diff options
-rw-r--r-- | arch/x86/kernel/cpu/perf_event_amd.c | 14 | ||||
-rw-r--r-- | arch/x86/kernel/cpu/perf_event_intel.c | 31 | ||||
-rw-r--r-- | include/linux/perf_event.h | 2 | ||||
-rw-r--r-- | tools/perf/builtin-stat.c | 284 | ||||
-rw-r--r-- | tools/perf/util/parse-events.c | 120 | ||||
-rw-r--r-- | tools/perf/util/python.c | 3 |
6 files changed, 352 insertions, 102 deletions
diff --git a/arch/x86/kernel/cpu/perf_event_amd.c b/arch/x86/kernel/cpu/perf_event_amd.c index cf4e369cea6..fe29c1d2219 100644 --- a/arch/x86/kernel/cpu/perf_event_amd.c +++ b/arch/x86/kernel/cpu/perf_event_amd.c @@ -96,12 +96,14 @@ static __initconst const u64 amd_hw_cache_event_ids */ static const u64 amd_perfmon_event_map[] = { - [PERF_COUNT_HW_CPU_CYCLES] = 0x0076, - [PERF_COUNT_HW_INSTRUCTIONS] = 0x00c0, - [PERF_COUNT_HW_CACHE_REFERENCES] = 0x0080, - [PERF_COUNT_HW_CACHE_MISSES] = 0x0081, - [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x00c2, - [PERF_COUNT_HW_BRANCH_MISSES] = 0x00c3, + [PERF_COUNT_HW_CPU_CYCLES] = 0x0076, + [PERF_COUNT_HW_INSTRUCTIONS] = 0x00c0, + [PERF_COUNT_HW_CACHE_REFERENCES] = 0x0080, + [PERF_COUNT_HW_CACHE_MISSES] = 0x0081, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x00c2, + [PERF_COUNT_HW_BRANCH_MISSES] = 0x00c3, + [PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = 0x00d0, /* "Decoder empty" event */ + [PERF_COUNT_HW_STALLED_CYCLES_BACKEND] = 0x00d1, /* "Dispatch stalls" event */ }; static u64 amd_pmu_event_map(int hw_event) diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c index 7cf2ec59c81..61cbf4831d7 100644 --- a/arch/x86/kernel/cpu/perf_event_intel.c +++ b/arch/x86/kernel/cpu/perf_event_intel.c @@ -36,7 +36,7 @@ static u64 intel_perfmon_event_map[PERF_COUNT_HW_MAX] __read_mostly = [PERF_COUNT_HW_BUS_CYCLES] = 0x013c, }; -static struct event_constraint intel_core_event_constraints[] = +static struct event_constraint intel_core_event_constraints[] __read_mostly = { INTEL_EVENT_CONSTRAINT(0x11, 0x2), /* FP_ASSIST */ INTEL_EVENT_CONSTRAINT(0x12, 0x2), /* MUL */ @@ -47,7 +47,7 @@ static struct event_constraint intel_core_event_constraints[] = EVENT_CONSTRAINT_END }; -static struct event_constraint intel_core2_event_constraints[] = +static struct event_constraint intel_core2_event_constraints[] __read_mostly = { FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ @@ -70,7 +70,7 @@ static struct event_constraint intel_core2_event_constraints[] = EVENT_CONSTRAINT_END }; -static struct event_constraint intel_nehalem_event_constraints[] = +static struct event_constraint intel_nehalem_event_constraints[] __read_mostly = { FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ @@ -86,19 +86,19 @@ static struct event_constraint intel_nehalem_event_constraints[] = EVENT_CONSTRAINT_END }; -static struct extra_reg intel_nehalem_extra_regs[] = +static struct extra_reg intel_nehalem_extra_regs[] __read_mostly = { INTEL_EVENT_EXTRA_REG(0xb7, MSR_OFFCORE_RSP_0, 0xffff), EVENT_EXTRA_END }; -static struct event_constraint intel_nehalem_percore_constraints[] = +static struct event_constraint intel_nehalem_percore_constraints[] __read_mostly = { INTEL_EVENT_CONSTRAINT(0xb7, 0), EVENT_CONSTRAINT_END }; -static struct event_constraint intel_westmere_event_constraints[] = +static struct event_constraint intel_westmere_event_constraints[] __read_mostly = { FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ @@ -110,7 +110,7 @@ static struct event_constraint intel_westmere_event_constraints[] = EVENT_CONSTRAINT_END }; -static struct event_constraint intel_snb_event_constraints[] = +static struct event_constraint intel_snb_event_constraints[] __read_mostly = { FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ @@ -123,21 +123,21 @@ static struct event_constraint intel_snb_event_constraints[] = EVENT_CONSTRAINT_END }; -static struct extra_reg intel_westmere_extra_regs[] = +static struct extra_reg intel_westmere_extra_regs[] __read_mostly = { INTEL_EVENT_EXTRA_REG(0xb7, MSR_OFFCORE_RSP_0, 0xffff), INTEL_EVENT_EXTRA_REG(0xbb, MSR_OFFCORE_RSP_1, 0xffff), EVENT_EXTRA_END }; -static struct event_constraint intel_westmere_percore_constraints[] = +static struct event_constraint intel_westmere_percore_constraints[] __read_mostly = { INTEL_EVENT_CONSTRAINT(0xb7, 0), INTEL_EVENT_CONSTRAINT(0xbb, 0), EVENT_CONSTRAINT_END }; -static struct event_constraint intel_gen_event_constraints[] = +static struct event_constraint intel_gen_event_constraints[] __read_mostly = { FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ @@ -1423,6 +1423,11 @@ static __init int intel_pmu_init(void) x86_pmu.enable_all = intel_pmu_nhm_enable_all; x86_pmu.extra_regs = intel_nehalem_extra_regs; + /* UOPS_ISSUED.STALLED_CYCLES */ + intel_perfmon_event_map[PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = 0x180010e; + /* UOPS_EXECUTED.CORE_ACTIVE_CYCLES,c=1,i=1 */ + intel_perfmon_event_map[PERF_COUNT_HW_STALLED_CYCLES_BACKEND] = 0x1803fb1; + if (ebx & 0x40) { /* * Erratum AAJ80 detected, we work it around by using @@ -1463,6 +1468,12 @@ static __init int intel_pmu_init(void) x86_pmu.enable_all = intel_pmu_nhm_enable_all; x86_pmu.pebs_constraints = intel_westmere_pebs_event_constraints; x86_pmu.extra_regs = intel_westmere_extra_regs; + + /* UOPS_ISSUED.STALLED_CYCLES */ + intel_perfmon_event_map[PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = 0x180010e; + /* UOPS_EXECUTED.CORE_ACTIVE_CYCLES,c=1,i=1 */ + intel_perfmon_event_map[PERF_COUNT_HW_STALLED_CYCLES_BACKEND] = 0x1803fb1; + pr_cont("Westmere events, "); break; diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 207c16976a1..3412684ce5d 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -52,6 +52,8 @@ enum perf_hw_id { PERF_COUNT_HW_BRANCH_INSTRUCTIONS = 4, PERF_COUNT_HW_BRANCH_MISSES = 5, PERF_COUNT_HW_BUS_CYCLES = 6, + PERF_COUNT_HW_STALLED_CYCLES_FRONTEND = 7, + PERF_COUNT_HW_STALLED_CYCLES_BACKEND = 8, PERF_COUNT_HW_MAX, /* non-ABI */ }; diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index 03f0e45f147..602c3c96fa1 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -46,6 +46,7 @@ #include "util/evlist.h" #include "util/evsel.h" #include "util/debug.h" +#include "util/color.h" #include "util/header.h" #include "util/cpumap.h" #include "util/thread.h" @@ -65,14 +66,56 @@ static struct perf_event_attr default_attrs[] = { { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_PAGE_FAULTS }, { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CPU_CYCLES }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_FRONTEND }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_BACKEND }, { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_INSTRUCTIONS }, { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_INSTRUCTIONS }, { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_MISSES }, - { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CACHE_REFERENCES }, - { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CACHE_MISSES }, }; +/* + * Detailed stats: + */ +static struct perf_event_attr detailed_attrs[] = { + + { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_TASK_CLOCK }, + { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CONTEXT_SWITCHES }, + { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CPU_MIGRATIONS }, + { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_PAGE_FAULTS }, + + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CPU_CYCLES }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_FRONTEND }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_BACKEND }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_INSTRUCTIONS }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_INSTRUCTIONS }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_MISSES }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_L1D << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_L1D << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_LL << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_LL << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) }, +}; + struct perf_evlist *evsel_list; static bool system_wide = false; @@ -86,6 +129,8 @@ static pid_t target_pid = -1; static pid_t target_tid = -1; static pid_t child_pid = -1; static bool null_run = false; +static bool detailed_run = false; +static bool sync_run = false; static bool big_num = true; static int big_num_opt = -1; static const char *cpu_list; @@ -156,7 +201,11 @@ static double stddev_stats(struct stats *stats) struct stats runtime_nsecs_stats[MAX_NR_CPUS]; struct stats runtime_cycles_stats[MAX_NR_CPUS]; +struct stats runtime_stalled_cycles_front_stats[MAX_NR_CPUS]; +struct stats runtime_stalled_cycles_back_stats[MAX_NR_CPUS]; struct stats runtime_branches_stats[MAX_NR_CPUS]; +struct stats runtime_cacherefs_stats[MAX_NR_CPUS]; +struct stats runtime_l1_dcache_stats[MAX_NR_CPUS]; struct stats walltime_nsecs_stats; static int create_perf_stat_counter(struct perf_evsel *evsel) @@ -193,6 +242,29 @@ static inline int nsec_counter(struct perf_evsel *evsel) } /* + * Update various tracking values we maintain to print + * more semantic information such as miss/hit ratios, + * instruction rates, etc: + */ +static void update_shadow_stats(struct perf_evsel *counter, u64 *count) +{ + if (perf_evsel__match(counter, SOFTWARE, SW_TASK_CLOCK)) + update_stats(&runtime_nsecs_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_CPU_CYCLES)) + update_stats(&runtime_cycles_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_STALLED_CYCLES_FRONTEND)) + update_stats(&runtime_stalled_cycles_front_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_STALLED_CYCLES_BACKEND)) + update_stats(&runtime_stalled_cycles_back_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_BRANCH_INSTRUCTIONS)) + update_stats(&runtime_branches_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_CACHE_REFERENCES)) + update_stats(&runtime_cacherefs_stats[0], count[0]); + else if (perf_evsel__match(counter, HW_CACHE, HW_CACHE_L1D)) + update_stats(&runtime_l1_dcache_stats[0], count[0]); +} + +/* * Read out the results