1 files changed, 263 insertions, 47 deletions
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index d22c5df1701e..bc144388f892 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -395,6 +395,7 @@ void evsel__init(struct evsel *evsel,
 	evsel->group_pmu_name = NULL;
 	evsel->skippable     = false;
 	evsel->alternate_hw_config = PERF_COUNT_HW_MAX;
+	evsel->script_output_type = -1; // FIXME: OUTPUT_TYPE_UNSET, see builtin-script.c
 }
 
 struct evsel *evsel__new_idx(struct perf_event_attr *attr, int idx)
@@ -454,7 +455,7 @@ static int evsel__copy_config_terms(struct evsel *dst, struct evsel *src)
  * The assumption is that @orig is not configured nor opened yet.
  * So we only care about the attributes that can be set while it's parsed.
  */
-struct evsel *evsel__clone(struct evsel *orig)
+struct evsel *evsel__clone(struct evsel *dest, struct evsel *orig)
 {
 	struct evsel *evsel;
 
@@ -467,7 +468,11 @@ struct evsel *evsel__clone(struct evsel *orig)
 	if (orig->bpf_obj)
 		return NULL;
 
-	evsel = evsel__new(&orig->core.attr);
+	if (dest)
+		evsel = dest;
+	else
+		evsel = evsel__new(&orig->core.attr);
+
 	if (evsel == NULL)
 		return NULL;
 
@@ -512,11 +517,12 @@ struct evsel *evsel__clone(struct evsel *orig)
 	evsel->core.leader = orig->core.leader;
 
 	evsel->max_events = orig->max_events;
-	free((char *)evsel->unit);
-	evsel->unit = strdup(orig->unit);
-	if (evsel->unit == NULL)
-		goto out_err;
-
+	zfree(&evsel->unit);
+	if (orig->unit) {
+		evsel->unit = strdup(orig->unit);
+		if (evsel->unit == NULL)
+			goto out_err;
+	}
 	evsel->scale = orig->scale;
 	evsel->snapshot = orig->snapshot;
 	evsel->per_pkg = orig->per_pkg;
@@ -544,53 +550,101 @@ out_err:
 	return NULL;
 }
 
+static int trace_event__id(const char *sys, const char *name)
+{
+	char *tp_dir = get_events_file(sys);
+	char path[PATH_MAX];
+	int id, err;
+
+	if (!tp_dir)
+		return -1;
+
+	scnprintf(path, PATH_MAX, "%s/%s/id", tp_dir, name);
+	put_events_file(tp_dir);
+	err = filename__read_int(path, &id);
+	if (err)
+		return err;
+
+	return id;
+}
+
 /*
  * Returns pointer with encoded error via <linux/err.h> interface.
  */
-#ifdef HAVE_LIBTRACEEVENT
 struct evsel *evsel__newtp_idx(const char *sys, const char *name, int idx, bool format)
 {
+	struct perf_event_attr attr = {
+		.type	       = PERF_TYPE_TRACEPOINT,
+		.sample_type   = (PERF_SAMPLE_RAW | PERF_SAMPLE_TIME |
+				PERF_SAMPLE_CPU | PERF_SAMPLE_PERIOD),
+	};
 	struct evsel *evsel = zalloc(perf_evsel__object.size);
-	int err = -ENOMEM;
+	int err = -ENOMEM, id = -1;
 
-	if (evsel == NULL) {
+	if (evsel == NULL)
 		goto out_err;
-	} else {
-		struct perf_event_attr attr = {
-			.type	       = PERF_TYPE_TRACEPOINT,
-			.sample_type   = (PERF_SAMPLE_RAW | PERF_SAMPLE_TIME |
-					  PERF_SAMPLE_CPU | PERF_SAMPLE_PERIOD),
-		};
 
-		if (asprintf(&evsel->name, "%s:%s", sys, name) < 0)
-			goto out_free;
 
-		event_attr_init(&attr);
+	if (asprintf(&evsel->name, "%s:%s", sys, name) < 0)
+		goto out_free;
 
