#include "svm.h"
#include "svm_ops.h"
+#include "kvm_onhyperv.h"
+#include "svm_onhyperv.h"
+
#define __ex(x) __kvm_handle_fault_on_reboot(x)
MODULE_AUTHOR("Qumranet");
static int vgif = true;
module_param(vgif, int, 0444);
+/*
+ * enable / disable AVIC. Because the defaults differ for APICv
+ * support between VMX and SVM we cannot use module_param_named.
+ */
+static bool avic;
+module_param(avic, bool, 0444);
+
bool __read_mostly dump_invalid_vmcb;
module_param(dump_invalid_vmcb, bool, 0644);
+
+bool intercept_smi = true;
+module_param(intercept_smi, bool, 0444);
+
+
static bool svm_gp_erratum_intercept = true;
static u8 rsm_ins_bytes[] = "\x0f\xaa";
write ? clear_bit(bit_write, &tmp) : set_bit(bit_write, &tmp);
msrpm[offset] = tmp;
+
+ svm_hv_vmcb_dirty_nested_enlightenments(vcpu);
+
}
void set_msr_interception(struct kvm_vcpu *vcpu, u32 *msrpm, u32 msr,
int r;
unsigned int order = get_order(IOPM_SIZE);
+ /*
+ * NX is required for shadow paging and for NPT if the NX huge pages
+ * mitigation is enabled.
+ */
+ if (!boot_cpu_has(X86_FEATURE_NX)) {
+ pr_err_ratelimited("NX (Execute Disable) not supported\n");
+ return -EOPNOTSUPP;
+ }
+ kvm_enable_efer_bits(EFER_NX);
+
iopm_pages = alloc_pages(GFP_KERNEL, order);
if (!iopm_pages)
supported_xcr0 &= ~(XFEATURE_MASK_BNDREGS | XFEATURE_MASK_BNDCSR);
- if (boot_cpu_has(X86_FEATURE_NX))
- kvm_enable_efer_bits(EFER_NX);
-
if (boot_cpu_has(X86_FEATURE_FXSR_OPT))
kvm_enable_efer_bits(EFER_FFXSR);
/* Note, SEV setup consumes npt_enabled. */
sev_hardware_setup();
+ svm_hv_hardware_setup();
+
svm_adjust_mmio_mask();
for_each_possible_cpu(cpu) {
nrips = false;
}
- if (avic) {
- if (!npt_enabled || !boot_cpu_has(X86_FEATURE_AVIC)) {
- avic = false;
- } else {
- pr_info("AVIC enabled\n");
+ enable_apicv = avic = avic && npt_enabled && boot_cpu_has(X86_FEATURE_AVIC);
- amd_iommu_register_ga_log_notifier(&avic_ga_log_notifier);
- }
+ if (enable_apicv) {
+ pr_info("AVIC enabled\n");
+
+ amd_iommu_register_ga_log_notifier(&avic_ga_log_notifier);
}
if (vls) {
seg->base = 0;
}
-static u64 svm_write_l1_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
+static u64 svm_get_l2_tsc_offset(struct kvm_vcpu *vcpu)
{
struct vcpu_svm *svm = to_svm(vcpu);
- u64 g_tsc_offset = 0;
- if (is_guest_mode(vcpu)) {
- /* Write L1's TSC offset. */
- g_tsc_offset = svm->vmcb->control.tsc_offset -
- svm->vmcb01.ptr->control.tsc_offset;
- svm->vmcb01.ptr->control.tsc_offset = offset;
- }
+ return svm->nested.ctl.tsc_offset;
+}
- trace_kvm_write_tsc_offset(vcpu->vcpu_id,
- svm->vmcb->control.tsc_offset - g_tsc_offset,
- offset);
+static u64 svm_get_l2_tsc_multiplier(struct kvm_vcpu *vcpu)
+{
+ return kvm_default_tsc_scaling_ratio;
+}
- svm->vmcb->control.tsc_offset = offset + g_tsc_offset;
+static void svm_write_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
+{
+ struct vcpu_svm *svm = to_svm(vcpu);
+ svm->vmcb01.ptr->control.tsc_offset = vcpu->arch.l1_tsc_offset;
+ svm->vmcb->control.tsc_offset = offset;
vmcb_mark_dirty(svm->vmcb, VMCB_INTERCEPTS);
- return svm->vmcb->control.tsc_offset;
+}
+
+static void svm_write_tsc_multiplier(struct kvm_vcpu *vcpu, u64 multiplier)
+{
+ wrmsrl(MSR_AMD64_TSC_RATIO, multiplier);
}
/* Evaluate instruction intercepts that depend on guest CPUID features. */
svm_set_intercept(svm, INTERCEPT_INTR);
svm_set_intercept(svm, INTERCEPT_NMI);
- svm_set_intercept(svm, INTERCEPT_SMI);
+
+ if (intercept_smi)
+ svm_set_intercept(svm, INTERCEPT_SMI);
+
svm_set_intercept(svm, INTERCEPT_SELECTIVE_CR0);
svm_set_intercept(svm, INTERCEPT_RDPMC);
svm_set_intercept(svm, INTERCEPT_CPUID);
}
}
+ svm_hv_init_vmcb(svm->vmcb);
