Re: [PATCH v3] crypto: x86/sm2 -add Zhaoxin SM2 algorithm implementation

[LeoLiu-oc <leoliu-oc@zhaoxin.com>]

在 2023/12/1 17:27, Herbert Xu 写道:
> On Wed, Nov 22, 2023 at 02:43:55PM +0800, LeoLiu-oc wrote:
>> From: LeoLiuoc <LeoLiu-oc@zhaoxin.com>
>>
>> Add support for SM2 (ShangMi 2) public key algorithm by Zhaoxin GMI
>> Instruction. The purpose of this driver is to ensure that the application
>> has both high performance and high security.
>>
>> ---
>>
>> v1 -> v2:
>> 1. The assembly code is modified to be embedded in the .c file.
>> 2. Optimize code style and details.
>>
>> v2 -> v3:
>> 1. Increase compatibility with i386 architecture.
>> 2. Optimize variable and return value types in some functions..
>>
>> Signed-off-by: LeoLiuoc <LeoLiu-oc@zhaoxin.com>
>> ---
>>   arch/x86/crypto/Kconfig                |  11 ++
>>   arch/x86/crypto/Makefile               |   2 +
>>   arch/x86/crypto/sm2-zhaoxin-gmi_glue.c | 158 +++++++++++++++++++++++++
>>   arch/x86/include/asm/cpufeatures.h     |   2 +
>>   4 files changed, 173 insertions(+)
>>   create mode 100644 arch/x86/crypto/sm2-zhaoxin-gmi_glue.c
>>
>> diff --git a/arch/x86/crypto/Kconfig b/arch/x86/crypto/Kconfig
>> index 9bbfd01cfa2f..974d4c3806ff 100644
>> --- a/arch/x86/crypto/Kconfig
>> +++ b/arch/x86/crypto/Kconfig
>> @@ -519,4 +519,15 @@ config CRYPTO_CRCT10DIF_PCLMUL
>>   	  Architecture: x86_64 using:
>>   	  - PCLMULQDQ (carry-less multiplication)
>>   
>> +config CRYPTO_SM2_ZHAOXIN_GMI
>> +	tristate "SM2 Cipher algorithm (Zhaoxin GMI Instruction)"
>> +	depends on X86 && (CPU_SUP_CENTAUR || CPU_SUP_ZHAOXIN)
>> +	select CRYPTO_AKCIPHER
>> +	select CRYPTO_MANAGER
> 
> Why does this depend on CRYPTO_MANAGER?
> 
Yes, this is redundant and will be removed in the next version.

>> +static int zhaoxin_sm2_verify(struct akcipher_request *req)
>> +{
>> +	struct crypto_akcipher *tfm = crypto_akcipher_reqtfm(req);
>> +	struct sm2_cipher_data *ec = akcipher_tfm_ctx(tfm);
>> +	unsigned char *buffer;
>> +	int ret, buf_len;
>> +
>> +	buf_len = req->src_len + req->dst_len;
> 
> What if this overflows? I know you copied this from the generic sm2
> code, but that's still broken and both should be fixed up.
> 
> Thanks,
Yes, you are right. Thank you for your advice. The variable types will 
be adjusted according to the actual needs of the code in the next version.

Sincerely,
LeoLiu-oc