[Bug tree-optimization/126417] [17 Regression] Incorrect floating-point re-association
"rguenth at gcc dot gnu.org via Gcc-bugs" <[email protected]>
| Newsgroups | gmane.comp.gcc.bugs |
|---|---|
| Message-ID | <[email protected]/bugzilla/> |
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126417
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |ASSIGNED
Assignee|unassigned at gcc dot gnu.org |rguenth at gcc dot gnu.org
Keywords| |wrong-code
Known to work| |16.1.1
Target Milestone|--- |17.0
Summary|Incorrect floating-point |[17 Regression] Incorrect
|re-association |floating-point
| |re-association
Ever confirmed|0 |1
Last reconfirmed| |2026-07-27
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Huh. I must somehow mis-written the in-order requirement check. Yep:
/* Two-element reductions do not need special-handling for fold-left,
other cases are not yet implemented. remain_defs also have to
be included here. */
|| (needs_fold_left_reduction_p (TREE_TYPE (vectype), reduc_code)
&& (!instance->remain_defs.is_empty ()
|| maybe_ne (TYPE_VECTOR_SUBPARTS (vectype), 2u))))
of course I need to check the total number of lanes here.