[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[James A. Markevitch]

I'm sending Brad a copy of a different OCR of the document that he
can use to check against the current one.  The file also includes
various fixes to bugs in the original document.

Presumably, he'll incorporate anything differences into the svn.

James Markevitch

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Brad Parker]

"James A. Markevitch" wrote:
>I'm sending Brad a copy of a different OCR of the document that he
>can use to check against the current one.  The file also includes
>various fixes to bugs in the original document.
>
>Presumably, he'll incorporate anything differences into the svn.

I made a diff and put all the changes to the kernel in svn.  I just
checked it in.

I tried "cold=1" and it seems to have written a file system to the RF
disk.  It looks correct from a dump anyway.

I think the init dies because "sys read" does not work and there are
still missing pages.

I'll add the missing pages from James' work next.

-brad

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Tim Newsham]

> I think the init dies because "sys read" does not work and there are
> still missing pages.

also you have "read" commented out of sys1.s?

> -brad

Tim Newsham
http://www.thenewsh.com/~newsham/

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Brad Parker]

Tim Newsham wrote:
>> I think the init dies because "sys read" does not work and there are
>> still missing pages.
>
>also you have "read" commented out of sys1.s?

yes, I think it has to be for this asm.

I changed the "sys read"'s to "sys _read" and then added "_read = 3."
as an experiment.  I think we'll have to sed that in eventually.

-brad

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Tim Newsham]

>> also you have "read" commented out of sys1.s?
>
> yes, I think it has to be for this asm.

There was previously a "read:" in u6.s, but it turned out to be
a typo for "readi".  I think its safe to use "read" now.

Btw, I built from latest sources and installed into simh and
ran and got some action on rf0 but I don't see any non-zero bytes
in the image after running.  It ended up in panic.  Is it possible
you have some changes that arent checked in?

I am using the clean sources with the addition of "read" in sys1.s,
an extra line of "4;4;unkni;0;fpsym;0" in u0.s and "cold = 1" instead
of zero.  My simh config is:

set cpu 11/20
set cpu 32K
set rk3 enabled
att rk3 rk0.dsk
set rf 2p
set rf enabled
attach rf rf0.dsk
set tc enabled
att tc tape.tc
set hk disabled
set rha disabled
set tm disabled
set rx disabled
set rl disabled
set tq disabled
set tc enabled
set rf enabled
set ke enabled
set cpu history=1000
d pc 400
[... lots of lines to deposit a.out into memory...]

I built as "as - sys1.s u?.s".

I run with "go" and then after the first halt "go" again.  Hitting
^E after waiting a while shows that its looping in panic.

> -brad

Tim Newsham
http://www.thenewsh.com/~newsham/

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Brad Parker]

Tim Newsham wrote:
>
>There was previously a "read:" in u6.s, but it turned out to be
>a typo for "readi".  I think its safe to use "read" now.

ah ok, thanks.

>Btw, I built from latest sources and installed into simh and
>ran and got some action on rf0 but I don't see any non-zero bytes
>in the image after running.  It ended up in panic.  Is it possible
>you have some changes that arent checked in?

I don't think so, but I'll check.  I think all I did was set cold=1
and rebuilt.  Make sure you use my "eae" version of simh.

I'm building with the v7 asm and using "ml" to process the a.out

>I am using the clean sources with the addition of "read" in sys1.s,
>an extra line of "4;4;unkni;0;fpsym;0" in u0.s and "cold = 1" instead
>of zero.

yes, that seems right.

>set cpu 11/20
>set cpu 32K
>set rk3 enabled
>att rk3 rk0.dsk
>set rf 2p
>set rf enabled
>attach rf rf0.dsk
>set tc enabled
>att tc tape.tc
>set hk disabled
>set rha disabled
>set tm disabled
>set rx disabled
>set rl disabled
>set tq disabled
>set tc enabled
>set rf enabled
>set ke enabled
>set cpu history=1000
>d pc 400
>[... lots of lines to deposit a.out into memory...]

That looks ok.  I don't have tc setup, but that should not matter.

>I built as "as - sys1.s u?.s".

same here, but I don't add the "-" (not sure what that does)

>I run with "go" and then after the first halt "go" again.  Hitting
>^E after waiting a while shows that its looping in panic.

Try "det rf".  I noticed that the RF device is buffered in ram
and I'm not sure it gets flushed until you detach.

-brad

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Tim Newsham]

>> I built as "as - sys1.s u?.s".
>
> same here, but I don't add the "-" (not sure what that does)

I tried without the dash and with the "det rf" and now I'm getting
an rf0 image.  The dash is supposed to ignore unresolved symbols.
(I added my assemv7 script to tools for building with the v7 "as"
using apout)

btw, what do we have to do about fpsym?

> -brad

Tim Newsham
http://www.thenewsh.com/~newsham/

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Brad Parker]

Tim Newsham wrote:
>
>btw, what do we have to do about fpsym?

I think we can ignore it for now (or just stub it out). It's just for
floating point emulation near as I can tell, which I don't think we'll
need for init & sh.

-brad

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Tim Newsham]

Btw, the file it tries to run at the end of u0.s's init is
</ect/init> [sic].  I entered the name as it appeared in the
printout, but its likely a typo -- have you tried running with
this fixed?

Tim Newsham
http://www.thenewsh.com/~newsham/

[Unix-jun72] OCR of PreliminaryUnixImplementationDocument_Jun72.pdf

[Brad Parker]

Tim Newsham wrote:
>Btw, the file it tries to run at the end of u0.s's init is
></ect/init> [sic].  I entered the name as it appeared in the
>printout, but its likely a typo -- have you tried running with
>this fixed?

yes, I tried it but no effect.  I don't think it's getting that far
to be honest.

if it where, I'd expect it to hang trying to read the dectape.  but
I didn't really look, maybe it errors out and does try it.

-brad

[Unix-jun72] some progress

[Brad Parker]

Apparently this instruction:

	mov    $[systm-inode]\/2,sb0+4 / sets up initial buffers per

is not doing the right thing.  I have not looked closely yet, but it should
put -109. in the sb0+4, but instead it puts a large positive number.

This is the top of the buffer, which has a WC and address in it; and it's
used to read the superblock from the RF.

anyway, when I change it to:

	mov    $-109.,sb0+4

things work much better.  I have run to a baseball game (little league :-)
but I'll retry the cold=1 later tonight; it may work better.

-brad

[Unix-jun72] some progress

[Tim Newsham]

> 	mov    $[systm-inode]\/2,sb0+4 / sets up initial buffers per

hmm, thats quite irritating.  It's computing $[systm-inode] to be
0177446 (correct) and then doing an unsigned divide by two to get
077623 instead of a signed divide by two to get 177623.

I verified that the v2 assembler behaves differently.
I can't think of a quick fix here...

> -brad

Tim Newsham
http://www.thenewsh.com/~newsham/