[WebAssembly] Nullify unnecessary setjmp invokes

[aheejin]

Summary

This is similar to D116619, but now it handles invokes. The reason we
didn't handle invokes back then was we didn't support Wasm EH + Wasm
SjLj together, and the only case SjLj transformation will see invokes
is when we are using Wasm EH. (In Emscripten EH, they would have been
transformed to calls to invoke wrappers.)

But after D117610 we support Wasm EH + Wasm SjLj together and we can
nullify invokes to setjmp when there is no other longjmpable calls
within the function. Actually this is very unlikely to happen in
practice, because we treat destructors as longjmpable and also treat
__cxa_end_catch as longjmpable even if it is not.

[aheejin]

rebase

[aheejin]

catch.longjmp.dispatch -> catch.dispatch.longjmp

[aheejin]

rebase

llvm/lib/Target/WebAssembly/WebAssemblyLowerEmscriptenEHSjLj.cpp

 // setjmp calls into "i32 0" so they don't cause link time errors. setjmp always
 // returns 0 when called directly.
 static void nullifySetjmp(Function *F) {
   Module &M = *F->getParent();
   IRBuilder<> IRB(M.getContext());
   Function *SetjmpF = M.getFunction("setjmp");
   SmallVector<Instruction *, 1> ToErase;
 
-  for (User *U : SetjmpF->users()) {
-    auto *CI = dyn_cast<CallInst>(U);
-    // FIXME 'invoke' to setjmp can happen when we use Wasm EH + Wasm SjLj, but
-    // we don't support two being used together yet.
-    if (!CI)
-      report_fatal_error("Wasm EH + Wasm SjLj is not fully supported yet");
-    BasicBlock *BB = CI->getParent();
+  for (User *U : make_early_inc_range(SetjmpF->users())) {
+    auto *CB = dyn_cast<CallBase>(U);
+    BasicBlock *BB = CB->getParent();
     if (BB->getParent() != F) // in other function
       continue;
+    CallInst *CI = nullptr;
+    // setjmp cannot throw. So if it is an invoke, lower it to a call
+    if (auto *II = dyn_cast<InvokeInst>(CB))
+      CI = llvm::changeToCall(II);
+    else
+      CI = cast<CallInst>(CB);
     ToErase.push_back(CI);
     CI->replaceAllUsesWith(IRB.getInt32(0));
   }
   for (auto *I : ToErase)
     I->eraseFromParent();
 }
 
 bool WebAssemblyLowerEmscriptenEHSjLj::runOnModule(Module &M) {

llvm/test/CodeGen/WebAssembly/lower-em-sjlj.ll

 ; CHECK-LABEL: @setjmp_only
 entry:
   %buf = alloca [1 x %struct.__jmp_buf_tag], align 16
   %arraydecay = getelementptr inbounds [1 x %struct.__jmp_buf_tag], [1 x %struct.__jmp_buf_tag]* %buf, i32 0, i32 0
   %call = call i32 @setjmp(%struct.__jmp_buf_tag* %arraydecay) #0
   ; free cannot longjmp
   call void @free(i8* %ptr)
   ret i32 %call
+; CHECK: entry:
 ; CHECK-NOT: @malloc
 ; CHECK-NOT: %setjmpTable
 ; CHECK-NOT: @saveSetjmp
 ; CHECK-NOT: @testSetjmp
+; The remaining setjmp call is converted to constant 0, because setjmp returns 0
+; when called directly.
 ; CHECK: ret i32 0
 }
 
 ; Test SSA validity
 define void @ssa(i32 %n) {
 ; CHECK-LABEL: @ssa
 entry:
   %buf = alloca [1 x %struct.__jmp_buf_tag], align 16

llvm/test/CodeGen/WebAssembly/lower-wasm-ehsjlj.ll

ehcleanup:                                        ; preds = %entry
   ; (cleanuppad), whose parent is 'none', so we should unwind directly to
   ; %catch.dispatch.longjmp.
   %call2 = call noundef %struct.Temp* @_ZN4TempD2Ev(%struct.Temp* noundef %t) #2 [ "funclet"(token %0) ]
 ; CHECK: %call13 = invoke {{.*}} %struct.Temp* @_ZN4TempD2Ev(%struct.Temp*
 ; CHECK-NEXT:    to label {{.*}} unwind label %catch.dispatch.longjmp
   cleanupret from %0 unwind to caller
 }
 
+; This case was adapted from @cleanuppad_no_parent by removing allocas and
+; destructor calls, to generate a situation that there's only 'invoke @setjmp'
+; and no other longjmpable calls.
+define i32 @setjmp_only() personality i8* bitcast (i32 (...)* @__gxx_wasm_personality_v0 to i8*) {
+; CHECK-LABEL: @setjmp_only
+entry:
+  %buf = alloca [1 x %struct.__jmp_buf_tag], align 16
+  %arraydecay = getelementptr inbounds [1 x %struct.__jmp_buf_tag], [1 x %struct.__jmp_buf_tag]* %buf, i32 0, i32 0
+  %call = invoke i32 @setjmp(%struct.__jmp_buf_tag* noundef %arraydecay) #0
+          to label %invoke.cont unwind label %ehcleanup
+
+invoke.cont:                                      ; preds = %entry
+  ret i32 %call
+; CHECK: invoke.cont:
+; The remaining setjmp call is converted to constant 0, because setjmp returns 0
+; when called directly.
+; CHECK:   ret i32 0
+
+ehcleanup:                                        ; preds = %entry
+  %0 = cleanuppad within none []
+  cleanupret from %0 unwind to caller
+}
+
 declare void @foo()
 ; Function Attrs: nounwind
 declare %struct.Temp* @_ZN4TempD2Ev(%struct.Temp* %this) #2
 ; Function Attrs: returns_twice
 declare i32 @setjmp(%struct.__jmp_buf_tag*) #0
 ; Function Attrs: noreturn
 declare void @longjmp(%struct.__jmp_buf_tag*, i32) #1
 declare i32 @__gxx_wasm_personality_v0(...)