Sega MD SMD ROM格式文件

Bart Trzynadlowski


==============================================================================
Sega MD SMD ROM FORMAT DOCUMENTATION
Written by Bart Trzynadlowski 1999
==============================================================================
精确的SMD格式技术描述,中级者使用

This document describes the SMD Genesis ROM format. I wrote it because I found
that XnaK's documentation is wrong. This text is best viewed in MS-DOS Editor
or any other decent ASCII text viewer.
Much thanks goes to Kuwanger (http://members.tripod.com/~Kuwanger) for
figuring out that the documentation I was using regarding SMD was incorrect.
Felipe XnaK (http://www.classicgaming.com/launchtool) wrote the original text
I used on the many Genesis ROM formats.

0. SUPER MAGICDRIVE INTERLEAVED ROM FORMAT
The SMD format is a Genesis ROM format which must be decoded to be manipulated
as normal Genesis cartridge data. SMD files consist of a series of 16KB blocks
of data with their odd bytes at the beginning, and even bytes at the end. They
also have a header which is 512 bytes long.

Important SMD Header Bytes:

Offset 00h: Number of 16KB blocks. This number may be incorrect in
some files, it is recommended you calculate it
manually: num_blocks = (sizeof(file) - 512) / 16384
Offset 02h: Indicates wether ROM is a part of a series of a split
ROM (1, or possibly just non-zero,) or wether it is
a standalone ROM or the last ROM in a split series (0)
Offset 08h: AAh
Offset 09h: BBh

* Note: Some documentation claims that byte 1 is always 3 and all
other bytes besides those mentioned as important are 0. This is not
always the case. The header information in this document should be
considered fairly accurate.

Each 16KB block has the data from each odd address at the beginning, and the
data for each even address at the end.

Decoding a 16KB SMD Block:

1. If the byte offset in the block is less than 8192, copy the byte
from the SMD block to the first unused odd offset in the decode
buffer.
2. Otherwise, put it in the first unused even offset in the decode
buffer.

Below is example C source code for an SMD-to-BIN converter:

/* convert smd to bin */
for (i = 0; i < header[0]; i++)
{
e = 0;
o = 1;
fread(smd_block, 1, 16384, in_fp);
for (j = 0; j < 16384; j++)
{
if (j < 8192)
{
bin_block[o] = smd_block[j];
o += 2;
}
else
{
bin_block[e] = smd_block[j];
e += 2;
}
}
fwrite(bin_block, 1, 16384, out_fp);
} 

In the code snippet, e represents the first unused even offset (which
will always start at 0) in the 16KB bin_block array. Thus o represents
the first unused odd offset (which will always start at 1) in the 16KB
bin_block array. The for (i) loop uses the information from the header
to determine how many blocks to decode. The for (j) loop actually does
the individual block decoding. It should be fairly self-explanatory
from there. The method shown is a linear way of doing it, there are
faster ways of doing it -- for example, you could decode two bytes in
one loop, an even and an odd.

Creating an SMD file is the reverse of decoding one.

高晶 □ 葵花宝典 □ 游戏杂谈