[H-GEN] PHP-11 emulator in Javascript!

[Russell Stuart]

On Thu, 2011-05-05 at 20:48 +1000, Robert Brockway wrote: 
> Amazing...
> 
> http://aiju.de/code/pdp11/

I confess to thinking it must be some sort of hoax.  But it isn't.

I presume Greg and I were one of the few who can comfortably use the
native v6 editor with referring to a manual.  I guess that is indicative
of being a certain age, however even I was mystified by the command dsw,
so I tracked it down.  Turns out it is the predecessor of rm -i, however
its usage is somewhat atypical.  From
http://tldp.org/LDP/LG/issue49/fischer.html:

        "To use the original dsw, you would set the front panel switches
        to a number and run dsw, which would then read that number of
        entries from the current directory, print on the terminal the
        name of the last one it had read, and crash itself, leaving a
        core dump. You could then, if you wanted, remove the file it had
        named by re-executing the core file."

Sadly the emulator doesn't have the front panel switches.

Seeing the instruction set of a PDP-11 reduced to 1000 odd lines of
javascript fired off mixed emotions.  I recall being very proud of
mastering it quickly, so it is deflating to see it made so simple.
Another experience a couple of months ago to puts it into a different
perspective. I was doing some kernel hacking and was forced to download
the Intel x86 Architecture Manuals, which comes in 5 volumes totalling
some 4000 pages.   I spent many days dredging through them, trying to
understand the finer details of how switch between ring 0 and ring 3
effects the memory mapping.  At the end of what felt like a journey to
hell and back, I wondered how we got to this point.

To get a feel for why, consider the pseudo code for the add instruction.
Here is it copy and pasted from the Intel manual:

  DEST ← DEST + SRC;

Notice how neatly the pseudo code abstracts away all the messy details
surrounding memory addressing, faults, flags and instruction decoding.
The pseudo code for the call instruction is written in a similar way.  I
won't list it here because it it around 350 lines long.

To put it another way, it takes roughly 1/3 of the code needed to
implement an entire PDP-11 to implement the call instruction on the x86.
Almost all of this pseudo code implements features no one uses, because
they are too slow.  If the features no one uses weren't there, the
pseudo code for call would be:

  Push(IP); IP ← DEST;

To me, the modern x86 architecture looks like it was designed using the
same mentality as dsw's user interface.