linux/tools/perf/util/pmus.c
Arnaldo Carvalho de Melo 82fe2e45cd perf pmus: Check if we can encode the PMU number in perf_event_attr.type
In some architectures we can't encode the PMU number in
perf_event_attr.type and thus can't just ask for the same event in
multiple CPUs (and thus PMUs), that is what we want in hybrid systems
but we can't when that encoding isn't understood by the kernel, such as
in ARM64's big.LITTLE.

If that is the case, fallback to the previous behaviour till we find a
better solution to have consistent output accross architectures with
hybrid CPU configurations.

Co-developed-with: Ian Rogers <irogers@google.com>
Cc: James Clark <james.clark@arm.com>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Sumanth Korikkar <sumanthk@linux.ibm.com>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: Thomas Richter <tmricht@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/linux-perf-users/ZIzYgImv61OGK1wA@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2023-06-16 19:01:34 -03:00

537 lines
13 KiB
C

// SPDX-License-Identifier: GPL-2.0
#include <linux/list.h>
#include <linux/zalloc.h>
#include <subcmd/pager.h>
#include <sys/types.h>
#include <dirent.h>
#include <pthread.h>
#include <string.h>
#include <unistd.h>
#include "debug.h"
#include "evsel.h"
#include "pmus.h"
#include "pmu.h"
#include "print-events.h"
/*
* core_pmus: A PMU belongs to core_pmus if it's name is "cpu" or it's sysfs
* directory contains "cpus" file. All PMUs belonging to core_pmus
* must have pmu->is_core=1. If there are more than one PMU in
* this list, perf interprets it as a heterogeneous platform.
* (FWIW, certain ARM platforms having heterogeneous cores uses
* homogeneous PMU, and thus they are treated as homogeneous
* platform by perf because core_pmus will have only one entry)
* other_pmus: All other PMUs which are not part of core_pmus list. It doesn't
* matter whether PMU is present per SMT-thread or outside of the
* core in the hw. For e.g., an instance of AMD ibs_fetch// and
* ibs_op// PMUs is present in each hw SMT thread, however they
* are captured under other_pmus. PMUs belonging to other_pmus
* must have pmu->is_core=0 but pmu->is_uncore could be 0 or 1.
*/
static LIST_HEAD(core_pmus);
static LIST_HEAD(other_pmus);
static bool read_sysfs_core_pmus;
static bool read_sysfs_all_pmus;
void perf_pmus__destroy(void)
{
struct perf_pmu *pmu, *tmp;
list_for_each_entry_safe(pmu, tmp, &core_pmus, list) {
list_del(&pmu->list);
perf_pmu__delete(pmu);
}
list_for_each_entry_safe(pmu, tmp, &other_pmus, list) {
list_del(&pmu->list);
perf_pmu__delete(pmu);
}
read_sysfs_core_pmus = false;
read_sysfs_all_pmus = false;
}
static struct perf_pmu *pmu_find(const char *name)
{
struct perf_pmu *pmu;
list_for_each_entry(pmu, &core_pmus, list) {
if (!strcmp(pmu->name, name) ||
(pmu->alias_name && !strcmp(pmu->alias_name, name)))
return pmu;
}
list_for_each_entry(pmu, &other_pmus, list) {
if (!strcmp(pmu->name, name) ||
(pmu->alias_name && !strcmp(pmu->alias_name, name)))
return pmu;
}
return NULL;
}
struct perf_pmu *perf_pmus__find(const char *name)
{
struct perf_pmu *pmu;
int dirfd;
bool core_pmu;
/*
* Once PMU is loaded it stays in the list,
* so we keep us from multiple reading/parsing
* the pmu format definitions.
*/
pmu = pmu_find(name);
if (pmu)
return pmu;
if (read_sysfs_all_pmus)
return NULL;
core_pmu = is_pmu_core(name);
if (core_pmu && read_sysfs_core_pmus)
return NULL;
dirfd = perf_pmu__event_source_devices_fd();
pmu = perf_pmu__lookup(core_pmu ? &core_pmus : &other_pmus, dirfd, name);
close(dirfd);
return pmu;
}
static struct perf_pmu *perf_pmu__find2(int dirfd, const char *name)
{
struct perf_pmu *pmu;
bool core_pmu;
/*
* Once PMU is loaded it stays in the list,
* so we keep us from multiple reading/parsing
* the pmu format definitions.
