Merge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

* 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (184 commits)
  perf probe: Clean up probe_point_lazy_walker() return value
  tracing: Fix irqoff selftest expanding max buffer
  tracing: Align 4 byte ints together in struct tracer
  tracing: Export trace_set_clr_event()
  tracing: Explain about unstable clock on resume with ring buffer warning
  ftrace/graph: Trace function entry before updating index
  ftrace: Add .ref.text as one of the safe areas to trace
  tracing: Adjust conditional expression latency formatting.
  tracing: Fix event alignment: skb:kfree_skb
  tracing: Fix event alignment: mce:mce_record
  tracing: Fix event alignment: kvm:kvm_hv_hypercall
  tracing: Fix event alignment: module:module_request
  tracing: Fix event alignment: ftrace:context_switch and ftrace:wakeup
  tracing: Remove lock_depth from event entry
  perf header: Stop using 'self'
  perf session: Use evlist/evsel for managing perf.data attributes
  perf top: Don't let events to eat up whole header line
  perf top: Fix events overflow in top command
  ring-buffer: Remove unused #include <linux/trace_irq.h>
  tracing: Add an 'overwrite' trace_option.
  ...

1 files changed, 169 insertions, 281 deletions

diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 60cac6f..6febcc1 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -18,11 +18,13 @@
 
 #include "util/header.h"
 #include "util/event.h"
+#include "util/evlist.h"
 #include "util/evsel.h"
 #include "util/debug.h"
 #include "util/session.h"
 #include "util/symbol.h"
 #include "util/cpumap.h"
+#include "util/thread_map.h"
 
 #include <unistd.h>
 #include <sched.h>
@@ -37,16 +39,14 @@ enum write_mode_t {
 
 static u64			user_interval			= ULLONG_MAX;
 static u64			default_interval		=      0;
-static u64			sample_type;
 
-static struct cpu_map		*cpus;
 static unsigned int		page_size;
 static unsigned int		mmap_pages			=    128;
 static unsigned int		user_freq 			= UINT_MAX;
 static int			freq				=   1000;
 static int			output;
 static int			pipe_output			=      0;
-static const char		*output_name			= "perf.data";
+static const char		*output_name			= NULL;
 static int			group				=      0;
 static int			realtime_prio			=      0;
 static bool			nodelay				=  false;
@@ -55,7 +55,6 @@ 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 struct thread_map	*threads;
 static pid_t			child_pid			=     -1;
 static bool			no_inherit			=  false;
 static enum write_mode_t	write_mode			= WRITE_FORCE;
@@ -66,51 +65,17 @@ static bool			sample_address			=  false;
 static bool			sample_time			=  false;
 static bool			no_buildid			=  false;
 static bool			no_buildid_cache		=  false;
+static struct perf_evlist	*evsel_list;
 
 static long			samples				=      0;
 static u64			bytes_written			=      0;
 
-static struct pollfd		*event_array;
-
-static int			nr_poll				=      0;
-static int			nr_cpu				=      0;
-
 static int			file_new			=      1;
 static off_t			post_processing_offset;
 
 static struct perf_session	*session;
 static const char		*cpu_list;
 
-struct mmap_data {
-	void			*base;
-	unsigned int		mask;
-	unsigned int		prev;
-};
-
-static struct mmap_data		mmap_array[MAX_NR_CPUS];
-
-static unsigned long mmap_read_head(struct mmap_data *md)
-{
-	struct perf_event_mmap_page *pc = md->base;
-	long head;
-
-	head = pc->data_head;
-	rmb();
-
-	return head;
-}
-
-static void mmap_write_tail(struct mmap_data *md, unsigned long tail)
-{
-	struct perf_event_mmap_page *pc = md->base;
-
-	/*
-	 * ensure all reads are done before we write the tail out.
-	 */
-	/* mb(); */
-	pc->data_tail = tail;
-}
-
 static void advance_output(size_t size)
 {
 	bytes_written += size;
@@ -131,42 +96,26 @@ static void write_output(void *buf, size_t size)
 	}
 }
 
