From owner-freebsd-mobile Tue Jan 18 13:14:40 2000 Delivered-To: freebsd-mobile@freebsd.org Received: from mass.cdrom.com (mass.cdrom.com [204.216.28.184]) by hub.freebsd.org (Postfix) with ESMTP id CDFB015066; Tue, 18 Jan 2000 13:14:31 -0800 (PST) (envelope-from msmith@mass.cdrom.com) Received: from mass.cdrom.com (localhost [127.0.0.1]) by mass.cdrom.com (8.9.3/8.9.3) with ESMTP id NAA01429; Tue, 18 Jan 2000 13:22:21 -0800 (PST) (envelope-from msmith@mass.cdrom.com) Message-Id: <200001182122.NAA01429@mass.cdrom.com> X-Mailer: exmh version 2.1.1 10/15/1999 To: Warner Losh Cc: Mike Smith , Tatsumi Hosokawa , mobile@FreeBSD.ORG Subject: Re: Working pcic polling mode patch (Re: Polling mode of pcic as default?) In-reply-to: Your message of "Mon, 17 Jan 2000 22:42:27 MST." <200001180542.WAA14505@harmony.village.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Tue, 18 Jan 2000 13:22:21 -0800 From: Mike Smith Sender: owner-freebsd-mobile@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > In message <200001170922.BAA02947@mass.cdrom.com> Mike Smith writes: > : Definitely not. The IRQ should be determined by examining the machine's > : actual resource configuration, with a manual override from the loader. > : Of all the possible sources of IRQ configuration information, the kernel > : configuration file is the most likely to be wrong. > > Why is pcic different than every other device in the entire system? > If I want to change the IRQ for a sio line, I do it in userconfig, not > some whacked out kernel loader env variable. Why should PCIC be > different? Er, because it _is_ different. With sio devices, you have a set of hardcoded configurations that are 'typical', and PnP information (which we mishandle in the case when it matches the hardcoded expectations). The pcic is a totally different animal in some respects; in particular, its IRQ needs to be selected based on what's actually available in the system; something that cannot possibly be determined correctly in a pre-built kernel. Using an environment variable is just a placeholder for userconfig becoming a real hint editor; making the pcic IRQ settable inside userconfig would require a heap more work than anyone wants to spend. > Also, this rusian roulette of picking an IRQ is unwise. It breaks too > many times. You must hardwire the IRQ, or use polling. Examining the > machine is too unreliable to work. Er, hardwiring is just persistent state based on examining the machine. This is an argument for a polling-only approach. -- \\ Give a man a fish, and you feed him for a day. \\ Mike Smith \\ Tell him he should learn how to fish himself, \\ msmith@freebsd.org \\ and he'll hate you for a lifetime. \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-mobile" in the body of the message