Re: [PATCH v6 09/60] i386/tdx: Initialize TDX before creating TD vcpus

["Edgecombe, Rick P" <rick.p.edgecombe@intel.com>]

+Tony

On Tue, 2024-11-05 at 01:23 -0500, Xiaoyao Li wrote:
> +int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
> +{
> +    X86CPU *x86cpu = X86_CPU(cpu);
> +    CPUX86State *env = &x86cpu->env;
> +    g_autofree struct kvm_tdx_init_vm *init_vm = NULL;
> +    int r = 0;
> +
> +    QEMU_LOCK_GUARD(&tdx_guest->lock);
> +    if (tdx_guest->initialized) {
> +        return r;
> +    }
> +
> +    init_vm = g_malloc0(sizeof(struct kvm_tdx_init_vm) +
> +                        sizeof(struct kvm_cpuid_entry2) * KVM_MAX_CPUID_ENTRIES);
> +
> +    r = setup_td_xfam(x86cpu, errp);
> +    if (r) {
> +        return r;
> +    }
> +
> +    init_vm->cpuid.nent = kvm_x86_build_cpuid(env, init_vm->cpuid.entries, 0);
> +    tdx_filter_cpuid(&init_vm->cpuid);
> +
> +    init_vm->attributes = tdx_guest->attributes;
> +    init_vm->xfam = tdx_guest->xfam;
> +
> +    do {
> +        r = tdx_vm_ioctl(KVM_TDX_INIT_VM, 0, init_vm);
> +    } while (r == -EAGAIN);

KVM_TDX_INIT_VM can also return EBUSY. This should check for it, or KVM should
standardize on one for both conditions. In KVM, both cases handle
TDX_RND_NO_ENTROPY, but one tries to save some of the initialization for the
next attempt. I don't know why userspace would need to differentiate between the
two cases though, which makes me think we should just change the KVM side.

> +    if (r < 0) {
> +        error_setg_errno(errp, -r, "KVM_TDX_INIT_VM failed");
> +        return r;
> +    }
> +
> +    tdx_guest->initialized = true;
> +
> +    return 0;
> +}