From owner-freebsd-mobile Mon Feb 21 1:39: 1 2000 Delivered-To: freebsd-mobile@freebsd.org Received: from mailgate.Cadence.COM (mailgate.Cadence.COM [158.140.2.1]) by hub.freebsd.org (Postfix) with ESMTP id 2B90337BC0B for ; Mon, 21 Feb 2000 01:38:59 -0800 (PST) (envelope-from dmlb@ragnet.demon.co.uk) Received: from symnt3.Cadence.COM (symnt3.Cadence.COM [194.32.101.100]) by mailgate.Cadence.COM (8.9.3/8.9.3) with ESMTP id BAA02838; Mon, 21 Feb 2000 01:38:49 -0800 (PST) Message-Id: <200002210938.BAA02838@mailgate.Cadence.COM> Received: from pc287-cam.cadence.com ([194.32.97.124]) by symnt3.Cadence.COM with SMTP (Microsoft Exchange Internet Mail Service Version 5.5.2650.21) id FA9FSHP5; Mon, 21 Feb 2000 09:38:28 -0000 Comments: Authenticated sender is From: "Duncan Barclay" To: Warner Losh , freebsd-mobile@FreeBSD.ORG, Scott Mitchell Date: Mon, 21 Feb 2000 09:38:24 +0000 MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Subject: Re: Raylink driver/Mapping attribute memory Reply-To: dmlb@ragnet.demon.co.uk Cc: Duncan Barclay In-reply-to: <200002210806.BAA19218@harmony.village.org> References: Your message of "Mon, 21 Feb 2000 08:00:41 GMT." X-mailer: Pegasus Mail for Win32 (v2.53/R1) X-Received: By mailgate.Cadence.COM as BAA02838 at Mon Feb 21 01:38:49 2000 Sender: owner-freebsd-mobile@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > In message Duncan > Barclay writes: : Can I urge you provide a mechanism to enable > mappng of both attribute and : common memory - i.e. at least two > mapped regions per card. The Raylink : needs this - it has the 4k of > attribute space in which a small (16bytes) : register set appears at > an offset of 0x0F00, and a 48k common memory section for : tx/rx > buffers and large structure based control. > > Yes. You should be able to map both attr and common memory at the > same time. However, there are limits in the hardware. Since the > memory doesn't need to be wirte for all time I may do a pass through > allocation. In the raylink case the atttribute memory is written to when the driver sends a command to the card. The basic idea is that the card or driver fills out a command structure in the common memory and then interrupts the driver or card. To interrupt the card, a bit is set in the attribute memory. This bit is cleared by the card when it has completed the comand. To interrupt the driver the card uses an IRQ and then monitors a completion bit in the attribute memory. This is written to by the driver when it has completed the action. I wouldn't want to have to keep changing the protection on the attribute memory. Hmm, just thought. Is the attribute memory mapped as R/O by pccardd in 3.x? This may explain a funny I saw late last night. I will check later today. > Warner Duncan To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-mobile" in the body of the message