This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] powerpc: fenvinline.h refactor
On 24/01/2020 06:50, Florian Weimer wrote:
> * Adhemerval Zanella:
>
>> I don't think it is a good practice to re-define bultins, use it along with
>> an extra indirection instead. Something as:
>>
>> # if !__GNUC_PREREQ(9, 0)
>> # define __mtfsb0(__b) __asm__ __volatile__ ("mtfsb0 %0" : : "i" (__b))
>> # define __mtfsb1(__b) __asm__ __volatile__ ("mtfsb1 %0" : : "i" (__b))
>> # else
>> # define __mtfsb0(__b) __builtin_mtfsb0 (__b)
>> # define __mtfsb1(__b) __builtin_mtfsb1 (__b)
>> # endif
>
> I think __mtfsb0 and __mtfsb1 conflict with gcc's installed
> pu_intrinsics.h header.
Right, in this case prepend with __glibc or any other related unique
identifier. Checking on ppu_intrinsics.h, I wonder if the asm inline
should not use the 'n' constraint as well.