lib/Target/WebAssembly/WebAssemblyLateEHPrepare.cpp

   1 //=== WebAssemblyLateEHPrepare.cpp - WebAssembly Exception Preparation -===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 ///
   9 /// \file
  10 /// \brief Does various transformations for exception handling.
  11 ///
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "MCTargetDesc/WebAssemblyMCTargetDesc.h"
  15 #include "WebAssembly.h"
  16 #include "WebAssemblySubtarget.h"
  17 #include "WebAssemblyUtilities.h"
  18 #include "llvm/ADT/SmallSet.h"
  19 #include "llvm/CodeGen/MachineInstrBuilder.h"
  20 #include "llvm/CodeGen/WasmEHFuncInfo.h"
  21 #include "llvm/MC/MCAsmInfo.h"
  22 using namespace llvm;
  23
  24 #define DEBUG_TYPE "wasm-late-eh-prepare"
  25
  26 namespace {
  27 class WebAssemblyLateEHPrepare final : public MachineFunctionPass {
  28   StringRef getPassName() const override {
  29     return "WebAssembly Late Prepare Exception";
  30   }
  31
  32   bool runOnMachineFunction(MachineFunction &MF) override;
  33   bool addCatches(MachineFunction &MF);
  34   bool replaceFuncletReturns(MachineFunction &MF);
  35   bool removeUnnecessaryUnreachables(MachineFunction &MF);
  36   bool addExceptionExtraction(MachineFunction &MF);
  37   bool restoreStackPointer(MachineFunction &MF);
  38
  39 public:
  40   static char ID; // Pass identification, replacement for typeid
  41   WebAssemblyLateEHPrepare() : MachineFunctionPass(ID) {}
  42 };
  43 } // end anonymous namespace
  44
  45 char WebAssemblyLateEHPrepare::ID = 0;
  46 INITIALIZE_PASS(WebAssemblyLateEHPrepare, DEBUG_TYPE,
  47                 "WebAssembly Late Exception Preparation", false, false)
  48
  49 FunctionPass *llvm::createWebAssemblyLateEHPrepare() {
  50   return new WebAssemblyLateEHPrepare();
  51 }
  52
  53 // Returns the nearest EH pad that dominates this instruction. This does not use
  54 // dominator analysis; it just does BFS on its predecessors until arriving at an
  55 // EH pad. This assumes valid EH scopes so the first EH pad it arrives in all
  56 // possible search paths should be the same.
  57 // Returns nullptr in case it does not find any EH pad in the search, or finds
  58 // multiple different EH pads.
  59 static MachineBasicBlock *getMatchingEHPad(MachineInstr *MI) {
  60   MachineFunction *MF = MI->getParent()->getParent();
  61   SmallVector<MachineBasicBlock *, 2> WL;
  62   SmallPtrSet<MachineBasicBlock *, 2> Visited;
  63   WL.push_back(MI->getParent());
  64   MachineBasicBlock *EHPad = nullptr;
  65   while (!WL.empty()) {
  66     MachineBasicBlock *MBB = WL.pop_back_val();
  67     if (Visited.count(MBB))
  68       continue;
  69     Visited.insert(MBB);
  70     if (MBB->isEHPad()) {
  71       if (EHPad && EHPad != MBB)
  72         return nullptr;
  73       EHPad = MBB;
  74       continue;
  75     }
  76     if (MBB == &MF->front())
  77       return nullptr;
  78     WL.append(MBB->pred_begin(), MBB->pred_end());
  79   }
  80   return EHPad;
  81 }
  82
  83 // Erase the specified BBs if the BB does not have any remaining predecessors,
  84 // and also all its dead children.
