diff --git a/tools/lib/traceevent/event-parse.c b/tools/lib/traceevent/event-parse.c
index b5b4d806ffa2..f4190b5764de 100644
--- a/tools/lib/traceevent/event-parse.c
+++ b/tools/lib/traceevent/event-parse.c
@@ -3889,8 +3889,11 @@ static void pretty_print(struct trace_seq *s, void *data, int size, struct event
 				goto cont_process;
 			case '*':
 				/* The argument is the length. */
-				if (!arg)
-					die("no argument match");
+				if (!arg) {
+					do_warning("no argument match");
+					event->flags |= EVENT_FL_FAILED;
+					goto out_failed;
+				}
 				len_arg = eval_num_arg(data, size, event, arg);
 				len_as_arg = 1;
 				arg = arg->next;
@@ -3923,15 +3926,21 @@ static void pretty_print(struct trace_seq *s, void *data, int size, struct event
 			case 'x':
 			case 'X':
 			case 'u':
-				if (!arg)
-					die("no argument match");
+				if (!arg) {
+					do_warning("no argument match");
+					event->flags |= EVENT_FL_FAILED;
+					goto out_failed;
+				}
 
 				len = ((unsigned long)ptr + 1) -
 					(unsigned long)saveptr;
 
 				/* should never happen */
-				if (len > 31)
-					die("bad format!");
+				if (len > 31) {
+					do_warning("bad format!");
+					event->flags |= EVENT_FL_FAILED;
+					len = 31;
+				}
 
 				memcpy(format, saveptr, len);
 				format[len] = 0;
@@ -3995,19 +4004,26 @@ static void pretty_print(struct trace_seq *s, void *data, int size, struct event
 						trace_seq_printf(s, format, (long long)val);
 					break;
 				default:
-					die("bad count (%d)", ls);
+					do_warning("bad count (%d)", ls);
+					event->flags |= EVENT_FL_FAILED;
 				}
 				break;
 			case 's':
-				if (!arg)
-					die("no matching argument");
+				if (!arg) {
+					do_warning("no matching argument");
+					event->flags |= EVENT_FL_FAILED;
+					goto out_failed;
+				}
 
 				len = ((unsigned long)ptr + 1) -
 					(unsigned long)saveptr;
 
 				/* should never happen */
-				if (len > 31)
-					die("bad format!");
+				if (len > 31) {
+					do_warning("bad format!");
+					event->flags |= EVENT_FL_FAILED;
+					len = 31;
+				}
 
 				memcpy(format, saveptr, len);
 				format[len] = 0;
@@ -4025,6 +4041,11 @@ static void pretty_print(struct trace_seq *s, void *data, int size, struct event
 			trace_seq_putc(s, *ptr);
 	}
 
+	if (event->flags & EVENT_FL_FAILED) {
+out_failed:
+		trace_seq_printf(s, "[FAILED TO PARSE]");
+	}
+
 	if (args) {
 		free_args(args);
 		free(bprint_fmt);
@@ -4812,8 +4833,8 @@ int pevent_strerror(struct pevent *pevent, enum pevent_errno errnum,
 		msg = strerror_r(errnum, buf, buflen);
 		if (msg != buf) {
 			size_t len = strlen(msg);
-			char *c = mempcpy(buf, msg, min(buflen-1, len));
-			*c = '\0';
+			memcpy(buf, msg, min(buflen - 1, len));
+			*(buf + min(buflen - 1, len)) = '\0';
 		}
 		return 0;
 	}
@@ -5059,6 +5080,7 @@ int pevent_register_print_function(struct pevent *pevent,
 	struct pevent_func_params *param;
 	enum pevent_func_arg_type type;
 	va_list ap;
+	int ret;
 
 	func_handle = find_func_handler(pevent, name);
 	if (func_handle) {
@@ -5071,14 +5093,21 @@ int pevent_register_print_function(struct pevent *pevent,
 		remove_func_handler(pevent, name);
 	}
 
