On Wed, Dec 28, 2022 at 6:36 PM Jim Mattson <jmattson@xxxxxxxxxx> wrote:
>
> On Tue, Dec 27, 2022 at 2:05 PM Aaron Lewis <aaronlewis@xxxxxxxxxx> wrote:
> >
> > The check in amx_test that ensures that XSAVE supports XTILE, doesn't
> > actually check anything. It simply returns a bool which the test does
> > nothing with.
> > Additionally, the check ensures that at least one of the XTILE bits are
> > set, XTILECFG or XTILEDATA, when it really should be checking that both
> > are set.
> >
> > Change both behaviors to:
> > 1) Asserting if the check fails.
> > 2) Fail if both XTILECFG and XTILEDATA aren't set.
>
> For (1), why not simply undo the damage caused by commit 5dc19f1c7dd3
> ("KVM: selftests: Convert AMX test to use X86_PROPRETY_XXX"), and
> restore the GUEST_ASSERT() at the call site?
I opted to add the assert in the check call to be consistent with the
others. I thought it would look odd to have 3 check calls being
called one after the other, where 2 of them made the call and did
nothing with the result and 1 was wrapped in an assert.
> Should this be two separate changes, since there are two separate bug fixes?
Sure, splitting this into two commits SGTM.
