The soft-fail path due to BB SLP unrolling can fail hard as well,
but there's no need to compute match[] when we just discard it
anyway.
Bootstrap and regtest running on x86_64-unknown-linux-gnu.
I'll probably re-order to the beginning of the series as well, or
drop as likely conflicting with predicated tails.
* tree-vect-slp.cc (vect_build_slp_tree_1): Hard-fail early
due to BB SLP unrolling.
---
gcc/tree-vect-slp.cc | 27 ++++++++++++---------------
1 file changed, 12 insertions(+), 15 deletions(-)
diff --git a/gcc/tree-vect-slp.cc b/gcc/tree-vect-slp.cc
index c0303627579..70140d72fdf 100644
--- a/gcc/tree-vect-slp.cc
+++ b/gcc/tree-vect-slp.cc
@@ -1685,7 +1685,7 @@ vect_build_slp_tree_1 (vec_info *vinfo, unsigned char
*swap,
return false;
}
bool maybe_soft_fail = false;
- tree soft_fail_nunits_vectype = NULL_TREE;
+ unsigned HOST_WIDE_INT const_nunits = 0;
/* Record nunits required but continue analysis, producing matches[]
as if nunits was not an issue. This allows splitting of groups
to happen. */
@@ -1694,8 +1694,14 @@ vect_build_slp_tree_1 (vec_info *vinfo, unsigned char
*swap,
nunits_vectype, max_nunits))
{
gcc_assert (is_a <bb_vec_info> (vinfo));
+ if (!TYPE_VECTOR_SUBPARTS (nunits_vectype).is_constant (&const_nunits)
+ || const_nunits > group_size)
+ {
+ /* Fatal mismatch. */
+ matches[0] = false;
+ return false;
+ }
maybe_soft_fail = true;
- soft_fail_nunits_vectype = nunits_vectype;
}
gcc_assert (vectype || !gimple_get_lhs (first_stmt_info->stmt));
@@ -1715,19 +1721,10 @@ vect_build_slp_tree_1 (vec_info *vinfo, unsigned char
*swap,
if (maybe_soft_fail)
{
- unsigned HOST_WIDE_INT const_nunits;
- if (!TYPE_VECTOR_SUBPARTS
- (soft_fail_nunits_vectype).is_constant (&const_nunits)
- || const_nunits > group_size)
- /* Fatal mismatch. */
- matches[0] = false;
- else
- {
- /* With constant vector elements simulate a mismatch at the
- point we need to split. */
- unsigned tail = group_size & (const_nunits - 1);
- memset (&matches[group_size - tail], 0, sizeof (bool) * tail);
- }
+ /* With constant vector elements simulate a mismatch at the
+ point we need to split. */
+ unsigned tail = group_size & (const_nunits - 1);
+ memset (&matches[group_size - tail], 0, sizeof (bool) * tail);
return false;
}
--
2.51.0
