[0/2] x86emul: vendor specific treatment adjustments

Message

Jan Beulich Sept. 16, 2019, 9:44 a.m. UTC

I've noticed the issue the 1st patch means to address only while
putting together the 2nd. Considering the other Hygon enablement
in this release cycle I think we want patch 1 for 4.13. Patch 2
should be considered too, but we've been effectively mis-emulating
MOVSXD on modern Intel hardware for quite some time, so it getting
delayed until after 4.13 (and then possible be backported) wouldn't
be overly bad.

1: treat Hygon guests like AMD ones
2: adjust MOVSXD source operand handling

Jan

Comments

Jürgen Groß Sept. 18, 2019, 5:31 a.m. UTC | #1

On 16.09.19 11:44, Jan Beulich wrote:
> I've noticed the issue the 1st patch means to address only while
> putting together the 2nd. Considering the other Hygon enablement
> in this release cycle I think we want patch 1 for 4.13. Patch 2
> should be considered too, but we've been effectively mis-emulating
> MOVSXD on modern Intel hardware for quite some time, so it getting
> delayed until after 4.13 (and then possible be backported) wouldn't
> be overly bad.
> 
> 1: treat Hygon guests like AMD ones
> 2: adjust MOVSXD source operand handling

For the series:

Release-acked-by: Juergen Gross <[e-mail]>


Juergen