diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2015-07-04 08:17:29 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2015-07-04 08:17:29 -0700 |
commit | c1776a18e3b5a3559f3dff5df0ecce570abd3a9f (patch) | |
tree | 5b97defe199f434f5d9829bf7d8df6124e6ce5f1 /tools/perf/builtin-stat.c | |
parent | 91cca0f0ffa3e485ad9d5b44744f23318c8f99ab (diff) | |
parent | b9df84fd7c05cc300d6d14f022b8a00773ebcf8c (diff) | |
download | linux-stable-c1776a18e3b5a3559f3dff5df0ecce570abd3a9f.tar.gz linux-stable-c1776a18e3b5a3559f3dff5df0ecce570abd3a9f.tar.bz2 linux-stable-c1776a18e3b5a3559f3dff5df0ecce570abd3a9f.zip |
Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf updates from Ingo Molnar:
"This tree includes an x86 PMU scheduling fix, but most changes are
late breaking tooling fixes and updates:
User visible fixes:
- Create config.detected into OUTPUT directory, fixing parallel
builds sharing the same source directory (Aaro Kiskinen)
- Allow to specify custom linker command, fixing some MIPS64 builds.
(Aaro Kiskinen)
- Fix to show proper convergence stats in 'perf bench numa' (Srikar
Dronamraju)
User visible changes:
- Validate syscall list passed via -e argument to 'perf trace'.
(Arnaldo Carvalho de Melo)
- Introduce 'perf stat --per-thread' (Jiri Olsa)
- Check access permission for --kallsyms and --vmlinux (Li Zhang)
- Move toggling event logic from 'perf top' and into hists browser,
allowing freeze/unfreeze with event lists with more than one entry
(Namhyung Kim)
- Add missing newlines when dumping PERF_RECORD_FINISHED_ROUND and
showing the Aggregated stats in 'perf report -D' (Adrian Hunter)
Infrastructure fixes:
- Add missing break for PERF_RECORD_ITRACE_START, which caused those
events samples to be parsed as well as PERF_RECORD_LOST_SAMPLES.
ITRACE_START only appears when Intel PT or BTS are present, so..
(Jiri Olsa)
- Call the perf_session destructor when bailing out in the inject,
kmem, report, kvm and mem tools (Taeung Song)
Infrastructure changes:
- Move stuff out of 'perf stat' and into the lib for further use
(Jiri Olsa)
- Reference count the cpu_map and thread_map classes (Jiri Olsa)
- Set evsel->{cpus,threads} from the evlist, if not set, allowing the
generalization of some 'perf stat' functions that previously were
accessing private static evlist variable (Jiri Olsa)
- Delete an unnecessary check before the calling free_event_desc()
(Markus Elfring)
- Allow auxtrace data alignment (Adrian Hunter)
- Allow events with dot (Andi Kleen)
- Fix failure to 'perf probe' events on arm (He Kuang)
- Add testing for Makefile.perf (Jiri Olsa)
- Add test for make install with prefix (Jiri Olsa)
- Fix single target build dependency check (Jiri Olsa)
- Access thread_map entries via accessors, prep patch to hold more
info per entry, for ongoing 'perf stat --per-thread' work (Jiri
Olsa)
- Use __weak definition from compiler.h (Sukadev Bhattiprolu)
- Split perf_pmu__new_alias() (Sukadev Bhattiprolu)"
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (54 commits)
perf tools: Allow to specify custom linker command
perf tools: Create config.detected into OUTPUT directory
perf mem: Fill in the missing session freeing after an error occurs
perf kvm: Fill in the missing session freeing after an error occurs
perf report: Fill in the missing session freeing after an error occurs
perf kmem: Fill in the missing session freeing after an error occurs
perf inject: Fill in the missing session freeing after an error occurs
perf tools: Add missing break for PERF_RECORD_ITRACE_START
perf/x86: Fix 'active_events' imbalance
perf symbols: Check access permission when reading symbol files
perf stat: Introduce --per-thread option
perf stat: Introduce print_counters function
perf stat: Using init_stats instead of memset
perf stat: Rename print_interval to process_interval
perf stat: Remove perf_evsel__read_cb function
perf stat: Move perf_stat initialization counter process code
perf stat: Move zero_per_pkg into counter process code
perf stat: Separate counters reading and processing
perf stat: Introduce read_counters function
perf stat: Introduce perf_evsel__read function
...
