![]() VAMPIRE |
eBACS: ECRYPT Benchmarking of Cryptographic Systems |
![]() ECRYPT II |
| General information: | Introduction | eBASH | eBASC | eBATS | SUPERCOP | XBX | Computers |
| How to submit new software: | Hash functions | Stream ciphers | DH functions | Public-key encryption | Public-key signatures |
| List of primitives measured: | SHA-3 candidates | All hash functions | Stream ciphers | DH functions | Public-key encryption | Public-key signatures |
| Measurements indexed by machine: | SHA-3 candidates | All hash functions | Stream ciphers | DH functions | Public-key encryption | Public-key signatures |
There is a separate page (SHA-3 excerpt) showing the hash functions submitted to eBASH so far, and another page (SHA-3 excerpt) reporting measurements of those functions.
The top-level directory name crypto_hash is required; it distinguishes hashing from other operations benchmarked by SUPERCOP, such as crypto_stream and crypto_sign.
The second-level directory name md7 should be a lowercase version of your hash-function name. Please omit dashes, dots, slashes, and other punctuation marks; the directory name should consist solely of digits (0123456789) and lowercase ASCII letters (abcdefghijklmnopqrstuvwxyz).
Different hash functions must be placed into different second-level directories, even if they are part of the same "family" of hash functions. For example, crypto_hash/sha224 is separate from crypto_hash/sha256. One submission tarball can include several hash functions. Directory names may be changed by the eBASH managers to resolve conflicts or confusion.
The third-level directory name ref is up to you. Different implementations must be placed into different third-level directories. You can use subdirectories here; for example, crypto_hash/md7/ref might be a reference implementation, crypto_hash/md7/smith/little might be John Smith's little-endian implementation, and crypto_hash/md7/smith/sse3 might be John Smith's SSE3-optimized implementation. One submission tarball can include several implementations.
After choosing the implementation name crypto_hash/md7/ref, create a directory by that name. Inside the crypto_hash/md7/ref directory, create a file named api.h with one line
#define CRYPTO_BYTES 64
indicating that your software produces a 64-byte hash output.
You can also add a line
#define CRYPTO_VERSION "3.01a"
indicating that this is version 3.01a of your software;
SUPERCOP will report this information in its database of measurements.
Next, inside the crypto_hash/md7/ref directory, create a file named hash.c that defines a crypto_hash function:
#include "crypto_hash.h"
int crypto_hash(
unsigned char *out,
const unsigned char *in,
unsigned long long inlen
)
{
...
... the code for your MD7 implementation goes here
...
return 0;
}
Your function must have exactly the prototype shown here:
first an unsigned char pointer for the output,
then a const unsigned char pointer for the input,
then an unsigned long long for the number of bytes of input.
Your function must return 0 to indicate success,
or a negative number to indicate failure (e.g., out of memory).
The file crypto_hash.h is not something for you to write; it is created automatically by SUPERCOP. It defines a crypto_hash_BYTES macro (converted from your CRYPTO_BYTES macro) and a prototype for the crypto_hash function.
You can use names other than hash.c. You can split your code across several files *.c defining various auxiliary functions; the files will be automatically compiled together. You can, for example, have files md7.c and md7.h using NIST's SHA-3 submission API, along with a file nist.c containing
#include "crypto_hash.h"
#include "md7.h"
int crypto_hash(unsigned char *out,const unsigned char *in,unsigned long long inlen)
{
if (Hash(crypto_hash_BYTES * 8,in,inlen * 8,out) == SUCCESS) return 0;
return -1;
}
If you want a 32-bit unsigned integer type, you can
#include "crypto_uint32.h"
and then use crypto_uint32.
This is somewhat more portable then using stdint.h or inttypes.h.
SUPERCOP provides similar facilities for
uint8, uint16, uint32, uint64,
int8, int16, int32, int64.
Instead of hash.c you can write hash.cc or hash.cpp in C++ or hash.s or hash.S in assembly language. Assembly-language implementations must define a function crypto_hash_md7_ref instead of a function crypto_hash. (This is also true for C implementations but is handled automatically by a macro in crypto_hash.h.)
Finally, create a tarball such as md7-ref-3.01a.tar.gz that contains your crypto_hash/md7/ref/api.h, crypto_hash/md7/ref/hash.c, etc. Put the tarball on the web, and send the URL to the eBACS/eBATS/eBASC/eBASH mailing list with a note requesting inclusion in SUPERCOP and subsequent benchmarking.
Your implementation is allowed to be unportable. If it doesn't compile on a particular computer, SUPERCOP will skip it and select a different implementation for that computer.
For example, SUPERCOP already includes two SHA-256 implementations. One implementation, crypto_hash/sha256/ref, is a reference implementation written from the SHA-256 specification. Another implementation, crypto_hash/sha256/openssl, is a simple wrapper around OpenSSL. You can submit another SHA-256 implementation such as crypto_hash/sha256/smith/sse3 in the same way that the designer of MD7 can submit crypto_hash/md7/ref; by using the existing name sha256 you indicate that your implementation computes the same SHA-256 hash function.
In particular, submitters are encouraged to clearly specify one of the following levels of copyright protection: