Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20211227091241.103-1-xieyongji@bytedance.com> <Ycycda8w/zHWGw9c@infradead.org>
 <CACycT3usfTdzmK=gOsBf3=-0e8HZ3_0ZiBJqkWb_r7nki7xzYA@mail.gmail.com>
 <YdMgCS1RMcb5V2RJ@localhost.localdomain> <CACycT3vYt0XNV2GdjKjDS1iyWieY_OV4h=W1qqk_AAAahRZowA@mail.gmail.com>
 <YdSMqKXv0PUkAwfl@localhost.localdomain> <CACycT3tPZOSkCXPz-oYCXRJ_EOBs3dC0+Juv=FYsa6qRS0GVCw@mail.gmail.com>
 <CACycT3tTKBpS_B5vVJ8MZ1iuaF2bf-01=9+tAdxUddziF2DQ-g@mail.gmail.com>
 <CACycT3thVwb466u2JR-oDRHLY5j_uxAx5uXXGmaoCZL5vs__mQ@mail.gmail.com>
 <Yg+5Wytvc2eG8uLD@localhost.localdomain> <CACycT3umMYfwVZRXinEBM=Kh+kQPYH5GBN6eKrt9unZSM8W0qw@mail.gmail.com>
In-Reply-To: <CACycT3umMYfwVZRXinEBM=Kh+kQPYH5GBN6eKrt9unZSM8W0qw@mail.gmail.com>
From:   Yongji Xie <xieyongji@bytedance.com>
Date:   Tue, 22 Mar 2022 16:10:01 +0800
Message-ID: <CACycT3t7=uXfu62WhahwaZpA15WbL42qB_dicB9kb2DS5-R9Qw@mail.gmail.com>
Subject: Re: [PATCH v2] nbd: Don't use workqueue to handle recv work
To:     Josef Bacik <josef@toxicpanda.com>
Cc:     Christoph Hellwig <hch@infradead.org>,
        Jens Axboe <axboe@kernel.dk>,
        Bart Van Assche <bvanassche@acm.org>,
        linux-block@vger.kernel.org, nbd@other.debian.org,
        linux-kernel <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Ping.

Hi Josef, any other comments for this approach.

On Sat, Feb 19, 2022 at 9:04 PM Yongji Xie <xieyongji@bytedance.com> wrote:
>
> On Fri, Feb 18, 2022 at 11:21 PM Josef Bacik <josef@toxicpanda.com> wrote:
> >
> > On Tue, Feb 15, 2022 at 09:17:37PM +0800, Yongji Xie wrote:
> > > Ping again.
> > >
> > > Hi Josef, could you take a look?
> >
> > Sorry Yongji this got lost.  Again in the reconnect case we're still setting up
> > a long running thread, so it's not like it'll happen during a normal reclaim
> > path thing, it'll be acted upon by userspace.  Thanks,
> >
>
> During creating this long running thread, we might trigger memory
> reclaim since workqueue will use GFP_KERNEL allocation for it. Then a
> deadlock can occur if the memory reclaim happens to hit the page cache
> on this reconnecting nbd device.
>
> Thanks,
> Yongji