Cryptographic Token Interface Standard |
PKCS#11 |
A number of the functions defined in Cryptoki return output produced by some cryptographic mechanism. The amount of output returned by these functions is returned in a variable-length application-supplied buffer. An example of a function of this sort is C_Encrypt, which takes some plaintext as an argument, and outputs a buffer full of ciphertext.
These functions have some common calling conventions, which we describe here. Two of the arguments to the function are a pointer to the output buffer (say pBuf) and a pointer to a location which will hold the length of the output produced (say pulBufLen). There are two ways for an application to call such a function:
Cryptographic functions which return output in a variable-length buffer should always return as much output as can be computed from what has been passed in to them thus far. As an example, consider a session which is performing a multiple-part decryption operation with DES in cipher-block chaining mode with PKCS padding. Suppose that, initially, 8 bytes of ciphertext are passed to the C_DecryptUpdate function. The blocksize of DES is 8 bytes, but the PKCS padding makes it unclear at this stage whether the ciphertext was produced from encrypting a 0-byte string, or from encrypting some string of length at least 8 bytes. Hence the call to C_DecryptUpdate should return 0 bytes of plaintext. If a single additional byte of ciphertext is supplied by a subsequent call to C_DecryptUpdate, then that call should return 8 bytes of plaintext (one full DES block).