*/
pmu = pmu_find(name);
if (pmu)
return pmu;
if (read_sysfs_all_pmus)
return NULL;
core_pmu = is_pmu_core(name);
if (core_pmu && read_sysfs_core_pmus)
return NULL;
return perf_pmu__lookup(core_pmu ? &core_pmus : &other_pmus, dirfd, name);
}
/* Add all pmus in sysfs to pmu list: */
static void pmu_read_sysfs(bool core_only)
{
int fd;
DIR *dir;
struct dirent *dent;
if (read_sysfs_all_pmus || (core_only && read_sysfs_core_pmus))
return;
fd = perf_pmu__event_source_devices_fd();
if (fd < 0)
return;
dir = fdopendir(fd);
if (!dir)
return;
while ((dent = readdir(dir))) {
if (!strcmp(dent->d_name, ".") || !strcmp(dent->d_name, ".."))
continue;
if (core_only && !is_pmu_core(dent->d_name))
continue;
/* add to static LIST_HEAD(core_pmus) or LIST_HEAD(other_pmus): */
perf_pmu__find2(fd, dent->d_name);
}
closedir(dir);
if (core_only) {
read_sysfs_core_pmus = true;
} else {
read_sysfs_core_pmus = true;
read_sysfs_all_pmus = true;
}
}
static struct perf_pmu *__perf_pmus__find_by_type(unsigned int type)
{
struct perf_pmu *pmu;
list_for_each_entry(pmu, &core_pmus, list) {
if (pmu->type == type)
return pmu;
}
list_for_each_entry(pmu, &other_pmus, list) {
if (pmu->type == type)
return pmu;
}
return NULL;
}
struct perf_pmu *perf_pmus__find_by_type(unsigned int type)
{
struct perf_pmu *pmu = __perf_pmus__find_by_type(type);
if (pmu || read_sysfs_all_pmus)
return pmu;
pmu_read_sysfs(/*core_only=*/false);
pmu = __perf_pmus__find_by_type(type);
return pmu;
}
/*
* pmu iterator: If pmu is NULL, we start at the begin, otherwise return the
* next pmu. Returns NULL on end.
*/
struct perf_pmu *perf_pmus__scan(struct perf_pmu *pmu)
{
bool use_core_pmus = !pmu || pmu->is_core;
if (!pmu) {
pmu_read_sysfs(/*core_only=*/false);
pmu = list_prepare_entry(pmu, &core_pmus, list);
}
if (use_core_pmus) {
list_for_each_entry_continue(pmu, &core_pmus, list)
return pmu;
pmu = NULL;
pmu = list_prepare_entry(pmu, &other_pmus, list);
}
list_for_each_entry_continue(pmu, &other_pmus, list)
return pmu;
return NULL;
}
struct perf_pmu *perf_pmus__scan_core(struct perf_pmu *pmu)
{
if (!pmu) {
pmu_read_sysfs(/*core_only=*/true);
pmu = list_prepare_entry(pmu, &core_pmus, list);
}
list_for_each_entry_continue(pmu, &core_pmus, list)
return pmu;
return NULL;
}
const struct perf_pmu *perf_pmus__pmu_for_pmu_filter(const char *str)
{
struct perf_pmu *pmu = NULL;
while ((pmu = perf_pmus__scan(pmu)) != NULL) {
if (!strcmp(pmu->name, str))
return pmu;
/* Ignore "uncore_" prefix. */
if (!strncmp(pmu->name, "uncore_", 7)) {
if (!strcmp(pmu->name + 7, str))
return pmu;
}
/* Ignore "cpu_" prefix on Intel hybrid PMUs. */
if (!strncmp(pmu->name, "cpu_", 4)) {
if (!strcmp(pmu->name + 4, str))
return pmu;
}
}
return NULL;
}
int __weak perf_pmus__num_mem_pmus(void)
{
/* All core PMUs are for mem events. */
return perf_pmus__num_core_pmus();
}
/** Struct for ordering events as output in perf list. */
struct sevent {
/** PMU for event. */
const struct perf_pmu *pmu;
/**
* Optional event for name, desc, etc. If not present then this is a
* selectable PMU and the event name is shown as "//".
