Re: [RFC[RAP] 1/2] block: fix race between set_blocksize and read paths

["Darrick J. Wong" <djwong@kernel.org>]

On Fri, Apr 18, 2025 at 07:51:58AM +0000, Shinichiro Kawasaki wrote:
> On Apr 15, 2025 / 22:14, Christoph Hellwig wrote:
> > On Tue, Apr 15, 2025 at 10:01:44PM -0700, Darrick J. Wong wrote:
> > > It's the same patch as:
> > > https://lore.kernel.org/linux-fsdevel/20250408175125.GL6266@frogsfrogsfrogs/
> > > 
> > > which is to say, xfs/032 with while true; do blkid; done running in the
> > > background to increase the chances of a collision.
> > 
> > I think the xfs-zoned CI actually hit this with 032 without any extra
> > action the.
> 
> I observed xfs/032 hanged using the kernel on linux-xfs/for-next branch with git
> hash 71700ac47ad8. Before the hang, kernel reported the messages below:
> 
>   Oops: general protection fault, probably for non-canonical address 0xdffffc0000000001: 0000 [#1] SMP KASAN NOPTI
>   KASAN: null-ptr-deref in range [0x0000000000000008-0x000000000000000f]
>   CPU: 21 UID: 0 PID: 3187783 Comm: (udev-worker) Not tainted 6.15.0-rc1-kts-xfs-g71700ac47ad+ #1 PREEMPT(lazy)
>   Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-3.fc41 04/01/2014
>   RIP: 0010:guard_bio_eod+0x52/0x5b0
> 
> The failure was recreated in stable manner. I applied this patch series, and
> confirmed the failure disappears. Good. (I needed to resolve conflicts, though)
> 
> This patch fixes block layer. So, IMO, it's the better to have a test case in
> blktests to confirm the fix. I created a blktests test case which recreates the
> failure using blockdev and fio commands. Will post it soon.

Ok.  I'll post a non-rfcrap version of the series shortly.  Thank you
for writing a regression test! :)

--D