Merge branch 'master' of /home/davem/src/GIT/linux-2.6/

32 files changed, 1956 insertions, 618 deletions

diff --git a/tools/perf/util/data_map.c b/tools/perf/util/data_map.c
index ca0bedf637c2..b557b836de3d 100644
--- a/tools/perf/util/data_map.c
+++ b/tools/perf/util/data_map.c
@@ -1,20 +1,17 @@
-#include "data_map.h"
 #include "symbol.h"
 #include "util.h"
 #include "debug.h"
+#include "thread.h"
+#include "session.h"
 
-
-static struct perf_file_handler *curr_handler;
-static unsigned long	mmap_window = 32;
-static char		__cwd[PATH_MAX];
-
-static int process_event_stub(event_t *event __used)
+static int process_event_stub(event_t *event __used,
+			      struct perf_session *session __used)
 {
 	dump_printf(": unhandled!\n");
 	return 0;
 }
 
-void register_perf_file_handler(struct perf_file_handler *handler)
+static void perf_event_ops__fill_defaults(struct perf_event_ops *handler)
 {
 	if (!handler->process_sample_event)
 		handler->process_sample_event = process_event_stub;
@@ -34,8 +31,6 @@ void register_perf_file_handler(struct perf_file_handler *handler)
 		handler->process_throttle_event = process_event_stub;
 	if (!handler->process_unthrottle_event)
 		handler->process_unthrottle_event = process_event_stub;
-
-	curr_handler = handler;
 }
 
 static const char *event__name[] = {
@@ -61,8 +56,9 @@ void event__print_totals(void)
 			event__name[i], event__total[i]);
 }
 
-static int
-process_event(event_t *event, unsigned long offset, unsigned long head)
+static int process_event(event_t *event, struct perf_session *session,
+			 struct perf_event_ops *ops,
+			 unsigned long offset, unsigned long head)
 {
 	trace_event(event);
 
@@ -77,34 +73,34 @@ process_event(event_t *event, unsigned long offset, unsigned long head)
 
 	switch (event->header.type) {
 	case PERF_RECORD_SAMPLE:
-		return curr_handler->process_sample_event(event);
+		return ops->process_sample_event(event, session);
 	case PERF_RECORD_MMAP:
-		return curr_handler->process_mmap_event(event);
+		return ops->process_mmap_event(event, session);
 	case PERF_RECORD_COMM:
-		return curr_handler->process_comm_event(event);
+		return ops->process_comm_event(event, session);
 	case PERF_RECORD_FORK:
-		return curr_handler->process_fork_event(event);
+		return ops->process_fork_event(event, session);
 	case PERF_RECORD_EXIT:
-		return curr_handler->process_exit_event(event);
+		return ops->process_exit_event(event, session);
 	case PERF_RECORD_LOST:
-		return curr_handler->process_lost_event(event);
+		return ops->process_lost_event(event, session);
 	case PERF_RECORD_READ:
-		return curr_handler->process_read_event(event);
+		return ops->process_read_event(event, session);
 	case PERF_RECORD_THROTTLE:
-		return curr_handler->process_throttle_event(event);
+		return ops->process_throttle_event(event, session);
 	case PERF_RECORD_UNTHROTTLE:
-		return curr_handler->process_unthrottle_event(event);
+		return ops->process_unthrottle_event(event, session);
 	default:
-		curr_handler->total_unknown++;
+		ops->total_unknown++;
 		return -1;
 	}
 }
 
-int perf_header__read_build_ids(int input, off_t offset, off_t size)
+int perf_header__read_build_ids(int input, u64 offset, u64 size)
 {
 	struct build_id_event bev;
 	char filename[PATH_MAX];
-	off_t limit = offset + size;
+	u64 limit = offset + size;
 	int err = -1;
 
 	while (offset < limit) {
@@ -129,88 +125,58 @@ out:
 	return err;
 }
 
-int mmap_dispatch_perf_file(struct perf_header **pheader,
-			    const char *input_name,
-			    int force,
-			    int full_paths,
-			    int *cwdlen,
-			    char **cwd)
+static struct thread *perf_session__register_idle_thread(struct perf_session *self)
+{
+	struct thread *thread = perf_session__findnew(self, 0);
+
+	if (!thread || thread__set_comm(thread, "swapper")) {
+		pr_err("problem inserting idle task.\n");
+		thread = NULL;
+	}
+
+	return thread;
+}
+
+int perf_session__process_events(struct perf_session *self,
+				 struct perf_event_ops *ops)
 {
 	int err;
-	struct perf_header *header;
 	unsigned long head, shift;
 	unsigned long offset = 0;
-	struct stat input_stat;
 	size_t	page_size;
-	u64 sample_type;
 	event_t *event;
 	uint32_t size;
-	int input;
 	char *buf;
 
-	if (curr_handler == NULL) {
-		pr_debug("Forgot to register perf file handler\n");
-		return -EINVAL;
-	}
-
-	page_size = getpagesize();
-
-	input = open(input_name, O_RDONLY);
-	if (input < 0) {
-		pr_err("Failed to open file: %s", input_name);
-		if (!strcmp(input_name, "perf.data"))
-			pr_err("  (try 'perf record' first)");
-		pr_err("\n");
-		return -errno;
-	}
-
-	if (fstat(input, &input_stat) < 0) {
-		pr_err("failed to stat file");
-		err = -errno;
-		goto out_close;
-	}
-
-	err = -EACCES;
-	if (!force && input_stat.st_uid && (input_stat.st_uid != geteuid())) {
-		pr_err("file: %s not owned by current user or root\n",
-			input_name);
-		goto out_close;
-	}
-
-	if (input_stat.st_size == 0) {
-		pr_info("zero-sized file, nothing to do!\n");
-		goto done;
-	}
+	if (perf_session__register_idle_thread(self) == NULL)
+		return -ENOMEM;
 
-	err = -ENOMEM;
-	header = perf_header__new();
-	if (header == NULL)
-		goto out_close;
+	perf_event_ops__fill_defaults(ops);
 
-	err = perf_header__read(header, input);
-	if (err < 0)
-		goto out_delete;
-	*pheader = header;
-	head = header->data_offset;
+	page_size = getpagesize();
 
-	sample_type = perf_header__sample_type(header);
+	head = self->header.data_offset;
+	self->sample_type = perf_header__sample_type(&self->header);
 
 	err = -EINVAL;
-	if (curr_handler->sample_type_check &&
-	    curr_handler->sample_type_check(sample_type) < 0)
-		goto out_delete;
+	if (ops->sample_type_check && ops->sample_type_check(self) < 0)
+		goto out_err;
 
-	if (!full_paths) {
-		if (getcwd(__cwd, sizeof(__cwd)) == NULL) {
-			pr_err("failed to get the current directory\n");
+	if (!ops->full_paths) {
+		char bf[PATH_MAX];
+
+		if (getcwd(bf, sizeof(bf)) == NULL) {
 			err = -errno;
-			goto out_delete;
+out_getcwd_err:
+			pr_err("failed to get the current directory\n");
+			goto out_err;
+		}
+		self->cwd = strdup(bf);
+		if (self->cwd == NULL) {
+			err = -ENOMEM;
+			goto out_getcwd_err;
 		}
-		*cwd = __cwd;
-		*cwdlen = strlen(*cwd);
-	} else {
-		*cwd = NULL;
-		*cwdlen = 0;
+		self->cwdlen = strlen(self->cwd);
 	}
 
 	shift = page_size * (head / page_size);
@@ -218,12 +184,12 @@ int mmap_dispatch_perf_file(struct perf_header **pheader,
 	head -= shift;
 
 remap:
-	buf = mmap(NULL, page_size * mmap_window, PROT_READ,
-		   MAP_SHARED, input, offset);
+	buf = mmap(NULL, page_size * self->mmap_window, PROT_READ,
+		   MAP_SHARED, self->fd, offset);
 	if (buf == MAP_FAILED) {
 		pr_err("failed to mmap file\n");
 		err = -errno;
-		goto out_delete;
+		goto out_err;
 	}
 
 more:
@@ -233,12 +199,12 @@ more:
 	if (!size)
 		size = 8;
 
-	if (head + event->header.size >= page_size * mmap_window) {
+	if (head + event->header.size >= page_size * self->mmap_window) {
 		int munmap_ret;
 
 		shift = page_size * (head / page_size);
 
-		munmap_ret = munmap(buf, page_size * mmap_window);
+		munmap_ret = munmap(buf, page_size * self->mmap_window);
 		assert(munmap_ret == 0);
 
 		offset += shift;
@@ -253,7 +219,7 @@ more:
 			(void *)(long)event->header.size,
 			event->header.type);
 
-	if (!size || process_event(event, offset, head) < 0) {
+	if (!size || process_event(event, self, ops, offset, head) < 0) {
 
 		dump_printf("%p [%p]: skipping unknown header type: %d\n",
 			(void *)(offset + head),
@@ -273,19 +239,14 @@ more:
 
 	head += size;
 
-	if (offset + head >= header->data_offset + header->data_size)
+	if (offset + head >= self->header.data_offset + self->header.data_size)
 		goto done;
 
-	if (offset + head < (unsigned long)input_stat.st_size)
+	if (offset + head < self->size)
 		goto more;
 
 done:
 	err = 0;
-out_close:
-	close(input);
-
+out_err:
 	return err;
-out_delete:
-	perf_header__delete(header);
-	goto out_close;
 }
diff --git a/tools/perf/util/data_map.h b/tools/perf/util/data_map.h
deleted file mode 100644
index 3180ff7e3633..000000000000
--- a/tools/perf/util/data_map.h
+++ /dev/null
@@ -1,32 +0,0 @@
-#ifndef __PERF_DATAMAP_H
-#define __PERF_DATAMAP_H
-
-#include "event.h"
-#include "header.h"
-
-typedef int (*event_type_handler_t)(event_t *);
-
-struct perf_file_handler {
-	event_type_handler_t	process_sample_event;
-	event_type_handler_t	process_mmap_event;
-	event_type_handler_t	process_comm_event;
-	event_type_handler_t	process_fork_event;
-	event_type_handler_t	process_exit_event;
-	event_type_handler_t	process_lost_event;
-	event_type_handler_t	process_read_event;
-	event_type_handler_t	process_throttle_event;
-	event_type_handler_t	process_unthrottle_event;
-	int			(*sample_type_check)(u64 sample_type);
-	unsigned long		total_unknown;
-};
-
-void register_perf_file_handler(struct perf_file_handler *handler);
-int mmap_dispatch_perf_file(struct perf_header **pheader,
-			    const char *input_name,
-			    int force,
-			    int full_paths,
-			    int *cwdlen,
-			    char **cwd);
-int perf_header__read_build_ids(int input, off_t offset, off_t file_size);
-
-#endif
diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index 414b89d1bde9..bb0fd6da2d56 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -1,11 +1,16 @@
 #include <linux/types.h>
 #include "event.h"
 #include "debug.h"
+#include "session.h"
+#include "sort.h"
 #include "string.h"
+#include "strlist.h"
 #include "thread.h"
 
 static pid_t event__synthesize_comm(pid_t pid, int full,
-				    int (*process)(event_t *event))
+				    int (*process)(event_t *event,
+						   struct perf_session *session),
+				    struct perf_session *session)
 {
 	event_t ev;
 	char filename[PATH_MAX];
@@ -54,7 +59,7 @@ out_race:
 	if (!full) {
 		ev.comm.tid = pid;
 
-		process(&ev);
+		process(&ev, session);
 		goto out_fclose;
 	}
 
@@ -72,7 +77,7 @@ out_race:
 
 		ev.comm.tid = pid;
 
-		process(&ev);
+		process(&ev, session);
 	}
 	closedir(tasks);
 
@@ -86,7 +91,9 @@ out_failure:
 }
 
 static int event__synthesize_mmap_events(pid_t pid, pid_t tgid,
-					 int (*process)(event_t *event))
+					 int (*process)(event_t *event,
+							struct perf_session *session),
+					 struct perf_session *session)
 {
 	char filename[PATH_MAX];
 	FILE *fp;
@@ -141,7 +148,7 @@ static int event__synthesize_mmap_events(pid_t pid, pid_t tgid,
 			ev.mmap.pid = tgid;
 			ev.mmap.tid = pid;
 
-			process(&ev);
+			process(&ev, session);
 		}
 	}
 
@@ -149,15 +156,20 @@ static int event__synthesize_mmap_events(pid_t pid, pid_t tgid,
 	return 0;
 }
 
-int event__synthesize_thread(pid_t pid, int (*process)(event_t *event))
+int event__synthesize_thread(pid_t pid,
+			     int (*process)(event_t *event,
+					    struct perf_session *session),
+			     struct perf_session *session)
 {
-	pid_t tgid = event__synthesize_comm(pid, 1, process);
+	pid_t tgid = event__synthesize_comm(pid, 1, process, session);
 	if (tgid == -1)
 		return -1;
-	return event__synthesize_mmap_events(pid, tgid, process);
+	return event__synthesize_mmap_events(pid, tgid, process, session);
 }
 
-void event__synthesize_threads(int (*process)(event_t *event))
+void event__synthesize_threads(int (*process)(event_t *event,
+					      struct perf_session *session),
+			       struct perf_session *session)
 {
 	DIR *proc;
 	struct dirent dirent, *next;
@@ -171,24 +183,47 @@ void event__synthesize_threads(int (*process)(event_t *event))
 		if (*end) /* only interested in proper numerical dirents */
 			continue;
 
-		event__synthesize_thread(pid, process);
+		event__synthesize_thread(pid, process, session);
 	}
 
 	closedir(proc);
 }
 
-char *event__cwd;
-int  event__cwdlen;
+static void thread__comm_adjust(struct thread *self)
+{
+	char *comm = self->comm;
+
+	if (!symbol_conf.col_width_list_str && !symbol_conf.field_sep &&
+	    (!symbol_conf.comm_list ||
+	     strlist__has_entry(symbol_conf.comm_list, comm))) {
+		unsigned int slen = strlen(comm);
+
+		if (slen > comms__col_width) {
+			comms__col_width = slen;
+			threads__col_width = slen + 6;
+		}
+	}
+}
+
+static int thread__set_comm_adjust(struct thread *self, const char *comm)
+{
+	int ret = thread__set_comm(self, comm);
+
+	if (ret)
+		return ret;
 
-struct events_stats event__stats;
+	thread__comm_adjust(self);
 
-int event__process_comm(event_t *self)
+	return 0;
+}
+
+int event__process_comm(event_t *self, struct perf_session *session)
 {
-	struct thread *thread = threads__findnew(self->comm.pid);
+	struct thread *thread = perf_session__findnew(session, self->comm.pid);
 
 	dump_printf(": %s:%d\n", self->comm.comm, self->comm.pid);
 
-	if (thread == NULL || thread__set_comm(thread, self->comm.comm)) {
+	if (thread == NULL || thread__set_comm_adjust(thread, self->comm.comm)) {
 		dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n");
 		return -1;
 	}
@@ -196,18 +231,18 @@ int event__process_comm(event_t *self)
 	return 0;
 }
 
-int event__process_lost(event_t *self)
+int event__process_lost(event_t *self, struct perf_session *session)
 {
 	dump_printf(": id:%Ld: lost:%Ld\n", self->lost.id, self->lost.lost);
-	event__stats.lost += self->lost.lost;
+	session->events_stats.lost += self->lost.lost;
 	return 0;
 }
 
-int event__process_mmap(event_t *self)
+int event__process_mmap(event_t *self, struct perf_session *session)
 {
-	struct thread *thread = threads__findnew(self->mmap.pid);
+	struct thread *thread = perf_session__findnew(session, self->mmap.pid);
 	struct map *map = map__new(&self->mmap, MAP__FUNCTION,
-				   event__cwd, event__cwdlen);
+				   session->cwd, session->cwdlen);
 
 	dump_printf(" %d/%d: [%p(%p) @ %p]: %s\n",
 		    self->mmap.pid, self->mmap.tid,
@@ -224,10 +259,10 @@ int event__process_mmap(event_t *self)
 	return 0;
 }
 
-int event__process_task(event_t *self)
+int event__process_task(event_t *self, struct perf_session *session)
 {
-	struct thread *thread = threads__findnew(self->fork.pid);
-	struct thread *parent = threads__findnew(self->fork.ppid);
+	struct thread *thread = perf_session__findnew(session, self->fork.pid);
+	struct thread *parent = perf_session__findnew(session, self->fork.ppid);
 
 	dump_printf("(%d:%d):(%d:%d)\n", self->fork.pid, self->fork.tid,
 		    self->fork.ppid, self->fork.ptid);
@@ -249,18 +284,20 @@ int event__process_task(event_t *self)
 	return 0;
 }
 
-void thread__find_addr_location(struct thread *self, u8 cpumode,
+void thread__find_addr_location(struct thread *self,
+				struct perf_session *session, u8 cpumode,
 				enum map_type type, u64 addr,
 				struct addr_location *al,
 				symbol_filter_t filter)
 {
-	struct thread *thread = al->thread = self;
+	struct map_groups *mg = &self->mg;
 
+	al->thread = self;
 	al->addr = addr;
 
 	if (cpumode & PERF_RECORD_MISC_KERNEL) {
 		al->level = 'k';
-		thread = kthread;
+		mg = &session->kmaps;
 	} else if (cpumode & PERF_RECORD_MISC_USER)
 		al->level = '.';
 	else {
@@ -270,7 +307,7 @@ void thread__find_addr_location(struct thread *self, u8 cpumode,
 		return;
 	}
 try_again:
-	al->map = thread__find_map(thread, type, al->addr);
+	al->map = map_groups__find(mg, type, al->addr);
 	if (al->map == NULL) {
 		/*
 		 * If this is outside of all known maps, and is a negative
@@ -281,32 +318,139 @@ try_again:
 		 * "[vdso]" dso, but for now lets use the old trick of looking
 		 * in the whole kernel symbol list.
 		 */
-		if ((long long)al->addr < 0 && thread != kthread) {
-			thread = kthread;
+		if ((long long)al->addr < 0 && mg != &session->kmaps) {
+			mg = &session->kmaps;
 			goto try_again;
 		}
 		al->sym = NULL;
 	} else {
 		al->addr = al->map->map_ip(al->map, al->addr);
-		al->sym = map__find_symbol(al->map, al->addr, filter);
+		al->sym = map__find_symbol(al->map, session, al->addr, filter);
 	}
 }
 
-int event__preprocess_sample(const event_t *self, struct addr_location *al,
-			     symbol_filter_t filter)
+static void dso__calc_col_width(struct dso *self)
+{
+	if (!symbol_conf.col_width_list_str && !symbol_conf.field_sep &&
+	    (!symbol_conf.dso_list ||
+	     strlist__has_entry(symbol_conf.dso_list, self->name))) {
+		unsigned int slen = strlen(self->name);
+		if (slen > dsos__col_width)
+			dsos__col_width = slen;
+	}
+
+	self->slen_calculated = 1;
+}
+
+int event__preprocess_sample(const event_t *self, struct perf_session *session,
+			     struct addr_location *al, symbol_filter_t filter)
 {
 	u8 cpumode = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
-	struct thread *thread = threads__findnew(self->ip.pid);
+	struct thread *thread = perf_session__findnew(session, self->ip.pid);
 
 	if (thread == NULL)
 		return -1;
 
+	if (symbol_conf.comm_list &&
+	    !strlist__has_entry(symbol_conf.comm_list, thread->comm))
+		goto out_filtered;
+
 	dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
 
-	thread__find_addr_location(thread, cpumode, MAP__FUNCTION,
+	thread__find_addr_location(thread, session, cpumode, MAP__FUNCTION,
 				   self->ip.ip, al, filter);
 	dump_printf(" ...... dso: %s\n",
 		    al->map ? al->map->dso->long_name :
 			al->level == 'H' ? "[hypervisor]" : "<not found>");
+	/*
+	 * We have to do this here as we may have a dso with no symbol hit that
+	 * has a name longer than the ones with symbols sampled.
+	 */
+	if (al->map && !sort_dso.elide && !al->map->dso->slen_calculated)
+		dso__calc_col_width(al->map->dso);
+
+	if (symbol_conf.dso_list &&
+	    (!al->map || !al->map->dso ||
+	     !(strlist__has_entry(symbol_conf.dso_list, al->map->dso->short_name) ||
+	       (al->map->dso->short_name != al->map->dso->long_name &&
+		strlist__has_entry(symbol_conf.dso_list, al->map->dso->long_name)))))
+		goto out_filtered;
+
+	if (symbol_conf.sym_list && al->sym &&
+	    !strlist__has_entry(symbol_conf.sym_list, al->sym->name))
+		goto out_filtered;
+
+	al->filtered = false;
+	return 0;
+
+out_filtered:
+	al->filtered = true;
+	return 0;
+}
+
+int event__parse_sample(event_t *event, u64 type, struct sample_data *data)
+{
+	u64 *array = event->sample.array;
+
+	if (type & PERF_SAMPLE_IP) {
+		data->ip = event->ip.ip;
+		array++;
+	}
+
+	if (type & PERF_SAMPLE_TID) {
+		u32 *p = (u32 *)array;
+		data->pid = p[0];
+		data->tid = p[1];
+		array++;
+	}
+
+	if (type & PERF_SAMPLE_TIME) {
+		data->time = *array;
+		array++;
+	}
+
+	if (type & PERF_SAMPLE_ADDR) {
+		data->addr = *array;
+		array++;
+	}
+
+	if (type & PERF_SAMPLE_ID) {
+		data->id = *array;
+		array++;
+	}
+
+	if (type & PERF_SAMPLE_STREAM_ID) {
+		data->stream_id = *array;
+		array++;
+	}
+
+	if (type & PERF_SAMPLE_CPU) {
+		u32 *p = (u32 *)array;
+		data->cpu = *p;
+		array++;
+	}
+
+	if (type & PERF_SAMPLE_PERIOD) {
+		data->period = *array;
+		array++;
+	}
+
+	if (type & PERF_SAMPLE_READ) {
+		pr_debug("PERF_SAMPLE_READ is unsuported for now\n");
+		return -1;
+	}
+
+	if (type & PERF_SAMPLE_CALLCHAIN) {
+		data->callchain = (struct ip_callchain *)array;
+		array += 1 + data->callchain->nr;
+	}
+
+	if (type & PERF_SAMPLE_RAW) {
+		u32 *p = (u32 *)array;
+		data->raw_size = *p;
+		p++;
+		data->raw_data = p;
+	}
+
 	return 0;
 }
diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h
index a4cc8105cf67..690a96d0467c 100644
--- a/tools/perf/util/event.h
+++ b/tools/perf/util/event.h
@@ -56,11 +56,25 @@ struct read_event {
 	u64 id;
 };
 
-struct sample_event{
+struct sample_event {
 	struct perf_event_header        header;
 	u64 array[];
 };
 
+struct sample_data {
+	u64 ip;
+	u32 pid, tid;
+	u64 time;
+	u64 addr;
+	u64 id;
+	u64 stream_id;
+	u32 cpu;
+	u64 period;
+	struct ip_callchain *callchain;
+	u32 raw_size;
+	void *raw_data;
+};
+
 #define BUILD_ID_SIZE 20
 
 struct build_id_event {
@@ -81,18 +95,19 @@ typedef union event_union {
 } event_t;
 
 struct events_stats {
-	unsigned long total;
-	unsigned long lost;
+	u64 total;
+	u64 lost;
 };
 
 void event__print_totals(void);
 
 enum map_type {
 	MAP__FUNCTION = 0,
-
-	MAP__NR_TYPES,
+	MAP__VARIABLE,
 };
 
+#define MAP__NR_TYPES (MAP__VARIABLE + 1)
+
 struct map {
 	union {
 		struct rb_node	rb_node;
@@ -134,26 +149,35 @@ void map__delete(struct map *self);
 struct map *map__clone(struct map *self);
 int map__overlap(struct map *l, struct map *r);
 size_t map__fprintf(struct map *self, FILE *fp);
-struct symbol *map__find_symbol(struct map *self, u64 addr,
-				symbol_filter_t filter);
+
+struct perf_session;
+
+int map__load(struct map *self, struct perf_session *session,
+	      symbol_filter_t filter);
+struct symbol *map__find_symbol(struct map *self, struct perf_session *session,
+				u64 addr, symbol_filter_t filter);
+struct symbol *map__find_symbol_by_name(struct map *self, const char *name,
+					struct perf_session *session,
+					symbol_filter_t filter);
 void map__fixup_start(struct map *self);
 void map__fixup_end(struct map *self);
 
-int event__synthesize_thread(pid_t pid, int (*process)(event_t *event));
-void event__synthesize_threads(int (*process)(event_t *event));
-
-extern char *event__cwd;
-extern int  event__cwdlen;
-extern struct events_stats event__stats;
-extern unsigned long event__total[PERF_RECORD_MAX];
+int event__synthesize_thread(pid_t pid,
+			     int (*process)(event_t *event,
+					    struct perf_session *session),
+			     struct perf_session *session);
+void event__synthesize_threads(int (*process)(event_t *event,
+					      struct perf_session *session),
+			       struct perf_session *session);
 
-int event__process_comm(event_t *self);
-int event__process_lost(event_t *self);
-int event__process_mmap(event_t *self);
-int event__process_task(event_t *self);
+int event__process_comm(event_t *self, struct perf_session *session);
+int event__process_lost(event_t *self, struct perf_session *session);
+int event__process_mmap(event_t *self, struct perf_session *session);
+int event__process_task(event_t *self, struct perf_session *session);
 
 struct addr_location;
-int event__preprocess_sample(const event_t *self, struct addr_location *al,
-			     symbol_filter_t filter);
+int event__preprocess_sample(const event_t *self, struct perf_session *session,
+			     struct addr_location *al, symbol_filter_t filter);
+int event__parse_sample(event_t *event, u64 type, struct sample_data *data);
 
 #endif /* __PERF_RECORD_H */
diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index 4805e6dfd23c..8a0bca55106f 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -8,8 +8,8 @@
 #include "header.h"
 #include "../perf.h"
 #include "trace-event.h"
+#include "session.h"
 #include "symbol.h"
-#include "data_map.h"
 #include "debug.h"
 
 /*
@@ -58,35 +58,19 @@ int perf_header_attr__add_id(struct perf_header_attr *self, u64 id)
 	return 0;
 }
 
-/*
- * Create new perf.data header:
- */
-struct perf_header *perf_header__new(void)
+int perf_header__init(struct perf_header *self)
 {
-	struct perf_header *self = zalloc(sizeof(*self));
-
-	if (self != NULL) {
-		self->size = 1;
-		self->attr = malloc(sizeof(void *));
-
-		if (self->attr == NULL) {
-			free(self);
-			self = NULL;
-		}
-	}
-
-	return self;
+	self->size = 1;
+	self->attr = malloc(sizeof(void *));
+	return self->attr == NULL ? -ENOMEM : 0;
 }
 
-void perf_header__delete(struct perf_header *self)
+void perf_header__exit(struct perf_header *self)
 {
 	int i;
-
 	for (i = 0; i < self->attrs; ++i)
-		perf_header_attr__delete(self->attr[i]);
-
+                perf_header_attr__delete(self->attr[i]);
 	free(self->attr);
-	free(self);
 }
 
 int perf_header__add_attr(struct perf_header *self,
@@ -187,7 +171,9 @@ static int do_write(int fd, const void *buf, size_t size)
 
 static int __dsos__write_buildid_table(struct list_head *head, int fd)
 {
+#define NAME_ALIGN	64
 	struct dso *pos;
+	static const char zero_buf[NAME_ALIGN];
 
 	list_for_each_entry(pos, head, node) {
 		int err;
@@ -197,14 +183,17 @@ static int __dsos__write_buildid_table(struct list_head *head, int fd)
 		if (!pos->has_build_id)
 			continue;
 		len = pos->long_name_len + 1;
-		len = ALIGN(len, 64);
+		len = ALIGN(len, NAME_ALIGN);
 		memset(&b, 0, sizeof(b));
 		memcpy(&b.build_id, pos->build_id, sizeof(pos->build_id));
 		b.header.size = sizeof(b) + len;
 		err = do_write(fd, &b, sizeof(b));
 		if (err < 0)
 			return err;
-		err = do_write(fd, pos->long_name, len);
+		err = do_write(fd, pos->long_name, pos->long_name_len + 1);
+		if (err < 0)
+			return err;
+		err = do_write(fd, zero_buf, len - pos->long_name_len - 1);
 		if (err < 0)
 			return err;
 	}
diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h
index d1dbe2b79c42..d118d05d3abe 100644
--- a/tools/perf/util/header.h
+++ b/tools/perf/util/header.h
@@ -55,8 +55,8 @@ struct perf_header {
 	DECLARE_BITMAP(adds_features, HEADER_FEAT_BITS);
 };
 
-struct perf_header *perf_header__new(void);
-void perf_header__delete(struct perf_header *self);
+int perf_header__init(struct perf_header *self);
+void perf_header__exit(struct perf_header *self);
 
 int perf_header__read(struct perf_header *self, int fd);
 int perf_header__write(struct perf_header *self, int fd, bool at_exit);
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index 0ebf6ee16caa..e8daf5ca6fd2 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -1,9 +1,7 @@
 #include "hist.h"
-
-struct rb_root hist;
-struct rb_root collapse_hists;
-struct rb_root output_hists;
-int callchain;
+#include "session.h"
+#include "sort.h"
+#include <math.h>
 
 struct callchain_param	callchain_param = {
 	.mode	= CHAIN_GRAPH_REL,
@@ -14,11 +12,12 @@ struct callchain_param	callchain_param = {
  * histogram, sorted on item, collects counts
  */
 
-struct hist_entry *__hist_entry__add(struct addr_location *al,
-				     struct symbol *sym_parent,
-				     u64 count, bool *hit)
+struct hist_entry *__perf_session__add_hist_entry(struct perf_session *self,
+						  struct addr_location *al,
+						  struct symbol *sym_parent,
+						  u64 count, bool *hit)
 {
-	struct rb_node **p = &hist.rb_node;
+	struct rb_node **p = &self->hists.rb_node;
 	struct rb_node *parent = NULL;
 	struct hist_entry *he;
 	struct hist_entry entry = {
@@ -54,7 +53,7 @@ struct hist_entry *__hist_entry__add(struct addr_location *al,
 		return NULL;
 	*he = entry;
 	rb_link_node(&he->rb_node, parent, p);
-	rb_insert_color(&he->rb_node, &hist);
+	rb_insert_color(&he->rb_node, &self->hists);
 	*hit = false;
 	return he;
 }
@@ -102,9 +101,9 @@ void hist_entry__free(struct hist_entry *he)
  * collapse the histogram
  */
 
-void collapse__insert_entry(struct hist_entry *he)
+static void collapse__insert_entry(struct rb_root *root, struct hist_entry *he)
 {
-	struct rb_node **p = &collapse_hists.rb_node;
+	struct rb_node **p = &root->rb_node;
 	struct rb_node *parent = NULL;
 	struct hist_entry *iter;
 	int64_t cmp;
@@ -128,38 +127,45 @@ void collapse__insert_entry(struct hist_entry *he)
 	}
 
 	rb_link_node(&he->rb_node, parent, p);
-	rb_insert_color(&he->rb_node, &collapse_hists);
+	rb_insert_color(&he->rb_node, root);
 }
 
-void collapse__resort(void)
+void perf_session__collapse_resort(struct perf_session *self)
 {
+	struct rb_root tmp;
 	struct rb_node *next;
 	struct hist_entry *n;
 
 	if (!sort__need_collapse)
 		return;
 
-	next = rb_first(&hist);
+	tmp = RB_ROOT;
+	next = rb_first(&self->hists);
+
 	while (next) {
 		n = rb_entry(next, struct hist_entry, rb_node);
 		next = rb_next(&n->rb_node);
 
-		rb_erase(&n->rb_node, &hist);
-		collapse__insert_entry(n);
+		rb_erase(&n->rb_node, &self->hists);
+		collapse__insert_entry(&tmp, n);
 	}
+
+	self->hists = tmp;
 }
 
 /*
  * reverse the map, sort on count.
  */
 
-void output__insert_entry(struct hist_entry *he, u64 min_callchain_hits)
+static void perf_session__insert_output_hist_entry(struct rb_root *root,
+						   struct hist_entry *he,
+						   u64 min_callchain_hits)
 {
-	struct rb_node **p = &output_hists.rb_node;
+	struct rb_node **p = &root->rb_node;
 	struct rb_node *parent = NULL;
 	struct hist_entry *iter;
 
-	if (callchain)
+	if (symbol_conf.use_callchain)
 		callchain_param.sort(&he->sorted_chain, &he->callchain,
 				      min_callchain_hits, &callchain_param);
 
@@ -174,29 +180,483 @@ void output__insert_entry(struct hist_entry *he, u64 min_callchain_hits)
 	}
 
 	rb_link_node(&he->rb_node, parent, p);
-	rb_insert_color(&he->rb_node, &output_hists);
+	rb_insert_color(&he->rb_node, root);
 }
 
-void output__resort(u64 total_samples)
+void perf_session__output_resort(struct perf_session *self, u64 total_samples)
 {
+	struct rb_root tmp;
 	struct rb_node *next;
 	struct hist_entry *n;
-	struct rb_root *tree = &hist;
 	u64 min_callchain_hits;
 
 	min_callchain_hits =
 		total_samples * (callchain_param.min_percent / 100);
 
-	if (sort__need_collapse)
-		tree = &collapse_hists;
-
-	next = rb_first(tree);
+	tmp = RB_ROOT;
+	next = rb_first(&self->hists);
 
 	while (next) {
 		n = rb_entry(next, struct hist_entry, rb_node);
 		next = rb_next(&n->rb_node);
 
