[haiku-commits] Re: haiku: hrev46863 - headers/private/kernel/arch/arm
- From: François Revol <revol@xxxxxxx>
- To: haiku-commits@xxxxxxxxxxxxx
- Date: Fri, 14 Feb 2014 18:31:59 +0100
On 14/02/2014 17:58, Stephan Aßmus wrote:
> However what about removing ports that have no perspective of ever being
> useful
> for anything to reduce this additional work? That includes IMHO the
M68K port.
>
>> There is also a FireBee clone with 274MHz ColdFire and 512M of RAM:
>> http://acp.atari.org/
>> However the coldfire would need some more code since it differs
>> significantly from m68k.
>
> The emulator argument is completely pointless, IMHO.
> Maybe Haiku would run on that FireBee clone you mention.
> But seriously, is that important enough to cause kernel
> devs more work (in the individual case of the m68k port)?
Well, I never really intended the m68k port to be useful for daily use,
but more as a proof of concept. Still, when I started it, Haiku booted
fine on QEMU with 40MB of RAM, which I'm not sure would still be the
case now, due to all the debug stuff we added (but that can be
disabled), and I'm not sure the PM stuff helps with that either.
Hopefully we still have room for optimization there.
I still intend to try and make it boot on m68k though.
Now, it's not like it caused that much more work except to me yet, I
recall spending nights rewriting the VM code after Ingo's changes there.
It surely helped making it cleaner and simpler to separate all the 68k
variants, but of course the rest staled as a result.
Next time I work on it I'll do that on a branch for a while though, just
like I did with the Sam port, as that limits the interference with the
rest of development. It makes it a bit harder to merge things later though.
François.
Other related posts:
