🚀 go-pugleaf

I have a rather hairy hardware/Linux problem here. I hope someone can
help me.

I have been having problems with the SCSI on my p90 since I purchased it. 
It will not reliably work with the ncr53c810 card I purchased with the
system, or the Adaptec 1542B I purchased to replace it.

The system specifications:

p90, 16MB of RAM
Gigabyte GA586IP Motherboard (ISA/PCI)
GA-410 ncr53c810 PCI SCSI controller (also tried aha1542B and aha1542CF)
Micropolis 1G HDD (also tried Connor 540M)
IDE hard disk controller
4-port serial card
ISA VGA card

This problem has been occurring since I tried Linux 1.1.59 in December
right up to recent kernels (1.3.15, plus 1.2.11 with Drew Eckhardt's latest 
alpha patches up to rel11). It can be fairly random in nature, but repeated 
runs of the hard disk benchmark "bonnie" will ALWAYS cause the problem to 
occur sooner or later.

I have stripped down the system to the minimum configuration (removed all
unneeded devices in both the kernel and the hardware), and tried all
permutations and combinations of motherboard/BIOS settings. I have also
tried, at various times, patches for "issue queue" problem that were posted
to the ncr53c810 mailing list (thanks to John Saunders helping me out
there). None of them have helped, though some cause the system to crash
faster than others.

The motherboard has an Award 4.50G BIOS, and initially had version 3.04 of
the NCR SCSI BIOS. This was later updated to version 3.06, which is the
latest version available for the GA586IP, according to the distributor.

As I need the system working I have been running the system with an IDE
drive (which works just fine), only trying the system out with SCSI when 
new versions of the ncr53c810 driver have come out. But now I have a 
pressing need to get this system up and running reliably with SCSI.

To get the system working with SCSI, I purchased an Adaptec 1542B card, as
this card is known to have a stable, reliable driver. This card does
not work reliably with the system either, despite working flawlessly on 
my 486. I have tried swapping hard drives, cables (all known to work on 
other systems), and even an aha1542CF, and none work reliably. The error
that is reported with Adaptec cards is:

aha1542_intr_handle: Unexpected interrupt
tarstat=0, hastat=0 idlun=88 ccb#=1

The 1542B is on IRQ 11, and there is no other device on 11. The system is
configured so that none of the PCI IRQ values are set to 11 either.

The company that sold me the system has ceased trading, and I am dealing
directly with the distributor. I have returned the system to them once
already, and they replaced the motherboard with one of the same type, which
also has the same problems. Before I take it back again (and probably have it
out of commission for another 2 weeks), I thought I'd try posting to the 
Linux newsgroups to see anyone has any ideas I haven't thought of yet, or 
any knowledge of the Gigabyte GA586IP motherboard. I cannot afford to 
continue purchasing SCSI adaptors (or motherboards) until I find a 
combination that works reliably, and this system will need to run soon.

I don't presently have net access, so I'm asking a friend to post this
for me. I'll have any followups or replies sent on to me though.

Regards,

Duncan Robertson

===============< assorted tech. info follows >=========================

Here is the contents of /proc/pci under Linux 1.3.15:

PCI devices found:
  Bus  0, device   3, function  0:
    Non-VGA device: NCR 53c810 (rev 1).
      Medium devsel.  IRQ 9.  Master Capable.  Latency=64.
      I/O at 0xd000.
      Non-prefetchable 32 bit memory at 0xfc000000.
  Bus  0, device   2, function  0:
    Non-VGA device: Intel 82378IB (rev 3).
      Medium devsel.  Master Capable.  No bursts.
  Bus  0, device   0, function  0:
    Host bridge: Intel 82434LX Mercury/Neptune (rev 17).
      Slow devsel.  Master Capable.  Latency=32.

Here is a sample run with the ncr53c810 controller, booting up and running
"bonnie" until the SCSI hangs:

Console: colour EGA+ 80x25, 1 virtual console (max 63)
bios32_init : BIOS32 Service Directory structure at 0x000fb550
bios32_init : BIOS32 Service Directory entry at 0xfb85c
pcibios_init : PCI BIOS revision 2.00 entry at 0xfb88c
Probing PCI hardware.
Calibrating delay loop.. ok - 36.08 BogoMips
Serial driver version 4.11 with no serial options enabled
tty00 at 0x03f8 (irq = 4) is a 16550A
tty01 at 0x02f8 (irq = 3) is a 16550A
tty02 at 0x03e8 (irq = 10) is a 16550A
tty03 at 0x02e8 (irq = 10) is a 16550A
tty04 at 0x02f0 (irq = 10) is a 16550A
tty05 at 0x03e0 (irq = 10) is a 16550A
PCSP-device 0.9 init:
  PCSP-Measurement: Maximal samplerate 149147 Hz, 18356 Hz used
  PC-Speaker, PCSP-Mixer installed
hda: WDC AC2340H, 325MB w/128KB Cache, CHS=1010/12/55, MaxMult=16
ide0: primary interface on irq 14
Floppy drive(s): fd0 is 1.44M
FDC 0 is a post-1991 82077
scsi-ncr53c7,8xx : at PCI bus 0, device 3,  function 0
scsi-ncr53c7,8xx : NCR53c810 at memory 0xfc000000, io 0xd000, irq 9
scsi0 : burst length 8
scsi0 : reset ccf to 3 from 0
scsi0 : NCR code relocated to 0x1e97e0
scsi0 : test 1 started
scsi0 : NCR53c{7,8}xx (rel 9)
scsi : 1 host.
  Vendor: MICROP    Model: 4110-09TBCU0322L  Rev: HT01
  Type:   Direct-Access                      ANSI SCSI revision: 02
Detected scsi disk sda at scsi0, id 4, lun 0
scsi : detected 1 SCSI disk total.
SCSI Hardware sector size is 512 bytes on device sda
Memory: 15032k/16384k available (692k kernel code, 384k reserved, 276k data)
This processor honours the WP bit even when in supervisor mode. Good.
Swansea University Computer Society NET3.019
Swansea University Computer Society TCP/IP for NET3.019
IP Protocols: ICMP, UDP, TCP
PPP: version 0.2.7 (4 channels) NEW_TTY_DRIVERS OPTIMIZE_FLAGS
TCP compression code copyright 1989 Regents of the University of California
PPP line discipline registered.
SLIP: version 0.8.3-NET3.019-NEWTTY (16 channels) (6 bit encapsulation enabled)
CSLIP: code copyright 1989 Regents of the University of California
Checking 386/387 coupling... Hmm, FDIV bug i586 system
Checking 'hlt' instruction... Ok.
Linux version 1.2.11 (root@darklands) (gcc version 2.5.8) #222 Sun Jul 23 23:47:46 EST 1995
Partition check:
  sda: sda1 sda2 sda3 < sda5 sda6 > sda4
  hda: multiple mode turned off
  hda: hda1 hda2 hda3 hda4
VFS: Mounted root (ext2 filesystem) readonly.
Adding Swap: 24744k swap-space
scsi0 : existing issue queue
scsi0 : disabled.  Unload and reload
scsi0 : driver disabled
SCSI disk error : host 0 id 4 lun 0 return code = 25040000
scsidisk I/O error: dev 0806, sector 105634
scsi0 : disabled target 4 lun 0
SCSI disk error : host 0 id 4 lun 0 return code = 25040000
scsidisk I/O error: dev 0806, sector 105636
scsi0 : disabled target 4 lun 0
scsi : aborting command due to timeout : pid 7457, scsi0, id 4, lun 0 Read (6) 12 5c c6 04 00 
scsi0 : disabled target 4 lun 0
SCSI disk error : host 0 id 4 lun 0 return code = 25040000
scsidisk I/O error: dev 0806, sector 105638
scsi0 : disabled target 4 lun 0
scsi0 : found command 7458 in Linux issue queue
SCSI disk error : host 0 id 4 lun 0 return code = 25040000
scsidisk I/O error: dev 0806, sector 105640
scsi0 : disabled target 4 lun 0


Here is a sample run of the system with the same hard drive but on the
1542B controller:

Console: 16 point font, 400 scans
Console: colour VGA+ 80x25, 1 virtual console (max 63)
pcibios_init : BIOS32 Service Directory structure at 0x000fb550
pcibios_init : BIOS32 Service Directory entry at 0xfb85c
pcibios_init : PCI BIOS revision 2.00 entry at 0xfb88c
Probing PCI hardware.
Calibrating delay loop.. ok - 36.08 BogoMips
Serial driver version 4.11 with no serial options enabled
tty00 at 0x03f8 (irq = 4) is a 16550A
tty01 at 0x02f8 (irq = 3) is a 16550A
tty02 at 0x03e8 (irq = 4) is a 16550A
tty03 at 0x02e8 (irq = 3) is a 16550A
hda: WDC AC2340H, 325MB w/128KB Cache, CHS=1010/12/55, MaxMult=16
ide0: primary interface on irq 14
Floppy drive(s): fd0 is 1.44M
Configuring Adaptec (SCSI-ID 7) at IO:330, IRQ 11, DMA priority 5
scsi-ncr53c7,8xx : at PCI bus 0, device 3,  function 0
scsi-ncr53c7,8xx : NCR53c810 at memory 0xfc000000, io 0xd000, irq 9
scsi1 : using io mapped access
scsi1 : using initiator ID 7
scsi1 : using level active interrupts
scsi1 : burst length 8
scsi1 : using 40MHz SCSI clock
scsi1 : NCR code relocated to 0x001edc0c
scsi1 : test 1 started
scsi0 : Adaptec 1542
scsi1 : NCR53c{7,8}xx (rel 4)
scsi : 2 hosts.
  Vendor: MICROP    Model: 4110-09TBCU0322L  Rev: HT01
  Type:   Direct-Access                      ANSI SCSI revision: 02
Detected scsi disk sda at scsi0, channel 0, id 4, lun 0
scsi : detected 1 SCSI disk total.
SCSI Hardware sector size is 512 bytes on device sda
Memory: 15016k/16384k available (744k kernel code, 384k reserved, 240k data)
This processor honours the WP bit even when in supervisor mode. Good.
Swansea University Computer Society NET3.030 Snap #1 for Linux 1.3.4
Swansea University Computer Society TCP/IP for NET3.030 (Snapshot #1)
IP Protocols: IGMP, ICMP, UDP, TCP
SLIP: version 0.8.3-NET3.019-NEWTTY (16 channels) (6 bit encapsulation enabled)
CSLIP: code copyright 1989 Regents of the University of California
Checking 386/387 coupling... Hmm, FDIV bug i586 system
Checking 'hlt' instruction... Ok.
Linux version 1.3.15 (root@darklands) (gcc version 2.5.8) #1 Sun Aug 13 17:46:26 EST 1995
Partition check:
  sda: sda1 sda2 sda3 < sda5 sda6 > sda4
  hda: hda1 hda2 hda3 hda4
VFS: Mounted root (ext2 filesystem) readonly.
Adding Swap: 24744k swap-space
aha1542_intr_handle: Unexpected interrupt
tarstat=0, hastat=0 idlun=88 ccb#=1