The following addresses
tree forward propagate : 12.41 ( 9%)
seen with the compile.i testcase of this PR which points at
the has_use_on_stmt function which, for SSA names with many
uses is slow. The solution is to instead of immediate uses,
look at stmt operands to identify whether a name has a use
on a stmt. That improves SSA forwprop to
tree forward propagate : 1.30 ( 0%)
for this testcase.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
PR tree-optimization/26854
* gimple-fold.cc (has_use_on_stmt): Look at stmt operands
instead of immediate uses.
---
gcc/gimple-fold.cc | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)
diff --git a/gcc/gimple-fold.cc b/gcc/gimple-fold.cc
index 379d2e930ea..935e8006413 100644
--- a/gcc/gimple-fold.cc
+++ b/gcc/gimple-fold.cc
@@ -5767,15 +5767,17 @@ gimple_fold_call (gimple_stmt_iterator *gsi, bool
inplace)
}
-/* Return true whether NAME has a use on STMT. */
+/* Return true whether NAME has a use on STMT. Note this can return
+ false even though there's a use on STMT if SSA operands are not
+ up-to-date. */
static bool
has_use_on_stmt (tree name, gimple *stmt)
{
- imm_use_iterator iter;
- use_operand_p use_p;
- FOR_EACH_IMM_USE_FAST (use_p, iter, name)
- if (USE_STMT (use_p) == stmt)
+ ssa_op_iter iter;
+ tree op;
+ FOR_EACH_SSA_TREE_OPERAND (op, stmt, iter, SSA_OP_USE)
+ if (op == name)
return true;
return false;
}
--
2.35.3
