Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
MIME-Version: 1.0
References: <20220511222910.635307-1-dlunev@chromium.org> <20220512082832.v2.2.I692165059274c30b59bed56940b54a573ccb46e4@changeid>
 <YoQfls6hFcP3kCaH@zeniv-ca.linux.org.uk> <YoQihi4OMjJj2Mj0@zeniv-ca.linux.org.uk>
 <CAJfpegtQUP045X5N8ib1rUTKzSj-giih0eL=jC5-MP7aVgyN_g@mail.gmail.com> <CAONX=-do9yvxW2gTak0WGbFVPiLbkM2xH5LReMZkvC-upOUVxg@mail.gmail.com>
In-Reply-To: <CAONX=-do9yvxW2gTak0WGbFVPiLbkM2xH5LReMZkvC-upOUVxg@mail.gmail.com>
From:   Daniil Lunev <dlunev@chromium.org>
Date:   Mon, 23 May 2022 10:25:02 +1000
Message-ID: <CAONX=-ehh=uGYAi++oV_uS23mp2yZcrUC+7U5H0rRz8q0h6OeQ@mail.gmail.com>
Subject: Re: [PATCH v2 2/2] FUSE: Retire superblock on force unmount
To:     Miklos Szeredi <miklos@szeredi.hu>
Cc:     Al Viro <viro@zeniv.linux.org.uk>, linux-fsdevel@vger.kernel.org,
        Christoph Hellwig <hch@infradead.org>,
        fuse-devel <fuse-devel@lists.sourceforge.net>,
        "Theodore Ts'o" <tytso@mit.edu>, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

So, I tried this patchset with open bdi elements during force unmount
and a random file open [1], and didn't see any major drama with
force unmounting the node, after re-mounting, read on sysfs node
returned "no such device", which is expected.
With private bdi flag patch, unless bdi is unregister on force unmount
in fuse, it will complain on name collision [2] (because the patch
actually doesn't do much but unregisters the bdi on unmount, which
seems to happen ok even if node is busy).
Let me know if I am missing anything or if there are any other
concerns, and advise what would be the best way to move this
forward.

Thanks,
Daniil.


[1] Python shell
>>> f1 = open('/sys/class/bdi/8:0-fuseblk/read_ahead_kb', 'r')
>>> f2 = open('/media/removable/USB Drive/m1', 'w')

[2]
[  149.826508] sysfs: cannot create duplicate filename
'/devices/virtual/bdi/8:0-fuseblk'

On Thu, May 19, 2022 at 8:55 AM Daniil Lunev <dlunev@chromium.org> wrote:
>
> > Yep, messing with the bdi doesn't look good.  Fuse always uses a
> > private bdi, so it's not even necessary.
>
> The reason I needed to remove the bdi is name collision - fuse
> generates a fixed name for its bdi based on the underlying block
> device. However, those collisions of mine were conducted on a
> version prior to the private bdi introduction, I am not sure if that
> is supposed to fix the collision issue. Need to check
>
> Thanks,
> Daniil