From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 28039 invoked by alias); 1 Sep 2006 00:20:32 -0000 Received: (qmail 28031 invoked by uid 22791); 1 Sep 2006 00:20:31 -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; Fri, 01 Sep 2006 00:20:29 +0000 Received: from drow by nevyn.them.org with local (Exim 4.54) id 1GIwlm-0005yQ-Gi; Thu, 31 Aug 2006 20:20:26 -0400 Date: Fri, 01 Sep 2006 00:20:00 -0000 From: Daniel Jacobowitz To: Chuck Simmons Cc: bartoschek@or.uni-bonn.de, gdb@sourceware.org Subject: Re: Cannot fetch general-purpose registers for thread 1342445920: generic error Message-ID: <20060901002026.GA22921@nevyn.them.org> Mail-Followup-To: Chuck Simmons , bartoschek@or.uni-bonn.de, gdb@sourceware.org References: <44F77934.3000305@oracle.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <44F77934.3000305@oracle.com> User-Agent: Mutt/1.5.11+cvs20060403 X-IsSubscribed: yes Mailing-List: contact gdb-help@sourceware.org; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-owner@sourceware.org X-SW-Source: 2006-09/txt/msg00001.txt.bz2 On Thu, Aug 31, 2006 at 05:05:08PM -0700, Chuck Simmons wrote: > Historically, in my experience, GDB has rarely handled threads well. > (There was a year or two about 10 to 15 years ago where things worked > fairly well, but since then...) Here is a short session on Suse Linux > using Posix threads: > > " > (gdb) run 44123:127.0.0.1 dsk log.3382 > Starting program: /home/csimmons/src/sysmon/sysmon2 44123:127.0.0.1 dsk > log.3382 > [Thread debugging using libthread_db enabled] > [New Thread 1075110560 (LWP 31423)] > main: policy=2, prio=1 > [New Thread 1077214128 (LWP 31426)] > ping_cpu: policy=2, prio=1 > [New Thread 1079315376 (LWP 31427)] > Couldn't get registers: No such process. > " > > This is probably a timing related bug in the implementation of GDB. > When debugging programs that spawn threads using back to back gdb > sessions, some runs will allow one to do debugging, and some runs won't. Sorry, but this is basically useless as a bug report. If you would like GDB to be more reliable debugging your applications, you need to provide us with enough information to reproduce and diagnose the problem. If you do, I'll be glad to take a look at it as soon as time permits. -- Daniel Jacobowitz CodeSourcery