Re: X server through the xsrc sets on PowerBook G4 with NVidia card

[Michael <macallan%netbsd.org@localhost>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On Mar 31, 2010, at 1:39 AM, Taylor R Campbell wrote:

>   Date: Wed, 31 Mar 2010 00:03:29 -0400
>   From: Michael <macallan%netbsd.org@localhost>
>
>   That looks like the vgahw module trying to map the legacy VGA
>   framebuffer - quite useless on macppc even if it was enabled.
>   There is a hack in the kernel to allow mmap()ing the first 128kB of
>   video memory at 0xa0000 - doesn't do any good but it shuts up that
>   particular error, enable it with
>   options GENFB_FAKE_VGA_FB
>
> Thanks.  It's spelled `WSFB_FAKE_VGA_FB',

Yes, I'm fluent in typo and brainfart ;)

> and enabling that got me further.
> I wonder why this wasn't a problem under NetBSD 4?  Was that
> option formerly enabled, or did the driver formerly not try to do
> anything with vgahw?

I'm not sure, the option was probably enabled since many drivers  
needed it, probably got overlooked with the transition to genfb since  
I fixed a few Xorg drivers to skip vgahw.

> Next, the X server complained about the keyboard, and instructed me to
> add `Options "Protocol" "wskbd"' and `Options "Device" "/dev/wskbd"'
> to the input device section of my xorg.conf, after which the X server
> launched, but seems to think I have a different keyboard model or
> layout from my actual one.

X -configure should have these options included by default. Did you  
generate the config file with an older Xserver?
See if using /dev/wskbd0 makes any difference.

>  Here is the stanza I have tried putting in
> my xorg.conf, the rest of which was generated by `X -configure':
>
> Section "InputDevice"
>         Identifier  "Keyboard0"
>         Driver      "kbd"
>         Option      "Protocol" "wskbd"
>         Option      "Device" "/dev/wskbd"
>         Option      "XkbRules" "xorg"
>         Option      "XkbModel" "macintosh"
>         Option      "XkbLayout" "us"
>         Option      "XkbOptions" "ctrl:nocaps"
> EndSection
>
> I have also tried omitting the last three lines (with no effect), and
> replacing `macintosh' by `macintosh_old' (which caused openbox to
> abort!) and `powerbook' (which seemed to have no effect).  X seems to
> see consistent keyboard input which is just wrong; for example, what
> should be read as `qwertyuiop[]' is instead read as `ijklnm352641'.
> Any ideas?

You shouldn't need any of the xkb options.

Wait a second. Does your powerbook use ADB or USB to talk to the  
keyboard? If it's ADB - are there any other wskbd devices? If you have  
more than one wskbd ( for example, a built-in ADB keyboard and an  
external USB numberpad ) then X will freak out. Mostly because the mux  
device ( /dev/wskbd, as opposed to /dev/wskbd0, 1 etc ) when put into  
raw mode will forward native scancodes for each device and return  
garbage when asked what scancodes it uses, and the actual events don't  
have any indication what scancode set they use either. Yes, that's a  
bug. The workaround is to use /dev/wskbd0 and if necessary have  
another section for each additional wskbd instance.

> Can the vgahw module in the X server just be replaced by a bunch of
> noops on macppc?  That would be simpler than going through every
> driver that uses vgahw stuff and #ifdef'ing it out.

Probably.

have fun
Michael

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)

iQEVAwUBS7LlLspnzkX8Yg2nAQKS8Af/QB+YbmO4rFu4SczXxNi+OxvPEl3KXUY3
bhz59+Qfc8vuz3QKuPt0iyDDqICvwt22JjyEtk3g6lyd8wHozpNIZeAqvsEVeXiT
rDyxpCZJlm8BEvkqSOvh04Ok5mH9PxAJrh7e4UI+KIPxSLCtbkXIa7ikyZvc3vfK
55i8Dr97Mlg9pi38Y0M1hlOgz7UcGhIovkbnCSbbOAtG1vL/Pr8oCcr2LJRalsTO
qRHFeZB97Qhs7r8IUwKHnKvKZmqGT3kAFauORNMi0XuFqj8DfhgmbymjYRNKhl/D
e7ninz4MzfInB8NFwALCYK3YgMMDN8hFm0LZ7fETqKJM2QwiOZsWog==
=VgjK
-----END PGP SIGNATURE-----