Re: [PATCH v1 12/15] perf tool_pmu: Switch to standard pmu functions and json descriptions

[Namhyung Kim <namhyung@kernel.org>]

On Fri, Sep 06, 2024 at 10:08:27PM -0700, Ian Rogers wrote:
> Use the regular PMU approaches with tool json events to reduce the
> amount of special tool_pmu code - tool_pmu__config_terms and
> tool_pmu__for_each_event_cb are removed. Some functions remain, like
> tool_pmu__str_to_event, as conveniences to metricgroups. Add
> tool_pmu__skip_event/tool_pmu__num_skip_events to handle the case that
> tool json events shouldn't appear on certain architectures. This isn't
> done in jevents.py due to complexity in the empty-pmu-events.c and
> when all vendor json is built into the tool.
> 
> Signed-off-by: Ian Rogers <irogers@google.com>
> ---
>  tools/perf/util/pmu.c      |  26 ++++-----
>  tools/perf/util/pmus.c     |   4 +-
>  tools/perf/util/tool_pmu.c | 111 ++++++++++++-------------------------
>  tools/perf/util/tool_pmu.h |  12 +---
>  4 files changed, 48 insertions(+), 105 deletions(-)
> 
> diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c
> index a412c9c62f9c..5ff6556292fd 100644
> --- a/tools/perf/util/pmu.c
> +++ b/tools/perf/util/pmu.c
> @@ -1549,9 +1549,6 @@ int perf_pmu__config(struct perf_pmu *pmu, struct perf_event_attr *attr,
>  {
>  	bool zero = !!pmu->perf_event_attr_init_default;
>  
> -	if (perf_pmu__is_tool(pmu))
> -		return tool_pmu__config_terms(attr, head_terms, err);
> -
>  	/* Fake PMU doesn't have proper terms so nothing to configure in attr. */
>  	if (perf_pmu__is_fake(pmu))
>  		return 0;
> @@ -1664,8 +1661,8 @@ int perf_pmu__check_alias(struct perf_pmu *pmu, struct parse_events_terms *head_
>  	info->scale    = 0.0;
>  	info->snapshot = false;
>  
> -	/* Tool/fake PMU doesn't rewrite terms. */
> -	if (perf_pmu__is_tool(pmu) || perf_pmu__is_fake(pmu))
> +	/* Fake PMU doesn't rewrite terms. */
> +	if (perf_pmu__is_fake(pmu))
>  		goto out;
>  
>  	list_for_each_entry_safe(term, h, &head_terms->terms, list) {
> @@ -1831,8 +1828,8 @@ bool perf_pmu__have_event(struct perf_pmu *pmu, const char *name)
>  {
>  	if (!name)
>  		return false;
> -	if (perf_pmu__is_tool(pmu))
> -		return tool_pmu__str_to_event(name) != TOOL_PMU__EVENT_NONE;
> +	if (perf_pmu__is_tool(pmu) && tool_pmu__skip_event(name))
> +		return false;
>  	if (perf_pmu__find_alias(pmu, name, /*load=*/ true) != NULL)
>  		return true;
>  	if (pmu->cpu_aliases_added || !pmu->events_table)
> @@ -1844,9 +1841,6 @@ size_t perf_pmu__num_events(struct perf_pmu *pmu)
>  {
>  	size_t nr;
>  
> -	if (perf_pmu__is_tool(pmu))
> -		return tool_pmu__num_events();
> -
>  	pmu_aliases_parse(pmu);
>  	nr = pmu->sysfs_aliases + pmu->sys_json_aliases;
>  
