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;
Date:   Mon, 31 Oct 2022 15:23:51 +0100
From:   David Sterba <dsterba@suse.cz>
To:     Matthew Wilcox <willy@infradead.org>
Cc:     syzbot <syzbot+8edfa01e46fd9fe3fbfb@syzkaller.appspotmail.com>,
        akpm@linux-foundation.org, clm@fb.com, dsterba@suse.com,
        hch@lst.de, josef@toxicpanda.com, linmiaohe@huawei.com,
        linux-btrfs@vger.kernel.org, linux-kernel@vger.kernel.org,
        syzkaller-bugs@googlegroups.com, torvalds@linux-foundation.org
Subject: Re: [syzbot] WARNING in btrfs_space_info_update_bytes_may_use
Message-ID: <20221031142351.GA5824@twin.jikos.cz>
Reply-To: dsterba@suse.cz
References: <0000000000000d9d6f05ec498263@google.com>
 <000000000000fa42c105ec5339ec@google.com>
 <Y1/SqIuP4tbszPAW@casper.infradead.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <Y1/SqIuP4tbszPAW@casper.infradead.org>
User-Agent: Mutt/1.5.23.1-rc1 (2014-03-12)
Precedence: bulk

On Mon, Oct 31, 2022 at 01:50:32PM +0000, Matthew Wilcox wrote:
> Dmitry, I don't see a way to tell syzbot that its bisection has gone
> astray.  Can you add one or document it if it already exists?
> 
> On Mon, Oct 31, 2022 at 04:51:22AM -0700, syzbot wrote:
> > syzbot has bisected this issue to:
> > 
> > commit 0c7c575df56b957390206deb018c41acbb412159
> > Author: Matthew Wilcox (Oracle) <willy@infradead.org>
> > Date:   Wed Feb 24 20:01:52 2021 +0000
> > 
> >     mm/filemap: remove dynamically allocated array from filemap_read
> 
> This change affects the read path.  The crash happens in the unmount
> path.  The data structure that's being checked is modified in the write
> path.  I just can't see how this commit is in any way related.

Yeah I agree the bisection hasn't identified the correct commit.

> > bisection log:  https://syzkaller.appspot.com/x/bisect.txt?x=119e21b6880000
> > start commit:   b229b6ca5abb Merge tag 'perf-tools-fixes-for-v6.1-2022-10-..

The starting commit is not related to btrfs but the bisection hit the
warning basically on each run so it's not completely random. There might
be some timing change that triggers the warning, likely it's caused by
some the space accounting bug.