[tip:perf/core] perf jit: add source line info support

[tip-bot for Stephane Eranian]

Commit-ID:  598b7c6919c7bbcc1243009721a01bc12275ff3e
Gitweb:     http://git.kernel.org/tip/598b7c6919c7bbcc1243009721a01bc12275ff3e
Author:     Stephane Eranian <eranian@xxxxxxxxxx>
AuthorDate: Mon, 30 Nov 2015 10:02:23 +0100
Committer:  Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
CommitDate: Fri, 5 Feb 2016 12:33:09 -0300

perf jit: add source line info support

This patch adds source line information support to perf for jitted code.

The source line info must be emitted by the runtime, such as JVMTI.

Perf injects extract the source line info from the jitdump file and adds
the corresponding .debug_lines section in the ELF image generated for
each jitted function.

The source line enables matching any address in the profile with a
source file and line number.

The improvement is visible in perf annotate with the source code
displayed alongside the assembly code.

The dwarf code leverages the support from OProfile which is also
released under GPLv2.  Copyright 2007 OProfile authors.

Signed-off-by: Stephane Eranian <eranian@xxxxxxxxxx>
Cc: Adrian Hunter <adrian.hunter@xxxxxxxxx>
Cc: Andi Kleen <ak@xxxxxxxxxxxxxxx>
Cc: Carl Love <cel@xxxxxxxxxx>
Cc: David Ahern <dsahern@xxxxxxxxx>
Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
Cc: John McCutchan <johnmccutchan@xxxxxxxxxx>
Cc: Namhyung Kim <namhyung@xxxxxxxxxx>
Cc: Pawel Moll <pawel.moll@xxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Sonny Rao <sonnyrao@xxxxxxxxxxxx>
Cc: Sukadev Bhattiprolu <sukadev@xxxxxxxxxxxxxxxxxx>
Link: http://lkml.kernel.org/r/1448874143-7269-5-git-send-email-eranian@xxxxxxxxxx
Signed-off-by: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
---
 tools/perf/jvmti/jvmti_agent.c |  32 +--
 tools/perf/jvmti/jvmti_agent.h |  11 +-
 tools/perf/jvmti/libjvmti.c    | 122 ++++++++-
 tools/perf/util/Build          |   3 +
 tools/perf/util/genelf.c       |  15 +-
 tools/perf/util/genelf.h       |   6 +-
 tools/perf/util/genelf_debug.c | 610 +++++++++++++++++++++++++++++++++++++++++
 tools/perf/util/jitdump.c      |   8 +-
 8 files changed, 768 insertions(+), 39 deletions(-)

diff --git a/tools/perf/jvmti/jvmti_agent.c b/tools/perf/jvmti/jvmti_agent.c
index cbab139..6461e02 100644
--- a/tools/perf/jvmti/jvmti_agent.c
+++ b/tools/perf/jvmti/jvmti_agent.c
@@ -374,20 +374,20 @@ jvmti_write_code(void *agent, char const *sym,
 
 int
 jvmti_write_debug_info(void *agent, uint64_t code, const char *file,
-		       jvmtiAddrLocationMap const *map,
-		       jvmtiLineNumberEntry *li, jint num)
+		       jvmti_line_info_t *li, int nr_lines)
 {
-	static const char *prev_str = "\xff";
 	struct jr_code_debug_info rec;
 	size_t sret, len, size, flen;
 	size_t padding_count;
+	uint64_t addr;
+	const char *fn = file;
 	FILE *fp = agent;
 	int i;
 
 	/*
 	 * no entry to write
 	 */
-	if (!num)
+	if (!nr_lines)
 		return 0;
 
 	if (!fp) {
@@ -401,17 +401,18 @@ jvmti_write_debug_info(void *agent, uint64_t code, const char *file,
 	size            = sizeof(rec);
 	rec.p.timestamp = perf_get_timestamp();
 	rec.code_addr   = (uint64_t)(uintptr_t)code;
-	rec.nr_entry    = num;
+	rec.nr_entry    = nr_lines;
 
 	/*
 	 * on disk source line info layout:
 	 * uint64_t : addr
 	 * int      : line number
+	 * int      : column discriminator
 	 * file[]   : source file name
 	 * padding  : pad to multiple of 8 bytes
 	 */
-	size += num * (sizeof(uint64_t) + sizeof(int));
-	size += flen + (num - 1) * 2;
+	size += nr_lines * sizeof(struct debug_entry);
+	size += flen * nr_lines;
 	/*
 	 * pad to 8 bytes
 	 */
@@ -429,28 +430,27 @@ jvmti_write_debug_info(void *agent, uint64_t code, const char *file,
 	if (sret != 1)
 		goto error;
 
-	for (i = 0; i < num; i++) {
-		uint64_t addr;
+	for (i = 0; i < nr_lines; i++) {
 
-		addr = (uint64_t)map[i].start_address;
+		addr = (uint64_t)li[i].pc;
 		len  = sizeof(addr);
 		sret = fwrite_unlocked(&addr, len, 1, fp);
 		if (sret != 1)
 			goto error;
 
