New submission from Robert Xiao <[email protected]>:
Environment: Several versions of Python (see below), macOS 10.12.6
The attached script creates an SSL echo server (fairly standard), connects to
the server, and spawns a read and write thread.
The write thread repeatedly shovels data into the connection, while the read
thread receives data and prints a dot for each successful read.
The socket has a timeout of 10 seconds set: if the timeout is 0, the script
blows up immediately due to blocking, and if the timeout is -1 nothing bad
happens.
On Linux and the default Mac Python 2.6, the script prints an endless series of
dots as expected.
On most other versions of Mac Python (2.7, 3.5, 3.6), the script dies quite
quickly (within 1-2 seconds) with an error like this:
$ /usr/bin/python2.7 test_ssl.py
Got connection from ('127.0.0.1', 49683)
..................................Exception in thread ReadThread:
Traceback (most recent call last):
File
"/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/threading.py",
line 810, in __bootstrap_inner
self.run()
File
"/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/threading.py",
line 763, in run
self.__target(*self.__args, **self.__kwargs)
File "test_ssl.py", line 93, in read_thread
csocket.recv()
File
"/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/ssl.py",
line 734, in recv
return self.read(buflen)
File
"/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/ssl.py",
line 621, in read
v = self._sslobj.read(len or 1024)
error: [Errno 35] Resource temporarily unavailable
The error can be one of the following:
[Py2.7] error: [Errno 35] Resource temporarily unavailable
[Py2.7] SSLWantReadError: The operation did not complete (read) (_ssl.c:1752)
[Py3.x] BlockingIOError: [Errno 35] Resource temporarily unavailable
[Py3.x] ssl.SSLWantReadError: The operation did not complete (read)
(_ssl.c:1974)
[Py3.6] ssl.SSLError: Invalid error code (_ssl.c:2217)
The last error occurs under much rarer circumstances, but appears to be
associated with the same underlying bug. The "invalid error code" is 0 when
tested with a debugger, indicating a successful completion (but somehow the
error logic gets triggered anyway).
This was tested with the following configurations:
macOS: /usr/bin/python2.6: Python 2.6.9 from Apple [ok]
macOS: /usr/bin/python2.7: Python 2.7.10 from Apple [crashes]
macOS: /usr/local/bin/python2.7: Python 2.7.13 from Python.org [crashes]
macOS: /usr/local/bin/python3.5: Python 3.5.1 from Python.org [crashes]
macOS: /usr/local/bin/python3.6: Python 3.6.4 from Python.org [crashes]
macOS: /opt/local/bin/python2.7: Python 2.7.14 from MacPorts [crashes]
A number of these were tested on a second machine (to rule out any strange
environment issues), and the same results were obtained.
----------
assignee: christian.heimes
components: SSL, macOS
files: test_ssl.py
messages: 313970
nosy: christian.heimes, ned.deily, nneonneo, ronaldoussoren
priority: normal
severity: normal
status: open
title: race condition between send and recv in _ssl with non-zero timeout
type: crash
versions: Python 2.7, Python 3.5, Python 3.6
Added file: https://bugs.python.org/file47489/test_ssl.py
_______________________________________
Python tracker <[email protected]>
<https://bugs.python.org/issue33090>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe:
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
