From owner-freebsd-questions Thu Jun 26 08:30:05 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id IAA26155 for questions-outgoing; Thu, 26 Jun 1997 08:30:05 -0700 (PDT) Received: from amaretto.csv.warwick.ac.uk (amaretto.csv.warwick.ac.uk [137.205.148.138]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id IAA26106 for ; Thu, 26 Jun 1997 08:29:49 -0700 (PDT) From: Mr M P Searle Message-Id: <6699.199706261525@amaretto.csv.warwick.ac.uk> Received: by amaretto.csv.warwick.ac.uk id QAA06699; Thu, 26 Jun 1997 16:25:49 +0100 (BST) Subject: Re: FreeBSD Info In-Reply-To: from 0000-Administrator at "Jun 25, 97 07:50:36 pm" To: root@counterintelligence.cdrom.com (0000-Administrator) Date: Thu, 26 Jun 1997 16:25:45 +0100 (BST) Cc: questions@freebsd.org X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-questions@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > > Is there a FreeBSD system call or better yet a device that can be > opened > (/dev/io just allows the process to directly do io) and read/write to the > io ports. I need to port some software from linux which to my own disgust > uses inb and outb macros (which are defined in some standard .h include > file there) for controlling a analog io board that has no driver. After I > used the io ports I found out that there was a /dev/port device that can > be opened and read/write to a the file pointer which corresponds to an io > port, anyway with the exception of writing a kernel driver (i really don't > have the patience for that what can I do) also is there some kind of > documentation I can get on ther kernel - particularily it looks like (from > calling usleep(1) in a loop that the system timer runs 50-100 ticks > /second I want to increase this to like 1000-4000 if that is safe (and > won't annoyingly screw up the date/time) > Well, I've increased the system timer from 100 to 1000 Hz (by including options HZ=1000 in the config file) - there's no measurable increase in overhead time on a Pentium but it does mess some things up. (top and similar utilities measuring CPU time etc - not the system clock itself.) It may be that recompiling these utilities with HZ=1000 solves the problem, I'd not tried anything like this. Of course, YMMV as I've not actually looked at exactly what HZ does in the kernel.