Re: /etc/mk.conf

[Andrew Daugherity <andrew_(_dot_)_daugherity_(_at_)_gmail_(_dot_)_com>]

On 4/17/05, Nick Holland <nick_(_at_)_holland-consulting_(_dot_)_net> wrote:
> > Is mac68k still being actively maintained?
> 
> The answer to that is "definitely".  Martin@ and Miod@ have done some
> incredible work to bring the mac68k platform to a much more usable
> level.  Users of OpenBSD/mac68k 3.6 and before will barely recognize the
> 3.7 release.  A *few* changes:
>   A usable installation system (bsd.rd-based, like other platforms)

Hallelujah!  Besides lack of support for ftp or nfs installation, the
Mac OS installer was just plain slow, and took about three hours to
just install base36, etc36, and kernel.  Installing the corresponding
NetBSD sets (with its 'sysinst' bsd.rd installer) took maybe 1/4 that
time (and that was installing via NFS, rather than local disk).

> Unfortunately, it sounds like OpenBSD and your Q605 are still not
> getting along.  I'm not overly surprised, as of all the machines that
> Martin and I have acquired to test, the Q605 is not one of them.  And,
> from a brief glance at the details of the machine, it sounds more like
> an LC (broke) than the other Quadras which do work.

Yep, the Q605, LC475, and Performa 475/476 have identical
motherboards.  There's actually a jumper that when closed, the machine
identifies itself as a Q605, and when open, it's an LC475.  After
replacing the 68LC040 with real '040, I was all set to run A/UX (for
shits and grins, if nothing useful), until I discovered it didn't
support the LC video hardware.  And the reason I have the Q605 in the
first place is because I was given an old LCII, which I promptly
discovered royally sucks, and after doing a bit of research, found
that the Q605/LC475 motherboard fits perfectly in the LC II/III case,
so off to eBay I went.

Perhaps I should post this to port-mac68k rather than misc, but I'll
be glad to test any new kernels on my Q605.

> > To veer back on-topic:
> >     You say -pipe is "safe" but then go onto to describe how it's
> > unsafe on mac68k... huh?  Or was that a gcc bug that's since been
> > fixed, and it's now OK to use -pipe on mac68k?
> 
> It was a GCC bug in 3.6 and before which is now fixed in 3.7
> It was a vivid demonstration of why you don't just blindly throw
> "optimizations" at OpenBSD and expect magic to happen.  The -pipe
> optimization should have been "safe" -- producing no code change, and
> yet, it doesn't work as desired in all places.  And people are expecting
> to mindlessly "rice" their OpenBSD box, "because I can, and you should
> help me".
Yep, sounds like the "Gentoo syndrome", as parodied at funroll-loops.org.

-Andrew