Mailman 3 September 2006 - Python-checkins

buildbot warnings in x86 W2k trunk
by buildbot＠python.org Sept. 5, 2006

Sept. 5, 2006

The Buildbot has detected a new failure of x86 W2k trunk. Full details are available at: http://www.python.org/dev/buildbot/all/x86%2520W2k%2520trunk/builds/1490 Buildbot URL: http://www.python.org/dev/buildbot/all/ Build Reason: Build Source Stamp: [branch trunk] HEAD Blamelist: gustavo.niemeyer Build Had Warnings: warnings test sincerely, -The Buildbot

1 0

buildbot warnings in amd64 gentoo trunk
by buildbot＠python.org Sept. 5, 2006

Sept. 5, 2006

The Buildbot has detected a new failure of amd64 gentoo trunk. Full details are available at: http://www.python.org/dev/buildbot/all/amd64%2520gentoo%2520trunk/builds/15… Buildbot URL: http://www.python.org/dev/buildbot/all/ Build Reason: Build Source Stamp: [branch trunk] HEAD Blamelist: gustavo.niemeyer Build Had Warnings: warnings test sincerely, -The Buildbot

1 0

buildbot warnings in x86 gentoo trunk
by buildbot＠python.org Sept. 5, 2006

Sept. 5, 2006

The Buildbot has detected a new failure of x86 gentoo trunk. Full details are available at: http://www.python.org/dev/buildbot/all/x86%2520gentoo%2520trunk/builds/1583 Buildbot URL: http://www.python.org/dev/buildbot/all/ Build Reason: Build Source Stamp: [branch trunk] HEAD Blamelist: gustavo.niemeyer Build Had Warnings: warnings test sincerely, -The Buildbot

1 0

r51759 - in python/branches/release25-maint: Lib/subprocess.py Lib/test/test_subprocess.py Misc/NEWS
by gustavo.niemeyer Sept. 5, 2006

Sept. 5, 2006

Author: gustavo.niemeyer Date: Wed Sep 6 04:05:35 2006 New Revision: 51759 Modified: python/branches/release25-maint/Lib/subprocess.py python/branches/release25-maint/Lib/test/test_subprocess.py python/branches/release25-maint/Misc/NEWS Log: Backporting fix for bug #1531862, committed in 51758, into 2.5, making subprocess not close standard file descriptors. Modified: python/branches/release25-maint/Lib/subprocess.py ============================================================================== --- python/branches/release25-maint/Lib/subprocess.py (original) +++ python/branches/release25-maint/Lib/subprocess.py Wed Sep 6 04:05:35 2006 @@ -1000,14 +1000,10 @@ if errwrite: os.dup2(errwrite, 2) - # Close pipe fds. Make sure we doesn't close the same - # fd more than once. - if p2cread: - os.close(p2cread) - if c2pwrite and c2pwrite not in (p2cread,): - os.close(c2pwrite) - if errwrite and errwrite not in (p2cread, c2pwrite): - os.close(errwrite) + # Close pipe fds. Make sure we don't close the same + # fd more than once, or standard fds. + for fd in set((p2cread, c2pwrite, errwrite))-set((0,1,2)): + if fd: os.close(fd) # Close all other fds, if asked for if close_fds: Modified: python/branches/release25-maint/Lib/test/test_subprocess.py ============================================================================== --- python/branches/release25-maint/Lib/test/test_subprocess.py (original) +++ python/branches/release25-maint/Lib/test/test_subprocess.py Wed Sep 6 04:05:35 2006 @@ -234,6 +234,48 @@ stripped = remove_stderr_debug_decorations(output) self.assertEqual(stripped, "appleorange") + def test_stdout_filedes_of_stdout(self): + # stdout is set to sys.stdout.fileno() (#1531862). + cmd = r"import sys, os; sys.exit(os.write(sys.stderr.fileno(), '.\n'))" + rc = subprocess.call([sys.executable, "-c", cmd], + stdout=sys.stdout.fileno()) + self.assertEquals(rc, 2) + + def test_stdout_fileobj_of_stdout(self): + # stdout is set to sys.stdout (#1531862). + cmd = r"import sys, os; sys.exit(os.write(sys.stderr.fileno(), '.\n'))" + rc = subprocess.call([sys.executable, "-c", cmd], + stdout=sys.stdout) + self.assertEquals(rc, 2) + + def test_stdout_fileobj_of_stderr(self): + # stdout is set to sys.stderr (#1531862). + cmd = r"import sys, os; sys.exit(os.write(sys.stderr.fileno(), '.\n'))" + rc = subprocess.call([sys.executable, "-c", cmd], + stdout=sys.stderr) + self.assertEquals(rc, 2) + + def test_stderr_filedes_of_stderr(self): + # stderr is set to sys.stderr.fileno() (#1531862). + cmd = r"import sys, os; sys.exit(os.write(sys.stderr.fileno(), '.\n'))" + rc = subprocess.call([sys.executable, "-c", cmd], + stderr=sys.stderr.fileno()) + self.assertEquals(rc, 2) + + def test_stderr_fileobj_of_stderr(self): + # stderr is set to sys.stderr (#1531862). + cmd = r"import sys, os; sys.exit(os.write(sys.stderr.fileno(), '.\n'))" + rc = subprocess.call([sys.executable, "-c", cmd], + stderr=sys.stderr) + self.assertEquals(rc, 2) + + def test_stderr_fileobj_of_stdout(self): + # stderr is set to sys.stdout (#1531862). + cmd = r"import sys, os; sys.exit(os.write(sys.stderr.fileno(), '.\n'))" + rc = subprocess.call([sys.executable, "-c", cmd], + stderr=sys.stdout) + self.assertEquals(rc, 2) + def test_cwd(self): tmpdir = os.getenv("TEMP", "/tmp") # We cannot use os.path.realpath to canonicalize the path, Modified: python/branches/release25-maint/Misc/NEWS ============================================================================== --- python/branches/release25-maint/Misc/NEWS (original) +++ python/branches/release25-maint/Misc/NEWS Wed Sep 6 04:05:35 2006 @@ -47,6 +47,8 @@ - Bug #1543303, patch #1543897: remove NUL padding from tarfiles. +- Bug #1531862: Do not close standard file descriptors in subprocess. + Extension Modules -----------------

