2 ways to fix Intel 430TX reboot problem

John Shifflett (jshiffle@netcom.com)
Sun, 5 Oct 1997 11:33:38 -0700 (PDT)


> I have a brand new Triton II TX motherboard and linux doesn't want to
> boot on it, it starts booting, recognizes the PCI chipsets, calculates
> the delay loops, starts the kernel threads (kswapd and bdflus), checks
> the parallel, checks the serials and reboots.

I had this trouble with my new Iwill P55UXW motherboard, which
uses the same chipset, I assume (430TX). On my system, it
reboots just like yours _if_ there is no PS2 mouse plugged
into the mouse port!!!! If I turn the computer off, plug in
a PS2 mouse, and turn it back on, Linux has no trouble! Not
sure why this happens, or whether the bug is in Linux or the
motherboard BIOS, because I have read messages claiming that
programming the flash BIOS with an _older_ version will also
cure the reboot.

I have also found another way to fix the problem, via an
article I got from dejanews. Someone discovered that writing
to the keyboard LED's is what actually triggers the reboot!!
If you have access to a working Linux machine, you can compile
a kernel that avoids touching the LEDs. In 2.1.57, comment
out line 868 in 'drivers/char/keyboard.c':

kbd_leds(leds);

In 2.0.31-pre9, comment out lines 1184 and 1185 in the same
file:

if (!send_data(0xed) || !send_data(leds))
send_data(0xf4); /* re-enable kbd if any errors */

For other kernel versions you may have to do some searching
for the place that writes to the LED's. Again, I am unclear
as to _why_ this small change cures the reboot - obviously
there is something new in the keyboard controller (firmware?
hardware?) involved with the motherboard reset circuitry. It
seems in this case that the old "standard" way of writing to
the keyboard is no longer the "correct" way. Rebooting when
there's no PS2 mouse - I haven't a clue what's going on there...

Anyway, good luck - hope this helps.

John Shifflett jshiffle@netcom.com