Update domination chain and RPO manually in MaybeAddExtraGotoBlocks
There's no need to recompute the whole graph as we know what changed.
As a drive-by, we now don't return false for graphs with irreducible
loops so we can remove that restriction from the builder. However,
if a graph with irreducible loops hits this path it means that:
A) it's being inlined
B) Has irreducible loops
We don't inline graphs with irreducible loops, and after building
for inline we don't remove them either because constant folding
and instruction simplifier don't remove them, and DCE doesn't
run for graphs with irreducible loops. So, in terms of dex2oat's
outputs nothing should change.
Test: art/test/testrunner/testrunner.py --host --64 --optimizing -b
Change-Id: I8cbf1b5f0518bb5dd14ffd751100ea81f5478863
diff --git a/compiler/optimizing/nodes.cc b/compiler/optimizing/nodes.cc
index 841576a..4a0ec93 100644
--- a/compiler/optimizing/nodes.cc
+++ b/compiler/optimizing/nodes.cc
@@ -569,6 +569,15 @@
}
}
+HBasicBlock* HGraph::SplitEdgeAndUpdateRPO(HBasicBlock* block, HBasicBlock* successor) {
+ HBasicBlock* new_block = SplitEdge(block, successor);
+ // In the RPO we have {... , block, ... , successor}. We want to insert `new_block` right after
+ // `block` to have a consistent RPO without recomputing the whole graph's RPO.
+ reverse_post_order_.insert(
+ reverse_post_order_.begin() + IndexOfElement(reverse_post_order_, block) + 1, new_block);
+ return new_block;
+}
+
// Reorder phi inputs to match reordering of the block's predecessors.
static void FixPhisAfterPredecessorsReodering(HBasicBlock* block, size_t first, size_t second) {
for (HInstructionIterator it(block->GetPhis()); !it.Done(); it.Advance()) {