[WebAssembly] Support clang -fwasm-exceptions for bitcode

[aheejin]

Summary

This supports bitcode compilation using clang -fwasm-exceptions.

---

The current situation:

Currently the backend requires two options for Wasm EH:
-wasm-enable-eh and -exception-model=wasm. Wasm SjLj requires two
options as well: -wasm-enable-sjlj and -exception-model=wasm. When
using Wasm EH via Emscripten, you only need to pass -fwasm-exceptions,
and these options will be added within the clang driver. This
description will focus on the case of Wasm EH going forward, but Wasm
SjLj's case is similar.

When you pass -fwasm-exceptions to emcc and clang driver, the clang
driver adds these options to the command line that calls the clang
frontend (clang -cc1): -mllvm -wasm-enable-eh and
-exception-model=wasm. -wasm-enable-eh is prefixed with -mllvm, so
it is passed as is to the backend. But -exception-model is parsed and
processed within the clang frontend and stored in LangOptions class.
This info is later transferred to TargetOptions class, and then
eventually passed to MCAsmInfo class. All LLVM code queries this
MCAsmInfo to get the exception model.

All LLVM code queries this MCAsmInfo to get the exception model. (
(e.g., here)

---

Problem:

The problem is the whole LangOptions processing is bypassed when
compiling bitcode, so the information transfer of LangOptions ->
TargetOptions -> MCAsmInfo does not happen. They are all set to
ExceptionHandling::None, which is the default value.

---

What other targets do, and why we can't do the same:

Other targets support bitcode compilation by the clang driver, but they
can do that by using different triples. For example, X86 target supports
multiple triples, each of which has its own subclass of MCAsmInfo, so
it can hardcode the appropriate exception model within those subclasses'
constructors. But we don't have separate triples for each exception
mode: none, emscripten, and wasm.

---

What this CL does:

If we can figure out whether -wasm-enable-eh is passed to the backend,
we can programatically set the exception model from the backend, rather
than requiring it to be passed.

So we check WasmEnableEH and WasmEnableSjLj variables, which are
cl::opt for -wasm-enable-eh and -wasm-enable-sjlj, in
WebAssemblyMCAsmInfo constructor, and if either of them is set, we set
MCAsmInfo.ExceptionType to Wasm. TargetOptions cannot be updated
there, so we make sure they are the same later.

Fixes https://github.com/emscripten-core/emscripten/issues/15712.

[aheejin]

Revert a function name change

[dschuff]

llvm/lib/Target/WebAssembly/WebAssemblyTargetMachine.cpp

37–38

We could put these declarations in WebAssemblyUtilities.h; then they wouldn't have to be duplicated here and in WebAssemblyMCAsmInfo.cpp

353

[aheejin]

Address comments

[aheejin]

llvm/lib/Target/WebAssembly/WebAssemblyTargetMachine.cpp

37–38

Done. But as a result they are not within WebAssembly namespace, which I think is actually better because we have been cluttering the llvm namespace so far. And I needed to attach WebAssembly:: to all the usages of these variables in other places. In this file I just used using WebAssembly::***, because there are too many usages of these options.

[aheejin]

Remove an unnecessary include

[dschuff]

llvm/lib/Target/WebAssembly/WebAssemblyTargetMachine.cpp

37–38

👍

clang/test/CodeGen/WebAssembly/wasm-eh.ll

+; REQUIRES: webassembly-registered-target
+; RUN: %clang %s -target wasm32-unknown-unknown -fwasm-exceptions -c -S -o - | FileCheck %s
+
+; This tests whether clang driver can take -fwasm-exceptions and compile bitcode
+; files using Wasm EH.
+
+; CHECK-LABEL: test
+; CHECK: try
+; CHECK:   call foo
+; CHECK: catch __cpp_exception
+; CHECK: end
+define void @test() personality i8* bitcast (i32 (...)* @__gxx_wasm_personality_v0 to i8*) {
+entry:
+  invoke void @foo()
+          to label %try.cont unwind label %catch.dispatch
+
+catch.dispatch:                                   ; preds = %entry
+  %0 = catchswitch within none [label %catch.start] unwind to caller
+
+catch.start:                                      ; preds = %catch.dispatch
+  %1 = catchpad within %0 [i8* null]
+  %2 = call i8* @llvm.wasm.get.exception(token %1)
+  %3 = call i32 @llvm.wasm.get.ehselector(token %1)
+  %4 = call i8* @__cxa_begin_catch(i8* %2) #2 [ "funclet"(token %1) ]
+  call void @__cxa_end_catch() [ "funclet"(token %1) ]
+  catchret from %1 to label %try.cont
+
+try.cont:                                         ; preds = %entry, %catch.start
+  ret void
+}
+
+declare void @foo()
+declare i32 @__gxx_wasm_personality_v0(...)
+declare i8* @llvm.wasm.get.exception(token)
+declare i32 @llvm.wasm.get.ehselector(token)
+declare i8* @__cxa_begin_catch(i8*)
+declare void @__cxa_end_catch()
+