-		if (format) {
-			evsel->tp_format = trace_event__tp_format(sys, name);
-			if (IS_ERR(evsel->tp_format)) {
-				err = PTR_ERR(evsel->tp_format);
-				goto out_free;
-			}
-			attr.config = evsel->tp_format->id;
-		} else {
-			attr.config = (__u64) -1;
-		}
+#ifdef HAVE_LIBTRACEEVENT
+	evsel->tp_sys = strdup(sys);
+	if (!evsel->tp_sys)
+		goto out_free;
 
+	evsel->tp_name = strdup(name);
+	if (!evsel->tp_name)
+		goto out_free;
+#endif
 
-		attr.sample_period = 1;
-		evsel__init(evsel, &attr, idx);
-	}
+	event_attr_init(&attr);
 
+	if (format) {
+		id = trace_event__id(sys, name);
+		if (id < 0) {
+			err = id;
+			goto out_free;
+		}
+	}
+	attr.config = (__u64)id;
+	attr.sample_period = 1;
+	evsel__init(evsel, &attr, idx);
 	return evsel;
 
 out_free:
 	zfree(&evsel->name);
+#ifdef HAVE_LIBTRACEEVENT
+	zfree(&evsel->tp_sys);
+	zfree(&evsel->tp_name);
+#endif
 	free(evsel);
 out_err:
 	return ERR_PTR(err);
 }
+
+#ifdef HAVE_LIBTRACEEVENT
+struct tep_event *evsel__tp_format(struct evsel *evsel)
+{
+	struct tep_event *tp_format = evsel->tp_format;
+
+	if (tp_format)
+		return tp_format;
+
+	if (evsel->core.attr.type != PERF_TYPE_TRACEPOINT)
+		return NULL;
+
+	tp_format = trace_event__tp_format(evsel->tp_sys, evsel->tp_name);
+	if (IS_ERR(tp_format)) {
+		int err = -PTR_ERR(evsel->tp_format);
+
+		pr_err("Error getting tracepoint format '%s' '%s'(%d)\n",
+			evsel__name(evsel), strerror(err), err);
+		return NULL;
+	}
+	evsel->tp_format = tp_format;
+	return evsel->tp_format;
+}
 #endif
 
 const char *const evsel__hw_names[PERF_COUNT_HW_MAX] = {
@@ -1103,6 +1157,9 @@ static void evsel__apply_config_terms(struct evsel *evsel,
 		case EVSEL__CONFIG_TERM_AUX_OUTPUT:
 			attr->aux_output = term->val.aux_output ? 1 : 0;
 			break;
+		case EVSEL__CONFIG_TERM_AUX_ACTION:
+			/* Already applied by auxtrace */
+			break;
 		case EVSEL__CONFIG_TERM_AUX_SAMPLE_SIZE:
 			/* Already applied by auxtrace */
 			break;
@@ -1587,6 +1644,10 @@ void evsel__exit(struct evsel *evsel)
 	perf_thread_map__put(evsel->core.threads);
 	zfree(&evsel->group_name);
 	zfree(&evsel->name);
+#ifdef HAVE_LIBTRACEEVENT
+	zfree(&evsel->tp_sys);
+	zfree(&evsel->tp_name);
+#endif
 	zfree(&evsel->filter);
 	zfree(&evsel->group_pmu_name);
 	zfree(&evsel->unit);
@@ -2090,16 +2151,17 @@ int evsel__prepare_open(struct evsel *evsel, struct perf_cpu_map *cpus,
 	return err;
 }
 
-static bool has_attr_feature(struct perf_event_attr *attr, unsigned long flags)
+static bool __has_attr_feature(struct perf_event_attr *attr,
+			       struct perf_cpu cpu, unsigned long flags)
 {
-	int fd = syscall(SYS_perf_event_open, attr, /*pid=*/0, /*cpu=*/-1,
+	int fd = syscall(SYS_perf_event_open, attr, /*pid=*/0, cpu.cpu,
 			 /*group_fd=*/-1, flags);
 	close(fd);
 
 	if (fd < 0) {
 		attr->exclude_kernel = 1;
 
-		fd = syscall(SYS_perf_event_open, attr, /*pid=*/0, /*cpu=*/-1,
+		fd = syscall(SYS_perf_event_open, attr, /*pid=*/0, cpu.cpu,
 			     /*group_fd=*/-1, flags);
 		close(fd);
 	}
@@ -2107,7 +2169,7 @@ static bool has_attr_feature(struct perf_event_attr *attr, unsigned long flags)
 	if (fd < 0) {
 		attr->exclude_hv = 1;
 
