[ocaml-ctypes] Lifecycle of a callback?

[Trevor Smith]

Thanks for the quick response Jeremy.

On Thursday, October 23, 2014, Jeremy Yallop <yallop at gmail.com> wrote:

> On 23 October 2014 11:48, Trevor Smith <trevorsummerssmith at gmail.com
> <javascript:;>> wrote:
> > I tried this out this morning. However I am seeing strange behavior with
> the
> > memory addresses. In native code, whenever I pass a pointer through the
> > accessor I receive a different address than the original address --
> however,
> > in valgrind the addresses are the same.
> >
> > Relevant code:
> > git clone git at github.com:trevorsummerssmith/ocaml-libuv.git && cd
> > ocaml-libuv && git checkout origin/test-lifecycle && make tests
> >
> > I see, eg:
> >
> > loop: 0x109d76800
> > Handle: 0x9d76800
> > Handle: 0x9d76800
> > F.
> >
> ==============================================================================
> > Failure: handle suite:0:accessors
> >
> > The loop address always starts with '10'. However in valgrind the
> addresses
> > are the same.
> >
> > The c accessors are in lib_gen/lib_accessors.c, the test is in
> > test/test_handle.ml
>
> The code generated by ctypes doesn't have declarations  for your
> accessor functions in scope, so the C compiler helpfully assumes that
> the functions return (32-bit) ints rather than (64-bit) pointers,
> leading to garbage in the upper bits.  You should be able to fix the
> problem by outputting suitable declarations (perhaps by including a
> header) in libuv_bindgen.ml.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ocaml.org/pipermail/ctypes/attachments/20141023/18a9815c/attachment.html>