> @@ -1857,6 +1851,9 @@ size_t perf_pmu__num_events(struct perf_pmu *pmu)
>  	else
>  		assert(pmu->cpu_json_aliases == 0);
>  
> +	if (perf_pmu__is_tool(pmu))
> +		nr -= tool_pmu__num_skip_events();
> +
>  	return pmu->selectable ? nr + 1 : nr;
>  }
>  
> @@ -1907,15 +1904,15 @@ int perf_pmu__for_each_event(struct perf_pmu *pmu, bool skip_duplicate_pmus,
>  	int ret = 0;
>  	struct strbuf sb;
>  
> -	if (perf_pmu__is_tool(pmu))
> -		return tool_pmu__for_each_event_cb(pmu, state, cb);
> -
>  	strbuf_init(&sb, /*hint=*/ 0);
>  	pmu_aliases_parse(pmu);
>  	pmu_add_cpu_aliases(pmu);
>  	list_for_each_entry(event, &pmu->aliases, list) {
>  		size_t buf_used, pmu_name_len;
>  
> +		if (perf_pmu__is_tool(pmu) && tool_pmu__skip_event(event->name))
> +			continue;
> +
>  		info.pmu_name = event->pmu_name ?: pmu->name;
>  		pmu_name_len = pmu_deduped_name_len(pmu, info.pmu_name,
>  						    skip_duplicate_pmus);
> @@ -2321,9 +2318,6 @@ const char *perf_pmu__name_from_config(struct perf_pmu *pmu, u64 config)
>  	if (!pmu)
>  		return NULL;
>  
> -	if (perf_pmu__is_tool(pmu))
> -		return tool_pmu__event_to_str(config);
> -
>  	pmu_aliases_parse(pmu);
>  	pmu_add_cpu_aliases(pmu);
>  	list_for_each_entry(event, &pmu->aliases, list) {
> diff --git a/tools/perf/util/pmus.c b/tools/perf/util/pmus.c
> index 36329bc77316..19673b9991c6 100644
> --- a/tools/perf/util/pmus.c
> +++ b/tools/perf/util/pmus.c
> @@ -440,6 +440,7 @@ static int perf_pmus__print_pmu_events__callback(void *vstate,
>  		pr_err("Unexpected event %s/%s/\n", info->pmu->name, info->name);
>  		return 1;
>  	}
> +	assert(info->pmu != NULL || info->name != NULL);
>  	s = &state->aliases[state->index];
>  	s->pmu = info->pmu;
>  #define COPY_STR(str) s->str = info->str ? strdup(info->str) : NULL
> @@ -590,9 +591,6 @@ void perf_pmus__print_raw_pmu_events(const struct print_callbacks *print_cb, voi
>  		int len = pmu_name_len_no_suffix(pmu->name);
>  		const char *desc = "(see 'man perf-list' or 'man perf-record' on how to encode it)";
>  
> -		if (perf_pmu__is_tool(pmu))
> -			continue;
> -
>  		if (!pmu->is_core)
>  			desc = NULL;
>  
> diff --git a/tools/perf/util/tool_pmu.c b/tools/perf/util/tool_pmu.c
> index 3dc7cc7b8d32..f2d26e9e51f5 100644
> --- a/tools/perf/util/tool_pmu.c
> +++ b/tools/perf/util/tool_pmu.c
> @@ -33,101 +33,54 @@ static const char *const tool_pmu__event_names[TOOL_PMU__EVENT_MAX] = {
>  	"system_tsc_freq",
>  };
>  
> -
> -const char *tool_pmu__event_to_str(enum tool_pmu_event ev)
> -{
> -	if (ev > TOOL_PMU__EVENT_NONE && ev < TOOL_PMU__EVENT_MAX)
> -		return tool_pmu__event_names[ev];
> -
> -	return NULL;
> -}
> -
> -enum tool_pmu_event tool_pmu__str_to_event(const char *str)
> +bool tool_pmu__skip_event(const char *name)

