In some mail I received from Ignatios Souvatzis, sie wrote
> 
> Okok...
> 
> as for the official way to boot:
> 
> please look at sys/arch/amiga/amiga/locore.s, line 2070 (in 1.2) and down.
> 
> Basically, we copy ourselves (if on an Amiga) to chipmem, map it, jump
> there, switch MMU off, then execute the official way to reboot an
> Amiga (tm).
> 
> I don't know why this wouldn't work on any Amiga, but basically, even
> the official way seems to have some voodoo components...
> 
> Note that we can't call the ColdReboot() vector of exec.library (which
> is the even more official way as of newer ROM version) because we
> destructed all of execbase... So we MIGHT have a problem.
> 
> However, the RESET, which is guaranteed to be executed by our code,
> should at least take care of the A3000 scsi problem. 
> 
> The rekick-A3000's (with the V36 ROMs) will do a nearly cold reboot,
> of course.

It was mentioned that once synchronous SCSI is negoitiated on, it stays
set until turned off.  Is it worth renegoiating drives back to async
after reboot and drives have been flushed ?

What is the default for SCSI HD's when they're powered on ?  Do any of
the amiga ROM's successfully negoiate and turn on synchronous SCSI ?

Darren