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;
MIME-Version: 1.0
References: <1599562954-87257-1-git-send-email-linyunsheng@huawei.com>
 <CAM_iQpX0_mz+McZdzZ7HFTjBihOKz5E6i4qJQSoFbZ=SZkVh=Q@mail.gmail.com>
 <830f85b5-ef29-c68e-c982-de20ac880bd9@huawei.com> <CAM_iQpU_tbRNO=Lznz_d6YjXmenYhowEfBoOiJgEmo9x8bEevw@mail.gmail.com>
 <1f8ebcde-f5ff-43df-960e-3661706e8d04@huawei.com> <CAM_iQpUm91x8Q0G=CXE7S43DKryABkyMTa4mz_oEfEOTFS7BgQ@mail.gmail.com>
 <db770012-f22c-dff4-5311-bf4d17cd08e3@huawei.com> <CAM_iQpUBytX3qim3rXLkwjdX3DSKeF8YhyX6o=Jwr-R9Onb-HA@mail.gmail.com>
 <5472023c-b50b-0cb3-4cb6-7bbea42d3612@huawei.com>
In-Reply-To: <5472023c-b50b-0cb3-4cb6-7bbea42d3612@huawei.com>
From:   Cong Wang <xiyou.wangcong@gmail.com>
Date:   Wed, 4 Nov 2020 22:04:38 -0800
Message-ID: <CAM_iQpVGm_Mz-yYUhhvn+p8H7mXHWHAuBNfyNj-251eY3Vr9iA@mail.gmail.com>
Subject: Re: [PATCH v2 net] net: sch_generic: aviod concurrent reset and
 enqueue op for lockless qdisc
To:     Yunsheng Lin <linyunsheng@huawei.com>
Cc:     Jamal Hadi Salim <jhs@mojatatu.com>, Jiri Pirko <jiri@resnulli.us>,
        David Miller <davem@davemloft.net>,
        Jakub Kicinski <kuba@kernel.org>,
        Linux Kernel Network Developers <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>, linuxarm@huawei.com,
        John Fastabend <john.fastabend@gmail.com>,
        Eric Dumazet <eric.dumazet@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Nov 2, 2020 at 11:24 PM Yunsheng Lin <linyunsheng@huawei.com> wrote:
> >> From my understanding, we can do anything about the old qdisc (including
> >> destorying the old qdisc) after some_qdisc_is_busy() return false.
> >
> > But the current code does the reset _before_ some_qdisc_is_busy(). ;)
>
> If lock is taken when doing reset, it does not matter if the reset is
> before some_qdisc_is_busy(), right?

Why not? How about the following scenario?

CPU0:                   CPU1:
dev_reset_queue()
                        net_tx_action()
                         -> sch_direct_xmit()
                           -> dev_requeue_skb()
some_qdisc_is_busy()
// waiting for TX action on CPU1
// now some packets are requeued