+
vmcb_mark_all_dirty(svm->vmcb);
enable_gif(svm);
{
struct vcpu_svm *svm = to_svm(vcpu);
- u64 fault_address = __sme_clr(svm->vmcb->control.exit_info_2);
+ u64 fault_address = svm->vmcb->control.exit_info_2;
u64 error_code = svm->vmcb->control.exit_info_1;
trace_kvm_page_fault(fault_address, error_code);
return 1;
}
+static int smi_interception(struct kvm_vcpu *vcpu)
+{
+ return 1;
+}
+
static int intr_interception(struct kvm_vcpu *vcpu)
{
++vcpu->stat.irq_exits;
svm_disable_lbrv(vcpu);
break;
case MSR_VM_HSAVE_PA:
- svm->nested.hsave_msr = data;
+ /*
+ * Old kernels did not validate the value written to
+ * MSR_VM_HSAVE_PA. Allow KVM_SET_MSR to set an invalid
+ * value to allow live migrating buggy or malicious guests
+ * originating from those kernels.
+ */
+ if (!msr->host_initiated && !page_address_valid(vcpu, data))
+ return 1;
+
+ svm->nested.hsave_msr = data & PAGE_MASK;
break;
case MSR_VM_CR:
return svm_set_vm_cr(vcpu, data);
[SVM_EXIT_EXCP_BASE + GP_VECTOR] = gp_interception,
[SVM_EXIT_INTR] = intr_interception,
[SVM_EXIT_NMI] = nmi_interception,
- [SVM_EXIT_SMI] = kvm_emulate_as_nop,
- [SVM_EXIT_INIT] = kvm_emulate_as_nop,
+ [SVM_EXIT_SMI] = smi_interception,
[SVM_EXIT_VINTR] = interrupt_window_interception,
[SVM_EXIT_RDPMC] = kvm_emulate_rdpmc,
[SVM_EXIT_CPUID] = kvm_emulate_cpuid,
return;
}
+ pr_err("VMCB %p, last attempted VMRUN on CPU %d\n",
+ svm->current_vmcb->ptr, vcpu->arch.last_vmentry_cpu);
pr_err("VMCB Control Area:\n");
pr_err("%-20s%04x\n", "cr_read:", control->intercepts[INTERCEPT_CR] & 0xffff);
pr_err("%-20s%04x\n", "cr_write:", control->intercepts[INTERCEPT_CR] >> 16);
}
svm->vmcb->save.cr2 = vcpu->arch.cr2;
+ svm_hv_update_vp_id(svm->vmcb, vcpu);
+
/*
* Run with all-zero DR6 unless needed, so that we can get the exact cause
* of a #DB.
svm->next_rip = 0;
if (is_guest_mode(vcpu)) {
nested_sync_control_from_vmcb02(svm);
+
+ /* Track VMRUNs that have made past consistency checking */
+ if (svm->nested.nested_run_pending &&
+ svm->vmcb->control.exit_code != SVM_EXIT_ERR)
+ ++vcpu->stat.nested_run;
+
svm->nested.nested_run_pending = 0;
}
vcpu->arch.apf.host_apf_flags =
kvm_read_and_reset_apf_flags();
- if (npt_enabled) {
- vcpu->arch.regs_avail &= ~(1 << VCPU_EXREG_PDPTR);
- vcpu->arch.regs_dirty &= ~(1 << VCPU_EXREG_PDPTR);
- }
+ if (npt_enabled)
+ kvm_register_clear_available(vcpu, VCPU_EXREG_PDPTR);
/*
* We need to handle MC intercepts here before the vcpu has a chance to
svm->vmcb->control.nested_cr3 = __sme_set(root_hpa);
vmcb_mark_dirty(svm->vmcb, VMCB_NPT);
+ hv_track_root_tdp(vcpu, root_hpa);
+
/* Loading L2's CR3 is handled by enter_svm_guest_mode. */
if (!test_bit(VCPU_EXREG_CR3, (ulong *)&vcpu->arch.regs_avail))
return;
return !svm_smi_blocked(vcpu);
}
-static int svm_pre_enter_smm(struct kvm_vcpu *vcpu, char *smstate)
+static int svm_enter_smm(struct kvm_vcpu *vcpu, char *smstate)
{
struct vcpu_svm *svm = to_svm(vcpu);
+ struct kvm_host_map map_save;
int ret;
if (is_guest_mode(vcpu)) {
ret = nested_svm_vmexit(svm);
if (ret)
return ret;
+
+ /*
+ * KVM uses VMCB01 to store L1 host state while L2 runs but
+ * VMCB01 is going to be used during SMM and thus the state will
+ * be lost. Temporary save non-VMLOAD/VMSAVE state to the host save
+ * area pointed to by MSR_VM_HSAVE_PA. APM guarantees that the
+ * format of the area is identical to guest save area offsetted
+ * by 0x400 (matches the offset of 'struct vmcb_save_area'
+ * within 'struct vmcb'). Note: HSAVE area may also be used by
+ * L1 hypervisor to save additional host context (e.g. KVM does
+ * that, see svm_prepare_guest_switch()) which must be
+ * preserved.
+ */
+ if (kvm_vcpu_map(vcpu, gpa_to_gfn(svm->nested.hsave_msr),
+ &map_save) == -EINVAL)
+ return 1;
+
+ BUILD_BUG_ON(offsetof(struct vmcb, save) != 0x400);
+
+ svm_copy_vmrun_state(&svm->vmcb01.ptr->save,
+ map_save.hva + 0x400);
+
+ kvm_vcpu_unmap(vcpu, &map_save, true);
}
return 0;
}
-static int svm_pre_leave_smm(struct kvm_vcpu *vcpu, const char *smstate)
+static int svm_leave_smm(struct kvm_vcpu *vcpu, const char *smstate)
{
struct vcpu_svm *svm = to_svm(vcpu);
- struct kvm_host_map map;
+ struct kvm_host_map map, map_save;
int ret = 0;
if (guest_cpuid_has(vcpu, X86_FEATURE_LM)) {
u64 saved_efer = GET_SMSTATE(u64, smstate, 0x7ed0);
u64 guest = GET_SMSTATE(u64, smstate, 0x7ed8);
u64 vmcb12_gpa = GET_SMSTATE(u64, smstate, 0x7ee0);
+ struct vmcb *vmcb12;
if (guest) {
if (!guest_cpuid_has(vcpu, X86_FEATURE_SVM))
if (svm_allocate_nested(svm))
return 1;
- ret = enter_svm_guest_mode(vcpu, vmcb12_gpa, map.hva);
+ vmcb12 = map.hva;
+
+ nested_load_control_from_vmcb12(svm, &vmcb12->control);
+
+ ret = enter_svm_guest_mode(vcpu, vmcb12_gpa, vmcb12);
kvm_vcpu_unmap(vcpu, &map, true);
+
+ /*
+ * Restore L1 host state from L1 HSAVE area as VMCB01 was
+ * used during SMM (see svm_enter_smm())
+ */
+ if (kvm_vcpu_map(vcpu, gpa_to_gfn(svm->nested.hsave_msr),
+ &map_save) == -EINVAL)
+ return 1;
+
+ svm_copy_vmrun_state(map_save.hva + 0x400,
+ &svm->vmcb01.ptr->save);
+
+ kvm_vcpu_unmap(vcpu, &map_save, true);
}
}
if (!pause_filter_count || !pause_filter_thresh)
kvm->arch.pause_in_guest = true;
- if (avic) {
+ if (enable_apicv) {
int ret = avic_vm_init(kvm);
if (ret)
return ret;
}
- kvm_apicv_init(kvm, avic);
return 0;
}
.has_wbinvd_exit = svm_has_wbinvd_exit,
- .write_l1_tsc_offset = svm_write_l1_tsc_offset,
+ .get_l2_tsc_offset = svm_get_l2_tsc_offset,
+ .get_l2_tsc_multiplier = svm_get_l2_tsc_multiplier,
+ .write_tsc_offset = svm_write_tsc_offset,
+ .write_tsc_multiplier = svm_write_tsc_multiplier,
.load_mmu_pgd = svm_load_mmu_pgd,
.setup_mce = svm_setup_mce,
.smi_allowed = svm_smi_allowed,
- .pre_enter_smm = svm_pre_enter_smm,
- .pre_leave_smm = svm_pre_leave_smm,
+ .enter_smm = svm_enter_smm,
+ .leave_smm = svm_leave_smm,
.enable_smi_window = svm_enable_smi_window,
.mem_enc_op = svm_mem_enc_op,