------- Comment #10 from sebastian dot pop at cri dot ensmp dot fr 2006-06-15
14:52 -------
Subject: Re: [4.2 Regression] segfault in fold_convert with -ftree-vectorize
You said that you had a fix in predcom, is that fix in your local
tree, or have you sent a patch to gcc-patches?
Here is a fix for this PR. I don't like the data-ref code that deals
with component_refs, as it seems very fragile.
Index: tree-data-ref.c
===================================================================
--- tree-data-ref.c (revision 114678)
+++ tree-data-ref.c (working copy)
@@ -1947,7 +1947,8 @@ create_data_ref (tree memref, tree stmt,
object_analysis divided by the size of data type.
*/
if (!DR_BASE_OBJECT (dr)
- || (TREE_CODE (memref) == COMPONENT_REF && DR_NUM_DIMENSIONS (dr) == 1))
+ || (TREE_CODE (memref) == COMPONENT_REF && DR_NUM_DIMENSIONS (dr) == 1
+ && !automatically_generated_chrec_p (DR_ACCESS_FN (dr, 0))))
{
tree access_fn;
tree new_step;
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27331
