]> www.infradead.org Git - users/jedix/linux-maple.git/commit
KVM: VMX: Don't modify guest XFD_ERR if CR0.TS=1
authorSean Christopherson <seanjc@google.com>
Tue, 1 Oct 2024 05:00:45 +0000 (22:00 -0700)
committerSean Christopherson <seanjc@google.com>
Tue, 25 Feb 2025 17:35:40 +0000 (09:35 -0800)
commit3ef0df3f760f438d275334bbe193f95886890fc4
treec558a3ef151cb5813cecfe7647c8da744834df84
parenta64dcfb451e254085a7daee5fe51bf22959d52d3
KVM: VMX: Don't modify guest XFD_ERR if CR0.TS=1

Don't update the guest's XFD_ERR MSR if CR0.TS is set; per the SDM,
XFD_ERR is not modified if CR0.TS=1.  Although it's not explicitly stated
in the SDM, conceptually it makes sense the CR0.TS check would be done
prior to the XFD_ERR check, e.g. CR0.TS=1 blocks all SIMD state, whereas
XFD blocks only XTILE state.

  Device-not-available exceptions that are not due to XFD - those
  resulting from setting CR0.TS to 1 - do not modify the IA32_XFD_ERR MSR.

Opportunistically update the comment to call out that XFD_ERR is updated
before the VM-Exit check occurs.  Nothing in the SDM explicitly calls out
this behavior, but logically it must be the behavior, otherwise reading
XFD_ERR in handle_nm_fault_irqoff() would return stale data, i.e. the
to-be-delivered XFD_ERR value would need to be saved in EXIT_QUALIFICATION,
a la DR6 for #DB and CR2 for #PF, so that software could capture the guest
value.

Fixes: ec5be88ab29f ("kvm: x86: Intercept #NM for saving IA32_XFD_ERR")
Signed-off-by: Xin Li (Intel) <xin@zytor.com>
Tested-by: Shan Kang <shan.kang@intel.com>
Link: https://lore.kernel.org/r/20241001050110.3643764-3-xin@zytor.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
arch/x86/kvm/vmx/vmx.c