[Python-checkins] r74063 - in python/branches/py3k: Doc/library/multiprocessing.rst Lib/multiprocessing/process.py Lib/test/test_multiprocessing.py Misc/ACKS

alexandre.vassalotti python-checkins at python.org
Fri Jul 17 14:07:01 CEST 2009


Author: alexandre.vassalotti
Date: Fri Jul 17 14:07:01 2009
New Revision: 74063

Log:
Merged revisions 73694,73708,73738 via svnmerge from 
svn+ssh://pythondev@svn.python.org/python/trunk

........
  r73694 | jesse.noller | 2009-06-29 14:24:26 -0400 (Mon, 29 Jun 2009) | 1 line
  
  Issue 5740: multiprocessing.connection.* authkey fixes
........
  r73708 | jesse.noller | 2009-06-30 13:11:52 -0400 (Tue, 30 Jun 2009) | 1 line
  
  Resolves issues 5155, 5313, 5331 - bad file descriptor error with processes in processes
........
  r73738 | r.david.murray | 2009-06-30 22:49:10 -0400 (Tue, 30 Jun 2009) | 2 lines
  
  Make punctuation prettier and break up run-on sentence.
........


Modified:
   python/branches/py3k/   (props changed)
   python/branches/py3k/Doc/library/multiprocessing.rst
   python/branches/py3k/Lib/multiprocessing/process.py
   python/branches/py3k/Lib/test/test_multiprocessing.py
   python/branches/py3k/Misc/ACKS

Modified: python/branches/py3k/Doc/library/multiprocessing.rst
==============================================================================
--- python/branches/py3k/Doc/library/multiprocessing.rst	(original)
+++ python/branches/py3k/Doc/library/multiprocessing.rst	Fri Jul 17 14:07:01 2009
@@ -1715,7 +1715,7 @@
    generally be omitted since it can usually be inferred from the format of
    *address*. (See :ref:`multiprocessing-address-formats`)
 
-   If *authentication* is ``True`` or *authkey* is a string then digest
+   If *authenticate* is ``True`` or *authkey* is a string then digest
    authentication is used.  The key used for authentication will be either
    *authkey* or ``current_process().authkey)`` if *authkey* is ``None``.
    If authentication fails then :exc:`AuthenticationError` is raised.  See
@@ -1757,7 +1757,7 @@
 
    If *authkey* is ``None`` and *authenticate* is ``True`` then
    ``current_process().authkey`` is used as the authentication key.  If
-   *authkey* is ``None`` and *authentication* is ``False`` then no
+   *authkey* is ``None`` and *authenticate* is ``False`` then no
    authentication is done.  If authentication fails then
    :exc:`AuthenticationError` is raised.  See :ref:`multiprocessing-auth-keys`.
 
@@ -2099,6 +2099,38 @@
            for i in range(10):
                 Process(target=f, args=(lock,)).start()
 
+Beware replacing sys.stdin with a "file like object"
+
+    :mod:`multiprocessing` originally unconditionally called::
+
+        os.close(sys.stdin.fileno())
+
+    in the :meth:`multiprocessing.Process._bootstrap` method --- this resulted
+    in issues with processes-in-processes. This has been changed to::
+
+        sys.stdin.close()
+        sys.stdin = open(os.devnull)
+
+    Which solves the fundamental issue of processes colliding with each other
+    resulting in a bad file descriptor error, but introduces a potential danger
+    to applications which replace :func:`sys.stdin` with a "file-like object"
+    with output buffering.  This danger is that if multiple processes call
+    :func:`close()` on this file-like object, it could result in the same
+    data being flushed to the object multiple times, resulting in corruption.
+
+    If you write a file-like object and implement your own caching, you can
+    make it fork-safe by storing the pid whenever you append to the cache,
+    and discarding the cache when the pid changes. For example::
+
+       @property
+       def cache(self):
+           pid = os.getpid()
+           if pid != self._pid:
+               self._pid = pid
+               self._cache = []
+           return self._cache
+
+    For more information, see :issue:`5155`, :issue:`5313` and :issue:`5331`
 
 Windows
 ~~~~~~~

Modified: python/branches/py3k/Lib/multiprocessing/process.py
==============================================================================
--- python/branches/py3k/Lib/multiprocessing/process.py	(original)
+++ python/branches/py3k/Lib/multiprocessing/process.py	Fri Jul 17 14:07:01 2009
@@ -221,7 +221,8 @@
             self._counter = itertools.count(1)
             if sys.stdin is not None:
                 try:
-                    os.close(sys.stdin.fileno())
+                    sys.stdin.close()
+                    sys.stdin = open(os.devnull)
                 except (OSError, ValueError):
                     pass
             _current_process = self

Modified: python/branches/py3k/Lib/test/test_multiprocessing.py
==============================================================================
--- python/branches/py3k/Lib/test/test_multiprocessing.py	(original)
+++ python/branches/py3k/Lib/test/test_multiprocessing.py	Fri Jul 17 14:07:01 2009
@@ -8,6 +8,7 @@
 import threading
 import queue as pyqueue
 import time
+import io
 import sys
 import os
 import gc
@@ -1868,7 +1869,74 @@
         p.join()
         self.assertEqual(self.ns.test, 1)
 
-testcases_other = [OtherTest, TestInvalidHandle, TestInitializers]
+#
+# Issue 5155, 5313, 5331: Test process in processes
+# Verifies os.close(sys.stdin.fileno) vs. sys.stdin.close() behavior
+#
+
+def _ThisSubProcess(q):
+    try:
+        item = q.get(block=False)
+    except pyqueue.Empty:
+        pass
+
+def _TestProcess(q):
+    queue = multiprocessing.Queue()
+    subProc = multiprocessing.Process(target=_ThisSubProcess, args=(queue,))
+    subProc.start()
+    subProc.join()
+
+def _afunc(x):
+    return x*x
+
+def pool_in_process():
+    pool = multiprocessing.Pool(processes=4)
+    x = pool.map(_afunc, [1, 2, 3, 4, 5, 6, 7])
+
+class _file_like(object):
+    def __init__(self, delegate):
+        self._delegate = delegate
+        self._pid = None
+
+    @property
+    def cache(self):
+        pid = os.getpid()
+        # There are no race conditions since fork keeps only the running thread
+        if pid != self._pid:
+            self._pid = pid
+            self._cache = []
+        return self._cache
+
+    def write(self, data):
+        self.cache.append(data)
+
+    def flush(self):
+        self._delegate.write(''.join(self.cache))
+        self._cache = []
+
+class TestStdinBadfiledescriptor(unittest.TestCase):
+
+    def test_queue_in_process(self):
+        queue = multiprocessing.Queue()
+        proc = multiprocessing.Process(target=_TestProcess, args=(queue,))
+        proc.start()
+        proc.join()
+
+    def test_pool_in_process(self):
+        p = multiprocessing.Process(target=pool_in_process)
+        p.start()
+        p.join()
+
+    def test_flushing(self):
+        sio = io.StringIO()
+        flike = _file_like(sio)
+        flike.write('foo')
+        proc = multiprocessing.Process(target=lambda: flike.flush())
+        flike.flush()
+        assert sio.getvalue() == 'foo'
+
+testcases_other = [OtherTest, TestInvalidHandle, TestInitializers,
+                   TestStdinBadfiledescriptor]
 
 #
 #

Modified: python/branches/py3k/Misc/ACKS
==============================================================================
--- python/branches/py3k/Misc/ACKS	(original)
+++ python/branches/py3k/Misc/ACKS	Fri Jul 17 14:07:01 2009
@@ -46,6 +46,7 @@
 Ulf Bartelt
 Nick Bastin
 Jeff Bauer
+Mike Bayer
 Michael R Bax
 Anthony Baxter
 Samuel L. Bayer
@@ -185,6 +186,7 @@
 Dean Draayer
 John DuBois
 Paul Dubois
+Graham Dumpleton
 Quinn Dunkan
 Robin Dunn
 Luke Dunstan
@@ -558,6 +560,7 @@
 Santiago Peresón
 Mark Perrego
 Trevor Perrin
+Gabriel de Perthuis
 Tim Peters
 Benjamin Peterson
 Chris Petrilli


More information about the Python-checkins mailing list