Closed stellarpower closed 4 weeks ago
@stellarpower Hmmm, just as a test case, maybe try unselecting “dbgmodel” when you start the target.
Thanks, I think I did. The little checkbox next to the path to the engine?
Yes, correct - the stack suggests you’re using generic containers, which, if I remember correctly, only happens if you’re using “dbgmodel” vs. straight “dbgeng”. Related: do you have the newest version of windbg installed, i.e. what was called Windbg Preview, and/or the current WDK or are you running from a base Windows 10 install?
Verified the error and put in a fix. Even with the fix, however, I would not recommend using "dbgmodel" with the default Windows 10 install as a good bit of the functionality in the default C:\Windows\System32\dbgmodel.dll is missing. If you can, install the latest windbg or WDK and point the dbgeng directory parameter at their versions. If not, maybe stick to "dbgmodel" off.
Describe the bug Trying to debug a Windows application analysed with Ghidra. I've set up Python (3.12; my application is 32-bit but capstone is a dependency of pybag and was not happy using 32-bit python, which I initially guessed might cause less friction given this is Windows) and installed pybag and protocol buffers as per the F1 help.
To Reproduce I've spent all afternoon trying to get the debugger to start - so I don't know enough about ghidra or the full setup to design an MRE.
I start the debugger and I get one of python's famously uneasy-on-the-eyes backtraces to the terminal:
Based on the contents of that file, I simply wrapped the line in an exception handler and return if not present. IDK if the check for None was meant to handle a bad access into the hashmap or conversely if the key is always expected but nillable and thus we have a problem (i.e. in ruby speak, if
[]
should really have been[]?
but the original author didn't hit the case when this was missing). I am one step further along for now so using as a workaround.Environment (please complete the following information):