This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Fix misuse of always_inline attribute in 3 hotpatch tests
- From: Jakub Jelinek <jakub at redhat dot com>
- To: gcc-patches at gcc dot gnu dot org
- Date: Mon, 1 Jun 2015 12:29:31 +0200
- Subject: Re: [PATCH] Fix misuse of always_inline attribute in 3 hotpatch tests
- Authentication-results: sourceware.org; auth=none
- References: <20150601083934 dot GH10247 at tucnak dot redhat dot com> <20150601102029 dot GA18892 at linux dot vnet dot ibm dot com>
- Reply-to: Jakub Jelinek <jakub at redhat dot com>
On Mon, Jun 01, 2015 at 11:20:29AM +0100, Dominik Vogt wrote:
> On Mon, Jun 01, 2015 at 10:39:34AM +0200, Jakub Jelinek wrote:
> > always_inline attribute is meaningful only for functions declared
> > inline, and the compiler even warns about this, so it is unclear
> > to me why the tests are attempting to use that.
>
> Since Gcc emits just a cryptic warning and not an error, many
> people seem to have missed that always_inline's behaviour is
> undefined for functions without "inline". There are many more
> test cases that forget the "inline" and filter the warning with
> "-Wno-attributes". I'll post an add-on patch later after testing
> it.
Well, we need some testcases that actually verify we don't ICE
when the inline keyword is not missed.
But if you mean tests like e.g. i386/mpx/, or some other s390/
tests, then indeed, those should be fixed.
Jakub