  85 template <typename Container>
  86 static void eraseDeadBBsAndChildren(const Container &MBBs) {
  87   SmallVector<MachineBasicBlock *, 8> WL(MBBs.begin(), MBBs.end());
  88   while (!WL.empty()) {
  89     MachineBasicBlock *MBB = WL.pop_back_val();
  90     if (!MBB->pred_empty())
  91       continue;
  92     SmallVector<MachineBasicBlock *, 4> Succs(MBB->succ_begin(),
  93                                               MBB->succ_end());
  94     WL.append(MBB->succ_begin(), MBB->succ_end());
  95     for (auto *Succ : Succs)
  96       MBB->removeSuccessor(Succ);
  97     MBB->eraseFromParent();
  98   }
  99 }
 100
 101 bool WebAssemblyLateEHPrepare::runOnMachineFunction(MachineFunction &MF) {
 102   LLVM_DEBUG(dbgs() << "********** Late EH Prepare **********\n"
 103                        "********** Function: "
 104                     << MF.getName() << '\n');
 105
 106   if (MF.getTarget().getMCAsmInfo()->getExceptionHandlingType() !=
 107       ExceptionHandling::Wasm)
 108     return false;
 109
 110   bool Changed = false;
 111   if (MF.getFunction().hasPersonalityFn()) {
 112     Changed |= addCatches(MF);
 113     Changed |= replaceFuncletReturns(MF);
 114   }
 115   Changed |= removeUnnecessaryUnreachables(MF);
 116   if (MF.getFunction().hasPersonalityFn()) {
 117     Changed |= addExceptionExtraction(MF);
 118     Changed |= restoreStackPointer(MF);
 119   }
 120   return Changed;
 121 }
 122
 123 // Add catch instruction to beginning of catchpads and cleanuppads.
 124 bool WebAssemblyLateEHPrepare::addCatches(MachineFunction &MF) {
 125   bool Changed = false;
 126   const auto &TII = *MF.getSubtarget<WebAssemblySubtarget>().getInstrInfo();
 127   MachineRegisterInfo &MRI = MF.getRegInfo();
 128   for (auto &MBB : MF) {
 129     if (MBB.isEHPad()) {
 130       Changed = true;
 131       auto InsertPos = MBB.begin();
 132       if (InsertPos->isEHLabel()) // EH pad starts with an EH label
 133         ++InsertPos;
 134       unsigned DstReg =
 135           MRI.createVirtualRegister(&WebAssembly::EXCEPT_REFRegClass);
 136       BuildMI(MBB, InsertPos, MBB.begin()->getDebugLoc(),
 137               TII.get(WebAssembly::CATCH), DstReg);
 138     }
 139   }
 140   return Changed;
 141 }
 142
 143 bool WebAssemblyLateEHPrepare::replaceFuncletReturns(MachineFunction &MF) {
 144   bool Changed = false;
 145   const auto &TII = *MF.getSubtarget<WebAssemblySubtarget>().getInstrInfo();
 146
 147   for (auto &MBB : MF) {
 148     auto Pos = MBB.getFirstTerminator();
 149     if (Pos == MBB.end())
 150       continue;
 151     MachineInstr *TI = &*Pos;
 152
 153     switch (TI->getOpcode()) {
 154     case WebAssembly::CATCHRET: {
 155       // Replace a catchret with a branch
 156       MachineBasicBlock *TBB = TI->getOperand(0).getMBB();
 157       if (!MBB.isLayoutSuccessor(TBB))
 158         BuildMI(MBB, TI, TI->getDebugLoc(), TII.get(WebAssembly::BR))
 159             .addMBB(TBB);
 160       TI->eraseFromParent();
 161       Changed = true;
 162       break;
 163     }
 164     case WebAssembly::CLEANUPRET:
 165     case WebAssembly::RETHROW_IN_CATCH: {
 166       // Replace a cleanupret/rethrow_in_catch with a rethrow
 167       auto *EHPad = getMatchingEHPad(TI);
 168       auto CatchPos = EHPad->begin();
 169       if (CatchPos->isEHLabel()) // EH pad starts with an EH label
 170         ++CatchPos;
 171       MachineInstr *Catch = &*CatchPos;
 172       unsigned ExnReg = Catch->getOperand(0).getReg();
 173       BuildMI(MBB, TI, TI->getDebugLoc(), TII.get(WebAssembly::RETHROW))
 174           .addReg(ExnReg);
 175       TI->eraseFromParent();
 176       Changed = true;
 177       break;
 178     }
 179     }
 180   }
 181   return Changed;
 182 }
 183
 184 bool WebAssemblyLateEHPrepare::removeUnnecessaryUnreachables(
 185     MachineFunction &MF) {
 186   bool Changed = false;
 187   for (auto &MBB : MF) {
 188     for (auto &MI : MBB) {
 189       if (MI.getOpcode() != WebAssembly::THROW &&
 190           MI.getOpcode() != WebAssembly::RETHROW)
 191         continue;
 192       Changed = true;
 193
 194       // The instruction after the throw should be an unreachable or a branch to
 195       // another BB that should eventually lead to an unreachable. Delete it
 196       // because throw itself is a terminator, and also delete successors if
 197       // any.
