Commit fd717f11 authored by Paolo Bonzini's avatar Paolo Bonzini

KVM: x86: apply guest MTRR virtualization on host reserved pages

Currently guest MTRR is avoided if kvm_is_reserved_pfn returns true.
However, the guest could prefer a different page type than UC for
such pages. A good example is that pass-throughed VGA frame buffer is
not always UC as host expected.

This patch enables full use of virtual guest MTRRs.
Suggested-by: default avatarXiao Guangrong <guangrong.xiao@linux.intel.com>
Tested-by: Joerg Roedel <jroedel@suse.de> (on AMD)
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
parent e098223b
...@@ -1166,14 +1166,11 @@ static u64 svm_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio) ...@@ -1166,14 +1166,11 @@ static u64 svm_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio)
u8 mtrr; u8 mtrr;
/* /*
* 1. MMIO: always map as UC * 1. MMIO: trust guest MTRR, so same as item 3.
* 2. No passthrough: always map as WB, and force guest PAT to WB as well * 2. No passthrough: always map as WB, and force guest PAT to WB as well
* 3. Passthrough: can't guarantee the result, try to trust guest. * 3. Passthrough: can't guarantee the result, try to trust guest.
*/ */
if (is_mmio) if (!is_mmio && !kvm_arch_has_assigned_device(vcpu->kvm))
return _PAGE_NOCACHE;
if (!kvm_arch_has_assigned_device(vcpu->kvm))
return 0; return 0;
mtrr = kvm_mtrr_get_guest_memory_type(vcpu, gfn); mtrr = kvm_mtrr_get_guest_memory_type(vcpu, gfn);
......
...@@ -8632,22 +8632,17 @@ static u64 vmx_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio) ...@@ -8632,22 +8632,17 @@ static u64 vmx_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio)
u64 ipat = 0; u64 ipat = 0;
/* For VT-d and EPT combination /* For VT-d and EPT combination
* 1. MMIO: always map as UC * 1. MMIO: guest may want to apply WC, trust it.
* 2. EPT with VT-d: * 2. EPT with VT-d:
* a. VT-d without snooping control feature: can't guarantee the * a. VT-d without snooping control feature: can't guarantee the
* result, try to trust guest. * result, try to trust guest. So the same as item 1.
* b. VT-d with snooping control feature: snooping control feature of * b. VT-d with snooping control feature: snooping control feature of
* VT-d engine can guarantee the cache correctness. Just set it * VT-d engine can guarantee the cache correctness. Just set it
* to WB to keep consistent with host. So the same as item 3. * to WB to keep consistent with host. So the same as item 3.
* 3. EPT without VT-d: always map as WB and set IPAT=1 to keep * 3. EPT without VT-d: always map as WB and set IPAT=1 to keep
* consistent with host MTRR * consistent with host MTRR
*/ */
if (is_mmio) { if (!is_mmio && !kvm_arch_has_noncoherent_dma(vcpu->kvm)) {
cache = MTRR_TYPE_UNCACHABLE;
goto exit;
}
if (!kvm_arch_has_noncoherent_dma(vcpu->kvm)) {
ipat = VMX_EPT_IPAT_BIT; ipat = VMX_EPT_IPAT_BIT;
cache = MTRR_TYPE_WRBACK; cache = MTRR_TYPE_WRBACK;
goto exit; goto exit;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment