13 Apr
2014
13 Apr
'14
3:38 a.m.
I have at last gotten Unix to boot to multi-user mode in FreeBee (Philip
Pemberton's 3B1/UNIX PC emulator). The fixes required are for bus error
handling (instructions were still sometimes changing memory and
registers after a page fault even though I had already partially fixed
it before; I now have it saving the registers before every instruction
and pushing those on page fault similar to what current MESS does - this
isn't actually what a real 68k does and may sometimes throw off register
dumps but it does work; I also have it using longjmp to immediately stop
any instruction that page faults), as well as zero page access (another
thing that I had tried to fix before; I had previously only allowed
reads from address 0, but some programs are expecting to read from
higher addresses in the zero page, so I now allow reads from all zero
page addresses). Another bug that I fixed was related to timing. The
emulator was way slower than real-time on the non-Linux OSes that I
tried (many of them impose a minimum on the amount of time a process can
sleep, and the emulator was depending on being able to sleep for very
short times, so I changed it to use longer sleeps). Unix would
presumably boot multi-user without this fix though.
However, the emulator is still not quite usable. The key mapping is not
quite right (one problem is that the UNIX PC has separate "return" and
"enter" keys, and there is no mapping for "enter", which is required
to
accept dialogs, and there might be other important keys that are
missing). Also, neither the mouse nor the RTC are emulated yet.
Here's a screenshot:
http://imgur.com/GzWkUbs
I'll post my patches very soon.
14 Apr
14 Apr
2:12 a.m.
Andrew Warkentin <andreww591 at gmail.com> writes:
I have at last gotten Unix to boot to multi-user mode
in FreeBee (Philip
Pemberton's 3B1/UNIX PC emulator).
Magnificent! This is excellent news. Since I've been diving deep into
3B1 stuff recently, I've been hoping that someone was working on the
emulator. I'm glad to know it's active.
Actually, since we're on the subject, I hope you won't mind if I pick
your brain about something 3B1 related...
This weekend I started working on a small daughterboard to replace the
socketed WD1010 and act as a hard drive emulator, somewhat similar to
the FreHD TRS-80 Hard Drive Emulator[1], but piggybacking on the
motherboard instead of sitting in a bus expansion slot.
Tonight, I realized that part of my understanding was wrong. I believed
that Drive Number and Head Number were selected by writing to the SDH
task file register of the WD1010 -- in fact, they CAN be, but I fear
that the 3B1 doesn't do it that way. The hardware reference manual
discusses the Disk Control Register (a 74LS373 seen in the schematics on
page 11/D4), which is directly addressed through location 4E0000, rather
than through the SDH task file register at E0000C. It handles generating
the ST-506 Disk Select and Head Select signals directly.
In fact, the Logic Board Test Procedures in the reference manual make it
clear that the SDH register is loaded from the disk's ID field after
performing a reset and a SEEK command. None of the procedures show
writing to the SDH register, only reading from it.
So, if my understanding is correct, simply replacing the WD1010 is not
enough to properly emulate a disk system, unless I'm pretending to be a
single disk with 1 head. Aiee!
IS my understanding correct? Do you (or anyone else) have any thoughts
on this?
(Copied to the list, in case anyone else is interested)
-Seth
[1] http://www.vecoven.com/trs80/trs80.html
3:03 a.m.
On Mon, Apr 14, 2014 at 12:12 AM, Seth Morabito <lists at loomcom.com> wrote:
So, if my understanding is correct, simply replacing
the WD1010 is not
enough to properly emulate a disk system, unless I'm pretending to be a
single disk with 1 head. Aiee!
IS my understanding correct? Do you (or anyone else) have any thoughts
on this?
Yes, that is correct.
7:59 a.m.
New subject: UNIX PC Hard Disk Emulator
Subject change to reflect my hijacking the thread (Sorry Andrew!)
...
Eric Smith <spacewar at gmail.com> writes:
On Mon, Apr 14, 2014 at 12:12 AM, Seth Morabito
<lists at loomcom.com> wrote:
Drat. So I have three options:
1. Enforce single-disk, single-head emulation. Not so bad, really. Max
disk size would be (for WD2010 behavior) 2048 cylinders by 1 head by 256
sectors/track by 512 bytes/sector, so 256 MB. That's pretty darn
respectable.
