Commit c410431c authored by Arnaldo Carvalho de Melo's avatar Arnaldo Carvalho de Melo
Browse files

perf tools: Reconstruct event with modifiers from perf_event_attr 

The modifiers:

  k		kernel space
  u		user space
  h		hypervisor
  G		guest
  H		host
  p, pp, ppp    precision level (PEBS)

that can be suffixed to an event were lost when tools used event_name()
to reconstruct them from the perf_event_attr entries in a perf.data
file.

Fix it by following the defaults used for these modifiers in the current
codebase, so:

 $ perf record -e instructions:u usleep 1 2> /dev/null
 $ perf evlist
 instructions:u
 $ perf record -e cycles:k usleep 1 2> /dev/null
 $ perf evlist
 cycles:k
 $ perf record -e cycles:kh usleep 1 2> /dev/null
 $ perf evlist
 cycles:kh
 $ perf record -e cache-misses:G usleep 1 2> /dev/null
 $ perf evlist
 cache-misses:G
 $ perf record -e cycles:ppk usleep 1 2> /dev/null
 $ perf evlist
 cycles:kpp
 $

Also works with 'top', 'report', etc.

More work needed to cover tracepoints and software events while not
dragging lots of baggage to the python binding, this is a minimal fix
for v3.5.

Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-4hl5glle0hxlklw4usva1mkt@git.kernel.org


Signed-off-by: default avatarArnaldo Carvalho de Melo <acme@redhat.com>
parent 895d9766

tools/perf/util/evsel.c

+90 −0
Original line number Diff line number Diff line
@@ -15,6 +15,7 @@ 
#include "cpumap.h"

#include "thread_map.h"

#include "target.h"

#include "../../include/linux/perf_event.h"



#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))

#define GROUP_FD(group_fd, cpu) (*(int *)xyarray__entry(group_fd, cpu, 0))
@@ -64,6 +65,95 @@ struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx) 
	return evsel;

}



static const char *perf_evsel__hw_names[PERF_COUNT_HW_MAX] = {

	"cycles",

	"instructions",

	"cache-references",

	"cache-misses",

	"branches",

	"branch-misses",

	"bus-cycles",

	"stalled-cycles-frontend",

	"stalled-cycles-backend",

	"ref-cycles",

};



const char *__perf_evsel__hw_name(u64 config)

{

	if (config < PERF_COUNT_HW_MAX && perf_evsel__hw_names[config])

		return perf_evsel__hw_names[config];



	return "unknown-hardware";

}



static int perf_evsel__hw_name(struct perf_evsel *evsel, char *bf, size_t size)

{

	int colon = 0;

	struct perf_event_attr *attr = &evsel->attr;

	int r = scnprintf(bf, size, "%s", __perf_evsel__hw_name(attr->config));

	bool exclude_guest_default = false;



#define MOD_PRINT(context, mod)	do {					\

		if (!attr->exclude_##context) {				\

			if (!colon) colon = r++;			\

			r += scnprintf(bf + r, size - r, "%c", mod);	\

		} } while(0)



	if (attr->exclude_kernel || attr->exclude_user || attr->exclude_hv) {

		MOD_PRINT(kernel, 'k');

		MOD_PRINT(user, 'u');

		MOD_PRINT(hv, 'h');

		exclude_guest_default = true;

	}



	if (attr->precise_ip) {

		if (!colon)

			colon = r++;

		r += scnprintf(bf + r, size - r, "%.*s", attr->precise_ip, "ppp");

		exclude_guest_default = true;

	}



	if (attr->exclude_host || attr->exclude_guest == exclude_guest_default) {

		MOD_PRINT(host, 'H');

		MOD_PRINT(guest, 'G');

	}

#undef MOD_PRINT

	if (colon)

		bf[colon] = ':';

	return r;

}



int perf_evsel__name(struct perf_evsel *evsel, char *bf, size_t size)

{

	int ret;



	switch (evsel->attr.type) {

	case PERF_TYPE_RAW:

		ret = scnprintf(bf, size, "raw 0x%" PRIx64, evsel->attr.config);

		break;



	case PERF_TYPE_HARDWARE:

		ret = perf_evsel__hw_name(evsel, bf, size);

		break;

	default:

		/*

		 * FIXME

 		 *

		 * This is the minimal perf_evsel__name so that we can

		 * reconstruct event names taking into account event modifiers.

		 *

		 * The old event_name uses it now for raw anr hw events, so that

		 * we don't drag all the parsing stuff into the python binding.

		 *

		 * On the next devel cycle the rest of the event naming will be

		 * brought here.

 		 */

		return 0;

	}



	return ret;

}



void perf_evsel__config(struct perf_evsel *evsel, struct perf_record_opts *opts,

			struct perf_evsel *first)

{

tools/perf/util/evsel.h

+3 −0
Original line number Diff line number Diff line
@@ -83,6 +83,9 @@ void perf_evsel__config(struct perf_evsel *evsel, 
			struct perf_record_opts *opts,

			struct perf_evsel *first);



const char* __perf_evsel__hw_name(u64 config);

int perf_evsel__name(struct perf_evsel *evsel, char *bf, size_t size);



int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads);

int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads);

int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus);

tools/perf/util/parse-events.c

+11 −16
Original line number Diff line number Diff line
@@ -62,19 +62,6 @@ static struct event_symbol event_symbols[] = { 
#define PERF_EVENT_TYPE(config)		__PERF_EVENT_FIELD(config, TYPE)

#define PERF_EVENT_ID(config)		__PERF_EVENT_FIELD(config, EVENT)



static const char *hw_event_names[PERF_COUNT_HW_MAX] = {

	"cycles",

	"instructions",

	"cache-references",

	"cache-misses",

	"branches",

	"branch-misses",

	"bus-cycles",

	"stalled-cycles-frontend",

	"stalled-cycles-backend",

	"ref-cycles",

};



static const char *sw_event_names[PERF_COUNT_SW_MAX] = {

	"cpu-clock",

	"task-clock",
@@ -300,6 +287,16 @@ const char *event_name(struct perf_evsel *evsel) 
	u64 config = evsel->attr.config;

	int type = evsel->attr.type;



	if (type == PERF_TYPE_RAW || type == PERF_TYPE_HARDWARE) {

		/*

 		 * XXX minimal fix, see comment on perf_evsen__name, this static buffer

 		 * will go away together with event_name in the next devel cycle.

 		 */

		static char bf[128];

		perf_evsel__name(evsel, bf, sizeof(bf));

		return bf;

	}



	if (evsel->name)

		return evsel->name;
@@ -317,9 +314,7 @@ const char *__event_name(int type, u64 config) 


	switch (type) {

	case PERF_TYPE_HARDWARE:

		if (config < PERF_COUNT_HW_MAX && hw_event_names[config])

			return hw_event_names[config];

		return "unknown-hardware";

		return __perf_evsel__hw_name(config);



	case PERF_TYPE_HW_CACHE: {

		u8 cache_type, cache_op, cache_result;

CRA Git | Maintained and supported by SUSTech CRA and CCSE