From mboxrd@z Thu Jan 1 00:00:00 1970 Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.remote-ip=209.85.210.71 ARC-Seal: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; cv=none; t=1794421935; b=hH0VfsSIOOduny5mfITRCX8MLDmym21lqt7Y9Am967almFgLfKZsT5GpBOOPmDCXODav NpO6WM+0VkZs7dTy2ABhGA5IWi7fXXcsa3yen97iwUd4Pia3jODEu8+PoH/vJBS2N4tZY 8GuVfTtqCU+0BPUvezovPTBHi/65CXC3v/oqqCzqmZigp06vCj99jgje9Rpfb13jNO1w7 jKI1kKb+Pxc2dMLFT7zeB1reab0KEWh+r75aW5An7H6R96kPIIu6fXx77PHpimftYtLr/ TdssWepEzxvs6DYMELm4h3UlIZwbsVtxpSLAdFEBTTXaC01J6bc4t772288VP3AD7tA== ARC-Message-Signature: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; c=relaxed/relaxed; t=1794421935; h=DMARC-Filter:Received:DKIM-Signature:X-Google-DKIM-Signature: X-Forwarded-Encrypted:X-Gm-Message-State:X-Received:Date:Mime-Version: X-Mailer:Message-ID:Subject:From:To:Cc:Content-Type: Content-Transfer-Encoding; bh=vZmzfUzzweDILEhd2+M+U3xfy6iAuTlSFuI18jyrNso=; b=WKvDSsJdAbgBOykKEDS46DxqsSl/WyJF1wvY49zmw2/c0PKnhNvhBuVI0Iy6XaxIntwW waZoGJ3HRzMvEgxVNWQO4tWCEgGCspjv1zpHYczp4ZlODFfVSAobJgrcksm/uOXidQrIi +A1ESNt0XWhVPJdW86Wq8CcrfGdYlGt8BILAL6ySiluUtPCdDH5bOiOla87oQFe+rSkAz 1S0ptIdStt9CkN6BE5HUAkM6xn78O2P3X+rjooBewdvuCMOZTScd7IOcOxiCj6cUeyUCs eDeOQ6GsZAqRL66UrVuwZEfjyyA1edmmFUFnsNaEA19CLylL2AHFmzNhD+QF6x6IBmA== ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass header.from=google.com; spf=pass smtp.mailfrom=flex--ctshao.bounces.google.com; dkim=pass header.d=google.com header.i=@google.com header.a=rsa-sha256 header.s=20260707 header.b=dS73KLWm; arc=none smtp.remote-ip=209.85.210.71 Received: from mail-ot1-f71.google.com (mail-ot1-f71.google.com [209.85.210.71]) (using TLSv1.3 with cipher TLS_AES_128_GCM_SHA256 (128/128 bits) key-exchange X25519 server-signature ECDSA (prime256v1) server-digest SHA256) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 6710C606A324 for ; Wed, 11 Nov 2026 18:32:14 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.4.2 smtp.subspace.kernel.org 6710C606A324 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--ctshao.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.a=rsa-sha256 header.s=20260707 header.b=dS73KLWm Received: by mail-ot1-f71.google.com with SMTP id 46e09a7af769-85c1135c289so2591595a34.3 for ; Wed, 11 Nov 2026 10:32:14 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20260707; t=1794421933; x=1795026733; darn=vger.kernel.org; h=content-transfer-encoding:content-type:cc:to:from:subject :message-id:mime-version:date:from:to:cc:subject:date:message-id :reply-to:content-type; bh=vZmzfUzzweDILEhd2+M+U3xfy6iAuTlSFuI18jyrNso=; b=dS73KLWm8Cj1YP2E11u/jG6biyWe+l4V1QVgIA+7xF2dJbbm+iX38lzP2aqChQCxw2 amuaAtY1lQWALElhJyn5xKZfOOy1tWLYPW9jT0jZJNkMysb/vSPTDuQggX7JnHKlKjka JmHIa9ysIusPvvyEPRwB8NUwI0CsXnPb+GWFMxUifkknyloPFsqUNC1JqmoAamo8+Nwb RJAwRDuygNKYAkkRBdaQPlgBmUX6BZrY39oxOPrkJZBabtMJy3adfMlnqJ5Tg2qpZN7F qyQkZJgNclY4ZvKt4eO942ua+E8hhxyvIjiI19IKfyj4S79HtZfmLNc75TCJCseVpKbE lJzg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20260707; t=1794421933; x=1795026733; h=content-transfer-encoding:content-type:cc:to:from:subject :message-id:mime-version:date:x-gm-message-state:from:to:cc:subject :date:message-id:reply-to:content-type; bh=vZmzfUzzweDILEhd2+M+U3xfy6iAuTlSFuI18jyrNso=; b=r2UiK0cjinH6NEov7gYfc8p2Ejv3ahFU6AbI03dYc3DH7eFzaFUOy0JTrdmJCLtmKT a3pTt9+QK8im9l4i1EMvtcXaPIErYVd0qcAMt028foFnj0XXBM6VlLsQr1uxdTaC6OaR aZcGeX96H5cx3930mLEjH/fQ3HvYC71rFY6nzSJNdKKWK4CptRSEMvDIvJXE0B7Izp2j gHSNuIYOnCxhPNCYIUVB9C7pUoDJvXbkjxyi7Hm+oHNeGH/k/cb8u3H8CZhVfCAY21TT MIAOA0JaPlexcJPmdIRp51a/zUhOWrlxBBjVrrTf3PacNvxMSBoH42EelA8RNq8fVtNC 2RJw== X-Forwarded-Encrypted: i=1; AL+hErGKrcIko9vNaY17SX9umsBlZNwVYXRCbnwJw/DSyK2twCquFkdz1UigdJKK5wtnTLzkP6mY24c2LdNgzuI=@vger.kernel.org X-Gm-Message-State: APuU4Z/WiCpDqE7zoFJ6d02MgSyL8JMUGI+Rc8jvlUjytqwVONUgm96q K2YgJ9ynZzaHlTh/9h04oWSPanYAkLnggnjCZC+bbRp6c0uwE5JWAIWHMx2p24/lY9tbSiBbqJ5 FJLWcKw== X-Received: from jahi21.prod.google.com ([2002:a05:6638:8b55:b0:615:c1f0:a936]) (user=ctshao job=prod-delivery.src-stubby-dispatcher) by 2002:a05:6808:5390:b0:4c3:e93e:e331 with SMTP id 5614622812f47-56b967c2a3fmr2150104b6e.30.1794421932812; Wed, 11 Nov 2026 10:32:12 -0800 (PST) Date: Wed, 11 Nov 2026 10:32:09 -0800 X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Post: List-Help: List-Subscribe: List-Unsubscribe: List-Owner: Precedence: list Mime-Version: 1.0 X-Mailer: git-send-email 2.56.0.567.gdb80ae8cd5-goog Message-ID: <20261111183209.1892421-1-ctshao@google.com> Subject: [PATCH v8] perf stat: Include PMU name and split uncore events per PMU in metric-only JSON output From: Chun-Tse Shao To: Peter Zijlstra , Ingo Molnar , Arnaldo Carvalho de Melo , Namhyung Kim Cc: Mark Rutland , Alexander Shishkin , Jiri Olsa , Ian Rogers , Adrian Hunter , James Clark , Hang Zhengyang , linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org, Chun-Tse Shao Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable When running `perf stat` with `-A` (--no-aggr) and `--metric-only` in JSON output mode (`-j`), `perf stat` evaluates metric expressions across all matching PMUs (including uncore PMUs like `uncore_iio_0`, `uncore_iio_1`, etc.). However, `perf stat` previously formatted JSON output by printing only "cpu" : "" and grouping all metric values on a single line per CPU without identifying which PMU instance evaluated each metric. As a result, when an uncore event spans multiple PMU boxes, `perf stat` printed repeated, ambiguous metric keys without PMU names. Fix this by: 1. Including "pmu" : "" in print_aggr_id_json in metric-only AGGR_NONE mode (-A) when evsel->pmu is a non-core hardware or hybrid PMU. Software and tool events don't belong to a hardware PMU instance and so are not labeled. 2. Starting a new JSON metric line in AGGR_NONE mode (-A) whenever the PMU name changes, so that a line only has metrics from the PMU it is labeled with. 3. Opening a metric line only for events that have a metric, as other events print nothing in metric-only mode. This avoids spurious lines like {"cpu" : "0", "pmu" : "tool"} for events like user_time, while metrics on tool events, like tma_info_system_time (duration_time), are still printed with the line prefix. 