From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Wed, 1 Aug 2001 13:51:22 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Wed, 1 Aug 2001 13:51:12 -0400 Received: from tisch.mail.mindspring.net ([207.69.200.157]:38707 "EHLO tisch.mail.mindspring.net") by vger.kernel.org with ESMTP id ; Wed, 1 Aug 2001 13:51:01 -0400 From: sduchene@mindspring.com Date: Wed, 1 Aug 2001 13:51:04 -0400 To: linux-kernel@vger.kernel.org Subject: Patch for Riva framebuffer driver Message-ID: <20010801135104.J10061@lapsony.mydomain.here> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.3.5i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org I sent the following E-mail to Ani Joshi but have never received a reply from him so I thought I would forward this info onto the linux-kernel list so this fix for a repeatable kernel Oops could be incorperated in future kernels. ----- Forwarded message from valsad ----- From: valsad Date: Sun, 15 Jul 2001 22:25:19 -0400 To: Ani Joshi Subject: Patch for Riva framebuffer driver User-Agent: Mutt/1.3.5i Ani: I'm not sure if you are the current maintainer of the Riva framebuffer driver however you are listed in the drivers/video/riva/fbdev.c file as the author. I have discovered a small problem with the code in the fbdev.c file that causes a kernel oops everytime. I was informed that the Riva 128 chip does not really support 16bpp and that I should be using 15 bpp instead however when I configure my video line in the lilo.conf file as follows: append="video=riva:1280x1024-15@74" I got an kernel oops and an assertion failed message from line 1120 in fbdev.c By adding the following code to fbdev.c everything works just fine now. *** linux/drivers/video/riva/fbdev.c Mon Jul 16 01:17:39 2001 --- linux_sad/drivers/video/riva/fbdev.c Mon Jul 16 01:17:12 2001 *************** *** 1107,1112 **** --- 1107,1115 ---- break; #endif #ifdef FBCON_HAS_CFB16 + case 15: + rc = 15; /* fix for 15 bpp depths on Riva 128 based cards */ + break; case 16: rc = 16; /* directcolor... 16 entries SW palette */ break; /* Mystique: truecolor, 16 entries SW palette, HW palette hardwired into 1:1 mapping */ BTW, the following is an exerpt from an E-mail conversation between myself and Bakonyi Ferenc regarding my problems with 16 bpp: > > The fbset output on this system says: > > > > mode "1280x1024-74" > > # D: 135.007 MHz, H: 78.859 kHz, V: 74.116 Hz > > geometry 1280 1024 1280 1024 16 > > timings 7407 256 32 34 3 144 3 > > accel true > > rgba 5/11,6/5,5/0,0/0 > ^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > endmode > > This line indicates you are using rivafb in 16 bpp mode. Unfortunetly > 16 bpp is not supported by Riva 128 hw. It's a bug, in a perfect > world rivafb should disable 16 bpp mode on Riva 128. In this case > hardware is set to 15 bpp, but rivafb thinks it's set to 16 bpp. (Or > something similar.) ----- End forwarded message ----- Also after implementing this change fbset now reports: mode "1280x1024-74" # D: 135.007 MHz, H: 78.859 kHz, V: 74.116 Hz geometry 1280 1024 1280 1024 16 timings 7407 256 32 34 3 144 3 accel true rgba 5/10,5/5,5/0,0/0 endmode -- Steven A. DuChene sad@ale.org linux-clusters@mindspring.com sduchene@mindspring.com http://www.mindspring.com/~sduchene/