From owner-freebsd-hackers@FreeBSD.ORG  Thu Oct  5 00:29:40 2006
Return-Path: <owner-freebsd-hackers@FreeBSD.ORG>
X-Original-To: freebsd-hackers@freebsd.org
Delivered-To: freebsd-hackers@freebsd.org
Received: from localhost.my.domain (localhost [127.0.0.1])
	by hub.freebsd.org (Postfix) with ESMTP id 741E316A407;
	Thu,  5 Oct 2006 00:29:40 +0000 (UTC)
	(envelope-from davidxu@freebsd.org)
From: David Xu <davidxu@freebsd.org>
To: freebsd-hackers@freebsd.org
Date: Thu, 5 Oct 2006 08:29:36 +0800
User-Agent: KMail/1.8.2
References: <20061004214040.GA6878@porthos.spock.org>
In-Reply-To: <20061004214040.GA6878@porthos.spock.org>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200610050829.36266.davidxu@freebsd.org>
Cc: Jonathan Chen <jon@freebsd.org>
Subject: Re: running code on all CPUs
X-BeenThere: freebsd-hackers@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Technical Discussions relating to FreeBSD
	<freebsd-hackers.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-hackers>, 
	<mailto:freebsd-hackers-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-hackers>
List-Post: <mailto:freebsd-hackers@freebsd.org>
List-Help: <mailto:freebsd-hackers-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-hackers>,
	<mailto:freebsd-hackers-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Thu, 05 Oct 2006 00:29:40 -0000

On Thursday 05 October 2006 05:40, Jonathan Chen wrote:
> Hello,
>
> I'm currently writing kernel code to use MCA (machine check architecture)
> in order to detect things like ECC errors.  As part of this, I need to run
> code on all CPUs periodically to check some status registers.  Where is the
> best place to do this?  It doesn't seem that I can use regular kernel
> threads since I can't specify a CPU binding with that interface.  I've
> thought about hooking a function call to either the beginning of
> idle_proc() or somewhere in mi_switch(), but neither solution seems optimal
> since there are no guarantees when idle_proc() is ran and mi_switch() seems
> like a really bad idea from a performance standpoint.  Suggestions?
>
> Also, are there any locking pitfalls or other issues I should be aware of
> when writing code to run in either idle_proc() or mi_switch()?
>
> -Jon

AFAIK, you can use scheduler API, the sched_bind() moves current thread to
a specific CPU, hope this helps.

David Xu