4. Updating perf_json_output_lint.py to recognize the new "pmu" key in the JSON test suite. Before the fix: $ perf stat -M iio_bandwidth_read -a -A --metric-only -j -I 1000 {"interval" : 1.001017947, "cpu" : "0", "MB/s iio_bandwidth_read" : "0.0= ", "MB/s iio_bandwidth_read" : "0.0", "MB/s iio_bandwidth_read" : "22.5",= "MB/s iio_bandwidth_read" : "0.0", "MB/s iio_bandwidth_read" : "0.2", "M= B/s iio_bandwidth_read" : "0.0", "MB/s iio_bandwidth_read" : "0.0", "MB/s= iio_bandwidth_read" : "0.0", "MB/s iio_bandwidth_read" : "0.0", "MB/s i= io_bandwidth_read" : "0.0"} There is no way to determine which uncore device generated the metrics. After: $ perf stat -M iio_bandwidth_read -a -A --metric-only -j -I 1000 {"interval" : 1.000314908, "cpu" : "0", "pmu" : "uncore_iio_0", "MB/s ii= o_bandwidth_read" : "0.0"} {"interval" : 1.000314908, "cpu" : "0", "pmu" : "uncore_iio_1", "MB/s ii= o_bandwidth_read" : "0.1"} {"interval" : 1.000314908, "cpu" : "0", "pmu" : "uncore_iio_11", "MB/s i= io_bandwidth_read" : "0.0"} ... {"interval" : 1.000314908, "cpu" : "56", "pmu" : "uncore_iio_0", "MB/s i= io_bandwidth_read" : "0.0"} {"interval" : 1.000314908, "cpu" : "56", "pmu" : "uncore_iio_1", "MB/s i= io_bandwidth_read" : "0.0"} {"interval" : 1.000314908, "cpu" : "56", "pmu" : "uncore_iio_11", "MB/s = iio_bandwidth_read" : "0.0"} Assisted-by: Gemini:gemini-3.1-pro-preview Signed-off-by: Chun-Tse Shao --- v8: - Gate "pmu" output in print_aggr_id_json() on metric-only mode and skip software PMUs via perf_pmu__is_software(), keeping plain `perf stat -a -A -j` unchanged and avoiding splitting default metrics onto a separate "software" line. (Hang Zhengyang) - Rebase onto the latest perf-tools-next. v7: lore.kernel.org/20261030192537.3079368-1-ctshao@google.com - Pass the evsel to print_metric_begin() explicitly from all callers instead of falling back to evlist__first(). (Ian) v6: lore.kernel.org/20261027203115.2647914-1-ctshao@google.com - Open metric lines for events that have a metric instead of skipping tool events, so metrics on tool events (e.g. tma_info_system_time) keep the line prefix and are terminated properly. (Sashiko) - Split JSON lines only when the shown PMU name changes. Tool event metrics are no longer mislabeled with an uncore PMU name, and the STD and CSV output are no longer split. - Rebase onto the latest perf-tools-next. v5: lore.kernel.org/20260811230518.3360-2-ctshao@google.com - Fix perf test "perf stat JSON output linter" failure. v4: lore.kernel.org/20260804210659.112192-1-ctshao@google.com - Fix the line splitting issue, causing spurious empty {"pmu": "tool"} objects and order-dependent PMU mislabeling. v3: lore.kernel.org/20260729170416.41904-1-ctshao@google.com - Add before the fix output. v2: lore.kernel.org/20260715203055.2981363-1-ctshao@google.com - Fix print_metric_begin() initialization in print_no_aggr_metric() to ensure tool events always open lines cleanly. .../tests/shell/lib/perf_json_output_lint.py | 1 + tools/perf/util/stat-display.c | 79 ++++++++++++++++--- 2 files changed, 68 insertions(+), 12 deletions(-) diff --git a/tools/perf/tests/shell/lib/perf_json_output_lint.py b/tools/pe= rf/tests/shell/lib/perf_json_output_lint.py index 8cde22ac9d97..0a16cae92766 100644 --- a/tools/perf/tests/shell/lib/perf_json_output_lint.py +++ b/tools/perf/tests/shell/lib/perf_json_output_lint.py @@ -65,6 +65,7 @@ def check_json_output(expected_count): 'metricgroup': lambda x: True, 'node': lambda x: True, 'pcnt-running': isfloat, + 'pmu': lambda x: True, 'socket': lambda x: True, 'thread': lambda x: True, 'unit': lambda x: True, diff --git a/tools/perf/util/stat-display.c b/tools/perf/util/stat-display.