From owner-freebsd-multimedia Fri Aug 14 06:31:33 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id GAA26045 for freebsd-multimedia-outgoing; Fri, 14 Aug 1998 06:31:33 -0700 (PDT) (envelope-from owner-freebsd-multimedia@FreeBSD.ORG) Received: from central.picker.com (central.picker.com [144.54.31.2]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id GAA26039 for ; Fri, 14 Aug 1998 06:31:32 -0700 (PDT) (envelope-from rhh@ct.picker.com) Received: from ct.picker.com (ct.picker.com [144.54.37.11]) by central.picker.com (8.9.1/8.9.1) with SMTP id JAA25991; Fri, 14 Aug 1998 09:41:13 -0400 (EDT) Received: from elmer.ct.picker.com ([144.54.57.34]) by ct.picker.com (4.1/SMI-4.1) id AA16997; Fri, 14 Aug 98 09:29:56 EDT Received: by elmer.ct.picker.com (SMI-8.6/SMI-SVR4) id JAA08034; Fri, 14 Aug 1998 09:29:24 -0400 Message-Id: <19980814092923.A6722@ct.picker.com> Date: Fri, 14 Aug 1998 09:29:23 -0400 From: Randall Hopper To: "Louis A. Mamakos" , Roger Hardiman Cc: Scott Mitchell , Amancio Hasty , multimedia@FreeBSD.ORG Subject: Re: FXTV and DGA Mail-Followup-To: "Louis A. Mamakos" , Roger Hardiman , Scott Mitchell , Amancio Hasty , multimedia@FreeBSD.ORG References: <35D2E663.7AF0@cs.strath.ac.uk> <19980813160216.B6924@dcs.qmw.ac.uk> <35D31170.2781@cs.strath.ac.uk> <199808140158.VAA11299@whizzo.transsys.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.91.1i In-Reply-To: <199808140158.VAA11299@whizzo.transsys.com>; from Louis A. Mamakos on Thu, Aug 13, 1998 at 09:58:03PM -0400 Sender: owner-freebsd-multimedia@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Louis A. Mamakos: |> The problem was basically this. |> You need to allocate configuous memory for the frame buffer. |> |> At boot time, this is easy. Memory is empty and you can allocate |> a 864k block of memory with no gaps in it. |> However, I was told that by the time the system has booted, |> and the LKM executed, there is quite possible you cannot get a |> contiguous block of memory as the memory is fragmented. ... |The alternative approach is to allocate some number of smaller (pagesize?) |buffers, and then build a command program for the Bt848 to do a scatter |read into the discountinuous buffers. This might obviate the need for |a static buffer in any case. Good idea. Some extra complexity building the RISC pgm, but a lot of benefit gained. Particularly if we could just alloc the necessary 4K pages as we need one by one on-the-fly. Don't have my bt manual here, but the only thing that comes to mind is splitting scan lines and/or pixel (e.g. packed 24bpp) on these subbuffer boundaries. I can't remember how flexible the bt848 is in dealing with this, though Amancio I'm sure knows. Worth a look though. Think the former is OK, but don't know about the latter. I wish I had the time to dig into it now, but I'm in the middle of a move/househunt/jobchg right now. Any takers? A bt848 LKM would be cool, as would not allocing the driver buffer until we open the device. A plus for BIGGENERIC inclusion. Randall To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-multimedia" in the body of the message