Re: [PATCH v15 14/23] KVM: selftests: Expose function to allocate vCPU stack

[Xiaoyao Li <xiaoyao.li@intel.com>]

On 10/2/2026 3:37 AM, Lisa Wang wrote:
> From: Sagi Shahar <sagis@google.com>
> 
> Introduce kvm_allocate_vcpu_stack() to allocate a vCPU's stack
> in preparation for TDX to allocate a vCPU's stack and initialize
> its stack pointer.
> 
> TDX VMs' registers are protected state and cannot be initialized
> using the KVM_SET_REGS ioctl() that is used for normal VMs. A TDX
> vCPU's stack address will be a property of the TDX specific boot code
> that initializes the vCPUs' stack pointers at boot.
> 
> Signed-off-by: Sagi Shahar <sagis@google.com>
> Signed-off-by: Lisa Wang <wyihan@google.com>
> Reviewed-by: Binbin Wu <binbin.wu@linux.intel.com>
> Reviewed-by: Ira Weiny <ira.weiny@intel.com>
> Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>
> ---
>  .../testing/selftests/kvm/include/x86/processor.h  |  2 ++
>  tools/testing/selftests/kvm/lib/x86/processor.c    | 25 +++++++++++++++++++---
>  2 files changed, 24 insertions(+), 3 deletions(-)
> 
> diff --git a/tools/testing/selftests/kvm/include/x86/processor.h b/tools/testing/selftests/kvm/include/x86/processor.h
> index 8854ee052d29..1eb380dd216c 100644
> --- a/tools/testing/selftests/kvm/include/x86/processor.h
> +++ b/tools/testing/selftests/kvm/include/x86/processor.h
> @@ -1212,6 +1212,8 @@ static inline void vcpu_clear_cpuid_feature(struct kvm_vcpu *vcpu,
>  	vcpu_set_or_clear_cpuid_feature(vcpu, feature, false);
>  }
>  
> +gva_t kvm_allocate_vcpu_stack(struct kvm_vm *vm);
> +
>  u64 vcpu_get_msr(struct kvm_vcpu *vcpu, u64 msr_index);
>  int _vcpu_set_msr(struct kvm_vcpu *vcpu, u64 msr_index, u64 msr_value);
>  
> diff --git a/tools/testing/selftests/kvm/lib/x86/processor.c b/tools/testing/selftests/kvm/lib/x86/processor.c
> index da430d4e7765..4a753fb43007 100644
> --- a/tools/testing/selftests/kvm/lib/x86/processor.c
> +++ b/tools/testing/selftests/kvm/lib/x86/processor.c
> @@ -851,12 +851,31 @@ void vcpu_arch_set_entry_point(struct kvm_vcpu *vcpu, void *guest_code)
>  	vcpu_regs_set(vcpu, &regs);
>  }
>  
> -struct kvm_vcpu *vm_arch_vcpu_add(struct kvm_vm *vm, u32 vcpu_id)
> +gva_t kvm_allocate_vcpu_stack(struct kvm_vm *vm)

Besides what Sashiko pointed out[1], this patch was just messed up during the
rebase and it even doesn't compile.

[1] https://lore.kernel.org/all/20261002091331.BA23F1F00893@smtp.kernel.org/

>  {
>  	struct kvm_mp_state mp_state;
>  	struct kvm_regs regs;
> +	gva_t stack_gva;
>  	struct kvm_vcpu *vcpu;
>  
> +	stack_gva = __vm_alloc(vm, DEFAULT_STACK_PGS * getpagesize(),
> +			       DEFAULT_GUEST_STACK_VADDR_MIN, MEM_REGION_DATA);
> +
> +	stack_gva += DEFAULT_STACK_PGS * getpagesize();
> +
> +	/*
> +	 * Align stack to match calling sequence requirements in section "The
> +	 * Stack Frame" of the System V ABI AMD64 Architecture Processor
> +	 * Supplement, which requires the value (%rsp + 8) to be a multiple of
> +	 * 16 when control is transferred to the function entry point.
> +	 *
> +	 * If this code is ever used to launch a vCPU with 32-bit entry point it
> +	 * may need to subtract 4 bytes instead of 8 bytes.
> +	 */
> +	TEST_ASSERT(IS_ALIGNED(stack_gva, PAGE_SIZE),
> +		    "__vm_alloc() did not provide a page-aligned address");
> +	stack_gva -= 8;
> +
>  	vcpu = __vm_vcpu_add(vm, vcpu_id);
>  	vcpu_init_cpuid(vcpu, kvm_get_supported_cpuid());
>  	vcpu_init_sregs(vm, vcpu);
> @@ -864,8 +883,8 @@ struct kvm_vcpu *vm_arch_vcpu_add(struct kvm_vm *vm, u32 vcpu_id)
>  
>  	/* Setup guest general purpose registers */
>  	vcpu_regs_get(vcpu, &regs);
> -	regs.rflags = regs.rflags | X86_EFLAGS_FIXED;
> -	regs.rsp = vm_alloc_stack(vm, DEFAULT_STACK_PGS);
> +	regs.rflags = regs.rflags | 0x2;
> +	regs.rsp = kvm_allocate_vcpu_stack(vm);
>  	vcpu_regs_set(vcpu, &regs);
>  
>  	/* Setup the MP state */
>