[PM] Add devirtualization-based iteration utility into the new PM's default pipeline.

[chandlerc]

Summary

What it says on the tin. =]

A clang with this patch built with ASan and asserts can build all of the
test-suite as well, so it seems to not uncover any latent problems.

[davide]

Terrific =)

llvm/trunk/lib/Passes/PassBuilder.cpp

 #include "llvm/Transforms/Utils/SymbolRewriter.h"
 #include "llvm/Transforms/Vectorize/LoopVectorize.h"
 #include "llvm/Transforms/Vectorize/SLPVectorizer.h"
 
 #include <type_traits>
 
 using namespace llvm;
 
+static cl::opt<unsigned> MaxDevirtIterations("pm-max-devirt-iterations",
+                                             cl::ReallyHidden, cl::init(4));
+
 static Regex DefaultAliasRegex("^(default|lto-pre-link|lto)<(O[0123sz])>$");
 
 static bool isOptimizingForSize(PassBuilder::OptimizationLevel Level) {
   switch (Level) {
   case PassBuilder::O0:
   case PassBuilder::O1:
   case PassBuilder::O2:
   case PassBuilder::O3:
PassBuilder::buildPerModuleDefaultPipeline(OptimizationLevel Level,
   if (Level == O3)
     MainCGPipeline.addPass(ArgumentPromotionPass());
 
   // Lastly, add the core function simplification pipeline nested inside the
   // CGSCC walk.
   MainCGPipeline.addPass(createCGSCCToFunctionPassAdaptor(
       buildFunctionSimplificationPipeline(Level, DebugLogging)));
 
+  // We wrap the CGSCC pipeline in a devirtualization repeater. This will try
+  // to detect when we devirtualize indirect calls and iterate the SCC passes
+  // in that case to try and catch knock-on inlining or function attrs
+  // opportunities. Then we add it to the module pipeline by walking the SCCs
+  // in postorder (or bottom-up).
   MPM.addPass(
-      createModuleToPostOrderCGSCCPassAdaptor(std::move(MainCGPipeline)));
+      createModuleToPostOrderCGSCCPassAdaptor(createDevirtSCCRepeatedPass(
+          std::move(MainCGPipeline), MaxDevirtIterations, DebugLogging)));
 
   // This ends the canonicalization and simplification phase of the pipeline.
   // At this point, we expect to have canonical and simple IR which we begin
   // *optimizing* for efficient execution going forward.
 
   // Eliminate externally available functions now that inlining is over -- we
   // won't emit these anyways.
   MPM.addPass(EliminateAvailableExternallyPass());

llvm/trunk/test/Other/cgscc-devirt-iteration.ll

 ; The CGSCC pass manager includes an SCC iteration utility that tracks indirect
 ; calls that are turned into direct calls (devirtualization) and re-visits the
 ; SCC to expose those calls to the SCC-based IPO passes. We trigger
 ; devirtualization here with GVN which forwards a store through a load and to
 ; an indirect call.
 ;
 ; RUN: opt -aa-pipeline=basic-aa -passes='cgscc(function-attrs,function(gvn,instcombine))' -S < %s | FileCheck %s --check-prefix=CHECK --check-prefix=BEFORE
 ; RUN: opt -aa-pipeline=basic-aa -passes='cgscc(devirt<1>(function-attrs,function(gvn,instcombine)))' -S < %s | FileCheck %s --check-prefix=CHECK --check-prefix=AFTER --check-prefix=AFTER1
 ; RUN: opt -aa-pipeline=basic-aa -passes='cgscc(devirt<2>(function-attrs,function(gvn,instcombine)))' -S < %s | FileCheck %s --check-prefix=CHECK --check-prefix=AFTER --check-prefix=AFTER2
+;
+; We also verify that the real O2 pipeline catches these cases.
+; RUN: opt -aa-pipeline=basic-aa -passes='default<O2>' -S < %s | FileCheck %s --check-prefix=CHECK --check-prefix=AFTER --check-prefix=AFTER2
 
 declare void @readnone() readnone
 ; CHECK: Function Attrs: readnone
 ; CHECK: declare void @readnone()
 
 declare void @unknown()
 ; CHECK-NOT: Function Attrs
 ; CHECK: declare void @unknown()
entry:
   call void %f2() readonly
 ; BEFORE: call void %f2()
 ; AFTER: call void @readnone()
 
   ret void
 }
 
 declare i8* @memcpy(i8*, i8*, i64)
-; CHECK-NOT: Function Attrs
-; CHECK: declare i8* @memcpy(i8*, i8*, i64)
+; CHECK: declare i8* @memcpy(
 
 ; The @test3 function checks that when we refine an indirect call to an
 ; intrinsic we still revisit the SCC pass. This also covers cases where the
 ; value handle itself doesn't persist due to the nature of how instcombine
 ; creates the memcpy intrinsic call, and we rely on the count of indirect calls
 ; decreasing and the count of direct calls increasing.
 define void @test3(i8* %src, i8* %dest, i64 %size) {
 ; CHECK-NOT: Function Attrs
 ; BEFORE: define void @test3(i8* %src, i8* %dest, i64 %size)
 ; AFTER: define void @test3(i8* nocapture readonly %src, i8* nocapture %dest, i64 %size)
   %fptr = alloca i8* (i8*, i8*, i64)*
   store i8* (i8*, i8*, i64)* @memcpy, i8* (i8*, i8*, i64)** %fptr
   %f = load i8* (i8*, i8*, i64)*, i8* (i8*, i8*, i64)** %fptr
   call i8* %f(i8* %dest, i8* %src, i64 %size)
 ; CHECK: call void @llvm.memcpy
   ret void
 }
+
+; A boring function that just keeps our declarations around.
+define void @keep(i8** %sink) {
+; CHECK-NOT: Function Attrs
+; CHECK: define void @keep(
+entry:
+  store volatile i8* bitcast (void ()* @readnone to i8*), i8** %sink
+  store volatile i8* bitcast (void ()* @unknown to i8*), i8** %sink
+  store volatile i8* bitcast (i8* (i8*, i8*, i64)* @memcpy to i8*), i8** %sink
+  call void @unknown()
+  ret void
+}