linux-kernel - Re: [PATCH v9 18/19] KVM: selftests: Add ucall support for TDX

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <18bf858c-e135-4a9b-bda8-a70be3b3720e@linux.intel.com>
Date: Wed, 27 Aug 2025 15:18:58 +0800
From: Binbin Wu <binbin.wu@...ux.intel.com>
To: Sagi Shahar <sagis@...gle.com>, Sean Christopherson <seanjc@...gle.com>
Cc: linux-kselftest@...r.kernel.org, Paolo Bonzini <pbonzini@...hat.com>,
 Shuah Khan <shuah@...nel.org>, Ackerley Tng <ackerleytng@...gle.com>,
 Ryan Afranji <afranji@...gle.com>, Andrew Jones <ajones@...tanamicro.com>,
 Isaku Yamahata <isaku.yamahata@...el.com>,
 Erdem Aktas <erdemaktas@...gle.com>,
 Rick Edgecombe <rick.p.edgecombe@...el.com>,
 Roger Wang <runanwang@...gle.com>, Oliver Upton <oliver.upton@...ux.dev>,
 "Pratik R. Sampat" <pratikrajesh.sampat@....com>,
 Reinette Chatre <reinette.chatre@...el.com>, Ira Weiny
 <ira.weiny@...el.com>, Chao Gao <chao.gao@...el.com>,
 Chenyi Qiang <chenyi.qiang@...el.com>, linux-kernel@...r.kernel.org,
 kvm@...r.kernel.org
Subject: Re: [PATCH v9 18/19] KVM: selftests: Add ucall support for TDX



On 8/21/2025 12:29 PM, Sagi Shahar wrote:
> From: Ackerley Tng <ackerleytng@...gle.com>
>
> ucalls for non-Coco VMs work by having the guest write to the rdi
> register, then perform an io instruction to exit to the host. The host
> then reads rdi using kvm_get_regs().
>
> CPU registers can't be read using kvm_get_regs() for TDX, so TDX
> guests use MMIO to pass the struct ucall's hva to the host. MMIO was
> chosen because it is one of the simplest (hence unlikely to fail)
> mechanisms that support passing 8 bytes from guest to host.
>
> Signed-off-by: Ackerley Tng <ackerleytng@...gle.com>
> Co-developed-by: Sagi Shahar <sagis@...gle.com>
> Signed-off-by: Sagi Shahar <sagis@...gle.com>
> ---
>   .../testing/selftests/kvm/include/x86/ucall.h |  4 +-
>   tools/testing/selftests/kvm/lib/x86/ucall.c   | 45 ++++++++++++++++---
>   2 files changed, 41 insertions(+), 8 deletions(-)
>
> diff --git a/tools/testing/selftests/kvm/include/x86/ucall.h b/tools/testing/selftests/kvm/include/x86/ucall.h
> index d3825dcc3cd9..0494a4a21557 100644
> --- a/tools/testing/selftests/kvm/include/x86/ucall.h
> +++ b/tools/testing/selftests/kvm/include/x86/ucall.h
> @@ -6,8 +6,6 @@
>   
>   #define UCALL_EXIT_REASON       KVM_EXIT_IO
>   
> -static inline void ucall_arch_init(struct kvm_vm *vm, vm_paddr_t mmio_gpa)
> -{
> -}
> +void ucall_arch_init(struct kvm_vm *vm, vm_paddr_t mmio_gpa);
>   
>   #endif
> diff --git a/tools/testing/selftests/kvm/lib/x86/ucall.c b/tools/testing/selftests/kvm/lib/x86/ucall.c
> index 1265cecc7dd1..0ad24baaa3c4 100644
> --- a/tools/testing/selftests/kvm/lib/x86/ucall.c
> +++ b/tools/testing/selftests/kvm/lib/x86/ucall.c
> @@ -5,11 +5,34 @@
>    * Copyright (C) 2018, Red Hat, Inc.
>    */
>   #include "kvm_util.h"
> +#include "tdx/tdx.h"
>   
>   #define UCALL_PIO_PORT ((uint16_t)0x1000)
>   
> +static uint8_t vm_type;
> +static vm_paddr_t host_ucall_mmio_gpa;
> +static vm_paddr_t ucall_mmio_gpa;
> +
> +void ucall_arch_init(struct kvm_vm *vm, vm_paddr_t mmio_gpa)
> +{
> +	vm_type = vm->type;
> +	sync_global_to_guest(vm, vm_type);
> +
> +	host_ucall_mmio_gpa = ucall_mmio_gpa = mmio_gpa;
> +
> +	if (vm_type == KVM_X86_TDX_VM)
> +		ucall_mmio_gpa |= vm->arch.s_bit;
> +
> +	sync_global_to_guest(vm, ucall_mmio_gpa);
> +}
> +
>   void ucall_arch_do_ucall(vm_vaddr_t uc)
>   {
> +	if (vm_type == KVM_X86_TDX_VM) {
> +		tdg_vp_vmcall_ve_request_mmio_write(ucall_mmio_gpa, 8, uc);
> +		return;
> +	}
> +
>   	/*
>   	 * FIXME: Revert this hack (the entire commit that added it) once nVMX
>   	 * preserves L2 GPRs across a nested VM-Exit.  If a ucall from L2, e.g.
> @@ -46,11 +69,23 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)
>   {
>   	struct kvm_run *run = vcpu->run;
>   
> -	if (run->exit_reason == KVM_EXIT_IO && run->io.port == UCALL_PIO_PORT) {
> -		struct kvm_regs regs;
> +	switch (vm_type) {
> +	case KVM_X86_TDX_VM:
> +		if (vcpu->run->exit_reason == KVM_EXIT_MMIO &&
> +		    vcpu->run->mmio.phys_addr == host_ucall_mmio_gpa &&
> +		    vcpu->run->mmio.len == 8 && vcpu->run->mmio.is_write) {
> +			uint64_t data = *(uint64_t *)vcpu->run->mmio.data;
> +
> +			return (void *)data;
> +		}
> +		return NULL;

My first thought was how did SEV_ES or SNP work for this since they are not
able to get RDI neither.
Then I had a check in sev_smoke_test.c, both guest_sev_es_code() and
guest_snp_code() call GUEST_ASSERT(), which finally calls ucall_assert(), but
in test_sev(), the code doesn't handle ucall for SEV_ES or SNP.
Does it mean GUEST_ASSERT() is currently not working and ignored for SEV_ES
and SNP? Or did I miss anything?

> +	default:
> +		if (run->exit_reason == KVM_EXIT_IO && run->io.port == UCALL_PIO_PORT) {
> +			struct kvm_regs regs;
>   
> -		vcpu_regs_get(vcpu, &regs);
> -		return (void *)regs.rdi;
> +			vcpu_regs_get(vcpu, &regs);
> +			return (void *)regs.rdi;
> +		}
> +		return NULL;
>   	}
> -	return NULL;
>   }