diff options
Diffstat (limited to 'tools/perf/util/session.c')
-rw-r--r-- | tools/perf/util/session.c | 570 |
1 files changed, 450 insertions, 120 deletions
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index eed1cb88900..8f83a183576 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -5,6 +5,7 @@ #include <byteswap.h> #include <unistd.h> #include <sys/types.h> +#include <sys/mman.h> #include "session.h" #include "sort.h" @@ -14,6 +15,16 @@ static int perf_session__open(struct perf_session *self, bool force) { struct stat input_stat; + if (!strcmp(self->filename, "-")) { + self->fd_pipe = true; + self->fd = STDIN_FILENO; + + if (perf_header__read(self, self->fd) < 0) + pr_err("incompatible file format"); + + return 0; + } + self->fd = open(self->filename, O_RDONLY); if (self->fd < 0) { pr_err("failed to open file: %s", self->filename); @@ -38,7 +49,7 @@ static int perf_session__open(struct perf_session *self, bool force) goto out_close; } - if (perf_header__read(&self->header, self->fd) < 0) { + if (perf_header__read(self, self->fd) < 0) { pr_err("incompatible file format"); goto out_close; } @@ -52,12 +63,21 @@ out_close: return -1; } -static inline int perf_session__create_kernel_maps(struct perf_session *self) +void perf_session__update_sample_type(struct perf_session *self) +{ + self->sample_type = perf_header__sample_type(&self->header); +} + +int perf_session__create_kernel_maps(struct perf_session *self) { - return map_groups__create_kernel_maps(&self->kmaps, self->vmlinux_maps); + int ret = machine__create_kernel_maps(&self->host_machine); + + if (ret >= 0) + ret = machines__create_guest_kernel_maps(&self->machines); + return ret; } -struct perf_session *perf_session__new(const char *filename, int mode, bool force) +struct perf_session *perf_session__new(const char *filename, int mode, bool force, bool repipe) { size_t len = filename ? strlen(filename) + 1 : 0; struct perf_session *self = zalloc(sizeof(*self) + len); @@ -70,13 +90,15 @@ struct perf_session *perf_session__new(const char *filename, int mode, bool forc memcpy(self->filename, filename, len); self->threads = RB_ROOT; - self->stats_by_id = RB_ROOT; + self->hists_tree = RB_ROOT; self->last_match = NULL; self->mmap_window = 32; self->cwd = NULL; self->cwdlen = 0; - self->unknown_events = 0; - map_groups__init(&self->kmaps); + self->machines = RB_ROOT; + self->repipe = repipe; + INIT_LIST_HEAD(&self->ordered_samples.samples_head); + machine__init(&self->host_machine, "", HOST_KERNEL_ID); if (mode == O_RDONLY) { if (perf_session__open(self, force) < 0) @@ -90,7 +112,7 @@ struct perf_session *perf_session__new(const char *filename, int mode, bool forc goto out_delete; } - self->sample_type = perf_header__sample_type(&self->header); + perf_session__update_sample_type(self); out: return self; out_free: @@ -117,22 +139,17 @@ static bool symbol__match_parent_regex(struct symbol *sym) return 0; } -struct symbol **perf_session__resolve_callchain(struct perf_session *self, - struct thread *thread, - struct ip_callchain *chain, - struct symbol **parent) +struct map_symbol *perf_session__resolve_callchain(struct perf_session *self, + struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent) { u8 cpumode = PERF_RECORD_MISC_USER; - struct symbol **syms = NULL; unsigned int i; + struct map_symbol *syms = calloc(chain->nr, sizeof(*syms)); - if (symbol_conf.use_callchain) { - syms = calloc(chain->nr, sizeof(*syms)); - if (!syms) { - fprintf(stderr, "Can't allocate memory for symbols\n"); - exit(-1); - } - } + if (!syms) + return NULL; for (i = 0; i < chain->nr; i++) { u64 ip = chain->ips[i]; @@ -152,15 +169,17 @@ struct symbol **perf_session__resolve_callchain(struct perf_session *self, continue; } + al.filtered = false; thread__find_addr_location(thread, self, cpumode, - MAP__FUNCTION, ip, &al, NULL); + MAP__FUNCTION, thread->pid, ip, &al, NULL); if (al.sym != NULL) { if (sort__has_parent && !