Just ran across this:
which isn't available online in this form. (This appears to be a different scan
>from the one on the Maine Coon site, split up into several TIFF's, as it has
the cover which that one doesn't show.)
As always, history shows that the best way not to lose things is to have
multiple independent copies! So download often!
I've received a couple of suggestions, thanks, but none seem right.
BTW, I'm sorry (Liam) that I didn't make it clearer that it was absolutely
a software bug, which excludes Spectre, Rowhammer, Meltdown.
Aside: the Meltdown and/or Spectre patch to macOS hurt performance ... the
elapsed time to compile 500 programs increased by about 12%. (At the time
of the patch, I tested (properly) before and after the patch ... then
forgot to publish, and since misplaced my notes.)
Although I mentioned 'code', I should have been more specific: C (or,
possibly, C++), but definitely no other language.
I don't recall it being a buffer overflow.
I *think* it was some kind of authentication failure (e.g., incorrectly
reporting "ok"), but I'm not sure.
I do know I wrote a several page article about it, and how certain coding
practices led to it, but I can't *find* the article now :(
My guess of 4-6 years ago is possibly narrower than it should be, but I'm
My hope is that by being reminded of the vulnerability name, I can search
my computer, and backups, for text files containing that name :)
(Or the name of function associated with the problem.)
I'm trying to remember the name (and some information about) a past
security bug, for an article.
Somewhere between 4 and 6 years ago (I think), there was a fairly major
security bug reported (probably in Linux, or in SSH code, but
something widely used).
IIRC, the bug was a single line that called a function (possibly along the
lines of CredentialsCheck), and may have involved a bit-wise or (or and)
instead of a logical one.
It may have been that either the routine wasn't getting called when it
should, or that the programmer misinterpreted what the return value meant.
Ring any bells?
> an adapter cable to go from a 9-pin male (shell; female pins) to a
> 15-pin female (shell; male pins)
Sigh, shouldn't try to type when I'm this tired. Female 9-pin (to plug into
the BA11-D) to male 15-pin (for the DD11-C/D to plug into).
> From: Chris Zach
> I'm guessing that the DD11-F is significantly different from the DD11-B?
I assume theat "DD11-F" is a typo; there is, AFAIK, no DD11-F, and a Web
search revealeddidn't turn anything up. (There are DD11-CF and -CK
backplanes, as well as -DF and -DK, but the -CF and -CK differ only in power
> the 11/24 used +12 on the +15 lines. No idea what was wrong with DEC
I'm guessing it used the parts that the IC vendors could provide.
> Don't know what would happen if you plugged a RL11 or other hex height
> card into one of those slots, probably blow everything up.
Not sure. Per:
These are _some_ of the MUD/EUB pin clashes:
Pin EUB MUD
AN1 - A21 - Parity P1
AP1 - A20 - Parity P0
BE1 - A19 - Internal SSYN
BE2 - A18 - Parity Detect
I don't think those would harm anything. Not sure about power pins - and have
no incentive to research it, as I have no need/interest in trying it.
> I know I ran it with two of these Plessy cards and a RX01 controller but
> now that I look at it that would be impossible as both were hex cards
Maybe Plessey designed them to go in a DD11-B? I know I've seen other
third-party cards that would go in oddball slots.
> and both could never fit in a 4 slot backplane with enough space for a
> quad spc
Why not? The two hexes in slots 2&3, the quad in 1 or 4.
> a DA11-F Unibus window
Wow; never heard of those. I'll have to do a CHWiki page for them. Luckily,
the maint manual is online. There's also a DA11-B.
> what would happen if I enabled the KT24 Unibus memory map.....
All that does is allow DMA devices on the -11/24's UNIBUS access to the
entire main memory:
So if you hooked up an -11/10 to an -11/24 with a DA11-F, then with the UNIBUS
Map on, the -11/10's CPU and/or DMA devices would have access to the entire
EUB memory via the 24's UNIBUS Map, is all.
> This is so much fun!
That _is_ why we collect old computers! ;-)
> From: Chris Zach
> the DD11-B is a MUD backplane
No, it's SPC; other sources, e.g.
So if you have a DD11-B, you must have a BA11-D, with the 9-pin power
The best thing to do is get a DD11-C or -D, and build an adapter cable to go
>from a 9-pin male (shell; female pins) to a 15-pin female (shell; male pins),
so you don't have to mess with the harness. Part numbers here:
Then you can plug in any memory you've got the right voltages for; the MS11-E
takes + and -15V (in addition to +5V, of course).
I'm working on my pdp11/10 getting it back together. One problem I think
I have is that the secondary memory (a Plessy 700101-100) may be
shorting the -15 line for some reason. Working on it, but does anyone
have a manual or anything like that for this kind of memory board?
Alternately, what kind of Unibus 16k memory board exists to get a 11/10
>from 16kw to 32kw of memory? Apparently I can't use a MM11-B as it
requires +20 and -5, both of which are not provided by the power supply
or sourced on a DD11-B backplane module. The 11/10 has +15, -15, and +5.
On a related note, where did +20 come from for Unibus and which systems
even supported it? Was it an 11/45,11/70 thing?
(Yes, I could build a regulator to take the 30 volts between +15 and -15
and create an independent 20 volts. Maybe. Likewise I could generate -5
>from the -15 and a 7815 regulator. Maybe.)
> From: Chris Zach
> the secondary memory (a Plessy 700101-100) may be shorting the -15 line
> for some reason. Working on it, but does anyone have a manual or
> anything like that for this kind of memory board?
I've got a Plessey core memory manual somewhere, but I can't find it, so I
don't know if it's the one you are looking for. I got it from Paul Birkel; it
was a duplicate, and he scanned his and sent the scan off, but I don't think
it made it online.
> Alternately, what kind of Unibus 16k memory board exists to get a 11/10
> from 16kw to 32kw of memory?
It all depends on what kind of -11/10 you have.
If yours is in a 5-1/4" box, you can't plug a DEC memory card into the SPC
slots that some of the CPU-holding backplane versions have because DEC
memories (other than the ones like the MM11-L and -U, which are multi-board
core systems that require custom backplanes) all require MUD slots, not SPC.
All of the CPU backplanes on that machine are for a _specific_ kind of core
memory (MM11-L or MM11-U), see here:
There are I think some third-party memories which can be used (Dataram,
maybe?) but I don't have time to go into them.
If you have a 10-1/2" box, you can mount a MUD backplane - but you might
still have an issue because the older BA11-D boxes use the old 9-pin power
connectors, and the MUD backplanes (DD11-C, -D, etc) all use the newer 15-pin
(Again, there are some oddball ones, and again, I don't have time to go into
If you're lucky enough to have one of the ones that will take a MUD backplane,
would be an option.
> On a related note, where did +20 come from for Unibus and which systems
> even supported it? Was it an 11/45,11/70 thing?
The later /05's, /40's and /45's were the first ones to provide +20V, for the
then-new MM11-U. On machines which took H744 'brick's, the _later_ harnesses
could take a H754 +20V, -5V regulator 'brick'. Alternatively, _some_ BA11-L's
(used for the /04 and /34) had the right version of H777:
to provide +20V.
To make testing of the H744 a bit easier I would like to try to make up some
connectors for the mate-n-lok connector to make it easier to connect power
and load. I know the H744 uses a mate-n-lok connector, but there seem to be
a lot of different types and I don't seem to be able to find a type that
would work. Does anyone know what the correct one is?