Skip to content

Commit

Permalink
Merge branch 'perf/test' of git://git.kernel.org/pub/scm/linux/kernel…
Browse files Browse the repository at this point in the history 
…/git/acme/linux-2.6 into perf/core
  • Loading branch information
Ingo Molnar committed Jan 4, 2011
2 parents cc22219 + d854861 commit 9285855
Show file tree
Hide file tree
Showing 24 changed files with 1,013 additions and 519 deletions.
4 changes: 4 additions & 0 deletions tools/perf/Makefile
View file
Original file line number Diff line number Diff line change
Expand Up @@ -396,6 +396,7 @@ LIB_H += util/build-id.h
LIB_H += util/debug.h
LIB_H += util/debugfs.h
LIB_H += util/event.h
LIB_H += util/evsel.h
LIB_H += util/exec_cmd.h
LIB_H += util/types.h
LIB_H += util/levenshtein.h
Expand All @@ -404,6 +405,7 @@ LIB_H += util/parse-options.h
LIB_H += util/parse-events.h
LIB_H += util/quote.h
LIB_H += util/util.h
LIB_H += util/xyarray.h
LIB_H += util/header.h
LIB_H += util/help.h
LIB_H += util/session.h
Expand Down Expand Up @@ -433,6 +435,7 @@ LIB_OBJS += $(OUTPUT)util/ctype.o
LIB_OBJS += $(OUTPUT)util/debugfs.o
LIB_OBJS += $(OUTPUT)util/environment.o
LIB_OBJS += $(OUTPUT)util/event.o
LIB_OBJS += $(OUTPUT)util/evsel.o
LIB_OBJS += $(OUTPUT)util/exec_cmd.o
LIB_OBJS += $(OUTPUT)util/help.o
LIB_OBJS += $(OUTPUT)util/levenshtein.o
Expand Down Expand Up @@ -470,6 +473,7 @@ LIB_OBJS += $(OUTPUT)util/sort.o
LIB_OBJS += $(OUTPUT)util/hist.o
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

BUILTIN_OBJS += $(OUTPUT)builtin-annotate.o
Expand Down
152 changes: 68 additions & 84 deletions tools/perf/builtin-record.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,7 @@

#include "util/header.h"
#include "util/event.h"
#include "util/evsel.h"
#include "util/debug.h"
#include "util/session.h"
#include "util/symbol.h"
Expand All @@ -27,18 +28,18 @@
#include <sched.h>
#include <sys/mman.h>

#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))

enum write_mode_t {
WRITE_FORCE,
WRITE_APPEND
};

static int *fd[MAX_NR_CPUS][MAX_COUNTERS];

static u64 user_interval = ULLONG_MAX;
static u64 default_interval = 0;
static u64 sample_type;

static int nr_cpus = 0;
static struct cpu_map *cpus;
static unsigned int page_size;
static unsigned int mmap_pages = 128;
static unsigned int user_freq = UINT_MAX;
Expand All @@ -53,8 +54,7 @@ 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 pid_t *all_tids = NULL;
static int thread_num = 0;
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;
Expand All @@ -81,7 +81,6 @@ static struct perf_session *session;
static const char *cpu_list;

struct mmap_data {
int counter;
void *base;
unsigned int mask;
unsigned int prev;
Expand Down Expand Up @@ -229,12 +228,12 @@ static struct perf_header_attr *get_header_attr(struct perf_event_attr *a, int n
return h_attr;
}

static void create_counter(int counter, int cpu)
static void create_counter(struct perf_evsel *evsel, int cpu)
{
char *filter = filters[counter];
struct perf_event_attr *attr = attrs + counter;
char *filter = evsel->filter;
struct perf_event_attr *attr = &evsel->attr;
struct perf_header_attr *h_attr;
int track = !counter; /* only the first counter needs these */
int track = !evsel->idx; /* only the first counter needs these */
int thread_index;
int ret;
struct {
Expand Down Expand Up @@ -318,12 +317,11 @@ static void create_counter(int counter, int cpu)
retry_sample_id:
attr->sample_id_all = sample_id_all_avail ? 1 : 0;

for (thread_index = 0; thread_index < thread_num; thread_index++) {
for (thread_index = 0; thread_index < threads->nr; thread_index++) {
try_again:
fd[nr_cpu][counter][thread_index] = sys_perf_event_open(attr,
all_tids[thread_index], cpu, group_fd, 0);
FD(evsel, nr_cpu, thread_index) = sys_perf_event_open(attr, threads->map[thread_index], cpu, group_fd, 0);

if (fd[nr_cpu][counter][thread_index] < 0) {
if (FD(evsel, nr_cpu, thread_index) < 0) {
int err = errno;

if (err == EPERM || err == EACCES)
Expand Down Expand Up @@ -360,7 +358,7 @@ static void create_counter(int counter, int cpu)
}
printf("\n");
error("sys_perf_event_open() syscall returned with %d (%s). /bin/dmesg may provide additional information.\n",
fd[nr_cpu][counter][thread_index], strerror(err));
FD(evsel, nr_cpu, thread_index), strerror(err));

