KVM: x86/mmu: nVMX: 5-level paging cleanup and enabling
v3 cover KVM: x86/mmu: nVMX: 5-level paging cleanup and enabling
Clean up MMU code related to 5 level paging, expose 5-level EPT to L1, and additional clean up on top (mostly renames of functions/variables that caused me no end of confusion when trying to figure out what was broken at various times).
v3:
- Dropped fixes for existing 5-level bugs (merged for 5.6).
- Use get_guest_pgd() instead of get_guest_cr3_or_eptp(). [Paolo]
- Add patches to fix MMU role calculation to play nice with 5-level paging without requiring additional CR4.LA_57 bit.
v2:
- Increase the nested EPT array sizes to accomodate 5-level paging in the patch that adds support for 5-level nested EPT, not in the bug fix for 5-level shadow paging.
Sean Christopherson (7):
KVM: x86/mmu: Don't drop level/direct from MMU role calculation
KVM: x86/mmu: Drop kvm_mmu_extended_role.cr4_la57 hack
KVM: nVMX: Allow L1 to use 5-level page walks for nested EPT
KVM: nVMX: Rename nested_ept_get_cr3() to nested_ept_get_eptp()
KVM: nVMX: Rename EPTP validity helper and associated variables
KVM: x86/mmu: Rename kvm_mmu->get_cr3() to ->get_guest_pgd()
KVM: nVMX: Drop unnecessary check on ept caps for execute-only
arch/x86/include/asm/kvm_host.h | 3 +- arch/x86/include/asm/vmx.h | 12 +++++++ arch/x86/kvm/mmu/mmu.c | 59 +++++++++++++++++---------------- arch/x86/kvm/mmu/paging_tmpl.h | 4 +-- arch/x86/kvm/svm.c | 2 +- arch/x86/kvm/vmx/nested.c | 52 ++++++++++++++++++----------- arch/x86/kvm/vmx/nested.h | 4 +-- arch/x86/kvm/vmx/vmx.c | 3 +- arch/x86/kvm/x86.c | 2 +- 9 files changed, 82 insertions(+), 59 deletions(-)
Signature · 1 line
2.24.1
PATCH v3 1/7 KVM: x86/mmu: Don't drop level/direct from MMU role calculation
Use the calculated role as-is when propagating it to kvm_mmu.mmu_role, i.e. stop masking off meaningful fields. The concept of masking off fields came from kvm_mmu_pte_write(), which (correctly) ignores certain fields when comparing kvm_mmu_page.role against kvm_mmu.mmu_role, e.g. the current mmu's access and level have no relation to a shadow page's access and level.
Masking off the level causes problems for 5-level paging, e.g. CR4.LA57 has its own redundant flag in the extended role, and nested EPT would need a similar hack to support 5-level paging for L2.
Opportunistically rework the mask for kvm_mmu_pte_write() to define the fields that should be ignored as opposed to the fields that should be checked, i.e. make it opt-out instead of opt-in so that new fields are automatically picked up. While doing so, stop ignoring "direct". The field is effectively ignored anyways because kvm_mmu_pte_write() is only reached with an indirect mmu and the loop only walks indirect shadow pages, but double checking "direct" literally costs nothing.
