linux-kernel - Re: [PATCH v19 085/130] KVM: TDX: Complete interrupts after tdexit

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20240617080729.j5nottky5bjmgdmf@yy-desk-7060>
Date: Mon, 17 Jun 2024 16:07:29 +0800
From: Yuan Yao <yuan.yao@...ux.intel.com>
To: isaku.yamahata@...el.com
Cc: kvm@...r.kernel.org, linux-kernel@...r.kernel.org,
	isaku.yamahata@...il.com, Paolo Bonzini <pbonzini@...hat.com>,
	erdemaktas@...gle.com, Sean Christopherson <seanjc@...gle.com>,
	Sagi Shahar <sagis@...gle.com>, Kai Huang <kai.huang@...el.com>,
	chen.bo@...el.com, hang.yuan@...el.com, tina.zhang@...el.com,
	Binbin Wu <binbin.wu@...ux.intel.com>
Subject: Re: [PATCH v19 085/130] KVM: TDX: Complete interrupts after tdexit

On Mon, Feb 26, 2024 at 12:26:27AM -0800, isaku.yamahata@...el.com wrote:
> From: Isaku Yamahata <isaku.yamahata@...el.com>
>
> This corresponds to VMX __vmx_complete_interrupts().  Because TDX
> virtualize vAPIC, KVM only needs to care NMI injection.
>
> Signed-off-by: Isaku Yamahata <isaku.yamahata@...el.com>
> Reviewed-by: Paolo Bonzini <pbonzini@...hat.com>
> Reviewed-by: Binbin Wu <binbin.wu@...ux.intel.com>
> ---
> v19:
> - move tdvps_management_check() to this patch
> - typo: complete -> Complete in short log
> ---
>  arch/x86/kvm/vmx/tdx.c | 10 ++++++++++
>  arch/x86/kvm/vmx/tdx.h |  4 ++++
>  2 files changed, 14 insertions(+)
>
> diff --git a/arch/x86/kvm/vmx/tdx.c b/arch/x86/kvm/vmx/tdx.c
> index 83dcaf5b6fbd..b8b168f74dfe 100644
> --- a/arch/x86/kvm/vmx/tdx.c
> +++ b/arch/x86/kvm/vmx/tdx.c
> @@ -535,6 +535,14 @@ void tdx_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event)
>  	 */
>  }
>
> +static void tdx_complete_interrupts(struct kvm_vcpu *vcpu)
> +{
> +	/* Avoid costly SEAMCALL if no nmi was injected */
> +	if (vcpu->arch.nmi_injected)
> +		vcpu->arch.nmi_injected = td_management_read8(to_tdx(vcpu),
> +							      TD_VCPU_PEND_NMI);
> +}

Looks this leads to NMI injection delay or even won't be
reinjected if KVM_REQ_EVENT is not set on the target cpu
when more than 1 NMIs are pending there.

On normal VM, KVM uses NMI window vmexit for injection
successful case to rasie the KVM_REQ_EVENT again for remain
pending NMIs, see handle_nmi_window(). KVM also checks
vectoring info after VMEXIT for case that the NMI is not
injected successfully in this vmentry vmexit round, and
raise KVM_REQ_EVENT to try again, see __vmx_complete_interrupts().

In TDX, consider there's no way to get vectoring info or
handle nmi window vmexit, below checking should cover both
scenarios for NMI injection:

diff --git a/arch/x86/kvm/vmx/tdx.c b/arch/x86/kvm/vmx/tdx.c
index e9c9a185bb7b..9edf446acd3b 100644
--- a/arch/x86/kvm/vmx/tdx.c
+++ b/arch/x86/kvm/vmx/tdx.c
@@ -835,9 +835,12 @@ void tdx_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event)
 static void tdx_complete_interrupts(struct kvm_vcpu *vcpu)
 {
        /* Avoid costly SEAMCALL if no nmi was injected */
-       if (vcpu->arch.nmi_injected)
+       if (vcpu->arch.nmi_injected) {
                vcpu->arch.nmi_injected = td_management_read8(to_tdx(vcpu),
                                                              TD_VCPU_PEND_NMI);
+               if (vcpu->arch.nmi_injected || vcpu->arch.nmi_pending)
+                       kvm_make_request(KVM_REQ_EVENT, vcpu);
+       }
 }

> +
>  struct tdx_uret_msr {
>  	u32 msr;
>  	unsigned int slot;
> @@ -663,6 +671,8 @@ fastpath_t tdx_vcpu_run(struct kvm_vcpu *vcpu)
>  	vcpu->arch.regs_avail &= ~VMX_REGS_LAZY_LOAD_SET;
>  	trace_kvm_exit(vcpu, KVM_ISA_VMX);
>
> +	tdx_complete_interrupts(vcpu);
> +
>  	return EXIT_FASTPATH_NONE;
>  }
>
> diff --git a/arch/x86/kvm/vmx/tdx.h b/arch/x86/kvm/vmx/tdx.h
> index 44eab734e702..0d8a98feb58e 100644
> --- a/arch/x86/kvm/vmx/tdx.h
> +++ b/arch/x86/kvm/vmx/tdx.h
> @@ -142,6 +142,8 @@ static __always_inline void tdvps_vmcs_check(u32 field, u8 bits)
>  			 "Invalid TD VMCS access for 16-bit field");
>  }
>
> +static __always_inline void tdvps_management_check(u64 field, u8 bits) {}
> +
>  #define TDX_BUILD_TDVPS_ACCESSORS(bits, uclass, lclass)				\
>  static __always_inline u##bits td_##lclass##_read##bits(struct vcpu_tdx *tdx,	\
>  							u32 field)		\
> @@ -200,6 +202,8 @@ TDX_BUILD_TDVPS_ACCESSORS(16, VMCS, vmcs);
>  TDX_BUILD_TDVPS_ACCESSORS(32, VMCS, vmcs);
>  TDX_BUILD_TDVPS_ACCESSORS(64, VMCS, vmcs);
>
> +TDX_BUILD_TDVPS_ACCESSORS(8, MANAGEMENT, management);
> +
>  static __always_inline u64 td_tdcs_exec_read64(struct kvm_tdx *kvm_tdx, u32 field)
>  {
>  	struct tdx_module_args out;
> --
> 2.25.1
>
>