[Qemu-devel] Qemu and PCI Interrupt Link?

[Qemu-devel] Qemu and PCI Interrupt Link?

[Peter Lieven]

Hi,

i noticed that newer guest kernels emit the following message at startup:

ACPI: No IRQ available for PCI Interrupt Link [LNKS]. Try pci=noacpi or acpi=off

Cmdline:
x86_64-softmmu/qemu-system-x86_64 -enable-kvm -m 2048 -cdrom ~/Downloads/ubuntu-16.04-desktop-amd64.iso

Is this expected?

Peter

Re: [Qemu-devel] Qemu and PCI Interrupt Link?

[Laszlo Ersek]

On 07/04/16 15:14, Peter Lieven wrote:
> Hi,
> 
> i noticed that newer guest kernels emit the following message at startup:
> 
> ACPI: No IRQ available for PCI Interrupt Link [LNKS]. Try pci=noacpi or
> acpi=off
> 
> Cmdline:
> x86_64-softmmu/qemu-system-x86_64 -enable-kvm -m 2048 -cdrom
> ~/Downloads/ubuntu-16.04-desktop-amd64.iso
> 
> Is this expected?

This seems to come from kernel commit 5ebc76035303. Thankfully the
commit message is quite detailed.

The commit message makes me think that the kernel patch is incorrect. It
certainly seems to conflict with the following SeaBIOS commit:

commit f64a472a481784231fbf8541825501df411b11d1
Author: Paolo Bonzini <pbonzini@redhat.com>
Date:   Thu Dec 13 08:42:02 2012 +0100

    acpi: reintroduce LNKS

That SeaBIOS patch is where we have the identical functionality in QEMU
from (see QEMU commits 74523b850189a and c35b6e8032105).

Thanks
Laszlo

Re: [Qemu-devel] Qemu and PCI Interrupt Link?

[Paolo Bonzini]

On 05/07/2016 20:30, Laszlo Ersek wrote:
> On 07/04/16 15:14, Peter Lieven wrote:
>> Hi,
>>
>> i noticed that newer guest kernels emit the following message at startup:
>>
>> ACPI: No IRQ available for PCI Interrupt Link [LNKS]. Try pci=noacpi or
>> acpi=off
>>
>> Cmdline:
>> x86_64-softmmu/qemu-system-x86_64 -enable-kvm -m 2048 -cdrom
>> ~/Downloads/ubuntu-16.04-desktop-amd64.iso
>>
>> Is this expected?
> 
> This seems to come from kernel commit 5ebc76035303. Thankfully the
> commit message is quite detailed.
> 
> The commit message makes me think that the kernel patch is incorrect. It
> certainly seems to conflict with the following SeaBIOS commit:
> 
> commit f64a472a481784231fbf8541825501df411b11d1
> Author: Paolo Bonzini <pbonzini@redhat.com>
> Date:   Thu Dec 13 08:42:02 2012 +0100
> 
>     acpi: reintroduce LNKS
> 
> That SeaBIOS patch is where we have the identical functionality in QEMU
> from (see QEMU commits 74523b850189a and c35b6e8032105).

The error is harmless.  LNKS has been there forever and I suspect no one
really knows why it was introduced in the first place.  The link for the
device that uses LNKS probably can be changed back to LNKA since the SCI
is specified in the FADT, but again---who knows why it was changed.

Paolo

Re: [Qemu-devel] Qemu and PCI Interrupt Link?

[Peter Lieven]

Am 05.07.2016 um 22:53 schrieb Paolo Bonzini:
>
> On 05/07/2016 20:30, Laszlo Ersek wrote:
>> On 07/04/16 15:14, Peter Lieven wrote:
>>> Hi,
>>>
>>> i noticed that newer guest kernels emit the following message at startup:
>>>
>>> ACPI: No IRQ available for PCI Interrupt Link [LNKS]. Try pci=noacpi or
>>> acpi=off
>>>
>>> Cmdline:
>>> x86_64-softmmu/qemu-system-x86_64 -enable-kvm -m 2048 -cdrom
>>> ~/Downloads/ubuntu-16.04-desktop-amd64.iso
>>>
>>> Is this expected?
>> This seems to come from kernel commit 5ebc76035303. Thankfully the
>> commit message is quite detailed.
>>
>> The commit message makes me think that the kernel patch is incorrect. It
>> certainly seems to conflict with the following SeaBIOS commit:
>>
>> commit f64a472a481784231fbf8541825501df411b11d1
>> Author: Paolo Bonzini <pbonzini@redhat.com>
>> Date:   Thu Dec 13 08:42:02 2012 +0100
>>
>>      acpi: reintroduce LNKS
>>
>> That SeaBIOS patch is where we have the identical functionality in QEMU
>> from (see QEMU commits 74523b850189a and c35b6e8032105).
> The error is harmless.  LNKS has been there forever and I suspect no one
> really knows why it was introduced in the first place.  The link for the
> device that uses LNKS probably can be changed back to LNKA since the SCI
> is specified in the FADT, but again---who knows why it was changed.

Okay, so just ignore this message for good or is there anything futher
I can/should do?

Peter