of a single counter: * aggregate counts across CPUs in system-wide mode */ @@ -217,12 +289,7 @@ static int read_counter_aggr(struct perf_evsel *counter) /* * Save the full runtime - to allow normalization during printout: */ - if (perf_evsel__match(counter, SOFTWARE, SW_TASK_CLOCK)) - update_stats(&runtime_nsecs_stats[0], count[0]); - if (perf_evsel__match(counter, HARDWARE, HW_CPU_CYCLES)) - update_stats(&runtime_cycles_stats[0], count[0]); - if (perf_evsel__match(counter, HARDWARE, HW_BRANCH_INSTRUCTIONS)) - update_stats(&runtime_branches_stats[0], count[0]); + update_shadow_stats(counter, count); return 0; } @@ -242,12 +309,7 @@ static int read_counter(struct perf_evsel *counter) count = counter->counts->cpu[cpu].values; - if (perf_evsel__match(counter, SOFTWARE, SW_TASK_CLOCK)) - update_stats(&runtime_nsecs_stats[cpu], count[0]); - if (perf_evsel__match(counter, HARDWARE, HW_CPU_CYCLES)) - update_stats(&runtime_cycles_stats[cpu], count[0]); - if (perf_evsel__match(counter, HARDWARE, HW_BRANCH_INSTRUCTIONS)) - update_stats(&runtime_branches_stats[cpu], count[0]); + update_shadow_stats(counter, count); } return 0; @@ -315,13 +377,18 @@ static int run_perf_stat(int argc __used, const char **argv) list_for_each_entry(counter, &evsel_list->entries, node) { if (create_perf_stat_counter(counter) < 0) { - if (errno == -EPERM || errno == -EACCES) { + if (errno == EINVAL || errno == ENOSYS || errno == ENOENT) { + if (verbose) + ui__warning("%s event is not supported by the kernel.\n", + event_name(counter)); + continue; + } + + if (errno == EPERM || errno == EACCES) { error("You may not have permission to collect %sstats.\n" "\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", @@ -372,6 +439,16 @@ static int run_perf_stat(int argc __used, const char **argv) return WEXITSTATUS(status); } +static void print_noise_pct(double total, double avg) +{ + double pct = 0.0; + + if (avg) + pct = 100.0*total/avg; + + fprintf(stderr, " ( +-%6.2f%% )", pct); +} + static void print_noise(struct perf_evsel *evsel, double avg) { struct perf_stat *ps; @@ -380,8 +457,7 @@ static void print_noise(struct perf_evsel *evsel, double avg) return; ps = evsel->priv; - fprintf(stderr, " ( +- %7.3f%% )", - 100 * stddev_stats(&ps->res_stats[0]) / avg); + print_noise_pct(stddev_stats(&ps->res_stats[0]), avg); } static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg) @@ -404,8 +480,99 @@ static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg) return; if (perf_evsel__match(evsel, SOFTWARE, SW_TASK_CLOCK)) - fprintf(stderr, " # %10.3f CPUs ", - avg / avg_stats(&walltime_nsecs_stats)); + fprintf(stderr, " # %8.3f CPUs utilized ", avg / avg_stats(&walltime_nsecs_stats)); +} + +static void print_stalled_cycles_frontend(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_cycles_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 50.0) + color = PERF_COLOR_RED; + else if (ratio > 30.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 10.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " frontend cycles idle "); +} + +static void print_stalled_cycles_backend(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_cycles_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 75.0) + color = PERF_COLOR_RED; + else if (ratio > 50.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 20.