[TUHS] asm helper routines in v7

[TUHS] asm helper routines in v7

[Will Senn]

[-- Attachment #1: Type: text/plain, Size: 1738 bytes --]

So,

in v6, it was possible to use the mesg function from the system library 
with:

         ed hello.s
/ hello world using external mesg routine

	.globl  mesg

         mov     sp,r5
         jsr     r5,mesg; <Hello, World!\n\0>; .even
         sys     exit


	as hello.sld -s a.out -l
	a.out
	Hello, World!

This was because v6 included mesg in the library, in v7, it doesn't look 
like mesg is included, so doing the same thing as above requires that 
code to write the message out be included and in addition system call 
names are not predefined, so exit and write have to be looked up in 
/usr/include/sys.s, resulting in the v7 equivalent file:

    ed hello2.s
    / hello world using internal mesg routine

             mov     sp,r5
             jsr     r5,mesg; <Hello, World!\n\0>; .even
             sys     1

    mesg:
             mov     r0,-(sp)
             mov     r5,r0
             mov     r5,0f
    1:
             tstb    (r5)+
             bne     1b
             sub     r5,r0
             com     r0
             mov     r0,0f+2
             mov     $1,r0
             sys     0; 9f
    .data
    9:
             sys     4; 0:..; ..
    .text
             inc     r5
             bic     $1,r5
             mov     (sp)+,r0
             rts     r5

    as hello2.s
    a.out
    Hello, World!

My questions are:

1. Is mesg or an equivalent available in v7?
2. If not, what was the v7 way of putting strings out?
3. Why aren't the system call names defined?
4. What was the v7 way of naming system calls?

Will


[-- Attachment #2: Type: text/html, Size: 2447 bytes --]

Re: [TUHS] asm helper routines in v7

[Dan Cross]

On Tue, Jan 4, 2022 at 3:38 PM Will Senn <will.senn@gmail.com> wrote:
> [snip]
> My questions are:
>
> 1. Is mesg or an equivalent available in v7?

Perhaps just call printf?

> 2. If not, what was the v7 way of putting strings out?

Here's what I did:

$ rm -f h h.o
$ cat h.s
.globl _printf, _exit

mov     sp, r5
mov     $hi,(sp)
jsr     pc,*$_printf
mov     $0, (sp)
jsr     pc,*$_exit

hi: <Hello, World!\n\0>
$ as -o h.o h.s
$ ld -o h h.o -lc
$ ./h
Hello, World!
$ rm h h.o

> 3. Why aren't the system call names defined?

If I had to hazard a guess, it would have been to de-emphasize the use
of assembler for user code, particularly as 7th Edition was starting
to be portable beyond the PDP-11.

> 4. What was the v7 way of naming system calls?

I imagine the canonical way to invoke system calls from assembler was
invoking calling functions in the C library and linking against that.

        - Dan C.

Re: [TUHS] asm helper routines in v7

[Will Senn]

[-- Attachment #1: Type: text/plain, Size: 1107 bytes --]

On 1/4/22 3:12 PM, Dan Cross wrote:
> On Tue, Jan 4, 2022 at 3:38 PM Will Senn<will.senn@gmail.com>  wrote:
>> [snip]
>> My questions are:
>>
>> 1. Is mesg or an equivalent available in v7?
> Perhaps just call printf?
>
>> 2. If not, what was the v7 way of putting strings out?
> Here's what I did:
>
> $ rm -f h h.o
> $ cat h.s
> .globl _printf, _exit
>
> mov     sp, r5
> mov     $hi,(sp)
> jsr     pc,*$_printf
> mov     $0, (sp)
> jsr     pc,*$_exit
>
> hi: <Hello, World!\n\0>
> $ as -o h.o h.s
> $ ld -o h h.o -lc
> $ ./h
> Hello, World!
> $ rm h h.o
>
>> 3. Why aren't the system call names defined?
> If I had to hazard a guess, it would have been to de-emphasize the use
> of assembler for user code, particularly as 7th Edition was starting
> to be portable beyond the PDP-11.
>
>> 4. What was the v7 way of naming system calls?
> I imagine the canonical way to invoke system calls from assembler was
> invoking calling functions in the C library and linking against that.
>
>          - Dan C.
Yep, that worked... you make it look so easy, and your rationales are 
definitely believable :).

Will

[-- Attachment #2: Type: text/html, Size: 2208 bytes --]

Re: [TUHS] asm helper routines in v7

[Warner Losh]

[-- Attachment #1: Type: text/plain, Size: 1739 bytes --]

On Tue, Jan 4, 2022 at 2:24 PM Will Senn <will.senn@gmail.com> wrote:

> On 1/4/22 3:12 PM, Dan Cross wrote:
>
> On Tue, Jan 4, 2022 at 3:38 PM Will Senn <will.senn@gmail.com> <will.senn@gmail.com> wrote:
>
> [snip]
> My questions are:
>
> 1. Is mesg or an equivalent available in v7?
>
> Perhaps just call printf?
>
>
> 2. If not, what was the v7 way of putting strings out?
>
> Here's what I did:
>
> $ rm -f h h.o
> $ cat h.s
> .globl _printf, _exit
>
> mov     sp, r5
> mov     $hi,(sp)
> jsr     pc,*$_printf
> mov     $0, (sp)
> jsr     pc,*$_exit
>
> hi: <Hello, World!\n\0>
> $ as -o h.o h.s
> $ ld -o h h.o -lc
> $ ./h
> Hello, World!
> $ rm h h.o
>
>
> 3. Why aren't the system call names defined?
>
> If I had to hazard a guess, it would have been to de-emphasize the use
> of assembler for user code, particularly as 7th Edition was starting
> to be portable beyond the PDP-11.
>
>
> 4. What was the v7 way of naming system calls?
>
> I imagine the canonical way to invoke system calls from assembler was
> invoking calling functions in the C library and linking against that.
>
>         - Dan C.
>
> Yep, that worked... you make it look so easy, and your rationales are
> definitely believable :).
>

You can use puts too, at least in v7.

The system call interface between V6 and V7 changed as well...  Some
internal AT&T branches
(cb unix for one) could run both V6 and V7 binaries. BSD uses the v7 system
call interface through
2.10.x. 2.11 moves to a new system call interface as well. There's niggles
in these statements, since
it isn't quite exactly the same, but for things like this it's ok.

V7 also moved to stdio completely. V6 had precursor routines to stdio and
mesg() was dropped in
that whole process.

Warner

[-- Attachment #2: Type: text/html, Size: 2820 bytes --]

Re: [TUHS] asm helper routines in v7

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 563 bytes --]

On Tue, Jan 4, 2022 at 3:38 PM Will Senn <will.senn@gmail.com> wrote:

> 1. Is mesg or an equivalent available in v7?
>
its not part of the C stdio library and by te

> 2. If not, what was the v7 way of putting strings out?
>
_puts   declared as an external

> 3. Why aren't the system call names defined?
>
Ah .. you want the DEC assembler ;-)    By the time of V5/V6 assembler is
starting to have less importance and by V7, little if anything is left.

> 4. What was the v7 way of naming system calls?
>
_XXX  where XXX is the system call declared as a global.

[-- Attachment #2: Type: text/html, Size: 1995 bytes --]

Re: [TUHS] asm helper routines in v7

[Dan Cross]

On Tue, Jan 4, 2022 at 4:28 PM Warner Losh <imp@bsdimp.com> wrote:
> On Tue, Jan 4, 2022 at 2:24 PM Will Senn <will.senn@gmail.com> wrote:
>> Yep, that worked... you make it look so easy, and your rationales are definitely believable :).

It would be nice to hear about the rationale from a primary source.
Perhaps Ken or Doug might chime in?

> You can use puts too, at least in v7.

Ah right; I always forget that puts was unequivocally a function in
7th Ed. It'll even add a newline for you!

        - Dan C.

Re: [TUHS] asm helper routines in v7

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 303 bytes --]

On Wed, Jan 5, 2022 at 10:27 PM Douglas McIlroy <
douglas.mcilroy@dartmouth.edu> wrote:

>  As I wrote in A Research Unix Reader, "Assembly language and magic
> constants gradually declined from the
> status of the 'real truth' (v4) to utterly forgotten (v8)."

Ah ... theory and practice in alignment.

[-- Attachment #2: Type: text/html, Size: 1006 bytes --]

Re: [TUHS] asm helper routines in v7

[Warner Losh]

[-- Attachment #1: Type: text/plain, Size: 796 bytes --]

On Wed, Jan 5, 2022, 8:27 PM Douglas McIlroy <douglas.mcilroy@dartmouth.edu>
wrote:

> > It would be nice to hear about the rationale from a primary source.
>
> Assembly language was deemed a last resort, especially as portability
> was coming to the fore. As I wrote in A Research Unix Reader,
> "Assembly language and magic constants gradually declined from the
> status of the 'real truth' (v4) to utterly forgotten (v8)." In v7,
> assembler usage was demoted to the bottom of syscall man pages. It
> could also be found in /usr/src/libc/sys/*.s
>


That makes a lot of sense. I've found errors and omissions in the assembler
documentation for the odd system call for V7 and System III ports. A little
sloppy I thought when I noticed.  I'll have to see if I can dig one up
again...

Warner

>

[-- Attachment #2: Type: text/html, Size: 1420 bytes --]

[TUHS] asm helper routines in v7

[Douglas McIlroy]

> It would be nice to hear about the rationale from a primary source.

Assembly language was deemed a last resort, especially as portability
was coming to the fore. As I wrote in A Research Unix Reader,
"Assembly language and magic constants gradually declined from the
status of the 'real truth' (v4) to utterly forgotten (v8)." In v7,
assembler usage was demoted to the bottom of syscall man pages. It
could also be found in /usr/src/libc/sys/*.s

Doug