From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 22808 invoked by alias); 29 Jan 2010 15:37:36 -0000 Received: (qmail 22797 invoked by uid 22791); 29 Jan 2010 15:37:36 -0000 X-SWARE-Spam-Status: No, hits=-2.5 required=5.0 tests=AWL,BAYES_00,SPF_PASS 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; Fri, 29 Jan 2010 15:37:31 +0000 Received: (qmail 8878 invoked from network); 29 Jan 2010 15:37:29 -0000 Received: from unknown (HELO caradoc.them.org) (dan@127.0.0.2) by mail.codesourcery.com with ESMTPA; 29 Jan 2010 15:37:29 -0000 Date: Fri, 29 Jan 2010 15:37:00 -0000 From: Daniel Jacobowitz To: Joel Brobecker Cc: gdb-patches@sourceware.org Subject: Re: RFC: Fix testsuite timeout clobbers Message-ID: <20100129153726.GC30542@caradoc.them.org> Mail-Followup-To: Joel Brobecker , gdb-patches@sourceware.org References: <20100128215305.GA2813@caradoc.them.org> <20100129035950.GB26827@adacore.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20100129035950.GB26827@adacore.com> User-Agent: Mutt/1.5.20 (2009-06-14) 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: 2010-01/txt/msg00643.txt.bz2 On Fri, Jan 29, 2010 at 07:59:50AM +0400, Joel Brobecker wrote: > Independently of that, do we want to try a different approache where > the timeout gets reset more systematically? For instance, I can propose: > everytime gdb_start is called, reset the timeout to the default value > (which itself should be configurable - through a site.exp or board file?) It sounds good - we'd have to audit everywhere that tests did set the timeout, though, since if it ends up before gdb_start or there are multiple calls to gdb_start, it'd get reset. Does DejaGNU give us a per-start-of-exp-file hook? Meanwhile, I'll check this patch in now. > On the same topic (timeouts), I ran the testsuite on sparc-solaris, > yesterday, and some tests were badly timing out, and each timeout > was taking what it felt like hours (the testsuite itself took more > than 2 hours, and that's after I justed killed -9 the inferiors from > the tests). With AdaCore's testsuite, a timeout means we've lost sync > with debugger anyway - is there really an advantage to continuing a > testcase when we get a timeout? Wouldn't it just as effective to abort > the testcase after the first timeout? IMO this is an excellent idea. I wonder if we shouldn't resurrect (but, working this time!) the suppression mechanism? -- Daniel Jacobowitz CodeSourcery