Re: kernel trap 12

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

:
:-On [20040716 23:32], Matthew Dillon (dillon@xxxxxxxxxxxxxxxxxxxx) wrote:
:>    Do a boot -v and write down the fault virtual address as well as a bit
:>    more of the backtrace (in particular, the function offsets).  Also I
:>    am assuming you used the default compiler and not gcc3.4.
:
:Yeah, 2.95.4, also removed any CPUTYPE or CXXFLAGS from make.conf.
:
:It's nexus_probe() that works, but when it reaches bus_generic_probe() at
:the end of that function, that's when it goes tits up.
:
:I always boot verbose, but that's not helping, trying BUS_DEBUG now.
:
:-- 
:Jeroen Ruigrok van der Werven <asmodai(at)wxs.nl> / asmodai / kita no mono

    Well, it's obviously a null function pointer fault but the question
    is what device was it trying to call through?  Try the following
    patch to see if better information can be gleaned.

					-Matt
					Matthew Dillon 
					<dillon@xxxxxxxxxxxxx>

Index: nexus.c
===================================================================
RCS file: /cvs/src/sys/i386/i386/nexus.c,v
retrieving revision 1.10
diff -u -r1.10 nexus.c
--- nexus.c	1 Mar 2004 06:33:16 -0000	1.10
+++ nexus.c	17 Jul 2004 01:04:52 -0000
@@ -171,7 +171,9 @@
 		return(ENXIO);
 #endif
 
+#if 0
 	device_quiet(dev);	/* suppress attach message for neatness */
+#endif
 
 	/*
 	 * IRQ's are on the mainboard on old systems, but on the ISA part
@@ -244,6 +246,8 @@
 	if (rman_init(&mem_rman)
 	    || rman_manage_region(&mem_rman, 0, ~0))
 		panic("nexus_probe mem_rman");
+
+	device_printf(dev, "nexus probe\n");
 
 	return bus_generic_probe(dev);
 }