This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] add alloc_size attribute


On 5/9/07, Jakub Jelinek <jakub@redhat.com> wrote:
On Wed, May 09, 2007 at 03:15:47PM +0200, Dirk Mueller wrote:
> On Wednesday, 25. April 2007, Richard Guenther wrote:
>
> > Otherwise this looks ok.  Please make the suggested changes and
> > re-post the patch.
>
> bootstrapped/regtested on x86_64-suse-linux
>
> 2007-04-25  Dirk Mueller  <dmueller@suse.de>
>         Marcus Meissner <meissner@suse.de>
>
>         * doc/extend.texi (alloc_size): New attribute.
>         * c-common.c (handle_alloc_size_attribute): New.
>         * tree-object-size.c (alloc_object_size): Use alloc_size
>         attribute, if available.
>
>         * testsuite/gcc.dg/attr-alloc_size.c: New.

Good idea, I just wonder if it wouldn't be better to extend the malloc
attribute for this instead (i.e. it could have no arguments as it
does now, or one or two as your alloc_size argument does).

I think we should not overload the malloc attribute, as we would be unable to handle realloc in this case.

Or are there functions which should not have malloc attribute set
and can have alloc_size?

realloc comes to my mind. Possibly no others.


Richard.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]