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " backend cycles idle "); +} + +static void print_branch_misses(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_branches_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 20.0) + color = PERF_COLOR_RED; + else if (ratio > 10.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 5.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " of all branches "); +} + +static void print_l1_dcache_misses(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_l1_dcache_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 20.0) + color = PERF_COLOR_RED; + else if (ratio > 10.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 5.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " of all L1-dcache hits "); } static void abs_printout(int cpu, struct perf_evsel *evsel, double avg) @@ -442,23 +609,55 @@ static void abs_printout(int cpu, struct perf_evsel *evsel, double avg) if (total) ratio = avg / total; - fprintf(stderr, " # %10.3f IPC ", ratio); + fprintf(stderr, " # %5.2f insns per cycle ", ratio); + + total = avg_stats(&runtime_stalled_cycles_front_stats[cpu]); + total = max(total, avg_stats(&runtime_stalled_cycles_back_stats[cpu])); + + if (total && avg) { + ratio = total / avg; + fprintf(stderr, "\n # %5.2f stalled cycles per insn", ratio); + } + } else if (perf_evsel__match(evsel, HARDWARE, HW_BRANCH_MISSES) && runtime_branches_stats[cpu].n != 0) { - total = avg_stats(&runtime_branches_stats[cpu]); + print_branch_misses(cpu, evsel, avg); + } else if ( + evsel->attr.type == PERF_TYPE_HW_CACHE && + evsel->attr.config == ( PERF_COUNT_HW_CACHE_L1D | + ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | + ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16)) && + runtime_l1_dcache_stats[cpu].n != 0) { + print_l1_dcache_misses(cpu, evsel, avg); + } else if (perf_evsel__match(evsel, HARDWARE, HW_CACHE_MISSES) && + runtime_cacherefs_stats[cpu].n != 0) { + total = avg_stats(&runtime_cacherefs_stats[cpu]); if (total) ratio = avg * 100 / total; - fprintf(stderr, " # %10.3f %% ", ratio); + fprintf(stderr, " # %8.3f %% of all cache refs ", ratio); + } else if (perf_evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_FRONTEND)) { + print_stalled_cycles_frontend(cpu, evsel, avg); + } else if (perf_evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_BACKEND)) { + print_stalled_cycles_backend(cpu, evsel, avg); + } else if (perf_evsel__match(evsel, HARDWARE, HW_CPU_CYCLES)) { + total = avg_stats(&runtime_nsecs_stats[cpu]); + + if (total) + ratio = 1.0 * avg / total; + + fprintf(stderr, " # %8.3f GHz ", ratio); } else if (runtime_nsecs_stats[cpu].n != 0) { total = avg_stats(&runtime_nsecs_stats[cpu]); if (total) ratio = 1000.0 * avg / total; - fprintf(stderr, " # %10.3f M/sec", ratio); + fprintf(stderr, " # %8.3f M/sec ", ratio); + } else { + fprintf(stderr, " "); } } @@ -505,8 +704,7 @@ static void print_counter_aggr(struct perf_evsel *counter) avg_enabled = avg_stats(&ps->res_stats[1]); avg_running = avg_stats(&ps->res_stats[2]); - fprintf(stderr, " (scaled from %.2f%%)", - 100 * avg_running / avg_enabled); + fprintf(stderr, " (%.2f%%)", 100 * avg_running / avg_enabled); } fprintf(stderr, "\n"); } @@ -548,10 +746,8 @@ static void print_counter(struct perf_evsel *counter) if (!csv_output) { print_noise(counter, 1.0); - if (run != ena) { - fprintf(stderr, " (scaled from %.2f%%)", - 100.0 * run / ena); - } + if (run != ena) + fprintf(stderr, " (%.2f%%)", 100.0 * run / ena); } fputc('\n', stderr); } @@ -595,9 +791,8 @@ static void print_stat(int argc, const char **argv) fprintf(stderr, " %18.9f seconds time elapsed", avg_stats(&walltime_nsecs_stats)/1e9); if (run_count > 1) { - fprintf(stderr, " ( +- %7.3f%% )", - 100*stddev_stats(&walltime_nsecs_stats) / - avg_stats(&walltime_nsecs_stats)); + print_noise_pct(stddev_stats(&walltime_nsecs_stats), + avg_stats(&walltime_nsecs_stats)); } fprintf(stderr, "\n\n"); } @@ -659,6 +854,10 @@ static const struct option options[] = { "repeat command and print average + stddev (max: 100)"), OPT_BOOLEAN('n', "null", &null_run, "null run - dont start any counters"), + OPT_BOOLEAN('d', "detailed", &detailed_run, + "detailed run - start a lot of events"), + OPT_BOOLEAN('S', "sync", &sync_run, + "call sync() before starting a run"), OPT_CALLBACK_NOOPT('B', "big-num", NULL, NULL, "print large numbers with thousands\' separators", stat__set_big_num), @@ -720,7 +919,18 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) } /* Set attrs and nr_counters if no event is selected and !null_run */ - if (!null_run && !evsel_list->nr_entries) { + if (detailed_run) { + size_t c; + + for (c = 0; c < ARRAY_SIZE(detailed_attrs); ++c) { + pos = perf_evsel__new(&detailed_attrs[c], c); + if (pos == NULL) + goto out; + perf_evlist__add(evsel_list, pos); + } + } + /* Set attrs and nr_counters if no event is selected and !null_run */ + if (!detailed_run && !null_run && !evsel_list->nr_entries) { size_t c; for (c = 0; c < ARRAY_SIZE(default_attrs); ++c) { @@ -773,6 +983,10 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) for (run_idx = 0; run_idx < run_count; run_idx++) { if (run_count != 1 && verbose) fprintf(stderr, "[ perf stat: executing run #%d ... ]\n", run_idx + 1); + + if (sync_run) + sync(); + status = run_perf_stat(argc, argv); } diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 952b4ae3d95..ffa493a2433 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -31,34 +31,36 @@ char debugfs_path[MAXPATHLEN]; #define CSW(x) .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_##x static struct event_symbol event_symbols[] = { - { CHW(CPU_CYCLES), "cpu-cycles", "cycles" }, - { CHW(INSTRUCTIONS), "instructions", "" }, - { CHW(CACHE_REFERENCES), "cache-references", "" }, - { CHW(CACHE_MISSES), "cache-misses", "" }, - { CHW(BRANCH_INSTRUCTIONS), "branch-instructions", "branches" }, - { CHW(BRANCH_MISSES), "branch-misses", "" }, - { CHW(BUS_CYCLES), "bus-cycles", "" }, - - { CSW(CPU_CLOCK), "cpu-clock", "" }, - { CSW(TASK_CLOCK), "task-clock", "" }, - { CSW(PAGE_FAULTS), "page-faults", "faults" }, - { CSW(PAGE_FAULTS_MIN), "minor-faults", "" }, - { CSW(PAGE_FAULTS_MAJ), "major-faults", "" }, - { CSW(CONTEXT_SWITCHES), "context-switches", "cs" }, - { CSW(CPU_MIGRATIONS), "cpu-migrations", "migrations" }, - { CSW(ALIGNMENT_FAULTS), "alignment-faults", "" }, - { CSW(EMULATION_FAULTS), "emulation-faults", "" }, + { CHW(CPU_CYCLES), "cpu-cycles", "cycles" }, + { CHW(STALLED_CYCLES_FRONTEND), "stalled-cycles-frontend", "idle-cycles-frontend" }, + { CHW(STALLED_CYCLES_BACKEND), "stalled-cycles-backend", "idle-cycles-backend" }, + { CHW(INSTRUCTIONS), "instructions", "" }, + { CHW(CACHE_REFERENCES), "cache