[Bug libc/19759] mempcpy shouldn't be inlined

["wdijkstr at arm dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=19759

--- Comment #5 from Wilco <wdijkstr at arm dot com> ---
(In reply to Carlos O'Donell from comment #2)
> (In reply to Carlos O'Donell from comment #1)
> > I believe this issue should also extend to the following arches which have
> > optimized mempcpy and would benefit from the compiler optimization which
> > reduces register pressure.
> > 
> > * i386/i586/i686
> > * ppc32/ppc64
> > 
> > For s390 we need an optimized mempcpy to be implemented to gain back that
> > callee saved register before we disable the inlining. For s390 I've filed
> > bug 19765 and asked Stefan Liebler (IBM) to review.
> 
> So for ppc3 and ppc64 the __builtin_mempcpy is not optimal.
> 
>    0:	38 a0 00 20 	li      r5,32
>    4:	48 00 00 00 	b       4 <foo+0x4>
> 			4: R_PPC_REL24	mempcpy
>    8:	60 00 00 00 	nop
>    c:	60 42 00 00 	ori     r2,r2,0
> 
> It doesn't inline the implementation, even at -O3. So it looks like only x86
> needs changing.

I created https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70140 to fix the mempcpy
inlining in GCC.

-- 
You are receiving this mail because:
You are on the CC list for the bug.