gabriel rosenkoetter on Tue, 8 Jul 2003 00:41:15 -0400

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] [OT] xmacppc, netbsd 1.6.1, ppc8500

On Mon, Jul 07, 2003 at 08:05:50PM -0400, Fred K Ollinger wrote:
> I thought I installed everything, but who knows. I'm probably going to
> reinstall anyway so I'll look out for it. Incidentally, I had scoured
> /usr/X11R6/bin/ for it, and I did not find it, though I did find regular
> X, which did not work as should be expected in this case.

X should just be a sym link to the local X server:

grappa:~% uname -a                                                          [1]
NetBSD grappa 1.6K NetBSD 1.6K (GRAPPA-$Revision: 1.533 $) #3: Tue Jan 21 22:32:43 EST 2003     gr@grappa:/new/src/netbsd/src/sys/arch/i386/compile/GRAPPA i386
grappa:~% ls -l `which X`                                                   [2]
lrwxr-xr-x  1 root  wheel  7 Apr 10  2002 /usr/X11R6/bin/X@ -> XFree86

uriel:~% uname -a                                                           [1]
NetBSD uriel 1.6.1_RC2 NetBSD 1.6.1_RC2 (URIEL) #0: Wed Mar 26 19:22:52 EST 2003     gr@grappa:/new/src/netbsd-1-6/src/sys/arch/macppc/compile/URIEL macppc
uriel:~% ls -l `which X`                                                    [2]
lrwxr-xr-x  1 root  wheel  7 Jan 15  2002 /usr/X11R6/bin/X@ -> Xmacppc

I'd be interested to see what your /usr/X11R6/bin/X is...

> I figured that what this means is that I can have all of X goodness except
> that it will really suck, be slow and so on. Is this correct?

Well, it means that you can have graphical windows and run a web
browser. I wouldn't go expecting any kind of 3D acceleration.

> Or will have something that is not quite X? I won't waste too much time on
> this unless I can pop up windows with tk code, which is the reason I want
> this box to run X, also, to run a simple web browser, xpdf, and maybe
> gnumeric.

You'll be able to do all of those things just fine. I X forward
things that complicated from my macppc machines all the time, but
seldom use their console for X.

Also, note that, depending on the system, you may be locked into
640x480 mode by OpenFirmward refusing to initialize the video
chipset to anything higher on boot. You might or might not be able
to fix this with some OF twiddling (which you could then define as
part of a Forth script that ended in calling the regular "boot",
store the script in the nvramrc, and call the script from the OF
environment variable boot-command).

> This is still the case and will probably remain so unless I grow another
> brain and code it myself.

Some later PowerMacs shipped with something closer to an industry
standard chip. My understanding, which may have been mistaken, was
that XF86 4 just didn't know to be looking for the chip on the other
side of the kind of bridge where it showed up in a Mac, as opposed
to a (IA32-)standard PCI bridge. If that's the case, generalizing
the XF86 code to deal with the situation (especially under NetBSD)
is ridiculously easy and the X folks are remiss for not doing so.

Oh, well, maybe. XF86 does its own probing of the PCI bus, for
better or for worse (this is frequently a good thing, see below), so
that stuff would perhaps have to be added on to it in a moderately
painful way...

> I might try the card trick out, but I heard that this is a lost cause
> unless one has of3.0, and I am at 1.x.

You're at 1.0.5, I think. I don't know if that's still the current
state of affairs, since I've a friend using a couple of PCI Ethernet
cards without OF on-board on an OF 1.0.5 PowerMac, which I also
thought was impossible.

The graphics card problem may be more complex, though, as if OF
doesn't recognize the card, you'll never get a console framebuffer
on it. Maybe that doesn't matter and you can still get X to operate
on it (and just use a serial console rather than a framebuffer).
Certainly, this is effectively what one does for second (or beyond)
monitors for use with Xinerama (the console, that is things like
wscons, only know about the first, and making a text console appear
on others is quite difficult, but having XF86 recognize all of them
is relatively easy, since it does its own probing of the PCI bus).

gabriel rosenkoetter

Attachment: pgpvn0hgM5EXN.pgp
Description: PGP signature