https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87361
--- Comment #7 from Eric Botcazou ---
> I think nonzero_bits should use SUBREG_PROMOTED_VAR_P and
> SUBREG_PROMOTED_UNSIGNED_P for REGs, not LOAD_EXTEND_OP, which is
> only for MEMs.
I don't see how SUBREG_PROMOTED_VAR_P/SUBREG_PROMOTED_UNSIGNED
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87361
--- Comment #6 from Segher Boessenkool ---
I think nonzero_bits should use SUBREG_PROMOTED_VAR_P and
SUBREG_PROMOTED_UNSIGNED_P for REGs, not LOAD_EXTEND_OP, which is
only for MEMs.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87361
--- Comment #5 from Eric Botcazou ---
> Ugh, I'm very sorry. I still think that patch is correct, but I didn't
> intend to commit it. Thanks for catching it.
No problem. I don't disagree with the correctness but it will bring back a lot
of re
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87361
Segher Boessenkool changed:
What|Removed |Added
CC||segher at gcc dot gnu.org
--- Comme
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87361
Eric Botcazou changed:
What|Removed |Added
Status|NEW |RESOLVED
Resolution|---
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87361
--- Comment #2 from Eric Botcazou ---
Author: ebotcazou
Date: Wed Sep 19 10:12:04 2018
New Revision: 264420
URL: https://gcc.gnu.org/viewcvs?rev=264420&root=gcc&view=rev
Log:
PR rtl-optimization/87361
* rtlanal.c (nonzero_bits1):
