diff mbox series

[V2,06/17] block: ublk_drv: mark device as LIVE before adding disk

Message ID 20230307141520.793891-7-ming.lei@redhat.com (mailing list archive)
State New
Headers show
Series io_uring/ublk: add IORING_OP_FUSED_CMD | expand

Commit Message

Ming Lei March 7, 2023, 2:15 p.m. UTC 
IO can be started before add_disk() returns, such as reading parititon table,
then the monitor work should work for making forward progress.

So mark device as LIVE before adding disk, meantime change to
DEAD if add_disk() fails.

Reviewed-by: Ziyang Zhang <ZiyangZhang@linux.alibaba.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
---
 drivers/block/ublk_drv.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Comments

Ziyang Zhang March 8, 2023, 3:48 a.m. UTC | #1 
On 2023/3/7 22:15, Ming Lei wrote:
> IO can be started before add_disk() returns, such as reading parititon table,
> then the monitor work should work for making forward progress.
> 
> So mark device as LIVE before adding disk, meantime change to
> DEAD if add_disk() fails.
> 
> Reviewed-by: Ziyang Zhang <ZiyangZhang@linux.alibaba.com>
> Signed-off-by: Ming Lei <ming.lei@redhat.com>
> ---

Hi Ming,

Without this patch, if we fail to read partition table(Could this
happen?)and EIO is returned, then START_DEV may hang forever, right?
I may have encountered such error before and I think this bug is introduced
by:
bbae8d1f526b(ublk_drv: consider recovery feature in aborting mechanism)
which change the behavior of monitor_work. So shall we add a fixes tag, such
as:
Fixes: bbae8d1f526b("ublk_drv: consider recovery feature in aborting mechanism")

Regards,
Zhang
Ming Lei March 8, 2023, 7:44 a.m. UTC | #2 
On Wed, Mar 08, 2023 at 11:48:15AM +0800, Ziyang Zhang wrote:
> On 2023/3/7 22:15, Ming Lei wrote:
> > IO can be started before add_disk() returns, such as reading parititon table,
> > then the monitor work should work for making forward progress.
> > 
> > So mark device as LIVE before adding disk, meantime change to
> > DEAD if add_disk() fails.
> > 
> > Reviewed-by: Ziyang Zhang <ZiyangZhang@linux.alibaba.com>
> > Signed-off-by: Ming Lei <ming.lei@redhat.com>
> > ---
> 
> Hi Ming,
> 
> Without this patch, if we fail to read partition table(Could this
> happen?)and EIO is returned, then START_DEV may hang forever, right?
> I may have encountered such error before and I think this bug is introduced
> by:
> bbae8d1f526b(ublk_drv: consider recovery feature in aborting mechanism)
> which change the behavior of monitor_work. So shall we add a fixes tag, such
> as:
> Fixes: bbae8d1f526b("ublk_drv: consider recovery feature in aborting mechanism")

Even without the above commit, monitor work still may not be started
because of ub->dev_info.state == UBLK_S_DEV_DEAD.

thanks, 
Ming
diff mbox series

Patch

diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
index 088d9efffe6b..2778c14a8407 100644
--- a/drivers/block/ublk_drv.c
+++ b/drivers/block/ublk_drv.c
@@ -1605,17 +1605,18 @@  static int ublk_ctrl_start_dev(struct ublk_device *ub, struct io_uring_cmd *cmd)
 		set_bit(GD_SUPPRESS_PART_SCAN, &disk->state);
 
 	get_device(&ub->cdev_dev);
+	ub->dev_info.state = UBLK_S_DEV_LIVE;
 	ret = add_disk(disk);
 	if (ret) {
 		/*
 		 * Has to drop the reference since ->free_disk won't be
 		 * called in case of add_disk failure.
 		 */
+		ub->dev_info.state = UBLK_S_DEV_DEAD;
 		ublk_put_device(ub);
 		goto out_put_disk;
 	}
 	set_bit(UB_STATE_USED, &ub->state);
-	ub->dev_info.state = UBLK_S_DEV_LIVE;
 out_put_disk:
 	if (ret)
 		put_disk(disk);