*/
const struct perf_pmu_alias *event;
/** Is the PMU for the CPU? */
bool is_cpu;
};
static int cmp_sevent(const void *a, const void *b)
{
const struct sevent *as = a;
const struct sevent *bs = b;
const char *a_pmu_name = NULL, *b_pmu_name = NULL;
const char *a_name = "//", *a_desc = NULL, *a_topic = "";
const char *b_name = "//", *b_desc = NULL, *b_topic = "";
int ret;
if (as->event) {
a_name = as->event->name;
a_desc = as->event->desc;
a_topic = as->event->topic ?: "";
a_pmu_name = as->event->pmu_name;
}
if (bs->event) {
b_name = bs->event->name;
b_desc = bs->event->desc;
b_topic = bs->event->topic ?: "";
b_pmu_name = bs->event->pmu_name;
}
/* Put extra events last. */
if (!!a_desc != !!b_desc)
return !!a_desc - !!b_desc;
/* Order by topics. */
ret = strcmp(a_topic, b_topic);
if (ret)
return ret;
/* Order CPU core events to be first */
if (as->is_cpu != bs->is_cpu)
return as->is_cpu ? -1 : 1;
/* Order by PMU name. */
if (as->pmu != bs->pmu) {
a_pmu_name = a_pmu_name ?: (as->pmu->name ?: "");
b_pmu_name = b_pmu_name ?: (bs->pmu->name ?: "");
ret = strcmp(a_pmu_name, b_pmu_name);
if (ret)
return ret;
}
/* Order by event name. */
return strcmp(a_name, b_name);
}
static bool pmu_alias_is_duplicate(struct sevent *alias_a,
struct sevent *alias_b)
{
const char *a_pmu_name = NULL, *b_pmu_name = NULL;
const char *a_name = "//", *b_name = "//";
if (alias_a->event) {
a_name = alias_a->event->name;
a_pmu_name = alias_a->event->pmu_name;
}
if (alias_b->event) {
b_name = alias_b->event->name;
b_pmu_name = alias_b->event->pmu_name;
}
/* Different names -> never duplicates */
if (strcmp(a_name, b_name))
return false;
/* Don't remove duplicates for different PMUs */
a_pmu_name = a_pmu_name ?: (alias_a->pmu->name ?: "");
b_pmu_name = b_pmu_name ?: (alias_b->pmu->name ?: "");
return strcmp(a_pmu_name, b_pmu_name) == 0;
}
static int sub_non_neg(int a, int b)
{
if (b > a)
return 0;
return a - b;
}
static char *format_alias(char *buf, int len, const struct perf_pmu *pmu,
const struct perf_pmu_alias *alias)
{
struct parse_events_term *term;
int used = snprintf(buf, len, "%s/%s", pmu->name, alias->name);
list_for_each_entry(term, &alias->terms, list) {
if (term->type_val == PARSE_EVENTS__TERM_TYPE_STR)
used += snprintf(buf + used, sub_non_neg(len, used),
",%s=%s", term->config,
term->val.str);
}
if (sub_non_neg(len, used) > 0) {
buf[used] = '/';
used++;
}
if (sub_non_neg(len, used) > 0) {
buf[used] = '\0';
used++;
} else
buf[len - 1] = '\0';
return buf;
}
void perf_pmus__print_pmu_events(const struct print_callbacks *print_cb, void *print_state)
{
struct perf_pmu *pmu;
struct perf_pmu_alias *event;
char buf[1024];
int printed = 0;
int len, j;
struct sevent *aliases;
pmu = NULL;
len = 0;
while ((pmu = perf_pmus__scan(pmu)) != NULL) {
list_for_each_entry(event, &pmu->aliases, list)
len++;
if (pmu->selectable)
len++;
}
aliases = zalloc(sizeof(struct sevent) * len);
if (!aliases) {
pr_err("FATAL: not enough memory to print PMU events\n");
return;
}
pmu = NULL;
j = 0;
while ((pmu = perf_pmus__scan(pmu)) != NULL) {
bool is_cpu = pmu->is_core;
list_for_each_entry(event, &pmu->aliases, list) {
