Re: [PATCH v4 2/3] perf sched stats: Fix SIGCHLD vs pause() race in schedstat_live()

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] The polling loop introduces unnecessary periodic wakeups when running system-wide without a workload.
--

commit 7bab731e30cd90b3c5ff6f388399fa3469974523
Author: Swapnil Sapkal <swapnil.sapkal@amd.com>

perf sched stats: Fix SIGCHLD vs pause() race in schedstat_live()

This commit addresses a lost-wakeup race in schedstat_live where a
short-lived workload's SIGCHLD could be consumed before entering pause().
It replaces the pause() call with a polling loop.

> diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
> --- a/tools/perf/builtin-sched.c
> +++ b/tools/perf/builtin-sched.c

[ ... ]

> @@ -4698,8 +4699,11 @@ static int perf_sched__schedstat_live(struct perf_sched *sched,
>  	if (argc)
>  		evlist__start_workload(evlist);
>  
> -	/* wait for signal */
> -	pause();
> +	while (!done) {
> +		if (argc && waitpid(evlist->workload.pid, NULL, WNOHANG) > 0)
> +			break;
> +		sleep(1);
> +	}

[Severity: Low]
When no workload is provided (argc == 0), there is no child process and thus
no risk of a SIGCHLD lost-wakeup race. Does replacing pause() with a 
sleep(1) loop unconditionally introduce unnecessary periodic wakeups for
the system-wide case?

>  
>  	if (reset) {
>  		err = disable_sched_schedstat();

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260520102017.293419-1-swapnil.sapkal@amd.com?part=2