= c index b337cc23f413..6c9358b64b38 100644 --- a/tools/perf/util/stat-display.c +++ b/tools/perf/util/stat-display.c @@ -1,6 +1,7 @@ #include #include #include +#include #include #include #include @@ -23,6 +24,7 @@ #include "pmu.h" #include "pmus.h" #include "tool_pmu.h" +#include "metricgroup.h" =20 #define CNTR_NOT_SUPPORTED "" #define CNTR_NOT_COUNTED "" @@ -361,9 +363,34 @@ static void print_aggr_id_csv(struct perf_stat_config = *config, } } =20 +/* + * In no-aggregation metric-only mode, JSON output includes the PMU name o= f + * non-core hardware and hybrid events so that metrics from different PMU + * instances, like uncore_iio_0 and uncore_iio_1, can be told apart. Softw= are + * and tool events don't belong to a hardware PMU instance and so are not + * labeled. + */ +static const char *json_pmu_name(const struct evsel *evsel) +{ + if (!evsel || !evsel->pmu || perf_pmu__is_software(evsel->pmu)) + return NULL; + if (evsel->pmu->is_core && !evsel__is_hybrid(evsel)) + return NULL; + return evsel->pmu->name; +} + +static bool json_pmu_name_changed(const char *name, const char *prev) +{ + if (!name || !prev) + return name !=3D prev; + return strcmp(name, prev) !=3D 0; +} + static void print_aggr_id_json(struct perf_stat_config *config, struct out= state *os, struct evsel *evsel, struct aggr_cpu_id id, int aggr_nr) { + const char *pmu_name; + switch (config->aggr_mode) { case AGGR_CORE: json_out(os, "\"core\" : \"S%d-D%d-C%d\", \"counters\" : %d", @@ -397,6 +424,11 @@ static void print_aggr_id_json(struct perf_stat_config= *config, struct outstate json_out(os, "\"cpu\" : \"%d\"", id.cpu.cpu); } + if (config->metric_only) { + pmu_name =3D json_pmu_name(evsel); + if (pmu_name) + json_out(os, "\"pmu\" : \"%s\"", pmu_name); + } break; case AGGR_THREAD: json_out(os, "\"thread\" : \"%s-%d\"", @@ -1010,13 +1042,16 @@ static void print_counter_aggrdata(struct perf_stat= _config *config, fputc('\n', output); } =20 +/* + * Start a metric-only line. @evsel only provides the line prefix: the + * aggregation id, the PMU name in JSON (see json_pmu_name()) and the cgro= up. + */ static void print_metric_begin(struct perf_stat_config *config, - struct evlist *evlist, + struct evsel *evsel, struct outstate *os, int aggr_idx) { struct perf_stat_aggr *aggr; struct aggr_cpu_id id; - struct evsel *evsel; =20 os->first =3D true; if (!config->metric_only) @@ -1031,7 +1066,6 @@ static void print_metric_begin(struct perf_stat_confi= g *config, else fprintf(config->output, "%s", os->timestamp); } - evsel =3D evlist__first(evlist); id =3D config->aggr_map->map[aggr_idx]; aggr =3D &evsel->stats->aggr[aggr_idx]; aggr_printout(config, os, evsel, id, aggr->nr); @@ -1069,7 +1103,7 @@ static void print_aggr(struct perf_stat_config *confi= g, * Without each counter