https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91680
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |easyhack,
| |missed-optimization
Status|UNCONFIRMED |NEW
Last reconfirmed| |2019-09-06
Ever confirmed|0 |1
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.
/* (A / (1 << B)) -> (A >> B).
Only for unsigned A. For signed A, this would not preserve rounding
toward zero.
For example: (-1 / ( 1 << B)) != -1 >> B. */
(simplify
(trunc_div @0 (lshift integer_onep@1 @2))
(if ((TYPE_UNSIGNED (type) || tree_expr_nonnegative_p (@0))
&& (!VECTOR_TYPE_P (type)
|| target_supports_op_p (type, RSHIFT_EXPR, optab_vector)
|| target_supports_op_p (type, RSHIFT_EXPR, optab_scalar)))
(rshift @0 @2)))
lacks a (convert? ...) on the divisor where at least sign-changes are OK.
