perf tools: Encode kernel module mappings in perf.data

We were always looking at the running machine /proc/modules,
even when processing a perf.data file, which only makes sense
when we're doing 'perf record' and 'perf report' on the same
machine, and in close sucession, or if we don't use modules at
all, right Peter? ;-)

Now, at 'perf record' time we read /proc/modules, find the long
path for modules, and put them as PERF_MMAP events, just like we
did to encode the reloc reference symbol for vmlinux. Talking
about that now it is encoded in .pgoff, so that we can use
.{start,len} to store the address boundaries for the kernel so
that when we reconstruct the kmaps tree we can do lookups right
away, without having to fixup the end of the kernel maps like we
did in the past (and now only in perf record).

One more step in the 'perf archive' direction when we'll finally
be able to collect data in one machine and analyse in another.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1263396139-4798-1-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

9 files changed, 212 insertions, 48 deletions

diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c
index 88c570c18e3..4af7199c5af 100644
--- a/tools/perf/builtin-kmem.c
+++ b/tools/perf/builtin-kmem.c
@@ -494,6 +494,11 @@ static int __cmd_kmem(void)
 	if (!perf_session__has_traces(session, "kmem record"))
 		goto out_delete;
 
+	if (perf_session__create_kernel_maps(session) < 0) {
+		pr_err("Problems creating kernel maps\n");
+		return -1;
+	}
+
 	setup_pager();
 	err = perf_session__process_events(session, &event_ops);
 	if (err != 0)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 8f88420e066..c130df2676f 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -465,6 +465,11 @@ static int __cmd_record(int argc, const char **argv)
 		return -1;
 	}
 
+	if (perf_session__create_kernel_maps(session) < 0) {
+		pr_err("Problems creating kernel maps\n");
+		return -1;
+	}
+
 	if (!file_new) {
 		err = perf_header__read(&session->header, output);
 		if (err < 0)
@@ -558,6 +563,12 @@ static int __cmd_record(int argc, const char **argv)
 		return err;
 	}
 
+	err = event__synthesize_modules(process_synthesized_event, session);
+	if (err < 0) {
+		pr_err("Couldn't record kernel reference relocation symbol.\n");
+		return err;
+	}
+
 	if (!system_wide && profile_cpu == -1)
 		event__synthesize_thread(pid, process_synthesized_event,
 					 session);
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index ddc584b6487..6822b44ca4f 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -1165,6 +1165,11 @@ static int __cmd_top(void)
 	if (session == NULL)
 		return -ENOMEM;
 
+	if (perf_session__create_kernel_maps(session) < 0) {
+		pr_err("Problems creating kernel maps\n");
+		return -1;
+	}
+
 	if (target_pid != -1)
 		event__synthesize_thread(target_pid, event__process, session);
 	else
diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index bfb3d872b9f..4f3e7ef33b8 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -154,6 +154,36 @@ static int event__synthesize_mmap_events(pid_t pid, pid_t tgid,
 	return 0;
 }
 
+int event__synthesize_modules(event__handler_t process,
+			      struct perf_session *session)
+{
+	struct rb_node *nd;
+
+	for (nd = rb_first(&session->kmaps.maps[MAP__FUNCTION]);
+	     nd; nd = rb_next(nd)) {
+		event_t ev;
+		size_t size;
+		struct map *pos = rb_entry(nd, struct map, rb_node);
+
+		if (pos->dso->kernel)
+			continue;
+
+		size = ALIGN(pos->dso->long_name_len + 1, sizeof(u64));
+		memset(&ev, 0, sizeof(ev));
+		ev.mmap.header.type = PERF_RECORD_MMAP;
+		ev.mmap.header.size = (sizeof(ev.mmap) -
+				        (sizeof(ev.mmap.filename) - size));
+		ev.mmap.start = pos->start;
+		ev.mmap.len   = pos->end - pos->start;
+
+		memcpy(ev.mmap.filename, pos->dso->long_name,
+		       pos->dso->long_name_len + 1);
+		process(&ev, session);
+	}
+
+	return 0;
+}
+
 int event__synthesize_thread(pid_t pid, event__handler_t process,
 			     struct perf_session *session)
 {
@@ -222,7 +252,9 @@ int event__synthesize_kernel_mmap(event__handler_t process,
 			"[kernel.kallsyms.%s]", symbol_name) + 1;
 	size = ALIGN(size, sizeof(u64));
 	ev.mmap.header.size = (sizeof(ev.mmap) - (sizeof(ev.mmap.filename) - size));
-	ev.mmap.start = args.start;
+	ev.mmap.pgoff = args.start;
+	ev.mmap.start = session->vmlinux_maps[MAP__FUNCTION]->start;
+	ev.mmap.len   = session->vmlinux_maps[MAP__FUNCTION]->end - ev.mmap.start ;
 
 	return process(&ev, session);
 }
@@ -280,7 +312,6 @@ int event__process_mmap(event_t *self, struct perf_session *session)
 {
 	struct thread *thread;
 	struct map *map;
-	static const char kmmap_prefix[] = "[kernel.kallsyms.";
 
 	dump_printf(" %d/%d: [%p(%p) @ %p]: %s\n",
 		    self->mmap.pid, self->mmap.tid,
@@ -289,13 +320,61 @@ int event__process_mmap(event_t *self, struct perf_session *session)
 		    (void *)(long)self->mmap.pgoff,
 		    self->mmap.filename);
 
-	if (self->mmap.pid == 0 &&
-	    memcmp(self->mmap.filename, kmmap_prefix,
-		   sizeof(kmmap_prefix) - 1) == 0) {
-		const char *symbol_name = (self->mmap.filename +
-					   sizeof(kmmap_prefix) - 1);
-		perf_session__set_kallsyms_ref_reloc_sym(session, symbol_name,
-							 self->mmap.start);
+	if (self->mmap.pid == 0) {
+		static const char kmmap_prefix[] = "[kernel.kallsyms.";
+
+		if (self->mmap.filename[0] == '/') {
+			char short_module_name[1024];
+			char *name = strrchr(self->mmap.filename, '/'), *dot;
+
+			if (name == NULL)
+				goto out_problem;
+
+			++name; /* skip / */
+			dot = strrchr(name, '.');
+			if (dot == NULL)
+				goto out_problem;
+
+			snprintf(short_module_name, sizeof(short_module_name),
+				 "[%.*s]", (int)(dot - name), name);
+			strxfrchar(short_module_name, '-', '_');
+
+			map = perf_session__new_module_map(session,
+							   self->mmap.start,
+							   short_module_name);
+			if (map == NULL)
+				goto out_problem;
+
+			name = strdup(self->mmap.filename);
+			if (name == NULL)
+				goto out_problem;
+
+			dso__set_long_name(map->dso, name);
+			map->end = map->start + self->mmap.len;
+		} else if (memcmp(self->mmap.filename, kmmap_prefix,
+				sizeof(kmmap_prefix) - 1) == 0) {
+			const char *symbol_name = (self->mmap.filename +
+						   sizeof(kmmap_prefix) - 1);
+			/*
+			 * Should be there already, from the build-id table in
+			 * the header.
+			 */
+			struct dso *kernel = __dsos__findnew(&dsos__kernel,
+							     "[kernel.kallsyms]");
+			if (kernel == NULL)
+				goto out_problem;
+
+			if (__map_groups__create_kernel_maps(&session->kmaps,
+							     session->vmlinux_maps,
+							     kernel) < 0)
+				goto out_problem;
+
+			session->vmlinux_maps[MAP__FUNCTION]->start = self->mmap.start;
+			session->vmlinux_maps[MAP__FUNCTION]->end   = self->mmap.start + self->mmap.len;
+
+			perf_session__set_kallsyms_ref_reloc_sym(session, symbol_name,
+								 self->mmap.pgoff);
+		}
 		return 0;
 	}
 
@@ -304,10 +383,13 @@ int event__process_mmap(event_t *self, struct perf_session *session)
 		       session->cwd, session->cwdlen);
 
 	if (thread == NULL || map == NULL)
-		dump_printf("problem processing PERF_RECORD_MMAP, skipping event.\n");
-	else
-		thread__insert_map(thread, map);
+		goto out_problem;
+
+	thread__insert_map(thread, map);
+	return 0;
 
+out_problem:
+	dump_printf("problem processing PERF_RECORD_MMAP, skipping event.\n");
 	return 0;
 }
 
diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h
index 80356da8216..50a7132887f 100644
--- a/tools/perf/util/event.h
+++ b/tools/perf/util/event.h
@@ -112,6 +112,8 @@ void event__synthesize_threads(event__handler_t process,
 int event__synthesize_kernel_mmap(event__handler_t process,
 				  struct perf_session *session,
 				  const char *symbol_name);
+int event__synthesize_modules(event__handler_t process,
+			      struct perf_session *session);
 
 int event__process_comm(event_t *self, struct perf_session *session);
 int event__process_lost(event_t *self, struct perf_session *session);
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index 378ac5422bc..fd1c5a39a5b 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -69,9 +69,6 @@ struct perf_session *perf_session__new(const char *filename, int mode, bool forc
 	self->unknown_events = 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;
 
@@ -268,8 +265,11 @@ int perf_header__read_build_ids(int input, u64 offset, u64 size)
 			head = &dsos__kernel;
 
 		dso = __dsos__findnew(head, filename);
-		if (dso != NULL)
+		if (dso != NULL) {
 			dso__set_build_id(dso, &bev.build_id);
+			if (head == &dsos__kernel && filename[0] == '[')
+				dso->kernel = 1;
+		}
 
 		offset += bev.header.size;
 	}
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index 8e6627e6b77..381999dd5c1 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -161,7 +161,7 @@ static size_t symbol__fprintf(struct symbol *self, FILE *fp)
 		       self->start, self->end, self->name);
 }
 
-static void dso__set_long_name(struct dso *self, char *name)
+void dso__set_long_name(struct dso *self, char *name)
 {
 	if (name == NULL)
 		return;
@@ -176,7 +176,7 @@ static void dso__set_basename(struct dso *self)
 
 struct dso *dso__new(const char *name)
 {
-	struct dso *self = malloc(sizeof(*self) + strlen(name) + 1);
+	struct dso *self = zalloc(sizeof(*self) + strlen(name) + 1);
 
 	if (self != NULL) {
 		int i;
@@ -500,13 +500,17 @@ static int dso__split_kallsyms(struct dso *self, struct map *map,
 
 			*module++ = '\0';
 
-			if (strcmp(self->name, module)) {
+			if (strcmp(curr_map->dso->short_name, module)) {
 				curr_map = map_groups__find_by_name(&session->kmaps, map->type, module);
 				if (curr_map == NULL) {
 					pr_debug("/proc/{kallsyms,modules} "
-					         "inconsistency!\n");
+					         "inconsistency while looking "
+						 "for \"%s\" module!\n", module);
 					return -1;
 				}
+
+				if (curr_map->dso->loaded)
+					goto discard_symbol;
 			}
 			/*
 			 * So that we look just like we get from .ko files,
@@ -1343,13 +1347,33 @@ struct map *map_groups__find_by_name(struct map_groups *self,
 	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)
+		if (map->dso && strcmp(map->dso->short_name, name) == 0)
 			return map;
 	}
 
 	return NULL;
 }
 
+static int dso__kernel_module_get_build_id(struct dso *self)
+{
+	char filename[PATH_MAX];
+	/*
+	 * kernel module short names are of the form "[module]" and
+	 * we need just "module" here.
+	 */
+	const char *name = self->short_name + 1;
+
+	snprintf(filename, sizeof(filename),
+		 "/sys/module/%.*s/notes/.note.gnu.build-id",
+		 (int)strlen(name - 1), name);
+
+	if (sysfs__read_build_id(filename, self->build_id,
+				 sizeof(self->build_id)) == 0)
+		self->has_build_id = true;
+
+	return 0;
+}
+
 static int perf_session__set_modules_path_dir(struct perf_session *self, char *dirname)
 {
 	struct dirent *dent;
@@ -1395,6 +1419,7 @@ static int perf_session__set_modules_path_dir(struct perf_session *self, char *d
 			if (long_name == NULL)
 				goto failure;
 			dso__set_long_name(map->dso, long_name);
+			dso__kernel_module_get_build_id(map->dso);
 		}
 	}
 
@@ -1437,6 +1462,24 @@ static struct map *map__new2(u64 start, struct dso *dso, enum map_type type)
 	return self;
 }
 
+struct map *perf_session__new_module_map(struct perf_session *self, u64 start,
+					 const char *filename)
+{
+	struct map *map;
+	struct dso *dso = __dsos__findnew(&dsos__kernel, filename);
+
+	if (dso == NULL)
+		return NULL;
+
+	map = map__new2(start, dso, MAP__FUNCTION);
+	if (map == NULL)
+		return NULL;
+
+	dso->origin = DSO__ORIG_KMODULE;
+	map_groups__insert(&self->kmaps, map);
+	return map;
+}
+
 static int perf_session__create_module_maps(struct perf_session *self)
 {
 	char *line = NULL;
@@ -1450,7 +1493,6 @@ static int perf_session__create_module_maps(struct perf_session *self)
 	while (!feof(file)) {
 		char name[PATH_MAX];
 		u64 start;
-		struct dso *dso;
 		char *sep;
 		int line_len;
 
@@ -1476,26 +1518,10 @@ static int perf_session__create_module_maps(struct perf_session *self)
 		*sep = '\0';
 
 		snprintf(name, sizeof(name), "[%s]", line);
-		dso = dso__new(name);
-
-		if (dso == NULL)
-			goto out_delete_line;
-
-		map = map__new2(start, dso, MAP__FUNCTION);
-		if (map == NULL) {
-			dso__delete(dso);
+		map = perf_session__new_module_map(self, start, name);
+		if (map == NULL)
 			goto out_delete_line;
-		}
-
-		snprintf(name, sizeof(name),
-			 "/sys/module/%s/notes/.note.gnu.build-id", line);
-		if (sysfs__read_build_id(name, dso->build_id,
-					 sizeof(dso->build_id)) == 0)
-			dso->has_build_id = true;
-
-		dso->origin = DSO__ORIG_KMODULE;
-		map_groups__insert(&self->kmaps, map);
-		dsos__add(&dsos__kernel, dso);
+		dso__kernel_module_get_build_id(map->dso);
 	}
 
 	free(line);
@@ -1573,10 +1599,28 @@ static int dso__load_kernel_sym(struct dso *self, struct map *map,
 		}
 	}
 
+	/*
+	 * Say the kernel DSO was created when processing the build-id header table,
+	 * we have a build-id, so check if it is the same as the running kernel,
+	 * using it if it is.
+	 */
+	if (self->has_build_id) {
+		u8 kallsyms_build_id[BUILD_ID_SIZE];
+
+		if (sysfs__read_build_id("/sys/kernel/notes", kallsyms_build_id,
+					 sizeof(kallsyms_build_id)) == 0)
+
+		is_kallsyms = dso__build_id_equal(self, kallsyms_build_id);
+		if (is_kallsyms)
+			goto do_kallsyms;
+		goto do_vmlinux;
+	}
+
 	is_kallsyms = self->long_name[0] == '[';
 	if (is_kallsyms)
 		goto do_kallsyms;
 
+do_vmlinux:
 	err = dso__load_vmlinux(self, map, session, self->long_name, filter);
 	if (err <= 0) {
 		pr_info("The file %s cannot be used, "
@@ -1694,16 +1738,12 @@ out_delete_kernel_dso:
 	return NULL;
 }
 
-static int map_groups__create_kernel_maps(struct map_groups *self,
-					  struct map *vmlinux_maps[MAP__NR_TYPES],
-					  const char *vmlinux)
+int __map_groups__create_kernel_maps(struct map_groups *self,
+				     struct map *vmlinux_maps[MAP__NR_TYPES],
+				     struct dso *kernel)
 {
-	struct dso *kernel = dsos__create_kernel(vmlinux);
 	enum map_type type;
 
-	if (kernel == NULL)
-		return -1;
-
 	for (type = 0; type < MAP__NR_TYPES; ++type) {
 		vmlinux_maps[type] = map__new2(0, kernel, type);
 		if (vmlinux_maps[type] == NULL)
@@ -1717,6 +1757,18 @@ static int map_groups__create_kernel_maps(struct map_groups *self,
 	return 0;
 }
 
+static int map_groups__create_kernel_maps(struct map_groups *self,
+					  struct map *vmlinux_maps[MAP__NR_TYPES],
+					  const char *vmlinux)
+{
+	struct dso *kernel = dsos__create_kernel(vmlinux);
+
+	if (kernel == NULL)
+		return -1;
+
+	return __map_groups__create_kernel_maps(self, vmlinux_maps, kernel);
+}
+
 static void vmlinux_path__exit(void)
 {
 	while (--vmlinux_path__nr_entries >= 0) {
diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h
index ee0b4593db7..594156e43b1 100644
--- a/tools/perf/util/symbol.h
+++ b/tools/perf/util/symbol.h
@@ -134,6 +134,7 @@ size_t dsos__fprintf_buildid(FILE *fp);
 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_long_name(struct dso *self, char *name);
 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,
@@ -151,5 +152,7 @@ bool symbol_type__is_a(char symbol_type, enum map_type map_type);
 
 int perf_session__create_kernel_maps(struct perf_session *self);
 
+struct map *perf_session__new_module_map(struct perf_session *self, u64 start,
+					 const char *filename);
 extern struct dso *vdso;
 #endif /* __PERF_SYMBOL */
diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h
index c206f72c888..c06c13535a7 100644
--- a/tools/perf/util/thread.h
+++ b/tools/perf/util/thread.h
@@ -67,4 +67,8 @@ map_groups__find_function(struct map_groups *self, struct perf_session *session,
 
 struct map *map_groups__find_by_name(struct map_groups *self,
 				     enum map_type type, const char *name);
+
+int __map_groups__create_kernel_maps(struct map_groups *self,
+				     struct map *vmlinux_maps[MAP__NR_TYPES],
+				     struct dso *kernel);
 #endif	/* __PERF_THREAD_H */