-		fd = syscall(SYS_perf_event_open, attr, /*pid=*/0, /*cpu=*/-1,
+		fd = syscall(SYS_perf_event_open, attr, /*pid=*/0, cpu.cpu,
 			     /*group_fd=*/-1, flags);
 		close(fd);
 	}
@@ -2115,7 +2177,7 @@ static bool has_attr_feature(struct perf_event_attr *attr, unsigned long flags)
 	if (fd < 0) {
 		attr->exclude_guest = 1;
 
-		fd = syscall(SYS_perf_event_open, attr, /*pid=*/0, /*cpu=*/-1,
+		fd = syscall(SYS_perf_event_open, attr, /*pid=*/0, cpu.cpu,
 			     /*group_fd=*/-1, flags);
 		close(fd);
 	}
@@ -2127,6 +2189,13 @@ static bool has_attr_feature(struct perf_event_attr *attr, unsigned long flags)
 	return fd >= 0;
 }
 
+static bool has_attr_feature(struct perf_event_attr *attr, unsigned long flags)
+{
+	struct perf_cpu cpu = {.cpu = -1};
+
+	return __has_attr_feature(attr, cpu, flags);
+}
+
 static void evsel__detect_missing_pmu_features(struct evsel *evsel)
 {
 	struct perf_event_attr attr = {
@@ -2215,7 +2284,65 @@ found:
 	errno = old_errno;
 }
 
-static bool evsel__detect_missing_features(struct evsel *evsel)
+static bool evsel__probe_aux_action(struct evsel *evsel, struct perf_cpu cpu)
+{
+	struct perf_event_attr attr = evsel->core.attr;
+	int old_errno = errno;
+
+	attr.disabled = 1;
+	attr.aux_start_paused = 1;
+
+	if (__has_attr_feature(&attr, cpu, /*flags=*/0)) {
+		errno = old_errno;
+		return true;
+	}
+
+	/*
+	 * EOPNOTSUPP means the kernel supports the feature but the PMU does
+	 * not, so keep that distinction if possible.
+	 */
+	if (errno != EOPNOTSUPP)
+		errno = old_errno;
+
+	return false;
+}
+
+static void evsel__detect_missing_aux_action_feature(struct evsel *evsel, struct perf_cpu cpu)
+{
+	static bool detection_done;
+	struct evsel *leader;
+
+	/*
+	 * Don't bother probing aux_action if it is not being used or has been
+	 * probed before.
+	 */
+	if (!evsel->core.attr.aux_action || detection_done)
+		return;
+
+	detection_done = true;
+
+	/*
+	 * The leader is an AUX area event. If it has failed, assume the feature
+	 * is not supported.
+	 */
+	leader = evsel__leader(evsel);
+	if (evsel == leader) {
+		perf_missing_features.aux_action = true;
+		return;
+	}
+
+	/*
+	 * AUX area event with aux_action must have been opened successfully
+	 * already, so feature is supported.
+	 */
+	if (leader->core.attr.aux_action)
+		return;
+
+	if (!evsel__probe_aux_action(leader, cpu))
+		perf_missing_features.aux_action = true;
+}
+
+static bool evsel__detect_missing_features(struct evsel *evsel, struct perf_cpu cpu)
 {
 	static bool detection_done = false;
 	struct perf_event_attr attr = {
@@ -2225,6 +2352,8 @@ static bool evsel__detect_missing_features(struct evsel *evsel)
 	};
 	int old_errno;
 
+	evsel__detect_missing_aux_action_feature(evsel, cpu);
+
 	evsel__detect_missing_pmu_features(evsel);
 
 	if (evsel__has_br_stack(evsel))
@@ -2439,6 +2568,7 @@ static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
 	int idx, thread, nthreads;
 	int pid = -1, err, old_errno;
 	enum rlimit_action set_rlimit = NO_CHANGE;
+	struct perf_cpu cpu;
 
 	if (evsel__is_retire_lat(evsel))
 		return tpebs_start(evsel->evlist);
@@ -2476,6 +2606,7 @@ fallback_missing_features:
 	}
 
 	for (idx = start_cpu_map_idx; idx < end_cpu_map_idx; idx++) {
+		cpu = perf_cpu_map__cpu(cpus, idx);
 
 		for (thread = 0; thread < nthreads; thread++) {
 			int fd, group_fd;
@@ -2496,10 +2627,9 @@ retry_open:
 