-	func_handle = malloc_or_die(sizeof(*func_handle));
+	func_handle = malloc(sizeof(*func_handle));
+	if (!func_handle) {
+		do_warning("Failed to allocate function handler");
+		return PEVENT_ERRNO__MEM_ALLOC_FAILED;
+	}
 	memset(func_handle, 0, sizeof(*func_handle));
 
 	func_handle->ret_type = ret_type;
 	func_handle->name = strdup(name);
 	func_handle->func = func;
-	if (!func_handle->name)
-		die("Failed to allocate function name");
+	if (!func_handle->name) {
+		do_warning("Failed to allocate function name");
+		free(func_handle);
+		return PEVENT_ERRNO__MEM_ALLOC_FAILED;
+	}
 
 	next_param = &(func_handle->params);
 	va_start(ap, name);
@@ -5088,11 +5117,17 @@ int pevent_register_print_function(struct pevent *pevent,
 			break;
 
 		if (type < 0 || type >= PEVENT_FUNC_ARG_MAX_TYPES) {
-			warning("Invalid argument type %d", type);
+			do_warning("Invalid argument type %d", type);
+			ret = PEVENT_ERRNO__INVALID_ARG_TYPE;
 			goto out_free;
 		}
 
-		param = malloc_or_die(sizeof(*param));
+		param = malloc(sizeof(*param));
+		if (!param) {
+			do_warning("Failed to allocate function param");
+			ret = PEVENT_ERRNO__MEM_ALLOC_FAILED;
+			goto out_free;
+		}
 		param->type = type;
 		param->next = NULL;
 
@@ -5110,7 +5145,7 @@ int pevent_register_print_function(struct pevent *pevent,
  out_free:
 	va_end(ap);
 	free_func_handle(func_handle);
-	return -1;
+	return ret;
 }
 
 /**
@@ -5162,7 +5197,12 @@ int pevent_register_event_handler(struct pevent *pevent,
 
  not_found:
 	/* Save for later use. */
-	handle = malloc_or_die(sizeof(*handle));
+	handle = malloc(sizeof(*handle));
+	if (!handle) {
+		do_warning("Failed to allocate event handler");
+		return PEVENT_ERRNO__MEM_ALLOC_FAILED;
+	}
+
 	memset(handle, 0, sizeof(*handle));
 	handle->id = id;
 	if (event_name)
@@ -5172,7 +5212,11 @@ int pevent_register_event_handler(struct pevent *pevent,
 
 	if ((event_name && !handle->event_name) ||
 	    (sys_name && !handle->sys_name)) {
-		die("Failed to allocate event/sys name");
+		do_warning("Failed to allocate event/sys name");
+		free((void *)handle->event_name);
+		free((void *)handle->sys_name);
+		free(handle);
+		return PEVENT_ERRNO__MEM_ALLOC_FAILED;
 	}
 
 	handle->func = func;
diff --git a/tools/lib/traceevent/event-parse.h b/tools/lib/traceevent/event-parse.h
index 863a0bbda7f1..3318963f1c98 100644
--- a/tools/lib/traceevent/event-parse.h
+++ b/tools/lib/traceevent/event-parse.h
@@ -351,7 +351,8 @@ enum pevent_flag {
 	_PE(READ_ID_FAILED,	"failed to read event id"),		      \
 	_PE(READ_FORMAT_FAILED,	"failed to read event format"),		      \
 	_PE(READ_PRINT_FAILED,	"failed to read event print fmt"), 	      \
-	_PE(OLD_FTRACE_ARG_FAILED,"failed to allocate field name for ftrace")
+	_PE(OLD_FTRACE_ARG_FAILED,"failed to allocate field name for ftrace"),\
+	_PE(INVALID_ARG_TYPE,	"invalid argument type")
 
 #undef _PE
 #define _PE(__code, __str) PEVENT_ERRNO__ ## __code
diff --git a/tools/perf/.gitignore b/tools/perf/.gitignore
index 26b823b61aa1..8f8fbc227a46 100644
--- a/tools/perf/.gitignore
+++ b/tools/perf/.gitignore
@@ -21,3 +21,5 @@ config.mak
 config.mak.autogen
 *-bison.*
 *-flex.*
+*.pyc
+*.pyo
diff --git a/tools/perf/Documentation/perf-diff.txt b/tools/perf/Documentation/perf-diff.txt
index 74d7481ed7a6..ab7f667de1b1 100644
--- a/tools/perf/Documentation/perf-diff.txt
+++ b/tools/perf/Documentation/perf-diff.txt
@@ -17,6 +17,9 @@ captured via perf record.
 
 If no parameters are passed it will assume perf.data.old and perf.data.
 
