Re: Dockstar (like SHEEVAPLUG) panics at init with -current after Aug 31, 2012

[David Laight <david%l8s.co.uk@localhost>]

On Wed, Dec 19, 2012 at 01:30:54PM -0700, Andy Ruhl wrote:
> > r4 is odd ... which is why it faulted.
> > Looks like it is actually 4 characters "ev (", find where it came from,
> > find the full string and you might have a lead.
> 
> I'm not really talented enough to debug this stuff.
> 
> Here's what the ascii characters look like I think:
> 
> root on sd0a dumps on sd0b
> WARNING: clock lost 4734 days
> WARNING: using filesystem time
> WARNING: CHECK AND RESET THE DATE!
> warning: no /dev/console
> init path (default /sbin/init):
> init: copying out path `/sbin/init' 11
> data_abort_handler: data_aborts fsr=0x1 far=0x28207697
> Fatal kernel mode data abort: 'Alignment Fault 1'
> trapframe: 0xc7efcd00
> FSR=00000001, FAR=28207697, spsr=20000013
> r0 =00000004, r1 =00000000, r2 =c207d010, r3 =00000000
> r4 =28207665, r5 =00000000, r6 =00000000, r7 =c207d010
> r8 =c04c40fc, r9 =00000000, r10=00000000, r11=c7efcd88
> r12=00000001, ssp=c7efcd50, slr=c02c5364, pc =c02c53f0
> 
> Stopped in pid 2.1 (sh) at      netbsd:uvm_pagealloc_pgfl+0xbc: ldrh    r3, 
> [r4,
>  #0x32]
> db> x/s 0xc207d010
> c207d010:       ev (622592 byte, 1184 inodes)\012
> db> x/s 0xc207d008
> c207d008:       tmpfs /dev (622592 byte, 1184 inodes)\012
> db> x/s 0xc207d004
> c207d004:       ted tmpfs /dev (622592 byte, 1184 inodes)\012
> db> x/s 0xc207d000
> c207d000:       Created tmpfs /dev (622592 byte, 1184 inodes)\012
> 
> Does that help?

I think that is a write() from init ?

The next thing to determine is whether that string has been written
to the wrong place, or whether the pointer used for the read is wrong.

This is all early enough that you might get away with adding a kernel
printf in the syscall entry path and the user space fault path.
I'd run staticly linked init and sh just to reduce the calls.

        David

-- 
David Laight: david%l8s.co.uk@localhost