Path: utzoo!mnetor!uunet!husc6!purdue!umd5!mimsy!chris
From: chris@mimsy.UUCP (Chris Torek)
Newsgroups: comp.os.minix
Subject: Re: Minix Scheduler (and tty performance)
Message-ID: <10721@mimsy.UUCP>
Date: 19 Mar 88 18:22:08 GMT
References: <1002@rtmvax.UUCP> <1432@runx.ips.oz>
Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742
Lines: 15

In article <1432@runx.ips.oz> brucee@runx.ips.oz (Bruce Evans) writes:
>(4) Changed system calls and interrupts to save the state on the user stacks
>    instead of in the process table.

Saving syscall state in the user stack is arguably correct (although
4BSD Unix, at least, leaves it in a per-process kernel stack).  Saving
interrupt state on the user stack, however, is `wrong'.  coroutines
(among others) may want to leave stack stuff undisturbed at times.

One reason not to use user stacks for syscalls is the old `what if you
run out of space' argument.  I have no idea whether Minix tries to
handle this, but it is something worth considering.
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163)
Domain:	chris@mimsy.umd.edu	Path:	uunet!mimsy!chris