Merge branch 'x86/urgent' into x86/asm, before applying dependent patches

Signed-off-by: Ingo Molnar <mingo@kernel.org>

9 files changed, 179 insertions, 34 deletions

diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
index 64dd46793099..2fbea2544f24 100644
--- a/arch/x86/kvm/cpuid.c
+++ b/arch/x86/kvm/cpuid.c
@@ -98,6 +98,8 @@ int kvm_update_cpuid(struct kvm_vcpu *vcpu)
 		best->ebx = xstate_required_size(vcpu->arch.xcr0, true);
 
 	vcpu->arch.eager_fpu = use_eager_fpu() || guest_cpuid_has_mpx(vcpu);
+	if (vcpu->arch.eager_fpu)
+		kvm_x86_ops->fpu_activate(vcpu);
 
 	/*
 	 * The existing code assumes virtual address is 48-bit in the canonical
diff --git a/arch/x86/kvm/iommu.c b/arch/x86/kvm/iommu.c
index 7dbced309ddb..5c520ebf6343 100644
--- a/arch/x86/kvm/iommu.c
+++ b/arch/x86/kvm/iommu.c
@@ -200,6 +200,7 @@ int kvm_assign_device(struct kvm *kvm, struct pci_dev *pdev)
 			goto out_unmap;
 	}
 
+	kvm_arch_start_assignment(kvm);
 	pci_set_dev_assigned(pdev);
 
 	dev_info(&pdev->dev, "kvm assign device\n");
@@ -224,6 +225,7 @@ int kvm_deassign_device(struct kvm *kvm, struct pci_dev *pdev)
 	iommu_detach_device(domain, &pdev->dev);
 
 	pci_clear_dev_assigned(pdev);
+	kvm_arch_end_assignment(kvm);
 
 	dev_info(&pdev->dev, "kvm deassign device\n");
 
diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
index 2f0ade48614f..946f52c43b8f 100644
--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -1595,7 +1595,7 @@ void kvm_lapic_reset(struct kvm_vcpu *vcpu, bool init_event)
 	for (i = 0; i < APIC_LVT_NUM; i++)
 		apic_set_reg(apic, APIC_LVTT + 0x10 * i, APIC_LVT_MASKED);
 	apic_update_lvtt(apic);
-	if (!(vcpu->kvm->arch.disabled_quirks & KVM_QUIRK_LINT0_REENABLED))
+	if (kvm_check_has_quirk(vcpu->kvm, KVM_X86_QUIRK_LINT0_REENABLED))
 		apic_set_reg(apic, APIC_LVT0,
 			     SET_APIC_DELIVERY_MODE(0, APIC_MODE_EXTINT));
 	apic_manage_nmi_watchdog(apic, kvm_apic_get_reg(apic, APIC_LVT0));
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index f807496b62c2..44171462bd2a 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -2479,6 +2479,14 @@ static int mmu_need_write_protect(struct kvm_vcpu *vcpu, gfn_t gfn,
 	return 0;
 }
 
+static bool kvm_is_mmio_pfn(pfn_t pfn)
+{
+	if (pfn_valid(pfn))
+		return !is_zero_pfn(pfn) && PageReserved(pfn_to_page(pfn));
+
+	return true;
+}
+
 static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
 		    unsigned pte_access, int level,
 		    gfn_t gfn, pfn_t pfn, bool speculative,
@@ -2506,7 +2514,7 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
 		spte |= PT_PAGE_SIZE_MASK;
 	if (tdp_enabled)
 		spte |= kvm_x86_ops->get_mt_mask(vcpu, gfn,
-			kvm_is_reserved_pfn(pfn));
+			kvm_is_mmio_pfn(pfn));
 
 	if (host_writable)
 		spte |= SPTE_HOST_WRITEABLE;
diff --git a/arch/x86/kvm/mtrr.c b/arch/x86/kvm/mtrr.c
index de1d2d8062e2..dc0a84a6f309 100644
--- a/arch/x86/kvm/mtrr.c
+++ b/arch/x86/kvm/mtrr.c
@@ -120,6 +120,16 @@ static u8 mtrr_default_type(struct kvm_mtrr *mtrr_state)
 	return mtrr_state->deftype & IA32_MTRR_DEF_TYPE_TYPE_MASK;
 }
 
