public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r13-3544] tree-optimization/107435 - ICE with recurrence vectorization
@ 2022-10-28 13:07 Richard Biener
  0 siblings, 0 replies; only message in thread
From: Richard Biener @ 2022-10-28 13:07 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:084128583212bd64308f50c2ab5f4c03be40e48c

commit r13-3544-g084128583212bd64308f50c2ab5f4c03be40e48c
Author: Richard Biener <rguenther@suse.de>
Date:   Fri Oct 28 13:50:57 2022 +0200

    tree-optimization/107435 - ICE with recurrence vectorization
    
    This implements the missed conversion from pointer to integer.
    
            PR tree-optimization/107435
            * tree-vect-loop.cc (vectorizable_recurr): Convert initial
            value to vector component type.
    
            * gcc.dg/torture/pr107435.c: New testcase.

Diff:
---
 gcc/testsuite/gcc.dg/torture/pr107435.c | 23 +++++++++++++++++++++++
 gcc/tree-vect-loop.cc                   |  6 ++++++
 2 files changed, 29 insertions(+)

diff --git a/gcc/testsuite/gcc.dg/torture/pr107435.c b/gcc/testsuite/gcc.dg/torture/pr107435.c
new file mode 100644
index 00000000000..10128969191
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/torture/pr107435.c
@@ -0,0 +1,23 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-mavx2" { target x86_64-*-* i?86-*-* } } */
+
+struct qlist_head {
+  struct qlist_head *next;
+};
+void
+qlist_add (struct qlist_head *new, struct qlist_head *head)
+{
+  struct qlist_head *prev = head;
+  new->next = head->next;
+  prev->next = new;
+}
+struct {
+  struct qlist_head queue_link;
+} free_list, prealloc[64];
+void
+dbpf_open_cache_initialize()
+{
+  int i = 0;
+  for (; i < 64; i++)
+    qlist_add(&prealloc[i].queue_link, &free_list.queue_link);
+}
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index d5c2bff80be..aacbb12580c 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -8469,6 +8469,12 @@ vectorizable_recurr (loop_vec_info loop_vinfo, stmt_vec_info stmt_info,
   edge pe = loop_preheader_edge (LOOP_VINFO_LOOP (loop_vinfo));
   basic_block bb = gimple_bb (phi);
   tree preheader = PHI_ARG_DEF_FROM_EDGE (phi, pe);
+  if (!useless_type_conversion_p (TREE_TYPE (vectype), TREE_TYPE (preheader)))
+    {
+      gimple_seq stmts = NULL;
+      preheader = gimple_convert (&stmts, TREE_TYPE (vectype), preheader);
+      gsi_insert_seq_on_edge_immediate (pe, stmts);
+    }
   tree vec_init = build_vector_from_val (vectype, preheader);
   vec_init = vect_init_vector (loop_vinfo, stmt_info, vec_init, vectype, NULL);

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2022-10-28 13:07 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-10-28 13:07 [gcc r13-3544] tree-optimization/107435 - ICE with recurrence vectorization Richard Biener

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).