Re: [RESEND PATCH v2 1/3] mm/damon: Introduce DAMOS_QUOTA_HUGEPAGE auto tuning

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 4 potential issue(s) to consider:

New issues:
- [High] Calculation of the huge page memory ratio in `damos_hugepage_mem_bp()` can trigger a divide-by-zero kernel panic if the calculated total memory evaluates to zero.
- [Medium] The new metric `DAMOS_QUOTA_HUGEPAGE_MEM_BP` is not exposed to the DAMON sysfs interface, making it unconfigurable from userspace.

Pre-existing issues:
- [High] Preexisting division by zero vulnerability in `damos_get_in_active_mem_bp()` when LRU memory is empty.
- [High] A user-controlled target_value of 0 will cause a divide-by-zero panic during quota scoring.
--

--- Patch [1]: [RESEND PATCH v2 1/3] mm/damon: Introduce DAMOS_QUOTA_HUGEPAGE auto tuning ---
commit bc154724e397943cbba0b2ca233b1c3995ce0802
Author: Asier Gutierrez <gutierrez.asier@huawei-partners.com>

mm/damon: Introduce DAMOS_QUOTA_HUGEPAGE auto tuning

Introduce DAMOS_QUOTA_HUGEPAGE_MEM_BP auto tuning. Add a new
DAMOS quota goal metric to measure the amount of huge page
consumption to total memory consumption ratio.

> diff --git a/include/linux/damon.h b/include/linux/damon.h
> index 02ac34537df9a..6f87ec40db8ca 100644
> --- a/include/linux/damon.h
> +++ b/include/linux/damon.h
> @@ -162,6 +162,7 @@ enum damos_action {
>   * @DAMOS_QUOTA_INACTIVE_MEM_BP:	Inactive to total LRU memory ratio.
>   * @DAMOS_QUOTA_NODE_ELIGIBLE_MEM_BP:	Scheme-eligible memory ratio of a
>   *					node in basis points (0-10000).
> + * @DAMOS_QUOTA_HUGEPAGE_MEM_BP:	Huge page to total used memory ratio.

[Severity: Medium]
Will this new metric be configurable from userspace?

It looks like damos_sysfs_qgoal_metric_names[] in mm/damon/sysfs-schemes.c
was not updated, which might cause target_metric_store() to return -EINVAL
when trying to configure this metric via sysfs.

[ ... ]

> diff --git a/mm/damon/core.c b/mm/damon/core.c
> index 7e4b9affc5b06..b001f80681b12 100644
> --- a/mm/damon/core.c
> +++ b/mm/damon/core.c
> @@ -2795,6 +2795,17 @@ static unsigned int damos_get_in_active_mem_bp(bool active_ratio)
>  	return mult_frac(inactive, 10000, total);
>  }

[Severity: High]
This isn't a bug introduced by this patch, but can total be zero here?

If the system has exactly 0 active and 0 inactive LRU pages (which could
happen if all memory is locked, used strictly by slab, or on tiny setups),
total would be 0. This might cause a divide-by-zero panic in mult_frac().

>  
> +static unsigned int damos_hugepage_mem_bp(void)
> +{
> +	unsigned long thp, total;
> +
> +	thp = global_node_page_state(NR_ANON_THPS) +
> +				global_node_page_state(NR_SHMEM_THPS) +
> +				global_node_page_state(NR_FILE_THPS);
> +	total = totalram_pages() - global_zone_page_state(NR_FREE_PAGES);
> +	return mult_frac(thp, 10000, total);

[Severity: High]
Could total evaluate to zero here?

Due to per-CPU drift in the vm_stat counters (global_zone_page_state),
NR_FREE_PAGES could momentarily equal totalram_pages(). Passing total == 0
to mult_frac() will trigger a divide-by-zero kernel panic when the schema
is evaluated. Should we add a check for !total before calculating the
fraction?

[ ... ]

[Severity: High]
This is a pre-existing issue, but I noticed another potential
divide-by-zero related to these quota goals.

In mm/damon/sysfs-schemes.c:target_value_store(), user input is parsed via
kstrtoul() directly into goal->target_value without validating that it is
non-zero:

    int err = kstrtoul(buf, 0, &goal->target_value);
    return err ? err : count;

Later, when the DAMON thread evaluates the quota score,
mm/damon/core.c:damos_quota_score() does:

    highest_score = max(highest_score,
            mult_frac(goal->current_value, 10000,
                goal->target_value));

Could a user writing 0 to the target_value sysfs file cause a predictable
and immediate system crash here?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260714150116.382521-1-gutierrez.asier@huawei-partners.com?part=1