On Mon, Mar 18, 2019 at 02:43:41PM +0000, David Laight wrote:
> From: Peter Zijlstra
> > Sent: 18 March 2019 09:14
> > On Fri, Mar 15, 2019 at 03:06:37PM -0700, Nick Desaulniers wrote:
> > > On Fri, Mar 15, 2019 at 1:54 PM Matthias Kaehlcke <m...@chromium.org> 
> > > wrote:
> > > >
> > > > The compiler may emit calls to __lshrti3 from the compiler runtime
> > > > library, which results in undefined references:
> > > >
> > > > arch/x86/kvm/x86.o: In function `mul_u64_u64_shr':
> > > >   include/linux/math64.h:186: undefined reference to `__lshrti3'
> > >
> > > Looks like Clang will emit this at -Oz (but not -O2):
> > > https://godbolt.org/z/w1_2YC
> > 
> > *OMG*, what is that compiler smoking and why do we want that?
> > 
> > It doesn't even do that for "-Os".
> 
> I like the way it moves %edx to %ecx, then %cl to %ecx and finally %ecx back 
> to %edx.
> I'm guessing this is all made worse by the prototype containing 'char' not 
> 'int'.
> 
> I'm sure the register tracking gets worse in every version of gcc.

This is clang, no GCC on that list comes even close to generating
anything as 'brilliant' as that.

Reply via email to