Re: [PATCH 12/13] loop: remove lo_refcount and avoid lo_mutex in ->open / ->release

[Christoph Hellwig <hch@lst.de>]

On Thu, Mar 24, 2022 at 11:24:43PM +0900, Tetsuo Handa wrote:
> On 2022/03/24 23:13, Jan Kara wrote:
> >> diff --git a/drivers/block/loop.c b/drivers/block/loop.c
> >> index b3170e8cdbe95..e1eb925d3f855 100644
> >> --- a/drivers/block/loop.c
> >> +++ b/drivers/block/loop.c
> >> @@ -1244,7 +1244,7 @@ static int loop_clr_fd(struct loop_device *lo)
> >>  	 * <dev>/do something like mkfs/losetup -d <dev> causing the losetup -d
> >>  	 * command to fail with EBUSY.
> >>  	 */
> >> -	if (atomic_read(&lo->lo_refcnt) > 1) {
> >> +	if (disk_openers(lo->lo_disk) > 1) {
> 
> This is sometimes "if (0) {" due to not holding disk->open_mutex.

Yes, as clearly documented in the commit log.  In fact turning it
into an explicit if 0 (that is removing this code) might be a not
bad idea either - the code was added to work around a

	if (lo->lo_refcnt > 1)  /* we needed one fd for the ioctl */
		return -EBUSY;

that already did not make much sense to start with (but goes
back to before git history).

But for now I'd really prefer to stop moving the goalpost further and
further.