Merge "simpleperf: add --binary_filter option in report_html.py."

am: f2307ae2a0

Change-Id: I5f9e6ad5f57f697b9d55bc1870523b9f2a0672a1

4 files changed, 115 insertions, 72 deletions

diff --git a/simpleperf/scripts/report_html.js b/simpleperf/scripts/report_html.js
index 4fc2a4a3..720f72f8 100644
--- a/simpleperf/scripts/report_html.js
+++ b/simpleperf/scripts/report_html.js
@@ -602,7 +602,7 @@ class SampleTableWeightSelectorView {
             if (e.clickEvent) {
                 let button = $(e.clickEvent.target);
                 let newOption = button.attr('key');
-                if (this.curOption != newOption) {
+                if (newOption && this.curOption != newOption) {
                     this.curOption = newOption;
                     divContainer.find(`#${id}`).text(options.get(this.curOption));
                     onSelectChange();
@@ -638,7 +638,7 @@ class SampleTableView {
         this.div = $('<div>', {id: this.id}).appendTo(divContainer);
         this.eventInfo = eventInfo;
         this.selectorView = null;
-        this.tableView = null;
+        this.tableDiv = null;
     }
 
     drawAsync(totalProgress) {
@@ -647,6 +647,7 @@ class SampleTableView {
                 this.div.empty();
                 this.selectorView = new SampleTableWeightSelectorView(
                     this.div, this.eventInfo, () => this.onSampleWeightChange());
+                this.tableDiv = $('<div>').appendTo(this.div);
             }))
             .then(() => this._drawSampleTable(totalProgress));
     }
@@ -657,7 +658,7 @@ class SampleTableView {
         let data = [];
         return createPromise()
             .then(wait(() => {
-                this.div.find('table').remove();
+                this.tableDiv.empty();
                 let getSampleWeight = this.selectorView.getSampleWeightFunction();
                 let sampleWeightSuffix = this.selectorView.getSampleWeightSuffix();
                 // Draw a table of 'Total', 'Self', 'Samples', 'Process', 'Thread', 'Library',
@@ -665,7 +666,7 @@ class SampleTableView {
                 let valueSuffix = sampleWeightSuffix.length > 0 ? `(in${sampleWeightSuffix})` : '';
                 let titles = ['Total' + valueSuffix, 'Self' + valueSuffix, 'Samples', 'Process',
                               'Thread', 'Library', 'Function', 'HideKey'];
-                this.div.append(`
+                this.tableDiv.append(`
                     <table cellspacing="0" class="table table-striped table-bordered"
                         style="width:100%">
                         <thead>${getTableRow(titles, 'th')}</thead>
@@ -691,7 +692,7 @@ class SampleTableView {
             }))
             .then(addProgress(totalProgress / 2))
             .then(wait(() => {
-                let table = this.div.find('table');
+                let table = this.tableDiv.find('table');
                 let dataTable = table.DataTable({
                     lengthMenu: [10, 20, 50, 100, -1],
                     order: [0, 'desc'],
@@ -1039,7 +1040,7 @@ class SampleWeightSelectorView {
             if (e.clickEvent) {
                 let button = $(e.clickEvent.target);
                 let newOption = button.attr('key');
-                if (this.curOption != newOption) {
+                if (newOption && this.curOption != newOption) {
                     this.curOption = newOption;
                     divContainer.find(`#${id}`).text(options.get(this.curOption));
                     onSelectChange();
diff --git a/simpleperf/scripts/report_html.py b/simpleperf/scripts/report_html.py
index c28a4299..c3ab0df2 100755
--- a/simpleperf/scripts/report_html.py
+++ b/simpleperf/scripts/report_html.py
@@ -635,7 +635,7 @@ class RecordData(object):
             self.events[event_name] = EventScope(event_name)
         return self.events[event_name]
 
