[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#556106: do_IRQ: 0.184 No irq handler for vector



On Tue, 2010-06-15 at 10:52 +0300, Andrejs Dubovskis wrote:
> >>>> Package: linux-image-2.6.26-2-amd64
> >>>>> Version: 2.6.26-19lenny2
> >>>>> Severity: critical
> >>>>> Justification: breaks the whole system
> >>>>>
> >>>>>
> >>>>> On some disk load (over qla2xxx attached disks) the server prints an 
> >>>>> error and stops (console output attached below).
> >>>>> The same kernel with additional parameter pci=nomsi runs with no any 
> >>>>> problem.
> >>
> >>> The next release of Debian (6.0, code name Squeeze) will be based
> >>> on 2.6.32. Please test the current 2.6.32 from unstable/testing and tell
> >>> us whether the problem persists. If so, we should report it upstream
> >>> to the kernel.org developers.
> >>>
> >>> The 2.6.32 kernel is available from packages.debian.org and can
> >>> be installed in both Debian stable, testing and unstable
> >>> installations.
> >>
> >> During few hours I was unable to reproduce the error on lenny with kernel from
> >> testing. A hardware was exactly the same as before.
> > 
> > I believe this bug is the same as #572322, which I fixed in version
> > 2.6.26-22.  Please can you check whether the current stable security
> > update (version 2.6.26-22lenny1) fixes this bug?
> 
> I do not think that the bug is the same.
> I had loaded kernel 2.6.32 and 2.6.26-22lenny1. With the same command line
> /proc/interrupts are different
> 
> # cat 2.6.32-3-amd64
[...]
>  57:         19         23         19         23   PCI-MSI-edge      qla2xxx
>  58:         27         32         26         27   PCI-MSI-edge      qla2xxx
[...]
> # cat 2.6.26-2-amd64
[...]
>  24:         27         27         28         26   IO-APIC-fasteoi   qla2xxx
>  25:         38         31         34         35   IO-APIC-fasteoi   qla2xxx
[...]
> Yes, the 2.6.26 does not have the problem. But it is because of disabled MSI.
> 2.6.32 has enabled MSI, at the same time the problem does not happen.

The fix I applied is to disable MSI automatically for this device.  I
don't know whether the driver somehow works reliably in 2.6.32 without
that change, but that change has been accepted by QLogic and by the
Linux SCSI maintainer.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: