[lug] Spurious 8259A interrupt: IRQ7, then hang

D. Stimits stimits at attbi.com
Mon Dec 2 20:58:21 MST 2002


Andrew Gilmore wrote:

> I've googled a bit for this one, but come up empty so far.
> If anyone has any ideas, that'd be great.
>
> This is my first seriously misbehaving system under Linux, and I'm a bit
> annoyed.
>
> Symptoms:
>
> ssh session over ethernet hangs after a few CRs.
> Console immediately says:
> Spurious 8259A interrupt: IRQ7

Perhaps this is not related at all...I once asked Alan Cox and some of 
the kernel people about the spurious interrupt message, it is normal and 
irrelevant, some kind of leftover message for developers. Whenever I got 
lost interrupt, it was related to the IO-APIC on Intel chipsets for SMP. 
Running with kernel option "noapic" fixed it (and also removed 
responsiveness that SMP is famous for). With the old system mentioned 
below, that couldn't possibly be an SMP issue, but I suspect the 
spurious interrupt is itself in no way related. People suggesting cables 
and similar possibilities are likely on target. Out of curiousity, what 
does /proc/interrupts contain? Are any of the related pci devices (like 
the ethernet card) in any way famous for not liking to share interrupts? 
There are a significant number of older pci devices that didn't share as 
well as newer ones...try changing slot of the ethernet card, possibly 
video card also.

D. Stimits, stimits AT attbi.com

> then about once a second or two:
> (without DMA enabled)
> hda: lost interrupt
> (with DMA enabled)
> idedma some error message about only supporting lostirq function 13
> hda: lost interrupt
>
> The console still responds, but starting anything that accesses the HD
> will hang with no recourse but reset.
>
> Box:
>
> Pentium 200, 32 MB ram, 2 gig HD on ide0, CDROM on ide1
> VIA VXPro chipset (chips on motherboard have "VXTwo" logo)
> Trident video card (96xx series) text mode only
> Netgear FA310 ethernet card
> RedHat 8.0, no patches yet (original kernel 2.4.18-14)
>
> Story so far:
>
> I originally suspected the ide DMA functions, due to some l-k traffic
> about lost interrupts on VPx (x<3) chipsets, but the problem still
> occurs with dma disabled.
>
> I have two identical ethernet cards (the FA310) and they are on the
> supported hardware list, both show this same result.
>
> I've been unable to reproduce this bug from the console. There are
> occassional occurances of the error message during the boot process, but
> the box doesn't hang at that time.
>
> Any ideas?
>
> Andrew
>
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> Join us on IRC: lug.boulder.co.us port=6667 channel=#colug
>





More information about the LUG mailing list