Diffstat (limited to 'tools/perf/builtin-stat.c')
-rw-r--r-- | tools/perf/builtin-stat.c | 412 |
1 files changed, 207 insertions, 205 deletions
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index fcf99bdeb19e..37e301a32f43 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -67,10 +67,7 @@ #define CNTR_NOT_SUPPORTED "<not supported>" #define CNTR_NOT_COUNTED "<not counted>" -static void print_stat(int argc, const char **argv); -static void print_counter_aggr(struct perf_evsel *counter, char *prefix); -static void print_counter(struct perf_evsel *counter, char *prefix); -static void print_aggr(char *prefix); +static void print_counters(struct timespec *ts, int argc, const char **argv); /* Default events used for perf stat -T */ static const char *transaction_attrs = { @@ -141,96 +138,9 @@ static inline void diff_timespec(struct timespec *r, struct timespec *a, } } -static inline struct cpu_map *perf_evsel__cpus(struct perf_evsel *evsel) +static void perf_stat__reset_stats(void) { - return (evsel->cpus && !target.cpu_list) ? evsel->cpus : evsel_list->cpus; -} - -static inline int perf_evsel__nr_cpus(struct perf_evsel *evsel) -{ - return perf_evsel__cpus(evsel)->nr; -} - -static void perf_evsel__reset_stat_priv(struct perf_evsel *evsel) -{ - int i; - struct perf_stat *ps = evsel->priv; - - for (i = 0; i < 3; i++) - init_stats(&ps->res_stats[i]); - - perf_stat_evsel_id_init(evsel); -} - -static int perf_evsel__alloc_stat_priv(struct perf_evsel *evsel) -{ - evsel->priv = zalloc(sizeof(struct perf_stat)); - if (evsel->priv == NULL) - return -ENOMEM; - perf_evsel__reset_stat_priv(evsel); - return 0; -} - -static void perf_evsel__free_stat_priv(struct perf_evsel *evsel) -{ - zfree(&evsel->priv); -} - -static int perf_evsel__alloc_prev_raw_counts(struct perf_evsel *evsel) -{ - struct perf_counts *counts; - - counts = perf_counts__new(perf_evsel__nr_cpus(evsel)); - if (counts) - evsel->prev_raw_counts = counts; - - return counts ? 0 : -ENOMEM; -} - -static void perf_evsel__free_prev_raw_counts(struct perf_evsel *evsel) -{ - perf_counts__delete(evsel->prev_raw_counts); - evsel->prev_raw_counts = NULL; -} - -static void perf_evlist__free_stats(struct perf_evlist *evlist) -{ - struct perf_evsel *evsel; - - evlist__for_each(evlist, evsel) { - perf_evsel__free_stat_priv(evsel); - perf_evsel__free_counts(evsel); - perf_evsel__free_prev_raw_counts(evsel); - } -} - -static int perf_evlist__alloc_stats(struct perf_evlist *evlist, bool alloc_raw) -{ - struct perf_evsel *evsel; - - evlist__for_each(evlist, evsel) { - if (perf_evsel__alloc_stat_priv(evsel) < 0 || - perf_evsel__alloc_counts(evsel, perf_evsel__nr_cpus(evsel)) < 0 || - (alloc_raw && perf_evsel__alloc_prev_raw_counts(evsel) < 0)) - goto out_free; - } - - return 0; - -out_free: - perf_evlist__free_stats(evlist); - return -1; -} - -static void perf_stat__reset_stats(struct perf_evlist *evlist) -{ - struct perf_evsel *evsel; - - evlist__for_each(evlist, evsel) { - perf_evsel__reset_stat_priv(evsel); - perf_evsel__reset_counts(evsel, perf_evsel__nr_cpus(evsel)); - } - + perf_evlist__reset_stats(evsel_list); perf_stat__reset_shadow_stats(); } @@ -304,8 +214,9 @@ static int check_per_pkg(struct perf_evsel *counter, int cpu, bool *skip) return 0; } -static int read_cb(struct perf_evsel *evsel, int cpu, int thread __maybe_unused, - struct perf_counts_values *count) +static int +process_counter_values(struct perf_evsel *evsel, int cpu, int thread, + struct perf_counts_values *count) { struct perf_counts_values *aggr = &evsel->counts->aggr; static struct perf_counts_values zero; @@ -320,13 +231,13 @@ static int read_cb(struct perf_evsel *evsel, int cpu, int thread __maybe_unused, count = &zero; switch (aggr_mode) { + case AGGR_THREAD: case AGGR_CORE: case AGGR_SOCKET: case AGGR_NONE: if (!evsel->snapshot) - perf_evsel__compute_deltas(evsel, cpu, count); + perf_evsel__compute_deltas(evsel, cpu, thread, count); perf_counts_values__scale(count, scale, NULL); - evsel->counts->cpu[cpu] = *count; if (aggr_mode == AGGR_NONE) perf_stat__update_shadow_stats(evsel, count->values, cpu); break; @@ -343,26 +254,48 @@ static int read_cb(struct perf_evsel *evsel, int cpu, int thread __maybe_unused, return 0; } -static int read_counter(struct perf_evsel *counter); +static int process_counter_maps(struct perf_evsel *counter) +{ + int nthreads = thread_map__nr(counter->threads); + int ncpus = perf_evsel__nr_cpus(counter); + int cpu, thread; -/* - * Read out the results of a single counter: - * aggregate counts across CPUs in system-wide mode - */ -static int read_counter_aggr(struct perf_evsel *counter) + if (counter->system_wide) + nthreads = 1; + + for (thread = 0; thread < nthreads; thread++) { + for (cpu = 0; cpu < ncpus; cpu++) { + if (process_counter_values(counter, cpu, thread, + perf_counts(counter->counts, cpu, thread))) + return -1; + } + } + + return 0; +} + +static int process_counter(struct perf_evsel *counter) { struct perf_counts_values *aggr = &counter->counts->aggr; struct perf_stat *ps = counter->priv; u64 *count = counter->counts->aggr.values; - int i; + int i, ret; aggr->val = aggr->ena = aggr->run = 0; + init_stats(ps->res_stats); - if (read_counter(counter)) - return -1; + if (counter->per_pkg) + zero_per_pkg(counter); + + ret = process_counter_maps(counter); + if (ret) + return ret; + + if (aggr_mode != AGGR_GLOBAL) + return 0; if (!counter->snapshot) - perf_evsel__compute_deltas(counter, -1, aggr); + perf_evsel__compute_deltas(counter, -1, -1, aggr); perf_counts_values__scale(aggr, scale, &counter->counts->scaled); for (i = 0; i < 3; i++) @@ -397,12 +330,12 @@ static int read_counter(struct perf_evsel *counter) if (counter->system_wide) nthreads = 1; - if (counter->per_pkg) - zero_per_pkg(counter); - for (thread = 0; thread < nthreads; thread++) { for (cpu = 0; cpu < ncpus; cpu++) { - if (perf_evsel__read_cb(counter, cpu, thread, read_cb)) + struct perf_counts_values *count; + + count = perf_counts(counter->counts, cpu, thread); + if (perf_evsel__read(counter, cpu, thread, count)) return -1; } } @@ -410,68 +343,34 @@ static int read_counter(struct perf_evsel *counter) return 0; } -static void print_interval(void) +static void read_counters(bool close) { - static int num_print_interval; struct perf_evsel *counter; - struct perf_stat *ps; - struct timespec ts, rs; - char prefix[64]; - if (aggr_mode == AGGR_GLOBAL) { - evlist__for_each(evsel_list, counter) { - ps = counter->priv; - memset(ps->res_stats, 0, sizeof(ps->res_stats)); - read_counter_aggr(counter); - } - } else { - evlist__for_each(evsel_list, counter) { - ps = counter->priv; - memset(ps->res_stats, 0, sizeof(ps->res_stats)); - read_counter(counter); - } - } + evlist__for_each(evsel_list, counter) { + if (read_counter(counter)) + pr_warning("failed to read counter %s\n", counter->name); - clock_gettime(CLOCK_MONOTONIC, &ts); - diff_timespec(&rs, &ts, &ref_time); - sprintf(prefix, "%6lu.