Re: Fix permute handling when vectorising scatters

[Richard Biener <richard dot guenther at gmail dot com>]

On Tue, Jan 9, 2018 at 12:34 PM, Richard Sandiford
<richard.sandiford@linaro.org> wrote:
> As mentioned in https://gcc.gnu.org/ml/gcc-patches/2017-11/msg01575.html ,
> the scatter handling in vectorizable_store seems to be dead code at the
> moment.  Enabling it with the vect_analyze_data_ref_access part of
> that patch triggered an ICE in the avx512f-scatter-*.c tests (which
> previously didn't use scatters).  The problem was that the NARROW
> and WIDEN handling uses permute_vec_elements to marshal the inputs,
> and permute_vec_elements expected the lhs of the stmt to be an SSA_NAME,
> which of course it isn't for stores.
>
> This patch makes permute_vec_elements create a fresh variable in this case.
>
> Tested on x86_64-linux-gnu.  OK to install?

Ok.

Richard.

> Richard
>
>
> 2018-01-09  Richard Sandiford  <richard.sandiford@linaro.org>
>
> gcc/
>         * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
>         if the destination isn't an SSA_NAME.
>
> Index: gcc/tree-vect-stmts.c
> ===================================================================
> --- gcc/tree-vect-stmts.c       2018-01-03 21:47:27.956862491 +0000
> +++ gcc/tree-vect-stmts.c       2018-01-09 11:30:00.578821631 +0000
> @@ -6585,7 +6585,11 @@ permute_vec_elements (tree x, tree y, tr
>    tree perm_dest, data_ref;
>    gimple *perm_stmt;
>
> -  perm_dest = vect_create_destination_var (gimple_get_lhs (stmt), vectype);
> +  tree scalar_dest = gimple_get_lhs (stmt);
> +  if (TREE_CODE (scalar_dest) == SSA_NAME)
> +    perm_dest = vect_create_destination_var (scalar_dest, vectype);
> +  else
> +    perm_dest = vect_get_new_vect_var (vectype, vect_simple_var, NULL);
>    data_ref = make_ssa_name (perm_dest);
>
>    /* Generate the permute statement.  */