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] Fix issues with flag_strict_aliasing changing during compilation (PR c/37106, take {2,3})


On Mon, Nov 3, 2008 at 9:23 PM, Jakub Jelinek <jakub@redhat.com> wrote:
> On Mon, Nov 03, 2008 at 05:01:45PM +0100, Richard Guenther wrote:
>> Hmm.  I do not like this too much.  For 4.4 I'd like us simply avoid
>> changing flag_strict_aliasing.
>
> Attached are 2 alternatives to do that.  One that just silently avoids
> changing flag_strict_aliasing, one that does that silently for
> __optimize__({0,1,2,3}), but errors on __optimize__("fstrict-aliasing")
> when -fno-strict-aliasing is given (or vice versa).

I don't have a strong preference here, so I'd go with the simpler version.

>>  For 4.5 we should move flag_strict_aliasing to struct function instead.
>
> How would that change anything?

It doesn't change anything if we query alias sets of types.  On the
alias-improvements branch I am trying to put alias sets to the
memory access statements instead.  But parts of your patch would
probably still apply.

Richard.

>        Jakub
>


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