1 //===-- WebAssemblyCFGStackify.cpp - CFG Stackification -------------------===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 /// 10 /// \file 11 /// This file implements a CFG stacking pass. 12 /// 13 /// This pass inserts BLOCK and LOOP markers to mark the start of scopes, since 14 /// scope boundaries serve as the labels for WebAssembly's control transfers. 15 /// 16 /// This is sufficient to convert arbitrary CFGs into a form that works on 17 /// WebAssembly, provided that all loops are single-entry. 18 /// 19 //===----------------------------------------------------------------------===// 20 21 #include "MCTargetDesc/WebAssemblyMCTargetDesc.h" 22 #include "WebAssembly.h" 23 #include "WebAssemblyMachineFunctionInfo.h" 24 #include "WebAssemblySubtarget.h" 25 #include "WebAssemblyUtilities.h" 26 #include "llvm/CodeGen/MachineDominators.h" 27 #include "llvm/CodeGen/MachineFunction.h" 28 #include "llvm/CodeGen/MachineInstrBuilder.h" 29 #include "llvm/CodeGen/MachineLoopInfo.h" 30 #include "llvm/CodeGen/MachineRegisterInfo.h" 31 #include "llvm/CodeGen/Passes.h" 32 #include "llvm/Support/Debug.h" 33 #include "llvm/Support/raw_ostream.h" 34 using namespace llvm; 35 36 #define DEBUG_TYPE "wasm-cfg-stackify" 37 38 namespace { 39 class WebAssemblyCFGStackify final : public MachineFunctionPass { 40 StringRef getPassName() const override { return "WebAssembly CFG Stackify"; } 41 42 void getAnalysisUsage(AnalysisUsage &AU) const override { 43 AU.setPreservesCFG(); 44 AU.addRequired<MachineDominatorTree>(); 45 AU.addPreserved<MachineDominatorTree>(); 46 AU.addRequired<MachineLoopInfo>(); 47 AU.addPreserved<MachineLoopInfo>(); 48 MachineFunctionPass::getAnalysisUsage(AU); 49 } 50 51 bool runOnMachineFunction(MachineFunction &MF) override; 52 53 public: 54 static char ID; // Pass identification, replacement for typeid 55 WebAssemblyCFGStackify() : MachineFunctionPass(ID) {} 56 }; 57 } // end anonymous namespace 58 59 char WebAssemblyCFGStackify::ID = 0; 60 INITIALIZE_PASS(WebAssemblyCFGStackify, DEBUG_TYPE, 61 "Insert BLOCK and LOOP markers for WebAssembly scopes", 62 false, false) 63 64 FunctionPass *llvm::createWebAssemblyCFGStackify() { 65 return new WebAssemblyCFGStackify(); 66 } 67 68 /// Test whether Pred has any terminators explicitly branching to MBB, as 69 /// opposed to falling through. Note that it's possible (eg. in unoptimized 70 /// code) for a branch instruction to both branch to a block and fallthrough 71 /// to it, so we check the actual branch operands to see if there are any 72 /// explicit mentions. 73 static bool ExplicitlyBranchesTo(MachineBasicBlock *Pred, 74 MachineBasicBlock *MBB) { 75 for (MachineInstr &MI : Pred->terminators()) 76 for (MachineOperand &MO : MI.explicit_operands()) 77 if (MO.isMBB() && MO.getMBB() == MBB) 78 return true; 79 return false; 80 } 81 82 /// Insert a BLOCK marker for branches to MBB (if needed). 83 static void PlaceBlockMarker( 84 MachineBasicBlock &MBB, MachineFunction &MF, 85 SmallVectorImpl<MachineBasicBlock *> &ScopeTops, 86 DenseMap<const MachineInstr *, MachineInstr *> &BlockTops, 87 DenseMap<const MachineInstr *, MachineInstr *> &LoopTops, 88 const WebAssemblyInstrInfo &TII, 89 const MachineLoopInfo &MLI, 90 MachineDominatorTree &MDT, 91 WebAssemblyFunctionInfo &MFI) { 92 // First compute the nearest common dominator of all forward non-fallthrough 93 // predecessors so that we minimize the time that the BLOCK is on the stack, 94 // which reduces overall stack height. 