From owner-freebsd-alpha Thu Oct 3 14:38:34 2002 Delivered-To: freebsd-alpha@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 27F0E37B401 for ; Thu, 3 Oct 2002 14:38:32 -0700 (PDT) Received: from gatekeeper.oremut01.us.wh.verio.net (gatekeeper.oremut01.us.wh.verio.net [198.65.168.16]) by mx1.FreeBSD.org (Postfix) with ESMTP id 9B0E243E75 for ; Thu, 3 Oct 2002 14:38:31 -0700 (PDT) (envelope-from fclift@verio.net) Received: from mx.dmz.orem.verio.net (mx.dmz.orem.verio.net [10.1.1.10]) by gatekeeper.oremut01.us.wh.verio.net (Postfix) with ESMTP id 1AACD3BF163 for ; Thu, 3 Oct 2002 15:38:31 -0600 (MDT) Received: from vespa.dmz.orem.verio.net (vespa.dmz.orem.verio.net [10.1.1.59]) by mx.dmz.orem.verio.net (8.11.6/8.11.6) with ESMTP id g93LcU343332 for ; Thu, 3 Oct 2002 15:38:30 -0600 (MDT) Date: Thu, 3 Oct 2002 15:43:01 -0600 (MDT) From: Fred Clift X-X-Sender: To: Subject: xdm build/runtime problems Message-ID: <20021003152717.M88080-100000@vespa.dmz.orem.verio.net> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org I cvsuped sometime on Sunday, and have been running 4.7-RC #0 on my miata (pws 433au) since then. Around the same time, I finally got my matrox millenium I card working (put in 64 bit slot, and fiddle with dip-switches on matrox card -- hours wasted by not looking at those two little switches...) I seem to remember that there was some discussion in July of xdm dumping core for some people, and at least one person claimed xdm has been broken for along time. At any rate, I was finally in a position to set up X - built from ports - mostly working now. Since I normally start X with xdm, once I had a kind of working XF86config, I started to play with xdm. Interestingly, when building the XFree86-4-clients port, the build of xdm fails (in greeter) with the following error: ---------------------------------------------------------------- rm -f libXdmGreet.so.1.0~ + cd . + cc -o ./libXdmGreet.so.1.0~ -shared -rpath /usr/X11R6/lib -Wl,-soname,libXdmGreet.so.1 greet.o verify.o Login.o -L/usr/ports/x11/XFree86-4-clients/work/xc/exports/lib -lXt -lSM -lICE -lXext -lX11 /usr/libexec/elf/ld: cannot find -lXt *** Error code 1 Stop in /usr/ports/x11/XFree86-4-clients/work/xc/programs/xdm/greeter. ---------------------------------------------------------------- The easy obvious fix is to include -L/usr/X11R6/lib on the command line which can be done in a thousand ways (ie edit the makefile...) The _right_ way to do this for this complicated build system eludes me, but for ports it could be as simple as patching the Imakefile to include this in the standard link options. So then libXdmGreet.so builds fine, and then xdm builds fine. Of course, it dumps core for me and was not working. -nodaemon -debug 100 provided little useful information. So, I rebuilt with debugging symbols and found that the core-dumps are hosed enough that you can't get a useful stack-trace out of them. Much binary-search with breakpoints revealed that the xdmauth code seems to totally screw up the stack somewhere - First I compiled it out (undef HASXDMAUTH). Then, I decided to get a bit smarter - I added the following lines to /usr/X11R6/lib/X11/xdm/xdm-config DisplayManager._0.authName: MIT-MAGIC-COOKIE-1 DisplayManager._1.authName: MIT-MAGIC-COOKIE-1 to explicitly exclude the xdmauth stuff and presto xdm was working for me. So now, xdm works for me. I don't really have the time to track down what exactly about the xdmauth stuff is hosed, so I guess I'm only bringing this up so that someone with enough time/motivation can really fix it, and for the rest of us, we can at least use xdm to start X on our local machines by using the config-file workaround to disable xdmauth. Should this be in a pr? Are others having this work for them without any flaws and my X config is just non-standard? I'll happily put all this info in a PR if anyone else can confirm that it is not just me and my crusty box. One of these days I might actually have time to try and track down what is really wrong. Fred -- Fred Clift - fclift@verio.net -- Remember: If brute force doesn't work, you're just not using enough. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message