* (or a CPU, or a PID) into the perf.data output file - for
* later analysis via perf report.
*/
+#define _FILE_OFFSET_BITS 64
+
#include "builtin.h"
#include "perf.h"
+#include "util/build-id.h"
#include "util/util.h"
#include "util/parse-options.h"
#include "util/parse-events.h"
#include "util/header.h"
#include "util/event.h"
#include "util/debug.h"
+#include "util/session.h"
#include "util/symbol.h"
+#include "util/cpumap.h"
#include <unistd.h>
#include <sched.h>
static int nr_cpu = 0;
static int file_new = 1;
+static off_t post_processing_offset;
-struct perf_header *header = NULL;
+static struct perf_session *session;
struct mmap_data {
int counter;
}
}
-static void write_event(event_t *buf, size_t size)
-{
- /*
- * Add it to the list of DSOs, so that when we finish this
- * record session we can pick the available build-ids.
- */
- if (buf->header.type == PERF_RECORD_MMAP)
- dsos__findnew(buf->mmap.filename);
-
- write_output(buf, size);
-}
-
-static int process_synthesized_event(event_t *event)
+static int process_synthesized_event(event_t *event,
+ struct perf_session *self __used)
{
- write_event(event, event->header.size);
+ write_output(event, event->header.size);
return 0;
}
size = md->mask + 1 - (old & md->mask);
old += size;
- write_event(buf, size);
+ write_output(buf, size);
}
buf = &data[old & md->mask];
size = head - old;
old += size;
- write_event(buf, size);
+ write_output(buf, size);
md->prev = old;
mmap_write_tail(md, old);
{
struct perf_header_attr *h_attr;
- if (nr < header->attrs) {
- h_attr = header->attr[nr];
+ 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(header, h_attr) < 0) {
+ if (perf_header__add_attr(&session->header, h_attr) < 0) {
perf_header_attr__delete(h_attr);
h_attr = NULL;
}
attr->sample_type |= PERF_SAMPLE_IP | PERF_SAMPLE_TID;
+ if (nr_counters > 1)
+ attr->sample_type |= PERF_SAMPLE_ID;
+
if (freq) {
attr->sample_type |= PERF_SAMPLE_PERIOD;
attr->freq = 1;
attr->mmap = track;
attr->comm = track;
- attr->inherit = (cpu < 0) && inherit;
+ attr->inherit = inherit;
attr->disabled = 1;
try_again:
printf("\n");
error("perfcounter syscall returned with %d (%s)\n",
fd[nr_cpu][counter], strerror(err));
+
+#if defined(__i386__) || defined(__x86_64__)
+ if (attr->type == PERF_TYPE_HARDWARE && err == EOPNOTSUPP)
+ die("No hardware sampling interrupt available. No APIC? If so then you can boot the kernel with the \"lapic\" boot parameter to force-enable it.\n");
+#endif
+
die("No CONFIG_PERF_EVENTS=y kernel support configured?\n");
exit(-1);
}
nr_cpu++;
}
+static int process_buildids(void)
+{
+ u64 size = lseek(output, 0, SEEK_CUR);
+
+ if (size == 0)
+ return 0;
+
+ session->fd = output;
+ return __perf_session__process_events(session, post_processing_offset,
+ size - post_processing_offset,
+ size, &build_id__mark_dso_hit_ops);
+}
+
static void atexit_header(void)
{
- header->data_size += bytes_written;
+ session->header.data_size += bytes_written;
- perf_header__write(header, output, true);
+ process_buildids();
+ perf_header__write(&session->header, output, true);
}
static int __cmd_record(int argc, const char **argv)
struct stat st;
pid_t pid = 0;
int flags;
- int ret;
+ int err;
unsigned long waking = 0;
+ int child_ready_pipe[2], go_pipe[2];
+ const bool forks = target_pid == -1 && argc > 0;
+ char buf;
page_size = sysconf(_SC_PAGE_SIZE);
- nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
- assert(nr_cpus <= MAX_NR_CPUS);
- assert(nr_cpus >= 0);
atexit(sig_atexit);
signal(SIGCHLD, sig_handler);
signal(SIGINT, sig_handler);
+ if (forks && (pipe(child_ready_pipe) < 0 || pipe(go_pipe) < 0)) {
+ perror("failed to create pipes");
+ exit(-1);
+ }
+
if (!stat(output_name, &st) && st.st_size) {
- if (!force && !append_file) {
- fprintf(stderr, "Error, output file %s exists, use -A to append or -f to overwrite.\n",
- output_name);
- exit(-1);
+ if (!force) {
+ if (!append_file) {
+ pr_err("Error, output file %s exists, use -A "
+ "to append or -f to overwrite.\n",
+ output_name);
+ exit(-1);
+ }
+ } else {
+ char oldname[PATH_MAX];
+ snprintf(oldname, sizeof(oldname), "%s.old",
+ output_name);
+ unlink(oldname);
+ rename(output_name, oldname);
}
} else {
append_file = 0;
exit(-1);
}
- if (!file_new)
- header = perf_header__read(output);
