Keyb/mouse scheduling issues

Hi,

I'm currently on
FreeBSD 5.3-STABLE #3: Tue Oct 26 13:24:36 CEST 2004 i386
and am still experiencing the same problems as I (and others) had months
and months (a year?) ago. At the time I thought it was a sched_ule
thing, but now I am running with sched_4bsd and am seeing the same:

When there is high disk I/O (very good example:
cd /usr/ports/www/firefox && make extract), I see the following
symptoms:
- Keyboard input gets erratic - keys are skipped, input is slow, keys
get 'stuck' repeating, etc. Typing is very difficult.
- Mouse movements become erratic - mouse pointer jumps around, mouse
buttons are pressed and released randomly (even though I'm not close to
any of them), basically the mouse is impossible to control. Moused is
running.

I am running with the default SMP kernel provided with the latest
RELENG_5_3 checkouts, apart from ACPI being disabled and mptable.c is
patched to force enablement of my second CPU. I have not yet found a way
to force-enable my second CPU when ACPI is enabled; any assistance would
be very welcome. (The BIOS has flagged it as unuseable, but this is
b.s.)

My hardware is a dual Athlon system on a MSI K7D-Master board. Only
using the standard IDE controller (yes, IDE drives) in ATA-100 mode:
atapci0: <AMD 768 UDMA100 controller> port
0xe000-0xe00f,0x376,0x170-0x177,0x3f6
,0x1f0-0x1f7 at device 7.1 on pci0

[root_at_eirik] ~# atacontrol list  
ATA channel 0:
    Master:  ad0 <WDC WD400BB-32CLB0/95.04E95> ATA/ATAPI revision 5
    Slave:   ad1 <WDC WD1200JB-75CRA0/16.06V16> ATA/ATAPI revision 5
ATA channel 1:
    Master: acd0 <FX4830T/R02C> ATA/ATAPI revision 0
    Slave:       no device present

[root_at_eirik] ~# atacontrol mode 0
Master = UDMA100 
Slave  = UDMA100
[root_at_eirik] ~# atacontrol mode 1
Master = UDMA33 
Slave  = BIOSPIO

Any takers?
/Eirik