From owner-freebsd-stable Fri Apr 27 22:17: 3 2001 Delivered-To: freebsd-stable@freebsd.org Received: from wyattearp.stanford.edu (wyattearp.Stanford.EDU [171.64.180.171]) by hub.freebsd.org (Postfix) with ESMTP id 3707C37B423 for ; Fri, 27 Apr 2001 22:17:00 -0700 (PDT) (envelope-from richw@wyattearp.stanford.edu) Received: (from richw@localhost) by wyattearp.stanford.edu (8.9.3/8.9.3) id WAA75566; Fri, 27 Apr 2001 22:16:11 -0700 (PDT) (envelope-from richw) Date: Fri, 27 Apr 2001 22:16:11 -0700 (PDT) From: Rich Wales X-Sender: richw@wyattearp.stanford.edu To: freebsd-stable@freebsd.org Subject: Followup on silo overflows In-Reply-To: <20010304065203.30478.richw@wyattearp.stanford.edu> Message-ID: <20010428045358.75024.richw@wyattearp.stanford.edu> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Some time ago, I reported "silo overflow" problems when directing high-speed serial input to an "xterm" window. I'm running 4.2-RELEASE and XFree86-4.0.1. After some other people pointed out a connection between this problem and the use of XFree86 4.x (as opposed to 3.3.6), I brought up the issue on the "xpert@xfree86.org" list, where someone suggested that some of the acceleration routines might be overly aggressive -- pipe- lining commands to the video card in such a way that the PCI bus got locked up, and thereby delaying serial interrupt handling. I experimented with the undocumented "XaaNo..." options to the video accelerator code, and found that the silo overflows went away if I included either one of the following two lines in the "Device" section of my /etc/X11/XF86Config file: Option "XaaNoSolidFillRect" "on" Option "XaaNoScanlineCPUToScreenColorExpandFill" "on" (Note: I didn't need to use both of the above options -- just one was enough in my case.) I'm using an ATI Xpert 128 video card, which is supported in XFree86 4.x via the "r128" driver. I suppose it's possible that a system with a different card might require a different "XaaNo..." option -- or that none of the "XaaNo..." options will get rid of the silo overflows for some cards. I've updated PR 26261 accordingly. I don't propose to close the PR yet, because there still might be a kernel bug that is allowing the bus to get locked up by the video driver. Rich Wales richw@webcom.com http://www.webcom.com/richw/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message