Check for rr traceback in stderr logs

grizzly/common/report.py

@@ -3,6 +3,7 @@
 # file, You can obtain one at http://mozilla.org/MPL/2.0/.
 from hashlib import sha1
 from logging import getLogger
+from mmap import ACCESS_READ, mmap
 from os import SEEK_END
 from pathlib import Path
 from platform import machine, system
@@ -76,6 +77,16 @@ def __init__(
             for log in log_path.iterdir():
                 if log.is_file() and log.stat().st_size > size_limit:
                     Report.tail(log, size_limit)
+        # check for rr traceback and warn
+        if self._logs.stderr and (log_path / "rr-traces").is_dir():
+            with self._logs.stderr.open("rb") as lfp:
+                try:
+                    with mmap(lfp.fileno(), 0, access=ACCESS_READ) as lmm:
+                        if lmm.find(b"=== Start rr backtrace:") != -1:
+                            LOG.warning("rr traceback detected in stderr log")
+                except ValueError:
+                    # cannot mmap an empty file on Windows
+                    pass
         # look through logs one by one until we find a stack
         for log_file in (x for x in self._logs if x is not None):
             stack = Stack.from_text(log_file.read_text("utf-8", errors="ignore"))

grizzly/common/status_reporter.py

@@ -9,7 +9,8 @@
 from functools import partial
 from itertools import zip_longest
 from logging import DEBUG, INFO, basicConfig, getLogger
-from os import SEEK_CUR, getenv
+from mmap import ACCESS_READ, mmap
+from os import getenv
 from pathlib import Path
 from platform import system
 from re import match
@@ -73,24 +74,18 @@ def from_file(
         token = b"Traceback (most recent call last):"
         assert len(token) < cls.READ_LIMIT
         try:
-            with log_file.open("rb") as in_fp:
-                for chunk in iter(partial(in_fp.read, cls.READ_LIMIT), b""):
-                    idx = chunk.find(token)
-                    if idx > -1:
-                        # calculate offset of data in the file
-                        pos = in_fp.tell() - len(chunk) + idx
-                        break
-                    if len(chunk) == cls.READ_LIMIT:
-                        # seek back to avoid missing beginning of token
-                        in_fp.seek(len(token) * -1, SEEK_CUR)
-                else:
-                    # no traceback here, move along
-                    return None
-                # seek back 2KB to collect preceding lines
-                in_fp.seek(max(pos - 2048, 0))
-                data = in_fp.read(cls.READ_LIMIT)
-        except OSError:  # pragma: no cover
-            # in case the file goes away
+            with log_file.open("rb") as lfp:
+                with mmap(lfp.fileno(), 0, access=ACCESS_READ) as lmm:
+                    idx = lmm.find(token)
+                    if idx == -1:
+                        # no traceback here, move along
+                        return None
+                    # seek back 2KB to collect preceding lines
+                    lmm.seek(max(idx - len(token) - 2048, 0))
+                    data = lmm.read(cls.READ_LIMIT)
+        except (OSError, ValueError):  # pragma: no cover
+            # OSError: in case the file goes away
+            # ValueError: cannot mmap an empty file on Windows
             return None
 
         data_lines = data.decode("ascii", errors="ignore").splitlines()

grizzly/common/test_report.py

@@ -412,3 +412,21 @@ def test_report_16(mocker, tmp_path, hang, has_log, expected):
     assert report.short_signature == expected
     if hang:
         assert report.crash_hash == "hang"
+
+
+@mark.parametrize(
+    "log_data",
+    [
+        # empty stderr log
+        b"",
+        # contains rr traceback
+        b"foo\n=== Start rr backtrace:\nfoo",
+        # no traceback
+        b"foo\nfoo",
+    ],
+)
+def test_report_17(tmp_path, log_data):
+    """test Report check for rr traceback"""
+    (tmp_path / "log_stderr.txt").write_bytes(log_data)
+    (tmp_path / "rr-traces").mkdir()
+    assert Report(tmp_path, Path("bin"))

grizzly/common/test_status_reporter.py

@@ -484,7 +484,7 @@ def test_status_reporter_09(mocker, tmp_path):
                 test_fp.write(
                     '  File "some/long/path/name/foobar.py", line 5000, in <module>\n'
                 )
-                test_fp.write(f"    some_long_name_for_a_func_{j:0>4d}()\n")
+                test_fp.write(f"    some_long_name_for_a_func_{j:04d}()\n")
             test_fp.write("IndexError: list index out of range\n")
     rptr = StatusReporter.load(db_file, tb_path=tmp_path)
     rptr._sys_info = _fake_sys_info
@@ -598,7 +598,7 @@ def test_traceback_report_04(tmp_path):
         test_fp.write("    second()\n")
         for i in reversed(range(TracebackReport.MAX_LINES)):
             test_fp.write('  File "foo.py", line 5, in <module>\n')
-            test_fp.write(f"    func_{i:0>2d}()\n")
+            test_fp.write(f"    func_{i:02d}()\n")
         test_fp.write("END_WITH_BLANK_LINE\n\n")
         test_fp.write("end junk\n")
     tbr = TracebackReport.from_file(test_log)