/* * Copyright (C) 2013 The Android Open Source Project * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ #include "mir_graph.h" #include #include "base/stl_util.h" #include "compiler_internals.h" #include "dex_file-inl.h" #include "dex/quick/dex_file_to_method_inliner_map.h" #include "dex/quick/dex_file_method_inliner.h" #include "leb128.h" namespace art { #define MAX_PATTERN_LEN 5 const char* MIRGraph::extended_mir_op_names_[kMirOpLast - kMirOpFirst] = { "Phi", "Copy", "FusedCmplFloat", "FusedCmpgFloat", "FusedCmplDouble", "FusedCmpgDouble", "FusedCmpLong", "Nop", "OpNullCheck", "OpRangeCheck", "OpDivZeroCheck", "Check1", "Check2", "Select", }; MIRGraph::MIRGraph(CompilationUnit* cu, ArenaAllocator* arena) : reg_location_(NULL), cu_(cu), ssa_base_vregs_(NULL), ssa_subscripts_(NULL), vreg_to_ssa_map_(NULL), ssa_last_defs_(NULL), is_constant_v_(NULL), constant_values_(NULL), use_counts_(arena, 256, kGrowableArrayMisc), raw_use_counts_(arena, 256, kGrowableArrayMisc), num_reachable_blocks_(0), dfs_order_(NULL), dfs_post_order_(NULL), dom_post_order_traversal_(NULL), i_dom_list_(NULL), def_block_matrix_(NULL), temp_dalvik_register_v_(NULL), temp_scoped_alloc_(), temp_insn_data_(nullptr), temp_bit_vector_size_(0u), temp_bit_vector_(nullptr), block_list_(arena, 100, kGrowableArrayBlockList), try_block_addr_(NULL), entry_block_(NULL), exit_block_(NULL), num_blocks_(0), current_code_item_(NULL), dex_pc_to_block_map_(arena, 0, kGrowableArrayMisc), current_method_(kInvalidEntry), current_offset_(kInvalidEntry), def_count_(0), opcode_count_(NULL), num_ssa_regs_(0), method_sreg_(0), attributes_(METHOD_IS_LEAF), // Start with leaf assumption, change on encountering invoke. checkstats_(NULL), arena_(arena), backward_branches_(0), forward_branches_(0), compiler_temps_(arena, 6, kGrowableArrayMisc), num_non_special_compiler_temps_(0), max_available_non_special_compiler_temps_(0), punt_to_interpreter_(false), merged_df_flags_(0u), ifield_lowering_infos_(arena, 0u), sfield_lowering_infos_(arena, 0u), method_lowering_infos_(arena, 0u) { try_block_addr_ = new (arena_) ArenaBitVector(arena_, 0, true /* expandable */); max_available_special_compiler_temps_ = std::abs(static_cast(kVRegNonSpecialTempBaseReg)) - std::abs(static_cast(kVRegTempBaseReg)); } MIRGraph::~MIRGraph() { STLDeleteElements(&m_units_); } /* * Parse an instruction, return the length of the instruction */ int MIRGraph::ParseInsn(const uint16_t* code_ptr, DecodedInstruction* decoded_instruction) { const Instruction* instruction = Instruction::At(code_ptr); *decoded_instruction = DecodedInstruction(instruction); return instruction->SizeInCodeUnits(); } /* Split an existing block from the specified code offset into two */ BasicBlock* MIRGraph::SplitBlock(DexOffset code_offset, BasicBlock* orig_block, BasicBlock** immed_pred_block_p) { DCHECK_GT(code_offset, orig_block->start_offset); MIR* insn = orig_block->first_mir_insn; MIR* prev = NULL; while (insn) { if (insn->offset == code_offset) break; prev = insn; insn = insn->next; } if (insn == NULL) { LOG(FATAL) << "Break split failed"; } BasicBlock *bottom_block = NewMemBB(kDalvikByteCode, num_blocks_++); block_list_.Insert(bottom_block); bottom_block->start_offset = code_offset; bottom_block->first_mir_insn = insn; bottom_block->last_mir_insn = orig_block->last_mir_insn; /* If this block was terminated by a return, the flag needs to go with the bottom block */ bottom_block->terminated_by_return = orig_block->terminated_by_return; orig_block->terminated_by_return = false; /* Handle the taken path */ bottom_block->taken = orig_block->taken; if (bottom_block->taken != NullBasicBlockId) { orig_block->taken = NullBasicBlockId; BasicBlock* bb_taken = GetBasicBlock(bottom_block->taken); bb_taken->predecessors->Delete(orig_block->id); bb_taken->predecessors->Insert(bottom_block->id); } /* Handle the fallthrough path */ bottom_block->fall_through = orig_block->fall_through; orig_block->fall_through = bottom_block->id; bottom_block->predecessors->Insert(orig_block->id); if (bottom_block->fall_through != NullBasicBlockId) { BasicBlock* bb_fall_through = GetBasicBlock(bottom_block->fall_through); bb_fall_through->predecessors->Delete(orig_block->id); bb_fall_through->predecessors->Insert(bottom_block->id); } /* Handle the successor list */ if (orig_block->successor_block_list_type != kNotUsed) { bottom_block->successor_block_list_type = orig_block->successor_block_list_type; bottom_block->successor_blocks = orig_block->successor_blocks; orig_block->successor_block_list_type = kNotUsed; orig_block->successor_blocks = NULL; GrowableArray::Iterator iterator(bottom_block->successor_blocks); while (true) { SuccessorBlockInfo *successor_block_info = iterator.Next(); if (successor_block_info == NULL) break; BasicBlock *bb = GetBasicBlock(successor_block_info->block); bb->predecessors->Delete(orig_block->id); bb->predecessors->Insert(bottom_block->id); } } orig_block->last_mir_insn = prev; prev->next = NULL; /* * Update the immediate predecessor block pointer so that outgoing edges * can be applied to the proper