-static int process_synthesized_event(event_t *event,
-				     struct sample_data *sample __used,
+static int process_synthesized_event(union perf_event *event,
+				     struct perf_sample *sample __used,
 				     struct perf_session *self __used)
 {
 	write_output(event, event->header.size);
 	return 0;
 }
 
-static void mmap_read(struct mmap_data *md)
+static void mmap_read(struct perf_mmap *md)
 {
-	unsigned int head = mmap_read_head(md);
+	unsigned int head = perf_mmap__read_head(md);
 	unsigned int old = md->prev;
 	unsigned char *data = md->base + page_size;
 	unsigned long size;
 	void *buf;
-	int diff;
 
-	/*
-	 * If we're further behind than half the buffer, there's a chance
-	 * the writer will bite our tail and mess up the samples under us.
-	 *
-	 * If we somehow ended up ahead of the head, we got messed up.
-	 *
-	 * In either case, truncate and restart at head.
-	 */
-	diff = head - old;
-	if (diff < 0) {
-		fprintf(stderr, "WARNING: failed to keep up with mmap data\n");
-		/*
-		 * head points to a known good entry, start there.
-		 */
-		old = head;
-	}
+	if (old == head)
+		return;
 
-	if (old != head)
-		samples++;
+	samples++;
 
 	size = head - old;
 
@@ -185,7 +134,7 @@ static void mmap_read(struct mmap_data *md)
 	write_output(buf, size);
 
 	md->prev = old;
-	mmap_write_tail(md, old);
+	perf_mmap__write_tail(md, old);
 }
 
 static volatile int done = 0;
@@ -209,53 +158,10 @@ static void sig_atexit(void)
 	kill(getpid(), signr);
 }
 
-static int group_fd;
-
-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)
+static void config_attr(struct perf_evsel *evsel, struct perf_evlist *evlist)
 {
-	char *filter = evsel->filter;
 	struct perf_event_attr *attr = &evsel->attr;
-	struct perf_header_attr *h_attr;
 	int track = !evsel->idx; /* only the first counter needs these */
-	int thread_index;
-	int ret;
-	struct {
-		u64 count;
-		u64 time_enabled;
-		u64 time_running;
-		u64 id;
-	} read_data;
-	/*
- 	 * Check if parse_single_tracepoint_event has already asked for
- 	 * PERF_SAMPLE_TIME.
- 	 *
-	 * XXX this is kludgy but short term fix for problems introduced by
-	 * eac23d1c that broke 'perf script' by having different sample_types
-	 * when using multiple tracepoint events when we use a perf binary
-	 * that tries to use sample_id_all on an older kernel.
- 	 *
- 	 * We need to move counter creation to perf_session, support
- 	 * different sample_types, etc.
- 	 */
-	bool time_needed = attr->sample_type & PERF_SAMPLE_TIME;
 
 	attr->read_format	= PERF_FORMAT_TOTAL_TIME_ENABLED |
 				  PERF_FORMAT_TOTAL_TIME_RUNNING |
@@ -263,7 +169,7 @@ static void create_counter(struct perf_evsel *evsel, int cpu)
 
 	attr->sample_type	|= PERF_SAMPLE_IP | PERF_SAMPLE_TID;
 
-	if (nr_counters > 1)
+	if (evlist->nr_entries > 1)
 		attr->sample_type |= PERF_SAMPLE_ID;
 
 	/*
@@ -315,19 +221,58 @@ static void create_counter(struct perf_evsel *evsel, int cpu)
 
 	attr->mmap		= track;
 	attr->comm		= track;
-	attr->inherit		= !no_inherit;
+
 	if (target_pid == -1 && target_tid == -1 && !system_wide) {
 		attr->disabled = 1;
 		attr->enable_on_exec = 1;
 	}
-retry_sample_id:
-	attr->sample_id_all = sample_id_all_avail ? 1 : 0;
+}
 
-	for (thread_index = 0; thread_index < threads->nr; thread_index++) {
-try_again:
-		FD(evsel, nr_cpu, thread_index) = sys_perf_event_open(attr, threads->map[thread_index], cpu, group_fd, 0);
+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);
 
