We lost software engineer Dick Hustvedt on this day in 2008, following
severe injuries in a vehicle accident. He contributed much to RSX-11 and
VMS, including the infamous "microfortnight" and the SD730 Fixed Head
Solar Horologue. An obituary of him can be found at
http://software.intel.com/en-us/blogs/2008/04/23/dick-hustvedt-the-consumma…
(and it's worth reading).
--
Dave Horsfall DTM (VK2KFU) "Those who don't understand security will suffer."
We lost Robert Taylor, computer scientist and Internet pioneer, on this
day in 2017. Amongst other things, he helped invent the mouse, pioneered
computer communications leading up to ARPAnet, developed the computer
science lab at Xerox...
--
Dave Horsfall DTM (VK2KFU) "Those who don't understand security will suffer."
Today I reached a minor milestone with my 'Venix restoration project' that
I talked about months ago. I ran a Venix 86 binary (sync) to successful
completion on my FreeBSD/amd64 box (though none of the code should be too
FreeBSD specific).
So, I hacked https://github.com/tkchia/reenigne.git to remove the DOS
loader and emulator and to add a Venix system call loader and emulator, or
at least the start of one. I've also added FP instruction parsing, but it's
100% wrong (it just parses the instructions and does nothing else to decode
or implement them). With this, I'm able to load OMAGIC binaries from the
extant venix 86 distributions and run them. The only one that runs
successfully is sync() since I've not yet implemented argument passing or
any of the other 58 system calls :). NMAGIC should be pretty quick after
this.
This is but a step on the road to getting the Venix compiler running so I
can see how much of the system I can recreate from the v7 and other sources
that are on TUHS.
Not sure who, if anybody, cares about this stuff. I thought people here
might be interested. I've pushed the results to
https://github.com/bsdimp/venix if you care. This program is in the
tools/86sim directory. There's also a doc directory where I document the
Venix 86 ABI, as well as doing a very deep-dive into a disassembled
/bin/sync to discover what I can from it (turns out, it's quite a lot).
So, I thought I'd share this here. Don't know if anybody else is
interested, but you never know until you tell people about stuff...
Warner
I was sure that I'd read a paper on the legal history of Unix. So I did a
Google search for it, and found a link to the PDF. The linked PDF was on
the TUHS website :-)
http://wiki.tuhs.org/lib/exe/fetch.php?media=publications:theses:gmp_thesis…
I'd better do a backup of my brain, as I've got a few flakey DRAM chips.
Cheers, Warren
> From: Clem Cole
> first of Jan 83 was the day the Arpanet was supposed to be turned off
Err, NCP, not the ARPANet. The latter kept running for quite some time,
serving as the Internet's wide-area backbone, and was only slowly turned off
(IMP by IMP) in the late 80's, with the very last remnants finally being
turned off in 1990.
> The truth is, it did not happen, there were a few exceptions granted for
> some sites that were not quite ready (I've forgotten now).
A few, yes, but NCP was indeed turned off for most hosts on January 1, 1983.
> From: "Erik E. Fair"
> as of the advent of TCP/IP, all those Ethernet and Chaosnet connected
> workstations became first class hosts on the Internet, which they
> could not be before.
Huh? As I just pointed out, TCP/IP (and the Internet) was a going concern well
_before_ January 1, 1983 - and one can confidently say that even had NCP _not_
been turned off, history would have proceeded much as it actually did, since
all the machines not on the ARPANET would have wanted to be connected to the
Internet.
(Also, to be technical, I'm not sure if TCP/IP ever really ran on CHAOSNet
hardware - I know I did a spec for it, and the C Gateway implemented it, and
there was a Unix machine at EECS that tried to use it, but it was not a great
success. Workstations connected to the CHAOSNet as of that date - AFAIK, just
LISP Machines - could only get access to the Internet via service gateways,
since at that point they all only implemented the CHAOS protocols; Symbolics
did TCP/IP somewhat later, IIRC, although I don't know the exact date.)
Noel
> I rewrote the article on the Software Tools project
An excellent job, Deborah.
> the Software Tools movement established one of the earliest traditions of open source
Would you be open to saying "reestablished"? Open source (not so called,
and in no way portable) was very much a tradition of SHARE in the late
1950s. Portability, as exemplified in ACM's collected algorithms, came
in at the same time that industry moved to a model of trade secrets and
intellectual property. Open source went into eclipse.
Doug
I rewrote the article on the Software Tools project and, thanks to Bruce
Borden's efforts to upload, they accepted it within 1 day. You can see
it here: https://en.wikipedia.org/wiki/Software_tools_users_group
The Usenix article in Wiki is pretty thin, in case anyone would like to
spiffy it up.
Deborah
> From: Steve Nickolas
> I thought the epoch of the Internet was January 1, 1983.
Turning off NCP was a significant step, but not that big a deal in terms of
its actual effects, really.
For those of us already on the Internet before that date (since as the number
of ARPANet ports was severely limited, for many non-ARPANet-connected machines
- which were almost all time-sharing systems, at that point in time, so lots
of actual users - there was a lot of value in an Internet connection, so there
were quite a few), it didn't produce any significant change - the universe of
machines we could talk to didn't change (since we could only talk to
ARPANet-connected machines with TCP), etc.
And for ARPANET-connected machines, there too, things didn't change much - the
services available (remote login, email, etc) remained the same - it was just
carried over TCP, not NCP.
I guess in some sense it marked 'coming of age' for TCP/IP, but I'd analogize
it to that, rather than a 'birth' date.
Noel
> From: Clem Cole
> Katie Hafner's: Where Wizards Stay Up Late: The Origins Of The Internet
> ...
> It's a great read
Yes, she did a great deal of careful research, and it's quite accurate.
It _is_ pointed toward a general readership, not a technical one, so it's not
necessarily the best _technical_ history (which she had the material at hand
to produce, had she wanted to - but did not). Still, very worthwhile.
Noel
A nerdy group on an Aussie list are discussing old Unix cracks, and the
infamous "SPL 0" brick-that-box came up. I first saw it in ";login:" (I
think), and, err, tried it (as did others)...
Can anyone reproduce the code? It went something like:
> [ SPL 0 ]
>
> I only did that once (and you should've heard what he said to me)...
> I'm still trying to find the source for it (it was published in a
> ";login:" journal) to see if SIMH is vulnerable.
The concept was simple enough - fill your entire memory space with an uninterruptible instruction. It would have gone something like:
opc = 000230 ; 000230 is the opcode for SPL 0
sys brk, -1 ; or whatever value got you all 64k of address space
mov #place, sp
jmp place
. = opc - 2 ; the -2 is to allow for the PC increment on an instruction fetch, which I believe happens before any execution
place:
jsr pc, -(pc)
Ring any bells, anyone?
--
Dave Horsfall DTM (VK2KFU) "Those who don't understand security will suffer."