From 361c99a661a78ed22264649440e87fe4fe8da1f2 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 11 Jan 2011 20:56:53 -0200 Subject: perf evsel: Introduce perf_evlist Killing two more perf wide global variables: nr_counters and evsel_list as a list_head. There are more operations that will need more fields in perf_evlist, like the pollfd for polling all the fds in a list of evsel instances. Use option->value to pass the evsel_list to parse_{events,filters}. LKML-Reference: Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 53 ++++++++++++++++++++++++++++++++++++ tools/perf/util/evlist.h | 19 +++++++++++++ tools/perf/util/header.c | 17 +++++++----- tools/perf/util/header.h | 7 +++-- tools/perf/util/include/linux/list.h | 1 + tools/perf/util/parse-events.c | 51 ++++++---------------------------- tools/perf/util/parse-events.h | 7 ----- 7 files changed, 97 insertions(+), 58 deletions(-) create mode 100644 tools/perf/util/evlist.c create mode 100644 tools/perf/util/evlist.h (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c new file mode 100644 index 000000000000..7b4faec23737 --- /dev/null +++ b/tools/perf/util/evlist.c @@ -0,0 +1,53 @@ +#include "evlist.h" +#include "evsel.h" +#include "util.h" + +struct perf_evlist *perf_evlist__new(void) +{ + struct perf_evlist *evlist = zalloc(sizeof(*evlist)); + + if (evlist != NULL) { + INIT_LIST_HEAD(&evlist->entries); + } + + return evlist; +} + +static void perf_evlist__purge(struct perf_evlist *evlist) +{ + struct perf_evsel *pos, *n; + + list_for_each_entry_safe(pos, n, &evlist->entries, node) { + list_del_init(&pos->node); + perf_evsel__delete(pos); + } + + evlist->nr_entries = 0; +} + +void perf_evlist__delete(struct perf_evlist *evlist) +{ + perf_evlist__purge(evlist); + free(evlist); +} + +void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry) +{ + list_add_tail(&entry->node, &evlist->entries); + ++evlist->nr_entries; +} + +int perf_evlist__add_default(struct perf_evlist *evlist) +{ + struct perf_event_attr attr = { + .type = PERF_TYPE_HARDWARE, + .config = PERF_COUNT_HW_CPU_CYCLES, + }; + struct perf_evsel *evsel = perf_evsel__new(&attr, 0); + + if (evsel == NULL) + return -ENOMEM; + + perf_evlist__add(evlist, evsel); + return 0; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h new file mode 100644 index 000000000000..48db91a8abf3 --- /dev/null +++ b/tools/perf/util/evlist.h @@ -0,0 +1,19 @@ +#ifndef __PERF_EVLIST_H +#define __PERF_EVLIST_H 1 + +#include + +struct perf_evlist { + struct list_head entries; + int nr_entries; +}; + +struct perf_evsel; + +struct perf_evlist *perf_evlist__new(void); +void perf_evlist__delete(struct perf_evlist *evlist); + +void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); +int perf_evlist__add_default(struct perf_evlist *evlist); + +#endif /* __PERF_EVLIST_H */ diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index f6a929e74981..f0138d472339 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -8,6 +8,7 @@ #include #include +#include "evlist.h" #include "util.h" #include "header.h" #include "../perf.h" @@ -428,7 +429,8 @@ static bool perf_session__read_build_ids(struct perf_session *self, bool with_hi return ret; } -static int perf_header__adds_write(struct perf_header *self, int fd) +static int perf_header__adds_write(struct perf_header *self, + struct perf_evlist *evlist, int fd) { int nr_sections; struct perf_session *session; @@ -463,7 +465,7 @@ static int perf_header__adds_write(struct perf_header *self, int fd) /* Write trace info */ trace_sec->offset = lseek(fd, 0, SEEK_CUR); - read_tracing_data(fd, &evsel_list); + read_tracing_data(fd, &evlist->entries); trace_sec->size = lseek(fd, 0, SEEK_CUR) - trace_sec->offset; } @@ -513,7 +515,8 @@ int perf_header__write_pipe(int fd) return 0; } -int perf_header__write(struct perf_header *self, int fd, bool at_exit) +int perf_header__write(struct perf_header *self, struct perf_evlist *evlist, + int fd, bool at_exit) { struct perf_file_header f_header; struct perf_file_attr f_attr; @@ -566,7 +569,7 @@ int perf_header__write(struct perf_header *self, int fd, bool at_exit) self->data_offset = lseek(fd, 0, SEEK_CUR); if (at_exit) { - err = perf_header__adds_write(self, fd); + err = perf_header__adds_write(self, evlist, fd); if (err < 0) return err; } @@ -1133,7 +1136,7 @@ int event__process_event_type(event_t *self, return 0; } -int event__synthesize_tracing_data(int fd, struct list_head *pattrs, +int event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, event__handler_t process, struct perf_session *session __unused) { @@ -1144,7 +1147,7 @@ int event__synthesize_tracing_data(int fd, struct list_head *pattrs, memset(&ev, 0, sizeof(ev)); ev.tracing_data.header.type = PERF_RECORD_HEADER_TRACING_DATA; - size = read_tracing_data_size(fd, pattrs); + size = read_tracing_data_size(fd, &evlist->entries); if (size <= 0) return size; aligned_size = ALIGN(size, sizeof(u64)); @@ -1154,7 +1157,7 @@ int event__synthesize_tracing_data(int fd, struct list_head *pattrs, process(&ev, NULL, session); - err = read_tracing_data(fd, pattrs); + err = read_tracing_data(fd, &evlist->entries); write_padded(fd, NULL, 0, padding); return aligned_size; diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 33f16be7b72f..65afd7f74e0d 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -65,8 +65,11 @@ struct perf_header { int perf_header__init(struct perf_header *self); void perf_header__exit(struct perf_header *self); +struct perf_evlist; + int perf_header__read(struct perf_session *session, int fd); -int perf_header__write(struct perf_header *self, int fd, bool at_exit); +int perf_header__write(struct perf_header *self, struct perf_evlist *evlist, + int fd, bool at_exit); int perf_header__write_pipe(int fd); int perf_header__add_attr(struct perf_header *self, @@ -113,7 +116,7 @@ int event__synthesize_event_types(event__handler_t process, int event__process_event_type(event_t *self, struct perf_session *session); -int event__synthesize_tracing_data(int fd, struct list_head *pattrs, +int event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, event__handler_t process, struct perf_session *session); int event__process_tracing_data(event_t *self, diff --git a/tools/perf/util/include/linux/list.h b/tools/perf/util/include/linux/list.h index f5ca26e53fbb..356c7e467b83 100644 --- a/tools/perf/util/include/linux/list.h +++ b/tools/perf/util/include/linux/list.h @@ -1,3 +1,4 @@ +#include #include "../../../../include/linux/list.h" #ifndef PERF_LIST_H diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 135f69baf966..d3086cecd2dd 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -1,6 +1,7 @@ #include "../../../include/linux/hw_breakpoint.h" #include "util.h" #include "../perf.h" +#include "evlist.h" #include "evsel.h" #include "parse-options.h" #include "parse-events.h" @@ -11,10 +12,6 @@ #include "header.h" #include "debugfs.h" -int nr_counters; - -LIST_HEAD(evsel_list); - struct event_symbol { u8 type; u64 config; @@ -778,8 +775,9 @@ modifier: return ret; } -int parse_events(const struct option *opt __used, const char *str, int unset __used) +int parse_events(const struct option *opt, const char *str, int unset __used) { + struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; struct perf_event_attr attr; enum event_result ret; @@ -794,12 +792,10 @@ 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, - nr_counters); + evsel = perf_evsel__new(&attr, evlist->nr_entries); if (evsel == NULL) return -1; - list_add_tail(&evsel->node, &evsel_list); - ++nr_counters; + perf_evlist__add(evlist, evsel); } if (*str == 0) @@ -813,13 +809,14 @@ int parse_events(const struct option *opt __used, const char *str, int unset __u return 0; } -int parse_filter(const struct option *opt __used, const char *str, +int parse_filter(const struct option *opt, const char *str, int unset __used) { + struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; struct perf_evsel *last = NULL; - if (!list_empty(&evsel_list)) - last = list_entry(evsel_list.prev, struct perf_evsel, node); + if (evlist->nr_entries > 0) + last = list_entry(evlist->entries.prev, struct perf_evsel, node); if (last == NULL || last->attr.type != PERF_TYPE_TRACEPOINT) { fprintf(stderr, @@ -981,33 +978,3 @@ void print_events(void) exit(129); } - -int perf_evsel_list__create_default(void) -{ - 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; - - list_add(&evsel->node, &evsel_list); - ++nr_counters; - return 0; -} - -void perf_evsel_list__delete(void) -{ - struct perf_evsel *pos, *n; - - list_for_each_entry_safe(pos, n, &evsel_list, node) { - list_del_init(&pos->node); - perf_evsel__delete(pos); - } - nr_counters = 0; -} diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h index 458e3ecf17af..cf7e94abb676 100644 --- a/tools/perf/util/parse-events.h +++ b/tools/perf/util/parse-events.h @@ -9,11 +9,6 @@ struct list_head; struct perf_evsel; -extern struct list_head evsel_list; - -int perf_evsel_list__create_default(void); -void perf_evsel_list__delete(void); - struct option; struct tracepoint_path { @@ -25,8 +20,6 @@ struct tracepoint_path { extern struct tracepoint_path *tracepoint_id_to_path(u64 config); extern bool have_tracepoints(struct list_head *evlist); -extern int nr_counters; - const char *event_name(struct perf_evsel *event); extern const char *__event_name(int type, u64 config); -- cgit v1.2.3 From 5c581041cf97aa7980b442de81ddea8273d6dcde Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 11 Jan 2011 22:30:02 -0200 Subject: perf evlist: Adopt the pollfd array Allocating just the space needed for nr_cpus * nr_threads * nr_evsels, not the MAX_NR_CPUS and counters. LKML-Reference: Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-record.c | 20 +++++++------------- tools/perf/builtin-top.c | 26 +++++++++++--------------- tools/perf/util/evlist.c | 9 +++++++++ tools/perf/util/evlist.h | 6 ++++++ 4 files changed, 33 insertions(+), 28 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 252ace873d32..1614d89b4765 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -72,9 +72,6 @@ static struct perf_evlist *evsel_list; static long samples = 0; static u64 bytes_written = 0; -static struct pollfd *event_array; - -static int nr_poll = 0; static int nr_cpu = 0; static int file_new = 1; @@ -432,9 +429,9 @@ try_again: exit(-1); } - event_array[nr_poll].fd = FD(evsel, nr_cpu, thread_index); - event_array[nr_poll].events = POLLIN; - nr_poll++; + evlist->pollfd[evlist->nr_fds].fd = FD(evsel, nr_cpu, thread_index); + evlist->pollfd[evlist->nr_fds].events = POLLIN; + evlist->nr_fds++; } if (filter != NULL) { @@ -793,7 +790,7 @@ static int __cmd_record(int argc, const char **argv) if (hits == samples) { if (done) break; - err = poll(event_array, nr_poll, -1); + err = poll(evsel_list->pollfd, evsel_list->nr_fds, -1); waking++; } @@ -948,9 +945,8 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) 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)); - if (!event_array) + + if (perf_evlist__alloc_pollfd(evsel_list, cpus->nr, threads->nr) < 0) goto out_free_fd; if (user_interval != ULLONG_MAX) @@ -968,13 +964,11 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) } else { fprintf(stderr, "frequency and count are zero, aborting\n"); err = -EINVAL; - goto out_free_event_array; + goto out_free_fd; } err = __cmd_record(argc, argv); -out_free_event_array: - free(event_array); out_free_fd: thread_map__delete(threads); threads = NULL; diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 216b62ed4b89..1bc465215fc6 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -1193,8 +1193,6 @@ static void perf_session__mmap_read_counter(struct perf_session *self, md->prev = old; } -static struct pollfd *event_array; - static void perf_session__mmap_read(struct perf_session *self) { struct perf_evsel *counter; @@ -1212,10 +1210,10 @@ static void perf_session__mmap_read(struct perf_session *self) } } -int nr_poll; int group_fd; -static void start_counter(int i, struct perf_evsel *evsel) +static void start_counter(int i, struct perf_evlist *evlist, + struct perf_evsel *evsel) { struct xyarray *mmap_array = evsel->priv; struct mmap_data *mm; @@ -1281,9 +1279,9 @@ try_again: if (group && group_fd == -1) group_fd = FD(evsel, i, thread_index); - event_array[nr_poll].fd = FD(evsel, i, thread_index); - event_array[nr_poll].events = POLLIN; - nr_poll++; + evlist->pollfd[evlist->nr_fds].fd = FD(evsel, i, thread_index); + evlist->pollfd[evlist->nr_fds].events = POLLIN; + evlist->nr_fds++; mm = xyarray__entry(mmap_array, i, thread_index); mm->prev = 0; @@ -1316,11 +1314,11 @@ static int __cmd_top(void) for (i = 0; i < cpus->nr; i++) { group_fd = -1; list_for_each_entry(counter, &evsel_list->entries, node) - start_counter(i, counter); + start_counter(i, evsel_list, counter); } /* Wait for a minimal set of events before starting the snapshot */ - poll(&event_array[0], nr_poll, 100); + poll(evsel_list->pollfd, evsel_list->nr_fds, 100); perf_session__mmap_read(session); @@ -1345,7 +1343,7 @@ static int __cmd_top(void) perf_session__mmap_read(session); if (hits == samples) - ret = poll(event_array, nr_poll, 100); + ret = poll(evsel_list->pollfd, evsel_list->nr_fds, 100); } return 0; @@ -1426,11 +1424,6 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) usage_with_options(top_usage, options); } - event_array = malloc((sizeof(struct pollfd) * - MAX_NR_CPUS * MAX_COUNTERS * threads->nr)); - if (!event_array) - return -ENOMEM; - /* CPU and PID are mutually exclusive */ if (target_tid > 0 && cpu_list) { printf("WARNING: PID switch overriding CPU\n"); @@ -1480,6 +1473,9 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) pos->attr.sample_period = default_interval; } + if (perf_evlist__alloc_pollfd(evsel_list, cpus->nr, threads->nr) < 0) + goto out_free_fd; + sym_evsel = list_entry(evsel_list->entries.next, struct perf_evsel, node); symbol_conf.priv_size = (sizeof(struct sym_entry) + diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 7b4faec23737..2abf949259d0 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -1,3 +1,4 @@ +#include #include "evlist.h" #include "evsel.h" #include "util.h" @@ -28,6 +29,7 @@ static void perf_evlist__purge(struct perf_evlist *evlist) void perf_evlist__delete(struct perf_evlist *evlist) { perf_evlist__purge(evlist); + free(evlist->pollfd); free(evlist); } @@ -51,3 +53,10 @@ int perf_evlist__add_default(struct perf_evlist *evlist) perf_evlist__add(evlist, evsel); return 0; } + +int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthreads) +{ + int nfds = ncpus * nthreads * evlist->nr_entries; + evlist->pollfd = malloc(sizeof(struct pollfd) * nfds); + return evlist->pollfd != NULL ? 0 : -ENOMEM; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 48db91a8abf3..a7d7e122e3c6 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -3,9 +3,13 @@ #include +struct pollfd; + struct perf_evlist { struct list_head entries; int nr_entries; + int nr_fds; + struct pollfd *pollfd; }; struct perf_evsel; @@ -16,4 +20,6 @@ void perf_evlist__delete(struct perf_evlist *evlist); void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); int perf_evlist__add_default(struct perf_evlist *evlist); +int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthreads); + #endif /* __PERF_EVLIST_H */ -- cgit v1.2.3 From f08199d314458610d4ca52f8e86e0a4ec7a7bc54 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 11 Jan 2011 23:42:19 -0200 Subject: perf evsel: Support event groups The perf_evsel__open now have an extra boolean argument specifying if event grouping is desired. The first file descriptor created on a CPU becomes the group leader. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-stat.c | 4 ++-- tools/perf/builtin-test.c | 4 ++-- tools/perf/util/evsel.c | 27 +++++++++++++++++---------- tools/perf/util/evsel.h | 10 ++++++---- 4 files changed, 27 insertions(+), 18 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index da9090245934..b5fe522f11dc 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -169,7 +169,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel) PERF_FORMAT_TOTAL_TIME_RUNNING; if (system_wide) - return perf_evsel__open_per_cpu(evsel, cpus); + return perf_evsel__open_per_cpu(evsel, cpus, false); attr->inherit = !no_inherit; if (target_pid == -1 && target_tid == -1) { @@ -177,7 +177,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel) attr->enable_on_exec = 1; } - return perf_evsel__open_per_thread(evsel, threads); + return perf_evsel__open_per_thread(evsel, threads, false); } /* diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index 5dcdba653d70..4282d671b161 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -289,7 +289,7 @@ static int test__open_syscall_event(void) goto out_thread_map_delete; } - if (perf_evsel__open_per_thread(evsel, threads) < 0) { + if (perf_evsel__open_per_thread(evsel, threads, false) < 0) { pr_debug("failed to open counter: %s, " "tweak /proc/sys/kernel/perf_event_paranoid?\n", strerror(errno)); @@ -364,7 +364,7 @@ static int test__open_syscall_event_on_all_cpus(void) goto out_thread_map_delete; } - if (perf_evsel__open(evsel, cpus, threads) < 0) { + if (perf_evsel__open(evsel, cpus, threads, false) < 0) { pr_debug("failed to open counter: %s, " "tweak /proc/sys/kernel/perf_event_paranoid?\n", strerror(errno)); diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index f5cfed60af98..da473ec93c75 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -128,7 +128,7 @@ int __perf_evsel__read(struct perf_evsel *evsel, } static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, - struct thread_map *threads) + struct thread_map *threads, bool group) { int cpu, thread; @@ -137,12 +137,18 @@ static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, return -1; for (cpu = 0; cpu < cpus->nr; cpu++) { + int group_fd = -1; + 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); + cpus->map[cpu], + group_fd, 0); if (FD(evsel, cpu, thread) < 0) goto out_close; + + if (group && group_fd == -1) + group_fd = FD(evsel, cpu, thread); } } @@ -175,10 +181,9 @@ static struct { .threads = { -1, }, }; -int perf_evsel__open(struct perf_evsel *evsel, - struct cpu_map *cpus, struct thread_map *threads) +int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, bool group) { - if (cpus == NULL) { /* Work around old compiler warnings about strict aliasing */ cpus = &empty_cpu_map.map; @@ -187,15 +192,17 @@ int perf_evsel__open(struct perf_evsel *evsel, if (threads == NULL) threads = &empty_thread_map.map; - return __perf_evsel__open(evsel, cpus, threads); + return __perf_evsel__open(evsel, cpus, threads, group); } -int perf_evsel__open_per_cpu(struct perf_evsel *evsel, struct cpu_map *cpus) +int perf_evsel__open_per_cpu(struct perf_evsel *evsel, + struct cpu_map *cpus, bool group) { - return __perf_evsel__open(evsel, cpus, &empty_thread_map.map); + return __perf_evsel__open(evsel, cpus, &empty_thread_map.map, group); } -int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *threads) +int perf_evsel__open_per_thread(struct perf_evsel *evsel, + struct thread_map *threads, bool group) { - return __perf_evsel__open(evsel, &empty_cpu_map.map, threads); + return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group); } diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index b2d755fe88a5..0962b500cb6d 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -45,10 +45,12 @@ int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus); void perf_evsel__free_fd(struct perf_evsel *evsel); void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads); -int perf_evsel__open_per_cpu(struct perf_evsel *evsel, struct cpu_map *cpus); -int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *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, bool group); +int perf_evsel__open_per_thread(struct perf_evsel *evsel, + struct thread_map *threads, bool group); +int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, bool group); #define perf_evsel__match(evsel, t, c) \ (evsel->attr.type == PERF_TYPE_##t && \ -- cgit v1.2.3 From 9d04f1781772e11bd58806391555fc23ebb54377 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 12 Jan 2011 00:08:18 -0200 Subject: perf evsel: Allow specifying if the inherit bit should be set As this is a per-cpu attribute, we can't set it up in advance and use it for all the calls. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-stat.c | 4 ++-- tools/perf/builtin-test.c | 4 ++-- tools/perf/util/evsel.c | 16 +++++++++------- tools/perf/util/evsel.h | 6 +++--- 4 files changed, 16 insertions(+), 14 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index b5fe522f11dc..e2a2d02c5c43 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -169,7 +169,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel) PERF_FORMAT_TOTAL_TIME_RUNNING; if (system_wide) - return perf_evsel__open_per_cpu(evsel, cpus, false); + return perf_evsel__open_per_cpu(evsel, cpus, false, false); attr->inherit = !no_inherit; if (target_pid == -1 && target_tid == -1) { @@ -177,7 +177,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel) attr->enable_on_exec = 1; } - return perf_evsel__open_per_thread(evsel, threads, false); + return perf_evsel__open_per_thread(evsel, threads, false, false); } /* diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index 4282d671b161..7287158c4830 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -289,7 +289,7 @@ static int test__open_syscall_event(void) goto out_thread_map_delete; } - if (perf_evsel__open_per_thread(evsel, threads, false) < 0) { + if (perf_evsel__open_per_thread(evsel, threads, false, false) < 0) { pr_debug("failed to open counter: %s, " "tweak /proc/sys/kernel/perf_event_paranoid?\n", strerror(errno)); @@ -364,7 +364,7 @@ static int test__open_syscall_event_on_all_cpus(void) goto out_thread_map_delete; } - if (perf_evsel__open(evsel, cpus, threads, false) < 0) { + if (perf_evsel__open(evsel, cpus, threads, false, false) < 0) { pr_debug("failed to open counter: %s, " "tweak /proc/sys/kernel/perf_event_paranoid?\n", strerror(errno)); diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index da473ec93c75..82a00536892a 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -128,7 +128,7 @@ int __perf_evsel__read(struct perf_evsel *evsel, } static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, - struct thread_map *threads, bool group) + struct thread_map *threads, bool group, bool inherit) { int cpu, thread; @@ -139,6 +139,8 @@ static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, for (cpu = 0; cpu < cpus->nr; cpu++) { int group_fd = -1; + evsel->attr.inherit = (cpus->map[cpu] < 0) && inherit; + for (thread = 0; thread < threads->nr; thread++) { FD(evsel, cpu, thread) = sys_perf_event_open(&evsel->attr, threads->map[thread], @@ -182,7 +184,7 @@ static struct { }; int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, - struct thread_map *threads, bool group) + struct thread_map *threads, bool group, bool inherit) { if (cpus == NULL) { /* Work around old compiler warnings about strict aliasing */ @@ -192,17 +194,17 @@ int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, if (threads == NULL) threads = &empty_thread_map.map; - return __perf_evsel__open(evsel, cpus, threads, group); + return __perf_evsel__open(evsel, cpus, threads, group, inherit); } int perf_evsel__open_per_cpu(struct perf_evsel *evsel, - struct cpu_map *cpus, bool group) + struct cpu_map *cpus, bool group, bool inherit) { - return __perf_evsel__open(evsel, cpus, &empty_thread_map.map, group); + return __perf_evsel__open(evsel, cpus, &empty_thread_map.map, group, inherit); } int perf_evsel__open_per_thread(struct perf_evsel *evsel, - struct thread_map *threads, bool group) + struct thread_map *threads, bool group, bool inherit) { - return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group); + return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group, inherit); } diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 0962b500cb6d..1594696bd127 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -46,11 +46,11 @@ void perf_evsel__free_fd(struct perf_evsel *evsel); void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads); int perf_evsel__open_per_cpu(struct perf_evsel *evsel, - struct cpu_map *cpus, bool group); + struct cpu_map *cpus, bool group, bool inherit); int perf_evsel__open_per_thread(struct perf_evsel *evsel, - struct thread_map *threads, bool group); + struct thread_map *threads, bool group, bool inherit); int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, - struct thread_map *threads, bool group); + struct thread_map *threads, bool group, bool inherit); #define perf_evsel__match(evsel, t, c) \ (evsel->attr.type == PERF_TYPE_##t && \ -- cgit v1.2.3 From 70082dd92c4b288bd723a77897e2b555f0e63113 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 12 Jan 2011 17:03:24 -0200 Subject: perf evsel: Introduce mmap support Out of the code in 'perf top'. Record is next in line. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-top.c | 71 ++++-------------------------------------------- tools/perf/perf.h | 14 ++++++++++ tools/perf/util/evlist.c | 8 ++++++ tools/perf/util/evlist.h | 1 + tools/perf/util/evsel.c | 71 ++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/evsel.h | 8 ++++++ 6 files changed, 107 insertions(+), 66 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 15d89bede2fb..7d723ad0bfa9 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -1095,43 +1095,12 @@ static void event__process_sample(const event_t *self, } } -struct mmap_data { - void *base; - int mask; - unsigned int prev; -}; - -static int perf_evsel__alloc_mmap_per_thread(struct perf_evsel *evsel, - int ncpus, int nthreads) -{ - evsel->priv = xyarray__new(ncpus, nthreads, sizeof(struct mmap_data)); - return evsel->priv != NULL ? 0 : -ENOMEM; -} - -static void perf_evsel__free_mmap(struct perf_evsel *evsel) -{ - xyarray__delete(evsel->priv); - evsel->priv = NULL; -} - -static unsigned int mmap_read_head(struct mmap_data *md) -{ - struct perf_event_mmap_page *pc = md->base; - int head; - - head = pc->data_head; - rmb(); - - return head; -} - static void perf_session__mmap_read_counter(struct perf_session *self, struct perf_evsel *evsel, int cpu, int thread_idx) { - struct xyarray *mmap_array = evsel->priv; - struct mmap_data *md = xyarray__entry(mmap_array, cpu, thread_idx); - unsigned int head = mmap_read_head(md); + struct perf_mmap *md = xyarray__entry(evsel->mmap, cpu, thread_idx); + unsigned int head = perf_mmap__read_head(md); unsigned int old = md->prev; unsigned char *data = md->base + page_size; struct sample_data sample; @@ -1210,35 +1179,9 @@ static void perf_session__mmap_read(struct perf_session *self) } } -static void start_counter(int i, struct perf_evlist *evlist, - struct perf_evsel *evsel) -{ - struct xyarray *mmap_array = evsel->priv; - struct mmap_data *mm; - int thread_index; - - for (thread_index = 0; thread_index < threads->nr; thread_index++) { - assert(FD(evsel, i, thread_index) >= 0); - fcntl(FD(evsel, i, thread_index), F_SETFL, O_NONBLOCK); - - evlist->pollfd[evlist->nr_fds].fd = FD(evsel, i, thread_index); - evlist->pollfd[evlist->nr_fds].events = POLLIN; - evlist->nr_fds++; - - mm = xyarray__entry(mmap_array, i, thread_index); - mm->prev = 0; - mm->mask = mmap_pages*page_size - 1; - mm->base = mmap(NULL, (mmap_pages+1)*page_size, - PROT_READ, MAP_SHARED, FD(evsel, i, thread_index), 0); - if (mm->base == MAP_FAILED) - die("failed to mmap with %d (%s)\n", errno, strerror(errno)); - } -} - static void start_counters(struct perf_evlist *evlist) { struct perf_evsel *counter; - int i; list_for_each_entry(counter, &evlist->entries, node) { struct perf_event_attr *attr = &counter->attr; @@ -1282,11 +1225,9 @@ try_again: die("No CONFIG_PERF_EVENTS=y kernel support configured?\n"); exit(-1); } - } - for (i = 0; i < cpus->nr; i++) { - list_for_each_entry(counter, &evlist->entries, node) - start_counter(i, evsel_list, counter); + if (perf_evsel__mmap(counter, cpus, threads, mmap_pages, evlist) < 0) + die("failed to mmap with %d (%s)\n", errno, strerror(errno)); } } @@ -1453,7 +1394,7 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) usage_with_options(top_usage, options); list_for_each_entry(pos, &evsel_list->entries, node) { - if (perf_evsel__alloc_mmap_per_thread(pos, cpus->nr, threads->nr) < 0 || + if (perf_evsel__alloc_mmap(pos, cpus->nr, threads->nr) < 0 || perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) goto out_free_fd; /* @@ -1485,8 +1426,6 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) status = __cmd_top(); out_free_fd: - list_for_each_entry(pos, &evsel_list->entries, node) - perf_evsel__free_mmap(pos); perf_evlist__delete(evsel_list); return status; diff --git a/tools/perf/perf.h b/tools/perf/perf.h index 95aaf565c704..5fb5e1f11d1c 100644 --- a/tools/perf/perf.h +++ b/tools/perf/perf.h @@ -94,6 +94,20 @@ void get_term_dimensions(struct winsize *ws); #include "util/types.h" #include +struct perf_mmap { + void *base; + int mask; + unsigned int prev; +}; + +static inline unsigned int perf_mmap__read_head(struct perf_mmap *mm) +{ + struct perf_event_mmap_page *pc = mm->base; + int head = pc->data_head; + rmb(); + return head; +} + /* * prctl(PR_TASK_PERF_EVENTS_DISABLE) will (cheaply) disable all * counters in the current task. diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 2abf949259d0..6d4129214ee8 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -60,3 +60,11 @@ int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthread evlist->pollfd = malloc(sizeof(struct pollfd) * nfds); return evlist->pollfd != NULL ? 0 : -ENOMEM; } + +void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd) +{ + fcntl(fd, F_SETFL, O_NONBLOCK); + evlist->pollfd[evlist->nr_fds].fd = fd; + evlist->pollfd[evlist->nr_fds].events = POLLIN; + evlist->nr_fds++; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index a7d7e122e3c6..16bbfcba8ca8 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -21,5 +21,6 @@ void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); int perf_evlist__add_default(struct perf_evlist *evlist); int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthreads); +void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); #endif /* __PERF_EVLIST_H */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 82a00536892a..f5006958f8da 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -1,9 +1,13 @@ #include "evsel.h" +#include "evlist.h" #include "../perf.h" #include "util.h" #include "cpumap.h" #include "thread.h" +#include +#include + #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx) @@ -49,10 +53,32 @@ void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads) } } +void perf_evsel__munmap(struct perf_evsel *evsel, int ncpus, int nthreads) +{ + struct perf_mmap *mm; + int cpu, thread; + + for (cpu = 0; cpu < ncpus; cpu++) + for (thread = 0; thread < nthreads; ++thread) { + mm = xyarray__entry(evsel->mmap, cpu, thread); + if (mm->base != NULL) { + munmap(mm->base, evsel->mmap_len); + mm->base = NULL; + } + } +} + +int perf_evsel__alloc_mmap(struct perf_evsel *evsel, int ncpus, int nthreads) +{ + evsel->mmap = xyarray__new(ncpus, nthreads, sizeof(struct perf_mmap)); + return evsel->mmap != NULL ? 0 : -ENOMEM; +} + void perf_evsel__delete(struct perf_evsel *evsel) { assert(list_empty(&evsel->node)); xyarray__delete(evsel->fd); + xyarray__delete(evsel->mmap); free(evsel); } @@ -208,3 +234,48 @@ int perf_evsel__open_per_thread(struct perf_evsel *evsel, { return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group, inherit); } + +int perf_evsel__mmap(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, int pages, + struct perf_evlist *evlist) +{ + unsigned int page_size = sysconf(_SC_PAGE_SIZE); + int mask = pages * page_size - 1, cpu; + struct perf_mmap *mm; + int thread; + + if (evsel->mmap == NULL && + perf_evsel__alloc_mmap(evsel, cpus->nr, threads->nr) < 0) + return -ENOMEM; + + evsel->mmap_len = (pages + 1) * page_size; + + for (cpu = 0; cpu < cpus->nr; cpu++) { + for (thread = 0; thread < threads->nr; thread++) { + mm = xyarray__entry(evsel->mmap, cpu, thread); + mm->prev = 0; + mm->mask = mask; + mm->base = mmap(NULL, evsel->mmap_len, PROT_READ, + MAP_SHARED, FD(evsel, cpu, thread), 0); + if (mm->base == MAP_FAILED) + goto out_unmap; + + if (evlist != NULL) + perf_evlist__add_pollfd(evlist, FD(evsel, cpu, thread)); + } + } + + return 0; + +out_unmap: + do { + while (--thread >= 0) { + mm = xyarray__entry(evsel->mmap, cpu, thread); + munmap(mm->base, evsel->mmap_len); + mm->base = NULL; + } + thread = threads->nr; + } while (--cpu >= 0); + + return -1; +} diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 1594696bd127..c8fbef299436 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -29,19 +29,23 @@ struct perf_evsel { struct perf_event_attr attr; char *filter; struct xyarray *fd; + struct xyarray *mmap; struct perf_counts *counts; + size_t mmap_len; int idx; void *priv; }; struct cpu_map; struct thread_map; +struct perf_evlist; 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); int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus); +int perf_evsel__alloc_mmap(struct perf_evsel *evsel, int ncpus, int nthreads); void perf_evsel__free_fd(struct perf_evsel *evsel); void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads); @@ -51,6 +55,10 @@ int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *threads, bool group, bool inherit); int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, struct thread_map *threads, bool group, bool inherit); +int perf_evsel__mmap(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, int pages, + struct perf_evlist *evlist); +void perf_evsel__munmap(struct perf_evsel *evsel, int ncpus, int nthreads); #define perf_evsel__match(evsel, t, c) \ (evsel->attr.type == PERF_TYPE_##t && \ -- cgit v1.2.3 From 70db7533caef02350ec8d6852e589491bca3a951 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 12 Jan 2011 22:39:13 -0200 Subject: perf evlist: Move the mmap array from perf_evsel Adopting the new model used in 'perf record', where we don't have a map per thread per cpu, instead we have an mmap per cpu, established on the first fd for that cpu and ask the kernel using the PERF_EVENT_IOC_SET_OUTPUT ioctl to send events for the other fds on that cpu for the one with the mmap. The methods moved from perf_evsel to perf_evlist, but for easing review they were modified in place, in evsel.c, the next patch will move the migrated methods to evlist.c. With this 'perf top' now uses the same mmap model used by 'perf record' and the next patches will make 'perf record' use these new routines, establishing a common codebase for both tools. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-top.c | 56 ++++++++--------- tools/perf/util/evlist.c | 27 ++++++++ tools/perf/util/evlist.h | 9 +++ tools/perf/util/evsel.c | 160 ++++++++++++++++++++++++++++++++++------------- tools/perf/util/evsel.h | 26 +++++--- 5 files changed, 201 insertions(+), 77 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 7d723ad0bfa9..df85c1f9417b 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -78,7 +78,7 @@ static struct cpu_map *cpus; static int realtime_prio = 0; static bool group = false; static unsigned int page_size; -static unsigned int mmap_pages = 16; +static unsigned int mmap_pages = 128; static int freq = 1000; /* 1 KHz */ static int delay_secs = 2; @@ -991,8 +991,7 @@ static int symbol_filter(struct map *map, struct symbol *sym) static void event__process_sample(const event_t *self, struct sample_data *sample, - struct perf_session *session, - struct perf_evsel *evsel) + struct perf_session *session) { u64 ip = self->ip.ip; struct sym_entry *syme; @@ -1085,8 +1084,12 @@ static void event__process_sample(const event_t *self, syme = symbol__priv(al.sym); if (!syme->skip) { - syme->count[evsel->idx]++; + struct perf_evsel *evsel; + syme->origin = origin; + evsel = perf_evlist__id2evsel(evsel_list, sample->id); + assert(evsel != NULL); + syme->count[evsel->idx]++; record_precise_ip(syme, evsel->idx, ip); pthread_mutex_lock(&active_symbols_lock); if (list_empty(&syme->node) || !syme->node.next) @@ -1095,11 +1098,9 @@ static void event__process_sample(const event_t *self, } } -static void perf_session__mmap_read_counter(struct perf_session *self, - struct perf_evsel *evsel, - int cpu, int thread_idx) +static void perf_session__mmap_read_cpu(struct perf_session *self, int cpu) { - struct perf_mmap *md = xyarray__entry(evsel->mmap, cpu, thread_idx); + struct perf_mmap *md = &evsel_list->mmap[cpu]; unsigned int head = perf_mmap__read_head(md); unsigned int old = md->prev; unsigned char *data = md->base + page_size; @@ -1153,7 +1154,7 @@ static void perf_session__mmap_read_counter(struct perf_session *self, event__parse_sample(event, self, &sample); if (event->header.type == PERF_RECORD_SAMPLE) - event__process_sample(event, &sample, self, evsel); + event__process_sample(event, &sample, self); else event__process(event, &sample, self); old += size; @@ -1164,19 +1165,10 @@ static void perf_session__mmap_read_counter(struct perf_session *self, static void perf_session__mmap_read(struct perf_session *self) { - struct perf_evsel *counter; - int i, thread_index; - - for (i = 0; i < cpus->nr; i++) { - list_for_each_entry(counter, &evsel_list->entries, node) { - for (thread_index = 0; - thread_index < threads->nr; - thread_index++) { - perf_session__mmap_read_counter(self, - counter, i, thread_index); - } - } - } + int i; + + for (i = 0; i < cpus->nr; i++) + perf_session__mmap_read_cpu(self, i); } static void start_counters(struct perf_evlist *evlist) @@ -1194,6 +1186,11 @@ static void start_counters(struct perf_evlist *evlist) attr->sample_freq = freq; } + if (evlist->nr_entries > 1) { + attr->sample_type |= PERF_SAMPLE_ID; + attr->read_format |= PERF_FORMAT_ID; + } + attr->mmap = 1; try_again: if (perf_evsel__open(counter, cpus, threads, group, inherit) < 0) { @@ -1225,15 +1222,16 @@ try_again: die("No CONFIG_PERF_EVENTS=y kernel support configured?\n"); exit(-1); } - - if (perf_evsel__mmap(counter, cpus, threads, mmap_pages, evlist) < 0) - die("failed to mmap with %d (%s)\n", errno, strerror(errno)); } + + if (perf_evlist__mmap(evlist, cpus, threads, mmap_pages, true) < 0) + die("failed to mmap with %d (%s)\n", errno, strerror(errno)); } static int __cmd_top(void) { pthread_t thread; + struct perf_evsel *first; int ret; /* * FIXME: perf_session__new should allow passing a O_MMAP, so that all this @@ -1249,6 +1247,8 @@ static int __cmd_top(void) event__synthesize_threads(event__process, session); start_counters(evsel_list); + first = list_entry(evsel_list->entries.next, struct perf_evsel, node); + perf_session__set_sample_type(session, first->attr.sample_type); /* Wait for a minimal set of events before starting the snapshot */ poll(evsel_list->pollfd, evsel_list->nr_fds, 100); @@ -1394,8 +1394,7 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) usage_with_options(top_usage, options); list_for_each_entry(pos, &evsel_list->entries, node) { - if (perf_evsel__alloc_mmap(pos, cpus->nr, threads->nr) < 0 || - perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) + if (perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) goto out_free_fd; /* * Fill in the ones not specifically initialized via -c: @@ -1406,7 +1405,8 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) pos->attr.sample_period = default_interval; } - if (perf_evlist__alloc_pollfd(evsel_list, cpus->nr, threads->nr) < 0) + if (perf_evlist__alloc_pollfd(evsel_list, cpus->nr, threads->nr) < 0 || + perf_evlist__alloc_mmap(evsel_list, cpus->nr) < 0) goto out_free_fd; sym_evsel = list_entry(evsel_list->entries.next, struct perf_evsel, node); diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 6d4129214ee8..deb82a4fc312 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -3,11 +3,18 @@ #include "evsel.h" #include "util.h" +#include +#include + struct perf_evlist *perf_evlist__new(void) { struct perf_evlist *evlist = zalloc(sizeof(*evlist)); if (evlist != NULL) { + int i; + + for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i) + INIT_HLIST_HEAD(&evlist->heads[i]); INIT_LIST_HEAD(&evlist->entries); } @@ -29,6 +36,7 @@ static void perf_evlist__purge(struct perf_evlist *evlist) void perf_evlist__delete(struct perf_evlist *evlist) { perf_evlist__purge(evlist); + free(evlist->mmap); free(evlist->pollfd); free(evlist); } @@ -68,3 +76,22 @@ void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd) evlist->pollfd[evlist->nr_fds].events = POLLIN; evlist->nr_fds++; } + +struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id) +{ + struct hlist_head *head; + struct hlist_node *pos; + struct perf_sample_id *sid; + int hash; + + if (evlist->nr_entries == 1) + return list_entry(evlist->entries.next, struct perf_evsel, node); + + hash = hash_64(id, PERF_EVLIST__HLIST_BITS); + head = &evlist->heads[hash]; + + hlist_for_each_entry(sid, pos, head, node) + if (sid->id == id) + return sid->evsel; + return NULL; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 16bbfcba8ca8..dbfcc79bb995 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -2,13 +2,20 @@ #define __PERF_EVLIST_H 1 #include +#include "../perf.h" struct pollfd; +#define PERF_EVLIST__HLIST_BITS 8 +#define PERF_EVLIST__HLIST_SIZE (1 << PERF_EVLIST__HLIST_BITS) + struct perf_evlist { struct list_head entries; + struct hlist_head heads[PERF_EVLIST__HLIST_SIZE]; int nr_entries; int nr_fds; + int mmap_len; + struct perf_mmap *mmap; struct pollfd *pollfd; }; @@ -23,4 +30,6 @@ int perf_evlist__add_default(struct perf_evlist *evlist); int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthreads); void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); +struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id); + #endif /* __PERF_EVLIST_H */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index f5006958f8da..ee490356c817 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -8,7 +8,11 @@ #include #include +#include +#include + #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) +#define SID(e, x, y) xyarray__entry(e->id, x, y) struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx) { @@ -29,6 +33,12 @@ int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads) return evsel->fd != NULL ? 0 : -ENOMEM; } +int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads) +{ + evsel->id = xyarray__new(ncpus, nthreads, sizeof(struct perf_sample_id)); + return evsel->id != NULL ? 0 : -ENOMEM; +} + int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus) { evsel->counts = zalloc((sizeof(*evsel->counts) + @@ -42,6 +52,12 @@ void perf_evsel__free_fd(struct perf_evsel *evsel) evsel->fd = NULL; } +void perf_evsel__free_id(struct perf_evsel *evsel) +{ + xyarray__delete(evsel->id); + evsel->id = NULL; +} + void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads) { int cpu, thread; @@ -53,32 +69,29 @@ void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads) } } -void perf_evsel__munmap(struct perf_evsel *evsel, int ncpus, int nthreads) +void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus) { - struct perf_mmap *mm; - int cpu, thread; + int cpu; - for (cpu = 0; cpu < ncpus; cpu++) - for (thread = 0; thread < nthreads; ++thread) { - mm = xyarray__entry(evsel->mmap, cpu, thread); - if (mm->base != NULL) { - munmap(mm->base, evsel->mmap_len); - mm->base = NULL; - } + for (cpu = 0; cpu < ncpus; cpu++) { + if (evlist->mmap[cpu].base != NULL) { + munmap(evlist->mmap[cpu].base, evlist->mmap_len); + evlist->mmap[cpu].base = NULL; } + } } -int perf_evsel__alloc_mmap(struct perf_evsel *evsel, int ncpus, int nthreads) +int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus) { - evsel->mmap = xyarray__new(ncpus, nthreads, sizeof(struct perf_mmap)); - return evsel->mmap != NULL ? 0 : -ENOMEM; + evlist->mmap = zalloc(ncpus * sizeof(struct perf_mmap)); + return evlist->mmap != NULL ? 0 : -ENOMEM; } void perf_evsel__delete(struct perf_evsel *evsel) { assert(list_empty(&evsel->node)); xyarray__delete(evsel->fd); - xyarray__delete(evsel->mmap); + xyarray__delete(evsel->id); free(evsel); } @@ -235,47 +248,110 @@ int perf_evsel__open_per_thread(struct perf_evsel *evsel, return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group, inherit); } -int perf_evsel__mmap(struct perf_evsel *evsel, struct cpu_map *cpus, - struct thread_map *threads, int pages, - struct perf_evlist *evlist) +static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot, + int mask, int fd) +{ + evlist->mmap[cpu].prev = 0; + evlist->mmap[cpu].mask = mask; + evlist->mmap[cpu].base = mmap(NULL, evlist->mmap_len, prot, + MAP_SHARED, fd, 0); + if (evlist->mmap[cpu].base == MAP_FAILED) + return -1; + + perf_evlist__add_pollfd(evlist, fd); + return 0; +} + +static int perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, int fd) +{ + struct perf_sample_id *sid; + u64 read_data[4] = { 0, }; + int hash, id_idx = 1; /* The first entry is the counter value */ + + if (!(evsel->attr.read_format & PERF_FORMAT_ID) || + read(fd, &read_data, sizeof(read_data)) == -1) + return -1; + + if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) + ++id_idx; + if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) + ++id_idx; + + sid = SID(evsel, cpu, thread); + sid->id = read_data[id_idx]; + sid->evsel = evsel; + hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS); + hlist_add_head(&sid->node, &evlist->heads[hash]); + return 0; +} + +/** perf_evlist__mmap - Create per cpu maps to receive events + * + * @evlist - list of events + * @cpus - cpu map being monitored + * @threads - threads map being monitored + * @pages - map length in pages + * @overwrite - overwrite older events? + * + * If overwrite is false the user needs to signal event consuption using: + * + * struct perf_mmap *m = &evlist->mmap[cpu]; + * unsigned int head = perf_mmap__read_head(m); + * + * perf_mmap__write_tail(m, head) + */ +int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads, int pages, bool overwrite) { unsigned int page_size = sysconf(_SC_PAGE_SIZE); int mask = pages * page_size - 1, cpu; - struct perf_mmap *mm; - int thread; + struct perf_evsel *first_evsel, *evsel; + int thread, prot = PROT_READ | (overwrite ? 0 : PROT_WRITE); - if (evsel->mmap == NULL && - perf_evsel__alloc_mmap(evsel, cpus->nr, threads->nr) < 0) + if (evlist->mmap == NULL && + perf_evlist__alloc_mmap(evlist, cpus->nr) < 0) return -ENOMEM; - evsel->mmap_len = (pages + 1) * page_size; + if (evlist->pollfd == NULL && + perf_evlist__alloc_pollfd(evlist, cpus->nr, threads->nr) < 0) + return -ENOMEM; - for (cpu = 0; cpu < cpus->nr; cpu++) { - for (thread = 0; thread < threads->nr; thread++) { - mm = xyarray__entry(evsel->mmap, cpu, thread); - mm->prev = 0; - mm->mask = mask; - mm->base = mmap(NULL, evsel->mmap_len, PROT_READ, - MAP_SHARED, FD(evsel, cpu, thread), 0); - if (mm->base == MAP_FAILED) - goto out_unmap; - - if (evlist != NULL) - perf_evlist__add_pollfd(evlist, FD(evsel, cpu, thread)); + evlist->mmap_len = (pages + 1) * page_size; + first_evsel = list_entry(evlist->entries.next, struct perf_evsel, node); + + list_for_each_entry(evsel, &evlist->entries, node) { + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + evsel->id == NULL && + perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0) + return -ENOMEM; + + for (cpu = 0; cpu < cpus->nr; cpu++) { + for (thread = 0; thread < threads->nr; thread++) { + int fd = FD(evsel, cpu, thread); + + if (evsel->idx || thread) { + if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, + FD(first_evsel, cpu, 0)) != 0) + goto out_unmap; + } else if (__perf_evlist__mmap(evlist, cpu, prot, mask, fd) < 0) + goto out_unmap; + + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + perf_evlist__id_hash(evlist, evsel, cpu, thread, fd) < 0) + goto out_unmap; + } } } return 0; out_unmap: - do { - while (--thread >= 0) { - mm = xyarray__entry(evsel->mmap, cpu, thread); - munmap(mm->base, evsel->mmap_len); - mm->base = NULL; + for (cpu = 0; cpu < cpus->nr; cpu++) { + if (evlist->mmap[cpu].base != NULL) { + munmap(evlist->mmap[cpu].base, evlist->mmap_len); + evlist->mmap[cpu].base = NULL; } - thread = threads->nr; - } while (--cpu >= 0); - + } return -1; } diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index c8fbef299436..667ee4e2e35e 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -24,14 +24,25 @@ struct perf_counts { struct perf_counts_values cpu[]; }; +struct perf_evsel; + +/* + * Per fd, to map back from PERF_SAMPLE_ID to evsel, only used when there are + * more than one entry in the evlist. + */ +struct perf_sample_id { + struct hlist_node node; + u64 id; + struct perf_evsel *evsel; +}; + struct perf_evsel { struct list_head node; struct perf_event_attr attr; char *filter; struct xyarray *fd; - struct xyarray *mmap; + struct xyarray *id; struct perf_counts *counts; - size_t mmap_len; int idx; void *priv; }; @@ -44,9 +55,11 @@ 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); +int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads); int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus); -int perf_evsel__alloc_mmap(struct perf_evsel *evsel, int ncpus, int nthreads); +int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus); void perf_evsel__free_fd(struct perf_evsel *evsel); +void perf_evsel__free_id(struct perf_evsel *evsel); void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads); int perf_evsel__open_per_cpu(struct perf_evsel *evsel, @@ -55,10 +68,9 @@ int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *threads, bool group, bool inherit); int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, struct thread_map *threads, bool group, bool inherit); -int perf_evsel__mmap(struct perf_evsel *evsel, struct cpu_map *cpus, - struct thread_map *threads, int pages, - struct perf_evlist *evlist); -void perf_evsel__munmap(struct perf_evsel *evsel, int ncpus, int nthreads); +int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads, int pages, bool overwrite); +void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus); #define perf_evsel__match(evsel, t, c) \ (evsel->attr.type == PERF_TYPE_##t && \ -- cgit v1.2.3 From 915fce20ecf8f7ff4189d0fff42b62aebf6a57cc Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 14 Jan 2011 16:19:12 -0200 Subject: perf tools: Add missing cpu_map__delete() Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/cpumap.c | 5 +++++ tools/perf/util/cpumap.h | 2 +- 2 files changed, 6 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c index 3ccaa1043383..6893eec693ab 100644 --- a/tools/perf/util/cpumap.c +++ b/tools/perf/util/cpumap.c @@ -177,3 +177,8 @@ struct cpu_map *cpu_map__dummy_new(void) return cpus; } + +void cpu_map__delete(struct cpu_map *map) +{ + free(map); +} diff --git a/tools/perf/util/cpumap.h b/tools/perf/util/cpumap.h index f7a4f42f6307..072c0a374794 100644 --- a/tools/perf/util/cpumap.h +++ b/tools/perf/util/cpumap.h @@ -8,6 +8,6 @@ struct cpu_map { struct cpu_map *cpu_map__new(const char *cpu_list); struct cpu_map *cpu_map__dummy_new(void); -void *cpu_map__delete(struct cpu_map *map); +void cpu_map__delete(struct cpu_map *map); #endif /* __PERF_CPUMAP_H */ -- cgit v1.2.3 From 04391debc3e1195222a4dbb162ace6542dd89c1c Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 15 Jan 2011 10:40:59 -0200 Subject: perf evlist: Steal mmap reading routine from 'perf top' Will be used in the upcoming 'perf test' entry for the evlist mmap routines. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-top.c | 56 +++---------------------------------------- tools/perf/util/evlist.c | 62 ++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/evlist.h | 4 ++++ 3 files changed, 69 insertions(+), 53 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index df85c1f9417b..58352ad807c7 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -1100,67 +1100,17 @@ static void event__process_sample(const event_t *self, static void perf_session__mmap_read_cpu(struct perf_session *self, int cpu) { - struct perf_mmap *md = &evsel_list->mmap[cpu]; - unsigned int head = perf_mmap__read_head(md); - unsigned int old = md->prev; - unsigned char *data = md->base + page_size; struct sample_data sample; - int diff; - - /* - * If we're further behind than half the buffer, there's a chance - * the writer will bite our tail and mess up the samples under us. - * - * If we somehow ended up ahead of the head, we got messed up. - * - * In either case, truncate and restart at head. - */ - diff = head - old; - if (diff > md->mask / 2 || diff < 0) { - fprintf(stderr, "WARNING: failed to keep up with mmap data.\n"); - - /* - * head points to a known good entry, start there. - */ - old = head; - } - - for (; old != head;) { - event_t *event = (event_t *)&data[old & md->mask]; - - event_t event_copy; - - size_t size = event->header.size; - - /* - * Event straddles the mmap boundary -- header should always - * be inside due to u64 alignment of output. - */ - if ((old & md->mask) + size != ((old + size) & md->mask)) { - unsigned int offset = old; - unsigned int len = min(sizeof(*event), size), cpy; - void *dst = &event_copy; - - do { - cpy = min(md->mask + 1 - (offset & md->mask), len); - memcpy(dst, &data[offset & md->mask], cpy); - offset += cpy; - dst += cpy; - len -= cpy; - } while (len); - - event = &event_copy; - } + event_t *event; + while ((event = perf_evlist__read_on_cpu(evsel_list, cpu)) != NULL) { event__parse_sample(event, self, &sample); + if (event->header.type == PERF_RECORD_SAMPLE) event__process_sample(event, &sample, self); else event__process(event, &sample, self); - old += size; } - - md->prev = old; } static void perf_session__mmap_read(struct perf_session *self) diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index deb82a4fc312..4b3b84cd71a1 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -95,3 +95,65 @@ struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id) return sid->evsel; return NULL; } + +event_t *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) +{ + /* XXX Move this to perf.c, making it generally available */ + unsigned int page_size = sysconf(_SC_PAGE_SIZE); + struct perf_mmap *md = &evlist->mmap[cpu]; + unsigned int head = perf_mmap__read_head(md); + unsigned int old = md->prev; + unsigned char *data = md->base + page_size; + event_t *event = NULL; + int diff; + + /* + * If we're further behind than half the buffer, there's a chance + * the writer will bite our tail and mess up the samples under us. + * + * If we somehow ended up ahead of the head, we got messed up. + * + * In either case, truncate and restart at head. + */ + diff = head - old; + if (diff > md->mask / 2 || diff < 0) { + fprintf(stderr, "WARNING: failed to keep up with mmap data.\n"); + + /* + * head points to a known good entry, start there. + */ + old = head; + } + + if (old != head) { + size_t size; + + event = (event_t *)&data[old & md->mask]; + size = event->header.size; + + /* + * Event straddles the mmap boundary -- header should always + * be inside due to u64 alignment of output. + */ + if ((old & md->mask) + size != ((old + size) & md->mask)) { + unsigned int offset = old; + unsigned int len = min(sizeof(*event), size), cpy; + void *dst = &evlist->event_copy; + + do { + cpy = min(md->mask + 1 - (offset & md->mask), len); + memcpy(dst, &data[offset & md->mask], cpy); + offset += cpy; + dst += cpy; + len -= cpy; + } while (len); + + event = &evlist->event_copy; + } + + old += size; + } + + md->prev = old; + return event; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index dbfcc79bb995..28712063db97 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -3,6 +3,7 @@ #include #include "../perf.h" +#include "event.h" struct pollfd; @@ -15,6 +16,7 @@ struct perf_evlist { int nr_entries; int nr_fds; int mmap_len; + event_t event_copy; struct perf_mmap *mmap; struct pollfd *pollfd; }; @@ -32,4 +34,6 @@ void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id); +event_t *perf_evlist__read_on_cpu(struct perf_evlist *self, int cpu); + #endif /* __PERF_EVLIST_H */ -- cgit v1.2.3 From 1b3a0e9592ebf174af934b3908a2bf6a6fa86169 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 14 Jan 2011 04:51:58 +0100 Subject: perf callchain: Feed callchains into a cursor The callchains are fed with an array of a fixed size. As a result we iterate over each callchains three times: - 1st to resolve symbols - 2nd to filter out context boundaries - 3rd for the insertion into the tree This also involves some pairs of memory allocation/deallocation everytime we insert a callchain, for the filtered out array of addresses and for the array of symbols that comes along. Instead, feed the callchains through a linked list with persistent allocations. It brings several pros like: - Merge the 1st and 2nd iterations in one. That was possible before but in a way that would involve allocating an array slightly taller than necessary because we don't know in advance the number of context boundaries to filter out. - Much lesser allocations/deallocations. The linked list keeps persistent empty entries for the next usages and is extendable at will. - Makes it easier for multiple sources of callchains to feed a stacktrace together. This is deemed to pave the way for cfi based callchains wherein traditional frame pointer based kernel stacktraces will precede cfi based user ones, producing an overall callchain which size is hardly predictable. This requirement makes the static array obsolete and makes a linked list based iterator a much more flexible fit. Basic testing on a big perf file containing callchains (~ 176 MB) has shown a throughput gain of about 11% with perf report. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra LKML-Reference: <1294977121-5700-2-git-send-email-fweisbec@gmail.com> Signed-off-by: Frederic Weisbecker Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-report.c | 25 +++--- tools/perf/util/callchain.c | 204 ++++++++++++++++++++++---------------------- tools/perf/util/callchain.h | 66 +++++++++++++- tools/perf/util/hist.c | 13 ++- tools/perf/util/hist.h | 2 + tools/perf/util/session.c | 22 ++--- tools/perf/util/session.h | 11 +-- 7 files changed, 204 insertions(+), 139 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index c27e31f289e6..c95599a82f9e 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -81,18 +81,17 @@ static int perf_session__add_hist_entry(struct perf_session *self, struct addr_location *al, struct sample_data *data) { - struct map_symbol *syms = NULL; struct symbol *parent = NULL; - int err = -ENOMEM; + int err = 0; struct hist_entry *he; struct hists *hists; struct perf_event_attr *attr; if ((sort__has_parent || symbol_conf.use_callchain) && data->callchain) { - syms = perf_session__resolve_callchain(self, al->thread, - data->callchain, &parent); - if (syms == NULL) - return -ENOMEM; + err = perf_session__resolve_callchain(self, al->thread, + data->callchain, &parent); + if (err) + return err; } attr = perf_header__find_attr(data->id, &self->header); @@ -101,16 +100,17 @@ static int perf_session__add_hist_entry(struct perf_session *self, else hists = perf_session__hists_findnew(self, data->id, 0, 0); if (hists == NULL) - goto out_free_syms; + return -ENOMEM; + he = __hists__add_entry(hists, al, parent, data->period); if (he == NULL) - goto out_free_syms; - err = 0; + return -ENOMEM; + if (symbol_conf.use_callchain) { - err = callchain_append(he->callchain, data->callchain, syms, + err = callchain_append(he->callchain, &self->callchain_cursor, data->period); if (err) - goto out_free_syms; + return err; } /* * Only in the newt browser we are doing integrated annotation, @@ -119,8 +119,7 @@ static int perf_session__add_hist_entry(struct perf_session *self, */ if (use_browser > 0) err = hist_entry__inc_addr_samples(he, al->addr); -out_free_syms: - free(syms); + return err; } diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index e12d539417b2..53a49e0cfc6c 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -1,5 +1,5 @@ /* - * Copyright (C) 2009-2010, Frederic Weisbecker + * Copyright (C) 2009-2011, Frederic Weisbecker * * Handle the callchains from the stream in an ad-hoc radix tree and then * sort them in an rbtree. @@ -195,26 +195,21 @@ create_child(struct callchain_node *parent, bool inherit_children) } -struct resolved_ip { - u64 ip; - struct map_symbol ms; -}; - -struct resolved_chain { - u64 nr; - struct resolved_ip ips[0]; -}; - - /* * Fill the node with callchain values */ static void -fill_node(struct callchain_node *node, struct resolved_chain *chain, int start) +fill_node(struct callchain_node *node, struct callchain_cursor *cursor) { - unsigned int i; + struct callchain_cursor_node *cursor_node; + + node->val_nr = cursor->nr - cursor->pos; + if (!node->val_nr) + pr_warning("Warning: empty node in callchain tree\n"); - for (i = start; i < chain->nr; i++) { + cursor_node = callchain_cursor_current(cursor); + + while (cursor_node) { struct callchain_list *call; call = zalloc(sizeof(*call)); @@ -222,23 +217,25 @@ fill_node(struct callchain_node *node, struct resolved_chain *chain, int start) perror("not enough memory for the code path tree"); return; } - call->ip = chain->ips[i].ip; - call->ms = chain->ips[i].ms; + call->ip = cursor_node->ip; + call->ms.sym = cursor_node->sym; + call->ms.map = cursor_node->map; list_add_tail(&call->list, &node->val); + + callchain_cursor_advance(cursor); + cursor_node = callchain_cursor_current(cursor); } - node->val_nr = chain->nr - start; - if (!node->val_nr) - pr_warning("Warning: empty node in callchain tree\n"); } static void -add_child(struct callchain_node *parent, struct resolved_chain *chain, - int start, u64 period) +add_child(struct callchain_node *parent, + struct callchain_cursor *cursor, + u64 period) { struct callchain_node *new; new = create_child(parent, false); - fill_node(new, chain, start); + fill_node(new, cursor); new->children_hit = 0; new->hit = period; @@ -250,9 +247,10 @@ add_child(struct callchain_node *parent, struct resolved_chain *chain, * Then create another child to host the given callchain of new branch */ static void -split_add_child(struct callchain_node *parent, struct resolved_chain *chain, - struct callchain_list *to_split, int idx_parents, int idx_local, - u64 period) +split_add_child(struct callchain_node *parent, + struct callchain_cursor *cursor, + struct callchain_list *to_split, + u64 idx_parents, u64 idx_local, u64 period) { struct callchain_node *new; struct list_head *old_tail; @@ -277,9 +275,9 @@ split_add_child(struct callchain_node *parent, struct resolved_chain *chain, parent->val_nr = idx_local; /* create a new child for the new branch if any */ - if (idx_total < chain->nr) { + if (idx_total < cursor->nr) { parent->hit = 0; - add_child(parent, chain, idx_total, period); + add_child(parent, cursor, period); parent->children_hit += period; } else { parent->hit = period; @@ -287,36 +285,41 @@ split_add_child(struct callchain_node *parent, struct resolved_chain *chain, } static int -append_chain(struct callchain_node *root, struct resolved_chain *chain, - unsigned int start, u64 period); +append_chain(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period); static void -append_chain_children(struct callchain_node *root, struct resolved_chain *chain, - unsigned int start, u64 period) +append_chain_children(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period) { struct callchain_node *rnode; /* lookup in childrens */ chain_for_each_child(rnode, root) { - unsigned int ret = append_chain(rnode, chain, start, period); + unsigned int ret = append_chain(rnode, cursor, period); if (!ret) goto inc_children_hit; } /* nothing in children, add to the current node */ - add_child(root, chain, start, period); + add_child(root, cursor, period); inc_children_hit: root->children_hit += period; } static int -append_chain(struct callchain_node *root, struct resolved_chain *chain, - unsigned int start, u64 period) +append_chain(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period) { + struct callchain_cursor_node *curr_snap = cursor->curr; struct callchain_list *cnode; - unsigned int i = start; + u64 start = cursor->pos; bool found = false; + u64 matches; /* * Lookup in the current node @@ -324,114 +327,95 @@ append_chain(struct callchain_node *root, struct resolved_chain *chain, * anywhere inside a function. */ list_for_each_entry(cnode, &root->val, list) { + struct callchain_cursor_node *node; struct symbol *sym; - if (i == chain->nr) + node = callchain_cursor_current(cursor); + if (!node) break; - sym = chain->ips[i].ms.sym; + sym = node->sym; if (cnode->ms.sym && sym) { if (cnode->ms.sym->start != sym->start) break; - } else if (cnode->ip != chain->ips[i].ip) + } else if (cnode->ip != node->ip) break; if (!found) found = true; - i++; + + callchain_cursor_advance(cursor); } /* matches not, relay on the parent */ - if (!found) + if (!found) { + cursor->curr = curr_snap; + cursor->pos = start; return -1; + } + + matches = cursor->pos - start; /* we match only a part of the node. Split it and add the new chain */ - if (i - start < root->val_nr) { - split_add_child(root, chain, cnode, start, i - start, period); + if (matches < root->val_nr) { + split_add_child(root, cursor, cnode, start, matches, period); return 0; } /* we match 100% of the path, increment the hit */ - if (i - start == root->val_nr && i == chain->nr) { + if (matches == root->val_nr && cursor->pos == cursor->nr) { root->hit += period; return 0; } /* We match the node and still have a part remaining */ - append_chain_children(root, chain, i, period); + append_chain_children(root, cursor, period); return 0; } -static void filter_context(struct ip_callchain *old, struct resolved_chain *new, - struct map_symbol *syms) -{ - int i, j = 0; - - for (i = 0; i < (int)old->nr; i++) { - if (old->ips[i] >= PERF_CONTEXT_MAX) - continue; - - new->ips[j].ip = old->ips[i]; - new->ips[j].ms = syms[i]; - j++; - } - - new->nr = j; -} - - -int callchain_append(struct callchain_root *root, struct ip_callchain *chain, - struct map_symbol *syms, u64 period) +int callchain_append(struct callchain_root *root, + struct callchain_cursor *cursor, + u64 period) { - struct resolved_chain *filtered; - - if (!chain->nr) + if (!cursor->nr) return 0; - filtered = zalloc(sizeof(*filtered) + - chain->nr * sizeof(struct resolved_ip)); - if (!filtered) - return -ENOMEM; - - filter_context(chain, filtered, syms); - - if (!filtered->nr) - goto end; + callchain_cursor_commit(cursor); - append_chain_children(&root->node, filtered, 0, period); + append_chain_children(&root->node, cursor, period); - if (filtered->nr > root->max_depth) - root->max_depth = filtered->nr; -end: - free(filtered); + if (cursor->nr > root->max_depth) + root->max_depth = cursor->nr; return 0; } static int -merge_chain_branch(struct callchain_node *dst, struct callchain_node *src, - struct resolved_chain *chain) +merge_chain_branch(struct callchain_cursor *cursor, + struct callchain_node *dst, struct callchain_node *src) { + struct callchain_cursor_node **old_last = cursor->last; struct callchain_node *child, *next_child; struct callchain_list *list, *next_list; - int old_pos = chain->nr; + int old_pos = cursor->nr; int err = 0; list_for_each_entry_safe(list, next_list, &src->val, list) { - chain->ips[chain->nr].ip = list->ip; - chain->ips[chain->nr].ms = list->ms; - chain->nr++; + callchain_cursor_append(cursor, list->ip, + list->ms.map, list->ms.sym); list_del(&list->list); free(list); } - if (src->hit) - append_chain_children(dst, chain, 0, src->hit); + if (src->hit) { + callchain_cursor_commit(cursor); + append_chain_children(dst, cursor, src->hit); + } chain_for_each_child_safe(child, next_child, src) { - err = merge_chain_branch(dst, child, chain); + err = merge_chain_branch(cursor, dst, child); if (err) break; @@ -439,26 +423,38 @@ merge_chain_branch(struct callchain_node *dst, struct callchain_node *src, free(child); } - chain->nr = old_pos; + cursor->nr = old_pos; + cursor->last = old_last; return err; } -int callchain_merge(struct callchain_root *dst, struct callchain_root *src) +int callchain_merge(struct callchain_cursor *cursor, + struct callchain_root *dst, struct callchain_root *src) +{ + return merge_chain_branch(cursor, &dst->node, &src->node); +} + +int callchain_cursor_append(struct callchain_cursor *cursor, + u64 ip, struct map *map, struct symbol *sym) { - struct resolved_chain *chain; - int err; + struct callchain_cursor_node *node = *cursor->last; - chain = malloc(sizeof(*chain) + - src->max_depth * sizeof(struct resolved_ip)); - if (!chain) - return -ENOMEM; + if (!node) { + node = calloc(sizeof(*node), 1); + if (!node) + return -ENOMEM; - chain->nr = 0; + *cursor->last = node; + } - err = merge_chain_branch(&dst->node, &src->node, chain); + node->ip = ip; + node->map = map; + node->sym = sym; - free(chain); + cursor->nr++; - return err; + cursor->last = &node->next; + + return 0; } diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index c15fb8c24ad2..d74a19af4a44 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -49,6 +49,27 @@ struct callchain_list { struct list_head list; }; +/* + * A callchain cursor is a single linked list that + * let one feed a callchain progressively. + * It keeps persitent allocated entries to minimize + * allocations. + */ +struct callchain_cursor_node { + u64 ip; + struct map *map; + struct symbol *sym; + struct callchain_cursor_node *next; +}; + +struct callchain_cursor { + u64 nr; + struct callchain_cursor_node *first; + struct callchain_cursor_node **last; + u64 pos; + struct callchain_cursor_node *curr; +}; + static inline void callchain_init(struct callchain_root *root) { INIT_LIST_HEAD(&root->node.brothers); @@ -67,9 +88,48 @@ static inline u64 cumul_hits(struct callchain_node *node) } int register_callchain_param(struct callchain_param *param); -int callchain_append(struct callchain_root *root, struct ip_callchain *chain, - struct map_symbol *syms, u64 period); -int callchain_merge(struct callchain_root *dst, struct callchain_root *src); +int callchain_append(struct callchain_root *root, + struct callchain_cursor *cursor, + u64 period); + +int callchain_merge(struct callchain_cursor *cursor, + struct callchain_root *dst, struct callchain_root *src); bool ip_callchain__valid(struct ip_callchain *chain, const event_t *event); + +/* + * Initialize a cursor before adding entries inside, but keep + * the previously allocated entries as a cache. + */ +static inline void callchain_cursor_reset(struct callchain_cursor *cursor) +{ + cursor->nr = 0; + cursor->last = &cursor->first; +} + +int callchain_cursor_append(struct callchain_cursor *cursor, u64 ip, + struct map *map, struct symbol *sym); + +/* Close a cursor writing session. Initialize for the reader */ +static inline void callchain_cursor_commit(struct callchain_cursor *cursor) +{ + cursor->curr = cursor->first; + cursor->pos = 0; +} + +/* Cursor reading iteration helpers */ +static inline struct callchain_cursor_node * +callchain_cursor_current(struct callchain_cursor *cursor) +{ + if (cursor->pos == cursor->nr) + return NULL; + + return cursor->curr; +} + +static inline void callchain_cursor_advance(struct callchain_cursor *cursor) +{ + cursor->curr = cursor->curr->next; + cursor->pos++; +} #endif /* __PERF_CALLCHAIN_H */ diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 32f4f1f2f6e4..a438a0652d23 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -211,7 +211,9 @@ void hist_entry__free(struct hist_entry *he) * collapse the histogram */ -static bool collapse__insert_entry(struct rb_root *root, struct hist_entry *he) +static bool hists__collapse_insert_entry(struct hists *self, + struct rb_root *root, + struct hist_entry *he) { struct rb_node **p = &root->rb_node; struct rb_node *parent = NULL; @@ -226,8 +228,11 @@ static bool collapse__insert_entry(struct rb_root *root, struct hist_entry *he) if (!cmp) { iter->period += he->period; - if (symbol_conf.use_callchain) - callchain_merge(iter->callchain, he->callchain); + if (symbol_conf.use_callchain) { + callchain_cursor_reset(&self->callchain_cursor); + callchain_merge(&self->callchain_cursor, iter->callchain, + he->callchain); + } hist_entry__free(he); return false; } @@ -262,7 +267,7 @@ void hists__collapse_resort(struct hists *self) next = rb_next(&n->rb_node); rb_erase(&n->rb_node, &self->entries); - if (collapse__insert_entry(&tmp, n)) + if (hists__collapse_insert_entry(self, &tmp, n)) hists__inc_nr_entries(self, n); } diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index ee789856a8c9..889559b86492 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -77,6 +77,8 @@ struct hists { u64 event_stream; u32 type; u16 col_len[HISTC_NR_COLS]; + /* Best would be to reuse the session callchain cursor */ + struct callchain_cursor callchain_cursor; }; struct hist_entry *__hists__add_entry(struct hists *self, diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 105f00bfd555..b58a48a5e5a9 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -242,17 +242,16 @@ static bool symbol__match_parent_regex(struct symbol *sym) return 0; } -struct map_symbol *perf_session__resolve_callchain(struct perf_session *self, - struct thread *thread, - struct ip_callchain *chain, - struct symbol **parent) +int perf_session__resolve_callchain(struct perf_session *self, + struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent) { u8 cpumode = PERF_RECORD_MISC_USER; unsigned int i; - struct map_symbol *syms = calloc(chain->nr, sizeof(*syms)); + int err; - if (!syms) - return NULL; + callchain_cursor_reset(&self->callchain_cursor); for (i = 0; i < chain->nr; i++) { u64 ip = chain->ips[i]; @@ -281,12 +280,15 @@ struct map_symbol *perf_session__resolve_callchain(struct perf_session *self, *parent = al.sym; if (!symbol_conf.use_callchain) break; - syms[i].map = al.map; - syms[i].sym = al.sym; } + + err = callchain_cursor_append(&self->callchain_cursor, + ip, al.map, al.sym); + if (err) + return err; } - return syms; + return 0; } static int process_event_synth_stub(event_t *event __used, diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index decd83f274fd..e815468eb888 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -51,7 +51,8 @@ struct perf_session { int cwdlen; char *cwd; struct ordered_samples ordered_samples; - char filename[0]; + struct callchain_cursor callchain_cursor; + char filename[0]; }; struct perf_event_ops; @@ -94,10 +95,10 @@ int __perf_session__process_events(struct perf_session *self, int perf_session__process_events(struct perf_session *self, struct perf_event_ops *event_ops); -struct map_symbol *perf_session__resolve_callchain(struct perf_session *self, - struct thread *thread, - struct ip_callchain *chain, - struct symbol **parent); +int perf_session__resolve_callchain(struct perf_session *self, + struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent); bool perf_session__has_traces(struct perf_session *self, const char *msg); -- cgit v1.2.3 From f08c3154ac439c4b5762a40107d84e839e08fbc5 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 14 Jan 2011 04:51:59 +0100 Subject: perf callchain: Rename cumul_hits into callchain_cumul_hits That makes the callchain API naming more consistent and reduce potential naming clashes. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra LKML-Reference: <1294977121-5700-3-git-send-email-fweisbec@gmail.com> Signed-off-by: Frederic Weisbecker Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/callchain.c | 10 +++++----- tools/perf/util/callchain.h | 2 +- tools/perf/util/hist.c | 2 +- tools/perf/util/ui/browsers/hists.c | 2 +- 4 files changed, 8 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index 53a49e0cfc6c..4c6360fc2d5b 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -38,14 +38,14 @@ rb_insert_callchain(struct rb_root *root, struct callchain_node *chain, struct rb_node **p = &root->rb_node; struct rb_node *parent = NULL; struct callchain_node *rnode; - u64 chain_cumul = cumul_hits(chain); + u64 chain_cumul = callchain_cumul_hits(chain); while (*p) { u64 rnode_cumul; parent = *p; rnode = rb_entry(parent, struct callchain_node, rb_node); - rnode_cumul = cumul_hits(rnode); + rnode_cumul = callchain_cumul_hits(rnode); switch (mode) { case CHAIN_FLAT: @@ -104,7 +104,7 @@ static void __sort_chain_graph_abs(struct callchain_node *node, chain_for_each_child(child, node) { __sort_chain_graph_abs(child, min_hit); - if (cumul_hits(child) >= min_hit) + if (callchain_cumul_hits(child) >= min_hit) rb_insert_callchain(&node->rb_root, child, CHAIN_GRAPH_ABS); } @@ -129,7 +129,7 @@ static void __sort_chain_graph_rel(struct callchain_node *node, chain_for_each_child(child, node) { __sort_chain_graph_rel(child, min_percent); - if (cumul_hits(child) >= min_hit) + if (callchain_cumul_hits(child) >= min_hit) rb_insert_callchain(&node->rb_root, child, CHAIN_GRAPH_REL); } @@ -270,7 +270,7 @@ split_add_child(struct callchain_node *parent, /* split the hits */ new->hit = parent->hit; new->children_hit = parent->children_hit; - parent->children_hit = cumul_hits(new); + parent->children_hit = callchain_cumul_hits(new); new->val_nr = parent->val_nr - idx_local; parent->val_nr = idx_local; diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index d74a19af4a44..07f71e3e0a71 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -82,7 +82,7 @@ static inline void callchain_init(struct callchain_root *root) root->max_depth = 0; } -static inline u64 cumul_hits(struct callchain_node *node) +static inline u64 callchain_cumul_hits(struct callchain_node *node) { return node->hit + node->children_hit; } diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index a438a0652d23..02ed318d7312 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -430,7 +430,7 @@ static size_t __callchain__fprintf_graph(FILE *fp, struct callchain_node *self, u64 cumul; child = rb_entry(node, struct callchain_node, rb_node); - cumul = cumul_hits(child); + cumul = callchain_cumul_hits(child); remaining -= cumul; /* diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c index 60c463c16028..86428239fa65 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -377,7 +377,7 @@ static int hist_browser__show_callchain_node_rb_tree(struct hist_browser *self, while (node) { struct callchain_node *child = rb_entry(node, struct callchain_node, rb_node); struct rb_node *next = rb_next(node); - u64 cumul = cumul_hits(child); + u64 cumul = callchain_cumul_hits(child); struct callchain_list *chain; char folded_sign = ' '; int first = true; -- cgit v1.2.3 From 16537f1355017a285b904bfb6bf767464293e69c Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 14 Jan 2011 04:52:00 +0100 Subject: perf callchain: Rename register_callchain_param into callchain_register_param To make the callchain API naming more consistent. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra LKML-Reference: <1294977121-5700-4-git-send-email-fweisbec@gmail.com> Signed-off-by: Frederic Weisbecker Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-report.c | 4 ++-- tools/perf/util/callchain.c | 2 +- tools/perf/util/callchain.h | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index c95599a82f9e..f6a43493d1d0 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -221,7 +221,7 @@ static int perf_session__setup_sample_type(struct perf_session *self) } else if (!dont_use_callchains && callchain_param.mode != CHAIN_NONE && !symbol_conf.use_callchain) { symbol_conf.use_callchain = true; - if (register_callchain_param(&callchain_param) < 0) { + if (callchain_register_param(&callchain_param) < 0) { fprintf(stderr, "Can't register callchain" " params\n"); return -EINVAL; @@ -423,7 +423,7 @@ parse_callchain_opt(const struct option *opt __used, const char *arg, if (tok2) callchain_param.print_limit = strtod(tok2, &endptr); setup: - if (register_callchain_param(&callchain_param) < 0) { + if (callchain_register_param(&callchain_param) < 0) { fprintf(stderr, "Can't register callchain params\n"); return -1; } diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index 4c6360fc2d5b..5b3f09dd137d 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -143,7 +143,7 @@ sort_chain_graph_rel(struct rb_root *rb_root, struct callchain_root *chain_root, rb_root->rb_node = chain_root->node.rb_root.rb_node; } -int register_callchain_param(struct callchain_param *param) +int callchain_register_param(struct callchain_param *param) { switch (param->mode) { case CHAIN_GRAPH_ABS: diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index 07f71e3e0a71..2bb5403010d2 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -87,7 +87,7 @@ static inline u64 callchain_cumul_hits(struct callchain_node *node) return node->hit + node->children_hit; } -int register_callchain_param(struct callchain_param *param); +int callchain_register_param(struct callchain_param *param); int callchain_append(struct callchain_root *root, struct callchain_cursor *cursor, u64 period); -- cgit v1.2.3 From 529363b76929beb85b81439c61063130af046a21 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 14 Jan 2011 04:52:01 +0100 Subject: perf callchain: Don't give arbitrary gender to callchain tree nodes Some little callchain tree nodes shyly asked me if they can have sisters. How cute! Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra LKML-Reference: <1294977121-5700-5-git-send-email-fweisbec@gmail.com> Signed-off-by: Frederic Weisbecker Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/callchain.c | 8 ++++---- tools/perf/util/callchain.h | 4 ++-- 2 files changed, 6 insertions(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index 5b3f09dd137d..f8c66d1435e0 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -26,10 +26,10 @@ bool ip_callchain__valid(struct ip_callchain *chain, const event_t *event) } #define chain_for_each_child(child, parent) \ - list_for_each_entry(child, &parent->children, brothers) + list_for_each_entry(child, &parent->children, siblings) #define chain_for_each_child_safe(child, next, parent) \ - list_for_each_entry_safe(child, next, &parent->children, brothers) + list_for_each_entry_safe(child, next, &parent->children, siblings) static void rb_insert_callchain(struct rb_root *root, struct callchain_node *chain, @@ -189,7 +189,7 @@ create_child(struct callchain_node *parent, bool inherit_children) chain_for_each_child(next, new) next->parent = new; } - list_add_tail(&new->brothers, &parent->children); + list_add_tail(&new->siblings, &parent->children); return new; } @@ -419,7 +419,7 @@ merge_chain_branch(struct callchain_cursor *cursor, if (err) break; - list_del(&child->brothers); + list_del(&child->siblings); free(child); } diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index 2bb5403010d2..67137256a1cd 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -16,7 +16,7 @@ enum chain_mode { struct callchain_node { struct callchain_node *parent; - struct list_head brothers; + struct list_head siblings; struct list_head children; struct list_head val; struct rb_node rb_node; /* to sort nodes in an rbtree */ @@ -72,7 +72,7 @@ struct callchain_cursor { static inline void callchain_init(struct callchain_root *root) { - INIT_LIST_HEAD(&root->node.brothers); + INIT_LIST_HEAD(&root->node.siblings); INIT_LIST_HEAD(&root->node.children); INIT_LIST_HEAD(&root->node.val); -- cgit v1.2.3 From 4cc9cec636e7f78aba7f17606ac13cac07ea5787 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 13 Jan 2011 21:45:58 +0900 Subject: perf probe: Introduce lines walker interface Introduce die_walk_lines() for walking on the line list of given die, and use it in line_range finder and probe point finder. Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Franck Bui-Huu Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Steven Rostedt LKML-Reference: <20110113124558.22426.48170.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu [ committer note: s/%ld/%zd/ for a size_t nlines var that broke f14 x86 build] Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-finder.c | 321 +++++++++++++++++++++-------------------- 1 file changed, 167 insertions(+), 154 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index ab83b6ac5d65..508c017f566a 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -458,6 +458,124 @@ static Dwarf_Die *die_find_inlinefunc(Dwarf_Die *sp_die, Dwarf_Addr addr, return die_find_child(sp_die, __die_find_inline_cb, &addr, die_mem); } +/* Walker on lines (Note: line number will not be sorted) */ +typedef int (* line_walk_handler_t) (const char *fname, int lineno, + Dwarf_Addr addr, void *data); + +struct __line_walk_param { + line_walk_handler_t handler; + void *data; + int retval; +}; + +/* Walk on decl lines in given DIE */ +static int __die_walk_funclines(Dwarf_Die *sp_die, + line_walk_handler_t handler, void *data) +{ + const char *fname; + Dwarf_Addr addr; + int lineno, ret = 0; + + /* Handle function declaration line */ + fname = dwarf_decl_file(sp_die); + if (fname && dwarf_decl_line(sp_die, &lineno) == 0 && + dwarf_entrypc(sp_die, &addr) == 0) { + ret = handler(fname, lineno, addr, data); + } + + return ret; +} + +static int __die_walk_culines_cb(Dwarf_Die *sp_die, void *data) +{ + struct __line_walk_param *lw = data; + + lw->retval = __die_walk_funclines(sp_die, lw->handler, lw->data); + if (lw->retval != 0) + return DWARF_CB_ABORT; + + return DWARF_CB_OK; +} + +/* + * Walk on lines inside given PDIE. If the PDIE is subprogram, walk only on + * the lines inside the subprogram, otherwise PDIE must be a CU DIE. + */ +static int die_walk_lines(Dwarf_Die *pdie, line_walk_handler_t handler, + void *data) +{ + Dwarf_Lines *lines; + Dwarf_Line *line; + Dwarf_Addr addr; + const char *fname; + int lineno, ret = 0; + Dwarf_Die die_mem, *cu_die; + size_t nlines, i; + + /* Get the CU die */ + if (dwarf_tag(pdie) == DW_TAG_subprogram) + cu_die = dwarf_diecu(pdie, &die_mem, NULL, NULL); + else + cu_die = pdie; + if (!cu_die) { + pr_debug2("Failed to get CU from subprogram\n"); + return -EINVAL; + } + + /* Get lines list in the CU */ + if (dwarf_getsrclines(cu_die, &lines, &nlines) != 0) { + pr_debug2("Failed to get source lines on this CU.