static int process_event(struct perf_evlist **pevlist, union perf_event *event)
{
struct perf_sample sample;
if (event->header.type == PERF_RECORD_HEADER_ATTR) {
if (perf_event__process_attr(NULL, event, pevlist)) {
pr_debug("perf_event__process_attr failed\n");
return -1;
}
return 0;
}
if (event->header.type >= PERF_RECORD_USER_TYPE_START)
return -1;
if (!*pevlist)
return -1;
if (perf_evlist__parse_sample(*pevlist, event, &sample)) {
pr_debug("perf_evlist__parse_sample failed\n");
return -1;
}
return 0;
}
static int process_sample_event(struct machine *machine,
struct perf_evlist *evlist,
union perf_event *event, struct state *state)
{
struct perf_sample sample;
struct thread *thread;
u8 cpumode;
int ret;
if (perf_evlist__parse_sample(evlist, event, &sample)) {
pr_debug("perf_evlist__parse_sample failed\n");
return -1;
}
thread = machine__findnew_thread(machine, sample.pid, sample.tid);
if (!thread) {
pr_debug("machine__findnew_thread failed\n");
return -1;
}
cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
ret = read_object_code(sample.ip, READLEN, cpumode, thread, state);
thread__put(thread);
return ret;
}
static int process_sample_event(struct machine *machine,
struct perf_evlist *evlist,
union perf_event *event, struct state *state)
{
struct perf_sample sample;
struct thread *thread;
int ret;
if (perf_evlist__parse_sample(evlist, event, &sample)) {
pr_debug("perf_evlist__parse_sample failed\n");
return -1;
}
thread = machine__findnew_thread(machine, sample.pid, sample.tid);
if (!thread) {
pr_debug("machine__findnew_thread failed\n");
return -1;
}
ret = read_object_code(sample.ip, READLEN, sample.cpumode, thread, state);
thread__put(thread);
return ret;
}
int test__PERF_RECORD(int subtest __maybe_unused)
{
struct record_opts opts = {
.target = {
.uid = UINT_MAX,
.uses_mmap = true,
},
.no_buffering = true,
.mmap_pages = 256,
};
cpu_set_t cpu_mask;
size_t cpu_mask_size = sizeof(cpu_mask);
struct perf_evlist *evlist = perf_evlist__new_dummy();
struct perf_evsel *evsel;
struct perf_sample sample;
const char *cmd = "sleep";
const char *argv[] = { cmd, "1", NULL, };
char *bname, *mmap_filename;
u64 prev_time = 0;
bool found_cmd_mmap = false,
found_libc_mmap = false,
found_vdso_mmap = false,
found_ld_mmap = false;
int err = -1, errs = 0, i, wakeups = 0;
u32 cpu;
int total_events = 0, nr_events[PERF_RECORD_MAX] = { 0, };
char sbuf[STRERR_BUFSIZE];
if (evlist == NULL) /* Fallback for kernels lacking PERF_COUNT_SW_DUMMY */
evlist = perf_evlist__new_default();
if (evlist == NULL || argv == NULL) {
pr_debug("Not enough memory to create evlist\n");
goto out;
}
/*
* Create maps of threads and cpus to monitor. In this case
* we start with all threads and cpus (-1, -1) but then in
* perf_evlist__prepare_workload we'll fill in the only thread
* we're monitoring, the one forked there.
*/
err = perf_evlist__create_maps(evlist, &opts.target);
if (err < 0) {
pr_debug("Not enough memory to create thread/cpu maps\n");
goto out_delete_evlist;
}
/*
* Prepare the workload in argv[] to run, it'll fork it, and then wait
* for perf_evlist__start_workload() to exec it. This is done this way
* so that we have time to open the evlist (calling sys_perf_event_open
* on all the fds) and then mmap them.
*/
err = perf_evlist__prepare_workload(evlist, &opts.target, argv, false, NULL);
if (err < 0) {
pr_debug("Couldn't run the workload!\n");
goto out_delete_evlist;
}
/*
* Config the evsels, setting attr->comm on the first one, etc.
*/
evsel = perf_evlist__first(evlist);
perf_evsel__set_sample_bit(evsel, CPU);
perf_evsel__set_sample_bit(evsel, TID);
perf_evsel__set_sample_bit(evsel, TIME);
perf_evlist__config(evlist, &opts);
err = sched__get_first_possible_cpu(evlist->workload.pid, &cpu_mask);
if (err < 0) {
pr_debug("sched__get_first_possible_cpu: %s\n",
strerror_r(errno, sbuf, sizeof(sbuf)));
goto out_delete_evlist;
}
cpu = err;
/*
* So that we can check perf_sample.cpu on all the samples.
