PERFORCE change 143121 for review

[Bernd Walter]

On Mon, Jun 09, 2008 at 09:17:54PM +0200, Hans Petter Selasky wrote:
> Hi Bernd,
> 
> When I re-enabled:
> 
> options         PSEUDOFS
> 
> in my kernel configuration file, everything worked again!

That sounds very much like you triggered the same well known problem.
It is not pseudofs, just the slight change in timing, memory layout or
whatever that made it go away.
I would really like to see this bug catched, but it is difficult to
debug.
I also have to admid that I'm also a bit to lazy, since it can be
worked around too easily...

> On Monday 09 June 2008, Bernd Walter wrote:
> > On Sun, Jun 08, 2008 at 11:13:03PM +0200, Hans Petter Selasky wrote:
> > > Hi,
> > >
> > > It might look like that my "init" executable in the file system is out of
> > > date. The kernel boots fine up to the following point:
> > >
> > > Trying to mount root from ufs:/dev/md0
> > > warning: no time-of-day clock registered, system time will not be set
> > > accurately
> > > Jun  4 20:32:47 init: login_getclass: unknown class 'daemon'
> > > Fatal kernel mode data abort: 'Alignment Fault 3'
> > > trapframe: 0xc67f7a24
> >
> > We've seen this at least on RM9200 from time to time - not sure about
> > other arm.
> > Try adding or removing a driver to your kernel and see if it helps.
> > It's some kind of heisenbug, because a small unrelated change and the
> > problem is away, which makes is hard to debug.
> > See arm@ mail archives for more details.
> > IIRC it happened earlier during boot, so this can still be something
> > different after all.
> > Even with an out of date init the kernel shouldn't panic with an
> > alignment fault.
> 
> 
> 

-- 
B.Walter <bernd at bwct.de> http://www.bwct.de
Modbus/TCP Ethernet I/O Baugruppen, ARM basierte FreeBSD Rechner uvm.