Re: [PATCH 5/8] xfs_scrub: fix estimate of work items for phase 4

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

On Fri, Jun 26, 2026 at 07:04:26AM +0200, Christoph Hellwig wrote:
> On Thu, Jun 25, 2026 at 03:55:31PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <djwong@kernel.org>
> > 
> > Codex complains that the number of work items computed when estimating
> > the amount of work for phase 4 doesn't include the FSCOUNTERS and
> > QUOTACHECK items.  Add them back in.
> > 
> > Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
> > ---
> >  scrub/phase4.c |    4 ++++
> >  1 file changed, 4 insertions(+)
> > 
> > 
> > diff --git a/scrub/phase4.c b/scrub/phase4.c
> > index 6bd16c23939eb4..744fd92ab195f4 100644
> > --- a/scrub/phase4.c
> > +++ b/scrub/phase4.c
> > @@ -297,6 +297,10 @@ phase4_estimate(
> >  	need_fixing = action_list_length(ctx->fs_repair_list) +
> >  		      action_list_length(ctx->file_repair_list);
> >  
> > +	/* fscounters and quotacheck */
> > +	if (need_fixing)
> > +		need_fixing += 2;
> > +
> 
> This matches the commit message, but there does this come from?
> Why are fscountれrs and quotachck special?

They're run explicitly by phase4_func so that repairs don't fail with
ENOSPC or EDQUOT if the summary counters (or quota usage numbers) are
wrong.  How about I expand on this in the comments?

phase4_func:

	/*
	 * Check the resource usage counters early.  Normally we do this
	 * during phase 7, but some of the cross-referencing requires
	 * fairly accurate summary counters.  Check and try to repair
	 * them now to minimize the chance that repairs of primary
	 * metadata fail due to secondary metadata or ENOSPC on broken
	 * counters.  If repairs fails, we'll come back during phase 7.
	 */
	scrub_item_init_fs(&sri);
	scrub_item_schedule(&sri, XFS_SCRUB_TYPE_FSCOUNTERS);

and:

	/*
	 * Try to fix the quota usage counts so that online repair
	 * doesn't fail with EDQUOT (or worse shut down the fs) due to
	 * bad counts.
	 */
	if (fsgeom.sick & XFS_FSOP_GEOM_SICK_QUOTACHECK)
		scrub_item_schedule(&sri, XFS_SCRUB_TYPE_QUOTACHECK);

phase4_estimate:

	/*
	 * fscounters and quotacheck are run directly by phase4_func
	 * independent of the repair lists, so put that in the item
	 * count.  See phase4_func for why.
	 */
	if (need_fixing)
		need_fixing += 2;

How does that sound?

--D