2. Piggyback the 74LS373 on the card, or vice-versa, and jumper to its
socket. Ugly. More modification than I wanted to do.
3. Give up and do a proper MFM emulation.
Option 3 is tempting but way beyond what I wanted to do, option 2 feels
ugly and not worth it. Option 1 still wins for now, I think.
-Seth
So, if my understanding is correct, simply
replacing the WD1010 is not
enough to properly emulate a disk system, unless I'm pretending to be a
single disk with 1 head. Aiee!
IS my understanding correct? Do you (or anyone else) have any thoughts
on this?
Yes, that is correct. 
4:23 a.m.
On 13/04/14 08:38, Andrew Warkentin wrote:
However, the emulator is still not quite usable. The
key mapping is not
quite right (one problem is that the UNIX PC has separate "return" and
"enter" keys, and there is no mapping for "enter", which is required
to
accept dialogs, and there might be other important keys that are
missing). Also, neither the mouse nor the RTC are emulated yet.
Andrew, all I have to say is "THANK YOU, THANK YOU, A MILLION TIMES -
THANK YOU!" The bus error logic has been a thorn in my side for ages,
and I just couldn't figure out what was going wrong. I'm really glad
someone managed to figure it out!
Get these patches to me and I'll merge them in ASAP and do my best to
fix the mouse, RTC and keyboard.
As you said, the main problem is that we don't have enough keys on a PC
keyboard to do a 1:1 mapping -- thankfully SDL supports Modifier keys,
so this isn't an insurmountable problem. The big problem is coming up
with a sane mapping :)
Truth be told, I'll probably move the repository over to Bitbucket so
people can fork it and work on it independently, then submit merge
requests. It takes a lot of work out of manually importing patches.
Is this based on the OldMMU branch or my revised MMU? I don't really
care which it is, I'll just deprecate the branch you aren't using so we
don't have a million and one build variants 8^)
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
5:06 a.m.
Great news ! My sincere congratulations for you and your work.
About your enquiry about how find a difference between the ENTER and RETURN
keys... I think that it would be possible to do it checking the keyboard
'scancode'. This is a concept from the early times of the PC but I suppose
that some applications have been using this technique even nowadays. It
must be easy to get some examples of use.
--
Gracias | Regards
Saludos - Greetings - Freundliche Gr??e - Salutations
--
*Sergio Pedraja*
twitter: @sergio_pedraja
skype: Sergio Pedraja
http://plus.google.com/u/0/101292256663392735405
http://www.linkedin.com/in/sergiopedraja
http://www.quora.com/Sergio-Pedraja
http://spedraja.wordpress.com
https://www.xing.com/profile/Sergio_Pedraja <http://spedraja.wordpress.com/>
-----
No crea todo lo que ve, ni crea que est? vi?ndolo todo
2014-04-14 10:23 GMT+02:00 Philip Pemberton <classiccmp at philpem.me.uk>:
On 13/04/14 08:38, Andrew Warkentin wrote:
?
However, the emulator is still not quite usable.
The key mapping is not
quite right (one problem is that the UNIX PC has separate "return" and
"enter" keys, and there is no mapping for "enter", which is required
to
accept dialogs, and there might be other important keys that are
missing). Also, neither the mouse nor the RTC are emulated yet.
Andrew, all I have to say is "THANK YOU, THANK YOU, A MILLION TIMES -
THANK YOU!" The bus error logic has been a thorn in my side for ages,
and I just couldn't figure out what was going wrong. I'm really glad
someone managed to figure it out!
Get these patches to me and I'll merge them in ASAP and do my best to
fix the mouse, RTC and keyboard.
As you said, the main problem is that we don't have enough keys on a PC
keyboard to do a 1:1 mapping -- thankfully SDL supports Modifier keys,
so this isn't an insurmountable problem. The big problem is coming up
with a sane mapping :)
Truth be told, I'll probably move the repository over to Bitbucket so
people can fork it and work on it independently, then submit merge
requests. It takes a lot of work out of manually importing patches.
