Open basavesh opened 1 month ago
One thing I noticed while comparing the llvm_disasm.cpp
with ghidra_disasm.cpp
is that llvm_disasm
calls the step
function recursively while ghidra_disasm
does not. I don't know if this relates to the exploration issue.
Edit: It looks like more handling the prefix case.
Hi,
Looks like the ghidra backend is not exploring the binary properly and missing a lot of instructions.
For example: Compile the attached mini word count file.
llvm-backend
However, the ghidra backend lifts much less instructions.
Source code:
@ivg