Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed, 19 May 2021 15:53:54 -0700
From:   Jakub Kicinski <kuba@kernel.org>
To:     Eric Dumazet <eric.dumazet@gmail.com>
Cc:     Stephen Hemminger <stephen@networkplumber.org>,
        Dave Taht <dave.taht@gmail.com>,
        Willem de Bruijn <willemdebruijn.kernel@gmail.com>,
        "Michael S. Tsirkin" <mst@redhat.com>,
        Xianting Tian <xianting.tian@linux.alibaba.com>,
        Linux Kernel Network Developers <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        virtualization <virtualization@lists.linux-foundation.org>,
        bloat <bloat@lists.bufferbloat.net>,
        "David S. Miller" <davem@davemloft.net>
Subject: Re: [Bloat] virtio_net: BQL?
Message-ID: <20210519155354.4438565e@kicinski-fedora-PC1C0HJN>
In-Reply-To: <a11eee78-b2a1-3dbc-4821-b5f4bfaae819@gmail.com>
References: <56270996-33a6-d71b-d935-452dad121df7@linux.alibaba.com>
        <CAA93jw6LUAnWZj0b5FvefpDKUyd6cajCNLoJ6OKrwbu-V_ffrA@mail.gmail.com>
        <CA+FuTSf0Af2RXEG=rCthNNEb5mwKTG37gpEBBZU16qKkvmF=qw@mail.gmail.com>
        <CAA93jw7Vr_pFMsPCrPadqaLGu0BdC-wtCmW2iyHFkHERkaiyWQ@mail.gmail.com>
        <20210517160036.4093d3f2@hermes.local>
        <a11eee78-b2a1-3dbc-4821-b5f4bfaae819@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Wed, 19 May 2021 16:31:10 +0200 Eric Dumazet wrote:
> On 5/18/21 1:00 AM, Stephen Hemminger wrote:
> > The Azure network driver (netvsc) also does not have BQL. Several years ago
> > I tried adding it but it benchmarked worse and there is the added complexity
> > of handling the accelerated networking VF path.
> 
> Note that NIC with many TX queues make BQL almost useless, only adding extra
> overhead.
> 
> We should probably make BQL something that can be manually turned on/off.

Ah, I've been pondering this. Are you thinking of a bit in
dev_queue->state? Not perfect, because with a careful driver design 
one can avoid most dev_queue accesses from the completion path.
It's still much better than recompiling the kernel to set BQL=n, tho.