*/
if (sched_setaffinity(evlist->workload.pid, cpu_mask_size, &cpu_mask) < 0) {
pr_debug("sched_setaffinity: %s\n",
strerror_r(errno, sbuf, sizeof(sbuf)));
goto out_delete_evlist;
}
/*
* Call sys_perf_event_open on all the fds on all the evsels,
* grouping them if asked to.
*/
err = perf_evlist__open(evlist);
if (err < 0) {
pr_debug("perf_evlist__open: %s\n",
strerror_r(errno, sbuf, sizeof(sbuf)));
goto out_delete_evlist;
}
/*
* mmap the first fd on a given CPU and ask for events for the other
* fds in the same CPU to be injected in the same mmap ring buffer
* (using ioctl(PERF_EVENT_IOC_SET_OUTPUT)).
*/
err = perf_evlist__mmap(evlist, opts.mmap_pages, false);
if (err < 0) {
pr_debug("perf_evlist__mmap: %s\n",
strerror_r(errno, sbuf, sizeof(sbuf)));
goto out_delete_evlist;
}
/*
* Now that all is properly set up, enable the events, they will
* count just on workload.pid, which will start...
*/
perf_evlist__enable(evlist);
/*
* Now!
*/
perf_evlist__start_workload(evlist);
while (1) {
int before = total_events;
for (i = 0; i < evlist->nr_mmaps; i++) {
union perf_event *event;
while ((event = perf_evlist__mmap_read(evlist, i)) != NULL) {
const u32 type = event->header.type;
const char *name = perf_event__name(type);
++total_events;
if (type < PERF_RECORD_MAX)
nr_events[type]++;
err = perf_evlist__parse_sample(evlist, event, &sample);
if (err < 0) {
if (verbose)
perf_event__fprintf(event, stderr);
pr_debug("Couldn't parse sample\n");
goto out_delete_evlist;
}
if (verbose) {
pr_info("%" PRIu64" %d ", sample.time, sample.cpu);
perf_event__fprintf(event, stderr);
}
if (prev_time > sample.time) {
pr_debug("%s going backwards in time, prev=%" PRIu64 ", curr=%" PRIu64 "\n",
name, prev_time, sample.time);
++errs;
}
prev_time = sample.time;
if (sample.cpu != cpu) {
pr_debug("%s with unexpected cpu, expected %d, got %d\n",
name, cpu, sample.cpu);
++errs;
}
if ((pid_t)sample.pid != evlist->workload.pid) {
pr_debug("%s with unexpected pid, expected %d, got %d\n",
name, evlist->workload.pid, sample.pid);
++errs;
}
if ((pid_t)sample.tid != evlist->workload.pid) {
pr_debug("%s with unexpected tid, expected %d, got %d\n",
name, evlist->workload.pid, sample.tid);
++errs;
}
if ((type == PERF_RECORD_COMM ||
type == PERF_RECORD_MMAP ||
type == PERF_RECORD_MMAP2 ||
type == PERF_RECORD_FORK ||
type == PERF_RECORD_EXIT) &&
(pid_t)event->comm.pid != evlist->workload.pid) {
pr_debug("%s with unexpected pid/tid\n", name);
++errs;
}
if ((type == PERF_RECORD_COMM ||
type == PERF_RECORD_MMAP ||
type == PERF_RECORD_MMAP2) &&
event->comm.pid != event->comm.tid) {
pr_debug("%s with different pid/tid!\n", name);
++errs;
}
switch (type) {
case PERF_RECORD_COMM:
if (strcmp(event->comm.comm, cmd)) {
pr_debug("%s with unexpected comm!\n", name);
++errs;
}
break;
case PERF_RECORD_EXIT:
goto found_exit;
case PERF_RECORD_MMAP:
mmap_filename = event->mmap.filename;
goto check_bname;
case PERF_RECORD_MMAP2:
mmap_filename = event->mmap2.filename;
check_bname:
bname = strrchr(mmap_filename, '/');
if (bname != NULL) {
if (!found_cmd_mmap)
found_cmd_mmap = !strcmp(bname + 1, cmd);
if (!found_libc_mmap)
found_libc_mmap = !strncmp(bname + 1, "libc", 4);
if (!found_ld_mmap)
found_ld_mmap = !strncmp(bname + 1, "ld", 2);
} else if (!found_vdso_mmap)
found_vdso_mmap = !strcmp(mmap_filename, "[vdso]");
break;
case PERF_RECORD_SAMPLE:
/* Just ignore samples for now */
break;
default:
pr_debug("Unexpected perf_event->header.type %d!\n",
type);
++errs;
}
perf_evlist__mmap_consume(evlist, i);
}
}
/*
* We don't use poll here because at least at 3.1 times the
* PERF_RECORD_{!SAMPLE} events don't honour
* perf_event_attr.wakeup_events, just PERF_EVENT_SAMPLE does.
