke xdp_do_flush() at the end.
>
> I'm wondering if we could detect (and WARN) in the net core e.g.
> net_rx_action() that a driver is missing a flush?
>
> The idea could be to check the per CPU (struct) bpf_redirect_info.
> Or check (per CPU) dev_flush_list.
>
> If some is worried about performance implications, then we can hide this
> under CONFIG_DEBUG_NET.
I had a WARN_ON in mind since the list has to be empty after the
completion of a NAPI callback. Now that you are bringing it up let me
actually do something…
> --Jesper
Sebastian
Powered by blists - more mailing lists