has its own line. */ cpu_aggr_map__for_each_idx(aggr_idx, config->aggr_map) { - print_metric_begin(config, evlist, os, aggr_idx); + print_metric_begin(config, evlist__first(evlist), os, aggr_idx); =20 evlist__for_each_entry(evlist, counter) { print_counter_aggrdata(config, counter, aggr_idx, os); @@ -1095,7 +1129,7 @@ static void print_aggr_cgroup(struct perf_stat_config= *config, os->cgrp =3D evsel->cgrp; =20 cpu_aggr_map__for_each_idx(aggr_idx, config->aggr_map) { - print_metric_begin(config, evlist, os, aggr_idx); + print_metric_begin(config, evsel, os, aggr_idx); =20 evlist__for_each_entry(evlist, counter) { if (counter->cgrp !=3D os->cgrp) @@ -1131,7 +1165,8 @@ static void print_no_aggr_metric(struct perf_stat_con= fig *config, =20 perf_cpu_map__for_each_cpu(cpu, all_idx, evlist__core(evlist)->user_reque= sted_cpus) { struct evsel *counter; - bool first =3D true; + const char *line_pmu =3D NULL; + bool line_open =3D false; =20 evlist__for_each_entry(evlist, counter) { u64 ena, run, val; @@ -1146,12 +1181,32 @@ static void print_no_aggr_metric(struct perf_stat_c= onfig *config, if (config->aggr_map->map[aggr_idx].cpu.cpu =3D=3D cpu.cpu) break; } + if (aggr_idx >=3D config->aggr_map->nr) + continue; =20 os->evsel =3D counter; os->id =3D aggr_cpu_id__cpu(cpu, /*data=3D*/NULL); - if (first) { - print_metric_begin(config, evlist, os, aggr_idx); - first =3D false; + + /* + * Only events with a metric print something in metric-only + * mode, so let the first of them open the line. For JSON, + * start a new line when the PMU name changes so that all + * metrics on a line share the same "cpu" and "pmu" prefix. + */ + if (config->iostat_run || + metricgroup__lookup(evlist__metric_events(evlist), counter, false))= { + const char *pmu_name =3D json_pmu_name(counter); + + if (line_open && config->json_output && + json_pmu_name_changed(pmu_name, line_pmu)) { + print_metric_end(config, os); + line_open =3D false; + } + if (!line_open) { + print_metric_begin(config, counter, os, aggr_idx); + line_pmu =3D pmu_name; + line_open =3D true; + } } val =3D ps->aggr[aggr_idx].counts.val; ena =3D ps->aggr[aggr_idx].counts.ena; @@ -1160,7 +1215,7 @@ static void print_no_aggr_metric(struct perf_stat_con= fig *config, uval =3D val * counter->scale; printout(config, os, uval, run, ena, 1.0, aggr_idx); } - if (!first) + if (line_open) print_metric_end(config, os); } } @@ -1523,7 +1578,7 @@ static void print_cgroup_counter(struct perf_stat_con= fig *config, struct evlist print_metric_end(config, os); =20 os->cgrp =3D counter->cgrp; - print_metric_begin(config, evlist, os, /*aggr_idx=3D*/0); + print_metric_begin(config, counter, os, /*aggr_idx=3D*/0); } =20 print_counter(config, counter, os); @@ -1573,7 +1628,7 @@ void evlist__print_counters(struct evlist *evlist, st= ruct perf_stat_config *conf } else if (config->cgroup_list) { print_cgroup_counter(config, evlist, &os); } else { - print_metric_begin(config, evlist, &os, /*aggr_idx=3D*/0); + print_metric_begin(config, evlist__first(evlist), &os, /*aggr_idx=3D*/0= ); evlist__for_each_entry(evlist, counter) { print_counter(config, counter, &os); } --=20 2.56.0.542.g3ef89b7020-goog