RE: [2.6.12rc4] PROBLEM: "drive appears confused" and "irq 18: nobody cared!"

From: Protasevich, Natalie
Date: Fri Jun 17 2005 - 12:33:27 EST

> Alan Cox wrote:
> >>Jun 17 12:07:49 orclex kernel: hdb: cdrom_pc_intr: The
> drive appears
> >>confused (ireason = 0x01) Jun 17 12:07:49 orclex kernel:
> irq 18: nobody cared (try booting with the "irqpoll" option.
> >
> > Something failed to clear IRQ 18, that typically means
> there are IRQ
> > routing problems rather than IDE ones and would explain your traces.
> >
> > Try booting with acpi=off and see what trace you get then.
> acpi=off makes linux hang and not continuing booting. Hm,
> syslog does not contain the trace until that crash but the
> last lines before the hanging are:
> ehci_hcd 0000:00:1d.7: USB 2.0 initialized, EHCI 1.00, driver
> 10 Dec 2004 hub 1-0:1.0: USB hub found
> I've tried booting the kernel with parameter irqpoll as you
> have suggested but it leads to a kernel panic.
> The last line was:
> kernel panic - not syncing: Aiee, killing interrupt handler!
> It's not saved in syslog too, so is there any way to get the
> trace to a file?

You can set up a serial console and capture the output.

I would also recommend booting with pci=routeirq, this will show the
pre-disposition of each GSI->IRQ pair, although sometimes it changes IRQ
distribution, and you may get different error. Also try using apic=debug
or acpi=verbose to see the IO-APIC lines setup. For debug failed IRQs, I
sometimes insert print_io_APIC() after each PCI device IRQ registration,
to see it got edge or level triggered and other details.
The other one that I saw causing problems especially for ISA devices is
ACPI PnP (and still does, I'm researching a similar problem on ES7000),
and in this case I get my system booted with pnpacpi=off.
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at