aliases[j].event = event;
aliases[j].pmu = pmu;
aliases[j].is_cpu = is_cpu;
j++;
}
if (pmu->selectable) {
aliases[j].event = NULL;
aliases[j].pmu = pmu;
aliases[j].is_cpu = is_cpu;
j++;
}
}
len = j;
qsort(aliases, len, sizeof(struct sevent), cmp_sevent);
for (j = 0; j < len; j++) {
const char *name, *alias = NULL, *scale_unit = NULL,
*desc = NULL, *long_desc = NULL,
*encoding_desc = NULL, *topic = NULL,
*pmu_name = NULL;
bool deprecated = false;
size_t buf_used;
/* Skip duplicates */
if (j > 0 && pmu_alias_is_duplicate(&aliases[j], &aliases[j - 1]))
continue;
if (!aliases[j].event) {
/* A selectable event. */
pmu_name = aliases[j].pmu->name;
buf_used = snprintf(buf, sizeof(buf), "%s//", pmu_name) + 1;
name = buf;
} else {
if (aliases[j].event->desc) {
name = aliases[j].event->name;
buf_used = 0;
} else {
name = format_alias(buf, sizeof(buf), aliases[j].pmu,
aliases[j].event);
if (aliases[j].is_cpu) {
alias = name;
name = aliases[j].event->name;
}
buf_used = strlen(buf) + 1;
}
pmu_name = aliases[j].event->pmu_name ?: (aliases[j].pmu->name ?: "");
if (strlen(aliases[j].event->unit) || aliases[j].event->scale != 1.0) {
scale_unit = buf + buf_used;
buf_used += snprintf(buf + buf_used, sizeof(buf) - buf_used,
"%G%s", aliases[j].event->scale,
aliases[j].event->unit) + 1;
}
desc = aliases[j].event->desc;
long_desc = aliases[j].event->long_desc;
topic = aliases[j].event->topic;
encoding_desc = buf + buf_used;
buf_used += snprintf(buf + buf_used, sizeof(buf) - buf_used,
"%s/%s/", pmu_name, aliases[j].event->str) + 1;
deprecated = aliases[j].event->deprecated;
}
print_cb->print_event(print_state,
pmu_name,
topic,
name,
alias,
scale_unit,
deprecated,
"Kernel PMU event",
desc,
long_desc,
encoding_desc);
}
if (printed && pager_in_use())
printf("\n");
zfree(&aliases);
}
bool perf_pmus__have_event(const char *pname, const char *name)
{
struct perf_pmu *pmu = perf_pmus__find(pname);
return pmu && perf_pmu__have_event(pmu, name);
}
int perf_pmus__num_core_pmus(void)
{
static int count;
if (!count) {
struct perf_pmu *pmu = NULL;
while ((pmu = perf_pmus__scan_core(pmu)) != NULL)
count++;
}
return count;
}
static bool __perf_pmus__supports_extended_type(void)
{
struct perf_pmu *pmu = NULL;
if (perf_pmus__num_core_pmus() <= 1)
return false;
while ((pmu = perf_pmus__scan_core(pmu)) != NULL) {
if (!is_event_supported(PERF_TYPE_HARDWARE, PERF_COUNT_HW_CPU_CYCLES | ((__u64)pmu->type << PERF_PMU_TYPE_SHIFT)))
return false;
}
return true;
}
static bool perf_pmus__do_support_extended_type;
static void perf_pmus__init_supports_extended_type(void)
{
perf_pmus__do_support_extended_type = __perf_pmus__supports_extended_type();
}
bool perf_pmus__supports_extended_type(void)
{
static pthread_once_t extended_type_once = PTHREAD_ONCE_INIT;
pthread_once(&extended_type_once, perf_pmus__init_supports_extended_type);
return perf_pmus__do_support_extended_type;
}
struct perf_pmu *evsel__find_pmu(const struct evsel *evsel)
{
struct perf_pmu *pmu = evsel->pmu;
if (!pmu) {
pmu = perf_pmus__find_by_type(evsel->core.attr.type);
((struct evsel *)evsel)->pmu = pmu;
}
return pmu;
}