+static u8 mtrr_disabled_type(void)
+{
+	/*
+	 * Intel SDM 11.11.2.2: all MTRRs are disabled when
+	 * IA32_MTRR_DEF_TYPE.E bit is cleared, and the UC
+	 * memory type is applied to all of physical memory.
+	 */
+	return MTRR_TYPE_UNCACHABLE;
+}
+
 /*
 * Three terms are used in the following code:
 * - segment, it indicates the address segments covered by fixed MTRRs.
@@ -434,6 +444,8 @@ struct mtrr_iter {
 
 	/* output fields. */
 	int mem_type;
+	/* mtrr is completely disabled? */
+	bool mtrr_disabled;
 	/* [start, end) is not fully covered in MTRRs? */
 	bool partial_map;
 
@@ -549,7 +561,7 @@ static void mtrr_lookup_var_next(struct mtrr_iter *iter)
 static void mtrr_lookup_start(struct mtrr_iter *iter)
 {
 	if (!mtrr_is_enabled(iter->mtrr_state)) {
-		iter->partial_map = true;
+		iter->mtrr_disabled = true;
 		return;
 	}
 
@@ -563,6 +575,7 @@ static void mtrr_lookup_init(struct mtrr_iter *iter,
 	iter->mtrr_state = mtrr_state;
 	iter->start = start;
 	iter->end = end;
+	iter->mtrr_disabled = false;
 	iter->partial_map = false;
 	iter->fixed = false;
 	iter->range = NULL;
@@ -656,15 +669,19 @@ u8 kvm_mtrr_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn)
 		return MTRR_TYPE_WRBACK;
 	}
 
-	/* It is not covered by MTRRs. */
-	if (iter.partial_map) {
-		/*
-		 * We just check one page, partially covered by MTRRs is
-		 * impossible.
-		 */
-		WARN_ON(type != -1);
-		type = mtrr_default_type(mtrr_state);
-	}
+	if (iter.mtrr_disabled)
+		return mtrr_disabled_type();
+
+	/*
+	 * We just check one page, partially covered by MTRRs is
+	 * impossible.
+	 */
+	WARN_ON(iter.partial_map);
+
+	/* not contained in any MTRRs. */
+	if (type == -1)
+		return mtrr_default_type(mtrr_state);
+
 	return type;
 }
 EXPORT_SYMBOL_GPL(kvm_mtrr_get_guest_memory_type);
@@ -689,6 +706,9 @@ bool kvm_mtrr_check_gfn_range_consistency(struct kvm_vcpu *vcpu, gfn_t gfn,
 			return false;
 	}
 
+	if (iter.mtrr_disabled)
+		return true;
+
 	if (!iter.partial_map)
 		return true;
 
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 8dfbad7a2c44..a4b8c8d57b96 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -865,6 +865,64 @@ static void svm_disable_lbrv(struct vcpu_svm *svm)
 	set_msr_interception(msrpm, MSR_IA32_LASTINTTOIP, 0, 0);
 }
 
+#define MTRR_TYPE_UC_MINUS	7
+#define MTRR2PROTVAL_INVALID 0xff
+
+static u8 mtrr2protval[8];
+
+static u8 fallback_mtrr_type(int mtrr)
+{
+	/*
+	 * WT and WP aren't always available in the host PAT.  Treat
+	 * them as UC and UC- respectively.  Everything else should be
+	 * there.
+	 */
+	switch (mtrr)
+	{
+	case MTRR_TYPE_WRTHROUGH:
+		return MTRR_TYPE_UNCACHABLE;
+	case MTRR_TYPE_WRPROT:
+		return MTRR_TYPE_UC_MINUS;
+	default:
+		BUG();
+	}
+}
+
+static void build_mtrr2protval(void)
+{
+	int i;
+	u64 pat;
+
+	for (i = 0; i < 8; i++)
+		mtrr2protval[i] = MTRR2PROTVAL_INVALID;
+
+	/* Ignore the invalid MTRR types.  */
+	mtrr2protval[2] = 0;
+	mtrr2protval[3] = 0;
+
+	/*
+	 * Use host PAT value to figure out the mapping from guest MTRR
+	 * values to nested page table PAT/PCD/PWT values.  We do not
+	 * want to change the host PAT value every time we enter the
+	 * guest.
+	 */
+	rdmsrl(MSR_IA32_CR_PAT, pat);
+	for (i = 0; i < 8; i++) {
+		u8 mtrr = pat >> (8 * i);
+
+		if (mtrr2protval[mtrr] == MTRR2PROTVAL_INVALID)
+			mtrr2protval[mtrr] = __cm_idx2pte(i);
+	}
+
+	for (i = 0; i < 8; i++) {
+		if (mtrr2protval[i] == MTRR2PROTVAL_INVALID) {
+			u8 fallback = fallback_mtrr_type(i);
+			mtrr2protval[i] = mtrr2protval[fallback];
+			BUG_ON(mtrr2protval[i] == MTRR2PROTVAL_INVALID);
+		}
+	}
+}
+
 static __init int svm_hardware_setup(void)
 {
 	int cpu;
@@ -931,6 +989,7 @@ static __init int svm_hardware_setup(void)
 	} else
 		kvm_disable_tdp();
 