-		if (FD(evsel, nr_cpu, thread_index) < 0) {
+	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;
+
+	list_for_each_entry(pos, &evlist->entries, node) {
+		struct perf_event_attr *attr = &pos->attr;
+		/*
+		 * Check if parse_single_tracepoint_event has already asked for
+		 * PERF_SAMPLE_TIME.
+		 *
+		 * XXX this is kludgy but short term fix for problems introduced by
+		 * eac23d1c that broke 'perf script' by having different sample_types
+		 * when using multiple tracepoint events when we use a perf binary
+		 * that tries to use sample_id_all on an older kernel.
+		 *
+		 * We need to move counter creation to perf_session, support
+		 * different sample_types, etc.
+		 */
+		bool time_needed = attr->sample_type & PERF_SAMPLE_TIME;
+
+		config_attr(pos, evlist);
+retry_sample_id:
+		attr->sample_id_all = sample_id_all_avail ? 1 : 0;
+try_again:
+		if (perf_evsel__open(pos, evlist->cpus, evlist->threads, group,
+				     !no_inherit) < 0) {
 			int err = errno;
 
 			if (err == EPERM || err == EACCES)
@@ -364,7 +309,7 @@ try_again:
 			}
 			printf("\n");
 			error("sys_perf_event_open() syscall returned with %d (%s).  /bin/dmesg may provide additional information.\n",
-			      FD(evsel, nr_cpu, thread_index), strerror(err));
+			      err, strerror(err));
 
 #if defined(__i386__) || defined(__x86_64__)
 			if (attr->type == PERF_TYPE_HARDWARE && err == EOPNOTSUPP)
@@ -375,90 +320,28 @@ try_again:
 #endif
 
 			die("No CONFIG_PERF_EVENTS=y kernel support configured?\n");
-			exit(-1);
 		}
+	}
 
-		h_attr = get_header_attr(attr, evsel->idx);
-		if (h_attr == NULL)
-			die("nomem\n");
+	if (perf_evlist__set_filters(evlist)) {
+		error("failed to set filter with %d (%s)\n", errno,
+			strerror(errno));
+		exit(-1);
+	}
 
-		if (!file_new) {
-			if (memcmp(&h_attr->attr, attr, sizeof(*attr))) {
-				fprintf(stderr, "incompatible append\n");
-				exit(-1);
-			}
-		}
+	if (perf_evlist__mmap(evlist, mmap_pages, false) < 0)
+		die("failed to mmap with %d (%s)\n", errno, strerror(errno));
 
-		if (read(FD(evsel, nr_cpu, thread_index), &read_data, sizeof(read_data)) == -1) {
-			perror("Unable to read perf file descriptor");
+	if (file_new)
+		session->evlist = evlist;
+	else {
+		if (!perf_evlist__equal(session->evlist, evlist)) {
+			fprintf(stderr, "incompatible append\n");
 			exit(-1);
 		}
+ 	}
 
-		if (perf_header_attr__add_id(h_attr, read_data.id) < 0) {
-			pr_warning("Not enough memory to add id\n");
-			exit(-1);
-		}
-
-		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(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].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(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(evsel, nr_cpu, thread_index);
-			event_array[nr_poll].events = POLLIN;
-			nr_poll++;
-		}
-
-		if (filter != NULL) {
-			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));
-				exit(-1);
-			}
-		}
-	}
-
-	if (!sample_type)
-		sample_type = attr->sample_type;
-}
-
-static void open_counters(int cpu)
-{
-	struct perf_evsel *pos;
-
-	group_fd = -1;
-
-	list_for_each_entry(pos, &evsel_list, node)
-		create_counter(pos, cpu);
-
-	nr_cpu++;
+	perf_session__update_sample_type(session);
 }
 
 static int process_buildids(void)
@@ -481,14 +364,14 @@ static void atexit_header(void)
 
 		if (!no_buildid)
 			process_buildids();
-		perf_header__write(&session->header, output, true);
+		perf_session__write_header(session, evsel_list, output, true);
 		perf_session__delete(session);
-		perf_evsel_list__delete();
+		perf_evlist__delete(evsel_list);
 		symbol__exit();
 	}
 }
 