Is this based on the OldMMU branch or my revised MMU? I don't really
care which it is, I'll just deprecate the branch you aren't using so we
don't have a million and one build variants 8^)
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
5:27 a.m.
Maybe "enter" and "shift+enter" to accomplish those tasks?
That's similar to the convention TN5250/TN3270 clients follow (e.g. F13 is
shift+F1). I think it would be the easiest for an end-user to have the
emulator work similar to other systems/clients out there.
2014-04-14 10:06 GMT+01:00 SPC <spedraja at gmail.com>:
Great news ! My sincere congratulations for you and
your work.
About your enquiry about how find a difference between the ENTER and RETURN
keys... I think that it would be possible to do it checking the keyboard
'scancode'. This is a concept from the early times of the PC but I suppose
that some applications have been using this technique even nowadays. It
must be easy to get some examples of use.
--
Gracias | Regards
Saludos - Greetings - Freundliche Gr??e - Salutations
--
*Sergio Pedraja*
twitter: @sergio_pedraja
skype: Sergio Pedraja
http://plus.google.com/u/0/101292256663392735405
http://www.linkedin.com/in/sergiopedraja
http://www.quora.com/Sergio-Pedraja
http://spedraja.wordpress.com
https://www.xing.com/profile/Sergio_Pedraja <
http://spedraja.wordpress.com/>
-----
No crea todo lo que ve, ni crea que est? vi?ndolo todo
2014-04-14 10:23 GMT+02:00 Philip Pemberton <classiccmp at philpem.me.uk>:
--
|_|0|_| Yvan Janssens |_|_|0| /* There are two ways to write
error-free programs.|0|0|0| Only the third one works. */
On 13/04/14 08:38, Andrew Warkentin wrote:
?
However, the emulator is still not quite usable.
The key mapping is not
quite right (one problem is that the UNIX PC has separate "return" and
"enter" keys, and there is no mapping for "enter", which is required
to
accept dialogs, and there might be other important keys that are
missing). Also, neither the mouse nor the RTC are emulated yet.
Andrew, all I have to say is "THANK YOU, THANK YOU, A MILLION TIMES -
THANK YOU!" The bus error logic has been a thorn in my side for ages,
and I just couldn't figure out what was going wrong. I'm really glad
someone managed to figure it out!
Get these patches to me and I'll merge them in ASAP and do my best to
fix the mouse, RTC and keyboard.
As you said, the main problem is that we don't have enough keys on a PC
keyboard to do a 1:1 mapping -- thankfully SDL supports Modifier keys,
so this isn't an insurmountable problem. The big problem is coming up
with a sane mapping :)
Truth be told, I'll probably move the repository over to Bitbucket so
people can fork it and work on it independently, then submit merge
requests. It takes a lot of work out of manually importing patches.
Is this based on the OldMMU branch or my revised MMU? I don't really
care which it is, I'll just deprecate the branch you aren't using so we
don't have a million and one build variants 8^)
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
5:33 a.m.
On 14/04/14 10:06, SPC wrote:
Great news ! My sincere congratulations for you and
your work.
About your enquiry about how find a difference between the ENTER and RETURN
keys... I think that it would be possible to do it checking the keyboard
'scancode'. This is a concept from the early times of the PC but I suppose
that some applications have been using this technique even nowadays. It
must be easy to get some examples of use.
The PC keyboard only has one combined "enter/return" key; the UNIX PC
has separate ones. The problem is that we have to somehow control a UNIX
PC with a PC keyboard -- when the UNIX PC keyboard has more keys than
the IBM PC.
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
5:55 a.m.
Of course. I was thinking on the Emulator behaviour, mostly under Windows
environment.
Kind regards.
Sergio Pedraja.
2014-04-14 11:33 GMT+02:00 Philip Pemberton <classiccmp at philpem.me.uk>:
On 14/04/14 10:06, SPC wrote:
Great news ! My sincere congratulations for you
and your work.
About your enquiry about how find a difference between the ENTER and
RETURN
keys... I think that it would be possible to do
it checking the keyboard
'scancode'. This is a concept from the early times of the PC but I
suppose
that some applications have been using this
technique even nowadays. It
must be easy to get some examples of use.
