12 Jun
2016
12 Jun
'16
8:08 p.m.
I have been able to use my Data General Nova 3's Program Load function to load in the
091-000036-01 self-loading bootstrap program, and then use that bootstrap to load in
absolute binary tape images from *.ab files, all over the TTI channel from a terminal
emulator.
Now, I'm trying to hack Toby Thain's Nova assembler to emit auto-loading images by
prepending the self-loading bootstrap to the assembled program, with a binary patch to
make the bootstrap jump straight into the absolute binary loader rather than halting at
address 000121. It seems to me that I should be able to use this scheme to assemble
programs which I can then Program Load through the TTI channel, and simply send the image
over the serial port from a terminal emulator.
I have the assembler hacked up to prepend the patched loader, and the patch I made seems
to cause the bootstrap to jump right into the absolute binary loader; I simply replaced
the HALT that ended up in 000120 with a JMP 000121. However, the binary loader continues
running after presumably digesting the data that followed the bootstrap.
Toby's assembler outputs "relocatable binary" code in .rb files, and I
don't understand if/how that differs from the "absolute binary" format in
the .ab tape images that I can successfully load. I'll continue researching, but
I'd appreciate it if anybody might be able to provide any clues to dispel my
confusion.
Regarding my hacked up assembler, I don't know whether it's advisable to post it
publicly after getting it working, since it'll have original DGC binary code embedded
in it. Opinions?
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
12 Jun
12 Jun
8:39 p.m.
Ok, I've found a description of the .rb format in manual 093-000081-02 for the Macro
Assembler, and a description of the .ab format in manual 093-000003-06 for the Binary
Loader. Now, off to dig into Toby's assembler some more to figure out how to make it
emit absolute binary!
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
11:02 p.m.
On 6/12/2016 5:39 PM, Mark J. Blair wrote:
Ok, I've found a description of the .rb format in
manual 093-000081-02 for the Macro Assembler, and a description of the .ab format in
manual 093-000003-06 for the Binary Loader. Now, off to dig into Toby's assembler some
more to figure out how to make it emit absolute binary!
Is there a linker included in his toolchain? That is where you need to
look. If it is designed to run for an OS, or with an smart loader, it
will take in object modules and emit them with relocatable addressing
resolution still required.
There should be some linker control to do placement like you wish to
have it do, which would be absolute binary with the loader directives
other than object addresses and text gone.
thanks
Jim
11:19 p.m.
On Jun 12, 2016, at 20:02, jwsmobile <jws at
jwsss.com> wrote:
Is there a linker included in his toolchain? That is where you need to look. If it is
designed to run for an OS, or with an smart loader, it will take in object modules and
emit them with relocatable addressing resolution still required.
There should be some linker control to do placement like you wish to have it do, which
would be absolute binary with the loader directives other than object addresses and text
gone.
There's no linker in the toolchain. There is an object dumper utility that the author
suggested I look at. I'm presently hacking that relocatable binary dumper into an
absolute binary converter, with the ability to prepend the auto program loader bootstrap,
and the constraint that it only supports absolute code. I think that should be sufficient
for my current purpose, which is writing short and simple test programs which I can then
bootstrap over the TTI channel. Adding proper linking capabilities may be a good thing in
the future.
Anyway, I've pulled out my hacks of the assembler itself, since the object dumper is
an easier place to splice in the absolute binary conversion.
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
11:19 p.m.
On 2016-06-12 11:02 PM, jwsmobile wrote:
On 6/12/2016 5:39 PM, Mark J. Blair wrote:
No, there is no linker. It was intended to be compatible with the RDOS
tools, but I never built a complete toolchain. I don't actually own a
Nova, although I think it's a fun machine -- this was inspired by a Nova
that I helped a friend obtain.
Ok, I've found a description of the .rb
format in manual 093-000081-02
for the Macro Assembler, and a description of the .ab format in manual
093-000003-06 for the Binary Loader. Now, off to dig into Toby's
assembler some more to figure out how to make it emit absolute binary!
Is there a linker included in his toolchain? That is where you need to look. If it is designed to run for an OS, or with an
smart loader, it
will take in object modules and emit them with relocatable addressing
resolution still required.
There should be some linker control to do placement like you wish to
have it do, which would be absolute binary with the loader directives
other than object addresses and text gone.
It's pretty easy to hack up a tool to do that much, I think. There is an
RB parsing utility in the source code.
--Toby
thanks
Jim
13 Jun
13 Jun
3:17 a.m.
On Jun 12, 2016, at 20:19, Toby Thain <toby at
telegraphics.com.au> wrote:
It's pretty easy to hack up a tool to do that much, I think. There is an RB parsing
utility in the source code.
I've modified that RB parser to output AB, for inputs that do not require any
relocation. Testing was very minimal... I managed to use Program Load to bootstrap the
little test program I shared yesterday in my 6045 drive thread after assembling it with
Toby's dga, then converting it to a self-loading, self-executing image with the
modified parser. I simply set the switches to 000010 (the TTI channel), hit PROGRAM LOAD,
then sent the image with a terminal emulator. After the image transferred, the drive
clunked as it did a full-stroke seek. I then halted the program and used the console to
verify that the program got loaded as intended.
Now, I can use this to write additional test programs to debug my disk drive! For example,
I'll probably write one that polls the DONE flag after a seek to see if the I/O status
flags are working despite status register reads returning zero.
It's early work, and likely to be buggy, but here's my fork:
https://github.com/NF6X/dpa
--
Mark J. Blair, NF6X <nf6x at nf6x.net>
http://www.nf6x.net/
2962
days inactive
2962
days old
5 comments
3 participants
participants (3)
-
jws＠jwsss.com -
nf6x＠nf6x.net -
toby＠telegraphics.com.au