From 10dc331ff5e7e4668c0f0c95b1a873aba9b70826 Mon Sep 17 00:00:00 2001 From: Xiao Guangrong Date: Thu, 16 Jul 2015 03:25:54 +0800 Subject: [PATCH 1/5] KVM: MTRR: fix memory type handling if MTRR is completely disabled Currently code uses default memory type if MTRR is fully disabled, fix it by using UC instead. Signed-off-by: Xiao Guangrong Tested-by: Alex Williamson Signed-off-by: Paolo Bonzini --- arch/x86/kvm/mtrr.c | 21 ++++++++++++++++++++- 1 file changed, 20 insertions(+), 1 deletion(-) diff --git a/arch/x86/kvm/mtrr.c b/arch/x86/kvm/mtrr.c index de1d2d8062e2..e2750134a22b 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; @@ -656,6 +669,9 @@ u8 kvm_mtrr_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn) return MTRR_TYPE_WRBACK; } + if (iter.mtrr_disabled) + return mtrr_disabled_type(); + /* It is not covered by MTRRs. */ if (iter.partial_map) { /* @@ -689,6 +705,9 @@ 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; From 3e5d2fdceda172554e681b68c853bf5d08205bbf Mon Sep 17 00:00:00 2001 From: Xiao Guangrong Date: Thu, 16 Jul 2015 03:25:55 +0800 Subject: [PATCH 2/5] KVM: MTRR: simplify kvm_mtrr_get_guest_memory_type kvm_mtrr_get_guest_memory_type never returns -1 which is implied in the current code since if @type = -1 (means no MTRR contains the range), iter.partial_map must be true Simplify the code to indicate this fact Signed-off-by: Xiao Guangrong Tested-by: Alex Williamson Signed-off-by: Paolo Bonzini --- arch/x86/kvm/mtrr.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) diff --git a/arch/x86/kvm/mtrr.c b/arch/x86/kvm/mtrr.c index e2750134a22b..dc0a84a6f309 100644 --- a/arch/x86/kvm/mtrr.c +++ b/arch/x86/kvm/mtrr.c @@ -672,15 +672,16 @@ u8 kvm_mtrr_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn) if (iter.mtrr_disabled) return mtrr_disabled_type(); - /* 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); - } + /* + * 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); From 41dbc6bcd9722bba8f09256655d060af30c63d34 Mon Sep 17 00:00:00 2001 From: Paolo Bonzini Date: Thu, 23 Jul 2015 08:22:45 +0200 Subject: [PATCH 3/5] KVM: x86: introduce kvm_check_has_quirk The logic of the disabled_quirks field usually results in a double negation. Wrap it in a simple function that checks the bit and negates it. Based on a patch from Xiao Guangrong. Signed-off-by: Paolo Bonzini --- arch/x86/kvm/lapic.c | 2 +- arch/x86/kvm/svm.c | 2 +- arch/x86/kvm/x86.h | 5 +++++ 3 files changed, 7 insertions(+), 2 deletions(-) diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c index 954e98a8c2e3..1c425443a41a 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_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/svm.c b/arch/x86/kvm/svm.c index bbc678a66b18..8cbec765b08d 100644 --- a/arch/x86/kvm/svm.c +++ b/arch/x86/kvm/svm.c @@ -1672,7 +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_QUIRK_CD_NW_CLEARED)) cr0 &= ~(X86_CR0_CD | X86_CR0_NW); svm->vmcb->save.cr0 = cr0; mark_dirty(svm->vmcb, VMCB_CR); 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); From fb279950ba02e3210a16b11ecfa8871f3ee0ca49 Mon Sep 17 00:00:00 2001 From: Xiao Guangrong Date: Thu, 16 Jul 2015 03:25:56 +0800 Subject: [PATCH 4/5] KVM: vmx: obey KVM_QUIRK_CD_NW_CLEARED OVMF depends on WB to boot fast, because it only clears caches after it has set up MTRRs---which is too late. Let's do writeback if CR0.CD is set to make it happy, similar to what SVM is already doing. Signed-off-by: Xiao Guangrong Tested-by: Alex Williamson Signed-off-by: Paolo Bonzini --- arch/x86/kvm/vmx.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c index 5b4e9384717a..e9e1fc517ab5 100644 --- a/arch/x86/kvm/vmx.c +++ b/arch/x86/kvm/vmx.c @@ -8650,7 +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_QUIRK_CD_NW_CLEARED)) + cache = MTRR_TYPE_WRBACK; + else + cache = MTRR_TYPE_UNCACHABLE; goto exit; } From 0da029ed7ee5fdf49a2a0e14160c3e9999be9292 Mon Sep 17 00:00:00 2001 From: Paolo Bonzini Date: Thu, 23 Jul 2015 08:24:42 +0200 Subject: [PATCH 5/5] KVM: x86: rename quirk constants to KVM_X86_QUIRK_* Make them clearly architecture-dependent; the capability is valid for all architectures, but the argument is not. Signed-off-by: Paolo Bonzini --- arch/x86/include/uapi/asm/kvm.h | 4 ++-- arch/x86/kvm/lapic.c | 2 +- arch/x86/kvm/svm.c | 2 +- arch/x86/kvm/vmx.c | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/arch/x86/include/uapi/asm/kvm.h b/arch/x86/include/uapi/asm/kvm.h index a4ae82eb82aa..cd54147cb365 100644 --- a/arch/x86/include/uapi/asm/kvm.h +++ b/arch/x86/include/uapi/asm/kvm.h @@ -354,7 +354,7 @@ struct kvm_xcrs { struct kvm_sync_regs { }; -#define KVM_QUIRK_LINT0_REENABLED (1 << 0) -#define KVM_QUIRK_CD_NW_CLEARED (1 << 1) +#define KVM_X86_QUIRK_LINT0_REENABLED (1 << 0) +#define KVM_X86_QUIRK_CD_NW_CLEARED (1 << 1) #endif /* _ASM_X86_KVM_H */ diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c index 1c425443a41a..2a5ca97c263b 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 (kvm_check_has_quirk(vcpu->kvm, 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/svm.c b/arch/x86/kvm/svm.c index 8cbec765b08d..8e0c0844c6b9 100644 --- a/arch/x86/kvm/svm.c +++ b/arch/x86/kvm/svm.c @@ -1672,7 +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 (kvm_check_has_quirk(vcpu->kvm, 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); diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c index e9e1fc517ab5..83b7b5cd75d5 100644 --- a/arch/x86/kvm/vmx.c +++ b/arch/x86/kvm/vmx.c @@ -8650,7 +8650,7 @@ 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; - if (kvm_check_has_quirk(vcpu->kvm, KVM_QUIRK_CD_NW_CLEARED)) + if (kvm_check_has_quirk(vcpu->kvm, KVM_X86_QUIRK_CD_NW_CLEARED)) cache = MTRR_TYPE_WRBACK; else cache = MTRR_TYPE_UNCACHABLE;