[Bug rtl-optimization/92342] [10 Regression] a small missed transformation into x?b:0
jakub at gcc dot gnu.org
gcc-bugzilla@gcc.gnu.org
Fri Feb 28 09:19:00 GMT 2020
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92342
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jakub at gcc dot gnu.org
--- Comment #15 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
For this, I agree we should canonicalize the second form to the first one
(well, any commutative x & -boolean_range into boolean_range ? x : 0 because at
least in 95% of cases it will be written the former way, the latter form looks
like a cute trick trying to workaround compiler deficiencies.
If the x & -boolean_range form results in better code than boolean_range ? x :
0 on some target, we should make sure to expand the ?: that way somewhere in
RTL.
More information about the Gcc-bugs
mailing list