block. */ if (immed_pred_block_p) { DCHECK_EQ(*immed_pred_block_p, orig_block); *immed_pred_block_p = bottom_block; } // Associate dex instructions in the bottom block with the new container. DCHECK(insn != nullptr); DCHECK(insn != orig_block->first_mir_insn); DCHECK(insn == bottom_block->first_mir_insn); DCHECK_EQ(insn->offset, bottom_block->start_offset); DCHECK(static_cast(insn->dalvikInsn.opcode) == kMirOpCheck || !IsPseudoMirOp(insn->dalvikInsn.opcode)); DCHECK_EQ(dex_pc_to_block_map_.Get(insn->offset), orig_block->id); MIR* p = insn; dex_pc_to_block_map_.Put(p->offset, bottom_block->id); while (p != bottom_block->last_mir_insn) { p = p->next; DCHECK(p != nullptr); int opcode = p->dalvikInsn.opcode; /* * Some messiness here to ensure that we only enter real opcodes and only the * first half of a potentially throwing instruction that has been split into * CHECK and work portions. Since the 2nd half of a split operation is always * the first in a BasicBlock, we can't hit it here. */ if ((opcode == kMirOpCheck) || !IsPseudoMirOp(opcode)) { DCHECK_EQ(dex_pc_to_block_map_.Get(p->offset), orig_block->id); dex_pc_to_block_map_.Put(p->offset, bottom_block->id); } } return bottom_block; } /* * Given a code offset, find out the block that starts with it. If the offset * is in the middle of an existing block, split it into two. If immed_pred_block_p * is not non-null and is the block being split, update *immed_pred_block_p to * point to the bottom block so that outgoing edges can be set up properly * (by the caller) * Utilizes a map for fast lookup of the typical cases. */ BasicBlock* MIRGraph::FindBlock(DexOffset code_offset, bool split, bool create, BasicBlock** immed_pred_block_p) { if (code_offset >= cu_->code_item->insns_size_in_code_units_) { return NULL; } int block_id = dex_pc_to_block_map_.Get(code_offset); BasicBlock* bb = (block_id == 0) ? NULL : block_list_.Get(block_id); if ((bb != NULL) && (bb->start_offset == code_offset)) { // Does this containing block start with the desired instruction? return bb; } // No direct hit. if (!create) { return NULL; } if (bb != NULL) { // The target exists somewhere in an existing block. return SplitBlock(code_offset, bb, bb == *immed_pred_block_p ? immed_pred_block_p : NULL); } // Create a new block. bb = NewMemBB(kDalvikByteCode, num_blocks_++); block_list_.Insert(bb); bb->start_offset = code_offset; dex_pc_to_block_map_.Put(bb->start_offset, bb->id); return bb; } /* Identify code range in try blocks and set up the empty catch blocks */ void MIRGraph::ProcessTryCatchBlocks() { int tries_size = current_code_item_->tries_size_; DexOffset offset; if (tries_size == 0) { return; } for (int i = 0; i < tries_size; i++) { const DexFile::TryItem* pTry = DexFile::GetTryItems(*current_code_item_, i); DexOffset start_offset = pTry->start_addr_; DexOffset end_offset = start_offset + pTry->insn_count_; for (offset = start_offset; offset < end_offset; offset++) { try_block_addr_->SetBit(offset); } } // Iterate over each of the handlers to enqueue the empty Catch blocks const byte* handlers_ptr = DexFile::GetCatchHandlerData(*current_code_item_, 0); uint32_t handlers_size = DecodeUnsignedLeb128(&handlers_ptr); for (uint32_t idx = 0; idx < handlers_size; idx++) { CatchHandlerIterator iterator(handlers_ptr); for (; iterator.HasNext(); iterator.Next()) { uint32_t address = iterator.GetHandlerAddress(); FindBlock(address, false /* split */, true /*create*/, /* immed_pred_block_p */ NULL); } handlers_ptr = iterator.EndDataPointer(); } } /* Process instructions with the kBranch flag */ BasicBlock* MIRGraph::ProcessCanBranch(BasicBlock* cur_block, MIR* insn, DexOffset cur_offset, int width, int flags, const uint16_t* code_ptr, const uint16_t* code_end) { DexOffset target = cur_offset; switch (insn->dalvikInsn.opcode) { case Instruction::GOTO: case Instruction::GOTO_16: case Instruction::GOTO_32: target += insn->dalvikInsn.vA; break; case Instruction::IF_EQ: case Instruction::IF_NE: case Instruction::IF_LT: case Instruction::IF_GE: case Instruction::IF_GT: case Instruction::IF_LE: cur_block->conditional_branch = true; target += insn->dalvikInsn.vC; break; case Instruction::IF_EQZ: case Instruction::IF_NEZ: case Instruction::IF_LTZ: case Instruction::IF_GEZ: case Instruction::IF_GTZ: case Instruction::IF_LEZ: cur_block->conditional_branch = true; target += insn->dalvikInsn.vB; break; default: LOG(FATAL) << "Unexpected opcode(" << insn->dalvikInsn.opcode << ") with kBranch set"; } CountBranch(target); BasicBlock *taken_block = FindBlock(target, /* split */ true, /* create */ true, /* immed_pred_block_p */ &cur_block); cur_block->taken = taken_block->id; taken_block->predecessors->Insert(cur_block->id); /* Always terminate the current block for conditional branches */ if (flags & Instruction::kContinue) { BasicBlock *fallthrough_block = FindBlock(cur_offset + width, /* * If the method is processed * in sequential order from the * beginning, we don't need to * specify split for continue * blocks. However, this * routine can be called by * compileLoop, which starts * parsing the method from an * arbitrary address in the * method body. */ true, /* create */ true, /* immed_pred_block_p */ &cur_block); cur_block->fall_through = fallthrough_block->id; fallthrough_block->predecessors->Insert(cur_block->id); } else if (code_ptr < code_end) { FindBlock(cur_offset + width, /* split */ false, /* create */ true, /* immed_pred_block_p */ NULL); } return cur_block; } /* Process instructions with the kSwitch flag */ BasicBlock* MIRGraph::ProcessCanSwitch(BasicBlock* cur_block, MIR* insn, DexOffset cur_offset, int width, int flags) { const uint16_t* switch_data = reinterpret_cast(GetCurrentInsns() + cur_offset + insn->dalvikInsn.vB); int size; const int* keyTable; const int* target_table; int i; int first_key; /* * Packed switch data format: * ushort ident = 0x0100 magic value * ushort size number of entries in the table * int first_key first (and lowest) switch case value * int targets[size] branch targets, relative to switch opcode * * Total size is (4+size*2) 16-bit code units. */ if (insn->dalvikInsn.opcode == Instruction::PACKED_SWITCH) { DCHECK_EQ(static_cast(switch_data[0]), static_cast(Instruction::kPackedSwitchSignature)); size = switch_data[1]; first_key = switch_data[2] | (switch_data[3] << 16); target_table = reinterpret_cast(&switch_data[4]); keyTable = NULL; // Make the compiler happy /* * Sparse switch data format: * ushort ident = 0x0200 magic value * ushort size number of entries in the table; > 0 * int keys[size] keys, sorted low-to-high; 32-bit aligned * int targets[size] branch targets, relative to switch opcode * * Total size is (2+size*4) 16-bit code units. */ } else { DCHECK_EQ(static_cast(switch_data[0]), static_cast(Instruction::kSparseSwitchSignature)); size = switch_data[1]; keyTable = reinterpret_cast(&switch_data[2]); target_table = reinterpret_cast(&switch_data[2 + size*2]); first_key = 0; // To make the compiler happy } if (cur_block->successor_block_list_type != kNotUsed) { LOG(FATAL) << "Successor block list already in use: " << static_cast(cur_block->successor_block_list_type); } cur_block->successor_block_list_type = (insn->dalvikInsn.opcode == Instruction::PACKED_SWITCH) ? kPackedSwitch : kSparseSwitch; cur_block->successor_blocks = new (arena_) GrowableArray(arena_, size, kGrowableArraySuccessorBlocks); for (i = 0; i < size; i++) { BasicBlock *case_block = FindBlock(cur_offset + target_table[i], /* split */ true, /* create */ true, /* immed_pred_block_p */ &cur_block); SuccessorBlockInfo *successor_block_info = static_cast(arena_->Alloc(sizeof(SuccessorBlockInfo), kArenaAllocSuccessor)); successor_block_info->block = case_block->id; successor_block_info->key = (insn->dalvikInsn.opcode == Instruction::PACKED_SWITCH) ? first_key + i : keyTable[i]; cur_block->successor_blocks->Insert(successor_block_info); case_block->predecessors->Insert(cur_block->id); } /* Fall-through case */ BasicBlock* fallthrough_block = FindBlock(cur_offset + width, /* split */ false, /* create */ true, /* immed_pred_block_p */ NULL); cur_block->fall_through = fallthrough_block->id; fallthrough_block->predecessors->Insert(cur_block->id); return cur_block; } /* Process instructions with the kThrow flag */ BasicBlock* MIRGraph::ProcessCanThrow(BasicBlock* cur_block, MIR* insn, DexOffset cur_offset, int width, int flags, ArenaBitVector* try_block_addr, const uint16_t* code_ptr, const uint16_t* code_end) { bool in_try_block = try_block_addr->IsBitSet(cur_offset); bool is_throw = (insn->dalvikInsn.opcode == Instruction::THROW); bool build_all_edges = (cu_->disable_opt & (1 << kSuppressExceptionEdges)) || is_throw || in_try_block; /* In try block */ if (in_try_block) { CatchHandlerIterator iterator(*current_code_item_, cur_offset); if (cur_block->successor_block_list_type != kNotUsed) { LOG(INFO) << PrettyMethod(cu_->method_idx, *cu_->dex_file); LOG(FATAL) << "Successor block list already in use: " << static_cast(cur_block->successor_block_list_type); } cur_block->successor_block_list_type = kCatch; cur_block->successor_blocks = new (arena_) GrowableArray(arena_, 2, kGrowableArraySuccessorBlocks); for (; iterator.HasNext(); iterator.Next()) { BasicBlock *catch_block = FindBlock(iterator.GetHandlerAddress(), false /* split*/, false /* creat */, NULL /* immed_pred_block_p */); catch_block->catch_entry = true; if (kIsDebugBuild) { catches_.insert(catch_block->start_offset); } SuccessorBlockInfo *successor_block_info = reinterpret_cast (arena_->Alloc(sizeof(SuccessorBlockInfo), kArenaAllocSuccessor)); successor_block_info->block = catch_block->id; successor_block_info->key = iterator.GetHandlerTypeIndex(); cur_block->successor_blocks->Insert(successor_block_info); catch_block->predecessors->Insert(cur_block->id); } } else if (build_all_edges) { BasicBlock *eh_block = NewMemBB(kExceptionHandling, num_blocks_++); cur_block->taken = eh_block->id; block_list_.Insert(eh_block); eh_block->start_offset = cur_offset; eh_block->predecessors->Insert(cur_block->id); } if (is_throw) { cur_block->explicit_throw = true; if (code_ptr < code_end) { // Force creation of new block following THROW via