From owner-freebsd-questions Wed Feb 28 22:17:59 1996 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id WAA02718 for questions-outgoing; Wed, 28 Feb 1996 22:17:59 -0800 (PST) Received: from pelican.com (pelican.com [206.16.90.21]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id WAA02709 for ; Wed, 28 Feb 1996 22:17:56 -0800 (PST) Received: by pelican.com (Smail3.1.29.1 #10) id m0ts1gL-0000SMC; Wed, 28 Feb 96 22:17 PST Message-Id: Date: Wed, 28 Feb 96 22:17 PST From: pete@pelican.com (Pete Carah) To: sreid@edmbbs.iceonline.com Subject: Re: X Windows crashes big-time In-Reply-To: <9602260503.D8475on@edmbbs.iceonline.com> Cc: questions@freebsd.org Sender: owner-questions@freebsd.org Precedence: bulk In article <9602260503.D8475on@edmbbs.iceonline.com> you write: >I have a problem with X Windows crashing on a FreeBSD 2.1 machine. It's >not just a little crash- the whole machine freezes, and we have to use >the reset button. Ctrl-Alt-Backspace and Ctrl-Alt-Delete don't have any >effect. >I type startx to go into a freshly installed Xwindows, and it works for >a few seconds. I can move the mouse around, but before I get a chance >to do anything, the machine dies. >The system is a Pentium-100, Asus Triton motherboard. 16 megs of 60ns >EDO ram. 256k burst pipeline cache. 1GB Quantum HD on an Adaptec 2940 >SCSI-2. Video card is Diamond Stealth 64 (2MB VRAM) on PCI, using >640x480x256 mode. I'm using the S3 server, and I selected the Diamond >Stealth Pro as the video card in the config (the vid card manual says >the 64 is backwards-compatible with the Pro). No clocks line or RAMDAC >selected. The mouse is a Logitech Mouseman, and I've selected Mouseman >in the config file. Also have an SMC EtherPower 10/100 in 10Mbps mode. Mine lasted longer; it crashed hard whenever I loaded very many pictures (pixmaps in general). I could surf the web for an average of 2 sites before it crashed :-( The solution for me was a combination; sup the 2.1-STABLE sys tree (may not be needed but DG said there were a few kernel-stack handling fixes), and *also* get the latest S3 server from ftp.xfree86.org. (it should be 3.1.2B-Beta or later; the A-Beta one didn't fix the problem at least for me. A more minor version of this happens to a friend on a 486 with a VLB S3 card (fahrenheit 1280+ VLB with 2mb ram) but he only crashes every other day or so. There is another later server now; I don't know much about either of these, but on ftp.xfree86.org: ftp> pwd 257 "/pub/XFree86/3.1.2D/FreeBSD-2.0.5" is current directory. ftp> dir *S3* 200 PORT command successful. 150 Opening ASCII mode data connection for /bin/ls. -rw-r--r-- 1 701 1190 1206787 Feb 25 07:31 X312D9NS3.tgz -rw-r--r-- 1 701 1190 1199083 Feb 25 09:51 X312DS3.tgz 226 Transfer complete. The X312BS3.tgz worked fine for me (in another directory, of course...) I don't know what a 9NS3 is either. Part of the 312B release notes say: ----- o Improved linear fb detection for PCI cards. In particular this should allow most Trio64V+ cards to use the linear fb, which seems to prevent a lot of the lockups that have been reported. ------ Though my old non-direct-mapped S3 card had lockups too, so I don't know if this is the whole story. BTW it also happened to me using an ISA S3 card (Fahrenheit 1280+) so the bug, whatever it was, didn't depend on the huge memory map or the Trio chip of the Stealth 64. I ended up changeing out my cache, ram, and the video card all to no avail :-( At least static cache is getting cheap... (and I ended up using it in the W95 machine I built up for my daughter) (I'm also using an ASUS Triton (XE) motherboard with a P100, back to pipeline burst cache, 16mb of non-edo ram. 16mb is shy of enough to run the accel server and netscape; it ends up swapping a bunch with nothing else but a few xterms :-( -- Pete