+The differential profile is displayed only for events matching both
+specified perf.data files.
+
 OPTIONS
 -------
 -M::
diff --git a/tools/perf/Makefile b/tools/perf/Makefile
index afd507574902..3eda49215730 100644
--- a/tools/perf/Makefile
+++ b/tools/perf/Makefile
@@ -755,6 +755,14 @@ else
 	endif
 endif
 
+ifdef NO_BACKTRACE
+       BASIC_CFLAGS += -DNO_BACKTRACE
+else
+       ifneq ($(call try-cc,$(SOURCE_BACKTRACE),),y)
+               BASIC_CFLAGS += -DNO_BACKTRACE
+       endif
+endif
+
 ifdef ASCIIDOC8
 	export ASCIIDOC8
 endif
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index d29d350fb2b7..e9933fdd256e 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
@@ -10,6 +10,7 @@
 #include "util/event.h"
 #include "util/hist.h"
 #include "util/evsel.h"
+#include "util/evlist.h"
 #include "util/session.h"
 #include "util/tool.h"
 #include "util/sort.h"
@@ -24,11 +25,6 @@ static char	  diff__default_sort_order[] = "dso,symbol";
 static bool  force;
 static bool show_displacement;
 
-struct perf_diff {
-	struct perf_tool tool;
-	struct perf_session *session;
-};
-
 static int hists__add_entry(struct hists *self,
 			    struct addr_location *al, u64 period)
 {
@@ -37,14 +33,12 @@ static int hists__add_entry(struct hists *self,
 	return -ENOMEM;
 }
 
-static int diff__process_sample_event(struct perf_tool *tool,
+static int diff__process_sample_event(struct perf_tool *tool __used,
 				      union perf_event *event,
 				      struct perf_sample *sample,
-				      struct perf_evsel *evsel __used,
+				      struct perf_evsel *evsel,
 				      struct machine *machine)
 {
-	struct perf_diff *_diff = container_of(tool, struct perf_diff, tool);
-	struct perf_session *session = _diff->session;
 	struct addr_location al;
 
 	if (perf_event__preprocess_sample(event, machine, &al, sample, NULL) < 0) {
@@ -56,26 +50,24 @@ static int diff__process_sample_event(struct perf_tool *tool,
 	if (al.filtered || al.sym == NULL)
 		return 0;
 
-	if (hists__add_entry(&session->hists, &al, sample->period)) {
+	if (hists__add_entry(&evsel->hists, &al, sample->period)) {
 		pr_warning("problem incrementing symbol period, skipping event\n");
 		return -1;
 	}
 
-	session->hists.stats.total_period += sample->period;
+	evsel->hists.stats.total_period += sample->period;
 	return 0;
 }
 
-static struct perf_diff diff = {
-	.tool = {
-		.sample	= diff__process_sample_event,
-		.mmap	= perf_event__process_mmap,
-		.comm	= perf_event__process_comm,
-		.exit	= perf_event__process_task,
-		.fork	= perf_event__process_task,
-		.lost	= perf_event__process_lost,
-		.ordered_samples = true,
-		.ordering_requires_timestamps = true,
-	},
+static struct perf_tool tool = {
+	.sample	= diff__process_sample_event,
+	.mmap	= perf_event__process_mmap,
+	.comm	= perf_event__process_comm,
+	.exit	= perf_event__process_task,
+	.fork	= perf_event__process_task,
+	.lost	= perf_event__process_lost,
+	.ordered_samples = true,
+	.ordering_requires_timestamps = true,
 };
 
 static void perf_session__insert_hist_entry_by_name(struct rb_root *root,
@@ -146,34 +138,71 @@ static void hists__match(struct hists *older, struct hists *newer)
 	}
 }
 
+static struct perf_evsel *evsel_match(struct perf_evsel *evsel,
+				      struct perf_evlist *evlist)
+{
+	struct perf_evsel *e;
+
+	list_for_each_entry(e, &evlist->entries, node)
+		if (perf_evsel__match2(evsel, e))
+			return e;
+
+	return NULL;
+}
+
 static int __cmd_diff(void)
 {
 	int ret, i;
 #define older (session[0])
 #define newer (session[1])
 	struct perf_session *session[2];
+	struct perf_evlist *evlist_new, *evlist_old;
+	struct perf_evsel *evsel;
+	bool first = true;
 
 	older = perf_session__new(input_old, O_RDONLY, force, false,
-				  &diff.tool);
+				  &tool);
 	newer = perf_session__new(input_new, O_RDONLY, force, false,
-				  &diff.tool);
+				  &tool);
 	if (session[0] == NULL || session[1] == NULL)
 		return -ENOMEM;
 
 	for (i = 0; i < 2; ++i) {
-		diff.session = session[i];
-		ret = perf_session__process_events(session[i], &diff.tool);
+		ret = perf_session__process_events(session[i], &tool);
 		if (ret)
 			goto out_delete;
-		hists__output_resort(&session[i]->hists);
 	}
 