--- arch/x86/kvm/mmu/mmu.c | 35 ++++++++++++++++++----------------- 1 file changed, 18 insertions(+), 17 deletions(-)
arch/x86/kvm/mmu/mmu.c +18 -17
@@ -215,17 +215,6 @@ struct kvm_shadow_walk_iterator {unsigned index; }; -static const union kvm_mmu_page_role mmu_base_role_mask = { - .cr0_wp =1 , - .gpte_is_8_bytes =1 , - .nxe =1 , - .smep_andnot_wp =1 , - .smap_andnot_wp =1 , - .smm =1 , - .guest_mode =1 , - .ad_disabled =1 , -}; -#define for_each_shadow_entry_using_root (_vcpu, _root, _addr, _walker) \ for (shadow_walk_init_using_root(&(_walker), (_vcpu), \ (_root), (_addr)); \
@@ -4919,7 +4908,6 @@ static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu)union kvm_mmu_role new_role =kvm_calc_tdp_mmu_root_page_role (vcpu, false); - new_role.base.word &= mmu_base_role_mask.word;if (new_role.as_u64 == context->mmu_role.as_u64)return ;
@@ -4991,7 +4979,6 @@ void kvm_init_shadow_mmu(struct kvm_vcpu *vcpu)union kvm_mmu_role new_role =kvm_calc_shadow_mmu_root_page_role (vcpu, false); - new_role.base.word &= mmu_base_role_mask.word;if (new_role.as_u64 == context->mmu_role.as_u64)return ;
@@ -5048,7 +5035,6 @@ void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly,__kvm_mmu_new_cr3 (vcpu, new_eptp, new_role.base, false); - new_role.base.word &= mmu_base_role_mask.word;if (new_role.as_u64 == context->mmu_role.as_u64)return ;
@@ -5089,7 +5075,6 @@ static void init_kvm_nested_mmu(struct kvm_vcpu *vcpu)union kvm_mmu_role new_role =kvm_calc_mmu_role_common (vcpu, false);struct kvm_mmu *g_context = &vcpu->arch.nested_mmu; - new_role.base.word &= mmu_base_role_mask.word;if (new_role.as_u64 == g_context->mmu_role.as_u64)return ;
@@ -5328,6 +5313,22 @@ static u64 *get_written_sptes(struct kvm_mmu_page *sp, gpa_t gpa, int *nspte)return spte; } +/* +* Ignore various flags when determining if a SPTE can be immediately +* overwritten for the current MMU. +* - level: explicitly checked in mmu_pte_write_new_pte(), and will never +* match the current MMU role, as MMU's level tracks the root level. +* - access: updated based on the new guest PTE +* - quadrant: handled by get_written_sptes() +* - invalid: always false (loop only walks valid shadow pages) +*/ +static const union kvm_mmu_page_role role_ign = { + .level =0xf , + .access =0x7 , + .quadrant =0x3 , + .invalid =0x1 , +}; +static void kvm_mmu_pte_write (struct kvm_vcpu *vcpu,gpa_t gpa,const u8 *new,int bytes,struct kvm_page_track_notifier_node *node)
@@ -5383,8 +5384,8 @@ static void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, entry = *spte;mmu_page_zap_pte (vcpu->kvm, sp, spte);if (gentry && - !((sp->role.word ^ base_role) - & mmu_base_role_mask.word) &&rmap_can_add (vcpu)) + !((sp->role.word ^ base_role) & ~role_ign.word) && +rmap_can_add (vcpu))mmu_pte_write_new_pte (vcpu, sp, spte, &gentry);if (need_remote_flush (entry, *spte)) remote_flush = true;
Signature · 1 line
2.24.1
PATCH v3 2/7 KVM: x86/mmu: Drop kvm_mmu_extended_role.cr4_la57 hack
In reply to Sean ChristophersonDrop kvm_mmu_extended_role.cr4_la57 now that mmu_role doesn't mask off level, which already incorporates the guest's CR4.LA57 for a shadow MMU by querying is_la57_mode().
--- arch/x86/include/asm/kvm_host.h | 1 - arch/x86/kvm/mmu/mmu.c | 1 - 2 files changed, 2 deletions(-)
arch/x86/include/asm/kvm_host.h +0 -1
@@ -297,7 +297,6 @@ union kvm_mmu_extended_role { unsigned int cr4_pke:1; unsigned int cr4_smap:1; unsigned int cr4_smep:1; - unsigned int cr4_la57:1; unsigned int maxphyaddr:6; }; };
arch/x86/kvm/mmu/mmu.c +0 -1
@@ -4862,7 +4862,6 @@ static union kvm_mmu_extended_role kvm_calc_mmu_role_ext(struct kvm_vcpu *vcpu) ext.cr4_smap = !!kvm_read_cr4_bits (vcpu,X86_CR4_SMAP ); ext.cr4_pse = !!is_pse (vcpu); ext.cr4_pke = !!kvm_read_cr4_bits (vcpu,X86_CR4_PKE ); - ext.cr4_la57 = !!kvm_read_cr4_bits (vcpu,X86_CR4_LA57 ); ext.maxphyaddr =cpuid_maxphyaddr (vcpu); ext.valid =1 ;
Signature · 1 line
2.24.1
PATCH v3 3/7 KVM: nVMX: Allow L1 to use 5-level page walks for nested EPT
In reply to Sean ChristophersonAdd support for 5-level nested EPT, and advertise said support in the EPT capabilities MSR. KVM's MMU can already handle 5-level legacy page tables, there's no reason to force an L1 VMM to use shadow paging if it wants to employ 5-level page tables.
