BUGZ is the worlds' smallest Z80 disassembler. It is written in C89 and should compile on just about any machine with a standard library.
One required function that might be missing in older implementations is the
bsearch. The GeeksForGeeks site provides the default implementation.
BUGZ is a table disassembler. Each instruction in the instruction table
has a name (a string, with printf compatible arguments for operands),
a 32 bit "opcode" (the longest instruction in Z80 is exactly 4 bytes),
and a 32 bit mask to tell which part of the "opcode" is instruction,
which the operand(s), and which can be ignored.
Opcode mask byte values can be:
| mask byte | meaning |
|---|---|
| 00 | Ignore this byte |
| 01 | Opcode (you have o match this part) |
| 02 | 8 bit operand |
| 03 | Low byte of 16 bit operand |
| 04 | High byte of 16 bit operand |
| 05 | 8 bit jump operand (JR, DJNZ) |
| 06 | Index register offset operand (+/-) |
An instruction DJNZ %hhd will have opcode of 0x00000010, and
a mask of 0x00000501. First byte of the mask is 0x01 which means
that first byte in memory must equal to first byte of the opcode
(i.e must be 0x10). Second byte of the mask is 0x05 which means
second byte in memory is an 8 bit jump offset value. Other bytes in
mask are 0x00 and are ignored. Hence if you are examining a memory
address and the first byte is 0x10 then take the second byte as
a signed byte and use it as argument to printf, together with the
DJNZ %hhd instruction format string.
For convenience, every instruction in the instruction table also contains instruction size (i.e. how many bytes are valid).
Make sure you have make and gcc tools installed. Open the terminal,
go to the bugz directory, and type -
make
The main.c file contains a main() function that accepts one argument -
the name of the binary file. It reads it and dump its disassembled code
on the standard output.
The 48.rom file is a sample Z80 binary (ZX Spectrum 48K ROM image).
After building BUGZ run it like this -
./bugz 48.rom
The instruction table also contains t-states for each instruction. This data is currently not used (I wanted to keep the source code small and simple!), but the disassembler can easily be extended to include this information into the final listing.