The following properly handles pattern matching generated COND_EXPRs
which can still have embedded compares in vectorizable_condition
which will always code generate the masked vector variant. We
were requiring vcond with embedded comparisons instead of also
allowing (as code generated) split compare and VEC_COND_EXPR.
This fixes some of the fallout when removing vcond{,u,eq} expanders
from the x86 backend.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
PR target/109955
* tree-vect-stmts.cc (vectorizable_condition): For
embedded comparisons also handle the case when the target
only provides vec_cmp and vcond_mask.
---
gcc/tree-vect-stmts.cc | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/gcc/tree-vect-stmts.cc b/gcc/tree-vect-stmts.cc
index 127b987cd62..bd3b07a3aa1 100644
--- a/gcc/tree-vect-stmts.cc
+++ b/gcc/tree-vect-stmts.cc
@@ -10836,7 +10836,12 @@ vectorizable_condition (vec_info *vinfo,
if (reduction_type == EXTRACT_LAST_REDUCTION)
/* Count one reduction-like operation per vector. */
kind = vec_to_scalar;
- else if (!expand_vec_cond_expr_p (vectype, comp_vectype, cond_code))
+ else if (!expand_vec_cond_expr_p (vectype, comp_vectype, cond_code)
+ && (masked
+ || (!expand_vec_cmp_expr_p (comp_vectype, vec_cmp_type,
+ cond_code)
+ || !expand_vec_cond_expr_p (vectype, vec_cmp_type,
+ ERROR_MARK))))
return false;
if (slp_node
--
2.35.3
