From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 7270 invoked by alias); 24 Oct 2007 20:12:33 -0000 Received: (qmail 7262 invoked by uid 22791); 24 Oct 2007 20:12:32 -0000 X-Spam-Check-By: sourceware.org Received: from heller.inter.net.il (HELO heller.inter.net.il) (213.8.233.23) by sourceware.org (qpsmtpd/0.31) with ESMTP; Wed, 24 Oct 2007 20:12:28 +0000 Received: from HOME-C4E4A596F7 (IGLD-84-229-232-46.inter.net.il [84.229.232.46]) by heller.inter.net.il (MOS 3.7.3a-GA) with ESMTP id DYG18238 (AUTH halo1); Wed, 24 Oct 2007 22:12:12 +0200 (IST) Date: Wed, 24 Oct 2007 20:15:00 -0000 Message-Id: From: Eli Zaretskii To: Daniel Jacobowitz CC: pedro_alves@portugalmail.pt, muller@ics.u-strasbg.fr, gdb-patches@sourceware.org In-reply-to: <20071024193735.GG10943@caradoc.them.org> (message from Daniel Jacobowitz on Wed, 24 Oct 2007 15:37:35 -0400) Subject: Re: [RFC] win32-nat.c: Handle EXCEPTION_INVALID_HANDLE as SIGSYS Reply-to: Eli Zaretskii References: <008101c814b1$9aeb2dd0$d0c18970$@u-strasbg.fr> <20071023214730.GB5570@ednor.casa.cgf.cx> <001301c81613$d5492730$7fdb7590$@u-strasbg.fr> <4053daab0710240139o1898369jc54fbb75c6c48cef@mail.gmail.com> <20071024193735.GG10943@caradoc.them.org> X-IsSubscribed: yes Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org X-SW-Source: 2007-10/txt/msg00620.txt.bz2 > Date: Wed, 24 Oct 2007 15:37:35 -0400 > From: Daniel Jacobowitz > Cc: Pedro Alves , muller@ics.u-strasbg.fr, > gdb-patches@sourceware.org > > > I think it's a good idea. In my experience, any serious program that > > wants to handle signals and exceptions on Windows cannot avoid > > supporting a large number of important EXCEPTION_* exceptions, because > > unlike on Posix platforms, most of them are not translated into SIG* > > style signals, at least in native Windows programs (as opposed to > > Cygwin). > > If whoever works on this is feeling ambitious, I would prefer a > solution that is not too tightly linked to Windows. I've worked with > at least two different embedded developers this past year who were > confused by the need to map platform-specific exceptions onto Unix > signals. An embedded PowerPC developer is likely to have a much > better idea what's going on if we tell him that his code triggered > a Machine Check Exception than if we report SIGBUS. I agree. (Actually, a careful reading of what I wrote would show that I never suggested to do such a mapping, I just said that it isn't done by the underlying library.)