 			/* Debug message used by test scripts */
 			pr_debug2_peo("sys_perf_event_open: pid %d  cpu %d  group_fd %d  flags %#lx",
-				pid, perf_cpu_map__cpu(cpus, idx).cpu, group_fd, evsel->open_flags);
+				pid, cpu.cpu, group_fd, evsel->open_flags);
 
-			fd = sys_perf_event_open(&evsel->core.attr, pid,
-						perf_cpu_map__cpu(cpus, idx).cpu,
+			fd = sys_perf_event_open(&evsel->core.attr, pid, cpu.cpu,
 						group_fd, evsel->open_flags);
 
 			FD(evsel, idx, thread) = fd;
@@ -2515,8 +2645,7 @@ retry_open:
 			bpf_counter__install_pe(evsel, idx, fd);
 
 			if (unlikely(test_attr__enabled())) {
-				test_attr__open(&evsel->core.attr, pid,
-						perf_cpu_map__cpu(cpus, idx),
+				test_attr__open(&evsel->core.attr, pid, cpu,
 						fd, group_fd, evsel->open_flags);
 			}
 
@@ -2571,7 +2700,7 @@ try_fallback:
 	if (err == -EMFILE && rlimit__increase_nofile(&set_rlimit))
 		goto retry_open;
 
-	if (err == -EINVAL && evsel__detect_missing_features(evsel))
+	if (err == -EINVAL && evsel__detect_missing_features(evsel, cpu))
 		goto fallback_missing_features;
 
 	if (evsel__precise_ip_fallback(evsel))
@@ -3218,12 +3347,16 @@ u16 evsel__id_hdr_size(const struct evsel *evsel)
 #ifdef HAVE_LIBTRACEEVENT
 struct tep_format_field *evsel__field(struct evsel *evsel, const char *name)
 {
-	return tep_find_field(evsel->tp_format, name);
+	struct tep_event *tp_format = evsel__tp_format(evsel);
+
+	return tp_format ? tep_find_field(tp_format, name) : NULL;
 }
 
 struct tep_format_field *evsel__common_field(struct evsel *evsel, const char *name)
 {
-	return tep_find_common_field(evsel->tp_format, name);
+	struct tep_event *tp_format = evsel__tp_format(evsel);
+
+	return tp_format ? tep_find_common_field(tp_format, name) : NULL;
 }
 
 void *evsel__rawptr(struct evsel *evsel, struct perf_sample *sample, const char *name)
@@ -3448,6 +3581,78 @@ static bool find_process(const char *name)
 	return ret ? false : true;
 }
 
