From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 1794 invoked by alias); 29 Apr 2006 15:06:02 -0000 Received: (qmail 1781 invoked by uid 22791); 29 Apr 2006 15:06:01 -0000 X-Spam-Check-By: sourceware.org Received: from nevyn.them.org (HELO nevyn.them.org) (66.93.172.17) by sourceware.org (qpsmtpd/0.31.1) with ESMTP; Sat, 29 Apr 2006 15:05:59 +0000 Received: from drow by nevyn.them.org with local (Exim 4.54) id 1FZr1B-0005h1-1V for gdb-patches@sourceware.org; Sat, 29 Apr 2006 11:05:57 -0400 Date: Sat, 29 Apr 2006 15:06:00 -0000 From: Daniel Jacobowitz To: gdb-patches@sourceware.org Subject: Re: [RFC/RFA/i386] pb reading insns if breakpoints still inserted Message-ID: <20060429150556.GA21836@nevyn.them.org> Mail-Followup-To: gdb-patches@sourceware.org References: <20060428171154.GP17613@adacore.com> <20060429142812.GA21067@nevyn.them.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.8i X-IsSubscribed: yes Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org X-SW-Source: 2006-04/txt/msg00391.txt.bz2 On Sat, Apr 29, 2006 at 05:51:28PM +0300, Eli Zaretskii wrote: > > Now, you could make a reasonable argument that the default when reading > > memory ought to be to ignore breakpoints. But that would be a much > > harder change to test; we'd need to catch all the places which do need > > to see breakpoints > > We don't need to catch all of them at once, if that's hard. We can > catch them one by one and fix them as we do. Well, this one's been there for years. I'd rather have slightly more confidence that we won't introduce a whole new set of such bugs... > > I can only think of one off the top of my head, but I'm sure there > > are more. So, I would suggest making it separately. > > It's fine with me, but I still don't understand why it has to be in an > architecture-specific file, not in some more general place. Am I > missing something? Right now, there are many ways to read memory; two of them skip inserted breakpoints, and the others don't. So, to fix this sort of problem, you have to deliberately use the functions which do skip breakpoints. Thus the change has to be at the site of the memory read, which is in this case in a target-specific file. We don't want to remove breakpoints before invoking the prologue analyzer; the removal may be unnecessary if a "step" command is in progress and not done yet. -- Daniel Jacobowitz CodeSourcery