Missing handlers for certain bitfield access patterns

[flammensturm]

Hi there, so I'm working on a project with PDB (so full types, classes etc.) and for some reason the plugin doesn't parse bitfields (as it should do?)

Here is an example in picture:

And here you can see the bits in question :

No error in console, so I don't really understand what is the issue

[JustasMasiulis]

I think I know what's wrong and will see if I can fix it it after work today.

As a temporary fix, can you try turning the bitfield inside the struct into an union that also has a non-bitfield union { struct { ... }; uint8 flags; } member? I reckon my function for matching access simply doesn't handle the form where you have the + offset which does not happen when you have the non-bitfield member in an union which IDA would access.

[flammensturm]

For some reason this won't save (tho it seems right imo?)

Seems like I can't modify directly the class in question because it contains Unreal Engine types like FVector (which got imported from the pdb) and IDA doesn't recognize them so it just drop an error and won't let me modify it in any way.

[flammensturm]

Would have loved to make the changes myself btw but I don't have 9.0 SDK :/

[JustasMasiulis]

Made some changes to try and handle the case you're running into. Can you try out the latest release and let me know if it's fixed or if I missed something?

[flammensturm]

PER-FECT, works like a charm!

Thank you for the reactivity, I'm closing the issue as it's resolved, take care and thanks again!!

[flammensturm]

There's still a few leftovers, especially for setters like here on the screenshot and for one getter (bHidden) but overall it seems fine

[JustasMasiulis]

I am assuming you have display of casts disabled in the screenshot. Can you post one with them enabled just for the sake of clarity so I can make an accurate repro?

[flammensturm]

Sure, here you go

[JustasMasiulis]

Made another update. The code to handle the |= is not great though, so I would advise making a backup of the idb just in case.

[flammensturm]

Alright, seems to work on my side, why the |= code is not great in your opinion ? It looks like what I made yesterday on my side to also handle my case and I had no issue 🤔

I'll see if I can make a PR tho to handle values with |= and also handle &=.

[JustasMasiulis]

why the |= code is not great in your opinion?

Had an INTERR on the first iteration of it and it's just kind of copy pasted from another place and not clean. I obviously don't have much time to test that there's no regression either plus it uses different syntax from original. Because I can't actually refer to a member |= doesn't work directly as I'd guess from the INTERR it only expects numeric types on both sides.

I'll see if I can make a PR tho to handle values with |= and also handle &=.

Always open to improvements. From my experience the only thing that bugs me is that in some cases I actually need to invert the generated bitmask for it to make more sense. I.E instead of & (A | B | C | D | ...) generate & ~F to reduce the noise.

[flammensturm]

Made my own handling for values here, it's showing it as a string like here

Didn't succeed to make it as a bool (tho it doesn't rly change anything to the result), I'm new to the IDA api and the documentation is kinda shit. Feel free if you have suggestions to make it cleaner so it can make it into a PR