linux-kernel - RE: [PATCH v1] perf intel-tpebs: Filter non-workload samples

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CO6PR11MB5635C6FF9A6C91B68FCE7937EE90A@CO6PR11MB5635.namprd11.prod.outlook.com>
Date: Thu, 15 May 2025 20:46:22 +0000
From: "Wang, Weilin" <weilin.wang@...el.com>
To: Ian Rogers <irogers@...gle.com>, Peter Zijlstra <peterz@...radead.org>,
	Ingo Molnar <mingo@...hat.com>, Arnaldo Carvalho de Melo <acme@...nel.org>,
	Namhyung Kim <namhyung@...nel.org>, Mark Rutland <mark.rutland@....com>,
	Alexander Shishkin <alexander.shishkin@...ux.intel.com>, Jiri Olsa
	<jolsa@...nel.org>, "Hunter, Adrian" <adrian.hunter@...el.com>, Kan Liang
	<kan.liang@...ux.intel.com>, James Clark <james.clark@...aro.org>,
	"linux-perf-users@...r.kernel.org" <linux-perf-users@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH v1] perf intel-tpebs: Filter non-workload samples



> -----Original Message-----
> From: Ian Rogers <irogers@...gle.com>
> Sent: Wednesday, April 30, 2025 1:01 PM
> To: Peter Zijlstra <peterz@...radead.org>; Ingo Molnar <mingo@...hat.com>;
> Arnaldo Carvalho de Melo <acme@...nel.org>; Namhyung Kim
> <namhyung@...nel.org>; Mark Rutland <mark.rutland@....com>;
> Alexander Shishkin <alexander.shishkin@...ux.intel.com>; Jiri Olsa
> <jolsa@...nel.org>; Ian Rogers <irogers@...gle.com>; Hunter, Adrian
> <adrian.hunter@...el.com>; Kan Liang <kan.liang@...ux.intel.com>; Wang,
> Weilin <weilin.wang@...el.com>; James Clark <james.clark@...aro.org>; linux-
> perf-users@...r.kernel.org; linux-kernel@...r.kernel.org
> Subject: [PATCH v1] perf intel-tpebs: Filter non-workload samples
> 
> If perf is running with a benchmark then we want the retirement
> latency samples associated with the benchmark rather than from the
> system as a whole. Use the workload's PID to filter out samples that
> aren't from the workload or its children.
> 
> Signed-off-by: Ian Rogers <irogers@...gle.com>
Tested-by: Weilin Wang <weilin.wang@...el.com>

> ---
>  tools/perf/util/intel-tpebs.c | 59
> ++++++++++++++++++++++++++++++++++-
>  1 file changed, 58 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/perf/util/intel-tpebs.c b/tools/perf/util/intel-tpebs.c
> index 7fd6cae1063e..4ad4bc118ea5 100644
> --- a/tools/perf/util/intel-tpebs.c
> +++ b/tools/perf/util/intel-tpebs.c
> @@ -3,7 +3,7 @@
>   * intel_tpebs.c: Intel TPEBS support
>   */
> 
> -
> +#include <api/fs/fs.h>
>  #include <sys/param.h>
>  #include <subcmd/run-command.h>
>  #include <thread.h>
> @@ -121,6 +121,59 @@ static int evsel__tpebs_start_perf_record(struct
> evsel *evsel)
>  	return ret;
>  }
> 
> +static bool is_child_pid(pid_t parent, pid_t child)
> +{
> +	if (parent < 0 || child < 0)
> +		return false;
> +
> +	while (true) {
> +		char path[PATH_MAX];
> +		char line[256];
> +		FILE *fp;
> +
> +new_child:
> +		if (parent == child)
> +			return true;
> +
> +		if (child <= 0)
> +			return false;
> +
> +		scnprintf(path, sizeof(path), "%s/%d/status",
> procfs__mountpoint(), child);
> +		fp = fopen(path, "r");
> +		if (!fp) {
> +			/* Presumably the process went away. Assume not a
> child. */
> +			return false;
> +		}
> +		while (fgets(line, sizeof(line), fp) != NULL) {
> +			if (strncmp(line, "PPid:", 5) == 0) {
> +				fclose(fp);
> +				if (sscanf(line + 5, "%d", &child) != 1) {
> +					/* Unexpected error parsing. */
> +					return false;
> +				}
> +				goto new_child;
> +			}
> +		}
> +		/* Unexpected EOF. */
> +		fclose(fp);
> +		return false;
> +	}
> +}
> +
> +static bool should_ignore_sample(const struct perf_sample *sample, const
> struct tpebs_retire_lat *t)
> +{
> +	pid_t workload_pid = t->evsel->evlist->workload.pid;
> +	pid_t sample_pid = sample->pid;
> +
> +	if (workload_pid < 0 || workload_pid == sample_pid)
> +		return false;
> +
> +	if (!t->evsel->core.attr.inherit)
> +		return true;
> +
> +	return !is_child_pid(workload_pid, sample_pid);
> +}
> +
>  static int process_sample_event(const struct perf_tool *tool
> __maybe_unused,
>  				union perf_event *event __maybe_unused,
>  				struct perf_sample *sample,
> @@ -140,6 +193,10 @@ static int process_sample_event(const struct
> perf_tool *tool __maybe_unused,
>  		mutex_unlock(tpebs_mtx_get());
>  		return -EINVAL;
>  	}
> +	if (should_ignore_sample(sample, t)) {
> +		mutex_unlock(tpebs_mtx_get());
> +		return 0;
> +	}
>  	/*
>  	 * Need to handle per core results? We are assuming average retire
>  	 * latency value will be used. Save the number of samples and the sum
> of
> --
> 2.49.0.906.g1f30a19c02-goog