llvm/lib/Target/WebAssembly/MCTargetDesc/WebAssemblyMCAsmInfo.cpp

 //===-- WebAssemblyMCAsmInfo.cpp - WebAssembly asm properties -------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 ///
 /// \file
 /// This file contains the declarations of the WebAssemblyMCAsmInfo
 /// properties.
 ///
 //===----------------------------------------------------------------------===//
 
 #include "WebAssemblyMCAsmInfo.h"
+#include "Utils/WebAssemblyUtilities.h"
 #include "llvm/ADT/Triple.h"
+#include "llvm/Support/CommandLine.h"
 
 using namespace llvm;
 
 #define DEBUG_TYPE "wasm-mc-asm-info"
 
 WebAssemblyMCAsmInfo::~WebAssemblyMCAsmInfo() = default; // anchor.
 
 WebAssemblyMCAsmInfo::WebAssemblyMCAsmInfo(const Triple &T,
WebAssemblyMCAsmInfo::WebAssemblyMCAsmInfo(const Triple &T,
   Data64bitsDirective = "\t.int64\t";
 
   AlignmentIsInBytes = false;
   COMMDirectiveAlignmentIsInBytes = false;
   LCOMMDirectiveAlignmentType = LCOMM::Log2Alignment;
 
   SupportsDebugInformation = true;
 
+  // When compilation is done on a cpp file by clang, the exception model info
+  // is stored in LangOptions, which is later used to set the info in
+  // TargetOptions and then MCAsmInfo in LLVMTargetMachine::initAsmInfo(). But
+  // this process does not happen when compiling bitcode directly with clang, so
+  // we make sure this info is set correctly.
+  if (WebAssembly::WasmEnableEH || WebAssembly::WasmEnableSjLj)
+    ExceptionsType = ExceptionHandling::Wasm;
+
   // TODO: UseIntegratedAssembler?
 }

llvm/lib/Target/WebAssembly/Utils/WebAssemblyUtilities.h

 /// utility functions.
 ///
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_LIB_TARGET_WEBASSEMBLY_UTILS_WEBASSEMBLYUTILITIES_H
 #define LLVM_LIB_TARGET_WEBASSEMBLY_UTILS_WEBASSEMBLYUTILITIES_H
 
 #include "llvm/IR/DerivedTypes.h"
+#include "llvm/Support/CommandLine.h"
 
 namespace llvm {
 
 class MachineBasicBlock;
 class MachineInstr;
 class MachineOperand;
 class MCContext;
 class MCSymbolWasm;

 }
 inline bool isRefType(const Type *Ty) {
   return isFuncrefType(Ty) || isExternrefType(Ty);
 }
 
 bool isChild(const MachineInstr &MI, const WebAssemblyFunctionInfo &MFI);
 bool mayThrow(const MachineInstr &MI);
 
+// Exception handling / setjmp-longjmp handling command-line options
+extern cl::opt<bool> WasmEnableEmEH;   // asm.js-style EH
+extern cl::opt<bool> WasmEnableEmSjLj; // asm.js-style SjLJ
+extern cl::opt<bool> WasmEnableEH;     // EH using Wasm EH instructions
+extern cl::opt<bool> WasmEnableSjLj;   // SjLj using Wasm EH instructions
+
 // Exception-related function names
 extern const char *const ClangCallTerminateFn;
 extern const char *const CxaBeginCatchFn;
 extern const char *const CxaRethrowFn;
 extern const char *const StdTerminateFn;
 extern const char *const PersonalityWrapperFn;
 
 /// Returns the operand number of a callee, assuming the argument is a call

llvm/lib/Target/WebAssembly/Utils/WebAssemblyUtilities.cpp

 
 #include "WebAssemblyUtilities.h"
 #include "WebAssemblyMachineFunctionInfo.h"
 #include "llvm/CodeGen/MachineInstr.h"
 #include "llvm/CodeGen/MachineLoopInfo.h"
 #include "llvm/MC/MCContext.h"
 using namespace llvm;
 
