On 11/5/2024 10:23 PM, Paolo Bonzini wrote:
On 11/5/24 12:25, Xiaoyao Li wrote:
On 11/5/2024 5:55 PM, Paolo Bonzini wrote:
On 11/5/24 07:23, Xiaoyao Li wrote:
From: Sean Christopherson <sean.j.christopherson@xxxxxxxxx>
Don't get/put state of TDX VMs since accessing/mutating guest state of
production TDs is not supported.
Note, it will be allowed for a debug TD. Corresponding support will be
introduced when debug TD support is implemented in the future.
Signed-off-by: Sean Christopherson <sean.j.christopherson@xxxxxxxxx>
Signed-off-by: Xiaoyao Li <xiaoyao.li@xxxxxxxxx>
Acked-by: Gerd Hoffmann <kraxel@xxxxxxxxxx>
This should be unnecessary now that QEMU has
kvm_mark_guest_state_protected().
Reverting this patch, we get:
tdx: tdx: error: failed to set MSR 0x174 to 0x0
tdx: ../../../go/src/tdx/tdx-qemu/target/i386/kvm/kvm.c:3859:
kvm_buf_set_msrs: Assertion `ret == cpu->kvm_msr_buf->nmsrs' failed.
error: failed to set MSR 0x174 to 0x0
tdx: ../../../go/src/tdx/tdx-qemu/target/i386/kvm/kvm.c:3859:
kvm_buf_set_msrs: Assertion `ret == cpu->kvm_msr_buf->nmsrs' failed.
Difficult to "debug" without even a backtrace, but you might be calling
kvm_mark_guest_state_protected() too late. For SNP, the entry values of
the registers are customizable, for TDX they're not. So for TDX I think
it should be called even before realize completes, whereas SNP only
calls it on the first transition to RUNNING.
TDX calls kvm_mark_guest_state_protected() very early in
kvm_arch_init() -> tdx_kvm_init()
I find the call site. It's caused by kvm_arch_put_register() called in
kvm_cpu_exec() because cpu->vcpu_dirty is set to true in kvm_create_vcpu().
Maybe we can do something like below?
8<<<<<<<<<<<<<<<<<<<<<<<<<<<<
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -457,7 +457,9 @@ int kvm_create_vcpu(CPUState *cpu)
cpu->kvm_fd = kvm_fd;
cpu->kvm_state = s;
- cpu->vcpu_dirty = true;
+ if (!s->guest_state_protected) {
+ cpu->vcpu_dirty = true;
+ }
cpu->dirty_pages = 0;
cpu->throttle_us_per_full = 0;
Paolo
