From owner-freebsd-python@FreeBSD.ORG Sat Jul 9 17:58:39 2011 Return-Path: Delivered-To: freebsd-python@FreeBSD.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 8DA3E106564A for ; Sat, 9 Jul 2011 17:58:39 +0000 (UTC) (envelope-from nox@jelal.kn-bremen.de) Received: from smtp.kn-bremen.de (gelbbaer.kn-bremen.de [78.46.108.116]) by mx1.freebsd.org (Postfix) with ESMTP id 503F28FC0A for ; Sat, 9 Jul 2011 17:58:39 +0000 (UTC) Received: by smtp.kn-bremen.de (Postfix, from userid 10) id 4815C1E0011C; Sat, 9 Jul 2011 19:43:27 +0200 (CEST) Received: from triton8.kn-bremen.de (noident@localhost [127.0.0.1]) by triton8.kn-bremen.de (8.14.4/8.14.3) with ESMTP id p69Hg2g9097479; Sat, 9 Jul 2011 19:42:02 +0200 (CEST) (envelope-from nox@triton8.kn-bremen.de) Received: (from nox@localhost) by triton8.kn-bremen.de (8.14.4/8.14.3/Submit) id p69Hg1As097478; Sat, 9 Jul 2011 19:42:01 +0200 (CEST) (envelope-from nox) From: Juergen Lock Date: Sat, 9 Jul 2011 19:42:01 +0200 To: freebsd-python@FreeBSD.org Message-ID: <20110709174201.GA96487@triton8.kn-bremen.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.21 (2010-09-15) Cc: Duncan Findlay , Olivier Cochard-Labbe , Benjamin Epitech Subject: Re: ports/153167: Problem with signals, threads, and subprocesses in lang/python X-BeenThere: freebsd-python@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: FreeBSD-specific Python issues List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 09 Jul 2011 17:58:39 -0000 Hi! (please Cc me with followups, I'm not subscribed) Has anyone looked at this problem again yet? I found the PR after debugging the same issue in the emulators/gns3 port, I'm not that good at python but maybe the workaround I came up with (patch to gns3's qemuwrapper.py, this is for an updated version of the port that Olivier is working on) still is useful as a starting point, maybe something like it could be added to subprocess.py, or the more portable C equivalent at the appropriate place? (or even generally in exec*()?) --- qemuwrapper/qemuwrapper.py.orig +++ qemuwrapper/qemuwrapper.py @@ -45,6 +45,7 @@ import SocketServer import time import random import pemubin +import ctypes __author__ = 'Thomas Pani and Jeremy Grossmann' @@ -118,6 +119,33 @@ class xEMUInstance(object): def unbase_disk(self): pass + def preexec(self): + # FreeBSD preexec_fn hack to unblock signals in child processes + # to work around the bug in this PR: + # http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/153167 + # inspired by: + # http://stackoverflow.com/questions/3791398/how-to-stop-python-from-propagating-signals-to-subprocesses + + # Get the size of the array used to + # represent the signal mask + SIGSET_NWORDS = 1024 / (8 * ctypes.sizeof(ctypes.c_uint)) + + # Define the sigset_t structure + class SIGSET(ctypes.Structure): + _fields_ = [ + ('val', ctypes.c_uint * SIGSET_NWORDS) + ] + + # Create a new sigset_t to mask out SIGINT + sigs = (ctypes.c_uint * SIGSET_NWORDS)() + mask = SIGSET(sigs) + + SIG_SETMASK = 3 + libc = ctypes.CDLL('libc.so') + + # Unblock all signals + libc.sigprocmask(SIG_SETMASK, ctypes.pointer(mask), 0) + def start(self): command = self._build_command() @@ -125,6 +152,7 @@ class xEMUInstance(object): try: self.process = subprocess.Popen(command, stdin=subprocess.PIPE, + preexec_fn=self.preexec, cwd=self.workdir) except OSError, e: print >> sys.stderr, "Unable to start instance", self.name, "of", self.__class__