#if defined(__i386__) || defined(__x86_64__)
if (attr->type == PERF_TYPE_HARDWARE && err == EOPNOTSUPP)
Expand All @@ -374,7 +372,7 @@ static void create_counter(int counter, int cpu)
exit(-1);
}

h_attr = get_header_attr(attr, counter);
h_attr = get_header_attr(attr, evsel->idx);
if (h_attr == NULL)
die("nomem\n");

Expand All @@ -385,7 +383,7 @@ static void create_counter(int counter, int cpu)
}
}

if (read(fd[nr_cpu][counter][thread_index], &read_data, sizeof(read_data)) == -1) {
if (read(FD(evsel, nr_cpu, thread_index), &read_data, sizeof(read_data)) == -1) {
perror("Unable to read perf file descriptor");
exit(-1);
}
Expand All @@ -395,43 +393,44 @@ static void create_counter(int counter, int cpu)
exit(-1);
}

assert(fd[nr_cpu][counter][thread_index] >= 0);
fcntl(fd[nr_cpu][counter][thread_index], F_SETFL, O_NONBLOCK);
assert(FD(evsel, nr_cpu, thread_index) >= 0);
fcntl(FD(evsel, nr_cpu, thread_index), F_SETFL, O_NONBLOCK);

/*
* First counter acts as the group leader:
*/
if (group && group_fd == -1)
group_fd = fd[nr_cpu][counter][thread_index];

if (counter || thread_index) {
ret = ioctl(fd[nr_cpu][counter][thread_index],
PERF_EVENT_IOC_SET_OUTPUT,
fd[nr_cpu][0][0]);
group_fd = FD(evsel, nr_cpu, thread_index);

if (evsel->idx || thread_index) {
struct perf_evsel *first;
first = list_entry(evsel_list.next, struct perf_evsel, node);
ret = ioctl(FD(evsel, nr_cpu, thread_index),
PERF_EVENT_IOC_SET_OUTPUT,
FD(first, nr_cpu, 0));
if (ret) {
error("failed to set output: %d (%s)\n", errno,
strerror(errno));
exit(-1);
}
} else {
mmap_array[nr_cpu].counter = counter;
mmap_array[nr_cpu].prev = 0;
mmap_array[nr_cpu].mask = mmap_pages*page_size - 1;
mmap_array[nr_cpu].base = mmap(NULL, (mmap_pages+1)*page_size,
PROT_READ|PROT_WRITE, MAP_SHARED, fd[nr_cpu][counter][thread_index], 0);
PROT_READ | PROT_WRITE, MAP_SHARED, FD(evsel, nr_cpu, thread_index), 0);
if (mmap_array[nr_cpu].base == MAP_FAILED) {
error("failed to mmap with %d (%s)\n", errno, strerror(errno));
exit(-1);
}

event_array[nr_poll].fd = fd[nr_cpu][counter][thread_index];
event_array[nr_poll].fd = FD(evsel, nr_cpu, thread_index);
event_array[nr_poll].events = POLLIN;
nr_poll++;
}

if (filter != NULL) {
ret = ioctl(fd[nr_cpu][counter][thread_index],
PERF_EVENT_IOC_SET_FILTER, filter);
ret = ioctl(FD(evsel, nr_cpu, thread_index),
PERF_EVENT_IOC_SET_FILTER, filter);
if (ret) {
error("failed to set filter with %d (%s)\n", errno,
strerror(errno));
Expand All @@ -446,11 +445,12 @@ static void create_counter(int counter, int cpu)

static void open_counters(int cpu)
{
int counter;
struct perf_evsel *pos;

group_fd = -1;
for (counter = 0; counter < nr_counters; counter++)
create_counter(counter, cpu);

list_for_each_entry(pos, &evsel_list, node)
create_counter(pos, cpu);

nr_cpu++;
}
Expand Down Expand Up @@ -537,7 +537,7 @@ static void mmap_read_all(void)

static int __cmd_record(int argc, const char **argv)
{
int i, counter;
int i;
struct stat st;
int flags;
int err;
Expand Down Expand Up @@ -604,7 +604,7 @@ static int __cmd_record(int argc, const char **argv)
goto out_delete_session;
}

if (have_tracepoints(attrs, nr_counters))
if (have_tracepoints(&evsel_list))
perf_header__set_feat(&session->header, HEADER_TRACE_INFO);

