https://gcc.gnu.org/g:39d429631ff7060238dc046db4340a304560227f
commit r16-2710-g39d429631ff7060238dc046db4340a304560227f
Author: Richard Biener <rguent...@suse.de>
Date: Fri Aug 1 13:27:00 2025 +0200
Properly record SLP node when costing a vectorized store
Even when we emit scalar stores we should pass down the SLP node.
PR tree-optimization/121350
* tree-vect-stmts.cc (vectorizable_store): Pass down SLP
node when costing scalar stores in vect_body.
Diff:
---
gcc/tree-vect-stmts.cc | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/gcc/tree-vect-stmts.cc b/gcc/tree-vect-stmts.cc
index 7f922c277996..97222f64b7eb 100644
--- a/gcc/tree-vect-stmts.cc
+++ b/gcc/tree-vect-stmts.cc
@@ -8233,7 +8233,7 @@ vectorizable_store (vec_info *vinfo,
else
inside_cost
+= record_stmt_cost (cost_vec, n_adjacent_stores,
- scalar_store, stmt_info, 0, vect_body);
+ scalar_store, slp_node, 0, vect_body);
/* Only need vector extracting when there are more
than one stores. */
if (nstores > 1)
