Very Computer

Can anyone here help me with this...

I've been lurking on the Linux SCSI and RAID discussion groups and can't
find anything related to the sepcific problem but the issue is probably
trivial or at least interesting.

Hardware:
---------

Dual Pentium III (SMP) w/ Intel motherboard Pheonix 4.0 BIOS

Symbios SYM53C875 embedded PCI SCSI controller on motherboard

Intel DAC960 (Mylex) RAID controller (adapter) w/ 8 MB cache

(2) IBM DNES-309170W SCSI hard disks (SCSI IDs set at 0 and 1)

Problem:
--------

The ncr53c8xx (SCSI) kernel module won't finish loading if the DAC960
controller is in the machine.

If I remove the Mylex DAC960 controller from the machine, then ncr53c8xx
(SCSI) Linux kernel module loads without any problems (detects both SCSI
drives with correct IDs), but when the DAC960 controller is in the
machine, the ncr53c8xx module will start to load but after the module
detects both SCSI drives (after downloading SCSI scripts from each
drive) the module then aborts like so:

ncr53c876-0: Downloading SCSI scripts

ncr53c876-1: Downloading SCSI scripts

scsi: 2 hosts detected

scsi: aborting command due to timeout: pid 0; scsi 0; channel 0; id 0;
lun 0

ncr53c8xx_abort: ...

...but again, if I remove the DAC960 controller, then the ncr53c8xx
module loads fine.

Notes
-----

The DAC960 kernel module loads fine.

The Mylex (DAC960) BIOS is not happy unless I disable the Symbios BIOS
(embedded PCI SCSI host) in the system (Pheonix) BIOS, otherwise the
DAC960 will complain about "Spinning up drives.... Target 0, Channel 0
not reponding" -- so apparently both the Mylex DAC960 BIOS and the
Symbios BIOS can not be both active at the same time.

We used the DOS utility that came with the DAC960 controller to
initialize the two SCSI disks as a RAID 1 array which seem go OK, but
either way it has no effect on the "aborting command due to timeout"
error coming from the ncr53c8xx module tries to load.

Software
--------

Tried installing Red Hat 6.0 (Kernel 2.2.11 I believe) and Suse 6.0 --
same behavior with both kernels.

My Guess
--------

Does this have something to do with the SCSI BIOS not being enabled when
the Mylex BIOS is enabled -- a basic hardware conflict, which is odd
since the motherboard and RAID controller are both made by Intel.

I'm stumped. Hope you can send me a clue.

Thanks,
Kurt

Sent via Deja.com http://www.deja.com/
Before you buy.

I found the answer to my own question: Don't load the SCSI driver if you
have a RAID controller installed.

First disable the embedded SCSI chipset in the system BIOS. The other
problem is when installing Red Hat 6.0 it will still sniff out the SCSI
chipset and try to load the SCSI module for it which you don't want, so
when installing RH Linux choose "expert mode", when asked if you have
SCSI devices, say "Yes", then choose the RAID controller, then when
asked if you have any more SCSI devices answer "No", and then rest of
the install goes as normal.

Quote:

> Can anyone here help me with this...

> I've been lurking on the Linux SCSI and RAID discussion groups and
can't
> find anything related to the sepcific problem but the issue is
probably
> trivial or at least interesting.

> Hardware:
> ---------

> Dual Pentium III (SMP) w/ Intel motherboard Pheonix 4.0 BIOS

> Symbios SYM53C875 embedded PCI SCSI controller on motherboard

> Intel DAC960 (Mylex) RAID controller (adapter) w/ 8 MB cache

> (2) IBM DNES-309170W SCSI hard disks (SCSI IDs set at 0 and 1)

> Problem:
> --------

> The ncr53c8xx (SCSI) kernel module won't finish loading if the DAC960
> controller is in the machine.

> If I remove the Mylex DAC960 controller from the machine, then
ncr53c8xx
> (SCSI) Linux kernel module loads without any problems (detects both
SCSI
> drives with correct IDs), but when the DAC960 controller is in the
> machine, the ncr53c8xx module will start to load but after the module
> detects both SCSI drives (after downloading SCSI scripts from each
> drive) the module then aborts like so:

> ncr53c876-0: Downloading SCSI scripts

> ncr53c876-1: Downloading SCSI scripts

> scsi: 2 hosts detected

> scsi: aborting command due to timeout: pid 0; scsi 0; channel 0; id 0;
> lun 0

> ncr53c8xx_abort: ...

> ...but again, if I remove the DAC960 controller, then the ncr53c8xx
> module loads fine.

> Notes
> -----

> The DAC960 kernel module loads fine.

> The Mylex (DAC960) BIOS is not happy unless I disable the Symbios BIOS
> (embedded PCI SCSI host) in the system (Pheonix) BIOS, otherwise the
> DAC960 will complain about "Spinning up drives.... Target 0, Channel 0
> not reponding" -- so apparently both the Mylex DAC960 BIOS and the
> Symbios BIOS can not be both active at the same time.

> We used the DOS utility that came with the DAC960 controller to
> initialize the two SCSI disks as a RAID 1 array which seem go OK, but
> either way it has no effect on the "aborting command due to timeout"
> error coming from the ncr53c8xx module tries to load.

> Software
> --------

> Tried installing Red Hat 6.0 (Kernel 2.2.11 I believe) and Suse 6.0 --
> same behavior with both kernels.

> My Guess
> --------

> Does this have something to do with the SCSI BIOS not being enabled
when
> the Mylex BIOS is enabled -- a basic hardware conflict, which is odd
> since the motherboard and RAID controller are both made by Intel.

> I'm stumped. Hope you can send me a clue.

> Thanks,
> Kurt

> Sent via Deja.com http://www.deja.com/
> Before you buy.