[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CA+mtBx-dYMRCTz651Nd4emO9gutynLvXzLRuqVkMbF+HhyGRPg@mail.gmail.com>
Date: Thu, 2 Oct 2014 08:27:36 -0700
From: Tom Herbert <therbert@...gle.com>
To: Eric Dumazet <eric.dumazet@...il.com>
Cc: Amir Vadai <amirv@...lanox.com>,
Jesper Dangaard Brouer <brouer@...hat.com>,
Linux Netdev List <netdev@...r.kernel.org>,
"David S. Miller" <davem@...emloft.net>,
Hannes Frederic Sowa <hannes@...essinduktion.org>,
Florian Westphal <fw@...len.de>,
Daniel Borkmann <dborkman@...hat.com>,
Jamal Hadi Salim <jhs@...atatu.com>,
Alexander Duyck <alexander.duyck@...il.com>,
John Fastabend <john.r.fastabend@...el.com>,
Dave Taht <dave.taht@...il.com>,
Toke Høiland-Jørgensen <toke@...e.dk>
Subject: Re: [net-next PATCH V6 0/2] qdisc: bulk dequeue support
On Thu, Oct 2, 2014 at 8:04 AM, Eric Dumazet <eric.dumazet@...il.com> wrote:
> On Thu, 2014-10-02 at 07:42 -0700, Tom Herbert wrote:
>
>
>> Unfortunately we probably still need something. If BQL were disabled
>> (by setting BQL min_limit to infinity) then we'll always dequeue all
>> the packets in the qdisc. Disabling BQL might be legitimate in
>> deployment if say a bug is found in a device that prevents prompt
>> transmit completions for some corner case.
>
>
> Unfortunately, there is nothing we can do if a ndo_start_xmit() function
> is buggy.
>
I was referring more to device bugs not driver bugs. But even so, if a
user hits a problem like this in the system, disabling something like
BQL would probably be a better recourse than bringing their network
down.
Minimally we should at least test with BQL disabled, it might not be
so horrible. I suppose in the worse case we're at least limited by the
qdisc limit.
> For example, if a prior packet is queued on NIC TX ring (with xmit_more
> set to 1)
>
> Then following packet given to ndo_start_xmit() returns infamous
> NETDEV_TX_BUSY, we are screwed.
>
> mlx4 driver for example has this bug.
>
>
> Thats why I advocate we have no artifical limit : We are going to catch
> driver bugs sooner and fix them before xmit_more support is validated,
> for each driver that claims to support it.
>
> Amir, the problem with NETDEV_TX_BUSY in mlx4 need to be fixed.
>
>
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists