On Tue, 18 Feb 2020, Jakub Jelinek wrote:
> Hi!
>
> The following testcase ICEs, because execute_update_addresses_taken attempts
> to create a VECTOR_TYPE with non-power of 2 number of elts.
> Fixed by guarding it with the corresponding predicate.
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
OK.
Richard.
> 2020-02-17 Jakub Jelinek <ja...@redhat.com>
>
> PR tree-optimization/93780
> * tree-ssa.c (non_rewritable_lvalue_p): Check valid_vector_subparts_p
> before calling build_vector_type.
> (execute_update_addresses_taken): Likewise.
>
> * gcc.dg/pr93780.c: New test.
>
> --- gcc/tree-ssa.c.jj 2020-01-12 11:54:38.515381696 +0100
> +++ gcc/tree-ssa.c 2020-02-17 10:54:32.481615050 +0100
> @@ -1550,7 +1550,8 @@ non_rewritable_lvalue_p (tree lhs)
> && multiple_p (lhs_bits,
> tree_to_uhwi
> (TYPE_SIZE (TREE_TYPE (TREE_TYPE (decl)))),
> - &nelts))
> + &nelts)
> + && valid_vector_subparts_p (nelts))
> {
> if (known_eq (nelts, 1u))
> return false;
> @@ -1925,7 +1926,8 @@ execute_update_addresses_taken (void)
> (TYPE_SIZE (TREE_TYPE
> (TREE_TYPE (sym)))),
> &nelts)
> - && maybe_ne (nelts, 1u))
> + && maybe_ne (nelts, 1u)
> + && valid_vector_subparts_p (nelts))
> temtype = build_vector_type (temtype, nelts);
> tree tem = make_ssa_name (temtype);
> gimple *pun
> --- gcc/testsuite/gcc.dg/pr93780.c.jj 2020-02-17 10:44:31.839583128 +0100
> +++ gcc/testsuite/gcc.dg/pr93780.c 2020-02-17 10:44:16.453812697 +0100
> @@ -0,0 +1,15 @@
> +/* PR tree-optimization/93780 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2" } */
> +/* { dg-additional-options "-mavx" { target avx } } */
> +
> +typedef float V __attribute__((vector_size (32)));
> +
> +float
> +foo (void)
> +{
> + const float init[6] = {};
> + V v = {};
> + __builtin_memcpy (&v, init, sizeof (init));
> + return v[0];
> +}
>
> Jakub
>
>
--
Richard Biener <rguent...@suse.de>
SUSE Software Solutions Germany GmbH, Maxfeldstrasse 5, 90409 Nuernberg,
Germany; GF: Felix Imendörffer; HRB 36809 (AG Nuernberg)