 198       MBB.erase(std::next(MI.getIterator()), MBB.end());
 199       SmallVector<MachineBasicBlock *, 8> Succs(MBB.succ_begin(),
 200                                                 MBB.succ_end());
 201       for (auto *Succ : Succs)
 202         if (!Succ->isEHPad())
 203           MBB.removeSuccessor(Succ);
 204       eraseDeadBBsAndChildren(Succs);
 205     }
 206   }
 207
 208   return Changed;
 209 }
 210
 211 // Wasm uses 'br_on_exn' instruction to check the tag of an exception. It takes
 212 // except_ref type object returned by 'catch', and branches to the destination
 213 // if it matches a given tag. We currently use __cpp_exception symbol to
 214 // represent the tag for all C++ exceptions.
 215 //
 216 // block $l (result i32)
 217 //   ...
 218 //   ;; except_ref $e is on the stack at this point
 219 //   br_on_exn $l $e ;; branch to $l with $e's arguments
 220 //   ...
 221 // end
 222 // ;; Here we expect the extracted values are on top of the wasm value stack
 223 // ... Handle exception using values ...
 224 //
 225 // br_on_exn takes an except_ref object and branches if it matches the given
 226 // tag. There can be multiple br_on_exn instructions if we want to match for
 227 // another tag, but for now we only test for __cpp_exception tag, and if it does
 228 // not match, i.e., it is a foreign exception, we rethrow it.
 229 //
 230 // In the destination BB that's the target of br_on_exn, extracted exception
 231 // values (in C++'s case a single i32, which represents an exception pointer)
 232 // are placed on top of the wasm stack. Because we can't model wasm stack in
 233 // LLVM instruction, we use 'extract_exception' pseudo instruction to retrieve
 234 // it. The pseudo instruction will be deleted later.
 235 bool WebAssemblyLateEHPrepare::addExceptionExtraction(MachineFunction &MF) {
 236   const auto &TII = *MF.getSubtarget<WebAssemblySubtarget>().getInstrInfo();
 237   auto *EHInfo = MF.getWasmEHFuncInfo();
 238   SmallVector<MachineInstr *, 16> ExtractInstrs;
 239   SmallVector<MachineInstr *, 8> ToDelete;
 240   for (auto &MBB : MF) {
 241     for (auto &MI : MBB) {
 242       if (MI.getOpcode() == WebAssembly::EXTRACT_EXCEPTION_I32) {
 243         if (MI.getOperand(0).isDead())
 244           ToDelete.push_back(&MI);
 245         else
 246           ExtractInstrs.push_back(&MI);
 247       }
 248     }
 249   }
 250   bool Changed = !ToDelete.empty() || !ExtractInstrs.empty();
 251   for (auto *MI : ToDelete)
 252     MI->eraseFromParent();
 253   if (ExtractInstrs.empty())
 254     return Changed;
 255
 256   // Find terminate pads.
 257   SmallSet<MachineBasicBlock *, 8> TerminatePads;
 258   for (auto &MBB : MF) {
 259     for (auto &MI : MBB) {
 260       if (MI.isCall()) {
 261         const MachineOperand &CalleeOp = MI.getOperand(0);
 262         if (CalleeOp.isGlobal() && CalleeOp.getGlobal()->getName() ==
 263                                        WebAssembly::ClangCallTerminateFn)
 264           TerminatePads.insert(getMatchingEHPad(&MI));
 265       }
 266     }
 267   }
 268
 269   for (auto *Extract : ExtractInstrs) {
 270     MachineBasicBlock *EHPad = getMatchingEHPad(Extract);
 271     assert(EHPad && "No matching EH pad for extract_exception");
 272     auto CatchPos = EHPad->begin();
 273     if (CatchPos->isEHLabel()) // EH pad starts with an EH label
 274       ++CatchPos;
 275     MachineInstr *Catch = &*CatchPos;
 276
 277     if (Catch->getNextNode() != Extract)
 278       EHPad->insert(Catch->getNextNode(), Extract->removeFromParent());
 279
 280     // - Before:
 281     // ehpad:
 282     //   %exnref:except_ref = catch
 283     //   %exn:i32 = extract_exception
 284     //   ... use exn ...
 285     //
 286     // - After:
 287     // ehpad:
 288     //   %exnref:except_ref = catch
 289     //   br_on_exn %thenbb, $__cpp_exception, %exnref
 290     //   br %elsebb
 291     // elsebb:
 292     //   rethrow
 293     // thenbb:
 294     //   %exn:i32 = extract_exception
 295     //   ... use exn ...
