Re: [RFC PATCH v3 8/9] mm/damon/core: cap the region merge threshold per target

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] mm/damon: region bound violation when probe weights exceed max_thres
--

commit 0342e0b2c0ad1539186d08fbb17c049c0f491a79
Author: Ravi Jonnalagadda <ravis.opensrc@gmail.com>

mm/damon/core: cap the region merge threshold per target

kdamond_merge_regions() applies one context-wide threshold, the highest
merge score across all targets divided by 10, to every target.  This commit
caps the threshold for each target by a tenth of its own maximum merge score
on the regular merge pass.

> diff --git a/mm/damon/core.c b/mm/damon/core.c
> index 1d25f820e3f93..c2c74d0daf957 100644
> --- a/mm/damon/core.c
> +++ b/mm/damon/core.c
> @@ -4175,12 +4175,36 @@ static noinline_for_stack void kdamond_merge_regions(struct damon_ctx *c,
>  	unsigned int nr_regions;
>  	unsigned int max_thres;
>  	bool count_age = true;
> +	bool use_probe_hits = damon_has_probe_weights(c);
>  
>  	max_thres = damon_nr_samples_per_aggr(&c->attrs);
>  	while (true) {
>  		nr_regions = 0;
>  		damon_for_each_target(t, c) {
> -			damon_merge_regions_of(t, threshold, sz_limit, c,
> +			struct damon_region *r;
> +			unsigned int t_max = 0, t_thres = threshold;
> +
> +			/*
> +			 * On the regular pass, cap the threshold by this
> +			 * target's own maximum merge score.

[ ... ]

> +			 */
> +			if (count_age) {
> +				damon_for_each_region(r, t)
> +					t_max = max(t_max, damon_merge_score(r,
> +							false, c, use_probe_hits));
> +				t_thres = min(threshold, t_max / 10);

[Severity: High]
Does this cap the merge threshold for the cold target to near-zero, ensuring it
is highly fragmented on the first pass?

> +			}
> +			damon_merge_regions_of(t, t_thres, sz_limit, c,
>  					count_age);
>  			nr_regions += damon_nr_regions(t);
>  		}
>  		count_age = false;
>  		if (nr_regions <= c->attrs.max_nr_regions ||
>  				max_thres <= threshold)
>  			break;

[Severity: High]
When DAMON is configured with event probes using high weights, does this break
the loop immediately if the weighted initial threshold exceeds the unweighted
max_thres?

If so, does this prevent the uncapped fallback passes from ever running to
satisfy the region count bound?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261003-damon-perf-rfc-v3-send-2026-10-03-v3-0-0f00417b41bc@gmail.com?part=8