side-effect FindBlock(cur_offset + width, /* split */ false, /* create */ true, /* immed_pred_block_p */ NULL); } if (!in_try_block) { // Don't split a THROW that can't rethrow - we're done. return cur_block; } } if (!build_all_edges) { /* * Even though there is an exception edge here, control cannot return to this * method. Thus, for the purposes of dataflow analysis and optimization, we can * ignore the edge. Doing this reduces compile time, and increases the scope * of the basic-block level optimization pass. */ return cur_block; } /* * Split the potentially-throwing instruction into two parts. * The first half will be a pseudo-op that captures the exception * edges and terminates the basic block. It always falls through. * Then, create a new basic block that begins with the throwing instruction * (minus exceptions). Note: this new basic block must NOT be entered into * the block_map. If the potentially-throwing instruction is the target of a * future branch, we need to find the check psuedo half. The new * basic block containing the work portion of the instruction should * only be entered via fallthrough from the block containing the * pseudo exception edge MIR. Note also that this new block is * not automatically terminated after the work portion, and may * contain following instructions. * * Note also that the dex_pc_to_block_map_ entry for the potentially * throwing instruction will refer to the original basic block. */ BasicBlock *new_block = NewMemBB(kDalvikByteCode, num_blocks_++); block_list_.Insert(new_block); new_block->start_offset = insn->offset; cur_block->fall_through = new_block->id; new_block->predecessors->Insert(cur_block->id); MIR* new_insn = static_cast(arena_->Alloc(sizeof(MIR), kArenaAllocMIR)); *new_insn = *insn; insn->dalvikInsn.opcode = static_cast(kMirOpCheck); // Associate the two halves insn->meta.throw_insn = new_insn; new_block->AppendMIR(new_insn); return new_block; } /* Parse a Dex method and insert it into the MIRGraph at the current insert point. */ void MIRGraph::InlineMethod(const DexFile::CodeItem* code_item, uint32_t access_flags, InvokeType invoke_type, uint16_t class_def_idx, uint32_t method_idx, jobject class_loader, const DexFile& dex_file) { current_code_item_ = code_item; method_stack_.push_back(std::make_pair(current_method_, current_offset_)); current_method_ = m_units_.size(); current_offset_ = 0; // TODO: will need to snapshot stack image and use that as the mir context identification. m_units_.push_back(new DexCompilationUnit(cu_, class_loader, Runtime::Current()->GetClassLinker(), dex_file, current_code_item_, class_def_idx, method_idx, access_flags, cu_->compiler_driver->GetVerifiedMethod(&dex_file, method_idx))); const uint16_t* code_ptr = current_code_item_->insns_; const uint16_t* code_end = current_code_item_->insns_ + current_code_item_->insns_size_in_code_units_; // TODO: need to rework expansion of block list & try_block_addr when inlining activated. // TUNING: use better estimate of basic blocks for following resize. block_list_.Resize(block_list_.Size() + current_code_item_->insns_size_in_code_units_); dex_pc_to_block_map_.SetSize(dex_pc_to_block_map_.Size() + current_code_item_->insns_size_in_code_units_); // TODO: replace with explicit resize routine. Using automatic extension side effect for now. try_block_addr_->SetBit(current_code_item_->insns_size_in_code_units_); try_block_addr_->ClearBit(current_code_item_->insns_size_in_code_units_); // If this is the first method, set up default entry and exit blocks. if (current_method_ == 0) { DCHECK(entry_block_ == NULL); DCHECK(exit_block_ == NULL); DCHECK_EQ(num_blocks_, 0); // Use id 0 to represent a null block. BasicBlock* null_block = NewMemBB(kNullBlock, num_blocks_++); DCHECK_EQ(null_block->id, NullBasicBlockId); null_block->hidden = true; block_list_.Insert(null_block); entry_block_ = NewMemBB(kEntryBlock, num_blocks_++); block_list_.Insert(entry_block_); exit_block_ = NewMemBB(kExitBlock, num_blocks_++); block_list_.Insert(exit_block_); // TODO: deprecate all "cu->" fields; move what's left to wherever CompilationUnit is allocated. cu_->dex_file = &dex_file; cu_->class_def_idx = class_def_idx; cu_->method_idx = method_idx; cu_->access_flags = access_flags; cu_->invoke_type = invoke_type; cu_->shorty = dex_file.GetMethodShorty(dex_file.GetMethodId(method_idx)); cu_->num_ins = current_code_item_->ins_size_; cu_->num_regs = current_code_item_->registers_size_ - cu_->num_ins; cu_->num_outs = current_code_item_->outs_size_; cu_->num_dalvik_registers = current_code_item_->registers_size_; cu_->insns = current_code_item_->insns_; cu_->code_item = current_code_item_; } else { UNIMPLEMENTED(FATAL) << "Nested inlining not implemented."; /* * Will need to manage storage for ins & outs, push prevous state and update * insert point. */ } /* Current block to record parsed instructions */ BasicBlock *cur_block = NewMemBB(kDalvikByteCode, num_blocks_++); DCHECK_EQ(current_offset_, 0U); cur_block->start_offset = current_offset_; block_list_.Insert(cur_block); // TODO: for inlining support, insert