+// Exception handling & setjmp-longjmp handling related options. These are
+// defined here to be shared between WebAssembly and its subdirectories.
+
+// Emscripten's asm.js-style exception handling
+cl::opt<bool> WebAssembly::WasmEnableEmEH(
+    "enable-emscripten-cxx-exceptions",
+    cl::desc("WebAssembly Emscripten-style exception handling"),
+    cl::init(false));
+// Emscripten's asm.js-style setjmp/longjmp handling
+cl::opt<bool> WebAssembly::WasmEnableEmSjLj(
+    "enable-emscripten-sjlj",
+    cl::desc("WebAssembly Emscripten-style setjmp/longjmp handling"),
+    cl::init(false));
+// Exception handling using wasm EH instructions
+cl::opt<bool>
+    WebAssembly::WasmEnableEH("wasm-enable-eh",
+                              cl::desc("WebAssembly exception handling"),
+                              cl::init(false));
+// setjmp/longjmp handling using wasm EH instrutions
+cl::opt<bool>
+    WebAssembly::WasmEnableSjLj("wasm-enable-sjlj",
+                                cl::desc("WebAssembly setjmp/longjmp handling"),
+                                cl::init(false));
+
+// Function names in libc++abi and libunwind
 const char *const WebAssembly::CxaBeginCatchFn = "__cxa_begin_catch";
 const char *const WebAssembly::CxaRethrowFn = "__cxa_rethrow";
 const char *const WebAssembly::StdTerminateFn = "_ZSt9terminatev";
 const char *const WebAssembly::PersonalityWrapperFn =
     "_Unwind_Wasm_CallPersonality";
 
 /// Test whether MI is a child of some other node in an expression tree.
 bool WebAssembly::isChild(const MachineInstr &MI,

llvm/lib/Target/WebAssembly/WebAssemblyAsmPrinter.cpp

 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
 
 using namespace llvm;
 
 #define DEBUG_TYPE "asm-printer"
 
 extern cl::opt<bool> WasmKeepRegisters;
-extern cl::opt<bool> WasmEnableEmEH;
-extern cl::opt<bool> WasmEnableEmSjLj;
 
 //===----------------------------------------------------------------------===//
 // Helpers.
 //===----------------------------------------------------------------------===//
 
 MVT WebAssemblyAsmPrinter::getRegType(unsigned RegNo) const {
   const TargetRegisterInfo *TRI = Subtarget->getRegisterInfo();
   const TargetRegisterClass *TRC = MRI->getRegClass(RegNo);
if (F.isDeclarationForLinker()) {
       SmallVector<MVT, 4> Params;
       computeSignatureVTs(F.getFunctionType(), &F, F, TM, Params, Results);
       // At this point these MCSymbols may or may not have been created already
       // and thus also contain a signature, but we need to get the signature
       // anyway here in case it is an invoke that has not yet been created. We
       // will discard it later if it turns out not to be necessary.
       auto Signature = signatureFromMVTs(Results, Params);
       bool InvokeDetected = false;
-      auto *Sym = getMCSymbolForFunction(&F, WasmEnableEmEH || WasmEnableEmSjLj,
-                                         Signature.get(), InvokeDetected);
+      auto *Sym = getMCSymbolForFunction(
+          &F, WebAssembly::WasmEnableEmEH || WebAssembly::WasmEnableEmSjLj,
 
       // Multiple functions can be mapped to the same invoke symbol. For
       // example, two IR functions '__invoke_void_i8*' and '__invoke_void_i32'
       // are both mapped to '__invoke_vi'. We keep them in a set once we emit an
       // Emscripten EH symbol so we don't emit the same symbol twice.
       if (InvokeDetected && !InvokeSymbols.insert(Sym).second)
         continue;
 

llvm/lib/Target/WebAssembly/WebAssemblyLowerEmscriptenEHSjLj.cpp

 ///   %val = load 'val' field from __WasmLongjmpArgs
 ///   %label = testSetjmp(mem[%env], setjmpTable, setjmpTableSize);
 ///   if (%label == 0)
 ///     __wasm_longjmp(%env, %val)
 ///   catchret to %setjmp.dispatch
 ///
 ///===----------------------------------------------------------------------===//
 
+#include "Utils/WebAssemblyUtilities.h"
 #include "WebAssembly.h"
 #include "WebAssemblyTargetMachine.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/CodeGen/TargetPassConfig.h"
 #include "llvm/CodeGen/WasmEHFuncInfo.h"
 #include "llvm/IR/DebugInfoMetadata.h"
 #include "llvm/IR/Dominators.h"
 #include "llvm/IR/IRBuilder.h"
 #include "llvm/IR/IntrinsicsWebAssembly.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
 #include "llvm/Transforms/Utils/SSAUpdater.h"
 #include "llvm/Transforms/Utils/SSAUpdaterBulk.h"
 
 using namespace llvm;
 