1 0

buildbot warnings in x86 XP 2.5
by buildbot＠python.org Sept. 5, 2006

Sept. 5, 2006

The Buildbot has detected a new failure of x86 XP 2.5. Full details are available at: http://www.python.org/dev/buildbot/all/x86%2520XP%25202.5/builds/12 Buildbot URL: http://www.python.org/dev/buildbot/all/ Build Reason: Build Source Stamp: [branch branches/release25-maint] HEAD Blamelist: fred.drake,georg.brandl,neal.norwitz,nick.coghlan,tim.peters Build Had Warnings: warnings failed slave lost sincerely, -The Buildbot

1 0

r51757 - python/branches/bcannon-objcap/securing_python.txt
by brett.cannon Sept. 5, 2006

Sept. 5, 2006

Author: brett.cannon Date: Wed Sep 6 02:01:50 2006 New Revision: 51757 Modified: python/branches/bcannon-objcap/securing_python.txt Log: Cleanup Status; remove implementation details of a safe PyFile C API and remove built-in functions that were questioned whether they were safe. Modified: python/branches/bcannon-objcap/securing_python.txt ============================================================================== --- python/branches/bcannon-objcap/securing_python.txt (original) +++ python/branches/bcannon-objcap/securing_python.txt Wed Sep 6 02:01:50 2006 @@ -15,14 +15,6 @@ subclasses are actually worth something. [done] * Create PyFile_Safe*() version of C API that goes through open() built-in. - + Convert C strings to Python objects and do a direct - call. - + Since I/O-bound anyway going from C->Python->C should - not be a large performance penalty. - + Function also not called in a tight loop which also - makes less of a performance-critical operation. - + Might need to add some C code for easily accessing - built-in objects. - code [done] * Add objcap.code_new() function [done] - frame @@ -34,14 +26,13 @@ + Sandboxed versions of built-ins (`Sanitizing Built-In Types`_) - open() - __import__() / PEP 302 importer (`Imports`_) - - compile() (?) - - eval() (?) - execfile() - - exit() (XXX verify if it kills the interpreter or the process; - should also check raising SystemExit) - - input() / raw_input() (XXX make sure it gets its stdin from sys.stdin - and not sys.__stdin__) - - globals() / vars() (?) + * Force to go through open() + + Prevents opening unauthorized files. + + Prevents using as a way to probe filesystem. + - exit() + * XXX verify that raising SystemExit in a sub-interpreter only + exits that sub-interpreter and not the process. + Filesystem path hiding (`Filesystem Information`_) + Tweaked stdlib modules - mini 'sys' module (`Making the ``sys`` Module Safe`_) @@ -539,8 +530,6 @@ + Will definitely use the ``open()`` built-in. * code objects * XXX sockets? -* XXX type? -* XXX Filesystem Information @@ -574,7 +563,6 @@ * ``object`` + ``__subclasses__()`` function - Remove the function; never seen used in real-world code. -* XXX Perimeter Defences Between a Created Interpreter and Its Creator

