From owner-freebsd-scsi Wed Dec 6 14:17:39 2000 From owner-freebsd-scsi@FreeBSD.ORG Wed Dec 6 14:17:38 2000 Return-Path: Delivered-To: freebsd-scsi@freebsd.org Received: from front1.grolier.fr (front1.grolier.fr [194.158.96.51]) by hub.freebsd.org (Postfix) with ESMTP id 440F837B400 for ; Wed, 6 Dec 2000 14:17:37 -0800 (PST) Received: from nas1-32.cgy.club-internet.fr (nas1-32.cgy.club-internet.fr [195.36.197.32]) by front1.grolier.fr (8.9.3/No_Relay+No_Spam_MGC990224) with ESMTP id XAA15725; Wed, 6 Dec 2000 23:17:22 +0100 (MET) Date: Wed, 6 Dec 2000 22:17:23 +0100 (CET) From: =?ISO-8859-1?Q?G=E9rard_Roudier?= X-Sender: groudier@linux.local To: Dan Langille Cc: David Kelly , freebsd-scsi@FreeBSD.ORG Subject: Re: sym SCSI card problems during settle wait In-Reply-To: <200012062132.KAA20067@ducky.nz.freebsd.org> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Sender: owner-freebsd-scsi@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Thu, 7 Dec 2000, Dan Langille wrote: > On 6 Dec 2000, at 15:30, David Kelly wrote: >=20 > > On Thu, Dec 07, 2000 at 09:22:31AM +1300, Dan Langille wrote: > > > On 6 Dec 2000, at 20:14, G=E9rard Roudier wrote: > > >=20 > > > > You must check that JP1/JP2 are configured for PCI INTA, which is n= ormally > > > > the default setting. > > >=20 > > > I checked it. It's set to INTA. Thanks. Could there be a correspon= ding=20 > > > item in the BIOS? > >=20 > > Dan, you did check that nothing was on IRQ 2 as IRQ 9 is really > > IRQ 2? >=20 > At the suggestion of someone offlist, I've had success with 3.5- > RELEASE boot disks. Now I'm putting the disks back into the box to=20 > do a proper install. Watch this space. In 3.5, it was probably the `ncr' driver and not the `sym' that attached your SCSI card. The `ncr' uses a clock and timely polls the interrupt status register of the PCI-SCSI chip. This was probably intended to reap lost interrupts in early time of broken PCI bridging implementations. But this has also the effect of silently band-aiding chips that have interrupt wiring misconfigured or just broken. The `sym' driver hasn't such an interrupt reaping clock. I donnot want unaware user to run such band-aiding for years instead of having caught and fixed such INT problem on day one. Instead, dummy PCI reads are theorically in proper place in both SCSI scripts and C codes for not losing interrupts in presence of posted transactions (not too broken PCI-HOST bridges assumed). This works so since 5 years under Linux, btw. G=E9rard. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message