Received-SPF: pass (google.com: domain of linux-kernel+bounces-34897-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Message-ID: <03e3eb4c-1d7e-4f24-8d5c-1f48f1d6f81f@linux.intel.com>
Date: Tue, 23 Jan 2024 16:32:24 +0800
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH v18 015/121] KVM: TDX: Retry SEAMCALL on the lack of
 entropy error
To: isaku.yamahata@intel.com, Kai Huang <kai.huang@intel.com>
Cc: kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
 isaku.yamahata@gmail.com, Paolo Bonzini <pbonzini@redhat.com>,
 erdemaktas@google.com, Sean Christopherson <seanjc@google.com>,
 Sagi Shahar <sagis@google.com>, chen.bo@intel.com, hang.yuan@intel.com,
 tina.zhang@intel.com
References: <cover.1705965634.git.isaku.yamahata@intel.com>
 <a4d44b01523c935595c8ee00622ab8766e269ef4.1705965634.git.isaku.yamahata@intel.com>
From: Binbin Wu <binbin.wu@linux.intel.com>
In-Reply-To: <a4d44b01523c935595c8ee00622ab8766e269ef4.1705965634.git.isaku.yamahata@intel.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit


On 1/23/2024 7:52 AM, isaku.yamahata@intel.com wrote:
> From: Isaku Yamahata <isaku.yamahata@intel.com>
>
> Some SEAMCALL may return TDX_RND_NO_ENTROPY error when the entropy is
> lacking.  Retry SEAMCALL on the error following rdrand_long() to retry
> RDRAND_RETRY_LOOPS times.
>
> Signed-off-by: Isaku Yamahata <isaku.yamahata@intel.com>
> ---
> v18:
> - update to use struct tdx_modules_args for inputs.
> ---
>   arch/x86/kvm/vmx/tdx_errno.h |  1 +
>   arch/x86/kvm/vmx/tdx_ops.h   | 15 +++++++++++----
>   2 files changed, 12 insertions(+), 4 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx/tdx_errno.h b/arch/x86/kvm/vmx/tdx_errno.h
> index 7f96696b8e7c..bb093e292fef 100644
> --- a/arch/x86/kvm/vmx/tdx_errno.h
> +++ b/arch/x86/kvm/vmx/tdx_errno.h
> @@ -14,6 +14,7 @@
>   #define TDX_OPERAND_INVALID			0xC000010000000000ULL
>   #define TDX_OPERAND_BUSY			0x8000020000000000ULL
>   #define TDX_PREVIOUS_TLB_EPOCH_BUSY		0x8000020100000000ULL
> +#define TDX_RND_NO_ENTROPY			0x8000020300000000ULL
>   #define TDX_VCPU_NOT_ASSOCIATED			0x8000070200000000ULL
>   #define TDX_KEY_GENERATION_FAILED		0x8000080000000000ULL
>   #define TDX_KEY_STATE_INCORRECT			0xC000081100000000ULL
> diff --git a/arch/x86/kvm/vmx/tdx_ops.h b/arch/x86/kvm/vmx/tdx_ops.h
> index 0e26cf22240e..f4c16e5265f0 100644
> --- a/arch/x86/kvm/vmx/tdx_ops.h
> +++ b/arch/x86/kvm/vmx/tdx_ops.h
> @@ -6,6 +6,7 @@
>   
>   #include <linux/compiler.h>
>   
> +#include <asm/archrandom.h>
>   #include <asm/cacheflush.h>
>   #include <asm/asm.h>
>   #include <asm/kvm_host.h>
> @@ -17,14 +18,20 @@
>   static inline u64 tdx_seamcall(u64 op, struct tdx_module_args *in,
>   			       struct tdx_module_args *out)
>   {
> +	struct tdx_module_args args;
> +	int retry;
>   	u64 ret;
>   
> -	if (out) {
> +	if (!out)
> +		out = &args;
> +
> +	/* Mimic the existing rdrand_long() to retry RDRAND_RETRY_LOOPS times. */
> +	retry = RDRAND_RETRY_LOOPS;
> +	do {
> +		/* As __seamcall_ret() overwrites out, init out on each loop. */
>   		*out = *in;
>   		ret = __seamcall_ret(op, out);
> -	} else
> -		ret = __seamcall(op, in);
> -
> +	} while (unlikely(ret == TDX_RND_NO_ENTROPY) && --retry);
>   	if (unlikely(ret == TDX_SEAMCALL_UD)) {
>   		/*
>   		 * SEAMCALLs fail with TDX_SEAMCALL_UD returned when VMX is off.
One question is when the return code is TDX_RND_NO_ENTROPY, does the input
tdx_module_args value will be modified or not?

As mentioned by Kai in previous review, there are seamcall*() variants which
handle TDX_RND_NO_ENTROPY already.

If the input tdx_module_args is not modified by a seamcall when the return
code is TDX_RND_NO_ENTROPY, the seamcall_ret() can be used directly.

But if the input tdx_module_args could be modified by a seamcall when the
return code is TDX_RND_NO_ENTROPY, the implementations of seamcall_ret() and
seamcall_saved_ret() have problem then, because the input tdx_module_args is
not re-initialized when retry.
They need to be fixed or deleted (no one is using them currently).