Landing: 13e24c75a812
Project / Subsystem
gcc / tree-optimization
Date
2026-09-21
Author
Richard Biener
Commit
13e24c75a812b4838242e8da7d6697d81534a30a
Source
github
Perf win
No
Breaking
No
All attributes
- project
- gcc
- subsystem
- tree-optimization
- patch_id
- —
- commit_hash
- 13e24c75a812b4838242e8da7d6697d81534a30a
- source_type
- github
- headline
- tree-optimization: Avoid get_same_sized_vectype in vectorizable_induction
- tldr
- Improves vector type selection in induction variable optimization.
- author
- Richard Biener
- outcome
- committed
- performance_win
- false
- breaking_change
- false
- series_id
- —
- series_parts
- []
- tags
-
- • tree-optimization
- • vectorization
- • bugfix
- discussion_id_link
- —
- bugzilla_pr
- —
- date
- 2026-09-21T00:00:00.000Z
This commit refactors the vectorizable_induction function within the tree optimization passes to avoid using get_same_sized_vectype. Instead, it now uses signed_or_unsigned_type_for to obtain signed or unsigned variants of vector types. This change prevents issues related to the handling of oddball scalar boolean types within get_same_sized_vectype, leading to more robust vectorization of induction variables.