Re: Writing a disassembler ?

Hey all, thanks for all for your replies.
I will likely consider all of them once I'll get to those stages, as
for now - I'm working on the decoder itself, which seems to be the
earliest stage of the disassembler itself, and I'm having a bit
trouble.


I've started building the structure of the decoder, starting from
reading raw opcodes stream(say, from the constructor
it has been yet decided)
I've created five classes, each for a specific "section" within an
instruction as follows:


1. A prefix class, with a byte prefix which identifies which prefix is
it (if at all)
2. An opcode class with byte[2] opcode, and boolean flags for
is_twobyte, is_modrm,has_disp,has_imm
3. A modrm class with a byte modrm and sib, along with boolean flags
for one, two, or four displacements
4. A displacement class with a byte displacement and boolean whether
it's two byte or four byte displacement (if all false it's obviously a
one byte)
5. An immediate class with the same variables as the previous one


and one class (instruction) which gathers them all together and
performs the checks with a result string and a long address
variable(for BB). I'm unsure whether where I should create the opcode
table (one byte opcodes for modrm/imm/disp fields, two byte etc ) and
how I'll represent it, either inside the opcode class or the
instruction. I got really lost in this side of the stream.
It seems that this is the main part of the decoder, afterwards the
representation (to the user screen) or something like it along with a
big while (1) loop goes into, if anyone could reference me or guide me
about this huge table I'd be fond

Return to the comp.compilers page.
Search the comp.compilers archives again.