I think you need __maybe_unused.


>  {
> -	int i;
> -
> -	tool_pmu__for_each_event(i) {
> -		if (!strcasecmp(str, tool_pmu__event_names[i])) {
>  #if !defined(__aarch64__)
> -			/* The slots event should only appear on arm64. */
> -			if (i == TOOL_PMU__EVENT_SLOTS)
> -				return TOOL_PMU__EVENT_NONE;
> +	/* The slots event should only appear on arm64. */
> +	if (strcasecmp(name, "slots") == 0)
> +		return true;
>  #endif
> -			return i;
> -		}
> -	}
> -	return TOOL_PMU__EVENT_NONE;
> +#if !defined(__i386__) && !defined(__x86_64__)
> +	/* The system_tsc_freq event should only appear on x86. */
> +	if (strcasecmp(name, "system_tsc_freq") == 0)
> +		return true;
> +#endif
> +	return false;
>  }
>  
> -static int tool_pmu__config_term(struct perf_event_attr *attr,
> -				 struct parse_events_term *term,
> -				 struct parse_events_error *err)
> +int tool_pmu__num_skip_events(void)
>  {
> -	if (term->type_term == PARSE_EVENTS__TERM_TYPE_USER) {
> -		enum tool_pmu_event ev = tool_pmu__str_to_event(term->config);
> +	int num = 0;
>  
> -		if (ev == TOOL_PMU__EVENT_NONE)
> -			goto err_out;
> -
> -		attr->config = ev;
> -		return 0;
> -	}
> -err_out:
> -	if (err) {
> -		char *err_str;
> -
> -		parse_events_error__handle(err, term->err_val,
> -					asprintf(&err_str,
> -						"unexpected tool event term (%s) %s",
> -						parse_events__term_type_str(term->type_term),
> -						term->config) < 0
> -					? strdup("unexpected tool event term")
> -					: err_str,
> -					NULL);
> -	}
> -	return -EINVAL;
> +#if !defined(__aarch64__)
> +	num++;
> +#endif
> +#if !defined(__i386__) && !defined(__x86_64__)
> +	num++;
> +#endif
> +	return num;
>  }
>  
> -int tool_pmu__config_terms(struct perf_event_attr *attr,
> -			   struct parse_events_terms *terms,
> -			   struct parse_events_error *err)
> +const char *tool_pmu__event_to_str(enum tool_pmu_event ev)
>  {
> -	struct parse_events_term *term;
> -
> -	list_for_each_entry(term, &terms->terms, list) {
> -		if (tool_pmu__config_term(attr, term, err))
> -			return -EINVAL;
> -	}
> -
> -	return 0;
> +	if (ev > TOOL_PMU__EVENT_NONE && ev < TOOL_PMU__EVENT_MAX)
> +		return tool_pmu__event_names[ev];
>  
> +	return NULL;
>  }
>  
> -int tool_pmu__for_each_event_cb(struct perf_pmu *pmu, void *state, pmu_event_callback cb)
> +enum tool_pmu_event tool_pmu__str_to_event(const char *str)
>  {
> -	struct pmu_event_info info = {
> -		.pmu = pmu,
> -		.event_type_desc = "Tool event",
> -	};
>  	int i;
>  
> +	if (tool_pmu__skip_event(str))
> +		return TOOL_PMU__EVENT_NONE;
> +
>  	tool_pmu__for_each_event(i) {
> -		int ret;
> -
> -		info.name = tool_pmu__event_to_str(i);
> -		info.alias = NULL;
> -		info.scale_unit = NULL;
> -		info.desc = NULL;
> -		info.long_desc = NULL;
> -		info.encoding_desc = NULL;
> -		info.topic = NULL;
> -		info.pmu_name = pmu->name;
> -		info.deprecated = false;
> -		ret = cb(state, &info);
> -		if (ret)
> -			return ret;
> +		if (!strcasecmp(str, tool_pmu__event_names[i]))
> +			return i;
>  	}
> -	return 0;
> +	return TOOL_PMU__EVENT_NONE;
>  }
>  
>  bool perf_pmu__is_tool(const struct perf_pmu *pmu)
> @@ -544,8 +497,12 @@ struct perf_pmu *perf_pmus__tool_pmu(void)
>  	static struct perf_pmu tool = {
>  		.name = "tool",
>  		.type = PERF_PMU_TYPE_TOOL,
> +		.aliases = LIST_HEAD_INIT(tool.aliases),
> +		.caps = LIST_HEAD_INIT(tool.caps),

This can be moved to the previous patch.

Thanks,
Namhyung


>  		.format = LIST_HEAD_INIT(tool.format),
>  	};
> +	if (!tool.events_table)
> +		tool.events_table = find_core_events_table("common", "common");
>  
>  	return &tool;
>  }
> diff --git a/tools/perf/util/tool_pmu.h b/tools/perf/util/tool_pmu.h
> index ecdf316525bb..a60184859080 100644
> --- a/tools/perf/util/tool_pmu.h
> +++ b/tools/perf/util/tool_pmu.h
> @@ -29,17 +29,11 @@ enum tool_pmu_event {
>  #define tool_pmu__for_each_event(ev)					\
>  	for ((ev) = TOOL_PMU__EVENT_DURATION_TIME; (ev) < TOOL_PMU__EVENT_MAX; ev++)
>  
> -static inline size_t tool_pmu__num_events(void)
> -{
> -	return TOOL_PMU__EVENT_MAX - 1;
> -}
> -
>  const char *tool_pmu__event_to_str(enum tool_pmu_event ev);
>  enum tool_pmu_event tool_pmu__str_to_event(const char *str);
> -int tool_pmu__config_terms(struct perf_event_attr *attr,
> -			   struct parse_events_terms *terms,
> -			   struct parse_events_error *err);
> -int tool_pmu__for_each_event_cb(struct perf_pmu *pmu, void *state, pmu_event_callback cb);
> +bool tool_pmu__skip_event(const char *name);
> +int tool_pmu__num_skip_events(void);
> +
>  bool tool_pmu__read_event(enum tool_pmu_event ev, u64 *result);
>  
>  u64 tool_pmu__cpu_slots_per_cycle(void);
> -- 
> 2.46.0.469.g59c65b2a67-goog
>