- else
- header = perf_header__new();
-
- if (header == NULL) {
+ session = perf_session__new(output_name, O_WRONLY, force);
+ if (session == NULL) {
pr_err("Not enough memory for reading perf file header\n");
return -1;
}
+ if (!file_new) {
+ err = perf_header__read(&session->header, output);
+ if (err < 0)
+ return err;
+ }
+
if (raw_samples) {
- perf_header__set_feat(header, HEADER_TRACE_INFO);
+ perf_header__set_feat(&session->header, HEADER_TRACE_INFO);
} else {
for (i = 0; i < nr_counters; i++) {
if (attrs[i].sample_type & PERF_SAMPLE_RAW) {
- perf_header__set_feat(header, HEADER_TRACE_INFO);
+ perf_header__set_feat(&session->header, HEADER_TRACE_INFO);
break;
}
}
atexit(atexit_header);
- if (!system_wide) {
- pid = target_pid;
- if (pid == -1)
- pid = getpid();
-
- open_counters(profile_cpu, pid);
- } else {
- if (profile_cpu != -1) {
- open_counters(profile_cpu, target_pid);
- } else {
- for (i = 0; i < nr_cpus; i++)
- open_counters(i, target_pid);
+ if (forks) {
+ pid = fork();
+ if (pid < 0) {
+ perror("failed to fork");
+ exit(-1);
}
- }
- if (file_new)
- perf_header__write(header, output, false);
+ if (!pid) {
+ close(child_ready_pipe[0]);
+ close(go_pipe[1]);
+ fcntl(go_pipe[0], F_SETFD, FD_CLOEXEC);
- if (!system_wide)
- event__synthesize_thread(pid, process_synthesized_event);
- else
- event__synthesize_threads(process_synthesized_event);
+ /*
+ * Do a dummy execvp to get the PLT entry resolved,
+ * so we avoid the resolver overhead on the real
+ * execvp call.
+ */
+ execvp("", (char **)argv);
- if (target_pid == -1 && argc) {
- pid = fork();
- if (pid < 0)
- die("failed to fork");
+ /*
+ * Tell the parent we're ready to go
+ */
+ close(child_ready_pipe[1]);
- if (!pid) {
- if (execvp(argv[0], (char **)argv)) {
- perror(argv[0]);
- exit(-1);
- }
- } else {
/*
- * Wait a bit for the execv'ed child to appear
- * and be updated in /proc
- * FIXME: Do you know a less heuristical solution?
+ * Wait until the parent tells us to go.
*/
- usleep(1000);
- event__synthesize_thread(pid,
- process_synthesized_event);
+ if (read(go_pipe[0], &buf, 1) == -1)
+ perror("unable to read pipe");
+
+ execvp(argv[0], (char **)argv);
+
+ perror(argv[0]);
+ exit(-1);
}
child_pid = pid;
+
+ if (!system_wide)
+ target_pid = pid;
+
+ close(child_ready_pipe[1]);
+ close(go_pipe[0]);
+ /*
+ * wait for child to settle
+ */
+ if (read(child_ready_pipe[0], &buf, 1) == -1) {
+ perror("unable to read pipe");
+ exit(-1);
+ }
+ close(child_ready_pipe[0]);
}
+
+ if ((!system_wide && !inherit) || profile_cpu != -1) {
+ open_counters(profile_cpu, target_pid);
+ } else {
+ nr_cpus = read_cpu_map();
+ for (i = 0; i < nr_cpus; i++)
+ open_counters(cpumap[i], target_pid);
+ }
+
+ if (file_new) {
+ err = perf_header__write(&session->header, output, false);
+ if (err < 0)
+ return err;
+ }
+
+ post_processing_offset = lseek(output, 0, SEEK_CUR);
+
+ err = event__synthesize_kernel_mmap(process_synthesized_event,
+ session, "_text");
+ if (err < 0)
+ err = event__synthesize_kernel_mmap(process_synthesized_event,
+ session, "_stext");
+ if (err < 0) {
+ pr_err("Couldn't record kernel reference relocation symbol.\n");
+ return err;
+ }
+
+ err = event__synthesize_modules(process_synthesized_event, session);
+ if (err < 0) {
+ pr_err("Couldn't record kernel reference relocation symbol.\n");
+ return err;
+ }
+
+ if (!system_wide && profile_cpu == -1)
+ event__synthesize_thread(target_pid, process_synthesized_event,
+ session);
+ else
+ event__synthesize_threads(process_synthesized_event, session);
+
if (realtime_prio) {
struct sched_param param;
}
}
+ /*
+ * Let the child rip
+ */
+ if (forks)
+ close(go_pipe[1]);
+
for (;;) {
int hits = samples;
if (hits == samples) {
if (done)
break;
- ret = poll(event_array, nr_poll, -1);
+ err = poll(event_array, nr_poll, -1);
waking++;
}
{
int counter;
- symbol__init(0);
-
argc = parse_options(argc, argv, options, record_usage,
- PARSE_OPT_STOP_AT_NON_OPTION);
- if (!argc && target_pid == -1 && !system_wide)
+ PARSE_OPT_STOP_AT_NON_OPTION);
+ if (!argc && target_pid == -1 && !system_wide && profile_cpu == -1)
usage_with_options(record_usage, options);
+ symbol__init();
+
if (!nr_counters) {
nr_counters = 1;
attrs[0].type = PERF_TYPE_HARDWARE;