Jennifer Schmitz <jschm...@nvidia.com> writes:
> @@ -3672,6 +3673,48 @@ gimple_folder::fold_pfalse ()
> return nullptr;
> }
>
> +/* Convert the lhs and all non-boolean vector-type operands to TYPE.
> + Pass the converted variables to the callback FP, and finally convert the
> + result back to the original type. Add the necessary conversion statements.
> + Return the new call. */
> +gimple *
> +gimple_folder::convert_and_fold (tree type,
> + gimple *(*fp) (gimple_folder &,
> + tree, vec<tree> &))
> +{
> + gcc_assert (VECTOR_TYPE_P (type)
> + && TYPE_MODE (type) != VNx16BImode);
> + tree old_ty = TREE_TYPE (lhs);
> + gimple_seq stmts = NULL;
> + tree lhs_conv, op, op_ty, t;
> + gimple *g, *new_stmt;
Sorry for the last-minute minor request, but: it would be nice to declare
these at the point of initialisation, for consistency with the rest of the
function.
> + bool convert_lhs_p = !useless_type_conversion_p (type, old_ty);
> + lhs_conv = convert_lhs_p ? create_tmp_var (type) : lhs;
> + unsigned int num_args = gimple_call_num_args (call);
> + auto_vec<tree, 16> args_conv;
> + args_conv.safe_grow (num_args);
> + for (unsigned int i = 0; i < num_args; ++i)
> + {
> + op = gimple_call_arg (call, i);
> + op_ty = TREE_TYPE (op);
> + args_conv[i] =
> + (VECTOR_TYPE_P (op_ty)
> + && TYPE_MODE (op_ty) != VNx16BImode
> + && !useless_type_conversion_p (op_ty, type))
> + ? gimple_build (&stmts, VIEW_CONVERT_EXPR, type, op) : op;
> + }
> +
> + new_stmt = fp (*this, lhs_conv, args_conv);
> + gsi_insert_seq_before (gsi, stmts, GSI_SAME_STMT);
> + if (convert_lhs_p)
> + {
> + t = build1 (VIEW_CONVERT_EXPR, old_ty, lhs_conv);
> + g = gimple_build_assign (lhs, VIEW_CONVERT_EXPR, t);
> + gsi_insert_after (gsi, g, GSI_SAME_STMT);
> + }
> + return new_stmt;
> +}
> +
> /* Fold the call to constant VAL. */
> gimple *
> gimple_folder::fold_to_cstu (poly_uint64 val)
> diff --git a/gcc/config/aarch64/aarch64-sve-builtins.h
> b/gcc/config/aarch64/aarch64-sve-builtins.h
> index 6f22868f9b3..02ae098ed32 100644
> --- a/gcc/config/aarch64/aarch64-sve-builtins.h
> +++ b/gcc/config/aarch64/aarch64-sve-builtins.h
> @@ -421,6 +421,7 @@ public:
> tree scalar_type (unsigned int) const;
> tree vector_type (unsigned int) const;
> tree tuple_type (unsigned int) const;
> + tree signed_type (unsigned int) const;
> unsigned int elements_per_vq (unsigned int) const;
> machine_mode vector_mode (unsigned int) const;
> machine_mode tuple_mode (unsigned int) const;
> @@ -649,6 +650,8 @@ public:
> gcall *redirect_call (const function_instance &);
> gimple *redirect_pred_x ();
> gimple *fold_pfalse ();
> + gimple *convert_and_fold (tree, gimple *(*) (gimple_folder &,
> + tree, vec<tree> &));
>
> gimple *fold_to_cstu (poly_uint64);
> gimple *fold_to_pfalse ();
> @@ -884,6 +887,20 @@ find_type_suffix (type_class_index tclass, unsigned int
> element_bits)
> gcc_unreachable ();
> }
>
> +/* Return the type suffix of the signed type of width ELEMENT_BITS. */
> +inline type_suffix_index
> +signed_type_suffix_index (unsigned int element_bits)
> +{
> + switch (element_bits)
> + {
> + case 8: return TYPE_SUFFIX_s8;
> + case 16: return TYPE_SUFFIX_s16;
> + case 32: return TYPE_SUFFIX_s32;
> + case 64: return TYPE_SUFFIX_s64;
> + }
> + gcc_unreachable ();
> +}
> +
We could drop this and instead replace calls with:
find_type_suffix (TYPE_signed, element_bits)
> /* Return the single field in tuple type TYPE. */
> inline tree
> tuple_type_field (tree type)
> @@ -1049,6 +1066,20 @@ function_instance::tuple_type (unsigned int i) const
> return acle_vector_types[num_vectors - 1][type_suffix (i).vector_type];
> }
>
> +/* Return the signed vector type of width ELEMENT_BITS. */
> +inline tree
> +function_instance::signed_type (unsigned int element_bits) const
> +{
> + switch (element_bits)
> + {
> + case 8: return acle_vector_types[0][VECTOR_TYPE_svint8_t];
> + case 16: return acle_vector_types[0][VECTOR_TYPE_svint16_t];
> + case 32: return acle_vector_types[0][VECTOR_TYPE_svint32_t];
> + case 64: return acle_vector_types[0][VECTOR_TYPE_svint64_t];
> + }
> + gcc_unreachable ();
> +}
> +
And for this, I think we should instead make:
/* Return the vector type associated with TYPE. */
static tree
get_vector_type (sve_type type)
{
auto vector_type = type_suffixes[type.type].vector_type;
return acle_vector_types[type.num_vectors - 1][vector_type];
}
public, or perhaps just define it inline in aarch64-sve-builtins.h.
OK with those changes, thanks.
Richard