 #define DEBUG_TYPE "wasm-lower-em-ehsjlj"
 
-// Emscripten's asm.js-style exception handling
-extern cl::opt<bool> WasmEnableEmEH;
-// Emscripten's asm.js-style setjmp/longjmp handling
-extern cl::opt<bool> WasmEnableEmSjLj;
-// Wasm setjmp/longjmp handling using wasm EH instructions
-extern cl::opt<bool> WasmEnableSjLj;
-
 static cl::list<std::string>
     EHAllowlist("emscripten-cxx-exceptions-allowed",
                 cl::desc("The list of function names in which Emscripten-style "
                          "exception handling is enabled (see emscripten "
                          "EMSCRIPTEN_CATCHING_ALLOWED options)"),
                 cl::CommaSeparated);
 
 namespace {
class WebAssemblyLowerEmscriptenEHSjLj final : public ModulePass {
   void replaceLongjmpWith(Function *LongjmpF, Function *NewF);
 
   void rebuildSSA(Function &F);
 
 public:
   static char ID;
 
   WebAssemblyLowerEmscriptenEHSjLj()
-      : ModulePass(ID), EnableEmEH(WasmEnableEmEH),
-        EnableEmSjLj(WasmEnableEmSjLj), EnableWasmSjLj(WasmEnableSjLj) {
+      : ModulePass(ID), EnableEmEH(WebAssembly::WasmEnableEmEH),
+        EnableEmSjLj(WebAssembly::WasmEnableEmSjLj),
+        EnableWasmSjLj(WebAssembly::WasmEnableSjLj) {
     assert(!(EnableEmSjLj && EnableWasmSjLj) &&
            "Two SjLj modes cannot be turned on at the same time");
     assert(!(EnableEmEH && EnableWasmSjLj) &&
            "Wasm SjLj should be only used with Wasm EH");
     EHAllowlistSet.insert(EHAllowlist.begin(), EHAllowlist.end());
   }
   bool runOnModule(Module &M) override;
 

llvm/lib/Target/WebAssembly/WebAssemblyMCInstLower.cpp

 // This disables the removal of registers when lowering into MC, as required
 // by some current tests.
 cl::opt<bool>
     WasmKeepRegisters("wasm-keep-registers", cl::Hidden,
                       cl::desc("WebAssembly: output stack registers in"
                                " instruction output for test purposes only."),
                       cl::init(false));
 
-extern cl::opt<bool> WasmEnableEmEH;
-extern cl::opt<bool> WasmEnableEmSjLj;
-
 static void removeRegisterOperands(const MachineInstr *MI, MCInst &OutMI);
 
 MCSymbol *
 WebAssemblyMCInstLower::GetGlobalAddressSymbol(const MachineOperand &MO) const {
   const GlobalValue *Global = MO.getGlobal();
   if (!isa<Function>(Global)) {
     auto *WasmSym = cast<MCSymbolWasm>(Printer.getSymbol(Global));
     // If the symbol doesn't have an explicit WasmSymbolType yet and the
WebAssemblyMCInstLower::GetGlobalAddressSymbol(const MachineOperand &MO) const {
   SmallVector<MVT, 1> ResultMVTs;
   SmallVector<MVT, 4> ParamMVTs;
   const auto *const F = dyn_cast<Function>(Global);
   computeSignatureVTs(FuncTy, F, CurrentFunc, TM, ParamMVTs, ResultMVTs);
   auto Signature = signatureFromMVTs(ResultMVTs, ParamMVTs);
 
   bool InvokeDetected = false;
   auto *WasmSym = Printer.getMCSymbolForFunction(
-      F, WasmEnableEmEH || WasmEnableEmSjLj, Signature.get(), InvokeDetected);
+      F, WebAssembly::WasmEnableEmEH || WebAssembly::WasmEnableEmSjLj,
+      Signature.get(), InvokeDetected);
   WasmSym->setSignature(Signature.get());
   Printer.addSignature(std::move(Signature));
   WasmSym->setType(wasm::WASM_SYMBOL_TYPE_FUNCTION);
   return WasmSym;
 }
 
 MCSymbol *WebAssemblyMCInstLower::GetExternalSymbolSymbol(
     const MachineOperand &MO) const {

llvm/lib/Target/WebAssembly/WebAssemblyTargetMachine.cpp

 //===- WebAssemblyTargetMachine.cpp - Define TargetMachine for WebAssembly -==//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 ///
 /// \file
 /// This file defines the WebAssembly-specific subclass of TargetMachine.
 ///
 //===----------------------------------------------------------------------===//
 