%09lu%s", rs.tv_sec, rs.tv_nsec, csv_sep); + if (process_counter(counter)) + pr_warning("failed to process counter %s\n", counter->name); - if (num_print_interval == 0 && !csv_output) { - switch (aggr_mode) { - case AGGR_SOCKET: - fprintf(output, "# time socket cpus counts %*s events\n", unit_width, "unit"); - break; - case AGGR_CORE: - fprintf(output, "# time core cpus counts %*s events\n", unit_width, "unit"); - break; - case AGGR_NONE: - fprintf(output, "# time CPU counts %*s events\n", unit_width, "unit"); - break; - case AGGR_GLOBAL: - default: - fprintf(output, "# time counts %*s events\n", unit_width, "unit"); + if (close) { + perf_evsel__close_fd(counter, perf_evsel__nr_cpus(counter), + thread_map__nr(evsel_list->threads)); } } +} - if (++num_print_interval == 25) - num_print_interval = 0; +static void process_interval(void) +{ + struct timespec ts, rs; - switch (aggr_mode) { - case AGGR_CORE: - case AGGR_SOCKET: - print_aggr(prefix); - break; - case AGGR_NONE: - evlist__for_each(evsel_list, counter) - print_counter(counter, prefix); - break; - case AGGR_GLOBAL: - default: - evlist__for_each(evsel_list, counter) - print_counter_aggr(counter, prefix); - } + read_counters(false); - fflush(output); + clock_gettime(CLOCK_MONOTONIC, &ts); + diff_timespec(&rs, &ts, &ref_time); + + print_counters(&rs, 0, NULL); } static void handle_initial_delay(void) @@ -586,7 +485,7 @@ static int __run_perf_stat(int argc, const char **argv) if (interval) { while (!waitpid(child_pid, &status, WNOHANG)) { nanosleep(&ts, NULL); - print_interval(); + process_interval(); } } wait(&status); @@ -604,7 +503,7 @@ static int __run_perf_stat(int argc, const char **argv) while (!done) { nanosleep(&ts, NULL); if (interval) - print_interval(); + process_interval(); } } @@ -612,18 +511,7 @@ static int __run_perf_stat(int argc, const char **argv) update_stats(&walltime_nsecs_stats, t1 - t0); - if (aggr_mode == AGGR_GLOBAL) { - evlist__for_each(evsel_list, counter) { - read_counter_aggr(counter); - perf_evsel__close_fd(counter, perf_evsel__nr_cpus(counter), - thread_map__nr(evsel_list->threads)); - } - } else { - evlist__for_each(evsel_list, counter) { - read_counter(counter); - perf_evsel__close_fd(counter, perf_evsel__nr_cpus(counter), 1); - } - } + read_counters(true); return WEXITSTATUS(status); } @@ -715,6 +603,14 @@ static void aggr_printout(struct perf_evsel *evsel, int id, int nr) csv_output ? 0 : -4, perf_evsel__cpus(evsel)->map[id], csv_sep); break; + case AGGR_THREAD: + fprintf(output, "%*s-%*d%s", + csv_output ? 0 : 16, + thread_map__comm(evsel->threads, id), + csv_output ? 