On Fri, Aug 18, 2023 at 05:49:57AM -0400, Xiaoyao Li wrote:
> Invoke KVM_TDX_INIT in kvm_arch_pre_create_vcpu() that KVM_TDX_INIT
> configures global TD configurations, e.g. the canonical CPUID config,
> and must be executed prior to creating vCPUs.
>
> Use kvm_x86_arch_cpuid() to setup the CPUID settings for TDX VM.
>
> Note, this doesn't address the fact that QEMU may change the CPUID
> configuration when creating vCPUs, i.e. punts on refactoring QEMU to
> provide a stable CPUID config prior to kvm_arch_init().
>
> Signed-off-by: Xiaoyao Li <xiaoyao.li@xxxxxxxxx>
> Acked-by: Gerd Hoffmann <kraxel@xxxxxxxxxx>
> ---
> accel/kvm/kvm-all.c | 9 +++++++-
> target/i386/kvm/kvm.c | 8 +++++++
> target/i386/kvm/tdx-stub.c | 5 +++++
> target/i386/kvm/tdx.c | 45 ++++++++++++++++++++++++++++++++++++++
> target/i386/kvm/tdx.h | 4 ++++
> 5 files changed, 70 insertions(+), 1 deletion(-)
>
> diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
> index 5071af917ae0..fceec7f2a83f 100644
> --- a/accel/kvm/kvm-all.c
> +++ b/accel/kvm/kvm-all.c
> @@ -435,10 +435,17 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)
>
> trace_kvm_init_vcpu(cpu->cpu_index, kvm_arch_vcpu_id(cpu));
>
> + /*
> + * tdx_pre_create_vcpu() may call cpu_x86_cpuid(). It in turn may call
> + * kvm_vm_ioctl(). Set cpu->kvm_state in advance to avoid NULL pointer
> + * dereference.
> + */
> + cpu->kvm_state = s;
> ret = kvm_arch_pre_create_vcpu(cpu);
> if (ret < 0) {
> error_setg_errno(errp, -ret, "%s: kvm_arch_pre_create_vcpu() failed",
> __func__);
> + cpu->kvm_state = NULL;
> goto err;
> }
>
> @@ -446,11 +453,11 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)
> if (ret < 0) {
> error_setg_errno(errp, -ret, "kvm_init_vcpu: kvm_get_vcpu failed (%lu)",
> kvm_arch_vcpu_id(cpu));
> + cpu->kvm_state = NULL;
> goto err;
> }
>
> cpu->kvm_fd = ret;
> - cpu->kvm_state = s;
> cpu->vcpu_dirty = true;
> cpu->dirty_pages = 0;
> cpu->throttle_us_per_full = 0;
> diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
> index 9ee41fffc445..d51067fdc12a 100644
> --- a/target/i386/kvm/kvm.c
> +++ b/target/i386/kvm/kvm.c
> @@ -2331,6 +2331,14 @@ int kvm_arch_init_vcpu(CPUState *cs)
> return r;
> }
>
> +int kvm_arch_pre_create_vcpu(CPUState *cpu)
> +{
> + if (is_tdx_vm())
> + return tdx_pre_create_vcpu(cpu);
Curly braces needed for coding style - run 'scripts/checkpatch.pl'
to validate it.
> +
> + return 0;
> +}
> +
> int kvm_arch_destroy_vcpu(CPUState *cs)
> {
> X86CPU *cpu = X86_CPU(cs);
> diff --git a/target/i386/kvm/tdx-stub.c b/target/i386/kvm/tdx-stub.c
> index 1d866d5496bf..61f70cc0d1d9 100644
> --- a/target/i386/kvm/tdx-stub.c
> +++ b/target/i386/kvm/tdx-stub.c
> @@ -6,3 +6,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp)
> {
> return -EINVAL;
> }
> +
> +int tdx_pre_create_vcpu(CPUState *cpu)
> +{
> + return -EINVAL;
> +}
> diff --git a/target/i386/kvm/tdx.c b/target/i386/kvm/tdx.c
> index 29f50fb9529e..3d313ed46bd1 100644
> --- a/target/i386/kvm/tdx.c
> +++ b/target/i386/kvm/tdx.c
> @@ -458,6 +458,49 @@ int tdx_kvm_init(MachineState *ms, Error **errp)
> return 0;
> }
>
> +int tdx_pre_create_vcpu(CPUState *cpu)
Add 'Error **errp' to this method
> +{
> + MachineState *ms = MACHINE(qdev_get_machine());
> + X86CPU *x86cpu = X86_CPU(cpu);
> + CPUX86State *env = &x86cpu->env;
> + struct kvm_tdx_init_vm *init_vm;
> + int r = 0;
> +
> + qemu_mutex_lock(&tdx_guest->lock);
> + if (tdx_guest->initialized) {
> + goto out;
> + }
> +
> + init_vm = g_malloc0(sizeof(struct kvm_tdx_init_vm) +
> + sizeof(struct kvm_cpuid_entry2) * KVM_MAX_CPUID_ENTRIES);
> +
> + r = kvm_vm_enable_cap(kvm_state, KVM_CAP_MAX_VCPUS, 0, ms->smp.cpus);
> + if (r < 0) {
> + error_report("Unable to set MAX VCPUS to %d", ms->smp.cpus);
Use error_setg / error_setg_errno in this method.
> + goto out_free;
> + }
> +
> + init_vm->cpuid.nent = kvm_x86_arch_cpuid(env, init_vm->cpuid.entries, 0);
> +
> + init_vm->attributes = tdx_guest->attributes;
> +
> + do {
> + r = tdx_vm_ioctl(KVM_TDX_INIT_VM, 0, init_vm);
> + } while (r == -EAGAIN);
> + if (r < 0) {
> + error_report("KVM_TDX_INIT_VM failed %s", strerror(-r));
> + goto out_free;
> + }
> +
> + tdx_guest->initialized = true;
> +
> +out_free:
> + g_free(init_vm);
> +out:
> + qemu_mutex_unlock(&tdx_guest->lock);
> + return r;
> +}
> +
> /* tdx guest */
> OBJECT_DEFINE_TYPE_WITH_INTERFACES(TdxGuest,
> tdx_guest,
> @@ -470,6 +513,8 @@ static void tdx_guest_init(Object *obj)
> {
> TdxGuest *tdx = TDX_GUEST(obj);
>
> + qemu_mutex_init(&tdx->lock);
> +
> tdx->attributes = 0;
> }
>
> diff --git a/target/i386/kvm/tdx.h b/target/i386/kvm/tdx.h
> index 06599b65b827..46a24ee8c7cc 100644
> --- a/target/i386/kvm/tdx.h
> +++ b/target/i386/kvm/tdx.h
> @@ -17,6 +17,9 @@ typedef struct TdxGuestClass {
> typedef struct TdxGuest {
> ConfidentialGuestSupport parent_obj;
>
> + QemuMutex lock;
> +
> + bool initialized;
> uint64_t attributes; /* TD attributes */
> } TdxGuest;
>
> @@ -29,5 +32,6 @@ bool is_tdx_vm(void);
> int tdx_kvm_init(MachineState *ms, Error **errp);
> void tdx_get_supported_cpuid(uint32_t function, uint32_t index, int reg,
> uint32_t *ret);
> +int tdx_pre_create_vcpu(CPUState *cpu);
>
> #endif /* QEMU_I386_TDX_H */
> --
> 2.34.1
>
With regards,
Daniel
--
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org -o- https://fstop138.berrange.com :|
|: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
