From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 28703 invoked by alias); 25 Nov 2010 19:19:26 -0000 Received: (qmail 28694 invoked by uid 22791); 25 Nov 2010 19:19:25 -0000 X-SWARE-Spam-Status: No, hits=-0.8 required=5.0 tests=AWL,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,FREEMAIL_ENVFROM_END_DIGIT,FREEMAIL_FROM,RCVD_IN_DNSWL_NONE,RFC_ABUSE_POST X-Spam-Check-By: sourceware.org Received: from mail-qy0-f169.google.com (HELO mail-qy0-f169.google.com) (209.85.216.169) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Thu, 25 Nov 2010 19:19:21 +0000 Received: by qyk4 with SMTP id 4so6595546qyk.0 for ; Thu, 25 Nov 2010 11:19:19 -0800 (PST) MIME-Version: 1.0 Received: by 10.229.240.4 with SMTP id ky4mr973520qcb.104.1290712758892; Thu, 25 Nov 2010 11:19:18 -0800 (PST) Received: by 10.229.85.195 with HTTP; Thu, 25 Nov 2010 11:19:18 -0800 (PST) In-Reply-To: <20101125163022.GI2634@adacore.com> References: <20101118172209.GE2634@adacore.com> <201011242118.oAOLIqZI030918@glazunov.sibelius.xs4all.nl> <20101124221451.GE2634@adacore.com> <20101124222555.GG2634@adacore.com> <201011251338.oAPDck2Y013525@glazunov.sibelius.xs4all.nl> <20101125163022.GI2634@adacore.com> Date: Thu, 25 Nov 2010 19:19:00 -0000 Message-ID: Subject: Re: [RFC] Improve amd64 prologue analysis From: Kai Tietz To: Joel Brobecker Cc: Mark Kettenis , pierre.muller@ics-cnrs.unistra.fr, gdb-patches@sourceware.org Content-Type: text/plain; charset=ISO-8859-1 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: 2010-11/txt/msg00434.txt.bz2 2010/11/25 Joel Brobecker : >> > One last bit of information - which I got second hand, but hopefully >> > accurate. I think that the GCC team is on its way to generating the >> > unwinding info in that format, rather than the usual DWARF-based >> > eh_frame/debug_frame. So the Microsoft format should become the default >> > even for code generated by GCC... So eventually, I think we are going >> > to need to add an unwinder for that, even for GCC code. >> >> I certainly hope you're wrong here, since it would mean a massive ABI >> break on platforms currently using .eh_frame, and presumably would be >> x86-specific. > > I should have said that this is only relevant to x86_64-windows. > All other platforms will continue to use the usual eh_frame, of > course. > > -- > Joel > Well, Joel is right. Beginning with 4.6 gcc there is a base support of SEH prologue unwind information for x64 windows target. By this it is possible to use API provided by imagehlp.dll for generating stack backtraces. At the moment SEH isn't used for exception unwinding, but this is planned for 4.7 gcc version. A test-implementation of it is already existing, but not sure if it will make it into 4.6 as optional unwind implementation for tests. Kai