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 gcc -v --help (PR middle-end/37576)


On Tue, Oct 7, 2008 at 1:48 PM, Jakub Jelinek <jakub@redhat.com> wrote:
> On Tue, Oct 07, 2008 at 07:35:00AM -0700, Ian Lance Taylor wrote:
>> Jakub Jelinek <jakub@redhat.com> writes:
>> Why is CL_SAVE listed with the option classes rather than the option
>> attributes?  It seems to me that CL_SAVE should move to the subsequent
>> list in opts.h (the one starting with CL_DISABLED) and that
>> CL_MIN_OPTION_CLASS should be set back to CL_PARAMS.  At least, I
>> can't see why on would ever want to set CL_SAVE without also setting
>> CL_TARGET.
>>
>> Moving CL_SAVE in that way seems right and should accomplish the same
>> goal as this patch.
>
> I've bootstrapped/regtested following on x86_64-linux.  Ok for trunk then?
>
> 2008-10-07  Jakub Jelinek  <jakub@redhat.com>
>
>        PR middle-end/37576
>        * opts.h (CL_SAVE): Move up to flags range.
>        (CL_PARAMS, CL_WARNING, CL_OPTIMIZATION, CL_TARGET,
>        CL_COMMON): Renumber.
>        (CL_MIN_OPTION_CLASS): Set to CL_PARAMS.
>        * opts.c (common_handle_option): Revert last change.

There are some fundamental issues with function specific optimizations.
One of them is

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37565

Can we fix it properly? I don't like

void foo (void) __attribute__
((__optimize__(2,"omit-frame-pointer,-mtune=core2")));

Why can't we do

void foo (void) __attribute__
((__optimize__("-O2,-fomit-frame-pointer,-mtune=core2")));

or

void foo (void) __attribute__
((__optimize__("-O2 -fomit-frame-pointer -mtune=core2")));


-- 
H.J.


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