--- arch/x86/include/asm/vmx.h | 12 ++++++++++++ arch/x86/kvm/mmu/mmu.c | 11 ++++++----- arch/x86/kvm/mmu/paging_tmpl.h | 2 +- arch/x86/kvm/vmx/nested.c | 21 +++++++++++++++++---- arch/x86/kvm/vmx/vmx.c | 3 +-- 5 files changed, 37 insertions(+), 12 deletions(-)
arch/x86/include/asm/vmx.h +12 -0
@@ -500,6 +500,18 @@ enum vmcs_field { VMX_EPT_EXECUTABLE_MASK)#define VMX_EPT_MT_MASK (7ull << VMX_EPT_MT_EPTE_SHIFT) +static inline u8 vmx_eptp_page_walk_level (u64 eptp) +{ +u64 encoded_level = eptp &VMX_EPTP_PWL_MASK ; + +if (encoded_level ==VMX_EPTP_PWL_5 ) +return 5 ; + +/* @eptp must be pre-validated by the caller. */ +WARN_ON_ONCE (encoded_level !=VMX_EPTP_PWL_4 ); +return 4 ; +} +/* The mask to use to trigger an EPT Misconfiguration in order to track MMIO */ #define VMX_EPT_MISCONFIG_WX_VALUE (VMX_EPT_WRITABLE_MASK | \ VMX_EPT_EXECUTABLE_MASK)
arch/x86/kvm/mmu/mmu.c +6 -5
@@ -4997,14 +4997,14 @@ EXPORT_SYMBOL_GPL(kvm_init_shadow_mmu);static union kvm_mmu_role kvm_calc_shadow_ept_root_page_role (struct kvm_vcpu *vcpu,bool accessed_dirty, -bool execonly) +bool execonly,u8 level) {union kvm_mmu_role role = {0 };/* SMM flag is inherited from root_mmu */ role.base.smm = vcpu->arch.root_mmu.mmu_role.base.smm; - role.base.level =PT64_ROOT_4LEVEL ; + role.base.level = level; role.base.gpte_is_8_bytes = true; role.base.direct = false; role.base.ad_disabled = !accessed_dirty;
@@ -5028,16 +5028,17 @@ void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly, bool accessed_dirty, gpa_t new_eptp) {struct kvm_mmu *context = vcpu->arch.mmu; +u8 level =vmx_eptp_page_walk_level (new_eptp);union kvm_mmu_role new_role =kvm_calc_shadow_ept_root_page_role (vcpu, accessed_dirty, - execonly); + execonly, level);__kvm_mmu_new_cr3 (vcpu, new_eptp, new_role.base, false);if (new_role.as_u64 == context->mmu_role.as_u64)return ; - context->shadow_root_level =PT64_ROOT_4LEVEL ; + context->shadow_root_level = level; context->nx = true; context->ept_ad = accessed_dirty;
@@ -5046,7 +5047,7 @@ void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly, context->sync_page = ept_sync_page; context->invlpg = ept_invlpg; context->update_pte = ept_update_pte; - context->root_level =PT64_ROOT_4LEVEL ; + context->root_level = level; context->direct_map = false; context->mmu_role.as_u64 = new_role.as_u64;
arch/x86/kvm/mmu/paging_tmpl.h +1 -1
@@ -66,7 +66,7 @@#define PT_GUEST_ACCESSED_SHIFT 8#define PT_HAVE_ACCESSED_DIRTY (mmu) ((mmu)->ept_ad)#define CMPXCHG cmpxchg64 -#define PT_MAX_FULL_LEVELS 4 +#define PT_MAX_FULL_LEVELS PT64_ROOT_MAX_LEVEL#else #error Invalid PTTYPE value#endif
arch/x86/kvm/vmx/nested.c +17 -4
@@ -2582,9 +2582,19 @@ static bool valid_ept_address(struct kvm_vcpu *vcpu, u64 address)return false; } -/* only 4 levels page-walk length are valid */ -if (CC ((address &VMX_EPTP_PWL_MASK ) !=VMX_EPTP_PWL_4 )) +/* Page-walk levels validity. */ +switch (address &VMX_EPTP_PWL_MASK ) { +case VMX_EPTP_PWL_5 : +if (CC (!(vmx->nested.msrs.ept_caps &VMX_EPT_PAGE_WALK_5_BIT ))) +return false; +break ; +case VMX_EPTP_PWL_4 : +if (CC (!(vmx->nested.msrs.ept_caps &VMX_EPT_PAGE_WALK_4_BIT ))) +return false; +break ; +default :return false; + }/* Reserved bits should not be set */ if (CC (address >> maxphyaddr || ((address >>7 ) &0x1f )))
