#lisp - Sun 8 Apr 2007 between 00:38 and 01:33

NY Lost Funds



nyefDoes anyone know of any documentation on this (sbcl/cmucl) disassembler that I'm just unable to think of right now?
slyrusnyef: looks like sb-show builds are busted
nyefOh, fun.
Binary chop time?
(The good thing from my perspective is that you're seeing it on x86-64.)
slyrusnyef: no, it's x86 too
nyefYes, but consider my two most recent commits.
slyrusbut, yeah, if I can't find anything obvious, I'll start chopping
tomorrow anyway...
nyefI don't have to worry if it's x86 too, so long as it's not x86-only.
actionnyef goes back to staring at the disassembler trying to figure out how to get it to do what he wants it to without having to modify the actual disassembler guts.
nyefgoes back to staring at the disassembler trying to figure out how to get it to do what he wants it to without having to modify the actual disassembler guts.
Hunh. Looks like I might not have to be quite so sneaky as I thought.
Now, is this functionality actually -used- anywhere?
Not obviously...
Hrm... Survived first genesis. That's promising...
Oops. Forgot a declare ignore. :-/
"This time for sure!"
rtoymnyef: I know a little about the disassembler. But the x86 version is just to complex for me to grok.
nyefrtoym: Know anything about "prefix instructions"?
rtoymHold on...
nyefI have something which I think might work, just waiting for a fresh build.
AFAICT, they're instructions without a printer.
And their control function can modify the dstate.
rtoymDo you have an example?
nyefI'll let you know once this build completes.
Wasn't able to find anything obvious, so I'm sortof working from first principles here.
rtoymThere's some function that prints out all the instructions nicely grouped. I've used that to figure out why my disassembler wasn't working.
nyefAnd reading the almost completely undocumented mess behind the :PRINTER option for DEFINE-INSTRUCTION.
Thing is, either an arg prefilter or a control has to do the actual tweaking of the dstate. And I haven't seen anything that does that yet.
At least, not in that way.
faxis "(unless (find-if-not ...", a bad thing?
rtoymnyef: print-inst-space prints out the instruction set.
nyeffax: It feels like a double negative.
faxI know :[
its just a bit clunky as well
do you think it matters?
nyeffax: But beyond that my head is so full of bits of disassembler goop that I can't think of anything better.
kpreidfax: are the elements of the sequence possibly nil?
faxit can happen, but I should probably just skip the whoel thing in tht case
nyefOkay, insts.lisp compiled this time, which is a good sign.

Page: 2 9 16 23 

IrcArchive

NY Lost Funds