[TUHS] Of /dev/tty8

[TUHS] Of /dev/tty8

[Dave Horsfall]

Evenin' all...

I have a vague recollection that /dev/tty8 was the console in Edition 5 
(we only used it briefly until Ed 6 appeared), but cannot find a reference 
to it; lots of stuff about Penguin/OS though...

Something to do with 0-7 being the mux, so "8" was left (remember that 
/dev/tty and /dev/console didn't exist back then), mayhaps?

Thanks.

-- Dave

[TUHS] Re: Of /dev/tty8

[Ron Natalie]

/dev/tty existed in Version 6 for sure.   It wasn't the console but 
rather a magic device that mapped to the processes "controlling 
terminal."

Just checked the V5 and V6 sources in the archive.   /dev/tty indeed 
showed up there.   The driver is in dmr/sys.c.



------ Original Message ------
From "Dave Horsfall" <dave@horsfall.org>
To "The Eunuchs Hysterical Society" <tuhs@tuhs.org>
Date 3/18/2024 7:13:45 AM
Subject [TUHS] Of /dev/tty8

>Evenin' all...
>
>I have a vague recollection that /dev/tty8 was the console in Edition 5
>(we only used it briefly until Ed 6 appeared), but cannot find a reference
>to it; lots of stuff about Penguin/OS though...
>
>Something to do with 0-7 being the mux, so "8" was left (remember that
>/dev/tty and /dev/console didn't exist back then), mayhaps?
>
>Thanks.
>
>-- Dave

[TUHS] Re: Of /dev/tty8

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 1601 bytes --]

Dave --  As Ron pointed /dev/ty was part of UNIX early in the game. I'm
pretty sure it was in V5 but it might even go back to V3 or V4.

To your point on naming the first serial mux for the PDP-11 was the
DH11/DM11, a 16 port [full 'system unit' in the backplate]. The single
hex-height DZ board did not appear from DEC until 1977 at the earliest
(although my memory is that it was late 78).  The DZ11 is an eight-port mux
(with short pinned modem control and no DMA - both causing issues). If you
remember, the original 1979 AT&T V7 release tape does not even have a
driver for the DZ ( it's in the v7 addendum, which came out 6-9 months
later).

I don't know, but maybe tty8 was just simply that the DC11s had originally
been to the first eight serial ports, and the first DL/KL was tty8

I'm not sure why I think this ... my hazy memory is that DC11 was 0-7, the
DLs 8-15, and DHs started 16
That memory may be because when I set up the the Teklabs 11/70 with 6 Able
DH/DMs [which were 16 ports in a hex board each], I used that convention to
remember which board controlled which port.
ᐧ
ᐧ
ᐧ

On Mon, Mar 18, 2024 at 7:14 AM Dave Horsfall <dave@horsfall.org> wrote:

> Evenin' all...
>
> I have a vague recollection that /dev/tty8 was the console in Edition 5
> (we only used it briefly until Ed 6 appeared), but cannot find a reference
> to it; lots of stuff about Penguin/OS though...
>
> Something to do with 0-7 being the mux, so "8" was left (remember that
> /dev/tty and /dev/console didn't exist back then), mayhaps?
>
> Thanks.
>
> -- Dave
>

[-- Attachment #2: Type: text/html, Size: 3478 bytes --]

[TUHS] Re: Of /dev/tty8

[Paul Winalski]

[-- Attachment #1: Type: text/plain, Size: 613 bytes --]

On Mon, Mar 18, 2024 at 9:57 AM Clem Cole <clemc@ccc.com> wrote:

> The DZ11 is an eight-port mux (with short pinned modem control and no DMA
> - both causing issues).
>

I'm pretty sure that when we beta tested the VAX-11/780 in the summer of
1977 it had a DZ11.  We only had a few LA36 printing terminals on it so we
never saw any issues with it.  Not so when I started work at DEC in 1980.
The VAX-11/780 I worked on had IIRC three DZ11a supporting 24 terminals.
The lack of DMA was a big issue.  We had to run our VT52s no faster than
300 baud or it brought the system to its knees.

-Paul W.

[-- Attachment #2: Type: text/html, Size: 925 bytes --]

[TUHS] Re: Of /dev/tty8

[Larry McVoy]

On Mon, Mar 18, 2024 at 10:58:04AM -0400, Paul Winalski wrote:
> On Mon, Mar 18, 2024 at 9:57???AM Clem Cole <clemc@ccc.com> wrote:
> 
> > The DZ11 is an eight-port mux (with short pinned modem control and no DMA
> > - both causing issues).
> >
> 
> I'm pretty sure that when we beta tested the VAX-11/780 in the summer of
> 1977 it had a DZ11.  We only had a few LA36 printing terminals on it so we
> never saw any issues with it.  Not so when I started work at DEC in 1980.
> The VAX-11/780 I worked on had IIRC three DZ11a supporting 24 terminals.
> The lack of DMA was a big issue.  We had to run our VT52s no faster than
> 300 baud or it brought the system to its knees.

I was one of about 40 students working on a terminal hooked to a 780.
I put up with it for a semester and then bought a 4mhz 128KB Z80
CPM machine.  It was not that fast but it was all mine.  I was more
productive on the Z80.

I got a $2000 loan to buy it.  Taught me that I really really hated to
owe the bank.

[TUHS] Re: Of /dev/tty8

[Dave Horsfall]

On Mon, 18 Mar 2024, Ron Natalie wrote:

> /dev/tty existed in Version 6 for sure.  It wasn't the console but 
> rather a magic device that mapped to the processes "controlling 
> terminal."

I was referring to /dev/tty8, not /dev/tty...

-- Dave

[TUHS] Re: Of /dev/tty8

[Ron Natalie]

The other problem with the DZ is it was one interrupt per character if I 
recall.
Thte DH you could get multiple (output) characters per interrupt.   
Greatly decreased the load on the system.

I do remember the KL kernel prints were not interrupt driven so the 
system pretty much froze while the kernel printfs were being output.   
There was a comment on the code saying this was “Not for idle chit 
chat.”

[TUHS] Re: Of /dev/tty8

[segaloco via TUHS]

On Monday, March 18th, 2024 at 12:36 PM, Dave Horsfall <dave@horsfall.org> wrote:

> On Mon, 18 Mar 2024, Ron Natalie wrote:
> 
> > /dev/tty existed in Version 6 for sure. It wasn't the console but
> > rather a magic device that mapped to the processes "controlling
> > terminal."
> 
> 
> I was referring to /dev/tty8, not /dev/tty...
> 
> -- Dave

ttys(V) in the Sixth Edition indicates the first digit of an /etc/ttys entry indicates a terminal line is active on init and the second indicates the final character of the /dev/ttyx entry[1].  Looking at both the Fifth and Sixth Edition /etc/ttys in the archive[2][3], both only have a 1 in the first column of entry 8, corresponding with /dev/tty8.

From the setup document distributed with the Sixth Edition[4]:

"The same goes for the character devices.
Here the names are arbitrary except that
devices meant to be used
for teletype access should be named
/dev/ttyX, where X is any character.
The files tty8 (console), mem, kmem, null are
already correctly configured."

From all of this it appears that by convention, tty8 was indeed the default console /dev entry, although this could be changed by editing the conf and ttys entries followed by regeneration of the system.  This would change in the Seventh Edition with the rearrangement of the ttys file to indicate longer /dev entry names and the establishment of a specific /dev/console entry.

- Matt G.

[1] - https://www.tuhs.org/cgi-bin/utree.pl?file=V6/usr/man/man5/ttys.5

[2] - https://www.tuhs.org/cgi-bin/utree.pl?file=V5/etc/ttys

[3] - https://www.tuhs.org/cgi-bin/utree.pl?file=V6/etc/ttys

[4] - https://www.tuhs.org/cgi-bin/utree.pl?file=V6/usr/doc/start/start

[TUHS] Re: Of /dev/tty8

[segaloco via TUHS]

On Monday, March 18th, 2024 at 1:08 PM, Ron Natalie <ron@ronnatalie.com> wrote:

> The other problem with the DZ is it was one interrupt per character if I
> recall.
> Thte DH you could get multiple (output) characters per interrupt.
> Greatly decreased the load on the system.
> 
> I do remember the KL kernel prints were not interrupt driven so the
> system pretty much froze while the kernel printfs were being output.
> There was a comment on the code saying this was “Not for idle chit
> chat.”

Yep, synchronous code that writes a character at a time to the transmit register then spins on a status bit awaiting transmit complete, or something like that.  I adapted the V6 kernel printf to a RISC-V board I was working on the past year to have a trustworthy print mechanism, only caveat being the transmitter status register was a dirty filthy liar and wouldn't flip the bit on transmit, so just had to put a delay, which it turns out all of the BSDs currently also do for Ti 16550-family UARTs as well.

- Matt G.

[TUHS] Re: Of /dev/tty8

[Bakul Shah via TUHS]

On Mar 18, 2024, at 1:21 PM, segaloco via TUHS <tuhs@tuhs.org> wrote:
> 
> only caveat being the transmitter status register was a dirty filthy liar and wouldn't flip the bit on transmit, so just had to put a delay, which it turns out all of the BSDs currently also do for Ti 16550-family UARTs as well.

NatSemi (later Ti) 16550s have a 16 byte fifo in each direction.
Any delays would be in getc(), putc(), used for system console io,
not general serial io. Ignoring any code for brokenness! The early
16660 versions had a bug preventing use of the fifo but this was
fixed ages ago. At least this is what I remember decades later!