5.4-RCx "jumping to kernel entry" hang on Ultra1/2/30

[Gavin Atkinson]

On Wed, 2005-04-27 at 12:04 -0700, Marcel Moolenaar wrote:
> On Wed, Apr 27, 2005 at 07:08:52AM -0800, Royce Williams wrote:
> > 
> > On Sun, 17 Apr 2005, IX at pandora.be wrote:
> > >
> > >
> > >On Thu, 14 Apr 2005, Gheorghe Ardelean wrote:
> > >
> > >>Hi,
> > >>
> > >>While trying to boot a plain U1 from a disk installed with
> > >>5.4-RC2 
> > >>(the disk was installed in a U5) the systems hangs just
> > >>after entering the kernel.
> > >
> > >
> > >I see the same hang on an ultra2, booting from the 5.4RC2
> > >bootonly iso. I don't know what details I need to provide.
> > >If it's useful, I can send a NetBSD dmesg output.
> > >
> > >Kind regards,
> > >dieter
> > 
> > 
> > This is happening to me too, 5.4-RC3 disc 1, Ultra 30.  I can't 
> > recreate using 5.2.1R or 5.3R disc 1 ISOs.  This is now reportedly 
> > happening to three different people with three different models. Is 
> > there a boot-only ISO with full debugging/verbosity available?
> 
> I think I have a fix for it, but it needs to be verified that to
> really solves the problem before I run off to re@ and shout: "stop
> the presses!"
> 
> The patch is attached. I just don't know a good way (short of
> creating a patched-up 5.4-RC install ISO for people to try it.
> 
> Ken: did you build 5.4-RC3 for sparc64 and if yes do you still have
> the $CHROOTDIR around?

I don't know if it's a useful datapoint for you, but I have a SunFire
280R (UltraSparc III processors, and therefore unsupported) which hangs
at this point as well.  Your patch does not help.  Reverting to
ofw_console fixes it and lets the boot continue, so I guess you're right
in the assumption that the problem is in the uart driver.

The serial chipset seems to be either a sab82532 or a 16550, depending
on which part of the device tree I should believe.

If this information is of use to you, I have access to a netbooted setup
where I can trivially test patches for you on the 280R.

Gavin