Re: [PATCH v9 20/20] KVM: selftests: Add APX state handling and XCR0 sanity checks
From: Christian Ludloff
Date: Thu Nov 19 2026 - 21:58:42 EST
On Thu, Nov 19, 2026 at 6:46 AM Sean Christopherson <seanjc@xxxxxxxxxx> wrote:
>
> On Wed, Nov 18, 2026, Christian Ludloff wrote:
> > > Since APX repurposes the XSAVE area previously used by MPX in the
> > > non-compacted format, add a check to ensure that MPX state is not set
> > > when APX is enabled.
> >
> > For the sake of completeness...
>
> For the sake of my sanity, fix your mail client to set In-Reply-To properly.
> It's not a gmail problem. I don't see this issue with mails from anyone else,
> and I just tested that sending a mail through the gmail web interface sets the
> header as expected.
True for a reply to a message that Gmail has (like here)
but not true for a reply to a message that Gmail doesn't
have (like when someone reads LKML via a web UI and
decides to reply by clicking Gmail's "Compose", pasting
the recipient(s), subject, and quoted text, etc.). There is
no way to paste an In-Reply-To into Gmail. And no sane
way to import an mbox file into Gmail either.
And yes, some mail clients fail to thread such messages
while others do not seem to have a problem.
Sorry for not having a working "git send-email" setup, as
well as for not just blindly handing it all to AI.
> > EVO processors do support simultaneously
> > enabled MPX and APX, with state that does
> > not overlap. (And several MSR control bits.)
>
> Until EVO processors are officially supported in KVM selftests, I don't care. I
> am a-ok supporting EVO processors, but if you want upstream to care about them,
> then someone needs to take the time to enable support.
I care that the MPX^APX narrative gets accompanied by
the "this is an Intel implementation-specific limitation", as
it is not a general x86 architecture limitation.
What code does with the ever growing ball of VMX/SVM
duct tape, I don't care.
--
C.