3

I'm using python2.6's gdb module while debugging a C program, and would like to convert a gdb.Value instance into a python numeral object (variable) based off the instance's '.Type'.

E.g. turn my C program's SomeStruct->some_float_val = 1./6; to a Python gdb.Value via sfv=gdb.parse_and_eval('SomeStruct->some_double_val'), but THEN turn this into a double precision floating point python variable -- knowing that str(sfv.type.strip_typedefs())=='double' and its size is 8B -- WITHOUT just converting through a string using dbl=float(str(sfv)) or Value.string() but rather something like unpacking the bytes using struct to get the correct double value.

Every link returned from my searches points https://sourceware.org/gdb/onlinedocs/gdb/Values-From-Inferior.html#Values-From-Inferior, but I can't see how to convert a Value instance into a python variable cleanly, say the Value wasn't even in C memory but represented a gdb.Value.address (so can't use Inferior.read_memory()), how would one turn this into a Python int without casting string values?

Improve this question

2 Answers 2

Reset to default
5

You can convert it directly from the Value using int or float:

(gdb) python print int(gdb.Value(0))
0
(gdb) python print float(gdb.Value(0.0))
0.0

There seems to be at least one glitch in the system, though, as float(gdb.Value(0)) does not work.

Improve this answer
Sign up to request clarification or add additional context in comments.

3 Comments

Thanks! This does appear to work on my C values but doesn't work all the time e.g. int(xxxValue.address) fails -- I have to use int(str()). Guess I'll have to catch exceptions. Unrelated, can you set a gdb.Value of a pointed to address to another gdb.Value or gdb.Parameter without using gdb.parse_and_eval(xxx=newval)?
I think you can convert a pointer using long(). I don't think there is any way to assign right now without using parse_and_eval.
actually with python 2.7 you are better off using long() instead of int, as on the 32bit builds int will truncate the value to 32bits. As such I always use long().
0

I stumbled upon this while trying to figure out how to do bitwise operations on a pointer. In my particular use case, I needed to calculate a page alignment offset. Python did not want to cast a pointer Value to int, however, the following worked:

int(ptr.cast(gdb.lookup_type("unsigned long long")))

We first make gdb cast our pointer to unsigned long long, and then the resulting gdb.Value can be cast to Python int.

Improve this answer

Comments

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.