Commit 137ee20d authored by Ingo Molnar's avatar Ingo Molnar

Merge branch 'perf/core' of...

Merge branch 'perf/core' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 into perf/core
parents 4a0b1665 1c0b04d1
......@@ -31,7 +31,6 @@
#include <sys/mman.h>
#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 @@ static void open_counters(struct perf_evlist *evlist)
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;
......
......@@ -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);
}
......
......@@ -883,7 +883,6 @@ static void start_counters(struct perf_evlist *evlist)
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);
......
......@@ -19,7 +19,7 @@
#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)
#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;
}
}
......
......@@ -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);
......
......@@ -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)
......
......@@ -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;
};
......
This diff is collapsed.
......@@ -9,13 +9,6 @@
#include <linux/bitmap.h>
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,
......@@ -46,14 +39,12 @@ 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 {
int frozen;
int attrs, size;
bool needs_swap;
struct perf_header_attr **attr;
s64 attr_offset;
u64 data_offset;
u64 data_size;
......@@ -62,35 +53,25 @@ 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);
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 *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_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);
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);
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));
......@@ -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,
......
......@@ -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 @@ static void perf_session__id_header_size(struct perf_session *session)
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);
......
......@@ -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
......
......@@ -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 +107,52 @@ 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",
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, "], ");
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" : "");
}
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment