🚀 go-pugleaf

david@wubios.wustl.edu (David J. Camp) writes:
>I have heard a story that a programmer at AT&T placed a virus in the
>cc compiler.  It would detect when cc itself was being compiled and
>reproduce itself.  I do not remember what it does to the rest of the
>system, but there was mention of a Happy Birthday greeting for the author.  

Wrong, it [DANGER--DANGER--WARNING FOR THE HUMOR IMPAIRED:  THIS IS SARCASM:::]
tells you to send a get-well card to Craig Shergold.

>It is possible that gcc is infected with this virus.  It is routinely
>compiled with the AT&T compiler.  Perhaps we need one assembler
>version of the C compiler, to bootstrap the others.  -David-

No.  No.  No.  No.

Ken Thompson's Turing Award lecture, published in the Communications of the ACM
(go look it up, I don't have an index handy; the title is "Reflections on
Trusting Trust", and I think it was between 1980 and 1985) describes his virus,
an interesting exercise in self-replicating code.  That version of the compiler
was NEVER shipped, though it almost got sent to a military site (he heard
of the impending shipment, and told them there was a last-minute bugfix that
had to go into the compiler before making the tape).

READ THE CACM ARTICLE.  Always assume that everything you read on USENET is
false (including this message here, but at least I've given you enough supposed
information that you can verify for yourself whether or not it's true).

(By the way, the virus did two things:  first, if it was compiling the login
program, it added a special magic password for "root"; second, if it was
compiling the C compiler, it inserted the code which checks for whether or
not it is compiling login and adds the magic password, or for whether it is
compiling the the C compiler, in which case it inserts the code for.......
How the latter is accomplished is the interesting exercise.)