https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116510
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed|2024-08-28 00:00:00 |
Target Milestone|15.0 |12.5
Component|tree-optimization |c
CC| |glisse at gcc dot gnu.org,
| |rguenth at gcc dot gnu.org
Summary|[15 Regression] ice in |[12/13/14/15 Regression]
|decompose, at |ice in decompose, at
|wide-int.h:1049 |wide-int.h:1049
--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed. This is likely from a new match.pd pattern.
* X == C (or X & Z == Y | C) is impossible if ~nonzero(X) & C != 0. */
(for cmp (eq ne)
(simplify
(cmp:c (with_possible_nonzero_bits2 @0) (with_certain_nonzero_bits2 @1))
(if (wi::bit_and_not (wi::to_wide (@1), get_nonzero_bits (@0)) != 0)
{ constant_boolean_node (cmp == NE_EXPR, type); })))
the issue is that with_possible_nonzero_bits2 also supports SSA_NAMEs, so
@1 cannot be used like this.
Caused by r8-910-g10bc8017a3028c - the pattern is quite old (and was
broken). I would suggest
(cmp (with_possible_nonzero_bits2 @0) INTEGER_CST@1)
as "trivial" fix.
The testcase exposes this only now.