-static void event__synthesize_guest_os(struct machine *machine, void *data)
+static void perf_event__synthesize_guest_os(struct machine *machine, void *data)
 {
 	int err;
 	struct perf_session *psession = data;
@@ -504,8 +387,8 @@ static void event__synthesize_guest_os(struct machine *machine, void *data)
 	 *method is used to avoid symbol missing when the first addr is
 	 *in module instead of in guest kernel.
 	 */
-	err = event__synthesize_modules(process_synthesized_event,
-					psession, machine);
+	err = perf_event__synthesize_modules(process_synthesized_event,
+					     psession, machine);
 	if (err < 0)
 		pr_err("Couldn't record guest kernel [%d]'s reference"
 		       " relocation symbol.\n", machine->pid);
@@ -514,11 +397,12 @@ static void event__synthesize_guest_os(struct machine *machine, void *data)
 	 * We use _stext for guest kernel because guest kernel's /proc/kallsyms
 	 * have no _text sometimes.
 	 */
-	err = event__synthesize_kernel_mmap(process_synthesized_event,
-					    psession, machine, "_text");
+	err = perf_event__synthesize_kernel_mmap(process_synthesized_event,
+						 psession, machine, "_text");
 	if (err < 0)
-		err = event__synthesize_kernel_mmap(process_synthesized_event,
-						    psession, machine, "_stext");
+		err = perf_event__synthesize_kernel_mmap(process_synthesized_event,
+							 psession, machine,
+							 "_stext");
 	if (err < 0)
 		pr_err("Couldn't record guest kernel [%d]'s reference"
 		       " relocation symbol.\n", machine->pid);
@@ -533,9 +417,9 @@ static void mmap_read_all(void)
 {
 	int i;
 
-	for (i = 0; i < nr_cpu; i++) {
-		if (mmap_array[i].base)
-			mmap_read(&mmap_array[i]);
+	for (i = 0; i < evsel_list->cpus->nr; i++) {
+		if (evsel_list->mmap[i].base)
+			mmap_read(&evsel_list->mmap[i]);
 	}
 
 	if (perf_header__has_feat(&session->header, HEADER_TRACE_INFO))
@@ -566,18 +450,26 @@ static int __cmd_record(int argc, const char **argv)
 		exit(-1);
 	}
 
-	if (!strcmp(output_name, "-"))
-		pipe_output = 1;
-	else if (!stat(output_name, &st) && st.st_size) {
-		if (write_mode == WRITE_FORCE) {
-			char oldname[PATH_MAX];
-			snprintf(oldname, sizeof(oldname), "%s.old",
-				 output_name);
-			unlink(oldname);
-			rename(output_name, oldname);
+	if (!output_name) {
+		if (!fstat(STDOUT_FILENO, &st) && S_ISFIFO(st.st_mode))
+			pipe_output = 1;
+		else
+			output_name = "perf.data";
+	}
+	if (output_name) {
+		if (!strcmp(output_name, "-"))
+			pipe_output = 1;
+		else if (!stat(output_name, &st) && st.st_size) {
+			if (write_mode == WRITE_FORCE) {
+				char oldname[PATH_MAX];
+				snprintf(oldname, sizeof(oldname), "%s.old",
+					 output_name);
+				unlink(oldname);
+				rename(output_name, oldname);
+			}
+		} else if (write_mode == WRITE_APPEND) {
+			write_mode = WRITE_FORCE;
 		}
-	} else if (write_mode == WRITE_APPEND) {
-		write_mode = WRITE_FORCE;
 	}
 
 	flags = O_CREAT|O_RDWR;
@@ -606,19 +498,14 @@ 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;
 	}
 
-	if (have_tracepoints(&evsel_list))
+	if (have_tracepoints(&evsel_list->entries))
 		perf_header__set_feat(&session->header, HEADER_TRACE_INFO);
 
-	/*
- 	 * perf_session__delete(session) will be called at atexit_header()
-	 */
-	atexit(atexit_header);
-
 	if (forks) {
 		child_pid = fork();
 		if (child_pid < 0) {
@@ -659,7 +546,7 @@ static int __cmd_record(int argc, const char **argv)
 		}
 
 		if (!system_wide && target_tid == -1 && target_pid == -1)
-			threads->map[0] = child_pid;
+			evsel_list->threads->map[0] = child_pid;
 
 		close(child_ready_pipe[1]);
 		close(go_pipe[0]);
@@ -673,46 +560,42 @@ static int __cmd_record(int argc, const char **argv)
 		close(child_ready_pipe[0]);
 	}
 
