Scott <scott.pig...@gmail.com> wrote:
> I think I found the answer to my own question. Can anyone spot any
> issues with the following solution? The application I'm writing will
> be hitting these callbacks pretty heavily so I'm nervous about mucking
> up the memory management and creating one of those bugs that passes
> undetected through testing but nails you in production.
>
> def my_callback(p_cstring):
> answer = 'foobar'
>
> address = VENDOR_malloc(len(answer)+1)
>
> cstring = c_char_p.from_address( address )
I think this allocates the pointer (the c_char_p) in the malloced
block, not the actual data...
> cstring.value = answer
And this overwrites the pointer
> p_cstring.contents = cstring
> return
If you try this, it gives all sorts of rubbish data / segfaults
memmove(address, address+1, 1)
Here is how I'd do it
from ctypes import *
from ctypes.util import find_library
c_lib = CDLL(find_library("c"))
malloc = c_lib.malloc
malloc.argtypes = [c_long]
malloc.restype = c_void_p
answer = 'foobar\0'
address = malloc(len(answer))
print address
cstring = c_char_p()
print addressof(cstring)
cstring.value = address
memmove(address, answer, len(answer))
print cstring.value
memmove(address, address+1, 1)
print cstring.value
Which prints
159611736
3084544552
foobar
ooobar
--
Nick Craig-Wood <n...@craig-wood.com> -- http://www.craig-wood.com/nick
--
http://mail.python.org/mailman/listinfo/python-list
