21 Jan
2014
21 Jan
'14
8:04 p.m.
I've got a PDP-8A with RK05's and a TC08/TU56.
I've verified the serial port is okay (via checkser), loaded focal with the
rim loader and run both read and write tests on a RK05 pack.
I've done a restrk05 with both
diagpack2.rk05<http://www.pdp8online.com/pdp8cgi/os8_html?act=dir;fn=ima…
and
haygood-osv3r.rk05<http://www.pdp8online.com/pdp8cgi/os8_html?act=dir;fn…me>.
It seems to progress ok, counting up to the last cylinder and leaving the
AC equal to 0 when done. However when I execute the two word bootstrap for
the RK8E I hear the drive seek once and then stop at location 0031.
I've also tried a restore with the TC08 but it never even gets to the point
that the drive gets selected (the SR is 0 on entry). However if I punch in
the TC08 bootstrap unit 0 does a rewind, then goes forward (with lots of
das blinken lights), goes past quite a few blocks and then halts (this is
on a tape that claims to have OS8 on it).
The RK05 is most frustrating since it passes the rw tests (9 passes) but
doesn't boot.
I guess I'm going to be loading in some maindecs unless someone out there
has any other ideas.
Thanks,
Marc
22 Jan
22 Jan
8:26 p.m.
On Tue, Jan 21, 2014 at 05:04:10PM -0800, Marc Howard wrote:
However when I execute the two word bootstrap for
the RK8E I hear the drive seek once and then stop at location 0031.
Halting at 0031 looks like the code didn't read properly into memory.
Try this with diagpack2.rk05
1000 6743
1001 5201
Start at 1000. The disk should read and the processor stay looping at 1001.
Halt the processor and see what is in memory.
You should have at
30 6741
31 5030
32 0035
If those look ok start the computer running at 31. OS/8 should start.
Are you hitting the init button to clear the registers before trying to
start the bootstrap? The bootstrap just tells the controller to go. It depends
on the registers being set to the default values.
I've also tried a restore with the TC08 but it
never even gets to the point
that the drive gets selected (the SR is 0 on entry).
It has never been tested on real hardware, only with emulator. If your
willing I can work with you to get it running. You did start the PC
program before starting the PDP-8 running? Its the opposite of how you do
restrk05.
10:23 p.m.
On 1/22/2014 6:26 PM, David Gesswein wrote:
On Tue, Jan 21, 2014 at 05:04:10PM -0800, Marc Howard
wrote:
The RK8E bootstrap is a little strange.
The bootstrap is located in low memory and typically ends with a
jump-to-self instruction (5031) at address 0031.
The strange thing is that the bootstrap program stores the contents of
disk sector 0 beginning at address 0 - which is right on top of the
bootstrap program. Eventually the disk read operation changes the
instruction at address 0031 to something else while it is in this
jump-to-self loop. Not quite self-modifying - but -
I had some difficulties emulating this behavior in my PDP8 FPGA! The
disk controller would overwrite the instruction at 0031 and the CPU
would execute the next instruction at 0032 before the disk had stored
/that/ instruction.
The bootstrap program /assumes/ that the disk is faster than the CPU.
When I sped up my PDP8, I exposed this race condition.
If the code is still executing the instruction at address 0031 after the
disk read operation, the disk wasn't read...
Rob.
However when I execute the two word bootstrap
for
the RK8E I hear the drive seek once and then stop at location 0031.
Halting at 0031 looks like the code didn't read properly into memory.
Try this with diagpack2.rk05
1000 6743
1001 5201
Start at 1000. The disk should read and the processor stay looping at 1001.
Halt the processor and see what is in memory.
You should have at
30 6741
31 5030
32 0035
If those look ok start the computer running at 31. OS/8 should start.
Are you hitting the init button to clear the registers before trying to
start the bootstrap? The bootstrap just tells the controller to go. It depends
on the registers being set to the default values.
I've also tried a restore with the TC08 but
it never even gets to the point
that the drive gets selected (the SR is 0 on entry).
It has never been tested on real hardware, only with emulator. If your
willing I can work with you to get it running. You did start the PC
program before starting the PDP-8 running? Its the opposite of how you do
restrk05.
10:44 p.m.
On 22/01/14 10:23 PM, Rob Doyle wrote:
...
The RK8E bootstrap is a little strange....
The bootstrap program /assumes/ that the disk is faster than the CPU.
When I sped up my PDP8, I exposed this race condition.
If the code is still executing the instruction at address 0031 after the
disk read operation, the disk wasn't read...
Thanks for sharing that little bit of knowledge. Awesome.
--Toby
Rob.
23 Jan
23 Jan
11:45 a.m.
One step forward, two steps back...
I got a few minutes to test last night and now I can't even boot the RK05
from 5000, let alone 0030.
No data is being xfered. Something flaky here. What a shock that 40 year
old hardware has intermittent problems.
I think I'm going to pull the TC08 interface cards (M8350/60) just to rule
out interference. Then I'll crank up the logic analyzer. Luckily I have
an extender card that has 68 of the more interesting omnibus signals
hard-wired to analyzer pods. Debugging the disk interface itself would be
a nightmare due to it's multi-board stackup.
Marc
On Wed, Jan 22, 2014 at 7:44 PM, Toby Thain <toby at telegraphics.com.au>wrote:
On 22/01/14 10:23 PM, Rob Doyle wrote:
...
The RK8E bootstrap is a little strange....
The bootstrap program /assumes/ that the disk is faster than the CPU.
When I sped up my PDP8, I exposed this race condition.
If the code is still executing the instruction at address 0031 after the
disk read operation, the disk wasn't read...
Thanks for sharing that little bit of knowledge. Awesome.
--Toby
Rob.
24 Jan
24 Jan
4:23 p.m.
Well last night I fired it back up and now everything works! Even boots
from 0030.
So the next question concerns running DECX8 off the diag pack. I do a ".R
DECX8" and it halts @ 0x200, fine. I enter 0040 to the switch register and
hit run. I don't get a message as the DECX8 document promises. It halts
again @210 and I enter 0030, run, still no output and it halts @ 2001 (or
was it 2100?).
Anything I'm doing wrong here?
Thanks,
Marc
On Thu, Jan 23, 2014 at 8:45 AM, Marc Howard <cramcram at gmail.com> wrote:
One step forward, two steps back...
I got a few minutes to test last night and now I can't even boot the RK05
from 5000, let alone 0030.
No data is being xfered. Something flaky here. What a shock that 40 year
old hardware has intermittent problems.
I think I'm going to pull the TC08 interface cards (M8350/60) just to rule
out interference. Then I'll crank up the logic analyzer. Luckily I have
an extender card that has 68 of the more interesting omnibus signals
hard-wired to analyzer pods. Debugging the disk interface itself would be
a nightmare due to it's multi-board stackup.
Marc
On Wed, Jan 22, 2014 at 7:44 PM, Toby Thain <toby at telegraphics.com.au>wrote:
On 22/01/14 10:23 PM, Rob Doyle wrote:
...
The RK8E bootstrap is a little strange....
The bootstrap program /assumes/ that the disk is faster than the CPU.
When I sped up my PDP8, I exposed this race condition.
If the code is still executing the instruction at address 0031 after the
disk read operation, the disk wasn't read...
Thanks for sharing that little bit of knowledge. Awesome.
--Toby
Rob.
3837
days inactive
3839
days old
5 comments
4 participants
participants (4)
-
cramcram＠gmail.com -
djg＠pdp8online.com -
radioengr＠gmail.com -
toby＠telegraphics.com.au