 296     unsigned ExnReg = Catch->getOperand(0).getReg();
 297     auto *ThenMBB = MF.CreateMachineBasicBlock();
 298     auto *ElseMBB = MF.CreateMachineBasicBlock();
 299     MF.insert(std::next(MachineFunction::iterator(EHPad)), ElseMBB);
 300     MF.insert(std::next(MachineFunction::iterator(ElseMBB)), ThenMBB);
 301     ThenMBB->splice(ThenMBB->end(), EHPad, Extract, EHPad->end());
 302     ThenMBB->transferSuccessors(EHPad);
 303     EHPad->addSuccessor(ThenMBB);
 304     EHPad->addSuccessor(ElseMBB);
 305
 306     DebugLoc DL = Extract->getDebugLoc();
 307     const char *CPPExnSymbol = MF.createExternalSymbolName("__cpp_exception");
 308     BuildMI(EHPad, DL, TII.get(WebAssembly::BR_ON_EXN))
 309         .addMBB(ThenMBB)
 310         .addExternalSymbol(CPPExnSymbol)
 311         .addReg(ExnReg);
 312     BuildMI(EHPad, DL, TII.get(WebAssembly::BR)).addMBB(ElseMBB);
 313
 314     // When this is a terminate pad with __clang_call_terminate() call, we don't
 315     // rethrow it anymore and call __clang_call_terminate() with a nullptr
 316     // argument, which will call std::terminate().
 317     //
 318     // - Before:
 319     // ehpad:
 320     //   %exnref:except_ref = catch
 321     //   %exn:i32 = extract_exception
 322     //   call @__clang_call_terminate(%exn)
 323     //   unreachable
 324     //
 325     // - After:
 326     // ehpad:
 327     //   %exnref:except_ref = catch
 328     //   br_on_exn %thenbb, $__cpp_exception, %exnref
 329     //   br %elsebb
 330     // elsebb:
 331     //   call @__clang_call_terminate(0)
 332     //   unreachable
 333     // thenbb:
 334     //   %exn:i32 = extract_exception
 335     //   call @__clang_call_terminate(%exn)
 336     //   unreachable
 337     if (TerminatePads.count(EHPad)) {
 338       Function *ClangCallTerminateFn =
 339           MF.getFunction().getParent()->getFunction(
 340               WebAssembly::ClangCallTerminateFn);
 341       assert(ClangCallTerminateFn &&
 342              "There is no __clang_call_terminate() function");
 343       BuildMI(ElseMBB, DL, TII.get(WebAssembly::CALL_VOID))
 344           .addGlobalAddress(ClangCallTerminateFn)
 345           .addImm(0);
 346       BuildMI(ElseMBB, DL, TII.get(WebAssembly::UNREACHABLE));
 347
 348     } else {
 349       BuildMI(ElseMBB, DL, TII.get(WebAssembly::RETHROW)).addReg(ExnReg);
 350       if (EHInfo->hasEHPadUnwindDest(EHPad))
 351         ElseMBB->addSuccessor(EHInfo->getEHPadUnwindDest(EHPad));
 352     }
 353   }
 354
 355   return true;
 356 }
 357
 358 // After the stack is unwound due to a thrown exception, the __stack_pointer
 359 // global can point to an invalid address. This inserts instructions that
 360 // restore __stack_pointer global.
 361 bool WebAssemblyLateEHPrepare::restoreStackPointer(MachineFunction &MF) {
 362   const auto *FrameLowering = static_cast<const WebAssemblyFrameLowering *>(
 363       MF.getSubtarget().getFrameLowering());
 364   if (!FrameLowering->needsPrologForEH(MF))
 365     return false;
 366   bool Changed = false;
 367
 368   for (auto &MBB : MF) {
 369     if (!MBB.isEHPad())
 370       continue;
 371     Changed = true;
 372
 373     // Insert __stack_pointer restoring instructions at the beginning of each EH
 374     // pad, after the catch instruction. Here it is safe to assume that SP32
 375     // holds the latest value of __stack_pointer, because the only exception for
 376     // this case is when a function uses the red zone, but that only happens
 377     // with leaf functions, and we don't restore __stack_pointer in leaf
 378     // functions anyway.
 379     auto InsertPos = MBB.begin();
 380     if (InsertPos->isEHLabel()) // EH pad starts with an EH label
 381       ++InsertPos;
 382     if (InsertPos->getOpcode() == WebAssembly::CATCH)
 383       ++InsertPos;
 384     FrameLowering->writeSPToGlobal(WebAssembly::SP32, MF, MBB, InsertPos,
 385                                    MBB.begin()->getDebugLoc());
 386   }
 387   return Changed;
 388 }