On Tue, Apr 15, 2025 at 09:05:20PM +0000, Qing Zhao wrote:
> > On Apr 15, 2025, at 16:35, Kees Cook <k...@kernel.org> wrote:
> > 1) When applying counted_by to pointer members, are out-of-order member
> > declarations expected to be handled? As in, is this expected to be valid?
> >
> > struct foo {
> > struct bar *p __attribute__((counted_by(count)));
> > int count;
> > };
>
> Yes, this is valid.
> Given a lone identifier, you will use “counted_by” attribute the same as the
> counted_by for FAM, and this identifier will be looked up inside the
> enclosing
> structure. (Even when the counted_by field is AFTER the pointer).
>
> The patch I sent a while ago on the counted_by for pointer filed of a
> structure:
> https://gcc.gnu.org/pipermail/gcc-patches/2025-January/673837.html
>
> Have supported this without issue.
Hurray! Okay, I think that will make this quite friendly to use.
> > 4) If there are type mismatches between the counted_by_expr struct
> > member declaration and the later actual struct member declaration, I
> > assume that will be a hard error. For example, this would fail to compile:
> >
> > struct foo {
> > struct bar *p __attribute__((counted_by_expr(int count; count)));
> > unsigned long count;
> > };
> >
> > Is that correct?
>
> I guess that the hard error will be issued when the parser sees “unsigned
> long count” which
> Is conflict with the previous seen “int count” inside counted_by_expr.
>
> But I am not sure on this.
I guess the mismatch could be ignored/warned as long as the actual
struct member type is used in the end.
Thanks!
--
Kees Cook
