From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 21902 invoked by alias); 31 Oct 2005 19:04:34 -0000 Mailing-List: contact ecos-discuss-help@ecos.sourceware.org; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: ecos-discuss-owner@ecos.sourceware.org Received: (qmail 21857 invoked by uid 22791); 31 Oct 2005 19:04:30 -0000 Received: from londo.lunn.ch (HELO londo.lunn.ch) (80.238.139.98) by sourceware.org (qpsmtpd/0.30-dev) with ESMTP; Mon, 31 Oct 2005 19:04:30 +0000 Received: from lunn by londo.lunn.ch with local (Exim 3.36 #1 (Debian)) id 1EWex2-0007gv-00; Mon, 31 Oct 2005 20:04:12 +0100 Date: Mon, 31 Oct 2005 19:04:00 -0000 To: Klaas Gadeyne Cc: ecos-discuss@ecos.sourceware.org Message-ID: <20051031190412.GQ4266@lunn.ch> Mail-Followup-To: Klaas Gadeyne , ecos-discuss@ecos.sourceware.org References: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.11 From: Andrew Lunn Subject: Re: [ECOS] ecos test infrastructure X-SW-Source: 2005-10/txt/msg00291.txt.bz2 On Mon, Oct 31, 2005 at 04:54:00PM +0100, Klaas Gadeyne wrote: > I am currently looking at the (host side) test infrastructure ecos currently > provides and I have several questions to which I couldn't find an > answer in the docs. > > I read in several places, e.g. > > > that the test infrastructure will be worked on in future ecos > versions. The last updates to README.host are dating from feb 2003. > I was wondering if anything is to be expected here in the near future > (say a couple of months)? No, i doubt any work will take place here in the near future. It is one of the selling points of eCosCentric that they have a testfarm. If they make it too easy for you to run the tests yourself you would not pay them money for the use of their test farm. Having said that, if somebody was to contribute code i would gladly incorpertate it, if it met the usual quality standards. This is open source after all.... > Furthermore, Chapter 22 of the user-guide describes how to build/run a test > suite from configtool. When I try that approach, building goes fine, > but when I try to run a test program, I get the same "failed to get > pty error" as described here > > However I have the LEGACY_PTYS option enabled in my kernel. Any > suggestions? That was just a guess. I had seen a similar problem with another application with 2.6. You need to debug the problem further. Try running strace and see what configtool is trying to actually use when it fails. If you can figure that out you might be able to adjust your kernel. Alternatively you might be able to fix configtool. There are instructions on how to compile it at http://www.ecoscentric.com/devzone/configtool.shtml I'll warn you thought, it is not easy. > Running a single test program from within gdb works fine. > Would it be possible with the current infrastructure to run a > testsuite on the target system without using the graphical configtool > and report the results to a database instead? [*] > > Is there any more documentation about the testing infrastructure > (e.g.~about the use of dejagnu) than what I found in user-, ref-, and > component writers guide and the above described README.host? Sure it is possible. You can look at the sources for the configtool. I can also make an educated guess at how it works. I think it will set breakpoints in cyg_test_exit() & cyg_assert_fail(). The breakpoints probably have gdb commands attached which causes gdb to exit. gdb will be started with a script which loads the image, starts login to a file, sets the breakpoints and then runs the program. I guess there is also a timeout machanism to detect stuck testcases. > [*] I guess the test farm at ecoscentric must do something similar... I've never seen eCosCentric's code, since it is there interlectual property, but i guess it does. Andrew -- Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss