Re: [RFC PATCH] mm/damon/core: filter out damos address filter-split failed region

[SJ Park <sj@kernel.org>]

On Sun, 18 Oct 2026 22:11:06 +0000 sashiko-bot@kernel.org wrote:

> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [Medium] mm/damon: unchecked damon_split_region_at() error in damos_filter_match()
> --
> 
> commit 881e287348e01011e3ec5442e7deb6f9c112db9c
> Author: SJ Park <sj@kernel.org>
> 
> mm/damon/core: filter out damos address filter-split failed region
> 
> This commit updates damos_filter_match() and damos_core_filter_out() to
> propagate errors from damon_split_region_at() and filter out the region if
> splitting fails during DAMOS_FILTER_TYPE_ADDR evaluation.
> 
> > diff --git a/mm/damon/core.c b/mm/damon/core.c
> > index b63e60ef89901..4d34587387456 100644
> > --- a/mm/damon/core.c
> > +++ b/mm/damon/core.c
> > @@ -2551,13 +2551,14 @@ static void damos_update_stat(struct damos *s,
> >  
> >  static bool damos_filter_match(struct damon_ctx *ctx, struct damon_target *t,
> >  		struct damon_region *r, struct damos_filter *filter,
> > -		unsigned long min_region_sz)
> > +		unsigned long min_region_sz, int *err)
> >  {
> >  	bool matched = false;
> >  	struct damon_target *ti;
> >  	int target_idx = 0;
> >  	unsigned long start, end, wsum;
> >  
> > +	*err = 0;
> >  	switch (filter->type) {
> >  	case DAMOS_FILTER_TYPE_TARGET:
> >  		damon_for_each_target(ti, ctx) {
> > @@ -2588,7 +2589,9 @@ static bool damos_filter_match(struct damon_ctx *ctx, struct damon_target *t,
> >  			break;
> >  		}
> >  		/* start inside the range */
> > -		damon_split_region_at(t, r, end - r->ar.start);
> > +		*err = damon_split_region_at(t, r, end - r->ar.start);
> > +		if (*err)
> > +			return false;
> >  		matched = true;
> >  		break;
> 
> [Severity: Medium]
> Should the earlier damon_split_region_at() call in damos_filter_match()
> when r->ar.start < start also propagate its return value via *err?
> 
> mm/damon/core.c:damos_filter_match() {
> 	case DAMOS_FILTER_TYPE_ADDR:
> 		...
> 		/* start before the range and overlap */
> 		if (r->ar.start < start) {
> 			damon_split_region_at(t, r, start - r->ar.start);
> 			matched = false;
> 			break;
> 		}
> }
> 
> When a region starts before the filter range and overlaps it (r->ar.start <
> start and r->ar.end > start), if damon_split_region_at(t, r, start -
> r->ar.start) fails with -ENOMEM, r remains unsplit across [r->ar.start,
> r->ar.end) while matched is set to false and *err remains 0.
> 
> If the filter is configured to exclude [start, end) (for example,
> filter->matching == true and filter->allow == false with
> s->core_filters_default_reject == false, or filter->matching == false and
> filter->allow == true), kdamond_apply_schemes() -> damos_apply_scheme() ->
> damos_core_filter_out() sees err == 0, does not reject r, and proceeds to
> call c->ops.apply_scheme() on the entire unsplit region including the
> excluded address range.

Good catch.  I will fix this in the next revision.

> 
> [ ... ]
> 
> -- 
> Sashiko AI review · https://sashiko.dev/#/patchset/20261018220253.158238-1-sj@kernel.org?part=1


Thanks,
SJ