-		len  = sizeof(int);
+		len  = sizeof(li[0].line_number);
 		sret = fwrite_unlocked(&li[i].line_number, len, 1, fp);
 		if (sret != 1)
 			goto error;
 
-		if (i == 0) {
-			sret = fwrite_unlocked(file, flen, 1, fp);
-		} else {
-			sret = fwrite_unlocked(prev_str, 2, 1, fp);
-		}
+		len  = sizeof(li[0].discrim);
+		sret = fwrite_unlocked(&li[i].discrim, len, 1, fp);
 		if (sret != 1)
 			goto error;
 
+		sret = fwrite_unlocked(fn, flen, 1, fp);
+		if (sret != 1)
+			goto error;
 	}
 	if (padding_count)
 		sret = fwrite_unlocked(pad_bytes, padding_count, 1, fp);
diff --git a/tools/perf/jvmti/jvmti_agent.h b/tools/perf/jvmti/jvmti_agent.h
index 8251a1c..bedf5d0 100644
--- a/tools/perf/jvmti/jvmti_agent.h
+++ b/tools/perf/jvmti/jvmti_agent.h
@@ -11,16 +11,23 @@
 extern "C" {
 #endif
 
+typedef struct {
+	unsigned long	pc;
+	int		line_number;
+	int		discrim; /* discriminator -- 0 for now */
+} jvmti_line_info_t;
+
 void *jvmti_open(void);
 int   jvmti_close(void *agent);
 int   jvmti_write_code(void *agent, char const *symbol_name,
 		       uint64_t vma, void const *code,
 		       const unsigned int code_size);
+
 int   jvmti_write_debug_info(void *agent,
 		             uint64_t code,
 			     const char *file,
-			     jvmtiAddrLocationMap const *map,
-			     jvmtiLineNumberEntry *tab, jint nr);
+			     jvmti_line_info_t *li,
+			     int nr_lines);
 
 #if defined(__cplusplus)
 }
diff --git a/tools/perf/jvmti/libjvmti.c b/tools/perf/jvmti/libjvmti.c
index 92ffbe4..ac12e4b 100644
--- a/tools/perf/jvmti/libjvmti.c
+++ b/tools/perf/jvmti/libjvmti.c
@@ -4,6 +4,7 @@
 #include <stdlib.h>
 #include <err.h>
 #include <jvmti.h>
+#include <jvmticmlr.h>
 #include <limits.h>
 
 #include "jvmti_agent.h"
@@ -11,6 +12,100 @@
 static int has_line_numbers;
 void *jvmti_agent;
 
+static jvmtiError
+do_get_line_numbers(jvmtiEnv *jvmti, void *pc, jmethodID m, jint bci,
+		    jvmti_line_info_t *tab, jint *nr)
+{
+	jint i, lines = 0;
+	jint nr_lines = 0;
+	jvmtiLineNumberEntry *loc_tab = NULL;
+	jvmtiError ret;
+
+	ret = (*jvmti)->GetLineNumberTable(jvmti, m, &nr_lines, &loc_tab);
+	if (ret != JVMTI_ERROR_NONE)
+		return ret;
+
+	for (i = 0; i < nr_lines; i++) {
+		if (loc_tab[i].start_location < bci) {
+			tab[lines].pc = (unsigned long)pc;
+			tab[lines].line_number = loc_tab[i].line_number;
+			tab[lines].discrim = 0; /* not yet used */
+			lines++;
+		} else {
+			break;
+		}
+	}
+	(*jvmti)->Deallocate(jvmti, (unsigned char *)loc_tab);
+	*nr = lines;
+	return JVMTI_ERROR_NONE;
+}
+
+static jvmtiError
+get_line_numbers(jvmtiEnv *jvmti, const void *compile_info, jvmti_line_info_t **tab, int *nr_lines)
+{
+	const jvmtiCompiledMethodLoadRecordHeader *hdr;
+	jvmtiCompiledMethodLoadInlineRecord *rec;
+	jvmtiLineNumberEntry *lne = NULL;
+	PCStackInfo *c;
+	jint nr, ret;
+	int nr_total = 0;
+	int i, lines_total = 0;
+
+	if (!(tab && nr_lines))
+		return JVMTI_ERROR_NULL_POINTER;
+
+	/*
+	 * Phase 1 -- get the number of lines necessary
+	 */
+	for (hdr = compile_info; hdr != NULL; hdr = hdr->next) {
+		if (hdr->kind == JVMTI_CMLR_INLINE_INFO) {
+			rec = (jvmtiCompiledMethodLoadInlineRecord *)hdr;
+			for (i = 0; i < rec->numpcs; i++) {
+				c = rec->pcinfo + i;
+				nr = 0;
+				/*
+				 * unfortunately, need a tab to get the number of lines!
+				 */
+				ret = (*jvmti)->GetLineNumberTable(jvmti, c->methods[0], &nr, &lne);
+				if (ret == JVMTI_ERROR_NONE) {
+					/* free what was allocated for nothing */
+					(*jvmti)->Deallocate(jvmti, (unsigned char *)lne);
+					nr_total += (int)nr;
+				}
+			}
+		}
+	}
+
+	if (nr_total == 0)
+		return JVMTI_ERROR_NOT_FOUND;
+
+	/*
+	 * Phase 2 -- allocate big enough line table
+	 */
+	*tab = malloc(nr_total * sizeof(**tab));
+	if (!*tab)
+		return JVMTI_ERROR_OUT_OF_MEMORY;
+
+	for (hdr = compile_info; hdr != NULL; hdr = hdr->next) {
+		if (hdr->kind == JVMTI_CMLR_INLINE_INFO) {
+			rec = (jvmtiCompiledMethodLoadInlineRecord *)hdr;
+			for (i = 0; i < rec->numpcs; i++) {
+				c = rec->pcinfo + i;
+				nr = 0;
+				ret = do_get_line_numbers(jvmti, c->pc,
+							  c->methods[0],
+							  c->bcis[0],
+							  *tab + lines_total,
+							  &nr);
+				if (ret == JVMTI_ERROR_NONE)
+					lines_total += nr;
+			}
+		}
+	}
+	*nr_lines = lines_total;
+	return JVMTI_ERROR_NONE;
+}
+
 static void JNICALL
 compiled_method_load_cb(jvmtiEnv *jvmti,
 			jmethodID method,
@@ -18,9 +113,9 @@ compiled_method_load_cb(jvmtiEnv *jvmti,
 			void const *code_addr,
 			jint map_length,
 			jvmtiAddrLocationMap const *map,
-			void const *compile_info __unused)
+			const void *compile_info)
 {
-	jvmtiLineNumberEntry *tab = NULL;
+	jvmti_line_info_t *line_tab = NULL;
 	jclass decl_class;
 	char *class_sign = NULL;
 	char *func_name = NULL;
@@ -29,7 +124,7 @@ compiled_method_load_cb(jvmtiEnv *jvmti,
 	char fn[PATH_MAX];
 	uint64_t addr = (uint64_t)(uintptr_t)code_addr;
 	jvmtiError ret;
-	jint nr_lines = 0;
+	int nr_lines = 0; /* in line_tab[] */
 	size_t len;
 
 	ret = (*jvmti)->GetMethodDeclaringClass(jvmti, method,
@@ -40,19 +135,19 @@ compiled_method_load_cb(jvmtiEnv *jvmti,
 	}
 
 	if (has_line_numbers && map && map_length) {
-
-		ret = (*jvmti)->GetLineNumberTable(jvmti, method, &nr_lines, &tab);
+		ret = get_line_numbers(jvmti, compile_info, &line_tab, &nr_lines);
 		if (ret != JVMTI_ERROR_NONE) {
 			warnx("jvmti: cannot get line table for method");
-		} else {
-			ret = (*jvmti)->GetSourceFileName(jvmti, decl_class, &file_name);
-			if (ret != JVMTI_ERROR_NONE) {
-				warnx("jvmti: cannot get source filename ret=%d", ret);
-				nr_lines = 0;
-			}
+			nr_lines = 0;
 		}
 	}
 
+	ret = (*jvmti)->GetSourceFileName(jvmti, decl_class, &file_name);
+	if (ret != JVMTI_ERROR_NONE) {
+		warnx("jvmti: cannot get source filename ret=%d", ret);
+		goto error;
+	}
+
 	ret = (*jvmti)->GetClassSignature(jvmti, decl_class,
 					  &class_sign, NULL);
 	if (ret != JVMTI_ERROR_NONE) {
@@ -92,13 +187,14 @@ compiled_method_load_cb(jvmtiEnv *jvmti,
 	/*
 	 * write source line info record if we have it
 	 */
-	if (jvmti_write_debug_info(jvmti_agent, addr, fn, map, tab, nr_lines))
+	if (jvmti_write_debug_info(jvmti_agent, addr, fn, line_tab, nr_lines))
 		warnx("jvmti: write_debug_info() failed");
 
 	len = strlen(func_name) + strlen(class_sign) + strlen(func_sign) + 2;
 	{
 		char str[len];
 		snprintf(str, len, "%s%s%s", class_sign, func_name, func_sign);
+
 		if (jvmti_write_code(jvmti_agent, str, addr, code_addr, code_size))
 			warnx("jvmti: write_code() failed");
 	}
@@ -106,8 +202,8 @@ error:
 	(*jvmti)->Deallocate(jvmti, (unsigned char *)func_name);
 	(*jvmti)->Deallocate(jvmti, (unsigned char *)func_sign);
 	(*jvmti)->Deallocate(jvmti, (unsigned char *)class_sign);
-	(*jvmti)->Deallocate(jvmti, (unsigned char *)tab);
 	(*jvmti)->Deallocate(jvmti, (unsigned char *)file_name);
+	free(line_tab);
 }
 
 static void JNICALL
diff --git a/tools/perf/util/Build b/tools/perf/util/Build
index 52a4a80..a34752d 100644
--- a/tools/perf/util/Build
+++ b/tools/perf/util/Build
@@ -108,8 +108,11 @@ libperf-$(CONFIG_LZMA) += lzma.o
 libperf-y += demangle-java.o
 libperf-$(CONFIG_LIBELF) += jitdump.o
 libperf-$(CONFIG_LIBELF) += genelf.o
+libperf-$(CONFIG_LIBELF) += genelf_debug.o
 
 CFLAGS_config.o   += -DETC_PERFCONFIG="BUILD_STR($(ETC_PERFCONFIG_SQ))"
+# avoid compiler warnings in 32-bit mode
+CFLAGS_genelf_debug.o  += -Wno-packed
 
 $(OUTPUT)util/parse-events-flex.c: util/parse-events.l $(OUTPUT)util/parse-events-bison.c
 	$(call rule_mkdir)
diff --git a/tools/perf/util/genelf.c b/tools/perf/util/genelf.c
index 145f811..c1ef805 100644
--- a/tools/perf/util/genelf.c
+++ b/tools/perf/util/genelf.c
@@ -156,7 +156,8 @@ gen_build_id(struct buildid_note *note, unsigned long load_addr, const void *cod
  */
 int
 jit_write_elf(int fd, uint64_t load_addr, const char *sym,
-	      const void *code, int csize)
+	      const void *code, int csize,
+	      void *debug, int nr_debug_entries)
 {
 	Elf *e;
 	Elf_Data *d;
@@ -385,9 +386,15 @@ jit_write_elf(int fd, uint64_t load_addr, const char *sym,
 	shdr->sh_size = sizeof(bnote);
 	shdr->sh_entsize = 0;
 
-	if (elf_update(e, ELF_C_WRITE) < 0) {
-		warnx("elf_update 4 failed");
-		goto error;
+	if (debug && nr_debug_entries) {
+		retval = jit_add_debug_info(e, load_addr, debug, nr_debug_entries);
+		if (retval)
+			goto error;
+	} else {
+		if (elf_update(e, ELF_C_WRITE) < 0) {
+			warnx("elf_update 4 failed");
+			goto error;
+		}
 	}
 
 	retval = 0;
diff --git a/tools/perf/util/genelf.h b/tools/perf/util/genelf.h
index d8e9ece..45bf9c6 100644
--- a/tools/perf/util/genelf.h
+++ b/tools/perf/util/genelf.h
@@ -3,7 +3,11 @@
 
 /* genelf.c */
 extern int jit_write_elf(int fd, uint64_t code_addr, const char *sym,
-			 const void *code, int csize);
+			 const void *code, int csize,
+			 void *debug, int nr_debug_entries);
+/* genelf_debug.c */
+extern int jit_add_debug_info(Elf *e, uint64_t code_addr,
+			      void *debug, int nr_debug_entries);
 
 #if   defined(__arm__)
 #define GEN_ELF_ARCH	EM_ARM
diff --git a/tools/perf/util/genelf_debug.c b/tools/perf/util/genelf_debug.c
new file mode 100644
index 0000000..5980f7d
--- /dev/null
+++ b/tools/perf/util/genelf_debug.c
@@ -0,0 +1,610 @@
+/*
+ * genelf_debug.c
+ * Copyright (C) 2015, Google, Inc
+ *
+ * Contributed by:
+ * 	Stephane Eranian <eranian@xxxxxxxxxx>
+ *
+ * Released under the GPL v2.
+ *
+ * based on GPLv2 source code from Oprofile
+ * @remark Copyright 2007 OProfile authors
+ * @author Philippe Elie
+ */
+#include <sys/types.h>
+#include <stdio.h>
+#include <getopt.h>
+#include <stddef.h>
+#include <libelf.h>
+#include <string.h>
+#include <stdlib.h>
+#include <inttypes.h>
+#include <limits.h>
+#include <fcntl.h>
+#include <err.h>
+#include <dwarf.h>
+
+#include "perf.h"
+#include "genelf.h"
+#include "../util/jitdump.h"
+
+#define BUFFER_EXT_DFL_SIZE	(4 * 1024)
+
+typedef uint32_t uword;
+typedef uint16_t uhalf;
+typedef int32_t  sword;
+typedef int16_t  shalf;
+typedef uint8_t  ubyte;
+typedef int8_t   sbyte;
+
+struct buffer_ext {
+	size_t cur_pos;
+	size_t max_sz;
+	void *data;
+};
+
+static void
+buffer_ext_dump(struct buffer_ext *be, const char *msg)
+{
+	size_t i;
+	warnx("DUMP for %s", msg);
+	for (i = 0 ; i < be->cur_pos; i++)
+		warnx("%4zu 0x%02x", i, (((char *)be->data)[i]) & 0xff);
+}
+
+static inline int
+buffer_ext_add(struct buffer_ext *be, void *addr, size_t sz)
+{
+	void *tmp;
+	size_t be_sz = be->max_sz;
+
+retry:
+	if ((be->cur_pos + sz) < be_sz) {
+		memcpy(be->data + be->cur_pos, addr, sz);
+		be->cur_pos += sz;
+		return 0;
+	}
+
+	if (!be_sz)
+		be_sz = BUFFER_EXT_DFL_SIZE;
+	else
+		be_sz <<= 1;
+
+	tmp = realloc(be->data, be_sz);
+	if (!tmp)
+		return -1;
+
+	be->data   = tmp;
+	be->max_sz = be_sz;
+
+	goto retry;
+}
+
+static void
+buffer_ext_init(struct buffer_ext *be)
+{
+	be->data = NULL;
+	be->cur_pos = 0;
+	be->max_sz = 0;
+}
+
+static inline size_t
+buffer_ext_size(struct buffer_ext *be)
+{
+	return be->cur_pos;
+}
+
+static inline void *
+buffer_ext_addr(struct buffer_ext *be)
+{
+	return be->data;
+}
+
+struct debug_line_header {
+	// Not counting this field
+	uword total_length;
+	// version number (2 currently)
+	uhalf version;
+	// relative offset from next field to
+	// program statement
+	uword prolog_length;
+	ubyte minimum_instruction_length;
+	ubyte default_is_stmt;
+	// line_base - see DWARF 2 specs
+	sbyte line_base;
+	// line_range - see DWARF 2 specs
+	ubyte line_range;
+	// number of opcode + 1
+	ubyte opcode_base;
+	/* follow the array of opcode args nr: ubytes [nr_opcode_base] */
+	/* follow the search directories index, zero terminated string
+	 * terminated by an empty string.
+	 */
+	/* follow an array of { filename, LEB128, LEB128, LEB128 }, first is
+	 * the directory index entry, 0 means current directory, then mtime
+	 * and filesize, last entry is followed by en empty string.
+	 */
+	/* follow the first program statement */
+} __attribute__((packed));
+
+/* DWARF 2 spec talk only about one possible compilation unit header while
+ * binutils can handle two flavours of dwarf 2, 32 and 64 bits, this is not
+ * related to the used arch, an ELF 32 can hold more than 4 Go of debug
+ * information. For now we handle only DWARF 2 32 bits comp unit. It'll only
+ * become a problem if we generate more than 4GB of debug information.
+ */
+struct compilation_unit_header {
+	uword total_length;
+	uhalf version;
+	uword debug_abbrev_offset;
+	ubyte pointer_size;
+} __attribute__((packed));
+
+#define DW_LNS_num_opcode (DW_LNS_set_isa + 1)
+
+/* field filled at run time are marked with -1 */
+static struct debug_line_header const default_debug_line_header = {
+	.total_length = -1,
+	.version = 2,
+	.prolog_length = -1,
+	.minimum_instruction_length = 1,	/* could be better when min instruction size != 1 */
+	.default_is_stmt = 1,	/* we don't take care about basic block */
+	.line_base = -5,	/* sensible value for line base ... */
+	.line_range = -14,     /* ... and line range are guessed statically */
+	.opcode_base = DW_LNS_num_opcode
+};
+
+static ubyte standard_opcode_length[] =
+{
+	0, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1
+};
+#if 0
+{
+	[DW_LNS_advance_pc]   = 1,
+	[DW_LNS_advance_line] = 1,
+	[DW_LNS_set_file] =  1,
+	[DW_LNS_set_column] = 1,
+	[DW_LNS_fixed_advance_pc] = 1,
+	[DW_LNS_set_isa] = 1,
+};
+#endif
+
+/* field filled at run time are marked with -1 */
+static struct compilation_unit_header default_comp_unit_header = {
+	.total_length = -1,
+	.version = 2,
+	.debug_abbrev_offset = 0,     /* we reuse the same abbrev entries for all comp unit */
+	.pointer_size = sizeof(void *)
+};
+
+static void emit_uword(struct buffer_ext *be, uword data)
+{
+	buffer_ext_add(be, &data, sizeof(uword));
+}
+
+static void emit_string(struct buffer_ext *be, const char *s)
+{
+	buffer_ext_add(be, (void *)s, strlen(s) + 1);
+}
+
+static void emit_unsigned_LEB128(struct buffer_ext *be,
+				 unsigned long data)
+{
+	do {
+		ubyte cur = data & 0x7F;
+		data >>= 7;
+		if (data)
+			cur |= 0x80;
+		buffer_ext_add(be, &cur, 1);
+	} while (data);
+}
+
+static void emit_signed_LEB128(struct buffer_ext *be, long data)
+{
+	int more = 1;
+	int negative = data < 0;
+	int size = sizeof(long) * CHAR_BIT;
+	while (more) {
+		ubyte cur = data & 0x7F;
+		data >>= 7;
+		if (negative)
+			data |= - (1 << (size - 7));
+		if ((data == 0 && !(cur & 0x40)) ||
+		    (data == -1l && (cur & 0x40)))
+			more = 0;
+		else
+			cur |= 0x80;
+		buffer_ext_add(be, &cur, 1);
+	}
+}
+
+static void emit_extended_opcode(struct buffer_ext *be, ubyte opcode,
+				 void *data, size_t data_len)
+{
+	buffer_ext_add(be, (char *)"", 1);
+
+	emit_unsigned_LEB128(be, data_len + 1);
+
+	buffer_ext_add(be, &opcode, 1);
+	buffer_ext_add(be, data, data_len);
+}
+
+static void emit_opcode(struct buffer_ext *be, ubyte opcode)
+{
+	buffer_ext_add(be, &opcode, 1);
+}
+
+static void emit_opcode_signed(struct buffer_ext  *be,
+			       ubyte opcode, long data)
+{
+	buffer_ext_add(be, &opcode, 1);
+	emit_signed_LEB128(be, data);
+}
+
+static void emit_opcode_unsigned(struct buffer_ext *be, ubyte opcode,
+				 unsigned long data)
+{
+	buffer_ext_add(be, &opcode, 1);
+	emit_unsigned_LEB128(be, data);
+}
+
+static void emit_advance_pc(struct buffer_ext *be, unsigned long delta_pc)
+{
+	emit_opcode_unsigned(be, DW_LNS_advance_pc, delta_pc);
+}
+
+static void emit_advance_lineno(struct buffer_ext  *be, long delta_lineno)
+{
+	emit_opcode_signed(be, DW_LNS_advance_line, delta_lineno);
+}
+
+static void emit_lne_end_of_sequence(struct buffer_ext *be)
+{
+	emit_extended_opcode(be, DW_LNE_end_sequence, NULL, 0);
+}
+
+static void emit_set_file(struct buffer_ext *be, unsigned long idx)
+{
+	emit_opcode_unsigned(be, DW_LNS_set_file, idx);
+}
+
+static void emit_lne_define_filename(struct buffer_ext *be,
+				     const char *filename)
+{
+	buffer_ext_add(be, (void *)"", 1);
+
+	/* LNE field, strlen(filename) + zero termination, 3 bytes for: the dir entry, timestamp, filesize */
+	emit_unsigned_LEB128(be, strlen(filename) + 5);
+	emit_opcode(be, DW_LNE_define_file);
+	emit_string(be, filename);
+	/* directory index 0=do not know */
+        emit_unsigned_LEB128(be, 0);
+	/* last modification date on file 0=do not know */
+        emit_unsigned_LEB128(be, 0);
+	/* filesize 0=do not know */
+        emit_unsigned_LEB128(be, 0);
+}
+
+static void emit_lne_set_address(struct buffer_ext *be,
+				 void *address)
+{
+	emit_extended_opcode(be, DW_LNE_set_address, &address, sizeof(unsigned long));
+}
+
+static ubyte get_special_opcode(struct debug_entry *ent,
+				unsigned int last_line,
+				unsigned long last_vma)
+{
+	unsigned int temp;
+	unsigned long delta_addr;
+
+	/*
+	 * delta from line_base
+	 */
+	temp = (ent->lineno - last_line) - default_debug_line_header.line_base;
+
+	if (temp >= default_debug_line_header.line_range)
+		return 0;
+
+	/*
+	 * delta of addresses
+	 */
+	delta_addr = (ent->addr - last_vma) / default_debug_line_header.minimum_instruction_length;
+
+	/* This is not sufficient to ensure opcode will be in [0-256] but
+	 * sufficient to ensure when summing with the delta lineno we will
+	 * not overflow the unsigned long opcode */
+
+	if (delta_addr <= 256 / default_debug_line_header.line_range) {
+		unsigned long opcode = temp +
+			(delta_addr * default_debug_line_header.line_range) +
+			default_debug_line_header.opcode_base;
+
+		return opcode <= 255 ? opcode : 0;
+	}
+	return 0;
+}
+
+static void emit_lineno_info(struct buffer_ext *be,
+			     struct debug_entry *ent, size_t nr_entry,
+			     unsigned long code_addr)
+{
+	size_t i;
+
+	/*
+	 * Machine state at start of a statement program
+	 * address = 0
+	 * file    = 1
+	 * line    = 1
+	 * column  = 0
+	 * is_stmt = default_is_stmt as given in the debug_line_header
+	 * basic block = 0
+	 * end sequence = 0
+	 */
+
+	/* start state of the state machine we take care of */
+	unsigned long last_vma = code_addr;
+	char const  *cur_filename = NULL;
+	unsigned long cur_file_idx = 0;
+	int last_line = 1;
+
+	emit_lne_set_address(be, (void *)code_addr);
+
+	for (i = 0; i < nr_entry; i++, ent = debug_entry_next(ent)) {
+		int need_copy = 0;
+		ubyte special_opcode;
+
+		/*
+		 * check if filename changed, if so add it
+		 */
+		if (!cur_filename || strcmp(cur_filename, ent->name)) {
+			emit_lne_define_filename(be, ent->name);
+			cur_filename = ent->name;
+			emit_set_file(be, ++cur_file_idx);
+			need_copy = 1;
+		}
+
+		special_opcode = get_special_opcode(ent, last_line, last_vma);
+		if (special_opcode != 0) {
+			last_line = ent->lineno;
+			last_vma  = ent->addr;
+			emit_opcode(be, special_opcode);
+		} else {
+			/*
+			 * lines differ, emit line delta
+			 */
+			if (last_line != ent->lineno) {
+				emit_advance_lineno(be, ent->lineno - last_line);
+				last_line = ent->lineno;
+				need_copy = 1;
+			}
+			/*
+			 * addresses differ, emit address delta
+			 */
+			if (last_vma != ent->addr) {
+				emit_advance_pc(be, ent->addr - last_vma);
+				last_vma = ent->addr;
+				need_copy = 1;
+			}
+			/*
+			 * add new row to matrix
+			 */
+			if (need_copy)
+				emit_opcode(be, DW_LNS_copy);
+		}
+	}
+}
+
+static void add_debug_line(struct buffer_ext *be,
+	struct debug_entry *ent, size_t nr_entry,
+	unsigned long code_addr)
+{
+	struct debug_line_header * dbg_header;
+	size_t old_size;
+
+	old_size = buffer_ext_size(be);
+
+	buffer_ext_add(be, (void *)&default_debug_line_header,
+		 sizeof(default_debug_line_header));
+
+	buffer_ext_add(be, &standard_opcode_length,  sizeof(standard_opcode_length));
+
+	// empty directory entry
+	buffer_ext_add(be, (void *)"", 1);
+
+	// empty filename directory
+	buffer_ext_add(be, (void *)"", 1);
+
+	dbg_header = buffer_ext_addr(be) + old_size;
+	dbg_header->prolog_length = (buffer_ext_size(be) - old_size) -
+		offsetof(struct debug_line_header, minimum_instruction_length);
+
+	emit_lineno_info(be, ent, nr_entry, code_addr);
+
+	emit_lne_end_of_sequence(be);
+
+	dbg_header = buffer_ext_addr(be) + old_size;
+	dbg_header->total_length = (buffer_ext_size(be) - old_size) -
+		offsetof(struct debug_line_header, version);
+}
+
+static void
+add_debug_abbrev(struct buffer_ext *be)
+{
+        emit_unsigned_LEB128(be, 1);
+        emit_unsigned_LEB128(be, DW_TAG_compile_unit);
+        emit_unsigned_LEB128(be, DW_CHILDREN_yes);
+        emit_unsigned_LEB128(be, DW_AT_stmt_list);
+        emit_unsigned_LEB128(be, DW_FORM_data4);
+        emit_unsigned_LEB128(be, 0);
+        emit_unsigned_LEB128(be, 0);
+        emit_unsigned_LEB128(be, 0);
+}
+
+static void
+add_compilation_unit(struct buffer_ext *be,
+		     size_t offset_debug_line)
+{
+	struct compilation_unit_header *comp_unit_header;
+	size_t old_size = buffer_ext_size(be);
+
+	buffer_ext_add(be, &default_comp_unit_header,
+		       sizeof(default_comp_unit_header));
+
+	emit_unsigned_LEB128(be, 1);
+	emit_uword(be, offset_debug_line);
+
+	comp_unit_header = buffer_ext_addr(be) + old_size;
+	comp_unit_header->total_length = (buffer_ext_size(be) - old_size) -
+		offsetof(struct compilation_unit_header, version);
+}
+
+static int
+jit_process_debug_info(uint64_t code_addr,
+		       void *debug, int nr_debug_entries,
+		       struct buffer_ext *dl,
