Re: NetBSD on Sun T5220

On Sun, 5 Oct 2025 09:53:09 +0000 (UTC)
Palle Lyckegaard <[email protected]> wrote:

> On Tue, 30 Sep 2025, Palle Lyckegaard wrote:
> 
> > Date: Tue, 30 Sep 2025 19:23:20 +0000 (UTC)
> >
> > More debugging to be done, but my understading of the issue has improved 
> > during the past weeks... :-)
> >
> >
> 
> Hi.
> 
> I have identified a potential issue in the locore.s file (pcbspill) where 
> %otherwin > 1 is not handles properly (at least in the sun4v issue I have 
> been debugging).
> 
> I should be able to provide a patch within a day or two - still unsure if 
> it fixes the issue 100% though and does not cause regression for sun4u.
> 
> Regards
> Palle
> 

OK, I can test when the patch is ready. SPARCv9 has a complex register
window handling mechanism and various ancillary registers to keep the
correct state. My idea was to start with the window trap handlers and
follow their execution, logging the register values. Sooner or later,
it would be possible to track the instructions which result in the
invalid state.