vi and heavy loads

[Robert_Toxen%anvil.UUCP at HARVARD.HARVARD.EDU]

> From: Guy Harris <guy%sun.uucp at BRL.ARPA>
> Subject: RE: Re: vi & heavy loads
> 
> > Steps should be taken to prevent subshells spawned from vi from also running
> > at -10 priority. Since we all used csh we just added 'nice 0' the our
> > .cshrc files.
> > 
> > Paranoid types and sh users can diddle vi appropriately.

> Since this "step" is voluntary, why "should" these steps be taken?  To be
> "nice" (pun only partially intended)?  Sorry, but most administrators are
> probably (to use your misapplied word) "paranoid".  Any admin who sets up
> "vi" to run at -10 priority (Heaven help you if it goes into an infinite
> loop, unless your shell is running at better than -10 priority when you type
> "kill" at it), and doesn't *enforce* that it won't give this priority to
> subshells, deserves to have the knowledgable users abuse the hell out of
> such a facility.

This step should be taken because there were only five guys on the system and
we knew each other and the system administrator (myself) used ps to make sure
everyone was being nice.  Not everyone in the world is obnoxious.  We had no
problems of vi going into an infinite loop.  (On rare occasions someone did
a global search on a very large file.)  I am aware that the method is not
perfect BUT all of the users at my site prefered this method to the
alternative of nothing.

As I stated, those requiring more security can modify the shell (or vi), to
do the "nice" automatically or not use this method; what are you complaining
about?  Because you personally don't find this method useful to you?   The
requestor asked for ideas not a guaranteed method that will satisfy all people
in all cases.

Actually, if someone does go into an infinite loop you just edit your .cshrc
to remove the nice and then start a subshell to kill the runaway process.
> -- 
> 	Guy Harris
> 	{ihnp4, decvax, seismo, decwrl, ...}!sun!guy
> 	guy at sun.com (or guy at sun.arpa)

> Subject: Re: vi & heavy loads
> 
> Great. The first thing I'll do when I log in is run vi, then
> escape to a shell. Then my compiles won't take a long time either.
> 
> ---rick

See above comments.

> From: harvard!seismo!c1east!convex!hosking (Doug Hosking)
> Subject: re: vi and heavy loads
> 
> Anyone who WRITES to /dev/mem in this way is playing with a loaded gun.
> It's bad enough that so many utilities go tromping around there to READ it,
> but writing is even worse news.  Various race conditions are likely to
> cause a lot more grief than they're worth.  The renice idea is reasonable,
> and is a LOT less risky.

In the usage that I mensioned, writing to /dev/mem is perfectly safe and there
are no race conditions.  To wit: since I'm writing to *my* proc table entry I
know that I'm not going to die and cause the slot suddenly be used for a
different process.  Also, since I'm not currently doing raw I/O or other such
things I know that the flag bits won't be diddled.  Also it worked!
the f

Bob Toxen
Stratus Computer
{ucbvax!ihnp4,harvard}!anvil!bob        (Please use THIS address to reply)
"panic: can't happen"