From owner-freebsd-questions Sat Dec 30 09:39:18 1995 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id JAA26420 for questions-outgoing; Sat, 30 Dec 1995 09:39:18 -0800 (PST) Received: from serv05.slac.stanford.edu (SERV05.SLAC.Stanford.EDU [134.79.16.135]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id JAA26414 for ; Sat, 30 Dec 1995 09:39:11 -0800 (PST) From: tomdean@MAILBOX.SLAC.Stanford.EDU Received: from mailbox.SLAC.Stanford.EDU by SERV05.SLAC.STANFORD.EDU (PMDF V5.0-4 #10979) id <01HZEWW436Z8000RVU@SERV05.SLAC.STANFORD.EDU> for questions@FreeBSD.org; Sat, 30 Dec 1995 09:42:37 -0800 (PST) Received: from hebe.SLAC.Stanford.EDU (HEBE.SLAC.Stanford.EDU [134.79.16.24]) by mailbox.SLAC.Stanford.EDU (8.6.11/8.6.11) with SMTP id JAA22735 for ; Sat, 30 Dec 1995 09:39:03 -0800 Received: by hebe.SLAC.Stanford.EDU (AIX 3.2/UCB 5.64/SLAC 920508) id AA97543; Sat, 30 Dec 1995 17:39:03 +0000 (GMT) Date: Sat, 30 Dec 1995 17:39:03 +0000 (GMT) To: questions@FreeBSD.org Message-id: <9512301739.AA97543@MAILBOX.SLAC.Stanford.EDU> X-Envelope-to: questions@FreeBSD.org Content-transfer-encoding: 7BIT Sender: owner-questions@FreeBSD.org Precedence: bulk I have two computer systems that I want to use for FreeBSD. One is a TI 4000E notebook. The other is a 486DX2/66. I encountered some problems in installing FreeBSD on both systems. First, the notebook. I have been running FreeBSD for a couple of years on this system. 1.5.1 worked ok. I installed 2.0.5, with some difficulty in accessing the CDROM over the network on a WINNT system (yeeeeeeeetch!). However, after some playing, this worked. When I upgraded to 2.1, there was one small problem. Installing from the Walnut Creek CDROM, over the network, requires both the boot and the root floppies. This is because the path to the dists must be supplied in the URL: ftp:///dists. The kernel is at the root level on the CDROM and the install cannot find it. Creating the root floppy solved the problem. The need to do this was not clear from the documentation. Second, the uuuuuuugly system. The 486DX2 system is several years old and has several needs. It has WINNT 3.51 installed on it. It is necessary to keep this for a year, or so, to support some applications. This system has some nice, but old, hardware, other than the CPU/MB. This system has grown and been upgraded several times. The latest being the mother board after a failure... It was decided for some obscure reason to get an EISA board, possibly with the idea of using EISA peripheral cards in the future. There are several WINNT applications that need to be supported for the next year, or, so. The DOS disk is used as a back-up medium, as 'NT does not support the Irwin drive... However, some time back, FreeBSD was on this machine and it supported the drive. This was before the SCSI interface and the EISA Motherboard. Nice EISA MB 20MB RAM Intel EtherExpress network adapter Adaptec 1542 SCSI adapter Quantum Empire 1080S 0 WINNT Maxtor LXT 340S 340MB disk 1 DOS Toshiba XM-3401TA 2 CDROM Fujitsu M1606S 1GB disk 3 2111/16/63 (new) I intended to install FreeBSD 2.1 on the Fujitsu disk, SCSI ID 3. I have not been successful in getting the drive to boot. In both of the first two attempts, I selected to use all of the disk, made the partition active (with "s"), and used the auto division. I had to enter the geometry as 2111/16/63. For some reason, when I attempted the second installation, the install process thought the disk was 2115/16/63. However, I labeled the disk as (sd2 listed for sd0, the SCSI ID was the only diff): Offset Size End Name PType Desc SubType Flag 0 63 62 - 6 unused 0 63 2131857 2131919 sd2s1 3 freebsd 165 CA 2131920 72 2131991 - 6 unused 0 This resulted in: Part Mount Size NewFS sd2s1a / 32MB UFS Y sd2s1b 51MB Swap sd2s1e /var 30MB UFS Y sd2s1f /usr 927MB UFS U First, I tried the drive at SCSI ID 3. I installed os-bs135 on the Quantum disk, SCSI ID 0. After installing FreeBSD 2.1 on the disk, os-bs could not see it. I could boot from the floppy and at the boot prompt, enter sd(2,a)/kernel and get the system to boot. Then, I tried SCSI ID 0. I removed the Quantum