I hadn't thought of these but at least added an assert which now
tripped. Fixed thus. There's also a latent issue with AVX512
mask types. The by-pieces reduction code used the wrong element
sizes.
Bootstrap and regtest running on x86_64-unknown-linux-gnu.
PR tree-optimization/122365
* tree-vect-loop.cc (vect_create_epilog_for_reduction):
Convert all inputs. Use the proper vector element sizes
for the elementwise reduction.
* gcc.dg/vect/vect-reduc-bool-9.c: New testcase.
---
gcc/testsuite/gcc.dg/vect/vect-reduc-bool-9.c | 27 +++++++++++
gcc/tree-vect-loop.cc | 48 ++++++++++---------
2 files changed, 53 insertions(+), 22 deletions(-)
create mode 100644 gcc/testsuite/gcc.dg/vect/vect-reduc-bool-9.c
diff --git a/gcc/testsuite/gcc.dg/vect/vect-reduc-bool-9.c
b/gcc/testsuite/gcc.dg/vect/vect-reduc-bool-9.c
new file mode 100644
index 00000000000..4ec141c5e69
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/vect/vect-reduc-bool-9.c
@@ -0,0 +1,27 @@
+/* PR122365 */
+/* { dg-do compile } */
+
+struct TDTI {
+ float V[4];
+};
+struct TDTI4D {
+ struct TDTI S[];
+};
+void bar();
+struct TDTI4D nii_readParRec_dti4D;
+int nii_readParRec_d_0_0;
+void nii_readParRec() {
+ for (int i;;) {
+ bool v1varies = false, v2varies = false, v3varies = false;
+ for (; i < nii_readParRec_d_0_0; i++) {
+ if (nii_readParRec_dti4D.S[i].V[1])
+ v1varies = true;
+ if (nii_readParRec_dti4D.S[i].V[2])
+ v2varies = true;
+ if (nii_readParRec_dti4D.S[i].V[3])
+ v3varies = true;
+ }
+ if (v1varies || v2varies || v3varies)
+ bar();
+ }
+}
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index be684a529db..15cb22023fc 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -5616,13 +5616,13 @@ vect_create_epilog_for_reduction (loop_vec_info
loop_vinfo,
&& VECT_REDUC_INFO_VECTYPE_FOR_MASK (reduc_info)
&& vectype != VECT_REDUC_INFO_VECTYPE_FOR_MASK (reduc_info))
{
- gcc_assert (reduc_inputs.length () == 1);
vectype = VECT_REDUC_INFO_VECTYPE_FOR_MASK (reduc_info);
gimple_seq stmts = NULL;
- reduc_inputs[0] = gimple_build (&stmts, VEC_COND_EXPR, vectype,
- reduc_inputs[0],
- build_one_cst (vectype),
- build_zero_cst (vectype));
+ for (unsigned i = 0; i < reduc_inputs.length (); ++i)
+ reduc_inputs[i] = gimple_build (&stmts, VEC_COND_EXPR, vectype,
+ reduc_inputs[i],
+ build_one_cst (vectype),
+ build_zero_cst (vectype));
gsi_insert_seq_before (&exit_gsi, stmts, GSI_SAME_STMT);
}
@@ -5963,25 +5963,29 @@ vect_create_epilog_for_reduction (loop_vec_info
loop_vinfo,
gcc_assert (exact_log2 (nunits1) != -1 && nunits1 <= nunits);
}
}
- if (!slp_reduc
- && (mode1 = targetm.vectorize.split_reduction (mode)) != mode)
+ else if (!slp_reduc
+ && (mode1 = targetm.vectorize.split_reduction (mode)) != mode)
nunits1 = GET_MODE_NUNITS (mode1).to_constant ();
- tree vectype1 = get_related_vectype_for_scalar_type (TYPE_MODE (vectype),
- stype, nunits1);
+ tree vectype1 = vectype;
+ if (mode1 != mode)
+ {
+ vectype1 = get_related_vectype_for_scalar_type (TYPE_MODE (vectype),
+ stype, nunits1);
+ /* First reduce the vector to the desired vector size we should
+ do shift reduction on by combining upper and lower halves. */
+ gimple_seq stmts = NULL;
+ new_temp = vect_create_partial_epilog (reduc_inputs[0], vectype1,
+ code, &stmts);
+ gsi_insert_seq_before (&exit_gsi, stmts, GSI_SAME_STMT);
+ reduc_inputs[0] = new_temp;
+ }
+
reduce_with_shift = have_whole_vector_shift (mode1);
if (!VECTOR_MODE_P (mode1)
|| !directly_supported_p (code, vectype1))
reduce_with_shift = false;
- /* First reduce the vector to the desired vector size we should
- do shift reduction on by combining upper and lower halves. */
- gimple_seq stmts = NULL;
- new_temp = vect_create_partial_epilog (reduc_inputs[0], vectype1,
- code, &stmts);
- gsi_insert_seq_before (&exit_gsi, stmts, GSI_SAME_STMT);
- reduc_inputs[0] = new_temp;
-
if (reduce_with_shift && (!slp_reduc || group_size == 1))
{
tree bitsize = TYPE_SIZE (TREE_TYPE (vectype1));
@@ -6009,7 +6013,7 @@ vect_create_epilog_for_reduction (loop_vec_info
loop_vinfo,
"Reduce using vector shifts\n");
gimple_seq stmts = NULL;
- new_temp = gimple_convert (&stmts, vectype1, new_temp);
+ new_temp = gimple_convert (&stmts, vectype1, reduc_inputs[0]);
for (elt_offset = nelements / 2;
elt_offset >= 1;
elt_offset /= 2)
@@ -6053,13 +6057,13 @@ vect_create_epilog_for_reduction (loop_vec_info
loop_vinfo,
"Reduce using scalar code.\n");
tree compute_type = TREE_TYPE (vectype1);
- tree bitsize = TYPE_SIZE (compute_type);
- int vec_size_in_bits = tree_to_uhwi (TYPE_SIZE (vectype1));
- int element_bitsize = tree_to_uhwi (bitsize);
+ unsigned vec_size_in_bits = tree_to_uhwi (TYPE_SIZE (vectype1));
+ unsigned element_bitsize = vector_element_bits (vectype1);
+ tree bitsize = bitsize_int (element_bitsize);
gimple_seq stmts = NULL;
FOR_EACH_VEC_ELT (reduc_inputs, i, vec_temp)
{
- int bit_offset;
+ unsigned bit_offset;
new_temp = gimple_build (&stmts, BIT_FIELD_REF, compute_type,
vec_temp, bitsize, bitsize_zero_node);
--
2.51.0
