Re: unexpected IRQ trap at vector

On 7/27/06, Andy Green <andy@xxxxxxxxxxx> wrote:
Lonni J Friedman wrote:
> I just performed a fresh FC5-x86 installation on a Toshiba Tecra M3
> notebook. All was good until I updated to the latest kernel. Now i'm
> seeing this error in dmesg about 10 times/second, non-stop:
> unexpected IRQ trap at vector 12
>
> Anyone have any ideas on what is going on here? I tried booting with
> both noapic and acpi=off, and neither made any difference.

Basically a device on IRQ 12 is spamming you with interrupts, Linux
believes that is an "Illegal vector":

/*
* 'what should we do if we get a hw irq event on an illegal
vector'.
* each architecture has to answer this themself.
*/

For i386 the answer is to print that dmesg message.

What is an "illegal vector"? I guess one that does not have a driver
ready to service the interrupts.

Normally hardware devices don't start issuing interrupts until they are
recognized and asked to do so by a driver.

What does

cat /proc/interrupts

say? What devices are on your box that you are not using under Linux?
How about a PS/2-style mouse-like device or somesuch? Anything in
/var/log/messages at boot-time about devices not being recognized properly?

Thanks for your reply. IRQ 12 is owned by:
12: 115 IO-APIC-edge i8042

However, the odd thing that I've discovered is that if the onboard
Marvell NIC using the sky2 driver doesn't come up at boot, then all of
these errors are eliminated completely. sky2 isn't using IRQ12:
169: 0 IO-APIC-level uhci_hcd:usb4, sky2

So i'm still rather confused on this.


--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
L. Friedman netllama@xxxxxxxxx
LlamaLand http://netllama.linux-sxs.org

--
fedora-list mailing list
fedora-list@xxxxxxxxxx
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list

Relevant Pages

  • [PATCH 2.6.16] Shared interrupts sometimes lost
    ... wireless cards - based on the RaLink 2560 chip set. ... This is using the "legacy" rt2500 driver derived from the vendor ... But Linux is definitely implicated in the loss of interrupts. ... This is my first real introduction to the IRQ handling code in Linux, ...
    (Linux-Kernel)
  • Re: [RFC/PATCH] Kdump: Disabling PCI interrupts in capture kernel
    ... device to have its final IRQ routing established. ... > In kdump, sometimes, general driver initialization issues seems to be cropping ... > not expecting any interrupts yet. ... > is booting and kernel disables that irq line. ...
    (Linux-Kernel)
  • Re: [PATCH] Prevent nested interrupts when the IRQ stack is near overflowing v2
    ... point to run those handlers with IRQs enabled at all? ... I think the idea was to minimize irq latency for other interrupts ... such a driver is broken by ...
    (Linux-Kernel)
  • Re: IST in C# application
    ... The current driver is doing that with ... you have to use DeviceIoControl() API to ... I do not want to handle the IRQ source identification in the ISR (to ... it is not a standard way to handle interrupts ...
    (microsoft.public.windowsce.platbuilder)
  • Re: 2.6.7-mm1 PCNet Problems under VMWare 4.5.2
    ... interrupts), and driver does not modify it in any way (f.e. ... irq fires. ... sure that all interrupts we handle in loop are catched), ...
    (Linux-Kernel)