diff options
| -rw-r--r-- | tools/perf/util/evlist.c | 142 | ||||
| -rw-r--r-- | tools/perf/util/evlist.h | 7 | ||||
| -rw-r--r-- | tools/perf/util/evsel.c | 144 | ||||
| -rw-r--r-- | tools/perf/util/evsel.h | 4 |
4 files changed, 158 insertions, 139 deletions
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 917fc18..dcd5932 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 <acme@redhat.com> + * + * 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 <poll.h> +#include "cpumap.h" +#include "thread_map.h" #include "evlist.h" #include "evsel.h" #include "util.h" +#include <sys/mman.h> + #include <linux/bitops.h> #include <linux/hash.h> +#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 022ae40..85aca6e 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 fddeb08..2720bc1 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 <acme@redhat.com> + * + * 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 <unistd.h> -#include <sys/mman.h> - -#include <linux/bitops.h> -#include <linux/hash.h> - #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 7962e75..eecdc3a 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 && \ |