[3190] in linux-scsi channel archive
Re: AHA1542CF suddenly stopped working! Help!
daemon@ATHENA.MIT.EDU (Jay Vassos-Libove)
Tue Feb 3 14:41:51 1998
Date: Tue, 3 Feb 1998 14:05:46 -0500 (EST)
From: Jay Vassos-Libove <libove@felines.org>
To: "Leonard N. Zubkoff" <lnz@dandelion.com>
cc: linux-scsi@vger.rutgers.edu, redhat-list@redhat.com
In-Reply-To: <199802031806.KAA10145@dandelion.com>
Thanks to Leonard for teaching me about this. I had not realized that
interrupts went unused if no explicit software driver was installed.
The question that this raises in my mind is: why would this have changed?
The SCSI card configuration is no different than it was before it stopped
working. Similarly, the BIOS configuration is unchanged. The RAM upgrade
was successfully completed and tested prior to the problem beginning. All
cables and boards have been removed and reseated.
I will re-check the BIOS settings -- PCI BIOSes are new to me.
Any other ideas?
Thanks -
> From: Jay Vassos-Libove <libove@felines.org>
> [snip]
> first installed the software to test it out) and it too would load the
> boot loader and kernel from the SCSI chain, and get the same SCSI bus
> screwup when the AHA1542 driver initialized.
> [snip]
> Another experiment: I booted a DOS 5.0 floppy, ran FDISK, wiped out the
> Linux (NON-DOS) partitions, made a single primary full-size (307M) DOS
> partition and made it bootable, rebooted the DOS floppy, formatted the
> drive with system files, and ... oh, right, gotta do FDISK/MBR or else it
> will still read the boot loader from Linux and load the kernel from the
> middle of the disk where the DOS "format" didn't actually erase it :)
> Okay, so I did FDISK/MBR and booted fine in to DOS.
> As far as I'm concerned, this proves that the hardware is good, and that
> the fault is something in the Linux kernel/driver.
>
> This proves nothing so far. Accesses made through the BIOS, which includes
> loading the boot loader and kernel, as well as DOS operation (possibly unless
> you've loaded the ASPI manager) do not require interrupts to be operational.
>
> The most likely cause of the problems you've reported is interrupts not getting
> from the host adapter to the Adaptec 1542 driver. Check the IRQ channel
> assignment on the host adapter and make sure that IRQ channel is assigned to
> "ISA" rather than "PCI" in the motherboard BIOS.
Jay Vassos-Libove libove@felines.org
+1 770 552 0543 home +1 404 705 2867 work
Roswell, GA 30075 U.S.A.
