Re: [PATCH 2/5] ublk: add UBLK_F_NO_AUTO_PART_SCAN feature flag

[Ming Lei <ming.lei@redhat.com>]

On Wed, Jan 28, 2026 at 08:44:58PM -0800, Caleb Sander Mateos wrote:
> On Wed, Jan 28, 2026 at 7:15 PM Ming Lei <ming.lei@redhat.com> wrote:
> >
> > Add a new feature flag UBLK_F_NO_AUTO_PART_SCAN to allow users to suppress
> > automatic partition scanning when starting a ublk device.
> >
> > This is useful for some cases in which use don't want to scan
> > partitions.
> >
> > Users still can manually trigger partition scanning later when appropriate
> > using standard tools (e.g., partprobe, blockdev --rereadpt).
> >
> > Reported-by: Yoav Cohen <yoav@nvidia.com>
> > Link: https://lore.kernel.org/linux-block/DM4PR12MB63280C5637917C071C2F0D65A9A8A@DM4PR12MB6328.namprd12.prod.outlook.com/
> > Signed-off-by: Ming Lei <ming.lei@redhat.com>
> > ---
> >  drivers/block/ublk_drv.c      | 7 ++++++-
> >  include/uapi/linux/ublk_cmd.h | 3 +++
> >  2 files changed, 9 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
> > index 94470a2ec05e..0b98ab0f8257 100644
> > --- a/drivers/block/ublk_drv.c
> > +++ b/drivers/block/ublk_drv.c
> > @@ -80,7 +80,8 @@
> >                 | UBLK_F_BUF_REG_OFF_DAEMON \
> >                 | (IS_ENABLED(CONFIG_BLK_DEV_INTEGRITY) ? UBLK_F_INTEGRITY : 0) \
> >                 | UBLK_F_SAFE_STOP_DEV \
> > -               | UBLK_F_BATCH_IO)
> > +               | UBLK_F_BATCH_IO \
> > +               | UBLK_F_NO_AUTO_PART_SCAN)
> >
> >  #define UBLK_F_ALL_RECOVERY_FLAGS (UBLK_F_USER_RECOVERY \
> >                 | UBLK_F_USER_RECOVERY_REISSUE \
> > @@ -2364,6 +2365,10 @@ static void ublk_partition_scan_work(struct work_struct *work)
> >                                              &disk->state)))
> >                 goto out;
> >
> > +       /* Skip partition scan if disabled by user */
> > +       if (ub->dev_info.flags & UBLK_F_NO_AUTO_PART_SCAN)
> > +               goto out;
> 
> Seems like it's not necessary to schedule ublk_partition_scan_work()
> in this case? Could just clear_bit(GD_SUPPRESS_PART_SCAN,
> &disk->state) in ublk_ctrl_start_dev()?

OK, we can do it by saving one time of wq schedule.

> Is it actually necessary to clear GD_SUPPRESS_PART_SCAN at all? Does
> that flag block partprobe too?

Yes, the flag is totally controlled by user side.


Thanks, 
Ming