We get similar problems to the ones you describe a lot on our PCs, but
not quite as bad, I don't think. Our X servers will "freeze up"--i.e.,
not accept any keyboard input, from time to time. Just keeping the space
bar down for a while seems to unstick them.. also switching to a different
virtual console, then switching back seems to do the trick. I have only
seen this with pcvt, and only running X.

P120--Intel MB, NetBSD 1.1B (it was the same on 1.1A), ATI Mach64,
3C595. But this problem still isn't nearly as annoying as the swapping
business :)

  Mika

Jonathan Stone writes:
>
>I've set up a nameless P120 running NetBSD 1.2-Beta.  I've recently
>stopped using the pcvt interface and started using X.  I find that
>after a few minutes use, all keyboard events stop.  I've run xev on
>another machine and verified that mouse events are being tracked, but
>key up/down events seem to totally disappear.
>
>I've tried switching keyboards to no avail.  Identical hardware works
>flawlessly with Linux 1.3.x, and Win95.  I've tried Xfree86 3.1.2
>and 3.1.2E, with and without COMPAT_10 (which is used in
>i386/isa/pccons.c).
>
>I've also occasionally seen the sw's idea of CAPSLOCK get out of
>stepwith the LEDs on the keyboard: capslock is in effect when
>the LED is off, and vice-versa.
>
>
>Killing and restarting the X server (or letting xdm restart it)
>causes the keyboard work again.   I haven't seen any such problems
>in glass-tty mode.
>
>
>Some things that may be relevant: I type really. _really__ fast.
>Typing more slowly seems to defer the symptoms (but not e.g., long
>enough to type as far as youve' read.)
>
>
>I'm using a Logitech serial  mouse.  If it matters, the video
>hardware is a PCI Diamond 64 Stealth with the 220 MHz TI RAMDAC,
>and the XF86_S3 server seems to identify the hardware OK.
>(The XF86 S3 server works just fine with Linux on identical hardware.)
>
>I can't easily try older kernels, as the NetBSD 1.1 doesn't support
>the peripherals on tehse machines  (they came with a 3c595 and a
>newish BusLogic PCI scsi controller.)
>
>Needless to say, this makes the console of this machine unusable in X.
>Has anyone else seen this or anything similar?  If so, please Cc:
>me, as I don't seem to be on the port-i386 anymore.
>
>thanks
>--Jonathan Stone