95 MachineBasicBlock *Header = nullptr; 96 bool IsBranchedTo = false; 97 int MBBNumber = MBB.getNumber(); 98 for (MachineBasicBlock *Pred : MBB.predecessors()) 99 if (Pred->getNumber() < MBBNumber) { 100 Header = Header ? MDT.findNearestCommonDominator(Header, Pred) : Pred; 101 if (ExplicitlyBranchesTo(Pred, &MBB)) 102 IsBranchedTo = true; 103 } 104 if (!Header) 105 return; 106 if (!IsBranchedTo) 107 return; 108 109 assert(&MBB != &MF.front() && "Header blocks shouldn't have predecessors"); 110 MachineBasicBlock *LayoutPred = &*std::prev(MachineFunction::iterator(&MBB)); 111 112 // If the nearest common dominator is inside a more deeply nested context, 113 // walk out to the nearest scope which isn't more deeply nested. 114 for (MachineFunction::iterator I(LayoutPred), E(Header); I != E; --I) { 115 if (MachineBasicBlock *ScopeTop = ScopeTops[I->getNumber()]) { 116 if (ScopeTop->getNumber() > Header->getNumber()) { 117 // Skip over an intervening scope. 118 I = std::next(MachineFunction::iterator(ScopeTop)); 119 } else { 120 // We found a scope level at an appropriate depth. 121 Header = ScopeTop; 122 break; 123 } 124 } 125 } 126 127 // Decide where in Header to put the BLOCK. 128 MachineBasicBlock::iterator InsertPos; 129 MachineLoop *HeaderLoop = MLI.getLoopFor(Header); 130 if (HeaderLoop && 131 MBB.getNumber() > WebAssembly::getBottom(HeaderLoop)->getNumber()) { 132 // Header is the header of a loop that does not lexically contain MBB, so 133 // the BLOCK needs to be above the LOOP, after any END constructs. 134 InsertPos = Header->begin(); 135 while (InsertPos->getOpcode() == WebAssembly::END_BLOCK || 136 InsertPos->getOpcode() == WebAssembly::END_LOOP) 137 ++InsertPos; 138 } else { 139 // Otherwise, insert the BLOCK as late in Header as we can, but before the 140 // beginning of the local expression tree and any nested BLOCKs. 141 InsertPos = Header->getFirstTerminator(); 142 while (InsertPos != Header->begin() && 143 WebAssembly::isChild(*std::prev(InsertPos), MFI) && 144 std::prev(InsertPos)->getOpcode() != WebAssembly::LOOP && 145 std::prev(InsertPos)->getOpcode() != WebAssembly::END_BLOCK && 146 std::prev(InsertPos)->getOpcode() != WebAssembly::END_LOOP) 147 --InsertPos; 148 } 149 150 // Add the BLOCK. 151 MachineInstr *Begin = 152 BuildMI(*Header, InsertPos, Header->findDebugLoc(InsertPos), 153 TII.get(WebAssembly::BLOCK)) 154 .addImm(int64_t(WebAssembly::ExprType::Void)); 155 156 // Mark the end of the block. 157 InsertPos = MBB.begin(); 158 while (InsertPos != MBB.end() && 159 InsertPos->getOpcode() == WebAssembly::END_LOOP && 160 LoopTops[&*InsertPos]->getParent()->getNumber() >= Header->getNumber()) 161 ++InsertPos; 162 MachineInstr *End = BuildMI(MBB, InsertPos, MBB.findPrevDebugLoc(InsertPos), 163 TII.get(WebAssembly::END_BLOCK)); 164 BlockTops[End] = Begin; 165 166 // Track the farthest-spanning scope that ends at this point. 167 int Number = MBB.getNumber(); 168 if (!ScopeTops[Number] || 169 ScopeTops[Number]->getNumber() > Header->getNumber()) 170 ScopeTops[Number] = Header; 171 } 172 173 /// Insert a LOOP marker for a loop starting at MBB (if it's a loop header). 174 static void PlaceLoopMarker( 175 MachineBasicBlock &MBB, MachineFunction &MF, 176 SmallVectorImpl<MachineBasicBlock *> &ScopeTops, 177 DenseMap<const MachineInstr *, MachineInstr *> &LoopTops, 178 const WebAssemblyInstrInfo &TII, const MachineLoopInfo &MLI) { 179 MachineLoop *Loop = MLI.getLoopFor(&MBB); 180 if (!Loop || Loop->getHeader() != &MBB) 181 return; 182 183 // The operand of a LOOP is the first block after the loop. If the loop is the 184 // bottom of the function, insert a dummy block at the end. 185 MachineBasicBlock *Bottom = WebAssembly::getBottom(Loop); 186 auto Iter = std::next(MachineFunction::iterator(Bottom)); 187 if (Iter == MF.end()) { 188 MachineBasicBlock *Label = MF.CreateMachineBasicBlock(); 189 // Give it a fake predecessor so that AsmPrinter prints its label. 190 Label->addSuccessor(Label); 191 MF.push_back(Label); 192 Iter = std::next(MachineFunction::iterator(Bottom)); 193 } 194 MachineBasicBlock *AfterLoop = &*Iter; 195 196 // Mark the beginning of the loop (after the end of any existing loop that 197 // ends here). 198 auto InsertPos = MBB.begin(); 199 while (InsertPos != MBB.end() && 200 InsertPos->getOpcode() == WebAssembly::END_LOOP) 201 ++InsertPos; 202 MachineInstr *Begin = BuildMI(MBB, InsertPos, MBB.findDebugLoc(InsertPos), 203 TII.get(WebAssembly::LOOP)) 204 .addImm(int64_t(WebAssembly::ExprType::Void)); 205 206 // Mark the end of the loop (using arbitrary debug location that branched 207 // to the loop end as its location). 208 DebugLoc EndDL = (*AfterLoop->pred_rbegin())->findBranchDebugLoc(); 209 MachineInstr *End = BuildMI(*AfterLoop, AfterLoop->begin(), EndDL, 210 TII.get(WebAssembly::END_LOOP)); 211 LoopTops[End] = Begin; 212 213 assert((!ScopeTops[AfterLoop->getNumber()] || 214 ScopeTops[AfterLoop->getNumber()]->getNumber() < MBB.getNumber()) && 215 "With block sorting the outermost loop for a block should be first."); 216 if (!ScopeTops[AfterLoop->getNumber()]) 217 ScopeTops[AfterLoop->getNumber()] = &MBB; 218 } 219 220 static unsigned 221 GetDepth(const SmallVectorImpl<const MachineBasicBlock *> &Stack, 222 const MachineBasicBlock *MBB) { 223 unsigned Depth = 0; 224 for (auto X : reverse(Stack)) { 225 if (X == MBB) 226 break; 227 ++Depth; 228 } 229 assert(Depth < Stack.size() && "Branch destination should be in scope"); 230 return Depth; 231 } 232 233 /// In normal assembly languages, when the end of a function is unreachable, 234 /// because the function ends in an infinite loop or a noreturn call or similar, 235 /// it isn't necessary to worry about the function return type at the end of 236 /// the function, because it's never reached. However, in WebAssembly, blocks 237 /// that end at the function end need to have a return type signature that 238 /// matches the function signature, even though it's unreachable. This function 239 /// checks for such cases and fixes up the signatures. 240 static void FixEndsAtEndOfFunction( 241 MachineFunction &MF, 242 const WebAssemblyFunctionInfo &MFI, 243 DenseMap<const MachineInstr *, MachineInstr *> &BlockTops, 244 DenseMap<const MachineInstr *, MachineInstr *> &LoopTops) { 245 assert(MFI.getResults().size() <= 1); 246 247 if (MFI.getResults().empty()) 248 return; 249 250 WebAssembly::ExprType retType; 251 switch (MFI.getResults().front().SimpleTy) { 252 case MVT::i32: retType = WebAssembly::ExprType::I32; break; 253 case MVT::i64: retType = WebAssembly::ExprType::I64; break; 254 case MVT::f32: retType = WebAssembly::ExprType::F32; break; 255 case MVT::f64: retType = WebAssembly::ExprType::F64; break; 256 case MVT::v16i8: 257 case MVT::v8i16: 258 case MVT::v4i32: 259 case MVT::v2i64: 260 case MVT::v4f32: 261 case MVT::v2f64: 262 retType = WebAssembly::ExprType::V128; 263 break; 264 case MVT::ExceptRef: retType = WebAssembly::ExprType::ExceptRef; break; 265 default: llvm_unreachable("unexpected return type"); 266 } 267 268 for (MachineBasicBlock &MBB : reverse(MF)) { 269 for (MachineInstr &MI : reverse(MBB)) { 270 if (MI.isPosition() || MI.isDebugInstr()) 271 continue; 272 if (MI.getOpcode() == WebAssembly::END_BLOCK) { 273 BlockTops[&MI]->getOperand(0).setImm(int32_t(retType)); 274 continue; 275 } 276 if (MI.getOpcode() == WebAssembly::END_LOOP) { 277 LoopTops[&MI]->getOperand(0).setImm(int32_t(retType)); 278 continue; 279 } 280 // Something other than an `end`. We're done. 281 return; 282 } 283 } 284 } 285 286 // WebAssembly functions end with an end instruction, as if the function body 287 // were a block. 