[9fans] 9atom and USB Keyboard Problem

[Gregory Pavelak]

I installed 9atom downloaded from Erik's site on 5/12 on a
Gigabyte GA-D525TUD
(http://www.gigabyte.com/products/product-page.aspx?pid=3549#ov).

When I have *nomp=1 in plan9.ini, I am able to log in and
things work OK (except the mouse wheel scroll direction is
opposite to what I would expect), When I remove *nomp=1 from
plan9.ini, 9atom reports four processors, and every keystroke
produces 6 or 8 or more tokens of the character I am trying to
type. If I'm lucky and I tap a key lightly and quickly enough, I
can get a single character, but it's not easy.

I tried the only other usb keyboard I have and had the same
problem.

Any suggestions?

Thanks.

Greg

[erik quanstrom]

9atom's usb is identical to sources, with the exception that i've
grabbed cinap's bios handoff code for ehci and enabled all
the controllers. this was necessary to get usb to work on some
new hardware.

i'd recommend doing pull (as glenda) and rebuilding the
kernel, first.

i do know there is a bug that causes spurious kb repeating.
but typically it's not quite as severe, and depends a lot on
the keyboard in question. it's interesting that you corellate
this to mp interrupts. does the same hapen without *nomp
and with *ncpu=1?

i haven't done much research on this, so although the usb/kb
repeat mechanism at first blush looks like a good place to start,
it's not clear to me that this is the issue.

- erik

[erik quanstrom]

can you try changing the default initial repeat from t = 160 to t = 500 (ms)?
you will need to do a mk install in /sys/src/cmd/usb, and then rebuild your kernel.

let me know if this makes the output a bit more reasonable.

- erik

[Gregory Pavelcak]

Thanks Erik. That helped. Now I only get multi-character
output periodically. I don't know what triggers it,
but at least I can type.

Turns out the mouse has the same problem. Button-3 click
in acme sends the text flying by. I don't know how many
clicks are being sent.