The PC keyboard only has one combined "enter/return" key; the UNIX PC
has separate ones. The problem is that we have to somehow control a UNIX
PC with a PC keyboard -- when the UNIX PC keyboard has more keys than
the IBM PC.
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
9:05 a.m.
The enter key on the main body of the keycode and the enter key on the
number pad produce different keycodes (According to showkey on my
Linux box, 28 and 96 respectively). You could probably use that,
although it wouldn't work on a laptop which lacks a seperate number
pad.
Mike
On Mon, Apr 14, 2014 at 5:33 AM, Philip Pemberton
<classiccmp at philpem.me.uk> wrote:
On 14/04/14 10:06, SPC wrote:
Great news ! My sincere congratulations for you
and your work.
About your enquiry about how find a difference between the ENTER and RETURN
keys... I think that it would be possible to do it checking the keyboard
'scancode'. This is a concept from the early times of the PC but I suppose
that some applications have been using this technique even nowadays. It
must be easy to get some examples of use.
The PC keyboard only has one combined "enter/return" key; the UNIX PC
has separate ones. The problem is that we have to somehow control a UNIX
PC with a PC keyboard -- when the UNIX PC keyboard has more keys than
the IBM PC.
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/ 
10:53 a.m.
On 14 April 2014 14:05, Michael Kerpan <mjkerpan at kerpan.com> wrote:
The enter key on the main body of the keycode and the
enter key on the
number pad produce different keycodes (According to showkey on my
Linux box, 28 and 96 respectively). You could probably use that,
although it wouldn't work on a laptop which lacks a seperate number
pad.
Pretty much all PC laptops have an embedded number pad on the main
keyboard, generally accessible with Fn (for individual keys) or
Fn-NumLock (for sustained use), precisely /because/ the numpad keys
are not the same as the main-block ones.
--
Liam Proven * Profile: http://lproven.livejournal.com/profile
Email: lproven at cix.co.uk * GMail/G+/Twitter/Flickr/Facebook: lproven
MSN: lproven at hotmail.com * Skype/AIM/Yahoo/LinkedIn: liamproven
Tel: +44 20-8685-0498 * Cell: +44 7939-087884
3:32 p.m.
On Mon, Apr 14, 2014 at 10:53 AM, Liam Proven <lproven at gmail.com> wrote:
Pretty much all PC laptops have an embedded number pad
on the main
keyboard, generally accessible with Fn (for individual keys) or
Fn-NumLock (for sustained use), precisely /because/ the numpad keys
are not the same as the main-block ones.
True enough. It's not as friendly to use as a real numpad, though.
Mike
5:17 a.m.
On 4/14/14, Seth Morabito <lists at loomcom.com> wrote:
Andrew Warkentin <andreww591 at gmail.com> writes:
Actually, Unix and the bootloader both write the correct values into
the SDH register (however, the ROM might not; it won't boot the hard
disk directly in the current version of the emulator - I'm booting
I have at last gotten Unix to boot to multi-user
mode in FreeBee (Philip
Pemberton's 3B1/UNIX PC emulator).
Magnificent! This is excellent news. Since I've been diving deep into
3B1 stuff recently, I've been hoping that someone was working on the
emulator. I'm glad to know it's active.
Actually, since we're on the subject, I hope you won't mind if I pick
your brain about something 3B1 related...
This weekend I started working on a small daughterboard to replace the
socketed WD1010 and act as a hard drive emulator, somewhat similar to
the FreHD TRS-80 Hard Drive Emulator[1], but piggybacking on the
motherboard instead of sitting in a bus expansion slot.
Tonight, I realized that part of my understanding was wrong. I believed
that Drive Number and Head Number were selected by writing to the SDH
task file register of the WD1010 -- in fact, they CAN be, but I fear
that the 3B1 doesn't do it that way. The hardware reference manual
discusses the Disk Control Register (a 74LS373 seen in the schematics on
page 11/D4), which is directly addressed through location 4E0000, rather
than through the SDH task file register at E0000C. It handles generating
the ST-506 Disk Select and Head Select signals directly.
