Re: [PATCH] Fixup for Amigas with memory at 0x200000

[Gunther Nikl <gni%gecko.de@localhost>]

On Mon, Jan 13, 2003 at 07:47:55PM +0100, Ignatios Souvatzis wrote:
> Personally, I'd rather see the bootblocks used.

  I understand the rationale behind it: only one program to care about (and
  buildable with the target toolchain)

> If somebody wants to contribute an AmigaOS-startable booter, I'd not object,
> but I'd prefer to have it NetBSD-buildable so that the base of the release
> can be built without access to AmigaOS (and without access to an Amiga).

  Is a cross-compiler acceptable? ;-)

On Mon, Jan 13, 2003 at 10:38:32PM +0100, Ignatios Souvatzis wrote:
> >   I know. However, you didn't answer the question :-/ (or I don't understood
> >   the anwser) Since you wrote that a you can implement a specific feature
> >   only because there is two-stage boot process, it sounds that booting
> >   directly into the kernel won't work after you implemented your change.
> 
> ah that . No - thats just a question of allowable code size. We managed 
> to squeeze the bootblock into the 8k available on 8k/1k bsd ffs for it, 
> but it was always a tight fit.

  Good, a size issue. That wasn't obvious for me.

  Another question about the bootloader: Why doesn't it support the p-flag?
  The handling of the memlist is completely different with the one of the
  old loadbsd. Is it easier to deal with that issue at boot time? 
  And yet another observation: the marks[MARK_ENTRY] is not consistent. At
  one place its used directly and at another MARK_START is subtracted. My
  guess because MARK_START is zero, it doesn't matter.

  Gunther

-- 
Linux is only free if your time has no value
 - Jamie Zawinsky