From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 11944 invoked by alias); 9 Oct 2008 03:11:19 -0000 Received: (qmail 11936 invoked by uid 22791); 9 Oct 2008 03:11:18 -0000 X-Spam-Check-By: sourceware.org Received: from ti-out-0910.google.com (HELO ti-out-0910.google.com) (209.85.142.187) by sourceware.org (qpsmtpd/0.31) with ESMTP; Thu, 09 Oct 2008 03:10:43 +0000 Received: by ti-out-0910.google.com with SMTP id d10so2699873tib.12 for ; Wed, 08 Oct 2008 20:10:41 -0700 (PDT) Received: by 10.110.52.1 with SMTP id z1mr10179381tiz.55.1223521840951; Wed, 08 Oct 2008 20:10:40 -0700 (PDT) Received: by 10.110.42.9 with HTTP; Wed, 8 Oct 2008 20:10:40 -0700 (PDT) Message-ID: Date: Thu, 09 Oct 2008 03:11:00 -0000 From: teawater To: "Pedro Alves" Subject: Re: [reverse] PATCH: Several interface changes Cc: gdb-patches@sourceware.org, "Michael Snyder" In-Reply-To: <200810090349.37968.pedro@codesourcery.com> MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline References: <200810071709.48346.pedro@codesourcery.com> <200810090027.54477.pedro@codesourcery.com> <200810090349.37968.pedro@codesourcery.com> 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: 2008-10/txt/msg00274.txt.bz2 On Thu, Oct 9, 2008 at 10:49, Pedro Alves wrote: > On Thursday 09 October 2008 03:31:47, teawater wrote: >> Maybe P record doesn't need make threads execute in different >> direction. Cause P record reverse executes base on replay mode. This >> mode will replay the memory change and the registers change. And all >> threads of one process share the memory. So... > > You're thinking single-inferior. What about threads of different inferiors > behind a single target_ops? Say, you're attached to process A, but you're > leaving it running (you'll hit internal breakpoints in forward mode), > while you're debugging/inspecting process B in reverse. There you have > your two threads, on a single target, where the single per-target > direction flag stops making sense. If there are multi-inferior use different memory, it must need a flag in resume. Sorry I am not make it clear in before. > > I understand the idealism behind this. I just posted the patch > to show the direction I think we will end up taking, instead of > trying to explain it by: "it would be nice if you did it the way > I'm saying". I always think GDB need it. :) Thanks, Hui