Re: [PATCH 4/4 v5] fortify: Add Clang support

[Nick Desaulniers]

On Tue, Feb 1, 2022 at 4:30 PM Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>
> --- a/security/Kconfig
> +++ b/security/Kconfig
> @@ -179,7 +179,7 @@ config FORTIFY_SOURCE
>         depends on ARCH_HAS_FORTIFY_SOURCE
>         # https://bugs.llvm.org/show_bug.cgi?id=50322
>         # https://bugs.llvm.org/show_bug.cgi?id=41459
> -       depends on !CC_IS_CLANG
> +       depends on !CC_IS_CLANG || CLANG_VERSION >= 130000

Are these comments still relevant, and is the clang version still correct?

In https://lore.kernel.org/llvm/CANiq72n1d7ouKNi+pbsy7chsg0DfCXxez27qqtS9XE1n3m5=8Q@xxxxxxxxxxxxxx/
Miguel notes that diagnose_as only exists in clang-14+.  If this
series relies on diagnose_as, then should this version check be for
clang-14+ rather than clang-13+?

https://bugs.llvm.org/show_bug.cgi?id=50322 is still open, but doesn't
signify why there's a version check. It makes sense if there's no
version check, but I'm not sure it's still relevant to this Kconfig
option after your series.

https://bugs.llvm.org/show_bug.cgi?id=41459 was fixed in clang-13, but
it was also backported to the clang 12.0.1 release.  Is it still
relevant if we're gated on diagnose_as from clang-14?

Perhaps a single comment, about the diagnose_as attribute or a link to
https://reviews.llvm.org/rGbc5f2d12cadce765620efc56a1ca815221db47af or
whatever, and updating the version check to be against clang-14 would
be more precise?
-- 
Thanks,
~Nick Desaulniers