In fact, the Logic Board Test Procedures in the reference manual make it
clear that the SDH register is loaded from the disk's ID field after
performing a reset and a SEEK command. None of the procedures show
writing to the SDH register, only reading from it.
So, if my understanding is correct, simply replacing the WD1010 is not
enough to properly emulate a disk system, unless I'm pretending to be a
single disk with 1 head. Aiee!
IS my understanding correct? Do you (or anyone else) have any thoughts
on this?
(Copied to the list, in case anyone else is interested)
-Seth
[1] http://www.vecoven.com/trs80/trs80.html
from a floppy with no kernel). The WD2010 emulation
that I contributed
a while back just uses the value written to the SDH register.
On 4/14/14, Philip Pemberton <classiccmp at philpem.me.uk> wrote:
On 13/04/14 08:38, Andrew Warkentin wrote:
You're welcome.
I already implemented the RTC a while ago, but my RTC implementation
was in an older copy of the source that I hadn't looked at for a
while. I forgot that I actually had it in a state where it mostly
worked (reading the date works, but attempts to set it are ignored).
Unix does have some Y2K issues though. I'll include the RTC in my
patches of course.
I did add a mapping of Alt+Return to Enter, and I found out that the
system sometimes still doesn't boot all the way. After I get through
the initial dialogs on boot, the system hangs after the "The screen is
being set up..." message. I just tried it now and instead of a hang I
got some messages relating to the modem, which I dismissed, and got to
a login prompt (I was able to log in and get a desktop), so I think at
least a stub of the serial controller will have to be implemented to
get it to boot reliably (I tried booting again and it just hung after
that message again).
Also, like I said before, the hard disk won't boot directly at the
moment; a floppy with a bootloader but no kernel (the install root
floppy works) is currently required. I think it might have to do with
the head select bits. IIRC you had implemented the Disk Control
Register properly in the new branch, and I'm using the OldMMU branch.
However, the emulator is still not quite usable.
The key mapping is not
quite right (one problem is that the UNIX PC has separate "return" and
"enter" keys, and there is no mapping for "enter", which is required
to
accept dialogs, and there might be other important keys that are
missing). Also, neither the mouse nor the RTC are emulated yet.
Andrew, all I have to say is "THANK YOU, THANK YOU, A MILLION TIMES -
THANK YOU!" The bus error logic has been a thorn in my side for ages,
and I just couldn't figure out what was going wrong. I'm really glad
someone managed to figure it out!
Get these patches to me and I'll merge them in ASAP and do my best to
fix the mouse, RTC and keyboard.
As you said, the main problem is that we don't have enough keys on a PC
keyboard to do a 1:1 mapping -- thankfully SDL supports Modifier keys,
so this isn't an insurmountable problem. The big problem is coming up
with a sane mapping :)
Truth be told, I'll probably move the repository over to Bitbucket so
people can fork it and work on it independently, then submit merge
requests. It takes a lot of work out of manually importing patches.
Is this based on the OldMMU branch or my revised MMU? I don't really
care which it is, I'll just deprecate the branch you aren't using so we
don't have a million and one build variants 8^)
8:45 a.m.
New subject: UNIX PC Hard Disk Emulator
Subject change to reflect my hijacking the thread (Sorry Andrew!)
Andrew Warkentin <andreww591 at gmail.com> writes:
Actually, Unix and the bootloader both write the correct values into
the SDH register (however, the ROM might not; it won't boot the hard
disk directly in the current version of the emulator - I'm booting
from a floppy with no kernel). The WD2010 emulation that I contributed
a while back just uses the value written to the SDH register.
Hmm, very interesting! I guess I should look further into the bootstrap
procedure (it must be in the manuals somewhere) to understand how the
hard disk is queried for geometry. Is it written somewhere on the disk,
something simple maybe? e.g. stored at C/H/S=0/0/0 in some basic format?
-Seth
3753
days inactive
3754
days old
14 comments
8 participants
participants (8)
-
andreww591＠gmail.com -
classiccmp＠philpem.me.uk -
ik＠yvanj.me -
lists＠loomcom.com -
lproven＠gmail.com -
mjkerpan＠kerpan.com -
spacewar＠gmail.com -
spedraja＠gmail.com