85 lines
2.8 KiB
Text
85 lines
2.8 KiB
Text
|
Rijndael
|
||
|
|
Joan Daemen
|
||
|
|
|
||
|
|
Optimised ANSI C v2.4
|
||
|
|
-----------------------------------------------------------
|
||
|
|
|
||
|
|
This archive contains the following files:
|
||
|
|
|
||
|
|
Makefile: A file that allows for easy compiling of the code with
|
||
|
|
Unix `make' (tested with GNU make).
|
||
|
|
README: This file.
|
||
|
|
boxes-fst.dat: Tables that are needed by the reference implementation.
|
||
|
|
The tables implement the S-box and its inverse, and also
|
||
|
|
some temporary tables needed for multiplying in the finite
|
||
|
|
field GF(2^8).
|
||
|
|
rijndael-alg-fst.c:
|
||
|
|
rijndael-alg-fst.h:
|
||
|
|
Algorithm implementation.
|
||
|
|
rijndael-api-fst.c:
|
||
|
|
rijndael-api-fst.h:
|
||
|
|
Interface to the C API.
|
||
|
|
rijndaeltest-fst.c:
|
||
|
|
Implementation of the KAT and MCT.
|
||
|
|
table.128:
|
||
|
|
table.192:
|
||
|
|
table.256:
|
||
|
|
Files needed for the KAT (for the Table Known Answer Test).
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
Instructions for the KAT and MCT software:
|
||
|
|
|
||
|
|
1) Compile the C code and put the executable in the same directory as the
|
||
|
|
table.??? files.
|
||
|
|
2) Run the executable. It generates all the tables in the NIST format.
|
||
|
|
3) Compare the generated tables with the original provided tables, e.g.
|
||
|
|
in Unix, with `diff'.
|
||
|
|
4) The code is independent of the ENDIANness of the machine.
|
||
|
|
However, the code casts char pointers to int pointers.
|
||
|
|
On some platforms, this casting can cause problems.
|
||
|
|
This will cause malfunctioning of the CBC Monte Carlo tests.
|
||
|
|
In this case, the compile parameter STRICT_ALIGN should be set to 1.
|
||
|
|
(Of course this results in slower code.)
|
||
|
|
|
||
|
|
|
||
|
|
Changes with respect to v1.0 (= round 1 submission)
|
||
|
|
---------------------------------------------------
|
||
|
|
|
||
|
|
1) Removed the parameter blockLen from makeKey() and cipherInit().
|
||
|
|
The parameter is still present in the structures keyInstance and
|
||
|
|
cipherInstance.
|
||
|
|
2) Removed some calls to sscan().
|
||
|
|
3) Corrected the CBC decryption Monte Carlo test code.
|
||
|
|
4) Provided code for platforms with pointer alignment problems.
|
||
|
|
|
||
|
|
----------------------------------------------------------------------
|
||
|
|
|
||
|
|
Changes from v2.0 (= round 2 submission)
|
||
|
|
----------------------------------------
|
||
|
|
|
||
|
|
Various bug fixes in CBC encryption and decryption mode.
|
||
|
|
|
||
|
|
----------------------------------------------------------------------
|
||
|
|
|
||
|
|
Changes from v2.2
|
||
|
|
-----------------
|
||
|
|
|
||
|
|
-- Moved number of rounds to the keyInstance structure
|
||
|
|
to make implementation thread-safe
|
||
|
|
-- Functions used only to generate the Intermediate Value
|
||
|
|
Known Answer Test are now conditionally compiled
|
||
|
|
via #define INTERMEDIATE_VALUE_KAT
|
||
|
|
-- Further optimized for the default 128-bit block size
|
||
|
|
-- Fixed CBC decryption bug (chained IV was lost if input
|
||
|
|
and outBuffer were the same)
|
||
|
|
-- Added padding capability to CBC mode (allows encryption
|
||
|
|
of any number of octets, not only full blocks)
|
||
|
|
|
||
|
|
----------------------------------------------------------------------
|
||
|
|
|
||
|
|
Changes from v2.3
|
||
|
|
-----------------
|
||
|
|
|
||
|
|
Fixed a small bug in CBC mode.
|