*parent && symbol__match_parent_regex(al.sym)) *parent = al.sym; if (!symbol_conf.use_callchain) break; - syms[i] = al.sym; + syms[i].map = al.map; + syms[i].sym = al.sym; } } @@ -174,6 +193,18 @@ static int process_event_stub(event_t *event __used, return 0; } +static int process_finished_round_stub(event_t *event __used, + struct perf_session *session __used, + struct perf_event_ops *ops __used) +{ + dump_printf(": unhandled!\n"); + return 0; +} + +static int process_finished_round(event_t *event, + struct perf_session *session, + struct perf_event_ops *ops); + static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) { if (handler->sample == NULL) @@ -194,29 +225,20 @@ static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) handler->throttle = process_event_stub; if (handler->unthrottle == NULL) handler->unthrottle = process_event_stub; -} - -static const char *event__name[] = { - [0] = "TOTAL", - [PERF_RECORD_MMAP] = "MMAP", - [PERF_RECORD_LOST] = "LOST", - [PERF_RECORD_COMM] = "COMM", - [PERF_RECORD_EXIT] = "EXIT", - [PERF_RECORD_THROTTLE] = "THROTTLE", - [PERF_RECORD_UNTHROTTLE] = "UNTHROTTLE", - [PERF_RECORD_FORK] = "FORK", - [PERF_RECORD_READ] = "READ", - [PERF_RECORD_SAMPLE] = "SAMPLE", -}; - -unsigned long event__total[PERF_RECORD_MAX]; - -void event__print_totals(void) -{ - int i; - for (i = 0; i < PERF_RECORD_MAX; ++i) - pr_info("%10s events: %10ld\n", - event__name[i], event__total[i]); + if (handler->attr == NULL) + handler->attr = process_event_stub; + if (handler->event_type == NULL) + handler->event_type = process_event_stub; + if (handler->tracing_data == NULL) + handler->tracing_data = process_event_stub; + if (handler->build_id == NULL) + handler->build_id = process_event_stub; + if (handler->finished_round == NULL) { + if (handler->ordered_samples) + handler->finished_round = process_finished_round; + else + handler->finished_round = process_finished_round_stub; + } } void mem_bswap_64(void *src, int byte_size) @@ -270,6 +292,37 @@ static void event__read_swap(event_t *self) self->read.id = bswap_64(self->read.id); } +static void event__attr_swap(event_t *self) +{ + 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); +} + +static void event__event_type_swap(event_t *self) +{ + self->event_type.event_type.event_id = + bswap_64(self->event_type.event_type.event_id); +} + +static void event__tracing_data_swap(event_t *self) +{ + self->tracing_data.size = bswap_32(self->tracing_data.size); +} + typedef void (*event__swap_op)(event_t *self); static event__swap_op event__swap_ops[] = { @@ -280,9 +333,212 @@ static event__swap_op event__swap_ops[] = { [PERF_RECORD_LOST] = event__all64_swap, [PERF_RECORD_READ] = event__read_swap, [PERF_RECORD_SAMPLE] = event__all64_swap, - [PERF_RECORD_MAX] = NULL, + [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, }; +struct sample_queue { + u64 timestamp; + struct sample_event *event; + struct list_head list; +}; + +static void flush_sample_queue(struct perf_session *s, + struct perf_event_ops *ops) +{ + struct list_head *head = &s->ordered_samples.samples_head; + u64 limit = s->ordered_samples.next_flush; + struct sample_queue *tmp, *iter; + + if (!ops->ordered_samples || !limit) + return; + + list_for_each_entry_safe(iter, tmp, head, list) { + if (iter->timestamp > limit) + return; + + if (iter == s->ordered_samples.last_inserted) + s->ordered_samples.last_inserted = NULL; + + ops->sample((event_t *)iter->event, s); + + s->ordered_samples.last_flush = iter->timestamp; + list_del(&iter->list); + free(iter->event); + free(iter); + } +} + +/* + * When perf record finishes a pass on every buffers, it records this pseudo + * event. + * We record the max timestamp t found in the pass n. + * Assuming these timestamps are monotonic across cpus, we know that if + * a buffer still has events with timestamps below t, they will be all + * available and then read in the pass n + 1. + * Hence when we start to read the pass n + 2, we can safely flush every + * events with timestamps below t. + * + * ============ PASS