*/
if (total_events == before && false)
perf_evlist__poll(evlist, -1);
sleep(1);
if (++wakeups > 5) {
pr_debug("No PERF_RECORD_EXIT event!\n");
break;
}
}
found_exit:
if (nr_events[PERF_RECORD_COMM] > 1) {
pr_debug("Excessive number of PERF_RECORD_COMM events!\n");
++errs;
}
if (nr_events[PERF_RECORD_COMM] == 0) {
pr_debug("Missing PERF_RECORD_COMM for %s!\n", cmd);
++errs;
}
if (!found_cmd_mmap) {
pr_debug("PERF_RECORD_MMAP for %s missing!\n", cmd);
++errs;
}
if (!found_libc_mmap) {
pr_debug("PERF_RECORD_MMAP for %s missing!\n", "libc");
++errs;
}
if (!found_ld_mmap) {
pr_debug("PERF_RECORD_MMAP for %s missing!\n", "ld");
++errs;
}
if (!found_vdso_mmap) {
pr_debug("PERF_RECORD_MMAP for %s missing!\n", "[vdso]");
++errs;
}
out_delete_evlist:
perf_evlist__delete(evlist);
out:
return (err < 0 || errs > 0) ? -1 : 0;
}
/*
* This test will generate random numbers of calls to some getpid syscalls,
* then establish an mmap for a group of events that are created to monitor
* the syscalls.
*
* It will receive the events, using mmap, use its PERF_SAMPLE_ID generated
* sample.id field to map back to its respective perf_evsel instance.
*
* Then it checks if the number of syscalls reported as perf events by
* the kernel corresponds to the number of syscalls made.
*/
int test__basic_mmap(void)
{
int err = -1;
union perf_event *event;
struct thread_map *threads;
struct cpu_map *cpus;
struct perf_evlist *evlist;
cpu_set_t cpu_set;
const char *syscall_names[] = { "getsid", "getppid", "getpgrp",
"getpgid", };
pid_t (*syscalls[])(void) = { (void *)getsid, getppid, getpgrp,
(void*)getpgid };
#define nsyscalls ARRAY_SIZE(syscall_names)
unsigned int nr_events[nsyscalls],
expected_nr_events[nsyscalls], i, j;
struct perf_evsel *evsels[nsyscalls], *evsel;
threads = thread_map__new(-1, getpid(), UINT_MAX);
if (threads == NULL) {
pr_debug("thread_map__new\n");
return -1;
}
cpus = cpu_map__new(NULL);
if (cpus == NULL) {
pr_debug("cpu_map__new\n");
goto out_free_threads;
}
CPU_ZERO(&cpu_set);
CPU_SET(cpus->map[0], &cpu_set);
sched_setaffinity(0, sizeof(cpu_set), &cpu_set);
if (sched_setaffinity(0, sizeof(cpu_set), &cpu_set) < 0) {
pr_debug("sched_setaffinity() failed on CPU %d: %s ",
cpus->map[0], strerror(errno));
goto out_free_cpus;
}
evlist = perf_evlist__new();
if (evlist == NULL) {
pr_debug("perf_evlist__new\n");
goto out_free_cpus;
}
perf_evlist__set_maps(evlist, cpus, threads);
for (i = 0; i < nsyscalls; ++i) {
char name[64];
snprintf(name, sizeof(name), "sys_enter_%s", syscall_names[i]);
evsels[i] = perf_evsel__newtp("syscalls", name);
if (evsels[i] == NULL) {
pr_debug("perf_evsel__new\n");
goto out_delete_evlist;
}
evsels[i]->attr.wakeup_events = 1;
perf_evsel__set_sample_id(evsels[i], false);
perf_evlist__add(evlist, evsels[i]);
if (perf_evsel__open(evsels[i], cpus, threads) < 0) {
pr_debug("failed to open counter: %s, "
"tweak /proc/sys/kernel/perf_event_paranoid?\n",
strerror(errno));
goto out_delete_evlist;
}
nr_events[i] = 0;
expected_nr_events[i] = 1 + rand() % 127;
}
if (perf_evlist__mmap(evlist, 128, true) < 0) {
pr_debug("failed to mmap events: %d (%s)\n", errno,
strerror(errno));
goto out_delete_evlist;
}
for (i = 0; i < nsyscalls; ++i)
for (j = 0; j < expected_nr_events[i]; ++j) {
int foo = syscalls[i]();
++foo;
}
while ((event = perf_evlist__mmap_read(evlist, 0)) != NULL) {
struct perf_sample sample;
if (event->header.type != PERF_RECORD_SAMPLE) {
pr_debug("unexpected %s event\n",
perf_event__name(event->header.type));
goto out_delete_evlist;
}
err = perf_evlist__parse_sample(evlist, event, &sample);
if (err) {
pr_err("Can't parse sample, err = %d\n", err);
goto out_delete_evlist;
}
err = -1;
evsel = perf_evlist__id2evsel(evlist, sample.id);
if (evsel == NULL) {
pr_debug("event with id %" PRIu64
" doesn't map to an evsel\n", sample.id);
goto out_delete_evlist;
}
nr_events[evsel->idx]++;
perf_evlist__mmap_consume(evlist, 0);
}
err = 0;
evlist__for_each(evlist, evsel) {
if (nr_events[evsel->idx] != expected_nr_events[evsel->idx]) {
pr_debug("expected %d %s events, got %d\n",
expected_nr_events[evsel->idx],
perf_evsel__name(evsel), nr_events[evsel->idx]);
err = -1;
goto out_delete_evlist;
}
}
out_delete_evlist:
perf_evlist__delete(evlist);
cpus = NULL;
threads = NULL;
out_free_cpus:
cpu_map__delete(cpus);
out_free_threads:
thread_map__delete(threads);
return err;
}
/*
* This test will open software clock events (cpu-clock, task-clock)
* then check their frequency -> period conversion has no artifact of
* setting period to 1 forcefully.