-references", "" }, + { CHW(CACHE_MISSES), "cache-misses", "" }, + { CHW(BRANCH_INSTRUCTIONS), "branch-instructions", "branches" }, + { CHW(BRANCH_MISSES), "branch-misses", "" }, + { CHW(BUS_CYCLES), "bus-cycles", "" }, + + { CSW(CPU_CLOCK), "cpu-clock", "" }, + { CSW(TASK_CLOCK), "task-clock", "" }, + { CSW(PAGE_FAULTS), "page-faults", "faults" }, + { CSW(PAGE_FAULTS_MIN), "minor-faults", "" }, + { CSW(PAGE_FAULTS_MAJ), "major-faults", "" }, + { CSW(CONTEXT_SWITCHES), "context-switches", "cs" }, + { CSW(CPU_MIGRATIONS), "cpu-migrations", "migrations" }, + { CSW(ALIGNMENT_FAULTS), "alignment-faults", "" }, + { CSW(EMULATION_FAULTS), "emulation-faults", "" }, }; #define __PERF_EVENT_FIELD(config, name) \ ((config & PERF_EVENT_##name##_MASK) >> PERF_EVENT_##name##_SHIFT) -#define PERF_EVENT_RAW(config) __PERF_EVENT_FIELD(config, RAW) +#define PERF_EVENT_RAW(config) __PERF_EVENT_FIELD(config, RAW) #define PERF_EVENT_CONFIG(config) __PERF_EVENT_FIELD(config, CONFIG) -#define PERF_EVENT_TYPE(config) __PERF_EVENT_FIELD(config, TYPE) +#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[] = { +static const char *hw_event_names[PERF_COUNT_HW_MAX] = { "cycles", "instructions", "cache-references", @@ -66,11 +68,13 @@ static const char *hw_event_names[] = { "branches", "branch-misses", "bus-cycles", + "stalled-cycles-frontend", + "stalled-cycles-backend", }; -static const char *sw_event_names[] = { - "cpu-clock-msecs", - "task-clock-msecs", +static const char *sw_event_names[PERF_COUNT_SW_MAX] = { + "cpu-clock", + "task-clock", "page-faults", "context-switches", "CPU-migrations", @@ -307,7 +311,7 @@ const char *__event_name(int type, u64 config) switch (type) { case PERF_TYPE_HARDWARE: - if (config < PERF_COUNT_HW_MAX) + if (config < PERF_COUNT_HW_MAX && hw_event_names[config]) return hw_event_names[config]; return "unknown-hardware"; @@ -333,7 +337,7 @@ const char *__event_name(int type, u64 config) } case PERF_TYPE_SOFTWARE: - if (config < PERF_COUNT_SW_MAX) + if (config < PERF_COUNT_SW_MAX && sw_event_names[config]) return sw_event_names[config]; return "unknown-software"; @@ -648,13 +652,15 @@ static int check_events(const char *str, unsigned int i) int n; n = strlen(event_symbols[i].symbol); - if (!strncmp(str, event_symbols[i].symbol, n)) + if (!strncasecmp(str, event_symbols[i].symbol, n)) return n; n = strlen(event_symbols[i].alias); - if (n) - if (!strncmp(str, event_symbols[i].alias, n)) + if (n) { + if (!strncasecmp(str, event_symbols[i].alias, n)) return n; + } + return 0; } @@ -718,15 +724,19 @@ parse_numeric_event(const char **strp, struct perf_event_attr *attr) return EVT_FAILED; } -static enum event_result +static int parse_event_modifier(const char **strp, struct perf_event_attr *attr) { const char *str = *strp; int exclude = 0; int eu = 0, ek = 0, eh = 0, precise = 0; - if (*str++ != ':') + if (!