At 9:26 Uhr +1000 10.7.2007, matthew green wrote:
   >has anyone tried netbsd-4 on these systems?  i've been reading
   >sparc changes between netbsd-2 and netbsd-3 and nothing seems
   >obviously at fault.
   
   I've got a dual-SM50 (no 2nd level cache) ss20 running as serial console
   server at work, and when I upgraded it to 4.0 beta the MP kernel showed
   exactly the same symptoms as before. So, yes, the problem remains. I
   haven't tried -current kernels, though.

ok, thanks.  that probably means it isn't fixed in -current
either, tho -current has worse problems on MP systems it
seems...

i have a couple of ideas for you to test, tho i seem to have
misplaced my notes.  one was to revert the data/stack size
limits that were changed in netbsd-3 times for sparc, and the
patch at the end should do this...

i'll send the other ideas this evening.


.mrg.


Index: vmparam.h
===================================================================
RCS file: /cvsroot/src/sys/arch/sparc/include/vmparam.h,v
retrieving revision 1.36.10.1
retrieving revision 1.35
diff -p -r1.36.10.1 -r1.35
*** vmparam.h	14 Feb 2006 13:23:22 -0000	1.36.10.1
--- vmparam.h	7 Aug 2003 16:29:42 -0000	1.35
***************
*** 88,100 ****
  #define	DFLDSIZ		(64*1024*1024)		/* initial data size limit */
  #endif
  #ifndef MAXDSIZ
! #define	MAXDSIZ		(512*1024*1024)		/* max data size */
  #endif
  #ifndef	DFLSSIZ
! #define	DFLSSIZ		(8*1024*1024)		/* initial stack size limit */
  #endif
  #ifndef	MAXSSIZ
! #define	MAXSSIZ		(32*1024*1024)		/* max stack size */
  #endif
  
  /*
--- 88,100 ----
  #define	DFLDSIZ		(64*1024*1024)		/* initial data size limit */
  #endif
  #ifndef MAXDSIZ
! #define	MAXDSIZ		(256*1024*1024)		/* max data size */
  #endif
  #ifndef	DFLSSIZ
! #define	DFLSSIZ		(2*1024*1024)		/* initial stack size limit */
  #endif
  #ifndef	MAXSSIZ
! #define	MAXSSIZ		MAXDSIZ			/* max stack size */
  #endif
  
  /*