From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 2943 invoked by alias); 7 Jun 2010 19:12:26 -0000 Received: (qmail 2935 invoked by uid 22791); 7 Jun 2010 19:12:26 -0000 X-SWARE-Spam-Status: No, hits=-2.1 required=5.0 tests=AWL,BAYES_00,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org Received: from mail.codesourcery.com (HELO mail.codesourcery.com) (38.113.113.100) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Mon, 07 Jun 2010 19:12:22 +0000 Received: (qmail 31458 invoked from network); 7 Jun 2010 19:12:20 -0000 Received: from unknown (HELO orlando.localnet) (pedro@127.0.0.2) by mail.codesourcery.com with ESMTPA; 7 Jun 2010 19:12:20 -0000 From: Pedro Alves To: gdb-patches@sourceware.org Subject: Re: [RFA] fix crasher on detach command Date: Mon, 07 Jun 2010 19:12:00 -0000 User-Agent: KMail/1.13.2 (Linux/2.6.32-22-generic; KDE/4.4.2; x86_64; ; ) Cc: Michael Snyder References: <4C0D3636.8040206@vmware.com> <4C0D41B4.3020008@vmware.com> <201006072008.51363.pedro@codesourcery.com> In-Reply-To: <201006072008.51363.pedro@codesourcery.com> MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-15" Content-Transfer-Encoding: 7bit Message-Id: <201006072012.18043.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: 2010-06/txt/msg00187.txt.bz2 On Monday 07 June 2010 20:08:51, Pedro Alves wrote: > On Monday 07 June 2010 20:00:04, Michael Snyder wrote: > > > Then we need to fix that, instead of adding workarounds in other > > > areas. > > > > I'm not sure how feasible that is. At this point I've had to use a > > ^C^C to get out of a failing attach (target remote), and I'm hoping to > > use the "detach" to cancel out any remaining inconsistent state. > > The ^C^C handler can't necessarily do it by itself, 'cause from its > > point of view, you don't really know what state you're in or what > > state you want to be in. > > It's feasible. remote_open_1 + remote_start_remote and remote_close > (usually from pop_target calls within remote.c) are all designed for > this to not happen, but clearly there's a bug somewhere. It just > sounds like there's something not exception safe that should be. To be a bit clearer -- you've said that the pid was left as 42000 (I assume you meant inferior_ptid, but that find_inferior no longer finds that inferior. Where is the current inferior getting it's pid cleared out? Why aren't we clearing inferior_ptid as well? -- Pedro Alves