NCR53c810 SCSI host - kernel panic

Steven Youngs (steve@ncgrafix.demon.co.uk)
Mon, 26 Jun 1995 12:07:06 +0100 (BST)


Hi Drew,
Firstly, this message has also been sent to the linux-kernel mailing list.

I'm currently trying to configure a Compaq Prosiginia 300 server with Linux
v1.2.10.

This has an NCR53c810 SCSI chipset built into the motherboard, and my machine
has an IBM DPES-31080 1.05GB SCSI hard disk with SCSI ID 0.

Without this drive connected to the SCSI bus, I can boot the machine from a
boot floppy and a magneto optical root device (this is a SCSI device also).
If I connect the SCSI hard disk to the SCSI bus, the kernel panics with a general protection fault. This fault is generated in the print_insn function on the line

dcmd = (insn[0] >> 24) & 0xff;

where insn has the value 0xffffff00. (A presumably invalid pointer in this
instance)
print_insn is being called from the intr_dma function, immediately after the
printk("scsi%d : illegal instruction", host->host_no) line. The EIP value
printed during the kernel panic supports this, after cross referencing the
system.map file.

This hard disk by default has no junper on the "Sync Nego" pin. With a jumper
on this pin, the machine boots fine. (Compaq supply no documentation with this
drive, so I'm making an educated guess that the sync nego ping is to do with
synchronous transfer negotiation)

Anyway, the output leading up to this is as follows

....
scsi0 : NCR53c{7,8}xx (rel 4)
scsi : 1 host
scsi0 : target 0 requesting synchronous transfer period 100ns, offset 15
scsi0 : setting target 0 to asynchronous SCSI
scsi0 : illegal instruction general protection: 0000

The output then gives details the kernel panic produced by derefencing the
insn pointer in print_insn, as I have described above.

I don't know very much about SCSI drivers, SCSI protocol etc. so do you have
any ideas what is going on here?? At the very least, I guess it would be nice
if the kernel did not panic before your diagnostics have finished printing.
If on the other hand, my hardware is not behaving properly with respect to the
SCSI protocol etc - I need to get some hardware that does work.

Anyway, any ideas, comments etc are appreciated.

Cheers,
Steve

-------------
Steven Youngs
steve@ncgrafix.demon.co.uk