disk. I installed FreeBSD and the boot manager. At the boot manager prompt, I seemed to hang at the F? prompt. The boot manager did see the dos disk, but, it is not bootable. The FAQ and other docs indicate that the F? problem is most likely a disk geometry problem. I checked the geometry with the manufacturer (actually sales - would they lie?) The brochure that came with the disk indicates that the geometry is 2111/16/63. NCA confirmed this. I verified that the AHA 1542 does not have the translation enabled - it does not. I tried again at SCSI ID 0. This time, I created a DOS partition on the disk and used the numbers install provided - 1041/64/32. This differs from the manufacturer's information of 2111/16/63. However, after the install completed, I could boot the drive. However, after booting with the -c switch and disabling unused devices, I still have a failure. After the fail message, the system goes into a look across the boot. The system crashed, then wanted to reboot. Again, I booted with the -c switch and after cleaning the file system, it encountered the same error: Fatal trap 12: page fault while in kernel mode. fault virtual address: = 0xef0dff80 fault code = supervisor write, page not present instruction pointer = 0x8:0xf019f216 code segment = base 0x0, limit 0xfffff, type 0x1b = DPL 0, pres 1, def32 1, gran 1 processor eflags = interrupt enabled, resume, IOPL = 0 current process = 32 (ifconfig) interrupt mask = net tty panic: page fault syncing disks... 7 7 5done Automatic reboot ( Fatal Trap, and loop forever ...) Most times, I see the message: "ixintr without being inited!!" just before the message "changing root device to sd0a" Some of the boot messages go by too fast to read. The Intel Ether Express was set to IRQ 10 0x300 0xd0000 32k. Changing the Ehter Express to IRQ5 eliminated (or hid) the ixintr message, but, then received an aha0 not responding message with the remark "Frozen?" Returned the Intel board to IRQ 10. This works fine with 'NT. I transferred, via ftp, several 100MB between the notebook and 'NT, without error... Still, could the Intel Ether Express be the problem? Checked the aha0 setup. The DMA clock was set to 5.7Mhz. Reset the clock to 5Mhz. Ran Adaptec diags 20 loops - OK. Rebooted. This time, saw the same fatal trap as above, except the fault virtual address was 0xef0dff00. Cycled power on the machine. - same fatal trap! Disabled shadow ROM - reboot - same fatal trap at a similar virtual address. Disabled cache - reboot - same problem. I tried installing a small WINNT partition on the disk. This resulted in a geometry of 132/256/63. I liked this better than the DOS, because of the 63 sectors/track, but, it also failed. I tried the Fujitsu disk at SCSI ID 1, I removed the Maxtor disk - same results. I tried having a small dos partition on the disk at the same time - same problem. I tried having a small 'NT partition on the disk at the same time - same problem. I have used both Winnt 3.51 and DOS on the Fujitsu disk with no problem. I can communicate via the Intel Ether Express to FreeBSD on my notebook, both with ftp and samba. I have ran all the diags I can find - everything seems OK. More on the hardware: Motherboard Nice EISA 1.2 CPU Intel 486DX2-66 RAM 20MB 70ns IDE Controller - used for floppy only and IRWIN 4752 floppy tape controller. Serial/Parallel Diamond Stealth 64 VRAM - I know diamond company has ugly policies, but the board was existing and works if no code is needed. I notice that XFree86 seems to support this board... Adaptec 1542C 0x330 ID Device OS 0 Quantum Empire 1080S 1GB Disk Winnt 3.51 1 Maxtor LXT-340S 340MB Disk Dos support - to be removed later 2 Toshiba XM-3401TA CDROM 3 Fujitsu M1606S-512 1GB Disk FreeBSD 2.1(?) Intel Ether Express IRQ 10 0x300 Late Thin 0xd0000 I would like to convert this machine to FreeBSD. Trashing 'NT is currently out of the question, but, in a year, or so, the machine will be entirely FreeBSD, if the install problems can be worked-out. I think I need this time to convert the applications... The machine is in Fremont and can be moved for observation, if desired. Any Ideas?