-		rb_erase(&n->rb_node, tree);
-		output__insert_entry(n, min_callchain_hits);
+		rb_erase(&n->rb_node, &self->hists);
+		perf_session__insert_output_hist_entry(&tmp, n,
+						       min_callchain_hits);
+	}
+
+	self->hists = tmp;
+}
+
+static size_t callchain__fprintf_left_margin(FILE *fp, int left_margin)
+{
+	int i;
+	int ret = fprintf(fp, "            ");
+
+	for (i = 0; i < left_margin; i++)
+		ret += fprintf(fp, " ");
+
+	return ret;
+}
+
+static size_t ipchain__fprintf_graph_line(FILE *fp, int depth, int depth_mask,
+					  int left_margin)
+{
+	int i;
+	size_t ret = callchain__fprintf_left_margin(fp, left_margin);
+
+	for (i = 0; i < depth; i++)
+		if (depth_mask & (1 << i))
+			ret += fprintf(fp, "|          ");
+		else
+			ret += fprintf(fp, "           ");
+
+	ret += fprintf(fp, "\n");
+
+	return ret;
+}
+
+static size_t ipchain__fprintf_graph(FILE *fp, struct callchain_list *chain,
+				     int depth, int depth_mask, int count,
+				     u64 total_samples, int hits,
+				     int left_margin)
+{
+	int i;
+	size_t ret = 0;
+
+	ret += callchain__fprintf_left_margin(fp, left_margin);
+	for (i = 0; i < depth; i++) {
+		if (depth_mask & (1 << i))
+			ret += fprintf(fp, "|");
+		else
+			ret += fprintf(fp, " ");
+		if (!count && i == depth - 1) {
+			double percent;
+
+			percent = hits * 100.0 / total_samples;
+			ret += percent_color_fprintf(fp, "--%2.2f%%-- ", percent);
+		} else
+			ret += fprintf(fp, "%s", "          ");
+	}
+	if (chain->sym)
+		ret += fprintf(fp, "%s\n", chain->sym->name);
+	else
+		ret += fprintf(fp, "%p\n", (void *)(long)chain->ip);
+
+	return ret;
+}
+
+static struct symbol *rem_sq_bracket;
+static struct callchain_list rem_hits;
+
+static void init_rem_hits(void)
+{
+	rem_sq_bracket = malloc(sizeof(*rem_sq_bracket) + 6);
+	if (!rem_sq_bracket) {
+		fprintf(stderr, "Not enough memory to display remaining hits\n");
+		return;
+	}
+
+	strcpy(rem_sq_bracket->name, "[...]");
+	rem_hits.sym = rem_sq_bracket;
+}
+
+static size_t __callchain__fprintf_graph(FILE *fp, struct callchain_node *self,
+					 u64 total_samples, int depth,
+					 int depth_mask, int left_margin)
+{
+	struct rb_node *node, *next;
+	struct callchain_node *child;
+	struct callchain_list *chain;
+	int new_depth_mask = depth_mask;
+	u64 new_total;
+	u64 remaining;
+	size_t ret = 0;
+	int i;
+
+	if (callchain_param.mode == CHAIN_GRAPH_REL)
+		new_total = self->children_hit;
+	else
+		new_total = total_samples;
+
+	remaining = new_total;
+
+	node = rb_first(&self->rb_root);
+	while (node) {
+		u64 cumul;
+
+		child = rb_entry(node, struct callchain_node, rb_node);
+		cumul = cumul_hits(child);
+		remaining -= cumul;
+
+		/*
+		 * The depth mask manages the output of pipes that show
+		 * the depth. We don't want to keep the pipes of the current
+		 * level for the last child of this depth.
+		 * Except if we have remaining filtered hits. They will
+		 * supersede the last child
+		 */
+		next = rb_next(node);
+		if (!next && (callchain_param.mode != CHAIN_GRAPH_REL || !remaining))
+			new_depth_mask &= ~(1 << (depth - 1));
+
+		/*
+		 * But we keep the older depth mask for the line seperator
+		 * to keep the level link until we reach the last child
+		 */
+		ret += ipchain__fprintf_graph_line(fp, depth, depth_mask,
+						   left_margin);
+		i = 0;
+		list_for_each_entry(chain, &child->val, list) {
+			if (chain->ip >= PERF_CONTEXT_MAX)
+				continue;
+			ret += ipchain__fprintf_graph(fp, chain, depth,
+						      new_depth_mask, i++,
+						      new_total,
+						      cumul,
+						      left_margin);
+		}
+		ret += __callchain__fprintf_graph(fp, child, new_total,
+						  depth + 1,
+						  new_depth_mask | (1 << depth),
+						  left_margin);
+		node = next;
+	}
+
+	if (callchain_param.mode == CHAIN_GRAPH_REL &&
+		remaining && remaining != new_total) {
+
+		if (!rem_sq_bracket)
+			return ret;
+
+		new_depth_mask &= ~(1 << (depth - 1));
+
+		ret += ipchain__fprintf_graph(fp, &rem_hits, depth,
+					      new_depth_mask, 0, new_total,
+					      remaining, left_margin);
+	}
+
+	return ret;
+}
+
+static size_t callchain__fprintf_graph(FILE *fp, struct callchain_node *self,
+				       u64 total_samples, int left_margin)
+{
+	struct callchain_list *chain;
+	bool printed = false;
+	int i = 0;
+	int ret = 0;
+
+	list_for_each_entry(chain, &self->val, list) {
+		if (chain->ip >= PERF_CONTEXT_MAX)
+			continue;
+
+		if (!i++ && sort__first_dimension == SORT_SYM)
+			continue;
+
+		if (!printed) {
+			ret += callchain__fprintf_left_margin(fp, left_margin);
+			ret += fprintf(fp, "|\n");
+			ret += callchain__fprintf_left_margin(fp, left_margin);
+			ret += fprintf(fp, "---");
+
+			left_margin += 3;
+			printed = true;
+		} else
+			ret += callchain__fprintf_left_margin(fp, left_margin);
+
+		if (chain->sym)
+			ret += fprintf(fp, " %s\n", chain->sym->name);
+		else
+			ret += fprintf(fp, " %p\n", (void *)(long)chain->ip);
+	}
+
+	ret += __callchain__fprintf_graph(fp, self, total_samples, 1, 1, left_margin);
+
+	return ret;
+}
+
+static size_t callchain__fprintf_flat(FILE *fp, struct callchain_node *self,
+				      u64 total_samples)
+{
+	struct callchain_list *chain;
+	size_t ret = 0;
+
+	if (!self)
+		return 0;
+
+	ret += callchain__fprintf_flat(fp, self->parent, total_samples);
+
+
+	list_for_each_entry(chain, &self->val, list) {
+		if (chain->ip >= PERF_CONTEXT_MAX)
+			continue;
+		if (chain->sym)
+			ret += fprintf(fp, "                %s\n", chain->sym->name);
+		else
+			ret += fprintf(fp, "                %p\n",
+					(void *)(long)chain->ip);
+	}
+
+	return ret;
+}
+
+static size_t hist_entry_callchain__fprintf(FILE *fp, struct hist_entry *self,
+					    u64 total_samples, int left_margin)
+{
+	struct rb_node *rb_node;
+	struct callchain_node *chain;
+	size_t ret = 0;
+
+	rb_node = rb_first(&self->sorted_chain);
+	while (rb_node) {
+		double percent;
+
+		chain = rb_entry(rb_node, struct callchain_node, rb_node);
+		percent = chain->hit * 100.0 / total_samples;
+		switch (callchain_param.mode) {
+		case CHAIN_FLAT:
+			ret += percent_color_fprintf(fp, "           %6.2f%%\n",
+						     percent);
+			ret += callchain__fprintf_flat(fp, chain, total_samples);
+			break;
+		case CHAIN_GRAPH_ABS: /* Falldown */
+		case CHAIN_GRAPH_REL:
+			ret += callchain__fprintf_graph(fp, chain, total_samples,
+							left_margin);
+		case CHAIN_NONE:
+		default:
+			break;
+		}
+		ret += fprintf(fp, "\n");
+		rb_node = rb_next(rb_node);
+	}
+
+	return ret;
+}
+
+static size_t hist_entry__fprintf(struct hist_entry *self,
+				  struct perf_session *session,
+				  struct perf_session *pair_session,
+				  bool show_displacement,
+				  long displacement, FILE *fp)
+{
+	struct sort_entry *se;
+	u64 count, total;
+	const char *sep = symbol_conf.field_sep;
+	size_t ret;
+
+	if (symbol_conf.exclude_other && !self->parent)
+		return 0;
+
+	if (pair_session) {
+		count = self->pair ? self->pair->count : 0;
+		total = pair_session->events_stats.total;
+	} else {
+		count = self->count;
+		total = session->events_stats.total;
+	}
+
+	if (total)
+		ret = percent_color_fprintf(fp, sep ? "%.2f" : "   %6.2f%%",
+					    (count * 100.0) / total);
+	else
+		ret = fprintf(fp, sep ? "%lld" : "%12lld ", count);
+
+	if (symbol_conf.show_nr_samples) {
+		if (sep)
+			fprintf(fp, "%c%lld", *sep, count);
+		else
+			fprintf(fp, "%11lld", count);
+	}
+
+	if (pair_session) {
+		char bf[32];
+		double old_percent = 0, new_percent = 0, diff;
+
+		if (total > 0)
+			old_percent = (count * 100.0) / total;
+		if (session->events_stats.total > 0)
+			new_percent = (self->count * 100.0) / session->events_stats.total;
+
+		diff = new_percent - old_percent;
+
+		if (fabs(diff) >= 0.01)
+			snprintf(bf, sizeof(bf), "%+4.2F%%", diff);
+		else
+			snprintf(bf, sizeof(bf), " ");
+
+		if (sep)
+			ret += fprintf(fp, "%c%s", *sep, bf);
+		else
+			ret += fprintf(fp, "%11.11s", bf);
+
+		if (show_displacement) {
+			if (displacement)
+				snprintf(bf, sizeof(bf), "%+4ld", displacement);
+			else
+				snprintf(bf, sizeof(bf), " ");
+
+			if (sep)
+				fprintf(fp, "%c%s", *sep, bf);
+			else
+				fprintf(fp, "%6.6s", bf);
+		}
+	}
+
+	list_for_each_entry(se, &hist_entry__sort_list, list) {
+		if (se->elide)
+			continue;
+
+		fprintf(fp, "%s", sep ?: "  ");
+		ret += se->print(fp, self, se->width ? *se->width : 0);
+	}
+
+	ret += fprintf(fp, "\n");
+
+	if (symbol_conf.use_callchain) {
+		int left_margin = 0;
+
+		if (sort__first_dimension == SORT_COMM) {
+			se = list_first_entry(&hist_entry__sort_list, typeof(*se),
+						list);
+			left_margin = se->width ? *se->width : 0;
+			left_margin -= thread__comm_len(self->thread);
+		}
+
+		hist_entry_callchain__fprintf(fp, self, session->events_stats.total,
+					      left_margin);
+	}
+
+	return ret;
+}
+
+size_t perf_session__fprintf_hists(struct perf_session *self,
+				   struct perf_session *pair,
+				   bool show_displacement, FILE *fp)
+{
+	struct sort_entry *se;
+	struct rb_node *nd;
+	size_t ret = 0;
+	unsigned long position = 1;
+	long displacement = 0;
+	unsigned int width;
+	const char *sep = symbol_conf.field_sep;
+	char *col_width = symbol_conf.col_width_list_str;
+
+	init_rem_hits();
+
+	fprintf(fp, "# %s", pair ? "Baseline" : "Overhead");
+
+	if (symbol_conf.show_nr_samples) {
+		if (sep)
+			fprintf(fp, "%cSamples", *sep);
+		else
+			fputs("  Samples  ", fp);
+	}
+
+	if (pair) {
+		if (sep)
+			ret += fprintf(fp, "%cDelta", *sep);
+		else
+			ret += fprintf(fp, "  Delta    ");
+
+		if (show_displacement) {
+			if (sep)
+				ret += fprintf(fp, "%cDisplacement", *sep);
+			else
+				ret += fprintf(fp, " Displ");
+		}
+	}
+
+	list_for_each_entry(se, &hist_entry__sort_list, list) {
+		if (se->elide)
+			continue;
+		if (sep) {
+			fprintf(fp, "%c%s", *sep, se->header);
+			continue;
+		}
+		width = strlen(se->header);
+		if (se->width) {
+			if (symbol_conf.col_width_list_str) {
+				if (col_width) {
+					*se->width = atoi(col_width);
+					col_width = strchr(col_width, ',');
+					if (col_width)
+						++col_width;
+				}
+			}
+			width = *se->width = max(*se->width, width);
+		}
+		fprintf(fp, "  %*s", width, se->header);
+	}
+	fprintf(fp, "\n");
+
+	if (sep)
+		goto print_entries;
+
+	fprintf(fp, "# ........");
+	if (symbol_conf.show_nr_samples)
+		fprintf(fp, " ..........");
+	if (pair) {
+		fprintf(fp, " ..........");
+		if (show_displacement)
+			fprintf(fp, " .....");
+	}
+	list_for_each_entry(se, &hist_entry__sort_list, list) {
+		unsigned int i;
+
+		if (se->elide)
+			continue;
+
+		fprintf(fp, "  ");
+		if (se->width)
+			width = *se->width;
+		else
+			width = strlen(se->header);
+		for (i = 0; i < width; i++)
+			fprintf(fp, ".");
+	}
+
+	fprintf(fp, "\n#\n");
+
+print_entries:
+	for (nd = rb_first(&self->hists); nd; nd = rb_next(nd)) {
+		struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node);
+
+		if (show_displacement) {
+			if (h->pair != NULL)
+				displacement = ((long)h->pair->position -
+					        (long)position);
+			else
+				displacement = 0;
+			++position;
+		}
+		ret += hist_entry__fprintf(h, self, pair, show_displacement,
+					   displacement, fp);
 	}
+
+	free(rem_sq_bracket);
+
+	return ret;
 }
diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h
index 3020db0c9292..e5f99b24048b 100644
--- a/tools/perf/util/hist.h
+++ b/tools/perf/util/hist.h
@@ -1,50 +1,27 @@
 #ifndef __PERF_HIST_H
 #define __PERF_HIST_H
-#include "../builtin.h"
 
-#include "util.h"
-
-#include "color.h"
-#include <linux/list.h>
-#include "cache.h"
-#include <linux/rbtree.h>
-#include "symbol.h"
-#include "string.h"
+#include <linux/types.h>
 #include "callchain.h"
-#include "strlist.h"
-#include "values.h"
-
-#include "../perf.h"
-#include "debug.h"
-#include "header.h"
-
-#include "parse-options.h"
-#include "parse-events.h"
 
-#include "thread.h"
-#include "sort.h"
-
-extern struct rb_root hist;
-extern struct rb_root collapse_hists;
-extern struct rb_root output_hists;
-extern int callchain;
 extern struct callchain_param callchain_param;
-extern unsigned long total;
-extern unsigned long total_mmap;
-extern unsigned long total_comm;
-extern unsigned long total_fork;
-extern unsigned long total_unknown;
-extern unsigned long total_lost;
 
-struct hist_entry *__hist_entry__add(struct addr_location *al,
-				     struct symbol *parent,
-				     u64 count, bool *hit);
+struct perf_session;
+struct hist_entry;
+struct addr_location;
+struct symbol;
+
+struct hist_entry *__perf_session__add_hist_entry(struct perf_session *self,
+						  struct addr_location *al,
+						  struct symbol *parent,
+						  u64 count, bool *hit);
 extern int64_t hist_entry__cmp(struct hist_entry *, struct hist_entry *);
 extern int64_t hist_entry__collapse(struct hist_entry *, struct hist_entry *);
-extern void hist_entry__free(struct hist_entry *);
-extern void collapse__insert_entry(struct hist_entry *);
-extern void collapse__resort(void);
-extern void output__insert_entry(struct hist_entry *, u64);
-extern void output__resort(u64);
+void hist_entry__free(struct hist_entry *);
 
+void perf_session__output_resort(struct perf_session *self, u64 total_samples);
+void perf_session__collapse_resort(struct perf_session *self);
+size_t perf_session__fprintf_hists(struct perf_session *self,
+				   struct perf_session *pair,
+				   bool show_displacement, FILE *fp);
 #endif	/* __PERF_HIST_H */
diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c
index 69f94fe9db20..c4d55a0da2ea 100644
--- a/tools/perf/util/map.c
+++ b/tools/perf/util/map.c
@@ -104,43 +104,70 @@ void map__fixup_end(struct map *self)
 
 #define DSO__DELETED "(deleted)"
 
-struct symbol *map__find_symbol(struct map *self, u64 addr,
-				symbol_filter_t filter)
+int map__load(struct map *self, struct perf_session *session,
+	      symbol_filter_t filter)
 {
-	if (!dso__loaded(self->dso, self->type)) {
-		int nr = dso__load(self->dso, self, filter);
-
-		if (nr < 0) {
-			if (self->dso->has_build_id) {
-				char sbuild_id[BUILD_ID_SIZE * 2 + 1];
-
-				build_id__sprintf(self->dso->build_id,
-						  sizeof(self->dso->build_id),
-						  sbuild_id);
-				pr_warning("%s with build id %s not found",
-					   self->dso->long_name, sbuild_id);
-			} else
-				pr_warning("Failed to open %s",
-					   self->dso->long_name);
-			pr_warning(", continuing without symbols\n");
-			return NULL;
-		} else if (nr == 0) {
-			const char *name = self->dso->long_name;
-			const size_t len = strlen(name);
-			const size_t real_len = len - sizeof(DSO__DELETED);
-
-			if (len > sizeof(DSO__DELETED) &&
-			    strcmp(name + real_len + 1, DSO__DELETED) == 0) {
-				pr_warning("%.*s was updated, restart the long running apps that use it!\n",
-					   (int)real_len, name);
-			} else {
-				pr_warning("no symbols found in %s, maybe install a debug package?\n", name);
-			}
-			return NULL;
+	const char *name = self->dso->long_name;
+	int nr;
+
+	if (dso__loaded(self->dso, self->type))
+		return 0;
+
+	nr = dso__load(self->dso, self, session, filter);
+	if (nr < 0) {
+		if (self->dso->has_build_id) {
+			char sbuild_id[BUILD_ID_SIZE * 2 + 1];
+
+			build_id__sprintf(self->dso->build_id,
+					  sizeof(self->dso->build_id),
+					  sbuild_id);
+			pr_warning("%s with build id %s not found",
+				   name, sbuild_id);
+		} else
+			pr_warning("Failed to open %s", name);
+
+		pr_warning(", continuing without symbols\n");
+		return -1;
+	} else if (nr == 0) {
+		const size_t len = strlen(name);
+		const size_t real_len = len - sizeof(DSO__DELETED);
+
+		if (len > sizeof(DSO__DELETED) &&
+		    strcmp(name + real_len + 1, DSO__DELETED) == 0) {
+			pr_warning("%.*s was updated, restart the long "
+				   "running apps that use it!\n",
+				   (int)real_len, name);
+		} else {
+			pr_warning("no symbols found in %s, maybe install "
+				   "a debug package?\n", name);
 		}
+
+		return -1;
 	}
 
-	return self->dso->find_symbol(self->dso, self->type, addr);
+	return 0;
+}
+
+struct symbol *map__find_symbol(struct map *self, struct perf_session *session,
+				u64 addr, symbol_filter_t filter)
+{
+	if (map__load(self, session, filter) < 0)
+		return NULL;
+
+	return dso__find_symbol(self->dso, self->type, addr);
+}
+
+struct symbol *map__find_symbol_by_name(struct map *self, const char *name,
+					struct perf_session *session,
+					symbol_filter_t filter)
+{
+	if (map__load(self, session, filter) < 0)
+		return NULL;
+
+	if (!dso__sorted_by_name(self->dso, self->type))
+		dso__sort_by_name(self->dso, self->type);
+
+	return dso__find_symbol_by_name(self->dso, self->type, name);
 }
 
 struct map *map__clone(struct map *self)
diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index 9e5dbd66d34d..e5bc0fb016b2 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -197,7 +197,7 @@ struct tracepoint_path *tracepoint_id_to_path(u64 config)
 			if (id == config) {
 				closedir(evt_dir);
 				closedir(sys_dir);
-				path = zalloc(sizeof(path));
+				path = zalloc(sizeof(*path));
 				path->system = malloc(MAX_EVENT_LENGTH);
 				if (!path->system) {
 					free(path);
@@ -467,7 +467,6 @@ parse_subsystem_tracepoint_event(char *sys_name, char *flags)
 	while ((evt_ent = readdir(evt_dir))) {
 		char event_opt[MAX_EVOPT_LEN + 1];
 		int len;
-		unsigned int rem = MAX_EVOPT_LEN;
 
 		if (!strcmp(evt_ent->d_name, ".")
 		    || !strcmp(evt_ent->d_name, "..")
@@ -475,20 +474,12 @@ parse_subsystem_tracepoint_event(char *sys_name, char *flags)
 		    || !strcmp(evt_ent->d_name, "filter"))
 			continue;
 
-		len = snprintf(event_opt, MAX_EVOPT_LEN, "%s:%s", sys_name,
-			       evt_ent->d_name);
+		len = snprintf(event_opt, MAX_EVOPT_LEN, "%s:%s%s%s", sys_name,
+			       evt_ent->d_name, flags ? ":" : "",
+			       flags ?: "");
 		if (len < 0)
 			return EVT_FAILED;
 
-		rem -= len;
-		if (flags) {
-			if (rem < strlen(flags) + 1)
-				return EVT_FAILED;
-
-			strcat(event_opt, ":");
-			strcat(event_opt, flags);
-		}
-
 		if (parse_events(NULL, event_opt, 0))
 			return EVT_FAILED;
 	}
diff --git a/tools/perf/util/parse-options.c b/tools/perf/util/parse-options.c
index 6d8af48c925e..efebd5b476b3 100644
--- a/tools/perf/util/parse-options.c
+++ b/tools/perf/util/parse-options.c
@@ -430,6 +430,9 @@ int usage_with_options_internal(const char * const *usagestr,
 		pos = fprintf(stderr, "    ");
 		if (opts->short_name)
 			pos += fprintf(stderr, "-%c", opts->short_name);
+		else
+			pos += fprintf(stderr, "    ");
+
 		if (opts->long_name && opts->short_name)
 			pos += fprintf(stderr, ", ");
 		if (opts->long_name)
diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c
index cd7fbda5e2a5..29465d440043 100644
--- a/tools/perf/util/probe-event.c
+++ b/tools/perf/util/probe-event.c
@@ -48,6 +48,9 @@
 
 /* If there is no space to write, returns -E2BIG. */
 static int e_snprintf(char *str, size_t size, const char *format, ...)
+	__attribute__((format(printf, 3, 4)));
+
+static int e_snprintf(char *str, size_t size, const char *format, ...)
 {
 	int ret;
 	va_list ap;
@@ -59,6 +62,18 @@ static int e_snprintf(char *str, size_t size, const char *format, ...)
 	return ret;
 }
 
+/* Check the name is good for event/group */
+static bool check_event_name(const char *name)
+{
+	if (!isalpha(*name) && *name != '_')
+		return false;
+	while (*++name != '\0') {
+		if (!isalpha(*name) && !isdigit(*name) && *name != '_')
+			return false;
+	}
+	return true;
+}
+
 /* Parse probepoint definition. */
 static void parse_perf_probe_probepoint(char *arg, struct probe_point *pp)
 {
@@ -66,10 +81,26 @@ static void parse_perf_probe_probepoint(char *arg, struct probe_point *pp)
 	char c, nc = 0;
 	/*
 	 * <Syntax>
-	 * perf probe SRC:LN
-	 * perf probe FUNC[+OFFS|%return][@SRC]
+	 * perf probe [EVENT=]SRC:LN
+	 * perf probe [EVENT=]FUNC[+OFFS|%return][@SRC]
+	 *
+	 * TODO:Group name support
 	 */
 
+	ptr = strchr(arg, '=');
+	if (ptr) {	/* Event name */
+		*ptr = '\0';
+		tmp = ptr + 1;
+		ptr = strchr(arg, ':');
+		if (ptr)	/* Group name is not supported yet. */
+			semantic_error("Group name is not supported yet.");
+		if (!check_event_name(arg))
+			semantic_error("%s is bad for event name -it must "
+				       "follow C symbol-naming rule.", arg);
+		pp->event = strdup(arg);
+		arg = tmp;
+	}
+
 	ptr = strpbrk(arg, ":+@%");
 	if (ptr) {
 		nc = *ptr;
@@ -147,10 +178,13 @@ static void parse_perf_probe_probepoint(char *arg, struct probe_point *pp)
 }
 
 /* Parse perf-probe event definition */
-int parse_perf_probe_event(const char *str, struct probe_point *pp)
+void parse_perf_probe_event(const char *str, struct probe_point *pp,
+			    bool *need_dwarf)
 {
 	char **argv;
-	int argc, i, need_dwarf = 0;
+	int argc, i;
+
+	*need_dwarf = false;
 
 	argv = argv_split(str, &argc);
 	if (!argv)
@@ -161,7 +195,7 @@ int parse_perf_probe_event(const char *str, struct probe_point *pp)
 	/* Parse probe point */
 	parse_perf_probe_probepoint(argv[0], pp);
 	if (pp->file || pp->line)
-		need_dwarf = 1;
+		*need_dwarf = true;
 
 	/* Copy arguments and ensure return probe has no C argument */
 	pp->nr_args = argc - 1;
@@ -174,17 +208,15 @@ int parse_perf_probe_event(const char *str, struct probe_point *pp)
 			if (pp->retprobe)
 				semantic_error("You can't specify local"
 						" variable for kretprobe");
-			need_dwarf = 1;
+			*need_dwarf = true;
 		}
 	}
 
 	argv_free(argv);
-	return need_dwarf;
 }
 
 /* Parse kprobe_events event into struct probe_point */
-void parse_trace_kprobe_event(const char *str, char **group, char **event,
-			      struct probe_point *pp)
+void parse_trace_kprobe_event(const char *str, struct probe_point *pp)
 {
 	char pr;
 	char *p;
@@ -200,18 +232,17 @@ void parse_trace_kprobe_event(const char *str, char **group, char **event,
 
 	/* Scan event and group name. */
 	ret = sscanf(argv[0], "%c:%a[^/ \t]/%a[^ \t]",
-		     &pr, (float *)(void *)group, (float *)(void *)event);
+		     &pr, (float *)(void *)&pp->group,
+		     (float *)(void *)&pp->event);
 	if (ret != 3)
 		semantic_error("Failed to parse event name: %s", argv[0]);
-	pr_debug("Group:%s Event:%s probe:%c\n", *group, *event, pr);
-
-	if (!pp)
-		goto end;
+	pr_debug("Group:%s Event:%s probe:%c\n", pp->group, pp->event, pr);
 
 	pp->retprobe = (pr == 'r');
 
 	/* Scan function name and offset */
-	ret = sscanf(argv[1], "%a[^+]+%d", (float *)(void *)&pp->function, &pp->offset);
+	ret = sscanf(argv[1], "%a[^+]+%d", (float *)(void *)&pp->function,
+		     &pp->offset);
 	if (ret == 1)
 		pp->offset = 0;
 
@@ -230,15 +261,15 @@ void parse_trace_kprobe_event(const char *str, char **group, char **event,
 			die("Failed to copy argument.");
 	}
 
-end:
 	argv_free(argv);
 }
 
-int synthesize_perf_probe_event(struct probe_point *pp)
+/* Synthesize only probe point (not argument) */
+int synthesize_perf_probe_point(struct probe_point *pp)
 {
 	char *buf;
 	char offs[64] = "", line[64] = "";
-	int i, len, ret;
+	int ret;
 
 	pp->probes[0] = buf = zalloc(MAX_CMDLEN);
 	if (!buf)
@@ -258,11 +289,25 @@ int synthesize_perf_probe_event(struct probe_point *pp)
 		ret = e_snprintf(buf, MAX_CMDLEN, "%s%s%s%s", pp->function,
 				 offs, pp->retprobe ? "%return" : "", line);
 	else
-		ret = e_snprintf(buf, MAX_CMDLEN, "%s%s%s%s", pp->file, line);
-	if (ret <= 0)
-		goto error;
-	len = ret;
+		ret = e_snprintf(buf, MAX_CMDLEN, "%s%s", pp->file, line);
+	if (ret <= 0) {
+error:
+		free(pp->probes[0]);
+		pp->probes[0] = NULL;
+	}
+	return ret;
+}
+
+int synthesize_perf_probe_event(struct probe_point *pp)
+{
+	char *buf;
+	int i, len, ret;
+
+	len = synthesize_perf_probe_point(pp);
+	if (len < 0)
+		return 0;
 
+	buf = pp->probes[0];
 	for (i = 0; i < pp->nr_args; i++) {
 		ret = e_snprintf(&buf[len], MAX_CMDLEN - len, " %s",
 				 pp->args[i]);
@@ -275,6 +320,7 @@ int synthesize_perf_probe_event(struct probe_point *pp)
 	return pp->found;
 error:
 	free(pp->probes[0]);
+	pp->probes[0] = NULL;
 
 	return ret;
 }
@@ -304,6 +350,7 @@ int synthesize_trace_kprobe_event(struct probe_point *pp)
 	return pp->found;
 error:
 	free(pp->probes[0]);
+	pp->probes[0] = NULL;
 
 	return ret;
 }
@@ -363,6 +410,10 @@ static void clear_probe_point(struct probe_point *pp)
 {
 	int i;
 
+	if (pp->event)
+		free(pp->event);
+	if (pp->group)
+		free(pp->group);
 	if (pp->function)
 		free(pp->function);
 	if (pp->file)
@@ -373,15 +424,33 @@ static void clear_probe_point(struct probe_point *pp)
 		free(pp->args);
 	for (i = 0; i < pp->found; i++)
 		free(pp->probes[i]);
-	memset(pp, 0, sizeof(pp));
+	memset(pp, 0, sizeof(*pp));
+}
+
+/* Show an event */
+static void show_perf_probe_event(const char *event, const char *place,
+				  struct probe_point *pp)
+{
+	int i, ret;
+	char buf[128];
+
+	ret = e_snprintf(buf, 128, "%s:%s", pp->group, event);
+	if (ret < 0)
+		die("Failed to copy event: %s", strerror(-ret));
+	printf("  %-40s (on %s", buf, place);
+
+	if (pp->nr_args > 0) {
+		printf(" with");
+		for (i = 0; i < pp->nr_args; i++)
+			printf(" %s", pp->args[i]);
+	}
+	printf(")\n");
 }
 
 /* List up current perf-probe events */
 void show_perf_probe_events(void)
 {
-	unsigned int i;
 	int fd;
-	char *group, *event;
 	struct probe_point pp;
 	struct strlist *rawlist;
 	struct str_node *ent;
@@ -390,13 +459,12 @@ void show_perf_probe_events(void)
 	rawlist = get_trace_kprobe_event_rawlist(fd);
 	close(fd);
 
-	for (i = 0; i < strlist__nr_entries(rawlist); i++) {
-		ent = strlist__entry(rawlist, i);
-		parse_trace_kprobe_event(ent->s, &group, &event, &pp);
-		synthesize_perf_probe_event(&pp);
-		printf("[%s:%s]\t%s\n", group, event, pp.probes[0]);
-		free(group);
-		free(event);
+	strlist__for_each(ent, rawlist) {
+		parse_trace_kprobe_event(ent->s, &pp);
+		/* Synthesize only event probe point */
+		synthesize_perf_probe_point(&pp);
+		/* Show an event */
+		show_perf_probe_event(pp.event, pp.probes[0], &pp);
 		clear_probe_point(&pp);
 	}
 
@@ -404,21 +472,27 @@ void show_perf_probe_events(void)
 }
 
 /* Get current perf-probe event names */
-static struct strlist *get_perf_event_names(int fd)
+static struct strlist *get_perf_event_names(int fd, bool include_group)
 {
-	unsigned int i;
-	char *group, *event;
+	char buf[128];
 	struct strlist *sl, *rawlist;
 	struct str_node *ent;
+	struct probe_point pp;
 
+	memset(&pp, 0, sizeof(pp));
 	rawlist = get_trace_kprobe_event_rawlist(fd);
 
-	sl = strlist__new(false, NULL);
-	for (i = 0; i < strlist__nr_entries(rawlist); i++) {
-		ent = strlist__entry(rawlist, i);
-		parse_trace_kprobe_event(ent->s, &group, &event, NULL);
-		strlist__add(sl, event);
-		free(group);
+	sl = strlist__new(true, NULL);
+	strlist__for_each(ent, rawlist) {
+		parse_trace_kprobe_event(ent->s, &pp);
+		if (include_group) {
+			if (e_snprintf(buf, 128, "%s:%s", pp.group,
+				       pp.event) < 0)
+				die("Failed to copy group:event name.");
+			strlist__add(sl, buf);
+		} else
+			strlist__add(sl, pp.event);
+		clear_probe_point(&pp);
 	}
 
 	strlist__delete(rawlist);
@@ -426,24 +500,36 @@ static struct strlist *get_perf_event_names(int fd)
 	return sl;
 }
 
-static int write_trace_kprobe_event(int fd, const char *buf)
+static void write_trace_kprobe_event(int fd, const char *buf)
 {
 	int ret;
 
+	pr_debug("Writing event: %s\n", buf);
 	ret = write(fd, buf, strlen(buf));
 	if (ret <= 0)
-		die("Failed to create event.");
-	else
-		printf("Added new event: %s\n", buf);
-
-	return ret;
+		die("Failed to write event: %s", strerror(errno));
 }
 
 static void get_new_event_name(char *buf, size_t len, const char *base,
-			       struct strlist *namelist)
+			       struct strlist *namelist, bool allow_suffix)
 {
 	int i, ret;
-	for (i = 0; i < MAX_EVENT_INDEX; i++) {
+
+	/* Try no suffix */
+	ret = e_snprintf(buf, len, "%s", base);
+	if (ret < 0)
+		die("snprintf() failed: %s", strerror(-ret));
+	if (!strlist__has_entry(namelist, buf))
+		return;
+
+	if (!allow_suffix) {
+		pr_warning("Error: event \"%s\" already exists. "
+			   "(Use -f to force duplicates.)\n", base);
+		die("Can't add new event.");
+	}
+
+	/* Try to add suffix */
+	for (i = 1; i < MAX_EVENT_INDEX; i++) {
 		ret = e_snprintf(buf, len, "%s_%d", base, i);
 		if (ret < 0)
 			die("snprintf() failed: %s", strerror(-ret));
@@ -454,31 +540,138 @@ static void get_new_event_name(char *buf, size_t len, const char *base,
 		die("Too many events are on the same function.");
 }
 
-void add_trace_kprobe_events(struct probe_point *probes, int nr_probes)
+void add_trace_kprobe_events(struct probe_point *probes, int nr_probes,
+			     bool force_add)
 {
 	int i, j, fd;
 	struct probe_point *pp;
 	char buf[MAX_CMDLEN];
 	char event[64];
 	struct strlist *namelist;
+	bool allow_suffix;
 
 	fd = open_kprobe_events(O_RDWR, O_APPEND);
 	/* Get current event names */
-	namelist = get_perf_event_names(fd);
+	namelist = get_perf_event_names(fd, false);
 
 	for (j = 0; j < nr_probes; j++) {
 		pp = probes + j;
+		if (!pp->event)
+			pp->event = strdup(pp->function);
+		if (!pp->group)
+			pp->group = strdup(PERFPROBE_GROUP);
+		DIE_IF(!pp->event || !pp->group);
+		/* If force_add is true, suffix search is allowed */
+		allow_suffix = force_add;
 		for (i = 0; i < pp->found; i++) {
 			/* Get an unused new event name */
-			get_new_event_name(event, 64, pp->function, namelist);
+			get_new_event_name(event, 64, pp->event, namelist,
+					   allow_suffix);
 			snprintf(buf, MAX_CMDLEN, "%c:%s/%s %s\n",
 				 pp->retprobe ? 'r' : 'p',
-				 PERFPROBE_GROUP, event,
+				 pp->group, event,
 				 pp->probes[i]);
 			write_trace_kprobe_event(fd, buf);
+			printf("Added new event:\n");
+			/* Get the first parameter (probe-point) */
+			sscanf(pp->probes[i], "%s", buf);
+			show_perf_probe_event(event, buf, pp);
 			/* Add added event name to namelist */
 			strlist__add(namelist, event);
+			/*
+			 * Probes after the first probe which comes from same
+			 * user input are always allowed to add suffix, because
+			 * there might be several addresses corresponding to
+			 * one code line.
+			 */
+			allow_suffix = true;
+		}
+	}
+	/* Show how to use the event. */
+	printf("\nYou can now use it on all perf tools, such as:\n\n");
+	printf("\tperf record -e %s:%s -a sleep 1\n\n", PERFPROBE_GROUP, event);
+
+	strlist__delete(namelist);
+	close(fd);
+}
+
+static void __del_trace_kprobe_event(int fd, struct str_node *ent)
+{
+	char *p;
+	char buf[128];
+
+	/* Convert from perf-probe event to trace-kprobe event */
+	if (e_snprintf(buf, 128, "-:%s", ent->s) < 0)
+		die("Failed to copy event.");
+	p = strchr(buf + 2, ':');
+	if (!p)
+		die("Internal error: %s should have ':' but not.", ent->s);
+	*p = '/';
+
+	write_trace_kprobe_event(fd, buf);
+	printf("Remove event: %s\n", ent->s);
+}
+
+static void del_trace_kprobe_event(int fd, const char *group,
+				   const char *event, struct strlist *namelist)
+{
+	char buf[128];
+	struct str_node *ent, *n;
+	int found = 0;
+
+	if (e_snprintf(buf, 128, "%s:%s", group, event) < 0)
+		die("Failed to copy event.");
+
+	if (strpbrk(buf, "*?")) { /* Glob-exp */
+		strlist__for_each_safe(ent, n, namelist)
+			if (strglobmatch(ent->s, buf)) {
+				found++;
+				__del_trace_kprobe_event(fd, ent);
+				strlist__remove(namelist, ent);
+			}
+	} else {
+		ent = strlist__find(namelist, buf);
+		if (ent) {
+			found++;
+			__del_trace_kprobe_event(fd, ent);
+			strlist__remove(namelist, ent);
 		}
 	}
+	if (found == 0)
+		pr_info("Info: event \"%s\" does not exist, could not remove it.\n", buf);
+}
+
+void del_trace_kprobe_events(struct strlist *dellist)
+{
+	int fd;
+	const char *group, *event;
+	char *p, *str;
+	struct str_node *ent;
+	struct strlist *namelist;
+
+	fd = open_kprobe_events(O_RDWR, O_APPEND);
+	/* Get current event names */
+	namelist = get_perf_event_names(fd, true);
+
+	strlist__for_each(ent, dellist) {
+		str = strdup(ent->s);
+		if (!str)
+			die("Failed to copy event.");
+		pr_debug("Parsing: %s\n", str);
+		p = strchr(str, ':');
+		if (p) {
+			group = str;
+			*p = '\0';
+			event = p + 1;
+		} else {
+			group = "*";
+			event = str;
+		}
+		pr_debug("Group: %s, Event: %s\n", group, event);
+		del_trace_kprobe_event(fd, group, event, namelist);
+		free(str);
+	}
+	strlist__delete(namelist);
 	close(fd);
 }
+
diff --git a/tools/perf/util/probe-event.h b/tools/perf/util/probe-event.h
index 0c6fe56fe38a..7f1d499118c0 100644
--- a/tools/perf/util/probe-event.h
+++ b/tools/perf/util/probe-event.h
@@ -1,15 +1,19 @@
 #ifndef _PROBE_EVENT_H
 #define _PROBE_EVENT_H
 
+#include <stdbool.h>
 #include "probe-finder.h"
 #include "strlist.h"
 
-extern int parse_perf_probe_event(const char *str, struct probe_point *pp);
+extern void parse_perf_probe_event(const char *str, struct probe_point *pp,
+				   bool *need_dwarf);
+extern int synthesize_perf_probe_point(struct probe_point *pp);
 extern int synthesize_perf_probe_event(struct probe_point *pp);
-extern void parse_trace_kprobe_event(const char *str, char **group,
-				     char **event, struct probe_point *pp);
+extern void parse_trace_kprobe_event(const char *str, struct probe_point *pp);
 extern int synthesize_trace_kprobe_event(struct probe_point *pp);
-extern void add_trace_kprobe_events(struct probe_point *probes, int nr_probes);
+extern void add_trace_kprobe_events(struct probe_point *probes, int nr_probes,
+				    bool force_add);
+extern void del_trace_kprobe_events(struct strlist *dellist);
 extern void show_perf_probe_events(void);
 
 /* Maximum index number of event-name postfix */
diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c
index 293cdfc1b8ca..4b852c0d16a5 100644
--- a/tools/perf/util/probe-finder.c
+++ b/tools/perf/util/probe-finder.c
@@ -106,7 +106,7 @@ static int strtailcmp(const char *s1, const char *s2)
 {
 	int i1 = strlen(s1);
 	int i2 = strlen(s2);
-	while (--i1 > 0 && --i2 > 0) {
+	while (--i1 >= 0 && --i2 >= 0) {
 		if (s1[i1] != s2[i2])
 			return s1[i1] - s2[i2];
 	}
@@ -687,10 +687,8 @@ int find_probepoint(int fd, struct probe_point *pp)
 	struct probe_finder pf = {.pp = pp};
 
 	ret = dwarf_init(fd, DW_DLC_READ, 0, 0, &__dw_debug, &__dw_error);
-	if (ret != DW_DLV_OK) {
-		pr_warning("No dwarf info found in the vmlinux - please rebuild with CONFIG_DEBUG_INFO.\n");
+	if (ret != DW_DLV_OK)
 		return -ENOENT;
-	}
 
 	pp->found = 0;
 	while (++cu_number) {
diff --git a/tools/perf/util/probe-finder.h b/tools/perf/util/probe-finder.h
index bdebca6697d2..a4086aaddb73 100644
--- a/tools/perf/util/probe-finder.h
+++ b/tools/perf/util/probe-finder.h
@@ -1,9 +1,9 @@
 #ifndef _PROBE_FINDER_H
 #define _PROBE_FINDER_H
 
-#define MAX_PATH_LEN 256
-#define MAX_PROBE_BUFFER 1024
-#define MAX_PROBES 128
+#define MAX_PATH_LEN		 256
+#define MAX_PROBE_BUFFER	1024
+#define MAX_PROBES		 128
 
 static inline int is_c_varname(const char *name)
 {
@@ -12,45 +12,53 @@ static inline int is_c_varname(const char *name)
 }
 
 struct probe_point {
+	char			*event;			/* Event name */
+	char			*group;			/* Event group */
+
 	/* Inputs */
-	char	*file;		/* File name */
-	int	line;		/* Line number */
+	char			*file;			/* File name */
+	int			line;			/* Line number */
 
-	char	*function;	/* Function name */
-	int	offset;		/* Offset bytes */
+	char			*function;		/* Function name */
+	int			offset;			/* Offset bytes */
 
-	int	nr_args;	/* Number of arguments */
-	char	**args;		/* Arguments */
+	int			nr_args;		/* Number of arguments */
+	char			**args;			/* Arguments */
 
-	int	retprobe;	/* Return probe */
+	int			retprobe;		/* Return probe */
 
 	/* Output */
-	int	found;		/* Number of found probe points */
-	char	*probes[MAX_PROBES];	/* Output buffers (will be allocated)*/
+	int			found;			/* Number of found probe points */
+	char			*probes[MAX_PROBES];	/* Output buffers (will be allocated)*/
 };
 
 #ifndef NO_LIBDWARF
 extern int find_probepoint(int fd, struct probe_point *pp);
 
-#include <libdwarf/dwarf.h>
-#include <libdwarf/libdwarf.h>
+/* Workaround for undefined _MIPS_SZLONG bug in libdwarf.h: */
+#ifndef _MIPS_SZLONG
+# define _MIPS_SZLONG		0
+#endif
+
+#include <dwarf.h>
+#include <libdwarf.h>
 
 struct probe_finder {
-	struct probe_point	*pp;	/* Target probe point */
+	struct probe_point	*pp;			/* Target probe point */
 
 	/* For function searching */
-	Dwarf_Addr	addr;		/* Address */
-	Dwarf_Unsigned	fno;		/* File number */
-	Dwarf_Unsigned	lno;		/* Line number */
-	Dwarf_Off	inl_offs;	/* Inline offset */
-	Dwarf_Die	cu_die;		/* Current CU */
+	Dwarf_Addr		addr;			/* Address */
+	Dwarf_Unsigned		fno;			/* File number */
+	Dwarf_Unsigned		lno;			/* Line number */
+	Dwarf_Off		inl_offs;		/* Inline offset */
+	Dwarf_Die		cu_die;			/* Current CU */
 
 	/* For variable searching */
-	Dwarf_Addr	cu_base;	/* Current CU base address */
-	Dwarf_Locdesc	fbloc;		/* Location of Current Frame Base */
-	const char	*var;		/* Current variable name */
-	char		*buf;		/* Current output buffer */
-	int		len;		/* Length of output buffer */
+	Dwarf_Addr		cu_base;		/* Current CU base address */
+	Dwarf_Locdesc		fbloc;			/* Location of Current Frame Base */
+	const char		*var;			/* Current variable name */
+	char			*buf;			/* Current output buffer */
+	int			len;			/* Length of output buffer */
 };
 #endif /* NO_LIBDWARF */
 
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
new file mode 100644
index 000000000000..ce3a6c8abe76
--- /dev/null
+++ b/tools/perf/util/session.c
@@ -0,0 +1,150 @@
+#include <linux/kernel.h>
+
+#include <unistd.h>
+#include <sys/types.h>
+
+#include "session.h"
+#include "sort.h"
+#include "util.h"
+
+static int perf_session__open(struct perf_session *self, bool force)
+{
+	struct stat input_stat;
+
+	self->fd = open(self->filename, O_RDONLY);
+	if (self->fd < 0) {
+		pr_err("failed to open file: %s", self->filename);
+		if (!strcmp(self->filename, "perf.data"))
+			pr_err("  (try 'perf record' first)");
+		pr_err("\n");
+		return -errno;
+	}
+
+	if (fstat(self->fd, &input_stat) < 0)
+		goto out_close;
+
+	if (!force && input_stat.st_uid && (input_stat.st_uid != geteuid())) {
+		pr_err("file %s not owned by current user or root\n",
+		       self->filename);
+		goto out_close;
+	}
+
+	if (!input_stat.st_size) {
+		pr_info("zero-sized file (%s), nothing to do!\n",
+			self->filename);
+		goto out_close;
+	}
+
+	if (perf_header__read(&self->header, self->fd) < 0) {
+		pr_err("incompatible file format");
+		goto out_close;
+	}
+
+	self->size = input_stat.st_size;
+	return 0;
+
+out_close:
+	close(self->fd);
+	self->fd = -1;
+	return -1;
+}
+
+struct perf_session *perf_session__new(const char *filename, int mode, bool force)
+{
+	size_t len = filename ? strlen(filename) + 1 : 0;
+	struct perf_session *self = zalloc(sizeof(*self) + len);
+
+	if (self == NULL)
+		goto out;
+
+	if (perf_header__init(&self->header) < 0)
+		goto out_free;
+
+	memcpy(self->filename, filename, len);
+	self->threads = RB_ROOT;
+	self->last_match = NULL;
+	self->mmap_window = 32;
+	self->cwd = NULL;
+	self->cwdlen = 0;
+	map_groups__init(&self->kmaps);
+
+	if (perf_session__create_kernel_maps(self) < 0)
+		goto out_delete;
+
+	if (mode == O_RDONLY && perf_session__open(self, force) < 0)
+		goto out_delete;
+out:
+	return self;
+out_free:
+	free(self);
+	return NULL;
+out_delete:
+	perf_session__delete(self);
+	return NULL;
+}
+
+void perf_session__delete(struct perf_session *self)
+{
+	perf_header__exit(&self->header);
+	close(self->fd);
+	free(self->cwd);
+	free(self);
+}
+
+static bool symbol__match_parent_regex(struct symbol *sym)
+{
+	if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
+		return 1;
+
+	return 0;
+}
+
+struct symbol **perf_session__resolve_callchain(struct perf_session *self,
+						struct thread *thread,
+						struct ip_callchain *chain,
+						struct symbol **parent)
+{
+	u8 cpumode = PERF_RECORD_MISC_USER;
+	struct symbol **syms = NULL;
+	unsigned int i;
+
+	if (symbol_conf.use_callchain) {
+		syms = calloc(chain->nr, sizeof(*syms));
+		if (!syms) {
+			fprintf(stderr, "Can't allocate memory for symbols\n");
+			exit(-1);
+		}
+	}
+
+	for (i = 0; i < chain->nr; i++) {
+		u64 ip = chain->ips[i];
+		struct addr_location al;
+
+		if (ip >= PERF_CONTEXT_MAX) {
+			switch (ip) {
+			case PERF_CONTEXT_HV:
+				cpumode = PERF_RECORD_MISC_HYPERVISOR;	break;
+			case PERF_CONTEXT_KERNEL:
+				cpumode = PERF_RECORD_MISC_KERNEL;	break;
+			case PERF_CONTEXT_USER:
+				cpumode = PERF_RECORD_MISC_USER;	break;
+			default:
+				break;
+			}
+			continue;
+		}
+
+		thread__find_addr_location(thread, self, cpumode,
+					   MAP__FUNCTION, ip, &al, NULL);
+		if (al.sym != NULL) {
+			if (sort__has_parent && !*parent &&
+			    symbol__match_parent_regex(al.sym))
+				*parent = al.sym;
+			if (!symbol_conf.use_callchain)
+				break;
+			syms[i] = al.sym;
+		}
+	}
+
+	return syms;
+}
diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
new file mode 100644
index 000000000000..32eaa1bada06
--- /dev/null
+++ b/tools/perf/util/session.h
@@ -0,0 +1,61 @@
+#ifndef __PERF_SESSION_H
+#define __PERF_SESSION_H
+
+#include "event.h"
+#include "header.h"
+#include "thread.h"
+#include <linux/rbtree.h>
+#include "../../../include/linux/perf_event.h"
+
+struct ip_callchain;
+struct thread;
+struct symbol;
+
+struct perf_session {
+	struct perf_header	header;
+	unsigned long		size;
+	unsigned long		mmap_window;
+	struct map_groups	kmaps;
+	struct rb_root		threads;
+	struct thread		*last_match;
+	struct events_stats	events_stats;
+	unsigned long		event_total[PERF_RECORD_MAX];
+	struct rb_root		hists;
+	u64			sample_type;
+	int			fd;
+	int			cwdlen;
+	char			*cwd;
+	char filename[0];
+};
+
+typedef int (*event_op)(event_t *self, struct perf_session *session);
+
+struct perf_event_ops {
+	event_op	process_sample_event;
+	event_op	process_mmap_event;
+	event_op	process_comm_event;
+	event_op	process_fork_event;
+	event_op	process_exit_event;
+	event_op	process_lost_event;
+	event_op	process_read_event;
+	event_op	process_throttle_event;
+	event_op	process_unthrottle_event;
+	int		(*sample_type_check)(struct perf_session *session);
+	unsigned long	total_unknown;
+	bool		full_paths;
+};
+
+struct perf_session *perf_session__new(const char *filename, int mode, bool force);
+void perf_session__delete(struct perf_session *self);
+
+int perf_session__process_events(struct perf_session *self,
+				 struct perf_event_ops *event_ops);
+
+struct symbol **perf_session__resolve_callchain(struct perf_session *self,
+						struct thread *thread,
+						struct ip_callchain *chain,
+						struct symbol **parent);
+
+int perf_header__read_build_ids(int input, u64 offset, u64 file_size);
+
+#endif /* __PERF_SESSION_H */
diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c
index b490354d1b23..cb0f327de9e8 100644
--- a/tools/perf/util/sort.c
+++ b/tools/perf/util/sort.c
@@ -288,3 +288,29 @@ int sort_dimension__add(const char *tok)
 
 	return -ESRCH;
 }
+
+void setup_sorting(const char * const usagestr[], const struct option *opts)
+{
+	char *tmp, *tok, *str = strdup(sort_order);
+
+	for (tok = strtok_r(str, ", ", &tmp);
+			tok; tok = strtok_r(NULL, ", ", &tmp)) {
+		if (sort_dimension__add(tok) < 0) {
+			error("Unknown --sort key: `%s'", tok);
+			usage_with_options(usagestr, opts);
+		}
+	}
+
+	free(str);
+}
+
+void sort_entry__setup_elide(struct sort_entry *self, struct strlist *list,
+			     const char *list_name, FILE *fp)
+{
+	if (list && strlist__nr_entries(list) == 1) {
+		if (fp != NULL)
+			fprintf(fp, "# %s: %s\n", list_name,
+				strlist__entry(list, 0)->s);
+		self->elide = true;
+	}
+}
diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h
index 333e664ff45f..753f9ea99fb0 100644
--- a/tools/perf/util/sort.h
+++ b/tools/perf/util/sort.h
@@ -49,9 +49,13 @@ struct hist_entry {
 	struct symbol		*sym;
 	u64			ip;
 	char			level;
-	struct symbol		*parent;
+	struct symbol	  *parent;
 	struct callchain_node	callchain;
-	struct rb_root		sorted_chain;
+	union {
+		unsigned long	  position;
+		struct hist_entry *pair;
+		struct rb_root	  sorted_chain;
+	};
 };
 
 enum sort_type {
@@ -81,6 +85,8 @@ struct sort_entry {
 extern struct sort_entry sort_thread;
 extern struct list_head hist_entry__sort_list;
 
+void setup_sorting(const char * const usagestr[], const struct option *opts);
+
 extern int repsep_fprintf(FILE *fp, const char *fmt, ...);
 extern size_t sort__thread_print(FILE *, struct hist_entry *, unsigned int);
 extern size_t sort__comm_print(FILE *, struct hist_entry *, unsigned int);
@@ -95,5 +101,7 @@ extern int64_t sort__sym_cmp(struct hist_entry *, struct hist_entry *);
 extern int64_t sort__parent_cmp(struct hist_entry *, struct hist_entry *);
 extern size_t sort__parent_print(FILE *, struct hist_entry *, unsigned int);
 extern int sort_dimension__add(const char *);
+void sort_entry__setup_elide(struct sort_entry *self, struct strlist *list,
+			     const char *list_name, FILE *fp);
 
 #endif	/* __PERF_SORT_H */
diff --git a/tools/perf/util/string.c b/tools/perf/util/string.c
index f24a8cc933d5..5352d7dccc61 100644
--- a/tools/perf/util/string.c
+++ b/tools/perf/util/string.c
@@ -226,3 +226,28 @@ fail:
 	argv_free(argv);
 	return NULL;
 }
+
+/* Glob expression pattern matching */
+bool strglobmatch(const char *str, const char *pat)
+{
+	while (*str && *pat && *pat != '*') {
+		if (*pat == '?') {
+			str++;
+			pat++;
+		} else
+			if (*str++ != *pat++)
+				return false;
+	}
+	/* Check wild card */
+	if (*pat == '*') {
+		while (*pat == '*')
+			pat++;
+		if (!*pat)	/* Tail wild card matches all */
+			return true;
+		while (*str)
+			if (strglobmatch(str++, pat))
+				return true;
+	}
+	return !*str && !*pat;
+}
+
diff --git a/tools/perf/util/string.h b/tools/perf/util/string.h
index bfecec265a1a..02ede58c54b4 100644
--- a/tools/perf/util/string.h
+++ b/tools/perf/util/string.h
@@ -1,6 +1,7 @@
 #ifndef __PERF_STRING_H_
 #define __PERF_STRING_H_
 
+#include <stdbool.h>
 #include "types.h"
 
 int hex2u64(const char *ptr, u64 *val);
@@ -8,6 +9,7 @@ char *strxfrchar(char *s, char from, char to);
 s64 perf_atoll(const char *str);
 char **argv_split(const char *str, int *argcp);
 void argv_free(char **argv);
+bool strglobmatch(const char *str, const char *pat);
 
 #define _STR(x) #x
 #define STR(x) _STR(x)
diff --git a/tools/perf/util/strlist.c b/tools/perf/util/strlist.c
index 7ad38171dc2b..6783a2043555 100644
--- a/tools/perf/util/strlist.c
+++ b/tools/perf/util/strlist.c
@@ -102,7 +102,7 @@ void strlist__remove(struct strlist *self, struct str_node *sn)
 	str_node__delete(sn, self->dupstr);
 }
 
-bool strlist__has_entry(struct strlist *self, const char *entry)
+struct str_node *strlist__find(struct strlist *self, const char *entry)
 {
 	struct rb_node **p = &self->entries.rb_node;
 	struct rb_node *parent = NULL;
@@ -120,10 +120,10 @@ bool strlist__has_entry(struct strlist *self, const char *entry)
 		else if (rc < 0)
 			p = &(*p)->rb_right;
 		else
-			return true;
+			return sn;
 	}
 
-	return false;
+	return NULL;
 }
 
 static int strlist__parse_list_entry(struct strlist *self, const char *s)
diff --git a/tools/perf/util/strlist.h b/tools/perf/util/strlist.h
index cb4659306d7b..3ba839007d2c 100644
--- a/tools/perf/util/strlist.h
+++ b/tools/perf/util/strlist.h
@@ -23,7 +23,12 @@ int strlist__load(struct strlist *self, const char *filename);
 int strlist__add(struct strlist *self, const char *str);
 
 struct str_node *strlist__entry(const struct strlist *self, unsigned int idx);
-bool strlist__has_entry(struct strlist *self, const char *entry);
+struct str_node *strlist__find(struct strlist *self, const char *entry);
+
+static inline bool strlist__has_entry(struct strlist *self, const char *entry)
+{
+	return strlist__find(self, entry) != NULL;
+}
 
 static inline bool strlist__empty(const struct strlist *self)
 {
@@ -35,5 +40,39 @@ static inline unsigned int strlist__nr_entries(const struct strlist *self)
 	return self->nr_entries;
 }
 
+/* For strlist iteration */
+static inline struct str_node *strlist__first(struct strlist *self)
+{
+	struct rb_node *rn = rb_first(&self->entries);
+	return rn ? rb_entry(rn, struct str_node, rb_node) : NULL;
+}
+static inline struct str_node *strlist__next(struct str_node *sn)
+{
+	struct rb_node *rn;
+	if (!sn)
+		return NULL;
+	rn = rb_next(&sn->rb_node);
+	return rn ? rb_entry(rn, struct str_node, rb_node) : NULL;
+}
+
+/**
+ * strlist_for_each      - iterate over a strlist
+ * @pos:	the &struct str_node to use as a loop cursor.
+ * @self:	the &struct strlist for loop.
+ */
+#define strlist__for_each(pos, self)	\
+	for (pos = strlist__first(self); pos; pos = strlist__next(pos))
+
+/**
+ * strlist_for_each_safe - iterate over a strlist safe against removal of
+ *                         str_node
+ * @pos:	the &struct str_node to use as a loop cursor.
+ * @n:		another &struct str_node to use as temporary storage.
+ * @self:	the &struct strlist for loop.
+ */
+#define strlist__for_each_safe(pos, n, self)	\
+	for (pos = strlist__first(self), n = strlist__next(pos); pos;\
+	     pos = n, n = strlist__next(n))
+
 int strlist__parse_list(struct strlist *self, const char *s);
 #endif /* __PERF_STRLIST_H */
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index fffcb937cdcb..ab92763edb03 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -1,5 +1,7 @@
 #include "util.h"
 #include "../perf.h"
+#include "session.h"
+#include "sort.h"
 #include "string.h"
 #include "symbol.h"
 #include "thread.h"
@@ -29,33 +31,50 @@ enum dso_origin {
 };
 
 static void dsos__add(struct list_head *head, struct dso *dso);
-static struct map *thread__find_map_by_name(struct thread *self, char *name);
 static struct map *map__new2(u64 start, struct dso *dso, enum map_type type);
-struct symbol *dso__find_symbol(struct dso *self, enum map_type type, u64 addr);
 static int dso__load_kernel_sym(struct dso *self, struct map *map,
-				struct thread *thread, symbol_filter_t filter);
-unsigned int symbol__priv_size;
+				struct perf_session *session, symbol_filter_t filter);
 static int vmlinux_path__nr_entries;
 static char **vmlinux_path;
 
-static struct symbol_conf symbol_conf__defaults = {
+struct symbol_conf symbol_conf = {
+	.exclude_other	  = true,
 	.use_modules	  = true,
 	.try_vmlinux_path = true,
 };
 
-static struct thread kthread_mem;
-struct thread *kthread = &kthread_mem;
-
 bool dso__loaded(const struct dso *self, enum map_type type)
 {
 	return self->loaded & (1 << type);
 }
 
+bool dso__sorted_by_name(const struct dso *self, enum map_type type)
+{
+	return self->sorted_by_name & (1 << type);
+}
+
 static void dso__set_loaded(struct dso *self, enum map_type type)
 {
 	self->loaded |= (1 << type);
 }
 
+static void dso__set_sorted_by_name(struct dso *self, enum map_type type)
+{
+	self->sorted_by_name |= (1 << type);
+}
+
+static bool symbol_type__is_a(char symbol_type, enum map_type map_type)
+{
+	switch (map_type) {
+	case MAP__FUNCTION:
+		return symbol_type == 'T' || symbol_type == 'W';
+	case MAP__VARIABLE:
+		return symbol_type == 'D' || symbol_type == 'd';
+	default:
+		return false;
+	}
+}
+
 static void symbols__fixup_end(struct rb_root *self)
 {
 	struct rb_node *nd, *prevnd = rb_first(self);
@@ -79,7 +98,7 @@ static void symbols__fixup_end(struct rb_root *self)
 		curr->end = roundup(curr->start, 4096);
 }
 
-static void __thread__fixup_maps_end(struct thread *self, enum map_type type)
+static void __map_groups__fixup_end(struct map_groups *self, enum map_type type)
 {
 	struct map *prev, *curr;
 	struct rb_node *nd, *prevnd = rb_first(&self->maps[type]);
@@ -102,23 +121,23 @@ static void __thread__fixup_maps_end(struct thread *self, enum map_type type)
 	curr->end = ~0UL;
 }
 
-static void thread__fixup_maps_end(struct thread *self)
+static void map_groups__fixup_end(struct map_groups *self)
 {
 	int i;
 	for (i = 0; i < MAP__NR_TYPES; ++i)
-		__thread__fixup_maps_end(self, i);
+		__map_groups__fixup_end(self, i);
 }
 
 static struct symbol *symbol__new(u64 start, u64 len, const char *name)
 {
 	size_t namelen = strlen(name) + 1;
-	struct symbol *self = zalloc(symbol__priv_size +
+	struct symbol *self = zalloc(symbol_conf.priv_size +
 				     sizeof(*self) + namelen);
 	if (self == NULL)
 		return NULL;
 
-	if (symbol__priv_size)
-		self = ((void *)self) + symbol__priv_size;
+	if (symbol_conf.priv_size)
+		self = ((void *)self) + symbol_conf.priv_size;
 
 	self->start = start;
 	self->end   = len ? start + len - 1 : start;
@@ -132,7 +151,7 @@ static struct symbol *symbol__new(u64 start, u64 len, const char *name)
 
 static void symbol__delete(struct symbol *self)
 {
-	free(((void *)self) - symbol__priv_size);
+	free(((void *)self) - symbol_conf.priv_size);
 }
 
 static size_t symbol__fprintf(struct symbol *self, FILE *fp)
@@ -164,11 +183,11 @@ struct dso *dso__new(const char *name)
 		dso__set_long_name(self, self->name);
 		self->short_name = self->name;
 		for (i = 0; i < MAP__NR_TYPES; ++i)
-			self->symbols[i] = RB_ROOT;
-		self->find_symbol = dso__find_symbol;
+			self->symbols[i] = self->symbol_names[i] = RB_ROOT;
 		self->slen_calculated = 0;
 		self->origin = DSO__ORIG_NOT_FOUND;
 		self->loaded = 0;
+		self->sorted_by_name = 0;
 		self->has_build_id = 0;
 	}
 
@@ -246,11 +265,85 @@ static struct symbol *symbols__find(struct rb_root *self, u64 ip)
 	return NULL;
 }
 
-struct symbol *dso__find_symbol(struct dso *self, enum map_type type, u64 addr)
+struct symbol_name_rb_node {
+	struct rb_node	rb_node;
+	struct symbol	sym;
+};
+
+static void symbols__insert_by_name(struct rb_root *self, struct symbol *sym)
+{
+	struct rb_node **p = &self->rb_node;
+	struct rb_node *parent = NULL;
+	struct symbol_name_rb_node *symn = ((void *)sym) - sizeof(*parent), *s;
+
+	while (*p != NULL) {
+		parent = *p;
+		s = rb_entry(parent, struct symbol_name_rb_node, rb_node);
+		if (strcmp(sym->name, s->sym.name) < 0)
+			p = &(*p)->rb_left;
+		else
+			p = &(*p)->rb_right;
+	}
+	rb_link_node(&symn->rb_node, parent, p);
+	rb_insert_color(&symn->rb_node, self);
+}
+
+static void symbols__sort_by_name(struct rb_root *self, struct rb_root *source)
+{
+	struct rb_node *nd;
+
+	for (nd = rb_first(source); nd; nd = rb_next(nd)) {
+		struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
+		symbols__insert_by_name(self, pos);
+	}
+}
+
+static struct symbol *symbols__find_by_name(struct rb_root *self, const char *name)
+{
+	struct rb_node *n;
+
+	if (self == NULL)
+		return NULL;
+
+	n = self->rb_node;
+
+	while (n) {
+		struct symbol_name_rb_node *s;
+		int cmp;
+
+		s = rb_entry(n, struct symbol_name_rb_node, rb_node);
+		cmp = strcmp(name, s->sym.name);
+
+		if (cmp < 0)
+			n = n->rb_left;
+		else if (cmp > 0)
+			n = n->rb_right;
+		else
+			return &s->sym;
+	}
+
+	return NULL;
+}
+
+struct symbol *dso__find_symbol(struct dso *self,
+				enum map_type type, u64 addr)
 {
 	return symbols__find(&self->symbols[type], addr);
 }
 
+struct symbol *dso__find_symbol_by_name(struct dso *self, enum map_type type,
+					const char *name)
+{
+	return symbols__find_by_name(&self->symbol_names[type], name);
+}
+
+void dso__sort_by_name(struct dso *self, enum map_type type)
+{
+	dso__set_sorted_by_name(self, type);
+	return symbols__sort_by_name(&self->symbol_names[type],
+				     &self->symbols[type]);
+}
+
 int build_id__sprintf(u8 *self, int len, char *bf)
 {
 	char *bid = bf;
@@ -327,10 +420,7 @@ static int dso__load_all_kallsyms(struct dso *self, struct map *map)
 			continue;
 
 		symbol_type = toupper(line[len]);
-		/*
-		 * We're interested only in code ('T'ext)
-		 */
-		if (symbol_type != 'T' && symbol_type != 'W')
+		if (!symbol_type__is_a(symbol_type, map->type))
 			continue;
 
 		symbol_name = line + len + 2;
@@ -364,8 +454,8 @@ out_failure:
  * kernel range is broken in several maps, named [kernel].N, as we don't have
  * the original ELF section names vmlinux have.
  */
-static int dso__split_kallsyms(struct dso *self, struct map *map, struct thread *thread,
-			       symbol_filter_t filter)
+static int dso__split_kallsyms(struct dso *self, struct map *map,
+			       struct perf_session *session, symbol_filter_t filter)
 {
 	struct map *curr_map = map;
 	struct symbol *pos;
@@ -382,13 +472,13 @@ static int dso__split_kallsyms(struct dso *self, struct map *map, struct thread
 
 		module = strchr(pos->name, '\t');
 		if (module) {
-			if (!thread->use_modules)
+			if (!symbol_conf.use_modules)
 				goto discard_symbol;
 
 			*module++ = '\0';
 
 			if (strcmp(self->name, module)) {
-				curr_map = thread__find_map_by_name(thread, module);
+				curr_map = map_groups__find_by_name(&session->kmaps, map->type, module);
 				if (curr_map == NULL) {
 					pr_debug("/proc/{kallsyms,modules} "
 					         "inconsistency!\n");
@@ -419,7 +509,7 @@ static int dso__split_kallsyms(struct dso *self, struct map *map, struct thread
 			}
 
 			curr_map->map_ip = curr_map->unmap_ip = identity__map_ip;
-			__thread__insert_map(thread, curr_map);
+			map_groups__insert(&session->kmaps, curr_map);
 			++kernel_range;
 		}
 
@@ -440,7 +530,7 @@ discard_symbol:		rb_erase(&pos->rb_node, root);
 
 
 static int dso__load_kallsyms(struct dso *self, struct map *map,
-			      struct thread *thread, symbol_filter_t filter)
+			      struct perf_session *session, symbol_filter_t filter)
 {
 	if (dso__load_all_kallsyms(self, map) < 0)
 		return -1;
@@ -448,14 +538,7 @@ static int dso__load_kallsyms(struct dso *self, struct map *map,
 	symbols__fixup_end(&self->symbols[map->type]);
 	self->origin = DSO__ORIG_KERNEL;
 
-	return dso__split_kallsyms(self, map, thread, filter);
-}
-
-size_t kernel_maps__fprintf(FILE *fp)
-{
-	size_t printed = fprintf(fp, "Kernel maps:\n");
-	printed += thread__fprintf_maps(kthread, fp);
-	return printed + fprintf(fp, "END kernel maps\n");
+	return dso__split_kallsyms(self, map, session, filter);
 }
 
 static int dso__load_perf_map(struct dso *self, struct map *map,
@@ -544,6 +627,13 @@ static inline int elf_sym__is_function(const GElf_Sym *sym)
 	       sym->st_shndx != SHN_UNDEF;
 }
 
+static inline bool elf_sym__is_object(const GElf_Sym *sym)
+{
+	return elf_sym__type(sym) == STT_OBJECT &&
+		sym->st_name != 0 &&
+		sym->st_shndx != SHN_UNDEF;
+}
+
 static inline int elf_sym__is_label(const GElf_Sym *sym)
 {
 	return elf_sym__type(sym) == STT_NOTYPE &&
@@ -564,6 +654,12 @@ static inline int elf_sec__is_text(const GElf_Shdr *shdr,
 	return strstr(elf_sec__name(shdr, secstrs), "text") != NULL;
 }
 
+static inline bool elf_sec__is_data(const GElf_Shdr *shdr,
+				    const Elf_Data *secstrs)
+{
+	return strstr(elf_sec__name(shdr, secstrs), "data") != NULL;
+}
+
 static inline const char *elf_sym__name(const GElf_Sym *sym,
 					const Elf_Data *symstrs)
 {
@@ -744,8 +840,32 @@ out:
 	return 0;
 }
 
+static bool elf_sym__is_a(GElf_Sym *self, enum map_type type)
+{
+	switch (type) {
+	case MAP__FUNCTION:
+		return elf_sym__is_function(self);
+	case MAP__VARIABLE:
+		return elf_sym__is_object(self);
+	default:
+		return false;
+	}
+}
+
+static bool elf_sec__is_a(GElf_Shdr *self, Elf_Data *secstrs, enum map_type type)
+{
+	switch (type) {
+	case MAP__FUNCTION:
+		return elf_sec__is_text(self, secstrs);
+	case MAP__VARIABLE:
+		return elf_sec__is_data(self, secstrs);
+	default:
+		return false;
+	}
+}
+
 static int dso__load_sym(struct dso *self, struct map *map,
-			 struct thread *thread, const char *name, int fd,
+			 struct perf_session *session, const char *name, int fd,
 			 symbol_filter_t filter, int kernel, int kmodule)
 {
 	struct map *curr_map = map;
@@ -818,7 +938,7 @@ static int dso__load_sym(struct dso *self, struct map *map,
 		int is_label = elf_sym__is_label(&sym);
 		const char *section_name;
 
-		if (!is_label && !elf_sym__is_function(&sym))
+		if (!is_label && !elf_sym__is_a(&sym, map->type))
 			continue;
 
 		sec = elf_getscn(elf, sym.st_shndx);
@@ -827,7 +947,7 @@ static int dso__load_sym(struct dso *self, struct map *map,
 
 		gelf_getshdr(sec, &shdr);
 
-		if (is_label && !elf_sec__is_text(&shdr, secstrs))
+		if (is_label && !elf_sec__is_a(&shdr, secstrs, map->type))
 			continue;
 
 		elf_name = elf_sym__name(&sym, symstrs);
@@ -849,7 +969,7 @@ static int dso__load_sym(struct dso *self, struct map *map,
 			snprintf(dso_name, sizeof(dso_name),
 				 "%s%s", self->short_name, section_name);
 
-			curr_map = thread__find_map_by_name(thread, dso_name);
+			curr_map = map_groups__find_by_name(&session->kmaps, map->type, dso_name);
 			if (curr_map == NULL) {
 				u64 start = sym.st_value;
 
