From owner-freebsd-current Wed Jun 13 23:50:25 2001 Delivered-To: freebsd-current@freebsd.org Received: from zibbi.icomtek.csir.co.za (zibbi.icomtek.csir.co.za [146.64.24.58]) by hub.freebsd.org (Postfix) with ESMTP id 6379C37B405 for ; Wed, 13 Jun 2001 23:49:50 -0700 (PDT) (envelope-from jhay@zibbi.icomtek.csir.co.za) Received: (from jhay@localhost) by zibbi.icomtek.csir.co.za (8.11.1/8.11.1) id f5E6mhZ52549; Thu, 14 Jun 2001 08:48:43 +0200 (SAT) (envelope-from jhay) From: John Hay Message-Id: <200106140648.f5E6mhZ52549@zibbi.icomtek.csir.co.za> Subject: Re: UDMA interfering with install In-Reply-To: <3B28590F.E8930BB7@mindspring.com> from Terry Lambert at "Jun 13, 2001 11:26:23 pm" To: tlambert2@mindspring.com Date: Thu, 14 Jun 2001 08:48:43 +0200 (SAT) Cc: current@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL54 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG > > > > That's good enough. :) Thanks.... > > > > Maybe _that_ will keep that ata code from over-riding > > the bios to disable dma (or maybe the bios just wasn't > > doing it's job right ;) > > > This won't work. What do you mean with this? The procedure that I described (barring typos) do work here and was used here to install and run FreeBSD on a silly A+ motherboard. Without disabling the DMA the install would fail and even if I installed with DMA disabled, but rebooted afterwards with DMA enabled, it would corrupt the filesystem to an almost unusable state. > > Someone was having the same problem the other day, and > I suggested the same soloution, but after probe, the > damn driver enabled UDMA at attach time anyway. > > So we removed it from the kernel config... and the damn > thing enabled it again. > > I don't know if the #ifdef was intended to only guard > in the boot case, but it doesn't help, because there > are several missign guards around the code, if that's > the case, and at least four places in the code ignore > the tuning variable, as well, if it isn't commented > out of the kernel at build time (thus disabling one of > the places). > > Look for the #ifdef, and then look for the function > call to do the enable, and the problem will be obvious. I'm not sure where the #ifdef comes into play. I didn't even recompile anything, so whatever #ifdef can be whatever it likes to be. Jun 5 18:42:51 d-5-71 /boot/kernel/kernel: ad0: 4104MB [8896/15/63] at ata0-master PIO4 John -- John Hay -- John.Hay@icomtek.csir.co.za To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message