When working on a smaller region like a loop version copy the main
time spent is now dominance fast query recompute which does a full
function DFS walk. The dominance queries within the region of
interest should be O(log n) without fast queries and we should do
on the order of O(n) of them which overall means reasonable
complexity.
For the artificial testcase I'm looking at this shaves off
considerable time again.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
* tree-into-ssa.cc (update_ssa): Do not forcefully
re-compute dominance fast queries for TODO_update_ssa_no_phi.
---
gcc/tree-into-ssa.cc | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/gcc/tree-into-ssa.cc b/gcc/tree-into-ssa.cc
index be71b629f97..d13fb720b37 100644
--- a/gcc/tree-into-ssa.cc
+++ b/gcc/tree-into-ssa.cc
@@ -3451,11 +3451,13 @@ update_ssa (unsigned update_flags)
phis_to_rewrite.create (last_basic_block_for_fn (cfun) + 1);
blocks_to_update = BITMAP_ALLOC (NULL);
- /* Ensure that the dominance information is up-to-date. */
- calculate_dominance_info (CDI_DOMINATORS);
-
insert_phi_p = (update_flags != TODO_update_ssa_no_phi);
+ /* Ensure that the dominance information is up-to-date and when we
+ are going to compute dominance frontiers fast queries are possible. */
+ if (insert_phi_p || dom_info_state (CDI_DOMINATORS) == DOM_NONE)
+ calculate_dominance_info (CDI_DOMINATORS);
+
/* If there are names defined in the replacement table, prepare
definition and use sites for all the names in NEW_SSA_NAMES and
OLD_SSA_NAMES. */
--
2.35.3
