KVM: x86: Get active PCID only when writing a CR3 value
authorSean Christopherson <sean.j.christopherson@intel.com>
Fri, 5 Mar 2021 18:31:13 +0000 (10:31 -0800)
committerPaolo Bonzini <pbonzini@redhat.com>
Mon, 15 Mar 2021 08:43:56 +0000 (04:43 -0400)
Retrieve the active PCID only when writing a guest CR3 value, i.e. don't
get the PCID when using EPT or NPT.  The PCID is especially problematic
for EPT as the bits have different meaning, and so the PCID and must be
manually stripped, which is annoying and unnecessary.  And on VMX,
getting the active PCID also involves reading the guest's CR3 and
CR4.PCIDE, i.e. may add pointless VMREADs.

Opportunistically rename the pgd/pgd_level params to root_hpa and
root_level to better reflect their new roles.  Keep the function names,
as "load the guest PGD" is still accurate/correct.

Last, and probably least, pass root_hpa as a hpa_t/u64 instead of an
unsigned long.  The EPTP holds a 64-bit value, even in 32-bit mode, so
in theory EPT could support HIGHMEM for 32-bit KVM.  Never mind that
doing so would require changing the MMU page allocators and reworking
the MMU to use kmap().

Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210305183123.3978098-2-seanjc@google.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/include/asm/kvm_host.h
arch/x86/kvm/mmu.h
arch/x86/kvm/svm/svm.c
arch/x86/kvm/vmx/vmx.c
arch/x86/kvm/vmx/vmx.h

index d05542645b5c521caecca05e0b6534daf89c9573..a52f973bdff6d5326ec3cccc8ddde06d2da27f80 100644 (file)
@@ -1252,8 +1252,8 @@ struct kvm_x86_ops {
        int (*set_identity_map_addr)(struct kvm *kvm, u64 ident_addr);
        u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
 
-       void (*load_mmu_pgd)(struct kvm_vcpu *vcpu, unsigned long pgd,
-                            int pgd_level);
+       void (*load_mmu_pgd)(struct kvm_vcpu *vcpu, hpa_t root_hpa,
+                            int root_level);
 
        bool (*has_wbinvd_exit)(void);
 
index 67e8c7c7a6ce1784109ce4a9ac37449804a99b9c..88d0ed5225a4dede6ecce24523928f80c2215117 100644 (file)
@@ -107,8 +107,8 @@ static inline void kvm_mmu_load_pgd(struct kvm_vcpu *vcpu)
        if (!VALID_PAGE(root_hpa))
                return;
 
-       static_call(kvm_x86_load_mmu_pgd)(vcpu, root_hpa | kvm_get_active_pcid(vcpu),
-                                vcpu->arch.mmu->shadow_root_level);
+       static_call(kvm_x86_load_mmu_pgd)(vcpu, root_hpa,
+                                         vcpu->arch.mmu->shadow_root_level);
 }
 
 int kvm_tdp_page_fault(struct kvm_vcpu *vcpu, gpa_t gpa, u32 error_code,
index 13205e9cb66667fa189b08c4729e3fe7e7aaf684..58f4dc0e7864266d1ae9514a65e5671d61e08678 100644 (file)
@@ -3901,21 +3901,27 @@ static __no_kcsan fastpath_t svm_vcpu_run(struct kvm_vcpu *vcpu)
        return svm_exit_handlers_fastpath(vcpu);
 }
 
-static void svm_load_mmu_pgd(struct kvm_vcpu *vcpu, unsigned long root,
+static void svm_load_mmu_pgd(struct kvm_vcpu *vcpu, hpa_t root_hpa,
                             int root_level)
 {
        struct vcpu_svm *svm = to_svm(vcpu);
        unsigned long cr3;
 
-       cr3 = __sme_set(root);
+       root_hpa = __sme_set(root_hpa);
        if (npt_enabled) {
-               svm->vmcb->control.nested_cr3 = cr3;
+               svm->vmcb->control.nested_cr3 = root_hpa;
                vmcb_mark_dirty(svm->vmcb, VMCB_NPT);
 
