[225] in linux-scsi channel archive
NCR 53c810 SCSI lockups
daemon@ATHENA.MIT.EDU (Eric Aardoom)
Sun Jun 4 08:15:39 1995
Date: Sun, 4 Jun 1995 12:42:11 +0100
From: eric@eenisn1.leeds.ac.uk (Eric Aardoom)
To: linux-scsi@vger.rutgers.edu
Hello all,
I've got a PCI pentium/66 system running linux 1.2.8. The motherboard
is a 1560/66 TIC-A ver.3, made in Taiwan. It has AWARD BIOS version
4.50G, and NCR BIOS version 3.004. The MB has an Intel mercury/neptune
chipset, 4 ISA slots and 4 PCI slots. I bought the machine from a
private owner, and didn't get any motherboard docs.
Recently I bought an NCR 53C810 SCSI controller. After about a month
of frustration I finally managed to get it running with linux. Here's
how I did it. As I don't have any docs with the motherboard I had to
trace the PCB to find out where the interrupt lines were going and
what the jumper settings should be. Eventually, I determined the
relation between the PCI interrupts and the ISA bus interrupts. From
this it finally became clear that on my system
INTA = IRQ12
INTB = IRQ 15
INTC = nothing (!?!?!)
INTD = IRQ14.
Now for the fun bit: I've installed the NCR card in PCI slot 0, and
set the interrupt to INT B, although I actually wanted to use
INTA. This was not possible, because the BIOS setup for the PCI bus
doesn't support IRQ12 in any of the PCI slots!!
In summary, I ended up with the card in PCI slot 0, jumper set to INTB
and the IRQ set to IRQ15. For the first time both IDE and SCSI drives
were detected at bootup! Major milestone :)
But now the next problem. I've repartitioned the SCSI drive, created
filesystems and a swap partition, and was ready to transfer the /usr
directory to one of the SCSI partitions by mounting the SCSI partition
on /tmpmount and doing
cd /usr
tar cf - | (cd /tmpmount; tar xvfp -)
I've tried this at least 6 times, and every single time the SCSI bus
locks up, reporting a BUS FAULT. On the earlier kernels I've tried
(1.2.1 and 1.2.3) the system still runs, and I can still log in on
another VC. Using kernel 1.2.8 it crashes with a kernel stack
overflow. After rebooting and running an fsck on the SCSI partition
there are lots of fs errors, even for the files that seemed to have
been transferred correctly before the SCSI bus locks up. Below is an
extract of what happens. I hope someone can enlighten me on this one.
Jun 4 08:59:47 tensai kernel: Kernel logging (proc) started.
Jun 4 08:59:47 tensai kernel: Console: colour EGA+ 80x25, 1 virtual console (max 63)
Jun 4 08:59:47 tensai kernel: bios32_init : BIOS32 Service Directory structure at 0x000fadd0
Jun 4 08:59:47 tensai kernel: bios32_init : BIOS32 Service Directory entry at 0xfb0b4
Jun 4 08:59:47 tensai kernel: pcibios_init : PCI BIOS revision 2.00 entry at 0xfb0e4
Jun 4 08:59:47 tensai kernel: Probing PCI hardware.
Jun 4 08:59:47 tensai kernel: Calibrating delay loop.. ok - 26.84 BogoMips
Jun 4 08:59:47 tensai kernel: Serial driver version 4.11 with no serial options enabled
Jun 4 08:59:47 tensai kernel: tty00 at 0x03f8 (irq = 4) is a 16450
Jun 4 08:59:47 tensai kernel: tty01 at 0x02f8 (irq = 3) is a 16450
Jun 4 08:59:47 tensai kernel: hda: H3342-A4, 327MB w/96KB Cache, CHS=872/16/48, MaxMult=16
Jun 4 08:59:47 tensai kernel: ide0: primary interface on irq 14
Jun 4 08:59:47 tensai kernel: Floppy drive(s): fd0 is 1.44M
Jun 4 08:59:47 tensai kernel: FDC 0 is a 8272A
Jun 4 08:59:47 tensai kernel: scsi-ncr53c7,8xx : at PCI bus 0, device 1, function 0
Jun 4 08:59:47 tensai kernel: scsi-ncr53c7,8xx : NCR53c810 at memory 0xfc000000, io 0xd000, irq 15
Jun 4 08:59:47 tensai kernel: scsi0 : using io mapped access
Jun 4 08:59:47 tensai kernel: scsi0 : using initiator ID 7
Jun 4 08:59:47 tensai kernel: scsi0 : using edge triggered interrupts
Jun 4 08:59:47 tensai kernel: scsi0 : burst length 8
Jun 4 08:59:47 tensai kernel: scsi0 : using 40MHz SCSI clock
Jun 4 08:59:47 tensai kernel: scsi0 : NCR code relocated to 0x1c1750
Jun 4 08:59:47 tensai kernel: scsi0 : test 1 started
Jun 4 08:59:47 tensai kernel: scsi0 : NCR53c{7,8}xx (rel 4)
Jun 4 08:59:47 tensai kernel: scsi : 1 host.
