12 May
1998
12 May
'98
4:48 a.m.
Tony Duell wrote:
The first requirement is that you're mad.
Now that you're foaming blue at the mouth, you could give the speech
about how to use the keyboard interface chip to bring the 286 into
real mode and back again... :-)
- John
Jefferson Computer Museum <http://www.threedee.com/jcm>
12 May
12 May
1:33 p.m.
A non-text attachment was scrubbed...
Name: not available
Type: text
Size: 1073 bytes
Desc: not available
Url :
http://www.classiccmp.org/pipermail/cctalk/attachments/19980512/6e41eba1/at…
12:56 p.m.
It was thus said that the Great Tony Duell once stated:
Well, there is at least one way through software, maybe two. Both are
undocumented, but work.
The first is to use the LOADALL instruction (different format between the
286 and 386. I think Intel finally removed that instruction on the
Pentiums). The second way is to triple fault the CPU (I know that works on
the 386 or higher, not exactly sure about the 286). Both are faster than
the hardware reset method.
-spc (Now getting a triple fault is a bit tricky ... )
Tony Duell wrote:
Sure, it's not _that_ hard...
The problem is that the i80286 can go from real mode to protected mode,
but not back again. At least not without a hardware reset. The first requirement is that you're mad.
Now that you're foaming blue at the mouth, you could give the speech
about how to use the keyboard interface chip to bring the 286 into
real mode and back again... :-) 
3:02 p.m.
[why does this thread exist?]
On Tue, 12 May 1998, Tony Duell wrote:
The problem is that the i80286 can go from real mode
to protected mode,
but not back again. At least not without a hardware reset.
Now, the keyboard controller output port bit 0 will, if pulsed low, reset
the microprocessor.
Wasn't there another trick? Tripple-fault will also cause a hardware
reset on most mobo's, and it's probably faster than diddling the keyboard
controller.
-- Doug
3:14 p.m.
A non-text attachment was scrubbed...
Name: not available
Type: text
Size: 1117 bytes
Desc: not available
Url :
http://www.classiccmp.org/pipermail/cctalk/attachments/19980513/9cb365da/at…
3:31 p.m.
On Wed, 13 May 1998, Tony Duell wrote:
If the triple-fault has to be detected by some piece
of hardware outside
the CPU (which would be unlikely) then it can't work on a genuine IBM
(which is what IMHO defines a PC/AT). The only ways to reset the CPU on
that machine are by the power good signal from the PSU or the keyboard
controller.
Tripple-fault is documented to put the CPU into the shutdown state. I
know many mobo's tied this to reset (which is why your machine would
mysteriously reboot when running buggy version of Windows, OS/2, etc).
But I don't know if *all* mobo's or the original AT did this.
-- Doug
4:05 p.m.
A non-text attachment was scrubbed...
Name: not available
Type: text
Size: 826 bytes
Desc: not available
Url :
http://www.classiccmp.org/pipermail/cctalk/attachments/19980513/2f5fcc3a/at…
5:11 p.m.
On Wed, 13 May 1998, Tony Duell wrote:
However, IMHO it's a lot harder to get a CPU to
tripple-fault than to
simply send a command to the keyboard controller. And you'd have to set
all the other things up no matter how you reset the CPU.
Harder? Microsoft did it all the time without even trying :-)
It only takes a few instructions, and doesn't require any slow I/O, so
it's a handy technique if you ever have to write a new mode switching O/S
for the 286.
-- Doug
13 May
13 May
9:44 a.m.
A non-text attachment was scrubbed...
Name: not available
Type: text
Size: 875 bytes
Desc: not available
Url :
http://www.classiccmp.org/pipermail/cctalk/attachments/19980513/9c7f0986/at…
12 May
12 May
4:13 p.m.
Tony Duell wrote:
If this is an undocumented feature of the CPU then it
should work. Alas,
being undocumented there's no way I can look it up in the data book.
What a special joy, though. I have several pounds (avoirdupois) of
material on features in several generations of Motorola processors.
Most of which I'm incompetent to use, and a few are documented to
invoke the legendary HCF instruction.
IBM certainly used the method I discribed in their BIOS. I've checked the
source listings.
--
Ward Griffiths
They say that politics makes strange bedfellows.
Of course, the main reason they cuddle up is to screw somebody else.
Michael Flynn, _Rogue Star_
-- Doug
-tony
4:08 p.m.
Tony Duell wrote:
Yes, Tony, obviously "The first requirement is that you're mad."
(We're
not talking _angry_ here, kids! "Mad" in English isn't the same thing
as in American.)
--
Ward Griffiths
They say that politics makes strange bedfellows.
Of course, the main reason they cuddle up is to screw somebody else.
Michael Flynn, _Rogue Star_
Tony Duell wrote:
Sure, it's not _that_ hard...
The problem is that the i80286 can go from real mode to protected mode,
but not back again. At least not without a hardware reset.
Now, the keyboard controller output port bit 0 will, if pulsed low, reset
the microprocessor. There's a keyboard controller command for that -
0xF1. You send that to output port 0x64. You also need to set the system
flag (send 0x60, 0x04 to port 0x64 (or maybe 0x60,0x05)), set the CMOS
shutdown bye (location 0x0F in the RTC) to 0x0A (to mean a user defined
reset), and set IO_ROM_INIT (far pointer at 0x0040:0067 (offset) and
0x0040:0069 (segment) to point to your code. Or at least that's what I
think you do - it's late and I'm readign the BIOS code as I type... The first requirement is that you're mad.
Now that you're foaming blue at the mouth, you could give the speech
about how to use the keyboard interface chip to bring the 286 into
real mode and back again... :-)
4:42 p.m.
A non-text attachment was scrubbed...
Name: not available
Type: text
Size: 379 bytes
Desc: not available
Url :
http://www.classiccmp.org/pipermail/cctalk/attachments/19980513/8371b9a6/at…
9568
days inactive
9569
days old
11 comments
5 participants
participants (5)
-
ard＠p850ug1.demon.co.uk -
gram＠cnct.com -
jfoust＠threedee.com -
spc＠armigeron.com -
yowza＠yowza.com