On 23 Apr 1997, Charles M. Hannum wrote:

> That VA is the Sun 4m locations for the microsecond counter.  This is
> almost certainly due to a bug in the recent changes to locore, which I
> checked in a fix for a couple of days ago, that caused a SUN4 or SUN4C
> kernel without SUN4M to attempt to read the counter from the wrong
> location.  (On Sun 4c models, the effect was to cause microtime() to
> fail to interpolate between ticks -- which I didn't notice until I
> started xntpd on one machine and it failed to sync.)
> 
> Hopefully this change fixes the last of your problems!
Unfortunately not :-( The crash seems to be in the same routine
(cpu_initclock) as before the patching.

data fault: pc=f80e2f34 addr=fe000004 ser=8080<WRITE,INVAL>
panic: kernel fault
syncing disks... done
Frame pointer is at 0xf8133b58
Call traceback:
  pc = f8111998  args = (0, 11401fe1, f813d000, f8133c78, f802edd8, 10,
f8133bc0) fp = 0xf8133bc0
  pc = f802df74  args = (100, 0, 1, f8133ce4, f802edd8, f8133c7c,
f8133c28) fp = 0xf8133c28
  pc = f8117f6c  args = (100, f8135c00, 1, f8133cf0, 0, 3, f8133c90) fp = 
0xf8133c90
  pc = f80062bc  args = (9, 8080, fe000004, f80e2f34, 119010c4, f8133d98,
f8133d38) fp = 0xf8133d38
  pc = f80e2ed8  args = (a0, 9c4400, fe000000, 2710, f813a000, 64,
f8133de8) fp = 0xf8133de8
  pc = f801d0c4  args = (f813ec00, 1, f8136000, f8138800, ffffffff, c8,
f8133e50) fp = 0xf8133e50
  pc = f801c474  args = (f814aa68, 499c000, d, 10000000, ffffffff, 3,
f8133eb8) fp = 0xf8133eb8
  pc = f80070ac  args = (0, 8, 1, f81340bf, ffffffff, 10, f8133f30) fp = 
0xf8133f30
  pc = 344efc  args = (347800, 340bd8, 33fefc, 33fea4, 342a34, 0,
f7ffffa0) fp = 0xf7ffffa0

Ralf
--
Ralf Sauther -- rhs@inka.de -- Keine Kommerz-Mailings erwuenscht !