Re: [RFC PATCH v1.2 11/11] mm/damon/sysfs-schemes: kobject_del() region for populate_region error

[SeongJae Park <sj@kernel.org>]

On Thu, 18 Jun 2026 15:38:37 +0000 sashiko-bot@kernel.org wrote:

> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> 
> Pre-existing issues:
> - [High] Memory leak in the error path of damon_sysfs_scheme_add_dirs() due to skipped cleanup of the 'stats' kobject.
> --
> 
> commit 73251d9a8366deae80fbc88ff7d3d42f79676704
> Author: SeongJae Park <sj@kernel.org>
> 
> mm/damon/sysfs-schemes: kobject_del() region for populate_region error
> 
> This patch fixes directory creation failures due to name conflicts when
> CONFIG_DEBUG_KOBJECT_RELEASE is enabled. It adds a missing kobject_del()
> call to properly clean up tried region directories in the error path of
> damos_sysfs_populate_region_dir().
> 
> > diff --git a/mm/damon/sysfs-schemes.c b/mm/damon/sysfs-schemes.c
> > index ebc692682481d..0b8a6f1f6b9fc 100644
> > --- a/mm/damon/sysfs-schemes.c
> > +++ b/mm/damon/sysfs-schemes.c
> [ ... ]
> 
> [Severity: High]
> This isn't a bug introduced by this patch, but while reviewing the kobject
> cleanup in this file, I noticed a memory leak in the error path of
> damon_sysfs_scheme_add_dirs():
> 
> mm/damon/sysfs-schemes.c:damon_sysfs_scheme_add_dirs() {
>     ...
> 	err = damon_sysfs_scheme_set_stats(scheme);
> 	if (err)
> 		goto put_filters_watermarks_quotas_access_pattern_out;
> 	err = damon_sysfs_scheme_set_tried_regions(scheme);
> 	if (err)
> 		goto put_tried_regions_out;
> 	return 0;
> 
> put_tried_regions_out:
> 	kobject_put(&scheme->tried_regions->kobj);
> 	scheme->tried_regions = NULL;
> put_filters_watermarks_quotas_access_pattern_out:
> 	kobject_put(&scheme->ops_filters->kobj);
>     ...
> }
> 
> If damon_sysfs_scheme_set_tried_regions() fails, it jumps to
> put_tried_regions_out. This attempts to clean up the failed tried_regions
> allocation, but then falls through to the next label.
> 
> Does this completely skip cleaning up the successfully allocated
> scheme->stats->kobj? Missing a call to kobject_put(&scheme->stats->kobj)
> here appears to permanently leak the stats kobject and its associated
> sysfs state.

Yes.  And I was separately working on this.  The patch is now added [1] to
mm-hotfixes-unstable.

[1] https://lore.kernel.org/20260618185830.978911F00A3A@smtp.kernel.org


Thanks,
SJ

[...]