-	if (!system_wide && no_inherit && !cpu_list) {
-		open_counters(-1);
-	} else {
-		for (i = 0; i < cpus->nr; i++)
-			open_counters(cpus->map[i]);
-	}
+	open_counters(evsel_list);
 
-	perf_session__set_sample_type(session, sample_type);
+	/*
+	 * perf_session__delete(session) will be called at atexit_header()
+	 */
+	atexit(atexit_header);
 
 	if (pipe_output) {
 		err = perf_header__write_pipe(output);
 		if (err < 0)
 			return err;
 	} else if (file_new) {
-		err = perf_header__write(&session->header, 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 = 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;
 		}
 
-		err = event__synthesize_event_types(process_synthesized_event,
-						    session);
+		err = perf_event__synthesize_event_types(process_synthesized_event,
+							 session);
 		if (err < 0) {
 			pr_err("Couldn't synthesize event_types.\n");
 			return err;
 		}
 
-		if (have_tracepoints(&evsel_list)) {
+		if (have_tracepoints(&evsel_list->entries)) {
 			/*
 			 * FIXME err <= 0 here actually means that
 			 * there were no tracepoints so its not really
@@ -721,9 +604,9 @@ 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, &evsel_list,
-							     process_synthesized_event,
-							     session);
+			err = perf_event__synthesize_tracing_data(output, evsel_list,
+								  process_synthesized_event,
+								  session);
 			if (err <= 0) {
 				pr_err("Couldn't record tracing data.\n");
 				return err;
@@ -738,31 +621,34 @@ static int __cmd_record(int argc, const char **argv)
 		return -1;
 	}
 
-	err = event__synthesize_kernel_mmap(process_synthesized_event,
-					    session, machine, "_text");
+	err = perf_event__synthesize_kernel_mmap(process_synthesized_event,
+						 session, machine, "_text");
 	if (err < 0)
-		err = event__synthesize_kernel_mmap(process_synthesized_event,
-						    session, machine, "_stext");
+		err = perf_event__synthesize_kernel_mmap(process_synthesized_event,
+							 session, machine, "_stext");
 	if (err < 0)
 		pr_err("Couldn't record kernel reference relocation symbol\n"
 		       "Symbol resolution may be skewed if relocation was used (e.g. kexec).\n"
 		       "Check /proc/kallsyms permission or run as root.\n");
 
-	err = event__synthesize_modules(process_synthesized_event,
-					session, machine);
+	err = perf_event__synthesize_modules(process_synthesized_event,
+					     session, machine);
 	if (err < 0)
 		pr_err("Couldn't record kernel module information.\n"
 		       "Symbol resolution may be skewed if relocation was used (e.g. kexec).\n"
 		       "Check /proc/modules permission or run as root.\n");
 
 	if (perf_guest)
-		perf_session__process_machines(session, event__synthesize_guest_os);
+		perf_session__process_machines(session,
+					       perf_event__synthesize_guest_os);
 
 	if (!system_wide)
-		event__synthesize_thread_map(threads, process_synthesized_event,
-					     session);
+		perf_event__synthesize_thread_map(evsel_list->threads,
+						  process_synthesized_event,
+						  session);
 	else
-		event__synthesize_threads(process_synthesized_event, session);
+		perf_event__synthesize_threads(process_synthesized_event,
+					       session);
 
 	if (realtime_prio) {
 		struct sched_param param;
@@ -789,17 +675,17 @@ static int __cmd_record(int argc, const char **argv)
 		if (hits == samples) {
 			if (done)
 				break;
-			err = poll(event_array, nr_poll, -1);
+			err = poll(evsel_list->pollfd, evsel_list->nr_fds, -1);
 			waking++;
 		}
 
 		if (done) {
-			for (i = 0; i < nr_cpu; i++) {
+			for (i = 0; i < evsel_list->cpus->nr; i++) {
 				struct perf_evsel *pos;
 