n ================= + * CPU 0 | CPU 1 + * | + * cnt1 timestamps | cnt2 timestamps + * 1 | 2 + * 2 | 3 + * - | 4 <--- max recorded + * + * ============ PASS n + 1 ============== + * CPU 0 | CPU 1 + * | + * cnt1 timestamps | cnt2 timestamps + * 3 | 5 + * 4 | 6 + * 5 | 7 <---- max recorded + * + * Flush every events below timestamp 4 + * + * ============ PASS n + 2 ============== + * CPU 0 | CPU 1 + * | + * cnt1 timestamps | cnt2 timestamps + * 6 | 8 + * 7 | 9 + * - | 10 + * + * Flush every events below timestamp 7 + * etc... + */ +static int process_finished_round(event_t *event __used, + struct perf_session *session, + struct perf_event_ops *ops) +{ + flush_sample_queue(session, ops); + session->ordered_samples.next_flush = session->ordered_samples.max_timestamp; + + return 0; +} + +static void __queue_sample_end(struct sample_queue *new, struct list_head *head) +{ + struct sample_queue *iter; + + list_for_each_entry_reverse(iter, head, list) { + if (iter->timestamp < new->timestamp) { + list_add(&new->list, &iter->list); + return; + } + } + + list_add(&new->list, head); +} + +static void __queue_sample_before(struct sample_queue *new, + struct sample_queue *iter, + struct list_head *head) +{ + list_for_each_entry_continue_reverse(iter, head, list) { + if (iter->timestamp < new->timestamp) { + list_add(&new->list, &iter->list); + return; + } + } + + list_add(&new->list, head); +} + +static void __queue_sample_after(struct sample_queue *new, + struct sample_queue *iter, + struct list_head *head) +{ + list_for_each_entry_continue(iter, head, list) { + if (iter->timestamp > new->timestamp) { + list_add_tail(&new->list, &iter->list); + return; + } + } + list_add_tail(&new->list, head); +} + +/* The queue is ordered by time */ +static void __queue_sample_event(struct sample_queue *new, + struct perf_session *s) +{ + struct sample_queue *last_inserted = s->ordered_samples.last_inserted; + struct list_head *head = &s->ordered_samples.samples_head; + + + if (!last_inserted) { + __queue_sample_end(new, head); + return; + } + + /* + * Most of the time the current event has a timestamp + * very close to the last event inserted, unless we just switched + * to another event buffer. Having a sorting based on a list and + * on the last inserted event that is close to the current one is + * probably more efficient than an rbtree based sorting. + */ + if (last_inserted->timestamp >= new->timestamp) + __queue_sample_before(new, last_inserted, head); + else + __queue_sample_after(new, last_inserted, head); +} + +static int queue_sample_event(event_t *event, struct sample_data *data, + struct perf_session *s) +{ + u64 timestamp = data->time; + struct sample_queue *new; + + + if (timestamp < s->ordered_samples.last_flush) { + printf("Warning: Timestamp below last timeslice flush\n"); + return -EINVAL; + } + + new = malloc(sizeof(*new)); + if (!new) + return -ENOMEM; + + new->timestamp = timestamp; + + new->event = malloc(event->header.size); + if (!new->event) { + free(new); + return -ENOMEM; + } + + memcpy(new->event, event, event->header.size); + + __queue_sample_event(new, s); + s->ordered_samples.last_inserted = new; + + if (new->timestamp > s->ordered_samples.max_timestamp) + s->ordered_samples.max_timestamp = new->timestamp; + + return 0; +} + +static int perf_session__process_sample(event_t *event, struct perf_session *s, + struct perf_event_ops *ops) +{ + struct sample_data data; + + if (!ops->ordered_samples) + return ops->sample(event, s); + + bzero(&data, sizeof(struct sample_data)); + event__parse_sample(event, s->sample_type, &data); + + queue_sample_event(event, &data, s); + + return 0; +} + static int perf_session__process_event(struct perf_session *self, event_t *event, struct perf_event_ops *ops, @@ -290,12 +546,11 @@ static int perf_session__process_event(struct perf_session *self, { trace_event(event); - if (event->header.type < PERF_RECORD_MAX) { + if (event->header.type < PERF_RECORD_HEADER_MAX) { dump_printf("%#Lx [%#x]: PERF_RECORD_%s", offset + head, event->header.size, event__name[event->header.type]); - ++event__total[0]; - ++event__total[event->header.type]; + hists__inc_nr_events(&self->hists, event->header.type); } if (self->header.needs_swap && event__swap_ops[event->header.type]) @@ -303,7 +558,7 @@ static int perf_session__process_event(struct perf_session *self, switch (event->header.type) { case PERF_RECORD_SAMPLE: - return ops->sample(event, self); + return perf_session__process_sample(event, self, ops); case PERF_RECORD_MMAP: return ops->mmap(event, self); case PERF_RECORD_COMM: @@ -320,8 +575,20 @@ static int perf_session__process_event(struct perf_session *self, return ops->throttle(event, self); case PERF_RECORD_UNTHROTTLE: return ops->unthrottle(event, self); + case PERF_RECORD_HEADER_ATTR: + return ops->attr(event, self); + case PERF_RECORD_HEADER_EVENT_TYPE: + return ops->event_type(event, self); + case PERF_RECORD_HEADER_TRACING_DATA: + /* setup for reading amidst mmap */ + lseek(self->fd, offset + head, SEEK_SET); + return ops->tracing_data(event, self); + case PERF_RECORD_HEADER_BUILD_ID: + return ops->build_id(event, self); + case PERF_RECORD_FINISHED_ROUND: + return ops->finished_round(event, self, ops); default: - self->unknown_events++; + ++self->hists.stats.nr_unknown_events; return -1; } } @@ -333,56 +600,114 @@ void perf_event_header__bswap(struct perf_event_header *self) self->size = bswap_16(self->size); } -int perf_header__read_build_ids(struct perf_header *self, - int input, u64 offset, u64 size) +static struct thread *perf_session__register_idle_thread(struct perf_session *self) { - struct build_id_event bev; - char filename[PATH_MAX]; - u64 limit = offset + size; - int err = -1; + struct thread *thread = perf_session__findnew(self, 0); - while (offset < limit) { - struct dso *dso; - ssize_t len; - struct list_head *head = &dsos__user; + if (thread == NULL || thread__set_comm(thread, "swapper")) { + pr_err("problem inserting idle task.\n"); + thread = NULL; + } - if (read(input, &bev, sizeof(bev)) != sizeof(bev)) - goto out; + return thread; +} - if (self->needs_swap) - perf_event_header__bswap(&bev.header); +int do_read(int fd, void *buf, size_t size) +{ + void *buf_start = buf; - len = bev.header.size - sizeof(bev); - if (read(input, filename, len) != len) - goto out; + while (size) { + int ret = read(fd, buf, size); - if (bev.header.misc & PERF_RECORD_MISC_KERNEL) - head = &dsos__kernel; + if (ret <= 0) + return ret; - dso = __dsos__findnew(head, filename); - if (dso != NULL) { - dso__set_build_id(dso, &bev.build_id); - if (head == &dsos__kernel && filename[0] == '[') - dso->kernel = 1; - } - - offset += bev.header.size; + size -= ret; + buf += ret; } - err = 0; -out: - return err; + + return buf - buf_start; } -static struct thread *perf_session__register_idle_thread(struct perf_session *self) +#define session_done() (*(volatile int *)(&session_done)) +volatile int session_done; + +static int __perf_session__process_pipe_events(struct perf_session *self, + struct perf_event_ops *ops) { - struct thread *thread = perf_session__findnew(self, 0); + event_t event; + uint32_t size; + int skip = 0; + u64 head; + int err; + void *p; - if (thread == NULL || thread__set_comm(thread, "swapper")) { - pr_err("problem inserting idle task.\n"); - thread = NULL; + perf_event_ops__fill_defaults(ops); + + head = 0; +more: + err = do_read(self->fd, &event, sizeof(struct perf_event_header)); + if (err <= 0) { + if (err == 0) + goto done; + + pr_err("failed to read event header\n"); + goto out_err; } - return thread; + if (self->header.needs_swap) + perf_event_header__bswap(&event.header); + + size = event.header.size; + if (size == 0) + size = 8; + + p = &event; + p += sizeof(struct perf_event_header); + + if (size - sizeof(struct perf_event_header)) { + err = do_read(self->fd, p, + size - sizeof(struct perf_event_header)); + if (err <= 0) { + if (err == 0) { + pr_err("unexpected end of event stream\n"); + goto