From owner-freebsd-stable Wed Sep 20 13:45:55 2000 Delivered-To: freebsd-stable@freebsd.org Received: from pcnet1.pcnet.com (pcnet1.pcnet.com [204.213.232.3]) by hub.freebsd.org (Postfix) with ESMTP id D149C37B424 for ; Wed, 20 Sep 2000 13:45:50 -0700 (PDT) Received: (from eischen@localhost) by pcnet1.pcnet.com (8.8.7/PCNet) id QAA21754; Wed, 20 Sep 2000 16:45:18 -0400 (EDT) Date: Wed, 20 Sep 2000 16:45:08 -0400 (EDT) From: Daniel Eischen To: "Justin T. Gibbs" Cc: Andy@silverbrook.com.au, Brad Knowles , stable@FreeBSD.ORG, "Brandon D. Valentine" Subject: Re: MFC of ahc driver updates (long-ish) In-Reply-To: <200009200428.WAA88102@pluto.plutotech.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Tue, 19 Sep 2000, Justin T. Gibbs wrote: > >Brandon, I'll take those patches if you've still got them lying > >around. If not, I'll try to make them myself or try to get the > >latest -current aic7xxx driver to work. > > Testing the stuff from -current would be the best option. There > are known issues for non-U2/U3 controllers in the code that Brandon > first proposed to have merged. OK, I got the -current driver integrated into -stable. The "ahc0: spurious interrupt..." messages that I was getting constantly in 4.1-stable were from option AHC_ALLOW_MEMIO in the kernel config file. I guess there's something strange about this motherboard/controller combination that doesn't let it do memory mapped IO. Removing this option makes the -stable driver bootable. As for the -current driver transplanted into -stable, this option has gone away and the check to see if memory mapped IO works evidently doesn't do the trick because I still get continuous "spurious interrupt" messages that prevent the system from booting. I added an #ifdef AHC_ALLOW_MEMIO around that section in ahc_pci.c and can boot the system. I also get the following messages (from the -current driver in -stable): ahc1:A:3: Message reject for 80 -- ignored These repeat pretty fast while the machine boots and also for a while after it has booted. There seem to be about 255 of them (dmesg | grep "Message reject" | wc -l). In this system, there is nothing (not even a cable) attached to ahc1. Here are the 2 controllers: pcib1: on motherboard pci1: on pcib1 ahc0: port 0xe400-0xe4ff mem 0xfebfe000-0xfebfefff irq 5 at device 2.0 on pci1 aic7892: Wide Channel A, SCSI Id=7, 32/255 SCBs ahc1: port 0xe800-0xe8ff mem 0xfebff000-0xfebfffff irq 10 at device 3.0 on pci1 aic7892: Wide Channel A, SCSI Id=7, 32/255 SCBs I'll see how the system holds up under load and get back to you if I can crash it. If you have anything else you ant to try in regards to memory-mapped IO or the message reject messages, let me know. -- Dan Eischen To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message