diff options
Diffstat (limited to 'tools/perf')
| -rw-r--r-- | tools/perf/Documentation/perf-record.txt | 3 | ||||
| -rw-r--r-- | tools/perf/Makefile | 2 | ||||
| -rw-r--r-- | tools/perf/builtin-record.c | 11 | ||||
| -rw-r--r-- | tools/perf/builtin-sched.c | 23 | ||||
| -rw-r--r-- | tools/perf/builtin-stat.c | 6 | ||||
| -rw-r--r-- | tools/perf/builtin-test.c | 116 | ||||
| -rw-r--r-- | tools/perf/builtin-top.c | 3 | ||||
| -rw-r--r-- | tools/perf/perf.c | 2 | ||||
| -rw-r--r-- | tools/perf/util/evsel.c | 87 | ||||
| -rw-r--r-- | tools/perf/util/evsel.h | 2 | ||||
| -rw-r--r-- | tools/perf/util/parse-events.c | 49 | ||||
| -rw-r--r-- | tools/perf/util/session.c | 2 |
12 files changed, 215 insertions, 91 deletions
diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt index 52462ae26455..e032716c839b 100644 --- a/tools/perf/Documentation/perf-record.txt +++ b/tools/perf/Documentation/perf-record.txt @@ -61,6 +61,9 @@ OPTIONS -r:: --realtime=:: Collect data with this RT SCHED_FIFO priority. +-D:: +--no-delay:: + Collect data without buffering. -A:: --append:: Append to the output file to do incremental profiling. diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 1b9b13ee2a72..2b5387d53ba5 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -227,7 +227,7 @@ ifndef PERF_DEBUG CFLAGS_OPTIMIZE = -O6 endif -CFLAGS = -ggdb3 -Wall -Wextra -std=gnu99 -Werror $(CFLAGS_OPTIMIZE) -D_FORTIFY_SOURCE=2 $(EXTRA_WARNINGS) $(EXTRA_CFLAGS) +CFLAGS = -fno-omit-frame-pointer -ggdb3 -Wall -Wextra -std=gnu99 -Werror $(CFLAGS_OPTIMIZE) -D_FORTIFY_SOURCE=2 $(EXTRA_WARNINGS) $(EXTRA_CFLAGS) EXTLIBS = -lpthread -lrt -lelf -lm ALL_CFLAGS = $(CFLAGS) -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 ALL_LDFLAGS = $(LDFLAGS) diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 7bc049035484..fcd29e8af29f 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -49,6 +49,7 @@ static int pipe_output = 0; static const char *output_name = "perf.data"; static int group = 0; static int realtime_prio = 0; +static bool nodelay = false; static bool raw_samples = false; static bool sample_id_all_avail = true; static bool system_wide = false; @@ -307,6 +308,11 @@ static void create_counter(struct perf_evsel *evsel, int cpu) attr->sample_type |= PERF_SAMPLE_CPU; } + if (nodelay) { + attr->watermark = 0; + attr->wakeup_events = 1; + } + attr->mmap = track; attr->comm = track; attr->inherit = !no_inherit; @@ -477,6 +483,7 @@ static void atexit_header(void) process_buildids(); perf_header__write(&session->header, output, true); perf_session__delete(session); + perf_evsel_list__delete(); symbol__exit(); } } @@ -842,6 +849,8 @@ const struct option record_options[] = { "record events on existing thread id"), OPT_INTEGER('r', "realtime", &realtime_prio, "collect data with this RT SCHED_FIFO priority"), + OPT_BOOLEAN('D', "no-delay", &nodelay, + "collect data without buffering"), OPT_BOOLEAN('R', "raw-samples", &raw_samples, "collect raw sample records from all opened counters"), OPT_BOOLEAN('a', "all-cpus", &system_wide, @@ -927,6 +936,8 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) list_for_each_entry(pos, &evsel_list, node) { if (perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) goto out_free_fd; + if (perf_header__push_event(pos->attr.config, event_name(pos))) + goto out_free_fd; } event_array = malloc((sizeof(struct pollfd) * MAX_NR_CPUS * MAX_COUNTERS * threads->nr)); diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c index 7a4ebeb8b016..29e7ffd85690 100644 --- a/tools/perf/builtin-sched.c +++ b/tools/perf/builtin-sched.c @@ -489,7 +489,8 @@ static void create_tasks(void) err = pthread_attr_init(&attr); BUG_ON(err); - err = pthread_attr_setstacksize(&attr, (size_t)(16*1024)); + err = pthread_attr_setstacksize(&attr, + (size_t) max(16 * 1024, PTHREAD_STACK_MIN)); BUG_ON(err); err = pthread_mutex_lock(&start_work_mutex); BUG_ON(err); @@ -1842,15 +1843,15 @@ static const char *record_args[] = { "-f", "-m", "1024", "-c", "1", - "-e", "sched:sched_switch:r", - "-e", "sched:sched_stat_wait:r", - "-e", "sched:sched_stat_sleep:r", - "-e", "sched:sched_stat_iowait:r", - "-e", "sched:sched_stat_runtime:r", - "-e", "sched:sched_process_exit:r", - "-e", "sched:sched_process_fork:r", - "-e", "sched:sched_wakeup:r", - "-e", "sched:sched_migrate_task:r", + "-e", "sched:sched_switch", + "-e", "sched:sched_stat_wait", + "-e", "sched:sched_stat_sleep", + "-e", "sched:sched_stat_iowait", + "-e", "sched:sched_stat_runtime", + "-e", "sched:sched_process_exit", + "-e", "sched:sched_process_fork", + "-e", "sched:sched_wakeup", + "-e", "sched:sched_migrate_task", }; static int __cmd_record(int argc, const char **argv) @@ -1861,7 +1862,7 @@ static int __cmd_record(int argc, const char **argv) rec_argc = ARRAY_SIZE(record_args) + argc - 1; rec_argv = calloc(rec_argc + 1, sizeof(char *)); - if (rec_argv) + if (rec_argv == NULL) return -ENOMEM; for (i = 0; i < ARRAY_SIZE(record_args); i++) diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index 02b2d8013a61..0ff11d9b13be 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -316,6 +316,8 @@ static int run_perf_stat(int argc __used, const char **argv) "\t Consider tweaking" " /proc/sys/kernel/perf_event_paranoid or running as root.", system_wide ? "system-wide " : ""); + } else if (errno == ENOENT) { + error("%s event is not supported. ", event_name(counter)); } else { error("open_counter returned with %d (%s). " "/bin/dmesg may provide additional information.\n", @@ -683,8 +685,7 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) nr_counters = ARRAY_SIZE(default_attrs); for (c = 0; c < ARRAY_SIZE(default_attrs); ++c) { - pos = perf_evsel__new(default_attrs[c].type, - default_attrs[c].config, + pos = perf_evsel__new(&default_attrs[c], nr_counters); if (pos == NULL) goto out; @@ -742,6 +743,7 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) out_free_fd: list_for_each_entry(pos, &evsel_list, node) perf_evsel__free_stat_priv(pos); + perf_evsel_list__delete(); out: thread_map__delete(threads); threads = NULL; diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index 1c984342a579..ed5696198d3d 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -234,6 +234,7 @@ out: return err; } +#include "util/cpumap.h" #include "util/evsel.h" #include <sys/types.h> @@ -264,6 +265,7 @@ static int test__open_syscall_event(void) int err = -1, fd; struct thread_map *threads; struct perf_evsel *evsel; + struct perf_event_attr attr; unsigned int nr_open_calls = 111, i; int id = trace_event__id("sys_enter_open"); @@ -278,7 +280,10 @@ static int test__open_syscall_event(void) return -1; } - evsel = perf_evsel__new(PERF_TYPE_TRACEPOINT, id, 0); + memset(&attr, 0, sizeof(attr)); + attr.type = PERF_TYPE_TRACEPOINT; + attr.config = id; + evsel = perf_evsel__new(&attr, 0); if (evsel == NULL) { pr_debug("perf_evsel__new\n"); goto out_thread_map_delete; @@ -317,6 +322,111 @@ out_thread_map_delete: return err; } +#include <sched.h> + +static int test__open_syscall_event_on_all_cpus(void) +{ + int err = -1, fd, cpu; + struct thread_map *threads; + struct cpu_map *cpus; + struct perf_evsel *evsel; + struct perf_event_attr attr; + unsigned int nr_open_calls = 111, i; + cpu_set_t *cpu_set; + size_t cpu_set_size; + int id = trace_event__id("sys_enter_open"); + + if (id < 0) { + pr_debug("is debugfs mounted on /sys/kernel/debug?