@@ -6129,8 +6139,11 @@ void nested_vmx_setup_ctls_msrs(struct nested_vmx_msrs *msrs, u32 ept_caps)/* nested EPT: emulate EPT also to L1 */ msrs->secondary_ctls_high |=SECONDARY_EXEC_ENABLE_EPT ; - msrs->ept_caps =VMX_EPT_PAGE_WALK_4_BIT | -VMX_EPTP_WB_BIT |VMX_EPT_INVEPT_BIT ; + msrs->ept_caps = +VMX_EPT_PAGE_WALK_4_BIT | +VMX_EPT_PAGE_WALK_5_BIT | +VMX_EPTP_WB_BIT | +VMX_EPT_INVEPT_BIT ;if (cpu_has_vmx_ept_execute_only ()) msrs->ept_caps |=VMX_EPT_EXECUTE_ONLY_BIT ;
arch/x86/kvm/vmx/vmx.c +1 -2
@@ -2985,9 +2985,8 @@ void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)static int get_ept_level (struct kvm_vcpu *vcpu) { -/* Nested EPT currently only supports 4-level walks. */ if (is_guest_mode (vcpu) &&nested_cpu_has_ept (get_vmcs12 (vcpu))) -return 4 ; +return vmx_eptp_page_walk_level (nested_ept_get_cr3 (vcpu));if (cpu_has_vmx_ept_5levels () && (cpuid_maxphyaddr (vcpu) >48 ))return 5 ;return 4 ;
Signature · 1 line
2.24.1
PATCH v3 4/7 KVM: nVMX: Rename nested_ept_get_cr3() to nested_ept_get_eptp()
In reply to Sean ChristophersonRename the accessor for vmcs12.EPTP to use "eptp" instead of "cr3". The accessor has no relation to cr3 whatsoever, other than it being assigned to the also poorly named kvm_mmu->get_cr3() hook.
No functional change intended.
--- arch/x86/kvm/vmx/nested.c | 4 ++-- arch/x86/kvm/vmx/nested.h | 4 ++-- arch/x86/kvm/vmx/vmx.c | 2 +- 3 files changed, 5 insertions(+), 5 deletions(-)
arch/x86/kvm/vmx/nested.c +2 -2
@@ -353,9 +353,9 @@ static void nested_ept_init_mmu_context(struct kvm_vcpu *vcpu)to_vmx (vcpu)->nested.msrs.ept_caps &VMX_EPT_EXECUTE_ONLY_BIT ,nested_ept_ad_enabled (vcpu), - nested_ept_get_cr3(vcpu)); + nested_ept_get_eptp(vcpu)); vcpu->arch.mmu->set_cr3 = vmx_set_cr3; - vcpu->arch.mmu->get_cr3 = nested_ept_get_cr3; + vcpu->arch.mmu->get_cr3 = nested_ept_get_eptp; vcpu->arch.mmu->inject_page_fault = nested_ept_inject_page_fault; vcpu->arch.mmu->get_pdptr = kvm_pdptr_read;
arch/x86/kvm/vmx/nested.h +2 -2
@@ -60,7 +60,7 @@ static inline int vmx_has_valid_vmcs12(struct kvm_vcpu *vcpu) vmx->nested.hv_evmcs; } -static inline unsigned long nested_ept_get_cr3 (struct kvm_vcpu *vcpu) +static inline unsigned long nested_ept_get_eptp (struct kvm_vcpu *vcpu) {/* return the page table to be shadowed - in our case, EPT12 */ return get_vmcs12 (vcpu)->ept_pointer;
@@ -68,7 +68,7 @@ static inline unsigned long nested_ept_get_cr3(struct kvm_vcpu *vcpu)static inline bool nested_ept_ad_enabled (struct kvm_vcpu *vcpu) { -return nested_ept_get_cr3 (vcpu) &VMX_EPTP_AD_ENABLE_BIT ; +return nested_ept_get_eptp (vcpu) &VMX_EPTP_AD_ENABLE_BIT ; }/*
arch/x86/kvm/vmx/vmx.c +1 -1
@@ -2986,7 +2986,7 @@ void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)static int get_ept_level (struct kvm_vcpu *vcpu) {if (is_guest_mode (vcpu) &&nested_cpu_has_ept (get_vmcs12 (vcpu))) -return vmx_eptp_page_walk_level (nested_ept_get_cr3 (vcpu)); +return vmx_eptp_page_walk_level (nested_ept_get_eptp (vcpu));if (cpu_has_vmx_ept_5levels () && (cpuid_maxphyaddr (vcpu) >48 ))return 5 ;return 4 ;
Signature · 1 line
2.24.1
PATCH v3 5/7 KVM: nVMX: Rename EPTP validity helper and associated variables
In reply to Sean ChristophersonRename valid_ept_address() to nested_vmx_check_eptp() to follow the nVMX nomenclature and to reflect that the function now checks a lot more than just the address contained in the EPTP. Rename address to new_eptp in associated code.
No functional change intended.
--- arch/x86/kvm/vmx/nested.c | 24 ++++++++++++------------ 1 file changed, 12 insertions(+), 12 deletions(-)
arch/x86/kvm/vmx/nested.c +12 -12
@@ -2563,13 +2563,13 @@ static int nested_vmx_check_nmi_controls(struct vmcs12 *vmcs12)return 0 ; } -static bool valid_ept_address (struct kvm_vcpu *vcpu,u64 address) +static bool nested_vmx_check_eptp (struct kvm_vcpu *vcpu,u64 new_eptp) {struct vcpu_vmx *vmx =to_vmx (vcpu);int maxphyaddr =cpuid_maxphyaddr (vcpu);/* Check for memory type validity */ -switch (address &VMX_EPTP_MT_MASK ) { +switch (new_eptp &VMX_EPTP_MT_MASK ) {case VMX_EPTP_MT_UC :if (CC (!(vmx->nested.msrs.ept_caps &VMX_EPTP_UC_BIT )))return false;
@@ -2583,7 +2583,7 @@ static bool valid_ept_address(struct kvm_vcpu *vcpu, u64 address) }/* Page-walk levels validity. */ -switch (address &VMX_EPTP_PWL_MASK ) { +switch (new_eptp &VMX_EPTP_PWL_MASK ) {case VMX_EPTP_PWL_5 :if (CC (!(vmx->nested.msrs.ept_caps &VMX_EPT_PAGE_WALK_5_BIT )))return false;
@@ -2597,11 +2597,11 @@ static bool valid_ept_address(struct kvm_vcpu *vcpu, u64 address) }/* Reserved bits should not be set */ -if (CC (address >> maxphyaddr || ((address >>7 ) &0x1f ))) +if (CC (new_eptp >> maxphyaddr || ((new_eptp >>7 ) &0x1f )))return false;/* AD, if set, should be supported */ -if (address &VMX_EPTP_AD_ENABLE_BIT ) { +if (new_eptp &VMX_EPTP_AD_ENABLE_BIT ) {if (CC (!(vmx->nested.msrs.ept_caps &VMX_EPT_AD_BIT )))return false; }
@@ -2650,7 +2650,7 @@ static int nested_check_vm_execution_controls(struct kvm_vcpu *vcpu,return -EINVAL ;if (nested_cpu_has_ept (vmcs12) && -CC (!valid_ept_address (vcpu, vmcs12->ept_pointer))) +CC (!nested_vmx_check_eptp (vcpu, vmcs12->ept_pointer)))return -EINVAL ;if (nested_cpu_has_vmfunc (vmcs12)) {
@@ -5244,7 +5244,7 @@ static int nested_vmx_eptp_switching(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12) {u32 index =kvm_rcx_read (vcpu); -u64 address; +u64 new_eptp;bool accessed_dirty;struct kvm_mmu *mmu = vcpu->arch.walk_mmu;
