diff --git a/Documentation/perf_counter/builtin-record.c b/Documentation/perf_counter/builtin-record.c index 6fa6ed66495..ec2b787b23b 100644 --- a/Documentation/perf_counter/builtin-record.c +++ b/Documentation/perf_counter/builtin-record.c @@ -4,7 +4,6 @@ #include "util/util.h" #include "util/parse-options.h" #include "util/parse-events.h" -#include "util/exec_cmd.h" #include @@ -400,7 +399,7 @@ static const char * const record_usage[] = { static char events_help_msg[EVENTS_HELP_MAX]; -const struct option options[] = { +static const struct option options[] = { OPT_CALLBACK('e', "event", NULL, "event", events_help_msg, parse_events), OPT_INTEGER('c', "count", &default_interval, diff --git a/Documentation/perf_counter/builtin-stat.c b/Documentation/perf_counter/builtin-stat.c index c1053d820c1..e7cb9412212 100644 --- a/Documentation/perf_counter/builtin-stat.c +++ b/Documentation/perf_counter/builtin-stat.c @@ -1,35 +1,5 @@ /* - * kerneltop.c: show top kernel functions - performance counters showcase - - Build with: - - cc -O6 -Wall -c -o kerneltop.o kerneltop.c -lrt - - Sample output: - ------------------------------------------------------------------------------- - KernelTop: 2669 irqs/sec [NMI, cache-misses/cache-refs], (all, cpu: 2) ------------------------------------------------------------------------------- - - weight RIP kernel function - ______ ________________ _______________ - - 35.20 - ffffffff804ce74b : skb_copy_and_csum_dev - 33.00 - ffffffff804cb740 : sock_alloc_send_skb - 31.26 - ffffffff804ce808 : skb_push - 22.43 - ffffffff80510004 : tcp_established_options - 19.00 - ffffffff8027d250 : find_get_page - 15.76 - ffffffff804e4fc9 : eth_type_trans - 15.20 - ffffffff804d8baa : dst_release - 14.86 - ffffffff804cf5d8 : skb_release_head_state - 14.00 - ffffffff802217d5 : read_hpet - 12.00 - ffffffff804ffb7f : __ip_local_out - 11.97 - ffffffff804fc0c8 : ip_local_deliver_finish - 8.54 - ffffffff805001a3 : ip_queue_xmit - */ - -/* - * perfstat: /usr/bin/time -alike performance counter statistics utility + * perf stat: /usr/bin/time -alike performance counter statistics utility It summarizes the counter events of all tasks (and child tasks), covering all CPUs that the command (or workload) executes on. @@ -38,59 +8,38 @@ Sample output: - $ ./perfstat -e 1 -e 3 -e 5 ls -lR /usr/include/ >/dev/null + $ perf stat -e 1 -e 3 -e 5 ls -lR /usr/include/ >/dev/null Performance counter stats for 'ls': 163516953 instructions 2295 cache-misses 2855182 branch-misses + * + * Copyright (C) 2008, Red Hat Inc, Ingo Molnar + * + * Improvements and fixes by: + * + * Arjan van de Ven + * Yanmin Zhang + * Wu Fengguang + * Mike Galbraith + * Paul Mackerras + * + * Released under the GPL v2. (and only v2, not any later version) */ - /* - * Copyright (C) 2008, Red Hat Inc, Ingo Molnar - * - * Improvements and fixes by: - * - * Arjan van de Ven - * Yanmin Zhang - * Wu Fengguang - * Mike Galbraith - * Paul Mackerras - * - * Released under the GPL v2. (and only v2, not any later version) - */ - #include "perf.h" #include "util/util.h" +#include "util/parse-options.h" +#include "util/parse-events.h" -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include #include -#include -#include -#include - -#include -#include - -#define EVENT_MASK_KERNEL 1 -#define EVENT_MASK_USER 2 static int system_wide = 0; +static int inherit = 1; -static int nr_counters = 0; -static __u64 event_id[MAX_COUNTERS] = { +static __u64 