@@ -868,7 +988,7 @@ static int dso__load_sym(struct dso *self, struct map *map,
 				curr_map->map_ip = identity__map_ip;
 				curr_map->unmap_ip = identity__map_ip;
 				curr_dso->origin = DSO__ORIG_KERNEL;
-				__thread__insert_map(kthread, curr_map);
+				map_groups__insert(&session->kmaps, curr_map);
 				dsos__add(&dsos__kernel, curr_dso);
 			} else
 				curr_dso = curr_map->dso;
@@ -938,8 +1058,9 @@ static bool __dsos__read_build_ids(struct list_head *head)
 
 bool dsos__read_build_ids(void)
 {
-	return __dsos__read_build_ids(&dsos__kernel) ||
-	       __dsos__read_build_ids(&dsos__user);
+	bool kbuildids = __dsos__read_build_ids(&dsos__kernel),
+	     ubuildids = __dsos__read_build_ids(&dsos__user);
+	return kbuildids || ubuildids;
 }
 
 /*
@@ -1082,7 +1203,8 @@ char dso__symtab_origin(const struct dso *self)
 	return origin[self->origin];
 }
 
-int dso__load(struct dso *self, struct map *map, symbol_filter_t filter)
+int dso__load(struct dso *self, struct map *map, struct perf_session *session,
+	      symbol_filter_t filter)
 {
 	int size = PATH_MAX;
 	char *name;
@@ -1093,7 +1215,7 @@ int dso__load(struct dso *self, struct map *map, symbol_filter_t filter)
 	dso__set_loaded(self, map->type);
 
 	if (self->kernel)
-		return dso__load_kernel_sym(self, map, kthread, filter);
+		return dso__load_kernel_sym(self, map, session, filter);
 
 	name = malloc(size);
 	if (!name)
@@ -1179,11 +1301,12 @@ out:
 	return ret;
 }
 
-static struct map *thread__find_map_by_name(struct thread *self, char *name)
+struct map *map_groups__find_by_name(struct map_groups *self,
+				     enum map_type type, const char *name)
 {
 	struct rb_node *nd;
 
-	for (nd = rb_first(&self->maps[MAP__FUNCTION]); nd; nd = rb_next(nd)) {
+	for (nd = rb_first(&self->maps[type]); nd; nd = rb_next(nd)) {
 		struct map *map = rb_entry(nd, struct map, rb_node);
 
 		if (map->dso && strcmp(map->dso->name, name) == 0)
@@ -1193,7 +1316,7 @@ static struct map *thread__find_map_by_name(struct thread *self, char *name)
 	return NULL;
 }
 
-static int dsos__set_modules_path_dir(char *dirname)
+static int perf_session__set_modules_path_dir(struct perf_session *self, char *dirname)
 {
 	struct dirent *dent;
 	DIR *dir = opendir(dirname);
@@ -1213,7 +1336,7 @@ static int dsos__set_modules_path_dir(char *dirname)
 
 			snprintf(path, sizeof(path), "%s/%s",
 				 dirname, dent->d_name);
-			if (dsos__set_modules_path_dir(path) < 0)
+			if (perf_session__set_modules_path_dir(self, path) < 0)
 				goto failure;
 		} else {
 			char *dot = strrchr(dent->d_name, '.'),
@@ -1227,7 +1350,7 @@ static int dsos__set_modules_path_dir(char *dirname)
 				 (int)(dot - dent->d_name), dent->d_name);
 
 			strxfrchar(dso_name, '-', '_');
-			map = thread__find_map_by_name(kthread, dso_name);
+			map = map_groups__find_by_name(&self->kmaps, MAP__FUNCTION, dso_name);
 			if (map == NULL)
 				continue;
 
@@ -1247,7 +1370,7 @@ failure:
 	return -1;
 }
 
-static int dsos__set_modules_path(void)
+static int perf_session__set_modules_path(struct perf_session *self)
 {
 	struct utsname uts;
 	char modules_path[PATH_MAX];
@@ -1258,7 +1381,7 @@ static int dsos__set_modules_path(void)
 	snprintf(modules_path, sizeof(modules_path), "/lib/modules/%s/kernel",
 		 uts.release);
 
-	return dsos__set_modules_path_dir(modules_path);
+	return perf_session__set_modules_path_dir(self, modules_path);
 }
 
 /*
@@ -1280,7 +1403,7 @@ static struct map *map__new2(u64 start, struct dso *dso, enum map_type type)
 	return self;
 }
 
-static int thread__create_module_maps(struct thread *self)
+static int perf_session__create_module_maps(struct perf_session *self)
 {
 	char *line = NULL;
 	size_t n;
@@ -1337,14 +1460,14 @@ static int thread__create_module_maps(struct thread *self)
 			dso->has_build_id = true;
 
 		dso->origin = DSO__ORIG_KMODULE;
-		__thread__insert_map(self, map);
+		map_groups__insert(&self->kmaps, map);
 		dsos__add(&dsos__kernel, dso);
 	}
 
 	free(line);
 	fclose(file);
 
-	return dsos__set_modules_path();
+	return perf_session__set_modules_path(self);
 
 out_delete_line:
 	free(line);
@@ -1352,7 +1475,8 @@ out_failure:
 	return -1;
 }
 
-static int dso__load_vmlinux(struct dso *self, struct map *map, struct thread *thread,
+static int dso__load_vmlinux(struct dso *self, struct map *map,
+			     struct perf_session *session,
 			     const char *vmlinux, symbol_filter_t filter)
 {
 	int err = -1, fd;
@@ -1386,14 +1510,14 @@ static int dso__load_vmlinux(struct dso *self, struct map *map, struct thread *t
 		return -1;
 
 	dso__set_loaded(self, map->type);
-	err = dso__load_sym(self, map, thread, self->long_name, fd, filter, 1, 0);
+	err = dso__load_sym(self, map, session, self->long_name, fd, filter, 1, 0);
 	close(fd);
 
 	return err;
 }
 
 static int dso__load_kernel_sym(struct dso *self, struct map *map,
-				struct thread *thread, symbol_filter_t filter)
+				struct perf_session *session, symbol_filter_t filter)
 {
 	int err;
 	bool is_kallsyms;
@@ -1403,7 +1527,7 @@ static int dso__load_kernel_sym(struct dso *self, struct map *map,
 		pr_debug("Looking at the vmlinux_path (%d entries long)\n",
 			 vmlinux_path__nr_entries);
 		for (i = 0; i < vmlinux_path__nr_entries; ++i) {
-			err = dso__load_vmlinux(self, map, thread,
+			err = dso__load_vmlinux(self, map, session,
 						vmlinux_path[i], filter);
 			if (err > 0) {
 				pr_debug("Using %s for symbols\n",
@@ -1419,12 +1543,12 @@ static int dso__load_kernel_sym(struct dso *self, struct map *map,
 	if (is_kallsyms)
 		goto do_kallsyms;
 
-	err = dso__load_vmlinux(self, map, thread, self->long_name, filter);
+	err = dso__load_vmlinux(self, map, session, self->long_name, filter);
 	if (err <= 0) {
 		pr_info("The file %s cannot be used, "
 			"trying to use /proc/kallsyms...", self->long_name);
 do_kallsyms:
-		err = dso__load_kallsyms(self, map, thread, filter);
+		err = dso__load_kallsyms(self, map, session, filter);
 		if (err > 0 && !is_kallsyms)
                         dso__set_long_name(self, strdup("[kernel.kallsyms]"));
 	}
@@ -1507,42 +1631,59 @@ size_t dsos__fprintf_buildid(FILE *fp)
 		__dsos__fprintf_buildid(&dsos__user, fp));
 }
 
-static int thread__create_kernel_map(struct thread *self, const char *vmlinux)
+static struct dso *dsos__create_kernel( const char *vmlinux)
 {
-	struct map *kmap;
 	struct dso *kernel = dso__new(vmlinux ?: "[kernel.kallsyms]");
 
 	if (kernel == NULL)
-		return -1;
-
-	kmap = map__new2(0, kernel, MAP__FUNCTION);
-	if (kmap == NULL)
-		goto out_delete_kernel_dso;
+		return NULL;
 
-	kmap->map_ip	   = kmap->unmap_ip = identity__map_ip;
 	kernel->short_name = "[kernel]";
 	kernel->kernel	   = 1;
 
 	vdso = dso__new("[vdso]");
 	if (vdso == NULL)
-		goto out_delete_kernel_map;
+		goto out_delete_kernel_dso;
 	dso__set_loaded(vdso, MAP__FUNCTION);
 
 	if (sysfs__read_build_id("/sys/kernel/notes", kernel->build_id,
 				 sizeof(kernel->build_id)) == 0)
 		kernel->has_build_id = true;
 
-	__thread__insert_map(self, kmap);
 	dsos__add(&dsos__kernel, kernel);
 	dsos__add(&dsos__user, vdso);
 
-	return 0;
+	return kernel;
 
-out_delete_kernel_map:
-	map__delete(kmap);
 out_delete_kernel_dso:
 	dso__delete(kernel);
-	return -1;
+	return NULL;
+}
+
+static int map_groups__create_kernel_maps(struct map_groups *self, const char *vmlinux)
+{
+	struct map *functions, *variables;
+	struct dso *kernel = dsos__create_kernel(vmlinux);
+
+	if (kernel == NULL)
+		return -1;
+
+	functions = map__new2(0, kernel, MAP__FUNCTION);
+	if (functions == NULL)
+		return -1;
+
+	variables = map__new2(0, kernel, MAP__VARIABLE);
+	if (variables == NULL) {
+		map__delete(functions);
+		return -1;
+	}
+
+	functions->map_ip = functions->unmap_ip =
+		variables->map_ip = variables->unmap_ip = identity__map_ip;
+	map_groups__insert(self, functions);
+	map_groups__insert(self, variables);
+
+	return 0;
 }
 
 static void vmlinux_path__exit(void)
@@ -1600,29 +1741,69 @@ out_fail:
 	return -1;
 }
 
-int symbol__init(struct symbol_conf *conf)
+static int setup_list(struct strlist **list, const char *list_str,
+		      const char *list_name)
 {
-	const struct symbol_conf *pconf = conf ?: &symbol_conf__defaults;
+	if (list_str == NULL)
+		return 0;
 
+	*list = strlist__new(true, list_str);
+	if (!*list) {
+		pr_err("problems parsing %s list\n", list_name);
+		return -1;
+	}
+	return 0;
+}
+
+int symbol__init(void)
+{
 	elf_version(EV_CURRENT);
-	symbol__priv_size = pconf->priv_size;
-	thread__init(kthread, 0);
+	if (symbol_conf.sort_by_name)
+		symbol_conf.priv_size += (sizeof(struct symbol_name_rb_node) -
+					  sizeof(struct symbol));
 
-	if (pconf->try_vmlinux_path && vmlinux_path__init() < 0)
+	if (symbol_conf.try_vmlinux_path && vmlinux_path__init() < 0)
 		return -1;
 
-	if (thread__create_kernel_map(kthread, pconf->vmlinux_name) < 0) {
-		vmlinux_path__exit();
+	if (symbol_conf.field_sep && *symbol_conf.field_sep == '.') {
+		pr_err("'.' is the only non valid --field-separator argument\n");
 		return -1;
 	}
 
-	kthread->use_modules = pconf->use_modules;
-	if (pconf->use_modules && thread__create_module_maps(kthread) < 0)
-		pr_debug("Failed to load list of modules in use, "
-			 "continuing...\n");
+	if (setup_list(&symbol_conf.dso_list,
+		       symbol_conf.dso_list_str, "dso") < 0)
+		return -1;
+
+	if (setup_list(&symbol_conf.comm_list,
+		       symbol_conf.comm_list_str, "comm") < 0)
+		goto out_free_dso_list;
+
+	if (setup_list(&symbol_conf.sym_list,
+		       symbol_conf.sym_list_str, "symbol") < 0)
+		goto out_free_comm_list;
+
+	return 0;
+
+out_free_dso_list:
+	strlist__delete(symbol_conf.dso_list);
+out_free_comm_list:
+	strlist__delete(symbol_conf.comm_list);
+	return -1;
+}
+
+int perf_session__create_kernel_maps(struct perf_session *self)
+{
+	if (map_groups__create_kernel_maps(&self->kmaps,
+					   symbol_conf.vmlinux_name) < 0)
+		return -1;
+
+	if (symbol_conf.use_modules &&
+	    perf_session__create_module_maps(self) < 0)
+		pr_debug("Failed to load list of modules for session %s, "
+			 "continuing...\n", self->filename);
 	/*
 	 * Now that we have all the maps created, just set the ->end of them:
 	 */
-	thread__fixup_maps_end(kthread);
+	map_groups__fixup_end(&self->kmaps);
 	return 0;
 }
diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h
index 17003efa0b39..8aded2356f79 100644
--- a/tools/perf/util/symbol.h
+++ b/tools/perf/util/symbol.h
@@ -49,18 +49,32 @@ struct symbol {
 	char		name[0];
 };
 
+struct strlist;
+
 struct symbol_conf {
 	unsigned short	priv_size;
 	bool		try_vmlinux_path,
-			use_modules;
-	const char	*vmlinux_name;
+			use_modules,
+			sort_by_name,
+			show_nr_samples,
+			use_callchain,
+			exclude_other;
+	const char	*vmlinux_name,
+			*field_sep;
+	char            *dso_list_str,
+			*comm_list_str,
+			*sym_list_str,
+			*col_width_list_str;
+       struct strlist	*dso_list,
+			*comm_list,
+			*sym_list;
 };
 
-extern unsigned int symbol__priv_size;
+extern struct symbol_conf symbol_conf;
 
 static inline void *symbol__priv(struct symbol *self)
 {
-	return ((void *)self) - symbol__priv_size;
+	return ((void *)self) - symbol_conf.priv_size;
 }
 
 struct addr_location {
@@ -69,18 +83,19 @@ struct addr_location {
 	struct symbol *sym;
 	u64	      addr;
 	char	      level;
+	bool	      filtered;
 };
 
 struct dso {
 	struct list_head node;
 	struct rb_root	 symbols[MAP__NR_TYPES];
-	struct symbol    *(*find_symbol)(struct dso *self,
-					 enum map_type type, u64 addr);
+	struct rb_root	 symbol_names[MAP__NR_TYPES];
 	u8		 adjust_symbols:1;
 	u8		 slen_calculated:1;
 	u8		 has_build_id:1;
 	u8		 kernel:1;
 	unsigned char	 origin;
+	u8		 sorted_by_name;
 	u8		 loaded;
 	u8		 build_id[BUILD_ID_SIZE];
 	u16		 long_name_len;
@@ -93,9 +108,15 @@ struct dso *dso__new(const char *name);
 void dso__delete(struct dso *self);
 
 bool dso__loaded(const struct dso *self, enum map_type type);
+bool dso__sorted_by_name(const struct dso *self, enum map_type type);
+
+void dso__sort_by_name(struct dso *self, enum map_type type);
+
+struct perf_session;
 
 struct dso *dsos__findnew(const char *name);
-int dso__load(struct dso *self, struct map *map, symbol_filter_t filter);
+int dso__load(struct dso *self, struct map *map, struct perf_session *session,
+	      symbol_filter_t filter);
 void dsos__fprintf(FILE *fp);
 size_t dsos__fprintf_buildid(FILE *fp);
 
@@ -103,18 +124,18 @@ size_t dso__fprintf_buildid(struct dso *self, FILE *fp);
 size_t dso__fprintf(struct dso *self, enum map_type type, FILE *fp);
 char dso__symtab_origin(const struct dso *self);
 void dso__set_build_id(struct dso *self, void *build_id);
+struct symbol *dso__find_symbol(struct dso *self, enum map_type type, u64 addr);
+struct symbol *dso__find_symbol_by_name(struct dso *self, enum map_type type,
+					const char *name);
 
 int filename__read_build_id(const char *filename, void *bf, size_t size);
 int sysfs__read_build_id(const char *filename, void *bf, size_t size);
 bool dsos__read_build_ids(void);
 int build_id__sprintf(u8 *self, int len, char *bf);
 
-size_t kernel_maps__fprintf(FILE *fp);
-
-int symbol__init(struct symbol_conf *conf);
+int symbol__init(void);
+int perf_session__create_kernel_maps(struct perf_session *self);
 
-struct thread;
-struct thread *kthread;
 extern struct list_head dsos__user, dsos__kernel;
 extern struct dso *vdso;
 #endif /* __PERF_SYMBOL */
diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c
index 603f5610861b..4a08dcf50b68 100644
--- a/tools/perf/util/thread.c
+++ b/tools/perf/util/thread.c
@@ -2,18 +2,14 @@
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
+#include "session.h"
 #include "thread.h"
 #include "util.h"
 #include "debug.h"
 
-static struct rb_root threads;
-static struct thread *last_match;
-
-void thread__init(struct thread *self, pid_t pid)
+void map_groups__init(struct map_groups *self)
 {
 	int i;
-	self->pid = pid;
-	self->comm = NULL;
 	for (i = 0; i < MAP__NR_TYPES; ++i) {
 		self->maps[i] = RB_ROOT;
 		INIT_LIST_HEAD(&self->removed_maps[i]);
@@ -25,7 +21,8 @@ static struct thread *thread__new(pid_t pid)
 	struct thread *self = zalloc(sizeof(*self));
 
 	if (self != NULL) {
-		thread__init(self, pid);
+		map_groups__init(&self->mg);
+		self->pid = pid;
 		self->comm = malloc(32);
 		if (self->comm)
 			snprintf(self->comm, 32, ":%d", self->pid);
@@ -55,10 +52,11 @@ int thread__comm_len(struct thread *self)
 
 static const char *map_type__name[MAP__NR_TYPES] = {
 	[MAP__FUNCTION] = "Functions",
+	[MAP__VARIABLE] = "Variables",
 };
 
-static size_t __thread__fprintf_maps(struct thread *self,
-				     enum map_type type, FILE *fp)
+static size_t __map_groups__fprintf_maps(struct map_groups *self,
+					 enum map_type type, FILE *fp)
 {
 	size_t printed = fprintf(fp, "%s:\n", map_type__name[type]);
 	struct rb_node *nd;
@@ -76,16 +74,16 @@ static size_t __thread__fprintf_maps(struct thread *self,
 	return printed;
 }
 
