From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 19720 invoked by alias); 4 Mar 2006 15:11:09 -0000 Received: (qmail 19712 invoked by uid 22791); 4 Mar 2006 15:11:09 -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, 04 Mar 2006 15:11:07 +0000 Received: from drow by nevyn.them.org with local (Exim 4.54) id 1FFYPR-0005Zi-I7; Sat, 04 Mar 2006 10:11:05 -0500 Date: Mon, 06 Mar 2006 18:38:00 -0000 From: Daniel Jacobowitz To: Alexandre Oliva Cc: gdb-patches@sourceware.org Subject: Re: Prelink.exp troubles Message-ID: <20060304151105.GE20187@nevyn.them.org> Mail-Followup-To: Alexandre Oliva , gdb-patches@sourceware.org References: <20060301200540.GE6465@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: 2006-03/txt/msg00102.txt.bz2 On Fri, Mar 03, 2006 at 11:04:21AM -0300, Alexandre Oliva wrote: > On Mar 1, 2006, Daniel Jacobowitz wrote: > > > First: you're using gdb_expect directly. I'm sorry I didn't notice > > this before. Don't do that; use gdb_test_multiple instead. > > Tricky to catch two independent messages triggered by the same command > that way while still producing a fail if either fails to match, so I > put them now in a single regexp. Actually, it's very easy: set a flag when you see the each message, exp_continue if you haven't seen the other. This is fine too. > > We need to either copy the libraries around (complicated, fragile), or > > punt if the system ld.so is not prelinked. > > Punting it is :-) Tricky, since prelink doesn't report an error in > this case. Fortunately prelink -u does. > > How's this? I've tested it on amd64-linux-gnu, with and without > prelinked system libraries. I got the warning in the latter case, and > a pass in the former. Hmm, two things: I don't think a WARNING is appropriate (there's nothing wrong), how about just an UNTESTED? And, when matching patterns in gdb_test_multiple (the same was true in gdb_expect) it's important to anchor the end of the pattern to the GDB prompt. Otherwise that prompt might not be consumed by expect, causing new tests added at the end of the file to fail. If you agree about the UNTESTED, with those changes it's fine to check in. My test runs thank you. -- Daniel Jacobowitz CodeSourcery