288 static void AppendEndToFunction( 289 MachineFunction &MF, 290 const WebAssemblyInstrInfo &TII) { 291 BuildMI(MF.back(), MF.back().end(), 292 MF.back().findPrevDebugLoc(MF.back().end()), 293 TII.get(WebAssembly::END_FUNCTION)); 294 } 295 296 /// Insert LOOP and BLOCK markers at appropriate places. 297 static void PlaceMarkers(MachineFunction &MF, const MachineLoopInfo &MLI, 298 const WebAssemblyInstrInfo &TII, 299 MachineDominatorTree &MDT, 300 WebAssemblyFunctionInfo &MFI) { 301 // For each block whose label represents the end of a scope, record the block 302 // which holds the beginning of the scope. This will allow us to quickly skip 303 // over scoped regions when walking blocks. We allocate one more than the 304 // number of blocks in the function to accommodate for the possible fake block 305 // we may insert at the end. 306 SmallVector<MachineBasicBlock *, 8> ScopeTops(MF.getNumBlockIDs() + 1); 307 308 // For each LOOP_END, the corresponding LOOP. 309 DenseMap<const MachineInstr *, MachineInstr *> LoopTops; 310 311 // For each END_BLOCK, the corresponding BLOCK. 312 DenseMap<const MachineInstr *, MachineInstr *> BlockTops; 313 314 for (auto &MBB : MF) { 315 // Place the LOOP for MBB if MBB is the header of a loop. 316 PlaceLoopMarker(MBB, MF, ScopeTops, LoopTops, TII, MLI); 317 318 // Place the BLOCK for MBB if MBB is branched to from above. 319 PlaceBlockMarker(MBB, MF, ScopeTops, BlockTops, LoopTops, TII, MLI, MDT, MFI); 320 } 321 322 // Now rewrite references to basic blocks to be depth immediates. 323 SmallVector<const MachineBasicBlock *, 8> Stack; 324 for (auto &MBB : reverse(MF)) { 325 for (auto &MI : reverse(MBB)) { 326 switch (MI.getOpcode()) { 327 case WebAssembly::BLOCK: 328 assert(ScopeTops[Stack.back()->getNumber()]->getNumber() <= MBB.getNumber() && 329 "Block should be balanced"); 330 Stack.pop_back(); 331 break; 332 case WebAssembly::LOOP: 333 assert(Stack.back() == &MBB && "Loop top should be balanced"); 334 Stack.pop_back(); 335 break; 336 case WebAssembly::END_BLOCK: 337 Stack.push_back(&MBB); 338 break; 339 case WebAssembly::END_LOOP: 340 Stack.push_back(LoopTops[&MI]->getParent()); 341 break; 342 default: 343 if (MI.isTerminator()) { 344 // Rewrite MBB operands to be depth immediates. 345 SmallVector<MachineOperand, 4> Ops(MI.operands()); 346 while (MI.getNumOperands() > 0) 347 MI.RemoveOperand(MI.getNumOperands() - 1); 348 for (auto MO : Ops) { 349 if (MO.isMBB()) 350 MO = MachineOperand::CreateImm(GetDepth(Stack, MO.getMBB())); 351 MI.addOperand(MF, MO); 352 } 353 } 354 break; 355 } 356 } 357 } 358 assert(Stack.empty() && "Control flow should be balanced"); 359 360 // Fix up block/loop signatures at the end of the function to conform to 361 // WebAssembly's rules. 362 FixEndsAtEndOfFunction(MF, MFI, BlockTops, LoopTops); 363 364 // Add an end instruction at the end of the function body. 365 AppendEndToFunction(MF, TII); 366 } 367 368 bool WebAssemblyCFGStackify::runOnMachineFunction(MachineFunction &MF) { 369 LLVM_DEBUG(dbgs() << "********** CFG Stackifying **********\n" 370 "********** Function: " 371 << MF.getName() << '\n'); 372 373 const auto &MLI = getAnalysis<MachineLoopInfo>(); 374 auto &MDT = getAnalysis<MachineDominatorTree>(); 375 // Liveness is not tracked for VALUE_STACK physreg. 376 const auto &TII = *MF.getSubtarget<WebAssemblySubtarget>().getInstrInfo(); 377 WebAssemblyFunctionInfo &MFI = *MF.getInfo<WebAssemblyFunctionInfo>(); 378 MF.getRegInfo().invalidateLiveness(); 379 380 // Place the BLOCK and LOOP markers to indicate the beginnings of scopes. 381 PlaceMarkers(MF, MLI, TII, MDT, MFI); 382 383 return true; 384 } 385