Merge "simpleperf: merge jitted and interpreted java methods."

3 files changed, 72 insertions, 0 deletions

diff --git a/simpleperf/report_lib_interface.cpp b/simpleperf/report_lib_interface.cpp
index d685ddf5..70a274f4 100644
--- a/simpleperf/report_lib_interface.cpp
+++ b/simpleperf/report_lib_interface.cpp
@@ -108,6 +108,7 @@ bool SetRecordFile(ReportLib* report_lib, const char* record_file) EXPORT;
 bool SetKallsymsFile(ReportLib* report_lib, const char* kallsyms_file) EXPORT;
 void ShowIpForUnknownSymbol(ReportLib* report_lib) EXPORT;
 void ShowArtFrames(ReportLib* report_lib, bool show) EXPORT;
+void MergeJavaMethods(ReportLib* report_lib, bool merge) EXPORT;
 
 Sample* GetNextSample(ReportLib* report_lib) EXPORT;
 Event* GetEventOfCurrentSample(ReportLib* report_lib) EXPORT;
@@ -154,6 +155,7 @@ class ReportLib {
 
   void ShowIpForUnknownSymbol() { thread_tree_.ShowIpForUnknownSymbol(); }
   void ShowArtFrames(bool show) { show_art_frames_ = show; }
+  void MergeJavaMethods(bool merge) { merge_java_methods_ = merge; }
 
   Sample* GetNextSample();
   Event* GetEventOfCurrentSample() { return &current_event_; }
@@ -192,6 +194,9 @@ class ReportLib {
   FeatureSection feature_section_;
   std::vector<char> feature_section_data_;
   bool show_art_frames_;
+  bool merge_java_methods_ = true;
+  // Map from a java method name to it's dex file, start_addr and len.
+  std::unordered_map<std::string, std::tuple<Dso*, uint64_t, uint64_t>> java_methods_;
   std::unique_ptr<Tracing> tracing_;
 };
 
@@ -227,6 +232,15 @@ bool ReportLib::OpenRecordFileIfNecessary() {
     if (auto it = meta_info.find("trace_offcpu"); it != meta_info.end()) {
       trace_offcpu_ = it->second == "true";
     }
+    if (merge_java_methods_) {
+      for (Dso* dso : thread_tree_.GetAllDsos()) {
+        if (dso->type() == DSO_DEX_FILE) {
+          for (auto& symbol : dso->GetSymbols()) {
+            java_methods_[symbol.Name()] = std::make_tuple(dso, symbol.addr, symbol.len);
+          }
+        }
+      }
+    }
   }
   return true;
 }
@@ -328,6 +342,23 @@ void ReportLib::SetCurrentSample() {
     entry.symbol.symbol_addr = symbol->addr;
     entry.symbol.symbol_len = symbol->len;
     entry.symbol.mapping = AddMapping(*map);
+
+    if (merge_java_methods_ && map->dso->type() == DSO_ELF_FILE && map->dso->IsForJavaMethod()) {
+      // This is a jitted java method, merge it with the interpreted java method having the same
+      // name if possible. Otherwise, merge it with other jitted java methods having the same name
+      // by assigning a common dso_name.
+      if (auto it = java_methods_.find(entry.symbol.symbol_name); it != java_methods_.end()) {
+        entry.symbol.dso_name = std::get<0>(it->second)->Path().c_str();
+        entry.symbol.symbol_addr = std::get<1>(it->second);
+        entry.symbol.symbol_len = std::get<2>(it->second);
+        // Not enough info to map an offset in a jitted method to an offset in a dex file. So just
+        // use the symbol_addr.
+        entry.symbol.vaddr_in_file = entry.symbol.symbol_addr;
+      } else {
+        entry.symbol.dso_name = "[JIT cache]";
+      }
+    }
+
     callchain_entries_.push_back(entry);
   }
   current_sample_.ip = callchain_entries_[0].ip;
@@ -460,6 +491,10 @@ void ShowArtFrames(ReportLib* report_lib, bool show) {
   return report_lib->ShowArtFrames(show);
 }
 
+void MergeJavaMethods(ReportLib* report_lib, bool merge) {
+  return report_lib->MergeJavaMethods(merge);
+}
+
 bool SetKallsymsFile(ReportLib* report_lib, const char* kallsyms_file) {
   return report_lib->SetKallsymsFile(kallsyms_file);
 }
diff --git a/simpleperf/scripts/simpleperf_report_lib.py b/simpleperf/scripts/simpleperf_report_lib.py
index ab45b67a..3e677be5 100644
--- a/simpleperf/scripts/simpleperf_report_lib.py
+++ b/simpleperf/scripts/simpleperf_report_lib.py
@@ -240,6 +240,7 @@ class ReportLib(object):
         self._SetKallsymsFileFunc = self._lib.SetKallsymsFile
         self._ShowIpForUnknownSymbolFunc = self._lib.ShowIpForUnknownSymbol
         self._ShowArtFramesFunc = self._lib.ShowArtFrames
+        self._MergeJavaMethodsFunc = self._lib.MergeJavaMethods
         self._GetNextSampleFunc = self._lib.GetNextSample
         self._GetNextSampleFunc.restype = ct.POINTER(SampleStruct)
         self._GetEventOfCurrentSampleFunc = self._lib.GetEventOfCurrentSample
@@ -294,6 +295,17 @@ class ReportLib(object):
         """ Show frames of internal methods of the Java interpreter. """
         self._ShowArtFramesFunc(self.getInstance(), show)
 
+    def MergeJavaMethods(self, merge=True):
+        """ This option merges jitted java methods with the same name but in different jit
+            symfiles. If possible, it also merges jitted methods with interpreted methods,
+            by mapping jitted methods to their corresponding dex files.
+            Side effects:
+              It only works at method level, not instruction level.
+              It makes symbol.vaddr_in_file and symbol.mapping not accurate for jitted methods.
+            Java methods are merged by default.
+        """
+        self._MergeJavaMethodsFunc(self.getInstance(), merge)
+
     def SetKallsymsFile(self, kallsym_file):
         """ Set the file path to a copy of the /proc/kallsyms file (for off device decoding) """
         cond = self._SetKallsymsFileFunc(self.getInstance(), _char_pt(kallsym_file))
diff --git a/simpleperf/scripts/test.py b/simpleperf/scripts/test.py
index e5b5957b..2adee0e0 100755
--- a/simpleperf/scripts/test.py
+++ b/simpleperf/scripts/test.py
@@ -990,6 +990,31 @@ class TestReportLib(unittest.TestCase):
         report_lib.ShowArtFrames(True)
         self.assertTrue(has_art_frame(report_lib))
 
+    def test_merge_java_methods(self):
+        def parse_dso_names(report_lib):
+            dso_names = set()
+            report_lib.SetRecordFile(os.path.join('testdata', 'perf_with_interpreter_frames.data'))
+            while report_lib.GetNextSample():
+                dso_names.add(report_lib.GetSymbolOfCurrentSample().dso_name)
+                callchain = report_lib.GetCallChainOfCurrentSample()
+                for i in range(callchain.nr):
+                    dso_names.add(callchain.entries[i].symbol.dso_name)
+            report_lib.Close()
+            has_jit_symfiles = any('TemporaryFile-' in name for name in dso_names)
+            has_jit_cache = '[JIT cache]' in dso_names
+            return has_jit_symfiles, has_jit_cache
+
+        report_lib = ReportLib()
+        self.assertEqual(parse_dso_names(report_lib), (False, True))
+
+        report_lib = ReportLib()
+        report_lib.MergeJavaMethods(True)
+        self.assertEqual(parse_dso_names(report_lib), (False, True))
+
+        report_lib = ReportLib()
+        report_lib.MergeJavaMethods(False)
+        self.assertEqual(parse_dso_names(report_lib), (True, False))
+
     def test_tracing_data(self):
         self.report_lib.SetRecordFile(os.path.join('testdata', 'perf_with_tracepoint_event.data'))
         has_tracing_data = False