[LTO][Pipelines][Coro] Handle coroutines in LTO pipeline (#126168)

ThinLTO delays handling of coroutines to ThinLTO backend.
However it's usually possible to use ThinLTO prelink objects for FullLTO.

In this case we have left-over coroutines which crash in codegen.

Issue #104525.

Author: vitalybuka

Diff for: llvm/lib/Passes/PassBuilderPipelines.cpp

@@ -1840,6 +1840,8 @@ PassBuilder::buildLTODefaultPipeline(OptimizationLevel Level,
     MPM.addPass(LowerTypeTestsPass(nullptr, nullptr,
                                    lowertypetests::DropTestKind::Assume));
 
+    MPM.addPass(buildCoroWrapper(ThinOrFullLTOPhase::FullLTOPostLink));
+
     invokeFullLinkTimeOptimizationLastEPCallbacks(MPM, Level);
 
     // Emit annotation remarks.
@@ -1924,6 +1926,8 @@ PassBuilder::buildLTODefaultPipeline(OptimizationLevel Level,
     MPM.addPass(LowerTypeTestsPass(nullptr, nullptr,
                                    lowertypetests::DropTestKind::Assume));
 
+    MPM.addPass(buildCoroWrapper(ThinOrFullLTOPhase::FullLTOPostLink));
+
     invokeFullLinkTimeOptimizationLastEPCallbacks(MPM, Level);
 
     // Emit annotation remarks.
@@ -1932,6 +1936,9 @@ PassBuilder::buildLTODefaultPipeline(OptimizationLevel Level,
     return MPM;
   }
 
+  // TODO: Skip to match buildCoroWrapper.
+  MPM.addPass(CoroEarlyPass());
+
   // Optimize globals to try and fold them into constants.
   MPM.addPass(GlobalOptPass());
 
@@ -1997,7 +2004,11 @@ PassBuilder::buildLTODefaultPipeline(OptimizationLevel Level,
 
   // If we didn't decide to inline a function, check to see if we can
   // transform it to pass arguments by value instead of by reference.
-  MPM.addPass(createModuleToPostOrderCGSCCPassAdaptor(ArgumentPromotionPass()));
+  CGSCCPassManager CGPM;
+  CGPM.addPass(ArgumentPromotionPass());
+  CGPM.addPass(CoroSplitPass(Level != OptimizationLevel::O0));
+  CGPM.addPass(CoroAnnotationElidePass());
+  MPM.addPass(createModuleToPostOrderCGSCCPassAdaptor(std::move(CGPM)));
 
   FunctionPassManager FPM;
   // The IPO Passes may leave cruft around. Clean up after them.
@@ -2149,6 +2160,8 @@ PassBuilder::buildLTODefaultPipeline(OptimizationLevel Level,
   if (PTO.CallGraphProfile)
     MPM.addPass(CGProfilePass(/*InLTOPostLink=*/true));
 
+  MPM.addPass(CoroCleanupPass());
+
   invokeFullLinkTimeOptimizationLastEPCallbacks(MPM, Level);
 
   // Emit annotation remarks.

Diff for: llvm/test/LTO/X86/coro.ll

@@ -0,0 +1,21 @@
+; RUN: llvm-as %s -o %t1.bc
+; RUN: llvm-lto2 run %t1.bc -o %t2.o -r=%t1.bc,test,plx -r=%t1.bc,extern_func,plx -save-temps
+; RUN: llvm-dis %t2.o.0.5.precodegen.bc -o - | FileCheck %s --implicit-check-not="call void @llvm.coro"
+
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-fuchsia"
+
+declare void @extern_func()
+
+; CHECK:      define {{.*}} void @test(
+define void @test(ptr %hdl) {
+  call void @llvm.coro.resume(ptr %hdl)
+  call void @llvm.coro.destroy(ptr %hdl)
+  call i1 @llvm.coro.done(ptr %hdl)
+  ret void
+}
+
+declare void @llvm.coro.resume(ptr)
+declare void @llvm.coro.destroy(ptr)
+declare i1 @llvm.coro.done(ptr)
+

Diff for: llvm/test/Other/new-pm-O0-defaults.ll

@@ -51,6 +51,7 @@
 ; CHECK-LTO-NEXT: Running analysis: InnerAnalysisManagerProxy
 ; CHECK-LTO-NEXT: Running pass: LowerTypeTestsPass
 ; CHECK-LTO-NEXT: Running pass: LowerTypeTestsPass
+; CHECK-LTO-NEXT: CoroConditionalWrapper
 ; CHECK-CORO-NEXT: Running pass: AnnotationRemarksPass
 ; CHECK-CORO-NEXT: Running analysis: TargetLibraryAnalysis
 ; CHECK-LTO-NEXT: Running pass: AnnotationRemarksPass

Diff for: llvm/test/Other/new-pm-lto-defaults.ll

@@ -67,6 +67,7 @@
 ; CHECK-O1-NEXT: Running analysis: TargetLibraryAnalysis
 ; CHECK-O-NEXT: Running pass: GlobalSplitPass
 ; CHECK-O-NEXT: Running pass: WholeProgramDevirtPass
+; CHECK-O23SZ-NEXT: Running pass: CoroEarlyPass
 ; CHECK-O1-NEXT: Running pass: LowerTypeTestsPass
 ; CHECK-O23SZ-NEXT: Running pass: GlobalOptPass
 ; CHECK-O23SZ-NEXT: Running pass: PromotePass
@@ -85,7 +86,9 @@
 ; CHECK-O23SZ-NEXT: Running pass: GlobalOptPass
 ; CHECK-O23SZ-NEXT: Running pass: OpenMPOptPass
 ; CHECK-O23SZ-NEXT: Running pass: GlobalDCEPass
-; CHECK-O23SZ-NEXT: Running pass: ArgumentPromotionPass
+; CHECK-O23SZ-NEXT: Running pass: ArgumentPromotionPass on (foo)
+; CHECK-O23SZ-NEXT: CoroSplitPass on (foo)
+; CHECK-O23SZ-NEXT: CoroAnnotationElidePass on (foo)
 ; CHECK-O23SZ-NEXT: Running pass: InstCombinePass
 ; CHECK-EP-PEEPHOLE-NEXT: Running pass: NoOpFunctionPass
 ; CHECK-O23SZ-NEXT: Running pass: ConstraintEliminationPass
@@ -156,6 +159,8 @@
 ; CHECK-O23SZ-NEXT: Running pass: GlobalDCEPass
 ; CHECK-O23SZ-NEXT: Running pass: RelLookupTableConverterPass
 ; CHECK-O23SZ-NEXT: Running pass: CGProfilePass
+; CHECK-O1-NEXT: Running pass: CoroConditionalWrapper
+; CHECK-O23SZ-NEXT: Running pass: CoroCleanupPass
 ; CHECK-EP-NEXT: Running pass: NoOpModulePass
 ; CHECK-O-NEXT: Running pass: AnnotationRemarksPass on foo
 ; CHECK-O-NEXT: Running pass: PrintModulePass