@@ -5257,23 +5257,23 @@ static int nested_vmx_eptp_switching(struct kvm_vcpu *vcpu,if (kvm_vcpu_read_guest_page (vcpu, vmcs12->eptp_list_address >>PAGE_SHIFT , - &address, index *8 ,8 )) + &new_eptp, index *8 ,8 ))return 1 ; - accessed_dirty = !!(address &VMX_EPTP_AD_ENABLE_BIT ); + accessed_dirty = !!(new_eptp &VMX_EPTP_AD_ENABLE_BIT );/* * If the (L2) guest does a vmfunc to the currently * active ept pointer, we don't have to do anything else */ -if (vmcs12->ept_pointer != address) { -if (!valid_ept_address (vcpu, address)) +if (vmcs12->ept_pointer != new_eptp) { +if (!nested_vmx_check_eptp (vcpu, new_eptp))return 1 ;kvm_mmu_unload (vcpu); mmu->ept_ad = accessed_dirty; mmu->mmu_role.base.ad_disabled = !accessed_dirty; - vmcs12->ept_pointer = address; + vmcs12->ept_pointer = new_eptp;/* * TODO: Check what's the correct approach in case * mmu reload fails. Currently, we just let the next
Signature · 1 line
2.24.1
PATCH v3 6/7 KVM: x86/mmu: Rename kvm_mmu->get_cr3() to ->get_guest_pgd()
In reply to Sean ChristophersonRename kvm_mmu->get_cr3() to call out that it is retrieving a guest value, as opposed to kvm_mmu->set_cr3(), which sets a host value, and to note that it will return something other than CR3 when nested EPT is in use. Hopefully the new name will also make it more obvious that L1's nested_cr3 is returned in SVM's nested NPT case.
No functional change intended.
--- arch/x86/include/asm/kvm_host.h | 2 +- arch/x86/kvm/mmu/mmu.c | 12 ++++++------ arch/x86/kvm/mmu/paging_tmpl.h | 2 +- arch/x86/kvm/svm.c | 2 +- arch/x86/kvm/vmx/nested.c | 2 +- arch/x86/kvm/x86.c | 2 +- 6 files changed, 11 insertions(+), 11 deletions(-)
arch/x86/include/asm/kvm_host.h +1 -1
@@ -382,7 +382,7 @@ struct kvm_mmu_root_info {*/ struct kvm_mmu {void (*set_cr3)(struct kvm_vcpu *vcpu,unsigned long root); -unsigned long (*get_cr3)(struct kvm_vcpu *vcpu); +unsigned long (*get_guest_pgd)(struct kvm_vcpu *vcpu);u64 (*get_pdptr)(struct kvm_vcpu *vcpu,int index);int (*page_fault)(struct kvm_vcpu *vcpu,gpa_t cr2_or_gpa,u32 err,bool prefault);
arch/x86/kvm/mmu/mmu.c +6 -6
@@ -3719,7 +3719,7 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu) vcpu->arch.mmu->root_hpa =__pa (vcpu->arch.mmu->pae_root); }else BUG (); - vcpu->arch.mmu->root_cr3 = vcpu->arch.mmu->get_cr3 (vcpu); + vcpu->arch.mmu->root_cr3 = vcpu->arch.mmu->get_guest_pgd (vcpu);return 0 ; }
@@ -3731,7 +3731,7 @@ static int mmu_alloc_shadow_roots(struct kvm_vcpu *vcpu)gfn_t root_gfn, root_cr3;int i; - root_cr3 = vcpu->arch.mmu->get_cr3 (vcpu); + root_cr3 = vcpu->arch.mmu->get_guest_pgd (vcpu); root_gfn = root_cr3 >>PAGE_SHIFT ;if (mmu_check_root (vcpu, root_gfn))
@@ -4068,7 +4068,7 @@ static int kvm_arch_setup_async_pf(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, arch.token = (vcpu->arch.apf.id++ <<12 ) | vcpu->vcpu_id; arch.gfn = gfn; arch.direct_map = vcpu->arch.mmu->direct_map; - arch.cr3 = vcpu->arch.mmu->get_cr3 (vcpu); + arch.cr3 = vcpu->arch.mmu->get_guest_pgd (vcpu);return kvm_setup_async_pf (vcpu, cr2_or_gpa,kvm_vcpu_gfn_to_hva (vcpu, gfn), &arch);
@@ -4918,7 +4918,7 @@ static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu) context->shadow_root_level = kvm_x86_ops->get_tdp_level (vcpu); context->direct_map = true; context->set_cr3 = kvm_x86_ops->set_tdp_cr3; - context->get_cr3 = get_cr3; + context->get_guest_pgd = get_cr3; context->get_pdptr = kvm_pdptr_read; context->inject_page_fault = kvm_inject_page_fault;