-size_t thread__fprintf_maps(struct thread *self, FILE *fp)
+size_t map_groups__fprintf_maps(struct map_groups *self, FILE *fp)
 {
 	size_t printed = 0, i;
 	for (i = 0; i < MAP__NR_TYPES; ++i)
-		printed += __thread__fprintf_maps(self, i, fp);
+		printed += __map_groups__fprintf_maps(self, i, fp);
 	return printed;
 }
 
-static size_t __thread__fprintf_removed_maps(struct thread *self,
-					     enum map_type type, FILE *fp)
+static size_t __map_groups__fprintf_removed_maps(struct map_groups *self,
+						 enum map_type type, FILE *fp)
 {
 	struct map *pos;
 	size_t printed = 0;
@@ -101,25 +99,30 @@ static size_t __thread__fprintf_removed_maps(struct thread *self,
 	return printed;
 }
 
-static size_t thread__fprintf_removed_maps(struct thread *self, FILE *fp)
+static size_t map_groups__fprintf_removed_maps(struct map_groups *self, FILE *fp)
 {
 	size_t printed = 0, i;
 	for (i = 0; i < MAP__NR_TYPES; ++i)
-		printed += __thread__fprintf_removed_maps(self, i, fp);
+		printed += __map_groups__fprintf_removed_maps(self, i, fp);
 	return printed;
 }
 
-static size_t thread__fprintf(struct thread *self, FILE *fp)
+static size_t map_groups__fprintf(struct map_groups *self, FILE *fp)
 {
-	size_t printed = fprintf(fp, "Thread %d %s\n", self->pid, self->comm);
-	printed += thread__fprintf_removed_maps(self, fp);
+	size_t printed = map_groups__fprintf_maps(self, fp);
 	printed += fprintf(fp, "Removed maps:\n");
-	return printed + thread__fprintf_removed_maps(self, fp);
+	return printed + map_groups__fprintf_removed_maps(self, fp);
+}
+
+static size_t thread__fprintf(struct thread *self, FILE *fp)
+{
+	return fprintf(fp, "Thread %d %s\n", self->pid, self->comm) +
+	       map_groups__fprintf(&self->mg, fp);
 }
 
-struct thread *threads__findnew(pid_t pid)
+struct thread *perf_session__findnew(struct perf_session *self, pid_t pid)
 {
-	struct rb_node **p = &threads.rb_node;
+	struct rb_node **p = &self->threads.rb_node;
 	struct rb_node *parent = NULL;
 	struct thread *th;
 
@@ -128,15 +131,15 @@ struct thread *threads__findnew(pid_t pid)
 	 * so most of the time we dont have to look up
 	 * the full rbtree:
 	 */
-	if (last_match && last_match->pid == pid)
-		return last_match;
+	if (self->last_match && self->last_match->pid == pid)
+		return self->last_match;
 
 	while (*p != NULL) {
 		parent = *p;
 		th = rb_entry(parent, struct thread, rb_node);
 
 		if (th->pid == pid) {
-			last_match = th;
+			self->last_match = th;
 			return th;
 		}
 
@@ -149,26 +152,15 @@ struct thread *threads__findnew(pid_t pid)
 	th = thread__new(pid);
 	if (th != NULL) {
 		rb_link_node(&th->rb_node, parent, p);
-		rb_insert_color(&th->rb_node, &threads);
-		last_match = th;
+		rb_insert_color(&th->rb_node, &self->threads);
+		self->last_match = th;
 	}
 
 	return th;
 }
 
-struct thread *register_idle_thread(void)
-{
-	struct thread *thread = threads__findnew(0);
-
-	if (!thread || thread__set_comm(thread, "swapper")) {
-		fprintf(stderr, "problem inserting idle task.\n");
-		exit(-1);
-	}
-
-	return thread;
-}
-
-static void thread__remove_overlappings(struct thread *self, struct map *map)
+static void map_groups__remove_overlappings(struct map_groups *self,
+					    struct map *map)
 {
 	struct rb_root *root = &self->maps[map->type];
 	struct rb_node *next = rb_first(root);
@@ -238,12 +230,15 @@ struct map *maps__find(struct rb_root *maps, u64 ip)
 
 void thread__insert_map(struct thread *self, struct map *map)
 {
-	thread__remove_overlappings(self, map);
-	maps__insert(&self->maps[map->type], map);
+	map_groups__remove_overlappings(&self->mg, map);
+	map_groups__insert(&self->mg, map);
 }
 
-static int thread__clone_maps(struct thread *self, struct thread *parent,
-			      enum map_type type)
+/*
+ * XXX This should not really _copy_ te maps, but refcount them.
+ */
+static int map_groups__clone(struct map_groups *self,
+			     struct map_groups *parent, enum map_type type)
 {
 	struct rb_node *nd;
 	for (nd = rb_first(&parent->maps[type]); nd; nd = rb_next(nd)) {
@@ -251,7 +246,7 @@ static int thread__clone_maps(struct thread *self, struct thread *parent,
 		struct map *new = map__clone(map);
 		if (new == NULL)
 			return -ENOMEM;
-		thread__insert_map(self, new);
+		map_groups__insert(self, new);
 	}
 	return 0;
 }
@@ -267,17 +262,17 @@ int thread__fork(struct thread *self, struct thread *parent)
 		return -ENOMEM;
 
 	for (i = 0; i < MAP__NR_TYPES; ++i)
-		if (thread__clone_maps(self, parent, i) < 0)
+		if (map_groups__clone(&self->mg, &parent->mg, i) < 0)
 			return -ENOMEM;
 	return 0;
 }
 
-size_t threads__fprintf(FILE *fp)
+size_t perf_session__fprintf(struct perf_session *self, FILE *fp)
 {
 	size_t ret = 0;
 	struct rb_node *nd;
 
-	for (nd = rb_first(&threads); nd; nd = rb_next(nd)) {
+	for (nd = rb_first(&self->threads); nd; nd = rb_next(nd)) {
 		struct thread *pos = rb_entry(nd, struct thread, rb_node);
 
 		ret += thread__fprintf(pos, fp);
@@ -286,14 +281,15 @@ size_t threads__fprintf(FILE *fp)
 	return ret;
 }
 
-struct symbol *thread__find_symbol(struct thread *self,
-				   enum map_type type, u64 addr,
-				   symbol_filter_t filter)
+struct symbol *map_groups__find_symbol(struct map_groups *self,
+				       struct perf_session *session,
+				       enum map_type type, u64 addr,
+				       symbol_filter_t filter)
 {
-	struct map *map = thread__find_map(self, type, addr);
+	struct map *map = map_groups__find(self, type, addr);
 
 	if (map != NULL)
-		return map__find_symbol(map, map->map_ip(map, addr), filter);
+		return map__find_symbol(map, session, map->map_ip(map, addr), filter);
 
 	return NULL;
 }
diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h
index 686d6e914d9e..c206f72c8881 100644
--- a/tools/perf/util/thread.h
+++ b/tools/perf/util/thread.h
@@ -5,52 +5,66 @@
 #include <unistd.h>
 #include "symbol.h"
 
-struct thread {
-	struct rb_node		rb_node;
+struct map_groups {
 	struct rb_root		maps[MAP__NR_TYPES];
 	struct list_head	removed_maps[MAP__NR_TYPES];
+};
+
+struct thread {
+	struct rb_node		rb_node;
+	struct map_groups	mg;
 	pid_t			pid;
-	bool			use_modules;
 	char			shortname[3];
 	char			*comm;
 	int			comm_len;
 };
 
-void thread__init(struct thread *self, pid_t pid);
+void map_groups__init(struct map_groups *self);
 int thread__set_comm(struct thread *self, const char *comm);
 int thread__comm_len(struct thread *self);
-struct thread *threads__findnew(pid_t pid);
-struct thread *register_idle_thread(void);
+struct thread *perf_session__findnew(struct perf_session *self, pid_t pid);
 void thread__insert_map(struct thread *self, struct map *map);
 int thread__fork(struct thread *self, struct thread *parent);
-size_t thread__fprintf_maps(struct thread *self, FILE *fp);
-size_t threads__fprintf(FILE *fp);
+size_t map_groups__fprintf_maps(struct map_groups *self, FILE *fp);
+size_t perf_session__fprintf(struct perf_session *self, FILE *fp);
 
 void maps__insert(struct rb_root *maps, struct map *map);
 struct map *maps__find(struct rb_root *maps, u64 addr);
 
-static inline struct map *thread__find_map(struct thread *self,
+static inline void map_groups__insert(struct map_groups *self, struct map *map)
+{
+	 maps__insert(&self->maps[map->type], map);
+}
+
+static inline struct map *map_groups__find(struct map_groups *self,
 					   enum map_type type, u64 addr)
 {
-	return self ? maps__find(&self->maps[type], addr) : NULL;
+	return maps__find(&self->maps[type], addr);
 }
 
-static inline void __thread__insert_map(struct thread *self, struct map *map)
+static inline struct map *thread__find_map(struct thread *self,
+					   enum map_type type, u64 addr)
 {
-	 maps__insert(&self->maps[map->type], map);
+	return self ? map_groups__find(&self->mg, type, addr) : NULL;
 }
 
-void thread__find_addr_location(struct thread *self, u8 cpumode,
+void thread__find_addr_location(struct thread *self,
+				struct perf_session *session, u8 cpumode,
 				enum map_type type, u64 addr,
 				struct addr_location *al,
 				symbol_filter_t filter);
-struct symbol *thread__find_symbol(struct thread *self,
-				   enum map_type type, u64 addr,
-				   symbol_filter_t filter);
+struct symbol *map_groups__find_symbol(struct map_groups *self,
+				       struct perf_session *session,
+				       enum map_type type, u64 addr,
+				       symbol_filter_t filter);
 
 static inline struct symbol *
-thread__find_function(struct thread *self, u64 addr, symbol_filter_t filter)
+map_groups__find_function(struct map_groups *self, struct perf_session *session,
+			  u64 addr, symbol_filter_t filter)
 {
-	return thread__find_symbol(self, MAP__FUNCTION, addr, filter);
+	return map_groups__find_symbol(self, session, MAP__FUNCTION, addr, filter);
 }
+
+struct map *map_groups__find_by_name(struct map_groups *self,
+				     enum map_type type, const char *name);
 #endif	/* __PERF_THREAD_H */
diff --git a/tools/perf/util/trace-event-parse.c b/tools/perf/util/trace-event-parse.c
index 0302405aa2ca..c5c32be040bf 100644
--- a/tools/perf/util/trace-event-parse.c
+++ b/tools/perf/util/trace-event-parse.c
@@ -177,7 +177,7 @@ void parse_proc_kallsyms(char *file, unsigned int size __unused)
 		func_count++;
 	}
 
-	func_list = malloc_or_die(sizeof(*func_list) * func_count + 1);
+	func_list = malloc_or_die(sizeof(*func_list) * (func_count + 1));
 
 	i = 0;
 	while (list) {
@@ -1477,7 +1477,7 @@ process_fields(struct event *event, struct print_flag_sym **list, char **tok)
 			goto out_free;
 
 		field = malloc_or_die(sizeof(*field));
-		memset(field, 0, sizeof(field));
+		memset(field, 0, sizeof(*field));
 
 		value = arg_eval(arg);
 		field->value = strdup(value);
diff --git a/tools/perf/util/trace-event-perl.c b/tools/perf/util/trace-event-perl.c
index 51e833fd58c3..6d6d76b8a21e 100644
--- a/tools/perf/util/trace-event-perl.c
+++ b/tools/perf/util/trace-event-perl.c
@@ -32,9 +32,6 @@
 
 void xs_init(pTHX);
 
-void boot_Perf__Trace__Context(pTHX_ CV *cv);
-void boot_DynaLoader(pTHX_ CV *cv);
-
 void xs_init(pTHX)
 {
 	const char *file = __FILE__;
@@ -270,7 +267,7 @@ int common_lock_depth(struct scripting_context *context)
 }
 
 static void perl_process_event(int cpu, void *data,
-			       int size __attribute((unused)),
+			       int size __unused,
 			       unsigned long long nsecs, char *comm)
 {
 	struct format_field *field;
@@ -362,28 +359,46 @@ static void run_start_sub(void)
 /*
  * Start trace script
  */
-static int perl_start_script(const char *script)
+static int perl_start_script(const char *script, int argc, const char **argv)
 {
-	const char *command_line[2] = { "", NULL };
+	const char **command_line;
+	int i, err = 0;
 
+	command_line = malloc((argc + 2) * sizeof(const char *));
+	command_line[0] = "";
 	command_line[1] = script;
+	for (i = 2; i < argc + 2; i++)
+		command_line[i] = argv[i - 2];
 
 	my_perl = perl_alloc();
 	perl_construct(my_perl);
 
-	if (perl_parse(my_perl, xs_init, 2, (char **)command_line,
-		       (char **)NULL))
-		return -1;
+	if (perl_parse(my_perl, xs_init, argc + 2, (char **)command_line,
+		       (char **)NULL)) {
+		err = -1;
+		goto error;
+	}
 
-	perl_run(my_perl);
-	if (SvTRUE(ERRSV))
-		return -1;
+	if (perl_run(my_perl)) {
+		err = -1;
+		goto error;
+	}
+
+	if (SvTRUE(ERRSV)) {
+		err = -1;
+		goto error;
+	}
 
 	run_start_sub();
 
+	free(command_line);
 	fprintf(stderr, "perf trace started with Perl script %s\n\n", script);
-
 	return 0;
+error:
+	perl_free(my_perl);
+	free(command_line);
+
+	return err;
 }
 
 /*
@@ -573,26 +588,74 @@ struct scripting_ops perl_scripting_ops = {
 	.generate_script = perl_generate_script,
 };
 
-#ifdef NO_LIBPERL
-void setup_perl_scripting(void)
+static void print_unsupported_msg(void)
 {
 	fprintf(stderr, "Perl scripting not supported."
-		"  Install libperl and rebuild perf to enable it.  e.g. "
-		"apt-get install libperl-dev (ubuntu), yum install "
-		"perl-ExtUtils-Embed (Fedora), etc.\n");
+		"  Install libperl and rebuild perf to enable it.\n"
+		"For example:\n  # apt-get install libperl-dev (ubuntu)"
+		"\n  # yum install perl-ExtUtils-Embed (Fedora)"
+		"\n  etc.\n");
 }
-#else
-void setup_perl_scripting(void)
+
+static int perl_start_script_unsupported(const char *script __unused,
+					 int argc __unused,
+					 const char **argv __unused)
+{
+	print_unsupported_msg();
+
+	return -1;
+}
+
+static int perl_stop_script_unsupported(void)
+{
+	return 0;
+}
+
+static void perl_process_event_unsupported(int cpu __unused,
+					   void *data __unused,
+					   int size __unused,
+					   unsigned long long nsecs __unused,
+					   char *comm __unused)
+{
+}
+
+static int perl_generate_script_unsupported(const char *outfile __unused)
+{
+	print_unsupported_msg();
+
+	return -1;
+}
+
+struct scripting_ops perl_scripting_unsupported_ops = {
+	.name = "Perl",
+	.start_script = perl_start_script_unsupported,
+	.stop_script = perl_stop_script_unsupported,
+	.process_event = perl_process_event_unsupported,
+	.generate_script = perl_generate_script_unsupported,
+};
+
+static void register_perl_scripting(struct scripting_ops *scripting_ops)
 {
 	int err;
-	err = script_spec_register("Perl", &perl_scripting_ops);
+	err = script_spec_register("Perl", scripting_ops);
 	if (err)
 		die("error registering Perl script extension");
 
-	err = script_spec_register("pl", &perl_scripting_ops);
+	err = script_spec_register("pl", scripting_ops);
 	if (err)
 		die("error registering pl script extension");
 
 	scripting_context = malloc(sizeof(struct scripting_context));
 }
+
+#ifdef NO_LIBPERL
+void setup_perl_scripting(void)
+{
+	register_perl_scripting(&perl_scripting_unsupported_ops);
+}
+#else
+void setup_perl_scripting(void)
+{
+	register_perl_scripting(&perl_scripting_ops);
+}
 #endif
diff --git a/tools/perf/util/trace-event-perl.h b/tools/perf/util/trace-event-perl.h
index 8fe0d866fe1a..e88fb26137bb 100644
--- a/tools/perf/util/trace-event-perl.h
+++ b/tools/perf/util/trace-event-perl.h
@@ -34,9 +34,13 @@ typedef int INTERP;
 #define dXSUB_SYS
 #define pTHX_
 static inline void newXS(const char *a, void *b, const char *c) {}
+static void boot_Perf__Trace__Context(pTHX_ CV *cv) {}
+static void boot_DynaLoader(pTHX_ CV *cv) {}
 #else
 #include <EXTERN.h>
 #include <perl.h>
+void boot_Perf__Trace__Context(pTHX_ CV *cv);
+void boot_DynaLoader(pTHX_ CV *cv);
 typedef PerlInterpreter * INTERP;
 #endif
 
diff --git a/tools/perf/util/trace-event-read.c b/tools/perf/util/trace-event-read.c
index 342dfdd43f87..1744422cafcb 100644
--- a/tools/perf/util/trace-event-read.c
+++ b/tools/perf/util/trace-event-read.c
@@ -145,8 +145,9 @@ static void read_proc_kallsyms(void)
 	if (!size)
 		return;
 
-	buf = malloc_or_die(size);
+	buf = malloc_or_die(size + 1);
 	read_or_die(buf, size);
+	buf[size] = '\0';
 
 	parse_proc_kallsyms(buf, size);
 
diff --git a/tools/perf/util/trace-event.h b/tools/perf/util/trace-event.h
index 81698d5e6503..6ad405620c9b 100644
--- a/tools/perf/util/trace-event.h
+++ b/tools/perf/util/trace-event.h
@@ -270,7 +270,7 @@ enum trace_flag_type {
 
 struct scripting_ops {
 	const char *name;
-	int (*start_script) (const char *);
+	int (*start_script) (const char *script, int argc, const char **argv);
 	int (*stop_script) (void);
 	void (*process_event) (int cpu, void *data, int size,
 			       unsigned long long nsecs, char *comm);