-				list_for_each_entry(pos, &evsel_list, node) {
+				list_for_each_entry(pos, &evsel_list->entries, node) {
 					for (thread = 0;
-						thread < threads->nr;
+						thread < evsel_list->threads->nr;
 						thread++)
 						ioctl(FD(pos, i, thread),
 							PERF_EVENT_IOC_DISABLE);
@@ -838,10 +724,10 @@ static const char * const record_usage[] = {
 static bool force, append_file;
 
 const struct option record_options[] = {
-	OPT_CALLBACK('e', "event", NULL, "event",
+	OPT_CALLBACK('e', "event", &evsel_list, "event",
 		     "event selector. use 'perf list' to list available events",
 		     parse_events),
-	OPT_CALLBACK(0, "filter", NULL, "filter",
+	OPT_CALLBACK(0, "filter", &evsel_list, "filter",
 		     "event filter", parse_filter),
 	OPT_INTEGER('p', "pid", &target_pid,
 		    "record events on existing process id"),
@@ -884,6 +770,9 @@ const struct option record_options[] = {
 		    "do not update the buildid cache"),
 	OPT_BOOLEAN('B', "no-buildid", &no_buildid,
 		    "do not collect buildids in perf.data"),
+	OPT_CALLBACK('G', "cgroup", &evsel_list, "name",
+		     "monitor event in cgroup name only",
+		     parse_cgroups),
 	OPT_END()
 };
 
@@ -892,6 +781,10 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
 	int err = -ENOMEM;
 	struct perf_evsel *pos;
 
+	evsel_list = perf_evlist__new(NULL, NULL);
+	if (evsel_list == NULL)
+		return -ENOMEM;
+
 	argc = parse_options(argc, argv, record_options, record_usage,
 			    PARSE_OPT_STOP_AT_NON_OPTION);
 	if (!argc && target_pid == -1 && target_tid == -1 &&
@@ -908,12 +801,19 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
 		write_mode = WRITE_FORCE;
 	}
 
+	if (nr_cgroups && !system_wide) {
+		fprintf(stderr, "cgroup monitoring only available in"
+			" system-wide mode\n");
+		usage_with_options(record_usage, record_options);
+	}
+
 	symbol__init();
 
 	if (no_buildid_cache || no_buildid)
 		disable_buildid_cache();
 
-	if (list_empty(&evsel_list) && perf_evsel_list__create_default() < 0) {
+	if (evsel_list->nr_entries == 0 &&
+	    perf_evlist__add_default(evsel_list) < 0) {
 		pr_err("Not enough memory for event selector list\n");
 		goto out_symbol_exit;
 	}
@@ -921,27 +821,19 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
 	if (target_pid != -1)
 		target_tid = target_pid;
 
-	threads = thread_map__new(target_pid, target_tid);
-	if (threads == NULL) {
-		pr_err("Problems finding threads of monitor\n");
+	if (perf_evlist__create_maps(evsel_list, target_pid,
+				     target_tid, cpu_list) < 0)
 		usage_with_options(record_usage, record_options);
-	}
 
-	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)
+	list_for_each_entry(pos, &evsel_list->entries, node) {
+		if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr,
+					 evsel_list->threads->nr) < 0)
 			goto out_free_fd;
 		if (perf_header__push_event(pos->attr.config, event_name(pos)))
 			goto out_free_fd;
 	}
-	event_array = malloc((sizeof(struct pollfd) * MAX_NR_CPUS *
-			      MAX_COUNTERS * threads->nr));
-	if (!event_array)
+
+	if (perf_evlist__alloc_pollfd(evsel_list) < 0)
 		goto out_free_fd;
 
 	if (user_interval != ULLONG_MAX)
@@ -959,16 +851,12 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
 	} else {
 		fprintf(stderr, "frequency and count are zero, aborting\n");
 		err = -EINVAL;
-		goto out_free_event_array;
+		goto out_free_fd;
 	}
 
 	err = __cmd_record(argc, argv);
-
-out_free_event_array:
-	free(event_array);
 out_free_fd:
-	thread_map__delete(threads);
-	threads = NULL;
+	perf_evlist__delete_maps(evsel_list);
 out_symbol_exit:
 	symbol__exit();
 	return err;