On 03/07/21 00:05, isaku.yamahata@xxxxxxxxx wrote:
From: Xiaoyao Li <xiaoyao.li@xxxxxxxxx>
Introduce a per-vm variable initial_tsc_khz to hold the default tsc_khz
for kvm_arch_vcpu_create().
This field is going to be used by TDX since TSC frequency for TD guest
is configured at TD VM initialization phase.
Signed-off-by: Xiaoyao Li <xiaoyao.li@xxxxxxxxx>
Signed-off-by: Isaku Yamahata <isaku.yamahata@xxxxxxxxx>
---
arch/x86/include/asm/kvm_host.h | 1 +
arch/x86/kvm/x86.c | 3 ++-
2 files changed, 3 insertions(+), 1 deletion(-)
So this means disabling TSC frequency scaling on TDX. Would it make
sense to delay VM creation to a separate ioctl, similar to
KVM_ARM_VCPU_FINALIZE (KVM_VM_FINALIZE)?
Paolo
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index a47e17892258..ae8b96e15e71 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -1030,6 +1030,7 @@ struct kvm_arch {
u64 last_tsc_nsec;
u64 last_tsc_write;
u32 last_tsc_khz;
+ u32 initial_tsc_khz;
u64 cur_tsc_nsec;
u64 cur_tsc_write;
u64 cur_tsc_offset;
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index a8299add443f..d3ebed784eac 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -10441,7 +10441,7 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
else
vcpu->arch.mp_state = KVM_MP_STATE_UNINITIALIZED;
- kvm_set_tsc_khz(vcpu, max_tsc_khz);
+ kvm_set_tsc_khz(vcpu, vcpu->kvm->arch.initial_tsc_khz);
r = kvm_mmu_create(vcpu);
if (r < 0)
@@ -10894,6 +10894,7 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
pvclock_update_vm_gtod_copy(kvm);
kvm->arch.guest_can_read_msr_platform_info = true;
+ kvm->arch.initial_tsc_khz = max_tsc_khz;
INIT_DELAYED_WORK(&kvm->arch.kvmclock_update_work, kvmclock_update_fn);
INIT_DELAYED_WORK(&kvm->arch.kvmclock_sync_work, kvmclock_sync_fn);