@@ -5065,7 +5065,7 @@ static void init_kvm_softmmu(struct kvm_vcpu *vcpu)kvm_init_shadow_mmu (vcpu); context->set_cr3 = kvm_x86_ops->set_cr3; - context->get_cr3 = get_cr3; + context->get_guest_pgd = get_cr3; context->get_pdptr = kvm_pdptr_read; context->inject_page_fault = kvm_inject_page_fault; }
@@ -5079,7 +5079,7 @@ static void init_kvm_nested_mmu(struct kvm_vcpu *vcpu)return ; g_context->mmu_role.as_u64 = new_role.as_u64; - g_context->get_cr3 = get_cr3; + g_context->get_guest_pgd = get_cr3; g_context->get_pdptr = kvm_pdptr_read; g_context->inject_page_fault = kvm_inject_page_fault;
arch/x86/kvm/mmu/paging_tmpl.h +1 -1
@@ -333,7 +333,7 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,trace_kvm_mmu_pagetable_walk (addr, access); retry_walk: walker->level = mmu->root_level; - pte = mmu->get_cr3 (vcpu); + pte = mmu->get_guest_pgd (vcpu); have_ad =PT_HAVE_ACCESSED_DIRTY (mmu);#if PTTYPE ==64
arch/x86/kvm/svm.c +1 -1
@@ -3012,7 +3012,7 @@ static void nested_svm_init_mmu_context(struct kvm_vcpu *vcpu) vcpu->arch.mmu = &vcpu->arch.guest_mmu;kvm_init_shadow_mmu (vcpu); vcpu->arch.mmu->set_cr3 = nested_svm_set_tdp_cr3; - vcpu->arch.mmu->get_cr3 = nested_svm_get_tdp_cr3; + vcpu->arch.mmu->get_guest_pgd = nested_svm_get_tdp_cr3; vcpu->arch.mmu->get_pdptr = nested_svm_get_tdp_pdptr; vcpu->arch.mmu->inject_page_fault = nested_svm_inject_npf_exit; vcpu->arch.mmu->shadow_root_level =get_npt_level (vcpu);
arch/x86/kvm/vmx/nested.c +1 -1
@@ -355,7 +355,7 @@ static void nested_ept_init_mmu_context(struct kvm_vcpu *vcpu)nested_ept_ad_enabled (vcpu), nested_ept_get_eptp(vcpu)); vcpu->arch.mmu->set_cr3 = vmx_set_cr3; - vcpu->arch.mmu->get_cr3 = nested_ept_get_eptp; + vcpu->arch.mmu->get_guest_pgd = nested_ept_get_eptp; vcpu->arch.mmu->inject_page_fault = nested_ept_inject_page_fault; vcpu->arch.mmu->get_pdptr = kvm_pdptr_read;
arch/x86/kvm/x86.c +1 -1
@@ -10153,7 +10153,7 @@ void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu, struct kvm_async_pf *work)return ;if (!vcpu->arch.mmu->direct_map && - work->arch.cr3 != vcpu->arch.mmu->get_cr3 (vcpu)) + work->arch.cr3 != vcpu->arch.mmu->get_guest_pgd (vcpu))return ;kvm_mmu_do_page_fault (vcpu, work->cr2_or_gpa,0 , true);
Signature · 1 line
2.24.1
PATCH v3 7/7 KVM: nVMX: Drop unnecessary check on ept caps for execute-only
In reply to Sean ChristophersonDrop the call to cpu_has_vmx_ept_execute_only() when calculating which EPT capabilities will be exposed to L1 for nested EPT. The resulting configuration is immediately sanitized by the passed in @ept_caps, and except for the call from vmx_check_processor_compat(), @ept_caps is the capabilities that are queried by cpu_has_vmx_ept_execute_only(). For vmx_check_processor_compat(), KVM *wants* to ignore vmx_capability.ept so that a divergence in EPT capabilities between CPUs is detected.
