From 2c97c89211cb50fe8a9827e187582337e4d5135d Mon Sep 17 00:00:00 2001
From: Tyson Smith <tyson.w.smith@gmail.com>
Date: Thu, 26 Oct 2023 11:58:41 -0700
Subject: [PATCH] Use the first stack in a sanitizer log

---
 grizzly/common/stack_hasher.py      | 54 ++++++++++++++---------------
 grizzly/common/test_stack_hasher.py | 32 +++++++++++++----
 2 files changed, 53 insertions(+), 33 deletions(-)

diff --git a/grizzly/common/stack_hasher.py b/grizzly/common/stack_hasher.py
index 14388f7f..29bf461c 100644
--- a/grizzly/common/stack_hasher.py
+++ b/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::",
@@ -347,9 +347,11 @@ def from_text(cls, input_text, major_depth=MAJOR_DEPTH, parse_mode=None):
         input_txt is the data to parse the trace from.
         """
 
-        frames = []
-        prev_line = None
-        for line in reversed(input_text.split("\n")):
+        # collection of parsed stacks
+        stacks = []
+        # current stack
+        current = []
+        for line in input_text.split("\n"):
             line = line.rstrip()
             if not line:
                 # skip empty lines
@@ -369,30 +371,28 @@ def from_text(cls, input_text, major_depth=MAJOR_DEPTH, parse_mode=None):
             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:
+                # check for new stack
+                if int(frame.stack_line) == 0 and current:
+                    stacks.append(current)
+                    current = []
+                # check for out of order frames
+                elif (
+                    current
+                    and current[-1].stack_line
+                    and int(frame.stack_line) <= int(current[-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,
-                )
-
-        return cls(frames=frames, major_depth=major_depth)
+            current.append(frame)
+        stacks.append(current)
+        LOG.debug("stacks detected: %d", sum(1 for x in stacks if x))
+
+        # select stack to use
+        if parse_mode == Mode.SANITIZER:
+            current = stacks[0]
+        else:
+            current = stacks[-1]
+        return cls(frames=current, major_depth=major_depth)
 
     @property
     def height_limit(self):
diff --git a/grizzly/common/test_stack_hasher.py b/grizzly/common/test_stack_hasher.py
index 238a32b6..30db181c 100644
--- a/grizzly/common/test_stack_hasher.py
+++ b/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
@@ -309,7 +316,7 @@ def test_stack_16():
         "    #2 0x0001233 in stack_1c() test/a.cpp:34\n"
     )
     assert len(stack.frames) == 2
-    assert stack.frames[0].function == "stack_1b"
+    assert stack.frames[0].function == "stack_1a"
     assert stack.frames[0].mode == Mode.SANITIZER
 
 
@@ -474,6 +481,19 @@ 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(