\n"); + return -1; + } + + threads = thread_map__new(-1, getpid()); + if (threads == NULL) { + pr_debug("thread_map__new\n"); + return -1; + } + + cpus = cpu_map__new(NULL); + if (threads == NULL) { + pr_debug("thread_map__new\n"); + return -1; + } + + cpu_set = CPU_ALLOC(cpus->nr); + + if (cpu_set == NULL) + goto out_thread_map_delete; + + cpu_set_size = CPU_ALLOC_SIZE(cpus->nr); + CPU_ZERO_S(cpu_set_size, cpu_set); + + memset(&attr, 0, sizeof(attr)); + attr.type = PERF_TYPE_TRACEPOINT; + attr.config = id; + evsel = perf_evsel__new(&attr, 0); + if (evsel == NULL) { + pr_debug("perf_evsel__new\n"); + goto out_cpu_free; + } + + if (perf_evsel__open(evsel, cpus, threads) < 0) { + pr_debug("failed to open counter: %s, " + "tweak /proc/sys/kernel/perf_event_paranoid?\n", + strerror(errno)); + goto out_evsel_delete; + } + + for (cpu = 0; cpu < cpus->nr; ++cpu) { + unsigned int ncalls = nr_open_calls + cpu; + + CPU_SET(cpu, cpu_set); + sched_setaffinity(0, cpu_set_size, cpu_set); + for (i = 0; i < ncalls; ++i) { + fd = open("/etc/passwd", O_RDONLY); + close(fd); + } + CPU_CLR(cpu, cpu_set); + } + + /* + * Here we need to explicitely preallocate the counts, as if + * we use the auto allocation it will allocate just for 1 cpu, + * as we start by cpu 0. + */ + if (perf_evsel__alloc_counts(evsel, cpus->nr) < 0) { + pr_debug("perf_evsel__alloc_counts(ncpus=%d)\n", cpus->nr); + goto out_close_fd; + } + + for (cpu = 0; cpu < cpus->nr; ++cpu) { + unsigned int expected; + + if (perf_evsel__read_on_cpu(evsel, cpu, 0) < 0) { + pr_debug("perf_evsel__open_read_on_cpu\n"); + goto out_close_fd; + } + + expected = nr_open_calls + cpu; + if (evsel->counts->cpu[cpu].val != expected) { + pr_debug("perf_evsel__read_on_cpu: expected to intercept %d calls on cpu %d, got %Ld\n", + expected, cpu, evsel->counts->cpu[cpu].val); + goto out_close_fd; + } + } + + err = 0; +out_close_fd: + perf_evsel__close_fd(evsel, 1, threads->nr); +out_evsel_delete: + perf_evsel__delete(evsel); +out_cpu_free: + CPU_FREE(cpu_set); +out_thread_map_delete: + thread_map__delete(threads); + return err; +} + static struct test { const char *desc; int (*func)(void); @@ -330,6 +440,10 @@ static struct test { .func = test__open_syscall_event, }, { + .desc = "detect open syscall event on all cpus", + .func = test__open_syscall_event_on_all_cpus, + }, + { .func = NULL, }, }; diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 1e67ab9c7ebc..05344c6210ac 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -1471,6 +1471,8 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) pos->attr.sample_period = default_interval; } + sym_evsel = list_entry(evsel_list.next, struct perf_evsel, node); + symbol_conf.priv_size = (sizeof(struct sym_entry) + (nr_counters + 1) * sizeof(unsigned long)); @@ -1488,6 +1490,7 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) out_free_fd: list_for_each_entry(pos, &evsel_list, node) perf_evsel__free_mmap(pos); + perf_evsel_list__delete(); return status; } diff --git a/tools/perf/perf.c b/tools/perf/perf.c index 5b1ecd66bb36..595d0f4a7103 100644 --- a/tools/perf/perf.c +++ b/tools/perf/perf.c @@ -286,8 +286,6 @@ static int run_builtin(struct cmd_struct *p, int argc, const char **argv) status = p->fn(argc, argv, prefix); exit_browser(status); - perf_evsel_list__delete(); - if (status) return status & 0xff; diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index c95267e63c5b..f5cfed60af98 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -6,14 +6,13 @@ #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) -struct perf_evsel *perf_evsel__new(u32 type, u64 config, int idx) +struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx) { struct perf_evsel *evsel = zalloc(sizeof(*evsel)); if (evsel != NULL) { evsel->idx = idx; - evsel->attr.type = type; - evsel->attr.config = config; + evsel->attr = *attr; INIT_LIST_HEAD(&evsel->node); } @@ -128,59 +127,75 @@ int __perf_evsel__read(struct perf_evsel *evsel, return 0; } -int perf_evsel__open_per_cpu(struct perf_evsel *evsel, struct cpu_map *cpus) +static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads) { - int cpu; + int cpu, thread; - if (evsel->fd == NULL && perf_evsel__alloc_fd(evsel, cpus->nr, 1) < 0) + if (evsel->fd == NULL && + perf_evsel__alloc_fd(evsel, cpus->nr, threads->nr) < 0) return -1; for (cpu = 0; cpu < cpus->nr; cpu++) { - FD(evsel, cpu, 0) = sys_perf_event_open(&evsel->attr, -1, - cpus->map[cpu], -1, 0); - if (FD(evsel, cpu, 0) < 0) - goto out_close; + for (thread = 0; thread < threads->nr; thread++) { + FD(evsel, cpu, thread) = sys_perf_event_open(&evsel->attr, + threads->map[thread], + cpus->map[cpu], -1, 0); + if (FD(evsel, cpu, thread) < 0) + goto out_close; + } } return 0; out_close: - while (--cpu >= 0) { - close(FD(evsel, cpu, 0)); - FD(evsel, cpu, 0) = -1; - } + do { + while (--thread >= 0) { + close(FD(evsel, cpu, thread)); + FD(evsel, cpu, thread) = -1; + } + thread = threads->nr; + } while (--cpu >= 0); return -1; } -int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *threads) +static struct { + struct cpu_map map; + int cpus[1]; +} empty_cpu_map = { + .map.nr = 1, + .cpus = { -1, }, +}; + +static struct { + struct thread_map map; + int threads[1]; +} empty_thread_map = { + .map.nr = 1, + .threads = { -1, }, +}; + +int perf_evsel__open(struct perf_evsel *evsel, + struct cpu_map *cpus, struct thread_map *threads) { - int thread; - - if (evsel->fd == NULL && perf_evsel__alloc_fd(evsel, 1, threads->nr)) - return -1; - for (thread = 0; thread < threads->nr; thread++) { - FD(evsel, 0, thread) = sys_perf_event_open(&evsel->attr, - threads->map[thread], -1, -1, 0); - if (FD(evsel, 0, thread) < 0) - goto out_close; + if (cpus == NULL) { + /* Work around old compiler warnings about strict aliasing */ + cpus = &empty_cpu_map.map; } - return 0; + if (threads == NULL) + threads = &empty_thread_map.map; -out_close: - while (--thread >= 0) { - close(FD(evsel, 0, thread)); - FD(evsel, 0, thread) = -1; - } - return -1; + return __perf_evsel__open(evsel, cpus, threads); } -int perf_evsel__open(struct perf_evsel *evsel, - struct cpu_map *cpus, struct thread_map *threads) +int perf_evsel__open_per_cpu(struct perf_evsel *evsel, struct cpu_map *cpus) { - if (threads == NULL) - return perf_evsel__open_per_cpu(evsel, cpus); + return __perf_evsel__open(evsel, cpus, &empty_thread_map.map); +} - return