\n"); + return -ENOENT; + } + pr_debug2("Get %zd lines from this CU\n", nlines); + + /* Walk on the lines on lines list */ + for (i = 0; i < nlines; i++) { + line = dwarf_onesrcline(lines, i); + if (line == NULL || + dwarf_lineno(line, &lineno) != 0 || + dwarf_lineaddr(line, &addr) != 0) { + pr_debug2("Failed to get line info. " + "Possible error in debuginfo.\n"); + continue; + } + /* Filter lines based on address */ + if (pdie != cu_die) + /* + * Address filtering + * The line is included in given function, and + * no inline block includes it. + */ + if (!dwarf_haspc(pdie, addr) || + die_find_inlinefunc(pdie, addr, &die_mem)) + continue; + /* Get source line */ + fname = dwarf_linesrc(line, NULL, NULL); + + ret = handler(fname, lineno, addr, data); + if (ret != 0) + return ret; + } + + /* + * Dwarf lines doesn't include function declarations and inlined + * subroutines. We have to check functions list or given function. + */ + if (pdie != cu_die) + ret = __die_walk_funclines(pdie, handler, data); + else { + struct __line_walk_param param = { + .handler = handler, + .data = data, + .retval = 0, + }; + dwarf_getfuncs(cu_die, __die_walk_culines_cb, ¶m, 0); + ret = param.retval; + } + + return ret; +} + struct __find_variable_param { const char *name; Dwarf_Addr addr; @@ -1050,43 +1168,26 @@ static int call_probe_finder(Dwarf_Die *sp_die, struct probe_finder *pf) return ret; } -/* Find probe point from its line number */ -static int find_probe_point_by_line(struct probe_finder *pf) +static int probe_point_line_walker(const char *fname, int lineno, + Dwarf_Addr addr, void *data) { - Dwarf_Lines *lines; - Dwarf_Line *line; - size_t nlines, i; - Dwarf_Addr addr; - int lineno; - int ret = 0; + struct probe_finder *pf = data; + int ret; - if (dwarf_getsrclines(&pf->cu_die, &lines, &nlines) != 0) { - pr_warning("No source lines found.\n"); - return -ENOENT; - } + if (lineno != pf->lno || strtailcmp(fname, pf->fname) != 0) + return 0; - for (i = 0; i < nlines && ret == 0; i++) { - line = dwarf_onesrcline(lines, i); - if (dwarf_lineno(line, &lineno) != 0 || - lineno != pf->lno) - continue; + pf->addr = addr; + ret = call_probe_finder(NULL, pf); - /* TODO: Get fileno from line, but how? */ - if (strtailcmp(dwarf_linesrc(line, NULL, NULL), pf->fname) != 0) - continue; - - if (dwarf_lineaddr(line, &addr) != 0) { - pr_warning("Failed to get the address of the line.\n"); - return -ENOENT; - } - pr_debug("Probe line found: line[%d]:%d addr:0x%jx\n", - (int)i, lineno, (uintmax_t)addr); - pf->addr = addr; + /* Continue if no error, because the line will be in inline function */ + return ret < 0 ?: 0; +} - ret = call_probe_finder(NULL, pf); - /* Continuing, because target line might be inlined. */ - } - return ret; +/* Find probe point from its line number */ +static int find_probe_point_by_line(struct probe_finder *pf) +{ + return die_walk_lines(&pf->cu_die, probe_point_line_walker, pf); } /* Find lines which match lazy pattern */ @@ -1140,15 +1241,31 @@ out_close: return nlines; } +static int probe_point_lazy_walker(const char *fname, int lineno, + Dwarf_Addr addr, void *data) +{ + struct probe_finder *pf = data; + int ret; + + if (!line_list__has_line(&pf->lcache, lineno) || + strtailcmp(fname, pf->fname) != 0) + return 0; + + pr_debug("Probe line found: line:%d addr:0x%llx\n", + lineno, (unsigned long long)addr); + pf->addr = addr; + ret = call_probe_finder(NULL, pf); + + /* + * Continue if no error, because the lazy pattern will match + * to other lines + */ + return ret < 0 ?: 0; +} + /* Find probe points from lazy pattern */ static int find_probe_point_lazy(Dwarf_Die *sp_die, struct probe_finder *pf) { - Dwarf_Lines *lines; - Dwarf_Line *line; - size_t nlines, i; - Dwarf_Addr addr; - Dwarf_Die die_mem; - int lineno; int ret = 0; if (list_empty(&pf->lcache)) { @@ -1162,45 +1279,7 @@ static int find_probe_point_lazy(Dwarf_Die *sp_die, struct probe_finder *pf) return ret; } - if (dwarf_getsrclines(&pf->cu_die, &lines, &nlines) != 0) { - pr_warning("No source lines found.\n"); - return -ENOENT; - } - - for (i = 0; i < nlines && ret >= 0; i++) { - line = dwarf_onesrcline(lines, i); - - if (dwarf_lineno(line, &lineno) != 0 || - !line_list__has_line(&pf->lcache, lineno)) - continue; - - /* TODO: Get fileno from line, but how? */ - if (strtailcmp(dwarf_linesrc(line, NULL, NULL), pf->fname) != 0) - continue; - - if (dwarf_lineaddr(line, &addr) != 0) { - pr_debug("Failed to get the address of line %d.\n", - lineno); - continue; - } - if (sp_die) { - /* Address filtering 1: does sp_die include addr? */ - if (!dwarf_haspc(sp_die, addr)) - continue; - /* Address filtering 2: No child include addr? */ - if (die_find_inlinefunc(sp_die, addr, &die_mem)) - continue; - } - - pr_debug("Probe line found: line[%d]:%d addr:0x%llx\n", - (int)i, lineno, (unsigned long long)addr); - pf->addr = addr; - - ret = call_probe_finder(sp_die, pf); - /* Continuing, because target line might be inlined. */ - } - /* TODO: deallocate lines, but how? */ - return ret; + return die_walk_lines(sp_die, probe_point_lazy_walker, pf); } /* Callback parameter with return value */ @@ -1644,91 +1723,28 @@ static int line_range_add_line(const char *src, unsigned int lineno, return line_list__add_line(&lr->line_list, lineno); } -/* Search function declaration lines */ -static int line_range_funcdecl_cb(Dwarf_Die *sp_die, void *data) +static int line_range_walk_cb(const char *fname, int lineno, + Dwarf_Addr addr __used, + void *data) { - struct dwarf_callback_param *param = data; - struct line_finder *lf = param->data; - const char *src; - int lineno; - - src = dwarf_decl_file(sp_die); - if (src && strtailcmp(src, lf->fname) != 0) - return DWARF_CB_OK; + struct line_finder *lf = data; - if (dwarf_decl_line(sp_die, &lineno) != 0 || + if ((strtailcmp(fname, lf->fname) != 0) || (lf->lno_s > lineno || lf->lno_e < lineno)) - return DWARF_CB_OK; + return 0; - param->retval = line_range_add_line(src, lineno, lf->lr); - if (param->retval < 0) - return DWARF_CB_ABORT; - return DWARF_CB_OK; -} + if (line_range_add_line(fname, lineno, lf->lr) < 0) + return -EINVAL; -static int find_line_range_func_decl_lines(struct line_finder *lf) -{ - struct dwarf_callback_param param = {.data = (void *)lf, .retval = 0}; - dwarf_getfuncs(&lf->cu_die, line_range_funcdecl_cb, ¶m, 0); - return param.retval; + return 0; } /* Find line range from its line number */ static int find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf) { - Dwarf_Lines *lines; - Dwarf_Line *line; - size_t nlines, i; - Dwarf_Addr addr; - int lineno, ret = 0; - const char *src; - Dwarf_Die die_mem; - - line_list__init(&lf->lr->line_list); - if (dwarf_getsrclines(&lf->cu_die, &lines, &nlines) != 0) { - pr_warning("No source lines found.\n"); - return -ENOENT; - } - - /* Search probable lines on lines list */ - for (i = 0; i < nlines; i++) { - line = dwarf_onesrcline(lines, i); - if (dwarf_lineno(line, &lineno) != 0 || - (lf->lno_s > lineno || lf->lno_e < lineno)) - continue; - - if (sp_die) { - /* Address filtering 1: does sp_die include addr? */ - if (dwarf_lineaddr(line, &addr) != 0 || - !dwarf_haspc(sp_die, addr)) - continue; - - /* Address filtering 2: No child include addr? */ - if (die_find_inlinefunc(sp_die, addr, &die_mem)) - continue; - } - - /* TODO: Get fileno from line, but how? */ - src = dwarf_linesrc(line, NULL, NULL); - if (strtailcmp(src, lf->fname) != 0) - continue; - - ret = line_range_add_line(src, lineno, lf->lr); - if (ret < 0) - return ret; - } + int ret; - /* - * Dwarf lines doesn't include function declarations. We have to - * check functions list or given function. - */ - if (sp_die) { - src = dwarf_decl_file(sp_die); - if (src && dwarf_decl_line(sp_die, &lineno) == 0 && - (lf->lno_s <= lineno && lf->lno_e >= lineno)) - ret = line_range_add_line(src, lineno, lf->lr); - } else - ret = find_line_range_func_decl_lines(lf); + ret = die_walk_lines(sp_die ?: &lf->cu_die, line_range_walk_cb, lf); /* Update status */ if (ret >= 0) @@ -1758,9 +1774,6 @@ static int line_range_search_cb(Dwarf_Die *sp_die, void *data) struct line_finder *lf = param->data; struct line_range *lr = lf->lr; - pr_debug("find (%llx) %s\n", - (unsigned long long)dwarf_dieoffset(sp_die), - dwarf_diename(sp_die)); if (dwarf_tag(sp_die) == DW_TAG_subprogram && die_compare_name(sp_die, lr->function)) { lf->fname = dwarf_decl_file(sp_die); -- cgit v1.2.3 From 5069ed86be3c2f28bcdf7fae1374ec0c325aafba Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 13 Jan 2011 21:46:05 +0900 Subject: perf probe: Enable to put probe inline function call site Enable to put probe inline function call site. This will increase line-based probe-ability. $ ./perf probe -L schedule:48 pre_schedule(rq, prev); 50 if (unlikely(!rq->nr_running)) idle_balance(cpu, rq); put_prev_task(rq, prev); next = pick_next_task(rq); 56 if (likely(prev != next)) { sched_info_switch(prev, next); trace_sched_switch_out(prev, next); perf_event_task_sched_out(prev, next); $ ./perf probe -L schedule:48 48 pre_schedule(rq, prev); 50 if (unlikely(!rq->nr_running)) 51 idle_balance(cpu, rq); 53 put_prev_task(rq, prev); 54 next = pick_next_task(rq); 56 if (likely(prev != next)) { 57 sched_info_switch(prev, next); 58 trace_sched_switch_out(prev, next); 59 perf_event_task_sched_out(prev, next); Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Franck Bui-Huu Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Steven Rostedt LKML-Reference: <20110113124604.22426.48873.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-finder.c | 56 ++++++++++++++++++++++++++++++++++++------ 1 file changed, 48 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index 508c017f566a..69215bff17e9 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -280,6 +280,19 @@ static bool die_compare_name(Dwarf_Die *dw_die, const char *tname) return name ? (strcmp(tname, name) == 0) : false; } +/* Get callsite line number of inline-function instance */ +static int die_get_call_lineno(Dwarf_Die *in_die) +{ + Dwarf_Attribute attr; + Dwarf_Word ret; + + if (!dwarf_attr(in_die, DW_AT_call_line, &attr)) + return -ENOENT; + + dwarf_formudata(&attr, &ret); + return (int)ret; +} + /* Get type die */ static Dwarf_Die *die_get_type(Dwarf_Die *vr_die, Dwarf_Die *die_mem) { @@ -463,27 +476,54 @@ typedef int (* line_walk_handler_t) (const char *fname, int lineno, Dwarf_Addr addr, void *data); struct __line_walk_param { + const char *fname; line_walk_handler_t handler; void *data; int retval; }; -/* Walk on decl lines in given DIE */ +static int __die_walk_funclines_cb(Dwarf_Die *in_die, void *data) +{ + struct __line_walk_param *lw = data; + Dwarf_Addr addr; + int lineno; + + if (dwarf_tag(in_die) == DW_TAG_inlined_subroutine) { + lineno = die_get_call_lineno(in_die); + if (lineno > 0 && dwarf_entrypc(in_die, &addr) == 0) { + lw->retval = lw->handler(lw->fname, lineno, addr, + lw->data); + if (lw->retval != 0) + return DIE_FIND_CB_FOUND; + } + } + return DIE_FIND_CB_SIBLING; +} + +/* Walk on lines of blocks included in given DIE */ static int __die_walk_funclines(Dwarf_Die *sp_die, line_walk_handler_t handler, void *data) { - const char *fname; + struct __line_walk_param lw = { + .handler = handler, + .data = data, + .retval = 0, + }; + Dwarf_Die die_mem; Dwarf_Addr addr; - int lineno, ret = 0; + int lineno; /* Handle function declaration line */ - fname = dwarf_decl_file(sp_die); - if (fname && dwarf_decl_line(sp_die, &lineno) == 0 && + lw.fname = dwarf_decl_file(sp_die); + if (lw.fname && dwarf_decl_line(sp_die, &lineno) == 0 && dwarf_entrypc(sp_die, &addr) == 0) { - ret = handler(fname, lineno, addr, data); + lw.retval = handler(lw.fname, lineno, addr, data); + if (lw.retval != 0) + goto done; } - - return ret; + die_find_child(sp_die, __die_walk_funclines_cb, &lw, &die_mem); +done: + return lw.retval; } static int __die_walk_culines_cb(Dwarf_Die *sp_die, void *data) -- cgit v1.2.3 From e80711ca8512c8586da0c3e18e2f1caf73c88731 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 13 Jan 2011 21:46:11 +0900 Subject: perf probe: Add --funcs to show available functions in symtab Add --funcs to show available functions in symtab. Originally this feature came from Srikar's uprobes patches ( http://lkml.org/lkml/2010/8/27/244 ) e.g. ... __ablkcipher_walk_complete __absent_pages_in_range __account_scheduler_latency __add_pages __alloc_pages_nodemask __alloc_percpu __alloc_reserved_percpu __alloc_skb __alloc_workqueue_key __any_online_cpu __ata_ehi_push_desc ... This also supports symbols in module, e.g. ... cleanup_module cpuid_maxphyaddr emulate_clts emulate_instruction emulate_int_real emulate_invlpg emulator_get_dr emulator_set_dr emulator_task_switch emulator_write_emulated emulator_write_phys fx_init ... Original-patch-from: Srikar Dronamraju Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Franck Bui-Huu Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Steven Rostedt LKML-Reference: <20110113124611.22426.10835.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu [ committer note: Add missing elf.h for STB_GLOBAL that broke a RHEL4 build ] Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Documentation/perf-probe.txt | 4 ++ tools/perf/builtin-probe.c | 29 +++++++++++++- tools/perf/util/probe-event.c | 68 ++++++++++++++++++++++++++++++++- tools/perf/util/probe-event.h | 1 + 4 files changed, 99 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/Documentation/perf-probe.txt b/tools/perf/Documentation/perf-probe.txt index 86b797a35aa6..fcc51fe0195c 100644 --- a/tools/perf/Documentation/perf-probe.txt +++ b/tools/perf/Documentation/perf-probe.txt @@ -73,6 +73,10 @@ OPTIONS (Only for --vars) Show external defined variables in addition to local variables. +-F:: +--funcs:: + Show available functions in given module or kernel. + -f:: --force:: Forcibly add events with existing name. diff --git a/tools/perf/builtin-probe.c b/tools/perf/builtin-probe.c index add163c9f0e7..6cf708aba7c9 100644 --- a/tools/perf/builtin-probe.c +++ b/tools/perf/builtin-probe.c @@ -52,6 +52,7 @@ static struct { bool show_lines; bool show_vars; bool show_ext_vars; + bool show_funcs; bool mod_events; int nevents; struct perf_probe_event events[MAX_PROBES]; @@ -221,6 +222,8 @@ static const struct option options[] = { OPT__DRY_RUN(&probe_event_dry_run), OPT_INTEGER('\0', "max-probes", ¶ms.max_probe_points, "Set how many probe points can be found for a probe."), + OPT_BOOLEAN('F', "funcs", ¶ms.show_funcs, + "Show potential probe-able functions."), OPT_END() }; @@ -246,7 +249,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) params.max_probe_points = MAX_PROBES; if ((!params.nevents && !params.dellist && !params.list_events && - !params.show_lines)) + !params.show_lines && !params.show_funcs)) usage_with_options(probe_usage, options); /* @@ -267,12 +270,36 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) pr_err(" Error: Don't use --list with --vars.\n"); usage_with_options(probe_usage, options); } + if (params.show_funcs) { + pr_err(" Error: Don't use --list with --funcs.\n"); + usage_with_options(probe_usage, options); + } ret = show_perf_probe_events(); if (ret < 0) pr_err(" Error: Failed to show event list. (%d)\n", ret); return ret; } + if (params.show_funcs) { + if (params.nevents != 0 || params.dellist) { + pr_err(" Error: Don't use --funcs with" + " --add/--del.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_lines) { + pr_err(" Error: Don't use --funcs with --line.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_vars) { + pr_err(" Error: Don't use --funcs with --vars.\n"); + usage_with_options(probe_usage, options); + } + ret = show_available_funcs(params.target_module); + if (ret < 0) + pr_err(" Error: Failed to show functions." + " (%d)\n", ret); + return ret; + } #ifdef DWARF_SUPPORT if (params.show_lines) { diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 6e29d9c9dccc..859d377a3df3 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -31,6 +31,7 @@ #include #include #include +#include #undef _GNU_SOURCE #include "util.h" @@ -111,7 +112,25 @@ static struct symbol *__find_kernel_function_by_name(const char *name, NULL); } -const char *kernel_get_module_path(const char *module) +static struct map *kernel_get_module_map(const char *module) +{ + struct rb_node *nd; + struct map_groups *grp = &machine.kmaps; + + if (!module) + module = "kernel"; + + for (nd = rb_first(&grp->maps[MAP__FUNCTION]); nd; nd = rb_next(nd)) { + struct map *pos = rb_entry(nd, struct map, rb_node); + if (strncmp(pos->dso->short_name + 1, module, + pos->dso->short_name_len - 2) == 0) { + return pos; + } + } + return NULL; +} + +static struct dso *kernel_get_module_dso(const char *module) { struct dso *dso; struct map *map; @@ -141,7 +160,13 @@ const char *kernel_get_module_path(const char *module) } } found: - return dso->long_name; + return dso; +} + +const char *kernel_get_module_path(const char *module) +{ + struct dso *dso = kernel_get_module_dso(module); + return (dso) ? dso->long_name : NULL; } #ifdef DWARF_SUPPORT @@ -1913,3 +1938,42 @@ int del_perf_probe_events(struct strlist *dellist) return ret; } +/* + * If a symbol corresponds to a function with global binding return 0. + * For all others return 1. + */ +static int filter_non_global_functions(struct map *map __unused, + struct symbol *sym) +{ + if (sym->binding != STB_GLOBAL) + return 1; + + return 0; +} + +int show_available_funcs(const char *module) +{ + struct map *map; + int ret; + + setup_pager(); + + ret = init_vmlinux(); + if (ret < 0) + return ret; + + map = kernel_get_module_map(module); + if (!map) { + pr_err("Failed to find %s map.\n", (module) ? : "kernel"); + return -EINVAL; + } + if (map__load(map, filter_non_global_functions)) { + pr_err("Failed to load map.\n"); + return -EINVAL; + } + if (!dso__sorted_by_name(map->dso, map->type)) + dso__sort_by_name(map->dso, map->type); + + dso__fprintf_symbols_by_name(map->dso, map->type, stdout); + return 0; +} diff --git a/tools/perf/util/probe-event.h b/tools/perf/util/probe-event.h index 5accbedfea37..1fb4f18337d3 100644 --- a/tools/perf/util/probe-event.h +++ b/tools/perf/util/probe-event.h @@ -127,6 +127,7 @@ extern int show_line_range(struct line_range *lr, const char *module); extern int show_available_vars(struct perf_probe_event *pevs, int npevs, int max_probe_points, const char *module, bool externs); +extern int show_available_funcs(const char *module); /* Maximum index number of event-name postfix */ -- cgit v1.2.3 From 17ea1b70a87e28457821318341bead2b45563092 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 17 Jan 2011 14:40:46 -0200 Subject: perf tools: Pass the struct opt to the wildcard parsing routine It is needed because it will call parse_event for each tracepoint name that matches, and we pass the perf_evlist via opt->value. Problem introduced in 4503fdd where my assumption about opt being always non NULL made me not look at callers of parse_events outside builtin-*.c. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.c | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index d3086cecd2dd..cf082daa43e3 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -446,8 +446,8 @@ parse_single_tracepoint_event(char *sys_name, /* sys + ':' + event + ':' + flags*/ #define MAX_EVOPT_LEN (MAX_EVENT_LENGTH * 2 + 2 + 128) static enum event_result -parse_multiple_tracepoint_event(char *sys_name, const char *evt_exp, - char *flags) +parse_multiple_tracepoint_event(const struct option *opt, char *sys_name, + const char *evt_exp, char *flags) { char evt_path[MAXPATHLEN]; struct dirent *evt_ent; @@ -480,15 +480,16 @@ parse_multiple_tracepoint_event(char *sys_name, const char *evt_exp, if (len < 0) return EVT_FAILED; - if (parse_events(NULL, event_opt, 0)) + if (parse_events(opt, event_opt, 0)) return EVT_FAILED; } return EVT_HANDLED_ALL; } -static enum event_result parse_tracepoint_event(const char **strp, - struct perf_event_attr *attr) +static enum event_result +parse_tracepoint_event(const struct option *opt, const char **strp, + struct perf_event_attr *attr) { const char *evt_name; char *flags = NULL, *comma_loc; @@ -527,7 +528,7 @@ static enum event_result parse_tracepoint_event(const char **strp, return EVT_FAILED; if (strpbrk(evt_name, "*?")) { *strp += strlen(sys_name) + evt_length + 1; /* 1 == the ':' */ - return parse_multiple_tracepoint_event(sys_name, evt_name, + return parse_multiple_tracepoint_event(opt, sys_name, evt_name, flags); } else { return parse_single_tracepoint_event(sys_name, evt_name, @@ -737,11 +738,12 @@ parse_event_modifier(const char **strp, struct perf_event_attr *attr) * Symbolic names are (almost) exactly matched. */ static enum event_result -parse_event_symbols(const char **str, struct perf_event_attr *attr) +parse_event_symbols(const struct option *opt, const char **str, + struct perf_event_attr *attr) { enum event_result ret; - ret = parse_tracepoint_event(str, attr); + ret = parse_tracepoint_event(opt, str, attr); if (ret != EVT_FAILED) goto modifier; @@ -783,7 +785,7 @@ int parse_events(const struct option *opt, const char *str, int unset __used) for (;;) { memset(&attr, 0, sizeof(attr)); - ret = parse_event_symbols(&str, &attr); + ret = parse_event_symbols(opt, &str, &attr); if (ret == EVT_FAILED) return -1; -- cgit v1.2.3 From fd78260b5376173faeb17127bd63b3c99a8e8bfb Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 18 Jan 2011 15:15:24 -0200 Subject: perf threads: Move thread_map to separate file To untangle it from struct thread handling, that is tied to symbols, etc. Right now in the python bindings I'm working on I need just a subset of the util/ files, untangling it allows me to do that. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Makefile | 2 ++ tools/perf/builtin-record.c | 1 + tools/perf/builtin-stat.c | 1 + tools/perf/builtin-test.c | 2 +- tools/perf/builtin-top.c | 1 + tools/perf/util/evsel.c | 2 +- tools/perf/util/thread.c | 55 ------------------------------------- tools/perf/util/thread.h | 14 ---------- tools/perf/util/thread_map.c | 64 ++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/thread_map.h | 15 +++++++++++ 10 files changed, 86 insertions(+), 71 deletions(-) create mode 100644 tools/perf/util/thread_map.c create mode 100644 tools/perf/util/thread_map.h (limited to 'tools/perf/util') diff --git a/tools/perf/Makefile b/tools/perf/Makefile index f20bc6f85611..638e8e146bb9 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -426,6 +426,7 @@ LIB_H += util/values.h LIB_H += util/sort.h LIB_H += util/hist.h LIB_H += util/thread.h +LIB_H += util/thread_map.h LIB_H += util/trace-event.h LIB_H += util/probe-finder.h LIB_H += util/probe-event.h @@ -471,6 +472,7 @@ LIB_OBJS += $(OUTPUT)util/map.o LIB_OBJS += $(OUTPUT)util/pstack.o LIB_OBJS += $(OUTPUT)util/session.o LIB_OBJS += $(OUTPUT)util/thread.o +LIB_OBJS += $(OUTPUT)util/thread_map.o LIB_OBJS += $(OUTPUT)util/trace-event-parse.o LIB_OBJS += $(OUTPUT)util/trace-event-read.o LIB_OBJS += $(OUTPUT)util/trace-event-info.o diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 1346d4230bc0..d7886307f6f4 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -24,6 +24,7 @@ #include "util/session.h" #include "util/symbol.h" #include "util/cpumap.h" +#include "util/thread_map.h" #include #include diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index e2a2d02c5c43..8906adfdbd8e 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -49,6 +49,7 @@ #include "util/header.h" #include "util/cpumap.h" #include "util/thread.h" +#include "util/thread_map.h" #include #include diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index 4fd34537c01d..dc91ee06a37c 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -12,7 +12,7 @@ #include "util/parse-events.h" #include "util/session.h" #include "util/symbol.h" -#include "util/thread.h" +#include "util/thread_map.h" static long page_size; diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 31fbaf38d9c1..d0b16d905405 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -26,6 +26,7 @@ #include "util/session.h" #include "util/symbol.h" #include "util/thread.h" +#include "util/thread_map.h" #include "util/util.h" #include #include "util/parse-options.h" diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index ee490356c817..9a6d94299ab8 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -3,7 +3,7 @@ #include "../perf.h" #include "util.h" #include "cpumap.h" -#include "thread.h" +#include "thread_map.h" #include #include diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c index 00f4eade2e3e..d5d3b22250f3 100644 --- a/tools/perf/util/thread.c +++ b/tools/perf/util/thread.c @@ -7,61 +7,6 @@ #include "util.h" #include "debug.h" -/* Skip "." and ".." directories */ -static int filter(const struct dirent *dir) -{ - if (dir->d_name[0] == '.') - return 0; - else - return 1; -} - -struct thread_map *thread_map__new_by_pid(pid_t pid) -{ - struct thread_map *threads; - char name[256]; - int items; - struct dirent **namelist = NULL; - int i; - - sprintf(name, "/proc/%d/task", pid); - items = scandir(name, &namelist, filter, NULL); - if (items <= 0) - return NULL; - - threads = malloc(sizeof(*threads) + sizeof(pid_t) * items); - if (threads != NULL) { - for (i = 0; i < items; i++) - threads->map[i] = atoi(namelist[i]->d_name); - threads->nr = items; - } - - for (i=0; imap[0] = tid; - threads->nr = 1; - } - - return threads; -} - -struct thread_map *thread_map__new(pid_t pid, pid_t tid) -{ - if (pid != -1) - return thread_map__new_by_pid(pid); - return thread_map__new_by_tid(tid); -} - static struct thread *thread__new(pid_t pid) { struct thread *self = zalloc(sizeof(*self)); diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h index d7574101054a..e5f2401c1b5e 100644 --- a/tools/perf/util/thread.h +++ b/tools/perf/util/thread.h @@ -18,24 +18,10 @@ struct thread { int comm_len; }; -struct thread_map { - int nr; - int map[]; -}; - struct perf_session; void thread__delete(struct thread *self); -struct thread_map *thread_map__new_by_pid(pid_t pid); -struct thread_map *thread_map__new_by_tid(pid_t tid); -struct thread_map *thread_map__new(pid_t pid, pid_t tid); - -static inline void thread_map__delete(struct thread_map *threads) -{ - free(threads); -} - int thread__set_comm(struct thread *self, const char *comm); int thread__comm_len(struct thread *self); struct thread *perf_session__findnew(struct perf_session *self, pid_t pid); diff --git a/tools/perf/util/thread_map.c b/tools/perf/util/thread_map.c new file mode 100644 index 000000000000..a5df131b77c3 --- /dev/null +++ b/tools/perf/util/thread_map.c @@ -0,0 +1,64 @@ +#include +#include +#include +#include "thread_map.h" + +/* Skip "." and ".." directories */ +static int filter(const struct dirent *dir) +{ + if (dir->d_name[0] == '.') + return 0; + else + return 1; +} + +struct thread_map *thread_map__new_by_pid(pid_t pid) +{ + struct thread_map *threads; + char name[256]; + int items; + struct dirent **namelist = NULL; + int i; + + sprintf(name, "/proc/%d/task", pid); + items = scandir(name, &namelist, filter, NULL); + if (items <= 0) + return NULL; + + threads = malloc(sizeof(*threads) + sizeof(pid_t) * items); + if (threads != NULL) { + for (i = 0; i < items; i++) + threads->map[i] = atoi(namelist[i]->d_name); + threads->nr = items; + } + + for (i=0; imap[0] = tid; + threads->nr = 1; + } + + return threads; +} + +struct thread_map *thread_map__new(pid_t pid, pid_t tid) +{ + if (pid != -1) + return thread_map__new_by_pid(pid); + return thread_map__new_by_tid(tid); +} + +void thread_map__delete(struct thread_map *threads) +{ + free(threads); +} diff --git a/tools/perf/util/thread_map.h b/tools/perf/util/thread_map.h new file mode 100644 index 000000000000..3cb907311409 --- /dev/null +++ b/tools/perf/util/thread_map.h @@ -0,0 +1,15 @@ +#ifndef __PERF_THREAD_MAP_H +#define __PERF_THREAD_MAP_H + +#include + +struct thread_map { + int nr; + int map[]; +}; + +struct thread_map *thread_map__new_by_pid(pid_t pid); +struct thread_map *thread_map__new_by_tid(pid_t tid); +struct thread_map *thread_map__new(pid_t pid, pid_t tid); +void thread_map__delete(struct thread_map *threads); +#endif /* __PERF_THREAD_MAP_H */ -- cgit v1.2.3 From d0dd74e853a0a6f37e8061d6d50be41c7034c54c Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 21 Jan 2011 13:46:41 -0200 Subject: perf tools: Move event__parse_sample to evsel.c To avoid linking more stuff in the python binding I'm working on, future csets will make the sample type be taken from the evsel itself, but for that we need to first have one file per cpu and per sample_type, not a single perf.data file. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-test.c | 11 +--- tools/perf/builtin-top.c | 2 +- tools/perf/util/event.c | 125 ---------------------------------------------- tools/perf/util/event.h | 5 +- tools/perf/util/evsel.c | 118 +++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/session.c | 4 +- tools/perf/util/session.h | 9 ++++ 7 files changed, 134 insertions(+), 140 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index dc91ee06a37c..231e3e21810c 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -10,7 +10,6 @@ #include "util/evlist.h" #include "util/parse-options.h" #include "util/parse-events.h" -#include "util/session.h" #include "util/symbol.h" #include "util/thread_map.h" @@ -457,7 +456,6 @@ static int test__basic_mmap(void) int err = -1; event_t *event; struct thread_map *threads; - struct perf_session session; struct cpu_map *cpus; struct perf_evlist *evlist; struct perf_event_attr attr = { @@ -521,13 +519,6 @@ static int test__basic_mmap(void) attr.wakeup_events = 1; attr.sample_period = 1; - /* - * FIXME: use evsel->attr.sample_type in event__parse_sample. - * This will nicely remove the requirement that we have - * all the events with the same sample_type. - */ - session.sample_type = attr.sample_type; - for (i = 0; i < nsyscalls; ++i) { attr.config = ids[i]; evsels[i] = perf_evsel__new(&attr, i); @@ -567,7 +558,7 @@ static int test__basic_mmap(void) goto out_munmap; } - event__parse_sample(event, &session, &sample); + event__parse_sample(event, attr.sample_type, false, &sample); evsel = perf_evlist__id2evsel(evlist, sample.id); if (evsel == NULL) { pr_debug("event with id %" PRIu64 diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index d0b16d905405..ce2e50c891c7 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -1106,7 +1106,7 @@ static void perf_session__mmap_read_cpu(struct perf_session *self, int cpu) event_t *event; while ((event = perf_evlist__read_on_cpu(evsel_list, cpu)) != NULL) { - event__parse_sample(event, self, &sample); + perf_session__parse_sample(self, event, &sample); if (event->header.type == PERF_RECORD_SAMPLE) event__process_sample(event, &sample, self); diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 1478ab4ee222..e4db8b888546 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -826,128 +826,3 @@ out_filtered: al->filtered = true; return 0; } - -static int event__parse_id_sample(const event_t *event, - struct perf_session *session, - struct sample_data *sample) -{ - const u64 *array; - u64 type; - - sample->cpu = sample->pid = sample->tid = -1; - sample->stream_id = sample->id = sample->time = -1ULL; - - if (!session->sample_id_all) - return 0; - - array = event->sample.array; - array += ((event->header.size - - sizeof(event->header)) / sizeof(u64)) - 1; - type = session->sample_type; - - if (type & PERF_SAMPLE_CPU) { - u32 *p = (u32 *)array; - sample->cpu = *p; - array--; - } - - if (type & PERF_SAMPLE_STREAM_ID) { - sample->stream_id = *array; - array--; - } - - if (type & PERF_SAMPLE_ID) { - sample->id = *array; - array--; - } - - if (type & PERF_SAMPLE_TIME) { - sample->time = *array; - array--; - } - - if (type & PERF_SAMPLE_TID) { - u32 *p = (u32 *)array; - sample->pid = p[0]; - sample->tid = p[1]; - } - - return 0; -} - -int event__parse_sample(const event_t *event, struct perf_session *session, - struct sample_data *data) -{ - const u64 *array; - u64 type; - - if (event->header.type != PERF_RECORD_SAMPLE) - return event__parse_id_sample(event, session, data); - - array = event->sample.array; - type = session->sample_type; - - if (type & PERF_SAMPLE_IP) { - data->ip = event->ip.ip; - array++; - } - - if (type & PERF_SAMPLE_TID) { - u32 *p = (u32 *)array; - data->pid = p[0]; - data->tid = p[1]; - array++; - } - - if (type & PERF_SAMPLE_TIME) { - data->time = *array; - array++; - } - - if (type & PERF_SAMPLE_ADDR) { - data->addr = *array; - array++; - } - - data->id = -1ULL; - if (type & PERF_SAMPLE_ID) { - data->id = *array; - array++; - } - - if (type & PERF_SAMPLE_STREAM_ID) { - data->stream_id = *array; - array++; - } - - if (type & PERF_SAMPLE_CPU) { - u32 *p = (u32 *)array; - data->cpu = *p; - array++; - } else - data->cpu = -1; - - if (type & PERF_SAMPLE_PERIOD) { - data->period = *array; - array++; - } - - if (type & PERF_SAMPLE_READ) { - pr_debug("PERF_SAMPLE_READ is unsuported for now\n"); - return -1; - } - - if (type & PERF_SAMPLE_CALLCHAIN) { - data->callchain = (struct ip_callchain *)array; - array += 1 + data->callchain->nr; - } - - if (type & PERF_SAMPLE_RAW) { - u32 *p = (u32 *)array; - data->raw_size = *p; - p++; - data->raw_data = p; - } - - return 0; -} diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 2b7e91902f10..d79e4edd82f9 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -169,9 +169,10 @@ struct addr_location; int event__preprocess_sample(const event_t *self, struct perf_session *session, struct addr_location *al, struct sample_data *data, symbol_filter_t filter); -int event__parse_sample(const event_t *event, struct perf_session *session, - struct sample_data *sample); const char *event__get_event_name(unsigned int id); +int event__parse_sample(const event_t *event, u64 type, bool sample_id_all, + struct sample_data *sample); + #endif /* __PERF_RECORD_H */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 9a6d94299ab8..a85ae12845ea 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -355,3 +355,121 @@ out_unmap: } return -1; } + +static int event__parse_id_sample(const event_t *event, u64 type, + struct sample_data *sample) +{ + const u64 *array = event->sample.array; + + array += ((event->header.size - + sizeof(event->header)) / sizeof(u64)) - 1; + + if (type & PERF_SAMPLE_CPU) { + u32 *p = (u32 *)array; + sample->cpu = *p; + array--; + } + + if (type & PERF_SAMPLE_STREAM_ID) { + sample->stream_id = *array; + array--; + } + + if (type & PERF_SAMPLE_ID) { + sample->id = *array; + array--; + } + + if (type & PERF_SAMPLE_TIME) { + sample->time = *array; + array--; + } + + if (type & PERF_SAMPLE_TID) { + u32 *p = (u32 *)array; + sample->pid = p[0]; + sample->tid = p[1]; + } + + return 0; +} + +int event__parse_sample(const event_t *event, u64 type, bool sample_id_all, + struct sample_data *data) +{ + const u64 *array; + + data->cpu = data->pid = data->tid = -1; + data->stream_id = data->id = data->time = -1ULL; + + if (event->header.type != PERF_RECORD_SAMPLE) { + if (!sample_id_all) + return 0; + return event__parse_id_sample(event, type, data); + } + + array = event->sample.array; + + if (type & PERF_SAMPLE_IP) { + data->ip = event->ip.ip; + array++; + } + + if (type & PERF_SAMPLE_TID) { + u32 *p = (u32 *)array; + data->pid = p[0]; + data->tid = p[1]; + array++; + } + + if (type & PERF_SAMPLE_TIME) { + data->time = *array; + array++; + } + + if (type & PERF_SAMPLE_ADDR) { + data->addr = *array; + array++; + } + + data->id = -1ULL; + if (type & PERF_SAMPLE_ID) { + data->id = *array; + array++; + } + + if (type & PERF_SAMPLE_STREAM_ID) { + data->stream_id = *array; + array++; + } + + if (type & PERF_SAMPLE_CPU) { + u32 *p = (u32 *)array; + data->cpu = *p; + array++; + } + + if (type & PERF_SAMPLE_PERIOD) { + data->period = *array; + array++; + } + + if (type & PERF_SAMPLE_READ) { + fprintf(stderr, "PERF_SAMPLE_READ is unsuported for now\n"); + return -1; + } + + if (type & PERF_SAMPLE_CALLCHAIN) { + data->callchain = (struct ip_callchain *)array; + array += 1 + data->callchain->nr; + } + + if (type & PERF_SAMPLE_RAW) { + u32 *p = (u32 *)array; + data->raw_size = *p; + p++; + data->raw_data = p; + } + + return 0; +} diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index b58a48a5e5a9..e6a07408669e 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -496,7 +496,7 @@ static void flush_sample_queue(struct perf_session *s, if (iter->timestamp > limit) break; - event__parse_sample(iter->event, s, &sample); + perf_session__parse_sample(s, iter->event, &sample); perf_session_deliver_event(s, iter->event, &sample, ops, iter->file_offset); @@ -806,7 +806,7 @@ static int perf_session__process_event(struct perf_session *session, /* * For all kernel events we get the sample data */ - event__parse_sample(event, session, &sample); + perf_session__parse_sample(session, event, &sample); /* Preprocess sample records - precheck callchains */ if (perf_session__preprocess_sample(session, event, &sample)) diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index e815468eb888..78239767011e 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -155,4 +155,13 @@ size_t perf_session__fprintf_nr_events(struct perf_session *self, FILE *fp) { return hists__fprintf_nr_events(&self->hists, fp); } + +static inline int perf_session__parse_sample(struct perf_session *session, + const event_t *event, + struct sample_data *sample) +{ + return event__parse_sample(event, session->sample_type, + session->sample_id_all, sample); +} + #endif /* __PERF_SESSION_H */ -- cgit v1.2.3 From ef1d1af28ca37fdbc2745da040529cd2953c1af5 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 18 Jan 2011 21:41:45 -0200 Subject: perf evsel: Introduce perf_evsel__{in,ex}it Out of the {con,des}structor, as in interpreted language bindings we will need to go back from the wrapper object to the real thing. In that case using container_of will save us to have an extra pointer in the perf_evsel struct. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 29 ++++++++++++++++++++--------- tools/perf/util/evlist.h | 2 ++ tools/perf/util/evsel.c | 22 ++++++++++++++++------ tools/perf/util/evsel.h | 3 +++ 4 files changed, 41 insertions(+), 15 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 4b3b84cd71a1..df0610e9c61b 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -6,17 +6,21 @@ #include #include +void perf_evlist__init(struct perf_evlist *evlist) +{ + int i; + + for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i) + INIT_HLIST_HEAD(&evlist->heads[i]); + INIT_LIST_HEAD(&evlist->entries); +} + struct perf_evlist *perf_evlist__new(void) { struct perf_evlist *evlist = zalloc(sizeof(*evlist)); - if (evlist != NULL) { - int i; - - for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i) - INIT_HLIST_HEAD(&evlist->heads[i]); - INIT_LIST_HEAD(&evlist->entries); - } + if (evlist != NULL) + perf_evlist__init(evlist); return evlist; } @@ -33,11 +37,18 @@ static void perf_evlist__purge(struct perf_evlist *evlist) evlist->nr_entries = 0; } -void perf_evlist__delete(struct perf_evlist *evlist) +void perf_evlist__exit(struct perf_evlist *evlist) { - perf_evlist__purge(evlist); free(evlist->mmap); free(evlist->pollfd); + evlist->mmap = NULL; + evlist->pollfd = NULL; +} + +void perf_evlist__delete(struct perf_evlist *evlist) +{ + perf_evlist__purge(evlist); + perf_evlist__exit(evlist); free(evlist); } diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 28712063db97..acbe48eac608 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -24,6 +24,8 @@ struct perf_evlist { struct perf_evsel; struct perf_evlist *perf_evlist__new(void); +void perf_evlist__init(struct perf_evlist *evlist); +void perf_evlist__exit(struct perf_evlist *evlist); void perf_evlist__delete(struct perf_evlist *evlist); void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index a85ae12845ea..76ab553637d6 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -14,15 +14,20 @@ #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) #define SID(e, x, y) xyarray__entry(e->id, x, y) +void perf_evsel__init(struct perf_evsel *evsel, + struct perf_event_attr *attr, int idx) +{ + evsel->idx = idx; + evsel->attr = *attr; + INIT_LIST_HEAD(&evsel->node); +} + 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 = *attr; - INIT_LIST_HEAD(&evsel->node); - } + if (evsel != NULL) + perf_evsel__init(evsel, attr, idx); return evsel; } @@ -87,11 +92,16 @@ int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus) return evlist->mmap != NULL ? 0 : -ENOMEM; } -void perf_evsel__delete(struct perf_evsel *evsel) +void perf_evsel__exit(struct perf_evsel *evsel) { assert(list_empty(&evsel->node)); xyarray__delete(evsel->fd); xyarray__delete(evsel->id); +} + +void perf_evsel__delete(struct perf_evsel *evsel) +{ + perf_evsel__exit(evsel); free(evsel); } diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 667ee4e2e35e..7962e7587dea 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -52,6 +52,9 @@ struct thread_map; struct perf_evlist; struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx); +void perf_evsel__init(struct perf_evsel *evsel, + struct perf_event_attr *attr, int idx); +void perf_evsel__exit(struct perf_evsel *evsel); void perf_evsel__delete(struct perf_evsel *evsel); int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads); -- cgit v1.2.3 From 68baa431ec2f14ba7510d4e79bceb6ceaf0d3b74 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 20 Jan 2011 23:15:30 +0900 Subject: perf tools: Add strfilter for general purpose string filter Add strfilter for general purpose string filter. Every filter rules are descrived by glob matching pattern and '!' prefix which means Logical NOT. A strfilter consists of those filter rules connected with '&' and '|'. A set of rules can be folded by using '(' and ')'. It also accepts spaces around rules and those operators. Format: ::= | "!" | | "(" ")" ::= "&" | "|" e.g.: "(add* | del*) & *timer" filter rules pass strings which start with add or del and end with timer. This will be used by perf probe --filter. Changes in V2: - Fix to check result of strdup() and strfilter__alloc(). - Encapsulate and simplify interfaces as like regex(3). Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Franck Bui-Huu Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Steven Rostedt LKML-Reference: <20110120141530.25915.12673.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Makefile | 2 + tools/perf/util/strfilter.c | 200 ++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/strfilter.h | 48 +++++++++++ 3 files changed, 250 insertions(+) create mode 100644 tools/perf/util/strfilter.c create mode 100644 tools/perf/util/strfilter.h (limited to 'tools/perf/util') diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 638e8e146bb9..eedcf9541572 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -417,6 +417,7 @@ LIB_H += util/help.h LIB_H += util/session.h LIB_H += util/strbuf.h LIB_H += util/strlist.h +LIB_H += util/strfilter.h LIB_H += util/svghelper.h LIB_H += util/run-command.h LIB_H += util/sigchain.h @@ -458,6 +459,7 @@ LIB_OBJS += $(OUTPUT)util/quote.o LIB_OBJS += $(OUTPUT)util/strbuf.o LIB_OBJS += $(OUTPUT)util/string.o LIB_OBJS += $(OUTPUT)util/strlist.o +LIB_OBJS += $(OUTPUT)util/strfilter.o LIB_OBJS += $(OUTPUT)util/usage.o LIB_OBJS += $(OUTPUT)util/wrapper.o LIB_OBJS += $(OUTPUT)util/sigchain.o diff --git a/tools/perf/util/strfilter.c b/tools/perf/util/strfilter.c new file mode 100644 index 000000000000..4064b7d708b8 --- /dev/null +++ b/tools/perf/util/strfilter.c @@ -0,0 +1,200 @@ +#include +#include "util.h" +#include "string.h" +#include "strfilter.h" + +/* Operators */ +static const char *OP_and = "&"; /* Logical AND */ +static const char *OP_or = "|"; /* Logical OR */ +static const char *OP_not = "!"; /* Logical NOT */ + +#define is_operator(c) ((c) == '|' || (c) == '&' || (c) == '!') +#define is_separator(c) (is_operator(c) || (c) == '(' || (c) == ')') + +static void strfilter_node__delete(struct strfilter_node *self) +{ + if (self) { + if (self->p && !is_operator(*self->p)) + free((char *)self->p); + strfilter_node__delete(self->l); + strfilter_node__delete(self->r); + free(self); + } +} + +void strfilter__delete(struct strfilter *self) +{ + if (self) { + strfilter_node__delete(self->root); + free(self); + } +} + +static const char *get_token(const char *s, const char **e) +{ + const char *p; + + while (isspace(*s)) /* Skip spaces */ + s++; + + if (*s == '\0') { + p = s; + goto end; + } + + p = s + 1; + if (!is_separator(*s)) { + /* End search */ +retry: + while (*p && !is_separator(*p) && !isspace(*p)) + p++; + /* Escape and special case: '!' is also used in glob pattern */ + if (*(p - 1) == '\\' || (*p == '!' && *(p - 1) == '[')) { + p++; + goto retry; + } + } +end: + *e = p; + return s; +} + +static struct strfilter_node *strfilter_node__alloc(const char *op, + struct strfilter_node *l, + struct strfilter_node *r) +{ + struct strfilter_node *ret = zalloc(sizeof(struct strfilter_node)); + + if (ret) { + ret->p = op; + ret->l = l; + ret->r = r; + } + + return ret; +} + +static struct strfilter_node *strfilter_node__new(const char *s, + const char **ep) +{ + struct strfilter_node root, *cur, *last_op; + const char *e; + + if (!s) + return NULL; + + memset(&root, 0, sizeof(root)); + last_op = cur = &root; + + s = get_token(s, &e); + while (*s != '\0' && *s != ')') { + switch (*s) { + case '&': /* Exchg last OP->r with AND */ + if (!cur->r || !last_op->r) + goto error; + cur = strfilter_node__alloc(OP_and, last_op->r, NULL); + if (!cur) + goto nomem; + last_op->r = cur; + last_op = cur; + break; + case '|': /* Exchg the root with OR */ + if (!cur->r || !root.r) + goto error; + cur = strfilter_node__alloc(OP_or, root.r, NULL); + if (!cur) + goto nomem; + root.r = cur; + last_op = cur; + break; + case '!': /* Add NOT as a leaf node */ + if (cur->r) + goto error; + cur->r = strfilter_node__alloc(OP_not, NULL, NULL); + if (!cur->r) + goto nomem; + cur = cur->r; + break; + case '(': /* Recursively parses inside the parenthesis */ + if (cur->r) + goto error; + cur->r = strfilter_node__new(s + 1, &s); + if (!s) + goto nomem; + if (!cur->r || *s != ')') + goto error; + e = s + 1; + break; + default: + if (cur->r) + goto error; + cur->r = strfilter_node__alloc(NULL, NULL, NULL); + if (!cur->r) + goto nomem; + cur->r->p = strndup(s, e - s); + if (!cur->r->p) + goto nomem; + } + s = get_token(e, &e); + } + if (!cur->r) + goto error; + *ep = s; + return root.r; +nomem: + s = NULL; +error: + *ep = s; + strfilter_node__delete(root.r); + return NULL; +} + +/* + * Parse filter rule and return new strfilter. + * Return NULL if fail, and *ep == NULL if memory allocation failed. + */ +struct strfilter *strfilter__new(const char *rules, const char **err) +{ + struct strfilter *ret = zalloc(sizeof(struct strfilter)); + const char *ep = NULL; + + if (ret) + ret->root = strfilter_node__new(rules, &ep); + + if (!ret || !ret->root || *ep != '\0') { + if (err) + *err = ep; + strfilter__delete(ret); + ret = NULL; + } + + return ret; +} + +static bool strfilter_node__compare(struct strfilter_node *self, + const char *str) +{ + if (!self || !self->p) + return false; + + switch (*self->p) { + case '|': /* OR */ + return strfilter_node__compare(self->l, str) || + strfilter_node__compare(self->r, str); + case '&': /* AND */ + return strfilter_node__compare(self->l, str) && + strfilter_node__compare(self->r, str); + case '!': /* NOT */ + return !strfilter_node__compare(self->r, str); + default: + return strglobmatch(str, self->p); + } +} + +/* Return true if STR matches the filter rules */ +bool strfilter__compare(struct strfilter *self, const char *str) +{ + if (!self) + return false; + return strfilter_node__compare(self->root, str); +} diff --git a/tools/perf/util/strfilter.h b/tools/perf/util/strfilter.h new file mode 100644 index 000000000000..00f58a7506de --- /dev/null +++ b/tools/perf/util/strfilter.h @@ -0,0 +1,48 @@ +#ifndef __PERF_STRFILTER_H +#define __PERF_STRFILTER_H +/* General purpose glob matching filter */ + +#include +#include + +/* A node of string filter */ +struct strfilter_node { + struct strfilter_node *l; /* Tree left branche (for &,|) */ + struct strfilter_node *r; /* Tree right branche (for !,&,|) */ + const char *p; /* Operator or rule */ +}; + +/* String filter */ +struct strfilter { + struct strfilter_node *root; +}; + +/** + * strfilter__new - Create a new string filter + * @rules: Filter rule, which is a combination of glob expressions. + * @err: Pointer which points an error detected on @rules + * + * Parse @rules and return new strfilter. Return NULL if an error detected. + * In that case, *@err will indicate where it is detected, and *@err is NULL + * if a memory allocation is failed. + */ +struct strfilter *strfilter__new(const char *rules, const char **err); + +/** + * strfilter__compare - compare given string and a string filter + * @self: String filter + * @str: target string + * + * Compare @str and @self. Return true if the str match the rule + */ +bool strfilter__compare(struct strfilter *self, const char *str); + +/** + * strfilter__delete - delete a string filter + * @self: String filter to delete + * + * Delete @self. + */ +void strfilter__delete(struct strfilter *self); + +#endif -- cgit v1.2.3 From bd09d7b5efeb13965b6725b4a3e9944908bca9d2 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 20 Jan 2011 23:15:39 +0900 Subject: perf probe: Add variable filter support Add filters support for available variable list. Default filter is "!__k???tab_*&!__crc_*" for filtering out automatically generated symbols. The format of filter rule is "[!]GLOBPATTERN", so you can use wild cards. If the filter rule starts with '!', matched variables are filter out. e.g.: # perf probe -V schedule --externs --filter=cpu* Available variables at schedule @ cpumask_var_t cpu_callout_mask cpumask_var_t cpu_core_map cpumask_var_t cpu_isolated_map cpumask_var_t cpu_sibling_map int cpu_number long unsigned int* cpu_bit_bitmap ... Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Chase Douglas Cc: Franck Bui-Huu Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Steven Rostedt LKML-Reference: <20110120141539.25915.43401.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu [ committer note: Removed the elf.h include as it was fixed up in e80711c] Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Documentation/perf-probe.txt | 14 +++++++ tools/perf/builtin-probe.c | 34 +++++++++++++++++ tools/perf/util/probe-event.c | 66 ++++++++++++++++++++------------- tools/perf/util/probe-event.h | 3 +- 4 files changed, 90 insertions(+), 27 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/Documentation/perf-probe.txt b/tools/perf/Documentation/perf-probe.txt index fcc51fe0195c..32fb18f1695d 100644 --- a/tools/perf/Documentation/perf-probe.txt +++ b/tools/perf/Documentation/perf-probe.txt @@ -77,6 +77,12 @@ OPTIONS --funcs:: Show available functions in given module or kernel. +--filter=FILTER:: + (Only for --vars) Set filter for variables. FILTER is a combination of + glob pattern, see FILTER PATTERN for details. + Default FILTER is "!__k???tab_* & !__crc_*". + If several filters are specified, only the last filter is valid. + -f:: --force:: Forcibly add events with existing name. @@ -139,6 +145,14 @@ e.g. This provides some sort of flexibility and robustness to probe point definitions against minor code changes. For example, actual 10th line of schedule() can be moved easily by modifying schedule(), but the same line matching 'rq=cpu_rq*' may still exist in the function.) +FILTER PATTERN +-------------- + The filter pattern is a glob matching pattern(s) to filter variables. + In addition, you can use "!" for specifying filter-out rule. You also can give several rules combined with "&" or "|", and fold those rules as one rule by using "(" ")". + +e.g. + With --filter "foo* | bar*", perf probe -V shows variables which start with "foo" or "bar". + With --filter "!foo* & *bar", perf probe -V shows variables which don't start with "foo" and end with "bar", like "fizzbar". But "foobar" is filtered out. EXAMPLES -------- diff --git a/tools/perf/builtin-probe.c b/tools/perf/builtin-probe.c index 6cf708aba7c9..abb423e164c8 100644 --- a/tools/perf/builtin-probe.c +++ b/tools/perf/builtin-probe.c @@ -36,6 +36,7 @@ #include "builtin.h" #include "util/util.h" #include "util/strlist.h" +#include "util/strfilter.h" #include "util/symbol.h" #include "util/debug.h" #include "util/debugfs.h" @@ -43,6 +44,7 @@ #include "util/probe-finder.h" #include "util/probe-event.h" +#define DEFAULT_VAR_FILTER "!__k???tab_* & !__crc_*" #define MAX_PATH_LEN 256 /* Session management structure */ @@ -60,6 +62,7 @@ static struct { struct line_range line_range; const char *target_module; int max_probe_points; + struct strfilter *filter; } params; /* Parse an event definition. Note that any error must die. */ @@ -156,6 +159,27 @@ static int opt_show_vars(const struct option *opt __used, return ret; } + +static int opt_set_filter(const struct option *opt __used, + const char *str, int unset __used) +{ + const char *err; + + if (str) { + pr_debug2("Set filter: %s\n", str); + if (params.filter) + strfilter__delete(params.filter); + params.filter = strfilter__new(str, &err); + if (!params.filter) { + pr_err("Filter parse error at %ld.\n", err - str + 1); + pr_err("Source: \"%s\"\n", str); + pr_err(" %*c\n", (int)(err - str + 1), '^'); + return -EINVAL; + } + } + + return 0; +} #endif static const char * const probe_usage[] = { @@ -212,6 +236,10 @@ static const struct option options[] = { "Show accessible variables on PROBEDEF", opt_show_vars), OPT_BOOLEAN('\0', "externs", ¶ms.show_ext_vars, "Show external variables too (with --vars only)"), + OPT_CALLBACK('\0', "filter", NULL, + "[!]FILTER", "Set a variable filter (with --vars only)\n" + "\t\t\t(default: \"" DEFAULT_VAR_FILTER "\")", + opt_set_filter), OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, "file", "vmlinux pathname"), OPT_STRING('s', "source", &symbol_conf.source_prefix, @@ -324,10 +352,16 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) " --add/--del.\n"); usage_with_options(probe_usage, options); } + if (!params.filter) + params.filter = strfilter__new(DEFAULT_VAR_FILTER, + NULL); + ret = show_available_vars(params.events, params.nevents, params.max_probe_points, params.target_module, + params.filter, params.show_ext_vars); + strfilter__delete(params.filter); if (ret < 0) pr_err(" Error: Failed to show vars. (%d)\n", ret); return ret; diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 859d377a3df3..077e0518f0f7 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -451,12 +451,14 @@ end: } static int show_available_vars_at(int fd, struct perf_probe_event *pev, - int max_vls, bool externs) + int max_vls, struct strfilter *_filter, + bool externs) { char *buf; - int ret, i; + int ret, i, nvars; struct str_node *node; struct variable_list *vls = NULL, *vl; + const char *var; buf = synthesize_perf_probe_point(&pev->point); if (!buf) @@ -464,36 +466,45 @@ static int show_available_vars_at(int fd, struct perf_probe_event *pev, pr_debug("Searching variables at %s\n", buf); ret = find_available_vars_at(fd, pev, &vls, max_vls, externs); - if (ret > 0) { - /* Some variables were found */ - fprintf(stdout, "Available variables at %s\n", buf); - for (i = 0; i < ret; i++) { - vl = &vls[i]; - /* - * A probe point might be converted to - * several trace points. - */ - fprintf(stdout, "\t@<%s+%lu>\n", vl->point.symbol, - vl->point.offset); - free(vl->point.symbol); - if (vl->vars) { - strlist__for_each(node, vl->vars) + if (ret <= 0) { + pr_err("Failed to find variables at %s (%d)\n", buf, ret); + goto end; + } + /* Some variables are found */ + fprintf(stdout, "Available variables at %s\n", buf); + for (i = 0; i < ret; i++) { + vl = &vls[i]; + /* + * A probe point might be converted to + * several trace points. + */ + fprintf(stdout, "\t@<%s+%lu>\n", vl->point.symbol, + vl->point.offset); + free(vl->point.symbol); + nvars = 0; + if (vl->vars) { + strlist__for_each(node, vl->vars) { + var = strchr(node->s, '\t') + 1; + if (strfilter__compare(_filter, var)) { fprintf(stdout, "\t\t%s\n", node->s); - strlist__delete(vl->vars); - } else - fprintf(stdout, "(No variables)\n"); + nvars++; + } + } + strlist__delete(vl->vars); } - free(vls); - } else - pr_err("Failed to find variables at %s (%d)\n", buf, ret); - + if (nvars == 0) + fprintf(stdout, "\t\t(No matched variables)\n"); + } + free(vls); +end: free(buf); return ret; } /* Show available variables on given probe point */ int show_available_vars(struct perf_probe_event *pevs, int npevs, - int max_vls, const char *module, bool externs) + int max_vls, const char *module, + struct strfilter *_filter, bool externs) { int i, fd, ret = 0; @@ -510,7 +521,8 @@ int show_available_vars(struct perf_probe_event *pevs, int npevs, setup_pager(); for (i = 0; i < npevs && ret >= 0; i++) - ret = show_available_vars_at(fd, &pevs[i], max_vls, externs); + ret = show_available_vars_at(fd, &pevs[i], max_vls, _filter, + externs); close(fd); return ret; @@ -556,7 +568,9 @@ int show_line_range(struct line_range *lr __unused, const char *module __unused) int show_available_vars(struct perf_probe_event *pevs __unused, int npevs __unused, int max_vls __unused, - const char *module __unused, bool externs __unused) + const char *module __unused, + struct strfilter *filter __unused, + bool externs __unused) { pr_warning("Debuginfo-analysis is not supported.\n"); return -ENOSYS; diff --git a/tools/perf/util/probe-event.h b/tools/perf/util/probe-event.h index 1fb4f18337d3..4e80b2bbc516 100644 --- a/tools/perf/util/probe-event.h +++ b/tools/perf/util/probe-event.h @@ -3,6 +3,7 @@ #include #include "strlist.h" +#include "strfilter.h" extern bool probe_event_dry_run; @@ -126,7 +127,7 @@ extern int show_perf_probe_events(void); extern int show_line_range(struct line_range *lr, const char *module); extern int show_available_vars(struct perf_probe_event *pevs, int npevs, int max_probe_points, const char *module, - bool externs); + struct strfilter *filter, bool externs); extern int show_available_funcs(const char *module); -- cgit v1.2.3 From 3c42258c9a4db70133fa6946a275b62a16792bb5 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 20 Jan 2011 23:15:45 +0900 Subject: perf probe: Add filters support for available functions Add filters support for available function list. Default filter is "!_*" for filtering out local-purpose symbols. e.g.: # perf probe --filter="add*" -F add_disk add_disk_randomness add_input_randomness add_interrupt_randomness add_memory add_page_to_unevictable_list add_page_wait_queue ... Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Chase Douglas Cc: Franck Bui-Huu Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Steven Rostedt LKML-Reference: <20110120141545.25915.85930.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Documentation/perf-probe.txt | 9 +++++---- tools/perf/builtin-probe.c | 19 +++++++++++++------ tools/perf/util/probe-event.c | 23 +++++++++++++---------- tools/perf/util/probe-event.h | 2 +- 4 files changed, 32 insertions(+), 21 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/Documentation/perf-probe.txt b/tools/perf/Documentation/perf-probe.txt index 32fb18f1695d..81c3220e04f3 100644 --- a/tools/perf/Documentation/perf-probe.txt +++ b/tools/perf/Documentation/perf-probe.txt @@ -78,10 +78,11 @@ OPTIONS Show available functions in given module or kernel. --filter=FILTER:: - (Only for --vars) Set filter for variables. FILTER is a combination of - glob pattern, see FILTER PATTERN for details. - Default FILTER is "!__k???tab_* & !__crc_*". - If several filters are specified, only the last filter is valid. + (Only for --vars and --funcs) Set filter. FILTER is a combination of glob + pattern, see FILTER PATTERN for detail. + Default FILTER is "!__k???tab_* & !__crc_*" for --vars, and "!_*" + for --funcs. + If several filters are specified, only the last filter is used. -f:: --force:: diff --git a/tools/perf/builtin-probe.c b/tools/perf/builtin-probe.c index abb423e164c8..fcde0031085f 100644 --- a/tools/perf/builtin-probe.c +++ b/tools/perf/builtin-probe.c @@ -45,6 +45,7 @@ #include "util/probe-event.h" #define DEFAULT_VAR_FILTER "!__k???tab_* & !__crc_*" +#define DEFAULT_FUNC_FILTER "!_*" #define MAX_PATH_LEN 256 /* Session management structure */ @@ -159,6 +160,7 @@ static int opt_show_vars(const struct option *opt __used, return ret; } +#endif static int opt_set_filter(const struct option *opt __used, const char *str, int unset __used) @@ -180,7 +182,6 @@ static int opt_set_filter(const struct option *opt __used, return 0; } -#endif static const char * const probe_usage[] = { "perf probe [] 'PROBEDEF' ['PROBEDEF' ...]", @@ -236,10 +237,6 @@ static const struct option options[] = { "Show accessible variables on PROBEDEF", opt_show_vars), OPT_BOOLEAN('\0', "externs", ¶ms.show_ext_vars, "Show external variables too (with --vars only)"), - OPT_CALLBACK('\0', "filter", NULL, - "[!]FILTER", "Set a variable filter (with --vars only)\n" - "\t\t\t(default: \"" DEFAULT_VAR_FILTER "\")", - opt_set_filter), OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, "file", "vmlinux pathname"), OPT_STRING('s', "source", &symbol_conf.source_prefix, @@ -252,6 +249,11 @@ static const struct option options[] = { "Set how many probe points can be found for a probe."), OPT_BOOLEAN('F', "funcs", ¶ms.show_funcs, "Show potential probe-able functions."), + OPT_CALLBACK('\0', "filter", NULL, + "[!]FILTER", "Set a filter (with --vars/funcs only)\n" + "\t\t\t(default: \"" DEFAULT_VAR_FILTER "\" for --vars,\n" + "\t\t\t \"" DEFAULT_FUNC_FILTER "\" for --funcs)", + opt_set_filter), OPT_END() }; @@ -322,7 +324,12 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) pr_err(" Error: Don't use --funcs with --vars.\n"); usage_with_options(probe_usage, options); } - ret = show_available_funcs(params.target_module); + if (!params.filter) + params.filter = strfilter__new(DEFAULT_FUNC_FILTER, + NULL); + ret = show_available_funcs(params.target_module, + params.filter); + strfilter__delete(params.filter); if (ret < 0) pr_err(" Error: Failed to show functions." " (%d)\n", ret); diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 077e0518f0f7..9d237e3cff5d 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -1951,21 +1951,23 @@ int del_perf_probe_events(struct strlist *dellist) return ret; } +/* TODO: don't use a global variable for filter ... */ +static struct strfilter *available_func_filter; /* - * If a symbol corresponds to a function with global binding return 0. - * For all others return 1. + * If a symbol corresponds to a function with global binding and + * matches filter return 0. For all others return 1. */ -static int filter_non_global_functions(struct map *map __unused, - struct symbol *sym) +static int filter_available_functions(struct map *map __unused, + struct symbol *sym) { - if (sym->binding != STB_GLOBAL) - return 1; - - return 0; + if (sym->binding == STB_GLOBAL && + strfilter__compare(available_func_filter, sym->name)) + return 0; + return 1; } -int show_available_funcs(const char *module) +int show_available_funcs(const char *module, struct strfilter *_filter) { struct map *map; int ret; @@ -1981,7 +1983,8 @@ int show_available_funcs(const char *module) pr_err("Failed to find %s map.\n", (module) ? : "kernel"); return -EINVAL; } - if (map__load(map, filter_non_global_functions)) { + available_func_filter = _filter; + if (map__load(map, filter_available_functions)) { pr_err("Failed to load map.\n"); return -EINVAL; } diff --git a/tools/perf/util/probe-event.h b/tools/perf/util/probe-event.h index 4e80b2bbc516..3434fc9d79d5 100644 --- a/tools/perf/util/probe-event.h +++ b/tools/perf/util/probe-event.h @@ -128,7 +128,7 @@ extern int show_line_range(struct line_range *lr, const char *module); extern int show_available_vars(struct perf_probe_event *pevs, int npevs, int max_probe_points, const char *module, struct strfilter *filter, bool externs); -extern int show_available_funcs(const char *module); +extern int show_available_funcs(const char *module, struct strfilter *filter); /* Maximum index number of event-name postfix */ -- cgit v1.2.3 From ef2bf6d043ac9bd4a6f38d862af407154a4754d9 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 29 Jan 2011 09:04:40 -0200 Subject: perf events: Account PERF_RECORD_LOST events in event__process Right now this function is only used by perf top, that uses PROT_READ only, i.e. overwrite mode, so no PERF_RECORD_LOST events are generated, but don't forget those events. The patch that moved this out of perf top was made so that this routine could be used by 'perf probe' in the uprobes patchset, so perhaps there they need to check for LOST events and warn the user, as will be done in the following patches that will switch 'perf top' to non overwrite mode (mmap with PROT_READ|PROT_WRITE). Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/event.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index e4db8b888546..b9d4ac87f9f7 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -646,6 +646,8 @@ int event__process(event_t *event, struct sample_data *sample, case PERF_RECORD_EXIT: event__process_task(event, sample, session); break; + case PERF_RECORD_LOST: + event__process_lost(event, sample, session); default: break; } -- cgit v1.2.3 From 7bb41152b9be7e31f10d8919bce5034135525d9d Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 29 Jan 2011 09:08:13 -0200 Subject: perf evlist: Support non overwrite mode in perf_evlist__read_on_cpu I.e. stash the overwrite mode in struct perf_evlist and act accordingly in perf_evlist__read_on_cpu, not checking for overwrites and touching the tail after consuming one event, like perf record does, for instance. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 35 ++++++++++++++++++++--------------- tools/perf/util/evlist.h | 1 + tools/perf/util/evsel.c | 1 + 3 files changed, 22 insertions(+), 15 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index df0610e9c61b..b498eecbe850 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -116,24 +116,25 @@ event_t *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) unsigned int old = md->prev; unsigned char *data = md->base + page_size; event_t *event = NULL; - int diff; - - /* - * If we're further behind than half the buffer, there's a chance - * the writer will bite our tail and mess up the samples under us. - * - * If we somehow ended up ahead of the head, we got messed up. - * - * In either case, truncate and restart at head. - */ - diff = head - old; - if (diff > md->mask / 2 || diff < 0) { - fprintf(stderr, "WARNING: failed to keep up with mmap data.\n"); + if (evlist->overwrite) { /* - * head points to a known good entry, start there. + * If we're further behind than half the buffer, there's a chance + * the writer will bite our tail and mess up the samples under us. + * + * If we somehow ended up ahead of the head, we got messed up. + * + * In either case, truncate and restart at head. */ - old = head; + int diff = head - old; + if (diff > md->mask / 2 || diff < 0) { + fprintf(stderr, "WARNING: failed to keep up with mmap data.\n"); + + /* + * head points to a known good entry, start there. + */ + old = head; + } } if (old != head) { @@ -166,5 +167,9 @@ event_t *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) } md->prev = old; + + if (!evlist->overwrite) + perf_mmap__write_tail(md, old); + return event; } diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index acbe48eac608..2706ae40c8b3 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -16,6 +16,7 @@ struct perf_evlist { int nr_entries; int nr_fds; int mmap_len; + bool overwrite; event_t event_copy; struct perf_mmap *mmap; struct pollfd *pollfd; diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 76ab553637d6..f98b3e585039 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -327,6 +327,7 @@ int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, perf_evlist__alloc_pollfd(evlist, cpus->nr, threads->nr) < 0) return -ENOMEM; + evlist->overwrite = overwrite; evlist->mmap_len = (pages + 1) * page_size; first_evsel = list_entry(evlist->entries.next, struct perf_evsel, node); -- cgit v1.2.3 From 8d50e5b4171a69cf48ca94a1e7c14033d0b4771d Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 29 Jan 2011 13:02:00 -0200 Subject: perf tools: Rename 'struct sample_data' to 'struct perf_sample' Making the namespace more uniform. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-annotate.c | 2 +- tools/perf/builtin-diff.c | 2 +- tools/perf/builtin-inject.c | 8 ++++---- tools/perf/builtin-kmem.c | 2 +- tools/perf/builtin-lock.c | 2 +- tools/perf/builtin-record.c | 2 +- tools/perf/builtin-report.c | 36 ++++++++++++++++++------------------ tools/perf/builtin-sched.c | 2 +- tools/perf/builtin-script.c | 2 +- tools/perf/builtin-test.c | 2 +- tools/perf/builtin-timechart.c | 8 ++++---- tools/perf/builtin-top.c | 4 ++-- tools/perf/util/build-id.c | 4 ++-- tools/perf/util/event.c | 16 ++++++++-------- tools/perf/util/event.h | 18 +++++++++--------- tools/perf/util/evsel.c | 4 ++-- tools/perf/util/session.c | 26 +++++++++++++------------- tools/perf/util/session.h | 4 ++-- 18 files changed, 72 insertions(+), 72 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index 8879463807e4..ef3675135414 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -58,7 +58,7 @@ static int hists__add_entry(struct hists *self, struct addr_location *al) return hist_entry__inc_addr_samples(he, al->addr); } -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(event_t *event, struct perf_sample *sample, struct perf_session *session) { struct addr_location al; diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c index 3153e492dbcc..0822149dc768 100644 --- a/tools/perf/builtin-diff.c +++ b/tools/perf/builtin-diff.c @@ -31,7 +31,7 @@ static int hists__add_entry(struct hists *self, } static int diff__process_sample_event(event_t *event, - struct sample_data *sample, + struct perf_sample *sample, struct perf_session *session) { struct addr_location al; diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c index 0c78ffa7bf67..4c9388ce878c 100644 --- a/tools/perf/builtin-inject.c +++ b/tools/perf/builtin-inject.c @@ -36,13 +36,13 @@ static int event__repipe_synth(event_t *event, return 0; } -static int event__repipe(event_t *event, struct sample_data *sample __used, +static int event__repipe(event_t *event, struct perf_sample *sample __used, struct perf_session *session) { return event__repipe_synth(event, session); } -static int event__repipe_mmap(event_t *self, struct sample_data *sample, +static int event__repipe_mmap(event_t *self, struct perf_sample *sample, struct perf_session *session) { int err; @@ -53,7 +53,7 @@ static int event__repipe_mmap(event_t *self, struct sample_data *sample, return err; } -static int event__repipe_task(event_t *self, struct sample_data *sample, +static int event__repipe_task(event_t *self, struct perf_sample *sample, struct perf_session *session) { int err; @@ -119,7 +119,7 @@ static int dso__inject_build_id(struct dso *self, struct perf_session *session) return 0; } -static int event__inject_buildid(event_t *event, struct sample_data *sample, +static int event__inject_buildid(event_t *event, struct perf_sample *sample, struct perf_session *session) { struct addr_location al; diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c index d97256d65980..3c1cdcf29902 100644 --- a/tools/perf/builtin-kmem.c +++ b/tools/perf/builtin-kmem.c @@ -304,7 +304,7 @@ process_raw_event(event_t *raw_event __used, void *data, } } -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(event_t *event, struct perf_sample *sample, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, event->ip.pid); diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c index 2b36defc5d73..c3f512791344 100644 --- a/tools/perf/builtin-lock.c +++ b/tools/perf/builtin-lock.c @@ -834,7 +834,7 @@ static void dump_info(void) die("Unknown type of information\n"); } -static int process_sample_event(event_t *self, struct sample_data *sample, +static int process_sample_event(event_t *self, struct perf_sample *sample, struct perf_session *s) { struct thread *thread = perf_session__findnew(s, sample->tid); diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index caf927978d92..5d3e4b32072b 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -101,7 +101,7 @@ static void write_output(void *buf, size_t size) } static int process_synthesized_event(event_t *event, - struct sample_data *sample __used, + struct perf_sample *sample __used, struct perf_session *self __used) { write_output(event, event->header.size); diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index f6a43493d1d0..bbbadcc04097 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -77,9 +77,9 @@ static struct hists *perf_session__hists_findnew(struct perf_session *self, return new; } -static int perf_session__add_hist_entry(struct perf_session *self, +static int perf_session__add_hist_entry(struct perf_session *session, struct addr_location *al, - struct sample_data *data) + struct perf_sample *sample) { struct symbol *parent = NULL; int err = 0; @@ -87,28 +87,28 @@ static int perf_session__add_hist_entry(struct perf_session *self, struct hists *hists; struct perf_event_attr *attr; - if ((sort__has_parent || symbol_conf.use_callchain) && data->callchain) { - err = perf_session__resolve_callchain(self, al->thread, - data->callchain, &parent); + if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) { + err = perf_session__resolve_callchain(session, al->thread, + sample->callchain, &parent); if (err) return err; } - attr = perf_header__find_attr(data->id, &self->header); + attr = perf_header__find_attr(sample->id, &session->header); if (attr) - hists = perf_session__hists_findnew(self, data->id, attr->type, attr->config); + hists = perf_session__hists_findnew(session, sample->id, attr->type, attr->config); else - hists = perf_session__hists_findnew(self, data->id, 0, 0); + hists = perf_session__hists_findnew(session, sample->id, 0, 0); if (hists == NULL) return -ENOMEM; - he = __hists__add_entry(hists, al, parent, data->period); + he = __hists__add_entry(hists, al, parent, sample->period); if (he == NULL) return -ENOMEM; if (symbol_conf.use_callchain) { - err = callchain_append(he->callchain, &self->callchain_cursor, - data->period); + err = callchain_append(he->callchain, &session->callchain_cursor, + sample->period); if (err) return err; } @@ -124,32 +124,32 @@ static int perf_session__add_hist_entry(struct perf_session *self, } static int add_event_total(struct perf_session *session, - struct sample_data *data, + struct perf_sample *sample, struct perf_event_attr *attr) { struct hists *hists; if (attr) - hists = perf_session__hists_findnew(session, data->id, + hists = perf_session__hists_findnew(session, sample->id, attr->type, attr->config); else - hists = perf_session__hists_findnew(session, data->id, 0, 0); + hists = perf_session__hists_findnew(session, sample->id, 0, 0); if (!hists) return -ENOMEM; - hists->stats.total_period += data->period; + hists->stats.total_period += sample->period; /* * FIXME: add_event_total should be moved from here to * perf_session__process_event so that the proper hist is passed to * the event_op methods. */ hists__inc_nr_events(hists, PERF_RECORD_SAMPLE); - session->hists.stats.total_period += data->period; + session->hists.stats.total_period += sample->period; return 0; } -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(event_t *event, struct perf_sample *sample, struct perf_session *session) { struct addr_location al; @@ -179,7 +179,7 @@ static int process_sample_event(event_t *event, struct sample_data *sample, return 0; } -static int process_read_event(event_t *event, struct sample_data *sample __used, +static int process_read_event(event_t *event, struct perf_sample *sample __used, struct perf_session *session __used) { struct perf_event_attr *attr; diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c index 29acb894e035..ff993c8b175d 100644 --- a/tools/perf/builtin-sched.c +++ b/tools/perf/builtin-sched.c @@ -1607,7 +1607,7 @@ process_raw_event(event_t *raw_event __used, struct perf_session *session, process_sched_migrate_task_event(data, session, event, cpu, timestamp, thread); } -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(event_t *event, struct perf_sample *sample, struct perf_session *session) { struct thread *thread; diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c index b766c2a9ac97..5c4c809008af 100644 --- a/tools/perf/builtin-script.c +++ b/tools/perf/builtin-script.c @@ -63,7 +63,7 @@ static int cleanup_scripting(void) static char const *input_name = "perf.data"; -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(event_t *event, struct perf_sample *sample, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, event->ip.pid); diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index 738830d298e8..df62433a34a7 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -550,7 +550,7 @@ static int test__basic_mmap(void) } while ((event = perf_evlist__read_on_cpu(evlist, 0)) != NULL) { - struct sample_data sample; + struct perf_sample sample; if (event->header.type != PERF_RECORD_SAMPLE) { pr_debug("unexpected %s event\n", diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c index 746cf03cb05d..01cf0c3771a6 100644 --- a/tools/perf/builtin-timechart.c +++ b/tools/perf/builtin-timechart.c @@ -276,21 +276,21 @@ static int cpus_cstate_state[MAX_CPUS]; static u64 cpus_pstate_start_times[MAX_CPUS]; static u64 cpus_pstate_state[MAX_CPUS]; -static int process_comm_event(event_t *event, struct sample_data *sample __used, +static int process_comm_event(event_t *event, struct perf_sample *sample __used, struct perf_session *session __used) { pid_set_comm(event->comm.tid, event->comm.comm); return 0; } -static int process_fork_event(event_t *event, struct sample_data *sample __used, +static int process_fork_event(event_t *event, struct perf_sample *sample __used, struct perf_session *session __used) { pid_fork(event->fork.pid, event->fork.ppid, event->fork.time); return 0; } -static int process_exit_event(event_t *event, struct sample_data *sample __used, +static int process_exit_event(event_t *event, struct perf_sample *sample __used, struct perf_session *session __used) { pid_exit(event->fork.pid, event->fork.time); @@ -487,7 +487,7 @@ static void sched_switch(int cpu, u64 timestamp, struct trace_entry *te) static int process_sample_event(event_t *event __used, - struct sample_data *sample, + struct perf_sample *sample, struct perf_session *session) { struct trace_entry *te; diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 7f92ab7696f7..d923127b41b6 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -997,7 +997,7 @@ static int symbol_filter(struct map *map, struct symbol *sym) } static void event__process_sample(const event_t *self, - struct sample_data *sample, + struct perf_sample *sample, struct perf_session *session) { u64 ip = self->ip.ip; @@ -1107,7 +1107,7 @@ static void event__process_sample(const event_t *self, static void perf_session__mmap_read_cpu(struct perf_session *self, int cpu) { - struct sample_data sample; + struct perf_sample sample; event_t *event; while ((event = perf_evlist__read_on_cpu(evsel_list, cpu)) != NULL) { diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c index deffb8c96071..b184a7fa0843 100644 --- a/tools/perf/util/build-id.c +++ b/tools/perf/util/build-id.c @@ -15,7 +15,7 @@ #include "debug.h" static int build_id__mark_dso_hit(event_t *event, - struct sample_data *sample __used, + struct perf_sample *sample __used, struct perf_session *session) { struct addr_location al; @@ -37,7 +37,7 @@ static int build_id__mark_dso_hit(event_t *event, return 0; } -static int event__exit_del_thread(event_t *self, struct sample_data *sample __used, +static int event__exit_del_thread(event_t *self, struct perf_sample *sample __used, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, self->fork.tid); diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index b9d4ac87f9f7..5c886fbd50ce 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -34,7 +34,7 @@ const char *event__get_event_name(unsigned int id) return event__name[id]; } -static struct sample_data synth_sample = { +static struct perf_sample synth_sample = { .pid = -1, .tid = -1, .time = -1, @@ -440,7 +440,7 @@ static int thread__set_comm_adjust(struct thread *self, const char *comm, return 0; } -int event__process_comm(event_t *self, struct sample_data *sample __used, +int event__process_comm(event_t *self, struct perf_sample *sample __used, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, self->comm.tid); @@ -456,7 +456,7 @@ int event__process_comm(event_t *self, struct sample_data *sample __used, return 0; } -int event__process_lost(event_t *self, struct sample_data *sample __used, +int event__process_lost(event_t *self, struct perf_sample *sample __used, struct perf_session *session) { dump_printf(": id:%" PRIu64 ": lost:%" PRIu64 "\n", @@ -567,7 +567,7 @@ out_problem: return -1; } -int event__process_mmap(event_t *self, struct sample_data *sample __used, +int event__process_mmap(event_t *self, struct perf_sample *sample __used, struct perf_session *session) { struct machine *machine; @@ -609,7 +609,7 @@ out_problem: return 0; } -int event__process_task(event_t *self, struct sample_data *sample __used, +int event__process_task(event_t *self, struct perf_sample *sample __used, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, self->fork.tid); @@ -632,7 +632,7 @@ int event__process_task(event_t *self, struct sample_data *sample __used, return 0; } -int event__process(event_t *event, struct sample_data *sample, +int event__process(event_t *event, struct perf_sample *sample, struct perf_session *session) { switch (event->header.type) { @@ -757,7 +757,7 @@ static void dso__calc_col_width(struct dso *self, struct hists *hists) } int event__preprocess_sample(const event_t *self, struct perf_session *session, - struct addr_location *al, struct sample_data *data, + struct addr_location *al, struct perf_sample *sample, symbol_filter_t filter) { u8 cpumode = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; @@ -788,7 +788,7 @@ int event__preprocess_sample(const event_t *self, struct perf_session *session, al->map ? al->map->dso->long_name : al->level == 'H' ? "[hypervisor]" : ""); al->sym = NULL; - al->cpu = data->cpu; + al->cpu = sample->cpu; if (al->map) { if (symbol_conf.dso_list && diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index d79e4edd82f9..84fd71ffd619 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -61,7 +61,7 @@ struct sample_event { u64 array[]; }; -struct sample_data { +struct perf_sample { u64 ip; u32 pid, tid; u64 time; @@ -138,7 +138,7 @@ struct perf_session; typedef int (*event__handler_synth_t)(event_t *event, struct perf_session *session); -typedef int (*event__handler_t)(event_t *event, struct sample_data *sample, +typedef int (*event__handler_t)(event_t *event, struct perf_sample *sample, struct perf_session *session); int event__synthesize_thread(pid_t pid, event__handler_t process, @@ -154,25 +154,25 @@ int event__synthesize_modules(event__handler_t process, struct perf_session *session, struct machine *machine); -int event__process_comm(event_t *self, struct sample_data *sample, +int event__process_comm(event_t *event, struct perf_sample *sample, struct perf_session *session); -int event__process_lost(event_t *self, struct sample_data *sample, +int event__process_lost(event_t *event, struct perf_sample *sample, struct perf_session *session); -int event__process_mmap(event_t *self, struct sample_data *sample, +int event__process_mmap(event_t *event, struct perf_sample *sample, struct perf_session *session); -int event__process_task(event_t *self, struct sample_data *sample, +int event__process_task(event_t *event, struct perf_sample *sample, struct perf_session *session); -int event__process(event_t *event, struct sample_data *sample, +int event__process(event_t *event, struct perf_sample *sample, struct perf_session *session); struct addr_location; int event__preprocess_sample(const event_t *self, struct perf_session *session, - struct addr_location *al, struct sample_data *data, + struct addr_location *al, struct perf_sample *sample, symbol_filter_t filter); const char *event__get_event_name(unsigned int id); int event__parse_sample(const event_t *event, u64 type, bool sample_id_all, - struct sample_data *sample); + struct perf_sample *sample); #endif /* __PERF_RECORD_H */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index f98b3e585039..a13488511880 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -368,7 +368,7 @@ out_unmap: } static int event__parse_id_sample(const event_t *event, u64 type, - struct sample_data *sample) + struct perf_sample *sample) { const u64 *array = event->sample.array; @@ -406,7 +406,7 @@ static int event__parse_id_sample(const event_t *event, u64 type, } int event__parse_sample(const event_t *event, u64 type, bool sample_id_all, - struct sample_data *data) + struct perf_sample *data) { const u64 *array; diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index e6a07408669e..ee0b61102571 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -67,7 +67,7 @@ out_close: static void perf_session__id_header_size(struct perf_session *session) { - struct sample_data *data; + struct perf_sample *data; u64 sample_type = session->sample_type; u16 size = 0; @@ -299,7 +299,7 @@ static int process_event_synth_stub(event_t *event __used, } static int process_event_stub(event_t *event __used, - struct sample_data *sample __used, + struct perf_sample *sample __used, struct perf_session *session __used) { dump_printf(": unhandled!\n"); @@ -475,7 +475,7 @@ static void perf_session_free_sample_buffers(struct perf_session *session) static int perf_session_deliver_event(struct perf_session *session, event_t *event, - struct sample_data *sample, + struct perf_sample *sample, struct perf_event_ops *ops, u64 file_offset); @@ -485,7 +485,7 @@ static void flush_sample_queue(struct perf_session *s, struct ordered_samples *os = &s->ordered_samples; struct list_head *head = &os->samples; struct sample_queue *tmp, *iter; - struct sample_data sample; + struct perf_sample sample; u64 limit = os->next_flush; u64 last_ts = os->last_sample ? os->last_sample->timestamp : 0ULL; @@ -610,11 +610,11 @@ static void __queue_event(struct sample_queue *new, struct perf_session *s) #define MAX_SAMPLE_BUFFER (64 * 1024 / sizeof(struct sample_queue)) static int perf_session_queue_event(struct perf_session *s, event_t *event, - struct sample_data *data, u64 file_offset) + struct perf_sample *sample, u64 file_offset) { struct ordered_samples *os = &s->ordered_samples; struct list_head *sc = &os->sample_cache; - u64 timestamp = data->time; + u64 timestamp = sample->time; struct sample_queue *new; if (!timestamp || timestamp == ~0ULL) @@ -650,7 +650,7 @@ static int perf_session_queue_event(struct perf_session *s, event_t *event, return 0; } -static void callchain__printf(struct sample_data *sample) +static void callchain__printf(struct perf_sample *sample) { unsigned int i; @@ -663,7 +663,7 @@ static void callchain__printf(struct sample_data *sample) static void perf_session__print_tstamp(struct perf_session *session, event_t *event, - struct sample_data *sample) + struct perf_sample *sample) { if (event->header.type != PERF_RECORD_SAMPLE && !session->sample_id_all) { @@ -679,7 +679,7 @@ static void perf_session__print_tstamp(struct perf_session *session, } static void dump_event(struct perf_session *session, event_t *event, - u64 file_offset, struct sample_data *sample) + u64 file_offset, struct perf_sample *sample) { if (!dump_trace) return; @@ -697,7 +697,7 @@ static void dump_event(struct perf_session *session, event_t *event, } static void dump_sample(struct perf_session *session, event_t *event, - struct sample_data *sample) + struct perf_sample *sample) { if (!dump_trace) return; @@ -712,7 +712,7 @@ static void dump_sample(struct perf_session *session, event_t *event, static int perf_session_deliver_event(struct perf_session *session, event_t *event, - struct sample_data *sample, + struct perf_sample *sample, struct perf_event_ops *ops, u64 file_offset) { @@ -745,7 +745,7 @@ static int perf_session_deliver_event(struct perf_session *session, } static int perf_session__preprocess_sample(struct perf_session *session, - event_t *event, struct sample_data *sample) + event_t *event, struct perf_sample *sample) { if (event->header.type != PERF_RECORD_SAMPLE || !(session->sample_type & PERF_SAMPLE_CALLCHAIN)) @@ -789,7 +789,7 @@ static int perf_session__process_event(struct perf_session *session, struct perf_event_ops *ops, u64 file_offset) { - struct sample_data sample; + struct perf_sample sample; int ret; if (session->header.needs_swap && event__swap_ops[event->header.type]) diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 78239767011e..365bf533a396 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -57,7 +57,7 @@ struct perf_session { struct perf_event_ops; -typedef int (*event_op)(event_t *self, struct sample_data *sample, +typedef int (*event_op)(event_t *self, struct perf_sample *sample, struct perf_session *session); typedef int (*event_synth_op)(event_t *self, struct perf_session *session); typedef int (*event_op2)(event_t *self, struct perf_session *session, @@ -158,7 +158,7 @@ size_t perf_session__fprintf_nr_events(struct perf_session *self, FILE *fp) static inline int perf_session__parse_sample(struct perf_session *session, const event_t *event, - struct sample_data *sample) + struct perf_sample *sample) { return event__parse_sample(event, session->sample_type, session->sample_id_all, sample); -- cgit v1.2.3 From 8115d60c323dd9931b95221c0a392aeddc1d6ef3 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 29 Jan 2011 14:01:45 -0200 Subject: perf tools: Kill event_t typedef, use 'union perf_event' instead And move the event_t methods to the perf_event__ too. No code changes, just namespace consistency. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-annotate.c | 11 +-- tools/perf/builtin-diff.c | 14 +-- tools/perf/builtin-inject.c | 82 +++++++++--------- tools/perf/builtin-kmem.c | 10 +-- tools/perf/builtin-lock.c | 6 +- tools/perf/builtin-record.c | 56 ++++++------ tools/perf/builtin-report.c | 30 ++++--- tools/perf/builtin-sched.c | 15 ++-- tools/perf/builtin-script.c | 17 ++-- tools/perf/builtin-test.c | 6 +- tools/perf/builtin-timechart.c | 11 ++- tools/perf/builtin-top.c | 33 ++++---- tools/perf/util/build-id.c | 19 +++-- tools/perf/util/callchain.c | 3 +- tools/perf/util/callchain.h | 4 +- tools/perf/util/debug.c | 2 +- tools/perf/util/debug.h | 2 +- tools/perf/util/event.c | 187 +++++++++++++++++++++-------------------- tools/perf/util/event.h | 69 +++++++-------- tools/perf/util/evlist.c | 6 +- tools/perf/util/evlist.h | 4 +- tools/perf/util/evsel.c | 10 +-- tools/perf/util/header.c | 83 +++++++++--------- tools/perf/util/header.h | 50 +++++------ tools/perf/util/hist.c | 2 +- tools/perf/util/session.c | 169 +++++++++++++++++++------------------ tools/perf/util/session.h | 13 +-- 27 files changed, 475 insertions(+), 439 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index ef3675135414..70067862e07f 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -58,12 +58,13 @@ static int hists__add_entry(struct hists *self, struct addr_location *al) return hist_entry__inc_addr_samples(he, al->addr); } -static int process_sample_event(event_t *event, struct perf_sample *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct addr_location al; - if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + if (perf_event__preprocess_sample(event, session, &al, sample, NULL) < 0) { pr_warning("problem processing %d event, skipping it.\n", event->header.type); return -1; @@ -372,9 +373,9 @@ find_next: static struct perf_event_ops event_ops = { .sample = process_sample_event, - .mmap = event__process_mmap, - .comm = event__process_comm, - .fork = event__process_task, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .fork = perf_event__process_task, .ordered_samples = true, .ordering_requires_timestamps = true, }; diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c index 0822149dc768..6b7d91160ecb 100644 --- a/tools/perf/builtin-diff.c +++ b/tools/perf/builtin-diff.c @@ -30,13 +30,13 @@ static int hists__add_entry(struct hists *self, return -ENOMEM; } -static int diff__process_sample_event(event_t *event, +static int diff__process_sample_event(union perf_event *event, struct perf_sample *sample, struct perf_session *session) { struct addr_location al; - if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + if (perf_event__preprocess_sample(event, session, &al, sample, NULL) < 0) { pr_warning("problem processing %d event, skipping it.\n", event->header.type); return -1; @@ -56,11 +56,11 @@ static int diff__process_sample_event(event_t *event, static struct perf_event_ops event_ops = { .sample = diff__process_sample_event, - .mmap = event__process_mmap, - .comm = event__process_comm, - .exit = event__process_task, - .fork = event__process_task, - .lost = event__process_lost, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .exit = perf_event__process_task, + .fork = perf_event__process_task, + .lost = perf_event__process_lost, .ordered_samples = true, .ordering_requires_timestamps = true, }; diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c index 4c9388ce878c..e29f04ed3396 100644 --- a/tools/perf/builtin-inject.c +++ b/tools/perf/builtin-inject.c @@ -16,8 +16,8 @@ static char const *input_name = "-"; static bool inject_build_ids; -static int event__repipe_synth(event_t *event, - struct perf_session *session __used) +static int perf_event__repipe_synth(union perf_event *event, + struct perf_session *session __used) { uint32_t size; void *buf = event; @@ -36,41 +36,44 @@ static int event__repipe_synth(event_t *event, return 0; } -static int event__repipe(event_t *event, struct perf_sample *sample __used, - struct perf_session *session) +static int perf_event__repipe(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { - return event__repipe_synth(event, session); + return perf_event__repipe_synth(event, session); } -static int event__repipe_mmap(event_t *self, struct perf_sample *sample, - struct perf_session *session) +static int perf_event__repipe_mmap(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) { int err; - err = event__process_mmap(self, sample, session); - event__repipe(self, sample, session); + err = perf_event__process_mmap(event, sample, session); + perf_event__repipe(event, sample, session); return err; } -static int event__repipe_task(event_t *self, struct perf_sample *sample, - struct perf_session *session) +static int perf_event__repipe_task(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) { int err; - err = event__process_task(self, sample, session); - event__repipe(self, sample, session); + err = perf_event__process_task(event, sample, session); + perf_event__repipe(event, sample, session); return err; } -static int event__repipe_tracing_data(event_t *self, - struct perf_session *session) +static int perf_event__repipe_tracing_data(union perf_event *event, + struct perf_session *session) { int err; - event__repipe_synth(self, session); - err = event__process_tracing_data(self, session); + perf_event__repipe_synth(event, session); + err = perf_event__process_tracing_data(event, session); return err; } @@ -109,8 +112,8 @@ static int dso__inject_build_id(struct dso *self, struct perf_session *session) if (self->kernel) misc = PERF_RECORD_MISC_KERNEL; - err = event__synthesize_build_id(self, misc, event__repipe, - machine, session); + err = perf_event__synthesize_build_id(self, misc, perf_event__repipe, + machine, session); if (err) { pr_err("Can't synthesize build_id event for %s\n", self->long_name); return -1; @@ -119,8 +122,9 @@ static int dso__inject_build_id(struct dso *self, struct perf_session *session) return 0; } -static int event__inject_buildid(event_t *event, struct perf_sample *sample, - struct perf_session *session) +static int perf_event__inject_buildid(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) { struct addr_location al; struct thread *thread; @@ -155,24 +159,24 @@ static int event__inject_buildid(event_t *event, struct perf_sample *sample, } repipe: - event__repipe(event, sample, session); + perf_event__repipe(event, sample, session); return 0; } struct perf_event_ops inject_ops = { - .sample = event__repipe, - .mmap = event__repipe, - .comm = event__repipe, - .fork = event__repipe, - .exit = event__repipe, - .lost = event__repipe, - .read = event__repipe, - .throttle = event__repipe, - .unthrottle = event__repipe, - .attr = event__repipe_synth, - .event_type = event__repipe_synth, - .tracing_data = event__repipe_synth, - .build_id = event__repipe_synth, + .sample = perf_event__repipe, + .mmap = perf_event__repipe, + .comm = perf_event__repipe, + .fork = perf_event__repipe, + .exit = perf_event__repipe, + .lost = perf_event__repipe, + .read = perf_event__repipe, + .throttle = perf_event__repipe, + .unthrottle = perf_event__repipe, + .attr = perf_event__repipe_synth, + .event_type = perf_event__repipe_synth, + .tracing_data = perf_event__repipe_synth, + .build_id = perf_event__repipe_synth, }; extern volatile int session_done; @@ -190,10 +194,10 @@ static int __cmd_inject(void) signal(SIGINT, sig_handler); if (inject_build_ids) { - inject_ops.sample = event__inject_buildid; - inject_ops.mmap = event__repipe_mmap; - inject_ops.fork = event__repipe_task; - inject_ops.tracing_data = event__repipe_tracing_data; + inject_ops.sample = perf_event__inject_buildid; + inject_ops.mmap = perf_event__repipe_mmap; + inject_ops.fork = perf_event__repipe_task; + inject_ops.tracing_data = perf_event__repipe_tracing_data; } session = perf_session__new(input_name, O_RDONLY, false, true, &inject_ops); diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c index 3c1cdcf29902..7f618f4e7b79 100644 --- a/tools/perf/builtin-kmem.c +++ b/tools/perf/builtin-kmem.c @@ -275,9 +275,8 @@ static void process_free_event(void *data, s_alloc->alloc_cpu = -1; } -static void -process_raw_event(event_t *raw_event __used, void *data, - int cpu, u64 timestamp, struct thread *thread) +static void process_raw_event(union perf_event *raw_event __used, void *data, + int cpu, u64 timestamp, struct thread *thread) { struct event *event; int type; @@ -304,7 +303,8 @@ process_raw_event(event_t *raw_event __used, void *data, } } -static int process_sample_event(event_t *event, struct perf_sample *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, event->ip.pid); @@ -325,7 +325,7 @@ static int process_sample_event(event_t *event, struct perf_sample *sample, static struct perf_event_ops event_ops = { .sample = process_sample_event, - .comm = event__process_comm, + .comm = perf_event__process_comm, .ordered_samples = true, }; diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c index c3f512791344..e00d93847c44 100644 --- a/tools/perf/builtin-lock.c +++ b/tools/perf/builtin-lock.c @@ -834,14 +834,14 @@ static void dump_info(void) die("Unknown type of information\n"); } -static int process_sample_event(event_t *self, struct perf_sample *sample, +static int process_sample_event(union perf_event *event, struct perf_sample *sample, struct perf_session *s) { struct thread *thread = perf_session__findnew(s, sample->tid); if (thread == NULL) { pr_debug("problem processing %d event, skipping it.\n", - self->header.type); + event->header.type); return -1; } @@ -852,7 +852,7 @@ static int process_sample_event(event_t *self, struct perf_sample *sample, static struct perf_event_ops eops = { .sample = process_sample_event, - .comm = event__process_comm, + .comm = perf_event__process_comm, .ordered_samples = true, }; diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 5d3e4b32072b..edc3555098c8 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -100,7 +100,7 @@ static void write_output(void *buf, size_t size) } } -static int process_synthesized_event(event_t *event, +static int process_synthesized_event(union perf_event *event, struct perf_sample *sample __used, struct perf_session *self __used) { @@ -404,7 +404,7 @@ static void atexit_header(void) } } -static void event__synthesize_guest_os(struct machine *machine, void *data) +static void perf_event__synthesize_guest_os(struct machine *machine, void *data) { int err; struct perf_session *psession = data; @@ -420,8 +420,8 @@ static void event__synthesize_guest_os(struct machine *machine, void *data) *method is used to avoid symbol missing when the first addr is *in module instead of in guest kernel. */ - err = event__synthesize_modules(process_synthesized_event, - psession, machine); + err = perf_event__synthesize_modules(process_synthesized_event, + psession, machine); if (err < 0) pr_err("Couldn't record guest kernel [%d]'s reference" " relocation symbol.\n", machine->pid); @@ -430,11 +430,12 @@ static void event__synthesize_guest_os(struct machine *machine, void *data) * We use _stext for guest kernel because guest kernel's /proc/kallsyms * have no _text sometimes. */ - err = event__synthesize_kernel_mmap(process_synthesized_event, - psession, machine, "_text"); + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + psession, machine, "_text"); if (err < 0) - err = event__synthesize_kernel_mmap(process_synthesized_event, - psession, machine, "_stext"); + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + psession, machine, + "_stext"); if (err < 0) pr_err("Couldn't record guest kernel [%d]'s reference" " relocation symbol.\n", machine->pid); @@ -617,16 +618,16 @@ static int __cmd_record(int argc, const char **argv) perf_session__set_sample_id_all(session, sample_id_all_avail); if (pipe_output) { - err = event__synthesize_attrs(&session->header, - process_synthesized_event, - session); + err = perf_event__synthesize_attrs(&session->header, + process_synthesized_event, + session); if (err < 0) { pr_err("Couldn't synthesize attrs.\n"); return err; } - err = event__synthesize_event_types(process_synthesized_event, - session); + err = perf_event__synthesize_event_types(process_synthesized_event, + session); if (err < 0) { pr_err("Couldn't synthesize event_types.\n"); return err; @@ -641,9 +642,9 @@ static int __cmd_record(int argc, const char **argv) * return this more properly and also * propagate errors that now are calling die() */ - err = event__synthesize_tracing_data(output, evsel_list, - process_synthesized_event, - session); + err = perf_event__synthesize_tracing_data(output, evsel_list, + process_synthesized_event, + session); if (err <= 0) { pr_err("Couldn't record tracing data.\n"); return err; @@ -658,31 +659,34 @@ static int __cmd_record(int argc, const char **argv) return -1; } - err = event__synthesize_kernel_mmap(process_synthesized_event, - session, machine, "_text"); + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + session, machine, "_text"); if (err < 0) - err = event__synthesize_kernel_mmap(process_synthesized_event, - session, machine, "_stext"); + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + session, machine, "_stext"); if (err < 0) pr_err("Couldn't record kernel reference relocation symbol\n" "Symbol resolution may be skewed if relocation was used (e.g. kexec).\n" "Check /proc/kallsyms permission or run as root.\n"); - err = event__synthesize_modules(process_synthesized_event, - session, machine); + err = perf_event__synthesize_modules(process_synthesized_event, + session, machine); if (err < 0) pr_err("Couldn't record kernel module information.\n" "Symbol resolution may be skewed if relocation was used (e.g. kexec).\n" "Check /proc/modules permission or run as root.\n"); if (perf_guest) - perf_session__process_machines(session, event__synthesize_guest_os); + perf_session__process_machines(session, + perf_event__synthesize_guest_os); if (!system_wide) - event__synthesize_thread(target_tid, process_synthesized_event, - session); + perf_event__synthesize_thread(target_tid, + process_synthesized_event, + session); else - event__synthesize_threads(process_synthesized_event, session); + perf_event__synthesize_threads(process_synthesized_event, + session); if (realtime_prio) { struct sched_param param; diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index bbbadcc04097..a6a4e5457b6f 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -149,13 +149,14 @@ static int add_event_total(struct perf_session *session, return 0; } -static int process_sample_event(event_t *event, struct perf_sample *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct addr_location al; struct perf_event_attr *attr; - if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + if (perf_event__preprocess_sample(event, session, &al, sample, NULL) < 0) { fprintf(stderr, "problem processing %d event, skipping it.\n", event->header.type); return -1; @@ -179,7 +180,8 @@ static int process_sample_event(event_t *event, struct perf_sample *sample, return 0; } -static int process_read_event(event_t *event, struct perf_sample *sample __used, +static int process_read_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session __used) { struct perf_event_attr *attr; @@ -232,17 +234,17 @@ static int perf_session__setup_sample_type(struct perf_session *self) } static struct perf_event_ops event_ops = { - .sample = process_sample_event, - .mmap = event__process_mmap, - .comm = event__process_comm, - .exit = event__process_task, - .fork = event__process_task, - .lost = event__process_lost, - .read = process_read_event, - .attr = event__process_attr, - .event_type = event__process_event_type, - .tracing_data = event__process_tracing_data, - .build_id = event__process_build_id, + .sample = process_sample_event, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .exit = perf_event__process_task, + .fork = perf_event__process_task, + .lost = perf_event__process_lost, + .read = process_read_event, + .attr = perf_event__process_attr, + .event_type = perf_event__process_event_type, + .tracing_data = perf_event__process_tracing_data, + .build_id = perf_event__process_build_id, .ordered_samples = true, .ordering_requires_timestamps = true, }; diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c index ff993c8b175d..ae2621182927 100644 --- a/tools/perf/builtin-sched.c +++ b/tools/perf/builtin-sched.c @@ -1580,9 +1580,9 @@ process_sched_migrate_task_event(void *data, struct perf_session *session, event, cpu, timestamp, thread); } -static void -process_raw_event(event_t *raw_event __used, struct perf_session *session, - void *data, int cpu, u64 timestamp, struct thread *thread) +static void process_raw_event(union perf_event *raw_event __used, + struct perf_session *session, void *data, int cpu, + u64 timestamp, struct thread *thread) { struct event *event; int type; @@ -1607,7 +1607,8 @@ process_raw_event(event_t *raw_event __used, struct perf_session *session, process_sched_migrate_task_event(data, session, event, cpu, timestamp, thread); } -static int process_sample_event(event_t *event, struct perf_sample *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct thread *thread; @@ -1635,9 +1636,9 @@ static int process_sample_event(event_t *event, struct perf_sample *sample, static struct perf_event_ops event_ops = { .sample = process_sample_event, - .comm = event__process_comm, - .lost = event__process_lost, - .fork = event__process_task, + .comm = perf_event__process_comm, + .lost = perf_event__process_lost, + .fork = perf_event__process_task, .ordered_samples = true, }; diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c index 5c4c809008af..5f40df635dcb 100644 --- a/tools/perf/builtin-script.c +++ b/tools/perf/builtin-script.c @@ -63,7 +63,8 @@ static int cleanup_scripting(void) static char const *input_name = "perf.data"; -static int process_sample_event(event_t *event, struct perf_sample *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, event->ip.pid); @@ -100,14 +101,14 @@ static int process_sample_event(event_t *event, struct perf_sample *sample, } static struct perf_event_ops event_ops = { - .sample = process_sample_event, - .comm = event__process_comm, - .attr = event__process_attr, - .event_type = event__process_event_type, - .tracing_data = event__process_tracing_data, - .build_id = event__process_build_id, - .ordering_requires_timestamps = true, + .sample = process_sample_event, + .comm = perf_event__process_comm, + .attr = perf_event__process_attr, + .event_type = perf_event__process_event_type, + .tracing_data = perf_event__process_tracing_data, + .build_id = perf_event__process_build_id, .ordered_samples = true, + .ordering_requires_timestamps = true, }; extern volatile int session_done; diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index df62433a34a7..845b9bd54ed4 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -454,7 +454,7 @@ out_thread_map_delete: static int test__basic_mmap(void) { int err = -1; - event_t *event; + union perf_event *event; struct thread_map *threads; struct cpu_map *cpus; struct perf_evlist *evlist; @@ -554,11 +554,11 @@ static int test__basic_mmap(void) if (event->header.type != PERF_RECORD_SAMPLE) { pr_debug("unexpected %s event\n", - event__get_event_name(event->header.type)); + perf_event__name(event->header.type)); goto out_munmap; } - event__parse_sample(event, attr.sample_type, false, &sample); + perf_event__parse_sample(event, attr.sample_type, false, &sample); evsel = perf_evlist__id2evsel(evlist, sample.id); if (evsel == NULL) { pr_debug("event with id %" PRIu64 diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c index 01cf0c3771a6..0801275c500e 100644 --- a/tools/perf/builtin-timechart.c +++ b/tools/perf/builtin-timechart.c @@ -276,21 +276,24 @@ static int cpus_cstate_state[MAX_CPUS]; static u64 cpus_pstate_start_times[MAX_CPUS]; static u64 cpus_pstate_state[MAX_CPUS]; -static int process_comm_event(event_t *event, struct perf_sample *sample __used, +static int process_comm_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session __used) { pid_set_comm(event->comm.tid, event->comm.comm); return 0; } -static int process_fork_event(event_t *event, struct perf_sample *sample __used, +static int process_fork_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session __used) { pid_fork(event->fork.pid, event->fork.ppid, event->fork.time); return 0; } -static int process_exit_event(event_t *event, struct perf_sample *sample __used, +static int process_exit_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session __used) { pid_exit(event->fork.pid, event->fork.time); @@ -486,7 +489,7 @@ static void sched_switch(int cpu, u64 timestamp, struct trace_entry *te) } -static int process_sample_event(event_t *event __used, +static int process_sample_event(union perf_event *event __used, struct perf_sample *sample, struct perf_session *session) { diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index d923127b41b6..2f4d1f244be1 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -401,7 +401,7 @@ static void show_details(struct sym_entry *syme) } /* - * Symbols will be added here in event__process_sample and will get out + * Symbols will be added here in perf_event__process_sample and will get out * after decayed. */ static LIST_HEAD(active_symbols); @@ -996,15 +996,15 @@ static int symbol_filter(struct map *map, struct symbol *sym) return 0; } -static void event__process_sample(const event_t *self, - struct perf_sample *sample, - struct perf_session *session) +static void perf_event__process_sample(const union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) { - u64 ip = self->ip.ip; + u64 ip = event->ip.ip; struct sym_entry *syme; struct addr_location al; struct machine *machine; - u8 origin = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + u8 origin = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; ++samples; @@ -1023,7 +1023,7 @@ static void event__process_sample(const event_t *self, break; case PERF_RECORD_MISC_GUEST_KERNEL: ++guest_kernel_samples; - machine = perf_session__find_machine(session, self->ip.pid); + machine = perf_session__find_machine(session, event->ip.pid); break; case PERF_RECORD_MISC_GUEST_USER: ++guest_us_samples; @@ -1038,15 +1038,15 @@ static void event__process_sample(const event_t *self, if (!machine && perf_guest) { pr_err("Can't find guest [%d]'s kernel information\n", - self->ip.pid); + event->ip.pid); return; } - if (self->header.misc & PERF_RECORD_MISC_EXACT_IP) + if (event->header.misc & PERF_RECORD_MISC_EXACT_IP) exact_samples++; - if (event__preprocess_sample(self, session, &al, sample, - symbol_filter) < 0 || + if (perf_event__preprocess_sample(event, session, &al, sample, + symbol_filter) < 0 || al.filtered) return; @@ -1108,15 +1108,15 @@ static void event__process_sample(const event_t *self, static void perf_session__mmap_read_cpu(struct perf_session *self, int cpu) { struct perf_sample sample; - event_t *event; + union perf_event *event; while ((event = perf_evlist__read_on_cpu(evsel_list, cpu)) != NULL) { perf_session__parse_sample(self, event, &sample); if (event->header.type == PERF_RECORD_SAMPLE) - event__process_sample(event, &sample, self); + perf_event__process_sample(event, &sample, self); else - event__process(event, &sample, self); + perf_event__process(event, &sample, self); } } @@ -1199,9 +1199,10 @@ static int __cmd_top(void) return -ENOMEM; if (target_tid != -1) - event__synthesize_thread(target_tid, event__process, session); + perf_event__synthesize_thread(target_tid, perf_event__process, + session); else - event__synthesize_threads(event__process, session); + perf_event__synthesize_threads(perf_event__process, session); start_counters(evsel_list); first = list_entry(evsel_list->entries.next, struct perf_evsel, node); diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c index b184a7fa0843..31f934af9861 100644 --- a/tools/perf/util/build-id.c +++ b/tools/perf/util/build-id.c @@ -14,7 +14,7 @@ #include #include "debug.h" -static int build_id__mark_dso_hit(event_t *event, +static int build_id__mark_dso_hit(union perf_event *event, struct perf_sample *sample __used, struct perf_session *session) { @@ -37,13 +37,14 @@ static int build_id__mark_dso_hit(event_t *event, return 0; } -static int event__exit_del_thread(event_t *self, struct perf_sample *sample __used, - struct perf_session *session) +static int perf_event__exit_del_thread(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { - struct thread *thread = perf_session__findnew(session, self->fork.tid); + struct thread *thread = perf_session__findnew(session, event->fork.tid); - dump_printf("(%d:%d):(%d:%d)\n", self->fork.pid, self->fork.tid, - self->fork.ppid, self->fork.ptid); + dump_printf("(%d:%d):(%d:%d)\n", event->fork.pid, event->fork.tid, + event->fork.ppid, event->fork.ptid); if (thread) { rb_erase(&thread->rb_node, &session->threads); @@ -56,9 +57,9 @@ static int event__exit_del_thread(event_t *self, struct perf_sample *sample __us struct perf_event_ops build_id__mark_dso_hit_ops = { .sample = build_id__mark_dso_hit, - .mmap = event__process_mmap, - .fork = event__process_task, - .exit = event__exit_del_thread, + .mmap = perf_event__process_mmap, + .fork = perf_event__process_task, + .exit = perf_event__exit_del_thread, }; char *dso__build_id_filename(struct dso *self, char *bf, size_t size) diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index f8c66d1435e0..9f7106a8d9a4 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -18,7 +18,8 @@ #include "util.h" #include "callchain.h" -bool ip_callchain__valid(struct ip_callchain *chain, const event_t *event) +bool ip_callchain__valid(struct ip_callchain *chain, + const union perf_event *event) { unsigned int chain_size = event->header.size; chain_size -= (unsigned long)&event->ip.__more_data - (unsigned long)event; diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index 67137256a1cd..1a79df9f739f 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -95,8 +95,8 @@ int callchain_append(struct callchain_root *root, int callchain_merge(struct callchain_cursor *cursor, struct callchain_root *dst, struct callchain_root *src); -bool ip_callchain__valid(struct ip_callchain *chain, const event_t *event); - +bool ip_callchain__valid(struct ip_callchain *chain, + const union perf_event *event); /* * Initialize a cursor before adding entries inside, but keep * the previously allocated entries as a cache. diff --git a/tools/perf/util/debug.c b/tools/perf/util/debug.c index 01bbe8ecec3f..d4536a9e0d8c 100644 --- a/tools/perf/util/debug.c +++ b/tools/perf/util/debug.c @@ -57,7 +57,7 @@ void ui__warning(const char *format, ...) } #endif -void trace_event(event_t *event) +void trace_event(union perf_event *event) { unsigned char *raw_event = (void *)event; const char *color = PERF_COLOR_BLUE; diff --git a/tools/perf/util/debug.h b/tools/perf/util/debug.h index ca35fd66b5df..93516cf4682c 100644 --- a/tools/perf/util/debug.h +++ b/tools/perf/util/debug.h @@ -9,7 +9,7 @@ extern int verbose; extern bool quiet, dump_trace; int dump_printf(const char *fmt, ...) __attribute__((format(printf, 1, 2))); -void trace_event(event_t *event); +void trace_event(union perf_event *event); struct ui_progress; diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 5c886fbd50ce..731265f4ad19 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -7,7 +7,7 @@ #include "strlist.h" #include "thread.h" -static const char *event__name[] = { +static const char *perf_event__names[] = { [0] = "TOTAL", [PERF_RECORD_MMAP] = "MMAP", [PERF_RECORD_LOST] = "LOST", @@ -25,13 +25,13 @@ static const char *event__name[] = { [PERF_RECORD_FINISHED_ROUND] = "FINISHED_ROUND", }; -const char *event__get_event_name(unsigned int id) +const char *perf_event__name(unsigned int id) { - if (id >= ARRAY_SIZE(event__name)) + if (id >= ARRAY_SIZE(perf_event__names)) return "INVALID"; - if (!event__name[id]) + if (!perf_event__names[id]) return "UNKNOWN"; - return event__name[id]; + return perf_event__names[id]; } static struct perf_sample synth_sample = { @@ -43,9 +43,9 @@ static struct perf_sample synth_sample = { .period = 1, }; -static pid_t event__synthesize_comm(event_t *event, pid_t pid, int full, - event__handler_t process, - struct perf_session *session) +static pid_t perf_event__synthesize_comm(union perf_event *event, pid_t pid, + int full, perf_event__handler_t process, + struct perf_session *session) { char filename[PATH_MAX]; char bf[BUFSIZ]; @@ -126,9 +126,10 @@ out: return tgid; } -static int event__synthesize_mmap_events(event_t *event, pid_t pid, pid_t tgid, - event__handler_t process, - struct perf_session *session) +static int perf_event__synthesize_mmap_events(union perf_event *event, + pid_t pid, pid_t tgid, + perf_event__handler_t process, + struct perf_session *session) { char filename[PATH_MAX]; FILE *fp; @@ -199,14 +200,14 @@ static int event__synthesize_mmap_events(event_t *event, pid_t pid, pid_t tgid, return 0; } -int event__synthesize_modules(event__handler_t process, - struct perf_session *session, - struct machine *machine) +int perf_event__synthesize_modules(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine) { struct rb_node *nd; struct map_groups *kmaps = &machine->kmaps; - event_t *event = zalloc(sizeof(event->mmap) + session->id_hdr_size); - + union perf_event *event = zalloc((sizeof(event->mmap) + + session->id_hdr_size)); if (event == NULL) { pr_debug("Not enough memory synthesizing mmap event " "for kernel modules\n"); @@ -251,22 +252,23 @@ int event__synthesize_modules(event__handler_t process, return 0; } -static int __event__synthesize_thread(event_t *comm_event, event_t *mmap_event, - pid_t pid, event__handler_t process, +static int __event__synthesize_thread(union perf_event *comm_event, + union perf_event *mmap_event, + pid_t pid, perf_event__handler_t process, struct perf_session *session) { - pid_t tgid = event__synthesize_comm(comm_event, pid, 1, process, + pid_t tgid = perf_event__synthesize_comm(comm_event, pid, 1, process, session); if (tgid == -1) return -1; - return event__synthesize_mmap_events(mmap_event, pid, tgid, + return perf_event__synthesize_mmap_events(mmap_event, pid, tgid, process, session); } -int event__synthesize_thread(pid_t pid, event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_thread(pid_t pid, perf_event__handler_t process, + struct perf_session *session) { - event_t *comm_event, *mmap_event; + union perf_event *comm_event, *mmap_event; int err = -1; comm_event = malloc(sizeof(comm_event->comm) + session->id_hdr_size); @@ -286,12 +288,12 @@ out: return err; } -int event__synthesize_threads(event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_threads(perf_event__handler_t process, + struct perf_session *session) { DIR *proc; struct dirent dirent, *next; - event_t *comm_event, *mmap_event; + union perf_event *comm_event, *mmap_event; int err = -1; comm_event = malloc(sizeof(comm_event->comm) + session->id_hdr_size); @@ -349,10 +351,10 @@ static int find_symbol_cb(void *arg, const char *name, char type, return 1; } -int event__synthesize_kernel_mmap(event__handler_t process, - struct perf_session *session, - struct machine *machine, - const char *symbol_name) +int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine, + const char *symbol_name) { size_t size; const char *filename, *mmap_name; @@ -366,8 +368,8 @@ int event__synthesize_kernel_mmap(event__handler_t process, * kernels. */ struct process_symbol_args args = { .name = symbol_name, }; - event_t *event = zalloc(sizeof(event->mmap) + session->id_hdr_size); - + union perf_event *event = zalloc((sizeof(event->mmap) + + session->id_hdr_size)); if (event == NULL) { pr_debug("Not enough memory synthesizing mmap event " "for kernel modules\n"); @@ -440,14 +442,15 @@ static int thread__set_comm_adjust(struct thread *self, const char *comm, return 0; } -int event__process_comm(event_t *self, struct perf_sample *sample __used, - struct perf_session *session) +int perf_event__process_comm(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { - struct thread *thread = perf_session__findnew(session, self->comm.tid); + struct thread *thread = perf_session__findnew(session, event->comm.tid); - dump_printf(": %s:%d\n", self->comm.comm, self->comm.tid); + dump_printf(": %s:%d\n", event->comm.comm, event->comm.tid); - if (thread == NULL || thread__set_comm_adjust(thread, self->comm.comm, + if (thread == NULL || thread__set_comm_adjust(thread, event->comm.comm, &session->hists)) { dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n"); return -1; @@ -456,19 +459,21 @@ int event__process_comm(event_t *self, struct perf_sample *sample __used, return 0; } -int event__process_lost(event_t *self, struct perf_sample *sample __used, - struct perf_session *session) +int perf_event__process_lost(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { dump_printf(": id:%" PRIu64 ": lost:%" PRIu64 "\n", - self->lost.id, self->lost.lost); - session->hists.stats.total_lost += self->lost.lost; + event->lost.id, event->lost.lost); + session->hists.stats.total_lost += event->lost.lost; return 0; } -static void event_set_kernel_mmap_len(struct map **maps, event_t *self) +static void perf_event__set_kernel_mmap_len(union perf_event *event, + struct map **maps) { - maps[MAP__FUNCTION]->start = self->mmap.start; - maps[MAP__FUNCTION]->end = self->mmap.start + self->mmap.len; + maps[MAP__FUNCTION]->start = event->mmap.start; + maps[MAP__FUNCTION]->end = event->mmap.start + event->mmap.len; /* * Be a bit paranoid here, some perf.data file came with * a zero sized synthesized MMAP event for the kernel. @@ -477,8 +482,8 @@ static void event_set_kernel_mmap_len(struct map **maps, event_t *self) maps[MAP__FUNCTION]->end = ~0ULL; } -static int event__process_kernel_mmap(event_t *self, - struct perf_session *session) +static int perf_event__process_kernel_mmap(union perf_event *event, + struct perf_session *session) { struct map *map; char kmmap_prefix[PATH_MAX]; @@ -486,9 +491,9 @@ static int event__process_kernel_mmap(event_t *self, enum dso_kernel_type kernel_type; bool is_kernel_mmap; - machine = perf_session__findnew_machine(session, self->mmap.pid); + machine = perf_session__findnew_machine(session, event->mmap.pid); if (!machine) { - pr_err("Can't find id %d's machine\n", self->mmap.pid); + pr_err("Can't find id %d's machine\n", event->mmap.pid); goto out_problem; } @@ -498,17 +503,17 @@ static int event__process_kernel_mmap(event_t *self, else kernel_type = DSO_TYPE_GUEST_KERNEL; - is_kernel_mmap = memcmp(self->mmap.filename, + is_kernel_mmap = memcmp(event->mmap.filename, kmmap_prefix, strlen(kmmap_prefix)) == 0; - if (self->mmap.filename[0] == '/' || - (!is_kernel_mmap && self->mmap.filename[0] == '[')) { + if (event->mmap.filename[0] == '/' || + (!is_kernel_mmap && event->mmap.filename[0] == '[')) { char short_module_name[1024]; char *name, *dot; - if (self->mmap.filename[0] == '/') { - name = strrchr(self->mmap.filename, '/'); + if (event->mmap.filename[0] == '/') { + name = strrchr(event->mmap.filename, '/'); if (name == NULL) goto out_problem; @@ -520,10 +525,10 @@ static int event__process_kernel_mmap(event_t *self, "[%.*s]", (int)(dot - name), name); strxfrchar(short_module_name, '-', '_'); } else - strcpy(short_module_name, self->mmap.filename); + strcpy(short_module_name, event->mmap.filename); - map = machine__new_module(machine, self->mmap.start, - self->mmap.filename); + map = machine__new_module(machine, event->mmap.start, + event->mmap.filename); if (map == NULL) goto out_problem; @@ -533,9 +538,9 @@ static int event__process_kernel_mmap(event_t *self, map->dso->short_name = name; map->dso->sname_alloc = 1; - map->end = map->start + self->mmap.len; + map->end = map->start + event->mmap.len; } else if (is_kernel_mmap) { - const char *symbol_name = (self->mmap.filename + + const char *symbol_name = (event->mmap.filename + strlen(kmmap_prefix)); /* * Should be there already, from the build-id table in @@ -550,10 +555,10 @@ static int event__process_kernel_mmap(event_t *self, if (__machine__create_kernel_maps(machine, kernel) < 0) goto out_problem; - event_set_kernel_mmap_len(machine->vmlinux_maps, self); + perf_event__set_kernel_mmap_len(event, machine->vmlinux_maps); perf_session__set_kallsyms_ref_reloc_sym(machine->vmlinux_maps, symbol_name, - self->mmap.pgoff); + event->mmap.pgoff); if (machine__is_default_guest(machine)) { /* * preload dso of guest kernel and modules @@ -567,22 +572,23 @@ out_problem: return -1; } -int event__process_mmap(event_t *self, struct perf_sample *sample __used, - struct perf_session *session) +int perf_event__process_mmap(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { struct machine *machine; struct thread *thread; struct map *map; - u8 cpumode = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; int ret = 0; dump_printf(" %d/%d: [%#" PRIx64 "(%#" PRIx64 ") @ %#" PRIx64 "]: %s\n", - self->mmap.pid, self->mmap.tid, self->mmap.start, - self->mmap.len, self->mmap.pgoff, self->mmap.filename); + event->mmap.pid, event->mmap.tid, event->mmap.start, + event->mmap.len, event->mmap.pgoff, event->mmap.filename); if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL || cpumode == PERF_RECORD_MISC_KERNEL) { - ret = event__process_kernel_mmap(self, session); + ret = perf_event__process_kernel_mmap(event, session); if (ret < 0) goto out_problem; return 0; @@ -591,12 +597,12 @@ int event__process_mmap(event_t *self, struct perf_sample *sample __used, machine = perf_session__find_host_machine(session); if (machine == NULL) goto out_problem; - thread = perf_session__findnew(session, self->mmap.pid); + thread = perf_session__findnew(session, event->mmap.pid); if (thread == NULL) goto out_problem; - map = map__new(&machine->user_dsos, self->mmap.start, - self->mmap.len, self->mmap.pgoff, - self->mmap.pid, self->mmap.filename, + map = map__new(&machine->user_dsos, event->mmap.start, + event->mmap.len, event->mmap.pgoff, + event->mmap.pid, event->mmap.filename, MAP__FUNCTION); if (map == NULL) goto out_problem; @@ -609,16 +615,17 @@ out_problem: return 0; } -int event__process_task(event_t *self, struct perf_sample *sample __used, - struct perf_session *session) +int perf_event__process_task(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { - struct thread *thread = perf_session__findnew(session, self->fork.tid); - struct thread *parent = perf_session__findnew(session, self->fork.ptid); + struct thread *thread = perf_session__findnew(session, event->fork.tid); + struct thread *parent = perf_session__findnew(session, event->fork.ptid); - dump_printf("(%d:%d):(%d:%d)\n", self->fork.pid, self->fork.tid, - self->fork.ppid, self->fork.ptid); + dump_printf("(%d:%d):(%d:%d)\n", event->fork.pid, event->fork.tid, + event->fork.ppid, event->fork.ptid); - if (self->header.type == PERF_RECORD_EXIT) { + if (event->header.type == PERF_RECORD_EXIT) { perf_session__remove_thread(session, thread); return 0; } @@ -632,22 +639,22 @@ int event__process_task(event_t *self, struct perf_sample *sample __used, return 0; } -int event__process(event_t *event, struct perf_sample *sample, - struct perf_session *session) +int perf_event__process(union perf_event *event, struct perf_sample *sample, + struct perf_session *session) { switch (event->header.type) { case PERF_RECORD_COMM: - event__process_comm(event, sample, session); + perf_event__process_comm(event, sample, session); break; case PERF_RECORD_MMAP: - event__process_mmap(event, sample, session); + perf_event__process_mmap(event, sample, session); break; case PERF_RECORD_FORK: case PERF_RECORD_EXIT: - event__process_task(event, sample, session); + perf_event__process_task(event, sample, session); break; case PERF_RECORD_LOST: - event__process_lost(event, sample, session); + perf_event__process_lost(event, sample, session); default: break; } @@ -756,12 +763,14 @@ static void dso__calc_col_width(struct dso *self, struct hists *hists) self->slen_calculated = 1; } -int event__preprocess_sample(const event_t *self, struct perf_session *session, - struct addr_location *al, struct perf_sample *sample, - symbol_filter_t filter) +int perf_event__preprocess_sample(const union perf_event *event, + struct perf_session *session, + struct addr_location *al, + struct perf_sample *sample, + symbol_filter_t filter) { - u8 cpumode = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; - struct thread *thread = perf_session__findnew(session, self->ip.pid); + u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + struct thread *thread = perf_session__findnew(session, event->ip.pid); if (thread == NULL) return -1; @@ -783,7 +792,7 @@ int event__preprocess_sample(const event_t *self, struct perf_session *session, machine__create_kernel_maps(&session->host_machine); thread__find_addr_map(thread, session, cpumode, MAP__FUNCTION, - self->ip.pid, self->ip.ip, al); + event->ip.pid, event->ip.ip, al); dump_printf(" ...... dso: %s\n", al->map ? al->map->dso->long_name : al->level == 'H' ? "[hypervisor]" : ""); diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 84fd71ffd619..eecb42273d59 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -117,7 +117,7 @@ struct tracing_data_event { u32 size; }; -typedef union event_union { +union perf_event { struct perf_event_header header; struct ip_event ip; struct mmap_event mmap; @@ -130,49 +130,52 @@ typedef union event_union { struct event_type_event event_type; struct tracing_data_event tracing_data; struct build_id_event build_id; -} event_t; +}; -void event__print_totals(void); +void perf_event__print_totals(void); struct perf_session; -typedef int (*event__handler_synth_t)(event_t *event, +typedef int (*perf_event__handler_synth_t)(union perf_event *event, + struct perf_session *session); +typedef int (*perf_event__handler_t)(union perf_event *event, + struct perf_sample *sample, struct perf_session *session); -typedef int (*event__handler_t)(event_t *event, struct perf_sample *sample, - struct perf_session *session); -int event__synthesize_thread(pid_t pid, event__handler_t process, +int perf_event__synthesize_thread(pid_t pid, perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_threads(perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine, + const char *symbol_name); + +int perf_event__synthesize_modules(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine); + +int perf_event__process_comm(union perf_event *event, struct perf_sample *sample, struct perf_session *session); -int event__synthesize_threads(event__handler_t process, - struct perf_session *session); -int event__synthesize_kernel_mmap(event__handler_t process, - struct perf_session *session, - struct machine *machine, - const char *symbol_name); - -int event__synthesize_modules(event__handler_t process, - struct perf_session *session, - struct machine *machine); - -int event__process_comm(event_t *event, struct perf_sample *sample, - struct perf_session *session); -int event__process_lost(event_t *event, struct perf_sample *sample, - struct perf_session *session); -int event__process_mmap(event_t *event, struct perf_sample *sample, - struct perf_session *session); -int event__process_task(event_t *event, struct perf_sample *sample, +int perf_event__process_lost(union perf_event *event, struct perf_sample *sample, + struct perf_session *session); +int perf_event__process_mmap(union perf_event *event, struct perf_sample *sample, + struct perf_session *session); +int perf_event__process_task(union perf_event *event, struct perf_sample *sample, + struct perf_session *session); +int perf_event__process(union perf_event *event, struct perf_sample *sample, struct perf_session *session); -int event__process(event_t *event, struct perf_sample *sample, - struct perf_session *session); struct addr_location; -int event__preprocess_sample(const event_t *self, struct perf_session *session, - struct addr_location *al, struct perf_sample *sample, - symbol_filter_t filter); +int perf_event__preprocess_sample(const union perf_event *self, + struct perf_session *session, + struct addr_location *al, + struct perf_sample *sample, + symbol_filter_t filter); -const char *event__get_event_name(unsigned int id); +const char *perf_event__name(unsigned int id); -int event__parse_sample(const event_t *event, u64 type, bool sample_id_all, - struct perf_sample *sample); +int perf_event__parse_sample(const union perf_event *event, u64 type, + bool sample_id_all, struct perf_sample *sample); #endif /* __PERF_RECORD_H */ diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index b498eecbe850..917fc18d0bed 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -107,7 +107,7 @@ struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id) return NULL; } -event_t *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) +union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) { /* XXX Move this to perf.c, making it generally available */ unsigned int page_size = sysconf(_SC_PAGE_SIZE); @@ -115,7 +115,7 @@ event_t *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) unsigned int head = perf_mmap__read_head(md); unsigned int old = md->prev; unsigned char *data = md->base + page_size; - event_t *event = NULL; + union perf_event *event = NULL; if (evlist->overwrite) { /* @@ -140,7 +140,7 @@ event_t *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) if (old != head) { size_t size; - event = (event_t *)&data[old & md->mask]; + event = (union perf_event *)&data[old & md->mask]; size = event->header.size; /* diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 2706ae40c8b3..022ae404b908 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -17,7 +17,7 @@ struct perf_evlist { int nr_fds; int mmap_len; bool overwrite; - event_t event_copy; + union perf_event event_copy; struct perf_mmap *mmap; struct pollfd *pollfd; }; @@ -37,6 +37,6 @@ void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id); -event_t *perf_evlist__read_on_cpu(struct perf_evlist *self, int cpu); +union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *self, int cpu); #endif /* __PERF_EVLIST_H */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index a13488511880..fddeb08f48a7 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -367,8 +367,8 @@ out_unmap: return -1; } -static int event__parse_id_sample(const event_t *event, u64 type, - struct perf_sample *sample) +static int perf_event__parse_id_sample(const union perf_event *event, u64 type, + struct perf_sample *sample) { const u64 *array = event->sample.array; @@ -405,8 +405,8 @@ static int event__parse_id_sample(const event_t *event, u64 type, return 0; } -int event__parse_sample(const event_t *event, u64 type, bool sample_id_all, - struct perf_sample *data) +int perf_event__parse_sample(const union perf_event *event, u64 type, + bool sample_id_all, struct perf_sample *data) { const u64 *array; @@ -416,7 +416,7 @@ int event__parse_sample(const event_t *event, u64 type, bool sample_id_all, if (event->header.type != PERF_RECORD_SAMPLE) { if (!sample_id_all) return 0; - return event__parse_id_sample(event, type, data); + return perf_event__parse_id_sample(event, type, data); } array = event->sample.array; diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index f0138d472339..c0de5ec44145 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -1000,11 +1000,11 @@ perf_header__find_attr(u64 id, struct perf_header *header) return NULL; } -int event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, - event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, + perf_event__handler_t process, + struct perf_session *session) { - event_t *ev; + union perf_event *ev; size_t size; int err; @@ -1031,8 +1031,9 @@ int event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, return err; } -int event__synthesize_attrs(struct perf_header *self, event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_attrs(struct perf_header *self, + perf_event__handler_t process, + struct perf_session *session) { struct perf_header_attr *attr; int i, err = 0; @@ -1040,8 +1041,8 @@ int event__synthesize_attrs(struct perf_header *self, event__handler_t process, for (i = 0; i < self->attrs; i++) { attr = self->attr[i]; - err = event__synthesize_attr(&attr->attr, attr->ids, attr->id, - process, session); + err = perf_event__synthesize_attr(&attr->attr, attr->ids, + attr->id, process, session); if (err) { pr_debug("failed to create perf header attribute\n"); return err; @@ -1051,21 +1052,22 @@ int event__synthesize_attrs(struct perf_header *self, event__handler_t process, return err; } -int event__process_attr(event_t *self, struct perf_session *session) +int perf_event__process_attr(union perf_event *event, + struct perf_session *session) { struct perf_header_attr *attr; unsigned int i, ids, n_ids; - attr = perf_header_attr__new(&self->attr.attr); + attr = perf_header_attr__new(&event->attr.attr); if (attr == NULL) return -ENOMEM; - ids = self->header.size; - ids -= (void *)&self->attr.id - (void *)self; + ids = event->header.size; + ids -= (void *)&event->attr.id - (void *)event; n_ids = ids / sizeof(u64); for (i = 0; i < n_ids; i++) { - if (perf_header_attr__add_id(attr, self->attr.id[i]) < 0) { + if (perf_header_attr__add_id(attr, event->attr.id[i]) < 0) { perf_header_attr__delete(attr); return -ENOMEM; } @@ -1081,11 +1083,11 @@ int event__process_attr(event_t *self, struct perf_session *session) return 0; } -int event__synthesize_event_type(u64 event_id, char *name, - event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_event_type(u64 event_id, char *name, + perf_event__handler_t process, + struct perf_session *session) { - event_t ev; + union perf_event ev; size_t size = 0; int err = 0; @@ -1106,8 +1108,8 @@ int event__synthesize_event_type(u64 event_id, char *name, return err; } -int event__synthesize_event_types(event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_event_types(perf_event__handler_t process, + struct perf_session *session) { struct perf_trace_event_type *type; int i, err = 0; @@ -1115,8 +1117,9 @@ int event__synthesize_event_types(event__handler_t process, for (i = 0; i < event_count; i++) { type = &events[i]; - err = event__synthesize_event_type(type->event_id, type->name, - process, session); + err = perf_event__synthesize_event_type(type->event_id, + type->name, process, + session); if (err) { pr_debug("failed to create perf header event type\n"); return err; @@ -1126,21 +1129,21 @@ int event__synthesize_event_types(event__handler_t process, return err; } -int event__process_event_type(event_t *self, - struct perf_session *session __unused) +int perf_event__process_event_type(union perf_event *event, + struct perf_session *session __unused) { - if (perf_header__push_event(self->event_type.event_type.event_id, - self->event_type.event_type.name) < 0) + if (perf_header__push_event(event->event_type.event_type.event_id, + event->event_type.event_type.name) < 0) return -ENOMEM; return 0; } -int event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, - event__handler_t process, +int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, + perf_event__handler_t process, struct perf_session *session __unused) { - event_t ev; + union perf_event ev; ssize_t size = 0, aligned_size = 0, padding; int err = 0; @@ -1163,10 +1166,10 @@ int event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, return aligned_size; } -int event__process_tracing_data(event_t *self, - struct perf_session *session) +int perf_event__process_tracing_data(union perf_event *event, + struct perf_session *session) { - ssize_t size_read, padding, size = self->tracing_data.size; + ssize_t size_read, padding, size = event->tracing_data.size; off_t offset = lseek(session->fd, 0, SEEK_CUR); char buf[BUFSIZ]; @@ -1192,12 +1195,12 @@ int event__process_tracing_data(event_t *self, return size_read + padding; } -int event__synthesize_build_id(struct dso *pos, u16 misc, - event__handler_t process, - struct machine *machine, - struct perf_session *session) +int perf_event__synthesize_build_id(struct dso *pos, u16 misc, + perf_event__handler_t process, + struct machine *machine, + struct perf_session *session) { - event_t ev; + union perf_event ev; size_t len; int err = 0; @@ -1220,11 +1223,11 @@ int event__synthesize_build_id(struct dso *pos, u16 misc, return err; } -int event__process_build_id(event_t *self, - struct perf_session *session) +int perf_event__process_build_id(union perf_event *event, + struct perf_session *session) { - __event_process_build_id(&self->build_id, - self->build_id.filename, + __event_process_build_id(&event->build_id, + event->build_id.filename, session); return 0; } diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 65afd7f74e0d..f042cebcec1e 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -100,32 +100,32 @@ int build_id_cache__add_s(const char *sbuild_id, const char *debugdir, const char *name, bool is_kallsyms); int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir); -int event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, - event__handler_t process, - struct perf_session *session); -int event__synthesize_attrs(struct perf_header *self, - event__handler_t process, - struct perf_session *session); -int event__process_attr(event_t *self, struct perf_session *session); - -int event__synthesize_event_type(u64 event_id, char *name, - event__handler_t process, +int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, + perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_attrs(struct perf_header *self, + perf_event__handler_t process, struct perf_session *session); -int event__synthesize_event_types(event__handler_t process, - struct perf_session *session); -int event__process_event_type(event_t *self, - struct perf_session *session); - -int event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, - event__handler_t process, +int perf_event__process_attr(union perf_event *event, struct perf_session *session); + +int perf_event__synthesize_event_type(u64 event_id, char *name, + perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_event_types(perf_event__handler_t process, + struct perf_session *session); +int perf_event__process_event_type(union perf_event *event, struct perf_session *session); -int event__process_tracing_data(event_t *self, - struct perf_session *session); - -int event__synthesize_build_id(struct dso *pos, u16 misc, - event__handler_t process, - struct machine *machine, - struct perf_session *session); -int event__process_build_id(event_t *self, struct perf_session *session); +int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, + perf_event__handler_t process, + struct perf_session *session); +int perf_event__process_tracing_data(union perf_event *event, + struct perf_session *session); + +int perf_event__synthesize_build_id(struct dso *pos, u16 misc, + perf_event__handler_t process, + struct machine *machine, + struct perf_session *session); +int perf_event__process_build_id(union perf_event *event, + struct perf_session *session); #endif /* __PERF_HEADER_H */ diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 02ed318d7312..95887804dc8e 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -1182,7 +1182,7 @@ size_t hists__fprintf_nr_events(struct hists *self, FILE *fp) size_t ret = 0; for (i = 0; i < PERF_RECORD_HEADER_MAX; ++i) { - const char *name = event__get_event_name(i); + const char *name = perf_event__name(i); if (!strcmp(name, "UNKNOWN")) continue; diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index ee0b61102571..a3a871f7bda3 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -165,7 +165,7 @@ struct perf_session *perf_session__new(const char *filename, int mode, } else if (mode == O_WRONLY) { /* * In O_RDONLY mode this will be performed when reading the - * kernel MMAP event, in event__process_mmap(). + * kernel MMAP event, in perf_event__process_mmap(). */ if (perf_session__create_kernel_maps(self) < 0) goto out_delete; @@ -291,14 +291,14 @@ int perf_session__resolve_callchain(struct perf_session *self, return 0; } -static int process_event_synth_stub(event_t *event __used, +static int process_event_synth_stub(union perf_event *event __used, struct perf_session *session __used) { dump_printf(": unhandled!\n"); return 0; } -static int process_event_stub(event_t *event __used, +static int process_event_stub(union perf_event *event __used, struct perf_sample *sample __used, struct perf_session *session __used) { @@ -306,7 +306,7 @@ static int process_event_stub(event_t *event __used, return 0; } -static int process_finished_round_stub(event_t *event __used, +static int process_finished_round_stub(union perf_event *event __used, struct perf_session *session __used, struct perf_event_ops *ops __used) { @@ -314,7 +314,7 @@ static int process_finished_round_stub(event_t *event __used, return 0; } -static int process_finished_round(event_t *event, +static int process_finished_round(union perf_event *event, struct perf_session *session, struct perf_event_ops *ops); @@ -331,7 +331,7 @@ static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) if (handler->exit == NULL) handler->exit = process_event_stub; if (handler->lost == NULL) - handler->lost = event__process_lost; + handler->lost = perf_event__process_lost; if (handler->read == NULL) handler->read = process_event_stub; if (handler->throttle == NULL) @@ -365,98 +365,98 @@ void mem_bswap_64(void *src, int byte_size) } } -static void event__all64_swap(event_t *self) +static void perf_event__all64_swap(union perf_event *event) { - struct perf_event_header *hdr = &self->header; - mem_bswap_64(hdr + 1, self->header.size - sizeof(*hdr)); + struct perf_event_header *hdr = &event->header; + mem_bswap_64(hdr + 1, event->header.size - sizeof(*hdr)); } -static void event__comm_swap(event_t *self) +static void perf_event__comm_swap(union perf_event *event) { - self->comm.pid = bswap_32(self->comm.pid); - self->comm.tid = bswap_32(self->comm.tid); + event->comm.pid = bswap_32(event->comm.pid); + event->comm.tid = bswap_32(event->comm.tid); } -static void event__mmap_swap(event_t *self) +static void perf_event__mmap_swap(union perf_event *event) { - self->mmap.pid = bswap_32(self->mmap.pid); - self->mmap.tid = bswap_32(self->mmap.tid); - self->mmap.start = bswap_64(self->mmap.start); - self->mmap.len = bswap_64(self->mmap.len); - self->mmap.pgoff = bswap_64(self->mmap.pgoff); + event->mmap.pid = bswap_32(event->mmap.pid); + event->mmap.tid = bswap_32(event->mmap.tid); + event->mmap.start = bswap_64(event->mmap.start); + event->mmap.len = bswap_64(event->mmap.len); + event->mmap.pgoff = bswap_64(event->mmap.pgoff); } -static void event__task_swap(event_t *self) +static void perf_event__task_swap(union perf_event *event) { - self->fork.pid = bswap_32(self->fork.pid); - self->fork.tid = bswap_32(self->fork.tid); - self->fork.ppid = bswap_32(self->fork.ppid); - self->fork.ptid = bswap_32(self->fork.ptid); - self->fork.time = bswap_64(self->fork.time); + event->fork.pid = bswap_32(event->fork.pid); + event->fork.tid = bswap_32(event->fork.tid); + event->fork.ppid = bswap_32(event->fork.ppid); + event->fork.ptid = bswap_32(event->fork.ptid); + event->fork.time = bswap_64(event->fork.time); } -static void event__read_swap(event_t *self) +static void perf_event__read_swap(union perf_event *event) { - self->read.pid = bswap_32(self->read.pid); - self->read.tid = bswap_32(self->read.tid); - self->read.value = bswap_64(self->read.value); - self->read.time_enabled = bswap_64(self->read.time_enabled); - self->read.time_running = bswap_64(self->read.time_running); - self->read.id = bswap_64(self->read.id); + event->read.pid = bswap_32(event->read.pid); + event->read.tid = bswap_32(event->read.tid); + event->read.value = bswap_64(event->read.value); + event->read.time_enabled = bswap_64(event->read.time_enabled); + event->read.time_running = bswap_64(event->read.time_running); + event->read.id = bswap_64(event->read.id); } -static void event__attr_swap(event_t *self) +static void perf_event__attr_swap(union perf_event *event) { size_t size; - self->attr.attr.type = bswap_32(self->attr.attr.type); - self->attr.attr.size = bswap_32(self->attr.attr.size); - self->attr.attr.config = bswap_64(self->attr.attr.config); - self->attr.attr.sample_period = bswap_64(self->attr.attr.sample_period); - self->attr.attr.sample_type = bswap_64(self->attr.attr.sample_type); - self->attr.attr.read_format = bswap_64(self->attr.attr.read_format); - self->attr.attr.wakeup_events = bswap_32(self->attr.attr.wakeup_events); - self->attr.attr.bp_type = bswap_32(self->attr.attr.bp_type); - self->attr.attr.bp_addr = bswap_64(self->attr.attr.bp_addr); - self->attr.attr.bp_len = bswap_64(self->attr.attr.bp_len); - - size = self->header.size; - size -= (void *)&self->attr.id - (void *)self; - mem_bswap_64(self->attr.id, size); + event->attr.attr.type = bswap_32(event->attr.attr.type); + event->attr.attr.size = bswap_32(event->attr.attr.size); + event->attr.attr.config = bswap_64(event->attr.attr.config); + event->attr.attr.sample_period = bswap_64(event->attr.attr.sample_period); + event->attr.attr.sample_type = bswap_64(event->attr.attr.sample_type); + event->attr.attr.read_format = bswap_64(event->attr.attr.read_format); + event->attr.attr.wakeup_events = bswap_32(event->attr.attr.wakeup_events); + event->attr.attr.bp_type = bswap_32(event->attr.attr.bp_type); + event->attr.attr.bp_addr = bswap_64(event->attr.attr.bp_addr); + event->attr.attr.bp_len = bswap_64(event->attr.attr.bp_len); + + size = event->header.size; + size -= (void *)&event->attr.id - (void *)event; + mem_bswap_64(event->attr.id, size); } -static void event__event_type_swap(event_t *self) +static void perf_event__event_type_swap(union perf_event *event) { - self->event_type.event_type.event_id = - bswap_64(self->event_type.event_type.event_id); + event->event_type.event_type.event_id = + bswap_64(event->event_type.event_type.event_id); } -static void event__tracing_data_swap(event_t *self) +static void perf_event__tracing_data_swap(union perf_event *event) { - self->tracing_data.size = bswap_32(self->tracing_data.size); + event->tracing_data.size = bswap_32(event->tracing_data.size); } -typedef void (*event__swap_op)(event_t *self); - -static event__swap_op event__swap_ops[] = { - [PERF_RECORD_MMAP] = event__mmap_swap, - [PERF_RECORD_COMM] = event__comm_swap, - [PERF_RECORD_FORK] = event__task_swap, - [PERF_RECORD_EXIT] = event__task_swap, - [PERF_RECORD_LOST] = event__all64_swap, - [PERF_RECORD_READ] = event__read_swap, - [PERF_RECORD_SAMPLE] = event__all64_swap, - [PERF_RECORD_HEADER_ATTR] = event__attr_swap, - [PERF_RECORD_HEADER_EVENT_TYPE] = event__event_type_swap, - [PERF_RECORD_HEADER_TRACING_DATA] = event__tracing_data_swap, - [PERF_RECORD_HEADER_BUILD_ID] = NULL, - [PERF_RECORD_HEADER_MAX] = NULL, +typedef void (*perf_event__swap_op)(union perf_event *event); + +static perf_event__swap_op perf_event__swap_ops[] = { + [PERF_RECORD_MMAP] = perf_event__mmap_swap, + [PERF_RECORD_COMM] = perf_event__comm_swap, + [PERF_RECORD_FORK] = perf_event__task_swap, + [PERF_RECORD_EXIT] = perf_event__task_swap, + [PERF_RECORD_LOST] = perf_event__all64_swap, + [PERF_RECORD_READ] = perf_event__read_swap, + [PERF_RECORD_SAMPLE] = perf_event__all64_swap, + [PERF_RECORD_HEADER_ATTR] = perf_event__attr_swap, + [PERF_RECORD_HEADER_EVENT_TYPE] = perf_event__event_type_swap, + [PERF_RECORD_HEADER_TRACING_DATA] = perf_event__tracing_data_swap, + [PERF_RECORD_HEADER_BUILD_ID] = NULL, + [PERF_RECORD_HEADER_MAX] = NULL, }; struct sample_queue { u64 timestamp; u64 file_offset; - event_t *event; + union perf_event *event; struct list_head list; }; @@ -474,7 +474,7 @@ static void perf_session_free_sample_buffers(struct perf_session *session) } static int perf_session_deliver_event(struct perf_session *session, - event_t *event, + union perf_event *event, struct perf_sample *sample, struct perf_event_ops *ops, u64 file_offset); @@ -552,7 +552,7 @@ static void flush_sample_queue(struct perf_session *s, * Flush every events below timestamp 7 * etc... */ -static int process_finished_round(event_t *event __used, +static int process_finished_round(union perf_event *event __used, struct perf_session *session, struct perf_event_ops *ops) { @@ -609,7 +609,7 @@ static void __queue_event(struct sample_queue *new, struct perf_session *s) #define MAX_SAMPLE_BUFFER (64 * 1024 / sizeof(struct sample_queue)) -static int perf_session_queue_event(struct perf_session *s, event_t *event, +static int perf_session_queue_event(struct perf_session *s, union perf_event *event, struct perf_sample *sample, u64 file_offset) { struct ordered_samples *os = &s->ordered_samples; @@ -662,7 +662,7 @@ static void callchain__printf(struct perf_sample *sample) } static void perf_session__print_tstamp(struct perf_session *session, - event_t *event, + union perf_event *event, struct perf_sample *sample) { if (event->header.type != PERF_RECORD_SAMPLE && @@ -678,7 +678,7 @@ static void perf_session__print_tstamp(struct perf_session *session, printf("%" PRIu64 " ", sample->time); } -static void dump_event(struct perf_session *session, event_t *event, +static void dump_event(struct perf_session *session, union perf_event *event, u64 file_offset, struct perf_sample *sample) { if (!dump_trace) @@ -693,10 +693,10 @@ static void dump_event(struct perf_session *session, event_t *event, perf_session__print_tstamp(session, event, sample); printf("%#" PRIx64 " [%#x]: PERF_RECORD_%s", file_offset, - event->header.size, event__get_event_name(event->header.type)); + event->header.size, perf_event__name(event->header.type)); } -static void dump_sample(struct perf_session *session, event_t *event, +static void dump_sample(struct perf_session *session, union perf_event *event, struct perf_sample *sample) { if (!dump_trace) @@ -711,7 +711,7 @@ static void dump_sample(struct perf_session *session, event_t *event, } static int perf_session_deliver_event(struct perf_session *session, - event_t *event, + union perf_event *event, struct perf_sample *sample, struct perf_event_ops *ops, u64 file_offset) @@ -745,7 +745,7 @@ static int perf_session_deliver_event(struct perf_session *session, } static int perf_session__preprocess_sample(struct perf_session *session, - event_t *event, struct perf_sample *sample) + union perf_event *event, struct perf_sample *sample) { if (event->header.type != PERF_RECORD_SAMPLE || !(session->sample_type & PERF_SAMPLE_CALLCHAIN)) @@ -760,7 +760,7 @@ static int perf_session__preprocess_sample(struct perf_session *session, return 0; } -static int perf_session__process_user_event(struct perf_session *session, event_t *event, +static int perf_session__process_user_event(struct perf_session *session, union perf_event *event, struct perf_event_ops *ops, u64 file_offset) { dump_event(session, event, file_offset, NULL); @@ -785,15 +785,16 @@ static int perf_session__process_user_event(struct perf_session *session, event_ } static int perf_session__process_event(struct perf_session *session, - event_t *event, + union perf_event *event, struct perf_event_ops *ops, u64 file_offset) { struct perf_sample sample; int ret; - if (session->header.needs_swap && event__swap_ops[event->header.type]) - event__swap_ops[event->header.type](event); + if (session->header.needs_swap && + perf_event__swap_ops[event->header.type]) + perf_event__swap_ops[event->header.type](event); if (event->header.type >= PERF_RECORD_HEADER_MAX) return -EINVAL; @@ -845,7 +846,7 @@ static struct thread *perf_session__register_idle_thread(struct perf_session *se static void perf_session__warn_about_errors(const struct perf_session *session, const struct perf_event_ops *ops) { - if (ops->lost == event__process_lost && + if (ops->lost == perf_event__process_lost && session->hists.stats.total_lost != 0) { ui__warning("Processed %" PRIu64 " events and LOST %" PRIu64 "!\n\nCheck IO/CPU overload!\n\n", @@ -877,7 +878,7 @@ volatile int session_done; static int __perf_session__process_pipe_events(struct perf_session *self, struct perf_event_ops *ops) { - event_t event; + union perf_event event; uint32_t size; int skip = 0; u64 head; @@ -958,7 +959,7 @@ int __perf_session__process_events(struct perf_session *session, struct ui_progress *progress; size_t page_size, mmap_size; char *buf, *mmaps[8]; - event_t *event; + union perf_event *event; uint32_t size; perf_event_ops__fill_defaults(ops); @@ -1003,7 +1004,7 @@ remap: file_pos = file_offset + head; more: - event = (event_t *)(buf + head); + event = (union perf_event *)(buf + head); if (session->header.needs_swap) perf_event_header__bswap(&event->header); diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 365bf533a396..977b3a1b14aa 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -57,10 +57,11 @@ struct perf_session { struct perf_event_ops; -typedef int (*event_op)(event_t *self, struct perf_sample *sample, +typedef int (*event_op)(union perf_event *self, struct perf_sample *sample, struct perf_session *session); -typedef int (*event_synth_op)(event_t *self, struct perf_session *session); -typedef int (*event_op2)(event_t *self, struct perf_session *session, +typedef int (*event_synth_op)(union perf_event *self, + struct perf_session *session); +typedef int (*event_op2)(union perf_event *self, struct perf_session *session, struct perf_event_ops *ops); struct perf_event_ops { @@ -157,11 +158,11 @@ size_t perf_session__fprintf_nr_events(struct perf_session *self, FILE *fp) } static inline int perf_session__parse_sample(struct perf_session *session, - const event_t *event, + const union perf_event *event, struct perf_sample *sample) { - return event__parse_sample(event, session->sample_type, - session->sample_id_all, sample); + return perf_event__parse_sample(event, session->sample_type, + session->sample_id_all, sample); } #endif /* __PERF_SESSION_H */ -- cgit v1.2.3 From 877108e42b1b9ba64857c4030cf356ecc120fd18 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 29 Jan 2011 15:44:29 -0200 Subject: perf tools: Initial python binding First clarifying that this kind of binding is not a replacement or an equivalent to the 'perf script' way of using python with perf. The 'perf script' way is to process events and look at a given script for some python function that matches the events to pass each event for processing. This is a python module, i.e. everything is driven from the python script, that merely uses "import perf" or "from perf import". perf script is focused on tracepoints, this binding is focused on profiling as an initial target. More work is needed to make available tracepoint specific variables as event variables accessible via this binding. There is one example of such usage model, in tools/perf/python/twatch.py, a tool to watch "cycles" events together with task (fork, exit) and comm perf events. For now, due to me not being able to grok how python distutils cope with building C extensions outside the sources dir the install target just builds it, I'm using it as: [root@emilia linux]# export PYTHONPATH=~acme/git/build/perf/lib.linux-x86_64-2.6/ [root@emilia linux]# tools/perf/python/twatch.py cpu: 4, pid: 30126, tid: 30126 { type: mmap, pid: 30126, tid: 30126, start: 0x4, length: 0x82e9ca03, offset: 0, filename: } cpu: 6, pid: 47, tid: 47 { type: mmap, pid: 47, tid: 47, start: 0x6, length: 0xbef87c36, offset: 0, filename: } cpu: 1, pid: 0, tid: 0 { type: mmap, pid: 0, tid: 0, start: 0x1, length: 0x775d1904, offset: 0, filename: } cpu: 7, pid: 0, tid: 0 { type: mmap, pid: 0, tid: 0, start: 0x7, length: 0xc750aeb6, offset: 0, filename: } cpu: 5, pid: 2255, tid: 2255 { type: mmap, pid: 2255, tid: 2255, start: 0x5, length: 0x76669635, offset: 0, filename: } cpu: 0, pid: 0, tid: 0 { type: mmap, pid: 0, tid: 0, start: 0, length: 0x6422ef6b, offset: 0, filename: } cpu: 2, pid: 2255, tid: 2255 { type: mmap, pid: 2255, tid: 2255, start: 0x2, length: 0xe078757a, offset: 0, filename: } cpu: 1, pid: 5769, tid: 5769 { type: fork, pid: 30127, ppid: 5769, tid: 30127, ptid: 5769, time: 103893991270534} cpu: 6, pid: 30127, tid: 30127 { type: comm, pid: 30127, tid: 30127, comm: ls } cpu: 6, pid: 30127, tid: 30127 { type: exit, pid: 30127, ppid: 30127, tid: 30127, ptid: 30127, time: 103893993273024} The first 8 mmap events in this 8 way machine are a mistery that is still being investigated. More of the tools/perf/util/ APIs will be exposed via this python binding as the need arises. For now the focus is on creating events and processing them, symbol resolution is an obvious next step, with tracepoint variables as a close second step. Cc: Clark Williams Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Makefile | 20 +- tools/perf/python/twatch.py | 41 ++ tools/perf/util/python.c | 888 ++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/setup.py | 18 + 4 files changed, 966 insertions(+), 1 deletion(-) create mode 100755 tools/perf/python/twatch.py create mode 100644 tools/perf/util/python.c create mode 100644 tools/perf/util/setup.py (limited to 'tools/perf/util') diff --git a/tools/perf/Makefile b/tools/perf/Makefile index eedcf9541572..36ff73c0af94 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -315,6 +315,7 @@ COMPAT_CFLAGS = COMPAT_OBJS = LIB_H = LIB_OBJS = +PYRF_OBJS = SCRIPT_PERL = SCRIPT_SH = TEST_PROGRAMS = @@ -324,6 +325,9 @@ SCRIPT_SH += perf-archive.sh grep-libs = $(filter -l%,$(1)) strip-libs = $(filter-out -l%,$(1)) +pyrf: $(PYRF_OBJS) + python util/setup.py build --build-base='$(OUTPUT)' + # # No Perl scripts right now: # @@ -349,7 +353,7 @@ PROGRAMS += $(OUTPUT)perf # # what 'all' will build and 'install' will install, in perfexecdir -ALL_PROGRAMS = $(PROGRAMS) $(SCRIPTS) +ALL_PROGRAMS = $(PROGRAMS) $(SCRIPTS) pyrf # what 'all' will build but not install in perfexecdir OTHER_PROGRAMS = $(OUTPUT)perf$X @@ -520,6 +524,20 @@ BUILTIN_OBJS += $(OUTPUT)builtin-inject.o PERFLIBS = $(LIB_FILE) +# Files needed for the python binding, perf.so +# pyrf is just an internal name needed for all those wrappers. +# This has to be in sync with what is in the 'sources' variable in +# tools/perf/util/setup.py + +PYRF_OBJS += $(OUTPUT)util/cpumap.o +PYRF_OBJS += $(OUTPUT)util/ctype.o +PYRF_OBJS += $(OUTPUT)util/evlist.o +PYRF_OBJS += $(OUTPUT)util/evsel.o +PYRF_OBJS += $(OUTPUT)util/python.o +PYRF_OBJS += $(OUTPUT)util/thread_map.o +PYRF_OBJS += $(OUTPUT)util/util.o +PYRF_OBJS += $(OUTPUT)util/xyarray.o + # # Platform specific tweaks # diff --git a/tools/perf/python/twatch.py b/tools/perf/python/twatch.py new file mode 100755 index 000000000000..5e9f3b7b7ee8 --- /dev/null +++ b/tools/perf/python/twatch.py @@ -0,0 +1,41 @@ +#! /usr/bin/python +# -*- python -*- +# -*- coding: utf-8 -*- +# twatch - Experimental use of the perf python interface +# Copyright (C) 2011 Arnaldo Carvalho de Melo +# +# This application is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License +# as published by the Free Software Foundation; version 2. +# +# This application is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# General Public License for more details. + +import perf + +def main(): + cpus = perf.cpu_map() + threads = perf.thread_map() + evsel = perf.evsel(task = 1, comm = 1, mmap = 0, + wakeup_events = 1, sample_period = 1, + sample_id_all = 1, + sample_type = perf.SAMPLE_PERIOD | perf.SAMPLE_TID | perf.SAMPLE_CPU | perf.SAMPLE_TID) + evsel.open(cpus = cpus, threads = threads); + evlist = perf.evlist() + evlist.add(evsel) + evlist.mmap(cpus = cpus, threads = threads) + while True: + evlist.poll(timeout = -1) + for cpu in cpus: + event = evlist.read_on_cpu(cpu) + if not event: + continue + print "cpu: %2d, pid: %4d, tid: %4d" % (event.sample_cpu, + event.sample_pid, + event.sample_tid), + print event + +if __name__ == '__main__': + main() diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c new file mode 100644 index 000000000000..88d47895a79c --- /dev/null +++ b/tools/perf/util/python.c @@ -0,0 +1,888 @@ +#include +#include +#include +#include +#include "evlist.h" +#include "evsel.h" +#include "event.h" +#include "cpumap.h" +#include "thread_map.h" + +struct throttle_event { + struct perf_event_header header; + u64 time; + u64 id; + u64 stream_id; +}; + +#define member_def(type, member, ptype, help) \ + { #member, ptype, \ + offsetof(struct pyrf_event, event) + offsetof(struct type, member), \ + 0, help } + +#define sample_member_def(name, member, ptype, help) \ + { #name, ptype, \ + offsetof(struct pyrf_event, sample) + offsetof(struct perf_sample, member), \ + 0, help } + +struct pyrf_event { + PyObject_HEAD + struct perf_sample sample; + union perf_event event; +}; + +#define T_ULONG_LONG T_ULONG + +#define sample_members \ + sample_member_def(sample_ip, ip, T_ULONG_LONG, "event type"), \ + sample_member_def(sample_pid, pid, T_INT, "event pid"), \ + sample_member_def(sample_tid, tid, T_INT, "event tid"), \ + sample_member_def(sample_time, time, T_ULONG_LONG, "event timestamp"), \ + sample_member_def(sample_addr, addr, T_ULONG_LONG, "event addr"), \ + sample_member_def(sample_id, id, T_ULONG_LONG, "event id"), \ + sample_member_def(sample_stream_id, stream_id, T_ULONG_LONG, "event stream id"), \ + sample_member_def(sample_period, period, T_ULONG_LONG, "event period"), \ + sample_member_def(sample_cpu, cpu, T_UINT, "event cpu"), + +static char pyrf_mmap_event__doc[] = PyDoc_STR("perf mmap event object."); + +static PyMemberDef pyrf_mmap_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(mmap_event, pid, T_UINT, "event pid"), + member_def(mmap_event, tid, T_UINT, "event tid"), + member_def(mmap_event, start, T_ULONG_LONG, "start of the map"), + member_def(mmap_event, len, T_ULONG_LONG, "map length"), + member_def(mmap_event, pgoff, T_ULONG_LONG, "page offset"), + member_def(mmap_event, filename, T_STRING_INPLACE, "backing store"), + { NULL, }, +}; + +static PyObject *pyrf_mmap_event__repr(struct pyrf_event *pevent) +{ + PyObject *ret; + char *s; + + if (asprintf(&s, "{ type: mmap, pid: %u, tid: %u, start: %#" PRIx64 ", " + "length: %#" PRIx64 ", offset: %#" PRIx64 ", " + "filename: %s }", + pevent->event.mmap.pid, pevent->event.mmap.tid, + pevent->event.mmap.start, pevent->event.mmap.len, + pevent->event.mmap.pgoff, pevent->event.mmap.filename) < 0) { + ret = PyErr_NoMemory(); + } else { + ret = PyString_FromString(s); + free(s); + } + return ret; +} + +static PyTypeObject pyrf_mmap_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.mmap_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_mmap_event__doc, + .tp_members = pyrf_mmap_event__members, + .tp_repr = (reprfunc)pyrf_mmap_event__repr, +}; + +static char pyrf_task_event__doc[] = PyDoc_STR("perf task (fork/exit) event object."); + +static PyMemberDef pyrf_task_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(fork_event, pid, T_UINT, "event pid"), + member_def(fork_event, ppid, T_UINT, "event ppid"), + member_def(fork_event, tid, T_UINT, "event tid"), + member_def(fork_event, ptid, T_UINT, "event ptid"), + member_def(fork_event, time, T_ULONG_LONG, "timestamp"), + { NULL, }, +}; + +static PyObject *pyrf_task_event__repr(struct pyrf_event *pevent) +{ + return PyString_FromFormat("{ type: %s, pid: %u, ppid: %u, tid: %u, " + "ptid: %u, time: %" PRIu64 "}", + pevent->event.header.type == PERF_RECORD_FORK ? "fork" : "exit", + pevent->event.fork.pid, + pevent->event.fork.ppid, + pevent->event.fork.tid, + pevent->event.fork.ptid, + pevent->event.fork.time); +} + +static PyTypeObject pyrf_task_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.task_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_task_event__doc, + .tp_members = pyrf_task_event__members, + .tp_repr = (reprfunc)pyrf_task_event__repr, +}; + +static char pyrf_comm_event__doc[] = PyDoc_STR("perf comm event object."); + +static PyMemberDef pyrf_comm_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(comm_event, pid, T_UINT, "event pid"), + member_def(comm_event, tid, T_UINT, "event tid"), + member_def(comm_event, comm, T_STRING_INPLACE, "process name"), + { NULL, }, +}; + +static PyObject *pyrf_comm_event__repr(struct pyrf_event *pevent) +{ + return PyString_FromFormat("{ type: comm, pid: %u, tid: %u, comm: %s }", + pevent->event.comm.pid, + pevent->event.comm.tid, + pevent->event.comm.comm); +} + +static PyTypeObject pyrf_comm_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.comm_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_comm_event__doc, + .tp_members = pyrf_comm_event__members, + .tp_repr = (reprfunc)pyrf_comm_event__repr, +}; + +static char pyrf_throttle_event__doc[] = PyDoc_STR("perf throttle event object."); + +static PyMemberDef pyrf_throttle_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(throttle_event, time, T_ULONG_LONG, "timestamp"), + member_def(throttle_event, id, T_ULONG_LONG, "event id"), + member_def(throttle_event, stream_id, T_ULONG_LONG, "event stream id"), + { NULL, }, +}; + +static PyObject *pyrf_throttle_event__repr(struct pyrf_event *pevent) +{ + struct throttle_event *te = (struct throttle_event *)(&pevent->event.header + 1); + + return PyString_FromFormat("{ type: %sthrottle, time: %" PRIu64 ", id: %" PRIu64 + ", stream_id: %" PRIu64 " }", + pevent->event.header.type == PERF_RECORD_THROTTLE ? "" : "un", + te->time, te->id, te->stream_id); +} + +static PyTypeObject pyrf_throttle_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.throttle_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_throttle_event__doc, + .tp_members = pyrf_throttle_event__members, + .tp_repr = (reprfunc)pyrf_throttle_event__repr, +}; + +static int pyrf_event__setup_types(void) +{ + int err; + pyrf_mmap_event__type.tp_new = + pyrf_task_event__type.tp_new = + pyrf_comm_event__type.tp_new = + pyrf_throttle_event__type.tp_new = PyType_GenericNew; + err = PyType_Ready(&pyrf_mmap_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_task_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_comm_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_throttle_event__type); + if (err < 0) + goto out; +out: + return err; +} + +static PyTypeObject *pyrf_event__type[] = { + [PERF_RECORD_MMAP] = &pyrf_mmap_event__type, + [PERF_RECORD_LOST] = &pyrf_mmap_event__type, + [PERF_RECORD_COMM] = &pyrf_comm_event__type, + [PERF_RECORD_EXIT] = &pyrf_task_event__type, + [PERF_RECORD_THROTTLE] = &pyrf_throttle_event__type, + [PERF_RECORD_UNTHROTTLE] = &pyrf_throttle_event__type, + [PERF_RECORD_FORK] = &pyrf_task_event__type, + [PERF_RECORD_READ] = &pyrf_mmap_event__type, + [PERF_RECORD_SAMPLE] = &pyrf_mmap_event__type, +}; + +static PyObject *pyrf_event__new(union perf_event *event) +{ + struct pyrf_event *pevent; + PyTypeObject *ptype; + + if (event->header.type < PERF_RECORD_MMAP || + event->header.type > PERF_RECORD_SAMPLE) + return NULL; + + ptype = pyrf_event__type[event->header.type]; + pevent = PyObject_New(struct pyrf_event, ptype); + if (pevent != NULL) + memcpy(&pevent->event, event, event->header.size); + return (PyObject *)pevent; +} + +struct pyrf_cpu_map { + PyObject_HEAD + + struct cpu_map *cpus; +}; + +static int pyrf_cpu_map__init(struct pyrf_cpu_map *pcpus, + PyObject *args, PyObject *kwargs) +{ + static char *kwlist[] = { "cpustr", NULL, NULL, }; + char *cpustr = NULL; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|s", + kwlist, &cpustr)) + return -1; + + pcpus->cpus = cpu_map__new(cpustr); + if (pcpus->cpus == NULL) + return -1; + return 0; +} + +static void pyrf_cpu_map__delete(struct pyrf_cpu_map *pcpus) +{ + cpu_map__delete(pcpus->cpus); + pcpus->ob_type->tp_free((PyObject*)pcpus); +} + +static Py_ssize_t pyrf_cpu_map__length(PyObject *obj) +{ + struct pyrf_cpu_map *pcpus = (void *)obj; + + return pcpus->cpus->nr; +} + +static PyObject *pyrf_cpu_map__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_cpu_map *pcpus = (void *)obj; + + if (i >= pcpus->cpus->nr) + return NULL; + + return Py_BuildValue("i", pcpus->cpus->map[i]); +} + +static PySequenceMethods pyrf_cpu_map__sequence_methods = { + .sq_length = pyrf_cpu_map__length, + .sq_item = pyrf_cpu_map__item, +}; + +static char pyrf_cpu_map__doc[] = PyDoc_STR("cpu map object."); + +static PyTypeObject pyrf_cpu_map__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.cpu_map", + .tp_basicsize = sizeof(struct pyrf_cpu_map), + .tp_dealloc = (destructor)pyrf_cpu_map__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_cpu_map__doc, + .tp_as_sequence = &pyrf_cpu_map__sequence_methods, + .tp_init = (initproc)pyrf_cpu_map__init, +}; + +static int pyrf_cpu_map__setup_types(void) +{ + pyrf_cpu_map__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_cpu_map__type); +} + +struct pyrf_thread_map { + PyObject_HEAD + + struct thread_map *threads; +}; + +static int pyrf_thread_map__init(struct pyrf_thread_map *pthreads, + PyObject *args, PyObject *kwargs) +{ + static char *kwlist[] = { "pid", "tid", NULL, NULL, }; + int pid = -1, tid = -1; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ii", + kwlist, &pid, &tid)) + return -1; + + pthreads->threads = thread_map__new(pid, tid); + if (pthreads->threads == NULL) + return -1; + return 0; +} + +static void pyrf_thread_map__delete(struct pyrf_thread_map *pthreads) +{ + thread_map__delete(pthreads->threads); + pthreads->ob_type->tp_free((PyObject*)pthreads); +} + +static Py_ssize_t pyrf_thread_map__length(PyObject *obj) +{ + struct pyrf_thread_map *pthreads = (void *)obj; + + return pthreads->threads->nr; +} + +static PyObject *pyrf_thread_map__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_thread_map *pthreads = (void *)obj; + + if (i >= pthreads->threads->nr) + return NULL; + + return Py_BuildValue("i", pthreads->threads->map[i]); +} + +static PySequenceMethods pyrf_thread_map__sequence_methods = { + .sq_length = pyrf_thread_map__length, + .sq_item = pyrf_thread_map__item, +}; + +static char pyrf_thread_map__doc[] = PyDoc_STR("thread map object."); + +static PyTypeObject pyrf_thread_map__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.thread_map", + .tp_basicsize = sizeof(struct pyrf_thread_map), + .tp_dealloc = (destructor)pyrf_thread_map__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_thread_map__doc, + .tp_as_sequence = &pyrf_thread_map__sequence_methods, + .tp_init = (initproc)pyrf_thread_map__init, +}; + +static int pyrf_thread_map__setup_types(void) +{ + pyrf_thread_map__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_thread_map__type); +} + +struct pyrf_evsel { + PyObject_HEAD + + struct perf_evsel evsel; +}; + +static int pyrf_evsel__init(struct pyrf_evsel *pevsel, + PyObject *args, PyObject *kwargs) +{ + struct perf_event_attr attr = { + .type = PERF_TYPE_HARDWARE, + .config = PERF_COUNT_HW_CPU_CYCLES, + .sample_type = PERF_SAMPLE_PERIOD | PERF_SAMPLE_TID, + }; + static char *kwlist[] = { + "type", + "config", + "sample_freq", + "sample_period", + "sample_type", + "read_format", + "disabled", + "inherit", + "pinned", + "exclusive", + "exclude_user", + "exclude_kernel", + "exclude_hv", + "exclude_idle", + "mmap", + "comm", + "freq", + "inherit_stat", + "enable_on_exec", + "task", + "watermark", + "precise_ip", + "mmap_data", + "sample_id_all", + "wakeup_events", + "bp_type", + "bp_addr", + "bp_len", NULL, NULL, }; + u64 sample_period = 0; + u32 disabled = 0, + inherit = 0, + pinned = 0, + exclusive = 0, + exclude_user = 0, + exclude_kernel = 0, + exclude_hv = 0, + exclude_idle = 0, + mmap = 0, + comm = 0, + freq = 1, + inherit_stat = 0, + enable_on_exec = 0, + task = 0, + watermark = 0, + precise_ip = 0, + mmap_data = 0, + sample_id_all = 1; + int idx = 0; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, + "|iKiKKiiiiiiiiiiiiiiiiiiiiiKK", kwlist, + &attr.type, &attr.config, &attr.sample_freq, + &sample_period, &attr.sample_type, + &attr.read_format, &disabled, &inherit, + &pinned, &exclusive, &exclude_user, + &exclude_kernel, &exclude_hv, &exclude_idle, + &mmap, &comm, &freq, &inherit_stat, + &enable_on_exec, &task, &watermark, + &precise_ip, &mmap_data, &sample_id_all, + &attr.wakeup_events, &attr.bp_type, + &attr.bp_addr, &attr.bp_len, &idx)) + return -1; + + /* union... */ + if (sample_period != 0) { + if (attr.sample_freq != 0) + return -1; /* FIXME: throw right exception */ + attr.sample_period = sample_period; + } + + /* Bitfields */ + attr.disabled = disabled; + attr.inherit = inherit; + attr.pinned = pinned; + attr.exclusive = exclusive; + attr.exclude_user = exclude_user; + attr.exclude_kernel = exclude_kernel; + attr.exclude_hv = exclude_hv; + attr.exclude_idle = exclude_idle; + attr.mmap = mmap; + attr.comm = comm; + attr.freq = freq; + attr.inherit_stat = inherit_stat; + attr.enable_on_exec = enable_on_exec; + attr.task = task; + attr.watermark = watermark; + attr.precise_ip = precise_ip; + attr.mmap_data = mmap_data; + attr.sample_id_all = sample_id_all; + + perf_evsel__init(&pevsel->evsel, &attr, idx); + return 0; +} + +static void pyrf_evsel__delete(struct pyrf_evsel *pevsel) +{ + perf_evsel__exit(&pevsel->evsel); + pevsel->ob_type->tp_free((PyObject*)pevsel); +} + +static PyObject *pyrf_evsel__open(struct pyrf_evsel *pevsel, + PyObject *args, PyObject *kwargs) +{ + struct perf_evsel *evsel = &pevsel->evsel; + struct cpu_map *cpus = NULL; + struct thread_map *threads = NULL; + PyObject *pcpus = NULL, *pthreads = NULL; + int group = 0, overwrite = 0; + static char *kwlist[] = {"cpus", "threads", "group", "overwrite", NULL, NULL}; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|OOii", kwlist, + &pcpus, &pthreads, &group, &overwrite)) + return NULL; + + if (pthreads != NULL) + threads = ((struct pyrf_thread_map *)pthreads)->threads; + + if (pcpus != NULL) + cpus = ((struct pyrf_cpu_map *)pcpus)->cpus; + + if (perf_evsel__open(evsel, cpus, threads, group, overwrite) < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyMethodDef pyrf_evsel__methods[] = { + { + .ml_name = "open", + .ml_meth = (PyCFunction)pyrf_evsel__open, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("open the event selector file descriptor table.") + }, + { NULL, } +}; + +static char pyrf_evsel__doc[] = PyDoc_STR("perf event selector list object."); + +static PyTypeObject pyrf_evsel__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.evsel", + .tp_basicsize = sizeof(struct pyrf_evsel), + .tp_dealloc = (destructor)pyrf_evsel__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_evsel__doc, + .tp_methods = pyrf_evsel__methods, + .tp_init = (initproc)pyrf_evsel__init, +}; + +static int pyrf_evsel__setup_types(void) +{ + pyrf_evsel__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_evsel__type); +} + +struct pyrf_evlist { + PyObject_HEAD + + struct perf_evlist evlist; +}; + +static int pyrf_evlist__init(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + perf_evlist__init(&pevlist->evlist); + return 0; +} + +static void pyrf_evlist__delete(struct pyrf_evlist *pevlist) +{ + perf_evlist__exit(&pevlist->evlist); + pevlist->ob_type->tp_free((PyObject*)pevlist); +} + +static PyObject *pyrf_evlist__mmap(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + PyObject *pcpus = NULL, *pthreads = NULL; + struct cpu_map *cpus = NULL; + struct thread_map *threads = NULL; + static char *kwlist[] = {"cpus", "threads", "pages", "overwrite", + NULL, NULL}; + int pages = 128, overwrite = false; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "OO|ii", kwlist, + &pcpus, &pthreads, &pages, &overwrite)) + return NULL; + + threads = ((struct pyrf_thread_map *)pthreads)->threads; + cpus = ((struct pyrf_cpu_map *)pcpus)->cpus; + + if (perf_evlist__mmap(evlist, cpus, threads, pages, overwrite) < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyObject *pyrf_evlist__poll(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + static char *kwlist[] = {"timeout", NULL, NULL}; + int timeout = -1, n; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|i", kwlist, &timeout)) + return NULL; + + n = poll(evlist->pollfd, evlist->nr_fds, timeout); + if (n < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + return Py_BuildValue("i", n); +} + +static PyObject *pyrf_evlist__get_pollfd(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + PyObject *list = PyList_New(0); + int i; + + for (i = 0; i < evlist->nr_fds; ++i) { + PyObject *file; + FILE *fp = fdopen(evlist->pollfd[i].fd, "r"); + + if (fp == NULL) + goto free_list; + + file = PyFile_FromFile(fp, "perf", "r", NULL); + if (file == NULL) + goto free_list; + + if (PyList_Append(list, file) != 0) { + Py_DECREF(file); + goto free_list; + } + + Py_DECREF(file); + } + + return list; +free_list: + return PyErr_NoMemory(); +} + + +static PyObject *pyrf_evlist__add(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + PyObject *pevsel; + struct perf_evsel *evsel; + + if (!PyArg_ParseTuple(args, "O", &pevsel)) + return NULL; + + Py_INCREF(pevsel); + evsel = &((struct pyrf_evsel *)pevsel)->evsel; + evsel->idx = evlist->nr_entries; + perf_evlist__add(evlist, evsel); + + return Py_BuildValue("i", evlist->nr_entries); +} + +static PyObject *pyrf_evlist__read_on_cpu(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + union perf_event *event; + int sample_id_all = 1, cpu; + static char *kwlist[] = {"sample_id_all", NULL, NULL}; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "i|i", kwlist, + &cpu, &sample_id_all)) + return NULL; + + event = perf_evlist__read_on_cpu(evlist, cpu); + if (event != NULL) { + struct perf_evsel *first; + PyObject *pyevent = pyrf_event__new(event); + struct pyrf_event *pevent = (struct pyrf_event *)pyevent; + + if (pyevent == NULL) + return PyErr_NoMemory(); + + first = list_entry(evlist->entries.next, struct perf_evsel, node); + perf_event__parse_sample(event, first->attr.sample_type, sample_id_all, + &pevent->sample); + return pyevent; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyMethodDef pyrf_evlist__methods[] = { + { + .ml_name = "mmap", + .ml_meth = (PyCFunction)pyrf_evlist__mmap, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("mmap the file descriptor table.") + }, + { + .ml_name = "poll", + .ml_meth = (PyCFunction)pyrf_evlist__poll, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("poll the file descriptor table.") + }, + { + .ml_name = "get_pollfd", + .ml_meth = (PyCFunction)pyrf_evlist__get_pollfd, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("get the poll file descriptor table.") + }, + { + .ml_name = "add", + .ml_meth = (PyCFunction)pyrf_evlist__add, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("adds an event selector to the list.") + }, + { + .ml_name = "read_on_cpu", + .ml_meth = (PyCFunction)pyrf_evlist__read_on_cpu, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("reads an event.") + }, + { NULL, } +}; + +static Py_ssize_t pyrf_evlist__length(PyObject *obj) +{ + struct pyrf_evlist *pevlist = (void *)obj; + + return pevlist->evlist.nr_entries; +} + +static PyObject *pyrf_evlist__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_evlist *pevlist = (void *)obj; + struct perf_evsel *pos; + + if (i >= pevlist->evlist.nr_entries) + return NULL; + + list_for_each_entry(pos, &pevlist->evlist.entries, node) + if (i-- == 0) + break; + + return Py_BuildValue("O", container_of(pos, struct pyrf_evsel, evsel)); +} + +static PySequenceMethods pyrf_evlist__sequence_methods = { + .sq_length = pyrf_evlist__length, + .sq_item = pyrf_evlist__item, +}; + +static char pyrf_evlist__doc[] = PyDoc_STR("perf event selector list object."); + +static PyTypeObject pyrf_evlist__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.evlist", + .tp_basicsize = sizeof(struct pyrf_evlist), + .tp_dealloc = (destructor)pyrf_evlist__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_as_sequence = &pyrf_evlist__sequence_methods, + .tp_doc = pyrf_evlist__doc, + .tp_methods = pyrf_evlist__methods, + .tp_init = (initproc)pyrf_evlist__init, +}; + +static int pyrf_evlist__setup_types(void) +{ + pyrf_evlist__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_evlist__type); +} + +static struct { + const char *name; + int value; +} perf__constants[] = { + { "TYPE_HARDWARE", PERF_TYPE_HARDWARE }, + { "TYPE_SOFTWARE", PERF_TYPE_SOFTWARE }, + { "TYPE_TRACEPOINT", PERF_TYPE_TRACEPOINT }, + { "TYPE_HW_CACHE", PERF_TYPE_HW_CACHE }, + { "TYPE_RAW", PERF_TYPE_RAW }, + { "TYPE_BREAKPOINT", PERF_TYPE_BREAKPOINT }, + + { "COUNT_HW_CPU_CYCLES", PERF_COUNT_HW_CPU_CYCLES }, + { "COUNT_HW_INSTRUCTIONS", PERF_COUNT_HW_INSTRUCTIONS }, + { "COUNT_HW_CACHE_REFERENCES", PERF_COUNT_HW_CACHE_REFERENCES }, + { "COUNT_HW_CACHE_MISSES", PERF_COUNT_HW_CACHE_MISSES }, + { "COUNT_HW_BRANCH_INSTRUCTIONS", PERF_COUNT_HW_BRANCH_INSTRUCTIONS }, + { "COUNT_HW_BRANCH_MISSES", PERF_COUNT_HW_BRANCH_MISSES }, + { "COUNT_HW_BUS_CYCLES", PERF_COUNT_HW_BUS_CYCLES }, + { "COUNT_HW_CACHE_L1D", PERF_COUNT_HW_CACHE_L1D }, + { "COUNT_HW_CACHE_L1I", PERF_COUNT_HW_CACHE_L1I }, + { "COUNT_HW_CACHE_LL", PERF_COUNT_HW_CACHE_LL }, + { "COUNT_HW_CACHE_DTLB", PERF_COUNT_HW_CACHE_DTLB }, + { "COUNT_HW_CACHE_ITLB", PERF_COUNT_HW_CACHE_ITLB }, + { "COUNT_HW_CACHE_BPU", PERF_COUNT_HW_CACHE_BPU }, + { "COUNT_HW_CACHE_OP_READ", PERF_COUNT_HW_CACHE_OP_READ }, + { "COUNT_HW_CACHE_OP_WRITE", PERF_COUNT_HW_CACHE_OP_WRITE }, + { "COUNT_HW_CACHE_OP_PREFETCH", PERF_COUNT_HW_CACHE_OP_PREFETCH }, + { "COUNT_HW_CACHE_RESULT_ACCESS", PERF_COUNT_HW_CACHE_RESULT_ACCESS }, + { "COUNT_HW_CACHE_RESULT_MISS", PERF_COUNT_HW_CACHE_RESULT_MISS }, + + { "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 }, + { "COUNT_SW_CONTEXT_SWITCHES", PERF_COUNT_SW_CONTEXT_SWITCHES }, + { "COUNT_SW_CPU_MIGRATIONS", PERF_COUNT_SW_CPU_MIGRATIONS }, + { "COUNT_SW_PAGE_FAULTS_MIN", PERF_COUNT_SW_PAGE_FAULTS_MIN }, + { "COUNT_SW_PAGE_FAULTS_MAJ", PERF_COUNT_SW_PAGE_FAULTS_MAJ }, + { "COUNT_SW_ALIGNMENT_FAULTS", PERF_COUNT_SW_ALIGNMENT_FAULTS }, + { "COUNT_SW_EMULATION_FAULTS", PERF_COUNT_SW_EMULATION_FAULTS }, + + { "SAMPLE_IP", PERF_SAMPLE_IP }, + { "SAMPLE_TID", PERF_SAMPLE_TID }, + { "SAMPLE_TIME", PERF_SAMPLE_TIME }, + { "SAMPLE_ADDR", PERF_SAMPLE_ADDR }, + { "SAMPLE_READ", PERF_SAMPLE_READ }, + { "SAMPLE_CALLCHAIN", PERF_SAMPLE_CALLCHAIN }, + { "SAMPLE_ID", PERF_SAMPLE_ID }, + { "SAMPLE_CPU", PERF_SAMPLE_CPU }, + { "SAMPLE_PERIOD", PERF_SAMPLE_PERIOD }, + { "SAMPLE_STREAM_ID", PERF_SAMPLE_STREAM_ID }, + { "SAMPLE_RAW", PERF_SAMPLE_RAW }, + + { "FORMAT_TOTAL_TIME_ENABLED", PERF_FORMAT_TOTAL_TIME_ENABLED }, + { "FORMAT_TOTAL_TIME_RUNNING", PERF_FORMAT_TOTAL_TIME_RUNNING }, + { "FORMAT_ID", PERF_FORMAT_ID }, + { "FORMAT_GROUP", PERF_FORMAT_GROUP }, + + { "RECORD_MMAP", PERF_RECORD_MMAP }, + { "RECORD_LOST", PERF_RECORD_LOST }, + { "RECORD_COMM", PERF_RECORD_COMM }, + { "RECORD_EXIT", PERF_RECORD_EXIT }, + { "RECORD_THROTTLE", PERF_RECORD_THROTTLE }, + { "RECORD_UNTHROTTLE", PERF_RECORD_UNTHROTTLE }, + { "RECORD_FORK", PERF_RECORD_FORK }, + { "RECORD_READ", PERF_RECORD_READ }, + { "RECORD_SAMPLE", PERF_RECORD_SAMPLE }, + { NULL, }, +}; + +static PyMethodDef perf__methods[] = { + { NULL, NULL } +}; + +PyMODINIT_FUNC initperf(void) +{ + PyObject *obj; + int i; + PyObject *dict, *module = Py_InitModule("perf", perf__methods); + + if (module == NULL || + pyrf_event__setup_types() < 0 || + pyrf_evlist__setup_types() < 0 || + pyrf_evsel__setup_types() < 0 || + pyrf_thread_map__setup_types() < 0 || + pyrf_cpu_map__setup_types() < 0) + return; + + Py_INCREF(&pyrf_evlist__type); + PyModule_AddObject(module, "evlist", (PyObject*)&pyrf_evlist__type); + + Py_INCREF(&pyrf_evsel__type); + PyModule_AddObject(module, "evsel", (PyObject*)&pyrf_evsel__type); + + Py_INCREF(&pyrf_thread_map__type); + PyModule_AddObject(module, "thread_map", (PyObject*)&pyrf_thread_map__type); + + Py_INCREF(&pyrf_cpu_map__type); + PyModule_AddObject(module, "cpu_map", (PyObject*)&pyrf_cpu_map__type); + + dict = PyModule_GetDict(module); + if (dict == NULL) + goto error; + + for (i = 0; perf__constants[i].name != NULL; i++) { + obj = PyInt_FromLong(perf__constants[i].value); + if (obj == NULL) + goto error; + PyDict_SetItemString(dict, perf__constants[i].name, obj); + Py_DECREF(obj); + } + +error: + if (PyErr_Occurred()) + PyErr_SetString(PyExc_ImportError, "perf: Init failed!"); +} diff --git a/tools/perf/util/setup.py b/tools/perf/util/setup.py new file mode 100644 index 000000000000..496d7f432fb8 --- /dev/null +++ b/tools/perf/util/setup.py @@ -0,0 +1,18 @@ +#!/usr/bin/python2 + +from distutils.core import setup, Extension + +perf = Extension('perf', + sources = ['util/python.c', 'util/ctype.c', 'util/evlist.c', + 'util/evsel.c', 'util/cpumap.c', 'util/thread_map.c', + 'util/util.c', 'util/xyarray.c'], + include_dirs = ['util/include']) + +setup(name='perf', + version='0.1', + description='Interface with the Linux profiling infrastructure', + author='Arnaldo Carvalho de Melo', + author_email='acme@redhat.com', + license='GPLv2', + url='http://perf.wiki.kernel.org', + ext_modules=[perf]) -- cgit v1.2.3 From f8a9530939ed87b9a1b1a038b90e355098b679a2 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sun, 30 Jan 2011 10:46:46 -0200 Subject: perf evlist: Move evlist methods to evlist.c They were on evsel.c because they came from refactoring existing evsel methods, so, to make reviewing the changes easier, I kept it there, now its a plain move. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 142 ++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/evlist.h | 7 +++ tools/perf/util/evsel.c | 144 +++-------------------------------------------- tools/perf/util/evsel.h | 4 -- 4 files changed, 158 insertions(+), 139 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 917fc18d0bed..dcd59328bb49 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -1,11 +1,26 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ #include +#include "cpumap.h" +#include "thread_map.h" #include "evlist.h" #include "evsel.h" #include "util.h" +#include + #include #include +#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) +#define SID(e, x, y) xyarray__entry(e->id, x, y) + void perf_evlist__init(struct perf_evlist *evlist) { int i; @@ -88,6 +103,30 @@ void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd) evlist->nr_fds++; } +static int perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, int fd) +{ + struct perf_sample_id *sid; + u64 read_data[4] = { 0, }; + int hash, id_idx = 1; /* The first entry is the counter value */ + + if (!(evsel->attr.read_format & PERF_FORMAT_ID) || + read(fd, &read_data, sizeof(read_data)) == -1) + return -1; + + if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) + ++id_idx; + if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) + ++id_idx; + + sid = SID(evsel, cpu, thread); + sid->id = read_data[id_idx]; + sid->evsel = evsel; + hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS); + hlist_add_head(&sid->node, &evlist->heads[hash]); + return 0; +} + struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id) { struct hlist_head *head; @@ -173,3 +212,106 @@ union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) return event; } + +void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus) +{ + int cpu; + + for (cpu = 0; cpu < ncpus; cpu++) { + if (evlist->mmap[cpu].base != NULL) { + munmap(evlist->mmap[cpu].base, evlist->mmap_len); + evlist->mmap[cpu].base = NULL; + } + } +} + +int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus) +{ + evlist->mmap = zalloc(ncpus * sizeof(struct perf_mmap)); + return evlist->mmap != NULL ? 0 : -ENOMEM; +} + +static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot, + int mask, int fd) +{ + evlist->mmap[cpu].prev = 0; + evlist->mmap[cpu].mask = mask; + evlist->mmap[cpu].base = mmap(NULL, evlist->mmap_len, prot, + MAP_SHARED, fd, 0); + if (evlist->mmap[cpu].base == MAP_FAILED) + return -1; + + perf_evlist__add_pollfd(evlist, fd); + return 0; +} + +/** perf_evlist__mmap - Create per cpu maps to receive events + * + * @evlist - list of events + * @cpus - cpu map being monitored + * @threads - threads map being monitored + * @pages - map length in pages + * @overwrite - overwrite older events? + * + * If overwrite is false the user needs to signal event consuption using: + * + * struct perf_mmap *m = &evlist->mmap[cpu]; + * unsigned int head = perf_mmap__read_head(m); + * + * perf_mmap__write_tail(m, head) + */ +int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads, int pages, bool overwrite) +{ + unsigned int page_size = sysconf(_SC_PAGE_SIZE); + int mask = pages * page_size - 1, cpu; + struct perf_evsel *first_evsel, *evsel; + int thread, prot = PROT_READ | (overwrite ? 0 : PROT_WRITE); + + if (evlist->mmap == NULL && + perf_evlist__alloc_mmap(evlist, cpus->nr) < 0) + return -ENOMEM; + + if (evlist->pollfd == NULL && + perf_evlist__alloc_pollfd(evlist, cpus->nr, threads->nr) < 0) + return -ENOMEM; + + evlist->overwrite = overwrite; + evlist->mmap_len = (pages + 1) * page_size; + first_evsel = list_entry(evlist->entries.next, struct perf_evsel, node); + + list_for_each_entry(evsel, &evlist->entries, node) { + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + evsel->id == NULL && + perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0) + return -ENOMEM; + + for (cpu = 0; cpu < cpus->nr; cpu++) { + for (thread = 0; thread < threads->nr; thread++) { + int fd = FD(evsel, cpu, thread); + + if (evsel->idx || thread) { + if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, + FD(first_evsel, cpu, 0)) != 0) + goto out_unmap; + } else if (__perf_evlist__mmap(evlist, cpu, prot, mask, fd) < 0) + goto out_unmap; + + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + perf_evlist__id_hash(evlist, evsel, cpu, thread, fd) < 0) + goto out_unmap; + } + } + } + + return 0; + +out_unmap: + for (cpu = 0; cpu < cpus->nr; cpu++) { + if (evlist->mmap[cpu].base != NULL) { + munmap(evlist->mmap[cpu].base, evlist->mmap_len); + evlist->mmap[cpu].base = NULL; + } + } + return -1; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 022ae404b908..85aca6eba16b 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -6,6 +6,8 @@ #include "event.h" struct pollfd; +struct thread_map; +struct cpu_map; #define PERF_EVLIST__HLIST_BITS 8 #define PERF_EVLIST__HLIST_SIZE (1 << PERF_EVLIST__HLIST_BITS) @@ -39,4 +41,9 @@ struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id); union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *self, int cpu); +int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus); +int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads, int pages, bool overwrite); +void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus); + #endif /* __PERF_EVLIST_H */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index fddeb08f48a7..2720bc1d578b 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -1,18 +1,19 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + #include "evsel.h" #include "evlist.h" -#include "../perf.h" #include "util.h" #include "cpumap.h" #include "thread_map.h" -#include -#include - -#include -#include - #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) -#define SID(e, x, y) xyarray__entry(e->id, x, y) void perf_evsel__init(struct perf_evsel *evsel, struct perf_event_attr *attr, int idx) @@ -74,24 +75,6 @@ void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads) } } -void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus) -{ - int cpu; - - for (cpu = 0; cpu < ncpus; cpu++) { - if (evlist->mmap[cpu].base != NULL) { - munmap(evlist->mmap[cpu].base, evlist->mmap_len); - evlist->mmap[cpu].base = NULL; - } - } -} - -int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus) -{ - evlist->mmap = zalloc(ncpus * sizeof(struct perf_mmap)); - return evlist->mmap != NULL ? 0 : -ENOMEM; -} - void perf_evsel__exit(struct perf_evsel *evsel) { assert(list_empty(&evsel->node)); @@ -258,115 +241,6 @@ int perf_evsel__open_per_thread(struct perf_evsel *evsel, return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group, inherit); } -static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot, - int mask, int fd) -{ - evlist->mmap[cpu].prev = 0; - evlist->mmap[cpu].mask = mask; - evlist->mmap[cpu].base = mmap(NULL, evlist->mmap_len, prot, - MAP_SHARED, fd, 0); - if (evlist->mmap[cpu].base == MAP_FAILED) - return -1; - - perf_evlist__add_pollfd(evlist, fd); - return 0; -} - -static int perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, - int cpu, int thread, int fd) -{ - struct perf_sample_id *sid; - u64 read_data[4] = { 0, }; - int hash, id_idx = 1; /* The first entry is the counter value */ - - if (!(evsel->attr.read_format & PERF_FORMAT_ID) || - read(fd, &read_data, sizeof(read_data)) == -1) - return -1; - - if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) - ++id_idx; - if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) - ++id_idx; - - sid = SID(evsel, cpu, thread); - sid->id = read_data[id_idx]; - sid->evsel = evsel; - hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS); - hlist_add_head(&sid->node, &evlist->heads[hash]); - return 0; -} - -/** perf_evlist__mmap - Create per cpu maps to receive events - * - * @evlist - list of events - * @cpus - cpu map being monitored - * @threads - threads map being monitored - * @pages - map length in pages - * @overwrite - overwrite older events? - * - * If overwrite is false the user needs to signal event consuption using: - * - * struct perf_mmap *m = &evlist->mmap[cpu]; - * unsigned int head = perf_mmap__read_head(m); - * - * perf_mmap__write_tail(m, head) - */ -int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, - struct thread_map *threads, int pages, bool overwrite) -{ - unsigned int page_size = sysconf(_SC_PAGE_SIZE); - int mask = pages * page_size - 1, cpu; - struct perf_evsel *first_evsel, *evsel; - int thread, prot = PROT_READ | (overwrite ? 0 : PROT_WRITE); - - if (evlist->mmap == NULL && - perf_evlist__alloc_mmap(evlist, cpus->nr) < 0) - return -ENOMEM; - - if (evlist->pollfd == NULL && - perf_evlist__alloc_pollfd(evlist, cpus->nr, threads->nr) < 0) - return -ENOMEM; - - evlist->overwrite = overwrite; - evlist->mmap_len = (pages + 1) * page_size; - first_evsel = list_entry(evlist->entries.next, struct perf_evsel, node); - - list_for_each_entry(evsel, &evlist->entries, node) { - if ((evsel->attr.read_format & PERF_FORMAT_ID) && - evsel->id == NULL && - perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0) - return -ENOMEM; - - for (cpu = 0; cpu < cpus->nr; cpu++) { - for (thread = 0; thread < threads->nr; thread++) { - int fd = FD(evsel, cpu, thread); - - if (evsel->idx || thread) { - if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, - FD(first_evsel, cpu, 0)) != 0) - goto out_unmap; - } else if (__perf_evlist__mmap(evlist, cpu, prot, mask, fd) < 0) - goto out_unmap; - - if ((evsel->attr.read_format & PERF_FORMAT_ID) && - perf_evlist__id_hash(evlist, evsel, cpu, thread, fd) < 0) - goto out_unmap; - } - } - } - - return 0; - -out_unmap: - for (cpu = 0; cpu < cpus->nr; cpu++) { - if (evlist->mmap[cpu].base != NULL) { - munmap(evlist->mmap[cpu].base, evlist->mmap_len); - evlist->mmap[cpu].base = NULL; - } - } - return -1; -} - static int perf_event__parse_id_sample(const union perf_event *event, u64 type, struct perf_sample *sample) { diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 7962e7587dea..eecdc3aabc14 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -60,7 +60,6 @@ void perf_evsel__delete(struct perf_evsel *evsel); int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads); int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads); int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus); -int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus); void perf_evsel__free_fd(struct perf_evsel *evsel); void perf_evsel__free_id(struct perf_evsel *evsel); void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads); @@ -71,9 +70,6 @@ int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *threads, bool group, bool inherit); int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, struct thread_map *threads, bool group, bool inherit); -int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, - struct thread_map *threads, int pages, bool overwrite); -void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus); #define perf_evsel__match(evsel, t, c) \ (evsel->attr.type == PERF_TYPE_##t && \ -- cgit v1.2.3 From 7e2ed097538c57ff5268e9a6bced7c0b885809c8 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sun, 30 Jan 2011 11:59:43 -0200 Subject: perf evlist: Store pointer to the cpu and thread maps So that we don't have to pass it around to the several methods that needs it, simplifying usage. There is one case where we don't have the thread/cpu map in advance, which is in the parsing routines used by top, stat, record, that we have to wait till all options are parsed to know if a cpu or thread list was passed to then create those maps. For that case consolidate the cpu and thread map creation via perf_evlist__create_maps() out of the code in top and record, while also providing a perf_evlist__set_maps() for cases where multiple evlists share maps or for when maps that represent CPU sockets, for instance, get crafted out of topology information or subsets of threads in a particular application are to be monitored, providing more granularity in specifying which cpus and threads to monitor. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-record.c | 44 +++++++++++------------------ tools/perf/builtin-stat.c | 45 +++++++++++++++--------------- tools/perf/builtin-test.c | 6 ++-- tools/perf/builtin-top.c | 47 +++++++++++++------------------ tools/perf/python/twatch.py | 4 +-- tools/perf/util/evlist.c | 67 +++++++++++++++++++++++++++++++++------------ tools/perf/util/evlist.h | 29 +++++++++++++++----- tools/perf/util/python.c | 25 +++++++++-------- 8 files changed, 148 insertions(+), 119 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index edc3555098c8..07f8d6d852c2 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -42,7 +42,6 @@ static u64 user_interval = ULLONG_MAX; static u64 default_interval = 0; static u64 sample_type; -static struct cpu_map *cpus; static unsigned int page_size; static unsigned int mmap_pages = 128; static unsigned int user_freq = UINT_MAX; @@ -58,7 +57,6 @@ static bool sample_id_all_avail = true; static bool system_wide = false; static pid_t target_pid = -1; static pid_t target_tid = -1; -static struct thread_map *threads; static pid_t child_pid = -1; static bool no_inherit = false; static enum write_mode_t write_mode = WRITE_FORCE; @@ -189,7 +187,7 @@ static void create_counter(struct perf_evsel *evsel, int cpu) int thread_index; int ret; - for (thread_index = 0; thread_index < threads->nr; thread_index++) { + for (thread_index = 0; thread_index < evsel_list->threads->nr; thread_index++) { h_attr = get_header_attr(attr, evsel->idx); if (h_attr == NULL) die("nomem\n"); @@ -317,7 +315,8 @@ static void open_counters(struct perf_evlist *evlist) retry_sample_id: attr->sample_id_all = sample_id_all_avail ? 1 : 0; try_again: - if (perf_evsel__open(pos, cpus, threads, group, !no_inherit) < 0) { + if (perf_evsel__open(pos, evlist->cpus, evlist->threads, group, + !no_inherit) < 0) { int err = errno; if (err == EPERM || err == EACCES) @@ -368,10 +367,10 @@ try_again: } } - if (perf_evlist__mmap(evlist, cpus, threads, mmap_pages, false) < 0) + if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) die("failed to mmap with %d (%s)\n", errno, strerror(errno)); - for (cpu = 0; cpu < cpus->nr; ++cpu) { + for (cpu = 0; cpu < evsel_list->cpus->nr; ++cpu) { list_for_each_entry(pos, &evlist->entries, node) create_counter(pos, cpu); } @@ -450,7 +449,7 @@ static void mmap_read_all(void) { int i; - for (i = 0; i < cpus->nr; i++) { + for (i = 0; i < evsel_list->cpus->nr; i++) { if (evsel_list->mmap[i].base) mmap_read(&evsel_list->mmap[i]); } @@ -584,7 +583,7 @@ static int __cmd_record(int argc, const char **argv) } if (!system_wide && target_tid == -1 && target_pid == -1) - threads->map[0] = child_pid; + evsel_list->threads->map[0] = child_pid; close(child_ready_pipe[1]); close(go_pipe[0]); @@ -718,12 +717,12 @@ static int __cmd_record(int argc, const char **argv) } if (done) { - for (i = 0; i < cpus->nr; i++) { + for (i = 0; i < evsel_list->cpus->nr; i++) { struct perf_evsel *pos; list_for_each_entry(pos, &evsel_list->entries, node) { for (thread = 0; - thread < threads->nr; + thread < evsel_list->threads->nr; thread++) ioctl(FD(pos, i, thread), PERF_EVENT_IOC_DISABLE); @@ -816,7 +815,7 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) int err = -ENOMEM; struct perf_evsel *pos; - evsel_list = perf_evlist__new(); + evsel_list = perf_evlist__new(NULL, NULL); if (evsel_list == NULL) return -ENOMEM; @@ -850,28 +849,19 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) if (target_pid != -1) target_tid = target_pid; - threads = thread_map__new(target_pid, target_tid); - if (threads == NULL) { - pr_err("Problems finding threads of monitor\n"); - usage_with_options(record_usage, record_options); - } - - if (target_tid != -1) - cpus = cpu_map__dummy_new(); - else - cpus = cpu_map__new(cpu_list); - - if (cpus == NULL) + if (perf_evlist__create_maps(evsel_list, target_pid, + target_tid, cpu_list) < 0) usage_with_options(record_usage, record_options); list_for_each_entry(pos, &evsel_list->entries, node) { - if (perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) + if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, + evsel_list->threads->nr) < 0) goto out_free_fd; if (perf_header__push_event(pos->attr.config, event_name(pos))) goto out_free_fd; } - if (perf_evlist__alloc_pollfd(evsel_list, cpus->nr, threads->nr) < 0) + if (perf_evlist__alloc_pollfd(evsel_list) < 0) goto out_free_fd; if (user_interval != ULLONG_MAX) @@ -893,10 +883,8 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) } err = __cmd_record(argc, argv); - out_free_fd: - thread_map__delete(threads); - threads = NULL; + perf_evlist__delete_maps(evsel_list); out_symbol_exit: symbol__exit(); return err; diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index 8906adfdbd8e..e0f95755361b 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -76,7 +76,6 @@ static struct perf_event_attr default_attrs[] = { struct perf_evlist *evsel_list; static bool system_wide = false; -static struct cpu_map *cpus; static int run_idx = 0; static int run_count = 1; @@ -85,7 +84,6 @@ static bool scale = true; static bool no_aggr = false; static pid_t target_pid = -1; static pid_t target_tid = -1; -static struct thread_map *threads; static pid_t child_pid = -1; static bool null_run = false; static bool big_num = true; @@ -170,7 +168,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel) PERF_FORMAT_TOTAL_TIME_RUNNING; if (system_wide) - return perf_evsel__open_per_cpu(evsel, cpus, false, false); + return perf_evsel__open_per_cpu(evsel, evsel_list->cpus, false, false); attr->inherit = !no_inherit; if (target_pid == -1 && target_tid == -1) { @@ -178,7 +176,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel) attr->enable_on_exec = 1; } - return perf_evsel__open_per_thread(evsel, threads, false, false); + return perf_evsel__open_per_thread(evsel, evsel_list->threads, false, false); } /* @@ -203,7 +201,8 @@ static int read_counter_aggr(struct perf_evsel *counter) u64 *count = counter->counts->aggr.values; int i; - if (__perf_evsel__read(counter, cpus->nr, threads->nr, scale) < 0) + if (__perf_evsel__read(counter, evsel_list->cpus->nr, + evsel_list->threads->nr, scale) < 0) return -1; for (i = 0; i < 3; i++) @@ -236,7 +235,7 @@ static int read_counter(struct perf_evsel *counter) u64 *count; int cpu; - for (cpu = 0; cpu < cpus->nr; cpu++) { + for (cpu = 0; cpu < evsel_list->cpus->nr; cpu++) { if (__perf_evsel__read_on_cpu(counter, cpu, 0, scale) < 0) return -1; @@ -301,7 +300,7 @@ static int run_perf_stat(int argc __used, const char **argv) } if (target_tid == -1 && target_pid == -1 && !system_wide) - threads->map[0] = child_pid; + evsel_list->threads->map[0] = child_pid; /* * Wait for the child to be ready to exec. @@ -353,12 +352,13 @@ static int run_perf_stat(int argc __used, const char **argv) if (no_aggr) { list_for_each_entry(counter, &evsel_list->entries, node) { read_counter(counter); - perf_evsel__close_fd(counter, cpus->nr, 1); + perf_evsel__close_fd(counter, evsel_list->cpus->nr, 1); } } else { list_for_each_entry(counter, &evsel_list->entries, node) { read_counter_aggr(counter); - perf_evsel__close_fd(counter, cpus->nr, threads->nr); + perf_evsel__close_fd(counter, evsel_list->cpus->nr, + evsel_list->threads->nr); } } @@ -386,7 +386,7 @@ static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg) if (no_aggr) sprintf(cpustr, "CPU%*d%s", csv_output ? 0 : -4, - cpus->map[cpu], csv_sep); + evsel_list->cpus->map[cpu], csv_sep); fprintf(stderr, fmt, cpustr, msecs, csv_sep, event_name(evsel)); @@ -414,7 +414,7 @@ static void abs_printout(int cpu, struct perf_evsel *evsel, double avg) if (no_aggr) sprintf(cpustr, "CPU%*d%s", csv_output ? 0 : -4, - cpus->map[cpu], csv_sep); + evsel_list->cpus->map[cpu], csv_sep); else cpu = 0; @@ -500,14 +500,14 @@ static void print_counter(struct perf_evsel *counter) u64 ena, run, val; int cpu; - for (cpu = 0; cpu < cpus->nr; cpu++) { + for (cpu = 0; cpu < evsel_list->cpus->nr; cpu++) { val = counter->counts->cpu[cpu].val; ena = counter->counts->cpu[cpu].ena; run = counter->counts->cpu[cpu].run; if (run == 0 || ena == 0) { fprintf(stderr, "CPU%*d%s%*s%s%-24s", csv_output ? 0 : -4, - cpus->map[cpu], csv_sep, + evsel_list->cpus->map[cpu], csv_sep, csv_output ? 0 : 18, "", csv_sep, event_name(counter)); @@ -652,7 +652,7 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) setlocale(LC_ALL, ""); - evsel_list = perf_evlist__new(); + evsel_list = perf_evlist__new(NULL, NULL); if (evsel_list == NULL) return -ENOMEM; @@ -701,18 +701,18 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) if (target_pid != -1) target_tid = target_pid; - threads = thread_map__new(target_pid, target_tid); - if (threads == NULL) { + evsel_list->threads = thread_map__new(target_pid, target_tid); + if (evsel_list->threads == NULL) { pr_err("Problems finding threads of monitor\n"); usage_with_options(stat_usage, options); } if (system_wide) - cpus = cpu_map__new(cpu_list); + evsel_list->cpus = cpu_map__new(cpu_list); else - cpus = cpu_map__dummy_new(); + evsel_list->cpus = cpu_map__dummy_new(); - if (cpus == NULL) { + if (evsel_list->cpus == NULL) { perror("failed to parse CPUs map"); usage_with_options(stat_usage, options); return -1; @@ -720,8 +720,8 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) list_for_each_entry(pos, &evsel_list->entries, node) { if (perf_evsel__alloc_stat_priv(pos) < 0 || - perf_evsel__alloc_counts(pos, cpus->nr) < 0 || - perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) + perf_evsel__alloc_counts(pos, evsel_list->cpus->nr) < 0 || + perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, evsel_list->threads->nr) < 0) goto out_free_fd; } @@ -750,7 +750,6 @@ out_free_fd: perf_evsel__free_stat_priv(pos); perf_evlist__delete(evsel_list); out: - thread_map__delete(threads); - threads = NULL; + perf_evlist__delete_maps(evsel_list); return status; } diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index 845b9bd54ed4..1b2106c58f66 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -509,7 +509,7 @@ static int test__basic_mmap(void) goto out_free_cpus; } - evlist = perf_evlist__new(); + evlist = perf_evlist__new(cpus, threads); if (evlist == NULL) { pr_debug("perf_evlist__new\n"); goto out_free_cpus; @@ -537,7 +537,7 @@ static int test__basic_mmap(void) } } - if (perf_evlist__mmap(evlist, cpus, threads, 128, true) < 0) { + if (perf_evlist__mmap(evlist, 128, true) < 0) { pr_debug("failed to mmap events: %d (%s)\n", errno, strerror(errno)); goto out_close_fd; @@ -579,7 +579,7 @@ static int test__basic_mmap(void) err = 0; out_munmap: - perf_evlist__munmap(evlist, 1); + perf_evlist__munmap(evlist); out_close_fd: for (i = 0; i < nsyscalls; ++i) perf_evsel__close_fd(evsels[i], 1, threads->nr); diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 2f4d1f244be1..599036b06730 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -73,9 +73,7 @@ static int print_entries; static int target_pid = -1; static int target_tid = -1; -static struct thread_map *threads; static bool inherit = false; -static struct cpu_map *cpus; static int realtime_prio = 0; static bool group = false; static unsigned int page_size; @@ -567,12 +565,13 @@ static void print_sym_table(struct perf_session *session) printf(" (all"); if (cpu_list) - printf(", CPU%s: %s)\n", cpus->nr > 1 ? "s" : "", cpu_list); + printf(", CPU%s: %s)\n", evsel_list->cpus->nr > 1 ? "s" : "", cpu_list); else { if (target_tid != -1) printf(")\n"); else - printf(", %d CPU%s)\n", cpus->nr, cpus->nr > 1 ? "s" : ""); + printf(", %d CPU%s)\n", evsel_list->cpus->nr, + evsel_list->cpus->nr > 1 ? "s" : ""); } printf("%-*.*s\n", win_width, win_width, graph_dotted_line); @@ -1124,7 +1123,7 @@ static void perf_session__mmap_read(struct perf_session *self) { int i; - for (i = 0; i < cpus->nr; i++) + for (i = 0; i < evsel_list->cpus->nr; i++) perf_session__mmap_read_cpu(self, i); } @@ -1150,7 +1149,8 @@ static void start_counters(struct perf_evlist *evlist) attr->mmap = 1; try_again: - if (perf_evsel__open(counter, cpus, threads, group, inherit) < 0) { + if (perf_evsel__open(counter, evsel_list->cpus, + evsel_list->threads, group, inherit) < 0) { int err = errno; if (err == EPERM || err == EACCES) @@ -1181,7 +1181,7 @@ try_again: } } - if (perf_evlist__mmap(evlist, cpus, threads, mmap_pages, false) < 0) + if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) die("failed to mmap with %d (%s)\n", errno, strerror(errno)); } @@ -1296,7 +1296,7 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) struct perf_evsel *pos; int status = -ENOMEM; - evsel_list = perf_evlist__new(); + evsel_list = perf_evlist__new(NULL, NULL); if (evsel_list == NULL) return -ENOMEM; @@ -1306,15 +1306,6 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) if (argc) usage_with_options(top_usage, options); - if (target_pid != -1) - target_tid = target_pid; - - threads = thread_map__new(target_pid, target_tid); - if (threads == NULL) { - pr_err("Problems finding threads of monitor\n"); - usage_with_options(top_usage, options); - } - /* CPU and PID are mutually exclusive */ if (target_tid > 0 && cpu_list) { printf("WARNING: PID switch overriding CPU\n"); @@ -1322,6 +1313,13 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) cpu_list = NULL; } + if (target_pid != -1) + target_tid = target_pid; + + if (perf_evlist__create_maps(evsel_list, target_pid, + target_tid, cpu_list) < 0) + usage_with_options(top_usage, options); + if (!evsel_list->nr_entries && perf_evlist__add_default(evsel_list) < 0) { pr_err("Not enough memory for event selector list\n"); @@ -1343,16 +1341,9 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) exit(EXIT_FAILURE); } - if (target_tid != -1) - cpus = cpu_map__dummy_new(); - else - cpus = cpu_map__new(cpu_list); - - if (cpus == NULL) - usage_with_options(top_usage, options); - list_for_each_entry(pos, &evsel_list->entries, node) { - if (perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) + if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, + evsel_list->threads->nr) < 0) goto out_free_fd; /* * Fill in the ones not specifically initialized via -c: @@ -1363,8 +1354,8 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) pos->attr.sample_period = default_interval; } - if (perf_evlist__alloc_pollfd(evsel_list, cpus->nr, threads->nr) < 0 || - perf_evlist__alloc_mmap(evsel_list, cpus->nr) < 0) + if (perf_evlist__alloc_pollfd(evsel_list) < 0 || + perf_evlist__alloc_mmap(evsel_list) < 0) goto out_free_fd; sym_evsel = list_entry(evsel_list->entries.next, struct perf_evsel, node); diff --git a/tools/perf/python/twatch.py b/tools/perf/python/twatch.py index 5e9f3b7b7ee8..df638c438a9f 100755 --- a/tools/perf/python/twatch.py +++ b/tools/perf/python/twatch.py @@ -23,9 +23,9 @@ def main(): sample_id_all = 1, sample_type = perf.SAMPLE_PERIOD | perf.SAMPLE_TID | perf.SAMPLE_CPU | perf.SAMPLE_TID) evsel.open(cpus = cpus, threads = threads); - evlist = perf.evlist() + evlist = perf.evlist(cpus, threads) evlist.add(evsel) - evlist.mmap(cpus = cpus, threads = threads) + evlist.mmap() while True: evlist.poll(timeout = -1) for cpu in cpus: diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index dcd59328bb49..95b21fece2ce 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -21,21 +21,24 @@ #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) #define SID(e, x, y) xyarray__entry(e->id, x, y) -void perf_evlist__init(struct perf_evlist *evlist) +void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads) { int i; for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i) INIT_HLIST_HEAD(&evlist->heads[i]); INIT_LIST_HEAD(&evlist->entries); + perf_evlist__set_maps(evlist, cpus, threads); } -struct perf_evlist *perf_evlist__new(void) +struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, + struct thread_map *threads) { struct perf_evlist *evlist = zalloc(sizeof(*evlist)); if (evlist != NULL) - perf_evlist__init(evlist); + perf_evlist__init(evlist, cpus, threads); return evlist; } @@ -88,9 +91,9 @@ int perf_evlist__add_default(struct perf_evlist *evlist) return 0; } -int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthreads) +int perf_evlist__alloc_pollfd(struct perf_evlist *evlist) { - int nfds = ncpus * nthreads * evlist->nr_entries; + int nfds = evlist->cpus->nr * evlist->threads->nr * evlist->nr_entries; evlist->pollfd = malloc(sizeof(struct pollfd) * nfds); return evlist->pollfd != NULL ? 0 : -ENOMEM; } @@ -213,11 +216,11 @@ union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) return event; } -void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus) +void perf_evlist__munmap(struct perf_evlist *evlist) { int cpu; - for (cpu = 0; cpu < ncpus; cpu++) { + for (cpu = 0; cpu < evlist->cpus->nr; cpu++) { if (evlist->mmap[cpu].base != NULL) { munmap(evlist->mmap[cpu].base, evlist->mmap_len); evlist->mmap[cpu].base = NULL; @@ -225,9 +228,9 @@ void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus) } } -int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus) +int perf_evlist__alloc_mmap(struct perf_evlist *evlist) { - evlist->mmap = zalloc(ncpus * sizeof(struct perf_mmap)); + evlist->mmap = zalloc(evlist->cpus->nr * sizeof(struct perf_mmap)); return evlist->mmap != NULL ? 0 : -ENOMEM; } @@ -248,8 +251,6 @@ static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot, /** perf_evlist__mmap - Create per cpu maps to receive events * * @evlist - list of events - * @cpus - cpu map being monitored - * @threads - threads map being monitored * @pages - map length in pages * @overwrite - overwrite older events? * @@ -259,21 +260,22 @@ static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot, * unsigned int head = perf_mmap__read_head(m); * * perf_mmap__write_tail(m, head) + * + * Using perf_evlist__read_on_cpu does this automatically. */ -int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, - struct thread_map *threads, int pages, bool overwrite) +int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite) { unsigned int page_size = sysconf(_SC_PAGE_SIZE); int mask = pages * page_size - 1, cpu; struct perf_evsel *first_evsel, *evsel; + const struct cpu_map *cpus = evlist->cpus; + const struct thread_map *threads = evlist->threads; int thread, prot = PROT_READ | (overwrite ? 0 : PROT_WRITE); - if (evlist->mmap == NULL && - perf_evlist__alloc_mmap(evlist, cpus->nr) < 0) + if (evlist->mmap == NULL && perf_evlist__alloc_mmap(evlist) < 0) return -ENOMEM; - if (evlist->pollfd == NULL && - perf_evlist__alloc_pollfd(evlist, cpus->nr, threads->nr) < 0) + if (evlist->pollfd == NULL && perf_evlist__alloc_pollfd(evlist) < 0) return -ENOMEM; evlist->overwrite = overwrite; @@ -315,3 +317,34 @@ out_unmap: } return -1; } + +int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid, + pid_t target_tid, const char *cpu_list) +{ + evlist->threads = thread_map__new(target_pid, target_tid); + + if (evlist->threads == NULL) + return -1; + + if (target_tid != -1) + evlist->cpus = cpu_map__dummy_new(); + else + evlist->cpus = cpu_map__new(cpu_list); + + if (evlist->cpus == NULL) + goto out_delete_threads; + + return 0; + +out_delete_threads: + thread_map__delete(evlist->threads); + return -1; +} + +void perf_evlist__delete_maps(struct perf_evlist *evlist) +{ + cpu_map__delete(evlist->cpus); + thread_map__delete(evlist->threads); + evlist->cpus = NULL; + evlist->threads = NULL; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 85aca6eba16b..c9884056097c 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -22,28 +22,43 @@ struct perf_evlist { union perf_event event_copy; struct perf_mmap *mmap; struct pollfd *pollfd; + struct thread_map *threads; + struct cpu_map *cpus; }; struct perf_evsel; -struct perf_evlist *perf_evlist__new(void); -void perf_evlist__init(struct perf_evlist *evlist); +struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, + struct thread_map *threads); +void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads); void perf_evlist__exit(struct perf_evlist *evlist); void perf_evlist__delete(struct perf_evlist *evlist); void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); int perf_evlist__add_default(struct perf_evlist *evlist); -int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthreads); +int perf_evlist__alloc_pollfd(struct perf_evlist *evlist); void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id); union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *self, int cpu); -int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus); -int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus, - struct thread_map *threads, int pages, bool overwrite); -void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus); +int perf_evlist__alloc_mmap(struct perf_evlist *evlist); +int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite); +void perf_evlist__munmap(struct perf_evlist *evlist); + +static inline void perf_evlist__set_maps(struct perf_evlist *evlist, + struct cpu_map *cpus, + struct thread_map *threads) +{ + evlist->cpus = cpus; + evlist->threads = threads; +} + +int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid, + pid_t target_tid, const char *cpu_list); +void perf_evlist__delete_maps(struct perf_evlist *evlist); #endif /* __PERF_EVLIST_H */ diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c index 88d47895a79c..d2d52175362e 100644 --- a/tools/perf/util/python.c +++ b/tools/perf/util/python.c @@ -553,7 +553,16 @@ struct pyrf_evlist { static int pyrf_evlist__init(struct pyrf_evlist *pevlist, PyObject *args, PyObject *kwargs) { - perf_evlist__init(&pevlist->evlist); + PyObject *pcpus = NULL, *pthreads = NULL; + struct cpu_map *cpus; + struct thread_map *threads; + + if (!PyArg_ParseTuple(args, "OO", &pcpus, &pthreads)) + return -1; + + threads = ((struct pyrf_thread_map *)pthreads)->threads; + cpus = ((struct pyrf_cpu_map *)pcpus)->cpus; + perf_evlist__init(&pevlist->evlist, cpus, threads); return 0; } @@ -567,21 +576,15 @@ static PyObject *pyrf_evlist__mmap(struct pyrf_evlist *pevlist, PyObject *args, PyObject *kwargs) { struct perf_evlist *evlist = &pevlist->evlist; - PyObject *pcpus = NULL, *pthreads = NULL; - struct cpu_map *cpus = NULL; - struct thread_map *threads = NULL; - static char *kwlist[] = {"cpus", "threads", "pages", "overwrite", + static char *kwlist[] = {"pages", "overwrite", NULL, NULL}; int pages = 128, overwrite = false; - if (!PyArg_ParseTupleAndKeywords(args, kwargs, "OO|ii", kwlist, - &pcpus, &pthreads, &pages, &overwrite)) + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ii", kwlist, + &pages, &overwrite)) return NULL; - threads = ((struct pyrf_thread_map *)pthreads)->threads; - cpus = ((struct pyrf_cpu_map *)pcpus)->cpus; - - if (perf_evlist__mmap(evlist, cpus, threads, pages, overwrite) < 0) { + if (perf_evlist__mmap(evlist, pages, overwrite) < 0) { PyErr_SetFromErrno(PyExc_OSError); return NULL; } -- cgit v1.2.3 From 8c3e10eb1968877d6a1957b7e790c6ce01bd56fc Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 31 Jan 2011 14:50:39 -0200 Subject: perf top: Move display agnostic routines to util/top.[ch] Paving the way for a slang browser a la 'perf report --tui'. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Makefile | 2 + tools/perf/builtin-top.c | 482 ++++++++++++++--------------------------------- tools/perf/util/top.c | 212 +++++++++++++++++++++ tools/perf/util/top.h | 67 +++++++ 4 files changed, 418 insertions(+), 345 deletions(-) create mode 100644 tools/perf/util/top.c create mode 100644 tools/perf/util/top.h (limited to 'tools/perf/util') diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 36ff73c0af94..edc660e0bc2f 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -437,6 +437,7 @@ LIB_H += util/probe-finder.h LIB_H += util/probe-event.h LIB_H += util/pstack.h LIB_H += util/cpumap.h +LIB_H += util/top.h LIB_H += $(ARCH_INCLUDE) LIB_OBJS += $(OUTPUT)util/abspath.o @@ -464,6 +465,7 @@ LIB_OBJS += $(OUTPUT)util/strbuf.o LIB_OBJS += $(OUTPUT)util/string.o LIB_OBJS += $(OUTPUT)util/strlist.o LIB_OBJS += $(OUTPUT)util/strfilter.o +LIB_OBJS += $(OUTPUT)util/top.o LIB_OBJS += $(OUTPUT)util/usage.o LIB_OBJS += $(OUTPUT)util/wrapper.o LIB_OBJS += $(OUTPUT)util/sigchain.o diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 599036b06730..3c9ba943aa48 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -27,6 +27,7 @@ #include "util/symbol.h" #include "util/thread.h" #include "util/thread_map.h" +#include "util/top.h" #include "util/util.h" #include #include "util/parse-options.h" @@ -47,7 +48,6 @@ #include #include #include -#include #include #include @@ -62,75 +62,35 @@ #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) -struct perf_evlist *evsel_list; +static struct perf_top top = { + .count_filter = 5, + .delay_secs = 2, + .display_weighted = -1, + .target_pid = -1, + .target_tid = -1, + .active_symbols = LIST_HEAD_INIT(top.active_symbols), + .active_symbols_lock = PTHREAD_MUTEX_INITIALIZER, + .freq = 1000, /* 1 KHz */ +}; static bool system_wide = false; static int default_interval = 0; -static int count_filter = 5; -static int print_entries; - -static int target_pid = -1; -static int target_tid = -1; static bool inherit = false; static int realtime_prio = 0; static bool group = false; static unsigned int page_size; static unsigned int mmap_pages = 128; -static int freq = 1000; /* 1 KHz */ -static int delay_secs = 2; -static bool zero = false; static bool dump_symtab = false; -static bool hide_kernel_symbols = false; -static bool hide_user_symbols = false; static struct winsize winsize; -/* - * Source - */ - -struct source_line { - u64 eip; - unsigned long count[MAX_COUNTERS]; - char *line; - struct source_line *next; -}; - static const char *sym_filter = NULL; struct sym_entry *sym_filter_entry = NULL; struct sym_entry *sym_filter_entry_sched = NULL; static int sym_pcnt_filter = 5; -static int sym_counter = 0; -static struct perf_evsel *sym_evsel = NULL; -static int display_weighted = -1; -static const char *cpu_list; - -/* - * Symbols - */ - -struct sym_entry_source { - struct source_line *source; - struct source_line *lines; - struct source_line **lines_tail; - pthread_mutex_t lock; -}; - -struct sym_entry { - struct rb_node rb_node; - struct list_head node; - unsigned long snap_count; - double weight; - int skip; - u16 name_len; - u8 origin; - struct map *map; - struct sym_entry_source *src; - unsigned long count[0]; -}; /* * Source functions @@ -165,10 +125,10 @@ void get_term_dimensions(struct winsize *ws) static void update_print_entries(struct winsize *ws) { - print_entries = ws->ws_row; + top.print_entries = ws->ws_row; - if (print_entries > 9) - print_entries -= 9; + if (top.print_entries > 9) + top.print_entries -= 9; } static void sig_winch_handler(int sig __used) @@ -269,7 +229,7 @@ static void __zero_source_counters(struct sym_entry *syme) line = syme->src->lines; while (line) { - for (i = 0; i < evsel_list->nr_entries; i++) + for (i = 0; i < top.evlist->nr_entries; i++) line->count[i] = 0; line = line->next; } @@ -331,9 +291,9 @@ static void show_lines(struct source_line *queue, int count, int total) line = queue; for (i = 0; i < count; i++) { - float pcnt = 100.0*(float)line->count[sym_counter]/(float)total; + float pcnt = 100.0*(float)line->count[top.sym_counter]/(float)total; - printf("%8li %4.1f%%\t%s\n", line->count[sym_counter], pcnt, line->line); + printf("%8li %4.1f%%\t%s\n", line->count[top.sym_counter], pcnt, line->line); line = line->next; } } @@ -358,13 +318,13 @@ static void show_details(struct sym_entry *syme) return; symbol = sym_entry__symbol(syme); - printf("Showing %s for %s\n", event_name(sym_evsel), symbol->name); + printf("Showing %s for %s\n", event_name(top.sym_evsel), symbol->name); printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter); pthread_mutex_lock(&syme->src->lock); line = syme->src->source; while (line) { - total += line->count[sym_counter]; + total += line->count[top.sym_counter]; line = line->next; } @@ -376,10 +336,10 @@ static void show_details(struct sym_entry *syme) line_queue = line; line_queue_count++; - if (line->count[sym_counter]) - pcnt = 100.0 * line->count[sym_counter] / (float)total; + if (line->count[top.sym_counter]) + pcnt = 100.0 * line->count[top.sym_counter] / (float)total; if (pcnt >= (float)sym_pcnt_filter) { - if (displayed <= print_entries) + if (displayed <= top.print_entries) show_lines(line_queue, line_queue_count, total); else more++; displayed += line_queue_count; @@ -390,7 +350,7 @@ static void show_details(struct sym_entry *syme) line_queue_count--; } - line->count[sym_counter] = zero ? 0 : line->count[sym_counter] * 7 / 8; + line->count[top.sym_counter] = top.zero ? 0 : line->count[top.sym_counter] * 7 / 8; line = line->next; } pthread_mutex_unlock(&syme->src->lock); @@ -398,181 +358,30 @@ static void show_details(struct sym_entry *syme) printf("%d lines not displayed, maybe increase display entries [e]\n", more); } -/* - * Symbols will be added here in perf_event__process_sample and will get out - * after decayed. - */ -static LIST_HEAD(active_symbols); -static pthread_mutex_t active_symbols_lock = PTHREAD_MUTEX_INITIALIZER; - -/* - * Ordering weight: count-1 * count-2 * ... / count-n - */ -static double sym_weight(const struct sym_entry *sym) -{ - double weight = sym->snap_count; - int counter; - - if (!display_weighted) - return weight; - - for (counter = 1; counter < evsel_list->nr_entries - 1; counter++) - weight *= sym->count[counter]; - - weight /= (sym->count[counter] + 1); - - return weight; -} - -static long samples; -static long kernel_samples, us_samples; -static long exact_samples; -static long guest_us_samples, guest_kernel_samples; static const char CONSOLE_CLEAR[] = ""; static void __list_insert_active_sym(struct sym_entry *syme) { - list_add(&syme->node, &active_symbols); -} - -static void list_remove_active_sym(struct sym_entry *syme) -{ - pthread_mutex_lock(&active_symbols_lock); - list_del_init(&syme->node); - pthread_mutex_unlock(&active_symbols_lock); -} - -static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se) -{ - struct rb_node **p = &tree->rb_node; - struct rb_node *parent = NULL; - struct sym_entry *iter; - - while (*p != NULL) { - parent = *p; - iter = rb_entry(parent, struct sym_entry, rb_node); - - if (se->weight > iter->weight) - p = &(*p)->rb_left; - else - p = &(*p)->rb_right; - } - - rb_link_node(&se->rb_node, parent, p); - rb_insert_color(&se->rb_node, tree); + list_add(&syme->node, &top.active_symbols); } static void print_sym_table(struct perf_session *session) { - int printed = 0, j; - struct perf_evsel *counter; - int snap = !display_weighted ? sym_counter : 0; - float samples_per_sec = samples/delay_secs; - float ksamples_per_sec = kernel_samples/delay_secs; - float us_samples_per_sec = (us_samples)/delay_secs; - float guest_kernel_samples_per_sec = (guest_kernel_samples)/delay_secs; - float guest_us_samples_per_sec = (guest_us_samples)/delay_secs; - float esamples_percent = (100.0*exact_samples)/samples; - float sum_ksamples = 0.0; - struct sym_entry *syme, *n; - struct rb_root tmp = RB_ROOT; + char bf[160]; + int printed = 0; struct rb_node *nd; - int sym_width = 0, dso_width = 0, dso_short_width = 0; + struct sym_entry *syme; + struct rb_root tmp = RB_ROOT; const int win_width = winsize.ws_col - 1; - - samples = us_samples = kernel_samples = exact_samples = 0; - guest_kernel_samples = guest_us_samples = 0; - - /* Sort the active symbols */ - pthread_mutex_lock(&active_symbols_lock); - syme = list_entry(active_symbols.next, struct sym_entry, node); - pthread_mutex_unlock(&active_symbols_lock); - - list_for_each_entry_safe_from(syme, n, &active_symbols, node) { - syme->snap_count = syme->count[snap]; - if (syme->snap_count != 0) { - - if ((hide_user_symbols && - syme->origin == PERF_RECORD_MISC_USER) || - (hide_kernel_symbols && - syme->origin == PERF_RECORD_MISC_KERNEL)) { - list_remove_active_sym(syme); - continue; - } - syme->weight = sym_weight(syme); - rb_insert_active_sym(&tmp, syme); - sum_ksamples += syme->snap_count; - - for (j = 0; j < evsel_list->nr_entries; j++) - syme->count[j] = zero ? 0 : syme->count[j] * 7 / 8; - } else - list_remove_active_sym(syme); - } + int sym_width, dso_width, dso_short_width; + float sum_ksamples = perf_top__decay_samples(&top, &tmp); puts(CONSOLE_CLEAR); - if (!perf_guest) { - printf(" PerfTop:%8.0f irqs/sec kernel:%4.1f%%" - " exact: %4.1f%% [", - samples_per_sec, - 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / - samples_per_sec)), - esamples_percent); - } else { - printf(" PerfTop:%8.0f irqs/sec kernel:%4.1f%% us:%4.1f%%" - " guest kernel:%4.1f%% guest us:%4.1f%%" - " exact: %4.1f%% [", - samples_per_sec, - 100.0 - (100.0 * ((samples_per_sec-ksamples_per_sec) / - samples_per_sec)), - 100.0 - (100.0 * ((samples_per_sec-us_samples_per_sec) / - samples_per_sec)), - 100.0 - (100.0 * ((samples_per_sec - - guest_kernel_samples_per_sec) / - samples_per_sec)), - 100.0 - (100.0 * ((samples_per_sec - - guest_us_samples_per_sec) / - samples_per_sec)), - esamples_percent); - } + perf_top__header_snprintf(&top, bf, sizeof(bf)); + printf("%s\n", bf); - if (evsel_list->nr_entries == 1 || !display_weighted) { - struct perf_evsel *first; - first = list_entry(evsel_list->entries.next, struct perf_evsel, node); - printf("%" PRIu64, (uint64_t)first->attr.sample_period); - if (freq) - printf("Hz "); - else - printf(" "); - } - - if (!display_weighted) - printf("%s", event_name(sym_evsel)); - else list_for_each_entry(counter, &evsel_list->entries, node) { - if (counter->idx) - printf("/"); - - printf("%s", event_name(counter)); - } - - printf( "], "); - - if (target_pid != -1) - printf(" (target_pid: %d", target_pid); - else if (target_tid != -1) - printf(" (target_tid: %d", target_tid); - else - printf(" (all"); - - if (cpu_list) - printf(", CPU%s: %s)\n", evsel_list->cpus->nr > 1 ? "s" : "", cpu_list); - else { - if (target_tid != -1) - printf(")\n"); - else - printf(", %d CPU%s)\n", evsel_list->cpus->nr, - evsel_list->cpus->nr > 1 ? "s" : ""); - } + perf_top__reset_sample_counters(&top); printf("%-*.*s\n", win_width, win_width, graph_dotted_line); @@ -587,26 +396,8 @@ static void print_sym_table(struct perf_session *session) return; } - /* - * Find the longest symbol name that will be displayed - */ - for (nd = rb_first(&tmp); nd; nd = rb_next(nd)) { - syme = rb_entry(nd, struct sym_entry, rb_node); - if (++printed > print_entries || - (int)syme->snap_count < count_filter) - continue; - - if (syme->map->dso->long_name_len > dso_width) - dso_width = syme->map->dso->long_name_len; - - if (syme->map->dso->short_name_len > dso_short_width) - dso_short_width = syme->map->dso->short_name_len; - - if (syme->name_len > sym_width) - sym_width = syme->name_len; - } - - printed = 0; + perf_top__find_widths(&top, &tmp, &dso_width, &dso_short_width, + &sym_width); if (sym_width + dso_width > winsize.ws_col - 29) { dso_width = dso_short_width; @@ -614,7 +405,7 @@ static void print_sym_table(struct perf_session *session) sym_width = winsize.ws_col - dso_width - 29; } putchar('\n'); - if (evsel_list->nr_entries == 1) + if (top.evlist->nr_entries == 1) printf(" samples pcnt"); else printf(" weight samples pcnt"); @@ -623,7 +414,7 @@ static void print_sym_table(struct perf_session *session) printf(" RIP "); printf(" %-*.*s DSO\n", sym_width, sym_width, "function"); printf(" %s _______ _____", - evsel_list->nr_entries == 1 ? " " : "______"); + top.evlist->nr_entries == 1 ? " " : "______"); if (verbose) printf(" ________________"); printf(" %-*.*s", sym_width, sym_width, graph_line); @@ -636,13 +427,14 @@ static void print_sym_table(struct perf_session *session) syme = rb_entry(nd, struct sym_entry, rb_node); sym = sym_entry__symbol(syme); - if (++printed > print_entries || (int)syme->snap_count < count_filter) + if (++printed > top.print_entries || + (int)syme->snap_count < top.count_filter) continue; pcnt = 100.0 - (100.0 * ((sum_ksamples - syme->snap_count) / sum_ksamples)); - if (evsel_list->nr_entries == 1 || !display_weighted) + if (top.evlist->nr_entries == 1 || !top.display_weighted) printf("%20.2f ", syme->weight); else printf("%9.1f %10ld ", syme->weight, syme->snap_count); @@ -715,11 +507,11 @@ static void prompt_symbol(struct sym_entry **target, const char *msg) if (p) *p = 0; - pthread_mutex_lock(&active_symbols_lock); - syme = list_entry(active_symbols.next, struct sym_entry, node); - pthread_mutex_unlock(&active_symbols_lock); + pthread_mutex_lock(&top.active_symbols_lock); + syme = list_entry(top.active_symbols.next, struct sym_entry, node); + pthread_mutex_unlock(&top.active_symbols_lock); - list_for_each_entry_safe_from(syme, n, &active_symbols, node) { + list_for_each_entry_safe_from(syme, n, &top.active_symbols, node) { struct symbol *sym = sym_entry__symbol(syme); if (!strcmp(buf, sym->name)) { @@ -749,28 +541,28 @@ static void print_mapped_keys(void) } fprintf(stdout, "\nMapped keys:\n"); - fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", delay_secs); - fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", print_entries); + fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", top.delay_secs); + fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", top.print_entries); - if (evsel_list->nr_entries > 1) - fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(sym_evsel)); + if (top.evlist->nr_entries > 1) + fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(top.sym_evsel)); - fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", count_filter); + fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", top.count_filter); fprintf(stdout, "\t[F] annotate display filter (percent). \t(%d%%)\n", sym_pcnt_filter); fprintf(stdout, "\t[s] annotate symbol. \t(%s)\n", name?: "NULL"); fprintf(stdout, "\t[S] stop annotation.\n"); - if (evsel_list->nr_entries > 1) - fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", display_weighted ? 1 : 0); + if (top.evlist->nr_entries > 1) + fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", top.display_weighted ? 1 : 0); fprintf(stdout, "\t[K] hide kernel_symbols symbols. \t(%s)\n", - hide_kernel_symbols ? "yes" : "no"); + top.hide_kernel_symbols ? "yes" : "no"); fprintf(stdout, "\t[U] hide user symbols. \t(%s)\n", - hide_user_symbols ? "yes" : "no"); - fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", zero ? 1 : 0); + top.hide_user_symbols ? "yes" : "no"); + fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", top.zero ? 1 : 0); fprintf(stdout, "\t[qQ] quit.\n"); } @@ -791,7 +583,7 @@ static int key_mapped(int c) return 1; case 'E': case 'w': - return evsel_list->nr_entries > 1 ? 1 : 0; + return top.evlist->nr_entries > 1 ? 1 : 0; default: break; } @@ -826,47 +618,47 @@ static void handle_keypress(struct perf_session *session, int c) switch (c) { case 'd': - prompt_integer(&delay_secs, "Enter display delay"); - if (delay_secs < 1) - delay_secs = 1; + prompt_integer(&top.delay_secs, "Enter display delay"); + if (top.delay_secs < 1) + top.delay_secs = 1; break; case 'e': - prompt_integer(&print_entries, "Enter display entries (lines)"); - if (print_entries == 0) { + prompt_integer(&top.print_entries, "Enter display entries (lines)"); + if (top.print_entries == 0) { sig_winch_handler(SIGWINCH); signal(SIGWINCH, sig_winch_handler); } else signal(SIGWINCH, SIG_DFL); break; case 'E': - if (evsel_list->nr_entries > 1) { + if (top.evlist->nr_entries > 1) { fprintf(stderr, "\nAvailable events:"); - list_for_each_entry(sym_evsel, &evsel_list->entries, node) - fprintf(stderr, "\n\t%d %s", sym_evsel->idx, event_name(sym_evsel)); + list_for_each_entry(top.sym_evsel, &top.evlist->entries, node) + fprintf(stderr, "\n\t%d %s", top.sym_evsel->idx, event_name(top.sym_evsel)); - prompt_integer(&sym_counter, "Enter details event counter"); + prompt_integer(&top.sym_counter, "Enter details event counter"); - if (sym_counter >= evsel_list->nr_entries) { - sym_evsel = list_entry(evsel_list->entries.next, struct perf_evsel, node); - sym_counter = 0; - fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(sym_evsel)); + if (top.sym_counter >= top.evlist->nr_entries) { + top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node); + top.sym_counter = 0; + fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(top.sym_evsel)); sleep(1); break; } - list_for_each_entry(sym_evsel, &evsel_list->entries, node) - if (sym_evsel->idx == sym_counter) + list_for_each_entry(top.sym_evsel, &top.evlist->entries, node) + if (top.sym_evsel->idx == top.sym_counter) break; - } else sym_counter = 0; + } else top.sym_counter = 0; break; case 'f': - prompt_integer(&count_filter, "Enter display event count filter"); + prompt_integer(&top.count_filter, "Enter display event count filter"); break; case 'F': prompt_percent(&sym_pcnt_filter, "Enter details display event filter (percent)"); break; case 'K': - hide_kernel_symbols = !hide_kernel_symbols; + top.hide_kernel_symbols = !top.hide_kernel_symbols; break; case 'q': case 'Q': @@ -890,13 +682,13 @@ static void handle_keypress(struct perf_session *session, int c) } break; case 'U': - hide_user_symbols = !hide_user_symbols; + top.hide_user_symbols = !top.hide_user_symbols; break; case 'w': - display_weighted = ~display_weighted; + top.display_weighted = ~top.display_weighted; break; case 'z': - zero = !zero; + top.zero = !top.zero; break; default: break; @@ -917,7 +709,7 @@ static void *display_thread(void *arg __used) tc.c_cc[VTIME] = 0; repeat: - delay_msecs = delay_secs * 1000; + delay_msecs = top.delay_secs * 1000; tcsetattr(0, TCSANOW, &tc); /* trash return*/ getc(stdin); @@ -1005,27 +797,27 @@ static void perf_event__process_sample(const union perf_event *event, struct machine *machine; u8 origin = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; - ++samples; + ++top.samples; switch (origin) { case PERF_RECORD_MISC_USER: - ++us_samples; - if (hide_user_symbols) + ++top.us_samples; + if (top.hide_user_symbols) return; machine = perf_session__find_host_machine(session); break; case PERF_RECORD_MISC_KERNEL: - ++kernel_samples; - if (hide_kernel_symbols) + ++top.kernel_samples; + if (top.hide_kernel_symbols) return; machine = perf_session__find_host_machine(session); break; case PERF_RECORD_MISC_GUEST_KERNEL: - ++guest_kernel_samples; + ++top.guest_kernel_samples; machine = perf_session__find_machine(session, event->ip.pid); break; case PERF_RECORD_MISC_GUEST_USER: - ++guest_us_samples; + ++top.guest_us_samples; /* * TODO: we don't process guest user from host side * except simple counting. @@ -1042,7 +834,7 @@ static void perf_event__process_sample(const union perf_event *event, } if (event->header.misc & PERF_RECORD_MISC_EXACT_IP) - exact_samples++; + top.exact_samples++; if (perf_event__preprocess_sample(event, session, &al, sample, symbol_filter) < 0 || @@ -1093,14 +885,14 @@ static void perf_event__process_sample(const union perf_event *event, struct perf_evsel *evsel; syme->origin = origin; - evsel = perf_evlist__id2evsel(evsel_list, sample->id); + evsel = perf_evlist__id2evsel(top.evlist, sample->id); assert(evsel != NULL); syme->count[evsel->idx]++; record_precise_ip(syme, evsel->idx, ip); - pthread_mutex_lock(&active_symbols_lock); + pthread_mutex_lock(&top.active_symbols_lock); if (list_empty(&syme->node) || !syme->node.next) __list_insert_active_sym(syme); - pthread_mutex_unlock(&active_symbols_lock); + pthread_mutex_unlock(&top.active_symbols_lock); } } @@ -1109,7 +901,7 @@ static void perf_session__mmap_read_cpu(struct perf_session *self, int cpu) struct perf_sample sample; union perf_event *event; - while ((event = perf_evlist__read_on_cpu(evsel_list, cpu)) != NULL) { + while ((event = perf_evlist__read_on_cpu(top.evlist, cpu)) != NULL) { perf_session__parse_sample(self, event, &sample); if (event->header.type == PERF_RECORD_SAMPLE) @@ -1123,7 +915,7 @@ static void perf_session__mmap_read(struct perf_session *self) { int i; - for (i = 0; i < evsel_list->cpus->nr; i++) + for (i = 0; i < top.evlist->cpus->nr; i++) perf_session__mmap_read_cpu(self, i); } @@ -1136,10 +928,10 @@ static void start_counters(struct perf_evlist *evlist) attr->sample_type = PERF_SAMPLE_IP | PERF_SAMPLE_TID; - if (freq) { + if (top.freq) { attr->sample_type |= PERF_SAMPLE_PERIOD; attr->freq = 1; - attr->sample_freq = freq; + attr->sample_freq = top.freq; } if (evlist->nr_entries > 1) { @@ -1149,8 +941,8 @@ static void start_counters(struct perf_evlist *evlist) attr->mmap = 1; try_again: - if (perf_evsel__open(counter, evsel_list->cpus, - evsel_list->threads, group, inherit) < 0) { + if (perf_evsel__open(counter, top.evlist->cpus, + top.evlist->threads, group, inherit) < 0) { int err = errno; if (err == EPERM || err == EACCES) @@ -1198,18 +990,18 @@ static int __cmd_top(void) if (session == NULL) return -ENOMEM; - if (target_tid != -1) - perf_event__synthesize_thread(target_tid, perf_event__process, + if (top.target_tid != -1) + perf_event__synthesize_thread(top.target_tid, perf_event__process, session); else perf_event__synthesize_threads(perf_event__process, session); - start_counters(evsel_list); - first = list_entry(evsel_list->entries.next, struct perf_evsel, node); + start_counters(top.evlist); + first = list_entry(top.evlist->entries.next, struct perf_evsel, node); perf_session__set_sample_type(session, first->attr.sample_type); /* Wait for a minimal set of events before starting the snapshot */ - poll(evsel_list->pollfd, evsel_list->nr_fds, 100); + poll(top.evlist->pollfd, top.evlist->nr_fds, 100); perf_session__mmap_read(session); @@ -1229,12 +1021,12 @@ static int __cmd_top(void) } while (1) { - int hits = samples; + u64 hits = top.samples; perf_session__mmap_read(session); - if (hits == samples) - ret = poll(evsel_list->pollfd, evsel_list->nr_fds, 100); + if (hits == top.samples) + ret = poll(top.evlist->pollfd, top.evlist->nr_fds, 100); } return 0; @@ -1246,31 +1038,31 @@ static const char * const top_usage[] = { }; static const struct option options[] = { - OPT_CALLBACK('e', "event", &evsel_list, "event", + OPT_CALLBACK('e', "event", &top.evlist, "event", "event selector. use 'perf list' to list available events", parse_events), OPT_INTEGER('c', "count", &default_interval, "event period to sample"), - OPT_INTEGER('p', "pid", &target_pid, + OPT_INTEGER('p', "pid", &top.target_pid, "profile events on existing process id"), - OPT_INTEGER('t', "tid", &target_tid, + OPT_INTEGER('t', "tid", &top.target_tid, "profile events on existing thread id"), OPT_BOOLEAN('a', "all-cpus", &system_wide, "system-wide collection from all CPUs"), - OPT_STRING('C', "cpu", &cpu_list, "cpu", + OPT_STRING('C', "cpu", &top.cpu_list, "cpu", "list of cpus to monitor"), OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, "file", "vmlinux pathname"), - OPT_BOOLEAN('K', "hide_kernel_symbols", &hide_kernel_symbols, + OPT_BOOLEAN('K', "hide_kernel_symbols", &top.hide_kernel_symbols, "hide kernel symbols"), OPT_UINTEGER('m', "mmap-pages", &mmap_pages, "number of mmap data pages"), OPT_INTEGER('r', "realtime", &realtime_prio, "collect data with this RT SCHED_FIFO priority"), - OPT_INTEGER('d', "delay", &delay_secs, + OPT_INTEGER('d', "delay", &top.delay_secs, "number of seconds to delay between refreshes"), OPT_BOOLEAN('D', "dump-symtab", &dump_symtab, "dump the symbol table used for profiling"), - OPT_INTEGER('f', "count-filter", &count_filter, + OPT_INTEGER('f', "count-filter", &top.count_filter, "only display functions with more events than this"), OPT_BOOLEAN('g', "group", &group, "put the counters into a counter group"), @@ -1278,13 +1070,13 @@ static const struct option options[] = { "child tasks inherit counters"), OPT_STRING('s', "sym-annotate", &sym_filter, "symbol name", "symbol to annotate"), - OPT_BOOLEAN('z', "zero", &zero, + OPT_BOOLEAN('z', "zero", &top.zero, "zero history across updates"), - OPT_INTEGER('F', "freq", &freq, + OPT_INTEGER('F', "freq", &top.freq, "profile at this frequency"), - OPT_INTEGER('E', "entries", &print_entries, + OPT_INTEGER('E', "entries", &top.print_entries, "display this many functions"), - OPT_BOOLEAN('U', "hide_user_symbols", &hide_user_symbols, + OPT_BOOLEAN('U', "hide_user_symbols", &top.hide_user_symbols, "hide user symbols"), OPT_INCR('v', "verbose", &verbose, "be more verbose (show counter open errors, etc)"), @@ -1296,8 +1088,8 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) struct perf_evsel *pos; int status = -ENOMEM; - evsel_list = perf_evlist__new(NULL, NULL); - if (evsel_list == NULL) + top.evlist = perf_evlist__new(NULL, NULL); + if (top.evlist == NULL) return -ENOMEM; page_size = sysconf(_SC_PAGE_SIZE); @@ -1307,43 +1099,43 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) usage_with_options(top_usage, options); /* CPU and PID are mutually exclusive */ - if (target_tid > 0 && cpu_list) { + if (top.target_tid > 0 && top.cpu_list) { printf("WARNING: PID switch overriding CPU\n"); sleep(1); - cpu_list = NULL; + top.cpu_list = NULL; } - if (target_pid != -1) - target_tid = target_pid; + if (top.target_pid != -1) + top.target_tid = top.target_pid; - if (perf_evlist__create_maps(evsel_list, target_pid, - target_tid, cpu_list) < 0) + if (perf_evlist__create_maps(top.evlist, top.target_pid, + top.target_tid, top.cpu_list) < 0) usage_with_options(top_usage, options); - if (!evsel_list->nr_entries && - perf_evlist__add_default(evsel_list) < 0) { + if (!top.evlist->nr_entries && + perf_evlist__add_default(top.evlist) < 0) { pr_err("Not enough memory for event selector list\n"); return -ENOMEM; } - if (delay_secs < 1) - delay_secs = 1; + if (top.delay_secs < 1) + top.delay_secs = 1; /* * User specified count overrides default frequency. */ if (default_interval) - freq = 0; - else if (freq) { - default_interval = freq; + top.freq = 0; + else if (top.freq) { + default_interval = top.freq; } else { fprintf(stderr, "frequency and count are zero, aborting\n"); exit(EXIT_FAILURE); } - list_for_each_entry(pos, &evsel_list->entries, node) { - if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, - evsel_list->threads->nr) < 0) + list_for_each_entry(pos, &top.evlist->entries, node) { + if (perf_evsel__alloc_fd(pos, top.evlist->cpus->nr, + top.evlist->threads->nr) < 0) goto out_free_fd; /* * Fill in the ones not specifically initialized via -c: @@ -1354,28 +1146,28 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) pos->attr.sample_period = default_interval; } - if (perf_evlist__alloc_pollfd(evsel_list) < 0 || - perf_evlist__alloc_mmap(evsel_list) < 0) + if (perf_evlist__alloc_pollfd(top.evlist) < 0 || + perf_evlist__alloc_mmap(top.evlist) < 0) goto out_free_fd; - sym_evsel = list_entry(evsel_list->entries.next, struct perf_evsel, node); + top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node); symbol_conf.priv_size = (sizeof(struct sym_entry) + - (evsel_list->nr_entries + 1) * sizeof(unsigned long)); + (top.evlist->nr_entries + 1) * sizeof(unsigned long)); symbol_conf.try_vmlinux_path = (symbol_conf.vmlinux_name == NULL); if (symbol__init() < 0) return -1; get_term_dimensions(&winsize); - if (print_entries == 0) { + if (top.print_entries == 0) { update_print_entries(&winsize); signal(SIGWINCH, sig_winch_handler); } status = __cmd_top(); out_free_fd: - perf_evlist__delete(evsel_list); + perf_evlist__delete(top.evlist); return status; } diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c new file mode 100644 index 000000000000..c06cc5386e7e --- /dev/null +++ b/tools/perf/util/top.c @@ -0,0 +1,212 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo + * + * Refactored from builtin-top.c, see that files for further copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + +#include "cpumap.h" +#include "event.h" +#include "evlist.h" +#include "evsel.h" +#include "parse-events.h" +#include "symbol.h" +#include "top.h" +#include + +/* + * Ordering weight: count-1 * count-2 * ... / count-n + */ +static double sym_weight(const struct sym_entry *sym, struct perf_top *top) +{ + double weight = sym->snap_count; + int counter; + + if (!top->display_weighted) + return weight; + + for (counter = 1; counter < top->evlist->nr_entries - 1; counter++) + weight *= sym->count[counter]; + + weight /= (sym->count[counter] + 1); + + return weight; +} + +static void perf_top__remove_active_sym(struct perf_top *top, struct sym_entry *syme) +{ + pthread_mutex_lock(&top->active_symbols_lock); + list_del_init(&syme->node); + pthread_mutex_unlock(&top->active_symbols_lock); +} + +static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se) +{ + struct rb_node **p = &tree->rb_node; + struct rb_node *parent = NULL; + struct sym_entry *iter; + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct sym_entry, rb_node); + + if (se->weight > iter->weight) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&se->rb_node, parent, p); + rb_insert_color(&se->rb_node, tree); +} + +size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size) +{ + struct perf_evsel *counter; + float samples_per_sec = top->samples / top->delay_secs; + float ksamples_per_sec = top->kernel_samples / top->delay_secs; + float esamples_percent = (100.0 * top->exact_samples) / top->samples; + size_t ret = 0; + + if (!perf_guest) { + ret = snprintf(bf, size, + " PerfTop:%8.0f irqs/sec kernel:%4.1f%%" + " exact: %4.1f%% [", samples_per_sec, + 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / + samples_per_sec)), + esamples_percent); + } else { + float us_samples_per_sec = top->us_samples / top->delay_secs; + float guest_kernel_samples_per_sec = top->guest_kernel_samples / top->delay_secs; + float guest_us_samples_per_sec = top->guest_us_samples / top->delay_secs; + + ret = snprintf(bf, size, + " PerfTop:%8.0f irqs/sec kernel:%4.1f%% us:%4.1f%%" + " guest kernel:%4.1f%% guest us:%4.1f%%" + " exact: %4.1f%% [", samples_per_sec, + 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - us_samples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - + guest_kernel_samples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - + guest_us_samples_per_sec) / + samples_per_sec)), + esamples_percent); + } + + if (top->evlist->nr_entries == 1 || !top->display_weighted) { + struct perf_evsel *first; + first = list_entry(top->evlist->entries.next, struct perf_evsel, node); + ret += snprintf(bf + ret, size - ret, "%" PRIu64 "%s ", + (uint64_t)first->attr.sample_period, + top->freq ? "Hz" : ""); + } + + if (!top->display_weighted) { + ret += snprintf(bf + ret, size - ret, "%s", + event_name(top->sym_evsel)); + } else list_for_each_entry(counter, &top->evlist->entries, node) { + ret += snprintf(bf + ret, size - ret, "%s%s", + counter->idx ? "/" : "", event_name(counter)); + } + + ret += snprintf(bf + ret, size - ret, "], "); + + if (top->target_pid != -1) + ret += snprintf(bf + ret, size - ret, " (target_pid: %d", + top->target_pid); + else if (top->target_tid != -1) + ret += snprintf(bf + ret, size - ret, " (target_tid: %d", + top->target_tid); + else + ret += snprintf(bf + ret, size - ret, " (all"); + + if (top->cpu_list) + ret += snprintf(bf + ret, size - ret, ", CPU%s: %s)", + top->evlist->cpus->nr > 1 ? "s" : "", top->cpu_list); + else { + if (top->target_tid != -1) + ret += snprintf(bf + ret, size - ret, ")"); + else + ret += snprintf(bf + ret, size - ret, ", %d CPU%s)", + top->evlist->cpus->nr, + top->evlist->cpus->nr > 1 ? "s" : ""); + } + + return ret; +} + +void perf_top__reset_sample_counters(struct perf_top *top) +{ + top->samples = top->us_samples = top->kernel_samples = + top->exact_samples = top->guest_kernel_samples = + top->guest_us_samples = 0; +} + +float perf_top__decay_samples(struct perf_top *top, struct rb_root *root) +{ + struct sym_entry *syme, *n; + float sum_ksamples = 0.0; + int snap = !top->display_weighted ? top->sym_counter : 0, j; + + /* Sort the active symbols */ + pthread_mutex_lock(&top->active_symbols_lock); + syme = list_entry(top->active_symbols.next, struct sym_entry, node); + pthread_mutex_unlock(&top->active_symbols_lock); + + list_for_each_entry_safe_from(syme, n, &top->active_symbols, node) { + syme->snap_count = syme->count[snap]; + if (syme->snap_count != 0) { + + if ((top->hide_user_symbols && + syme->origin == PERF_RECORD_MISC_USER) || + (top->hide_kernel_symbols && + syme->origin == PERF_RECORD_MISC_KERNEL)) { + perf_top__remove_active_sym(top, syme); + continue; + } + syme->weight = sym_weight(syme, top); + rb_insert_active_sym(root, syme); + sum_ksamples += syme->snap_count; + + for (j = 0; j < top->evlist->nr_entries; j++) + syme->count[j] = top->zero ? 0 : syme->count[j] * 7 / 8; + } else + perf_top__remove_active_sym(top, syme); + } + + return sum_ksamples; +} + +/* + * Find the longest symbol name that will be displayed + */ +void perf_top__find_widths(struct perf_top *top, struct rb_root *root, + int *dso_width, int *dso_short_width, int *sym_width) +{ + struct rb_node *nd; + int printed = 0; + + *sym_width = *dso_width = *dso_short_width = 0; + + for (nd = rb_first(root); nd; nd = rb_next(nd)) { + struct sym_entry *syme = rb_entry(nd, struct sym_entry, rb_node); + + if (++printed > top->print_entries || + (int)syme->snap_count < top->count_filter) + continue; + + if (syme->map->dso->long_name_len > *dso_width) + *dso_width = syme->map->dso->long_name_len; + + if (syme->map->dso->short_name_len > *dso_short_width) + *dso_short_width = syme->map->dso->short_name_len; + + if (syme->name_len > *sym_width) + *sym_width = syme->name_len; + } +} diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h new file mode 100644 index 000000000000..0467b26dd8d8 --- /dev/null +++ b/tools/perf/util/top.h @@ -0,0 +1,67 @@ +#ifndef __PERF_TOP_H +#define __PERF_TOP_H 1 + +#include "types.h" +#include "../perf.h" +#include +#include +#include +#include + +struct perf_evlist; +struct perf_evsel; + +struct source_line { + u64 eip; + unsigned long count[MAX_COUNTERS]; /* FIXME */ + char *line; + struct source_line *next; +}; + +struct sym_entry_source { + struct source_line *source; + struct source_line *lines; + struct source_line **lines_tail; + pthread_mutex_t lock; +}; + +struct sym_entry { + struct rb_node rb_node; + struct list_head node; + unsigned long snap_count; + double weight; + int skip; + u16 name_len; + u8 origin; + struct map *map; + struct sym_entry_source *src; + unsigned long count[0]; +}; + +struct perf_top { + struct perf_evlist *evlist; + /* + * Symbols will be added here in perf_event__process_sample and will + * get out after decayed. + */ + struct list_head active_symbols; + pthread_mutex_t active_symbols_lock; + u64 samples; + u64 kernel_samples, us_samples; + u64 exact_samples; + u64 guest_us_samples, guest_kernel_samples; + int print_entries, count_filter, delay_secs; + int display_weighted, freq; + int sym_counter, target_pid, target_tid; + bool hide_kernel_symbols, hide_user_symbols, zero; + const char *cpu_list; + struct perf_evsel *sym_evsel; +}; + +size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size); +void perf_top__reset_sample_counters(struct perf_top *top); +float perf_top__decay_samples(struct perf_top *top, struct rb_root *root); +void perf_top__find_widths(struct perf_top *top, struct rb_root *root, + int *dso_width, int *dso_short_width, int *sym_width); + +#endif /* __PERF_TOP_H */ -- cgit v1.2.3 From 229ade9ba36341f7369ecb4f134bcec9133520bf Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 31 Jan 2011 18:08:39 -0200 Subject: perf tools: Don't fallback to setup_pager unconditionally Because in tools like 'top' we don't want the pager. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-annotate.c | 2 +- tools/perf/builtin-report.c | 2 +- tools/perf/util/cache.h | 7 ++++--- tools/perf/util/ui/setup.c | 5 +++-- 4 files changed, 9 insertions(+), 7 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index 70067862e07f..cd9dec46c19f 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -452,7 +452,7 @@ int cmd_annotate(int argc, const char **argv, const char *prefix __used) else if (use_tui) use_browser = 1; - setup_browser(); + setup_browser(true); symbol_conf.priv_size = sizeof(struct sym_priv); symbol_conf.try_vmlinux_path = true; diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index a6a4e5457b6f..080937c3a656 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -499,7 +499,7 @@ int cmd_report(int argc, const char **argv, const char *prefix __used) use_browser = 1; if (strcmp(input_name, "-") != 0) - setup_browser(); + setup_browser(true); else use_browser = 0; /* diff --git a/tools/perf/util/cache.h b/tools/perf/util/cache.h index a7729797fd96..fc5e5a09d5b9 100644 --- a/tools/perf/util/cache.h +++ b/tools/perf/util/cache.h @@ -34,13 +34,14 @@ extern int pager_use_color; extern int use_browser; #ifdef NO_NEWT_SUPPORT -static inline void setup_browser(void) +static inline void setup_browser(bool fallback_to_pager) { - setup_pager(); + if (fallback_to_pager) + setup_pager(); } static inline void exit_browser(bool wait_for_ok __used) {} #else -void setup_browser(void); +void setup_browser(bool fallback_to_pager); void exit_browser(bool wait_for_ok); #endif diff --git a/tools/perf/util/ui/setup.c b/tools/perf/util/ui/setup.c index 662085032eb7..fbf1a145492f 100644 --- a/tools/perf/util/ui/setup.c +++ b/tools/perf/util/ui/setup.c @@ -14,11 +14,12 @@ static void newt_suspend(void *d __used) newtResume(); } -void setup_browser(void) +void setup_browser(bool fallback_to_pager) { if (!isatty(1) || !use_browser || dump_trace) { use_browser = 0; - setup_pager(); + if (fallback_to_pager) + setup_pager(); return; } -- cgit v1.2.3 From c0443df1b69b59675fc6790e0ddce87c8ca00abf Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 31 Jan 2011 18:19:33 -0200 Subject: perf top: Introduce slang based TUI Disabled by default as there are features found in the stdio based one that aren't implemented, like live annotation, filtering knobs data entry. Annotation hopefully will get somehow merged with the 'perf annotate' code. To use it: perf top --tui Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Makefile | 4 ++ tools/perf/builtin-top.c | 35 ++++++++-- tools/perf/util/top.c | 7 +- tools/perf/util/top.h | 15 ++++- tools/perf/util/ui/browsers/top.c | 136 ++++++++++++++++++++++++++++++++++++++ 5 files changed, 189 insertions(+), 8 deletions(-) create mode 100644 tools/perf/util/ui/browsers/top.c (limited to 'tools/perf/util') diff --git a/tools/perf/Makefile b/tools/perf/Makefile index edc660e0bc2f..67a9f4d805de 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -621,6 +621,7 @@ else LIB_OBJS += $(OUTPUT)util/ui/browsers/annotate.o LIB_OBJS += $(OUTPUT)util/ui/browsers/hists.o LIB_OBJS += $(OUTPUT)util/ui/browsers/map.o + LIB_OBJS += $(OUTPUT)util/ui/browsers/top.o LIB_OBJS += $(OUTPUT)util/ui/helpline.o LIB_OBJS += $(OUTPUT)util/ui/progress.o LIB_OBJS += $(OUTPUT)util/ui/util.o @@ -1050,6 +1051,9 @@ $(OUTPUT)util/ui/browser.o: util/ui/browser.c $(OUTPUT)PERF-CFLAGS $(OUTPUT)util/ui/browsers/annotate.o: util/ui/browsers/annotate.c $(OUTPUT)PERF-CFLAGS $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< +$(OUTPUT)util/ui/browsers/top.o: util/ui/browsers/top.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< + $(OUTPUT)util/ui/browsers/hists.o: util/ui/browsers/hists.c $(OUTPUT)PERF-CFLAGS $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 3c9ba943aa48..104de9ab314c 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -20,6 +20,7 @@ #include "perf.h" +#include "util/cache.h" #include "util/color.h" #include "util/evlist.h" #include "util/evsel.h" @@ -75,6 +76,8 @@ static struct perf_top top = { static bool system_wide = false; +static bool use_tui, use_stdio; + static int default_interval = 0; static bool inherit = false; @@ -96,11 +99,6 @@ static int sym_pcnt_filter = 5; * Source functions */ -static inline struct symbol *sym_entry__symbol(struct sym_entry *self) -{ - return ((void *)self) + symbol_conf.priv_size; -} - void get_term_dimensions(struct winsize *ws) { char *s = getenv("LINES"); @@ -695,6 +693,14 @@ static void handle_keypress(struct perf_session *session, int c) } } +static void *display_thread_tui(void *arg __used) +{ + perf_top__tui_browser(&top); + exit_browser(0); + exit(0); + return NULL; +} + static void *display_thread(void *arg __used) { struct pollfd stdin_poll = { .fd = 0, .events = POLLIN }; @@ -1005,7 +1011,8 @@ static int __cmd_top(void) perf_session__mmap_read(session); - if (pthread_create(&thread, NULL, display_thread, session)) { + if (pthread_create(&thread, NULL, (use_browser > 0 ? display_thread_tui : + display_thread), session)) { printf("Could not create display thread.\n"); exit(-1); } @@ -1078,6 +1085,8 @@ static const struct option options[] = { "display this many functions"), OPT_BOOLEAN('U', "hide_user_symbols", &top.hide_user_symbols, "hide user symbols"), + OPT_BOOLEAN(0, "tui", &use_tui, "Use the TUI interface"), + OPT_BOOLEAN(0, "stdio", &use_stdio, "Use the stdio interface"), OPT_INCR('v', "verbose", &verbose, "be more verbose (show counter open errors, etc)"), OPT_END() @@ -1098,6 +1107,20 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) if (argc) usage_with_options(top_usage, options); + /* + * XXX For now start disabled, only using TUI if explicitely asked for. + * Change that when handle_keys equivalent gets written, live annotation + * done, etc. + */ + use_browser = 0; + + if (use_stdio) + use_browser = 0; + else if (use_tui) + use_browser = 1; + + setup_browser(false); + /* CPU and PID are mutually exclusive */ if (top.target_tid > 0 && top.cpu_list) { printf("WARNING: PID switch overriding CPU\n"); diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c index c06cc5386e7e..1d2e2652cd68 100644 --- a/tools/perf/util/top.c +++ b/tools/perf/util/top.c @@ -158,6 +158,7 @@ float perf_top__decay_samples(struct perf_top *top, struct rb_root *root) syme = list_entry(top->active_symbols.next, struct sym_entry, node); pthread_mutex_unlock(&top->active_symbols_lock); + top->rb_entries = 0; list_for_each_entry_safe_from(syme, n, &top->active_symbols, node) { syme->snap_count = syme->count[snap]; if (syme->snap_count != 0) { @@ -170,7 +171,11 @@ float perf_top__decay_samples(struct perf_top *top, struct rb_root *root) continue; } syme->weight = sym_weight(syme, top); - rb_insert_active_sym(root, syme); + + if ((int)syme->snap_count >= top->count_filter) { + rb_insert_active_sym(root, syme); + ++top->rb_entries; + } sum_ksamples += syme->snap_count; for (j = 0; j < top->evlist->nr_entries; j++) diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index 0467b26dd8d8..611370fa7df8 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -38,6 +38,11 @@ struct sym_entry { unsigned long count[0]; }; +static inline struct symbol *sym_entry__symbol(struct sym_entry *self) +{ + return ((void *)self) + symbol_conf.priv_size; +} + struct perf_top { struct perf_evlist *evlist; /* @@ -51,7 +56,7 @@ struct perf_top { u64 exact_samples; u64 guest_us_samples, guest_kernel_samples; int print_entries, count_filter, delay_secs; - int display_weighted, freq; + int display_weighted, freq, rb_entries; int sym_counter, target_pid, target_tid; bool hide_kernel_symbols, hide_user_symbols, zero; const char *cpu_list; @@ -64,4 +69,12 @@ float perf_top__decay_samples(struct perf_top *top, struct rb_root *root); void perf_top__find_widths(struct perf_top *top, struct rb_root *root, int *dso_width, int *dso_short_width, int *sym_width); +#ifdef NO_NEWT_SUPPORT +static inline int perf_top__tui_browser(struct perf_top *top __used) +{ + return 0; +} +#else +int perf_top__tui_browser(struct perf_top *top); +#endif #endif /* __PERF_TOP_H */ diff --git a/tools/perf/util/ui/browsers/top.c b/tools/perf/util/ui/browsers/top.c new file mode 100644 index 000000000000..ca6062483a8f --- /dev/null +++ b/tools/perf/util/ui/browsers/top.c @@ -0,0 +1,136 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ +#include "../browser.h" +#include "../helpline.h" +#include "../libslang.h" +#include "../../evlist.h" +#include "../../hist.h" +#include "../../sort.h" +#include "../../symbol.h" +#include "../../top.h" + +struct perf_top_browser { + struct ui_browser b; + struct rb_root root; + float sum_ksamples; + int dso_width; + int dso_short_width; + int sym_width; +}; + +static void perf_top_browser__write(struct ui_browser *browser, void *entry, int row) +{ + struct perf_top_browser *top_browser = container_of(browser, struct perf_top_browser, b); + struct sym_entry *syme = rb_entry(entry, struct sym_entry, rb_node); + bool current_entry = ui_browser__is_current_entry(browser, row); + struct symbol *symbol = sym_entry__symbol(syme); + struct perf_top *top = browser->priv; + int width = browser->width; + double pcnt; + + pcnt = 100.0 - (100.0 * ((top_browser->sum_ksamples - syme->snap_count) / + top_browser->sum_ksamples)); + ui_browser__set_percent_color(browser, pcnt, current_entry); + + if (top->evlist->nr_entries == 1 || !top->display_weighted) { + slsmg_printf("%20.2f ", syme->weight); + width -= 24; + } else { + slsmg_printf("%9.1f %10ld ", syme->weight, syme->snap_count); + width -= 23; + } + + slsmg_printf("%4.1f%%", pcnt); + width -= 7; + + if (verbose) { + slsmg_printf(" %016" PRIx64, symbol->start); + width -= 17; + } + + slsmg_printf(" %-*.*s ", top_browser->sym_width, top_browser->sym_width, + symbol->name); + width -= top_browser->sym_width; + slsmg_write_nstring(width >= syme->map->dso->long_name_len ? + syme->map->dso->long_name : + syme->map->dso->short_name, width); +} + +static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser) +{ + struct perf_top *top = browser->b.priv; + + browser->root = RB_ROOT; + browser->b.top = NULL; + browser->sum_ksamples = perf_top__decay_samples(top, &browser->root); + perf_top__find_widths(top, &browser->root, &browser->dso_width, + &browser->dso_short_width, + &browser->sym_width); + if (browser->sym_width + browser->dso_width > browser->b.width - 29) { + browser->dso_width = browser->dso_short_width; + if (browser->sym_width + browser->dso_width > browser->b.width - 29) + browser->sym_width = browser->b.width - browser->dso_width - 29; + } + browser->b.nr_entries = top->rb_entries; +} + +static int perf_top_browser__run(struct perf_top_browser *browser) +{ + int key; + char title[160]; + struct perf_top *top = browser->b.priv; + int delay_msecs = top->delay_secs * 1000; + + perf_top_browser__update_rb_tree(browser); + perf_top__header_snprintf(top, title, sizeof(title)); + perf_top__reset_sample_counters(top); + + if (ui_browser__show(&browser->b, title, "ESC: exit") < 0) + return -1; + + newtFormSetTimer(browser->b.form, delay_msecs); + + while (1) { + key = ui_browser__run(&browser->b); + + switch (key) { + case -1: + /* FIXME we need to check if it was es.reason == NEWT_EXIT_TIMER */ + perf_top_browser__update_rb_tree(browser); + perf_top__header_snprintf(top, title, sizeof(title)); + perf_top__reset_sample_counters(top); + ui_browser__set_color(&browser->b, NEWT_COLORSET_ROOT); + SLsmg_gotorc(0, 0); + slsmg_write_nstring(title, browser->b.width); + break; + case NEWT_KEY_TAB: + default: + goto out; + } + } +out: + ui_browser__hide(&browser->b); + return key; +} + +int perf_top__tui_browser(struct perf_top *top) +{ + struct perf_top_browser browser = { + .b = { + .entries = &browser.root, + .refresh = ui_browser__rb_tree_refresh, + .seek = ui_browser__rb_tree_seek, + .write = perf_top_browser__write, + .priv = top, + }, + }; + + ui_helpline__push("Press <- or ESC to exit"); + return perf_top_browser__run(&browser); +} -- cgit v1.2.3 From f6bbc1daac964da551130dbf01809d3fbd178b2d Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 31 Jan 2011 20:56:27 -0200 Subject: perf python: Fix build on 32-bit Where there are lots of errors related to python methods receiving 'char *' for things like file open mode, which break the build, also disable strict aliasing and fixup some other warnings. Now builds on both 32-bit and 64-bit fedora systems. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/python.c | 52 ++++++++++++++++++++++++------------------------ tools/perf/util/setup.py | 3 ++- 2 files changed, 28 insertions(+), 27 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c index d2d52175362e..5317ef229041 100644 --- a/tools/perf/util/python.c +++ b/tools/perf/util/python.c @@ -15,6 +15,8 @@ struct throttle_event { u64 stream_id; }; +PyMODINIT_FUNC initperf(void); + #define member_def(type, member, ptype, help) \ { #member, ptype, \ offsetof(struct pyrf_event, event) + offsetof(struct type, member), \ @@ -31,17 +33,15 @@ struct pyrf_event { union perf_event event; }; -#define T_ULONG_LONG T_ULONG - #define sample_members \ - sample_member_def(sample_ip, ip, T_ULONG_LONG, "event type"), \ + sample_member_def(sample_ip, ip, T_ULONGLONG, "event type"), \ sample_member_def(sample_pid, pid, T_INT, "event pid"), \ sample_member_def(sample_tid, tid, T_INT, "event tid"), \ - sample_member_def(sample_time, time, T_ULONG_LONG, "event timestamp"), \ - sample_member_def(sample_addr, addr, T_ULONG_LONG, "event addr"), \ - sample_member_def(sample_id, id, T_ULONG_LONG, "event id"), \ - sample_member_def(sample_stream_id, stream_id, T_ULONG_LONG, "event stream id"), \ - sample_member_def(sample_period, period, T_ULONG_LONG, "event period"), \ + sample_member_def(sample_time, time, T_ULONGLONG, "event timestamp"), \ + sample_member_def(sample_addr, addr, T_ULONGLONG, "event addr"), \ + sample_member_def(sample_id, id, T_ULONGLONG, "event id"), \ + sample_member_def(sample_stream_id, stream_id, T_ULONGLONG, "event stream id"), \ + sample_member_def(sample_period, period, T_ULONGLONG, "event period"), \ sample_member_def(sample_cpu, cpu, T_UINT, "event cpu"), static char pyrf_mmap_event__doc[] = PyDoc_STR("perf mmap event object."); @@ -51,11 +51,11 @@ static PyMemberDef pyrf_mmap_event__members[] = { member_def(perf_event_header, type, T_UINT, "event type"), member_def(mmap_event, pid, T_UINT, "event pid"), member_def(mmap_event, tid, T_UINT, "event tid"), - member_def(mmap_event, start, T_ULONG_LONG, "start of the map"), - member_def(mmap_event, len, T_ULONG_LONG, "map length"), - member_def(mmap_event, pgoff, T_ULONG_LONG, "page offset"), + member_def(mmap_event, start, T_ULONGLONG, "start of the map"), + member_def(mmap_event, len, T_ULONGLONG, "map length"), + member_def(mmap_event, pgoff, T_ULONGLONG, "page offset"), member_def(mmap_event, filename, T_STRING_INPLACE, "backing store"), - { NULL, }, + { .name = NULL, }, }; static PyObject *pyrf_mmap_event__repr(struct pyrf_event *pevent) @@ -96,8 +96,8 @@ static PyMemberDef pyrf_task_event__members[] = { member_def(fork_event, ppid, T_UINT, "event ppid"), member_def(fork_event, tid, T_UINT, "event tid"), member_def(fork_event, ptid, T_UINT, "event ptid"), - member_def(fork_event, time, T_ULONG_LONG, "timestamp"), - { NULL, }, + member_def(fork_event, time, T_ULONGLONG, "timestamp"), + { .name = NULL, }, }; static PyObject *pyrf_task_event__repr(struct pyrf_event *pevent) @@ -130,7 +130,7 @@ static PyMemberDef pyrf_comm_event__members[] = { member_def(comm_event, pid, T_UINT, "event pid"), member_def(comm_event, tid, T_UINT, "event tid"), member_def(comm_event, comm, T_STRING_INPLACE, "process name"), - { NULL, }, + { .name = NULL, }, }; static PyObject *pyrf_comm_event__repr(struct pyrf_event *pevent) @@ -156,10 +156,10 @@ static char pyrf_throttle_event__doc[] = PyDoc_STR("perf throttle event object." static PyMemberDef pyrf_throttle_event__members[] = { sample_members member_def(perf_event_header, type, T_UINT, "event type"), - member_def(throttle_event, time, T_ULONG_LONG, "timestamp"), - member_def(throttle_event, id, T_ULONG_LONG, "event id"), - member_def(throttle_event, stream_id, T_ULONG_LONG, "event stream id"), - { NULL, }, + member_def(throttle_event, time, T_ULONGLONG, "timestamp"), + member_def(throttle_event, id, T_ULONGLONG, "event id"), + member_def(throttle_event, stream_id, T_ULONGLONG, "event stream id"), + { .name = NULL, }, }; static PyObject *pyrf_throttle_event__repr(struct pyrf_event *pevent) @@ -522,7 +522,7 @@ static PyMethodDef pyrf_evsel__methods[] = { .ml_flags = METH_VARARGS | METH_KEYWORDS, .ml_doc = PyDoc_STR("open the event selector file descriptor table.") }, - { NULL, } + { .ml_name = NULL, } }; static char pyrf_evsel__doc[] = PyDoc_STR("perf event selector list object."); @@ -551,7 +551,7 @@ struct pyrf_evlist { }; static int pyrf_evlist__init(struct pyrf_evlist *pevlist, - PyObject *args, PyObject *kwargs) + PyObject *args, PyObject *kwargs __used) { PyObject *pcpus = NULL, *pthreads = NULL; struct cpu_map *cpus; @@ -613,7 +613,7 @@ static PyObject *pyrf_evlist__poll(struct pyrf_evlist *pevlist, } static PyObject *pyrf_evlist__get_pollfd(struct pyrf_evlist *pevlist, - PyObject *args, PyObject *kwargs) + PyObject *args __used, PyObject *kwargs __used) { struct perf_evlist *evlist = &pevlist->evlist; PyObject *list = PyList_New(0); @@ -645,7 +645,7 @@ free_list: static PyObject *pyrf_evlist__add(struct pyrf_evlist *pevlist, - PyObject *args, PyObject *kwargs) + PyObject *args, PyObject *kwargs __used) { struct perf_evlist *evlist = &pevlist->evlist; PyObject *pevsel; @@ -724,7 +724,7 @@ static PyMethodDef pyrf_evlist__methods[] = { .ml_flags = METH_VARARGS | METH_KEYWORDS, .ml_doc = PyDoc_STR("reads an event.") }, - { NULL, } + { .ml_name = NULL, } }; static Py_ssize_t pyrf_evlist__length(PyObject *obj) @@ -840,11 +840,11 @@ static struct { { "RECORD_FORK", PERF_RECORD_FORK }, { "RECORD_READ", PERF_RECORD_READ }, { "RECORD_SAMPLE", PERF_RECORD_SAMPLE }, - { NULL, }, + { .name = NULL, }, }; static PyMethodDef perf__methods[] = { - { NULL, NULL } + { .ml_name = NULL, } }; PyMODINIT_FUNC initperf(void) diff --git a/tools/perf/util/setup.py b/tools/perf/util/setup.py index 496d7f432fb8..1947b0430c94 100644 --- a/tools/perf/util/setup.py +++ b/tools/perf/util/setup.py @@ -6,7 +6,8 @@ perf = Extension('perf', sources = ['util/python.c', 'util/ctype.c', 'util/evlist.c', 'util/evsel.c', 'util/cpumap.c', 'util/thread_map.c', 'util/util.c', 'util/xyarray.c'], - include_dirs = ['util/include']) + include_dirs = ['util/include'], + extra_compile_args = ['-fno-strict-aliasing', '-Wno-write-strings']) setup(name='perf', version='0.1', -- cgit v1.2.3 From cdb0861c85c03fe80f4da033aab69df949579dc6 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 1 Feb 2011 10:51:23 -0800 Subject: perf top: Fix TUI compilation > + slsmg_write_nstring(width >= syme->map->dso->long_name_len ? > + syme->map->dso->long_name : > + syme->map->dso->short_name, width); need update macro for that calling util/ui/browsers/top.c: In function ‘perf_top_browser__write’: util/ui/browsers/top.c:60:2: error: cast to pointer from integer of different size util/ui/browsers/top.c:60:2: error: comparison between pointer and integer util/ui/browsers/top.c:60:2: error: passing argument 1 of ‘SLsmg_write_nstring’ discards qualifiers from pointer target type /usr/include/slang.h:1728:16: note: expected ‘char *’ but argument is of type ‘const char *’ make: *** [util/ui/browsers/top.o] Error 1 Cc: Frederic Weisbecker Cc: H. Peter Anvin Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Thomas Gleixner Cc: Tom Zanussi LKML-Reference: <4D48562B.20006@kernel.org> Signed-off-by: Yinghai Lu Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/libslang.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/libslang.h b/tools/perf/util/ui/libslang.h index 5623da8e8080..2b63e1c9b181 100644 --- a/tools/perf/util/ui/libslang.h +++ b/tools/perf/util/ui/libslang.h @@ -13,11 +13,11 @@ #if SLANG_VERSION < 20104 #define slsmg_printf(msg, args...) \ - SLsmg_printf((char *)msg, ##args) + SLsmg_printf((char *)(msg), ##args) #define slsmg_write_nstring(msg, len) \ - SLsmg_write_nstring((char *)msg, len) + SLsmg_write_nstring((char *)(msg), len) #define sltt_set_color(obj, name, fg, bg) \ - SLtt_set_color(obj,(char *)name, (char *)fg, (char *)bg) + SLtt_set_color(obj,(char *)(name), (char *)(fg), (char *)(bg)) #else #define slsmg_printf SLsmg_printf #define slsmg_write_nstring SLsmg_write_nstring -- cgit v1.2.3 From 764328d3209dd81b02a55722556b07b6f35e3ca0 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 4 Feb 2011 07:33:24 -0200 Subject: perf top: Remove superfluous name_len field From the sym_entry struct, struct symbol already has this field. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-top.c | 3 --- tools/perf/util/top.c | 5 +++-- tools/perf/util/top.h | 1 - 3 files changed, 3 insertions(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 104de9ab314c..154e088588bc 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -787,9 +787,6 @@ static int symbol_filter(struct map *map, struct symbol *sym) } } - if (!syme->skip) - syme->name_len = strlen(sym->name); - return 0; } diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c index 1d2e2652cd68..70a9c13f4ad5 100644 --- a/tools/perf/util/top.c +++ b/tools/perf/util/top.c @@ -200,6 +200,7 @@ void perf_top__find_widths(struct perf_top *top, struct rb_root *root, for (nd = rb_first(root); nd; nd = rb_next(nd)) { struct sym_entry *syme = rb_entry(nd, struct sym_entry, rb_node); + struct symbol *sym = sym_entry__symbol(syme); if (++printed > top->print_entries || (int)syme->snap_count < top->count_filter) @@ -211,7 +212,7 @@ void perf_top__find_widths(struct perf_top *top, struct rb_root *root, if (syme->map->dso->short_name_len > *dso_short_width) *dso_short_width = syme->map->dso->short_name_len; - if (syme->name_len > *sym_width) - *sym_width = syme->name_len; + if (sym->namelen > *sym_width) + *sym_width = sym->namelen; } } diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index 611370fa7df8..500950818d2f 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -31,7 +31,6 @@ struct sym_entry { unsigned long snap_count; double weight; int skip; - u16 name_len; u8 origin; struct map *map; struct sym_entry_source *src; -- cgit v1.2.3 From 78f7defedbb4da73b9a07635c357c1afcaa55c8f Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 4 Feb 2011 09:45:46 -0200 Subject: perf annotate: Move annotate functions to util/ They will be used by perf top, so that we have just one set of routines to do annotation. Rename "struct sym_priv" to "struct annotation", etc, to clarify this code a bit. Rename "struct sym_ext" to "struct source_line", to give it a meaningful name, that clarifies that it is a the result of an addr2line call, that is sorted by percentage one particular source code line appeared in the annotation. And since we're moving things around also rename 'sym_hist->ip' to 'sym_hist->addr' as we want to do data structure annotation at some point. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Makefile | 2 + tools/perf/builtin-annotate.c | 255 +----------------- tools/perf/builtin-report.c | 3 +- tools/perf/util/annotate.c | 467 +++++++++++++++++++++++++++++++++ tools/perf/util/annotate.h | 65 +++++ tools/perf/util/hist.c | 219 +--------------- tools/perf/util/hist.h | 27 -- tools/perf/util/ui/browsers/annotate.c | 43 +-- 8 files changed, 578 insertions(+), 503 deletions(-) create mode 100644 tools/perf/util/annotate.c create mode 100644 tools/perf/util/annotate.h (limited to 'tools/perf/util') diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 4c9499cb4398..be3eb1dc9a5a 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -401,6 +401,7 @@ LIB_H += util/include/dwarf-regs.h LIB_H += util/include/asm/dwarf2.h LIB_H += util/include/asm/cpufeature.h LIB_H += perf.h +LIB_H += util/annotate.h LIB_H += util/cache.h LIB_H += util/callchain.h LIB_H += util/build-id.h @@ -444,6 +445,7 @@ LIB_H += $(ARCH_INCLUDE) LIB_OBJS += $(OUTPUT)util/abspath.o LIB_OBJS += $(OUTPUT)util/alias.o +LIB_OBJS += $(OUTPUT)util/annotate.o LIB_OBJS += $(OUTPUT)util/build-id.o LIB_OBJS += $(OUTPUT)util/config.o LIB_OBJS += $(OUTPUT)util/ctype.o diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index cd9dec46c19f..9072ef44cfcb 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -9,6 +9,7 @@ #include "util/util.h" +#include "util/util.h" #include "util/color.h" #include #include "util/cache.h" @@ -18,6 +19,7 @@ #include "perf.h" #include "util/debug.h" +#include "util/annotate.h" #include "util/event.h" #include "util/parse-options.h" #include "util/parse-events.h" @@ -79,245 +81,10 @@ static int process_sample_event(union perf_event *event, return 0; } -static int objdump_line__print(struct objdump_line *self, - struct list_head *head, - struct hist_entry *he, u64 len) -{ - struct symbol *sym = he->ms.sym; - static const char *prev_line; - static const char *prev_color; - - if (self->offset != -1) { - const char *path = NULL; - unsigned int hits = 0; - double percent = 0.0; - const char *color; - struct sym_priv *priv = symbol__priv(sym); - struct sym_ext *sym_ext = priv->ext; - struct sym_hist *h = priv->hist; - s64 offset = self->offset; - struct objdump_line *next = objdump__get_next_ip_line(head, self); - - while (offset < (s64)len && - (next == NULL || offset < next->offset)) { - if (sym_ext) { - if (path == NULL) - path = sym_ext[offset].path; - percent += sym_ext[offset].percent; - } else - hits += h->ip[offset]; - - ++offset; - } - - if (sym_ext == NULL && h->sum) - percent = 100.0 * hits / h->sum; - - color = get_percent_color(percent); - - /* - * Also color the filename and line if needed, with - * the same color than the percentage. Don't print it - * twice for close colored ip with the same filename:line - */ - if (path) { - if (!prev_line || strcmp(prev_line, path) - || color != prev_color) { - color_fprintf(stdout, color, " %s", path); - prev_line = path; - prev_color = color; - } - } - - color_fprintf(stdout, color, " %7.2f", percent); - printf(" : "); - color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", self->line); - } else { - if (!*self->line) - printf(" :\n"); - else - printf(" : %s\n", self->line); - } - - return 0; -} - -static struct rb_root root_sym_ext; - -static void insert_source_line(struct sym_ext *sym_ext) -{ - struct sym_ext *iter; - struct rb_node **p = &root_sym_ext.rb_node; - struct rb_node *parent = NULL; - - while (*p != NULL) { - parent = *p; - iter = rb_entry(parent, struct sym_ext, node); - - if (sym_ext->percent > iter->percent) - p = &(*p)->rb_left; - else - p = &(*p)->rb_right; - } - - rb_link_node(&sym_ext->node, parent, p); - rb_insert_color(&sym_ext->node, &root_sym_ext); -} - -static void free_source_line(struct hist_entry *he, int len) -{ - struct sym_priv *priv = symbol__priv(he->ms.sym); - struct sym_ext *sym_ext = priv->ext; - int i; - - if (!sym_ext) - return; - - for (i = 0; i < len; i++) - free(sym_ext[i].path); - free(sym_ext); - - priv->ext = NULL; - root_sym_ext = RB_ROOT; -} - -/* Get the filename:line for the colored entries */ -static void -get_source_line(struct hist_entry *he, int len, const char *filename) -{ - struct symbol *sym = he->ms.sym; - u64 start; - int i; - char cmd[PATH_MAX * 2]; - struct sym_ext *sym_ext; - struct sym_priv *priv = symbol__priv(sym); - struct sym_hist *h = priv->hist; - - if (!h->sum) - return; - - sym_ext = priv->ext = calloc(len, sizeof(struct sym_ext)); - if (!priv->ext) - return; - - start = he->ms.map->unmap_ip(he->ms.map, sym->start); - - for (i = 0; i < len; i++) { - char *path = NULL; - size_t line_len; - u64 offset; - FILE *fp; - - sym_ext[i].percent = 100.0 * h->ip[i] / h->sum; - if (sym_ext[i].percent <= 0.5) - continue; - - offset = start + i; - sprintf(cmd, "addr2line -e %s %016" PRIx64, filename, offset); - fp = popen(cmd, "r"); - if (!fp) - continue; - - if (getline(&path, &line_len, fp) < 0 || !line_len) - goto next; - - sym_ext[i].path = malloc(sizeof(char) * line_len + 1); - if (!sym_ext[i].path) - goto next; - - strcpy(sym_ext[i].path, path); - insert_source_line(&sym_ext[i]); - - next: - pclose(fp); - } -} - -static void print_summary(const char *filename) -{ - struct sym_ext *sym_ext; - struct rb_node *node; - - printf("\nSorted summary for file %s\n", filename); - printf("----------------------------------------------\n\n"); - - if (RB_EMPTY_ROOT(&root_sym_ext)) { - printf(" Nothing higher than %1.1f%%\n", MIN_GREEN); - return; - } - - node = rb_first(&root_sym_ext); - while (node) { - double percent; - const char *color; - char *path; - - sym_ext = rb_entry(node, struct sym_ext, node); - percent = sym_ext->percent; - color = get_percent_color(percent); - path = sym_ext->path; - - color_fprintf(stdout, color, " %7.2f %s", percent, path); - node = rb_next(node); - } -} - -static void hist_entry__print_hits(struct hist_entry *self) -{ - struct symbol *sym = self->ms.sym; - struct sym_priv *priv = symbol__priv(sym); - struct sym_hist *h = priv->hist; - u64 len = sym->end - sym->start, offset; - - for (offset = 0; offset < len; ++offset) - if (h->ip[offset] != 0) - printf("%*" PRIx64 ": %" PRIu64 "\n", BITS_PER_LONG / 2, - sym->start + offset, h->ip[offset]); - printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); -} - static int hist_entry__tty_annotate(struct hist_entry *he) { - struct map *map = he->ms.map; - struct dso *dso = map->dso; - struct symbol *sym = he->ms.sym; - const char *filename = dso->long_name, *d_filename; - u64 len; - LIST_HEAD(head); - struct objdump_line *pos, *n; - - if (hist_entry__annotate(he, &head, 0) < 0) - return -1; - - if (full_paths) - d_filename = filename; - else - d_filename = basename(filename); - - len = sym->end - sym->start; - - if (print_line) { - get_source_line(he, len, filename); - print_summary(filename); - } - - printf("\n\n------------------------------------------------\n"); - printf(" Percent | Source code & Disassembly of %s\n", d_filename); - printf("------------------------------------------------\n"); - - if (verbose) - hist_entry__print_hits(he); - - list_for_each_entry_safe(pos, n, &head, node) { - objdump_line__print(pos, &head, he, len); - list_del(&pos->node); - objdump_line__free(pos); - } - - if (print_line) - free_source_line(he, len); - - return 0; + return symbol__tty_annotate(he->ms.sym, he->ms.map, + print_line, full_paths); } static void hists__find_annotations(struct hists *self) @@ -327,13 +94,13 @@ static void hists__find_annotations(struct hists *self) while (nd) { struct hist_entry *he = rb_entry(nd, struct hist_entry, rb_node); - struct sym_priv *priv; + struct annotation *notes; if (he->ms.sym == NULL || he->ms.map->dso->annotate_warned) goto find_next; - priv = symbol__priv(he->ms.sym); - if (priv->hist == NULL) { + notes = symbol__annotation(he->ms.sym); + if (notes->histogram == NULL) { find_next: if (key == KEY_LEFT) nd = rb_prev(nd); @@ -362,11 +129,11 @@ find_next: nd = rb_next(nd); /* * Since we have a hist_entry per IP for the same - * symbol, free he->ms.sym->hist to signal we already + * symbol, free he->ms.sym->histogram to signal we already * processed this symbol. */ - free(priv->hist); - priv->hist = NULL; + free(notes->histogram); + notes->histogram = NULL; } } } @@ -454,7 +221,7 @@ int cmd_annotate(int argc, const char **argv, const char *prefix __used) setup_browser(true); - symbol_conf.priv_size = sizeof(struct sym_priv); + symbol_conf.priv_size = sizeof(struct annotation); symbol_conf.try_vmlinux_path = true; if (symbol__init() < 0) diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index 080937c3a656..91e4cdba933b 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -9,6 +9,7 @@ #include "util/util.h" +#include "util/annotate.h" #include "util/color.h" #include #include "util/cache.h" @@ -508,7 +509,7 @@ int cmd_report(int argc, const char **argv, const char *prefix __used) * implementation. */ if (use_browser > 0) { - symbol_conf.priv_size = sizeof(struct sym_priv); + symbol_conf.priv_size = sizeof(struct annotation); /* * For searching by name on the "Browse map details". * providing it only in verbose mode not to bloat too diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c new file mode 100644 index 000000000000..9b25575b980c --- /dev/null +++ b/tools/perf/util/annotate.c @@ -0,0 +1,467 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo + * + * Parts came from builtin-annotate.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + +#include "util.h" +#include "build-id.h" +#include "color.h" +#include "cache.h" +#include "symbol.h" +#include "debug.h" +#include "annotate.h" + +static int symbol__alloc_hist(struct symbol *sym) +{ + struct annotation *notes = symbol__annotation(sym); + const int size = (sizeof(*notes->histogram) + + (sym->end - sym->start) * sizeof(u64)); + + notes->histogram = zalloc(size); + return notes->histogram == NULL ? -1 : 0; +} + +int symbol__inc_addr_samples(struct symbol *sym, struct map *map, u64 addr) +{ + unsigned int sym_size, offset; + struct annotation *notes; + struct sym_hist *h; + + if (!sym || !map) + return 0; + + notes = symbol__annotation(sym); + if (notes->histogram == NULL && symbol__alloc_hist(sym) < 0) + return -ENOMEM; + + sym_size = sym->end - sym->start; + offset = addr - sym->start; + + pr_debug3("%s: addr=%#" PRIx64 "\n", __func__, map->unmap_ip(map, addr)); + + if (offset >= sym_size) + return 0; + + h = notes->histogram; + h->sum++; + h->addr[offset]++; + + pr_debug3("%#" PRIx64 " %s: period++ [addr: %#" PRIx64 ", %#" PRIx64 + "] => %" PRIu64 "\n", sym->start, sym->name, + addr, addr - sym->start, h->addr[offset]); + return 0; +} + +static struct objdump_line *objdump_line__new(s64 offset, char *line, size_t privsize) +{ + struct objdump_line *self = malloc(sizeof(*self) + privsize); + + if (self != NULL) { + self->offset = offset; + self->line = line; + } + + return self; +} + +void objdump_line__free(struct objdump_line *self) +{ + free(self->line); + free(self); +} + +static void objdump__add_line(struct list_head *head, struct objdump_line *line) +{ + list_add_tail(&line->node, head); +} + +struct objdump_line *objdump__get_next_ip_line(struct list_head *head, + struct objdump_line *pos) +{ + list_for_each_entry_continue(pos, head, node) + if (pos->offset >= 0) + return pos; + + return NULL; +} + +static void objdump_line__print(struct objdump_line *oline, + struct list_head *head, + struct symbol *sym, u64 len) +{ + static const char *prev_line; + static const char *prev_color; + + if (oline->offset != -1) { + const char *path = NULL; + unsigned int hits = 0; + double percent = 0.0; + const char *color; + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src_line; + struct sym_hist *h = notes->histogram; + s64 offset = oline->offset; + struct objdump_line *next = objdump__get_next_ip_line(head, oline); + + while (offset < (s64)len && + (next == NULL || offset < next->offset)) { + if (src_line) { + if (path == NULL) + path = src_line[offset].path; + percent += src_line[offset].percent; + } else + hits += h->addr[offset]; + + ++offset; + } + + if (src_line == NULL && h->sum) + percent = 100.0 * hits / h->sum; + + color = get_percent_color(percent); + + /* + * Also color the filename and line if needed, with + * the same color than the percentage. Don't print it + * twice for close colored addr with the same filename:line + */ + if (path) { + if (!prev_line || strcmp(prev_line, path) + || color != prev_color) { + color_fprintf(stdout, color, " %s", path); + prev_line = path; + prev_color = color; + } + } + + color_fprintf(stdout, color, " %7.2f", percent); + printf(" : "); + color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", oline->line); + } else { + if (!*oline->line) + printf(" :\n"); + else + printf(" : %s\n", oline->line); + } +} + +static int symbol__parse_objdump_line(struct symbol *sym, struct map *map, FILE *file, + struct list_head *head, size_t privsize) +{ + struct objdump_line *objdump_line; + char *line = NULL, *tmp, *tmp2, *c; + size_t line_len; + s64 line_ip, offset = -1; + + if (getline(&line, &line_len, file) < 0) + return -1; + + if (!line) + return -1; + + while (line_len != 0 && isspace(line[line_len - 1])) + line[--line_len] = '\0'; + + c = strchr(line, '\n'); + if (c) + *c = 0; + + line_ip = -1; + + /* + * Strip leading spaces: + */ + tmp = line; + while (*tmp) { + if (*tmp != ' ') + break; + tmp++; + } + + if (*tmp) { + /* + * Parse hexa addresses followed by ':' + */ + line_ip = strtoull(tmp, &tmp2, 16); + if (*tmp2 != ':' || tmp == tmp2 || tmp2[1] == '\0') + line_ip = -1; + } + + if (line_ip != -1) { + u64 start = map__rip_2objdump(map, sym->start), + end = map__rip_2objdump(map, sym->end); + + offset = line_ip - start; + if (offset < 0 || (u64)line_ip > end) + offset = -1; + } + + objdump_line = objdump_line__new(offset, line, privsize); + if (objdump_line == NULL) { + free(line); + return -1; + } + objdump__add_line(head, objdump_line); + + return 0; +} + +int symbol__annotate(struct symbol *sym, struct map *map, + struct list_head *head, size_t privsize) +{ + struct dso *dso = map->dso; + char *filename = dso__build_id_filename(dso, NULL, 0); + bool free_filename = true; + char command[PATH_MAX * 2]; + FILE *file; + int err = 0; + u64 len; + char symfs_filename[PATH_MAX]; + + if (filename) { + snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", + symbol_conf.symfs, filename); + } + + if (filename == NULL) { + if (dso->has_build_id) { + pr_err("Can't annotate %s: not enough memory\n", + sym->name); + return -ENOMEM; + } + goto fallback; + } else if (readlink(symfs_filename, command, sizeof(command)) < 0 || + strstr(command, "[kernel.kallsyms]") || + access(symfs_filename, R_OK)) { + free(filename); +fallback: + /* + * If we don't have build-ids or the build-id file isn't in the + * cache, or is just a kallsyms file, well, lets hope that this + * DSO is the same as when 'perf record' ran. + */ + filename = dso->long_name; + snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", + symbol_conf.symfs, filename); + free_filename = false; + } + + if (dso->origin == DSO__ORIG_KERNEL) { + if (dso->annotate_warned) + goto out_free_filename; + err = -ENOENT; + dso->annotate_warned = 1; + pr_err("Can't annotate %s: No vmlinux file was found in the " + "path\n", sym->name); + goto out_free_filename; + } + + pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__, + filename, sym->name, map->unmap_ip(map, sym->start), + map->unmap_ip(map, sym->end)); + + len = sym->end - sym->start; + + pr_debug("annotating [%p] %30s : [%p] %30s\n", + dso, dso->long_name, sym, sym->name); + + snprintf(command, sizeof(command), + "objdump --start-address=0x%016" PRIx64 + " --stop-address=0x%016" PRIx64 " -dS -C %s|grep -v %s|expand", + map__rip_2objdump(map, sym->start), + map__rip_2objdump(map, sym->end), + symfs_filename, filename); + + pr_debug("Executing: %s\n", command); + + file = popen(command, "r"); + if (!file) + goto out_free_filename; + + while (!feof(file)) + if (symbol__parse_objdump_line(sym, map, file, head, privsize) < 0) + break; + + pclose(file); +out_free_filename: + if (free_filename) + free(filename); + return err; +} + +static void insert_source_line(struct rb_root *root, struct source_line *src_line) +{ + struct source_line *iter; + struct rb_node **p = &root->rb_node; + struct rb_node *parent = NULL; + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct source_line, node); + + if (src_line->percent > iter->percent) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&src_line->node, parent, p); + rb_insert_color(&src_line->node, root); +} + +static void symbol__free_source_line(struct symbol *sym, int len) +{ + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src_line; + int i; + + for (i = 0; i < len; i++) + free(src_line[i].path); + + free(src_line); + notes->src_line = NULL; +} + +/* Get the filename:line for the colored entries */ +static int symbol__get_source_line(struct symbol *sym, struct map *map, + struct rb_root *root, int len, + const char *filename) +{ + u64 start; + int i; + char cmd[PATH_MAX * 2]; + struct source_line *src_line; + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = notes->histogram; + + if (!h->sum) + return 0; + + src_line = notes->src_line = calloc(len, sizeof(struct source_line)); + if (!notes->src_line) + return -1; + + start = map->unmap_ip(map, sym->start); + + for (i = 0; i < len; i++) { + char *path = NULL; + size_t line_len; + u64 offset; + FILE *fp; + + src_line[i].percent = 100.0 * h->addr[i] / h->sum; + if (src_line[i].percent <= 0.5) + continue; + + offset = start + i; + sprintf(cmd, "addr2line -e %s %016" PRIx64, filename, offset); + fp = popen(cmd, "r"); + if (!fp) + continue; + + if (getline(&path, &line_len, fp) < 0 || !line_len) + goto next; + + src_line[i].path = malloc(sizeof(char) * line_len + 1); + if (!src_line[i].path) + goto next; + + strcpy(src_line[i].path, path); + insert_source_line(root, &src_line[i]); + + next: + pclose(fp); + } + + return 0; +} + +static void print_summary(struct rb_root *root, const char *filename) +{ + struct source_line *src_line; + struct rb_node *node; + + printf("\nSorted summary for file %s\n", filename); + printf("----------------------------------------------\n\n"); + + if (RB_EMPTY_ROOT(root)) { + printf(" Nothing higher than %1.1f%%\n", MIN_GREEN); + return; + } + + node = rb_first(root); + while (node) { + double percent; + const char *color; + char *path; + + src_line = rb_entry(node, struct source_line, node); + percent = src_line->percent; + color = get_percent_color(percent); + path = src_line->path; + + color_fprintf(stdout, color, " %7.2f %s", percent, path); + node = rb_next(node); + } +} + +static void symbol__annotate_hits(struct symbol *sym) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = notes->histogram; + u64 len = sym->end - sym->start, offset; + + for (offset = 0; offset < len; ++offset) + if (h->addr[offset] != 0) + printf("%*" PRIx64 ": %" PRIu64 "\n", BITS_PER_LONG / 2, + sym->start + offset, h->addr[offset]); + printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); +} + +int symbol__tty_annotate(struct symbol *sym, struct map *map, bool print_lines, + bool full_paths) +{ + struct dso *dso = map->dso; + const char *filename = dso->long_name, *d_filename; + struct rb_root source_line = RB_ROOT; + struct objdump_line *pos, *n; + LIST_HEAD(head); + u64 len; + + if (symbol__annotate(sym, map, &head, 0) < 0) + return -1; + + if (full_paths) + d_filename = filename; + else + d_filename = basename(filename); + + len = sym->end - sym->start; + + if (print_lines) { + symbol__get_source_line(sym, map, &source_line, len, filename); + print_summary(&source_line, filename); + } + + printf("\n\n------------------------------------------------\n"); + printf(" Percent | Source code & Disassembly of %s\n", d_filename); + printf("------------------------------------------------\n"); + + if (verbose) + symbol__annotate_hits(sym); + + list_for_each_entry_safe(pos, n, &head, node) { + objdump_line__print(pos, &head, sym, len); + list_del(&pos->node); + objdump_line__free(pos); + } + + if (print_lines) + symbol__free_source_line(sym, len); + + return 0; +} diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h new file mode 100644 index 000000000000..6e2fbc205299 --- /dev/null +++ b/tools/perf/util/annotate.h @@ -0,0 +1,65 @@ +#ifndef __PERF_ANNOTATE_H +#define __PERF_ANNOTATE_H + +#include +#include "types.h" +#include "symbol.h" +#include +#include + +struct objdump_line { + struct list_head node; + s64 offset; + char *line; +}; + +void objdump_line__free(struct objdump_line *self); +struct objdump_line *objdump__get_next_ip_line(struct list_head *head, + struct objdump_line *pos); + +struct sym_hist { + u64 sum; + u64 addr[0]; +}; + +struct source_line { + struct rb_node node; + double percent; + char *path; +}; + +struct annotation { + struct sym_hist *histogram; + struct source_line *src_line; +}; + +struct sannotation { + struct annotation annotation; + struct symbol symbol; +}; + +static inline struct annotation *symbol__annotation(struct symbol *sym) +{ + struct sannotation *a = container_of(sym, struct sannotation, symbol); + return &a->annotation; +} + +int symbol__inc_addr_samples(struct symbol *sym, struct map *map, u64 addr); + +int symbol__annotate(struct symbol *sym, struct map *map, + struct list_head *head, size_t privsize); + +int symbol__tty_annotate(struct symbol *sym, struct map *map, + bool print_lines, bool full_paths); + +#ifdef NO_NEWT_SUPPORT +static inline int symbol__tui_annotate(symbol *sym __used, + struct map *map __used) +{ + return 0; +} +#else +int symbol__tui_annotate(struct symbol *sym, struct map *map); +#endif + +#endif /* __PERF_ANNOTATE_H */ diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 95887804dc8e..6d9c92c3d7cb 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -1,3 +1,4 @@ +#include "annotate.h" #include "util.h" #include "build-id.h" #include "hist.h" @@ -949,225 +950,15 @@ void hists__filter_by_thread(struct hists *self, const struct thread *thread) } } -static int symbol__alloc_hist(struct symbol *self) +int hist_entry__inc_addr_samples(struct hist_entry *he, u64 ip) { - struct sym_priv *priv = symbol__priv(self); - const int size = (sizeof(*priv->hist) + - (self->end - self->start) * sizeof(u64)); - - priv->hist = zalloc(size); - return priv->hist == NULL ? -1 : 0; -} - -int hist_entry__inc_addr_samples(struct hist_entry *self, u64 ip) -{ - unsigned int sym_size, offset; - struct symbol *sym = self->ms.sym; - struct sym_priv *priv; - struct sym_hist *h; - - if (!sym || !self->ms.map) - return 0; - - priv = symbol__priv(sym); - if (priv->hist == NULL && symbol__alloc_hist(sym) < 0) - return -ENOMEM; - - sym_size = sym->end - sym->start; - offset = ip - sym->start; - - pr_debug3("%s: ip=%#" PRIx64 "\n", __func__, self->ms.map->unmap_ip(self->ms.map, ip)); - - if (offset >= sym_size) - return 0; - - h = priv->hist; - h->sum++; - h->ip[offset]++; - - pr_debug3("%#" PRIx64 " %s: period++ [ip: %#" PRIx64 ", %#" PRIx64 - "] => %" PRIu64 "\n", self->ms.sym->start, self->ms.sym->name, - ip, ip - self->ms.sym->start, h->ip[offset]); - return 0; -} - -static struct objdump_line *objdump_line__new(s64 offset, char *line, size_t privsize) -{ - struct objdump_line *self = malloc(sizeof(*self) + privsize); - - if (self != NULL) { - self->offset = offset; - self->line = line; - } - - return self; -} - -void objdump_line__free(struct objdump_line *self) -{ - free(self->line); - free(self); -} - -static void objdump__add_line(struct list_head *head, struct objdump_line *line) -{ - list_add_tail(&line->node, head); -} - -struct objdump_line *objdump__get_next_ip_line(struct list_head *head, - struct objdump_line *pos) -{ - list_for_each_entry_continue(pos, head, node) - if (pos->offset >= 0) - return pos; - - return NULL; + return symbol__inc_addr_samples(he->ms.sym, he->ms.map, ip); } -static int hist_entry__parse_objdump_line(struct hist_entry *self, FILE *file, - struct list_head *head, size_t privsize) -{ - struct symbol *sym = self->ms.sym; - struct objdump_line *objdump_line; - char *line = NULL, *tmp, *tmp2, *c; - size_t line_len; - s64 line_ip, offset = -1; - - if (getline(&line, &line_len, file) < 0) - return -1; - - if (!line) - return -1; - - while (line_len != 0 && isspace(line[line_len - 1])) - line[--line_len] = '\0'; - - c = strchr(line, '\n'); - if (c) - *c = 0; - - line_ip = -1; - - /* - * Strip leading spaces: - */ - tmp = line; - while (*tmp) { - if (*tmp != ' ') - break; - tmp++; - } - - if (*tmp) { - /* - * Parse hexa addresses followed by ':' - */ - line_ip = strtoull(tmp, &tmp2, 16); - if (*tmp2 != ':' || tmp == tmp2 || tmp2[1] == '\0') - line_ip = -1; - } - - if (line_ip != -1) { - u64 start = map__rip_2objdump(self->ms.map, sym->start), - end = map__rip_2objdump(self->ms.map, sym->end); - - offset = line_ip - start; - if (offset < 0 || (u64)line_ip > end) - offset = -1; - } - - objdump_line = objdump_line__new(offset, line, privsize); - if (objdump_line == NULL) { - free(line); - return -1; - } - objdump__add_line(head, objdump_line); - - return 0; -} - -int hist_entry__annotate(struct hist_entry *self, struct list_head *head, +int hist_entry__annotate(struct hist_entry *he, struct list_head *head, size_t privsize) { - struct symbol *sym = self->ms.sym; - struct map *map = self->ms.map; - struct dso *dso = map->dso; - char *filename = dso__build_id_filename(dso, NULL, 0); - bool free_filename = true; - char command[PATH_MAX * 2]; - FILE *file; - int err = 0; - u64 len; - char symfs_filename[PATH_MAX]; - - if (filename) { - snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", - symbol_conf.symfs, filename); - } - - if (filename == NULL) { - if (dso->has_build_id) { - pr_err("Can't annotate %s: not enough memory\n", - sym->name); - return -ENOMEM; - } - goto fallback; - } else if (readlink(symfs_filename, command, sizeof(command)) < 0 || - strstr(command, "[kernel.kallsyms]") || - access(symfs_filename, R_OK)) { - free(filename); -fallback: - /* - * If we don't have build-ids or the build-id file isn't in the - * cache, or is just a kallsyms file, well, lets hope that this - * DSO is the same as when 'perf record' ran. - */ - filename = dso->long_name; - snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", - symbol_conf.symfs, filename); - free_filename = false; - } - - if (dso->origin == DSO__ORIG_KERNEL) { - if (dso->annotate_warned) - goto out_free_filename; - err = -ENOENT; - dso->annotate_warned = 1; - pr_err("Can't annotate %s: No vmlinux file was found in the " - "path\n", sym->name); - goto out_free_filename; - } - - pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__, - filename, sym->name, map->unmap_ip(map, sym->start), - map->unmap_ip(map, sym->end)); - - len = sym->end - sym->start; - - pr_debug("annotating [%p] %30s : [%p] %30s\n", - dso, dso->long_name, sym, sym->name); - - snprintf(command, sizeof(command), - "objdump --start-address=0x%016" PRIx64 " --stop-address=0x%016" PRIx64 " -dS -C %s|grep -v %s|expand", - map__rip_2objdump(map, sym->start), - map__rip_2objdump(map, sym->end), - symfs_filename, filename); - - pr_debug("Executing: %s\n", command); - - file = popen(command, "r"); - if (!file) - goto out_free_filename; - - while (!feof(file)) - if (hist_entry__parse_objdump_line(self, file, head, privsize) < 0) - break; - - pclose(file); -out_free_filename: - if (free_filename) - free(filename); - return err; + return symbol__annotate(he->ms.sym, he->ms.map, head, privsize); } void hists__inc_nr_events(struct hists *self, u32 type) diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 889559b86492..8a201f755534 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -9,33 +9,6 @@ extern struct callchain_param callchain_param; struct hist_entry; struct addr_location; struct symbol; -struct rb_root; - -struct objdump_line { - struct list_head node; - s64 offset; - char *line; -}; - -void objdump_line__free(struct objdump_line *self); -struct objdump_line *objdump__get_next_ip_line(struct list_head *head, - struct objdump_line *pos); - -struct sym_hist { - u64 sum; - u64 ip[0]; -}; - -struct sym_ext { - struct rb_node node; - double percent; - char *path; -}; - -struct sym_priv { - struct sym_hist *hist; - struct sym_ext *ext; -}; /* * The kernel collects the number of events it couldn't send in a stretch and diff --git a/tools/perf/util/ui/browsers/annotate.c b/tools/perf/util/ui/browsers/annotate.c index 82b78f99251b..daa7138d8015 100644 --- a/tools/perf/util/ui/browsers/annotate.c +++ b/tools/perf/util/ui/browsers/annotate.c @@ -1,9 +1,11 @@ #include "../browser.h" #include "../helpline.h" #include "../libslang.h" +#include "../../annotate.h" #include "../../hist.h" #include "../../sort.h" #include "../../symbol.h" +#include "../../annotate.h" static void ui__error_window(const char *fmt, ...) { @@ -66,24 +68,26 @@ static double objdump_line__calc_percent(struct objdump_line *self, if (self->offset != -1) { int len = sym->end - sym->start; unsigned int hits = 0; - struct sym_priv *priv = symbol__priv(sym); - struct sym_ext *sym_ext = priv->ext; - struct sym_hist *h = priv->hist; + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src_line; + struct sym_hist *h = notes->histogram; s64 offset = self->offset; struct objdump_line *next = objdump__get_next_ip_line(head, self); - while (offset < (s64)len && (next == NULL || offset < next->offset)) { - if (sym_ext) { - percent += sym_ext[offset].percent; + if (src_line) { + percent += src_line[offset].percent; } else - hits += h->ip[offset]; + hits += h->addr[offset]; ++offset; } - - if (sym_ext == NULL && h->sum) + /* + * If the percentage wasn't already calculated in + * symbol__get_source_line, do it now: + */ + if (src_line == NULL && h->sum) percent = 100.0 * hits / h->sum; } @@ -136,10 +140,10 @@ static void annotate_browser__set_top(struct annotate_browser *self, static int annotate_browser__run(struct annotate_browser *self) { struct rb_node *nd; - struct hist_entry *he = self->b.priv; + struct symbol *sym = self->b.priv; int key; - if (ui_browser__show(&self->b, he->ms.sym->name, + if (ui_browser__show(&self->b, sym->name, "<-, -> or ESC: exit, TAB/shift+TAB: cycle thru samples") < 0) return -1; /* @@ -179,7 +183,12 @@ out: return key; } -int hist_entry__tui_annotate(struct hist_entry *self) +int hist_entry__tui_annotate(struct hist_entry *he) +{ + return symbol__tui_annotate(he->ms.sym, he->ms.map); +} + +int symbol__tui_annotate(struct symbol *sym, struct map *map) { struct objdump_line *pos, *n; struct objdump_line_rb_node *rbpos; @@ -190,18 +199,18 @@ int hist_entry__tui_annotate(struct hist_entry *self) .refresh = ui_browser__list_head_refresh, .seek = ui_browser__list_head_seek, .write = annotate_browser__write, - .priv = self, + .priv = sym, }, }; int ret; - if (self->ms.sym == NULL) + if (sym == NULL) return -1; - if (self->ms.map->dso->annotate_warned) + if (map->dso->annotate_warned) return -1; - if (hist_entry__annotate(self, &head, sizeof(*rbpos)) < 0) { + if (symbol__annotate(sym, map, &head, sizeof(*rbpos)) < 0) { ui__error_window(ui_helpline__last_msg); return -1; } @@ -214,7 +223,7 @@ int hist_entry__tui_annotate(struct hist_entry *self) browser.b.width = line_len; rbpos = objdump_line__rb(pos); rbpos->idx = browser.b.nr_entries++; - rbpos->percent = objdump_line__calc_percent(pos, &head, self->ms.sym); + rbpos->percent = objdump_line__calc_percent(pos, &head, sym); if (rbpos->percent < 0.01) continue; objdump__insert_line(&browser.entries, rbpos); -- cgit v1.2.3 From 2f525d0148ef2734c8a172201e5e1e9167a8a5fd Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 4 Feb 2011 13:43:24 -0200 Subject: perf annotate: Support multiple histograms in annotation The perf annotate tool continues aggregating everything on just one histograms, but to support the top model add support for one histogram perf evsel in the evlist. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-annotate.c | 29 ++++++++++++----- tools/perf/builtin-report.c | 15 +++++++-- tools/perf/util/annotate.c | 57 ++++++++++++++++------------------ tools/perf/util/annotate.h | 29 ++++++++++++++--- tools/perf/util/hist.c | 4 +-- tools/perf/util/hist.h | 16 +++++----- tools/perf/util/ui/browsers/annotate.c | 12 +++---- tools/perf/util/ui/browsers/hists.c | 9 +++--- 8 files changed, 106 insertions(+), 65 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index 9072ef44cfcb..f3e44231b10d 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -57,7 +57,18 @@ static int hists__add_entry(struct hists *self, struct addr_location *al) if (he == NULL) return -ENOMEM; - return hist_entry__inc_addr_samples(he, al->addr); + if (he->ms.sym != NULL) { + /* + * All aggregated on the first sym_hist. + */ + struct annotation *notes = symbol__annotation(he->ms.sym); + if (notes->histograms == NULL && symbol__alloc_hist(he->ms.sym, 1) < 0) + return -ENOMEM; + + return hist_entry__inc_addr_samples(he, 0, al->addr); + } + + return 0; } static int process_sample_event(union perf_event *event, @@ -81,9 +92,9 @@ static int process_sample_event(union perf_event *event, return 0; } -static int hist_entry__tty_annotate(struct hist_entry *he) +static int hist_entry__tty_annotate(struct hist_entry *he, int evidx) { - return symbol__tty_annotate(he->ms.sym, he->ms.map, + return symbol__tty_annotate(he->ms.sym, he->ms.map, evidx, print_line, full_paths); } @@ -100,7 +111,7 @@ static void hists__find_annotations(struct hists *self) goto find_next; notes = symbol__annotation(he->ms.sym); - if (notes->histogram == NULL) { + if (notes->histograms == NULL) { find_next: if (key == KEY_LEFT) nd = rb_prev(nd); @@ -110,7 +121,8 @@ find_next: } if (use_browser > 0) { - key = hist_entry__tui_annotate(he); + /* For now all is aggregated on the first */ + key = hist_entry__tui_annotate(he, 0); switch (key) { case KEY_RIGHT: next = rb_next(nd); @@ -125,15 +137,16 @@ find_next: if (next != NULL) nd = next; } else { - hist_entry__tty_annotate(he); + /* For now all is aggregated on the first */ + hist_entry__tty_annotate(he, 0); nd = rb_next(nd); /* * Since we have a hist_entry per IP for the same * symbol, free he->ms.sym->histogram to signal we already * processed this symbol. */ - free(notes->histogram); - notes->histogram = NULL; + free(notes->histograms); + notes->histograms = NULL; } } } diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index 91e4cdba933b..de06bf55efff 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -118,8 +118,17 @@ static int perf_session__add_hist_entry(struct perf_session *session, * so we don't allocated the extra space needed because the stdio * code will not use it. */ - if (use_browser > 0) - err = hist_entry__inc_addr_samples(he, al->addr); + if (al->sym != NULL && use_browser > 0) { + /* + * All aggregated on the first sym_hist. + */ + struct annotation *notes = symbol__annotation(he->ms.sym); + if (notes->histograms == NULL && + symbol__alloc_hist(he->ms.sym, 1) < 0) + err = -ENOMEM; + else + err = hist_entry__inc_addr_samples(he, 0, al->addr); + } return err; } @@ -349,7 +358,7 @@ static int __cmd_report(void) } if (use_browser > 0) - hists__tui_browse_tree(&session->hists_tree, help); + hists__tui_browse_tree(&session->hists_tree, help, 0); else hists__tty_browse_tree(&session->hists_tree, help); diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 9b25575b980c..7488fe99502c 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -15,44 +15,40 @@ #include "debug.h" #include "annotate.h" -static int symbol__alloc_hist(struct symbol *sym) +int symbol__alloc_hist(struct symbol *sym, int nevents) { struct annotation *notes = symbol__annotation(sym); - const int size = (sizeof(*notes->histogram) + - (sym->end - sym->start) * sizeof(u64)); - notes->histogram = zalloc(size); - return notes->histogram == NULL ? -1 : 0; + notes->sizeof_sym_hist = (sizeof(*notes->histograms) + + (sym->end - sym->start) * sizeof(u64)); + notes->histograms = calloc(nevents, notes->sizeof_sym_hist); + return notes->histograms == NULL ? -1 : 0; } -int symbol__inc_addr_samples(struct symbol *sym, struct map *map, u64 addr) +int symbol__inc_addr_samples(struct symbol *sym, struct map *map, + int evidx, u64 addr) { - unsigned int sym_size, offset; + unsigned offset; struct annotation *notes; struct sym_hist *h; - if (!sym || !map) - return 0; - notes = symbol__annotation(sym); - if (notes->histogram == NULL && symbol__alloc_hist(sym) < 0) + if (notes->histograms == NULL) return -ENOMEM; - sym_size = sym->end - sym->start; - offset = addr - sym->start; - pr_debug3("%s: addr=%#" PRIx64 "\n", __func__, map->unmap_ip(map, addr)); - if (offset >= sym_size) + if (addr >= sym->end) return 0; - h = notes->histogram; + offset = addr - sym->start; + h = annotation__histogram(notes, evidx); h->sum++; h->addr[offset]++; pr_debug3("%#" PRIx64 " %s: period++ [addr: %#" PRIx64 ", %#" PRIx64 - "] => %" PRIu64 "\n", sym->start, sym->name, - addr, addr - sym->start, h->addr[offset]); + ", evidx=%d] => %" PRIu64 "\n", sym->start, sym->name, + addr, addr - sym->start, evidx, h->addr[offset]); return 0; } @@ -90,8 +86,8 @@ struct objdump_line *objdump__get_next_ip_line(struct list_head *head, } static void objdump_line__print(struct objdump_line *oline, - struct list_head *head, - struct symbol *sym, u64 len) + struct list_head *head, struct symbol *sym, + int evidx, u64 len) { static const char *prev_line; static const char *prev_color; @@ -103,7 +99,7 @@ static void objdump_line__print(struct objdump_line *oline, const char *color; struct annotation *notes = symbol__annotation(sym); struct source_line *src_line = notes->src_line; - struct sym_hist *h = notes->histogram; + struct sym_hist *h = annotation__histogram(notes, evidx); s64 offset = oline->offset; struct objdump_line *next = objdump__get_next_ip_line(head, oline); @@ -328,7 +324,7 @@ static void symbol__free_source_line(struct symbol *sym, int len) /* Get the filename:line for the colored entries */ static int symbol__get_source_line(struct symbol *sym, struct map *map, - struct rb_root *root, int len, + int evidx, struct rb_root *root, int len, const char *filename) { u64 start; @@ -336,7 +332,7 @@ static int symbol__get_source_line(struct symbol *sym, struct map *map, char cmd[PATH_MAX * 2]; struct source_line *src_line; struct annotation *notes = symbol__annotation(sym); - struct sym_hist *h = notes->histogram; + struct sym_hist *h = annotation__histogram(notes, evidx); if (!h->sum) return 0; @@ -409,10 +405,10 @@ static void print_summary(struct rb_root *root, const char *filename) } } -static void symbol__annotate_hits(struct symbol *sym) +static void symbol__annotate_hits(struct symbol *sym, int evidx) { struct annotation *notes = symbol__annotation(sym); - struct sym_hist *h = notes->histogram; + struct sym_hist *h = annotation__histogram(notes, evidx); u64 len = sym->end - sym->start, offset; for (offset = 0; offset < len; ++offset) @@ -422,8 +418,8 @@ static void symbol__annotate_hits(struct symbol *sym) printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); } -int symbol__tty_annotate(struct symbol *sym, struct map *map, bool print_lines, - bool full_paths) +int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, + bool print_lines, bool full_paths) { struct dso *dso = map->dso; const char *filename = dso->long_name, *d_filename; @@ -443,7 +439,8 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, bool print_lines, len = sym->end - sym->start; if (print_lines) { - symbol__get_source_line(sym, map, &source_line, len, filename); + symbol__get_source_line(sym, map, evidx, &source_line, + len, filename); print_summary(&source_line, filename); } @@ -452,10 +449,10 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, bool print_lines, printf("------------------------------------------------\n"); if (verbose) - symbol__annotate_hits(sym); + symbol__annotate_hits(sym, evidx); list_for_each_entry_safe(pos, n, &head, node) { - objdump_line__print(pos, &head, sym, len); + objdump_line__print(pos, &head, sym, evidx, len); list_del(&pos->node); objdump_line__free(pos); } diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index 6e2fbc205299..0a5069ca6dd7 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -28,9 +28,21 @@ struct source_line { char *path; }; +/** struct annotation - symbols with hits have this attached as in sannotation + * + * @histogram: Array of addr hit histograms per event being monitored + * @src_line: If 'print_lines' is specified, per source code line percentages + * + * src_line is allocated, percentages calculated and all sorted by percentage + * when the annotation is about to be presented, so the percentages are for + * one of the entries in the histogram array, i.e. for the event/counter being + * presented. It is deallocated right after symbol__{tui,tty,etc}_annotate + * returns. + */ struct annotation { - struct sym_hist *histogram; struct source_line *src_line; + struct sym_hist *histograms; + int sizeof_sym_hist; }; struct sannotation { @@ -38,28 +50,35 @@ struct sannotation { struct symbol symbol; }; +static inline struct sym_hist *annotation__histogram(struct annotation *notes, int idx) +{ + return ((void *)notes->histograms) + (notes->sizeof_sym_hist * idx); +} + static inline struct annotation *symbol__annotation(struct symbol *sym) { struct sannotation *a = container_of(sym, struct sannotation, symbol); return &a->annotation; } -int symbol__inc_addr_samples(struct symbol *sym, struct map *map, u64 addr); +int symbol__inc_addr_samples(struct symbol *sym, struct map *map, + int evidx, u64 addr); +int symbol__alloc_hist(struct symbol *sym, int nevents); int symbol__annotate(struct symbol *sym, struct map *map, struct list_head *head, size_t privsize); -int symbol__tty_annotate(struct symbol *sym, struct map *map, +int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, bool print_lines, bool full_paths); #ifdef NO_NEWT_SUPPORT static inline int symbol__tui_annotate(symbol *sym __used, - struct map *map __used) + struct map *map __used, int evidx __used) { return 0; } #else -int symbol__tui_annotate(struct symbol *sym, struct map *map); +int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx); #endif #endif /* __PERF_ANNOTATE_H */ diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 6d9c92c3d7cb..bac5ab684967 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -950,9 +950,9 @@ void hists__filter_by_thread(struct hists *self, const struct thread *thread) } } -int hist_entry__inc_addr_samples(struct hist_entry *he, u64 ip) +int hist_entry__inc_addr_samples(struct hist_entry *he, int evidx, u64 ip) { - return symbol__inc_addr_samples(he->ms.sym, he->ms.map, ip); + return symbol__inc_addr_samples(he->ms.sym, he->ms.map, evidx, ip); } int hist_entry__annotate(struct hist_entry *he, struct list_head *head, diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 8a201f755534..2c6cdae6a764 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -77,7 +77,7 @@ size_t hists__fprintf_nr_events(struct hists *self, FILE *fp); size_t hists__fprintf(struct hists *self, struct hists *pair, bool show_displacement, FILE *fp); -int hist_entry__inc_addr_samples(struct hist_entry *self, u64 ip); +int hist_entry__inc_addr_samples(struct hist_entry *self, int evidx, u64 addr); int hist_entry__annotate(struct hist_entry *self, struct list_head *head, size_t privsize); @@ -91,18 +91,20 @@ bool hists__new_col_len(struct hists *self, enum hist_column col, u16 len); #ifdef NO_NEWT_SUPPORT static inline int hists__browse(struct hists *self __used, const char *helpline __used, - const char *ev_name __used) + const char *ev_name __used, int evidx __used) { return 0; } static inline int hists__tui_browse_tree(struct rb_root *self __used, - const char *help __used) + const char *help __used, + int evidx __used) { return 0; } -static inline int hist_entry__tui_annotate(struct hist_entry *self __used) +static inline int hist_entry__tui_annotate(struct hist_entry *self __used, + int evidx __used) { return 0; } @@ -111,13 +113,13 @@ static inline int hist_entry__tui_annotate(struct hist_entry *self __used) #else #include int hists__browse(struct hists *self, const char *helpline, - const char *ev_name); -int hist_entry__tui_annotate(struct hist_entry *self); + const char *ev_name, int evidx); +int hist_entry__tui_annotate(struct hist_entry *self, int evidx); #define KEY_LEFT NEWT_KEY_LEFT #define KEY_RIGHT NEWT_KEY_RIGHT -int hists__tui_browse_tree(struct rb_root *self, const char *help); +int hists__tui_browse_tree(struct rb_root *self, const char *help, int evidx); #endif unsigned int hists__sort_list_width(struct hists *self); diff --git a/tools/perf/util/ui/browsers/annotate.c b/tools/perf/util/ui/browsers/annotate.c index daa7138d8015..8d8a16895af7 100644 --- a/tools/perf/util/ui/browsers/annotate.c +++ b/tools/perf/util/ui/browsers/annotate.c @@ -61,7 +61,7 @@ static void annotate_browser__write(struct ui_browser *self, void *entry, int ro static double objdump_line__calc_percent(struct objdump_line *self, struct list_head *head, - struct symbol *sym) + struct symbol *sym, int evidx) { double percent = 0.0; @@ -70,7 +70,7 @@ static double objdump_line__calc_percent(struct objdump_line *self, unsigned int hits = 0; struct annotation *notes = symbol__annotation(sym); struct source_line *src_line = notes->src_line; - struct sym_hist *h = notes->histogram; + struct sym_hist *h = annotation__histogram(notes, evidx); s64 offset = self->offset; struct objdump_line *next = objdump__get_next_ip_line(head, self); @@ -183,12 +183,12 @@ out: return key; } -int hist_entry__tui_annotate(struct hist_entry *he) +int hist_entry__tui_annotate(struct hist_entry *he, int evidx) { - return symbol__tui_annotate(he->ms.sym, he->ms.map); + return symbol__tui_annotate(he->ms.sym, he->ms.map, evidx); } -int symbol__tui_annotate(struct symbol *sym, struct map *map) +int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx) { struct objdump_line *pos, *n; struct objdump_line_rb_node *rbpos; @@ -223,7 +223,7 @@ int symbol__tui_annotate(struct symbol *sym, struct map *map) browser.b.width = line_len; rbpos = objdump_line__rb(pos); rbpos->idx = browser.b.nr_entries++; - rbpos->percent = objdump_line__calc_percent(pos, &head, sym); + rbpos->percent = objdump_line__calc_percent(pos, &head, sym, evidx); if (rbpos->percent < 0.01) continue; objdump__insert_line(&browser.entries, rbpos); diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c index 86428239fa65..294b49538522 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -797,7 +797,8 @@ static int hists__browser_title(struct hists *self, char *bf, size_t size, return printed; } -int hists__browse(struct hists *self, const char *helpline, const char *ev_name) +int hists__browse(struct hists *self, const char *helpline, + const char *ev_name, int evidx) { struct hist_browser *browser = hist_browser__new(self); struct pstack *fstack; @@ -935,7 +936,7 @@ do_annotate: if (he == NULL) continue; - hist_entry__tui_annotate(he); + hist_entry__tui_annotate(he, evidx); } else if (choice == browse_map) map__browse(browser->selection->map); else if (choice == zoom_dso) { @@ -984,7 +985,7 @@ out: return key; } -int hists__tui_browse_tree(struct rb_root *self, const char *help) +int hists__tui_browse_tree(struct rb_root *self, const char *help, int evidx) { struct rb_node *first = rb_first(self), *nd = first, *next; int key = 0; @@ -993,7 +994,7 @@ int hists__tui_browse_tree(struct rb_root *self, const char *help) struct hists *hists = rb_entry(nd, struct hists, rb_node); const char *ev_name = __event_name(hists->type, hists->config); - key = hists__browse(hists, help, ev_name); + key = hists__browse(hists, help, ev_name, evidx); switch (key) { case NEWT_KEY_TAB: next = rb_next(nd); -- cgit v1.2.3 From d040bd363824f9f0ad6610b91ee6c65f292c066c Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 5 Feb 2011 15:37:31 -0200 Subject: perf annotate: Config options for symbol__tty_annotate Max line# that should be printed, minimum percentage filter, just like 'perf top', alas, due to it :-) Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-annotate.c | 2 +- tools/perf/util/annotate.c | 14 ++++++++++---- tools/perf/util/annotate.h | 3 ++- 3 files changed, 13 insertions(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index f3e44231b10d..ea6a1165956f 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -95,7 +95,7 @@ static int process_sample_event(union perf_event *event, static int hist_entry__tty_annotate(struct hist_entry *he, int evidx) { return symbol__tty_annotate(he->ms.sym, he->ms.map, evidx, - print_line, full_paths); + print_line, full_paths, 0, 0); } static void hists__find_annotations(struct hists *self) diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 7488fe99502c..072bc8d91aa1 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -87,7 +87,7 @@ struct objdump_line *objdump__get_next_ip_line(struct list_head *head, static void objdump_line__print(struct objdump_line *oline, struct list_head *head, struct symbol *sym, - int evidx, u64 len) + int evidx, u64 len, int min_pcnt) { static const char *prev_line; static const char *prev_color; @@ -118,6 +118,9 @@ static void objdump_line__print(struct objdump_line *oline, if (src_line == NULL && h->sum) percent = 100.0 * hits / h->sum; + if (percent < min_pcnt) + return; + color = get_percent_color(percent); /* @@ -419,13 +422,15 @@ static void symbol__annotate_hits(struct symbol *sym, int evidx) } int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, - bool print_lines, bool full_paths) + bool print_lines, bool full_paths, int min_pcnt, + int max_lines) { struct dso *dso = map->dso; const char *filename = dso->long_name, *d_filename; struct rb_root source_line = RB_ROOT; struct objdump_line *pos, *n; LIST_HEAD(head); + int printed = 2; u64 len; if (symbol__annotate(sym, map, &head, 0) < 0) @@ -444,7 +449,6 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, print_summary(&source_line, filename); } - printf("\n\n------------------------------------------------\n"); printf(" Percent | Source code & Disassembly of %s\n", d_filename); printf("------------------------------------------------\n"); @@ -452,9 +456,11 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, symbol__annotate_hits(sym, evidx); list_for_each_entry_safe(pos, n, &head, node) { - objdump_line__print(pos, &head, sym, evidx, len); + objdump_line__print(pos, &head, sym, evidx, len, min_pcnt); list_del(&pos->node); objdump_line__free(pos); + if (max_lines && ++printed >= max_lines) + break; } if (print_lines) diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index 0a5069ca6dd7..6b707324e66f 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -69,7 +69,8 @@ int symbol__annotate(struct symbol *sym, struct map *map, struct list_head *head, size_t privsize); int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, - bool print_lines, bool full_paths); + bool print_lines, bool full_paths, int min_pcnt, + int max_lines); #ifdef NO_NEWT_SUPPORT static inline int symbol__tui_annotate(symbol *sym __used, -- cgit v1.2.3 From f1e2701de02cff6d988b1dd49960620d5720cb89 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 5 Feb 2011 18:51:38 -0200 Subject: perf annotate: Separate objdump parsing from actual screen rendering Because in 'perf top' we'll need to parse just once and then, as samples come, render multiple times with evolving counter values. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/annotate.c | 62 ++++++++++++++++++++++++++++++++-------------- tools/perf/util/annotate.h | 4 +++ 2 files changed, 47 insertions(+), 19 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 072bc8d91aa1..10cdbad76058 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -421,21 +421,16 @@ static void symbol__annotate_hits(struct symbol *sym, int evidx) printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); } -int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, - bool print_lines, bool full_paths, int min_pcnt, - int max_lines) +void symbol__annotate_printf(struct symbol *sym, struct map *map, + struct list_head *head, int evidx, bool full_paths, + int min_pcnt, int max_lines) { struct dso *dso = map->dso; const char *filename = dso->long_name, *d_filename; - struct rb_root source_line = RB_ROOT; - struct objdump_line *pos, *n; - LIST_HEAD(head); + struct objdump_line *pos; int printed = 2; u64 len; - if (symbol__annotate(sym, map, &head, 0) < 0) - return -1; - if (full_paths) d_filename = filename; else @@ -443,28 +438,57 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, len = sym->end - sym->start; - if (print_lines) { - symbol__get_source_line(sym, map, evidx, &source_line, - len, filename); - print_summary(&source_line, filename); - } - printf(" Percent | Source code & Disassembly of %s\n", d_filename); printf("------------------------------------------------\n"); if (verbose) symbol__annotate_hits(sym, evidx); - list_for_each_entry_safe(pos, n, &head, node) { - objdump_line__print(pos, &head, sym, evidx, len, min_pcnt); - list_del(&pos->node); - objdump_line__free(pos); + list_for_each_entry(pos, head, node) { + objdump_line__print(pos, head, sym, evidx, len, min_pcnt); if (max_lines && ++printed >= max_lines) break; + + } +} + +void objdump_line_list__purge(struct list_head *head) +{ + struct objdump_line *pos, *n; + + list_for_each_entry_safe(pos, n, head, node) { + list_del(&pos->node); + objdump_line__free(pos); + } +} + +int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, + bool print_lines, bool full_paths, int min_pcnt, + int max_lines) +{ + struct dso *dso = map->dso; + const char *filename = dso->long_name; + struct rb_root source_line = RB_ROOT; + LIST_HEAD(head); + u64 len; + + if (symbol__annotate(sym, map, &head, 0) < 0) + return -1; + + len = sym->end - sym->start; + + if (print_lines) { + symbol__get_source_line(sym, map, evidx, &source_line, + len, filename); + print_summary(&source_line, filename); } + symbol__annotate_printf(sym, map, &head, evidx, full_paths, + min_pcnt, max_lines); if (print_lines) symbol__free_source_line(sym, len); + objdump_line_list__purge(&head); + return 0; } diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index 6b707324e66f..53dd92de2615 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -67,6 +67,10 @@ int symbol__alloc_hist(struct symbol *sym, int nevents); int symbol__annotate(struct symbol *sym, struct map *map, struct list_head *head, size_t privsize); +void symbol__annotate_printf(struct symbol *sym, struct map *map, + struct list_head *head, int evidx, bool full_paths, + int min_pcnt, int max_lines); +void objdump_line_list__purge(struct list_head *head); int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, bool print_lines, bool full_paths, int min_pcnt, -- cgit v1.2.3 From 36532461a0f60bb36c5470a0326f7394f19db23c Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sun, 6 Feb 2011 14:54:44 -0200 Subject: perf top: Ditch private annotation code, share perf annotate's Next step: Live TUI annotation in perf top, just press enter on a symbol line. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-top.c | 181 ++++++++------------------------------------- tools/perf/util/annotate.c | 76 ++++++++++++++++--- tools/perf/util/annotate.h | 11 ++- tools/perf/util/top.h | 11 +-- 4 files changed, 106 insertions(+), 173 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 154e088588bc..716118a3b3e4 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -20,6 +20,7 @@ #include "perf.h" +#include "util/annotate.h" #include "util/cache.h" #include "util/color.h" #include "util/evlist.h" @@ -140,10 +141,7 @@ static int parse_source(struct sym_entry *syme) struct symbol *sym; struct sym_entry_source *source; struct map *map; - FILE *file; - char command[PATH_MAX*2]; - const char *path; - u64 len; + int err = -1; if (!syme) return -1; @@ -162,197 +160,80 @@ static int parse_source(struct sym_entry *syme) if (syme->src == NULL) return -1; pthread_mutex_init(&syme->src->lock, NULL); + INIT_LIST_HEAD(&syme->src->head); } source = syme->src; - if (source->lines) { + if (symbol__annotation(sym)->histograms != NULL) { pthread_mutex_lock(&source->lock); goto out_assign; } - path = map->dso->long_name; - - len = sym->end - sym->start; - - sprintf(command, - "objdump --start-address=%#0*" PRIx64 " --stop-address=%#0*" PRIx64 " -dS %s", - BITS_PER_LONG / 4, map__rip_2objdump(map, sym->start), - BITS_PER_LONG / 4, map__rip_2objdump(map, sym->end), path); - - file = popen(command, "r"); - if (!file) - return -1; pthread_mutex_lock(&source->lock); - source->lines_tail = &source->lines; - while (!feof(file)) { - struct source_line *src; - size_t dummy = 0; - char *c, *sep; - - src = malloc(sizeof(struct source_line)); - assert(src != NULL); - memset(src, 0, sizeof(struct source_line)); - - if (getline(&src->line, &dummy, file) < 0) - break; - if (!src->line) - break; - - c = strchr(src->line, '\n'); - if (c) - *c = 0; - src->next = NULL; - *source->lines_tail = src; - source->lines_tail = &src->next; - - src->eip = strtoull(src->line, &sep, 16); - if (*sep == ':') - src->eip = map__objdump_2ip(map, src->eip); - else /* this line has no ip info (e.g. source line) */ - src->eip = 0; + if (symbol__alloc_hist(sym, top.evlist->nr_entries) < 0) { + pr_err("Not enough memory for annotating '%s' symbol!