Hi Hans,
It's a really good question why that message is printed at all. As Frank
pointed out, the apic_check_deadline_errata() function only goes by the CPU
model and does not check whether the feature is available.
It is generally bad form to assume specific CPU features based solely on the
CPU model, and Intel pretty strongly advises against it. I would think that
apic_check_deadline_errata() should check for X86_FEATURE_TSC_DEADLINE_TIMER
first, because if it's not there, there is nothing to do. That would get rid of
the message on VirtualBox and would save a couple of cycles on older CPUs that
don't have the APIC deadline timer (and therefore don't need to have their
model figured out).
We will probably expose the microcode revision to the guest at some point, but
as I mentioned, that's something which could have unpredictable side effects.
Regards,
Michal
----- Original Message -----
From: [email protected]
To: [email protected], [email protected]
Sent: Wednesday, July 26, 2017 10:17:26 PM GMT +01:00 Amsterdam / Berlin / Bern
/ Rome / Stockholm / Vienna
Subject: Re: [vbox-dev] New kernel error message when running 4.13 as guest
Hi Michal,
On 26-07-17 17:18, Michal Necasek wrote:
>
> Hi Hans,
>
> The message is harmless. It's triggered because VirtualBox currently does
> not expose the host CPU microcode revision to guest operating systems. I
> don't think we want to do that without fully understanding what sorts of
> assumptions operating systems might make when they see the actual revision.
>
> At any rate, VirtualBox currently doesn't support TSC deadline timers so
> the message is simply noise.
Thank you for your reaction. I understand that the message is just noise, but
it would
still be nice to do something about it, Linux distributions have gone through
great
efforts to offer a smooth boot experience without scary text-messages and an
error
like this breaks that part of the user-experience.
I wonder why the kernel prints the message at all if vbox does not support
TSC deadline timers, could it be that somehow the capability bit for that
gets leaked from the host to the guest even though it is not supported ?
Regards,
Hans
> ----- Original Message -----
> From: [email protected]
> To: [email protected]
> Sent: Monday, July 24, 2017 9:49:56 AM GMT +01:00 Amsterdam / Berlin / Bern /
> Rome / Stockholm / Vienna
> Subject: [vbox-dev] New kernel error message when running 4.13 as guest
>
> Hi all,
>
> I just upgraded a vbox guest to 4.13, the good news is
> everything works fine, but one thing which I noticed, which
> would be nice to fix is this new error message:
>
> [ 0.000000] [Firmware Bug]: TSC_DEADLINE disabled due to Errata; please
> update microcode to version: 0xb2 (or later)
>
> This is on a Skylake host with no special settings done
> to the guest (I selected the Fedora 64 bit template
> when installing).
>
> Regards,
>
> Hans
> _______________________________________________
> vbox-dev mailing list
> [email protected]
> https://www.virtualbox.org/mailman/listinfo/vbox-dev
> _______________________________________________
> vbox-dev mailing list
> [email protected]
> https://www.virtualbox.org/mailman/listinfo/vbox-dev
>
_______________________________________________
vbox-dev mailing list
[email protected]
https://www.virtualbox.org/mailman/listinfo/vbox-dev