0 : -8, + thread_map__pid(evsel->threads, id), + csv_sep); + break; case AGGR_GLOBAL: default: break; @@ -815,9 +711,9 @@ static void print_aggr(char *prefix) s2 = aggr_get_id(evsel_list->cpus, cpu2); if (s2 != id) continue; - val += counter->counts->cpu[cpu].val; - ena += counter->counts->cpu[cpu].ena; - run += counter->counts->cpu[cpu].run; + val += perf_counts(counter->counts, cpu, 0)->val; + ena += perf_counts(counter->counts, cpu, 0)->ena; + run += perf_counts(counter->counts, cpu, 0)->run; nr++; } if (prefix) @@ -863,6 +759,40 @@ static void print_aggr(char *prefix) } } +static void print_aggr_thread(struct perf_evsel *counter, char *prefix) +{ + int nthreads = thread_map__nr(counter->threads); + int ncpus = cpu_map__nr(counter->cpus); + int cpu, thread; + double uval; + + for (thread = 0; thread < nthreads; thread++) { + u64 ena = 0, run = 0, val = 0; + + for (cpu = 0; cpu < ncpus; cpu++) { + val += perf_counts(counter->counts, cpu, thread)->val; + ena += perf_counts(counter->counts, cpu, thread)->ena; + run += perf_counts(counter->counts, cpu, thread)->run; + } + + if (prefix) + fprintf(output, "%s", prefix); + + uval = val * counter->scale; + + if (nsec_counter(counter)) + nsec_printout(thread, 0, counter, uval); + else + abs_printout(thread, 0, counter, uval); + + if (!csv_output) + print_noise(counter, 1.0); + + print_running(run, ena); + fputc('\n', output); + } +} + /* * Print out the results of a single counter: * aggregated counts in system-wide mode @@ -925,9 +855,9 @@ static void print_counter(struct perf_evsel *counter, char *prefix) int cpu; for (cpu = 0; cpu < perf_evsel__nr_cpus(counter); cpu++) { - val = counter->counts->cpu[cpu].val; - ena = counter->counts->cpu[cpu].ena; - run = counter->counts->cpu[cpu].run; + val = perf_counts(counter->counts, cpu, 0)->val; + ena = perf_counts(counter->counts, cpu, 0)->ena; + run = perf_counts(counter->counts, cpu, 0)->run; if (prefix) fprintf(output, "%s", prefix); @@ -972,9 +902,38 @@ static void print_counter(struct perf_evsel *counter, char *prefix) } } -static void print_stat(int argc, const char **argv) +static void print_interval(char *prefix, struct timespec *ts) +{ + static int num_print_interval; + + sprintf(prefix, "%6lu.%09lu%s", ts->tv_sec, ts->tv_nsec, csv_sep); + + if (num_print_interval == 0 && !csv_output) { + switch (aggr_mode) { + case AGGR_SOCKET: + fprintf(output, "# time socket cpus counts %*s events\n", unit_width, "unit"); + break; + case AGGR_CORE: + fprintf(output, "# time core cpus counts %*s events\n", unit_width, "unit"); + break; + case AGGR_NONE: + fprintf(output, "# time CPU counts %*s events\n", unit_width, "unit"); + break; + case AGGR_THREAD: + fprintf(output, "# time comm-pid counts %*s events\n", unit_width, "unit"); + break; + case AGGR_GLOBAL: + default: + fprintf(output, "# time counts %*s events\n", unit_width, "unit"); + } + } + + if (++num_print_interval == 25) + num_print_interval = 0; +} + +static void print_header(int argc, const char **argv) { - struct perf_evsel *counter; int i; fflush(stdout); @@ -1000,36 +959,57 @@ static void print_stat(int argc, const char **argv) fprintf(output, " (%d runs)", run_count); fprintf(output, ":\n\n"); } +} + +static void print_footer(void) +{ + if (!null_run) + fprintf(output, "\n"); + fprintf(output, " %17.9f seconds time elapsed", + avg_stats(&walltime_nsecs_stats)/1e9); + if (run_count > 1) { + fprintf(output, " "); + print_noise_pct(stddev_stats(&walltime_nsecs_stats), + avg_stats(&walltime_nsecs_stats)); + } + fprintf(output, "\n\n"); +} + +static void print_counters(struct timespec *ts, int argc, const char **argv) +{ + struct perf_evsel *counter; + char buf[64], *prefix = NULL; + + if (interval) + print_interval(prefix = buf, ts); + else + print_header(argc, argv); switch (aggr_mode) { case AGGR_CORE: case AGGR_SOCKET: - print_aggr(NULL); + print_aggr(prefix); + break; + case AGGR_THREAD: + evlist__for_each(evsel_list, counter) + print_aggr_thread(counter, prefix); break; case AGGR_GLOBAL: evlist__for_each(evsel_list, counter) - print_counter_aggr(counter, NULL); + print_counter_aggr(counter, prefix); break; case AGGR_NONE: evlist__for_each(evsel_list, counter) - print_counter(counter, NULL); + print_counter(counter, prefix); break; default: break; } - if (!csv_output) { - if (!null_run) - fprintf(output, "\n"); - fprintf(output, " %17.9f seconds time elapsed", - avg_stats(&walltime_nsecs_stats)/1e9); - if (run_count > 1) { - fprintf(output, " "); - print_noise_pct(stddev_stats(&walltime_nsecs_stats), - avg_stats(&walltime_nsecs_stats)); - } - fprintf(output, "\n\n"); - } + if (!interval && !csv_output) + print_footer(); + + fflush(output); } static volatile int signr = -1; @@ -1101,6 +1081,7 @@ static int perf_stat_init_aggr_mode(void) break; case AGGR_NONE: case AGGR_GLOBAL: + case AGGR_THREAD: default: break; } @@ -1325,6 +1306,8 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused) "aggregate counts per processor socket", AGGR_SOCKET), OPT_SET_UINT(0, "per-core", &aggr_mode, "aggregate counts per physical processor core", AGGR_CORE), + OPT_SET_UINT(0, "per-thread", &aggr_mode, + "aggregate counts per thread", AGGR_THREAD), OPT_UINTEGER('D', "delay", &initial_delay, "ms to wait before starting measurement after program start"), OPT_END() @@ -1416,8 +1399,19 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused) run_count = 1; } - /* no_aggr, cgroup are for system-wide only */ - if ((aggr_mode != AGGR_GLOBAL || nr_cgroups) && + if ((aggr_mode == AGGR_THREAD) && !target__has_task(&target)) { + fprintf(stderr, "The --per-thread option is only available " + "when monitoring via -p -t options.\n"); + parse_options_usage(NULL, options, "p", 1); + parse_options_usage(NULL, options, "t", 1); + goto out; + } + + /* + * no_aggr, cgroup are for system-wide only + * --per-thread is aggregated per thread, we dont mix it with cpu mode + */ + if (((aggr_mode != AGGR_GLOBAL && aggr_mode != AGGR_THREAD) || nr_cgroups) && !target__has_cpu(&target)) { fprintf(stderr, "both cgroup and no-aggregation " "modes only available in system-wide mode\n"); @@ -1445,6 +1439,14 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused) } goto out; } + + /* + * Initialize thread_map with comm names, + * so we could print it out on output. + */ + if (aggr_mode == AGGR_THREAD) + thread_map__read_comms(evsel_list->threads); + if (interval && interval < 100) { pr_err("print interval must be >= 100ms\n"); parse_options_usage(stat_usage, options, "I", 1); @@ -1478,13 +1480,13 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused) status = run_perf_stat(argc, argv); if (forever && status != -1) { - print_stat(argc, argv); - perf_stat__reset_stats(evsel_list); + print_counters(NULL, argc, argv); + perf_stat__reset_stats(); } } if (!forever && status != -1 && !interval) - print_stat(argc, argv); + print_counters(NULL, argc, argv); perf_evlist__free_stats(evsel_list); out: |