\n", + sym->name); + goto out_unlock; } - pclose(file); + + err = symbol__annotate(sym, syme->map, &source->head, 0); + if (err == 0) { out_assign: sym_filter_entry = syme; + } +out_unlock: pthread_mutex_unlock(&source->lock); - return 0; + return err; } static void __zero_source_counters(struct sym_entry *syme) { - int i; - struct source_line *line; - - line = syme->src->lines; - while (line) { - for (i = 0; i < top.evlist->nr_entries; i++) - line->count[i] = 0; - line = line->next; - } + struct symbol *sym = sym_entry__symbol(syme); + symbol__annotate_zero_histograms(sym); } static void record_precise_ip(struct sym_entry *syme, int counter, u64 ip) { - struct source_line *line; - if (syme != sym_filter_entry) return; if (pthread_mutex_trylock(&syme->src->lock)) return; - if (syme->src == NULL || syme->src->source == NULL) - goto out_unlock; - - for (line = syme->src->lines; line; line = line->next) { - /* skip lines without IP info */ - if (line->eip == 0) - continue; - if (line->eip == ip) { - line->count[counter]++; - break; - } - if (line->eip > ip) - break; - } -out_unlock: - pthread_mutex_unlock(&syme->src->lock); -} - -#define PATTERN_LEN (BITS_PER_LONG / 4 + 2) + ip = syme->map->map_ip(syme->map, ip); + symbol__inc_addr_samples(sym_entry__symbol(syme), syme->map, counter, ip); -static void lookup_sym_source(struct sym_entry *syme) -{ - struct symbol *symbol = sym_entry__symbol(syme); - struct source_line *line; - char pattern[PATTERN_LEN + 1]; - - sprintf(pattern, "%0*" PRIx64 " <", BITS_PER_LONG / 4, - map__rip_2objdump(syme->map, symbol->start)); - - pthread_mutex_lock(&syme->src->lock); - for (line = syme->src->lines; line; line = line->next) { - if (memcmp(line->line, pattern, PATTERN_LEN) == 0) { - syme->src->source = line; - break; - } - } pthread_mutex_unlock(&syme->src->lock); } -static void show_lines(struct source_line *queue, int count, int total) -{ - int i; - struct source_line *line; - - line = queue; - for (i = 0; i < count; i++) { - float pcnt = 100.0*(float)line->count[top.sym_counter]/(float)total; - - printf("%8li %4.1f%%\t%s\n", line->count[top.sym_counter], pcnt, line->line); - line = line->next; - } -} - -#define TRACE_COUNT 3 - static void show_details(struct sym_entry *syme) { struct symbol *symbol; - struct source_line *line; - struct source_line *line_queue = NULL; - int displayed = 0; - int line_queue_count = 0, total = 0, more = 0; + int more; if (!syme) return; - if (!syme->src->source) - lookup_sym_source(syme); - - if (!syme->src->source) + symbol = sym_entry__symbol(syme); + if (!syme->src || symbol__annotation(symbol)->histograms == NULL) return; - symbol = sym_entry__symbol(syme); printf("Showing %s for %s\n", event_name(top.sym_evsel), symbol->name); printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter); pthread_mutex_lock(&syme->src->lock); - line = syme->src->source; - while (line) { - total += line->count[top.sym_counter]; - line = line->next; - } - - line = syme->src->source; - while (line) { - float pcnt = 0.0; - - if (!line_queue_count) - line_queue = line; - line_queue_count++; - - if (line->count[top.sym_counter]) - pcnt = 100.0 * line->count[top.sym_counter] / (float)total; - if (pcnt >= (float)sym_pcnt_filter) { - if (displayed <= top.print_entries) - show_lines(line_queue, line_queue_count, total); - else more++; - displayed += line_queue_count; - line_queue_count = 0; - line_queue = NULL; - } else if (line_queue_count > TRACE_COUNT) { - line_queue = line_queue->next; - line_queue_count--; - } - - line->count[top.sym_counter] = top.zero ? 0 : line->count[top.sym_counter] * 7 / 8; - line = line->next; - } + more = symbol__annotate_printf(symbol, syme->map, &syme->src->head, + top.sym_evsel->idx, 0, sym_pcnt_filter, + top.print_entries); + if (top.zero) + symbol__annotate_zero_histogram(symbol, top.sym_evsel->idx); + else + symbol__annotate_decay_histogram(symbol, &syme->src->head, + top.sym_evsel->idx); pthread_mutex_unlock(&syme->src->lock); - if (more) + if (more != 0) printf("%d lines not displayed, maybe increase display entries [e]\n", more); } @@ -1172,7 +1053,7 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node); - symbol_conf.priv_size = (sizeof(struct sym_entry) + + symbol_conf.priv_size = (sizeof(struct sym_entry) + sizeof(struct annotation) + (top.evlist->nr_entries + 1) * sizeof(unsigned long)); symbol_conf.try_vmlinux_path = (symbol_conf.vmlinux_name == NULL); diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 10cdbad76058..297337649c21 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -22,9 +22,19 @@ int symbol__alloc_hist(struct symbol *sym, int nevents) notes->sizeof_sym_hist = (sizeof(*notes->histograms) + (sym->end - sym->start) * sizeof(u64)); notes->histograms = calloc(nevents, notes->sizeof_sym_hist); + notes->nr_histograms = nevents; return notes->histograms == NULL ? -1 : 0; } +void symbol__annotate_zero_histograms(struct symbol *sym) +{ + struct annotation *notes = symbol__annotation(sym); + + if (notes->histograms != NULL) + memset(notes->histograms, 0, + notes->nr_histograms * notes->sizeof_sym_hist); +} + int symbol__inc_addr_samples(struct symbol *sym, struct map *map, int evidx, u64 addr) { @@ -85,9 +95,10 @@ struct objdump_line *objdump__get_next_ip_line(struct list_head *head, return NULL; } -static void objdump_line__print(struct objdump_line *oline, - struct list_head *head, struct symbol *sym, - int evidx, u64 len, int min_pcnt) +static int objdump_line__print(struct objdump_line *oline, + struct list_head *head, struct symbol *sym, + int evidx, u64 len, int min_pcnt, + int printed, int max_lines) { static const char *prev_line; static const char *prev_color; @@ -119,7 +130,10 @@ static void objdump_line__print(struct objdump_line *oline, percent = 100.0 * hits / h->sum; if (percent < min_pcnt) - return; + return -1; + + if (printed >= max_lines) + return 1; color = get_percent_color(percent); @@ -140,12 +154,16 @@ static void objdump_line__print(struct objdump_line *oline, color_fprintf(stdout, color, " %7.2f", percent); printf(" : "); color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", oline->line); - } else { + } else if (printed >= max_lines) + return 1; + else { if (!*oline->line) printf(" :\n"); else printf(" : %s\n", oline->line); } + + return 0; } static int symbol__parse_objdump_line(struct symbol *sym, struct map *map, FILE *file, @@ -421,14 +439,15 @@ static void symbol__annotate_hits(struct symbol *sym, int evidx) printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); } -void symbol__annotate_printf(struct symbol *sym, struct map *map, - struct list_head *head, int evidx, bool full_paths, - int min_pcnt, int max_lines) +int symbol__annotate_printf(struct symbol *sym, struct map *map, + struct list_head *head, int evidx, bool full_paths, + int min_pcnt, int max_lines) { struct dso *dso = map->dso; const char *filename = dso->long_name, *d_filename; struct objdump_line *pos; int printed = 2; + int more = 0; u64 len; if (full_paths) @@ -445,10 +464,47 @@ void symbol__annotate_printf(struct symbol *sym, struct map *map, symbol__annotate_hits(sym, evidx); list_for_each_entry(pos, head, node) { - objdump_line__print(pos, head, sym, evidx, len, min_pcnt); - if (max_lines && ++printed >= max_lines) + switch (objdump_line__print(pos, head, sym, evidx, len, min_pcnt, + printed, max_lines)) { + case 0: + ++printed; + break; + case 1: + /* filtered by max_lines */ + ++more; break; + case -1: + default: + /* filtered by min_pcnt */ + break; + } + } + + return more; +} +void symbol__annotate_zero_histogram(struct symbol *sym, int evidx) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + + memset(h, 0, notes->sizeof_sym_hist); +} + +void symbol__annotate_decay_histogram(struct symbol *sym, + struct list_head *head, int evidx) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + struct objdump_line *pos; + + h->sum = 0; + + list_for_each_entry(pos, head, node) { + if (pos->offset != -1) { + h->addr[pos->offset] = h->addr[pos->offset] * 7 / 8; + h->sum += h->addr[pos->offset]; + } } } diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index 53dd92de2615..b1253aadf340 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -42,6 +42,7 @@ struct source_line { struct annotation { struct source_line *src_line; struct sym_hist *histograms; + int nr_histograms; int sizeof_sym_hist; }; @@ -64,12 +65,16 @@ static inline struct annotation *symbol__annotation(struct symbol *sym) int symbol__inc_addr_samples(struct symbol *sym, struct map *map, int evidx, u64 addr); int symbol__alloc_hist(struct symbol *sym, int nevents); +void symbol__annotate_zero_histograms(struct symbol *sym); int symbol__annotate(struct symbol *sym, struct map *map, struct list_head *head, size_t privsize); -void symbol__annotate_printf(struct symbol *sym, struct map *map, - struct list_head *head, int evidx, bool full_paths, - int min_pcnt, int max_lines); +int symbol__annotate_printf(struct symbol *sym, struct map *map, + struct list_head *head, int evidx, bool full_paths, + int min_pcnt, int max_lines); +void symbol__annotate_zero_histogram(struct symbol *sym, int evidx); +void symbol__annotate_decay_histogram(struct symbol *sym, + struct list_head *head, int evidx); void objdump_line_list__purge(struct list_head *head); int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index 500950818d2f..fe44afb69985 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -11,17 +11,8 @@ struct perf_evlist; struct perf_evsel; -struct source_line { - u64 eip; - unsigned long count[MAX_COUNTERS]; /* FIXME */ - char *line; - struct source_line *next; -}; - struct sym_entry_source { - struct source_line *source; - struct source_line *lines; - struct source_line **lines_tail; + struct list_head head; pthread_mutex_t lock; }; -- cgit v1.2.3 From ef4d001d79ac4bab6c2d81e9986a42059f877ec3 Mon Sep 17 00:00:00 2001 From: Denis Kirjanov Date: Sat, 5 Feb 2011 20:39:38 +0000 Subject: perf top: Use pid_t for target_{pid|tid} Use pid_t data type for target_{pid|tid} vars. Cc: Ingo Molnar LKML-Reference: <20110205203938.GA15328@hera.kernel.org> Signed-off-by: Denis Kirjanov [ committer note: those variables are now in struct perf_top, fixed ] Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/top.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index fe44afb69985..62e32939f3d1 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -46,8 +46,8 @@ struct perf_top { u64 exact_samples; u64 guest_us_samples, guest_kernel_samples; int print_entries, count_filter, delay_secs; - int display_weighted, freq, rb_entries; - int sym_counter, target_pid, target_tid; + int display_weighted, freq, rb_entries, sym_counter; + pid_t target_pid, target_tid; bool hide_kernel_symbols, hide_user_symbols, zero; const char *cpu_list; struct perf_evsel *sym_evsel; -- cgit v1.2.3 From f50c2169bd054984e976e67e8651d28f3caf6ba3 Mon Sep 17 00:00:00 2001 From: Franck Bui-Huu Date: Thu, 13 Jan 2011 11:18:30 +0100 Subject: perf probe: Rewrite find_lazy_match_lines() by using getline(3) Acked-by: Masami Hiramatsu Cc: Masami Hiramatsu Cc: lkml LKML-Reference: Signed-off-by: Franck Bui-Huu Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-finder.c | 72 ++++++++++++++++-------------------------- 1 file changed, 28 insertions(+), 44 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index 69215bff17e9..46addfb3183e 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -1234,51 +1234,38 @@ static int find_probe_point_by_line(struct probe_finder *pf) static int find_lazy_match_lines(struct list_head *head, const char *fname, const char *pat) { - char *fbuf, *p1, *p2; - int fd, line, nlines = -1; - struct stat st; - - fd = open(fname, O_RDONLY); - if (fd < 0) { - pr_warning("Failed to open %s: %s\n", fname, strerror(-fd)); + FILE *fp; + char *line = NULL; + size_t line_len; + ssize_t len; + int count = 0, linenum = 1; + + fp = fopen(fname, "r"); + if (!fp) { + pr_warning("Failed to open %s: %s\n", fname, strerror(errno)); return -errno; } - if (fstat(fd, &st) < 0) { - pr_warning("Failed to get the size of %s: %s\n", - fname, strerror(errno)); - nlines = -errno; - goto out_close; - } - - nlines = -ENOMEM; - fbuf = malloc(st.st_size + 2); - if (fbuf == NULL) - goto out_close; - if (read(fd, fbuf, st.st_size) < 0) { - pr_warning("Failed to read %s: %s\n", fname, strerror(errno)); - nlines = -errno; - goto out_free_fbuf; - } - fbuf[st.st_size] = '\n'; /* Dummy line */ - fbuf[st.st_size + 1] = '\0'; - p1 = fbuf; - line = 1; - nlines = 0; - while ((p2 = strchr(p1, '\n')) != NULL) { - *p2 = '\0'; - if (strlazymatch(p1, pat)) { - line_list__add_line(head, line); - nlines++; + while ((len = getline(&line, &line_len, fp)) > 0) { + + if (line[len - 1] == '\n') + line[len - 1] = '\0'; + + if (strlazymatch(line, pat)) { + line_list__add_line(head, linenum); + count++; } - line++; - p1 = p2 + 1; + linenum++; } -out_free_fbuf: - free(fbuf); -out_close: - close(fd); - return nlines; + + if (ferror(fp)) + count = -errno; + free(line); + fclose(fp); + + if (count == 0) + pr_debug("No matched lines found in %s.\n", fname); + return count; } static int probe_point_lazy_walker(const char *fname, int lineno, @@ -1312,10 +1299,7 @@ static int find_probe_point_lazy(Dwarf_Die *sp_die, struct probe_finder *pf) /* Matching lazy line pattern */ ret = find_lazy_match_lines(&pf->lcache, pf->fname, pf->pev->point.lazy_line); - if (ret == 0) { - pr_debug("No matched lines found in %s.\n", pf->fname); - return 0; - } else if (ret < 0) + if (ret <= 0) return ret; } -- cgit v1.2.3 From fb7d0b3cefb80a105f7fd26bbc62e0cbf9192822 Mon Sep 17 00:00:00 2001 From: Kyle McMartin Date: Mon, 24 Jan 2011 11:13:04 -0500 Subject: perf tool: Fix gcc 4.6.0 issues GCC 4.6.0 in Fedora rawhide turned up some compile errors in tools/perf due to the -Werror=unused-but-set-variable flag. I've gone through and annotated some of the assignments that had side effects (ie: return value from a function) with the __used annotation, and in some cases, just removed unused code. In a few cases, we were assigning something useful, but not using it in later parts of the function. kyle@dreadnought:~/src% gcc --version gcc (GCC) 4.6.0 20110122 (Red Hat 4.6.0-0.3) Cc: Ingo Molnar LKML-Reference: <20110124161304.GK27353@bombadil.infradead.org> Signed-off-by: Kyle McMartin [ committer note: Fixed up the annotation fixes, as that code moved recently ] Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/bench/sched-pipe.c | 2 +- tools/perf/builtin-sched.c | 12 +++--------- tools/perf/builtin-top.c | 2 +- tools/perf/util/annotate.c | 3 --- tools/perf/util/header.c | 2 +- tools/perf/util/scripting-engines/trace-event-python.c | 3 +-- tools/perf/util/symbol.c | 4 ++-- tools/perf/util/trace-event-parse.c | 2 +- tools/perf/util/ui/browsers/map.c | 2 +- 9 files changed, 11 insertions(+), 21 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/bench/sched-pipe.c b/tools/perf/bench/sched-pipe.c index d9ab3ce446ac..0c7454f8b8a9 100644 --- a/tools/perf/bench/sched-pipe.c +++ b/tools/perf/bench/sched-pipe.c @@ -55,7 +55,7 @@ int bench_sched_pipe(int argc, const char **argv, * discarding returned value of read(), write() * causes error in building environment for perf */ - int ret, wait_stat; + int __used ret, wait_stat; pid_t pid, retpid; argc = parse_options(argc, argv, options, diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c index ae2621182927..a32f411faeac 100644 --- a/tools/perf/builtin-sched.c +++ b/tools/perf/builtin-sched.c @@ -369,11 +369,6 @@ static void process_sched_event(struct task_desc *this_task __used, struct sched_atom *atom) { int ret = 0; - u64 now; - long long delta; - - now = get_nsecs(); - delta = start_time + atom->timestamp - now; switch (atom->type) { case SCHED_EVENT_RUN: @@ -562,7 +557,7 @@ static void wait_for_tasks(void) static void run_one_test(void) { - u64 T0, T1, delta, avg_delta, fluct, std_dev; + u64 T0, T1, delta, avg_delta, fluct; T0 = get_nsecs(); wait_for_tasks(); @@ -578,7 +573,6 @@ static void run_one_test(void) else fluct = delta - avg_delta; sum_fluct += fluct; - std_dev = sum_fluct / nr_runs / sqrt(nr_runs); if (!run_avg) run_avg = delta; run_avg = (run_avg*9 + delta)/10; @@ -799,7 +793,7 @@ replay_switch_event(struct trace_switch_event *switch_event, u64 timestamp, struct thread *thread __used) { - struct task_desc *prev, *next; + struct task_desc *prev, __used *next; u64 timestamp0; s64 delta; @@ -1404,7 +1398,7 @@ map_switch_event(struct trace_switch_event *switch_event, u64 timestamp, struct thread *thread __used) { - struct thread *sched_out, *sched_in; + struct thread *sched_out __used, *sched_in; int new_shortname; u64 timestamp0; s64 delta; diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 716118a3b3e4..b790673cb0aa 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -865,7 +865,7 @@ static int __cmd_top(void) { pthread_t thread; struct perf_evsel *first; - int ret; + int ret __used; /* * FIXME: perf_session__new should allow passing a O_MMAP, so that all this * mmap reading, etc is encapsulated in it. Use O_WRONLY for now. diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 297337649c21..1012841835a3 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -236,7 +236,6 @@ int symbol__annotate(struct symbol *sym, struct map *map, char command[PATH_MAX * 2]; FILE *file; int err = 0; - u64 len; char symfs_filename[PATH_MAX]; if (filename) { @@ -281,8 +280,6 @@ fallback: filename, sym->name, map->unmap_ip(map, sym->start), map->unmap_ip(map, sym->end)); - len = sym->end - sym->start; - pr_debug("annotating [%p] %30s : [%p] %30s\n", dso, dso->long_name, sym, sym->name); diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index c0de5ec44145..72c124dc5781 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -1145,7 +1145,7 @@ int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, { union perf_event ev; ssize_t size = 0, aligned_size = 0, padding; - int err = 0; + int err __used = 0; memset(&ev, 0, sizeof(ev)); diff --git a/tools/perf/util/scripting-engines/trace-event-python.c b/tools/perf/util/scripting-engines/trace-event-python.c index c6d99334bdfa..2040b8538527 100644 --- a/tools/perf/util/scripting-engines/trace-event-python.c +++ b/tools/perf/util/scripting-engines/trace-event-python.c @@ -248,8 +248,7 @@ static void python_process_event(int cpu, void *data, context = PyCObject_FromVoidPtr(scripting_context, NULL); PyTuple_SetItem(t, n++, PyString_FromString(handler_name)); - PyTuple_SetItem(t, n++, - PyCObject_FromVoidPtr(scripting_context, NULL)); + PyTuple_SetItem(t, n++, context); if (handler) { PyTuple_SetItem(t, n++, PyInt_FromLong(cpu)); diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 7821d0e6866f..3e193f8e3061 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -1525,8 +1525,8 @@ int dso__load(struct dso *self, struct map *map, symbol_filter_t filter) symbol_conf.symfs, self->long_name); break; case DSO__ORIG_GUEST_KMODULE: - if (map->groups && map->groups->machine) - root_dir = map->groups->machine->root_dir; + if (map->groups && machine) + root_dir = machine->root_dir; else root_dir = ""; snprintf(name, size, "%s%s%s", symbol_conf.symfs, diff --git a/tools/perf/util/trace-event-parse.c b/tools/perf/util/trace-event-parse.c index 73a02223c629..d8e622dd738a 100644 --- a/tools/perf/util/trace-event-parse.c +++ b/tools/perf/util/trace-event-parse.c @@ -153,7 +153,7 @@ void parse_proc_kallsyms(char *file, unsigned int size __unused) char *next = NULL; char *addr_str; char ch; - int ret; + int ret __used; int i; line = strtok_r(file, "\n", &next); diff --git a/tools/perf/util/ui/browsers/map.c b/tools/perf/util/ui/browsers/map.c index e5158369106e..8462bffe20bc 100644 --- a/tools/perf/util/ui/browsers/map.c +++ b/tools/perf/util/ui/browsers/map.c @@ -41,7 +41,7 @@ static int ui_entry__read(const char *title, char *bf, size_t size, int width) out_free_form: newtPopWindow(); newtFormDestroy(form); - return 0; + return err; } struct map_browser { -- cgit v1.2.3 From a2221796256ea7b236cec6bf027c1c1de5b8ccd7 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Mon, 7 Feb 2011 15:32:18 +0100 Subject: perf annotate: Fix build error A small fix for when NO_NEWT_SUPPORT is defined. Add a missing "struct" to the function prototype. Cc: Frederic Weisbecker Cc: H. Peter Anvin Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Thomas Gleixner Cc: Tom Zanussi LKML-Reference: <20110207143218.GA31197@kryptos.osrc.amd.com> Signed-off-by: Borislav Petkov Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/annotate.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index b1253aadf340..bc08b36a713a 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -82,7 +82,7 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, int max_lines); #ifdef NO_NEWT_SUPPORT -static inline int symbol__tui_annotate(symbol *sym __used, +static inline int symbol__tui_annotate(struct symbol *sym __used, struct map *map __used, int evidx __used) { return 0; -- cgit v1.2.3 From 124bb83cd7de4d851af7595650233fb9e9279d5d Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Fri, 4 Feb 2011 21:52:11 +0900 Subject: perf probe: Add bitfield member support Add bitfield member accessing support to probe arguments. Suggested-by: Arnaldo Carvalho de Melo Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Steven Rostedt LKML-Reference: <20110204125211.9507.60265.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu [ committer note: Fixed up '%lu' use for return of BYTES_TO_BITS ('%zd') ] Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-finder.c | 95 ++++++++++++++++++++++++++++++------------ 1 file changed, 68 insertions(+), 27 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index 46addfb3183e..fe461f6559f1 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -33,6 +33,7 @@ #include #include +#include #include "event.h" #include "debug.h" #include "util.h" @@ -333,13 +334,23 @@ static Dwarf_Die *die_get_real_type(Dwarf_Die *vr_die, Dwarf_Die *die_mem) return vr_die; } -static bool die_is_signed_type(Dwarf_Die *tp_die) +static int die_get_attr_udata(Dwarf_Die *tp_die, unsigned int attr_name, + Dwarf_Word *result) { Dwarf_Attribute attr; + + if (dwarf_attr(tp_die, attr_name, &attr) == NULL || + dwarf_formudata(&attr, result) != 0) + return -ENOENT; + + return 0; +} + +static bool die_is_signed_type(Dwarf_Die *tp_die) +{ Dwarf_Word ret; - if (dwarf_attr(tp_die, DW_AT_encoding, &attr) == NULL || - dwarf_formudata(&attr, &ret) != 0) + if (die_get_attr_udata(tp_die, DW_AT_encoding, &ret)) return false; return (ret == DW_ATE_signed_char || ret == DW_ATE_signed || @@ -348,11 +359,29 @@ static bool die_is_signed_type(Dwarf_Die *tp_die) static int die_get_byte_size(Dwarf_Die *tp_die) { - Dwarf_Attribute attr; Dwarf_Word ret; - if (dwarf_attr(tp_die, DW_AT_byte_size, &attr) == NULL || - dwarf_formudata(&attr, &ret) != 0) + if (die_get_attr_udata(tp_die, DW_AT_byte_size, &ret)) + return 0; + + return (int)ret; +} + +static int die_get_bit_size(Dwarf_Die *tp_die) +{ + Dwarf_Word ret; + + if (die_get_attr_udata(tp_die, DW_AT_bit_size, &ret)) + return 0; + + return (int)ret; +} + +static int die_get_bit_offset(Dwarf_Die *tp_die) +{ + Dwarf_Word ret; + + if (die_get_attr_udata(tp_die, DW_AT_bit_offset, &ret)) return 0; return (int)ret; @@ -827,6 +856,8 @@ static_var: return 0; } +#define BYTES_TO_BITS(nb) ((nb) * BITS_PER_LONG / sizeof(long)) + static int convert_variable_type(Dwarf_Die *vr_die, struct probe_trace_arg *tvar, const char *cast) @@ -843,6 +874,14 @@ static int convert_variable_type(Dwarf_Die *vr_die, return (tvar->type == NULL) ? -ENOMEM : 0; } + if (die_get_bit_size(vr_die) != 0) { + /* This is a bitfield */ + ret = snprintf(buf, 16, "b%d@%d/%zd", die_get_bit_size(vr_die), + die_get_bit_offset(vr_die), + BYTES_TO_BITS(die_get_byte_size(vr_die))); + goto formatted; + } + if (die_get_real_type(vr_die, &type) == NULL) { pr_warning("Failed to get a type information of %s.\n", dwarf_diename(vr_die)); @@ -887,29 +926,31 @@ static int convert_variable_type(Dwarf_Die *vr_die, return (tvar->type == NULL) ? -ENOMEM : 0; } - ret = die_get_byte_size(&type) * 8; - if (ret) { - /* Check the bitwidth */ - if (ret > MAX_BASIC_TYPE_BITS) { - pr_info("%s exceeds max-bitwidth." - " Cut down to %d bits.\n", - dwarf_diename(&type), MAX_BASIC_TYPE_BITS); - ret = MAX_BASIC_TYPE_BITS; - } + ret = BYTES_TO_BITS(die_get_byte_size(&type)); + if (!ret) + /* No size ... try to use default type */ + return 0; - ret = snprintf(buf, 16, "%c%d", - die_is_signed_type(&type) ? 's' : 'u', ret); - if (ret < 0 || ret >= 16) { - if (ret >= 16) - ret = -E2BIG; - pr_warning("Failed to convert variable type: %s\n", - strerror(-ret)); - return ret; - } - tvar->type = strdup(buf); - if (tvar->type == NULL) - return -ENOMEM; + /* Check the bitwidth */ + if (ret > MAX_BASIC_TYPE_BITS) { + pr_info("%s exceeds max-bitwidth. Cut down to %d bits.\n", + dwarf_diename(&type), MAX_BASIC_TYPE_BITS); + ret = MAX_BASIC_TYPE_BITS; + } + ret = snprintf(buf, 16, "%c%d", + die_is_signed_type(&type) ? 's' : 'u', ret); + +formatted: + if (ret < 0 || ret >= 16) { + if (ret >= 16) + ret = -E2BIG; + pr_warning("Failed to convert variable type: %s\n", + strerror(-ret)); + return ret; } + tvar->type = strdup(buf); + if (tvar->type == NULL) + return -ENOMEM; return 0; } -- cgit v1.2.3 From e3087b80aa0bceda9863f33307460f3ba79f2b15 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 8 Feb 2011 15:01:39 -0200 Subject: perf annotate: Fix --stdio rendering The checks for not using a max_lines parameter were b0rked, problem introduced in 3653246. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/annotate.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 1012841835a3..6db435167d74 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -132,7 +132,7 @@ static int objdump_line__print(struct objdump_line *oline, if (percent < min_pcnt) return -1; - if (printed >= max_lines) + if (max_lines && printed >= max_lines) return 1; color = get_percent_color(percent); @@ -154,7 +154,7 @@ static int objdump_line__print(struct objdump_line *oline, color_fprintf(stdout, color, " %7.2f", percent); printf(" : "); color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", oline->line); - } else if (printed >= max_lines) + } else if (max_lines && printed >= max_lines) return 1; else { if (!*oline->line) -- cgit v1.2.3 From ce6f4fab4059cd72638a0cfa596a8ee2c79c1c8e Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 8 Feb 2011 13:27:39 -0200 Subject: perf annotate: Move locking to struct annotation Since we'll need it when implementing the live annotate TUI browser. This also simplifies things a bit by having the list head for the source code to be in the dynamicly allocated part of struct annotation, that way we don't have to pass it around, it can be found from the struct symbol that is passed everywhere. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-annotate.c | 14 +++--- tools/perf/builtin-report.c | 5 +- tools/perf/builtin-top.c | 67 ++++++++++++++------------- tools/perf/util/annotate.c | 85 ++++++++++++++++++++-------------- tools/perf/util/annotate.h | 34 ++++++++------ tools/perf/util/hist.c | 5 +- tools/perf/util/hist.h | 3 +- tools/perf/util/top.h | 6 --- tools/perf/util/ui/browsers/annotate.c | 18 +++---- 9 files changed, 126 insertions(+), 111 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index ea6a1165956f..427182953fd7 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -62,7 +62,8 @@ static int hists__add_entry(struct hists *self, struct addr_location *al) * All aggregated on the first sym_hist. */ struct annotation *notes = symbol__annotation(he->ms.sym); - if (notes->histograms == NULL && symbol__alloc_hist(he->ms.sym, 1) < 0) + if (notes->src == NULL && + symbol__alloc_hist(he->ms.sym, 1) < 0) return -ENOMEM; return hist_entry__inc_addr_samples(he, 0, al->addr); @@ -77,7 +78,8 @@ static int process_sample_event(union perf_event *event, { struct addr_location al; - if (perf_event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + if (perf_event__preprocess_sample(event, session, &al, sample, + symbol__annotate_init) < 0) { pr_warning("problem processing %d event, skipping it.\n", event->header.type); return -1; @@ -111,7 +113,7 @@ static void hists__find_annotations(struct hists *self) goto find_next; notes = symbol__annotation(he->ms.sym); - if (notes->histograms == NULL) { + if (notes->src == NULL) { find_next: if (key == KEY_LEFT) nd = rb_prev(nd); @@ -142,11 +144,11 @@ find_next: nd = rb_next(nd); /* * Since we have a hist_entry per IP for the same - * symbol, free he->ms.sym->histogram to signal we already + * symbol, free he->ms.sym->src to signal we already * processed this symbol. */ - free(notes->histograms); - notes->histograms = NULL; + free(notes->src); + notes->src = NULL; } } } diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index de06bf55efff..f403aced4cba 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -123,7 +123,7 @@ static int perf_session__add_hist_entry(struct perf_session *session, * All aggregated on the first sym_hist. */ struct annotation *notes = symbol__annotation(he->ms.sym); - if (notes->histograms == NULL && + if (notes->src == NULL && symbol__alloc_hist(he->ms.sym, 1) < 0) err = -ENOMEM; else @@ -166,7 +166,8 @@ static int process_sample_event(union perf_event *event, struct addr_location al; struct perf_event_attr *attr; - if (perf_event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + if (perf_event__preprocess_sample(event, session, &al, sample, + symbol__annotate_init) < 0) { fprintf(stderr, "problem processing %d event, skipping it.\n", event->header.type); return -1; diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index b790673cb0aa..7dbf22d096b8 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -139,7 +139,7 @@ static void sig_winch_handler(int sig __used) static int parse_source(struct sym_entry *syme) { struct symbol *sym; - struct sym_entry_source *source; + struct annotation *notes; struct map *map; int err = -1; @@ -152,39 +152,35 @@ static int parse_source(struct sym_entry *syme) /* * We can't annotate with just /proc/kallsyms */ - if (map->dso->origin == DSO__ORIG_KERNEL) + if (map->dso->origin == DSO__ORIG_KERNEL) { + pr_err("Can't annotate %s: No vmlinux file was found in the " + "path\n", sym->name); + sleep(1); return -1; - - if (syme->src == NULL) { - syme->src = zalloc(sizeof(*source)); - if (syme->src == NULL) - return -1; - pthread_mutex_init(&syme->src->lock, NULL); - INIT_LIST_HEAD(&syme->src->head); } - source = syme->src; - - if (symbol__annotation(sym)->histograms != NULL) { - pthread_mutex_lock(&source->lock); + notes = symbol__annotation(sym); + if (notes->src != NULL) { + pthread_mutex_lock(¬es->lock); goto out_assign; } - pthread_mutex_lock(&source->lock); + pthread_mutex_lock(¬es->lock); if (symbol__alloc_hist(sym, top.evlist->nr_entries) < 0) { pr_err("Not enough memory for annotating '%s' symbol!\n", sym->name); + sleep(1); goto out_unlock; } - err = symbol__annotate(sym, syme->map, &source->head, 0); + err = symbol__annotate(sym, syme->map, 0); if (err == 0) { out_assign: sym_filter_entry = syme; } out_unlock: - pthread_mutex_unlock(&source->lock); + pthread_mutex_unlock(¬es->lock); return err; } @@ -196,20 +192,27 @@ static void __zero_source_counters(struct sym_entry *syme) static void record_precise_ip(struct sym_entry *syme, int counter, u64 ip) { + struct annotation *notes; + struct symbol *sym; + if (syme != sym_filter_entry) return; - if (pthread_mutex_trylock(&syme->src->lock)) + sym = sym_entry__symbol(syme); + notes = symbol__annotation(sym); + + if (pthread_mutex_trylock(¬es->lock)) return; ip = syme->map->map_ip(syme->map, ip); - symbol__inc_addr_samples(sym_entry__symbol(syme), syme->map, counter, ip); + symbol__inc_addr_samples(sym, syme->map, counter, ip); - pthread_mutex_unlock(&syme->src->lock); + pthread_mutex_unlock(¬es->lock); } static void show_details(struct sym_entry *syme) { + struct annotation *notes; struct symbol *symbol; int more; @@ -217,24 +220,26 @@ static void show_details(struct sym_entry *syme) return; symbol = sym_entry__symbol(syme); - if (!syme->src || symbol__annotation(symbol)->histograms == NULL) - return; + notes = symbol__annotation(symbol); + + pthread_mutex_lock(¬es->lock); + + if (notes->src == NULL) + goto out_unlock; printf("Showing %s for %s\n", event_name(top.sym_evsel), symbol->name); printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter); - pthread_mutex_lock(&syme->src->lock); - more = symbol__annotate_printf(symbol, syme->map, &syme->src->head, - top.sym_evsel->idx, 0, sym_pcnt_filter, - top.print_entries); + more = symbol__annotate_printf(symbol, syme->map, top.sym_evsel->idx, + 0, sym_pcnt_filter, top.print_entries); if (top.zero) symbol__annotate_zero_histogram(symbol, top.sym_evsel->idx); else - symbol__annotate_decay_histogram(symbol, &syme->src->head, - top.sym_evsel->idx); - pthread_mutex_unlock(&syme->src->lock); + symbol__annotate_decay_histogram(symbol, top.sym_evsel->idx); if (more != 0) printf("%d lines not displayed, maybe increase display entries [e]\n", more); +out_unlock: + pthread_mutex_unlock(¬es->lock); } static const char CONSOLE_CLEAR[] = ""; @@ -372,10 +377,8 @@ static void prompt_symbol(struct sym_entry **target, const char *msg) /* zero counters of active symbol */ if (syme) { - pthread_mutex_lock(&syme->src->lock); __zero_source_counters(syme); *target = NULL; - pthread_mutex_unlock(&syme->src->lock); } fprintf(stdout, "\n%s: ", msg); @@ -554,10 +557,8 @@ static void handle_keypress(struct perf_session *session, int c) else { struct sym_entry *syme = sym_filter_entry; - pthread_mutex_lock(&syme->src->lock); sym_filter_entry = NULL; __zero_source_counters(syme); - pthread_mutex_unlock(&syme->src->lock); } break; case 'U': @@ -653,7 +654,7 @@ static int symbol_filter(struct map *map, struct symbol *sym) syme = symbol__priv(sym); syme->map = map; - syme->src = NULL; + symbol__annotate_init(map, sym); if (!sym_filter_entry && sym_filter && !strcmp(name, sym_filter)) { /* schedule initial sym_filter_entry setup */ diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 6db435167d74..c777bdaf91da 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -14,25 +14,39 @@ #include "symbol.h" #include "debug.h" #include "annotate.h" +#include -int symbol__alloc_hist(struct symbol *sym, int nevents) +int symbol__annotate_init(struct map *map __used, struct symbol *sym) { struct annotation *notes = symbol__annotation(sym); + pthread_mutex_init(¬es->lock, NULL); + return 0; +} - notes->sizeof_sym_hist = (sizeof(*notes->histograms) + +int symbol__alloc_hist(struct symbol *sym, int nevents) +{ + struct annotation *notes = symbol__annotation(sym); + size_t sizeof_sym_hist = (sizeof(struct sym_hist) + (sym->end - sym->start) * sizeof(u64)); - notes->histograms = calloc(nevents, notes->sizeof_sym_hist); - notes->nr_histograms = nevents; - return notes->histograms == NULL ? -1 : 0; + + notes->src = zalloc(sizeof(*notes->src) + nevents * sizeof_sym_hist); + if (notes->src == NULL) + return -1; + notes->src->sizeof_sym_hist = sizeof_sym_hist; + notes->src->nr_histograms = nevents; + INIT_LIST_HEAD(¬es->src->source); + return 0; } void symbol__annotate_zero_histograms(struct symbol *sym) { struct annotation *notes = symbol__annotation(sym); - if (notes->histograms != NULL) - memset(notes->histograms, 0, - notes->nr_histograms * notes->sizeof_sym_hist); + pthread_mutex_lock(¬es->lock); + if (notes->src != NULL) + memset(notes->src->histograms, 0, + notes->src->nr_histograms * notes->src->sizeof_sym_hist); + pthread_mutex_unlock(¬es->lock); } int symbol__inc_addr_samples(struct symbol *sym, struct map *map, @@ -43,7 +57,7 @@ int symbol__inc_addr_samples(struct symbol *sym, struct map *map, struct sym_hist *h; notes = symbol__annotation(sym); - if (notes->histograms == NULL) + if (notes->src == NULL) return -ENOMEM; pr_debug3("%s: addr=%#" PRIx64 "\n", __func__, map->unmap_ip(map, addr)); @@ -95,8 +109,7 @@ struct objdump_line *objdump__get_next_ip_line(struct list_head *head, return NULL; } -static int objdump_line__print(struct objdump_line *oline, - struct list_head *head, struct symbol *sym, +static int objdump_line__print(struct objdump_line *oline, struct symbol *sym, int evidx, u64 len, int min_pcnt, int printed, int max_lines) { @@ -109,10 +122,12 @@ static int objdump_line__print(struct objdump_line *oline, double percent = 0.0; const char *color; struct annotation *notes = symbol__annotation(sym); - struct source_line *src_line = notes->src_line; + struct source_line *src_line = notes->src->lines; struct sym_hist *h = annotation__histogram(notes, evidx); s64 offset = oline->offset; - struct objdump_line *next = objdump__get_next_ip_line(head, oline); + struct objdump_line *next; + + next = objdump__get_next_ip_line(¬es->src->source, oline); while (offset < (s64)len && (next == NULL || offset < next->offset)) { @@ -166,9 +181,10 @@ static int objdump_line__print(struct objdump_line *oline, return 0; } -static int symbol__parse_objdump_line(struct symbol *sym, struct map *map, FILE *file, - struct list_head *head, size_t privsize) +static int symbol__parse_objdump_line(struct symbol *sym, struct map *map, + FILE *file, size_t privsize) { + struct annotation *notes = symbol__annotation(sym); struct objdump_line *objdump_line; char *line = NULL, *tmp, *tmp2, *c; size_t line_len; @@ -222,13 +238,12 @@ static int symbol__parse_objdump_line(struct symbol *sym, struct map *map, FILE free(line); return -1; } - objdump__add_line(head, objdump_line); + objdump__add_line(¬es->src->source, objdump_line); return 0; } -int symbol__annotate(struct symbol *sym, struct map *map, - struct list_head *head, size_t privsize) +int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize) { struct dso *dso = map->dso; char *filename = dso__build_id_filename(dso, NULL, 0); @@ -297,7 +312,7 @@ fallback: goto out_free_filename; while (!feof(file)) - if (symbol__parse_objdump_line(sym, map, file, head, privsize) < 0) + if (symbol__parse_objdump_line(sym, map, file, privsize) < 0) break; pclose(file); @@ -330,14 +345,14 @@ static void insert_source_line(struct rb_root *root, struct source_line *src_lin static void symbol__free_source_line(struct symbol *sym, int len) { struct annotation *notes = symbol__annotation(sym); - struct source_line *src_line = notes->src_line; + struct source_line *src_line = notes->src->lines; int i; for (i = 0; i < len; i++) free(src_line[i].path); free(src_line); - notes->src_line = NULL; + notes->src->lines = NULL; } /* Get the filename:line for the colored entries */ @@ -355,8 +370,8 @@ static int symbol__get_source_line(struct symbol *sym, struct map *map, if (!h->sum) return 0; - src_line = notes->src_line = calloc(len, sizeof(struct source_line)); - if (!notes->src_line) + src_line = notes->src->lines = calloc(len, sizeof(struct source_line)); + if (!notes->src->lines) return -1; start = map->unmap_ip(map, sym->start); @@ -436,12 +451,12 @@ static void symbol__annotate_hits(struct symbol *sym, int evidx) printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); } -int symbol__annotate_printf(struct symbol *sym, struct map *map, - struct list_head *head, int evidx, bool full_paths, - int min_pcnt, int max_lines) +int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, + bool full_paths, int min_pcnt, int max_lines) { struct dso *dso = map->dso; const char *filename = dso->long_name, *d_filename; + struct annotation *notes = symbol__annotation(sym); struct objdump_line *pos; int printed = 2; int more = 0; @@ -460,8 +475,8 @@ int symbol__annotate_printf(struct symbol *sym, struct map *map, if (verbose) symbol__annotate_hits(sym, evidx); - list_for_each_entry(pos, head, node) { - switch (objdump_line__print(pos, head, sym, evidx, len, min_pcnt, + list_for_each_entry(pos, ¬es->src->source, node) { + switch (objdump_line__print(pos, sym, evidx, len, min_pcnt, printed, max_lines)) { case 0: ++printed; @@ -485,11 +500,10 @@ void symbol__annotate_zero_histogram(struct symbol *sym, int evidx) struct annotation *notes = symbol__annotation(sym); struct sym_hist *h = annotation__histogram(notes, evidx); - memset(h, 0, notes->sizeof_sym_hist); + memset(h, 0, notes->src->sizeof_sym_hist); } -void symbol__annotate_decay_histogram(struct symbol *sym, - struct list_head *head, int evidx) +void symbol__annotate_decay_histogram(struct symbol *sym, int evidx) { struct annotation *notes = symbol__annotation(sym); struct sym_hist *h = annotation__histogram(notes, evidx); @@ -497,7 +511,7 @@ void symbol__annotate_decay_histogram(struct symbol *sym, h->sum = 0; - list_for_each_entry(pos, head, node) { + list_for_each_entry(pos, ¬es->src->source, node) { if (pos->offset != -1) { h->addr[pos->offset] = h->addr[pos->offset] * 7 / 8; h->sum += h->addr[pos->offset]; @@ -522,10 +536,9 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, struct dso *dso = map->dso; const char *filename = dso->long_name; struct rb_root source_line = RB_ROOT; - LIST_HEAD(head); u64 len; - if (symbol__annotate(sym, map, &head, 0) < 0) + if (symbol__annotate(sym, map, 0) < 0) return -1; len = sym->end - sym->start; @@ -536,12 +549,12 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, print_summary(&source_line, filename); } - symbol__annotate_printf(sym, map, &head, evidx, full_paths, + symbol__annotate_printf(sym, map, evidx, full_paths, min_pcnt, max_lines); if (print_lines) symbol__free_source_line(sym, len); - objdump_line_list__purge(&head); + objdump_line_list__purge(&symbol__annotation(sym)->src->source); return 0; } diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index bc08b36a713a..b237c8678c22 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -28,22 +28,29 @@ struct source_line { char *path; }; -/** struct annotation - symbols with hits have this attached as in sannotation +/** struct annotated_source - symbols with hits have this attached as in sannotation * * @histogram: Array of addr hit histograms per event being monitored - * @src_line: If 'print_lines' is specified, per source code line percentages + * @lines: If 'print_lines' is specified, per source code line percentages + * @source: source parsed from objdump -dS * - * src_line is allocated, percentages calculated and all sorted by percentage + * lines is allocated, percentages calculated and all sorted by percentage * when the annotation is about to be presented, so the percentages are for * one of the entries in the histogram array, i.e. for the event/counter being * presented. It is deallocated right after symbol__{tui,tty,etc}_annotate * returns. */ -struct annotation { - struct source_line *src_line; - struct sym_hist *histograms; +struct annotated_source { + struct list_head source; + struct source_line *lines; int nr_histograms; int sizeof_sym_hist; + struct sym_hist histograms[0]; +}; + +struct annotation { + pthread_mutex_t lock; + struct annotated_source *src; }; struct sannotation { @@ -53,7 +60,8 @@ struct sannotation { static inline struct sym_hist *annotation__histogram(struct annotation *notes, int idx) { - return ((void *)notes->histograms) + (notes->sizeof_sym_hist * idx); + return (((void *)¬es->src->histograms) + + (notes->src->sizeof_sym_hist * idx)); } static inline struct annotation *symbol__annotation(struct symbol *sym) @@ -67,14 +75,12 @@ int symbol__inc_addr_samples(struct symbol *sym, struct map *map, int symbol__alloc_hist(struct symbol *sym, int nevents); void symbol__annotate_zero_histograms(struct symbol *sym); -int symbol__annotate(struct symbol *sym, struct map *map, - struct list_head *head, size_t privsize); -int symbol__annotate_printf(struct symbol *sym, struct map *map, - struct list_head *head, int evidx, bool full_paths, - int min_pcnt, int max_lines); +int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize); +int symbol__annotate_init(struct map *map __used, struct symbol *sym); +int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, + bool full_paths, int min_pcnt, int max_lines); void symbol__annotate_zero_histogram(struct symbol *sym, int evidx); -void symbol__annotate_decay_histogram(struct symbol *sym, - struct list_head *head, int evidx); +void symbol__annotate_decay_histogram(struct symbol *sym, int evidx); void objdump_line_list__purge(struct list_head *head); int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index bac5ab684967..3f437236f193 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -955,10 +955,9 @@ int hist_entry__inc_addr_samples(struct hist_entry *he, int evidx, u64 ip) return symbol__inc_addr_samples(he->ms.sym, he->ms.map, evidx, ip); } -int hist_entry__annotate(struct hist_entry *he, struct list_head *head, - size_t privsize) +int hist_entry__annotate(struct hist_entry *he, size_t privsize) { - return symbol__annotate(he->ms.sym, he->ms.map, head, privsize); + return symbol__annotate(he->ms.sym, he->ms.map, privsize); } void hists__inc_nr_events(struct hists *self, u32 type) diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 2c6cdae6a764..37c79089de09 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -78,8 +78,7 @@ size_t hists__fprintf(struct hists *self, struct hists *pair, bool show_displacement, FILE *fp); int hist_entry__inc_addr_samples(struct hist_entry *self, int evidx, u64 addr); -int hist_entry__annotate(struct hist_entry *self, struct list_head *head, - size_t privsize); +int hist_entry__annotate(struct hist_entry *self, size_t privsize); void hists__filter_by_dso(struct hists *self, const struct dso *dso); void hists__filter_by_thread(struct hists *self, const struct thread *thread); diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index 62e32939f3d1..4f769f47e19a 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -11,11 +11,6 @@ struct perf_evlist; struct perf_evsel; -struct sym_entry_source { - struct list_head head; - pthread_mutex_t lock; -}; - struct sym_entry { struct rb_node rb_node; struct list_head node; @@ -24,7 +19,6 @@ struct sym_entry { int skip; u8 origin; struct map *map; - struct sym_entry_source *src; unsigned long count[0]; }; diff --git a/tools/perf/util/ui/browsers/annotate.c b/tools/perf/util/ui/browsers/annotate.c index 8d8a16895af7..1aa39658539c 100644 --- a/tools/perf/util/ui/browsers/annotate.c +++ b/tools/perf/util/ui/browsers/annotate.c @@ -60,7 +60,6 @@ static void annotate_browser__write(struct ui_browser *self, void *entry, int ro } static double objdump_line__calc_percent(struct objdump_line *self, - struct list_head *head, struct symbol *sym, int evidx) { double percent = 0.0; @@ -69,11 +68,12 @@ static double objdump_line__calc_percent(struct objdump_line *self, int len = sym->end - sym->start; unsigned int hits = 0; struct annotation *notes = symbol__annotation(sym); - struct source_line *src_line = notes->src_line; + struct source_line *src_line = notes->src->lines; struct sym_hist *h = annotation__histogram(notes, evidx); s64 offset = self->offset; - struct objdump_line *next = objdump__get_next_ip_line(head, self); + struct objdump_line *next; + next = objdump__get_next_ip_line(¬es->src->source, self); while (offset < (s64)len && (next == NULL || offset < next->offset)) { if (src_line) { @@ -192,10 +192,10 @@ int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx) { struct objdump_line *pos, *n; struct objdump_line_rb_node *rbpos; - LIST_HEAD(head); + struct annotation *notes = symbol__annotation(sym); struct annotate_browser browser = { .b = { - .entries = &head, + .entries = ¬es->src->source, .refresh = ui_browser__list_head_refresh, .seek = ui_browser__list_head_seek, .write = annotate_browser__write, @@ -210,20 +210,20 @@ int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx) if (map->dso->annotate_warned) return -1; - if (symbol__annotate(sym, map, &head, sizeof(*rbpos)) < 0) { + if (symbol__annotate(sym, map, sizeof(*rbpos)) < 0) { ui__error_window(ui_helpline__last_msg); return -1; } ui_helpline__push("Press <- or ESC to exit"); - list_for_each_entry(pos, &head, node) { + list_for_each_entry(pos, ¬es->src->source, node) { size_t line_len = strlen(pos->line); if (browser.b.width < line_len) browser.b.width = line_len; rbpos = objdump_line__rb(pos); rbpos->idx = browser.b.nr_entries++; - rbpos->percent = objdump_line__calc_percent(pos, &head, sym, evidx); + rbpos->percent = objdump_line__calc_percent(pos, sym, evidx); if (rbpos->percent < 0.01) continue; objdump__insert_line(&browser.entries, rbpos); @@ -238,7 +238,7 @@ int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx) browser.b.width += 18; /* Percentage */ ret = annotate_browser__run(&browser); - list_for_each_entry_safe(pos, n, &head, node) { + list_for_each_entry_safe(pos, n, ¬es->src->source, node) { list_del(&pos->node); objdump_line__free(pos); } -- cgit v1.2.3 From d5e3d747007fdb541e57ed72e020ff0b94db3470 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 8 Feb 2011 15:29:25 -0200 Subject: perf annotate: Fix annotate context lines regression The live annotation done in 'perf top' needs to limit the context before lines that aren't filtered out by the min percent filter, if we don't do that, the screen in a tty often is not enough for showing what is interesting: lines with hits and a few source code lines before it. Reported-by: Mike Galbraith Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-top.c | 2 +- tools/perf/util/annotate.c | 47 +++++++++++++++++++++++++++++++++++++++------- tools/perf/util/annotate.h | 3 ++- 3 files changed, 43 insertions(+), 9 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 7dbf22d096b8..210c736e6db4 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -231,7 +231,7 @@ static void show_details(struct sym_entry *syme) printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter); more = symbol__annotate_printf(symbol, syme->map, top.sym_evsel->idx, - 0, sym_pcnt_filter, top.print_entries); + 0, sym_pcnt_filter, top.print_entries, 4); if (top.zero) symbol__annotate_zero_histogram(symbol, top.sym_evsel->idx); else diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index c777bdaf91da..02976b895f27 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -111,7 +111,8 @@ struct objdump_line *objdump__get_next_ip_line(struct list_head *head, static int objdump_line__print(struct objdump_line *oline, struct symbol *sym, int evidx, u64 len, int min_pcnt, - int printed, int max_lines) + int printed, int max_lines, + struct objdump_line *queue) { static const char *prev_line; static const char *prev_color; @@ -150,6 +151,15 @@ static int objdump_line__print(struct objdump_line *oline, struct symbol *sym, if (max_lines && printed >= max_lines) return 1; + if (queue != NULL) { + list_for_each_entry_from(queue, ¬es->src->source, node) { + if (queue == oline) + break; + objdump_line__print(queue, sym, evidx, len, + 0, 0, 1, NULL); + } + } + color = get_percent_color(percent); /* @@ -172,6 +182,9 @@ static int objdump_line__print(struct objdump_line *oline, struct symbol *sym, } else if (max_lines && printed >= max_lines) return 1; else { + if (queue) + return -1; + if (!*oline->line) printf(" :\n"); else @@ -452,13 +465,14 @@ static void symbol__annotate_hits(struct symbol *sym, int evidx) } int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, - bool full_paths, int min_pcnt, int max_lines) + bool full_paths, int min_pcnt, int max_lines, + int context) { struct dso *dso = map->dso; const char *filename = dso->long_name, *d_filename; struct annotation *notes = symbol__annotation(sym); - struct objdump_line *pos; - int printed = 2; + struct objdump_line *pos, *queue = NULL; + int printed = 2, queue_len = 0; int more = 0; u64 len; @@ -476,10 +490,20 @@ int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, symbol__annotate_hits(sym, evidx); list_for_each_entry(pos, ¬es->src->source, node) { + if (context && queue == NULL) { + queue = pos; + queue_len = 0; + } + switch (objdump_line__print(pos, sym, evidx, len, min_pcnt, - printed, max_lines)) { + printed, max_lines, queue)) { case 0: ++printed; + if (context) { + printed += queue_len; + queue = NULL; + queue_len = 0; + } break; case 1: /* filtered by max_lines */ @@ -487,7 +511,16 @@ int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, break; case -1: default: - /* filtered by min_pcnt */ + /* + * Filtered by min_pcnt or non IP lines when + * context != 0 + */ + if (!context) + break; + if (queue_len == context) + queue = list_entry(queue->node.next, typeof(*queue), node); + else + ++queue_len; break; } } @@ -550,7 +583,7 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, } symbol__annotate_printf(sym, map, evidx, full_paths, - min_pcnt, max_lines); + min_pcnt, max_lines, 0); if (print_lines) symbol__free_source_line(sym, len); diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index b237c8678c22..e848803fcd48 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -78,7 +78,8 @@ void symbol__annotate_zero_histograms(struct symbol *sym); int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize); int symbol__annotate_init(struct map *map __used, struct symbol *sym); int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, - bool full_paths, int min_pcnt, int max_lines); + bool full_paths, int min_pcnt, int max_lines, + int context); void symbol__annotate_zero_histogram(struct symbol *sym, int evidx); void symbol__annotate_decay_histogram(struct symbol *sym, int evidx); void objdump_line_list__purge(struct list_head *head); -- cgit v1.2.3 From 023695d96ee06f36cf5014e286edcd623e9fb847 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Mon, 14 Feb 2011 11:20:01 +0200 Subject: perf tool: Add cgroup support This patch adds the ability to filter monitoring based on container groups (cgroups) for both perf stat and perf record. It is possible to monitor multiple cgroup in parallel. There is one cgroup per event. The cgroups to monitor are passed via a new -G option followed by a comma separated list of cgroup names. The cgroup filesystem has to be mounted. Given a cgroup name, the perf tool finds the corresponding directory in the cgroup filesystem and opens it. It then passes that file descriptor to the kernel. Example: $ perf stat -B -a -e cycles:u,cycles:u,cycles:u -G test1,,test2 -- sleep 1 Performance counter stats for 'sleep 1': 2,368,667,414 cycles test1 2,369,661,459 cycles cycles test2 1.001856890 seconds time elapsed Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra LKML-Reference: <4d590290.825bdf0a.7d0a.4890@mx.google.com> Signed-off-by: Ingo Molnar --- tools/perf/Documentation/perf-record.txt | 11 ++ tools/perf/Documentation/perf-stat.txt | 11 ++ tools/perf/Makefile | 2 + tools/perf/builtin-record.c | 9 ++ tools/perf/builtin-stat.c | 40 +++++-- tools/perf/util/cgroup.c | 178 +++++++++++++++++++++++++++++++ tools/perf/util/cgroup.h | 17 +++ tools/perf/util/evsel.c | 16 ++- tools/perf/util/evsel.h | 2 + 9 files changed, 276 insertions(+), 10 deletions(-) create mode 100644 tools/perf/util/cgroup.c create mode 100644 tools/perf/util/cgroup.h (limited to 'tools/perf/util') diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt index e032716c839b..5a520f825295 100644 --- a/tools/perf/Documentation/perf-record.txt +++ b/tools/perf/Documentation/perf-record.txt @@ -137,6 +137,17 @@ Do not update the builid cache. This saves some overhead in situations where the information in the perf.data file (which includes buildids) is sufficient. +-G name,...:: +--cgroup name,...:: +monitor only in the container (cgroup) called "name". This option is available only +in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to +container "name" are monitored when they run on the monitored CPUs. Multiple cgroups +can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup +to first event, second cgroup to second event and so on. It is possible to provide +an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have +corresponding events, i.e., they always refer to events defined earlier on the command +line. + SEE ALSO -------- linkperf:perf-stat[1], linkperf:perf-list[1] diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt index b6da7affbbee..918cc38ee6d1 100644 --- a/tools/perf/Documentation/perf-stat.txt +++ b/tools/perf/Documentation/perf-stat.txt @@ -83,6 +83,17 @@ This option is only valid in system-wide mode. print counts using a CSV-style output to make it easy to import directly into spreadsheets. Columns are separated by the string specified in SEP. +-G name:: +--cgroup name:: +monitor only in the container (cgroup) called "name". This option is available only +in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to +container "name" are monitored when they run on the monitored CPUs. Multiple cgroups +can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup +to first event, second cgroup to second event and so on. It is possible to provide +an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have +corresponding events, i.e., they always refer to events defined earlier on the command +line. + EXAMPLES -------- diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 94f73abdc56a..bc4d9bf8a556 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -442,6 +442,7 @@ LIB_H += util/pstack.h LIB_H += util/cpumap.h LIB_H += util/top.h LIB_H += $(ARCH_INCLUDE) +LIB_H += util/cgroup.h LIB_OBJS += $(OUTPUT)util/abspath.o LIB_OBJS += $(OUTPUT)util/alias.o @@ -496,6 +497,7 @@ LIB_OBJS += $(OUTPUT)util/probe-event.o LIB_OBJS += $(OUTPUT)util/util.o LIB_OBJS += $(OUTPUT)util/xyarray.o LIB_OBJS += $(OUTPUT)util/cpumap.o +LIB_OBJS += $(OUTPUT)util/cgroup.o BUILTIN_OBJS += $(OUTPUT)builtin-annotate.o diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 12e0e41696d9..a4aaadcb4c8b 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -807,6 +807,9 @@ const struct option record_options[] = { "do not update the buildid cache"), OPT_BOOLEAN('B', "no-buildid", &no_buildid, "do not collect buildids in perf.data"), + OPT_CALLBACK('G', "cgroup", &evsel_list, "name", + "monitor event in cgroup name only", + parse_cgroups), OPT_END() }; @@ -835,6 +838,12 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) write_mode = WRITE_FORCE; } + if (nr_cgroups && !system_wide) { + fprintf(stderr, "cgroup monitoring only available in" + " system-wide mode\n"); + usage_with_options(record_usage, record_options); + } + symbol__init(); if (no_buildid_cache || no_buildid) diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index 806a9998fcd5..21c025222496 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -390,6 +390,9 @@ static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg) fprintf(stderr, fmt, cpustr, msecs, csv_sep, event_name(evsel)); + if (evsel->cgrp) + fprintf(stderr, "%s%s", csv_sep, evsel->cgrp->name); + if (csv_output) return; @@ -420,6 +423,9 @@ static void abs_printout(int cpu, struct perf_evsel *evsel, double avg) fprintf(stderr, fmt, cpustr, avg, csv_sep, event_name(evsel)); + if (evsel->cgrp) + fprintf(stderr, "%s%s", csv_sep, evsel->cgrp->name); + if (csv_output) return; @@ -460,9 +466,17 @@ static void print_counter_aggr(struct perf_evsel *counter) int scaled = counter->counts->scaled; if (scaled == -1) { - fprintf(stderr, "%*s%s%-24s\n", + fprintf(stderr, "%*s%s%*s", csv_output ? 0 : 18, - "", csv_sep, event_name(counter)); + "", + csv_sep, + csv_output ? 0 : -24, + event_name(counter)); + + if (counter->cgrp) + fprintf(stderr, "%s%s", csv_sep, counter->cgrp->name); + + fputc('\n', stderr); return; } @@ -487,7 +501,6 @@ static void print_counter_aggr(struct perf_evsel *counter) fprintf(stderr, " (scaled from %.2f%%)", 100 * avg_running / avg_enabled); } - fprintf(stderr, "\n"); } @@ -505,14 +518,18 @@ static void print_counter(struct perf_evsel *counter) ena = counter->counts->cpu[cpu].ena; run = counter->counts->cpu[cpu].run; if (run == 0 || ena == 0) { - fprintf(stderr, "CPU%*d%s%*s%s%-24s", + fprintf(stderr, "CPU%*d%s%*s%s%*s", csv_output ? 0 : -4, evsel_list->cpus->map[cpu], csv_sep, csv_output ? 0 : 18, "", csv_sep, + csv_output ? 0 : -24, event_name(counter)); - fprintf(stderr, "\n"); + if (counter->cgrp) + fprintf(stderr, "%s%s", csv_sep, counter->cgrp->name); + + fputc('\n', stderr); continue; } @@ -529,7 +546,7 @@ static void print_counter(struct perf_evsel *counter) 100.0 * run / ena); } } - fprintf(stderr, "\n"); + fputc('\n', stderr); } } @@ -642,6 +659,9 @@ static const struct option options[] = { "disable CPU count aggregation"), OPT_STRING('x', "field-separator", &csv_sep, "separator", "print counts with custom separator"), + OPT_CALLBACK('G', "cgroup", &evsel_list, "name", + "monitor event in cgroup name only", + parse_cgroups), OPT_END() }; @@ -682,9 +702,13 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) if (run_count <= 0) usage_with_options(stat_usage, options); - /* no_aggr is for system-wide only */ - if (no_aggr && !system_wide) + /* no_aggr, cgroup are for system-wide only */ + if ((no_aggr || nr_cgroups) && !system_wide) { + fprintf(stderr, "both cgroup and no-aggregation " + "modes only available in system-wide mode\n"); + usage_with_options(stat_usage, options); + } /* Set attrs and nr_counters if no event is selected and !null_run */ if (!null_run && !evsel_list->nr_entries) { diff --git a/tools/perf/util/cgroup.c b/tools/perf/util/cgroup.c new file mode 100644 index 000000000000..9fea75535221 --- /dev/null +++ b/tools/perf/util/cgroup.c @@ -0,0 +1,178 @@ +#include "util.h" +#include "../perf.h" +#include "parse-options.h" +#include "evsel.h" +#include "cgroup.h" +#include "debugfs.h" /* MAX_PATH, STR() */ +#include "evlist.h" + +int nr_cgroups; + +static int +cgroupfs_find_mountpoint(char *buf, size_t maxlen) +{ + FILE *fp; + char mountpoint[MAX_PATH+1], tokens[MAX_PATH+1], type[MAX_PATH+1]; + char *token, *saved_ptr; + int found = 0; + + fp = fopen("/proc/mounts", "r"); + if (!fp) + return -1; + + /* + * in order to handle split hierarchy, we need to scan /proc/mounts + * and inspect every cgroupfs mount point to find one that has + * perf_event subsystem + */ + while (fscanf(fp, "%*s %"STR(MAX_PATH)"s %"STR(MAX_PATH)"s %" + STR(MAX_PATH)"s %*d %*d\n", + mountpoint, type, tokens) == 3) { + + if (!strcmp(type, "cgroup")) { + + token = strtok_r(tokens, ",", &saved_ptr); + + while (token != NULL) { + if (!strcmp(token, "perf_event")) { + found = 1; + break; + } + token = strtok_r(NULL, ",", &saved_ptr); + } + } + if (found) + break; + } + fclose(fp); + if (!found) + return -1; + + if (strlen(mountpoint) < maxlen) { + strcpy(buf, mountpoint); + return 0; + } + return -1; +} + +static int open_cgroup(char *name) +{ + char path[MAX_PATH+1]; + char mnt[MAX_PATH+1]; + int fd; + + + if (cgroupfs_find_mountpoint(mnt, MAX_PATH+1)) + return -1; + + snprintf(path, MAX_PATH, "%s/%s", mnt, name); + + fd = open(path, O_RDONLY); + if (fd == -1) + fprintf(stderr, "no access to cgroup %s\n", path); + + return fd; +} + +static int add_cgroup(struct perf_evlist *evlist, char *str) +{ + struct perf_evsel *counter; + struct cgroup_sel *cgrp = NULL; + int n; + /* + * check if cgrp is already defined, if so we reuse it + */ + list_for_each_entry(counter, &evlist->entries, node) { + cgrp = counter->cgrp; + if (!cgrp) + continue; + if (!strcmp(cgrp->name, str)) + break; + + cgrp = NULL; + } + + if (!cgrp) { + cgrp = zalloc(sizeof(*cgrp)); + if (!cgrp) + return -1; + + cgrp->name = str; + + cgrp->fd = open_cgroup(str); + if (cgrp->fd == -1) { + free(cgrp); + return -1; + } + } + + /* + * find corresponding event + * if add cgroup N, then need to find event N + */ + n = 0; + list_for_each_entry(counter, &evlist->entries, node) { + if (n == nr_cgroups) + goto found; + n++; + } + if (cgrp->refcnt == 0) + free(cgrp); + + return -1; +found: + cgrp->refcnt++; + counter->cgrp = cgrp; + return 0; +} + +void close_cgroup(struct cgroup_sel *cgrp) +{ + if (!cgrp) + return; + + /* XXX: not reentrant */ + if (--cgrp->refcnt == 0) { + close(cgrp->fd); + free(cgrp->name); + free(cgrp); + } +} + +int parse_cgroups(const struct option *opt __used, const char *str, + int unset __used) +{ + struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; + const char *p, *e, *eos = str + strlen(str); + char *s; + int ret; + + if (list_empty(&evlist->entries)) { + fprintf(stderr, "must define events before cgroups\n"); + return -1; + } + + for (;;) { + p = strchr(str, ','); + e = p ? p : eos; + + /* allow empty cgroups, i.e., skip */ + if (e - str) { + /* termination added */ + s = strndup(str, e - str); + if (!s) + return -1; + ret = add_cgroup(evlist, s); + if (ret) { + free(s); + return -1; + } + } + /* nr_cgroups is increased een for empty cgroups */ + nr_cgroups++; + if (!p) + break; + str = p+1; + } + return 0; +} diff --git a/tools/perf/util/cgroup.h b/tools/perf/util/cgroup.h new file mode 100644 index 000000000000..89acd6debdc5 --- /dev/null +++ b/tools/perf/util/cgroup.h @@ -0,0 +1,17 @@ +#ifndef __CGROUP_H__ +#define __CGROUP_H__ + +struct option; + +struct cgroup_sel { + char *name; + int fd; + int refcnt; +}; + + +extern int nr_cgroups; /* number of explicit cgroups defined */ +extern void close_cgroup(struct cgroup_sel *cgrp); +extern int parse_cgroups(const struct option *opt, const char *str, int unset); + +#endif /* __CGROUP_H__ */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 211063eed474..c974e08d07ab 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -85,6 +85,7 @@ void perf_evsel__exit(struct perf_evsel *evsel) void perf_evsel__delete(struct perf_evsel *evsel) { perf_evsel__exit(evsel); + close_cgroup(evsel->cgrp); free(evsel); } @@ -163,21 +164,32 @@ static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, struct thread_map *threads, bool group, bool inherit) { int cpu, thread; + unsigned long flags = 0; + int pid = -1; if (evsel->fd == NULL && perf_evsel__alloc_fd(evsel, cpus->nr, threads->nr) < 0) return -1; + if (evsel->cgrp) { + flags = PERF_FLAG_PID_CGROUP; + pid = evsel->cgrp->fd; + } + for (cpu = 0; cpu < cpus->nr; cpu++) { int group_fd = -1; evsel->attr.inherit = (cpus->map[cpu] < 0) && inherit; for (thread = 0; thread < threads->nr; thread++) { + + if (!evsel->cgrp) + pid = threads->map[thread]; + FD(evsel, cpu, thread) = sys_perf_event_open(&evsel->attr, - threads->map[thread], + pid, cpus->map[cpu], - group_fd, 0); + group_fd, flags); if (FD(evsel, cpu, thread) < 0) goto out_close; diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index eecdc3aabc14..1d3d5a3dbe60 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -6,6 +6,7 @@ #include "../../../include/linux/perf_event.h" #include "types.h" #include "xyarray.h" +#include "cgroup.h" struct perf_counts_values { union { @@ -45,6 +46,7 @@ struct perf_evsel { struct perf_counts *counts; int idx; void *priv; + struct cgroup_sel *cgrp; }; struct cpu_map; -- cgit v1.2.3 From 5c35d69fb60b1dc49595f5b9a2c7158283e9eaf3 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 9 Feb 2011 11:38:43 -0200 Subject: perf ui: Serialize screen updates The ui operations so far were used by just one thread, but 'perf top --tui' now has two threads updating the screen, so we need to use a mutex to avoid garbling the screen. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Makefile | 1 + tools/perf/util/ui/browser.c | 7 +++++++ tools/perf/util/ui/helpline.c | 5 ++++- tools/perf/util/ui/setup.c | 3 +++ tools/perf/util/ui/ui.h | 8 ++++++++ 5 files changed, 23 insertions(+), 1 deletion(-) create mode 100644 tools/perf/util/ui/ui.h (limited to 'tools/perf/util') diff --git a/tools/perf/Makefile b/tools/perf/Makefile index bc4d9bf8a556..ffd1047cd93b 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -637,6 +637,7 @@ else LIB_H += util/ui/libslang.h LIB_H += util/ui/progress.h LIB_H += util/ui/util.h + LIB_H += util/ui/ui.h endif endif diff --git a/tools/perf/util/ui/browser.c b/tools/perf/util/ui/browser.c index 8bc010edca25..60d6c815e1db 100644 --- a/tools/perf/util/ui/browser.c +++ b/tools/perf/util/ui/browser.c @@ -1,4 +1,5 @@ #include "libslang.h" +#include "ui.h" #include #include #include @@ -178,6 +179,7 @@ int ui_browser__show(struct ui_browser *self, const char *title, if (self->sb == NULL) return -1; + pthread_mutex_lock(&ui__lock); SLsmg_gotorc(0, 0); ui_browser__set_color(self, NEWT_COLORSET_ROOT); slsmg_write_nstring(title, self->width); @@ -188,25 +190,30 @@ int ui_browser__show(struct ui_browser *self, const char *title, va_start(ap, helpline); ui_helpline__vpush(helpline, ap); va_end(ap); + pthread_mutex_unlock(&ui__lock); return 0; } void ui_browser__hide(struct ui_browser *self) { + pthread_mutex_lock(&ui__lock); newtFormDestroy(self->form); self->form = NULL; ui_helpline__pop(); + pthread_mutex_unlock(&ui__lock); } int ui_browser__refresh(struct ui_browser *self) { int row; + pthread_mutex_lock(&ui__lock); newtScrollbarSet(self->sb, self->index, self->nr_entries - 1); row = self->refresh(self); ui_browser__set_color(self, HE_COLORSET_NORMAL); SLsmg_fill_region(self->y + row, self->x, self->height - row, self->width, ' '); + pthread_mutex_unlock(&ui__lock); return 0; } diff --git a/tools/perf/util/ui/helpline.c b/tools/perf/util/ui/helpline.c index 8d79daa4458a..f36d2ff509ed 100644 --- a/tools/perf/util/ui/helpline.c +++ b/tools/perf/util/ui/helpline.c @@ -5,6 +5,7 @@ #include "../debug.h" #include "helpline.h" +#include "ui.h" void ui_helpline__pop(void) { @@ -55,7 +56,8 @@ int ui_helpline__show_help(const char *format, va_list ap) int ret; static int backlog; - ret = vsnprintf(ui_helpline__last_msg + backlog, + pthread_mutex_lock(&ui__lock); + ret = vsnprintf(ui_helpline__last_msg + backlog, sizeof(ui_helpline__last_msg) - backlog, format, ap); backlog += ret; @@ -64,6 +66,7 @@ int ui_helpline__show_help(const char *format, va_list ap) newtRefresh(); backlog = 0; } + pthread_mutex_unlock(&ui__lock); return ret; } diff --git a/tools/perf/util/ui/setup.c b/tools/perf/util/ui/setup.c index fbf1a145492f..ee46d671db59 100644 --- a/tools/perf/util/ui/setup.c +++ b/tools/perf/util/ui/setup.c @@ -6,6 +6,9 @@ #include "../debug.h" #include "browser.h" #include "helpline.h" +#include "ui.h" + +pthread_mutex_t ui__lock = PTHREAD_MUTEX_INITIALIZER; static void newt_suspend(void *d __used) { diff --git a/tools/perf/util/ui/ui.h b/tools/perf/util/ui/ui.h new file mode 100644 index 000000000000..d264e059c829 --- /dev/null +++ b/tools/perf/util/ui/ui.h @@ -0,0 +1,8 @@ +#ifndef _PERF_UI_H_ +#define _PERF_UI_H_ 1 + +#include + +extern pthread_mutex_t ui__lock; + +#endif /* _PERF_UI_H_ */ -- cgit v1.2.3 From 289c082044643e55f65c6a16bb3621cf3f35a454 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 9 Feb 2011 13:56:28 -0200 Subject: perf annotate: Check if offset is less than symbol size Just like done on symbol__inc_addr_samples to catch misparsed offsets from objdump. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/annotate.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 02976b895f27..70ec422ddb64 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -541,11 +541,12 @@ void symbol__annotate_decay_histogram(struct symbol *sym, int evidx) struct annotation *notes = symbol__annotation(sym); struct sym_hist *h = annotation__histogram(notes, evidx); struct objdump_line *pos; + int len = sym->end - sym->start; h->sum = 0; list_for_each_entry(pos, ¬es->src->source, node) { - if (pos->offset != -1) { + if (pos->offset != -1 && pos->offset < len) { h->addr[pos->offset] = h->addr[pos->offset] * 7 / 8; h->sum += h->addr[pos->offset]; } -- cgit v1.2.3 From b99976e2d277c963138e090ae17bf835f8a07680 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 9 Feb 2011 13:59:14 -0200 Subject: perf annotate browser: Use the percent color for the whole line Not just for the percentage number, to see the hot lines more easily. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/browsers/annotate.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/browsers/annotate.c b/tools/perf/util/ui/browsers/annotate.c index 1aa39658539c..cfb5a27f15d2 100644 --- a/tools/perf/util/ui/browsers/annotate.c +++ b/tools/perf/util/ui/browsers/annotate.c @@ -44,8 +44,6 @@ static void annotate_browser__write(struct ui_browser *self, void *entry, int ro struct objdump_line_rb_node *olrb = objdump_line__rb(ol); ui_browser__set_percent_color(self, olrb->percent, current_entry); slsmg_printf(" %7.2f ", olrb->percent); - if (!current_entry) - ui_browser__set_color(self, HE_COLORSET_CODE); } else { ui_browser__set_percent_color(self, 0, current_entry); slsmg_write_nstring(" ", 9); @@ -57,6 +55,9 @@ static void annotate_browser__write(struct ui_browser *self, void *entry, int ro slsmg_write_nstring(" ", width - 18); else slsmg_write_nstring(ol->line, width - 18); + + if (!current_entry) + ui_browser__set_color(self, HE_COLORSET_CODE); } static double objdump_line__calc_percent(struct objdump_line *self, -- cgit v1.2.3 From e116dfa1c357da49f55e1555767ec991225a8321 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 10 Feb 2011 18:08:10 +0900 Subject: perf probe: Support function@filename syntax for --line Since "perf probe --add" supports function@filename syntax, --line option should also support it. Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Franck Bui-Huu Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: linux-kernel@vger.kernel.org LKML-Reference: <20110210090810.1809.26913.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Documentation/perf-probe.txt | 7 ++++--- tools/perf/util/probe-event.c | 15 ++++++++++++--- 2 files changed, 16 insertions(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/Documentation/perf-probe.txt b/tools/perf/Documentation/perf-probe.txt index 81c3220e04f3..02bafce4b341 100644 --- a/tools/perf/Documentation/perf-probe.txt +++ b/tools/perf/Documentation/perf-probe.txt @@ -16,7 +16,7 @@ or or 'perf probe' --list or -'perf probe' [options] --line='FUNC[:RLN[+NUM|:RLN2]]|SRC:ALN[+NUM|:ALN2]' +'perf probe' [options] --line='LINE' or 'perf probe' [options] --vars='PROBEPOINT' @@ -128,13 +128,14 @@ LINE SYNTAX ----------- Line range is described by following syntax. - "FUNC[:RLN[+NUM|-RLN2]]|SRC[:ALN[+NUM|-ALN2]]" + "FUNC[@SRC][:RLN[+NUM|-RLN2]]|SRC[:ALN[+NUM|-ALN2]]" FUNC specifies the function name of showing lines. 'RLN' is the start line number from function entry line, and 'RLN2' is the end line number. As same as probe syntax, 'SRC' means the source file path, 'ALN' is start line number, and 'ALN2' is end line number in the file. It is also possible to specify how -many lines to show by using 'NUM'. +many lines to show by using 'NUM'. Moreover, 'FUNC@SRC' combination is good +for searching a specific function when several functions share same name. So, "source.c:100-120" shows lines between 100th to l20th in source.c file. And "func:10+20" shows 20 lines from 10th line of func function. LAZY MATCHING diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 9d237e3cff5d..cbd76507a56f 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -595,11 +595,11 @@ static int parse_line_num(char **ptr, int *val, const char *what) * The line range syntax is described by: * * SRC[:SLN[+NUM|-ELN]] - * FNC[:SLN[+NUM|-ELN]] + * FNC[@SRC][:SLN[+NUM|-ELN]] */ int parse_line_range_desc(const char *arg, struct line_range *lr) { - char *range, *name = strdup(arg); + char *range, *file, *name = strdup(arg); int err; if (!name) @@ -649,7 +649,16 @@ int parse_line_range_desc(const char *arg, struct line_range *lr) } } - if (strchr(name, '.')) + file = strchr(name, '@'); + if (file) { + *file = '\0'; + lr->file = strdup(++file); + if (lr->file == NULL) { + err = -ENOMEM; + goto err; + } + lr->function = name; + } else if (strchr(name, '.')) lr->file = name; else lr->function = name; -- cgit v1.2.3 From 8737ebdea02315eaffaebb3b73d55f2f726a4fe0 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 10 Feb 2011 18:08:16 +0900 Subject: perf probe: Show filename which contains target function Show filename which contains a target function with the function name on "--lines" mode, because perf-probe just shows the first function even if there are many same-name functions. Originally adopted by Franck Bui-Huu's patch which shows file name instead of function name. I've just modified it to show both of function name and file name, because of completeness of output. E.g.) $ perf probe -L t_show 0 static int t_show(struct seq_file *m, void *v) 1 { 2 struct ftrace_iterator *iter = m->private; ... $ perf probe -L t_show@trace/trace.c 0 static int t_show(struct seq_file *m, void *v) 1 { struct tracer *t = v; ... Original-patch-by: Franck Bui-Huu Cc: 2nddept-manager@sdl.hitachi.co.jp Cc: Franck Bui-Huu Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra LKML-Reference: <20110210090816.1809.43426.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-event.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index cbd76507a56f..0e3ea1321103 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -409,7 +409,7 @@ int show_line_range(struct line_range *lr, const char *module) setup_pager(); if (lr->function) - fprintf(stdout, "<%s:%d>\n", lr->function, + fprintf(stdout, "<%s@%s:%d>\n", lr->function, lr->path, lr->start - lr->offset); else fprintf(stdout, "<%s:%d>\n", lr->path, lr->start); -- cgit v1.2.3 From 4498062e72fd55b2a9a4ac1b44fab8cb44ad5367 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 17 Feb 2011 10:07:42 -0200 Subject: perf python: Add cgroup.c to setup.py to get it building again The 023695d cset added a new file, util/cgroup.c, that is referenced from util/evsel.c, so it needs to be present in util/setup.py so that the python shared object binding works, fixing this: [root@emilia linux]# export PYTHONPATH=~acme/git/build/perf/python/ [root@emilia linux]# ./tools/perf/python/twatch.py Traceback (most recent call last): File "./tools/perf/python/twatch.py", line 16, in import perf ImportError: /home/acme/git/build/perf/python/perf.so: undefined symbol: close_cgroup Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/setup.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/setup.py b/tools/perf/util/setup.py index 1947b0430c94..e24ffadb20b2 100644 --- a/tools/perf/util/setup.py +++ b/tools/perf/util/setup.py @@ -5,7 +5,7 @@ from distutils.core import setup, Extension perf = Extension('perf', sources = ['util/python.c', 'util/ctype.c', 'util/evlist.c', 'util/evsel.c', 'util/cpumap.c', 'util/thread_map.c', - 'util/util.c', 'util/xyarray.c'], + 'util/util.c', 'util/xyarray.c', 'util/cgroup.c'], include_dirs = ['util/include'], extra_compile_args = ['-fno-strict-aliasing', '-Wno-write-strings']) -- cgit v1.2.3 From f0c55bcf4aa41b4b1dbee826513b1acb01bf65e1 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Wed, 16 Feb 2011 15:10:01 +0200 Subject: perf: make perf stat print user provided full event names This patch changes the way perf stat prints event names at the end of a run. Until now, it was trying to reconstruct the event name from its encoding. The problem is that it would only print generic events without their modifiers (u, k, pp). This patch saves the event name as passed by the user in the evsel struct and uses it to print the final event name. This would also work in case perf is linked with a library (such as libpfm4) which provides full PMU event tables. $ perf stat -e cycles:u,cycles:k date Wed Feb 16 14:58:52 CET 2011 Performance counter stats for 'date': 568600 cycles:u 2779715 cycles:k 0.001908182 seconds time elapsed Cc: Arun Sharma Cc: David S. Miller Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Robert Richter Cc: Stephane Eranian LPU-Reference: <4d5bdc64.98a1df0a.7aa3.06c2@mx.google.com> Signed-off-by: Stephane Eranian [ committer note: Fixed a merge problem with 023695d "Add cgroup support" ] Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evsel.c | 1 + tools/perf/util/evsel.h | 7 +++++++ tools/perf/util/parse-events.c | 10 ++++++++++ 3 files changed, 18 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index c974e08d07ab..63cadaf3e208 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -86,6 +86,7 @@ void perf_evsel__delete(struct perf_evsel *evsel) { perf_evsel__exit(evsel); close_cgroup(evsel->cgrp); + free(evsel->name); free(evsel); } diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 1d3d5a3dbe60..f6fc8f651a25 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -37,6 +37,12 @@ struct perf_sample_id { struct perf_evsel *evsel; }; +/** struct perf_evsel - event selector + * + * @name - Can be set to retain the original event name passed by the user, + * so that when showing results in tools such as 'perf stat', we + * show the name used, not some alias. + */ struct perf_evsel { struct list_head node; struct perf_event_attr attr; @@ -45,6 +51,7 @@ struct perf_evsel { struct xyarray *id; struct perf_counts *counts; int idx; + char *name; void *priv; struct cgroup_sel *cgrp; }; diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index cf082daa43e3..80a3dd5ef573 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -268,6 +268,9 @@ const char *event_name(struct perf_evsel *evsel) u64 config = evsel->attr.config; int type = evsel->attr.type; + if (evsel->name) + return evsel->name; + return __event_name(type, config); } @@ -782,8 +785,10 @@ int parse_events(const struct option *opt, const char *str, int unset __used) struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; struct perf_event_attr attr; enum event_result ret; + const char *ostr; for (;;) { + ostr = str; memset(&attr, 0, sizeof(attr)); ret = parse_event_symbols(opt, &str, &attr); if (ret == EVT_FAILED) @@ -798,6 +803,11 @@ int parse_events(const struct option *opt, const char *str, int unset __used) if (evsel == NULL) return -1; perf_evlist__add(evlist, evsel); + + evsel->name = calloc(str - ostr + 1, 1); + if (!evsel->name) + return -1; + strncpy(evsel->name, ostr, str - ostr); } if (*str == 0) -- cgit v1.2.3 From fec9cbd15b9e99bab9bc50f1ed7e20a1087d7c6d Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 17 Feb 2011 10:37:23 -0200 Subject: perf hists: Print number of samples, not the period sum So that we match the header where we state the number of events with the "Samples" column when using 'perf report -n/--show-nr-samples': [root@emilia ~]# perf record -a sleep 1 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.111 MB perf.data (~4860 samples) ] [root@emilia ~]# perf report --stdio --show-nr-samples # Events: 11 cycles # # Overhead Samples Command Shared Object Symbol # ........ .......... ........... .................. ............................ # 16.65% 1 sleep [kernel.kallsyms] [k] unmap_vmas 16.10% 1 perf libpthread-2.12.so [.] __pthread_cleanup_push_defer 15.79% 2 perf [kernel.kallsyms] [k] format_decode 12.88% 1 kworker/1:2 [kernel.kallsyms] [k] cache_reap 10.69% 1 swapper [kernel.kallsyms] [k] _raw_spin_lock 7.55% 1 sleep [kernel.kallsyms] [k] prepare_exec_creds 6.00% 1 perf [jbd2] [k] start_this_handle 5.29% 1 perf [kernel.kallsyms] [k] seq_read 4.75% 1 perf [kernel.kallsyms] [k] get_pid_task 4.30% 1 perf [kernel.kallsyms] [k] _raw_spin_unlock_irqrestore # # (For a higher level overview, try: perf report --sort comm,dso) # [root@emilia ~]# Reported-by: Stephane Eranian Acked-by: Stephane Eranian Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/hist.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 3f437236f193..da2899e8c6f8 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -591,6 +591,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size, { struct sort_entry *se; u64 period, total, period_sys, period_us, period_guest_sys, period_guest_us; + u64 nr_events; const char *sep = symbol_conf.field_sep; int ret; @@ -599,6 +600,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size, if (pair_hists) { period = self->pair ? self->pair->period : 0; + nr_events = self->pair ? self->pair->nr_events : 0; total = pair_hists->stats.total_period; period_sys = self->pair ? self->pair->period_sys : 0; period_us = self->pair ? self->pair->period_us : 0; @@ -606,6 +608,7 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size, period_guest_us = self->pair ? self->pair->period_guest_us : 0; } else { period = self->period; + nr_events = self->nr_events; total = session_total; period_sys = self->period_sys; period_us = self->period_us; @@ -646,9 +649,9 @@ int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size, if (symbol_conf.show_nr_samples) { if (sep) - ret += snprintf(s + ret, size - ret, "%c%" PRIu64, *sep, period); + ret += snprintf(s + ret, size - ret, "%c%" PRIu64, *sep, nr_events); else - ret += snprintf(s + ret, size - ret, "%11" PRIu64, period); + ret += snprintf(s + ret, size - ret, "%11" PRIu64, nr_events); } if (pair_hists) { -- cgit v1.2.3 From 668b8788f497b2386402daeca583d6300240d41d Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 17 Feb 2011 15:38:58 -0200 Subject: perf list: Allow filtering list of events The man page has the details, here are some examples: [root@emilia ~]# perf list *fault* *:*wait* List of pre-defined events (to be used in -e): page-faults OR faults [Software event] minor-faults [Software event] major-faults [Software event] alignment-faults [Software event] emulation-faults [Software event] radeon:radeon_fence_wait_begin [Tracepoint event] radeon:radeon_fence_wait_end [Tracepoint event] writeback:wbc_writeback_wait [Tracepoint event] writeback:wbc_balance_dirty_wait [Tracepoint event] writeback:writeback_congestion_wait [Tracepoint event] writeback:writeback_wait_iff_congested [Tracepoint event] sched:sched_wait_task [Tracepoint event] sched:sched_process_wait [Tracepoint event] sched:sched_stat_wait [Tracepoint event] sched:sched_stat_iowait [Tracepoint event] syscalls:sys_enter_epoll_wait [Tracepoint event] syscalls:sys_exit_epoll_wait [Tracepoint event] syscalls:sys_enter_epoll_pwait [Tracepoint event] syscalls:sys_exit_epoll_pwait [Tracepoint event] syscalls:sys_enter_rt_sigtimedwait [Tracepoint event] syscalls:sys_exit_rt_sigtimedwait [Tracepoint event] syscalls:sys_enter_waitid [Tracepoint event] syscalls:sys_exit_waitid [Tracepoint event] syscalls:sys_enter_wait4 [Tracepoint event] syscalls:sys_exit_wait4 [Tracepoint event] syscalls:sys_enter_waitpid [Tracepoint event] syscalls:sys_exit_waitpid [Tracepoint event] [root@emilia ~]# Suggested-by: Ingo Molnar Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Documentation/perf-list.txt | 23 ++++++++- tools/perf/builtin-list.c | 43 +++++++++++++++- tools/perf/util/parse-events.c | 94 +++++++++++++++++++++++++++------- tools/perf/util/parse-events.h | 5 +- 4 files changed, 142 insertions(+), 23 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/Documentation/perf-list.txt b/tools/perf/Documentation/perf-list.txt index 399751befeed..7a527f7e9da9 100644 --- a/tools/perf/Documentation/perf-list.txt +++ b/tools/perf/Documentation/perf-list.txt @@ -8,7 +8,7 @@ perf-list - List all symbolic event types SYNOPSIS -------- [verse] -'perf list' +'perf list' [hw|sw|cache|tracepoint|event_glob] DESCRIPTION ----------- @@ -63,7 +63,26 @@ details. Some of them are referenced in the SEE ALSO section below. OPTIONS ------- -None + +Without options all known events will be listed. + +To limit the list use: + +. 