Richard Ball <richard.b...@arm.com> writes:
> Adds a check to ensure that the input vector arguments
> to a function are not variable length. Previously, only the
> output vector of a function was checked.
>
> The ICE in question is within the neon-sve-bridge.c test,
> and is related to https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111268
>
> gcc/ChangeLog:
>
> * tree-vect-slp.cc (vectorizable_slp_permutation_1):
> Add variable-length check for vector input arguments
> to a function.
OK, thanks. Sorry, I'd misunderstood the op_vectype suggestion and
thought it involved returning early from the existing loop.
Richard
> diff --git a/gcc/tree-vect-slp.cc b/gcc/tree-vect-slp.cc
> index
> 086377a9ac08528880c90bee3b1d1e08341a6288..7cf9504398c98cccfdba3b91c3e995f73f8b5d50
> 100644
> --- a/gcc/tree-vect-slp.cc
> +++ b/gcc/tree-vect-slp.cc
> @@ -8987,7 +8987,8 @@ vectorizable_slp_permutation_1 (vec_info *vinfo,
> gimple_stmt_iterator *gsi,
> {
> /* Calculate every element of every permute mask vector explicitly,
> instead of relying on the pattern described above. */
> - if (!nunits.is_constant (&npatterns))
> + if (!nunits.is_constant (&npatterns)
> + || !TYPE_VECTOR_SUBPARTS (op_vectype).is_constant ())
> return -1;
> nelts_per_pattern = ncopies = 1;
> if (loop_vec_info linfo = dyn_cast <loop_vec_info> (vinfo))