+		       struct buffer_ext *da,
+		       struct buffer_ext *di)
+{
+	struct debug_entry *ent = debug;
+	int i;
+
+	for (i = 0; i < nr_debug_entries; i++) {
+		ent->addr = ent->addr - code_addr;
+		ent = debug_entry_next(ent);
+	}
+	add_compilation_unit(di, buffer_ext_size(dl));
+	add_debug_line(dl, debug, nr_debug_entries, 0);
+	add_debug_abbrev(da);
+	if (0) buffer_ext_dump(da, "abbrev");
+
+	return 0;
+}
+
+int
+jit_add_debug_info(Elf *e, uint64_t code_addr, void *debug, int nr_debug_entries)
+{
+	Elf_Data *d;
+	Elf_Scn *scn;
+	Elf_Shdr *shdr;
+	struct buffer_ext dl, di, da;
+	int ret;
+
+	buffer_ext_init(&dl);
+	buffer_ext_init(&di);
+	buffer_ext_init(&da);
+
+	ret = jit_process_debug_info(code_addr, debug, nr_debug_entries, &dl, &da, &di);
+	if (ret)
+		return -1;
+	/*
+	 * setup .debug_line section
+	 */
+	scn = elf_newscn(e);
+	if (!scn) {
+		warnx("cannot create section");
+		return -1;
+	}
+
+	d = elf_newdata(scn);
+	if (!d) {
+		warnx("cannot get new data");
+		return -1;
+	}
+
+	d->d_align = 1;
+	d->d_off = 0LL;
+	d->d_buf = buffer_ext_addr(&dl);
+	d->d_type = ELF_T_BYTE;
+	d->d_size = buffer_ext_size(&dl);
+	d->d_version = EV_CURRENT;
+
+	shdr = elf_getshdr(scn);
+	if (!shdr) {
+		warnx("cannot get section header");
+		return -1;
+	}
+
+	shdr->sh_name = 52; /* .debug_line */
+	shdr->sh_type = SHT_PROGBITS;
+	shdr->sh_addr = 0; /* must be zero or == sh_offset -> dynamic object */
+	shdr->sh_flags = 0;
+	shdr->sh_entsize = 0;
+
+	/*
+	 * setup .debug_info section
+	 */
+	scn = elf_newscn(e);
+	if (!scn) {
+		warnx("cannot create section");
+		return -1;
+	}
+
+	d = elf_newdata(scn);
+	if (!d) {
+		warnx("cannot get new data");
+		return -1;
+	}
+
+	d->d_align = 1;
+	d->d_off = 0LL;
+	d->d_buf = buffer_ext_addr(&di);
+	d->d_type = ELF_T_BYTE;
+	d->d_size = buffer_ext_size(&di);
+	d->d_version = EV_CURRENT;
+
+	shdr = elf_getshdr(scn);
+	if (!shdr) {
+		warnx("cannot get section header");
+		return -1;
+	}
+
+	shdr->sh_name = 64; /* .debug_info */
+	shdr->sh_type = SHT_PROGBITS;
+	shdr->sh_addr = 0; /* must be zero or == sh_offset -> dynamic object */
+	shdr->sh_flags = 0;
+	shdr->sh_entsize = 0;
+
+	/*
+	 * setup .debug_abbrev section
+	 */
+	scn = elf_newscn(e);
+	if (!scn) {
+		warnx("cannot create section");
+		return -1;
+	}
+
+	d = elf_newdata(scn);
+	if (!d) {
+		warnx("cannot get new data");
+		return -1;
+	}
+
+	d->d_align = 1;
+	d->d_off = 0LL;
+	d->d_buf = buffer_ext_addr(&da);
+	d->d_type = ELF_T_BYTE;
+	d->d_size = buffer_ext_size(&da);
+	d->d_version = EV_CURRENT;
+
+	shdr = elf_getshdr(scn);
+	if (!shdr) {
+		warnx("cannot get section header");
+		return -1;
+	}
+
+	shdr->sh_name = 76; /* .debug_info */
+	shdr->sh_type = SHT_PROGBITS;
+	shdr->sh_addr = 0; /* must be zero or == sh_offset -> dynamic object */
+	shdr->sh_flags = 0;
+	shdr->sh_entsize = 0;
+
+	/*
+	 * now we update the ELF image with all the sections
+	 */
+	if (elf_update(e, ELF_C_WRITE) < 0) {
+		warnx("elf_update debug failed");
+		return -1;
+	}
+	return 0;
+}
diff --git a/tools/perf/util/jitdump.c b/tools/perf/util/jitdump.c
index 9f7a012..99fa5ee 100644
--- a/tools/perf/util/jitdump.c
+++ b/tools/perf/util/jitdump.c
@@ -63,7 +63,9 @@ jit_emit_elf(char *filename,
 	     const char *sym,
 	     uint64_t code_addr,
 	     const void *code,
-	     int csize)
+	     int csize,
+	     void *debug,
+	     int nr_debug_entries)
 {
 	int ret, fd;
 
@@ -76,7 +78,7 @@ jit_emit_elf(char *filename,
 		return -1;
 	}
 
-        ret = jit_write_elf(fd, code_addr, sym, (const void *)code, csize);
+        ret = jit_write_elf(fd, code_addr, sym, (const void *)code, csize, debug, nr_debug_entries);
 
         close(fd);
 
@@ -347,7 +349,7 @@ static int jit_repipe_code_load(struct jit_buf_desc *jd, union jr_entry *jr)
 
 	size = PERF_ALIGN(size, sizeof(u64));
 	uaddr = (uintptr_t)code;
-	ret = jit_emit_elf(filename, sym, addr, (const void *)uaddr, csize);
+	ret = jit_emit_elf(filename, sym, addr, (const void *)uaddr, csize, jd->debug_data, jd->nr_debug_entries);
 
 	if (jd->debug_data && jd->nr_debug_entries) {
 		free(jd->debug_data);