Merge "simpleperf: use path in binary_cache in pprof_proto_generator.py."

3 files changed, 46 insertions, 32 deletions

diff --git a/simpleperf/scripts/pprof_proto_generator.py b/simpleperf/scripts/pprof_proto_generator.py
index fc3ef357..2a4e1d2a 100755
--- a/simpleperf/scripts/pprof_proto_generator.py
+++ b/simpleperf/scripts/pprof_proto_generator.py
@@ -281,7 +281,9 @@ class PprofProfileGenerator(object):
         self.mapping_list = []
         self.function_map = {}
         self.function_list = []
-        self.build_id_map = {}  # map from filename to build_id
+
+        # Map from dso_name in perf.data to (binary path, build_id).
+        self.binary_map = {}
         self.read_elf = ReadElf(self.config['ndk_path'])
 
     def gen(self):
@@ -375,10 +377,10 @@ class PprofProfileGenerator(object):
         return sample_type_id
 
     def get_location_id(self, ip, symbol):
-        mapping_id = self.get_mapping_id(symbol.mapping[0], symbol.dso_name)
+        binary_path, build_id = self.get_binary(symbol.dso_name)
+        mapping_id = self.get_mapping_id(symbol.mapping[0], binary_path, build_id)
         location = Location(mapping_id, ip, symbol.vaddr_in_file)
-        function_id = self.get_function_id(symbol.symbol_name, symbol.dso_name,
-                                           symbol.symbol_addr)
+        function_id = self.get_function_id(symbol.symbol_name, binary_path, symbol.symbol_addr)
         if function_id:
             # Add Line only when it has a valid function id, see http://b/36988814.
             # Default line info only contains the function name
@@ -395,9 +397,8 @@ class PprofProfileGenerator(object):
         self.location_map[location.key] = location
         return location.id
 
-    def get_mapping_id(self, report_mapping, filename):
+    def get_mapping_id(self, report_mapping, filename, build_id):
         filename_id = self.get_string_id(filename)
-        build_id = self.get_build_id(filename)
         build_id_id = self.get_string_id(build_id)
         mapping = Mapping(report_mapping.start, report_mapping.end,
                           report_mapping.pgoff, filename_id, build_id_id)
@@ -410,28 +411,36 @@ class PprofProfileGenerator(object):
         self.mapping_map[mapping.key] = mapping
         return mapping.id
 
-    def get_build_id(self, filename):
-        build_id = self.build_id_map.get(filename)
-        if build_id is None:
-            build_id = ''
-            # The build ids in perf.data are padded to 20 bytes, but pprof needs without padding.
-            # So read build id from the binary in binary_cache, and check it with build id in
-            # perf.data.
-            build_id_in_perf_data = self.lib.GetBuildIdForPath(filename)
-            if build_id_in_perf_data:
-                elf_path = find_real_dso_path(filename, self.config['binary_cache_dir'])
-                if elf_path:
-                    elf_build_id = self.read_elf.get_build_id(elf_path, False)
-                    if build_id_in_perf_data == self.read_elf.pad_build_id(elf_build_id):
-                        build_id = elf_build_id
-                if not build_id and build_id_in_perf_data.startswith('0x'):
-                    # Fallback to the way used by TrimZeroesFromBuildIDString() in quipper.
-                    build_id = build_id_in_perf_data[2:]  # remove '0x'
-                    padding = '0' * 8
-                    while build_id.endswith(padding):
-                        build_id = build_id[:-len(padding)]
-            self.build_id_map[filename] = build_id
-        return build_id
+    def get_binary(self, dso_name):
+        """ Return (binary_path, build_id) for a given dso_name. """
+        value = self.binary_map.get(dso_name)
+        if value:
+            return value
+
+        binary_path = dso_name
+        build_id = ''
+
+        # The build ids in perf.data are padded to 20 bytes, but pprof needs without padding.
+        # So read build id from the binary in binary_cache, and check it with build id in
+        # perf.data.
+        build_id_in_perf_data = self.lib.GetBuildIdForPath(dso_name)
+        if build_id_in_perf_data:
+            # Try elf_path in binary cache.
+            elf_path = find_real_dso_path(dso_name, self.config['binary_cache_dir'])
+            if elf_path:
+                elf_build_id = self.read_elf.get_build_id(elf_path, False)
+                if build_id_in_perf_data == self.read_elf.pad_build_id(elf_build_id):
+                    build_id = elf_build_id
+                    binary_path = elf_path
+
+            if not build_id and build_id_in_perf_data.startswith('0x'):
+                # Fallback to the way used by TrimZeroesFromBuildIDString() in quipper.
+                build_id = build_id_in_perf_data[2:]  # remove '0x'
+                padding = '0' * 8
+                while build_id.endswith(padding):
+                    build_id = build_id[:-len(padding)]
+        self.binary_map[dso_name] = (binary_path, build_id)
+        return (binary_path, build_id)
 
     def get_mapping(self, mapping_id):
         return self.mapping_list[mapping_id - 1] if mapping_id > 0 else None
@@ -468,7 +477,9 @@ class PprofProfileGenerator(object):
         if not find_tool_path('llvm-symbolizer', self.config['ndk_path']):
             log_info("Can't generate line information because can't find llvm-symbolizer.")
             return
-        addr2line = Addr2Nearestline(self.config['ndk_path'], self.config['binary_cache_dir'], True)
+        # We have changed dso names to paths in binary_cache in self.get_binary(). So no need to
+        # pass binary_cache_dir to addr2line.
+        addr2line = Addr2Nearestline(self.config['ndk_path'], None, True)
 
         # 2. Put all needed addresses to it.
         for location in self.location_list:
diff --git a/simpleperf/scripts/test.py b/simpleperf/scripts/test.py
index c5ecd93e..91f5f044 100755
--- a/simpleperf/scripts/test.py
+++ b/simpleperf/scripts/test.py
@@ -578,8 +578,13 @@ class TestExampleWithNative(TestExampleBase):
              "__start_thread"])
 
     def test_pprof_proto_generator(self):
+        check_strings_with_lines = [
+            "native-lib.cpp",
+            "BusyLoopThread",
+            # Check if dso name in perf.data is replaced by binary path in binary_cache.
+            'filename: binary_cache/data/app/com.example.simpleperf.simpleperfexamplewithnative-']
         self.common_test_pprof_proto_generator(
-            check_strings_with_lines=["native-lib.cpp", "BusyLoopThread"],
+            check_strings_with_lines,
             check_strings_without_lines=["BusyLoopThread"])
 
     def test_inferno(self):
diff --git a/simpleperf/scripts/utils.py b/simpleperf/scripts/utils.py
index 4e2855e3..e0d40b24 100644
--- a/simpleperf/scripts/utils.py
+++ b/simpleperf/scripts/utils.py
@@ -380,8 +380,6 @@ def is_elf_file(path):
 
 def find_real_dso_path(dso_path_in_record_file, binary_cache_path):
     """ Given the path of a shared library in perf.data, find its real path in the file system. """
-    if dso_path_in_record_file[0] != '/' or dso_path_in_record_file == '//anon':
-        return None
     if binary_cache_path:
         tmp_path = os.path.join(binary_cache_path, dso_path_in_record_file[1:])
         if is_elf_file(tmp_path):