Re: oskit-mach won't boot

[Roland McGrath]

> After rebooting a lot of times (I should really get a serial cable to do 
> remote debugging) I found out that the sti instruction near the of the 
> function is causing the troubles. At the moment I've no idea what's wrong, I
> have to read a lot of documentation and code before knowing what's going on. 
> :)

This probably the first time interrupts have been enabled since early in
the boot process.  There will immediately be a clock interrupt, since one
surely fired a little bit earlier and was blocked until the "sti" insn.
There may be other interrupts pending too.  So, not surprisingly, this
suggests that the interrupt handler is not set up properly.  If the setup
of the interrupt/trap table, and all the various x86 hoo-ha that's
necessary for it to be properly arranged to run in protected mode, is
futzed somehow then many of those errors will cause the CPU to reset.  I
don't think any of that stuff has really changed, so it's hopefully just
the case that the interrupt is jumping to the handler but the
oskit/oskit-mach interrupt handler code itself is somehow broken.
If that's the case, you might be able to catch the code at all_intrs (locore.S)
and see how far it gets from there.