Subject: Re: NetBSD X11 client doesn't like SGI X11 Server
To: Jochen Kunz <jkunz@unixag-kl.fh-kl.de>
From: Frederick Bruckman <fredb@immanent.net>
List: tech-x11
Date: 12/13/2004 09:15:13
In article <20041213100007.2bdd7261.jkunz@unixag-kl.fh-kl.de>,
	jkunz@unixag-kl.fh-kl.de (Jochen Kunz) writes:
> Hi.
> Since NetBSD switched to XF86 4.x I can't use my SGI machines as remote
> display for NetBSD X11 clients. The NetBSD machine is a Sun Ultra 10
> running a 32 Bit kernel with NetBSD/sparc userland. The SGI is an Octane
> with ESSI + ESI dual head GFX running xinerama. I had no problems using
> the Octane as remote display with NetBSD when the NetBSD X11 was based
> on XF86 3.x.
> Symptome:
> [jkunz@Zimbo jkunz]$ export DISPLAY=MrPomeroy:0.0
> [jkunz@Zimbo jkunz]$ xclock
> Xlib: Did not parse entire setup message: parsed: 848, message: 1600
> Error: Can't open display: MrPomeroy:0.0

It looks like you're coming up against this XFree86 revision 3.13:

  http://cvsweb.xfree86.org/cvsweb/*checkout*/xc/lib/X11/OpenDis.c?rev=3.13

which was in response to this bug:

  http://bugs.xfree86.org/show_bug.cgi?id=40

The part that's bothering you is just a small part of the patch, with
a comment that says "sanity check". Most of the changes check that the
server message was long enough, to prevent operating on uninitialized
memory, but this is just saying that the server returned too much data,
is what it looks like to me.

Perhaps we could make this just a warning?


Index: xfree/xc/lib/X11/OpenDis.c
===================================================================
RCS file: /cvsroot/xsrc/xfree/xc/lib/X11/OpenDis.c,v
retrieving revision 1.1.1.5
diff -u -r1.1.1.5 OpenDis.c
--- xfree/xc/lib/X11/OpenDis.c	5 Mar 2004 14:24:08 -0000	1.1.1.5
+++ xfree/xc/lib/X11/OpenDis.c	13 Dec 2004 15:03:15 -0000
@@ -582,8 +582,10 @@
 	    fprintf(stderr, "Xlib: Did not parse entire setup message: "
 	                    "parsed: %ld, message: %ld\n",
 		    usedbytes, setuplength);
+#if 0 /* Non-fatal ??? */
 	    OutOfMemory(dpy, setup);
 	    return(NULL);
+#endif
 	}
 
 /*

In any case, would you please file a problem report (with NetBSD)?  Even
if the fix turns out to be that simple, we'll still need a record for the
upstream maintainers.


Frederick