/*
Expand Down Expand Up @@ -652,7 +652,7 @@ static int __cmd_record(int argc, const char **argv)
}

if (!system_wide && target_tid == -1 && target_pid == -1)
all_tids[0] = child_pid;
threads->map[0] = child_pid;

close(child_ready_pipe[1]);
close(go_pipe[0]);
Expand All @@ -666,17 +666,11 @@ static int __cmd_record(int argc, const char **argv)
close(child_ready_pipe[0]);
}

nr_cpus = read_cpu_map(cpu_list);
if (nr_cpus < 1) {
perror("failed to collect number of CPUs");
return -1;
}

if (!system_wide && no_inherit && !cpu_list) {
open_counters(-1);
} else {
for (i = 0; i < nr_cpus; i++)
open_counters(cpumap[i]);
for (i = 0; i < cpus->nr; i++)
open_counters(cpus->map[i]);
}

perf_session__set_sample_type(session, sample_type);
Expand Down Expand Up @@ -711,7 +705,7 @@ static int __cmd_record(int argc, const char **argv)
return err;
}

if (have_tracepoints(attrs, nr_counters)) {
if (have_tracepoints(&evsel_list)) {
/*
* FIXME err <= 0 here actually means that
* there were no tracepoints so its not really
Expand All @@ -720,8 +714,7 @@ 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, attrs,
nr_counters,
err = event__synthesize_tracing_data(output, &evsel_list,
process_synthesized_event,
session);
if (err <= 0) {
Expand Down Expand Up @@ -795,13 +788,13 @@ static int __cmd_record(int argc, const char **argv)

if (done) {
for (i = 0; i < nr_cpu; i++) {
for (counter = 0;
counter < nr_counters;
counter++) {
struct perf_evsel *pos;

list_for_each_entry(pos, &evsel_list, node) {
for (thread = 0;
thread < thread_num;
thread < threads->nr;
thread++)
ioctl(fd[i][counter][thread],
ioctl(FD(pos, i, thread),
PERF_EVENT_IOC_DISABLE);
}
}
Expand Down Expand Up @@ -887,7 +880,8 @@ const struct option record_options[] = {

int cmd_record(int argc, const char **argv, const char *prefix __used)
{
int i, j, err = -ENOMEM;
int err = -ENOMEM;
struct perf_evsel *pos;

argc = parse_options(argc, argv, record_options, record_usage,
PARSE_OPT_STOP_AT_NON_OPTION);
Expand All @@ -910,38 +904,32 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
if (no_buildid_cache || no_buildid)
disable_buildid_cache();

if (!nr_counters) {
nr_counters = 1;
attrs[0].type = PERF_TYPE_HARDWARE;
attrs[0].config = PERF_COUNT_HW_CPU_CYCLES;
if (list_empty(&evsel_list) && perf_evsel_list__create_default() < 0) {
pr_err("Not enough memory for event selector list\n");
goto out_symbol_exit;
}

if (target_pid != -1) {
if (target_pid != -1)
target_tid = target_pid;
thread_num = find_all_tid(target_pid, &all_tids);
if (thread_num <= 0) {
fprintf(stderr, "Can't find all threads of pid %d\n",
target_pid);
usage_with_options(record_usage, record_options);
}
} else {
all_tids=malloc(sizeof(pid_t));
if (!all_tids)
goto out_symbol_exit;

all_tids[0] = target_tid;
thread_num = 1;
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);
}

for (i = 0; i < MAX_NR_CPUS; i++) {
for (j = 0; j < MAX_COUNTERS; j++) {
fd[i][j] = malloc(sizeof(int)*thread_num);
if (!fd[i][j])
goto out_free_fd;
}
cpus = cpu_map__new(cpu_list);
if (cpus == NULL) {
perror("failed to parse CPUs map");
return -1;
}

list_for_each_entry(pos, &evsel_list, node) {
if (perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0)
goto out_free_fd;
}
event_array = malloc(
sizeof(struct pollfd)*MAX_NR_CPUS*MAX_COUNTERS*thread_num);
event_array = malloc((sizeof(struct pollfd) * MAX_NR_CPUS *
MAX_COUNTERS * threads->nr));
if (!event_array)
goto out_free_fd;

Expand All @@ -968,12 +956,8 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
out_free_event_array:
free(event_array);
out_free_fd:
for (i = 0; i < MAX_NR_CPUS; i++) {
for (j = 0; j < MAX_COUNTERS; j++)
free(fd[i][j]);
}
free(all_tids);
all_tids = NULL;
thread_map__delete(threads);
threads = NULL;
out_symbol_exit:
symbol__exit();
return err;
Expand Down
Loading

0 comments on commit 9285855

Please sign in to comment.