Just FYI, I showed the paniced kernel to MRG who passed it down the 
   line, the suspect is altavec.  (But it shouldn't panic the kernel anyway :-)


charles said that trap was maybe altivec, but in any case, it is a
user trap and it shouldn't cause a kernel panic in any case.  someone
should try this (untested) patch:


Index: trap.c
===================================================================
RCS file: /cvsroot/syssrc/sys/arch/powerpc/powerpc/trap.c,v
retrieving revision 1.53
diff -p -r1.53 trap.c
*** trap.c	2001/10/18 01:33:48	1.53
--- trap.c	2001/11/08 04:35:56
*************** syscall_bad:
*** 409,414 ****
--- 409,420 ----
  	}
  
  	default:
+ 		if (type & EXC_USER) {
+ #ifdef DIAGNOSTIC
+ 			printf("user trap type %x at %x\n", type, frame->srr0);
+ #endif
+ 			break;
+ 		}
  brain_damage:
  		printf("trap type %x at %x\n", type, frame->srr0);
  brain_damage2: