Sprinter/mame
2
0
Fork 0
mirror of https://github.com/holub/mame synced 2025-05-08 23:31:54 +03:00
Code Issues Actions Packages Projects Releases Wiki Activity

removed some obsolete commentaries. Not worth.

Browse Source
This commit is contained in:
Andreas Naive 2011-06-17 23:34:53 +00:00
parent 697eff192e
commit 4e4ff5da98
1 changed files with 2 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
src/mame/machine/naomibd.c
View File

@ -1436,7 +1436,7 @@ M1: DMA read of protected ROM area
M2: special read of ROM area which supplies decryption key first
M3: normal read followed by write to cart's decryption buffer (up to 64kB), followed by M2 but from buffer area
M1's working is still unclear (more on this later), so we will be speaking of M2 & M3 most of the time.
Notes below refer to M2 & M3.
The encryption is done by a stream cipher operating in counter mode, which use a 16-bits internal block cipher.
@ -1454,7 +1454,7 @@ internal block-cipher. So, at a given step, the internal block cipher will outpu
given plaintext word, and the remaining 2 to the next plaintext word.
The underlying block cipher consists of two 4-round Feistel Networks (FN): the first one takes the counter (16 bits),
the game-key (>=25 bits) and the sequence-key (16 bits) and output a middle result (16 bits) which will act as another key
the game-key (>=26 bits) and the sequence-key (16 bits) and output a middle result (16 bits) which will act as another key
for the second one. The second FN will take the encrypted word (16 bits), the game-key, the sequence-key and the result
from the first FN and will output the decrypted word (16 bits).
@ -1484,19 +1484,6 @@ It can be observed that a couple of sboxes have incomplete tables (a 255 value i
as of december/2010 show small randomness and big correlations, making possible that some unseen bits could make the
decryption need those incomplete parts.
When bit #1 of the heading control bits is set to 1, an additional decompression step seems to be carried out. As of
february/2010, Deunan Knute has put some work on analyzing the decompression algorithm, but probably much more work will
be needed to completely reverse engineer it. Interested devs with quick access to hardware tests are welcomed to join
the task.
Guilty Gear X & Sega Tetris are examples of games using the decompression ingame.
Due to technical details, it's more difficult to get custom decryption data from M1 carts, which hinders some types of
analysis. The only M1 cart which have received attention until now have been AH! MY GODDESS. The available data clearly
doesn't have the same structure than M2&M3 carts using only the decryption step. However, due to that some hardware tests
show cycled structures similar to the ones returned by M2&M3 carts using the decompression algo, it's conjectured that M1
carts will be using the same decompression algorithm seen in M2&M3 ones.
****************************************************************************************/
struct sbox