Machine code parsers (entropy of machine code)

Hello,


Initially I posted the following to comp.compression but noone seems to
have any information. I don't really expect you guys to have it but it
doesn't hurt to ask:


Does anyone know what is the usual entropy of the following:


1. Assembly code generated by widely used compilers for Pentium and PowerPC
platforms. Different compiler optimization options that may affect this
parameter are consideration. For Pentium only 32-bit code is interesting.
2. Code parts in the operating system which are likely to be written in
Assembly (not generated by compiler).


The alphabets of interest are:


1. Instruction opcodes + data set (all intgr numbers in [0..2^32-1]). The
data set may be different.
2. Instruction opcodes only. Data associated with the instructions is
completely ignored (skipped).
3. Data only. Instruction opcodes are completely ignored (skipped).
4. Code treated as 32-bit data. No instructions or data are recognized.


Alternatively, if someone can point me to a free-ware/share-ware machine
code parsers (if there are any) I can try to build these statistics myself.


I am also interested in the relation between the entropies of a line of C
code and its machine code representation. For this I would need a compiler
for the above mentioned platforms. I don't believe there are compilers for
such "hot" platforms distributed as source code, but can you guys confirm
that?


If anyone has this data or can point to a reference, it will be greatly
appreciated.


Thanks,
Andrey
[For a code parser, try taking the dissasembler our of gdb or any other
debugger you have at hand. -John]






--

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