[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20180115114402.46090d88@xeon-e3>
Date: Mon, 15 Jan 2018 11:44:02 -0800
From: Stephen Hemminger <stephen@...workplumber.org>
To: Marcelo Ricardo Leitner <marcelo.leitner@...il.com>
Cc: Daniel Axtens <dja@...ens.net>, netdev@...r.kernel.org,
Thomas Falcon <tlfalcon@...ux.vnet.ibm.com>,
Yuval Mintz <Yuval.Mintz@...ium.com>
Subject: Re: [PATCH v2] bnx2x: disable GSO where gso_size is too big for
hardware
On Fri, 12 Jan 2018 09:53:23 -0200
Marcelo Ricardo Leitner <marcelo.leitner@...il.com> wrote:
> On Fri, Jan 12, 2018 at 09:48:44AM -0200, Marcelo Ricardo Leitner wrote:
> > On Fri, Jan 12, 2018 at 10:59:05AM +1100, Daniel Axtens wrote:
> > > If a bnx2x card is passed a GSO packet with a gso_size larger than
> > > ~9700 bytes, it will cause a firmware error that will bring the card
> > > down:
> >
> > Why not use netif_set_gso_max_size() instead?
> > Some drivers are using it to avoid such larger than supported packets.
>
> Scratch that, sorry. It doesn't take effect in forwarding path.
>
> Marcelo
It still makes sense to tell TCP not to generate huge GSO packets.
For the forwarding path, there needs to be some software recovery
logic in net transmit path.
Powered by blists - more mailing lists