Em Mon, Jun 25, 2018 at 02:58:56PM +0200, Thomas Richter escreveu:
> perf stat command line flag -T to display transaction counters is
> currently supported for x86 only.
> Add support for s390. It is based on the metrics flag -M transaction
> using the architecture dependend json files. This requires a metric
> named "transaction" in the json files for the platform.
You forgot to CC Andi in this post, Andi, is everything ok now? May I
have your reviewed-by or acked-by?
Thanks,
- Arnaldo
> v2 --> v3:
> Introduce new function metricgroup__has_metric() to check for
> the existance of a metric named transaction.
>
> As suggested by Andi Kleen, this is the new approach to support
> transactions counters. Other architectures will follow.
>
> Output before:
> [root@p23lp27 perf]# ./perf stat -T -- sleep 1
> Cannot set up transaction events
> [root@p23lp27 perf]#
>
> Output after:
> [root@s35lp76 perf]# ./perf stat -T -- ~/mytesttx 1 >/tmp/111
>
> Performance counter stats for '/root/mytesttx 1':
>
> 1 tx_c_tend # 13.0 transaction
> 1 tx_nc_tend
> 11 tx_nc_tabort
> 0 tx_c_tabort_special
> 0 tx_c_tabort_no_special
>
> 0.001070109 seconds time elapsed
>
> [root@s35lp76 perf]#
>
> Suggested-by: Andi Kleen <a...@linux.intel.com>
> Signed-off-by: Thomas Richter <tmri...@linux.ibm.com>
> Suggested-by: Hendrik Brueckner <brueck...@linux.ibm.com>
> Reviewed-by: Hendrik Brueckner <brueck...@linux.ibm.com>
> ---
> tools/perf/builtin-stat.c | 12 ++++++++++++
> tools/perf/util/metricgroup.c | 23 +++++++++++++++++++++++
> tools/perf/util/metricgroup.h | 1 +
> 3 files changed, 36 insertions(+)
>
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index a4f662a462c6..e309279e7b04 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -2425,6 +2425,18 @@ static int add_default_attributes(void)
> if (transaction_run) {
> struct parse_events_error errinfo;
>
> + /* Handle -T as -M transaction. Once platform specific metrics
> + * support has been added to the json files, all archictures
> + * will use this approach. To determine transaction support
> + * on an architecture test for such a metric name.
> + */
> + if (metricgroup__has_metric("transaction")) {
> + struct option opt = { .value = &evsel_list };
> +
> + return metricgroup__parse_groups(&opt, "transaction",
> + &metric_events);
> + }
> +
> if (pmu_have_event("cpu", "cycles-ct") &&
> pmu_have_event("cpu", "el-start"))
> err = parse_events(evsel_list, transaction_attrs,
> diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c
> index 1ddc3d1d0147..90cff5226ecc 100644
> --- a/tools/perf/util/metricgroup.c
> +++ b/tools/perf/util/metricgroup.c
> @@ -490,3 +490,26 @@ int metricgroup__parse_groups(const struct option *opt,
> metricgroup__free_egroups(&group_list);
> return ret;
> }
> +
> +bool metricgroup__has_metric(const char *metric)
> +{
> + struct pmu_events_map *map = perf_pmu__find_map(NULL);
> + struct pmu_event *pe;
> + int i;
> +
> + if (!map)
> + return false;
> +
> + for (i = 0; ; i++) {
> + pe = &map->table[i];
> +
> + if (!pe->name && !pe->metric_group && !pe->metric_name)
> + break;
> + if (!pe->metric_expr)
> + continue;
> + if (match_metric(pe->metric_group, metric) ||
> + match_metric(pe->metric_name, metric))
> + return true;
> + }
> + return false;
> +}
> diff --git a/tools/perf/util/metricgroup.h b/tools/perf/util/metricgroup.h
> index 06854e125ee7..8a155dba0581 100644
> --- a/tools/perf/util/metricgroup.h
> +++ b/tools/perf/util/metricgroup.h
> @@ -28,4 +28,5 @@ int metricgroup__parse_groups(const struct option *opt,
> struct rblist *metric_events);
>
> void metricgroup__print(bool metrics, bool groups, char *filter, bool raw);
> +bool metricgroup__has_metric(const char *metric);
> #endif
> --
> 2.14.3