-	if (show_displacement)
-		hists__resort_entries(&older->hists);
+	evlist_old = older->evlist;
+	evlist_new = newer->evlist;
+
+	list_for_each_entry(evsel, &evlist_new->entries, node)
+		hists__output_resort(&evsel->hists);
+
+	list_for_each_entry(evsel, &evlist_old->entries, node) {
+		hists__output_resort(&evsel->hists);
+
+		if (show_displacement)
+			hists__resort_entries(&evsel->hists);
+	}
+
+	list_for_each_entry(evsel, &evlist_new->entries, node) {
+		struct perf_evsel *evsel_old;
+
+		evsel_old = evsel_match(evsel, evlist_old);
+		if (!evsel_old)
+			continue;
+
+		fprintf(stdout, "%s# Event '%s'\n#\n", first ? "" : "\n",
+			perf_evsel__name(evsel));
+
+		first = false;
+
+		hists__match(&evsel_old->hists, &evsel->hists);
+		hists__fprintf(&evsel->hists, &evsel_old->hists,
+			       show_displacement, true, 0, 0, stdout);
+	}
 
-	hists__match(&older->hists, &newer->hists);
-	hists__fprintf(&newer->hists, &older->hists,
-		       show_displacement, true, 0, 0, stdout);
 out_delete:
 	for (i = 0; i < 2; ++i)
 		perf_session__delete(session[i]);
diff --git a/tools/perf/config/feature-tests.mak b/tools/perf/config/feature-tests.mak
index 2f1156a62ab7..116690a669d2 100644
--- a/tools/perf/config/feature-tests.mak
+++ b/tools/perf/config/feature-tests.mak
@@ -179,3 +179,17 @@ int main(void)
 }
 endef
 endif
+
+ifndef NO_BACKTRACE
+define SOURCE_BACKTRACE
+#include <execinfo.h>
+#include <stdio.h>
+
+int main(void)
+{
+	backtrace(NULL, 0);
+	backtrace_symbols(NULL, 0);
+	return 0;
+}
+endef
+endif
diff --git a/tools/perf/perf.c b/tools/perf/perf.c
index e7840e500715..fb8578cfa03c 100644
--- a/tools/perf/perf.c
+++ b/tools/perf/perf.c
@@ -14,6 +14,7 @@
 #include "util/run-command.h"
 #include "util/parse-events.h"
 #include "util/debugfs.h"
+#include <pthread.h>
 
 const char perf_usage_string[] =
 	"perf [--version] [--help] COMMAND [ARGS]";
diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h
index a6d6bc5d7164..62a6e7a7365d 100644
--- a/tools/perf/util/annotate.h
+++ b/tools/perf/util/annotate.h
@@ -7,6 +7,7 @@
 #include "symbol.h"
 #include <linux/list.h>
 #include <linux/rbtree.h>
+#include <pthread.h>
 
 struct ins;
 
diff --git a/tools/perf/util/dso-test-data.c b/tools/perf/util/dso-test-data.c
index 541cdc72c7df..c6caedeb1d6b 100644
--- a/tools/perf/util/dso-test-data.c
+++ b/tools/perf/util/dso-test-data.c
@@ -23,7 +23,7 @@ static char *test_file(int size)
 	int fd, i;
 	unsigned char *buf;
 
-	fd = mkostemp(templ, O_CREAT|O_WRONLY|O_TRUNC);
+	fd = mkstemp(templ);
 
 	buf = malloc(size);
 	if (!buf) {
diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index a3f562cec433..390690eb8781 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -124,6 +124,13 @@ void perf_evsel__close(struct perf_evsel *evsel, int ncpus, int nthreads);
 	(evsel->attr.type == PERF_TYPE_##t &&	\
 	 evsel->attr.config == PERF_COUNT_##c)
 
+static inline bool perf_evsel__match2(struct perf_evsel *e1,
+				      struct perf_evsel *e2)
+{
+	return (e1->attr.type == e2->attr.type) &&
+	       (e1->attr.config == e2->attr.config);
+}
+
 int __perf_evsel__read_on_cpu(struct perf_evsel *evsel,
 			      int cpu, int thread, bool scale);
 
diff --git a/tools/perf/util/help.c b/tools/perf/util/help.c
index 6f2975a00358..4fa764d8f7d7 100644
--- a/tools/perf/util/help.c
+++ b/tools/perf/util/help.c
@@ -3,6 +3,7 @@
 #include "exec_cmd.h"
 #include "levenshtein.h"
 #include "help.h"
+#include <termios.h>
 
 void add_cmdname(struct cmdnames *cmds, const char *name, size_t len)
 {
diff --git a/tools/perf/util/include/linux/rbtree.h b/tools/perf/util/include/linux/rbtree.h
index 7a243a143037..2a030c5af3aa 100644
--- a/tools/perf/util/include/linux/rbtree.h
+++ b/tools/perf/util/include/linux/rbtree.h
@@ -1 +1,2 @@
+#include <stdbool.h>
 #include "../../../../include/linux/rbtree.h"
diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
index 176a60902f56..aab414fbb64b 100644
--- a/tools/perf/util/session.h
+++ b/tools/perf/util/session.h
@@ -36,9 +36,7 @@ struct perf_session {
 	struct pevent		*pevent;
 	/*
 	 * FIXME: Need to split this up further, we need global
-	 *	  stats + per event stats. 'perf diff' also needs
-	 *	  to properly support multiple events in a single
-	 *	  perf.data file.
+	 *	  stats + per event stats.
 	 */
 	struct hists		hists;
 	int			fd;
diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c
index 0f5a0a496bc4..7a2fbd8855b7 100644
--- a/tools/perf/util/sort.c
+++ b/tools/perf/util/sort.c
@@ -12,8 +12,6 @@ int		sort__branch_mode = -1; /* -1 = means not set */
 
 enum sort_type	sort__first_dimension;
 
-char * field_sep;
-
 LIST_HEAD(hist_entry__sort_list);
 
 static int repsep_snprintf(char *bf, size_t size, const char *fmt, ...)
@@ -23,11 +21,11 @@ static int repsep_snprintf(char *bf, size_t size, const char *fmt, ...)
 
 	va_start(ap, fmt);
 	n = vsnprintf(bf, size, fmt, ap);
-	if (field_sep && n > 0) {
+	if (symbol_conf.field_sep && n > 0) {
 		char *sep = bf;
 
 		while (1) {
-			sep = strchr(sep, *field_sep);
+			sep = strchr(sep, *symbol_conf.field_sep);
 			if (sep == NULL)
 				break;
 			*sep = '.';
diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h
index e724b26acd51..e459c981b039 100644
--- a/tools/perf/util/sort.h
+++ b/tools/perf/util/sort.h
@@ -32,7 +32,6 @@ extern const char default_sort_order[];
 extern int sort__need_collapse;
 extern int sort__has_parent;
 extern int sort__branch_mode;
-extern char *field_sep;
 extern struct sort_entry sort_comm;
 extern struct sort_entry sort_dso;
 extern struct sort_entry sort_sym;
diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h
index fc4b1e630fd9..d3b330cbc3e0 100644
--- a/tools/perf/util/symbol.h
+++ b/tools/perf/util/symbol.h
@@ -10,6 +10,9 @@
 #include <linux/rbtree.h>
 #include <stdio.h>
 #include <byteswap.h>
+#if defined(__BIONIC__)
+#include <libgen.h>
+#endif
 
 #ifndef NO_LIBELF_SUPPORT
 #include <libelf.h>
diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h
index 33347ca89ee4..86ff1b15059b 100644
--- a/tools/perf/util/top.h
+++ b/tools/perf/util/top.h
@@ -5,6 +5,7 @@
 #include "types.h"
 #include <stddef.h>
 #include <stdbool.h>
+#include <termios.h>
 
 struct perf_evlist;
 struct perf_evsel;
diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c
index 1b8775c3707d..2055cf38041c 100644
--- a/tools/perf/util/util.c
+++ b/tools/perf/util/util.c
@@ -1,7 +1,9 @@
 #include "../perf.h"
 #include "util.h"
 #include <sys/mman.h>
+#ifndef NO_BACKTRACE
 #include <execinfo.h>
+#endif
 #include <stdio.h>
 #include <stdlib.h>
 
@@ -163,6 +165,7 @@ size_t hex_width(u64 v)
 }
 
 /* Obtain a backtrace and print it to stdout. */
+#ifndef NO_BACKTRACE
 void dump_stack(void)
 {
 	void *array[16];
@@ -177,3 +180,6 @@ void dump_stack(void)
 
 	free(strings);
 }
+#else
+void dump_stack(void) {}
+#endif