https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83389
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2017-12-12
CC| |jakub at gcc dot gnu.org,
| |rguenth at gcc dot gnu.org
Component|c++ |tree-optimization
Ever confirmed|0 |1
--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
This is fold_truth_andor / optimize_bit_field_compare at work combining
d1.x == d2.x && d1.y == d2.y very early if exposed in such form in the IL.
The uglification via tuples makes this not happen.
The optimization is premature at its current place and lacks a suitable
implementation later (in some GIMPLE optimization pass).
We must have a dozen related / dups of this.
Maybe store-merging/bswap infrastructure can come to the rescue here
somehow ...