perf_evsel__open_per_thread(evsel, threads); +int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *threads) +{ + return __perf_evsel__open(evsel, &empty_cpu_map.map, threads); } diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index a0ccd69c3fc2..b2d755fe88a5 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -37,7 +37,7 @@ struct perf_evsel { struct cpu_map; struct thread_map; -struct perf_evsel *perf_evsel__new(u32 type, u64 config, int idx); +struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx); void perf_evsel__delete(struct perf_evsel *evsel); int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads); diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 649083f27e08..bc2732ee23eb 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -490,7 +490,6 @@ parse_multiple_tracepoint_event(char *sys_name, const char *evt_exp, return EVT_HANDLED_ALL; } - static enum event_result parse_tracepoint_event(const char **strp, struct perf_event_attr *attr) { @@ -530,12 +529,13 @@ static enum event_result parse_tracepoint_event(const char **strp, if (evt_length >= MAX_EVENT_LENGTH) return EVT_FAILED; if (strpbrk(evt_name, "*?")) { - *strp += strlen(sys_name) + evt_length; + *strp += strlen(sys_name) + evt_length + 1; /* 1 == the ':' */ return parse_multiple_tracepoint_event(sys_name, evt_name, flags); - } else + } else { return parse_single_tracepoint_event(sys_name, evt_name, evt_length, attr, strp); + } } static enum event_result @@ -778,41 +778,11 @@ modifier: return ret; } -static int store_event_type(const char *orgname) -{ - char filename[PATH_MAX], *c; - FILE *file; - int id, n; - - sprintf(filename, "%s/", debugfs_path); - strncat(filename, orgname, strlen(orgname)); - strcat(filename, "/id"); - - c = strchr(filename, ':'); - if (c) - *c = '/'; - - file = fopen(filename, "r"); - if (!file) - return 0; - n = fscanf(file, "%i", &id); - fclose(file); - if (n < 1) { - pr_err("cannot store event ID\n"); - return -EINVAL; - } - return perf_header__push_event(id, orgname); -} - int parse_events(const struct option *opt __used, const char *str, int unset __used) { struct perf_event_attr attr; enum event_result ret; - if (strchr(str, ':')) - if (store_event_type(str) < 0) - return -1; - for (;;) { memset(&attr, 0, sizeof(attr)); ret = parse_event_symbols(&str, &attr); @@ -824,7 +794,7 @@ int parse_events(const struct option *opt __used, const char *str, int unset __u if (ret != EVT_HANDLED_ALL) { struct perf_evsel *evsel; - evsel = perf_evsel__new(attr.type, attr.config, + evsel = perf_evsel__new(&attr, nr_counters); if (evsel == NULL) return -1; @@ -1014,8 +984,15 @@ void print_events(void) int perf_evsel_list__create_default(void) { - struct perf_evsel *evsel = perf_evsel__new(PERF_TYPE_HARDWARE, - PERF_COUNT_HW_CPU_CYCLES, 0); + struct perf_evsel *evsel; + struct perf_event_attr attr; + + memset(&attr, 0, sizeof(attr)); + attr.type = PERF_TYPE_HARDWARE; + attr.config = PERF_COUNT_HW_CPU_CYCLES; + + evsel = perf_evsel__new(&attr, 0); + if (evsel == NULL) return -ENOMEM; diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 6fb4694d05fa..313dac2d94ce 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -1007,7 +1007,7 @@ more: if (size == 0) size = 8; - if (head + event->header.size >= mmap_size) { + if (head + event->header.size > mmap_size) { if (mmaps[map_idx]) { munmap(mmaps[map_idx], mmap_size); mmaps[map_idx] = NULL; |