Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1a 12/4/83; site rlgvax.UUCP
Path: utzoo!linus!decvax!harpo!seismo!rlgvax!guy
From: guy@rlgvax.UUCP (Guy Harris)
Newsgroups: net.unix-wizards
Subject: Re: horror stories
Message-ID: <1732@rlgvax.UUCP>
Date: Sun, 19-Feb-84 13:15:25 EST
Article-I.D.: rlgvax.1732
Posted: Sun Feb 19 13:15:25 1984
Date-Received: Mon, 20-Feb-84 02:18:15 EST
References: <16771@sri-arpa.UUCP>
Organization: CCI Office Systems Group, Reston, VA
Lines: 21

>     "Anyone ever think of swapping user processes to the CRT screen (in
>     hex or octal, of course), and requiring the user to type them back
>     in at the proper time?                      :-)"

> Well, if you type "RUN TT:" to VMS, it calmly waits for you to type in
> a VMS executable image at your terminal.  I've never managed to type
> one that it liked, but if I did, VMS likes to page the executable
> against the file it came from, so...

Fortunately, UNIX at least refuses to run a file which isn't a plain file, so
turning on execute permission for "/dev/ttyXX" or "/dev/tty" and running
it as a command would fail.  If that restriction were removed from the code,
the Berkeley VMUNIX would die a horrible death; it assumes that the inode that
the file is being paged from is a regular inode containing the usual file map,
and it uses the file map to set up the disk addresses of the text pages.  (I
can't speak for Bell's virtual memory implementation(s).)  I'd be curious to
see what VMS does in the same situation; presumably, it gets the file map by
asking the Files-11 ACP, but if there's no ACP involved, what does it do then?

	Guy Harris
	{seismo,ihnp4,allegra}!rlgvax!guy