*/
static int __test__sw_clock_freq(enum perf_sw_ids clock_id)
{
int i, err = -1;
volatile int tmp = 0;
u64 total_periods = 0;
int nr_samples = 0;
char sbuf[STRERR_BUFSIZE];
union perf_event *event;
struct perf_evsel *evsel;
struct perf_evlist *evlist;
struct perf_event_attr attr = {
.type = PERF_TYPE_SOFTWARE,
.config = clock_id,
.sample_type = PERF_SAMPLE_PERIOD,
.exclude_kernel = 1,
.disabled = 1,
.freq = 1,
};
struct cpu_map *cpus;
struct thread_map *threads;
attr.sample_freq = 500;
evlist = perf_evlist__new();
if (evlist == NULL) {
pr_debug("perf_evlist__new\n");
return -1;
}
evsel = perf_evsel__new(&attr);
if (evsel == NULL) {
pr_debug("perf_evsel__new\n");
goto out_delete_evlist;
}
perf_evlist__add(evlist, evsel);
cpus = cpu_map__dummy_new();
threads = thread_map__new_by_tid(getpid());
if (!cpus || !threads) {
err = -ENOMEM;
pr_debug("Not enough memory to create thread/cpu maps\n");
goto out_free_maps;
}
perf_evlist__set_maps(evlist, cpus, threads);
cpus = NULL;
threads = NULL;
if (perf_evlist__open(evlist)) {
const char *knob = "/proc/sys/kernel/perf_event_max_sample_rate";
err = -errno;
pr_debug("Couldn't open evlist: %s\nHint: check %s, using %" PRIu64 " in this test.\n",
str_error_r(errno, sbuf, sizeof(sbuf)),
knob, (u64)attr.sample_freq);
goto out_delete_evlist;
}
err = perf_evlist__mmap(evlist, 128);
if (err < 0) {
pr_debug("failed to mmap event: %d (%s)\n", errno,
str_error_r(errno, sbuf, sizeof(sbuf)));
goto out_delete_evlist;
}
perf_evlist__enable(evlist);
/* collect samples */
for (i = 0; i < NR_LOOPS; i++)
tmp++;
perf_evlist__disable(evlist);
while ((event = perf_evlist__mmap_read(evlist, 0)) != NULL) {
struct perf_sample sample;
if (event->header.type != PERF_RECORD_SAMPLE)
goto next_event;
err = perf_evlist__parse_sample(evlist, event, &sample);
if (err < 0) {
pr_debug("Error during parse sample\n");
goto out_delete_evlist;
}
total_periods += sample.period;
nr_samples++;
next_event:
perf_evlist__mmap_consume(evlist, 0);
}
if ((u64) nr_samples == total_periods) {
pr_debug("All (%d) samples have period value of 1!\n",
nr_samples);
err = -1;
}
out_free_maps:
cpu_map__put(cpus);
thread_map__put(threads);
out_delete_evlist:
perf_evlist__delete(evlist);
return err;
}
int test__sw_clock_freq(struct test *test __maybe_unused, int subtest __maybe_unused)
{
int ret;
ret = __test__sw_clock_freq(PERF_COUNT_SW_CPU_CLOCK);
if (!ret)
ret = __test__sw_clock_freq(PERF_COUNT_SW_TASK_CLOCK);
return ret;
}