*str) return 0; + + if (*str++ != ':') + return -1; + while (*str) { if (*str == 'u') { if (!exclude) @@ -747,14 +757,16 @@ parse_event_modifier(const char **strp, struct perf_event_attr *attr) ++str; } - if (str >= *strp + 2) { - *strp = str; - attr->exclude_user = eu; - attr->exclude_kernel = ek; - attr->exclude_hv = eh; - attr->precise_ip = precise; - return 1; - } + if (str < *strp + 2) + return -1; + + *strp = str; + + attr->exclude_user = eu; + attr->exclude_kernel = ek; + attr->exclude_hv = eh; + attr->precise_ip = precise; + return 0; } @@ -797,7 +809,12 @@ parse_event_symbols(const struct option *opt, const char **str, return EVT_FAILED; modifier: - parse_event_modifier(str, attr); + if (parse_event_modifier(str, attr) < 0) { + fprintf(stderr, "invalid event modifier: '%s'\n", *str); + fprintf(stderr, "Run 'perf list' for a list of valid events and modifiers\n"); + + return EVT_FAILED; + } return ret; } @@ -912,7 +929,7 @@ void print_tracepoint_events(const char *subsys_glob, const char *event_glob) snprintf(evt_path, MAXPATHLEN, "%s:%s", sys_dirent.d_name, evt_dirent.d_name); - printf(" %-42s [%s]\n", evt_path, + printf(" %-50s [%s]\n", evt_path, event_type_descriptors[PERF_TYPE_TRACEPOINT]); } closedir(evt_dir); @@ -977,7 +994,7 @@ void print_events_type(u8 type) else snprintf(name, sizeof(name), "%s", syms->symbol); - printf(" %-42s [%s]\n", name, + printf(" %-50s [%s]\n", name, event_type_descriptors[type]); } } @@ -995,11 +1012,10 @@ int print_hwcache_events(const char *event_glob) for (i = 0; i < PERF_COUNT_HW_CACHE_RESULT_MAX; i++) { char *name = event_cache_name(type, op, i); - if (event_glob != NULL && - !strglobmatch(name, event_glob)) + if (event_glob != NULL && !strglobmatch(name, event_glob)) continue; - printf(" %-42s [%s]\n", name, + printf(" %-50s [%s]\n", name, event_type_descriptors[PERF_TYPE_HW_CACHE]); ++printed; } @@ -1009,14 +1025,16 @@ int print_hwcache_events(const char *event_glob) return printed; } +#define MAX_NAME_LEN 100 + /* * Print the help text for the event symbols: */ void print_events(const char *event_glob) { - struct event_symbol *syms = event_symbols; unsigned int i, type, prev_type = -1, printed = 0, ntypes_printed = 0; - char name[40]; + struct event_symbol *syms = event_symbols; + char name[MAX_NAME_LEN]; printf("\n"); printf("List of pre-defined events (to be used in -e):\n"); @@ -1036,10 +1054,10 @@ void print_events(const char *event_glob) continue; if (strlen(syms->alias)) - sprintf(name, "%s OR %s", syms->symbol, syms->alias); + snprintf(name, MAX_NAME_LEN, "%s OR %s", syms->symbol, syms->alias); else - strcpy(name, syms->symbol); - printf(" %-42s [%s]\n", name, + strncpy(name, syms->symbol, MAX_NAME_LEN); + printf(" %-50s [%s]\n", name, event_type_descriptors[type]); prev_type = type; @@ -1056,12 +1074,12 @@ void print_events(const char *event_glob) return; printf("\n"); - printf(" %-42s [%s]\n", + printf(" %-50s [%s]\n", "rNNN (see 'perf list --help' on how to encode it)", event_type_descriptors[PERF_TYPE_RAW]); printf("\n"); - printf(" %-42s [%s]\n", + printf(" %-50s [%s]\n", "mem:<addr>[:access]", event_type_descriptors[PERF_TYPE_BREAKPOINT]); printf("\n"); diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c index f5e38451fdc..8b0eff8b828 100644 --- a/tools/perf/util/python.c +++ b/tools/perf/util/python.c @@ -810,6 +810,9 @@ static struct { { "COUNT_HW_CACHE_RESULT_ACCESS", PERF_COUNT_HW_CACHE_RESULT_ACCESS }, { "COUNT_HW_CACHE_RESULT_MISS", PERF_COUNT_HW_CACHE_RESULT_MISS }, + { "COUNT_HW_STALLED_CYCLES_FRONTEND", PERF_COUNT_HW_STALLED_CYCLES_FRONTEND }, + { "COUNT_HW_STALLED_CYCLES_BACKEND", PERF_COUNT_HW_STALLED_CYCLES_BACKEND }, + { "COUNT_SW_CPU_CLOCK", PERF_COUNT_SW_CPU_CLOCK }, { "COUNT_SW_TASK_CLOCK", PERF_COUNT_SW_TASK_CLOCK }, { "COUNT_SW_PAGE_FAULTS", PERF_COUNT_SW_PAGE_FAULTS }, |