Re: target attributes, pragmas and preprocessor macros

[Kyrill Tkachov <kyrylo dot tkachov at arm dot com>]

Hi Christian,

On 18/05/15 07:26, Christian Bruel wrote:
> Hi Kyrill,
>
> On 05/13/2015 05:43 PM, Kyrill Tkachov wrote:
> > Hi all,
> >
> > Are target attributes supposed to redefine the preprocessor macros available?
> > For example, on aarch64 if the file is compiled with floating point support
> > the __ARM_FEATURE_FMA predefine is available. If the user adds to a function
> > a target attribute disabling floating point, then is __ARM_FEATURE_FMA supposed
> > to be undefined in the body of that function?
> >
> > Looking at some backends, it seems that only #pragmas are supposed to have that effect,
> > but I just wanted to confirm.
> yes they do, But careful, even the "inherited" macros should be redefined...

Thanks.
Implementation question then.
Is TARGET_PRAGMA_PARSE actually called during preprocessing time?
I built an x86_64 toolchain, compiled a file with target pragmas in it,
put a breakpoint on the hook entry and it only triggered later on during parsing.
Same for an aarch64 implementation of target pragmas that I'm working on.
Am I just doing something wrong? Or is there some other place where I should
be looking?

Thanks,
Kyrill

> for instance for arm/thumb we can have something like:
>
> #pragma GCC target ("thumb")
>
> #ifndef __thumb__
> #error "__thumb__ is not defined"
> #endif
>
> #ifdef __thumb2__
> #ifndef __ARM_32BIT_STATE
> #error  "__ARM_32BIT_STATE is not defined"
> #endif
> #else /* thumb1 */
> #ifdef __ARM_32BIT_STATE
> #error  "__ARM_32BIT_STATE is defined"
> #endif
> #endif /* thumb1 */
>
> ...
>
> > Thanks,
> > Kyrill
> Cheers
>
> Christian