Re: [syzbot] [fs?] BUG: sleeping function called from invalid context in __getblk_gfp

[Jan Kara]

On Mon 29-01-24 17:15:05, syzbot wrote:
> syzbot suspects this issue was fixed by commit:
> 
> commit 6f861765464f43a71462d52026fbddfc858239a5
> Author: Jan Kara <jack@xxxxxxx>
> Date:   Wed Nov 1 17:43:10 2023 +0000
> 
>     fs: Block writes to mounted block devices
> 
> bisection log:  https://syzkaller.appspot.com/x/bisect.txt?x=116642dfe80000
> start commit:   d88520ad73b7 Merge tag 'pull-nfsd-fix' of git://git.kernel..
> git tree:       upstream
> kernel config:  https://syzkaller.appspot.com/x/.config?x=174a257c5ae6b4fd
> dashboard link: https://syzkaller.appspot.com/bug?extid=69b40dc5fd40f32c199f
> syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=16a77593680000
> C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=1104a593680000
> 
> If the result looks correct, please mark the issue as fixed by replying with:
> 
> #syz fix: fs: Block writes to mounted block devices

This doesn't look correct. The problem here really is that sysv is calling
sb_bread() under a RWLOCK - pointers_lock - in get_block(). Which more or
less shows nobody has run sysv in ages because otherwise the chances of
sleeping in atomic context are really high? Perhaps another candidate for
removal?

								Honza
-- 
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR