bpo-31530: fix crash when multiple threads iterate over a file, round 2 (#5060)
Multiple threads iterating over a file can corrupt the file's internal readahead
buffer resulting in crashes. To fix this, cache buffer state thread-locally for
the duration of a file_iternext call and only update the file's internal state
after reading completes.
No attempt is made to define or provide "reasonable" semantics for iterating
over a file on multiple threads. (Non-crashing) races are still
present. Duplicated, corrupt, and missing data will happen.
This was originally fixed by 6401e5671781eb217ee1afb4603cc0d1b0367ae6, which
raised an exception from seek() and next() when concurrent operations were
detected. Alas, this simpler solution breaks legitimate use cases such as
capturing the standard streams when multiple threads are logging.
diff --git a/Lib/test/test_file2k.py b/Lib/test/test_file2k.py
index d8966e0..c73e8d8 100644
--- a/Lib/test/test_file2k.py
+++ b/Lib/test/test_file2k.py
@@ -653,18 +653,15 @@
self._test_close_open_io(io_func)
def test_iteration_torture(self):
- # bpo-31530: Crash when concurrently iterate over a file.
+ # bpo-31530
with open(self.filename, "wb") as fp:
for i in xrange(2**20):
fp.write(b"0"*50 + b"\n")
with open(self.filename, "rb") as f:
- def iterate():
- try:
- for l in f:
- pass
- except IOError:
+ def it():
+ for l in f:
pass
- self._run_workers(iterate, 10)
+ self._run_workers(it, 10)
def test_iteration_seek(self):
# bpo-31530: Crash when concurrently seek and iterate over a file.
@@ -674,17 +671,15 @@
with open(self.filename, "rb") as f:
it = iter([1] + [0]*10) # one thread reads, others seek
def iterate():
- try:
- if next(it):
- for l in f:
- pass
- else:
- for i in range(100):
- f.seek(i*100, 0)
- except IOError:
- pass
+ if next(it):
+ for l in f:
+ pass
+ else:
+ for i in xrange(100):
+ f.seek(i*100, 0)
self._run_workers(iterate, 10)
+
@unittest.skipUnless(os.name == 'posix', 'test requires a posix system.')
class TestFileSignalEINTR(unittest.TestCase):
def _test_reading(self, data_to_write, read_and_verify_code, method_name,