97b3db1
This exact patch gives 2% performance decrease on my RISC-V processor.
test:

openssl speed -evp kuznyechik-ecb -seconds 5

I do not understand how a function that is called once may affect performance that much.