at the insert point rather than entry block. entry_block_->fall_through = cur_block->id; cur_block->predecessors->Insert(entry_block_->id); /* Identify code range in try blocks and set up the empty catch blocks */ ProcessTryCatchBlocks(); uint64_t merged_df_flags = 0u; /* Parse all instructions and put them into containing basic blocks */ while (code_ptr < code_end) { MIR *insn = static_cast(arena_->Alloc(sizeof(MIR), kArenaAllocMIR)); insn->offset = current_offset_; insn->m_unit_index = current_method_; int width = ParseInsn(code_ptr, &insn->dalvikInsn); insn->width = width; Instruction::Code opcode = insn->dalvikInsn.opcode; if (opcode_count_ != NULL) { opcode_count_[static_cast(opcode)]++; } int flags = Instruction::FlagsOf(insn->dalvikInsn.opcode); int verify_flags = Instruction::VerifyFlagsOf(insn->dalvikInsn.opcode); uint64_t df_flags = oat_data_flow_attributes_[insn->dalvikInsn.opcode]; merged_df_flags |= df_flags; if (df_flags & DF_HAS_DEFS) { def_count_ += (df_flags & DF_A_WIDE) ? 2 : 1; } if (df_flags & DF_LVN) { cur_block->use_lvn = true; // Run local value numbering on this basic block. } // Check for inline data block signatures if (opcode == Instruction::NOP) { // A simple NOP will have a width of 1 at this point, embedded data NOP > 1. if ((width == 1) && ((current_offset_ & 0x1) == 0x1) && ((code_end - code_ptr) > 1)) { // Could be an aligning nop. If an embedded data NOP follows, treat pair as single unit. uint16_t following_raw_instruction = code_ptr[1]; if ((following_raw_instruction == Instruction::kSparseSwitchSignature) || (following_raw_instruction == Instruction::kPackedSwitchSignature) || (following_raw_instruction == Instruction::kArrayDataSignature)) { width += Instruction::At(code_ptr + 1)->SizeInCodeUnits(); } } if (width == 1) { // It is a simple nop - treat normally. cur_block->AppendMIR(insn); } else { DCHECK(cur_block->fall_through == NullBasicBlockId); DCHECK(cur_block->taken == NullBasicBlockId); // Unreachable instruction, mark for no continuation. flags &= ~Instruction::kContinue; } } else { cur_block->AppendMIR(insn); } // Associate the starting dex_pc for this opcode with its containing basic block. dex_pc_to_block_map_.Put(insn->offset, cur_block->id); code_ptr += width; if (flags & Instruction::kBranch) { cur_block = ProcessCanBranch(cur_block, insn, current_offset_, width, flags, code_ptr, code_end); } else if (flags & Instruction::kReturn) { cur_block->terminated_by_return = true; cur_block->fall_through = exit_block_->id; exit_block_->predecessors->Insert(cur_block->id); /* * Terminate the current block if there are instructions * afterwards. */ if (code_ptr < code_end) { /* * Create a fallthrough block for real instructions * (incl. NOP). */ FindBlock(current_offset_ + width, /* split */ false, /* create */ true, /* immed_pred_block_p */ NULL); } } else if (flags & Instruction::kThrow) { cur_block = ProcessCanThrow(cur_block, insn, current_offset_, width, flags, try_block_addr_, code_ptr, code_end); } else if (flags & Instruction::kSwitch) { cur_block = ProcessCanSwitch(cur_block, insn, current_offset_, width, flags); } if (verify_flags & Instruction::kVerifyVarArgRange) { /* * The Quick backend's runtime model includes a gap between a method's * argument ("in") vregs and the rest of its vregs. Handling a range instruction * which spans the gap is somewhat complicated, and should not happen * in normal usage of dx. Punt to the interpreter. */ int first_reg_in_range = insn->dalvikInsn.vC; int last_reg_in_range = first_reg_in_range + insn->dalvikInsn.vA - 1; if (IsInVReg(first_reg_in_range) != IsInVReg(last_reg_in_range)) { punt_to_interpreter_ = true; } } current_offset_ += width; BasicBlock *next_block = FindBlock(current_offset_, /* split */ false, /* create */ false, /* immed_pred_block_p */ NULL); if (next_block) { /* * The next instruction could be the target of a previously parsed * forward branch so a block is already created. If the current * instruction is not an unconditional branch, connect them through * the fall-through link. */ DCHECK(cur_block->fall_through == NullBasicBlockId || GetBasicBlock(cur_block->fall_through) == next_block || GetBasicBlock(cur_block->fall_through) == exit_block_); if ((cur_block->fall_through == NullBasicBlockId) && (flags & Instruction::kContinue)) { cur_block->fall_through = next_block->id; next_block->predecessors->Insert(cur_block->id); } cur_block = next_block; } } merged_df_flags_ = merged_df_flags; if (cu_->enable_debug & (1 << kDebugDumpCFG)) { DumpCFG("/sdcard/1_post_parse_cfg/", true); } if (cu_->verbose) { DumpMIRGraph(); } } void MIRGraph::ShowOpcodeStats() { DCHECK(opcode_count_ != NULL); LOG(INFO) << "Opcode Count"; for (int i = 0; i < kNumPackedOpcodes; i++) { if (opcode_count_[i] != 0) { LOG(INFO) << "-C- " << Instruction::Name(static_cast(i)) << " " << opcode_count_[i]; } } } // TODO: use a configurable base prefix, and adjust callers to supply pass name. /* Dump the CFG into a DOT graph */ void MIRGraph::DumpCFG(const char* dir_prefix, bool all_blocks, const char *suffix) { FILE* file; std::string