Ack, I gotta get in on this too...
On Sun, Apr 01, 2001 at 07:03:40PM -0400, AMD-760MP wrote:
> > To further perpetuate this insanely off-topic thread, using a USB mouse
> gets finer control. I'm not sure about lag, but there's certainly a lot
> more precision. As a matter of fact, when I switched from my PS/2 trackball
> to a USB trackball using the PS/2 *adapter*, the precision went up. It's
> the same model (Logitech Trackman, PS/2 was 3-button, USB was with wheel),
> but they respond very differently. One day soon I'll get the balls to
> actually set up USB support...
You're imagining things here with the USB -> PS/2 adapter. The USB models
were remade slightly and as a result thei handle a little better. They
don't talk USB with that adapter, they talk native PS/2.
> Further abject apologies for perpetuating this, but USB is getting pretty
> easy to do, with recent 2.4.X kernels. GPM is still hit and miss for me, but
> X is very easy, and gives an even bigger boost. The finer control evidenced
> with a USB pointer and a PS/2 adapter is due to the device still measuring
> and moving at a high rate internally, and then downsampling to PS/2 rates.
> I'm not sure, but USB on PS/2 adapters may increase the default PS/2 polling
> rate to adapt... Basically, from lowest performance to highest, it currently
> goes serial-PS/2-USBw/adapter-USBnative
What a coincidence, my USB testing setup kept simply because it's easier
than changing things happens to be exactly what you need to see how to set
up gpm for repeating and USB, and X for the same! Yes, thank you, thank
you, I'm good and all that nonsense. =)
--8<-- gpm.conf
device=/dev/input/mice
responsiveness=
repeat_type=ms3
type=ps2
append=""
--8<--
For a non-Debian system, let's put that in perspective:
gpm -- -m $device -t $type -R$repeat_type
Debian obviously does a parameter test/append, but no need to overly
complicate things. =) USB offers multiple opens, so you can remove that
repeat_type parameter and everything will just work without it, regardless
of gpm interfering with the device. X is of course very simple:
Section "InputDevice"
Identifier "Configured Mouse"
Driver "mouse"
Option "Device" "/dev/gpmdata"
Option "Protocol" "IntelliMouse"
Option "SendCoreEvents" "true"
Option "ZAxisMapping" "4 5"
EndSection
Note you MUST use ms2/IntelliMouse for this or it will not work with gpm.
However, just point it at /dev/input/mice and it will work with USB for
you. Protocols IMPS/2, IntelliMouse, and PS/2 are supported. I suppose
MouseSystems could have been done but nobody would care if it had been
since the kernel's version of IMPS/2 is very sane and usable.
> I have a Logitech Trackman Marble+ Wheel working perfectly since 2.4.0-test2
> or so. I'd be happy to try and help anyone who wants to get USB pointers
> working. I'll be quiet now and let the OT topic die.
Kernel-based /dev/input for USB (for now - the rest will happen in 2.5):
#
# Input core support
#
CONFIG_INPUT=y
CONFIG_INPUT_KEYBDEV=y
CONFIG_INPUT_MOUSEDEV=y
CONFIG_INPUT_JOYDEV=y
CONFIG_INPUT_EVDEV=y
#
# USB support
#
CONFIG_USB=y
CONFIG_USB_DEVICEFS=y
# You might have OHCI - see kernel docs
CONFIG_USB_UHCI=y
CONFIG_USB_HID=y
Those devices I have compiled in are necessary for booting with a USB
keyboard, you could change them and use modules for a USB mouse only, but
as I said I won't go back to legacy devices personally.
If that isn't EVERYTHING that can be said on the subject of USB input with
Linux and X and gpm short of a heaping pile of total newbie documentation
including info on how to compile a kernel...
And so, having RE-RE-disposed of the topic, exit our hero through the
front door, stage left.
--
Joseph Carter <[EMAIL PROTECTED]> Free software developer
There is no snooze button on a cat who wants breakfast.
PGP signature
