*BSD News Article 74142


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!spool.mu.edu!howland.reston.ans.net!gatech!news.mathworks.com!news.kei.com!nntp.coast.net!dispatch.news.demon.net!demon!awfulhak.demon.co.uk!awfulhak.demon.co.uk!awfulhak.demon.co.uk!not-for-mail
From: brian@awfulhak.demon.co.uk (Brian Somers)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: Re: User PPP connection is s......l......o.......w......!
Date: 17 Jul 1996 16:51:17 +0100
Organization: Coverform Ltd.
Lines: 36
Message-ID: <4sj25l$k6@anorak.coverform.lan>
References: <01bb6e37.014cfd20$38673fcb@simonh.addease.com.au> <4sfus4$meq@uriah.heep.sax.de>
NNTP-Posting-Host: localhost.coverform.lan
X-NNTP-Posting-Host: awfulhak.demon.co.uk
X-Newsreader: TIN [version 1.2 PL2]

J Wunsch (j@uriah.heep.sax.de) wrote:
: brian@awfulhak.demon.co.uk (Brian Somers) wrote:

: > Does this mean that FreeBSD allows the interrupt handler to get
: > preempted ?  If so, I'm *very* impressed.  Or is it a FreeBSD
: > standard that says that interrupt handlers need as many preemption
: > points as possible ?

: Why "preemption points"?

: If i read Bruce's posting right, they are being preempted without any
: influence by the driver writer.  Of course, they run at their IPL, so
: they will only be preempted by higher priorized interrupts.
: (Actually, interrupts of another class.)

As I said, if the interrupt handler can get preempted - this is brilliant.

I wondered if he *meant* calling something like preempt() (or whatever)
every now and again to allow higher priorities in.  If interrupt
handlers can be preempted by higher level IPL routines (by swtch),
this implies (I think) that FreeBSD has to somehow lock any relevent
data structures (ones shared by interrupts of a different IPL) in a given
interrupt routine.  Surely this is an overhead on a uniprocessor
machine....

I realize there is flaw in the previous paragraph - I'm fishing for what
that flaw is.....  Is it the case that (say) there are no such occurences
- ie. a given kernel data structure "belongs" to one and only one IPL.

BTW, I *am* reading about this - Unix Internals - I just havn't got the
-current sources (and havn't looked for this sort of stuff in my latest
sources (960323-SNAP).

--
Brian <brian@awfulhak.demon.co.uk>
Don't _EVER_ lose your sense of humour....