default_event_id[MAX_COUNTERS] = { EID(PERF_TYPE_SOFTWARE, PERF_COUNT_TASK_CLOCK), EID(PERF_TYPE_SOFTWARE, PERF_COUNT_CONTEXT_SWITCHES), EID(PERF_TYPE_SOFTWARE, PERF_COUNT_CPU_MIGRATIONS), @@ -101,20 +50,15 @@ static __u64 event_id[MAX_COUNTERS] = { EID(PERF_TYPE_HARDWARE, PERF_COUNT_CACHE_REFERENCES), EID(PERF_TYPE_HARDWARE, PERF_COUNT_CACHE_MISSES), }; + static int default_interval = 100000; static int event_count[MAX_COUNTERS]; static int fd[MAX_NR_CPUS][MAX_COUNTERS]; -static int event_mask[MAX_COUNTERS]; -static int tid = -1; -static int profile_cpu = -1; +static int target_pid = -1; static int nr_cpus = 0; -static int nmi = 1; -static int group = 0; static unsigned int page_size; -static int zero; - static int scale = 1; static const unsigned int default_count[] = { @@ -126,197 +70,6 @@ static const unsigned int default_count[] = { 10000, }; -static char *hw_event_names[] = { - "CPU cycles", - "instructions", - "cache references", - "cache misses", - "branches", - "branch misses", - "bus cycles", -}; - -static char *sw_event_names[] = { - "cpu clock ticks", - "task clock ticks", - "pagefaults", - "context switches", - "CPU migrations", - "minor faults", - "major faults", -}; - -struct event_symbol { - __u64 event; - char *symbol; -}; - -static struct event_symbol event_symbols[] = { - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_CPU_CYCLES), "cpu-cycles", }, - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_CPU_CYCLES), "cycles", }, - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_INSTRUCTIONS), "instructions", }, - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_CACHE_REFERENCES), "cache-references", }, - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_CACHE_MISSES), "cache-misses", }, - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_BRANCH_INSTRUCTIONS), "branch-instructions", }, - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_BRANCH_INSTRUCTIONS), "branches", }, - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_BRANCH_MISSES), "branch-misses", }, - {EID(PERF_TYPE_HARDWARE, PERF_COUNT_BUS_CYCLES), "bus-cycles", }, - - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_CPU_CLOCK), "cpu-clock", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_TASK_CLOCK), "task-clock", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_PAGE_FAULTS), "page-faults", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_PAGE_FAULTS), "faults", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_PAGE_FAULTS_MIN), "minor-faults", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_PAGE_FAULTS_MAJ), "major-faults", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_CONTEXT_SWITCHES), "context-switches", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_CONTEXT_SWITCHES), "cs", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_CPU_MIGRATIONS), "cpu-migrations", }, - {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_CPU_MIGRATIONS), "migrations", }, -}; - -#define __PERF_COUNTER_FIELD(config, name) \ - ((config & PERF_COUNTER_##name##_MASK) >> PERF_COUNTER_##name##_SHIFT) - -#define PERF_COUNTER_RAW(config) __PERF_COUNTER_FIELD(config, RAW) -#define PERF_COUNTER_CONFIG(config) __PERF_COUNTER_FIELD(config, CONFIG) -#define PERF_COUNTER_TYPE(config) __PERF_COUNTER_FIELD(config, TYPE) -#define PERF_COUNTER_ID(config) __PERF_COUNTER_FIELD(config, EVENT) - -static void display_events_help(void) -{ - unsigned int i; - __u64 e; - - printf( - " -e EVENT --event=EVENT # symbolic-name abbreviations"); - - for (i = 0; i < ARRAY_SIZE(event_symbols); i++) { - int type, id; - - e = event_symbols[i].event; - type = PERF_COUNTER_TYPE(e); - id = PERF_COUNTER_ID(e); - - printf("\n %d:%d: %-20s", - type, id, event_symbols[i].symbol); - } - - printf("\n" - " rNNN: raw PMU events (eventsel+umask)\n\n"); -} - -static void display_help(void) -{ - printf( - "Usage: perfstat [] \n\n" - "PerfStat Options (up to %d event types can be specified):\n\n", - MAX_COUNTERS); - - display_events_help(); - - printf( - " -l # scale counter values\n" - " -a # system-wide collection\n"); - exit(0); -} - -static char *event_name(int ctr) -{ - __u64 config = event_id[ctr]; - int type = PERF_COUNTER_TYPE(config); - int id = PERF_COUNTER_ID(config); - static char buf[32]; - - if (PERF_COUNTER_RAW(config)) { - sprintf(buf, "raw 0x%llx", PERF_COUNTER_CONFIG(config)); - return buf; - } - - switch (type) { - case PERF_TYPE_HARDWARE: - if (id < PERF_HW_EVENTS_MAX) - return hw_event_names[id]; - return "unknown-hardware"; - - case PERF_TYPE_SOFTWARE: - if (id < PERF_SW_EVENTS_MAX) - return sw_event_names[id]; - return "unknown-software"; - - default: - break; - } - - return "unknown"; -} - -/* - * Each event can have multiple symbolic names. - * Symbolic names are (almost) exactly matched. - */ -static __u64 match_event_symbols(char *str) -{ - __u64 config, id; - int type; - unsigned int i; - char mask_str[4]; - - if (sscanf(str, "r%llx", &config) == 1) - return config | PERF_COUNTER_RAW_MASK; - - switch (sscanf(str, "%d:%llu:%2s", &type, &id, mask_str)) { - case 3: - if (strchr(mask_str, 'k')) - event_mask[nr_counters] |= EVENT_MASK_USER; - if (strchr(mask_str, 'u')) - event_mask[nr_counters] |= EVENT_MASK_KERNEL; - case 2: - return EID(type, id); - - default: - break; - } - - for (i = 0; i < ARRAY_SIZE(event_symbols); i++) { - if (!strncmp(str, event_symbols[i].symbol, - strlen(event_symbols[i].symbol))) - return event_symbols[i].event; - } - - return ~0ULL; -} - -static int parse_events(char *str) -{ - __u64 config; - -again: - if (nr_counters == MAX_COUNTERS) - return -1; - - config = match_event_symbols(str); - if (config == ~0ULL) - return -1; - - event_id[nr_counters] = config; - nr_counters++; - - str = strstr(str, ","); - if (str) { - str++; - goto again; - } - - return 0; -} - - -/* - * perfstat - */ - -char fault_here[1000000]; - static void create_perfstat_counter(int counter) { struct perf_counter_hw_event hw_event; @@ -324,7 +77,7 @@ static void create_perfstat_counter(int counter) memset(&hw_event, 0, sizeof(hw_event)); hw_event.config = event_id[counter]; hw_event.record_type = 0; - hw_event.nmi = 0; + hw_event.nmi = 1; hw_event.exclude_kernel = event_mask[counter] & EVENT_MASK_KERNEL; hw_event.exclude_user = event_mask[counter] & EVENT_MASK_USER; @@ -343,7 +96,7 @@ static void create_perfstat_counter(int counter) } } } else { - hw_event.inherit = 1; + hw_event.inherit = inherit; hw_event.disabled = 1; fd[0][counter] = sys_perf_counter_open(&hw_event, 0, -1, -1, 0); @@ -355,7 +108,7 @@ static void create_perfstat_counter(int counter) } } -int do_perfstat(int argc, char *argv[]) +int do_perfstat(int argc, const char **argv) { unsigned long long t0, t1; int counter; @@ -369,12 +122,6 @@ int do_perfstat(int argc, char *argv[]) for (counter = 0; counter < nr_counters; counter++) create_perfstat_counter(counter); - argc -= optind; - argv += optind; - - if (!argc) - display_help(); - /* * Enable counters and exec the command: */ @@ -384,7 +131,7 @@ int do_perfstat(int argc, char *argv[]) if ((pid = fork()) < 0) perror("failed to fork"); if (!pid) { - if (execvp(argv[0], argv)) { + if (execvp(argv[0], (char **)argv)) { perror(argv[0]); exit(-1); } @@ -458,70 +205,45 @@ int do_perfstat(int argc, char *argv[]) return 0; } -static void process_options(int argc, char **argv) +static void skip_signal(int signo) { - int error = 0, counter; +} - for (;;) { - int option_index = 0; - /** Options for getopt */ - static struct option long_options[] = { - {"count", required_argument, NULL, 'c'}, - {"cpu", required_argument, NULL, 'C'}, - {"delay", required_argument, NULL, 'd'}, - {"dump_symtab", no_argument, NULL, 'D'}, - {"event", required_argument, NULL, 'e'}, - {"filter", required_argument, NULL, 'f'}, - {"group", required_argument, NULL, 'g'}, - {"help", no_argument, NULL, 'h'}, - {"nmi", required_argument, NULL, 'n'}, - {"munmap_info", no_argument, NULL, 'U'}, - {"pid", required_argument, NULL, 'p'}, - {"realtime", required_argument, NULL, 'r'}, - {"scale", no_argument, NULL, 'l'}, - {"symbol", required_argument, NULL, 's'}, - {"stat", no_argument, NULL, 'S'}, - {"vmlinux", required_argument, NULL, 'x'}, - {"zero", no_argument, NULL, 'z'}, - {NULL, 0, NULL, 0 } - }; - int c = getopt_long(argc, argv, "+:ac:C:d:De:f:g:hln:m:p:r:s:Sx:zMU", - long_options, &option_index); - if (c == -1) - break; +static const char * const stat_usage[] = { + "perf stat [] ", + NULL +}; - switch (c) { - case 'a': system_wide = 1; break; - case 'c': default_interval = atoi(optarg); break; - case 'C': - /* CPU and PID are mutually exclusive */ - if (tid != -1) { - printf("WARNING: CPU switch overriding PID\n"); - sleep(1); - tid = -1; - } - profile_cpu = atoi(optarg); break; +static char events_help_msg[EVENTS_HELP_MAX]; - case 'e': error = parse_events(optarg); break; +static const struct option options[] = { + OPT_CALLBACK('e', "event", NULL, "event", + events_help_msg, parse_events), + OPT_INTEGER('c', "count", &default_interval, + "event period to sample"), + OPT_BOOLEAN('i', "inherit", &inherit, + "child tasks inherit counters"), + OPT_INTEGER('p', "pid", &target_pid, + "stat events on existing pid"), + OPT_BOOLEAN('a', "all-cpus", &system_wide, + "system-wide collection from all CPUs"), + OPT_BOOLEAN('l', "scale", &scale, + "scale/normalize counters"), + OPT_END() +}; - case 'g': group = atoi(optarg); break; - case 'h': display_help(); break; - case 'l': scale = 1; break; - case 'n': nmi = atoi(optarg); break; - case 'p': - /* CPU and PID are mutually exclusive */ - if (profile_cpu != -1) { - printf("WARNING: PID switch overriding CPU\n"); - sleep(1); - profile_cpu = -1; - } - tid = atoi(optarg); break; - case 'z': zero = 1; break; - default: error = 1; break; - } - } - if (error) - display_help(); +int cmd_stat(int argc, const char **argv, const char *prefix) +{ + int counter; + + page_size = sysconf(_SC_PAGE_SIZE); + + create_events_help(events_help_msg); + memcpy(event_id, default_event_id, sizeof(default_event_id)); + + argc = parse_options(argc, argv, options, stat_usage, 0); + if (!argc) + usage_with_options(stat_usage, options); if (!nr_counters) { nr_counters = 8; @@ -533,18 +255,6 @@ static void process_options(int argc, char **argv) event_count[counter] = default_interval; } -} - -static void skip_signal(int signo) -{ -} - -int cmd_stat(int argc, char **argv, const char *prefix) -{ - page_size = sysconf(_SC_PAGE_SIZE); - - process_options(argc, argv); - nr_cpus = sysconf(_SC_NPROCESSORS_ONLN); assert(nr_cpus <= MAX_NR_CPUS); assert(nr_cpus >= 0); diff --git a/Documentation/perf_counter/util/parse-events.c b/Documentation/perf_counter/util/parse-events.c index 77d0917d55d..88c903eb260 100644 --- a/Documentation/perf_counter/util/parse-events.c +++ b/Documentation/perf_counter/util/parse-events.c @@ -8,6 +8,7 @@ int nr_counters; __u64 event_id[MAX_COUNTERS] = { }; +int event_mask[MAX_COUNTERS]; struct event_symbol { __u64 event; @@ -37,6 +38,64 @@ static struct event_symbol event_symbols[] = { {EID(PERF_TYPE_SOFTWARE, PERF_COUNT_CPU_MIGRATIONS), "migrations", }, }; +#define __PERF_COUNTER_FIELD(config, name) \ + ((config & PERF_COUNTER_##name##_MASK) >> PERF_COUNTER_##name##_SHIFT) + +#define PERF_COUNTER_RAW(config) __PERF_COUNTER_FIELD(config, RAW) +#define PERF_COUNTER_CONFIG(config) __PERF_COUNTER_FIELD(config, CONFIG) +#define PERF_COUNTER_TYPE(config) __PERF_COUNTER_FIELD(config, TYPE) +#define PERF_COUNTER_ID(config) __PERF_COUNTER_FIELD(config, EVENT) + +static char *hw_event_names[] = { + "CPU cycles", + "instructions", + "cache references", + "cache misses", + "branches", + "branch misses", + "bus cycles", +}; + +static char *sw_event_names[] = { + "cpu clock ticks", + "task clock ticks", + "pagefaults", + "context switches", + "CPU migrations", + "minor faults", + "major faults", +}; + +char *event_name(int ctr) +{ + __u64 config = event_id[ctr]; + int type = PERF_COUNTER_TYPE(config); + int id = PERF_COUNTER_ID(config); + static char buf[32]; + + if (PERF_COUNTER_RAW(config)) { + sprintf(buf, "raw 0x%llx", PERF_COUNTER_CONFIG(config)); + return buf; + } + + switch (type) { + case PERF_TYPE_HARDWARE: + if (id < PERF_HW_EVENTS_MAX) + return hw_event_names[id]; + return "unknown-hardware"; + + case PERF_TYPE_SOFTWARE: + if (id < PERF_SW_EVENTS_MAX) + return sw_event_names[id]; + return "unknown-software"; + + default: + break; + } + + return "unknown"; +} + /* * Each event can have multiple symbolic names. * Symbolic names are (almost) exactly matched. @@ -46,12 +105,23 @@ static __u64 match_event_symbols(const char *str) __u64 config, id; int type; unsigned int i; + char mask_str[4]; if (sscanf(str, "r%llx", &config) == 1) return config | PERF_COUNTER_RAW_MASK; - if (sscanf(str, "%d:%llu", &type, &id) == 2) - return EID(type, id); + switch (sscanf(str, "%d:%llu:%2s", &type, &id, mask_str)) { + case 3: + if (strchr(mask_str, 'k')) + event_mask[nr_counters] |= EVENT_MASK_USER; + if (strchr(mask_str, 'u')) + event_mask[nr_counters] |= EVENT_MASK_KERNEL; + case 2: + return EID(type, id); + + default: + break; + } for (i = 0; i < ARRAY_SIZE(event_symbols); i++) { if (!strncmp(str, event_symbols[i].symbol, @@ -86,14 +156,6 @@ again: return 0; } -#define __PERF_COUNTER_FIELD(config, name) \ - ((config & PERF_COUNTER_##name##_MASK) >> PERF_COUNTER_##name##_SHIFT) - -#define PERF_COUNTER_RAW(config) __PERF_COUNTER_FIELD(config, RAW) -#define PERF_COUNTER_CONFIG(config) __PERF_COUNTER_FIELD(config, CONFIG) -#define PERF_COUNTER_TYPE(config) __PERF_COUNTER_FIELD(config, TYPE) -#define PERF_COUNTER_ID(config) __PERF_COUNTER_FIELD(config, EVENT) - /* * Create the help text for the event symbols: */ diff --git a/Documentation/perf_counter/util/parse-events.h b/Documentation/perf_counter/util/parse-events.h index 6e2ebe5ff7d..0da306bb902 100644 --- a/Documentation/perf_counter/util/parse-events.h +++ b/Documentation/perf_counter/util/parse-events.h @@ -1,6 +1,16 @@ +/* + * Parse symbolic events/counts passed in as options: + */ + extern int nr_counters; extern __u64 event_id[MAX_COUNTERS]; +extern int event_mask[MAX_COUNTERS]; + +#define EVENT_MASK_KERNEL 1 +#define EVENT_MASK_USER 2 + +extern char *event_name(int ctr); extern int parse_events(const struct option *opt, const char *str, int unset);