Improve code sinking's sorting

Our previous sorting used StrictlyDominates which doesn't guarantee
the Transitivity of incomparability. We can use some extra
bookkeping to make sure we are sorting the instructions in reverse
domination order.

Bug: 249752167
Test: art/test/testrunner/testrunner.py --host --64 --optimizing -b
Change-Id: Ia01866b2073e2f7853ecba4df55919d1dd3cdea5

1 files changed, 53 insertions, 11 deletions

diff --git a/compiler/optimizing/code_sinking.cc b/compiler/optimizing/code_sinking.cc
index d759a16f48..33b5bd5169 100644
--- a/compiler/optimizing/code_sinking.cc
+++ b/compiler/optimizing/code_sinking.cc
@@ -16,6 +16,9 @@
 
 #include "code_sinking.h"
 
+#include <sstream>
+
+#include "android-base/logging.h"
 #include "base/arena_bit_vector.h"
 #include "base/array_ref.h"
 #include "base/bit_vector-inl.h"
@@ -335,10 +338,6 @@ void CodeSinking::SinkCodeToUncommonBranch(HBasicBlock* end_block) {
   processed_instructions.ClearAllBits();
   ArenaBitVector post_dominated(&allocator, graph_->GetBlocks().size(), /* expandable= */ false);
   post_dominated.ClearAllBits();
-  ArenaBitVector instructions_that_can_move(
-      &allocator, number_of_instructions, /* expandable= */ false);
-  instructions_that_can_move.ClearAllBits();
-  ScopedArenaVector<HInstruction*> move_in_order(allocator.Adapter(kArenaAllocMisc));
 
   // Step (1): Visit post order to get a subset of blocks post dominated by `end_block`.
   // TODO(ngeoffray): Getting the full set of post-dominated should be done by
@@ -411,6 +410,13 @@ void CodeSinking::SinkCodeToUncommonBranch(HBasicBlock* end_block) {
   HBasicBlock* common_dominator = finder.Get();
 
   // Step (2): iterate over the worklist to find sinking candidates.
+  ArenaBitVector instructions_that_can_move(
+      &allocator, number_of_instructions, /* expandable= */ false);
+  instructions_that_can_move.ClearAllBits();
+  ScopedArenaVector<ScopedArenaVector<HInstruction*>> instructions_to_move(
+      graph_->GetBlocks().size(),
+      ScopedArenaVector<HInstruction*>(allocator.Adapter(kArenaAllocMisc)),
+      allocator.Adapter(kArenaAllocMisc));
   while (!worklist.empty()) {
     HInstruction* instruction = worklist.back();
     if (processed_instructions.IsBitSet(instruction->GetId())) {
@@ -467,7 +473,7 @@ void CodeSinking::SinkCodeToUncommonBranch(HBasicBlock* end_block) {
       // Instruction is a candidate for being sunk. Mark it as such, remove it from the
       // work list, and add its inputs to the work list.
       instructions_that_can_move.SetBit(instruction->GetId());
-      move_in_order.push_back(instruction);
+      instructions_to_move[instruction->GetBlock()->GetBlockId()].push_back(instruction);
       processed_instructions.SetBit(instruction->GetId());
       worklist.pop_back();
       AddInputs(instruction, processed_instructions, post_dominated, &worklist);
@@ -493,14 +499,50 @@ void CodeSinking::SinkCodeToUncommonBranch(HBasicBlock* end_block) {
     }
   }
 
-  // Make sure we process instructions in dominated order. This is required for heap
-  // stores.
-  std::sort(move_in_order.begin(), move_in_order.end(), [](HInstruction* a, HInstruction* b) {
-    return b->StrictlyDominates(a);
-  });
+  // We want to process the instructions in reverse dominated order. This is required for heap
+  // stores. To guarantee this (including the transitivity of incomparability) we have some extra
+  // bookkeeping.
+  ScopedArenaVector<HInstruction*> instructions_to_move_sorted(allocator.Adapter(kArenaAllocMisc));
+  for (HBasicBlock* block : graph_->GetPostOrder()) {
+    const int block_id = block->GetBlockId();
+
+    // Order the block itself first.
+    std::sort(instructions_to_move[block_id].begin(),
+              instructions_to_move[block_id].end(),
+              [&block](HInstruction* a, HInstruction* b) {
+                return block->GetInstructions().FoundBefore(b, a);
+              });
+
+    for (HInstruction* instruction : instructions_to_move[block_id]) {
+      instructions_to_move_sorted.push_back(instruction);
+    }
+  }
+
+  if (kIsDebugBuild) {
+    // We should have ordered the instructions in reverse dominated order. This means that
+    // instructions shouldn't dominate instructions that come after it in the vector.
+    for (size_t i = 0; i < instructions_to_move_sorted.size(); ++i) {
+      for (size_t j = i + 1; j < instructions_to_move_sorted.size(); ++j) {
+        if (instructions_to_move_sorted[i]->StrictlyDominates(instructions_to_move_sorted[j])) {
+          std::stringstream ss;
+          graph_->Dump(ss, nullptr);
+          ss << "\n"
+             << "{";
+          for (HInstruction* instr : instructions_to_move_sorted) {
+            ss << *instr << " in block: " << instr->GetBlock() << ", ";
+          }
+          ss << "}\n";
+          ss << "i = " << i << " which is " << *instructions_to_move_sorted[i]
+             << "strictly dominates j = " << j << " which is " << *instructions_to_move_sorted[j]
+             << "\n";
+          LOG(FATAL) << "Unexpected ordering of code sinking instructions: " << ss.str();
+        }
+      }
+    }
+  }
 
   // Step (3): Try to move sinking candidates.
-  for (HInstruction* instruction : move_in_order) {
+  for (HInstruction* instruction : instructions_to_move_sorted) {
     HInstruction* position = nullptr;
     if (instruction->IsArraySet()
             || instruction->IsInstanceFieldSet()