Use the first stack in a sanitizer log

grizzly/common/stack_hasher.py

@@ -25,7 +25,7 @@
 # These entries pad out the stack and make bucketing more difficult
 IGNORED_FRAMES = (
     "core::panicking::",
-    "mozglue_static::panic_hook::",
+    "mozglue_static::panic_hook",
     "rust_begin_unwind",
     "RustMozCrash",
     "std::panicking::",
@@ -342,14 +342,21 @@ def from_file(self, file_name):  # pragma: no cover
 
     @classmethod
     def from_text(cls, input_text, major_depth=MAJOR_DEPTH, parse_mode=None):
-        """
-        parse a stack trace from text.
-        input_txt is the data to parse the trace from.
+        """Parse a stack trace from text. This is intended to parse the output
+        from a single result. Some debuggers such as ASan and TSan can include
+        multiple stacks per result.
+
+        Args:
+            input_text: Data to parse.
+            major_depth: Number of frames use to calculate the major hash.
+            parse_mode: Format to use. If None the format is detected automatically.
+
+        Returns:
+            Stack
         """
 
         frames = []
-        prev_line = None
-        for line in reversed(input_text.split("\n")):
+        for line in input_text.split("\n"):
             line = line.rstrip()
             if not line:
                 # skip empty lines
@@ -368,29 +375,19 @@ def from_text(cls, input_text, major_depth=MAJOR_DEPTH, parse_mode=None):
                 LOG.debug("parser mode: %s", parse_mode.name)
             assert frame.mode == parse_mode
 
-            if frame.stack_line is not None:
-                stack_line = int(frame.stack_line)
-                # check if we've found a different stack in the data
-                if prev_line is not None and prev_line <= stack_line:
+            if frame.stack_line is not None and frames:
+                num = int(frame.stack_line)
+                # check for new stack
+                if num == 0:
+                    # select stack to use
+                    if parse_mode in (Mode.SANITIZER, Mode.TSAN):
+                        break
+                    frames.clear()
+                # check for out of order or missing frames
+                elif frames[-1].stack_line and num - 1 != int(frames[-1].stack_line):
+                    LOG.debug("scrambled logs?")
                     break
-                frames.insert(0, frame)
-                if stack_line < 1:
-                    break
-                prev_line = stack_line
-            else:
-                frames.insert(0, frame)
-
-        # sanity check
-        if frames and prev_line is not None:
-            # assuming the first frame is 0
-            if int(frames[0].stack_line) != 0:
-                LOG.warning("First stack frame is %s not 0", frames[0].stack_line)
-            if int(frames[-1].stack_line) != len(frames) - 1:
-                LOG.warning(
-                    "Missing frames? Last frame is %s, expected %d (frames-1)",
-                    frames[-1].stack_line,
-                    len(frames) - 1,
-                )
+            frames.append(frame)
 
         return cls(frames=frames, major_depth=major_depth)
 

grizzly/common/test_stack_hasher.py

@@ -125,13 +125,20 @@ def test_stack_07():
 
 
 def test_stack_08():
-    """test creating a Stack by calling from_text() with text containing 2 stacks"""
+    """test creating a Stack by calling from_text() with text multiple stacks"""
     input_txt = (
-        ""
-        "    #0 0x0bad0bad in bad::frame0(nsA*, nsB*) /aa/a.cpp:12:1\n"
-        "    #1 0x0bad0bad in bad::frame1(mz::d::EE*, int) /aa/a.cpp:12:1\n"
+        "=================================================================\n"
+        "==5540==ERROR: AddressSanitizer: ...\n"
         "    #0 0x1badf00d in good::frame0(nsA*, nsB*) /aa/a.cpp:12:1\n"
-        "    #1 0xdeadbeef in good::frame1(mz::d::EE*, int) /aa/a.cpp:12:1\n"
+        "    #1 0xdeadbeef in good::frame1(mz::d::EE*, int) /aa/a.cpp:12:1\n\n"
+        "0x12876 is located 1024 bytes after 4096-byte region [0x12876,0x12876)\n"
+        "freed by thread T5 here:\n"
+        "    #0 0x0bad0bad in bad::frame0(nsA*, nsB*) /aa/a.cpp:12:1\n"
+        "    #1 0x0bad0bad in bad::frame1(mz::d::EE*, int) /aa/a.cpp:12:1\n\n"
+        "0x12876 is located 1024 bytes after 4096-byte region [0x12876,0x12876)\n"
+        "previously allocated by thread T0 here:\n"
+        "    #0 0x0bad0bad in bad::frame0(nsA*, nsB*) /aa/a.cpp:12:1\n"
+        "    #1 0x0bad0bad in bad::frame1(mz::d::EE*, int) /aa/a.cpp:12:1\n\n"
     )
     stack = Stack.from_text(input_txt)
     assert len(stack.frames) == 2
@@ -308,8 +315,8 @@ def test_stack_16():
         "    #1 0x0001232 in stack_1b() test/a.cpp:23\n"
         "    #2 0x0001233 in stack_1c() test/a.cpp:34\n"
     )
-    assert len(stack.frames) == 2
-    assert stack.frames[0].function == "stack_1b"
+    assert len(stack.frames) == 1
+    assert stack.frames[0].function == "stack_1a"
     assert stack.frames[0].mode == Mode.SANITIZER
 
 
@@ -336,12 +343,31 @@ def test_stack_17():
     )
     stack02 = Stack.from_text(st_02, major_depth=3)
     assert len(stack02.frames) == 6
-
     assert stack01.minor != stack02.minor
     assert stack01.major != stack02.major
     assert stack01.frames[0].mode == Mode.SANITIZER
 
 
+def test_stack_18():
+    """test creating a Stack from two minidump stackwalk traces"""
+    stack = Stack.from_text(
+        "CPU|x86|GenuineIntel family 6 model 85 stepping 4|8\n"
+        "Crash|EXCEPTION_BREAKPOINT|0x67a30091|44\n"
+        "42|0|xul.dll|foo_a()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+        "42|1|xul.dll|foo_b()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+        "42|2|xul.dll|foo_c()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+        "43|0|xul.dll|bar_a()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+        "43|1|xul.dll|bar_b()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+        "43|2|xul.dll|bar_c()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+        "44|0|xul.dll|good_a()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+        "44|1|xul.dll|good_a()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+        "44|2|xul.dll|good_c()|s3:g-g-sources:e/a.cpp:|14302|0x3f1\n"
+    )
+    assert len(stack.frames) == 3
+    assert stack.frames[0].function == "good_a()"
+    assert stack.frames[0].mode == Mode.MINIDUMP
+
+
 def test_stackframe_01():
     """test creating an empty StackFrame"""
     stack = StackFrame()
@@ -474,6 +500,19 @@ def test_sanitizer_stackframe_09():
     assert frame.mode == Mode.SANITIZER
 
 
+def test_sanitizer_stackframe_10():
+    """test creating a StackFrame from a line with build id"""
+    frame = StackFrame.from_line(
+        "    #0 0x7f76d25b7fc0  (/usr/lib/x86_64-linux-gnu/dri/swrast_dri.so+0x704fc0) "
+        "(BuildId: d04a40e4062a8d444ff6f23d4fe768215b2e32c7)"
+    )
+    assert frame.stack_line == "0"
+    assert frame.function is None
+    assert frame.location == "swrast_dri.so"
+    assert frame.offset == "0x704fc0"
+    assert frame.mode == Mode.SANITIZER
+
+
 def test_gdb_stackframe_01():
     """test creating a StackFrame from a GDB line with symbols"""
     frame = StackFrame.from_line(
@@ -560,6 +599,18 @@ def test_minidump_stackframe_03():
     assert frame.mode == Mode.MINIDUMP
 
 
+def test_minidump_stackframe_04():
+    """test creating a StackFrame from a Minidump line with s3 repo info"""
+    frame = StackFrame.from_line(
+        "42|0|xul.dll|foo_a() const|s3:g-g-sources:e/a.cpp:|14302|0x3f1"
+    )
+    assert frame.stack_line == "0"
+    assert frame.function == "foo_a() const"
+    assert frame.location == "a.cpp"
+    assert frame.offset == "14302"
+    assert frame.mode == Mode.MINIDUMP
+
+
 def test_tsan_stackframe_01():
     """test creating a StackFrame from a symbolized TSan line"""
     frame = StackFrame.from_line("    #0 main race.c:10 (exe+0xa3b4)")