done; + } + + pr_err("failed to read event data\n"); + goto out_err; + } + } + + if (size == 0 || + (skip = perf_session__process_event(self, &event, ops, + 0, head)) < 0) { + dump_printf("%#Lx [%#x]: skipping unknown header type: %d\n", + head, event.header.size, event.header.type); + /* + * assume we lost track of the stream, check alignment, and + * increment a single u64 in the hope to catch on again 'soon'. + */ + if (unlikely(head & 7)) + head &= ~7ULL; + + size = 8; + } + + head += size; + + dump_printf("\n%#Lx [%#x]: event: %d\n", + head, event.header.size, event.header.type); + + if (skip > 0) + head += skip; + + if (!session_done()) + goto more; +done: + err = 0; +out_err: + return err; } int __perf_session__process_events(struct perf_session *self, @@ -396,6 +721,10 @@ int __perf_session__process_events(struct perf_session *self, event_t *event; uint32_t size; char *buf; + struct ui_progress *progress = ui_progress__new("Processing events...", + self->size); + if (progress == NULL) + return -1; perf_event_ops__fill_defaults(ops); @@ -424,6 +753,7 @@ remap: more: event = (event_t *)(buf + head); + ui_progress__update(progress, offset); if (self->header.needs_swap) perf_event_header__bswap(&event->header); @@ -473,7 +803,11 @@ more: goto more; done: err = 0; + /* do the final flush for ordered samples */ + self->ordered_samples.next_flush = ULLONG_MAX; + flush_sample_queue(self, ops); out_err: + ui_progress__delete(progress); return err; } @@ -502,9 +836,13 @@ out_getcwd_err: self->cwdlen = strlen(self->cwd); } - err = __perf_session__process_events(self, self->header.data_offset, - self->header.data_size, - self->size, ops); + if (!self->fd_pipe) + err = __perf_session__process_events(self, + self->header.data_offset, + self->header.data_size, + self->size, ops); + else + err = __perf_session__process_pipe_events(self, ops); out_err: return err; } @@ -519,56 +857,48 @@ bool perf_session__has_traces(struct perf_session *self, const char *msg) return true; } -int perf_session__set_kallsyms_ref_reloc_sym(struct perf_session *self, +int perf_session__set_kallsyms_ref_reloc_sym(struct map **maps, const char *symbol_name, u64 addr) { char *bracket; enum map_type i; + struct ref_reloc_sym *ref; + + ref = zalloc(sizeof(struct ref_reloc_sym)); + if (ref == NULL) + return -ENOMEM; - self->ref_reloc_sym.name = strdup(symbol_name); - if (self->ref_reloc_sym.name == NULL) + ref->name = strdup(symbol_name); + if (ref->name == NULL) { + free(ref); return -ENOMEM; + } - bracket = strchr(self->ref_reloc_sym.name, ']'); + bracket = strchr(ref->name, ']'); if (bracket) *bracket = '\0'; - self->ref_reloc_sym.addr = addr; + ref->addr = addr; for (i = 0; i < MAP__NR_TYPES; ++i) { - struct kmap *kmap = map__kmap(self->vmlinux_maps[i]); - kmap->ref_reloc_sym = &self->ref_reloc_sym; + struct kmap *kmap = map__kmap(maps[i]); + kmap->ref_reloc_sym = ref; } return 0; } -static u64 map__reloc_map_ip(struct map *map, u64 ip) -{ - return ip + (s64)map->pgoff; -} - -static u64 map__reloc_unmap_ip(struct map *map, u64 ip) +size_t perf_session__fprintf_dsos(struct perf_session *self, FILE *fp) { - return ip - (s64)map->pgoff; + return __dsos__fprintf(&self->host_machine.kernel_dsos, fp) + + __dsos__fprintf(&self->host_machine.user_dsos, fp) + + machines__fprintf_dsos(&self->machines, fp); } -void map__reloc_vmlinux(struct map *self) +size_t perf_session__fprintf_dsos_buildid(struct perf_session *self, FILE *fp, + bool with_hits) { - struct kmap *kmap = map__kmap(self); - s64 reloc; - - if (!kmap->ref_reloc_sym || !kmap->ref_reloc_sym->unrelocated_addr) - return; - - reloc = (kmap->ref_reloc_sym->unrelocated_addr - - kmap->ref_reloc_sym->addr); - - if (!reloc) - return; - - self->map_ip = map__reloc_map_ip; - self->unmap_ip = map__reloc_unmap_ip; - self->pgoff = reloc; + size_t ret = machine__fprintf_dsos_buildid(&self->host_machine, fp, with_hits); + return ret + machines__fprintf_dsos_buildid(&self->machines, fp, with_hits); } |