https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64700
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed|2015-01-21 00:00:00 |2016-9-3
--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
here is a much simpler example:
float f(int a, float b, float c, float d)
{
if (a)
return c + d;
return b + d;
}
float f1(int a, float b, float c, float d)
{
float e;
if (a)
e = c;
else
e = b;
return e + d;
}
For aarch64 we get:
f:
cbnz w0, .L5
fadd s0, s2, s0
ret
.p2align 3
.L5:
fadd s0, s1, s2
ret
...
f1:
cmp w0, 0
fcsel s0, s1, s0, ne
fadd s0, s0, s2
ret
These two functions should produce the same code.