-    def add_source_code(self, source_dirs):
+    def add_source_code(self, source_dirs, filter_lib):
         """ Collect source code information:
             1. Find line ranges for each function in FunctionSet.
             2. Find line for each addr in FunctionScope.addr_hit_map.
@@ -647,17 +647,19 @@ class RecordData(object):
             if function.func_name == 'unknown':
                 continue
             lib_name = self.libs.get_lib_name(function.lib_id)
-            addr2line.add_addr(lib_name, function.start_addr, function.start_addr)
-            addr2line.add_addr(lib_name, function.start_addr,
-                               function.start_addr + function.addr_len - 1)
+            if filter_lib(lib_name):
+                addr2line.add_addr(lib_name, function.start_addr, function.start_addr)
+                addr2line.add_addr(lib_name, function.start_addr,
+                                   function.start_addr + function.addr_len - 1)
         # Request line for each addr in FunctionScope.addr_hit_map.
         for event in self.events.values():
             for lib in event.libraries:
                 lib_name = self.libs.get_lib_name(lib.lib_id)
-                for function in lib.functions.values():
-                    func_addr = self.functions.id_to_func[function.func_id].start_addr
-                    for addr in function.addr_hit_map:
-                        addr2line.add_addr(lib_name, func_addr, addr)
+                if filter_lib(lib_name):
+                    for function in lib.functions.values():
+                        func_addr = self.functions.id_to_func[function.func_id].start_addr
+                        for addr in function.addr_hit_map:
+                            addr2line.add_addr(lib_name, func_addr, addr)
         addr2line.convert_addrs_to_lines()
 
         # Set line range for each function.
@@ -665,6 +667,8 @@ class RecordData(object):
             if function.func_name == 'unknown':
                 continue
             dso = addr2line.get_dso(self.libs.get_lib_name(function.lib_id))
+            if not dso:
+                continue
             start_source = addr2line.get_addr_source(dso, function.start_addr)
             end_source = addr2line.get_addr_source(dso, function.start_addr + function.addr_len - 1)
             if not start_source or not end_source:
@@ -681,6 +685,8 @@ class RecordData(object):
         for event in self.events.values():
             for lib in event.libraries:
                 dso = addr2line.get_dso(self.libs.get_lib_name(lib.lib_id))
+                if not dso:
+                    continue
                 for function in lib.functions.values():
                     for addr in function.addr_hit_map:
                         source = addr2line.get_addr_source(dso, addr)
@@ -697,18 +703,29 @@ class RecordData(object):
         # Collect needed source code in SourceFileSet.
         self.source_files.load_source_code(source_dirs)
 
-    def add_disassembly(self):
+    def add_disassembly(self, filter_lib):
         """ Collect disassembly information:
             1. Use objdump to collect disassembly for each function in FunctionSet.
             2. Set flag to dump addr_hit_map when generating record info.
         """
         objdump = Objdump(self.ndk_path, self.binary_cache_path)
-        for function in self.functions.id_to_func.values():
+        cur_lib_name = None
+        dso_info = None
+        for function in sorted(self.functions.id_to_func.values(), key=lambda a: a.lib_id):
             if function.func_name == 'unknown':
                 continue
             lib_name = self.libs.get_lib_name(function.lib_id)
-            code = objdump.disassemble_code(lib_name, function.start_addr, function.addr_len)
-            function.disassembly = code
+            if lib_name != cur_lib_name:
+                cur_lib_name = lib_name
+                if filter_lib(lib_name):
+                    dso_info = objdump.get_dso_info(lib_name)
+                else:
+                    dso_info = None
+                if dso_info:
+                    log_info('Disassemble %s' % dso_info[0])
+            if dso_info:
+                code = objdump.disassemble_code(dso_info, function.start_addr, function.addr_len)
+                function.disassembly = code
 
         self.gen_addr_hit_map_in_record_info = True
 
@@ -866,6 +883,8 @@ def main():
     parser.add_argument('--add_source_code', action='store_true', help='Add source code.')
     parser.add_argument('--source_dirs', nargs='+', help='Source code directories.')
     parser.add_argument('--add_disassembly', action='store_true', help='Add disassembled code.')
+    parser.add_argument('--binary_filter', nargs='+', help="""Annotate source code and disassembly
+                        only for selected binaries.""")
     parser.add_argument('--ndk_path', nargs=1, help='Find tools in the ndk path.')
     parser.add_argument('--no_browser', action='store_true', help="Don't open report in browser.")
     parser.add_argument('--show_art_frames', action='store_true',
@@ -892,10 +911,18 @@ def main():
     for record_file in args.record_file:
         record_data.load_record_file(record_file, args.show_art_frames)
     record_data.limit_percents(args.min_func_percent, args.min_callchain_percent)
+
+    def filter_lib(lib_name):
+        if not args.binary_filter:
+            return True
+        for binary in args.binary_filter:
+            if binary in lib_name:
+                return True
+        return False
     if args.add_source_code:
-        record_data.add_source_code(args.source_dirs)
+        record_data.add_source_code(args.source_dirs, filter_lib)
     if args.add_disassembly:
-        record_data.add_disassembly()
+        record_data.add_disassembly(filter_lib)
 
     # 3. Generate report html.
     report_generator = ReportGenerator(args.report_path)
diff --git a/simpleperf/scripts/test.py b/simpleperf/scripts/test.py
index 7eed5112..8424085b 100644
--- a/simpleperf/scripts/test.py
+++ b/simpleperf/scripts/test.py
@@ -577,6 +577,8 @@ class TestExampleWithNative(TestExampleBase):
 
     def test_report_html(self):
         self.common_test_report_html()
+        self.run_cmd(['report_html.py', '--add_source_code', '--source_dirs', 'testdata',
+                      '--add_disassembly', '--binary_filter', "libnative-lib.so"])
 
 
 class TestExampleWithNativeRoot(TestExampleBase):
@@ -1063,11 +1065,12 @@ class TestTools(unittest.TestCase):
         }
         objdump = Objdump(None, binary_cache_path)
         for dso_path in test_map:
-            dso_info = test_map[dso_path]
-            disassemble_code = objdump.disassemble_code(dso_path, dso_info['start_addr'],
-                                                        dso_info['len'])
+            dso = test_map[dso_path]
+            dso_info = objdump.get_dso_info(dso_path)
+            self.assertIsNotNone(dso_info)
+            disassemble_code = objdump.disassemble_code(dso_info, dso['start_addr'], dso['len'])
             self.assertTrue(disassemble_code)
-            for item in dso_info['expected_items']:
+            for item in dso['expected_items']:
                 self.assertTrue(item in disassemble_code)
 
     def test_readelf(self):
diff --git a/simpleperf/scripts/utils.py b/simpleperf/scripts/utils.py
index 81f29a33..22554830 100644
--- a/simpleperf/scripts/utils.py
+++ b/simpleperf/scripts/utils.py
@@ -354,6 +354,13 @@ def open_report_in_browser(report_path):
         # webbrowser.get() doesn't work well on darwin/windows.
         webbrowser.open_new_tab(report_path)
 
+def is_elf_file(path):
+    if os.path.isfile(path):
+        with open(path, 'rb') as fh:
+            data = fh.read(4)
+            if len(data) == 4 and bytes_to_str(data) == '\x7fELF':
+                return True
+    return False
 
 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. """
@@ -361,9 +368,9 @@ def find_real_dso_path(dso_path_in_record_file, binary_cache_path):
         return None
     if binary_cache_path:
         tmp_path = os.path.join(binary_cache_path, dso_path_in_record_file[1:])
-        if os.path.isfile(tmp_path):
+        if is_elf_file(tmp_path):
             return tmp_path
-    if os.path.isfile(dso_path_in_record_file):
+    if is_elf_file(dso_path_in_record_file):
         return dso_path_in_record_file
     return None
 
@@ -559,19 +566,20 @@ class Objdump(object):
         self.readelf = ReadElf(ndk_path)
         self.objdump_paths = {}
 
-    def disassemble_code(self, dso_path, start_addr, addr_len):
-        """ Disassemble [start_addr, start_addr + addr_len] of dso_path.
-            Return a list of pair (disassemble_code_line, addr).
-        """
-        # 1. Find real path.
+    def get_dso_info(self, dso_path):
         real_path = find_real_dso_path(dso_path, self.binary_cache_path)
-        if real_path is None:
+        if not real_path:
             return None
-
-        # 2. Get path of objdump.
         arch = self.readelf.get_arch(real_path)
         if arch == 'unknown':
             return None
+        return (real_path, arch)
+
+    def disassemble_code(self, dso_info, start_addr, addr_len):
+        """ Disassemble [start_addr, start_addr + addr_len] of dso_path.
+            Return a list of pair (disassemble_code_line, addr).
+        """
+        real_path, arch = dso_info
         objdump_path = self.objdump_paths.get(arch)
         if not objdump_path:
             objdump_path = find_tool_path('objdump', self.ndk_path, arch)
@@ -614,53 +622,57 @@ class ReadElf(object):
 
     def get_arch(self, elf_file_path):
         """ Get arch of an elf file. """
-        try:
-            output = subprocess.check_output([self.readelf_path, '-h', elf_file_path])
-            if output.find('AArch64') != -1:
-                return 'arm64'
-            if output.find('ARM') != -1:
-                return 'arm'
-            if output.find('X86-64') != -1:
-                return 'x86_64'
-            if output.find('80386') != -1:
-                return 'x86'
-        except subprocess.CalledProcessError:
-            pass
+        if is_elf_file(elf_file_path):
+            try:
+                output = subprocess.check_output([self.readelf_path, '-h', elf_file_path])
+                output = bytes_to_str(output)
+                if output.find('AArch64') != -1:
+                    return 'arm64'
+                if output.find('ARM') != -1:
+                    return 'arm'
+                if output.find('X86-64') != -1:
+                    return 'x86_64'
+                if output.find('80386') != -1:
+                    return 'x86'
+            except subprocess.CalledProcessError:
+                pass
         return 'unknown'
 
     def get_build_id(self, elf_file_path):
         """ Get build id of an elf file. """
-        try:
-            output = subprocess.check_output([self.readelf_path, '-n', elf_file_path])
-            output = bytes_to_str(output)
-            result = re.search(r'Build ID:\s*(\S+)', output)
-            if result:
-                build_id = result.group(1)
-                if len(build_id) < 40:
-                    build_id += '0' * (40 - len(build_id))
-                else:
-                    build_id = build_id[:40]
-                build_id = '0x' + build_id
-                return build_id
-        except subprocess.CalledProcessError:
-            pass
+        if is_elf_file(elf_file_path):
+            try:
+                output = subprocess.check_output([self.readelf_path, '-n', elf_file_path])
+                output = bytes_to_str(output)
+                result = re.search(r'Build ID:\s*(\S+)', output)
+                if result:
+                    build_id = result.group(1)
+                    if len(build_id) < 40:
+                        build_id += '0' * (40 - len(build_id))
+                    else:
+                        build_id = build_id[:40]
+                    build_id = '0x' + build_id
+                    return build_id
+            except subprocess.CalledProcessError:
+                pass
         return ""
 
     def get_sections(self, elf_file_path):
         """ Get sections of an elf file. """
         section_names = []
-        try:
-            output = subprocess.check_output([self.readelf_path, '-SW', elf_file_path])
-            output = bytes_to_str(output)
-            for line in output.split('\n'):
-                # Parse line like:" [ 1] .note.android.ident NOTE  0000000000400190 ...".
-                result = re.search(r'^\s+\[\s*\d+\]\s(.+?)\s', line)
-                if result:
-                    section_name = result.group(1).strip()
-                    if section_name:
-                        section_names.append(section_name)
-        except subprocess.CalledProcessError:
-            pass
+        if is_elf_file(elf_file_path):
+            try:
+                output = subprocess.check_output([self.readelf_path, '-SW', elf_file_path])
+                output = bytes_to_str(output)
+                for line in output.split('\n'):
+                    # Parse line like:" [ 1] .note.android.ident NOTE  0000000000400190 ...".
+                    result = re.search(r'^\s+\[\s*\d+\]\s(.+?)\s', line)
+                    if result:
+                        section_name = result.group(1).strip()
+                        if section_name:
+                            section_names.append(section_name)
+            except subprocess.CalledProcessError:
+                pass
         return section_names
 
 def extant_dir(arg):