Jun 4 08:59:47 tensai kernel: Vendor: DEC Model: RZ26L (C) DEC Rev: 440C
Jun 4 08:59:47 tensai kernel: Type: Direct-Access ANSI SCSI revision: 02
Jun 4 08:59:47 tensai kernel: Detected scsi disk sda at scsi0, id 0, lun 0
Jun 4 08:59:47 tensai kernel: scsi : detected 1 SCSI disk total.
Jun 4 08:59:47 tensai kernel: SCSI Hardware sector size is 512 bytes on device sda
Jun 4 08:59:47 tensai kernel: Memory: 7000k/8192k available (620k kernel code, 384k reserved, 188k data)
Jun 4 08:59:47 tensai kernel: This processor honours the WP bit even when in supervisor mode. Good.
Jun 4 08:59:47 tensai kernel: Swansea University Computer Society NET3.019
Jun 4 08:59:47 tensai kernel: Swansea University Computer Society TCP/IP for NET3.019
Jun 4 08:59:47 tensai kernel: IP Protocols: ICMP, UDP, TCP
Jun 4 08:59:47 tensai kernel: Checking 386/387 coupling... Hmm, FDIV bug i586 system
Jun 4 08:59:47 tensai kernel: Checking 'hlt' instruction... Ok.
Jun 4 08:59:47 tensai kernel: Linux version 1.2.8 (root@tensai) (gcc version 2.6.3) #2 Wed May 10 19:39:55 BST 1995
Jun 4 08:59:47 tensai kernel: Partition check:
Jun 4 08:59:47 tensai kernel: sda: sda1 sda2 sda3
Jun 4 08:59:47 tensai kernel: hda: hda1 hda2
Jun 4 08:59:47 tensai kernel: VFS: Mounted root (ext2 filesystem) readonly.
Jun 4 08:59:47 tensai kernel: Adding Swap: 16124k swap-space
Jun 4 09:00:24 tensai login: ROOT LOGIN ON tty1
Jun 4 09:12:08 tensai kernel: scsi0 : BUS FAULT, DSP=0x1c2298, DCMD|DBC=0x8000320, DSA=0x1c20a8
Jun 4 09:12:08 tensai kernel: DSPS=0x307c00, TEMP=0x1c19c4, DMODE=0x80
Jun 4 09:12:08 tensai kernel: scsi0 : dsa at 0x1c20a8
Jun 4 09:12:08 tensai kernel: + 64 : dsa_msgout length = 1, data = 0x1c1fdc
Jun 4 09:12:08 tensai kernel: Identify disconnect not allowed lun 0
Jun 4 09:12:08 tensai kernel: scsi0 : DSP->
Jun 4 09:12:08 tensai kernel: 001c2298 : 0x08000400 0x00307c00
Jun 4 09:12:08 tensai kernel: 001c22a0 : 0x8f0b0000 0x001c1aa4
Jun 4 09:12:08 tensai kernel: scsi0 : DMA FIFO not empty
Jun 4 09:12:14 tensai kernel: scsi : aborting command due to timeout : pid 2313, scsi0, id 0, lun 0 Write (6) 00 46 ef f4 00
Jun 4 09:12:14 tensai kernel: scsi0 : DANGER : command in running list, can not abort.
cheers,
Eric Aardoom
CAA Institute of Satellite Navigation
Dept. of Electronic and Electrical Engineering
University of Leeds, LS2 9JT, UK
E-mail: E.Aardoom@CAAISN.leeds.ac.uk
Phone: +44 113 233 2090
Fax: +44 113 233 2032