+static int dump_perf_event_processes(char *msg, size_t size)
+{
+	DIR *proc_dir;
+	struct dirent *proc_entry;
+	int printed = 0;
+
+	proc_dir = opendir(procfs__mountpoint());
+	if (!proc_dir)
+		return 0;
+
+	/* Walk through the /proc directory. */
+	while ((proc_entry = readdir(proc_dir)) != NULL) {
+		char buf[256];
+		DIR *fd_dir;
+		struct dirent *fd_entry;
+		int fd_dir_fd;
+
+		if (proc_entry->d_type != DT_DIR ||
+		    !isdigit(proc_entry->d_name[0]) ||
+		    strlen(proc_entry->d_name) > sizeof(buf) - 4)
+			continue;
+
+		scnprintf(buf, sizeof(buf), "%s/fd", proc_entry->d_name);
+		fd_dir_fd = openat(dirfd(proc_dir), buf, O_DIRECTORY);
+		if (fd_dir_fd == -1)
+			continue;
+		fd_dir = fdopendir(fd_dir_fd);
+		if (!fd_dir) {
+			close(fd_dir_fd);
+			continue;
+		}
+		while ((fd_entry = readdir(fd_dir)) != NULL) {
+			ssize_t link_size;
+
+			if (fd_entry->d_type != DT_LNK)
+				continue;
+			link_size = readlinkat(fd_dir_fd, fd_entry->d_name, buf, sizeof(buf));
+			if (link_size < 0)
+				continue;
+			/* Take care as readlink doesn't null terminate the string. */
+			if (!strncmp(buf, "anon_inode:[perf_event]", link_size)) {
+				int cmdline_fd;
+				ssize_t cmdline_size;
+
+				scnprintf(buf, sizeof(buf), "%s/cmdline", proc_entry->d_name);
+				cmdline_fd = openat(dirfd(proc_dir), buf, O_RDONLY);
+				if (cmdline_fd == -1)
+					continue;
+				cmdline_size = read(cmdline_fd, buf, sizeof(buf) - 1);
+				close(cmdline_fd);
+				if (cmdline_size < 0)
+					continue;
+				buf[cmdline_size] = '\0';
+				for (ssize_t i = 0; i < cmdline_size; i++) {
+					if (buf[i] == '\0')
+						buf[i] = ' ';
+				}
+
+				if (printed == 0)
+					printed += scnprintf(msg, size, "Possible processes:\n");
+
+				printed += scnprintf(msg + printed, size - printed,
+						"%s %s\n", proc_entry->d_name, buf);
+				break;
+			}
+		}
+		closedir(fd_dir);
+	}
+	closedir(proc_dir);
+	return printed;
+}
+
 int __weak arch_evsel__open_strerror(struct evsel *evsel __maybe_unused,
 				     char *msg __maybe_unused,
 				     size_t size __maybe_unused)
@@ -3481,7 +3686,7 @@ int evsel__open_strerror(struct evsel *evsel, struct target *target,
 			printed += scnprintf(msg, size,
 				"No permission to enable %s event.\n\n", evsel__name(evsel));
 
-		return scnprintf(msg + printed, size - printed,
+		return printed + scnprintf(msg + printed, size - printed,
 		 "Consider adjusting /proc/sys/kernel/perf_event_paranoid setting to open\n"
 		 "access to performance monitoring and observability operations for processes\n"
 		 "without CAP_PERFMON, CAP_SYS_PTRACE or CAP_SYS_ADMIN Linux capability.\n"
@@ -3526,6 +3731,10 @@ int evsel__open_strerror(struct evsel *evsel, struct target *target,
 			return scnprintf(msg, size,
 	"%s: PMU Hardware doesn't support 'aux_output' feature",
 					 evsel__name(evsel));
+		if (evsel->core.attr.aux_action)
+			return scnprintf(msg, size,
+	"%s: PMU Hardware doesn't support 'aux_action' feature",
+					evsel__name(evsel));
 		if (evsel->core.attr.sample_period != 0)
 			return scnprintf(msg, size,
 	"%s: PMU Hardware doesn't support sampling/overflow-interrupts. Try 'perf stat'",
@@ -3544,6 +3753,11 @@ int evsel__open_strerror(struct evsel *evsel, struct target *target,
 			return scnprintf(msg, size,
 	"The PMU counters are busy/taken by another profiler.\n"
 	"We found oprofile daemon running, please stop it and try again.");
+		printed += scnprintf(
+			msg, size,
+			"The PMU %s counters are busy and in use by another process.\n",
+			evsel->pmu ? evsel->pmu->name : "");
+		return printed + dump_perf_event_processes(msg + printed, size - printed);
 		break;
 	case EINVAL:
 		if (evsel->core.attr.sample_type & PERF_SAMPLE_CODE_PAGE_SIZE && perf_missing_features.code_page_size)
@@ -3556,6 +3770,8 @@ int evsel__open_strerror(struct evsel *evsel, struct target *target,
 			return scnprintf(msg, size, "clockid feature not supported.");
 		if (perf_missing_features.clockid_wrong)
 			return scnprintf(msg, size, "wrong clockid (%d).", clockid);
+		if (perf_missing_features.aux_action)
+			return scnprintf(msg, size, "The 'aux_action' feature is not supported, update the kernel.");
 		if (perf_missing_features.aux_output)
 			return scnprintf(msg, size, "The 'aux_output' feature is not supported, update the kernel.");
 		if (!target__has_cpu(target))