Re: [PATCH 3/3] block: skip blkcg walk in blk_cgroup_congested() when nothing throttled

Hello,

On Thu, Aug 06, 2026 at 11:37:20AM -0700, Usama Arif wrote:
> @@ -53,6 +53,9 @@ EXPORT_SYMBOL_GPL(blkcg_root);
>  struct cgroup_subsys_state * const blkcg_root_css = &blkcg_root.css;
>  EXPORT_SYMBOL_GPL(blkcg_root_css);
>  
> +/* number of blkcgs with a non-zero congestion_count */
> +atomic_t blkcg_congested_blkcgs __read_mostly = ATOMIC_INIT(0);

Maybe name it to clarify that this is a count. nr_ or _cnt or whatever.

> @@ -1350,6 +1353,16 @@ static void blkcg_css_free(struct cgroup_subsys_state *css)
>  	struct blkcg *blkcg = css_to_blkcg(css);
>  	int i;
>  
> +	/*
> +	 * Every blkg holds a reference on this css and drops any delay it
> +	 * still has from pd_free_fn(), so this is expected to be zero.  Should
> +	 * a policy ever leave one behind, drop it here rather than let it pin
> +	 * blkcg_congested_blkcgs and disable the fast path for the rest of the
> +	 * boot.  Nothing can race with us at this point.
> +	 */
> +	if (atomic_xchg(&blkcg->congestion_count, 0) > 0)
> +		atomic_dec(&blkcg_congested_blkcgs);

Maybe put WARN_ON_ONCE() on it?

> +static inline void blkcg_dec_congestion_count(struct blkcg *blkcg)
> +{
> +	if (atomic_dec_return(&blkcg->congestion_count) == 0)
> +		atomic_dec(&blkcg_congested_blkcgs);

atomic_dec_and_test() is minutely more efficient.

Thanks.

-- 
tejun