From owner-freebsd-current@FreeBSD.ORG Thu Jan 8 17:39:15 2009 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 80CD110658F9 for ; Thu, 8 Jan 2009 17:39:15 +0000 (UTC) (envelope-from sam@freebsd.org) Received: from ebb.errno.com (ebb.errno.com [69.12.149.25]) by mx1.freebsd.org (Postfix) with ESMTP id 0DE258FC1F for ; Thu, 8 Jan 2009 17:39:14 +0000 (UTC) (envelope-from sam@freebsd.org) Received: from trouble.errno.com (trouble.errno.com [10.0.0.248]) (authenticated bits=0) by ebb.errno.com (8.13.6/8.12.6) with ESMTP id n08HdE10071513 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Thu, 8 Jan 2009 09:39:14 -0800 (PST) (envelope-from sam@freebsd.org) Message-ID: <49663A42.7010101@freebsd.org> Date: Thu, 08 Jan 2009 09:39:14 -0800 From: Sam Leffler Organization: FreeBSD Project User-Agent: Thunderbird 2.0.0.18 (X11/20081209) MIME-Version: 1.0 To: Garrett Cooper References: <7d6fde3d0901072349i50d04d2ch34f308388c9fd414@mail.gmail.com> In-Reply-To: <7d6fde3d0901072349i50d04d2ch34f308388c9fd414@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-DCC-Rhyolite-Metrics: ebb.errno.com; whitelist Cc: FreeBSD Current Subject: Re: X becomes unresponsive with nvidia / xscreensaver and desktop panics X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 08 Jan 2009 17:39:19 -0000 Garrett Cooper wrote: > Hello folks, > As many probably know, I recently installed FreeBSD 8-CURRENT on my desktop. > One of the things I'm noting is that when I use dual displays with > the nvidia driver, and xscreensaver kicks in and keeps going for a > period of time, the machine's X.org console eats up a core, and when I > try and do anything like gdb Xorg, the system hangs, attempts to panic > (I assume that's the case because I hear it beep and attempt to > restart), then I have to give it a warm boot. truss(1)'ing Xorg showed > that there were a lot of SIGALARM's being fired and masked. > This is incredibly reminiscent of the days when I used to run Linux and > I forgot to configure my dump device (doh), but this problem appears > to be easily reproducible with my environment. I'm fixing that issue > right now, so I should have a working crashdump soon... > Just wondering if anyone else has seen this issue, or if it's just me > once again. > I have been battling with the nvidia binary driver on HEAD for a long time. I've got a Quadro NVS card: vgapci0@pci0:1:0:0: class=0x030000 card=0x019010de chip=0x018a10de rev=0xa2 hdr=0x00 vendor = 'Nvidia Corp' device = 'Quadro NVS with AGP 8x [NV18GL.2]' class = display subclass = VGA and run the older port because the card support was yanked at some point: nvidia-driver-96.43.07 NVidia graphics card binary drivers for hardware OpenGL ren I run dual-dvi displays w/ xinerama. I have not seen panics but I do see unacceptable performance. The most obvious problems are impossible mouse and rendering lag. I also see large memory consumption when running something like firefox w/ lots of hidden/off-screen images in a page (e.g. p4web). I am now 100% certain the nvidia driver is at fault as I swapped in an ATI card over the holiday break and the change was like night and day. Unfortunately the xrandr support for the ATI card was too busted to use so I gave up and stuck the nvidia card back in the box (for those that care the card is a Radeon 9600 and the problem is that after a while one display would blank randomly). This is a dual-core AMD (i386) desktop but performance actually dropped when I went from UP to SMP which makes me believe there's something braindead in the nvidia driver. The problems have only gotten worse as FreeBSD has evolved. I've not replaced the video card because I need dual-dvi in an AGP format which is rare these days. Moving to PCIE means a new mb/box which I'm trying to avoid. Sam