From owner-cvs-all Tue Dec 17 15:57:32 2002 Delivered-To: cvs-all@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 9D77C37B401 for ; Tue, 17 Dec 2002 15:57:31 -0800 (PST) Received: from rootlabs.com (root.org [67.118.192.226]) by mx1.FreeBSD.org (Postfix) with SMTP id A253F43ED8 for ; Tue, 17 Dec 2002 15:57:30 -0800 (PST) (envelope-from nate@rootlabs.com) Received: (qmail 51052 invoked by uid 1000); 17 Dec 2002 23:57:31 -0000 Date: Tue, 17 Dec 2002 15:57:31 -0800 (PST) From: Nate Lawson To: Archie Cobbs Cc: cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/i386/i386 i386-gdbstub.c In-Reply-To: <200212171831.gBHIVeD4012682@arch20m.dellroad.org> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-cvs-all@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG On Tue, 17 Dec 2002, Archie Cobbs wrote: > Nate Lawson wrote: > > Julian wrote: > > > NOOOOOOO!!!!!!! > > > previously a 'detach' would drop me back to ddb. > > > please don't do this! (or provide a way for me to get back to ddb) > > > > I have backed it out. The original patch I submitted for review had this > > behavior covered by an option + sysctl. The only feedback I got was that > > the current behavior was considered a bug, hence the commit without any > > protection. How would you prefer this work? > > Hmm, there seems to be some confusion (possibly on my part). > > What I "reviewed" and approved of was fixing the GDB stub so that > when it received the 'detach' command, it would (a) go back to DDB mode > instead of staying in GDB mode waiting for another GDB command, > *not* (b) continue execution. It currently works like (a) and the patch caused it to do (b). While I agree that it is important to be able to toggle between gdb and ddb, I also would like to be able to continue+detach from gdb. This behavior makes sense when you think of a box with only one serial port doing debugging + console. (I'm not going to address "remotechat" because it is flakey and nonportable). With the other part of the patch, when the gdb port receives a valid gdb command, it pops into the debugger in gdb mode. Being able to detach/continue when done is useful (vs. detach, then fire up cu on the same line, then hit "c" in ddb). I'm open to suggestions how to implement this. Perhaps a "default to gdb" option that skips the unnecessary transition through ddb both when attaching and detaching for those that are primarily using ddb. Ideas? -Nate To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message