+	build_mtrr2protval();
 	return 0;
 
 err:
@@ -1085,6 +1144,39 @@ static u64 svm_compute_tsc_offset(struct kvm_vcpu *vcpu, u64 target_tsc)
 	return target_tsc - tsc;
 }
 
+static void svm_set_guest_pat(struct vcpu_svm *svm, u64 *g_pat)
+{
+	struct kvm_vcpu *vcpu = &svm->vcpu;
+
+	/* Unlike Intel, AMD takes the guest's CR0.CD into account.
+	 *
+	 * AMD does not have IPAT.  To emulate it for the case of guests
+	 * with no assigned devices, just set everything to WB.  If guests
+	 * have assigned devices, however, we cannot force WB for RAM
+	 * pages only, so use the guest PAT directly.
+	 */
+	if (!kvm_arch_has_assigned_device(vcpu->kvm))
+		*g_pat = 0x0606060606060606;
+	else
+		*g_pat = vcpu->arch.pat;
+}
+
+static u64 svm_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio)
+{
+	u8 mtrr;
+
+	/*
+	 * 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
+	 * 3. Passthrough: can't guarantee the result, try to trust guest.
+	 */
+	if (!is_mmio && !kvm_arch_has_assigned_device(vcpu->kvm))
+		return 0;
+
+	mtrr = kvm_mtrr_get_guest_memory_type(vcpu, gfn);
+	return mtrr2protval[mtrr];
+}
+
 static void init_vmcb(struct vcpu_svm *svm, bool init_event)
 {
 	struct vmcb_control_area *control = &svm->vmcb->control;
@@ -1180,6 +1272,7 @@ static void init_vmcb(struct vcpu_svm *svm, bool init_event)
 		clr_cr_intercept(svm, INTERCEPT_CR3_READ);
 		clr_cr_intercept(svm, INTERCEPT_CR3_WRITE);
 		save->g_pat = svm->vcpu.arch.pat;
+		svm_set_guest_pat(svm, &save->g_pat);
 		save->cr3 = 0;
 		save->cr4 = 0;
 	}
@@ -1579,7 +1672,7 @@ static void svm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
 	 * does not do it - this results in some delay at
 	 * reboot
 	 */
-	if (!(vcpu->kvm->arch.disabled_quirks & KVM_QUIRK_CD_NW_CLEARED))
+	if (kvm_check_has_quirk(vcpu->kvm, KVM_X86_QUIRK_CD_NW_CLEARED))
 		cr0 &= ~(X86_CR0_CD | X86_CR0_NW);
 	svm->vmcb->save.cr0 = cr0;
 	mark_dirty(svm->vmcb, VMCB_CR);
@@ -3254,6 +3347,16 @@ static int svm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr)
 	case MSR_VM_IGNNE:
 		vcpu_unimpl(vcpu, "unimplemented wrmsr: 0x%x data 0x%llx\n", ecx, data);
 		break;
+	case MSR_IA32_CR_PAT:
+		if (npt_enabled) {
+			if (!kvm_mtrr_valid(vcpu, MSR_IA32_CR_PAT, data))
+				return 1;
+			vcpu->arch.pat = data;
+			svm_set_guest_pat(svm, &svm->vmcb->save.g_pat);
+			mark_dirty(svm->vmcb, VMCB_NPT);
+			break;
+		}
+		/* fall through */
 	default:
 		return kvm_set_msr_common(vcpu, msr);
 	}
@@ -4088,11 +4191,6 @@ static bool svm_has_high_real_mode_segbase(void)
 	return true;
 }
 
