From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 6451 invoked by alias); 5 Oct 2005 20:31:34 -0000 Mailing-List: contact gdb-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-owner@sources.redhat.com Received: (qmail 6442 invoked by uid 22791); 5 Oct 2005 20:31:32 -0000 Received: from webmail.streamline-computing.com (HELO webmail.streamline-computing.com) (82.133.39.164) by sourceware.org (qpsmtpd/0.30-dev) with ESMTP; Wed, 05 Oct 2005 20:31:32 +0000 Received: from [82.15.10.199] (helo=cpc2-oxfd8-3-0-cust199.oxfd.cable.ntl.com) by webmail.streamline-computing.com with esmtpa (Exim 4.52) id 1ENFq7-0005C2-BY; Wed, 05 Oct 2005 21:26:11 +0100 Subject: Re: ptrace PEEKTEXT IO error?! From: David Lecomber To: Daniel Jacobowitz Cc: gdb In-Reply-To: <20051005195438.GA13350@nevyn.them.org> References: <1128531841.21837.96.camel@delmo.priv.wark.uk.streamline-computing.com> <20051005171106.GA7835@nevyn.them.org> <1128536411.21837.123.camel@delmo.priv.wark.uk.streamline-computing.com> <20051005184121.GA10564@nevyn.them.org> <1128542780.3071.7.camel@cpc2-oxfd8-3-0-cust199.oxfd.cable.ntl.com> <20051005195438.GA13350@nevyn.them.org> Content-Type: text/plain Date: Wed, 05 Oct 2005 20:31:00 -0000 Message-Id: <1128545245.3071.9.camel@cpc2-oxfd8-3-0-cust199.oxfd.cable.ntl.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit X-SW-Source: 2005-10/txt/msg00019.txt.bz2 On Wed, 2005-10-05 at 15:54 -0400, Daniel Jacobowitz wrote: > On Wed, Oct 05, 2005 at 09:06:20PM +0100, David Lecomber wrote: > > > > > > > > > > It seems ok at reading memory from addresses on the heap and stack -- > > > > but it barfs at reading memory from the text segment. [snip] > > > > > No, this is most likely a kernel bug. This is mapped code, right, not > > > some mmaped device file (which is not generally accessible by ptrace)? > > > > The code I'm trying to read is an ordinary subroutine in an ordinary C > > file. The program does call a library handling infiniband network > > cards, and there are loads of kernel modules on this machine looking > > after that. > > The infiniband code does all sorts of funky things to VM. It's > extremely likely that this is repsonsible. Great, I bet this bug'll cheer them up! Thanks for the info, I'll chase their mailing list now. Cheers David