                /* Loading L2's CR3 is handled by enter_svm_guest_mode.  */
                if (!test_bit(VCPU_EXREG_CR3, (ulong *)&vcpu->arch.regs_avail))
                        return;
                cr3 = vcpu->arch.cr3;
+       } else if (vcpu->arch.mmu->shadow_root_level >= PT64_ROOT_4LEVEL) {
+               cr3 = root_hpa | kvm_get_active_pcid(vcpu);
+       } else {
+               /* PCID in the guest should be impossible with a 32-bit MMU. */
+               WARN_ON_ONCE(kvm_get_active_pcid(vcpu));
+               cr3 = root_hpa;
        }
 
        svm->vmcb->save.cr3 = cr3;
index 2a64df2fd55c99714d75dba5dea20738ef2271b7..0c36cb688ca9a78fe9a8e85f3dcd2f8dc01464c2 100644 (file)
@@ -3088,8 +3088,7 @@ static int vmx_get_max_tdp_level(void)
        return 4;
 }
 
-u64 construct_eptp(struct kvm_vcpu *vcpu, unsigned long root_hpa,
-                  int root_level)
+u64 construct_eptp(struct kvm_vcpu *vcpu, hpa_t root_hpa, int root_level)
 {
        u64 eptp = VMX_EPTP_MT_WB;
 
@@ -3098,13 +3097,13 @@ u64 construct_eptp(struct kvm_vcpu *vcpu, unsigned long root_hpa,
        if (enable_ept_ad_bits &&
            (!is_guest_mode(vcpu) || nested_ept_ad_enabled(vcpu)))
                eptp |= VMX_EPTP_AD_ENABLE_BIT;
-       eptp |= (root_hpa & PAGE_MASK);
+       eptp |= root_hpa;
 
        return eptp;
 }
 
-static void vmx_load_mmu_pgd(struct kvm_vcpu *vcpu, unsigned long pgd,
-                            int pgd_level)
+static void vmx_load_mmu_pgd(struct kvm_vcpu *vcpu, hpa_t root_hpa,
+                            int root_level)
 {
        struct kvm *kvm = vcpu->kvm;
        bool update_guest_cr3 = true;
@@ -3112,7 +3111,7 @@ static void vmx_load_mmu_pgd(struct kvm_vcpu *vcpu, unsigned long pgd,
        u64 eptp;
 
        if (enable_ept) {
-               eptp = construct_eptp(vcpu, pgd, pgd_level);
+               eptp = construct_eptp(vcpu, root_hpa, root_level);
                vmcs_write64(EPT_POINTER, eptp);
 
                if (kvm_x86_ops.tlb_remote_flush) {
@@ -3131,7 +3130,7 @@ static void vmx_load_mmu_pgd(struct kvm_vcpu *vcpu, unsigned long pgd,
                        update_guest_cr3 = false;
                vmx_ept_load_pdptrs(vcpu);
        } else {
-               guest_cr3 = pgd;
+               guest_cr3 = root_hpa | kvm_get_active_pcid(vcpu);
        }
 
        if (update_guest_cr3)
index 89da5e1251f18c115c8436aca484b3656151a0cf..4795955490cb226851bb4ea0d5f23eba63611537 100644 (file)
@@ -376,8 +376,7 @@ void set_cr4_guest_host_mask(struct vcpu_vmx *vmx);
 void ept_save_pdptrs(struct kvm_vcpu *vcpu);
 void vmx_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
 void vmx_set_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
-u64 construct_eptp(struct kvm_vcpu *vcpu, unsigned long root_hpa,
-                  int root_level);
+u64 construct_eptp(struct kvm_vcpu *vcpu, hpa_t root_hpa, int root_level);
 
 void vmx_update_exception_bitmap(struct kvm_vcpu *vcpu);
 void vmx_update_msr_bitmap(struct kvm_vcpu *vcpu);