Open mumbel opened 4 years ago
At the moment there is no work-around or pointer datatype attribute on the java-side which would identify a type as a code
pointer. We are looking into ways we can incorporate this and other datatype attributes into the API.
At the moment there is no work-around or pointer datatype attribute on the java-side which would identify a type as a
code
pointer. We are looking into ways we can incorporate this and other datatype attributes into the API.
Will these new datatype attributes contain the ability to have: 1) different sized pointers (address sizes) - for NEAR & FAR attribute (in my case 16- to 32-bit); and 2) ability to "widen" a NEAR pointer to a FAR one, by your choice of address offset - for passing across functions.
Will these new datatype attributes contain the ability to have:
- different sized pointers (address sizes) - for NEAR & FAR attribute (in my case 16- to 32-bit); and
- ability to "widen" a NEAR pointer to a FAR one, by your choice of address offset - for passing across functions.
Hopefully, but at this point it is just a concept.
Through analysis ghidra can decide that a pointer is code and will cast function pointers sometimes as
code
.This issue is simply readability in the decompiler only (though maybe could help label functions through type propagation). You may see something like
thing->func = (undefined *)0x12345678
instead ofthing->func = FUN_12345678
even though the listing windows does have it listed asFUN_12345678
.code
however is not a selectable DataType. The only other generic options arepointer
,void *
, andundefined *
; but none of these would explicitly saythis is a pointer to code
ascode
would. Short of creating function definitions is there a built-in data type that I'm just not aware of to help with this.