--- arch/x86/kvm/vmx/nested.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-)
arch/x86/kvm/vmx/nested.c +3 -4
@@ -6143,10 +6143,9 @@ void nested_vmx_setup_ctls_msrs(struct nested_vmx_msrs *msrs, u32 ept_caps)VMX_EPT_PAGE_WALK_4_BIT |VMX_EPT_PAGE_WALK_5_BIT |VMX_EPTP_WB_BIT | -VMX_EPT_INVEPT_BIT ; -if (cpu_has_vmx_ept_execute_only ()) - msrs->ept_caps |= -VMX_EPT_EXECUTE_ONLY_BIT ; +VMX_EPT_INVEPT_BIT | +VMX_EPT_EXECUTE_ONLY_BIT ; + msrs->ept_caps &= ept_caps; msrs->ept_caps |=VMX_EPT_EXTENT_GLOBAL_BIT | VMX_EPT_EXTENT_CONTEXT_BIT | VMX_EPT_2MB_PAGE_BIT |
Signature · 1 line
2.24.1
Paolo Bonzini
maintainerQuoted 25 lines · Sean Christopherson
Clean up MMU code related to 5 level paging, expose 5-level EPT to L1, and additional clean up on top (mostly renames of functions/variables that caused me no end of confusion when trying to figure out what was broken at various times).
v3:
- Dropped fixes for existing 5-level bugs (merged for 5.6).
- Useget_guest_pgd()instead ofget_guest_cr3_or_eptp(). [Paolo]
- Add patches to fix MMU role calculation to play nice with 5-level paging without requiring additional CR4.LA_57 bit.v2:
- Increase the nested EPT array sizes to accomodate 5-level paging in the patch that adds support for 5-level nested EPT, not in the bug fix for 5-level shadow paging.Sean Christopherson (7):
KVM: x86/mmu: Don't drop level/direct from MMU role calculation
KVM: x86/mmu: Drop kvm_mmu_extended_role.cr4_la57 hack
KVM: nVMX: Allow L1 to use 5-level page walks for nested EPT
KVM: nVMX: Renamenested_ept_get_cr3()tonested_ept_get_eptp()
KVM: nVMX: Rename EPTP validity helper and associated variables
KVM: x86/mmu: Renamekvm_mmu->get_cr3()to ->get_guest_pgd()
KVM: nVMX: Drop unnecessary check on ept caps for execute-only
Quoted 11 lines
arch/x86/include/asm/kvm_host.h | 3 +- arch/x86/include/asm/vmx.h | 12 +++++++ arch/x86/kvm/mmu/mmu.c | 59 +++++++++++++++++---------------- arch/x86/kvm/mmu/paging_tmpl.h | 4 +-- arch/x86/kvm/svm.c | 2 +- arch/x86/kvm/vmx/nested.c | 52 ++++++++++++++++++----------- arch/x86/kvm/vmx/nested.h | 4 +-- arch/x86/kvm/vmx/vmx.c | 3 +- arch/x86/kvm/x86.c | 2 +- 9 files changed, 82 insertions(+), 59 deletions(-)
Queued, thanks. I have a cleanup on top to unify set_cr3 and set_tdp_cr3, which I'll post after testing it.
Paolo