'hw' or 'hardware' to list hardware events such as cache-misses, etc. + +. 'sw' or 'software' to list software events such as context switches, etc. + +. 'cache' or 'hwcache' to list hardware cache events such as L1-dcache-loads, etc. + +. 'tracepoint' to list all tracepoint events, alternatively use + 'subsys_glob:event_glob' to filter by tracepoint subsystems such as sched, + block, etc. + +. If none of the above is matched, it will apply the supplied glob to all + events, printing the ones that match. + +One or more types can be used at the same time, listing the events for the +types specified. SEE ALSO -------- diff --git a/tools/perf/builtin-list.c b/tools/perf/builtin-list.c index d88c6961274c..6313b6eb3ebb 100644 --- a/tools/perf/builtin-list.c +++ b/tools/perf/builtin-list.c @@ -5,6 +5,7 @@ * * Copyright (C) 2009, Thomas Gleixner * Copyright (C) 2008-2009, Red Hat Inc, Ingo Molnar + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo */ #include "builtin.h" @@ -13,9 +14,47 @@ #include "util/parse-events.h" #include "util/cache.h" -int cmd_list(int argc __used, const char **argv __used, const char *prefix __used) +int cmd_list(int argc, const char **argv, const char *prefix __used) { setup_pager(); - print_events(); + + if (argc == 1) + print_events(NULL); + else { + int i; + + for (i = 1; i < argc; ++i) { + if (i > 1) + putchar('\n'); + if (strncmp(argv[i], "tracepoint", 10) == 0) + print_tracepoint_events(NULL, NULL); + else if (strcmp(argv[i], "hw") == 0 || + strcmp(argv[i], "hardware") == 0) + print_events_type(PERF_TYPE_HARDWARE); + else if (strcmp(argv[i], "sw") == 0 || + strcmp(argv[i], "software") == 0) + print_events_type(PERF_TYPE_SOFTWARE); + else if (strcmp(argv[i], "cache") == 0 || + strcmp(argv[i], "hwcache") == 0) + print_hwcache_events(NULL); + else { + char *sep = strchr(argv[i], ':'), *s; + int sep_idx; + + if (sep == NULL) { + print_events(argv[i]); + continue; + } + sep_idx = sep - argv[i]; + s = strdup(argv[i]); + if (s == NULL) + return -1; + + s[sep_idx] = '\0'; + print_tracepoint_events(s, s + sep_idx + 1); + free(s); + } + } + } return 0; } diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 80a3dd5ef573..54a7e2634d58 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -858,7 +858,7 @@ static const char * const event_type_descriptors[] = { * Print the events from /tracing/events */ -static void print_tracepoint_events(void) +void print_tracepoint_events(const char *subsys_glob, const char *event_glob) { DIR *sys_dir, *evt_dir; struct dirent *sys_next, *evt_next, sys_dirent, evt_dirent; @@ -873,6 +873,9 @@ static void print_tracepoint_events(void) return; for_each_subsystem(sys_dir, sys_dirent, sys_next) { + if (subsys_glob != NULL && + !strglobmatch(sys_dirent.d_name, subsys_glob)) + continue; snprintf(dir_path, MAXPATHLEN, "%s/%s", debugfs_path, sys_dirent.d_name); @@ -881,6 +884,10 @@ static void print_tracepoint_events(void) continue; for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next) { + if (event_glob != NULL && + !strglobmatch(evt_dirent.d_name, event_glob)) + continue; + snprintf(evt_path, MAXPATHLEN, "%s:%s", sys_dirent.d_name, evt_dirent.d_name); printf(" %-42s [%s]\n", evt_path, @@ -932,13 +939,61 @@ int is_valid_tracepoint(const char *event_string) return 0; } +void print_events_type(u8 type) +{ + struct event_symbol *syms = event_symbols; + unsigned int i; + char name[64]; + + for (i = 0; i < ARRAY_SIZE(event_symbols); i++, syms++) { + if (type != syms->type) + continue; + + if (strlen(syms->alias)) + snprintf(name, sizeof(name), "%s OR %s", + syms->symbol, syms->alias); + else + snprintf(name, sizeof(name), "%s", syms->symbol); + + printf(" %-42s [%s]\n", name, + event_type_descriptors[type]); + } +} + +int print_hwcache_events(const char *event_glob) +{ + unsigned int type, op, i, printed = 0; + + for (type = 0; type < PERF_COUNT_HW_CACHE_MAX; type++) { + for (op = 0; op < PERF_COUNT_HW_CACHE_OP_MAX; op++) { + /* skip invalid cache type */ + if (!is_cache_op_valid(type, op)) + continue; + + 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)) + continue; + + printf(" %-42s [%s]\n", name, + event_type_descriptors[PERF_TYPE_HW_CACHE]); + ++printed; + } + } + } + + return printed; +} + /* * Print the help text for the event symbols: */ -void print_events(void) +void print_events(const char *event_glob) { struct event_symbol *syms = event_symbols; - unsigned int i, type, op, prev_type = -1; + unsigned int i, type, prev_type = -1, printed = 0, ntypes_printed = 0; char name[40]; printf("\n"); @@ -947,8 +1002,16 @@ void print_events(void) for (i = 0; i < ARRAY_SIZE(event_symbols); i++, syms++) { type = syms->type; - if (type != prev_type) + if (type != prev_type && printed) { printf("\n"); + printed = 0; + ntypes_printed++; + } + + if (event_glob != NULL && + !(strglobmatch(syms->symbol, event_glob) || + (syms->alias && strglobmatch(syms->alias, event_glob)))) + continue; if (strlen(syms->alias)) sprintf(name, "%s OR %s", syms->symbol, syms->alias); @@ -958,22 +1021,17 @@ void print_events(void) event_type_descriptors[type]); prev_type = type; + ++printed; } - printf("\n"); - for (type = 0; type < PERF_COUNT_HW_CACHE_MAX; type++) { - for (op = 0; op < PERF_COUNT_HW_CACHE_OP_MAX; op++) { - /* skip invalid cache type */ - if (!is_cache_op_valid(type, op)) - continue; - - for (i = 0; i < PERF_COUNT_HW_CACHE_RESULT_MAX; i++) { - printf(" %-42s [%s]\n", - event_cache_name(type, op, i), - event_type_descriptors[PERF_TYPE_HW_CACHE]); - } - } + if (ntypes_printed) { + printed = 0; + printf("\n"); } + print_hwcache_events(event_glob); + + if (event_glob != NULL) + return; printf("\n"); printf(" %-42s [%s]\n", @@ -986,7 +1044,7 @@ void print_events(void) event_type_descriptors[PERF_TYPE_BREAKPOINT]); printf("\n"); - print_tracepoint_events(); + print_tracepoint_events(NULL, NULL); exit(129); } diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h index cf7e94abb676..212f88e07a9c 100644 --- a/tools/perf/util/parse-events.h +++ b/tools/perf/util/parse-events.h @@ -28,7 +28,10 @@ extern int parse_filter(const struct option *opt, const char *str, int unset); #define EVENTS_HELP_MAX (128*1024) -extern void print_events(void); +void print_events(const char *event_glob); +void print_events_type(u8 type); +void print_tracepoint_events(const char *subsys_glob, const char *event_glob); +int print_hwcache_events(const char *event_glob); extern int is_valid_tracepoint(const char *event_string); extern char debugfs_path[]; -- cgit v1.2.3 From 8796cb9d7dc028945af4b2ea858ae8f8f2ecbe8c Mon Sep 17 00:00:00 2001 From: Michael Witten Date: Wed, 2 Feb 2011 11:57:41 -0600 Subject: perf tools: Makefile: Remove platform-specific cruft While it makes sense that this tool could be used on other platforms at least to parse data, there doesn't appear to be any real support for such usage. This commit squashes several commits that remove: SNPRINTF_RETURNS_BOGUS FREAD_READS_DIRECTORIES NO_D_{INO,TYPE}_IN_DIRENT NO_STRCASESTR NO_MEMMEM NO_STRTOUMAX and NO_STRTOULL NO_SETENV NO_UNSETENV NO_MKDTEMP NEEDS_LIBICONV NEEDS_SOCKET NO_MMAP NO_PTHREADS NO_PREAD NO_TRUSTABLE_FILEMODE NO_IPV6 and NO_SOCKADDR_STORAGE NO_ICONV and OLD_ICONV NO_NSEC, USE_NSEC, and USE_ST_TIMESPEC NO_ST_BLOCKS_IN_STRUCT_STAT NO_FINK and NO_DARWIN_PORTS NO_SYS_SELECT_H NO_HSTRERROR DIR_HAS_BSD_GROUP_SEMANTICS and FORCE_DIR_SET_GID NEEDS_NSL, NO_UINTMAX_T, NO_INET_{N,P}TON COMPAT_{CFLAGS,OBJS} Executable extension `X' Signed-off-by: Michael Witten LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/Makefile | 232 ++----------------------------------------------- tools/perf/util/util.h | 26 ------ 2 files changed, 8 insertions(+), 250 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 544367cf6d58..53c1e93ac432 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -14,103 +14,23 @@ endif # Define V=1 to have a more verbose compile. # Define V=2 to have an even more verbose compile. # -# Define SNPRINTF_RETURNS_BOGUS if your are on a system which snprintf() -# or vsnprintf() return -1 instead of number of characters which would -# have been written to the final string if enough space had been available. -# -# Define FREAD_READS_DIRECTORIES if your are on a system which succeeds -# when attempting to read from an fopen'ed directory. -# # Define CURLDIR=/foo/bar if your curl header and library files are in # /foo/bar/include and /foo/bar/lib directories. # # Define EXPATDIR=/foo/bar if your expat header and library files are in # /foo/bar/include and /foo/bar/lib directories. # -# Define NO_D_INO_IN_DIRENT if you don't have d_ino in your struct dirent. -# -# Define NO_D_TYPE_IN_DIRENT if your platform defines DT_UNKNOWN but lacks -# d_type in struct dirent (latest Cygwin -- will be fixed soonish). -# # Define NO_C99_FORMAT if your formatted IO functions (printf/scanf et.al.) # do not support the 'size specifiers' introduced by C99, namely ll, hh, # j, z, t. (representing long long int, char, intmax_t, size_t, ptrdiff_t). # some C compilers supported these specifiers prior to C99 as an extension. # -# Define NO_STRCASESTR if you don't have strcasestr. -# -# Define NO_MEMMEM if you don't have memmem. -# -# Define NO_STRTOUMAX if you don't have strtoumax in the C library. -# If your compiler also does not support long long or does not have -# strtoull, define NO_STRTOULL. -# -# Define NO_SETENV if you don't have setenv in the C library. -# -# Define NO_UNSETENV if you don't have unsetenv in the C library. -# -# Define NO_MKDTEMP if you don't have mkdtemp in the C library. -# -# Define NO_SYS_SELECT_H if you don't have sys/select.h. -# -# Define NO_FINK if you are building on Darwin/Mac OS X, have Fink -# installed in /sw, but don't want PERF to link against any libraries -# installed there. If defined you may specify your own (or Fink's) -# include directories and library directories by defining CFLAGS -# and LDFLAGS appropriately. -# -# Define NO_DARWIN_PORTS if you are building on Darwin/Mac OS X, -# have DarwinPorts installed in /opt/local, but don't want PERF to -# link against any libraries installed there. If defined you may -# specify your own (or DarwinPort's) include directories and -# library directories by defining CFLAGS and LDFLAGS appropriately. -# -# Define NEEDS_LIBICONV if linking with libc is not enough (Darwin). -# -# Define NEEDS_SOCKET if linking with libc is not enough (SunOS, -# Patrick Mauritz). -# -# Define NO_MMAP if you want to avoid mmap. -# -# Define NO_PTHREADS if you do not have or do not want to use Pthreads. -# -# Define NO_PREAD if you have a problem with pread() system call (e.g. -# cygwin.dll before v1.5.22). -# -# Define NO_TRUSTABLE_FILEMODE if your filesystem may claim to support -# the executable mode bit, but doesn't really do so. -# -# Define NO_IPV6 if you lack IPv6 support and getaddrinfo(). -# -# Define NO_SOCKADDR_STORAGE if your platform does not have struct -# sockaddr_storage. -# -# Define NO_ICONV if your libc does not properly support iconv. -# -# Define OLD_ICONV if your library has an old iconv(), where the second -# (input buffer pointer) parameter is declared with type (const char **). -# # Define NO_DEFLATE_BOUND if your zlib does not have deflateBound. # # Define NO_R_TO_GCC_LINKER if your gcc does not like "-R/path/lib" # that tells runtime paths to dynamic libraries; # "-Wl,-rpath=/path/lib" is used instead. # -# Define USE_NSEC below if you want perf to care about sub-second file mtimes -# and ctimes. Note that you need recent glibc (at least 2.2.4) for this, and -# it will BREAK YOUR LOCAL DIFFS! show-diff and anything using it will likely -# randomly break unless your underlying filesystem supports those sub-second -# times (my ext3 doesn't). -# -# Define USE_ST_TIMESPEC if your "struct stat" uses "st_ctimespec" instead of -# "st_ctim" -# -# Define NO_NSEC if your "struct stat" does not have "st_ctim.tv_nsec" -# available. This automatically turns USE_NSEC off. -# -# Define NO_ST_BLOCKS_IN_STRUCT_STAT if your platform does not have st_blocks -# field that counts the on-disk footprint in 512-byte blocks. -# # Define ASCIIDOC8 if you want to format documentation with AsciiDoc 8 # # Define DOCBOOK_XSL_172 if you want to format man pages with DocBook XSL v1.72. @@ -282,8 +202,6 @@ BASIC_LDFLAGS = # Guard against environment variables BUILTIN_OBJS = BUILT_INS = -COMPAT_CFLAGS = -COMPAT_OBJS = LIB_H = LIB_OBJS = PYRF_OBJS = @@ -329,7 +247,7 @@ LANG_BINDINGS = ALL_PROGRAMS = $(PROGRAMS) $(SCRIPTS) # what 'all' will build but not install in perfexecdir -OTHER_PROGRAMS = $(OUTPUT)perf$X +OTHER_PROGRAMS = $(OUTPUT)perf # Set paths to tools early so that they can be used for version tests. ifndef SHELL_PATH @@ -538,22 +456,6 @@ endif # NO_DWARF -include arch/$(ARCH)/Makefile -ifeq ($(uname_S),Darwin) - ifndef NO_FINK - ifeq ($(shell test -d /sw/lib && echo y),y) - BASIC_CFLAGS += -I/sw/include - BASIC_LDFLAGS += -L/sw/lib - endif - endif - ifndef NO_DARWIN_PORTS - ifeq ($(shell test -d /opt/local/lib && echo y),y) - BASIC_CFLAGS += -I/opt/local/include - BASIC_LDFLAGS += -L/opt/local/lib - endif - endif - PTHREAD_LIBS = -endif - ifneq ($(OUTPUT),) BASIC_CFLAGS += -I$(OUTPUT) endif @@ -707,110 +609,9 @@ ifndef CC_LD_DYNPATH endif endif -ifdef NEEDS_SOCKET - EXTLIBS += -lsocket -endif -ifdef NEEDS_NSL - EXTLIBS += -lnsl -endif -ifdef NO_D_TYPE_IN_DIRENT - BASIC_CFLAGS += -DNO_D_TYPE_IN_DIRENT -endif -ifdef NO_D_INO_IN_DIRENT - BASIC_CFLAGS += -DNO_D_INO_IN_DIRENT -endif -ifdef NO_ST_BLOCKS_IN_STRUCT_STAT - BASIC_CFLAGS += -DNO_ST_BLOCKS_IN_STRUCT_STAT -endif -ifdef USE_NSEC - BASIC_CFLAGS += -DUSE_NSEC -endif -ifdef USE_ST_TIMESPEC - BASIC_CFLAGS += -DUSE_ST_TIMESPEC -endif -ifdef NO_NSEC - BASIC_CFLAGS += -DNO_NSEC -endif ifdef NO_C99_FORMAT BASIC_CFLAGS += -DNO_C99_FORMAT endif -ifdef SNPRINTF_RETURNS_BOGUS - COMPAT_CFLAGS += -DSNPRINTF_RETURNS_BOGUS - COMPAT_OBJS += $(OUTPUT)compat/snprintf.o -endif -ifdef FREAD_READS_DIRECTORIES - COMPAT_CFLAGS += -DFREAD_READS_DIRECTORIES - COMPAT_OBJS += $(OUTPUT)compat/fopen.o -endif -ifdef NO_STRCASESTR - COMPAT_CFLAGS += -DNO_STRCASESTR - COMPAT_OBJS += $(OUTPUT)compat/strcasestr.o -endif -ifdef NO_STRTOUMAX - COMPAT_CFLAGS += -DNO_STRTOUMAX - COMPAT_OBJS += $(OUTPUT)compat/strtoumax.o -endif -ifdef NO_STRTOULL - COMPAT_CFLAGS += -DNO_STRTOULL -endif -ifdef NO_SETENV - COMPAT_CFLAGS += -DNO_SETENV - COMPAT_OBJS += $(OUTPUT)compat/setenv.o -endif -ifdef NO_MKDTEMP - COMPAT_CFLAGS += -DNO_MKDTEMP - COMPAT_OBJS += $(OUTPUT)compat/mkdtemp.o -endif -ifdef NO_UNSETENV - COMPAT_CFLAGS += -DNO_UNSETENV - COMPAT_OBJS += $(OUTPUT)compat/unsetenv.o -endif -ifdef NO_SYS_SELECT_H - BASIC_CFLAGS += -DNO_SYS_SELECT_H -endif -ifdef NO_MMAP - COMPAT_CFLAGS += -DNO_MMAP - COMPAT_OBJS += $(OUTPUT)compat/mmap.o -else - ifdef USE_WIN32_MMAP - COMPAT_CFLAGS += -DUSE_WIN32_MMAP - COMPAT_OBJS += $(OUTPUT)compat/win32mmap.o - endif -endif -ifdef NO_PREAD - COMPAT_CFLAGS += -DNO_PREAD - COMPAT_OBJS += $(OUTPUT)compat/pread.o -endif -ifdef NO_TRUSTABLE_FILEMODE - BASIC_CFLAGS += -DNO_TRUSTABLE_FILEMODE -endif -ifdef NO_IPV6 - BASIC_CFLAGS += -DNO_IPV6 -endif -ifdef NO_UINTMAX_T - BASIC_CFLAGS += -Duintmax_t=uint32_t -endif -ifdef NO_SOCKADDR_STORAGE -ifdef NO_IPV6 - BASIC_CFLAGS += -Dsockaddr_storage=sockaddr_in -else - BASIC_CFLAGS += -Dsockaddr_storage=sockaddr_in6 -endif -endif -ifdef NO_INET_NTOP - LIB_OBJS += $(OUTPUT)compat/inet_ntop.o -endif -ifdef NO_INET_PTON - LIB_OBJS += $(OUTPUT)compat/inet_pton.o -endif - -ifdef NO_ICONV - BASIC_CFLAGS += -DNO_ICONV -endif - -ifdef OLD_ICONV - BASIC_CFLAGS += -DOLD_ICONV -endif ifdef NO_DEFLATE_BOUND BASIC_CFLAGS += -DNO_DEFLATE_BOUND @@ -819,14 +620,6 @@ endif ifdef NO_PERL_MAKEMAKER export NO_PERL_MAKEMAKER endif -ifdef NO_HSTRERROR - COMPAT_CFLAGS += -DNO_HSTRERROR - COMPAT_OBJS += $(OUTPUT)compat/hstrerror.o -endif -ifdef NO_MEMMEM - COMPAT_CFLAGS += -DNO_MEMMEM - COMPAT_OBJS += $(OUTPUT)compat/memmem.o -endif ifdef INTERNAL_QSORT COMPAT_CFLAGS += -DINTERNAL_QSORT COMPAT_OBJS += $(OUTPUT)compat/qsort.o @@ -835,9 +628,6 @@ ifdef RUNTIME_PREFIX COMPAT_CFLAGS += -DRUNTIME_PREFIX endif -ifdef DIR_HAS_BSD_GROUP_SEMANTICS - COMPAT_CFLAGS += -DDIR_HAS_BSD_GROUP_SEMANTICS -endif ifdef NO_EXTERNAL_GREP BASIC_CFLAGS += -DNO_EXTERNAL_GREP endif @@ -895,9 +685,6 @@ PERL_PATH_SQ = $(subst ','\'',$(PERL_PATH)) LIBS = -Wl,--whole-archive $(PERFLIBS) -Wl,--no-whole-archive $(EXTLIBS) -BASIC_CFLAGS += $(COMPAT_CFLAGS) -LIB_OBJS += $(COMPAT_OBJS) - ALL_CFLAGS += $(BASIC_CFLAGS) ALL_CFLAGS += $(ARCH_CFLAGS) ALL_LDFLAGS += $(BASIC_LDFLAGS) @@ -910,9 +697,6 @@ export TAR INSTALL DESTDIR SHELL_PATH SHELL = $(SHELL_PATH) all:: shell_compatibility_test $(ALL_PROGRAMS) $(LANG_BINDINGS) $(BUILT_INS) $(OTHER_PROGRAMS) $(OUTPUT)PERF-BUILD-OPTIONS -ifneq (,$X) - $(foreach p,$(patsubst %$X,%,$(filter %$X,$(ALL_PROGRAMS) $(BUILT_INS) perf$X)), test '$p' -ef '$p$X' || $(RM) '$p';) -endif all:: @@ -921,15 +705,15 @@ please_set_SHELL_PATH_to_a_more_modern_shell: shell_compatibility_test: please_set_SHELL_PATH_to_a_more_modern_shell -strip: $(PROGRAMS) $(OUTPUT)perf$X - $(STRIP) $(STRIP_OPTS) $(PROGRAMS) $(OUTPUT)perf$X +strip: $(PROGRAMS) $(OUTPUT)perf + $(STRIP) $(STRIP_OPTS) $(PROGRAMS) $(OUTPUT)perf $(OUTPUT)perf.o: perf.c $(OUTPUT)common-cmds.h $(OUTPUT)PERF-CFLAGS $(QUIET_CC)$(CC) -DPERF_VERSION='"$(PERF_VERSION)"' \ '-DPERF_HTML_PATH="$(htmldir_SQ)"' \ $(ALL_CFLAGS) -c $(filter %.c,$^) -o $@ -$(OUTPUT)perf$X: $(OUTPUT)perf.o $(BUILTIN_OBJS) $(PERFLIBS) +$(OUTPUT)perf: $(OUTPUT)perf.o $(BUILTIN_OBJS) $(PERFLIBS) $(QUIET_LINK)$(CC) $(ALL_CFLAGS) $(ALL_LDFLAGS) $(OUTPUT)perf.o \ $(BUILTIN_OBJS) $(LIBS) -o $@ @@ -1027,11 +811,11 @@ $(OUTPUT)util/scripting-engines/trace-event-python.o: util/scripting-engines/tra $(OUTPUT)scripts/python/Perf-Trace-Util/Context.o: scripts/python/Perf-Trace-Util/Context.c $(OUTPUT)PERF-CFLAGS $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) $(PYTHON_EMBED_CCOPTS) -Wno-redundant-decls -Wno-strict-prototypes -Wno-unused-parameter -Wno-nested-externs $< -$(OUTPUT)perf-%$X: %.o $(PERFLIBS) +$(OUTPUT)perf-%: %.o $(PERFLIBS) $(QUIET_LINK)$(CC) $(ALL_CFLAGS) -o $@ $(ALL_LDFLAGS) $(filter %.o,$^) $(LIBS) $(LIB_OBJS) $(BUILTIN_OBJS): $(LIB_H) -$(patsubst perf-%$X,%.o,$(PROGRAMS)): $(LIB_H) $(wildcard */*.h) +$(patsubst perf-%,%.o,$(PROGRAMS)): $(LIB_H) $(wildcard */*.h) # we compile into subdirectories. if the target directory is not the source directory, they might not exists. So # we depend the various files onto their directories. @@ -1168,7 +952,7 @@ export perfexec_instdir install: all $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(bindir_SQ)' - $(INSTALL) $(OUTPUT)perf$X '$(DESTDIR_SQ)$(bindir_SQ)' + $(INSTALL) $(OUTPUT)perf '$(DESTDIR_SQ)$(bindir_SQ)' $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/perl/Perf-Trace-Util/lib/Perf/Trace' $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/perl/bin' $(INSTALL) $(OUTPUT)perf-archive -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)' @@ -1267,7 +1051,7 @@ distclean: clean clean: $(RM) $(OUTPUT){*.o,*/*.o,*/*/*.o,*/*/*/*.o,$(LIB_FILE),perf-archive} - $(RM) $(ALL_PROGRAMS) $(BUILT_INS) perf$X + $(RM) $(ALL_PROGRAMS) $(BUILT_INS) perf $(RM) $(TEST_PROGRAMS) $(RM) *.spec *.pyc *.pyo */*.pyc */*.pyo $(OUTPUT)common-cmds.h TAGS tags cscope* $(RM) -r $(PERF_TARNAME) .doc-tmp-dir diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h index e833f26f3bfc..fc784284ac8b 100644 --- a/tools/perf/util/util.h +++ b/tools/perf/util/util.h @@ -70,9 +70,7 @@ #include #include #include -#ifndef NO_SYS_SELECT_H #include -#endif #include #include #include @@ -83,10 +81,6 @@ #include "types.h" #include -#ifndef NO_ICONV -#include -#endif - extern const char *graph_line; extern const char *graph_dotted_line; extern char buildid_dir[]; @@ -236,26 +230,6 @@ static inline int sane_case(int x, int high) return x; } -#ifndef DIR_HAS_BSD_GROUP_SEMANTICS -# define FORCE_DIR_SET_GID S_ISGID -#else -# define FORCE_DIR_SET_GID 0 -#endif - -#ifdef NO_NSEC -#undef USE_NSEC -#define ST_CTIME_NSEC(st) 0 -#define ST_MTIME_NSEC(st) 0 -#else -#ifdef USE_ST_TIMESPEC -#define ST_CTIME_NSEC(st) ((unsigned int)((st).st_ctimespec.tv_nsec)) -#define ST_MTIME_NSEC(st) ((unsigned int)((st).st_mtimespec.tv_nsec)) -#else -#define ST_CTIME_NSEC(st) ((unsigned int)((st).st_ctim.tv_nsec)) -#define ST_MTIME_NSEC(st) ((unsigned int)((st).st_mtim.tv_nsec)) -#endif -#endif - int mkdir_p(char *path, mode_t mode); int copyfile(const char *from, const char *to); -- cgit v1.2.3 From a3d1ee10d1bf4520af3d44c1aa6cd46956ec4fd7 Mon Sep 17 00:00:00 2001 From: Michael Witten Date: Wed, 2 Feb 2011 14:22:08 -0600 Subject: perf tools: Makefile: Remove various and sundry cruft This commit squashes several commits that remove: unnecessary uname calls `sh -c' BUILT_INS and QUIET_BUILT_IN They have no effect, and the `fixup-builtins' and `check-builtins.sh' scripts don't even exist. RUNTIME_PREFIX It's currently never anything but unset, and it's apparently only meaningful when Microsoft Windows is the operating system (according to the source for git). TEST_PROGRAMS EXTRA_PROGRAMS unused SHELL_PATH_SQ portions unused test for V=2 useless exports Only when `V' is undefined (that is, only when the value of `V' is empty) is `export V' performed, which just has the effect of placing the empty-valued variable `V' in the environment. The only other script to make use of `V' is `Documentation/Makefile', which only checks whether `V' is undefined (that is, whether the value of `V' is empty); hence, the `export V' has no effect whatsoever. Similarly, `export QUIET_GEN' is useless because it will only have a non-empty value when `V' has an empty-value, and when `V' has an empty-value, `QUIET_GEN' is always explicitly set in every script in which it is used. `DESTDIR' is only ever defined by the user via the environment or the command line, both of which are automatically exported to sub-make processes. Furthermore, no non-make sub-scripts make use of `DESTDIR' as an environment variable. No other scripts use `perfexec_instdir'. unused QUIET_SUBDIR{0,1} TAR and RPMBUILD PTHREAD_LIBS Maintainer's dist rules and commands distclean target Test suite coverage testing PRINT_DIR and NO_SUBDIR `configure' target NO_CURL @@PERF_VERSION@@ substitution Without the sed command, all of the rule's commands can be reduced to a single line that copies a file and sets the permissions properly in the process. `make test' echo line template_instdir PERF-BUILD-OPTIONS double-colon rules The use of double-colon rules seems misguided or vestigial git. Essentially hard-coded $(SCRIPTS) expansion Signed-off-by: Michael Witten LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/.gitignore | 1 - tools/perf/Makefile | 213 +++------------------------------------------ tools/perf/util/exec_cmd.c | 19 ---- 3 files changed, 10 insertions(+), 223 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/.gitignore b/tools/perf/.gitignore index cb43289e447f..416684be0ad3 100644 --- a/tools/perf/.gitignore +++ b/tools/perf/.gitignore @@ -1,4 +1,3 @@ -PERF-BUILD-OPTIONS PERF-CFLAGS PERF-GUI-VARS PERF-VERSION-FILE diff --git a/tools/perf/Makefile b/tools/perf/Makefile index fde196fe9366..9b8421805c5c 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -3,7 +3,7 @@ ifeq ("$(origin O)", "command line") endif # The default target of this Makefile is... -all:: +all: ifneq ($(OUTPUT),) # check that the output directory actually exists @@ -11,8 +11,7 @@ OUTDIR := $(shell cd $(OUTPUT) && /bin/pwd) $(if $(OUTDIR),, $(error output directory "$(OUTPUT)" does not exist)) endif -# Define V=1 to have a more verbose compile. -# Define V=2 to have an even more verbose compile. +# Define V to have a more verbose compile. # # Define ASCIIDOC8 if you want to format documentation with AsciiDoc 8 # @@ -28,12 +27,7 @@ $(OUTPUT)PERF-VERSION-FILE: .FORCE-PERF-VERSION-FILE @$(SHELL_PATH) util/PERF-VERSION-GEN $(OUTPUT) -include $(OUTPUT)PERF-VERSION-FILE -uname_S := $(shell sh -c 'uname -s 2>/dev/null || echo not') -uname_M := $(shell sh -c 'uname -m 2>/dev/null || echo not') -uname_O := $(shell sh -c 'uname -o 2>/dev/null || echo not') -uname_R := $(shell sh -c 'uname -r 2>/dev/null || echo not') -uname_P := $(shell sh -c 'uname -p 2>/dev/null || echo not') -uname_V := $(shell sh -c 'uname -v 2>/dev/null || echo not') +uname_M := $(shell uname -m 2>/dev/null || echo not) ARCH ?= $(shell echo $(uname_M) | sed -e s/i.86/i386/ -e s/sun4u/sparc64/ \ -e s/arm.*/arm/ -e s/sa110/arm/ \ @@ -52,8 +46,6 @@ ifeq ($(ARCH),x86_64) ARCH_INCLUDE = ../../arch/x86/lib/memcpy_64.S endif -# CFLAGS and LDFLAGS are for the users to override from the command line. - # # Include saner warnings here, which can catch bugs: # @@ -131,22 +123,13 @@ CC = $(CROSS_COMPILE)gcc AR = $(CROSS_COMPILE)ar RM = rm -f MKDIR = mkdir -TAR = tar FIND = find INSTALL = install -RPMBUILD = rpmbuild -PTHREAD_LIBS = -lpthread # sparse is architecture-neutral, which means that we need to tell it # explicitly what architecture to check for. Fix this up for yours.. SPARSE_FLAGS = -D__BIG_ENDIAN__ -D__powerpc__ -ifeq ($(V), 2) - QUIET_STDERR = ">/dev/null" -else - QUIET_STDERR = ">/dev/null 2>&1" -endif - -include feature-tests.mak ifeq ($(call try-cc,$(SOURCE_HELLO),-Werror -fstack-protector-all),y) @@ -171,12 +154,10 @@ BASIC_LDFLAGS = # Guard against environment variables BUILTIN_OBJS = -BUILT_INS = LIB_H = LIB_OBJS = PYRF_OBJS = SCRIPT_SH = -TEST_PROGRAMS = SCRIPT_SH += perf-archive.sh @@ -192,12 +173,6 @@ $(OUTPUT)python/perf.so: $(PYRF_OBJS) SCRIPTS = $(patsubst %.sh,%,$(SCRIPT_SH)) -# Empty... -EXTRA_PROGRAMS = - -# ... and all the rest that could be moved out of bindir to perfexecdir -PROGRAMS += $(EXTRA_PROGRAMS) - # # Single 'perf' binary right now: # @@ -205,10 +180,6 @@ PROGRAMS += $(OUTPUT)perf LANG_BINDINGS = -# List built-in command $C whose implementation cmd_$C() is not in -# builtin-$C.o but is linked in as part of some other command. -# - # what 'all' will build and 'install' will install, in perfexecdir ALL_PROGRAMS = $(PROGRAMS) $(SCRIPTS) @@ -565,33 +536,13 @@ else endif endif -ifdef RUNTIME_PREFIX - COMPAT_CFLAGS += -DRUNTIME_PREFIX -endif - -QUIET_SUBDIR0 = +$(MAKE) -C # space to separate -C and subdir -QUIET_SUBDIR1 = - -ifneq ($(findstring $(MAKEFLAGS),w),w) -PRINT_DIR = --no-print-directory -else # "make -w" -NO_SUBDIR = : -endif - ifneq ($(findstring $(MAKEFLAGS),s),s) ifndef V QUIET_CC = @echo ' ' CC $@; QUIET_AR = @echo ' ' AR $@; QUIET_LINK = @echo ' ' LINK $@; QUIET_MKDIR = @echo ' ' MKDIR $@; - QUIET_BUILT_IN = @echo ' ' BUILTIN $@; QUIET_GEN = @echo ' ' GEN $@; - QUIET_SUBDIR0 = +@subdir= - QUIET_SUBDIR1 = ;$(NO_SUBDIR) echo ' ' SUBDIR $$subdir; \ - $(MAKE) $(PRINT_DIR) -C $$subdir - export V - export QUIET_GEN - export QUIET_BUILT_IN endif endif @@ -621,16 +572,14 @@ ALL_CFLAGS += $(BASIC_CFLAGS) ALL_CFLAGS += $(ARCH_CFLAGS) ALL_LDFLAGS += $(BASIC_LDFLAGS) -export TAR INSTALL DESTDIR SHELL_PATH +export INSTALL SHELL_PATH ### Build rules SHELL = $(SHELL_PATH) -all:: shell_compatibility_test $(ALL_PROGRAMS) $(LANG_BINDINGS) $(BUILT_INS) $(OTHER_PROGRAMS) $(OUTPUT)PERF-BUILD-OPTIONS - -all:: +all: shell_compatibility_test $(ALL_PROGRAMS) $(LANG_BINDINGS) $(OTHER_PROGRAMS) please_set_SHELL_PATH_to_a_more_modern_shell: @$$(:) @@ -661,37 +610,17 @@ $(OUTPUT)builtin-timechart.o: builtin-timechart.c $(OUTPUT)common-cmds.h $(OUTPU '-DPERF_MAN_PATH="$(mandir_SQ)"' \ '-DPERF_INFO_PATH="$(infodir_SQ)"' $< -$(BUILT_INS): $(OUTPUT)perf$X - $(QUIET_BUILT_IN)$(RM) $@ && \ - ln perf$X $@ 2>/dev/null || \ - ln -s perf$X $@ 2>/dev/null || \ - cp perf$X $@ - $(OUTPUT)common-cmds.h: util/generate-cmdlist.sh command-list.txt $(OUTPUT)common-cmds.h: $(wildcard Documentation/perf-*.txt) $(QUIET_GEN). util/generate-cmdlist.sh > $@+ && mv $@+ $@ -$(patsubst %.sh,%,$(SCRIPT_SH)) : % : %.sh - $(QUIET_GEN)$(RM) $(OUTPUT)$@ $(OUTPUT)$@+ && \ - sed -e '1s|#!.*/sh|#!$(SHELL_PATH_SQ)|' \ - -e 's|@SHELL_PATH@|$(SHELL_PATH_SQ)|' \ - -e 's/@@PERF_VERSION@@/$(PERF_VERSION)/g' \ - -e 's/@@NO_CURL@@/$(NO_CURL)/g' \ - $@.sh > $(OUTPUT)$@+ && \ - chmod +x $(OUTPUT)$@+ && \ - mv $(OUTPUT)$@+ $(OUTPUT)$@ - -configure: configure.ac - $(QUIET_GEN)$(RM) $@ $<+ && \ - sed -e 's/@@PERF_VERSION@@/$(PERF_VERSION)/g' \ - $< > $<+ && \ - autoconf -o $@ $<+ && \ - $(RM) $<+ +$(SCRIPTS) : % : %.sh + $(QUIET_GEN)$(INSTALL) '$@.sh' '$(OUTPUT)$@' # These can record PERF_VERSION $(OUTPUT)perf.o perf.spec \ - $(patsubst %.sh,%,$(SCRIPT_SH)) \ + $(SCRIPTS) \ : $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)%.o: %.c $(OUTPUT)PERF-CFLAGS @@ -826,23 +755,8 @@ $(OUTPUT)PERF-CFLAGS: .FORCE-PERF-CFLAGS echo "$$FLAGS" >$(OUTPUT)PERF-CFLAGS; \ fi -# We need to apply sq twice, once to protect from the shell -# that runs $(OUTPUT)PERF-BUILD-OPTIONS, and then again to protect it -# and the first level quoting from the shell that runs "echo". -$(OUTPUT)PERF-BUILD-OPTIONS: .FORCE-PERF-BUILD-OPTIONS - @echo SHELL_PATH=\''$(subst ','\'',$(SHELL_PATH_SQ))'\' >$@ - @echo TAR=\''$(subst ','\'',$(subst ','\'',$(TAR)))'\' >>$@ - @echo NO_CURL=\''$(subst ','\'',$(subst ','\'',$(NO_CURL)))'\' >>$@ - ### Testing rules -# -# None right now: -# -# TEST_PROGRAMS += test-something$X - -all:: $(TEST_PROGRAMS) - # GNU make supports exporting all variables by "export" without parameters. # However, the environment gets quite big, and some programs have problems # with that. @@ -855,29 +769,17 @@ check: $(OUTPUT)common-cmds.h sparse $(ALL_CFLAGS) $(SPARSE_FLAGS) $$i || exit; \ done; \ else \ - echo 2>&1 "Did you mean 'make test'?"; \ exit 1; \ fi -remove-dashes: - ./fixup-builtins $(BUILT_INS) $(PROGRAMS) $(SCRIPTS) - ### Installation rules -ifneq ($(filter /%,$(firstword $(template_dir))),) -template_instdir = $(template_dir) -else -template_instdir = $(prefix)/$(template_dir) -endif -export template_instdir - ifneq ($(filter /%,$(firstword $(perfexecdir))),) perfexec_instdir = $(perfexecdir) else perfexec_instdir = $(prefix)/$(perfexecdir) endif perfexec_instdir_SQ = $(subst ','\'',$(perfexec_instdir)) -export perfexec_instdir install: all $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(bindir_SQ)' @@ -894,14 +796,6 @@ install: all $(INSTALL) scripts/python/*.py -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/python' $(INSTALL) scripts/python/bin/* -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/python/bin' -ifdef BUILT_INS - $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)' - $(INSTALL) $(BUILT_INS) '$(DESTDIR_SQ)$(perfexec_instdir_SQ)' -ifneq (,$X) - $(foreach p,$(patsubst %$X,%,$(filter %$X,$(ALL_PROGRAMS) $(BUILT_INS) $(OUTPUT)perf$X)), $(RM) '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/$p';) -endif -endif - install-doc: $(MAKE) -C Documentation install @@ -926,104 +820,17 @@ quick-install-man: quick-install-html: $(MAKE) -C Documentation quick-install-html - -### Maintainer's dist rules -# -# None right now -# -# -# perf.spec: perf.spec.in -# sed -e 's/@@VERSION@@/$(PERF_VERSION)/g' < $< > $@+ -# mv $@+ $@ -# -# PERF_TARNAME=perf-$(PERF_VERSION) -# dist: perf.spec perf-archive$(X) configure -# ./perf-archive --format=tar \ -# --prefix=$(PERF_TARNAME)/ HEAD^{tree} > $(PERF_TARNAME).tar -# @mkdir -p $(PERF_TARNAME) -# @cp perf.spec configure $(PERF_TARNAME) -# @echo $(PERF_VERSION) > $(PERF_TARNAME)/version -# $(TAR) rf $(PERF_TARNAME).tar \ -# $(PERF_TARNAME)/perf.spec \ -# $(PERF_TARNAME)/configure \ -# $(PERF_TARNAME)/version -# @$(RM) -r $(PERF_TARNAME) -# gzip -f -9 $(PERF_TARNAME).tar -# -# htmldocs = perf-htmldocs-$(PERF_VERSION) -# manpages = perf-manpages-$(PERF_VERSION) -# dist-doc: -# $(RM) -r .doc-tmp-dir -# mkdir .doc-tmp-dir -# $(MAKE) -C Documentation WEBDOC_DEST=../.doc-tmp-dir install-webdoc -# cd .doc-tmp-dir && $(TAR) cf ../$(htmldocs).tar . -# gzip -n -9 -f $(htmldocs).tar -# : -# $(RM) -r .doc-tmp-dir -# mkdir -p .doc-tmp-dir/man1 .doc-tmp-dir/man5 .doc-tmp-dir/man7 -# $(MAKE) -C Documentation DESTDIR=./ \ -# man1dir=../.doc-tmp-dir/man1 \ -# man5dir=../.doc-tmp-dir/man5 \ -# man7dir=../.doc-tmp-dir/man7 \ -# install -# cd .doc-tmp-dir && $(TAR) cf ../$(manpages).tar . -# gzip -n -9 -f $(manpages).tar -# $(RM) -r .doc-tmp-dir -# -# rpm: dist -# $(RPMBUILD) -ta $(PERF_TARNAME).tar.gz - ### Cleaning rules -distclean: clean -# $(RM) configure - clean: $(RM) $(OUTPUT){*.o,*/*.o,*/*/*.o,*/*/*/*.o,$(LIB_FILE),perf-archive} - $(RM) $(ALL_PROGRAMS) $(BUILT_INS) perf - $(RM) $(TEST_PROGRAMS) + $(RM) $(ALL_PROGRAMS) perf $(RM) *.spec *.pyc *.pyo */*.pyc */*.pyo $(OUTPUT)common-cmds.h TAGS tags cscope* - $(RM) -r $(PERF_TARNAME) .doc-tmp-dir - $(RM) $(PERF_TARNAME).tar.gz perf-core_$(PERF_VERSION)-*.tar.gz - $(RM) $(htmldocs).tar.gz $(manpages).tar.gz $(MAKE) -C Documentation/ clean - $(RM) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)PERF-CFLAGS $(OUTPUT)PERF-BUILD-OPTIONS + $(RM) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)PERF-CFLAGS @python util/setup.py clean --build-lib='$(OUTPUT)python' \ --build-temp='$(OUTPUT)python/temp' .PHONY: all install clean strip .PHONY: shell_compatibility_test please_set_SHELL_PATH_to_a_more_modern_shell .PHONY: .FORCE-PERF-VERSION-FILE TAGS tags cscope .FORCE-PERF-CFLAGS -.PHONY: .FORCE-PERF-BUILD-OPTIONS - -### Make sure built-ins do not have dups and listed in perf.c -# -check-builtins:: - ./check-builtins.sh - -### Test suite coverage testing -# -# None right now -# -# .PHONY: coverage coverage-clean coverage-build coverage-report -# -# coverage: -# $(MAKE) coverage-build -# $(MAKE) coverage-report -# -# coverage-clean: -# rm -f *.gcda *.gcno -# -# COVERAGE_CFLAGS = $(CFLAGS) -O0 -ftest-coverage -fprofile-arcs -# COVERAGE_LDFLAGS = $(CFLAGS) -O0 -lgcov -# -# coverage-build: coverage-clean -# $(MAKE) CFLAGS="$(COVERAGE_CFLAGS)" LDFLAGS="$(COVERAGE_LDFLAGS)" all -# $(MAKE) CFLAGS="$(COVERAGE_CFLAGS)" LDFLAGS="$(COVERAGE_LDFLAGS)" \ -# -j1 test -# -# coverage-report: -# gcov -b *.c */*.c -# grep '^function.*called 0 ' *.c.gcov */*.c.gcov \ -# | sed -e 's/\([^:]*\)\.gcov: *function \([^ ]*\) called.*/\1: \2/' \ -# | tee coverage-untested-functions diff --git a/tools/perf/util/exec_cmd.c b/tools/perf/util/exec_cmd.c index 67eeff571568..7adf4ad15d8f 100644 --- a/tools/perf/util/exec_cmd.c +++ b/tools/perf/util/exec_cmd.c @@ -11,31 +11,12 @@ static const char *argv0_path; const char *system_path(const char *path) { -#ifdef RUNTIME_PREFIX - static const char *prefix; -#else static const char *prefix = PREFIX; -#endif struct strbuf d = STRBUF_INIT; if (is_absolute_path(path)) return path; -#ifdef RUNTIME_PREFIX - assert(argv0_path); - assert(is_absolute_path(argv0_path)); - - if (!prefix && - !(prefix = strip_path_suffix(argv0_path, PERF_EXEC_PATH)) && - !(prefix = strip_path_suffix(argv0_path, BINDIR)) && - !(prefix = strip_path_suffix(argv0_path, "perf"))) { - prefix = PREFIX; - fprintf(stderr, "RUNTIME_PREFIX requested, " - "but prefix computation failed. " - "Using static fallback '%s'.\n", prefix); - } -#endif - strbuf_addf(&d, "%s/%s", prefix, path); path = strbuf_detach(&d, NULL); return path; -- cgit v1.2.3 From fbee632d0ca9f4073a3fefb9a843eac8af036b0f Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 21 Feb 2011 13:23:57 -0300 Subject: perf probe: Fix error propagation leading to segfault There are two hunks in this patch that stops probe processing as soon as one error is found, breaking out of loops, the other fix an error propagation that should return a negative error number but instead was returning the result of "ret < 0", which is 1 and thus made several error checks fail because they test agains < 0. The problem could be triggered by asking for a variable that was optimized out, fact that should stop the whole probe processing but instead was segfaulting while installing broken probes: [root@emilia ~]# probe perf_mmap:55 user_lock_limit Failed to find the location of user_lock_limit at this address. Perhaps, it has been optimized out. Failed to find 'user_lock_limit' in this function. Add new events: probe:perf_mmap (on perf_mmap:55 with user_lock_limit) probe:perf_mmap_1 (on perf_mmap:55 with user_lock_limit) Segmentation fault (core dumped) [root@emilia ~]# perf probe -l probe:perf_mmap (on perf_mmap:55@git/linux/kernel/perf_event.c with user_lock_limit) probe:perf_mmap_1 (on perf_mmap:55@git/linux/kernel/perf_event.c with user_lock_limit) [root@emilia ~]# After the fix: [root@emilia ~]# probe perf_mmap:55 user_lock_limit Failed to find the location of user_lock_limit at this address. Perhaps, it has been optimized out. Failed to find 'user_lock_limit' in this function. Error: Failed to add events. (-2) [root@emilia ~]# Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Masami Hiramatsu Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-event.c | 5 ++++- tools/perf/util/probe-finder.c | 4 +++- 2 files changed, 7 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 0e3ea1321103..369ddc64bbb6 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -1832,9 +1832,12 @@ int add_perf_probe_events(struct perf_probe_event *pevs, int npevs, } /* Loop 2: add all events */ - for (i = 0; i < npevs && ret >= 0; i++) + for (i = 0; i < npevs && ret >= 0; i++) { ret = __add_probe_trace_events(pkgs[i].pev, pkgs[i].tevs, pkgs[i].ntevs, force_add); + if (ret < 0) + break; + } end: /* Loop 3: cleanup and free trace events */ for (i = 0; i < npevs; i++) { diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index fe461f6559f1..eecbdca5c0d5 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -1262,7 +1262,7 @@ static int probe_point_line_walker(const char *fname, int lineno, ret = call_probe_finder(NULL, pf); /* Continue if no error, because the line will be in inline function */ - return ret < 0 ?: 0; + return ret < 0 ? ret : 0; } /* Find probe point from its line number */ @@ -1484,6 +1484,8 @@ static int find_probes(int fd, struct probe_finder *pf) pf->lno = pp->line; ret = find_probe_point_by_line(pf); } + if (ret != DWARF_CB_OK) + break; } off = noff; } -- cgit v1.2.3 From e603dc15072c7fec0ae263597e6dabc3bb4c5c5b Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 21 Feb 2011 16:05:50 -0300 Subject: perf evsel: Fix inverted test for fixing up attr.inherit flag The kernel refuses mmapping an event with the inherit flag set for something that is systemwide (cpu == -1), and the evsel layer got this reversed at some point, fix it. The symtom was that the --pid and --tid parameters for 'perf record' and 'perf top' returned with -EINVAL, like: # /tmp/build-perf/perf record -v -fo/tmp/perf.data -p 1042 Warning: ... trying to fall back to cpu-clock-ticks Fatal: failed to mmap with 22 (Invalid argument) Reported-by: David Ahern Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evsel.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 63cadaf3e208..8083d5126fca 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -179,8 +179,19 @@ static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, for (cpu = 0; cpu < cpus->nr; cpu++) { int group_fd = -1; - - evsel->attr.inherit = (cpus->map[cpu] < 0) && inherit; + /* + * Don't allow mmap() of inherited per-task counters. This + * would create a performance issue due to all children writing + * to the same buffer. + * + * FIXME: + * Proper fix is not to pass 'inherit' to perf_evsel__open*, + * but a 'flags' parameter, with 'group' folded there as well, + * then introduce a PERF_O_{MMAP,GROUP,INHERIT} enum, and if + * O_MMAP is set, emit a warning if cpu < 0 and O_INHERIT is + * set. Lets go for the minimal fix first tho. + */ + evsel->attr.inherit = (cpus->map[cpu] >= 0) && inherit; for (thread = 0; thread < threads->nr; thread++) { -- cgit v1.2.3 From 8635bf6ea3402154eec64763e6ed14972013c1c1 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 22 Feb 2011 06:56:18 -0300 Subject: perf probe: Remove redundant checks While fixing an error propagating problem in f809b25 I added two redundant checks. I did that because I didn't expect the checks to be on the while and for loop condition expression, where they are tested before we run the loop, where the 'ret' variable is set. So remove it from there and leave it just after it is actually set, eliminating unneded tests. Reported-by: Masami Hiramatsu Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Masami Hiramatsu Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-event.c | 2 +- tools/perf/util/probe-finder.c | 5 ++--- 2 files changed, 3 insertions(+), 4 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 369ddc64bbb6..5ddee66020a7 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -1832,7 +1832,7 @@ int add_perf_probe_events(struct perf_probe_event *pevs, int npevs, } /* Loop 2: add all events */ - for (i = 0; i < npevs && ret >= 0; i++) { + for (i = 0; i < npevs; i++) { ret = __add_probe_trace_events(pkgs[i].pev, pkgs[i].tevs, pkgs[i].ntevs, force_add); if (ret < 0) diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index eecbdca5c0d5..17f9c4a66ddd 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -1462,8 +1462,7 @@ static int find_probes(int fd, struct probe_finder *pf) off = 0; line_list__init(&pf->lcache); /* Loop on CUs (Compilation Unit) */ - while (!dwarf_nextcu(dbg, off, &noff, &cuhl, NULL, NULL, NULL) && - ret >= 0) { + while (!dwarf_nextcu(dbg, off, &noff, &cuhl, NULL, NULL, NULL)) { /* Get the DIE(Debugging Information Entry) of this CU */ diep = dwarf_offdie(dbg, off + cuhl, &pf->cu_die); if (!diep) @@ -1484,7 +1483,7 @@ static int find_probes(int fd, struct probe_finder *pf) pf->lno = pp->line; ret = find_probe_point_by_line(pf); } - if (ret != DWARF_CB_OK) + if (ret < 0) break; } off = noff; -- cgit v1.2.3 From c97cf42219b7b6037d2f96c27a5f114f2383f828 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 22 Feb 2011 12:02:07 -0300 Subject: perf top: Live TUI Annotation Now one has just to press the right key, 'a' or Enter on the main 'perf top --tui' screen to live annotate the symbol under the cursor. The annotate window starts centered on the hottest line (the one with most samples so far) then TAB and shift+TAB can be used to go to the prev/next hot line. Pressing 'H' at any point will center again the screen on the hottest line. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-top.c | 36 +++++----- tools/perf/util/annotate.h | 6 +- tools/perf/util/top.h | 1 + tools/perf/util/ui/browsers/annotate.c | 126 +++++++++++++++++++++++---------- tools/perf/util/ui/browsers/top.c | 45 +++++++++++- 5 files changed, 156 insertions(+), 58 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index c9fd66d4a082..f88a2630e1fc 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -92,7 +92,6 @@ static bool dump_symtab = false; static struct winsize winsize; static const char *sym_filter = NULL; -struct sym_entry *sym_filter_entry = NULL; struct sym_entry *sym_filter_entry_sched = NULL; static int sym_pcnt_filter = 5; @@ -168,18 +167,19 @@ static int parse_source(struct sym_entry *syme) pthread_mutex_lock(¬es->lock); if (symbol__alloc_hist(sym, top.evlist->nr_entries) < 0) { + pthread_mutex_unlock(¬es->lock); pr_err("Not enough memory for annotating '%s' symbol!\n", sym->name); sleep(1); - goto out_unlock; + return err; } err = symbol__annotate(sym, syme->map, 0); if (err == 0) { out_assign: - sym_filter_entry = syme; + top.sym_filter_entry = syme; } -out_unlock: + pthread_mutex_unlock(¬es->lock); return err; } @@ -195,7 +195,7 @@ static void record_precise_ip(struct sym_entry *syme, int counter, u64 ip) struct annotation *notes; struct symbol *sym; - if (syme != sym_filter_entry) + if (syme != top.sym_filter_entry) return; sym = sym_entry__symbol(syme); @@ -275,8 +275,8 @@ static void print_sym_table(struct perf_session *session) session->hists.stats.total_lost); } - if (sym_filter_entry) { - show_details(sym_filter_entry); + if (top.sym_filter_entry) { + show_details(top.sym_filter_entry); return; } @@ -417,8 +417,8 @@ static void print_mapped_keys(void) { char *name = NULL; - if (sym_filter_entry) { - struct symbol *sym = sym_entry__symbol(sym_filter_entry); + if (top.sym_filter_entry) { + struct symbol *sym = sym_entry__symbol(top.sym_filter_entry); name = sym->name; } @@ -549,15 +549,15 @@ static void handle_keypress(struct perf_session *session, int c) perf_session__fprintf_dsos(session, stderr); exit(0); case 's': - prompt_symbol(&sym_filter_entry, "Enter details symbol"); + prompt_symbol(&top.sym_filter_entry, "Enter details symbol"); break; case 'S': - if (!sym_filter_entry) + if (!top.sym_filter_entry) break; else { - struct sym_entry *syme = sym_filter_entry; + struct sym_entry *syme = top.sym_filter_entry; - sym_filter_entry = NULL; + top.sym_filter_entry = NULL; __zero_source_counters(syme); } break; @@ -656,7 +656,7 @@ static int symbol_filter(struct map *map, struct symbol *sym) syme->map = map; symbol__annotate_init(map, sym); - if (!sym_filter_entry && sym_filter && !strcmp(name, sym_filter)) { + if (!top.sym_filter_entry && sym_filter && !strcmp(name, sym_filter)) { /* schedule initial sym_filter_entry setup */ sym_filter_entry_sched = syme; sym_filter = NULL; @@ -750,13 +750,13 @@ static void perf_event__process_sample(const union perf_event *event, /* let's see, whether we need to install initial sym_filter_entry */ if (sym_filter_entry_sched) { - sym_filter_entry = sym_filter_entry_sched; + top.sym_filter_entry = sym_filter_entry_sched; sym_filter_entry_sched = NULL; - if (parse_source(sym_filter_entry) < 0) { - struct symbol *sym = sym_entry__symbol(sym_filter_entry); + if (parse_source(top.sym_filter_entry) < 0) { + struct symbol *sym = sym_entry__symbol(top.sym_filter_entry); pr_err("Can't annotate %s", sym->name); - if (sym_filter_entry->map->dso->origin == DSO__ORIG_KERNEL) { + if (top.sym_filter_entry->map->dso->origin == DSO__ORIG_KERNEL) { pr_err(": No vmlinux file was found in the path:\n"); machine__fprintf_vmlinux_path(machine, stderr); } else diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index e848803fcd48..c2c286896801 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -90,12 +90,14 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, #ifdef NO_NEWT_SUPPORT static inline int symbol__tui_annotate(struct symbol *sym __used, - struct map *map __used, int evidx __used) + struct map *map __used, + int evidx __used, int refresh __used) { return 0; } #else -int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx); +int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx, + int refresh); #endif #endif /* __PERF_ANNOTATE_H */ diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index 4f769f47e19a..e8d28e2ecdba 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -44,6 +44,7 @@ struct perf_top { pid_t target_pid, target_tid; bool hide_kernel_symbols, hide_user_symbols, zero; const char *cpu_list; + struct sym_entry *sym_filter_entry; struct perf_evsel *sym_evsel; }; diff --git a/tools/perf/util/ui/browsers/annotate.c b/tools/perf/util/ui/browsers/annotate.c index cfb5a27f15d2..8c17a8730e4a 100644 --- a/tools/perf/util/ui/browsers/annotate.c +++ b/tools/perf/util/ui/browsers/annotate.c @@ -6,6 +6,7 @@ #include "../../sort.h" #include "../../symbol.h" #include "../../annotate.h" +#include static void ui__error_window(const char *fmt, ...) { @@ -138,46 +139,108 @@ static void annotate_browser__set_top(struct annotate_browser *self, self->curr_hot = nd; } -static int annotate_browser__run(struct annotate_browser *self) +static void annotate_browser__calc_percent(struct annotate_browser *browser, + int evidx) { - struct rb_node *nd; + struct symbol *sym = browser->b.priv; + struct annotation *notes = symbol__annotation(sym); + struct objdump_line *pos; + + browser->entries = RB_ROOT; + + pthread_mutex_lock(¬es->lock); + + list_for_each_entry(pos, ¬es->src->source, node) { + struct objdump_line_rb_node *rbpos = objdump_line__rb(pos); + rbpos->percent = objdump_line__calc_percent(pos, sym, evidx); + if (rbpos->percent < 0.01) { + RB_CLEAR_NODE(&rbpos->rb_node); + continue; + } + objdump__insert_line(&browser->entries, rbpos); + } + pthread_mutex_unlock(¬es->lock); + + browser->curr_hot = rb_last(&browser->entries); +} + +static int annotate_browser__run(struct annotate_browser *self, int evidx, + int refresh) +{ + struct rb_node *nd = NULL; struct symbol *sym = self->b.priv; + /* + * RIGHT To allow builtin-annotate to cycle thru multiple symbols by + * examining the exit key for this function. + */ + int exit_keys[] = { 'H', NEWT_KEY_TAB, NEWT_KEY_UNTAB, + NEWT_KEY_RIGHT, 0 }; int key; if (ui_browser__show(&self->b, sym->name, - "<-, -> or ESC: exit, TAB/shift+TAB: cycle thru samples") < 0) + "<-, -> or ESC: exit, TAB/shift+TAB: " + "cycle hottest lines, H: Hottest") < 0) return -1; - /* - * To allow builtin-annotate to cycle thru multiple symbols by - * examining the exit key for this function. - */ - ui_browser__add_exit_key(&self->b, NEWT_KEY_RIGHT); + + ui_browser__add_exit_keys(&self->b, exit_keys); + annotate_browser__calc_percent(self, evidx); + + if (self->curr_hot) + annotate_browser__set_top(self, self->curr_hot); nd = self->curr_hot; - if (nd) { - int tabs[] = { NEWT_KEY_TAB, NEWT_KEY_UNTAB, 0 }; - ui_browser__add_exit_keys(&self->b, tabs); - } + + if (refresh != 0) + newtFormSetTimer(self->b.form, refresh); while (1) { key = ui_browser__run(&self->b); + if (refresh != 0) { + annotate_browser__calc_percent(self, evidx); + /* + * Current line focus got out of the list of most active + * lines, NULL it so that if TAB|UNTAB is pressed, we + * move to curr_hot (current hottest line). + */ + if (nd != NULL && RB_EMPTY_NODE(nd)) + nd = NULL; + } + switch (key) { + case -1: + /* + * FIXME we need to check if it was + * es.reason == NEWT_EXIT_TIMER + */ + if (refresh != 0) + symbol__annotate_decay_histogram(sym, evidx); + continue; case NEWT_KEY_TAB: - nd = rb_prev(nd); - if (nd == NULL) - nd = rb_last(&self->entries); - annotate_browser__set_top(self, nd); + if (nd != NULL) { + nd = rb_prev(nd); + if (nd == NULL) + nd = rb_last(&self->entries); + } else + nd = self->curr_hot; break; case NEWT_KEY_UNTAB: - nd = rb_next(nd); - if (nd == NULL) - nd = rb_first(&self->entries); - annotate_browser__set_top(self, nd); + if (nd != NULL) + nd = rb_next(nd); + if (nd == NULL) + nd = rb_first(&self->entries); + else + nd = self->curr_hot; + break; + case 'H': + nd = self->curr_hot; break; default: goto out; } + + if (nd != NULL) + annotate_browser__set_top(self, nd); } out: ui_browser__hide(&self->b); @@ -186,13 +249,13 @@ out: int hist_entry__tui_annotate(struct hist_entry *he, int evidx) { - return symbol__tui_annotate(he->ms.sym, he->ms.map, evidx); + return symbol__tui_annotate(he->ms.sym, he->ms.map, evidx, 0); } -int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx) +int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx, + int refresh) { struct objdump_line *pos, *n; - struct objdump_line_rb_node *rbpos; struct annotation *notes = symbol__annotation(sym); struct annotate_browser browser = { .b = { @@ -211,7 +274,7 @@ int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx) if (map->dso->annotate_warned) return -1; - if (symbol__annotate(sym, map, sizeof(*rbpos)) < 0) { + if (symbol__annotate(sym, map, sizeof(struct objdump_line_rb_node)) < 0) { ui__error_window(ui_helpline__last_msg); return -1; } @@ -219,26 +282,17 @@ int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx) ui_helpline__push("Press <- or ESC to exit"); list_for_each_entry(pos, ¬es->src->source, node) { + struct objdump_line_rb_node *rbpos; size_t line_len = strlen(pos->line); + if (browser.b.width < line_len) browser.b.width = line_len; rbpos = objdump_line__rb(pos); rbpos->idx = browser.b.nr_entries++; - rbpos->percent = objdump_line__calc_percent(pos, sym, evidx); - if (rbpos->percent < 0.01) - continue; - objdump__insert_line(&browser.entries, rbpos); } - /* - * Position the browser at the hottest line. - */ - browser.curr_hot = rb_last(&browser.entries); - if (browser.curr_hot) - annotate_browser__set_top(&browser, browser.curr_hot); - browser.b.width += 18; /* Percentage */ - ret = annotate_browser__run(&browser); + ret = annotate_browser__run(&browser, evidx, refresh); list_for_each_entry_safe(pos, n, ¬es->src->source, node) { list_del(&pos->node); objdump_line__free(pos); diff --git a/tools/perf/util/ui/browsers/top.c b/tools/perf/util/ui/browsers/top.c index ca6062483a8f..377ff58c9ae9 100644 --- a/tools/perf/util/ui/browsers/top.c +++ b/tools/perf/util/ui/browsers/top.c @@ -7,6 +7,7 @@ * Released under the GPL v2. (and only v2, not any later version) */ #include "../browser.h" +#include "../../annotate.h" #include "../helpline.h" #include "../libslang.h" #include "../../evlist.h" @@ -18,6 +19,7 @@ struct perf_top_browser { struct ui_browser b; struct rb_root root; + struct sym_entry *selection; float sum_ksamples; int dso_width; int dso_short_width; @@ -60,6 +62,9 @@ static void perf_top_browser__write(struct ui_browser *browser, void *entry, int slsmg_write_nstring(width >= syme->map->dso->long_name_len ? syme->map->dso->long_name : syme->map->dso->short_name, width); + + if (current_entry) + top_browser->selection = syme; } static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser) @@ -80,21 +85,52 @@ static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser) browser->b.nr_entries = top->rb_entries; } +static void perf_top_browser__annotate(struct perf_top_browser *browser) +{ + struct sym_entry *syme = browser->selection; + struct symbol *sym = sym_entry__symbol(syme); + struct annotation *notes = symbol__annotation(sym); + struct perf_top *top = browser->b.priv; + + if (notes->src != NULL) + goto do_annotation; + + pthread_mutex_lock(¬es->lock); + + top->sym_filter_entry = NULL; + + if (symbol__alloc_hist(sym, top->evlist->nr_entries) < 0) { + pr_err("Not enough memory for annotating '%s' symbol!\n", + sym->name); + pthread_mutex_unlock(¬es->lock); + return; + } + + top->sym_filter_entry = syme; + + pthread_mutex_unlock(¬es->lock); +do_annotation: + symbol__tui_annotate(sym, syme->map, 0, top->delay_secs * 1000); +} + static int perf_top_browser__run(struct perf_top_browser *browser) { int key; char title[160]; struct perf_top *top = browser->b.priv; int delay_msecs = top->delay_secs * 1000; + int exit_keys[] = { 'a', NEWT_KEY_ENTER, NEWT_KEY_RIGHT, 0, }; perf_top_browser__update_rb_tree(browser); perf_top__header_snprintf(top, title, sizeof(title)); perf_top__reset_sample_counters(top); - if (ui_browser__show(&browser->b, title, "ESC: exit") < 0) + if (ui_browser__show(&browser->b, title, + "ESC: exit, ENTER|->|a: Live Annotate") < 0) return -1; newtFormSetTimer(browser->b.form, delay_msecs); + ui_browser__add_exit_keys(&browser->b, exit_keys); while (1) { key = ui_browser__run(&browser->b); @@ -109,7 +145,12 @@ static int perf_top_browser__run(struct perf_top_browser *browser) SLsmg_gotorc(0, 0); slsmg_write_nstring(title, browser->b.width); break; - case NEWT_KEY_TAB: + case 'a': + case NEWT_KEY_RIGHT: + case NEWT_KEY_ENTER: + if (browser->selection) + perf_top_browser__annotate(browser); + break; default: goto out; } -- cgit v1.2.3 From 6435a5e39d3e01a1a73a925ed53ee18619b0a368 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 23 Feb 2011 07:25:02 -0300 Subject: perf top browser: Adjust the browser indexes when refreshing This is not a problem when we're not at the bottom of the active symbols list, so was not noticed, but at the end of the screen it falls apart. Fix it by adjusting the ui_browser indexes according to the new number of entries in the rb_tree and by seeking from the start of the rb_tree to find the new symbol at the top of the screen. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/browsers/top.c | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/browsers/top.c b/tools/perf/util/ui/browsers/top.c index 377ff58c9ae9..2f47224426b6 100644 --- a/tools/perf/util/ui/browsers/top.c +++ b/tools/perf/util/ui/browsers/top.c @@ -70,6 +70,7 @@ static void perf_top_browser__write(struct ui_browser *browser, void *entry, int static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser) { struct perf_top *top = browser->b.priv; + u64 top_idx = browser->b.top_idx; browser->root = RB_ROOT; browser->b.top = NULL; @@ -82,7 +83,29 @@ static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser) if (browser->sym_width + browser->dso_width > browser->b.width - 29) browser->sym_width = browser->b.width - browser->dso_width - 29; } + + /* + * Adjust the ui_browser indexes since the entries in the browser->root + * rb_tree may have changed, then seek it from start, so that we get a + * possible new top of the screen. + */ browser->b.nr_entries = top->rb_entries; + + if (top_idx >= browser->b.nr_entries) { + if (browser->b.height >= browser->b.nr_entries) + top_idx = browser->b.nr_entries - browser->b.height; + else + top_idx = 0; + } + + if (browser->b.index >= top_idx + browser->b.height) + browser->b.index = top_idx + browser->b.index - browser->b.top_idx; + + if (browser->b.index >= browser->b.nr_entries) + browser->b.index = browser->b.nr_entries - 1; + + browser->b.top_idx = top_idx; + browser->b.seek(&browser->b, top_idx, SEEK_SET); } static void perf_top_browser__annotate(struct perf_top_browser *browser) -- cgit v1.2.3 From 170ae6bc24e1d7f9bd921a484ec9ea2825497970 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 23 Feb 2011 11:08:59 -0300 Subject: perf annotate: Show better message when no vmlinux is found In both --tui and --stdio, in 'annotate', 'top', 'report' when trying to annotate a kernel symbol having just access to a kallsyms file, that doesn't have the DWARF info needed for annotation. Suggested-by: Ingo Molnar Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/annotate.c | 15 +++++++++++++-- tools/perf/util/ui/browsers/hists.c | 8 -------- 2 files changed, 13 insertions(+), 10 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 70ec422ddb64..0d0830c98cd7 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -295,12 +295,23 @@ fallback: } if (dso->origin == DSO__ORIG_KERNEL) { + char bf[BUILD_ID_SIZE * 2 + 16] = " with build id "; + char *build_id_msg = NULL; + if (dso->annotate_warned) goto out_free_filename; + + if (dso->has_build_id) { + build_id__sprintf(dso->build_id, + sizeof(dso->build_id), bf + 15); + build_id_msg = bf; + } err = -ENOENT; dso->annotate_warned = 1; - pr_err("Can't annotate %s: No vmlinux file was found in the " - "path\n", sym->name); + pr_err("Can't annotate %s: No vmlinux file%s was found in the " + "path.\nPlease use 'perf buildid-cache -av vmlinux' or " + "--vmlinux vmlinux.\n", + sym->name, build_id_msg ?: ""); goto out_free_filename; } diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c index 294b49538522..497b3c4076a3 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -924,14 +924,6 @@ int hists__browse(struct hists *self, const char *helpline, if (choice == annotate) { struct hist_entry *he; do_annotate: - if (browser->selection->map->dso->origin == DSO__ORIG_KERNEL) { - browser->selection->map->dso->annotate_warned = 1; - ui_helpline__puts("No vmlinux file found, can't " - "annotate with just a " - "kallsyms file"); - continue; - } - he = hist_browser__selected_entry(browser); if (he == NULL) continue; -- cgit v1.2.3 From c16bfe9ac389b13a37ff617a09682ecc0685960f Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 25 Feb 2011 09:30:29 -0300 Subject: perf top browser: Fix up exit keys The left key was exiting 'perf top --tui' when it really shouldn't, it was too easy to leave the live annotation window and then press one too many <- and get out of the tool altogether. Do just like the report TUI does, ignore the left key for exit and also ask the user when pressing ESC if that is really what is wanted. Reported-by: Mike Galbraith Suggested-by: Ingo Molnar Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/browsers/top.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/browsers/top.c b/tools/perf/util/ui/browsers/top.c index 2f47224426b6..e9381ec9bfb1 100644 --- a/tools/perf/util/ui/browsers/top.c +++ b/tools/perf/util/ui/browsers/top.c @@ -10,6 +10,7 @@ #include "../../annotate.h" #include "../helpline.h" #include "../libslang.h" +#include "../util.h" #include "../../evlist.h" #include "../../hist.h" #include "../../sort.h" @@ -174,6 +175,12 @@ static int perf_top_browser__run(struct perf_top_browser *browser) if (browser->selection) perf_top_browser__annotate(browser); break; + case NEWT_KEY_LEFT: + continue; + case NEWT_KEY_ESCAPE: + if (!ui__dialog_yesno("Do you really want to exit?")) + continue; + /* Fall thru */ default: goto out; } -- cgit v1.2.3 From b210b3bb1b002f27165325a5edb6ebce3c168e92 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 25 Feb 2011 11:33:31 -0300 Subject: perf ui browser: Introduce ui_browser__show_title Needed because we were only showing the title in ui_browser__show, not in ui_browser__run, and in the run loop we may be calling other browsers that would then change the title, when we go back to the previous browser, we need to redraw the title. We could have done this as the Newt help line, with pop, etc, but I don't think its worth, doing it explicitely, when needed (some browsers may not use the title area at all) seems enough/more flexible. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/browser.c | 18 +++++++++++++++--- tools/perf/util/ui/browser.h | 3 ++- 2 files changed, 17 insertions(+), 4 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/browser.c b/tools/perf/util/ui/browser.c index 60d6c815e1db..611219f80680 100644 --- a/tools/perf/util/ui/browser.c +++ b/tools/perf/util/ui/browser.c @@ -157,6 +157,20 @@ void ui_browser__add_exit_keys(struct ui_browser *self, int keys[]) } } +void __ui_browser__show_title(struct ui_browser *browser, const char *title) +{ + SLsmg_gotorc(0, 0); + ui_browser__set_color(browser, NEWT_COLORSET_ROOT); + slsmg_write_nstring(title, browser->width); +} + +void ui_browser__show_title(struct ui_browser *browser, const char *title) +{ + pthread_mutex_lock(&ui__lock); + __ui_browser__show_title(browser, title); + pthread_mutex_unlock(&ui__lock); +} + int ui_browser__show(struct ui_browser *self, const char *title, const char *helpline, ...) { @@ -180,9 +194,7 @@ int ui_browser__show(struct ui_browser *self, const char *title, return -1; pthread_mutex_lock(&ui__lock); - SLsmg_gotorc(0, 0); - ui_browser__set_color(self, NEWT_COLORSET_ROOT); - slsmg_write_nstring(title, self->width); + __ui_browser__show_title(self, title); ui_browser__add_exit_keys(self, keys); newtFormAddComponent(self->form, self->sb); diff --git a/tools/perf/util/ui/browser.h b/tools/perf/util/ui/browser.h index 0dc7e4da36f5..fc63dda10910 100644 --- a/tools/perf/util/ui/browser.h +++ b/tools/perf/util/ui/browser.h @@ -24,7 +24,6 @@ struct ui_browser { u32 nr_entries; }; - void ui_browser__set_color(struct ui_browser *self, int color); void ui_browser__set_percent_color(struct ui_browser *self, double percent, bool current); @@ -35,6 +34,8 @@ void ui_browser__reset_index(struct ui_browser *self); void ui_browser__gotorc(struct ui_browser *self, int y, int x); void ui_browser__add_exit_key(struct ui_browser *self, int key); void ui_browser__add_exit_keys(struct ui_browser *self, int keys[]); +void __ui_browser__show_title(struct ui_browser *browser, const char *title); +void ui_browser__show_title(struct ui_browser *browser, const char *title); int ui_browser__show(struct ui_browser *self, const char *title, const char *helpline, ...); void ui_browser__hide(struct ui_browser *self); -- cgit v1.2.3 From 3166fc8fb6a2f52273d545e970297524e02c3e39 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 1 Mar 2011 10:21:44 -0300 Subject: perf top browser: Handle empty active symbols list Fixing a SEGV. An empty list could happen when not being able to resolve symbols, for instance when --vmlinux invalid-file is used. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/browsers/top.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/browsers/top.c b/tools/perf/util/ui/browsers/top.c index e9381ec9bfb1..5a06538532af 100644 --- a/tools/perf/util/ui/browsers/top.c +++ b/tools/perf/util/ui/browsers/top.c @@ -76,6 +76,12 @@ static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser) browser->root = RB_ROOT; browser->b.top = NULL; browser->sum_ksamples = perf_top__decay_samples(top, &browser->root); + /* + * No active symbols + */ + if (top->rb_entries == 0) + return; + perf_top__find_widths(top, &browser->root, &browser->dso_width, &browser->dso_short_width, &browser->sym_width); -- cgit v1.2.3 From a1ceb741cf86ef433006379742db81c00b450bae Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 1 Mar 2011 10:24:43 -0300 Subject: perf tui: Make ui__warning modal By taking the ui__lock so that no other screen updates take place while waiting for the user. That was happening when handling an invalid --vmlinux parameter in 'perf top --tui', with the screen refresh routine repainting the screen and removing the warning window. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/util.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/util.c b/tools/perf/util/ui/util.c index 7b5a8926624e..fdf1fc8f08bc 100644 --- a/tools/perf/util/ui/util.c +++ b/tools/perf/util/ui/util.c @@ -9,6 +9,7 @@ #include "../debug.h" #include "browser.h" #include "helpline.h" +#include "ui.h" #include "util.h" static void newt_form__set_exit_keys(newtComponent self) @@ -118,10 +119,12 @@ void ui__warning(const char *format, ...) va_list args; va_start(args, format); - if (use_browser > 0) + if (use_browser > 0) { + pthread_mutex_lock(&ui__lock); newtWinMessagev((char *)warning_str, (char *)ok, (char *)format, args); - else + pthread_mutex_unlock(&ui__lock); + } else vfprintf(stderr, format, args); va_end(args); } -- cgit v1.2.3 From 5807806a92450fd57f8063868efae9d4af74db02 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 1 Mar 2011 10:43:03 -0300 Subject: perf top tui: Wait till the first sample to refresh the screen. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-top.c | 21 +++++++++++++++++++-- tools/perf/util/top.h | 1 + 2 files changed, 20 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 0b07cc30b669..417f757e3cbe 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -72,6 +72,7 @@ static struct perf_top top = { .target_tid = -1, .active_symbols = LIST_HEAD_INIT(top.active_symbols), .active_symbols_lock = PTHREAD_MUTEX_INITIALIZER, + .active_symbols_cond = PTHREAD_COND_INITIALIZER, .freq = 1000, /* 1 KHz */ }; @@ -577,7 +578,17 @@ static void handle_keypress(struct perf_session *session, int c) static void *display_thread_tui(void *arg __used) { - perf_top__tui_browser(&top); + int err = 0; + pthread_mutex_lock(&top.active_symbols_lock); + while (list_empty(&top.active_symbols)) { + err = pthread_cond_wait(&top.active_symbols_cond, + &top.active_symbols_lock); + if (err) + break; + } + pthread_mutex_unlock(&top.active_symbols_lock); + if (!err) + perf_top__tui_browser(&top); exit_browser(0); exit(0); return NULL; @@ -776,8 +787,14 @@ static void perf_event__process_sample(const union perf_event *event, syme->count[evsel->idx]++; record_precise_ip(syme, evsel->idx, ip); pthread_mutex_lock(&top.active_symbols_lock); - if (list_empty(&syme->node) || !syme->node.next) + if (list_empty(&syme->node) || !syme->node.next) { + static bool first = true; __list_insert_active_sym(syme); + if (first) { + pthread_cond_broadcast(&top.active_symbols_cond); + first = false; + } + } pthread_mutex_unlock(&top.active_symbols_lock); } } diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index e8d28e2ecdba..96d1cb78af01 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -35,6 +35,7 @@ struct perf_top { */ struct list_head active_symbols; pthread_mutex_t active_symbols_lock; + pthread_cond_t active_symbols_cond; u64 samples; u64 kernel_samples, us_samples; u64 exact_samples; -- cgit v1.2.3 From 0a10247914a5cad3caf7ef8a255c54c4d3ed2062 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 26 Feb 2011 04:51:54 +0100 Subject: perf: Set filters before mmaping events We currently set the filters after we mmap the events, this is a race that let undesired events record themselves in the buffer before we had the time to set the filters. So set the filters before they can be recorded. That also librarizes the filters setting so that filtering can be done more easily from other tools than perf record later. Signed-off-by: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi Cc: Arnaldo Carvalho de Melo Cc: Steven Rostedt --- tools/perf/builtin-record.c | 18 ++++++------------ tools/perf/util/evlist.c | 28 ++++++++++++++++++++++++++++ tools/perf/util/evlist.h | 1 + 3 files changed, 35 insertions(+), 12 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index db4cd1e7b51a..d40a81e8cc56 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -180,12 +180,10 @@ static struct perf_header_attr *get_header_attr(struct perf_event_attr *a, int n static void create_counter(struct perf_evsel *evsel, int cpu) { - char *filter = evsel->filter; struct perf_event_attr *attr = &evsel->attr; struct perf_header_attr *h_attr; struct perf_sample_id *sid; int thread_index; - int ret; for (thread_index = 0; thread_index < evsel_list->threads->nr; thread_index++) { h_attr = get_header_attr(attr, evsel->idx); @@ -204,16 +202,6 @@ static void create_counter(struct perf_evsel *evsel, int cpu) pr_warning("Not enough memory to add id\n"); exit(-1); } - - if (filter != NULL) { - ret = ioctl(FD(evsel, cpu, thread_index), - PERF_EVENT_IOC_SET_FILTER, filter); - if (ret) { - error("failed to set filter with %d (%s)\n", errno, - strerror(errno)); - exit(-1); - } - } } if (!sample_type) @@ -367,6 +355,12 @@ try_again: } } + if (perf_evlist__set_filters(evlist)) { + error("failed to set filter with %d (%s)\n", errno, + strerror(errno)); + exit(-1); + } + if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) die("failed to mmap with %d (%s)\n", errno, strerror(errno)); diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 95b21fece2ce..030ae7f05e03 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -348,3 +348,31 @@ void perf_evlist__delete_maps(struct perf_evlist *evlist) evlist->cpus = NULL; evlist->threads = NULL; } + +int perf_evlist__set_filters(struct perf_evlist *evlist) +{ + const struct thread_map *threads = evlist->threads; + const struct cpu_map *cpus = evlist->cpus; + struct perf_evsel *evsel; + char *filter; + int thread; + int cpu; + int err; + int fd; + + list_for_each_entry(evsel, &evlist->entries, node) { + filter = evsel->filter; + if (!filter) + continue; + for (cpu = 0; cpu < cpus->nr; cpu++) { + for (thread = 0; thread < threads->nr; thread++) { + fd = FD(evsel, cpu, thread); + err = ioctl(fd, PERF_EVENT_IOC_SET_FILTER, filter); + if (err) + return err; + } + } + } + + return 0; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index c9884056097c..b75805aeb7e4 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -60,5 +60,6 @@ static inline void perf_evlist__set_maps(struct perf_evlist *evlist, int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid, pid_t target_tid, const char *cpu_list); void perf_evlist__delete_maps(struct perf_evlist *evlist); +int perf_evlist__set_filters(struct perf_evlist *evlist); #endif /* __PERF_EVLIST_H */ -- cgit v1.2.3 From ff9ae1babd8ce88c3f90db6278ea5f55bdcb4624 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 25 Feb 2011 21:57:04 +0100 Subject: perf: Fix missing strndup declaration is included first without _GNU_SOURCE, so it ends up including without declaring strndup(). And further declarations, even with _GNU_SOURCE defined, are of course without effect. Therefore: util/strfilter.c: Dans la fonction «strfilter_node__new» : util/strfilter.c:134: attention : déclaration implicite de la fonction « «strndup» » util/strfilter.c:134: attention : incompatible implicit declaration of built-in function «strndup» make: *** [util/strfilter.o] Erreur 1 Just don't include ctype.h as it doesn't appear to be necessary anyway. Signed-off-by: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi Cc: Arnaldo Carvalho de Melo --- tools/perf/util/strfilter.c | 1 - 1 file changed, 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/strfilter.c b/tools/perf/util/strfilter.c index 4064b7d708b8..834c8ebfe38e 100644 --- a/tools/perf/util/strfilter.c +++ b/tools/perf/util/strfilter.c @@ -1,4 +1,3 @@ -#include #include "util.h" #include "string.h" #include "strfilter.h" -- cgit v1.2.3 From cfff2d909cbdaf8c467bd321aa0502a548ec8f7e Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 25 Feb 2011 21:30:16 +0100 Subject: perf: Fix undefined PyVarObject_HEAD_INIT in python 2.5 PyVarObject_HEAD_INIT is undefined in python 2.5, resulting in a build crash: util/python.c:81: attention : déclaration implicite de la fonction « «PyVarObject_HEAD_INIT» » util/python.c:82: erreur: request for member «tp_name» in something not a structure or union util/python.c:117: erreur: request for member «tp_name» in something not a structure or union util/python.c:146: erreur: request for member «tp_name» in something not a structure or union util/python.c:177: erreur: request for member «tp_name» in something not a structure or union util/python.c:290: erreur: request for member «tp_name» in something not a structure or union util/python.c:359: erreur: request for member «tp_name» in something not a structure or union util/python.c:532: erreur: request for member «tp_name» in something not a structure or union util/python.c:761: erreur: request for member «tp_name» in something not a structure or union error: command 'gcc' failed with exit status 1 make: *** [python/perf.so] Erreur 1 We can fix that by defining PyVarObject_HEAD_INIT as a wrapper on PyObject_HEAD_INIT, thanks to a trick found on biopython: https://github.com/biopython/biopython/commit/d4eaf57946c7b4c32eca8d18821edf32f83e300d Signed-off-by: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi Cc: Arnaldo Carvalho de Melo --- tools/perf/util/python.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c index 5317ef229041..a9f2d7e1204d 100644 --- a/tools/perf/util/python.c +++ b/tools/perf/util/python.c @@ -8,6 +8,11 @@ #include "cpumap.h" #include "thread_map.h" +/* Define PyVarObject_HEAD_INIT for python 2.5 */ +#ifndef PyVarObject_HEAD_INIT +# define PyVarObject_HEAD_INIT(type, size) PyObject_HEAD_INIT(type) size, +#endif + struct throttle_event { struct perf_event_header header; u64 time; -- cgit v1.2.3 From a03f35ceeb3d279da35c5a914ac01a4b1effb0a1 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 3 Mar 2011 16:43:03 -0300 Subject: perf report tui: Fix multi event switching TAB/UNTAB were not hotkeys, so didn't exit hists__browse back to hists__tui_browse_tree, allowing just the first event to be browsed. Reported-by: William Cohen Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi Cc: William Cohen LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/browsers/hists.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c index 497b3c4076a3..c27ab35042f1 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -292,7 +292,8 @@ static int hist_browser__run(struct hist_browser *self, const char *title) { int key; int exit_keys[] = { 'a', '?', 'h', 'C', 'd', 'D', 'E', 't', - NEWT_KEY_ENTER, NEWT_KEY_RIGHT, NEWT_KEY_LEFT, 0, }; + NEWT_KEY_ENTER, NEWT_KEY_RIGHT, NEWT_KEY_LEFT, + NEWT_KEY_TAB, NEWT_KEY_UNTAB, 0, }; self->b.entries = &self->hists->entries; self->b.nr_entries = self->hists->nr_entries; @@ -859,6 +860,7 @@ int hists__browse(struct hists *self, const char *helpline, "E Expand all callchains\n" "d Zoom into current DSO\n" "t Zoom into current Thread\n" + "TAB/UNTAB Switch events\n" "q/CTRL+C Exit browser"); continue; case NEWT_KEY_ENTER: @@ -997,6 +999,7 @@ int hists__tui_browse_tree(struct rb_root *self, const char *help, int evidx) if (nd == first) continue; nd = rb_prev(nd); + break; default: return key; } -- cgit v1.2.3 From d7603d5122d9700fb8f36fa08b04f4e900fef059 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 4 Mar 2011 14:51:33 -0300 Subject: perf hists: Remove needless global col lenght calcs To support multiple events we need to do these calcs per 'struct hists' instance, and it turns out we already do that at: __hists__add_entry hists__inc_nr_entries hists__calc_col_len for all the unfiltered hist_entry instances we stash in the rb tree, so trow away the dead code. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/event.c | 57 +----------------------------------------------- tools/perf/util/hist.c | 9 ++++++++ tools/perf/util/symbol.c | 1 - tools/perf/util/symbol.h | 1 - 4 files changed, 10 insertions(+), 58 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index fbf5754c8866..2b15c362ef56 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -424,33 +424,6 @@ int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, return err; } -static void thread__comm_adjust(struct thread *self, struct hists *hists) -{ - char *comm = self->comm; - - if (!symbol_conf.col_width_list_str && !symbol_conf.field_sep && - (!symbol_conf.comm_list || - strlist__has_entry(symbol_conf.comm_list, comm))) { - u16 slen = strlen(comm); - - if (hists__new_col_len(hists, HISTC_COMM, slen)) - hists__set_col_len(hists, HISTC_THREAD, slen + 6); - } -} - -static int thread__set_comm_adjust(struct thread *self, const char *comm, - struct hists *hists) -{ - int ret = thread__set_comm(self, comm); - - if (ret) - return ret; - - thread__comm_adjust(self, hists); - - return 0; -} - int perf_event__process_comm(union perf_event *event, struct perf_sample *sample __used, struct perf_session *session) @@ -459,8 +432,7 @@ int perf_event__process_comm(union perf_event *event, dump_printf(": %s:%d\n", event->comm.comm, event->comm.tid); - if (thread == NULL || thread__set_comm_adjust(thread, event->comm.comm, - &session->hists)) { + if (thread == NULL || thread__set_comm(thread, event->comm.comm)) { dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n"); return -1; } @@ -760,18 +732,6 @@ void thread__find_addr_location(struct thread *self, al->sym = NULL; } -static void dso__calc_col_width(struct dso *self, struct hists *hists) -{ - if (!symbol_conf.col_width_list_str && !symbol_conf.field_sep && - (!symbol_conf.dso_list || - strlist__has_entry(symbol_conf.dso_list, self->name))) { - u16 slen = dso__name_len(self); - hists__new_col_len(hists, HISTC_DSO, slen); - } - - self->slen_calculated = 1; -} - int perf_event__preprocess_sample(const union perf_event *event, struct perf_session *session, struct addr_location *al, @@ -817,23 +777,8 @@ int perf_event__preprocess_sample(const union perf_event *event, strlist__has_entry(symbol_conf.dso_list, al->map->dso->long_name))))) goto out_filtered; - /* - * We have to do this here as we may have a dso with no symbol - * hit that has a name longer than the ones with symbols - * sampled. - */ - if (!sort_dso.elide && !al->map->dso->slen_calculated) - dso__calc_col_width(al->map->dso, &session->hists); al->sym = map__find_symbol(al->map, al->addr, filter); - } else { - const unsigned int unresolved_col_width = BITS_PER_LONG / 4; - - if (hists__col_len(&session->hists, HISTC_DSO) < unresolved_col_width && - !symbol_conf.col_width_list_str && !symbol_conf.field_sep && - !symbol_conf.dso_list) - hists__set_col_len(&session->hists, HISTC_DSO, - unresolved_col_width); } if (symbol_conf.sym_list && al->sym && diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index da2899e8c6f8..f7ad6bdbc667 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -50,6 +50,15 @@ static void hists__calc_col_len(struct hists *self, struct hist_entry *h) if (h->ms.sym) hists__new_col_len(self, HISTC_SYMBOL, h->ms.sym->namelen); + else { + const unsigned int unresolved_col_width = BITS_PER_LONG / 4; + + if (hists__col_len(self, HISTC_DSO) < unresolved_col_width && + !symbol_conf.col_width_list_str && !symbol_conf.field_sep && + !symbol_conf.dso_list) + hists__set_col_len(self, HISTC_DSO, + unresolved_col_width); + } len = thread__comm_len(h->thread); if (hists__new_col_len(self, HISTC_COMM, len)) diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index ba6d48949092..00014e32c288 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -207,7 +207,6 @@ struct dso *dso__new(const char *name) dso__set_short_name(self, self->name); for (i = 0; i < MAP__NR_TYPES; ++i) self->symbols[i] = self->symbol_names[i] = RB_ROOT; - self->slen_calculated = 0; self->origin = DSO__ORIG_NOT_FOUND; self->loaded = 0; self->sorted_by_name = 0; diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h index 670cd1c88f54..4d7ed09fe332 100644 --- a/tools/perf/util/symbol.h +++ b/tools/perf/util/symbol.h @@ -132,7 +132,6 @@ struct dso { struct rb_root symbol_names[MAP__NR_TYPES]; enum dso_kernel_type kernel; u8 adjust_symbols:1; - u8 slen_calculated:1; u8 has_build_id:1; u8 hit:1; u8 annotate_warned:1; -- cgit v1.2.3 From 60098917c06d154d06ce030c125266eab9e60768 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 4 Mar 2011 21:19:21 -0300 Subject: perf hists browser: Handle browsing empty hists tree Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/browsers/hists.c | 22 +++++++++++++++------- 1 file changed, 15 insertions(+), 7 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c index c27ab35042f1..c98e6f81d285 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -639,6 +639,9 @@ static void ui_browser__hists_seek(struct ui_browser *self, struct rb_node *nd; bool first = true; + if (self->nr_entries == 0) + return; + switch (whence) { case SEEK_SET: nd = hists__filter_entries(rb_first(self->entries)); @@ -820,8 +823,8 @@ int hists__browse(struct hists *self, const char *helpline, hists__browser_title(self, msg, sizeof(msg), ev_name, dso_filter, thread_filter); while (1) { - const struct thread *thread; - const struct dso *dso; + const struct thread *thread = NULL; + const struct dso *dso = NULL; char *options[16]; int nr_options = 0, choice = 0, i, annotate = -2, zoom_dso = -2, zoom_thread = -2, @@ -829,8 +832,10 @@ int hists__browse(struct hists *self, const char *helpline, key = hist_browser__run(browser, msg); - thread = hist_browser__selected_thread(browser); - dso = browser->selection->map ? browser->selection->map->dso : NULL; + if (browser->he_selection != NULL) { + thread = hist_browser__selected_thread(browser); + dso = browser->selection->map ? browser->selection->map->dso : NULL; + } switch (key) { case NEWT_KEY_TAB: @@ -841,7 +846,8 @@ int hists__browse(struct hists *self, const char *helpline, */ goto out_free_stack; case 'a': - if (browser->selection->map == NULL && + if (browser->selection == NULL || + browser->selection->map == NULL || browser->selection->map->dso->annotate_warned) continue; goto do_annotate; @@ -887,7 +893,8 @@ int hists__browse(struct hists *self, const char *helpline, goto out_free_stack; } - if (browser->selection->sym != NULL && + if (browser->selection != NULL && + browser->selection->sym != NULL && !browser->selection->map->dso->annotate_warned && asprintf(&options[nr_options], "Annotate %s", browser->selection->sym->name) > 0) @@ -906,7 +913,8 @@ int hists__browse(struct hists *self, const char *helpline, (dso->kernel ? "the Kernel" : dso->short_name)) > 0) zoom_dso = nr_options++; - if (browser->selection->map != NULL && + if (browser->selection != NULL && + browser->selection->map != NULL && asprintf(&options[nr_options], "Browse map details") > 0) browse_map = nr_options++; -- cgit v1.2.3 From 3d3b5e95997208067c963923db90ed1517565d14 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 4 Mar 2011 22:29:39 -0300 Subject: perf evlist: Split perf_evlist__id_hash The previous situation was to receive an fd from where to read the event ID. Spin off a routine for when we have the ID handy, not having to read it from some fd. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 28 ++++++++++++++++++---------- tools/perf/util/evlist.h | 3 +++ 2 files changed, 21 insertions(+), 10 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 030ae7f05e03..190c64c6e266 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -106,12 +106,24 @@ void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd) evlist->nr_fds++; } -static int perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, - int cpu, int thread, int fd) +void perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, u64 id) +{ + int hash; + struct perf_sample_id *sid = SID(evsel, cpu, thread); + + sid->id = id; + sid->evsel = evsel; + hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS); + hlist_add_head(&sid->node, &evlist->heads[hash]); +} + +static int perf_evlist__id_hash_fd(struct perf_evlist *evlist, + struct perf_evsel *evsel, + int cpu, int thread, int fd) { - struct perf_sample_id *sid; u64 read_data[4] = { 0, }; - int hash, id_idx = 1; /* The first entry is the counter value */ + int id_idx = 1; /* The first entry is the counter value */ if (!(evsel->attr.read_format & PERF_FORMAT_ID) || read(fd, &read_data, sizeof(read_data)) == -1) @@ -122,11 +134,7 @@ static int perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *e if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) ++id_idx; - sid = SID(evsel, cpu, thread); - sid->id = read_data[id_idx]; - sid->evsel = evsel; - hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS); - hlist_add_head(&sid->node, &evlist->heads[hash]); + perf_evlist__id_hash(evlist, evsel, cpu, thread, read_data[id_idx]); return 0; } @@ -300,7 +308,7 @@ int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite) goto out_unmap; if ((evsel->attr.read_format & PERF_FORMAT_ID) && - perf_evlist__id_hash(evlist, evsel, cpu, thread, fd) < 0) + perf_evlist__id_hash_fd(evlist, evsel, cpu, thread, fd) < 0) goto out_unmap; } } diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index b75805aeb7e4..078d51256085 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -38,6 +38,9 @@ void perf_evlist__delete(struct perf_evlist *evlist); void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); int perf_evlist__add_default(struct perf_evlist *evlist); +void perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, u64 id); + int perf_evlist__alloc_pollfd(struct perf_evlist *evlist); void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); -- cgit v1.2.3 From e248de331a452f8771eda6ed4bb30d92c82df28b Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 5 Mar 2011 21:40:06 -0300 Subject: perf tools: Improve support for sessions with multiple events By creating an perf_evlist out of the attributes in the perf.data file header, so that we can use evlists and evsels when reading recorded sessions in addition to when we record sessions. More work is needed to allow tools to allow the user to select which events are wanted when browsing sessions, be it just one or a subset of them, aggregated or showed at the same time but with different indications on the UI to allow seeing workloads thru different views at the same time. But the overall goal/trend is to more uniformly use evsels and evlists. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-annotate.c | 80 +++++++++++++----- tools/perf/builtin-report.c | 160 +++++++++++------------------------- tools/perf/util/evsel.h | 2 + tools/perf/util/header.c | 31 ------- tools/perf/util/header.h | 2 - tools/perf/util/hist.c | 6 +- tools/perf/util/hist.h | 12 ++- tools/perf/util/session.c | 63 +++++++++++++- tools/perf/util/session.h | 16 ++-- tools/perf/util/ui/browsers/hists.c | 32 ++++---- 10 files changed, 208 insertions(+), 196 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index 427182953fd7..695de4b5ae63 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -19,6 +19,8 @@ #include "perf.h" #include "util/debug.h" +#include "util/evlist.h" +#include "util/evsel.h" #include "util/annotate.h" #include "util/event.h" #include "util/parse-options.h" @@ -38,9 +40,13 @@ static bool print_line; static const char *sym_hist_filter; -static int hists__add_entry(struct hists *self, struct addr_location *al) +static int perf_evlist__add_sample(struct perf_evlist *evlist, + struct perf_sample *sample, + struct addr_location *al) { + struct perf_evsel *evsel; struct hist_entry *he; + int ret; if (sym_hist_filter != NULL && (al->sym == NULL || strcmp(sym_hist_filter, al->sym->name) != 0)) { @@ -53,23 +59,35 @@ static int hists__add_entry(struct hists *self, struct addr_location *al) return 0; } - he = __hists__add_entry(self, al, NULL, 1); + evsel = perf_evlist__id2evsel(evlist, sample->id); + if (evsel == NULL) { + /* + * FIXME: Propagate this back, but at least we're in a builtin, + * where exit() is allowed. ;-) + */ + ui__warning("Invalid %s file, contains samples with id not in " + "its header!\n", input_name); + exit_browser(0); + exit(1); + } + + he = __hists__add_entry(&evsel->hists, al, NULL, 1); if (he == NULL) return -ENOMEM; + ret = 0; if (he->ms.sym != NULL) { - /* - * All aggregated on the first sym_hist. - */ struct annotation *notes = symbol__annotation(he->ms.sym); if (notes->src == NULL && - symbol__alloc_hist(he->ms.sym, 1) < 0) + symbol__alloc_hist(he->ms.sym, evlist->nr_entries) < 0) return -ENOMEM; - return hist_entry__inc_addr_samples(he, 0, al->addr); + ret = hist_entry__inc_addr_samples(he, evsel->idx, al->addr); } - return 0; + evsel->hists.stats.total_period += sample->period; + hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE); + return ret; } static int process_sample_event(union perf_event *event, @@ -85,7 +103,7 @@ static int process_sample_event(union perf_event *event, return -1; } - if (!al.filtered && hists__add_entry(&session->hists, &al)) { + if (!al.filtered && perf_evlist__add_sample(session->evlist, sample, &al)) { pr_warning("problem incrementing symbol count, " "skipping event\n"); return -1; @@ -100,7 +118,7 @@ static int hist_entry__tty_annotate(struct hist_entry *he, int evidx) print_line, full_paths, 0, 0); } -static void hists__find_annotations(struct hists *self) +static void hists__find_annotations(struct hists *self, int evidx) { struct rb_node *nd = rb_first(&self->entries), *next; int key = KEY_RIGHT; @@ -123,8 +141,7 @@ find_next: } if (use_browser > 0) { - /* For now all is aggregated on the first */ - key = hist_entry__tui_annotate(he, 0); + key = hist_entry__tui_annotate(he, evidx); switch (key) { case KEY_RIGHT: next = rb_next(nd); @@ -139,8 +156,7 @@ find_next: if (next != NULL) nd = next; } else { - /* For now all is aggregated on the first */ - hist_entry__tty_annotate(he, 0); + hist_entry__tty_annotate(he, evidx); nd = rb_next(nd); /* * Since we have a hist_entry per IP for the same @@ -166,6 +182,8 @@ static int __cmd_annotate(void) { int ret; struct perf_session *session; + struct perf_evsel *pos; + u64 total_nr_samples; session = perf_session__new(input_name, O_RDONLY, force, false, &event_ops); if (session == NULL) @@ -186,12 +204,36 @@ static int __cmd_annotate(void) if (verbose > 2) perf_session__fprintf_dsos(session, stdout); - hists__collapse_resort(&session->hists); - hists__output_resort(&session->hists); - hists__find_annotations(&session->hists); -out_delete: - perf_session__delete(session); + total_nr_samples = 0; + list_for_each_entry(pos, &session->evlist->entries, node) { + struct hists *hists = &pos->hists; + u32 nr_samples = hists->stats.nr_events[PERF_RECORD_SAMPLE]; + + if (nr_samples > 0) { + total_nr_samples += nr_samples; + hists__collapse_resort(hists); + hists__output_resort(hists); + hists__find_annotations(hists, pos->idx); + } + } + if (total_nr_samples == 0) { + ui__warning("The %s file has no samples!\n", input_name); + goto out_delete; + } +out_delete: + /* + * Speed up the exit process, for large files this can + * take quite a while. + * + * XXX Enable this when using valgrind or if we ever + * librarize this command. + * + * Also experiment with obstacks to see how much speed + * up we'll get here. + * + * perf_session__delete(session); + */ return ret; } diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index dddcc7ea2bec..1c399eae5f7b 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -21,6 +21,8 @@ #include "perf.h" #include "util/debug.h" +#include "util/evlist.h" +#include "util/evsel.h" #include "util/header.h" #include "util/session.h" @@ -46,39 +48,6 @@ static const char *pretty_printing_style = default_pretty_printing_style; static char callchain_default_opt[] = "fractal,0.5"; static symbol_filter_t annotate_init; -static struct hists *perf_session__hists_findnew(struct perf_session *self, - u64 event_stream, u32 type, - u64 config) -{ - struct rb_node **p = &self->hists_tree.rb_node; - struct rb_node *parent = NULL; - struct hists *iter, *new; - - while (*p != NULL) { - parent = *p; - iter = rb_entry(parent, struct hists, rb_node); - if (iter->config == config) - return iter; - - - if (config > iter->config) - p = &(*p)->rb_right; - else - p = &(*p)->rb_left; - } - - new = malloc(sizeof(struct hists)); - if (new == NULL) - return NULL; - memset(new, 0, sizeof(struct hists)); - new->event_stream = event_stream; - new->config = config; - new->type = type; - rb_link_node(&new->rb_node, parent, p); - rb_insert_color(&new->rb_node, &self->hists_tree); - return new; -} - static int perf_session__add_hist_entry(struct perf_session *session, struct addr_location *al, struct perf_sample *sample) @@ -86,8 +55,7 @@ static int perf_session__add_hist_entry(struct perf_session *session, struct symbol *parent = NULL; int err = 0; struct hist_entry *he; - struct hists *hists; - struct perf_event_attr *attr; + struct perf_evsel *evsel; if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) { err = perf_session__resolve_callchain(session, al->thread, @@ -96,15 +64,19 @@ static int perf_session__add_hist_entry(struct perf_session *session, return err; } - attr = perf_header__find_attr(sample->id, &session->header); - if (attr) - hists = perf_session__hists_findnew(session, sample->id, attr->type, attr->config); - else - hists = perf_session__hists_findnew(session, sample->id, 0, 0); - if (hists == NULL) - return -ENOMEM; + evsel = perf_evlist__id2evsel(session->evlist, sample->id); + if (evsel == NULL) { + /* + * FIXME: Propagate this back, but at least we're in a builtin, + * where exit() is allowed. ;-) + */ + ui__warning("Invalid %s file, contains samples with id not in " + "its header!\n", input_name); + exit_browser(0); + exit(1); + } - he = __hists__add_entry(hists, al, parent, sample->period); + he = __hists__add_entry(&evsel->hists, al, parent, sample->period); if (he == NULL) return -ENOMEM; @@ -120,52 +92,30 @@ static int perf_session__add_hist_entry(struct perf_session *session, * code will not use it. */ if (al->sym != NULL && use_browser > 0) { - /* - * All aggregated on the first sym_hist. - */ struct annotation *notes = symbol__annotation(he->ms.sym); + + assert(evsel != NULL); + + err = -ENOMEM; if (notes->src == NULL && - symbol__alloc_hist(he->ms.sym, 1) < 0) - err = -ENOMEM; - else - err = hist_entry__inc_addr_samples(he, 0, al->addr); + symbol__alloc_hist(he->ms.sym, session->evlist->nr_entries) < 0) + goto out; + + err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr); } + evsel->hists.stats.total_period += sample->period; + hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE); +out: return err; } -static int add_event_total(struct perf_session *session, - struct perf_sample *sample, - struct perf_event_attr *attr) -{ - struct hists *hists; - - if (attr) - hists = perf_session__hists_findnew(session, sample->id, - attr->type, attr->config); - else - hists = perf_session__hists_findnew(session, sample->id, 0, 0); - - if (!hists) - return -ENOMEM; - - hists->stats.total_period += sample->period; - /* - * FIXME: add_event_total should be moved from here to - * perf_session__process_event so that the proper hist is passed to - * the event_op methods. - */ - hists__inc_nr_events(hists, PERF_RECORD_SAMPLE); - session->hists.stats.total_period += sample->period; - return 0; -} static int process_sample_event(union perf_event *event, struct perf_sample *sample, struct perf_session *session) { struct addr_location al; - struct perf_event_attr *attr; if (perf_event__preprocess_sample(event, session, &al, sample, annotate_init) < 0) { @@ -182,27 +132,17 @@ static int process_sample_event(union perf_event *event, return -1; } - attr = perf_header__find_attr(sample->id, &session->header); - - if (add_event_total(session, sample, attr)) { - pr_debug("problem adding event period\n"); - return -1; - } - return 0; } static int process_read_event(union perf_event *event, struct perf_sample *sample __used, - struct perf_session *session __used) + struct perf_session *session) { - struct perf_event_attr *attr; - - attr = perf_header__find_attr(event->read.id, &session->header); - + struct perf_evsel *evsel = perf_evlist__id2evsel(session->evlist, + event->read.id); if (show_threads) { - const char *name = attr ? __event_name(attr->type, attr->config) - : "unknown"; + const char *name = evsel ? event_name(evsel) : "unknown"; perf_read_values_add_value(&show_threads_values, event->read.pid, event->read.tid, event->read.id, @@ -211,7 +151,7 @@ static int process_read_event(union perf_event *event, } dump_printf(": %d %d %s %" PRIu64 "\n", event->read.pid, event->read.tid, - attr ? __event_name(attr->type, attr->config) : "FAIL", + evsel ? event_name(evsel) : "FAIL", event->read.value); return 0; @@ -282,21 +222,20 @@ static size_t hists__fprintf_nr_sample_events(struct hists *self, return ret + fprintf(fp, "\n#\n"); } -static int hists__tty_browse_tree(struct rb_root *tree, const char *help) +static int hists__tty_browse_tree(struct perf_evlist *evlist, const char *help) { - struct rb_node *next = rb_first(tree); + struct perf_evsel *pos; - while (next) { - struct hists *hists = rb_entry(next, struct hists, rb_node); + list_for_each_entry(pos, &evlist->entries, node) { + struct hists *hists = &pos->hists; const char *evname = NULL; if (rb_first(&hists->entries) != rb_last(&hists->entries)) - evname = __event_name(hists->type, hists->config); + evname = event_name(pos); hists__fprintf_nr_sample_events(hists, evname, stdout); hists__fprintf(hists, NULL, false, stdout); fprintf(stdout, "\n\n"); - next = rb_next(&hists->rb_node); } if (sort_order == default_sort_order && @@ -317,8 +256,9 @@ static int hists__tty_browse_tree(struct rb_root *tree, const char *help) static int __cmd_report(void) { int ret = -EINVAL; + u64 nr_samples; struct perf_session *session; - struct rb_node *next; + struct perf_evsel *pos; const char *help = "For a higher level overview, try: perf report --sort comm,dso"; signal(SIGINT, sig_handler); @@ -349,26 +289,24 @@ static int __cmd_report(void) if (verbose > 2) perf_session__fprintf_dsos(session, stdout); - next = rb_first(&session->hists_tree); - - if (next == NULL) { - ui__warning("The %s file has no samples!\n", input_name); - goto out_delete; - } - - while (next) { - struct hists *hists; + nr_samples = 0; + list_for_each_entry(pos, &session->evlist->entries, node) { + struct hists *hists = &pos->hists; - hists = rb_entry(next, struct hists, rb_node); hists__collapse_resort(hists); hists__output_resort(hists); - next = rb_next(&hists->rb_node); + nr_samples += hists->stats.nr_events[PERF_RECORD_SAMPLE]; + } + + if (nr_samples == 0) { + ui__warning("The %s file has no samples!\n", input_name); + goto out_delete; } if (use_browser > 0) - hists__tui_browse_tree(&session->hists_tree, help, 0); + hists__tui_browse_tree(session->evlist, help); else - hists__tty_browse_tree(&session->hists_tree, help); + hists__tty_browse_tree(session->evlist, help); out_delete: /* diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index f6fc8f651a25..281b60e5fc7b 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -7,6 +7,7 @@ #include "types.h" #include "xyarray.h" #include "cgroup.h" +#include "hist.h" struct perf_counts_values { union { @@ -51,6 +52,7 @@ struct perf_evsel { struct xyarray *id; struct perf_counts *counts; int idx; + struct hists hists; char *name; void *priv; struct cgroup_sel *cgrp; diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 72c124dc5781..108b0db7bbef 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -969,37 +969,6 @@ bool perf_header__sample_id_all(const struct perf_header *header) return value; } -struct perf_event_attr * -perf_header__find_attr(u64 id, struct perf_header *header) -{ - int i; - - /* - * We set id to -1 if the data file doesn't contain sample - * ids. This can happen when the data file contains one type - * of event and in that case, the header can still store the - * event attribute information. Check for this and avoid - * walking through the entire list of ids which may be large. - */ - if (id == -1ULL) { - if (header->attrs > 0) - return &header->attr[0]->attr; - return NULL; - } - - for (i = 0; i < header->attrs; i++) { - struct perf_header_attr *attr = header->attr[i]; - int j; - - for (j = 0; j < attr->ids; j++) { - if (attr->id[j] == id) - return &attr->attr; - } - } - - return NULL; -} - int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, perf_event__handler_t process, struct perf_session *session) diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index f042cebcec1e..2fab13348aab 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -85,8 +85,6 @@ int perf_header_attr__add_id(struct perf_header_attr *self, u64 id); u64 perf_header__sample_type(struct perf_header *header); bool perf_header__sample_id_all(const struct perf_header *header); -struct perf_event_attr * -perf_header__find_attr(u64 id, struct perf_header *header); void perf_header__set_feat(struct perf_header *self, int feat); void perf_header__clear_feat(struct perf_header *self, int feat); bool perf_header__has_feat(const struct perf_header *self, int feat); diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index f7ad6bdbc667..627a02e03c57 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -984,8 +984,12 @@ size_t hists__fprintf_nr_events(struct hists *self, FILE *fp) size_t ret = 0; for (i = 0; i < PERF_RECORD_HEADER_MAX; ++i) { - const char *name = perf_event__name(i); + const char *name; + if (self->stats.nr_events[i] == 0) + continue; + + name = perf_event__name(i); if (!strcmp(name, "UNKNOWN")) continue; diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 37c79089de09..0d38b435827b 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -42,13 +42,10 @@ enum hist_column { }; struct hists { - struct rb_node rb_node; struct rb_root entries; u64 nr_entries; struct events_stats stats; - u64 config; u64 event_stream; - u32 type; u16 col_len[HISTC_NR_COLS]; /* Best would be to reuse the session callchain cursor */ struct callchain_cursor callchain_cursor; @@ -87,6 +84,8 @@ u16 hists__col_len(struct hists *self, enum hist_column col); void hists__set_col_len(struct hists *self, enum hist_column col, u16 len); bool hists__new_col_len(struct hists *self, enum hist_column col, u16 len); +struct perf_evlist; + #ifdef NO_NEWT_SUPPORT static inline int hists__browse(struct hists *self __used, const char *helpline __used, @@ -95,9 +94,8 @@ static inline int hists__browse(struct hists *self __used, return 0; } -static inline int hists__tui_browse_tree(struct rb_root *self __used, - const char *help __used, - int evidx __used) +static inline int hists__tui_browse_tree(struct perf_evlist *evlist __used, + const char *help __used) { return 0; } @@ -118,7 +116,7 @@ int hist_entry__tui_annotate(struct hist_entry *self, int evidx); #define KEY_LEFT NEWT_KEY_LEFT #define KEY_RIGHT NEWT_KEY_RIGHT -int hists__tui_browse_tree(struct rb_root *self, const char *help, int evidx); +int hists__tui_browse_tree(struct perf_evlist *evlist, const char *help); #endif unsigned int hists__sort_list_width(struct hists *self); diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index a3a871f7bda3..0d414199889d 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -7,10 +7,52 @@ #include #include +#include "evlist.h" +#include "evsel.h" #include "session.h" #include "sort.h" #include "util.h" +static int perf_session__read_evlist(struct perf_session *session) +{ + int i, j; + + session->evlist = perf_evlist__new(NULL, NULL); + if (session->evlist == NULL) + return -ENOMEM; + + for (i = 0; i < session->header.attrs; ++i) { + struct perf_header_attr *hattr = session->header.attr[i]; + struct perf_evsel *evsel = perf_evsel__new(&hattr->attr, i); + + if (evsel == NULL) + goto out_delete_evlist; + /* + * Do it before so that if perf_evsel__alloc_id fails, this + * entry gets purged too at perf_evlist__delete(). + */ + perf_evlist__add(session->evlist, evsel); + /* + * We don't have the cpu and thread maps on the header, so + * for allocating the perf_sample_id table we fake 1 cpu and + * hattr->ids threads. + */ + if (perf_evsel__alloc_id(evsel, 1, hattr->ids)) + goto out_delete_evlist; + + for (j = 0; j < hattr->ids; ++j) + perf_evlist__id_hash(session->evlist, evsel, 0, j, + hattr->id[j]); + } + + return 0; + +out_delete_evlist: + perf_evlist__delete(session->evlist); + session->evlist = NULL; + return -ENOMEM; +} + static int perf_session__open(struct perf_session *self, bool force) { struct stat input_stat; @@ -56,6 +98,11 @@ static int perf_session__open(struct perf_session *self, bool force) goto out_close; } + if (perf_session__read_evlist(self) < 0) { + pr_err("Not enough memory to read the event selector list\n"); + goto out_close; + } + self->size = input_stat.st_size; return 0; @@ -141,7 +188,6 @@ struct perf_session *perf_session__new(const char *filename, int mode, memcpy(self->filename, filename, len); self->threads = RB_ROOT; INIT_LIST_HEAD(&self->dead_threads); - self->hists_tree = RB_ROOT; self->last_match = NULL; /* * On 64bit we can mmap the data file in one go. No need for tiny mmap @@ -1137,3 +1183,18 @@ size_t perf_session__fprintf_dsos_buildid(struct perf_session *self, FILE *fp, size_t ret = machine__fprintf_dsos_buildid(&self->host_machine, fp, with_hits); return ret + machines__fprintf_dsos_buildid(&self->machines, fp, with_hits); } + +size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp) +{ + struct perf_evsel *pos; + size_t ret = fprintf(fp, "Aggregated stats:\n"); + + ret += hists__fprintf_nr_events(&session->hists, fp); + + list_for_each_entry(pos, &session->evlist->entries, node) { + ret += fprintf(fp, "%s stats:\n", event_name(pos)); + ret += hists__fprintf_nr_events(&pos->hists, fp); + } + + return ret; +} diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 977b3a1b14aa..05dd7bcb9453 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -34,12 +34,12 @@ struct perf_session { struct thread *last_match; struct machine host_machine; struct rb_root machines; - struct rb_root hists_tree; + struct perf_evlist *evlist; /* - * FIXME: should point to the first entry in hists_tree and - * be a hists instance. Right now its only 'report' - * that is using ->hists_tree while all the rest use - * ->hists. + * FIXME: Need to split this up further, we need global + * stats + per event stats. 'perf diff' also needs + * to properly support multiple events in a single + * perf.data file. */ struct hists hists; u64 sample_type; @@ -151,11 +151,7 @@ size_t perf_session__fprintf_dsos(struct perf_session *self, FILE *fp); size_t perf_session__fprintf_dsos_buildid(struct perf_session *self, FILE *fp, bool with_hits); -static inline -size_t perf_session__fprintf_nr_events(struct perf_session *self, FILE *fp) -{ - return hists__fprintf_nr_events(&self->hists, fp); -} +size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp); static inline int perf_session__parse_sample(struct perf_session *session, const union perf_event *event, diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c index c98e6f81d285..f3af4fe5cdc4 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -7,6 +7,8 @@ #include #include +#include "../../evsel.h" +#include "../../evlist.h" #include "../../hist.h" #include "../../pstack.h" #include "../../sort.h" @@ -987,31 +989,33 @@ out: return key; } -int hists__tui_browse_tree(struct rb_root *self, const char *help, int evidx) +int hists__tui_browse_tree(struct perf_evlist *evlist, const char *help) { - struct rb_node *first = rb_first(self), *nd = first, *next; - int key = 0; + struct perf_evsel *pos; - while (nd) { - struct hists *hists = rb_entry(nd, struct hists, rb_node); - const char *ev_name = __event_name(hists->type, hists->config); + pos = list_entry(evlist->entries.next, struct perf_evsel, node); + while (pos) { + struct hists *hists = &pos->hists; + const char *ev_name = event_name(pos); + int key = hists__browse(hists, help, ev_name, pos->idx); - key = hists__browse(hists, help, ev_name, evidx); switch (key) { case NEWT_KEY_TAB: - next = rb_next(nd); - if (next) - nd = next; + if (pos->node.next == &evlist->entries) + pos = list_entry(evlist->entries.next, struct perf_evsel, node); + else + pos = list_entry(pos->node.next, struct perf_evsel, node); break; case NEWT_KEY_UNTAB: - if (nd == first) - continue; - nd = rb_prev(nd); + if (pos->node.prev == &evlist->entries) + pos = list_entry(evlist->entries.prev, struct perf_evsel, node); + else + pos = list_entry(pos->node.prev, struct perf_evsel, node); break; default: return key; } } - return key; + return 0; } -- cgit v1.2.3 From 7f0030b211579939461468f25b80c73e293c46e0 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sun, 6 Mar 2011 13:07:30 -0300 Subject: perf report tui: Improve multi event session support When multiple events were used in 'perf record', allow the user to choose which one is wanted before showing the per event histograms. Annotations will be performed on the chosen event. Allow going back and forth from event to event quickly using just the arrow keys and enter. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi Cc: William Cohen LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-report.c | 7 +- tools/perf/util/hist.h | 16 +--- tools/perf/util/ui/browsers/hists.c | 147 ++++++++++++++++++++++++++++++++---- 3 files changed, 140 insertions(+), 30 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index 1c399eae5f7b..e9b5d513333a 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -222,7 +222,8 @@ static size_t hists__fprintf_nr_sample_events(struct hists *self, return ret + fprintf(fp, "\n#\n"); } -static int hists__tty_browse_tree(struct perf_evlist *evlist, const char *help) +static int perf_evlist__tty_browse_hists(struct perf_evlist *evlist, + const char *help) { struct perf_evsel *pos; @@ -304,9 +305,9 @@ static int __cmd_report(void) } if (use_browser > 0) - hists__tui_browse_tree(session->evlist, help); + perf_evlist__tui_browse_hists(session->evlist, help); else - hists__tty_browse_tree(session->evlist, help); + perf_evlist__tty_browse_hists(session->evlist, help); out_delete: /* diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 0d38b435827b..cb6858a2f9a3 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -87,15 +87,9 @@ bool hists__new_col_len(struct hists *self, enum hist_column col, u16 len); struct perf_evlist; #ifdef NO_NEWT_SUPPORT -static inline int hists__browse(struct hists *self __used, - const char *helpline __used, - const char *ev_name __used, int evidx __used) -{ - return 0; -} - -static inline int hists__tui_browse_tree(struct perf_evlist *evlist __used, - const char *help __used) +static inline +int perf_evlist__tui_browse_hists(struct perf_evlist *evlist __used, + const char *help __used) { return 0; } @@ -109,14 +103,12 @@ static inline int hist_entry__tui_annotate(struct hist_entry *self __used, #define KEY_RIGHT -2 #else #include -int hists__browse(struct hists *self, const char *helpline, - const char *ev_name, int evidx); int hist_entry__tui_annotate(struct hist_entry *self, int evidx); #define KEY_LEFT NEWT_KEY_LEFT #define KEY_RIGHT NEWT_KEY_RIGHT -int hists__tui_browse_tree(struct perf_evlist *evlist, const char *help); +int perf_evlist__tui_browse_hists(struct perf_evlist *evlist, const char *help); #endif unsigned int hists__sort_list_width(struct hists *self); diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c index f3af4fe5cdc4..798efdca3ead 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -803,9 +803,11 @@ static int hists__browser_title(struct hists *self, char *bf, size_t size, return printed; } -int hists__browse(struct hists *self, const char *helpline, - const char *ev_name, int evidx) +static int perf_evsel__hists_browse(struct perf_evsel *evsel, + const char *helpline, const char *ev_name, + bool left_exits) { + struct hists *self = &evsel->hists; struct hist_browser *browser = hist_browser__new(self); struct pstack *fstack; const struct thread *thread_filter = NULL; @@ -878,8 +880,14 @@ int hists__browse(struct hists *self, const char *helpline, case NEWT_KEY_LEFT: { const void *top; - if (pstack__empty(fstack)) + if (pstack__empty(fstack)) { + /* + * Go back to the perf_evsel_menu__run or other user + */ + if (left_exits) + goto out_free_stack; continue; + } top = pstack__pop(fstack); if (top == &dso_filter) goto zoom_out_dso; @@ -888,7 +896,8 @@ int hists__browse(struct hists *self, const char *helpline, continue; } case NEWT_KEY_ESCAPE: - if (!ui__dialog_yesno("Do you really want to exit?")) + if (!left_exits && + !ui__dialog_yesno("Do you really want to exit?")) continue; /* Fall thru */ default: @@ -940,7 +949,7 @@ do_annotate: if (he == NULL) continue; - hist_entry__tui_annotate(he, evidx); + hist_entry__tui_annotate(he, evsel->idx); } else if (choice == browse_map) map__browse(browser->selection->map); else if (choice == zoom_dso) { @@ -989,15 +998,71 @@ out: return key; } -int hists__tui_browse_tree(struct perf_evlist *evlist, const char *help) +struct perf_evsel_menu { + struct ui_browser b; + struct perf_evsel *selection; +}; + +static void perf_evsel_menu__write(struct ui_browser *browser, + void *entry, int row) +{ + struct perf_evsel_menu *menu = container_of(browser, + struct perf_evsel_menu, b); + struct perf_evsel *evsel = list_entry(entry, struct perf_evsel, node); + bool current_entry = ui_browser__is_current_entry(browser, row); + unsigned long nr_events = evsel->hists.stats.nr_events[PERF_RECORD_SAMPLE]; + const char *ev_name = event_name(evsel); + char bf[256], unit; + + ui_browser__set_color(browser, current_entry ? HE_COLORSET_SELECTED : + HE_COLORSET_NORMAL); + + nr_events = convert_unit(nr_events, &unit); + snprintf(bf, sizeof(bf), "%lu%c%s%s", nr_events, + unit, unit == ' ' ? "" : " ", ev_name); + slsmg_write_nstring(bf, browser->width); + + if (current_entry) + menu->selection = evsel; +} + +static int perf_evsel_menu__run(struct perf_evsel_menu *menu, const char *help) { + int exit_keys[] = { NEWT_KEY_ENTER, NEWT_KEY_RIGHT, 0, }; + struct perf_evlist *evlist = menu->b.priv; struct perf_evsel *pos; + const char *ev_name, *title = "Available samples"; + int key; - pos = list_entry(evlist->entries.next, struct perf_evsel, node); - while (pos) { - struct hists *hists = &pos->hists; - const char *ev_name = event_name(pos); - int key = hists__browse(hists, help, ev_name, pos->idx); + if (ui_browser__show(&menu->b, title, + "ESC: exit, ENTER|->: Browse histograms") < 0) + return -1; + + ui_browser__add_exit_keys(&menu->b, exit_keys); + + while (1) { + key = ui_browser__run(&menu->b); + + switch (key) { + case NEWT_KEY_RIGHT: + case NEWT_KEY_ENTER: + if (!menu->selection) + continue; + pos = menu->selection; +browse_hists: + ev_name = event_name(pos); + key = perf_evsel__hists_browse(pos, help, ev_name, true); + ui_browser__show_title(&menu->b, title); + break; + case NEWT_KEY_LEFT: + continue; + case NEWT_KEY_ESCAPE: + if (!ui__dialog_yesno("Do you really want to exit?")) + continue; + /* Fall thru */ + default: + goto out; + } switch (key) { case NEWT_KEY_TAB: @@ -1005,17 +1070,69 @@ int hists__tui_browse_tree(struct perf_evlist *evlist, const char *help) pos = list_entry(evlist->entries.next, struct perf_evsel, node); else pos = list_entry(pos->node.next, struct perf_evsel, node); - break; + goto browse_hists; case NEWT_KEY_UNTAB: if (pos->node.prev == &evlist->entries) pos = list_entry(evlist->entries.prev, struct perf_evsel, node); else pos = list_entry(pos->node.prev, struct perf_evsel, node); - break; + goto browse_hists; + case 'q': + case CTRL('c'): + goto out; default: - return key; + break; } } - return 0; +out: + ui_browser__hide(&menu->b); + return key; +} + +static int __perf_evlist__tui_browse_hists(struct perf_evlist *evlist, + const char *help) +{ + struct perf_evsel *pos; + struct perf_evsel_menu menu = { + .b = { + .entries = &evlist->entries, + .refresh = ui_browser__list_head_refresh, + .seek = ui_browser__list_head_seek, + .write = perf_evsel_menu__write, + .nr_entries = evlist->nr_entries, + .priv = evlist, + }, + }; + + ui_helpline__push("Press ESC to exit"); + + list_for_each_entry(pos, &evlist->entries, node) { + const char *ev_name = event_name(pos); + size_t line_len = strlen(ev_name) + 7; + + if (menu.b.width < line_len) + menu.b.width = line_len; + /* + * Cache the evsel name, tracepoints have a _high_ cost per + * event_name() call. + */ + if (pos->name == NULL) + pos->name = strdup(ev_name); + } + + return perf_evsel_menu__run(&menu, help); +} + +int perf_evlist__tui_browse_hists(struct perf_evlist *evlist, const char *help) +{ + + if (evlist->nr_entries == 1) { + struct perf_evsel *first = list_entry(evlist->entries.next, + struct perf_evsel, node); + const char *ev_name = event_name(first); + return perf_evsel__hists_browse(first, help, ev_name, false); + } + + return __perf_evlist__tui_browse_hists(evlist, help); } -- cgit v1.2.3 From b9a46bba88001504235459c8410f17e6a7e38008 Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Mon, 7 Mar 2011 21:13:40 +0100 Subject: perf top: Fix events overflow in top command The snprintf function returns number of printed characters even if it cross the size parameter. So passing enough events via '-e' parameter will cause segmentation fault. It's reproduced by following command: perf top -e `perf list | grep Tracepoint | awk -F'[' '\ {gsub(/[[:space:]]+/,"",$1);array[FNR]=$1}END{outputs=array[1];\ for (i=2;i<=FNR;i++){ outputs=outputs "," array[i];};print outputs}'` Attached patch is adding SNPRINTF macro that provides the overflow check and returns actuall number of printed characters. Reported-by: Han Pingtian Cc: Han Pingtian Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra LKML-Reference: <1299528821-17521-2-git-send-email-jolsa@redhat.com> Signed-off-by: Jiri Olsa Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/top.c | 30 ++++++++++++++++++------------ 1 file changed, 18 insertions(+), 12 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c index 70a9c13f4ad5..4f869da4e9c7 100644 --- a/tools/perf/util/top.c +++ b/tools/perf/util/top.c @@ -61,6 +61,12 @@ static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se) rb_insert_color(&se->rb_node, tree); } +#define SNPRINTF(buf, size, fmt, args...) \ +({ \ + size_t r = snprintf(buf, size, fmt, ## args); \ + r > size ? size : r; \ +}) + size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size) { struct perf_evsel *counter; @@ -70,7 +76,7 @@ size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size) size_t ret = 0; if (!perf_guest) { - ret = snprintf(bf, size, + ret = SNPRINTF(bf, size, " PerfTop:%8.0f irqs/sec kernel:%4.1f%%" " exact: %4.1f%% [", samples_per_sec, 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / @@ -81,7 +87,7 @@ size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size) float guest_kernel_samples_per_sec = top->guest_kernel_samples / top->delay_secs; float guest_us_samples_per_sec = top->guest_us_samples / top->delay_secs; - ret = snprintf(bf, size, + ret = SNPRINTF(bf, size, " PerfTop:%8.0f irqs/sec kernel:%4.1f%% us:%4.1f%%" " guest kernel:%4.1f%% guest us:%4.1f%%" " exact: %4.1f%% [", samples_per_sec, @@ -101,38 +107,38 @@ size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size) if (top->evlist->nr_entries == 1 || !top->display_weighted) { struct perf_evsel *first; first = list_entry(top->evlist->entries.next, struct perf_evsel, node); - ret += snprintf(bf + ret, size - ret, "%" PRIu64 "%s ", + ret += SNPRINTF(bf + ret, size - ret, "%" PRIu64 "%s ", (uint64_t)first->attr.sample_period, top->freq ? "Hz" : ""); } if (!top->display_weighted) { - ret += snprintf(bf + ret, size - ret, "%s", + ret += SNPRINTF(bf + ret, size - ret, "%s", event_name(top->sym_evsel)); } else list_for_each_entry(counter, &top->evlist->entries, node) { - ret += snprintf(bf + ret, size - ret, "%s%s", + ret += SNPRINTF(bf + ret, size - ret, "%s%s", counter->idx ? "/" : "", event_name(counter)); } - ret += snprintf(bf + ret, size - ret, "], "); + ret += SNPRINTF(bf + ret, size - ret, "], "); if (top->target_pid != -1) - ret += snprintf(bf + ret, size - ret, " (target_pid: %d", + ret += SNPRINTF(bf + ret, size - ret, " (target_pid: %d", top->target_pid); else if (top->target_tid != -1) - ret += snprintf(bf + ret, size - ret, " (target_tid: %d", + ret += SNPRINTF(bf + ret, size - ret, " (target_tid: %d", top->target_tid); else - ret += snprintf(bf + ret, size - ret, " (all"); + ret += SNPRINTF(bf + ret, size - ret, " (all"); if (top->cpu_list) - ret += snprintf(bf + ret, size - ret, ", CPU%s: %s)", + ret += SNPRINTF(bf + ret, size - ret, ", CPU%s: %s)", top->evlist->cpus->nr > 1 ? "s" : "", top->cpu_list); else { if (top->target_tid != -1) - ret += snprintf(bf + ret, size - ret, ")"); + ret += SNPRINTF(bf + ret, size - ret, ")"); else - ret += snprintf(bf + ret, size - ret, ", %d CPU%s)", + ret += SNPRINTF(bf + ret, size - ret, ", %d CPU%s)", top->evlist->cpus->nr, top->evlist->cpus->nr > 1 ? "s" : ""); } -- cgit v1.2.3 From 6547250381eb315acff3d52b4872ad775359407c Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Mon, 7 Mar 2011 21:13:41 +0100 Subject: perf top: Don't let events to eat up whole header line Passing multiple events might force out information about pid/tid/cpu. Attached patch leaves 30 characters for this info at the expense of the events' names. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Han Pingtian LKML-Reference: <1299528821-17521-3-git-send-email-jolsa@redhat.com> Signed-off-by: Jiri Olsa Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/top.c | 20 +++++++++++++++++--- 1 file changed, 17 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c index 4f869da4e9c7..75cfe4d45119 100644 --- a/tools/perf/util/top.c +++ b/tools/perf/util/top.c @@ -115,9 +115,23 @@ size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size) if (!top->display_weighted) { ret += SNPRINTF(bf + ret, size - ret, "%s", event_name(top->sym_evsel)); - } else list_for_each_entry(counter, &top->evlist->entries, node) { - ret += SNPRINTF(bf + ret, size - ret, "%s%s", - counter->idx ? "/" : "", event_name(counter)); + } else { + /* + * Don't let events eat all the space. Leaving 30 bytes + * for the rest should be enough. + */ + size_t last_pos = size - 30; + + list_for_each_entry(counter, &top->evlist->entries, node) { + ret += SNPRINTF(bf + ret, size - ret, "%s%s", + counter->idx ? "/" : "", + event_name(counter)); + if (ret > last_pos) { + sprintf(bf + last_pos - 3, ".."); + ret = last_pos - 1; + break; + } + } } ret += SNPRINTF(bf + ret, size - ret, "], "); -- cgit v1.2.3 From a91e5431d54f5359fccb5ec2512f252eb217707e Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 10 Mar 2011 11:15:54 -0300 Subject: perf session: Use evlist/evsel for managing perf.data attributes So that we can reuse things like the id to attr lookup routine (perf_evlist__id2evsel) that uses a hash table instead of the linear lookup done in the older perf_header_attr routines, etc. Also to make evsels/evlist more pervasive an API, simplyfing using the emerging perf lib. cc: Arun Sharma Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Zanussi LKML-Reference: Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-record.c | 101 +++++++------------- tools/perf/builtin-report.c | 4 +- tools/perf/builtin-top.c | 5 +- tools/perf/util/evlist.c | 26 +++-- tools/perf/util/evlist.h | 4 +- tools/perf/util/evsel.c | 21 ++++- tools/perf/util/evsel.h | 9 +- tools/perf/util/header.c | 225 +++++++++++++++++--------------------------- tools/perf/util/header.h | 36 ++----- tools/perf/util/session.c | 74 +-------------- tools/perf/util/session.h | 2 - 11 files changed, 179 insertions(+), 328 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index d40a81e8cc56..6febcc168a8c 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -31,7 +31,6 @@ #include #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) -#define SID(e, x, y) xyarray__entry(e->id, x, y) enum write_mode_t { WRITE_FORCE, @@ -40,7 +39,6 @@ enum write_mode_t { static u64 user_interval = ULLONG_MAX; static u64 default_interval = 0; -static u64 sample_type; static unsigned int page_size; static unsigned int mmap_pages = 128; @@ -160,54 +158,6 @@ static void sig_atexit(void) kill(getpid(), signr); } -static struct perf_header_attr *get_header_attr(struct perf_event_attr *a, int nr) -{ - struct perf_header_attr *h_attr; - - if (nr < session->header.attrs) { - h_attr = session->header.attr[nr]; - } else { - h_attr = perf_header_attr__new(a); - if (h_attr != NULL) - if (perf_header__add_attr(&session->header, h_attr) < 0) { - perf_header_attr__delete(h_attr); - h_attr = NULL; - } - } - - return h_attr; -} - -static void create_counter(struct perf_evsel *evsel, int cpu) -{ - struct perf_event_attr *attr = &evsel->attr; - struct perf_header_attr *h_attr; - struct perf_sample_id *sid; - int thread_index; - - for (thread_index = 0; thread_index < evsel_list->threads->nr; thread_index++) { - h_attr = get_header_attr(attr, evsel->idx); - if (h_attr == NULL) - die("nomem\n"); - - if (!file_new) { - if (memcmp(&h_attr->attr, attr, sizeof(*attr))) { - fprintf(stderr, "incompatible append\n"); - exit(-1); - } - } - - sid = SID(evsel, cpu, thread_index); - if (perf_header_attr__add_id(h_attr, sid->id) < 0) { - pr_warning("Not enough memory to add id\n"); - exit(-1); - } - } - - if (!sample_type) - sample_type = attr->sample_type; -} - static void config_attr(struct perf_evsel *evsel, struct perf_evlist *evlist) { struct perf_event_attr *attr = &evsel->attr; @@ -278,10 +228,28 @@ static void config_attr(struct perf_evsel *evsel, struct perf_evlist *evlist) } } +static bool perf_evlist__equal(struct perf_evlist *evlist, + struct perf_evlist *other) +{ + struct perf_evsel *pos, *pair; + + if (evlist->nr_entries != other->nr_entries) + return false; + + pair = list_entry(other->entries.next, struct perf_evsel, node); + + list_for_each_entry(pos, &evlist->entries, node) { + if (memcmp(&pos->attr, &pair->attr, sizeof(pos->attr) != 0)) + return false; + pair = list_entry(pair->node.next, struct perf_evsel, node); + } + + return true; +} + static void open_counters(struct perf_evlist *evlist) { struct perf_evsel *pos; - int cpu; list_for_each_entry(pos, &evlist->entries, node) { struct perf_event_attr *attr = &pos->attr; @@ -364,10 +332,16 @@ try_again: if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) die("failed to mmap with %d (%s)\n", errno, strerror(errno)); - for (cpu = 0; cpu < evsel_list->cpus->nr; ++cpu) { - list_for_each_entry(pos, &evlist->entries, node) - create_counter(pos, cpu); - } + if (file_new) + session->evlist = evlist; + else { + if (!perf_evlist__equal(session->evlist, evlist)) { + fprintf(stderr, "incompatible append\n"); + exit(-1); + } + } + + perf_session__update_sample_type(session); } static int process_buildids(void) @@ -390,7 +364,7 @@ static void atexit_header(void) if (!no_buildid) process_buildids(); - perf_header__write(&session->header, evsel_list, output, true); + perf_session__write_header(session, evsel_list, output, true); perf_session__delete(session); perf_evlist__delete(evsel_list); symbol__exit(); @@ -524,7 +498,7 @@ static int __cmd_record(int argc, const char **argv) perf_header__set_feat(&session->header, HEADER_BUILD_ID); if (!file_new) { - err = perf_header__read(session, output); + err = perf_session__read_header(session, output); if (err < 0) goto out_delete_session; } @@ -588,8 +562,6 @@ static int __cmd_record(int argc, const char **argv) open_counters(evsel_list); - perf_session__set_sample_type(session, sample_type); - /* * perf_session__delete(session) will be called at atexit_header() */ @@ -600,20 +572,17 @@ static int __cmd_record(int argc, const char **argv) if (err < 0) return err; } else if (file_new) { - err = perf_header__write(&session->header, evsel_list, - output, false); + err = perf_session__write_header(session, evsel_list, + output, false); if (err < 0) return err; } post_processing_offset = lseek(output, 0, SEEK_CUR); - perf_session__set_sample_id_all(session, sample_id_all_avail); - if (pipe_output) { - err = perf_event__synthesize_attrs(&session->header, - process_synthesized_event, - session); + err = perf_session__synthesize_attrs(session, + process_synthesized_event); if (err < 0) { pr_err("Couldn't synthesize attrs.\n"); return err; diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index e9b5d513333a..b1b82009ab9b 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -70,8 +70,8 @@ static int perf_session__add_hist_entry(struct perf_session *session, * FIXME: Propagate this back, but at least we're in a builtin, * where exit() is allowed. ;-) */ - ui__warning("Invalid %s file, contains samples with id not in " - "its header!\n", input_name); + ui__warning("Invalid %s file, contains samples with id %" PRIu64 " not in " + "its header!\n", input_name, sample->id); exit_browser(0); exit(1); } diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 417f757e3cbe..80c9e062bd5b 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -883,7 +883,6 @@ try_again: static int __cmd_top(void) { pthread_t thread; - struct perf_evsel *first; int ret __used; /* * FIXME: perf_session__new should allow passing a O_MMAP, so that all this @@ -900,8 +899,8 @@ static int __cmd_top(void) perf_event__synthesize_threads(perf_event__process, session); start_counters(top.evlist); - first = list_entry(top.evlist->entries.next, struct perf_evsel, node); - perf_session__set_sample_type(session, first->attr.sample_type); + session->evlist = top.evlist; + perf_session__update_sample_type(session); /* Wait for a minimal set of events before starting the snapshot */ poll(top.evlist->pollfd, top.evlist->nr_fds, 100); diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 190c64c6e266..d852cefa20de 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -19,7 +19,7 @@ #include #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) -#define SID(e, x, y) xyarray__entry(e->id, x, y) +#define SID(e, x, y) xyarray__entry(e->sample_id, x, y) void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, struct thread_map *threads) @@ -106,8 +106,9 @@ void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd) evlist->nr_fds++; } -void perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, - int cpu, int thread, u64 id) +static void perf_evlist__id_hash(struct perf_evlist *evlist, + struct perf_evsel *evsel, + int cpu, int thread, u64 id) { int hash; struct perf_sample_id *sid = SID(evsel, cpu, thread); @@ -118,9 +119,16 @@ void perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, hlist_add_head(&sid->node, &evlist->heads[hash]); } -static int perf_evlist__id_hash_fd(struct perf_evlist *evlist, - struct perf_evsel *evsel, - int cpu, int thread, int fd) +void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, u64 id) +{ + perf_evlist__id_hash(evlist, evsel, cpu, thread, id); + evsel->id[evsel->ids++] = id; +} + +static int perf_evlist__id_add_fd(struct perf_evlist *evlist, + struct perf_evsel *evsel, + int cpu, int thread, int fd) { u64 read_data[4] = { 0, }; int id_idx = 1; /* The first entry is the counter value */ @@ -134,7 +142,7 @@ static int perf_evlist__id_hash_fd(struct perf_evlist *evlist, if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) ++id_idx; - perf_evlist__id_hash(evlist, evsel, cpu, thread, read_data[id_idx]); + perf_evlist__id_add(evlist, evsel, cpu, thread, read_data[id_idx]); return 0; } @@ -292,7 +300,7 @@ int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite) list_for_each_entry(evsel, &evlist->entries, node) { if ((evsel->attr.read_format & PERF_FORMAT_ID) && - evsel->id == NULL && + evsel->sample_id == NULL && perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0) return -ENOMEM; @@ -308,7 +316,7 @@ int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite) goto out_unmap; if ((evsel->attr.read_format & PERF_FORMAT_ID) && - perf_evlist__id_hash_fd(evlist, evsel, cpu, thread, fd) < 0) + perf_evlist__id_add_fd(evlist, evsel, cpu, thread, fd) < 0) goto out_unmap; } } diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 078d51256085..8b1cb7a4c5f1 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -38,8 +38,8 @@ void perf_evlist__delete(struct perf_evlist *evlist); void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); int perf_evlist__add_default(struct perf_evlist *evlist); -void perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, - int cpu, int thread, u64 id); +void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, u64 id); int perf_evlist__alloc_pollfd(struct perf_evlist *evlist); void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 8083d5126fca..662596afd7f1 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -41,8 +41,18 @@ int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads) int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads) { - evsel->id = xyarray__new(ncpus, nthreads, sizeof(struct perf_sample_id)); - return evsel->id != NULL ? 0 : -ENOMEM; + evsel->sample_id = xyarray__new(ncpus, nthreads, sizeof(struct perf_sample_id)); + if (evsel->sample_id == NULL) + return -ENOMEM; + + evsel->id = zalloc(ncpus * nthreads * sizeof(u64)); + if (evsel->id == NULL) { + xyarray__delete(evsel->sample_id); + evsel->sample_id = NULL; + return -ENOMEM; + } + + return 0; } int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus) @@ -60,7 +70,9 @@ void perf_evsel__free_fd(struct perf_evsel *evsel) void perf_evsel__free_id(struct perf_evsel *evsel) { - xyarray__delete(evsel->id); + xyarray__delete(evsel->sample_id); + evsel->sample_id = NULL; + free(evsel->id); evsel->id = NULL; } @@ -79,7 +91,8 @@ void perf_evsel__exit(struct perf_evsel *evsel) { assert(list_empty(&evsel->node)); xyarray__delete(evsel->fd); - xyarray__delete(evsel->id); + xyarray__delete(evsel->sample_id); + free(evsel->id); } void perf_evsel__delete(struct perf_evsel *evsel) diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 281b60e5fc7b..6710ab538342 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -49,12 +49,17 @@ struct perf_evsel { struct perf_event_attr attr; char *filter; struct xyarray *fd; - struct xyarray *id; + struct xyarray *sample_id; + u64 *id; struct perf_counts *counts; int idx; + int ids; struct hists hists; char *name; - void *priv; + union { + void *priv; + off_t id_offset; + }; struct cgroup_sel *cgrp; }; diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 108b0db7bbef..40b10e4d3927 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -9,6 +9,7 @@ #include #include "evlist.h" +#include "evsel.h" #include "util.h" #include "header.h" #include "../perf.h" @@ -19,89 +20,6 @@ static bool no_buildid_cache = false; -/* - * Create new perf.data header attribute: - */ -struct perf_header_attr *perf_header_attr__new(struct perf_event_attr *attr) -{ - struct perf_header_attr *self = malloc(sizeof(*self)); - - if (self != NULL) { - self->attr = *attr; - self->ids = 0; - self->size = 1; - self->id = malloc(sizeof(u64)); - if (self->id == NULL) { - free(self); - self = NULL; - } - } - - return self; -} - -void perf_header_attr__delete(struct perf_header_attr *self) -{ - free(self->id); - free(self); -} - -int perf_header_attr__add_id(struct perf_header_attr *self, u64 id) -{ - int pos = self->ids; - - self->ids++; - if (self->ids > self->size) { - int nsize = self->size * 2; - u64 *nid = realloc(self->id, nsize * sizeof(u64)); - - if (nid == NULL) - return -1; - - self->size = nsize; - self->id = nid; - } - self->id[pos] = id; - return 0; -} - -int perf_header__init(struct perf_header *self) -{ - self->size = 1; - self->attr = malloc(sizeof(void *)); - return self->attr == NULL ? -ENOMEM : 0; -} - -void perf_header__exit(struct perf_header *self) -{ - int i; - for (i = 0; i < self->attrs; ++i) - perf_header_attr__delete(self->attr[i]); - free(self->attr); -} - -int perf_header__add_attr(struct perf_header *self, - struct perf_header_attr *attr) -{ - if (self->frozen) - return -1; - - if (self->attrs == self->size) { - int nsize = self->size * 2; - struct perf_header_attr **nattr; - - nattr = realloc(self->attr, nsize * sizeof(void *)); - if (nattr == NULL) - return -1; - - self->size = nsize; - self->attr = nattr; - } - - self->attr[self->attrs++] = attr; - return 0; -} - static int event_count; static struct perf_trace_event_type *events; @@ -515,33 +433,41 @@ int perf_header__write_pipe(int fd) return 0; } -int perf_header__write(struct perf_header *self, struct perf_evlist *evlist, - int fd, bool at_exit) +int perf_session__write_header(struct perf_session *session, + struct perf_evlist *evlist, + int fd, bool at_exit) { struct perf_file_header f_header; struct perf_file_attr f_attr; - struct perf_header_attr *attr; - int i, err; + struct perf_header *self = &session->header; + struct perf_evsel *attr, *pair = NULL; + int err; lseek(fd, sizeof(f_header), SEEK_SET); - for (i = 0; i < self->attrs; i++) { - attr = self->attr[i]; + if (session->evlist != evlist) + pair = list_entry(session->evlist->entries.next, struct perf_evsel, node); + list_for_each_entry(attr, &evlist->entries, node) { attr->id_offset = lseek(fd, 0, SEEK_CUR); err = do_write(fd, attr->id, attr->ids * sizeof(u64)); if (err < 0) { +out_err_write: pr_debug("failed to write perf header\n"); return err; } + if (session->evlist != evlist) { + err = do_write(fd, pair->id, pair->ids * sizeof(u64)); + if (err < 0) + goto out_err_write; + attr->ids += pair->ids; + pair = list_entry(pair->node.next, struct perf_evsel, node); + } } - self->attr_offset = lseek(fd, 0, SEEK_CUR); - for (i = 0; i < self->attrs; i++) { - attr = self->attr[i]; - + list_for_each_entry(attr, &evlist->entries, node) { f_attr = (struct perf_file_attr){ .attr = attr->attr, .ids = { @@ -580,7 +506,7 @@ int perf_header__write(struct perf_header *self, struct perf_evlist *evlist, .attr_size = sizeof(f_attr), .attrs = { .offset = self->attr_offset, - .size = self->attrs * sizeof(f_attr), + .size = evlist->nr_entries * sizeof(f_attr), }, .data = { .offset = self->data_offset, @@ -861,7 +787,7 @@ static int perf_header__read_pipe(struct perf_session *session, int fd) return 0; } -int perf_header__read(struct perf_session *session, int fd) +int perf_session__read_header(struct perf_session *session, int fd) { struct perf_header *self = &session->header; struct perf_file_header f_header; @@ -869,6 +795,10 @@ int perf_header__read(struct perf_session *session, int fd) u64 f_id; int nr_attrs, nr_ids, i, j; + session->evlist = perf_evlist__new(NULL, NULL); + if (session->evlist == NULL) + return -ENOMEM; + if (session->fd_pipe) return perf_header__read_pipe(session, fd); @@ -881,33 +811,39 @@ int perf_header__read(struct perf_session *session, int fd) lseek(fd, f_header.attrs.offset, SEEK_SET); for (i = 0; i < nr_attrs; i++) { - struct perf_header_attr *attr; + struct perf_evsel *evsel; off_t tmp; if (perf_header__getbuffer64(self, fd, &f_attr, sizeof(f_attr))) goto out_errno; tmp = lseek(fd, 0, SEEK_CUR); + evsel = perf_evsel__new(&f_attr.attr, i); - attr = perf_header_attr__new(&f_attr.attr); - if (attr == NULL) - return -ENOMEM; + if (evsel == NULL) + goto out_delete_evlist; + /* + * Do it before so that if perf_evsel__alloc_id fails, this + * entry gets purged too at perf_evlist__delete(). + */ + perf_evlist__add(session->evlist, evsel); nr_ids = f_attr.ids.size / sizeof(u64); + /* + * We don't have the cpu and thread maps on the header, so + * for allocating the perf_sample_id table we fake 1 cpu and + * hattr->ids threads. + */ + if (perf_evsel__alloc_id(evsel, 1, nr_ids)) + goto out_delete_evlist; + lseek(fd, f_attr.ids.offset, SEEK_SET); for (j = 0; j < nr_ids; j++) { if (perf_header__getbuffer64(self, fd, &f_id, sizeof(f_id))) goto out_errno; - if (perf_header_attr__add_id(attr, f_id) < 0) { - perf_header_attr__delete(attr); - return -ENOMEM; - } - } - if (perf_header__add_attr(self, attr) < 0) { - perf_header_attr__delete(attr); - return -ENOMEM; + perf_evlist__id_add(session->evlist, evsel, 0, j, f_id); } lseek(fd, tmp, SEEK_SET); @@ -932,37 +868,38 @@ int perf_header__read(struct perf_session *session, int fd) return 0; out_errno: return -errno; + +out_delete_evlist: + perf_evlist__delete(session->evlist); + session->evlist = NULL; + return -ENOMEM; } -u64 perf_header__sample_type(struct perf_header *header) +u64 perf_evlist__sample_type(struct perf_evlist *evlist) { + struct perf_evsel *pos; u64 type = 0; - int i; - - for (i = 0; i < header->attrs; i++) { - struct perf_header_attr *attr = header->attr[i]; + list_for_each_entry(pos, &evlist->entries, node) { if (!type) - type = attr->attr.sample_type; - else if (type != attr->attr.sample_type) + type = pos->attr.sample_type; + else if (type != pos->attr.sample_type) die("non matching sample_type"); } return type; } -bool perf_header__sample_id_all(const struct perf_header *header) +bool perf_evlist__sample_id_all(const struct perf_evlist *evlist) { bool value = false, first = true; - int i; - - for (i = 0; i < header->attrs; i++) { - struct perf_header_attr *attr = header->attr[i]; + struct perf_evsel *pos; + list_for_each_entry(pos, &evlist->entries, node) { if (first) { - value = attr->attr.sample_id_all; + value = pos->attr.sample_id_all; first = false; - } else if (value != attr->attr.sample_id_all) + } else if (value != pos->attr.sample_id_all) die("non matching sample_id_all"); } @@ -1000,16 +937,13 @@ int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, return err; } -int perf_event__synthesize_attrs(struct perf_header *self, - perf_event__handler_t process, - struct perf_session *session) +int perf_session__synthesize_attrs(struct perf_session *session, + perf_event__handler_t process) { - struct perf_header_attr *attr; - int i, err = 0; - - for (i = 0; i < self->attrs; i++) { - attr = self->attr[i]; + struct perf_evsel *attr; + int err = 0; + list_for_each_entry(attr, &session->evlist->entries, node) { err = perf_event__synthesize_attr(&attr->attr, attr->ids, attr->id, process, session); if (err) { @@ -1024,27 +958,36 @@ int perf_event__synthesize_attrs(struct perf_header *self, int perf_event__process_attr(union perf_event *event, struct perf_session *session) { - struct perf_header_attr *attr; unsigned int i, ids, n_ids; + struct perf_evsel *evsel; - attr = perf_header_attr__new(&event->attr.attr); - if (attr == NULL) + if (session->evlist == NULL) { + session->evlist = perf_evlist__new(NULL, NULL); + if (session->evlist == NULL) + return -ENOMEM; + } + + evsel = perf_evsel__new(&event->attr.attr, + session->evlist->nr_entries); + if (evsel == NULL) return -ENOMEM; + perf_evlist__add(session->evlist, evsel); + ids = event->header.size; ids -= (void *)&event->attr.id - (void *)event; n_ids = ids / sizeof(u64); + /* + * We don't have the cpu and thread maps on the header, so + * for allocating the perf_sample_id table we fake 1 cpu and + * hattr->ids threads. + */ + if (perf_evsel__alloc_id(evsel, 1, n_ids)) + return -ENOMEM; for (i = 0; i < n_ids; i++) { - if (perf_header_attr__add_id(attr, event->attr.id[i]) < 0) { - perf_header_attr__delete(attr); - return -ENOMEM; - } - } - - if (perf_header__add_attr(&session->header, attr) < 0) { - perf_header_attr__delete(attr); - return -ENOMEM; + perf_evlist__id_add(session->evlist, evsel, 0, i, + event->attr.id[i]); } perf_session__update_sample_type(session); diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 2fab13348aab..4cc267559bb6 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -9,13 +9,6 @@ #include -struct perf_header_attr { - struct perf_event_attr attr; - int ids, size; - u64 *id; - off_t id_offset; -}; - enum { HEADER_TRACE_INFO = 1, HEADER_BUILD_ID, @@ -51,9 +44,7 @@ int perf_file_header__read(struct perf_file_header *self, struct perf_header { int frozen; - int attrs, size; bool needs_swap; - struct perf_header_attr **attr; s64 attr_offset; u64 data_offset; u64 data_size; @@ -62,29 +53,19 @@ struct perf_header { DECLARE_BITMAP(adds_features, HEADER_FEAT_BITS); }; -int perf_header__init(struct perf_header *self); -void perf_header__exit(struct perf_header *self); - struct perf_evlist; -int perf_header__read(struct perf_session *session, int fd); -int perf_header__write(struct perf_header *self, struct perf_evlist *evlist, - int fd, bool at_exit); +int perf_session__read_header(struct perf_session *session, int fd); +int perf_session__write_header(struct perf_session *session, + struct perf_evlist *evlist, + int fd, bool at_exit); int perf_header__write_pipe(int fd); -int perf_header__add_attr(struct perf_header *self, - struct perf_header_attr *attr); - int perf_header__push_event(u64 id, const char *name); char *perf_header__find_event(u64 id); -struct perf_header_attr *perf_header_attr__new(struct perf_event_attr *attr); -void perf_header_attr__delete(struct perf_header_attr *self); - -int perf_header_attr__add_id(struct perf_header_attr *self, u64 id); - -u64 perf_header__sample_type(struct perf_header *header); -bool perf_header__sample_id_all(const struct perf_header *header); +u64 perf_evlist__sample_type(struct perf_evlist *evlist); +bool perf_evlist__sample_id_all(const struct perf_evlist *evlist); void perf_header__set_feat(struct perf_header *self, int feat); void perf_header__clear_feat(struct perf_header *self, int feat); bool perf_header__has_feat(const struct perf_header *self, int feat); @@ -101,9 +82,8 @@ int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir); int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, perf_event__handler_t process, struct perf_session *session); -int perf_event__synthesize_attrs(struct perf_header *self, - perf_event__handler_t process, - struct perf_session *session); +int perf_session__synthesize_attrs(struct perf_session *session, + perf_event__handler_t process); int perf_event__process_attr(union perf_event *event, struct perf_session *session); int perf_event__synthesize_event_type(u64 event_id, char *name, diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 0d414199889d..f26639fa0fb3 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -13,46 +13,6 @@ #include "sort.h" #include "util.h" -static int perf_session__read_evlist(struct perf_session *session) -{ - int i, j; - - session->evlist = perf_evlist__new(NULL, NULL); - if (session->evlist == NULL) - return -ENOMEM; - - for (i = 0; i < session->header.attrs; ++i) { - struct perf_header_attr *hattr = session->header.attr[i]; - struct perf_evsel *evsel = perf_evsel__new(&hattr->attr, i); - - if (evsel == NULL) - goto out_delete_evlist; - /* - * Do it before so that if perf_evsel__alloc_id fails, this - * entry gets purged too at perf_evlist__delete(). - */ - perf_evlist__add(session->evlist, evsel); - /* - * We don't have the cpu and thread maps on the header, so - * for allocating the perf_sample_id table we fake 1 cpu and - * hattr->ids threads. - */ - if (perf_evsel__alloc_id(evsel, 1, hattr->ids)) - goto out_delete_evlist; - - for (j = 0; j < hattr->ids; ++j) - perf_evlist__id_hash(session->evlist, evsel, 0, j, - hattr->id[j]); - } - - return 0; - -out_delete_evlist: - perf_evlist__delete(session->evlist); - session->evlist = NULL; - return -ENOMEM; -} - static int perf_session__open(struct perf_session *self, bool force) { struct stat input_stat; @@ -61,7 +21,7 @@ static int perf_session__open(struct perf_session *self, bool force) self->fd_pipe = true; self->fd = STDIN_FILENO; - if (perf_header__read(self, self->fd) < 0) + if (perf_session__read_header(self, self->fd) < 0) pr_err("incompatible file format"); return 0; @@ -93,16 +53,11 @@ static int perf_session__open(struct perf_session *self, bool force) goto out_close; } - if (perf_header__read(self, self->fd) < 0) { + if (perf_session__read_header(self, self->fd) < 0) { pr_err("incompatible file format"); goto out_close; } - if (perf_session__read_evlist(self) < 0) { - pr_err("Not enough memory to read the event selector list\n"); - goto out_close; - } - self->size = input_stat.st_size; return 0; @@ -139,21 +94,10 @@ out: session->id_hdr_size = size; } -void perf_session__set_sample_id_all(struct perf_session *session, bool value) -{ - session->sample_id_all = value; - perf_session__id_header_size(session); -} - -void perf_session__set_sample_type(struct perf_session *session, u64 type) -{ - session->sample_type = type; -} - void perf_session__update_sample_type(struct perf_session *self) { - self->sample_type = perf_header__sample_type(&self->header); - self->sample_id_all = perf_header__sample_id_all(&self->header); + self->sample_type = perf_evlist__sample_type(self->evlist); + self->sample_id_all = perf_evlist__sample_id_all(self->evlist); perf_session__id_header_size(self); } @@ -182,9 +126,6 @@ struct perf_session *perf_session__new(const char *filename, int mode, if (self == NULL) goto out; - if (perf_header__init(&self->header) < 0) - goto out_free; - memcpy(self->filename, filename, len); self->threads = RB_ROOT; INIT_LIST_HEAD(&self->dead_threads); @@ -208,6 +149,7 @@ struct perf_session *perf_session__new(const char *filename, int mode, if (mode == O_RDONLY) { if (perf_session__open(self, force) < 0) goto out_delete; + perf_session__update_sample_type(self); } else if (mode == O_WRONLY) { /* * In O_RDONLY mode this will be performed when reading the @@ -217,8 +159,6 @@ struct perf_session *perf_session__new(const char *filename, int mode, goto out_delete; } - perf_session__update_sample_type(self); - if (ops && ops->ordering_requires_timestamps && ops->ordered_samples && !self->sample_id_all) { dump_printf("WARNING: No sample_id_all support, falling back to unordered processing\n"); @@ -227,9 +167,6 @@ struct perf_session *perf_session__new(const char *filename, int mode, out: return self; -out_free: - free(self); - return NULL; out_delete: perf_session__delete(self); return NULL; @@ -260,7 +197,6 @@ static void perf_session__delete_threads(struct perf_session *self) void perf_session__delete(struct perf_session *self) { - perf_header__exit(&self->header); perf_session__destroy_kernel_maps(self); perf_session__delete_dead_threads(self); perf_session__delete_threads(self); diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 05dd7bcb9453..b5b148b0aaca 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -112,8 +112,6 @@ void mem_bswap_64(void *src, int byte_size); int perf_session__create_kernel_maps(struct perf_session *self); void perf_session__update_sample_type(struct perf_session *self); -void perf_session__set_sample_id_all(struct perf_session *session, bool value); -void perf_session__set_sample_type(struct perf_session *session, u64 type); void perf_session__remove_thread(struct perf_session *self, struct thread *th); static inline -- cgit v1.2.3 From 1c0b04d10bbe35279c50e3b36cf5b8ec2a0050d8 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 9 Mar 2011 08:13:19 -0300 Subject: perf header: Stop using 'self' Stop using this python/OOP convention, doesn't really helps. Will do more from time to time till we get it cleaned up in all of tools/perf. Suggested-by: Thomas Gleixner LKML-Reference: Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Thomas Gleixner Cc: Tom Zanussi Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 199 +++++++++++++++++++++++------------------------ tools/perf/util/header.h | 12 +-- 2 files changed, 105 insertions(+), 106 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 40b10e4d3927..5a72d421e211 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -66,19 +66,19 @@ struct perf_file_attr { struct perf_file_section ids; }; -void perf_header__set_feat(struct perf_header *self, int feat) +void perf_header__set_feat(struct perf_header *header, int feat) { - set_bit(feat, self->adds_features); + set_bit(feat, header->adds_features); } -void perf_header__clear_feat(struct perf_header *self, int feat) +void perf_header__clear_feat(struct perf_header *header, int feat) { - clear_bit(feat, self->adds_features); + clear_bit(feat, header->adds_features); } -bool perf_header__has_feat(const struct perf_header *self, int feat) +bool perf_header__has_feat(const struct perf_header *header, int feat) { - return test_bit(feat, self->adds_features); + return test_bit(feat, header->adds_features); } static int do_write(int fd, const void *buf, size_t size) @@ -147,22 +147,22 @@ static int __dsos__write_buildid_table(struct list_head *head, pid_t pid, return 0; } -static int machine__write_buildid_table(struct machine *self, int fd) +static int machine__write_buildid_table(struct machine *machine, int fd) { int err; u16 kmisc = PERF_RECORD_MISC_KERNEL, umisc = PERF_RECORD_MISC_USER; - if (!machine__is_host(self)) { + if (!machine__is_host(machine)) { kmisc = PERF_RECORD_MISC_GUEST_KERNEL; umisc = PERF_RECORD_MISC_GUEST_USER; } - err = __dsos__write_buildid_table(&self->kernel_dsos, self->pid, + err = __dsos__write_buildid_table(&machine->kernel_dsos, machine->pid, kmisc, fd); if (err == 0) - err = __dsos__write_buildid_table(&self->user_dsos, - self->pid, umisc, fd); + err = __dsos__write_buildid_table(&machine->user_dsos, + machine->pid, umisc, fd); return err; } @@ -280,12 +280,12 @@ out_free: return err; } -static int dso__cache_build_id(struct dso *self, const char *debugdir) +static int dso__cache_build_id(struct dso *dso, const char *debugdir) { - bool is_kallsyms = self->kernel && self->long_name[0] != '/'; + bool is_kallsyms = dso->kernel && dso->long_name[0] != '/'; - return build_id_cache__add_b(self->build_id, sizeof(self->build_id), - self->long_name, debugdir, is_kallsyms); + return build_id_cache__add_b(dso->build_id, sizeof(dso->build_id), + dso->long_name, debugdir, is_kallsyms); } static int __dsos__cache_build_ids(struct list_head *head, const char *debugdir) @@ -300,14 +300,14 @@ static int __dsos__cache_build_ids(struct list_head *head, const char *debugdir) return err; } -static int machine__cache_build_ids(struct machine *self, const char *debugdir) +static int machine__cache_build_ids(struct machine *machine, const char *debugdir) { - int ret = __dsos__cache_build_ids(&self->kernel_dsos, debugdir); - ret |= __dsos__cache_build_ids(&self->user_dsos, debugdir); + int ret = __dsos__cache_build_ids(&machine->kernel_dsos, debugdir); + ret |= __dsos__cache_build_ids(&machine->user_dsos, debugdir); return ret; } -static int perf_session__cache_build_ids(struct perf_session *self) +static int perf_session__cache_build_ids(struct perf_session *session) { struct rb_node *nd; int ret; @@ -318,28 +318,28 @@ static int perf_session__cache_build_ids(struct perf_session *self) if (mkdir(debugdir, 0755) != 0 && errno != EEXIST) return -1; - ret = machine__cache_build_ids(&self->host_machine, debugdir); + ret = machine__cache_build_ids(&session->host_machine, debugdir); - for (nd = rb_first(&self->machines); nd; nd = rb_next(nd)) { + for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { struct machine *pos = rb_entry(nd, struct machine, rb_node); ret |= machine__cache_build_ids(pos, debugdir); } return ret ? -1 : 0; } -static bool machine__read_build_ids(struct machine *self, bool with_hits) +static bool machine__read_build_ids(struct machine *machine, bool with_hits) { - bool ret = __dsos__read_build_ids(&self->kernel_dsos, with_hits); - ret |= __dsos__read_build_ids(&self->user_dsos, with_hits); + bool ret = __dsos__read_build_ids(&machine->kernel_dsos, with_hits); + ret |= __dsos__read_build_ids(&machine->user_dsos, with_hits); return ret; } -static bool perf_session__read_build_ids(struct perf_session *self, bool with_hits) +static bool perf_session__read_build_ids(struct perf_session *session, bool with_hits) { struct rb_node *nd; - bool ret = machine__read_build_ids(&self->host_machine, with_hits); + bool ret = machine__read_build_ids(&session->host_machine, with_hits); - for (nd = rb_first(&self->machines); nd; nd = rb_next(nd)) { + for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { struct machine *pos = rb_entry(nd, struct machine, rb_node); ret |= machine__read_build_ids(pos, with_hits); } @@ -347,7 +347,7 @@ static bool perf_session__read_build_ids(struct perf_session *self, bool with_hi return ret; } -static int perf_header__adds_write(struct perf_header *self, +static int perf_header__adds_write(struct perf_header *header, struct perf_evlist *evlist, int fd) { int nr_sections; @@ -357,13 +357,13 @@ static int perf_header__adds_write(struct perf_header *self, u64 sec_start; int idx = 0, err; - session = container_of(self, struct perf_session, header); + session = container_of(header, struct perf_session, header); - if (perf_header__has_feat(self, HEADER_BUILD_ID && + if (perf_header__has_feat(header, HEADER_BUILD_ID && !perf_session__read_build_ids(session, true))) - perf_header__clear_feat(self, HEADER_BUILD_ID); + perf_header__clear_feat(header, HEADER_BUILD_ID); - nr_sections = bitmap_weight(self->adds_features, HEADER_FEAT_BITS); + nr_sections = bitmap_weight(header->adds_features, HEADER_FEAT_BITS); if (!nr_sections) return 0; @@ -373,10 +373,10 @@ static int perf_header__adds_write(struct perf_header *self, sec_size = sizeof(*feat_sec) * nr_sections; - sec_start = self->data_offset + self->data_size; + sec_start = header->data_offset + header->data_size; lseek(fd, sec_start + sec_size, SEEK_SET); - if (perf_header__has_feat(self, HEADER_TRACE_INFO)) { + if (perf_header__has_feat(header, HEADER_TRACE_INFO)) { struct perf_file_section *trace_sec; trace_sec = &feat_sec[idx++]; @@ -387,14 +387,14 @@ static int perf_header__adds_write(struct perf_header *self, trace_sec->size = lseek(fd, 0, SEEK_CUR) - trace_sec->offset; } - if (perf_header__has_feat(self, HEADER_BUILD_ID)) { + if (perf_header__has_feat(header, HEADER_BUILD_ID)) { struct perf_file_section *buildid_sec; buildid_sec = &feat_sec[idx++]; /* Write build-ids */ buildid_sec->offset = lseek(fd, 0, SEEK_CUR); - err = dsos__write_buildid_table(self, fd); + err = dsos__write_buildid_table(header, fd); if (err < 0) { pr_debug("failed to write buildid table\n"); goto out_free; @@ -439,7 +439,7 @@ int perf_session__write_header(struct perf_session *session, { struct perf_file_header f_header; struct perf_file_attr f_attr; - struct perf_header *self = &session->header; + struct perf_header *header = &session->header; struct perf_evsel *attr, *pair = NULL; int err; @@ -465,7 +465,7 @@ out_err_write: } } - self->attr_offset = lseek(fd, 0, SEEK_CUR); + header->attr_offset = lseek(fd, 0, SEEK_CUR); list_for_each_entry(attr, &evlist->entries, node) { f_attr = (struct perf_file_attr){ @@ -482,20 +482,20 @@ out_err_write: } } - self->event_offset = lseek(fd, 0, SEEK_CUR); - self->event_size = event_count * sizeof(struct perf_trace_event_type); + header->event_offset = lseek(fd, 0, SEEK_CUR); + header->event_size = event_count * sizeof(struct perf_trace_event_type); if (events) { - err = do_write(fd, events, self->event_size); + err = do_write(fd, events, header->event_size); if (err < 0) { pr_debug("failed to write perf header events\n"); return err; } } - self->data_offset = lseek(fd, 0, SEEK_CUR); + header->data_offset = lseek(fd, 0, SEEK_CUR); if (at_exit) { - err = perf_header__adds_write(self, evlist, fd); + err = perf_header__adds_write(header, evlist, fd); if (err < 0) return err; } @@ -505,20 +505,20 @@ out_err_write: .size = sizeof(f_header), .attr_size = sizeof(f_attr), .attrs = { - .offset = self->attr_offset, + .offset = header->attr_offset, .size = evlist->nr_entries * sizeof(f_attr), }, .data = { - .offset = self->data_offset, - .size = self->data_size, + .offset = header->data_offset, + .size = header->data_size, }, .event_types = { - .offset = self->event_offset, - .size = self->event_size, + .offset = header->event_offset, + .size = header->event_size, }, }; - memcpy(&f_header.adds_features, &self->adds_features, sizeof(self->adds_features)); + memcpy(&f_header.adds_features, &header->adds_features, sizeof(header->adds_features)); lseek(fd, 0, SEEK_SET); err = do_write(fd, &f_header, sizeof(f_header)); @@ -526,26 +526,26 @@ out_err_write: pr_debug("failed to write perf header\n"); return err; } - lseek(fd, self->data_offset + self->data_size, SEEK_SET); + lseek(fd, header->data_offset + header->data_size, SEEK_SET); - self->frozen = 1; + header->frozen = 1; return 0; } -static int perf_header__getbuffer64(struct perf_header *self, +static int perf_header__getbuffer64(struct perf_header *header, int fd, void *buf, size_t size) { if (readn(fd, buf, size) <= 0) return -1; - if (self->needs_swap) + if (header->needs_swap) mem_bswap_64(buf, size); return 0; } -int perf_header__process_sections(struct perf_header *self, int fd, - int (*process)(struct perf_file_section *self, +int perf_header__process_sections(struct perf_header *header, int fd, + int (*process)(struct perf_file_section *section, struct perf_header *ph, int feat, int fd)) { @@ -555,7 +555,7 @@ int perf_header__process_sections(struct perf_header *self, int fd, int idx = 0; int err = -1, feat = 1; - nr_sections = bitmap_weight(self->adds_features, HEADER_FEAT_BITS); + nr_sections = bitmap_weight(header->adds_features, HEADER_FEAT_BITS); if (!nr_sections) return 0; @@ -565,17 +565,17 @@ int perf_header__process_sections(struct perf_header *self, int fd, sec_size = sizeof(*feat_sec) * nr_sections; - lseek(fd, self->data_offset + self->data_size, SEEK_SET); + lseek(fd, header->data_offset + header->data_size, SEEK_SET); - if (perf_header__getbuffer64(self, fd, feat_sec, sec_size)) + if (perf_header__getbuffer64(header, fd, feat_sec, sec_size)) goto out_free; err = 0; while (idx < nr_sections && feat < HEADER_LAST_FEATURE) { - if (perf_header__has_feat(self, feat)) { + if (perf_header__has_feat(header, feat)) { struct perf_file_section *sec = &feat_sec[idx++]; - err = process(sec, self, feat, fd); + err = process(sec, header, feat, fd); if (err < 0) break; } @@ -586,35 +586,35 @@ out_free: return err; } -int perf_file_header__read(struct perf_file_header *self, +int perf_file_header__read(struct perf_file_header *header, struct perf_header *ph, int fd) { lseek(fd, 0, SEEK_SET); - if (readn(fd, self, sizeof(*self)) <= 0 || - memcmp(&self->magic, __perf_magic, sizeof(self->magic))) + if (readn(fd, header, sizeof(*header)) <= 0 || + memcmp(&header->magic, __perf_magic, sizeof(header->magic))) return -1; - if (self->attr_size != sizeof(struct perf_file_attr)) { - u64 attr_size = bswap_64(self->attr_size); + if (header->attr_size != sizeof(struct perf_file_attr)) { + u64 attr_size = bswap_64(header->attr_size); if (attr_size != sizeof(struct perf_file_attr)) return -1; - mem_bswap_64(self, offsetof(struct perf_file_header, + mem_bswap_64(header, offsetof(struct perf_file_header, adds_features)); ph->needs_swap = true; } - if (self->size != sizeof(*self)) { + if (header->size != sizeof(*header)) { /* Support the previous format */ - if (self->size == offsetof(typeof(*self), adds_features)) - bitmap_zero(self->adds_features, HEADER_FEAT_BITS); + if (header->size == offsetof(typeof(*header), adds_features)) + bitmap_zero(header->adds_features, HEADER_FEAT_BITS); else return -1; } - memcpy(&ph->adds_features, &self->adds_features, + memcpy(&ph->adds_features, &header->adds_features, sizeof(ph->adds_features)); /* * FIXME: hack that assumes that if we need swap the perf.data file @@ -628,10 +628,10 @@ int perf_file_header__read(struct perf_file_header *self, perf_header__set_feat(ph, HEADER_BUILD_ID); } - ph->event_offset = self->event_types.offset; - ph->event_size = self->event_types.size; - ph->data_offset = self->data.offset; - ph->data_size = self->data.size; + ph->event_offset = header->event_types.offset; + ph->event_size = header->event_types.size; + ph->data_offset = header->data.offset; + ph->data_size = header->data.size; return 0; } @@ -690,11 +690,10 @@ out: return err; } -static int perf_header__read_build_ids(struct perf_header *self, - int input, u64 offset, u64 size) +static int perf_header__read_build_ids(struct perf_header *header, + int input, u64 offset, u64 size) { - struct perf_session *session = container_of(self, - struct perf_session, header); + struct perf_session *session = container_of(header, struct perf_session, header); struct build_id_event bev; char filename[PATH_MAX]; u64 limit = offset + size; @@ -706,7 +705,7 @@ static int perf_header__read_build_ids(struct perf_header *self, if (read(input, &bev, sizeof(bev)) != sizeof(bev)) goto out; - if (self->needs_swap) + if (header->needs_swap) perf_event_header__bswap(&bev.header); len = bev.header.size - sizeof(bev); @@ -722,13 +721,13 @@ out: return err; } -static int perf_file_section__process(struct perf_file_section *self, +static int perf_file_section__process(struct perf_file_section *section, struct perf_header *ph, int feat, int fd) { - if (lseek(fd, self->offset, SEEK_SET) == (off_t)-1) { + if (lseek(fd, section->offset, SEEK_SET) == (off_t)-1) { pr_debug("Failed to lseek to %" PRIu64 " offset for feature " - "%d, continuing...\n", self->offset, feat); + "%d, continuing...\n", section->offset, feat); return 0; } @@ -738,7 +737,7 @@ static int perf_file_section__process(struct perf_file_section *self, break; case HEADER_BUILD_ID: - if (perf_header__read_build_ids(ph, fd, self->offset, self->size)) + if (perf_header__read_build_ids(ph, fd, section->offset, section->size)) pr_debug("Failed to read buildids, continuing...\n"); break; default: @@ -748,21 +747,21 @@ static int perf_file_section__process(struct perf_file_section *self, return 0; } -static int perf_file_header__read_pipe(struct perf_pipe_file_header *self, +static int perf_file_header__read_pipe(struct perf_pipe_file_header *header, struct perf_header *ph, int fd, bool repipe) { - if (readn(fd, self, sizeof(*self)) <= 0 || - memcmp(&self->magic, __perf_magic, sizeof(self->magic))) + if (readn(fd, header, sizeof(*header)) <= 0 || + memcmp(&header->magic, __perf_magic, sizeof(header->magic))) return -1; - if (repipe && do_write(STDOUT_FILENO, self, sizeof(*self)) < 0) + if (repipe && do_write(STDOUT_FILENO, header, sizeof(*header)) < 0) return -1; - if (self->size != sizeof(*self)) { - u64 size = bswap_64(self->size); + if (header->size != sizeof(*header)) { + u64 size = bswap_64(header->size); - if (size != sizeof(*self)) + if (size != sizeof(*header)) return -1; ph->needs_swap = true; @@ -773,10 +772,10 @@ static int perf_file_header__read_pipe(struct perf_pipe_file_header *self, static int perf_header__read_pipe(struct perf_session *session, int fd) { - struct perf_header *self = &session->header; + struct perf_header *header = &session->header; struct perf_pipe_file_header f_header; - if (perf_file_header__read_pipe(&f_header, self, fd, + if (perf_file_header__read_pipe(&f_header, header, fd, session->repipe) < 0) { pr_debug("incompatible file format\n"); return -EINVAL; @@ -789,7 +788,7 @@ static int perf_header__read_pipe(struct perf_session *session, int fd) int perf_session__read_header(struct perf_session *session, int fd) { - struct perf_header *self = &session->header; + struct perf_header *header = &session->header; struct perf_file_header f_header; struct perf_file_attr f_attr; u64 f_id; @@ -802,7 +801,7 @@ int perf_session__read_header(struct perf_session *session, int fd) if (session->fd_pipe) return perf_header__read_pipe(session, fd); - if (perf_file_header__read(&f_header, self, fd) < 0) { + if (perf_file_header__read(&f_header, header, fd) < 0) { pr_debug("incompatible file format\n"); return -EINVAL; } @@ -814,7 +813,7 @@ int perf_session__read_header(struct perf_session *session, int fd) struct perf_evsel *evsel; off_t tmp; - if (perf_header__getbuffer64(self, fd, &f_attr, sizeof(f_attr))) + if (perf_header__getbuffer64(header, fd, &f_attr, sizeof(f_attr))) goto out_errno; tmp = lseek(fd, 0, SEEK_CUR); @@ -840,7 +839,7 @@ int perf_session__read_header(struct perf_session *session, int fd) lseek(fd, f_attr.ids.offset, SEEK_SET); for (j = 0; j < nr_ids; j++) { - if (perf_header__getbuffer64(self, fd, &f_id, sizeof(f_id))) + if (perf_header__getbuffer64(header, fd, &f_id, sizeof(f_id))) goto out_errno; perf_evlist__id_add(session->evlist, evsel, 0, j, f_id); @@ -854,17 +853,17 @@ int perf_session__read_header(struct perf_session *session, int fd) events = malloc(f_header.event_types.size); if (events == NULL) return -ENOMEM; - if (perf_header__getbuffer64(self, fd, events, + if (perf_header__getbuffer64(header, fd, events, f_header.event_types.size)) goto out_errno; event_count = f_header.event_types.size / sizeof(struct perf_trace_event_type); } - perf_header__process_sections(self, fd, perf_file_section__process); + perf_header__process_sections(header, fd, perf_file_section__process); - lseek(fd, self->data_offset, SEEK_SET); + lseek(fd, header->data_offset, SEEK_SET); - self->frozen = 1; + header->frozen = 1; return 0; out_errno: return -errno; diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 4cc267559bb6..456661d7f10e 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -39,7 +39,7 @@ struct perf_pipe_file_header { struct perf_header; -int perf_file_header__read(struct perf_file_header *self, +int perf_file_header__read(struct perf_file_header *header, struct perf_header *ph, int fd); struct perf_header { @@ -66,12 +66,12 @@ char *perf_header__find_event(u64 id); u64 perf_evlist__sample_type(struct perf_evlist *evlist); bool perf_evlist__sample_id_all(const struct perf_evlist *evlist); -void perf_header__set_feat(struct perf_header *self, int feat); -void perf_header__clear_feat(struct perf_header *self, int feat); -bool perf_header__has_feat(const struct perf_header *self, int feat); +void perf_header__set_feat(struct perf_header *header, int feat); +void perf_header__clear_feat(struct perf_header *header, int feat); +bool perf_header__has_feat(const struct perf_header *header, int feat); -int perf_header__process_sections(struct perf_header *self, int fd, - int (*process)(struct perf_file_section *self, +int perf_header__process_sections(struct perf_header *header, int fd, + int (*process)(struct perf_file_section *section, struct perf_header *ph, int feat, int fd)); -- cgit v1.2.3 From 5e814dd597c42daeb8d2a276e64a6ec986ad0e2a Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 15 Mar 2011 20:51:09 +0100 Subject: perf probe: Clean up probe_point_lazy_walker() return value Newer compilers (gcc 4.6) complains about: return ret < 0 ?: 0; For the following reason: util/probe-finder.c: In function ‘probe_point_lazy_walker’: util/probe-finder.c:1331:18: error: the omitted middle operand in ?: will always be ‘true’, suggest explicit middle operand [-Werror=parentheses] And indeed the return value is a somewhat obscure (but correct) value of 'true', so return 'ret' instead - this is cleaner and unconfuses GCC as well. Cc: Arnaldo Carvalho de Melo Cc: Masami Hiramatsu Cc: Frederic Weisbecker Cc: Masami Hiramatsu Cc: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- tools/perf/util/probe-finder.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index 17f9c4a66ddd..194f9e2a3285 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -1328,7 +1328,7 @@ static int probe_point_lazy_walker(const char *fname, int lineno, * Continue if no error, because the lazy pattern will match * to other lines */ - return ret < 0 ?: 0; + return ret < 0 ? ret : 0; } /* Find probe points from lazy pattern */ -- cgit v1.2.3