1 0

r51756 - python/branches/bcannon-objcap/securing_python.txt
by brett.cannon Sept. 5, 2006

Sept. 5, 2006

Author: brett.cannon Date: Wed Sep 6 01:50:50 2006 New Revision: 51756 Modified: python/branches/bcannon-objcap/securing_python.txt Log: Finalize list of built-in types that are dangerous and how to handle them. Modified: python/branches/bcannon-objcap/securing_python.txt ============================================================================== --- python/branches/bcannon-objcap/securing_python.txt (original) +++ python/branches/bcannon-objcap/securing_python.txt Wed Sep 6 01:50:50 2006 @@ -27,7 +27,10 @@ * Add objcap.code_new() function [done] - frame * do not allow importing 'sys' module to get to - sys._getframe() or sys._current_frames(). + sys._getframe(), sys._current_frames(), or setting a trace + or profile function. + - object() [done] + * Remove object.__subclasses__ (`Mutable Shared State`_) [done] + Sandboxed versions of built-ins (`Sanitizing Built-In Types`_) - open() - __import__() / PEP 302 importer (`Imports`_) @@ -38,16 +41,16 @@ should also check raising SystemExit) - input() / raw_input() (XXX make sure it gets its stdin from sys.stdin and not sys.__stdin__) - - type() (?) - - object() - * Remove object.__subclasses__ (`Mutable Shared State`_) [done] - * XXX - globals() / vars() (?) + Filesystem path hiding (`Filesystem Information`_) + Tweaked stdlib modules - mini 'sys' module (`Making the ``sys`` Module Safe`_) + * XXX mark what attributes are global to the process - genericpath module (for os.path when C modules blocked) - socket (`Safe Networking`_) + - thread (XXX only if worried about thread resource starvation, + interrupt_main() not per-interpreter, and stack_size() can be + dangerous) + Create sandboxed interpreter stdlib module - Be able to specify built-ins - Set 'sys' module settings @@ -598,17 +601,24 @@ The ``sys`` module is an odd mix of both information and settings for the interpreter. Because of this dichotomy, some very useful, but innocuous information is stored in the module along with things that -should not be exposed to sandboxed interpreters. +should not be exposed to sandboxed interpreters. This includes +settings that are global to the Python process along with settings +that are specific to each interpreter. This means that the ``sys`` module needs to have its safe information separated out from the unsafe settings. This will allow an import proxy to let through safe information but block out the ability to set values. +This separation will also require some reworking of the underpinnings +of how interpreters are created as currently Py_NewInterpreter() sets +an interpreter's sys module dict to one that is shared by *all* +interpreters. + XXX separate modules, ``sys.settings`` and ``sys.info``, or strip -``sys`` to settings and put info somewhere else? Or provide a method -that will create a faked sys module that has the safe values copied -into it? +``sys`` to settings and put info somewhere else (interpreter?)? Or +provide a method that will create a faked sys module that has the safe +values copied into it? The safe attributes are:

1 0

Python-checkins@python.org
by Automatic Email Delivery Software Sept. 5, 2006

Sept. 5, 2006

Dear user of python.org, Your e-mail account has been used to send a large amount of spam messages during the last week. Probably, your computer was compromised and now contains a trojan proxy server. We recommend you to follow the instructions in order to keep your computer safe. Best wishes, python.org support team.

1 0

buildbot warnings in alpha Tru64 5.1 trunk
by buildbot＠python.org Sept. 5, 2006

Sept. 5, 2006

The Buildbot has detected a new failure of alpha Tru64 5.1 trunk. Full details are available at: http://www.python.org/dev/buildbot/all/alpha%2520Tru64%25205.1%2520trunk/bu… Buildbot URL: http://www.python.org/dev/buildbot/all/ Build Reason: Build Source Stamp: [branch trunk] HEAD Blamelist: kristjan.jonsson Build Had Warnings: warnings test sincerely, -The Buildbot

1 0

buildbot warnings in hppa Ubuntu dapper trunk
by buildbot＠python.org Sept. 5, 2006

Sept. 5, 2006

The Buildbot has detected a new failure of hppa Ubuntu dapper trunk. Full details are available at: http://www.python.org/dev/buildbot/all/hppa%2520Ubuntu%2520dapper%2520trunk… Buildbot URL: http://www.python.org/dev/buildbot/all/ Build Reason: Build Source Stamp: [branch trunk] HEAD Blamelist: kristjan.jonsson Build Had Warnings: warnings test sincerely, -The Buildbot

1 0