 #include "WebAssemblyTargetMachine.h"
 #include "MCTargetDesc/WebAssemblyMCTargetDesc.h"
 #include "TargetInfo/WebAssemblyTargetInfo.h"
+#include "Utils/WebAssemblyUtilities.h"
 #include "WebAssembly.h"
 #include "WebAssemblyMachineFunctionInfo.h"
 #include "WebAssemblyTargetObjectFile.h"
 #include "WebAssemblyTargetTransformInfo.h"
 #include "llvm/CodeGen/MIRParser/MIParser.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
 #include "llvm/CodeGen/Passes.h"
 #include "llvm/CodeGen/RegAllocRegistry.h"
 #include "llvm/CodeGen/TargetPassConfig.h"
 #include "llvm/IR/Function.h"
+#include "llvm/MC/MCAsmInfo.h"
 #include "llvm/MC/TargetRegistry.h"
 #include "llvm/Target/TargetOptions.h"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/Transforms/Scalar/LowerAtomic.h"
 #include "llvm/Transforms/Utils.h"
 using namespace llvm;
 
 #define DEBUG_TYPE "wasm"
 
-// Emscripten's asm.js-style exception handling
-cl::opt<bool>
-    WasmEnableEmEH("enable-emscripten-cxx-exceptions",
-                   cl::desc("WebAssembly Emscripten-style exception handling"),
-                   cl::init(false));
-
-// Emscripten's asm.js-style setjmp/longjmp handling
-cl::opt<bool> WasmEnableEmSjLj(
-    "enable-emscripten-sjlj",
-    cl::desc("WebAssembly Emscripten-style setjmp/longjmp handling"),
-    cl::init(false));
-
-// Exception handling using wasm EH instructions
-cl::opt<bool> WasmEnableEH("wasm-enable-eh",
-                           cl::desc("WebAssembly exception handling"),
-                           cl::init(false));
-
-// setjmp/longjmp handling using wasm EH instrutions
-cl::opt<bool> WasmEnableSjLj("wasm-enable-sjlj",
-                             cl::desc("WebAssembly setjmp/longjmp handling"),
-                             cl::init(false));
-
 // A command-line option to keep implicit locals
 // for the purpose of testing with lit/llc ONLY.
 // This produces output which is not valid WebAssembly, and is not supported
 // by assemblers/disassemblers and other MC based tools.
 static cl::opt<bool> WasmDisableExplicitLocals(
     "wasm-disable-explicit-locals", cl::Hidden,
     cl::desc("WebAssembly: output implicit locals in"
              " instruction output for test purposes only."),
     cl::init(false));

 WebAssemblyTargetMachine::createPassConfig(PassManagerBase &PM) {
   return new WebAssemblyPassConfig(*this, PM);
 }
 
 FunctionPass *WebAssemblyPassConfig::createTargetRegisterAllocator(bool) {
   return nullptr; // No reg alloc
 }
 
-static void basicCheckForEHAndSjLj(const TargetMachine *TM) {
+using WebAssembly::WasmEnableEH;
+using WebAssembly::WasmEnableEmEH;
+using WebAssembly::WasmEnableEmSjLj;
+using WebAssembly::WasmEnableSjLj;
+
+static void basicCheckForEHAndSjLj(TargetMachine *TM) {
+  // Before checking, we make sure TargetOptions.ExceptionModel is the same as
+  // MCAsmInfo.ExceptionsType. Normally these have to be the same, because clang
+  // stores the exception model info in LangOptions, which is later transferred
+  // to TargetOptions and MCAsmInfo. But when clang compiles bitcode directly,
+  // clang's LangOptions is not used and thus the exception model info is not
+  // correctly transferred to TargetOptions and MCAsmInfo, so we make sure we
+  // have the correct exception model in in WebAssemblyMCAsmInfo constructor.
+  // But in this case TargetOptions is still not updated, so we make sure they
+  // are the same.
+  TM->Options.ExceptionModel = TM->getMCAsmInfo()->getExceptionHandlingType();
+
   // Basic Correctness checking related to -exception-model
   if (TM->Options.ExceptionModel != ExceptionHandling::None &&
       TM->Options.ExceptionModel != ExceptionHandling::Wasm)
     report_fatal_error("-exception-model should be either 'none' or 'wasm'");
   if (WasmEnableEmEH && TM->Options.ExceptionModel == ExceptionHandling::Wasm)
     report_fatal_error("-exception-model=wasm not allowed with "
                        "-enable-emscripten-cxx-exceptions");
   if (WasmEnableEH && TM->Options.ExceptionModel != ExceptionHandling::Wasm)