Subject: Re: XF401 now working on 6500! [somewhat long]
From: Stefan Jeglinski (jeglin@4pi.com)
Date: Wed Dec 20 2000 - 11:57:50 MST
> > title:
>> How-to, XFree86 4.0.1 on an ADB Mac (the 6500) [including input layer]
>
>Um. It looks like you're using the new mouse bit but not the keycodes. :)
yes, see below.
> > Note: the link mouse->adbmouse can be changed but it doesn't matter
>> (see further below). The link can be changed with mouseconfig BUT
>> mouseconfig will mess with your XF86Config file in ways that you
>> don't want for XF4. I'm a little unclear on this point, but basically
>> forget about /dev/mouse, and make sure /dev/input/mice exists.
>
>The mouseconfig in Halloween should set everything up properly. Also,
>mouseconfig/kbdconfig only modify /etc/X11/XF86Config, _not_ XF86Config-4.
I'm not using Halloween. I have a LinuxPPC 1999 system that has
probably been upgraded 90% (manually bit by bit as needed) to YDL
1.2.1. So I've had to work on the little tricks that arise as a
result. Question: Is XF401 even looking at XF86Config when it starts?
How then would mouseconfig even be relevant any more? It's true that
I wasn't too clear on this point when I wrote my post. I'm also sure
I don't quite get this, please clarify?
> > 4. Get and update to Xautoconfig-0.12-1 by Tom Rini, at
>> <ftp://penguinppc.org/users/trini/Xautoconfig-0.12/>. Run
>> Xautoconfig4, which will save the file /etc/X11/Xautoconfig-4. In
>> this saved file, make the following changes:
>>
>> a. comment out 'Load "v4l"' in the Modules section.
> > b. change "macintosh" to "macintosh_old" in the keyboard Input Device
>> section.
>
>Fixed.
in 0.13? Is v4l something that LinuxPPC users generally won't
want/need anyway? And is "macintosh_old" really _generally_ correct
in light of this whole keyboard mapping question? Or just for old ADB
macs?
> > c. change "BusMouse" to "IMPS/2" in the Protocol section
>> of the mouse InputDevice section.
> > d. change "/dev/mouse" to "/dev/input/mice" in the Device section
>> of the mouse InputDevice section.
>
>Hmm. I know why this happened. Did X actually work with BusMouse and
>/dev/mouse (pointing at adbmouse).
X started but the ADB mouse was dead until I changed it to IMPS/2
-and- /dev/input/mice. Again another point of confusion for me: it
appears that I have to configure everything as if I had a USB
keyboard and mouse, even though there is no USB. This just to get the
input layer for ADB compiled. Is that really correct? I've spent more
than a few hours messing with this, and it seems to be the only way
it will work. Fine by me, because adding a USB board and real USB
mouse should just trivially work at this point. On my PowerTower, the
one with the OrangeMicro and Logitech mouse and the Adaptec card with
the shared interrupt that messes with the USB mouse, this same setup
allows me to just use the ADB mouse instead while the USB mouse is
plugged in and waiting to be used. I can tell the USB mouse works
fine, but only when there is disk activity. Clearing up that
interrupt issue will make it work perfect.
> Basically, Xautoconfig said you aren't
>using the new input layer (which at least imho) is the new keycodes, not the
>mouse bit.
>
snip
> > 5. Wait, you're not done. In /etc/sysctl.conf, make sure you have
>> this line present:
>> # Enables new keycodes on PPC
> > dev.mac_hid.keyboard_sends_linux_keycodes = 0
>
>...This is turning _off_ the new keycodes. You should be keeping this at 1
>and then re-running kbdconfig/mouseconfig from Halloween. You should then
>have a working keymap/etc again (and Xautoconfig{3,4} should find and setup
>your keyboard properly).
I can't do this. If I boot with keyboard_sends_linux_keycodes = 1, my
console keyboard is messed up. Maybe after a couple of hours of
trying, I could actually run mouseconfig. Sounds like it should all
work for those who install a pristine Halloween, but who does that?
:-) What's the best plan for those of us that manually upgrade. And
are you saying that people with YDL will have a harder time of it? Is
there an order these three routines (kbdconfig, mouseconfig,
Xautoconfig4) should be run?
And while I've got your attention, here's another question: what does
the compile option ADB_RAW_KEYCODES (from memory, in the Mac Device
Driver section of .config) have to do with the setting for
keyboard_sends_linux_keycodes, if anything? Do these need to be
paired?
And to wrap it up, a dumb question: I never even knew there was a
kbdconfig routine, I've never run it in my history of Linux. It this
a stronger requirement now, in light of the keymap issues that people
are having?
Thanks for the help.
Stefan Jeglinski
This archive was generated by hypermail 2a24 : Wed Dec 20 2000 - 11:58:22 MST