-static u64 svm_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio)
-{
-	return 0;
-}
-
 static void svm_cpuid_update(struct kvm_vcpu *vcpu)
 {
 }
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 10d69a6df14f..af30c265412c 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -8632,22 +8632,17 @@ static u64 vmx_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio)
 	u64 ipat = 0;
 
 	/* 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:
 	 *   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
 	 *	VT-d engine can guarantee the cache correctness. Just set it
 	 *	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
 	 *    consistent with host MTRR
 	 */
-	if (is_mmio) {
-		cache = MTRR_TYPE_UNCACHABLE;
-		goto exit;
-	}
-
-	if (!kvm_arch_has_noncoherent_dma(vcpu->kvm)) {
+	if (!is_mmio && !kvm_arch_has_noncoherent_dma(vcpu->kvm)) {
 		ipat = VMX_EPT_IPAT_BIT;
 		cache = MTRR_TYPE_WRBACK;
 		goto exit;
@@ -8655,7 +8650,10 @@ static u64 vmx_get_mt_mask(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio)
 
 	if (kvm_read_cr0(vcpu) & X86_CR0_CD) {
 		ipat = VMX_EPT_IPAT_BIT;
-		cache = MTRR_TYPE_UNCACHABLE;
+		if (kvm_check_has_quirk(vcpu->kvm, KVM_X86_QUIRK_CD_NW_CLEARED))
+			cache = MTRR_TYPE_WRBACK;
+		else
+			cache = MTRR_TYPE_UNCACHABLE;
 		goto exit;
 	}
 
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 8d73ec8a2364..a51f94d3193d 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -3145,8 +3145,7 @@ static void load_xsave(struct kvm_vcpu *vcpu, u8 *src)
 			cpuid_count(XSTATE_CPUID, index,
 				    &size, &offset, &ecx, &edx);
 			memcpy(dest, src + offset, size);
-		} else
-			WARN_ON_ONCE(1);
+		}
 
 		valid -= feature;
 	}
@@ -7303,11 +7302,6 @@ struct kvm_vcpu *kvm_arch_vcpu_create(struct kvm *kvm,
 
 	vcpu = kvm_x86_ops->vcpu_create(kvm, id);
 
-	/*
-	 * Activate fpu unconditionally in case the guest needs eager FPU.  It will be
-	 * deactivated soon if it doesn't.
-	 */
-	kvm_x86_ops->fpu_activate(vcpu);
 	return vcpu;
 }
 
@@ -8206,6 +8200,24 @@ bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu)
 			kvm_x86_ops->interrupt_allowed(vcpu);
 }
 
+void kvm_arch_start_assignment(struct kvm *kvm)
+{
+	atomic_inc(&kvm->arch.assigned_device_count);
+}
+EXPORT_SYMBOL_GPL(kvm_arch_start_assignment);
+
+void kvm_arch_end_assignment(struct kvm *kvm)
+{
+	atomic_dec(&kvm->arch.assigned_device_count);
+}
+EXPORT_SYMBOL_GPL(kvm_arch_end_assignment);
+
+bool kvm_arch_has_assigned_device(struct kvm *kvm)
+{
+	return atomic_read(&kvm->arch.assigned_device_count);
+}
+EXPORT_SYMBOL_GPL(kvm_arch_has_assigned_device);
+
 void kvm_arch_register_noncoherent_dma(struct kvm *kvm)
 {
 	atomic_inc(&kvm->arch.noncoherent_dma_count);
diff --git a/arch/x86/kvm/x86.h b/arch/x86/kvm/x86.h
index edc8cdcd786b..0ca2f3e4803c 100644
--- a/arch/x86/kvm/x86.h
+++ b/arch/x86/kvm/x86.h
@@ -147,6 +147,11 @@ static inline void kvm_register_writel(struct kvm_vcpu *vcpu,
 	return kvm_register_write(vcpu, reg, val);
 }
 
+static inline bool kvm_check_has_quirk(struct kvm *kvm, u64 quirk)
+{
+	return !(kvm->arch.disabled_quirks & quirk);
+}
+
 void kvm_before_handle_nmi(struct kvm_vcpu *vcpu);
 void kvm_after_handle_nmi(struct kvm_vcpu *vcpu);
 void kvm_set_pending_timer(struct kvm_vcpu *vcpu);