panics after updating to RELENG_4 aug 25 from May 17th

[Doug White]

On Sun, 29 Aug 2004, Mike Tancsa wrote:

>
> The box was quite stable prior to this with well burnt in hardware. Both
> panics seem to be in the same location.  The box functions as an outbound
> smtp smarthost running stock sendmail.  dmesg and kernel config at the end

The double faults aren't helping to debug this. But at first glance it
kinda looks like some sort of memory corruption.

Here's the meat of the first trace:

> #16 0xc014013e in statclock (frame=0xc023a130) at
> /usr/src/sys/kern/kern_clock.c:439
> #17 0xc01ff66d in rtcintr (frame={cf_vec = 0, cf_ppl = 0, cf_fs = 16, cf_es
> = -65520, cf_ds = 16, cf_edi = -1, cf_esi = 0,
>        cf_ebp = -1071406728, -1071406748, cf_ebx = 0, cf_edx = 2517, cf_ecx
> = -425803104, cf_eax = 0, 0, 0, cf_eip = -1071702170,
>        cf_cs = 8, cf_eflags = 514, cf_esp = 0, cf_ss = -1071670027}) at
> /usr/src/sys/i386/isa/clock.c:385
> #18 0xc01f1f66 in cpu_idle () at /usr/src/sys/i386/i386/machdep.c:1000

statclock() doesn't do a whole lot. Line 439 (in RELENG_4 at least) is:

439                 if (--pscnt > 0)
440                         return;

pscnt is a static local variable.  Not likely thats going to fly off into
space.  Please check this on your install.  If its the same thing I'd
start suspecting hardware failure.

Dumping lists of the generic trap code isn't telling us anything new :)

-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite at gumbysoft.com          |  www.FreeBSD.org