diff mbox series

tree-optimization/112677 - stack corruption with .COND_* reduction

Message ID 20231124102537.34DF8132E2@imap2.dmz-prg2.suse.org
State Accepted
Headers
Series tree-optimization/112677 - stack corruption with .COND_* reduction |

Checks

Context Check Description
snail/gcc-patch-check success Github commit url

Commit Message

Richard Biener Nov. 24, 2023, 10:25 a.m. UTC 
  The following makes sure to allocate enough space for vectype_op
in vectorizable_reduction.

Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.

	PR tree-optimization/112677
	* tree-vect-loop.cc (vectorizable_reduction): Use alloca
	to allocate vectype_op.
---
 gcc/tree-vect-loop.cc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
diff mbox series

Patch

diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index caaed977c89..3df020d2228 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -7385,7 +7385,6 @@  vectorizable_reduction (loop_vec_info loop_vinfo,
 			stmt_vector_for_cost *cost_vec)
 {
   tree vectype_in = NULL_TREE;
-  tree vectype_op[3] = { NULL_TREE, NULL_TREE, NULL_TREE };
   class loop *loop = LOOP_VINFO_LOOP (loop_vinfo);
   enum vect_def_type cond_reduc_dt = vect_unknown_def_type;
   stmt_vec_info cond_stmt_vinfo = NULL;
@@ -7617,6 +7616,7 @@  vectorizable_reduction (loop_vec_info loop_vinfo,
      assumption is not true: we use reduc_index to record the index of the
      reduction variable.  */
   slp_tree *slp_op = XALLOCAVEC (slp_tree, op.num_ops);
+  tree *vectype_op = XALLOCAVEC (tree, op.num_ops);
   /* We need to skip an extra operand for COND_EXPRs with embedded
      comparison.  */
   unsigned opno_adjust = 0;