Binary File Decompiler To CPG

[hac425xxx]

currently joern support ghidra2cpg, with use ghidra to load binary to cpg.

but it seems only use the assembly instruction, why not use the decompiler infomation?

  Call(
    argumentIndex = -1,
    argumentName = None,
    code = "SUB RSP,0x30",
    columnNumber = None,
    dispatchType = "STATIC_DISPATCH",
    dynamicTypeHintFullName = IndexedSeq(),
    lineNumber = Some(value = 1053724),
    methodFullName = "<operator>.subtraction",
    name = "<operator>.subtraction",
    order = 0,
    possibleTypes = IndexedSeq(),
    signature = "",
    typeFullName = "<empty>"
  ),
  Call(
    argumentIndex = -1,
    argumentName = None,
    code = "MOV dword ptr [RBP + -0x14],EDI",
    columnNumber = None,
    dispatchType = "STATIC_DISPATCH",
    dynamicTypeHintFullName = IndexedSeq(),
    lineNumber = Some(value = 1053728),
    methodFullName = "<operator>.assignment",
    name = "<operator>.assignment",
    order = 0,
    possibleTypes = IndexedSeq(),
    signature = "",
    typeFullName = "<empty>"
  ),

[itsacoderepo]

Hello hac425xx,

This approach is intentional because some people read assembly, searching for patterns and other specific details. While this may not be your use case, it's perfectly fine :)

However, you can use decompiled code as input for c2cpg; there's no reason it shouldn't work, aside from potential bugs or missing information.

[hac425xxx]

Thank you very much, I understand what you mean, if use the decompiled output c file as the input of c2cpg, there may be many errors.

My idea is to build CPG directly using the decompiler's AST, so the result should be more accurate