Jonathan Salwan
Jonathan Salwan
I hope you recovered well! > he example that @aris-bb gave is sufficient for me. Let's close this issue then
Thanks for all those feedback
> yet Triton still call the operand callbacks. In a concrete point of view I agree. But if we want to provide a correct symbolic expression, we need to know...
You can use some modes: ```python setMode(ALIGNED_MEMORY, True) setMode(CONSTANT_FOLDING, True) setMode(ONLY_ON_SYMBOLIZED, True) ```
Yeah, everything depend on what you symbolize. More you symbolize more you will consume memory and on million instructions, you pay the cost on your memory. The max I did,...
That's exactly what I want in long term. * ctx.processing(inst) * ctx.disassembly(inst) * ctx.buildSemantics(inst) * ctx.executeSemantics(inst)
Ok, for a first try it looks not too bad. Below, a patch for the aarch64 semantics. The idea is to make all expressions volatile, and then, assign these expressions...
Some brainstorming with @RobinDavid * `API::disassembly(inst)` returns a static disassembly information and a static IR without any assignment (standalone object). * `API::buildSemantics(inst)` takes the static IR crafted in `disassembly` and...
Note to myself: To implement a static IR we must not used `getConcrete*()` in `*Semantics.cpp`. So, we have to implement LOAD and STORE nodes (#806).
> It would be useful if the Python binding has the execCallbacks parameter for me to specify this behavior. Indeed. I'm in vacation atm but as soon as I came...