From mboxrd@z Thu Jan  1 00:00:00 1970
Date: Tue,  7 Dec 2004 10:37:55 -0800
From: Micah Stetson <micah@cnm-vra.com>
To: Fans of the OS Plan 9 from Bell Labs <9fans@cse.psu.edu>
Subject: Re: [9fans] plan9ports to NetBSD
Message-ID: <20041207183755.GA18071@epaphras>
References: <038021ee700f4c13187b3e750f329d99@proxima.alt.za>
	<20041207182242.GB21003@mero.morphisms.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20041207182242.GB21003@mero.morphisms.net>
User-Agent: Mutt/1.4.1i
Topicbox-Message-UUID: 122d76be-eace-11e9-9e20-41e7f4b1d025

> > I can get most of the way there, but
> > 	<sched.h>
> > and
> > 	sched_yield()
> > concern me.  The former seems quite unrelated to the 9ports altogether
> > and the latter does not appear in NetBSD (1.6.1, nearly up to date) at
> > all.
> 
> What do you mean by ``unrelated''?  As I recall it is there so that
> the lock routines can yield the processor.  I expect that sleep() is
> sufficient, but it is there for a reason.

I don't know if this will help, but in attempting an
Inferno port to NetBSD a few months ago, replacing
sched_yield() with sleep(0) did weird things to interactive
performance.  I don't recall the exact symptoms, but I
remember it made the system (at least the GUI) a pain to
use.  At the time, I asked about alternatives on NetBSD's
tech-kern mailing list and was told that sched_yield()
was implemented in -current.

http://mail-index.netbsd.org/tech-kern/2004/03/10/0019.html

After that, my free time disappeared, so I haven't played
with it since.

Micah