New submission from Owen Lin:
If we call two subprocess.Popen simultaneously, the second one is blocked until
the first one is finished.
The attached file is a code snippet to reproduce this bug. I can reproduce the
bug in version 2.7.3 and 2.7.6 very easily (in few seconds with the code). But
it works fine on python3.
Here is the backtrace of python
==========================================
#0 0x00007f0eba954d2d in read () at ../sysdeps/unix/syscall-template.S:82#1
0x00000000005d8d10 in posix_read (self=0x0, args=(5, 1048576)) at
../Modules/posixmodule.c:6628#2 0x0000000000486896 in PyCFunction_Call
(func=<built-in function read>, arg=(5, 1048576), kw=0x0) at
../Objects/methodobject.c:81#3 0x00000000005278e4 in ext_do_call
(func=<built-in function read>, pp_stack=0x7fff1fc0ac80, flags=1, na=0, nk=0)
at ../Python/ceval.c:4331
#4 0x00000000005215cd in PyEval_EvalFrameEx (
f=Frame 0x298f800, for file /usr/lib/python2.7/subprocess.py, line 478, in
_eintr_retry_call (func=<built-in funct
ion read>, args=(5, 1048576)), throwflag=0) at ../Python/ceval.c:2705
#5 0x0000000000523c2e in PyEval_EvalCodeEx (co=0x294b880,
globals={'STDOUT': -2, '_has_poll': True, 'gc': <module at remote 0x29672d0>,
'check_call': <function at remote 0x29c4450>, 'mswindows': False, 'select':
<module at remote 0x29676e0>, 'list2cmdline': <function at remote 0x29c45a0>,
'__all__': ['Popen', 'PIPE', 'STDOUT', 'call', 'check_call', 'check_output',
'CalledProcessError'], 'errno': <module at remote 0x272d4d8>, '_demo_posix':
<function at remote 0x29c4648>, '__package__': None, 'PIPE': -1, '_cleanup':
<func
===================================
The fd 5 is actually a pipe. But I cannot find the other end of the pipe. A
workaround is using lock around all the Popen()s.
----------
files: test.py
messages: 204568
nosy: owenlin
priority: normal
severity: normal
status: open
title: Python get stuck in second Popen call
type: behavior
versions: Python 2.7
Added file: http://bugs.python.org/file32868/test.py
_______________________________________
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue19809>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe:
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