fname(PrettyMethod(cu_->method_idx, *cu_->dex_file)); ReplaceSpecialChars(fname); fname = StringPrintf("%s%s%x%s.dot", dir_prefix, fname.c_str(), GetBasicBlock(GetEntryBlock()->fall_through)->start_offset, suffix == nullptr ? "" : suffix); file = fopen(fname.c_str(), "w"); if (file == NULL) { return; } fprintf(file, "digraph G {\n"); fprintf(file, " rankdir=TB\n"); int num_blocks = all_blocks ? GetNumBlocks() : num_reachable_blocks_; int idx; for (idx = 0; idx < num_blocks; idx++) { int block_idx = all_blocks ? idx : dfs_order_->Get(idx); BasicBlock *bb = GetBasicBlock(block_idx); if (bb == NULL) continue; if (bb->block_type == kDead) continue; if (bb->block_type == kEntryBlock) { fprintf(file, " entry_%d [shape=Mdiamond];\n", bb->id); } else if (bb->block_type == kExitBlock) { fprintf(file, " exit_%d [shape=Mdiamond];\n", bb->id); } else if (bb->block_type == kDalvikByteCode) { fprintf(file, " block%04x_%d [shape=record,label = \"{ \\\n", bb->start_offset, bb->id); const MIR *mir; fprintf(file, " {block id %d\\l}%s\\\n", bb->id, bb->first_mir_insn ? " | " : " "); for (mir = bb->first_mir_insn; mir; mir = mir->next) { int opcode = mir->dalvikInsn.opcode; fprintf(file, " {%04x %s %s %s\\l}%s\\\n", mir->offset, mir->ssa_rep ? GetDalvikDisassembly(mir) : (opcode < kMirOpFirst) ? Instruction::Name(mir->dalvikInsn.opcode) : extended_mir_op_names_[opcode - kMirOpFirst], (mir->optimization_flags & MIR_IGNORE_RANGE_CHECK) != 0 ? " no_rangecheck" : " ", (mir->optimization_flags & MIR_IGNORE_NULL_CHECK) != 0 ? " no_nullcheck" : " ", mir->next ? " | " : " "); } fprintf(file, " }\"];\n\n"); } else if (bb->block_type == kExceptionHandling) { char block_name[BLOCK_NAME_LEN]; GetBlockName(bb, block_name); fprintf(file, " %s [shape=invhouse];\n", block_name); } char block_name1[BLOCK_NAME_LEN], block_name2[BLOCK_NAME_LEN]; if (bb->taken != NullBasicBlockId) { GetBlockName(bb, block_name1); GetBlockName(GetBasicBlock(bb->taken), block_name2); fprintf(file, " %s:s -> %s:n [style=dotted]\n", block_name1, block_name2); } if (bb->fall_through != NullBasicBlockId) { GetBlockName(bb, block_name1); GetBlockName(GetBasicBlock(bb->fall_through), block_name2); fprintf(file, " %s:s -> %s:n\n", block_name1, block_name2); } if (bb->successor_block_list_type != kNotUsed) { fprintf(file, " succ%04x_%d [shape=%s,label = \"{ \\\n", bb->start_offset, bb->id, (bb->successor_block_list_type == kCatch) ? "Mrecord" : "record"); GrowableArray::Iterator iterator(bb->successor_blocks); SuccessorBlockInfo *successor_block_info = iterator.Next(); int succ_id = 0; while (true) { if (successor_block_info == NULL) break; BasicBlock *dest_block = GetBasicBlock(successor_block_info->block); SuccessorBlockInfo *next_successor_block_info = iterator.Next(); fprintf(file, " { %04x: %04x\\l}%s\\\n", succ_id++, successor_block_info->key, dest_block->start_offset, (next_successor_block_info != NULL) ? " | " : " "); successor_block_info = next_successor_block_info; } fprintf(file, " }\"];\n\n"); GetBlockName(bb, block_name1); fprintf(file, " %s:s -> succ%04x_%d:n [style=dashed]\n", block_name1, bb->start_offset, bb->id); // Link the successor pseudo-block with all of its potential targets. GrowableArray::Iterator iter(bb->successor_blocks); succ_id = 0; while (true) { SuccessorBlockInfo *successor_block_info = iter.Next(); if (successor_block_info == NULL) break; BasicBlock* dest_block = GetBasicBlock(successor_block_info->block); GetBlockName(dest_block, block_name2); fprintf(file, " succ%04x_%d:f%d:e -> %s:n\n", bb->start_offset, bb->id, succ_id++, block_name2); } } fprintf(file, "\n"); if (cu_->verbose) { /* Display the dominator tree */ GetBlockName(bb, block_name1); fprintf(file, " cfg%s [label=\"%s\", shape=none];\n", block_name1, block_name1); if (bb->i_dom) { GetBlockName(GetBasicBlock(bb->i_dom), block_name2); fprintf(file, " cfg%s:s -> cfg%s:n\n\n", block_name2, block_name1); } } } fprintf(file, "}\n"); fclose(file); } /* Insert an MIR instruction to the end of a basic block */ void BasicBlock::AppendMIR(MIR* mir) { if (first_mir_insn == nullptr) { DCHECK(last_mir_insn == nullptr); last_mir_insn = first_mir_insn = mir; mir->next = nullptr; } else { last_mir_insn->next = mir; mir->next = nullptr; last_mir_insn = mir; } } /* Insert an MIR instruction to the head of a basic block */ void BasicBlock::PrependMIR(MIR* mir) { if (first_mir_insn == nullptr) { DCHECK(last_mir_insn == nullptr); last_mir_insn = first_mir_insn = mir; mir->next = nullptr; } else { mir->next = first_mir_insn; first_mir_insn = mir; } } /* Insert a MIR instruction after the specified MIR */ void BasicBlock::InsertMIRAfter(MIR* current_mir, MIR* new_mir) { new_mir->next = current_mir->next; current_mir->next = new_mir; if (last_mir_insn == current_mir) { /* Is the last MIR in the block */ last_mir_insn = new_mir; } } MIR* BasicBlock::GetNextUnconditionalMir(MIRGraph* mir_graph, MIR* current) { MIR* next_mir = nullptr; if (current != nullptr) { next_mir = current->next; } if (next_mir == nullptr) { // Only look for next MIR that follows unconditionally. if ((taken == NullBasicBlockId) && (fall_through != NullBasicBlockId)) { next_mir = mir_graph->GetBasicBlock(fall_through)->first_mir_insn; } } return next_mir; } char* MIRGraph::GetDalvikDisassembly(const MIR* mir) { DecodedInstruction insn = mir->dalvikInsn; std::string str; int flags = 0; int opcode = insn.opcode; char* ret; bool nop = false; SSARepresentation* ssa_rep = mir->ssa_rep; Instruction::Format dalvik_format = Instruction::k10x; // Default to no-operand format int defs = (ssa_rep != NULL) ? ssa_rep->num_defs : 0; int uses = (ssa_rep != NULL) ? ssa_rep->num_uses : 0; // Handle special cases. if ((opcode == kMirOpCheck) || (opcode == kMirOpCheckPart2)) { str.append(extended_mir_op_names_[opcode - kMirOpFirst]); str.append(": "); // Recover the original Dex instruction insn = mir->meta.throw_insn->dalvikInsn; ssa_rep = mir->meta.throw_insn->ssa_rep; defs = ssa_rep->num_defs; uses = ssa_rep->num_uses; opcode = insn.opcode; } else if (opcode == kMirOpNop) { str.append("["); // Recover original opcode. insn.opcode = Instruction::At(current_code_item_->insns_ + mir->offset)->Opcode(); opcode = insn.opcode; nop = true; } if (opcode >= kMirOpFirst) { str.append(extended_mir_op_names_[opcode - kMirOpFirst]); } else { dalvik_format = Instruction::FormatOf(insn.opcode); flags = Instruction::FlagsOf(insn.opcode); str.append(Instruction::Name(insn.opcode)); } if (opcode == kMirOpPhi) { BasicBlockId* incoming = mir->meta.phi_incoming; str.append(StringPrintf(" %s = (%s", GetSSANameWithConst(ssa_rep->defs[0], true).c_str(), GetSSANameWithConst(ssa_rep->uses[0], true).c_str())); str.append(StringPrintf(":%d", incoming[0])); int i; for (i = 1; i < uses; i++) { str.append(StringPrintf(", %s:%d", GetSSANameWithConst(ssa_rep->uses[i], true).c_str(), incoming[i])); } str.append(")"); } else if ((flags & Instruction::kBranch) != 0) { // For branches, decode the instructions to print out the branch targets. int offset = 0; switch (dalvik_format) { case Instruction::k21t: str.append(StringPrintf(" %s,", GetSSANameWithConst(ssa_rep->uses[0], false).c_str())); offset = insn.vB; break; case Instruction::k22t: str.append(StringPrintf(" %s, %s,", GetSSANameWithConst(ssa_rep->uses[0], false).c_str(), GetSSANameWithConst(ssa_rep->uses[1], false).c_str())); offset = insn.vC; break; case Instruction::k10t: case Instruction::k20t: case Instruction::k30t: offset = insn.vA; break; default: LOG(FATAL) << "Unexpected branch format " << dalvik_format << " from " << insn.opcode; } str.append(StringPrintf(" 0x%x (%c%x)", mir->offset + offset, offset > 0 ? '+' : '-', offset > 0 ? offset : -offset)); } else { // For invokes-style formats, treat wide regs as a pair of singles bool show_singles = ((dalvik_format == Instruction::k35c) || (dalvik_format == Instruction::k3rc)); if (defs != 0) { str.append(StringPrintf(" %s", GetSSANameWithConst(ssa_rep->defs[0], false).c_str())); if (uses != 0) { str.append(", "); } } for (int i = 0; i < uses; i++) { str.append( StringPrintf(" %s", GetSSANameWithConst(ssa_rep->uses[i], show_singles).c_str())); if (!show_singles && (reg_location_ != NULL) && reg_location_[i].wide) { // For the listing, skip the high sreg. i++; } if (i != (uses -1)) { str.append(","); } } switch (dalvik_format) { case Instruction::k11n: // Add one immediate from vB case Instruction::k21s: case Instruction::k31i: case Instruction::k21h: str.append(StringPrintf(", #%d", insn.vB)); break; case Instruction::k51l: // Add one wide immediate str.append(StringPrintf(", #%" PRId64, insn.vB_wide)); break; case Instruction::k21c: // One register, one string/type/method index case Instruction::k31c: str.append(StringPrintf(", index #%d", insn.vB)); break; case Instruction::k22c: // Two registers, one string/type/method index str.append(StringPrintf(", index #%d", insn.vC)); break; case Instruction::k22s: // Add one immediate from vC case Instruction::k22b: str.append(StringPrintf(", #%d", insn.vC)); break; default: { // Nothing left to print } } } if (nop) { str.append("]--optimized away"); } int length = str.length() + 1; ret = static_cast(arena_->Alloc(length, kArenaAllocDFInfo)); strncpy(ret, str.c_str(), length); return ret; } /* Turn method name into a legal Linux file name */ void MIRGraph::ReplaceSpecialChars(std::string& str) { static const struct { const char before; const char after; } match[] = { {'/', '-'}, {';', '#'}, {' ', '#'}, {'$', '+'}, {'(', '@'}, {')', '@'}, {'<', '='}, {'>', '='} }; for (unsigned int i = 0; i < sizeof(match)/sizeof(match[0]); i++) { std::replace(str.begin(), str.end(), match[i].before, match[i].after); } } std::string MIRGraph::GetSSAName(int ssa_reg) { // TODO: This value is needed for LLVM and debugging. Currently, we compute this and then copy to // the arena. We should be smarter and just place straight into the arena, or compute the // value more lazily. return StringPrintf("v%d_%d", SRegToVReg(ssa_reg), GetSSASubscript(ssa_reg)); } // Similar to GetSSAName, but if ssa name represents an immediate show that as well. std::string MIRGraph::GetSSANameWithConst(int ssa_reg, bool singles_only) { if (reg_location_ == NULL) { // Pre-SSA - just use the standard name return GetSSAName(ssa_reg); } if (IsConst(reg_location_[ssa_reg])) { if (!singles_only && reg_location_[ssa_reg].wide) { return StringPrintf("v%d_%d#0x%" PRIx64, SRegToVReg(ssa_reg), GetSSASubscript(ssa_reg), ConstantValueWide(reg_location_[ssa_reg])); } else { return StringPrintf("v%d_%d#0x%x", SRegToVReg(ssa_reg), GetSSASubscript(ssa_reg), ConstantValue(reg_location_[ssa_reg])); } } else { return StringPrintf("v%d_%d", SRegToVReg(ssa_reg), GetSSASubscript(ssa_reg)); } } void MIRGraph::GetBlockName(BasicBlock* bb, char* name) { switch (bb->block_type) { case kEntryBlock: snprintf(name, BLOCK_NAME_LEN, "entry_%d", bb->id); break; case kExitBlock: snprintf(name, BLOCK_NAME_LEN, "exit_%d", bb->id); break; case kDalvikByteCode: snprintf(name, BLOCK_NAME_LEN, "block%04x_%d", bb->start_offset, bb->id); break; case kExceptionHandling: snprintf(name, BLOCK_NAME_LEN, "exception%04x_%d", bb->start_offset, bb->id); break; default: snprintf(name, BLOCK_NAME_LEN, "_%d", bb->id); break; } } const char* MIRGraph::GetShortyFromTargetIdx(int target_idx) { // TODO: for inlining support, use current code unit. const DexFile::MethodId& method_id = cu_->dex_file->GetMethodId(target_idx); return cu_->dex_file->GetShorty(method_id.proto_idx_); } /* Debug Utility - dump a compilation unit */ void MIRGraph::DumpMIRGraph() { BasicBlock* bb; const char* block_type_names[] = { "Null Block", "Entry Block", "Code Block", "Exit Block", "Exception Handling", "Catch Block" }; LOG(INFO) << "Compiling " << PrettyMethod(cu_->method_idx, *cu_->dex_file); LOG(INFO) << cu_->insns << " insns"; LOG(INFO) << GetNumBlocks() << " blocks in total"; GrowableArray::Iterator iterator(&block_list_); while (true) { bb = iterator.Next(); if (bb == NULL) break; LOG(INFO) << StringPrintf("Block %d (%s) (insn %04x - %04x%s)", bb->id, block_type_names[bb->block_type], bb->start_offset, bb->last_mir_insn ? bb->last_mir_insn->offset : bb->start_offset, bb->last_mir_insn ? "" : " empty"); if (bb->taken != NullBasicBlockId) { LOG(INFO) << " Taken branch: block " << bb->taken << "(0x" << std::hex << GetBasicBlock(bb->taken)->start_offset << ")"; } if (bb->fall_through != NullBasicBlockId) { LOG(INFO) << " Fallthrough : block " << bb->fall_through << " (0x" << std::hex << GetBasicBlock(bb->fall_through)->start_offset << ")"; } } } /* * Build an array of location records for the incoming arguments. * Note: one location record per word of arguments, with dummy * high-word loc for wide arguments. Also pull up any following * MOVE_RESULT and incorporate it into the invoke. */ CallInfo* MIRGraph::NewMemCallInfo(BasicBlock* bb, MIR* mir, InvokeType type, bool is_range) { CallInfo* info = static_cast(arena_->Alloc(sizeof(CallInfo), kArenaAllocMisc)); MIR* move_result_mir = FindMoveResult(bb, mir); if (move_result_mir == NULL) { info->result.location = kLocInvalid; } else { info->result = GetRawDest(move_result_mir); move_result_mir->dalvikInsn.opcode = static_cast(kMirOpNop); } info->num_arg_words = mir->ssa_rep->num_uses; info->args = (info->num_arg_words == 0) ? NULL : static_cast (arena_->Alloc(sizeof(RegLocation) * info->num_arg_words, kArenaAllocMisc)); for (int i = 0; i < info->num_arg_words; i++) { info->args[i] = GetRawSrc(mir, i); } info->opt_flags = mir->optimization_flags; info->type = type; info->is_range = is_range; info->index = mir->dalvikInsn.vB; info->offset = mir->offset; info->mir = mir; return info; } // Allocate a new basic block. BasicBlock* MIRGraph::NewMemBB(BBType block_type, int block_id) { BasicBlock* bb = static_cast(arena_->Alloc(sizeof(BasicBlock), kArenaAllocBB)); bb->block_type = block_type; bb->id = block_id; // TUNING: better estimate of the exit block predecessors? bb->predecessors = new (arena_) GrowableArray(arena_, (block_type == kExitBlock) ? 2048 : 2, kGrowableArrayPredecessors); bb->successor_block_list_type = kNotUsed; block_id_map_.Put(block_id, block_id); return bb; } void MIRGraph::InitializeConstantPropagation() { is_constant_v_ = new (arena_) ArenaBitVector(arena_, GetNumSSARegs(), false); constant_values_ = static_cast(arena_->Alloc(sizeof(int) * GetNumSSARegs(), kArenaAllocDFInfo)); } void MIRGraph::InitializeMethodUses() { // The gate starts by initializing the use counts int num_ssa_regs = GetNumSSARegs(); use_counts_.Resize(num_ssa_regs + 32); raw_use_counts_.Resize(num_ssa_regs + 32); // Initialize list for (int i = 0; i < num_ssa_regs; i++) { use_counts_.Insert(0); raw_use_counts_.Insert(0); } } void MIRGraph::InitializeSSATransformation() { /* Compute the DFS order */ ComputeDFSOrders(); /* Compute the dominator info */ ComputeDominators(); /* Allocate data structures in preparation for SSA conversion */ CompilerInitializeSSAConversion(); /* Find out the "Dalvik reg def x block" relation */ ComputeDefBlockMatrix(); /* Insert phi nodes to dominance frontiers for all variables */ InsertPhiNodes(); /* Rename register names by local defs and phi nodes */ ClearAllVisitedFlags(); DoDFSPreOrderSSARename(GetEntryBlock()); } } // namespace art