Re: [PATCH for-4.15] vtd: make sure QI/IR are disabled before initialisation

[Igor Druzhinin <igor.druzhinin@xxxxxxxxxx>]

On 08/03/2021 08:18, Jan Beulich wrote:
> On 08.03.2021 08:00, Igor Druzhinin wrote:
> > BIOS might pass control to Xen leaving QI and/or IR in enabled and/or
> > partially configured state. In case of x2APIC code path where EIM is
> > enabled early in boot - those are correctly disabled by Xen before any
> > attempt to configure. But for xAPIC that step is missing which was
> > proven to cause QI initialization failures on some ICX based platforms
> > where QI is left pre-enabled and partially configured by BIOS.
> And those systems then tell us to avoid use of x2APIC? I would have
> expected that on modern systems we wouldn't see such quirky firmware
> behavior anymore. Anyway, half a sentence to this effect might help
> here, as without such firmware behavior the only way to run into
> this ought to be use of "no-x2apic" on the command line. Which in
> turn might require justification (and potentially a fix elsewhere in
> the code to make use of that option unnecessary).
>
> > Unify the behaviour between x2APIC and xAPIC code paths keeping that in
> > line with what Linux does.
> >
> > Signed-off-by: Igor Druzhinin <igor.druzhinin@xxxxxxxxxx>
> Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>
> with some editing of the description. If no other need for a v2
> arises, I suppose whatever you come up with could be folded in
> while committing.
How about:

"... But for xAPIC that step is missing which was proven to cause QI 
initialization failures on some ICX based platforms where QI is left 
pre-enabled and partially configured by BIOS. That problem becomes hard 
to avoid since those platforms are shipped with x2APIC opt out being 
advertised by default at the same time by firmware.
..."

Igor