Re: [PATCH v2 1/2] mm/damon/core: split a fraction of regions when nr_regions exceeds max/2

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Data race on the globally shared static variable `rotation` causes undefined behavior and inter-context interference.
--

commit 2895b546218b4ccc570b82e50e62f788bac4157b
Author: Jiayuan Chen <jiayuan.chen@shopee.com>

mm/damon/core: split a fraction of regions when nr_regions exceeds max/2

This commit modifies kdamond_split_regions() to split only a fraction of
regions rather than returning early when nr_regions is above max_nr_regions / 2.
It uses a rotating offset to ensure different regions are picked for splitting
over time as the region count smoothly approaches max_nr_regions.

> diff --git a/mm/damon/core.c b/mm/damon/core.c
> --- a/mm/damon/core.c
> +++ b/mm/damon/core.c

[ ... ]

> +/* Split one in every @split_step regions into two, from a rotating offset */
> +static void damon_split_some_regions(struct damon_ctx *ctx,
> +				     unsigned long split_step)
> +{
> +	static unsigned long rotation;
> +	struct damon_target *t;
> +	struct damon_region *r, *next;
> +	unsigned long offset = rotation++ % split_step;

[Severity: Medium]
Will this cause a data race if multiple DAMON contexts are running concurrently?

Since rotation is declared static and locklessly incremented, independent
kdamond worker threads could read and mutate it simultaneously.

This unsynchronized read-modify-write on shared state might trigger automated
data race reports. Additionally, does this mean different contexts will step
on each other's rotation offsets, breaking the intended round-robin behavior
for each individual context?

[ ... ]

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260626085851.70754-1-jiayuan.chen@linux.dev?part=1