Re: Creating a dis-assembler on my own - Project killed ...

From: R.Wieser (rwieser-killthis-_at_xs4all.nl)
Date: 01/20/04

  • Next message: R.Wieser: "Re: Creating a dis-assembler on my own - Project killed ..." 
    Date: Tue, 20 Jan 2004 01:41:16 +0000 (UTC)

    Gerhard W. Gruber <sparhawk@gmx.at> schreef in berichtnieuws
    88co0090vojv8qi91tsl2t9u307o6q4rsa@4ax.com...

    Hello Gerhard,

    > On Sun, 18 Jan 2004 22:15:23 +0000 (UTC) wrote "R.Wieser"
    > <rwieser-killthis-@xs4all.nl> in comp.lang.asm.x86 with
    > <400b0399$0$142$e4fe514c@dreader5.news.xs4all.nl>
    >
    > > Thanks for all that took the time to think with me. But alas, due to
    not
    > >being able to find/generate/cajole a reliable list of commands (both in
    > >binary and readable form), I decided to stop.
    >
    > This was a very short project lifetime. :)

    In day's perhaps, not in the hours that I have put in it :-) Don't forget
    that programming was not the problem. Finding accurate, complete data is.

    > >As long as I can't check the output of the disassembler I'm trying/tried
    to
    > >write against known correct data all my coding is in vain ...
    >
    > But you can. You can always launch a debugger, or several to cross check,
    and
    > see what they are disassmebling. That's how I verified it when I was not
    sure.

    I hate the idea to have to check everything by hand. If god would have
    wanted that, he would not have given us the computer :-)

    What you suggest is exactly what I've done. I've taken two of them, next to
    the paper-data I allso have. All three of them had blatant errors. And
    yes, I've looked at Intel's papers too. Do you know I can't even *find* the
    FFh-section ? Oh, well. Maybe it's just too late for things like that
    (2.17 in the morning) :-)

    > Of course you have to read the manuals and really understand them. It took
    me
    > quite some time until I understood how the various bits and pieces in an
    > instruction worked, but the Intel Manuals are all that I needed for this.
    > I only had to read them up and again and again until it finally was clear
    to me.
    > Of course you can also ask in a newsgroup like this and use other tools to
    > verify your output. For consistency checks I would recommend a
    testfunctions
    > like the one I posted.

    Actually, I'm aiming at a diff-check between a known good source, and my own
    disassembler. That should flag any problems at the spot.

    > >So, If *anyone* knows where to get a good *full* and *uncontaminated*
    list
    > >of possible commands (preferrable sorted or marked by processor), or
    knows
    >
    > It would be really nice to have something like that. I found some list
    > somewhere on the net, but I don't remeber exactly where. Of course you
    always
    > have to expect that these can also contain errors.

    Errors I do not really mind. But having to look at commands like "call far
    al/ax" (output of both of the debuggers), and realize that there could be
    more there that just could slip by .... Most of the lists do not even go as
    far as to display details of an FFh-code.

    > >of a way to generate one (have allready tried Debug.exe & Borland's
    > >Turbo-Debugger. Both emitted, at one point or another, crap), I would
    sure
    > >like to hear about it ... :-)
    >
    > Both of these are not 32 Bit and quite outdated.

    I know. But do they need to be ? And if yes, why ?

    Actually, I was trying to create a list with 8086 commands first, and see
    what changes with the next processor.

    > I think Borland can do 32 Bit
    > protected mode instructions but it still is quite outdated, so I would
    > recommedn something a little bit newer.

    TD32 perhaps ?

    > Either SoftICE, GDB, or windbg should
    > do fine. My favourite is SoftICE but it is not exactly cheap, so GDB or
    windbg
    > should be a good solution because they are free.

    Actually, I never bother using a debugger. Most problems will reveal
    themselves nicely using some extra output. The only reason I now use them
    (and will probably use them at all) is to generate a reference list.

    Regards,
      Rudy Wieser

  • Next message: R.Wieser: "Re: Creating a dis-assembler on my own - Project killed ..."

    Relevant Pages