Re: regression: LTP isofs testcase failure - bisection fingers "loop: make autoclear operation asynchronous"

[Mike Galbraith]

On Tue, 2022-01-18 at 06:57 +0900, Tetsuo Handa wrote:
> On 2022/01/18 0:34, Mike Galbraith wrote:
> > Greetings,
> >
> > LTP's isofs testcase began failing this cycle, and bisected as below,
> > which a revert then confirmed.  Full testcase output attached.
> >
>
> Thanks for reporting.
> Discussion is continued at https://lkml.kernel.org/r/cdaf1346-2885-f0da-8878-12264bd48348@xxxxxxxxxxxxxxxxxxx ;.

FWIW, simply immediately flushing that shiny new rundown_work turned
LTP's thumb back to upward pointing.

---
 drivers/block/loop.c |    1 +
 1 file changed, 1 insertion(+)

--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -1197,6 +1197,7 @@ static void loop_schedule_rundown(struct
 	kobject_get(&bdev->bd_device.kobj);
 	INIT_WORK(&lo->rundown_work, loop_rundown_workfn);
 	queue_work(system_long_wq, &lo->rundown_work);
+	flush_work(&lo->rundown_work);
 }

 static int loop_clr_fd(struct loop_device *lo)