AMDGPU: Add pass to replace out arguments

It is better to return arguments directly in registers
if we are making a call rather than introducing expensive
stack usage. In one of sample compile from one of
Blender's many kernel variants, this fires on about
~20 different functions. Future improvements may be to
recognize simple cases where the pointer is indexing a small
array. This also fails when the store to the out argument
is in a separate block from the return, which happens in
a few of the Blender functions. This should also probably
be using MemorySSA which might help with that.

I'm not sure this is correct as a FunctionPass, but
MemoryDependenceAnalysis seems to not work with
a ModulePass.

I'm also not sure where it should run.I think it should
run  before DeadArgumentElimination, so maybe either
EP_CGSCCOptimizerLate or EP_ScalarOptimizerLate.

llvm-svn: 309416
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURewriteOutArguments.cpp b/llvm/lib/Target/AMDGPU/AMDGPURewriteOutArguments.cpp
new file mode 100644
index 0000000..4d03040
--- /dev/null
+++ b/llvm/lib/Target/AMDGPU/AMDGPURewriteOutArguments.cpp
@@ -0,0 +1,375 @@
+//===-- AMDGPURewriteOutArgumentsPass.cpp - Create struct returns ---------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+/// \file This pass attempts to replace out argument usage with a return of a
+/// struct.
+///
+/// We can support returning a lot of values directly in registers, but
+/// idiomatic C code frequently uses a pointer argument to return a second value
+/// rather than returning a struct by value. GPU stack access is also quite
+/// painful, so we want to avoid that if possible. Passing a stack object
+/// pointer to a function also requires an additional address expansion code
+/// sequence to convert the pointer to be relative to the kernel's scratch wave
+/// offset register since the callee doesn't know what stack frame the incoming
+/// pointer is relative to.
+///
+/// The goal is to try rewriting code that looks like this:
+///
+///  int foo(int a, int b, int* out) {
+///     *out = bar();
+///     return a + b;
+/// }
+///
+/// into something like this:
+///
+///  std::pair<int, int> foo(int a, int b) {
+///     return std::make_pair(a + b, bar());
+/// }
+///
+/// Typically the incoming pointer is a simple alloca for a temporary variable
+/// to use the API, which if replaced with a struct return will be easily SROA'd
+/// out when the stub function we create is inlined
+///
+/// This pass introduces the struct return, but leaves the unused pointer
+/// arguments and introduces a new stub function calling the struct returning
+/// body. DeadArgumentElimination should be run after this to clean these up.
+//
+//===----------------------------------------------------------------------===//
+
+#include "AMDGPU.h"
+#include "Utils/AMDGPUBaseInfo.h"
+
+#include "llvm/Analysis/MemoryDependenceAnalysis.h"
+#include "llvm/ADT/BitVector.h"
+#include "llvm/ADT/SetVector.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/IRBuilder.h"
+#include "llvm/IR/Module.h"
+#include "llvm/Transforms/Utils/Cloning.h"
+#include "llvm/Support/Debug.h"
+
+#define DEBUG_TYPE "amdgpu-rewrite-out-arguments"
+
+using namespace llvm;
+
+namespace {
+
+static cl::opt<bool> AnyAddressSpace(
+  "amdgpu-any-address-space-out-arguments",
+  cl::desc("Replace pointer out arguments with "
+           "struct returns for non-private address space"),
+  cl::Hidden,
+  cl::init(false));
+
+static cl::opt<unsigned> MaxNumRetRegs(
+  "amdgpu-max-return-arg-num-regs",
+  cl::desc("Approximately limit number of return registers for replacing out arguments"),
+  cl::Hidden,
+  cl::init(16));
+
+STATISTIC(NumOutArgumentsReplaced,
+          "Number out arguments moved to struct return values");
+STATISTIC(NumOutArgumentFunctionsReplaced,
+          "Number of functions with out arguments moved to struct return values");
+
+class AMDGPURewriteOutArguments : public FunctionPass {
+private:
+  const DataLayout *DL = nullptr;
+  MemoryDependenceResults *MDA = nullptr;
+
+  bool isOutArgumentCandidate(Argument &Arg) const;
+
+public:
+  static char ID;
+
+  AMDGPURewriteOutArguments() :
+    FunctionPass(ID) {}
+
+  void getAnalysisUsage(AnalysisUsage &AU) const override {
+    AU.addRequired<MemoryDependenceWrapperPass>();
+    FunctionPass::getAnalysisUsage(AU);
+  }
+
+  bool doInitialization(Module &M) override;
+  bool runOnFunction(Function &M) override;
+};
+
+} // End anonymous namespace
+
+INITIALIZE_PASS_BEGIN(AMDGPURewriteOutArguments, DEBUG_TYPE,
+                      "AMDGPU Rewrite Out Arguments", false, false)
+INITIALIZE_PASS_DEPENDENCY(MemoryDependenceWrapperPass)
+INITIALIZE_PASS_END(AMDGPURewriteOutArguments, DEBUG_TYPE,
+                    "AMDGPU Rewrite Out Arguments", false, false)
+
+char AMDGPURewriteOutArguments::ID = 0;
+
+bool AMDGPURewriteOutArguments::isOutArgumentCandidate(Argument &Arg) const {
+  const int MaxUses = 10;
+  const unsigned MaxOutArgSizeBytes = 4 * MaxNumRetRegs;
+  int UseCount = 0;
+
+  PointerType *ArgTy = dyn_cast<PointerType>(Arg.getType());
+
+  // TODO: It might be useful for any out arguments, not just privates.
+  if (!ArgTy || (ArgTy->getAddressSpace() != DL->getAllocaAddrSpace() &&
+                 !AnyAddressSpace) ||
+      Arg.hasByValAttr() || Arg.hasStructRetAttr() ||
+      DL->getTypeStoreSize(ArgTy->getPointerElementType()) > MaxOutArgSizeBytes) {
+    return false;
+  }
+
+  for (Use &U : Arg.uses()) {
+    StoreInst *SI = dyn_cast<StoreInst>(U.getUser());
+    if (UseCount > MaxUses)
+      return false;
+
+    if (!SI || !SI->isSimple() ||
+        U.getOperandNo() != StoreInst::getPointerOperandIndex())
+      return false;
+
+    ++UseCount;
+  }
+
+  // Skip unused arguments.
+  return UseCount > 0;
+}
+
+bool AMDGPURewriteOutArguments::doInitialization(Module &M) {
+  DL = &M.getDataLayout();
+  return false;
+}
+
+bool AMDGPURewriteOutArguments::runOnFunction(Function &F) {
+  if (skipFunction(F))
+    return false;
+
+  // TODO: Could probably handle variadic functions.
+  if (F.isVarArg() || F.hasStructRetAttr() ||
+      AMDGPU::isEntryFunctionCC(F.getCallingConv()))
+    return false;
+
+  MDA = &getAnalysis<MemoryDependenceWrapperPass>().getMemDep();
+
+  unsigned ReturnNumRegs = 0;
+  SmallSet<int, 4> OutArgIndexes;
+  SmallVector<Type *, 4> ReturnTypes;
+  Type *RetTy = F.getReturnType();
+  if (!RetTy->isVoidTy()) {
+    ReturnNumRegs = DL->getTypeStoreSize(RetTy) / 4;
+
+    if (ReturnNumRegs >= MaxNumRetRegs)
+      return false;
+
+    ReturnTypes.push_back(RetTy);
+  }
+
+  SmallVector<Argument *, 4> OutArgs;
+  for (Argument &Arg : F.args()) {
+    if (isOutArgumentCandidate(Arg)) {
+      DEBUG(dbgs() << "Found possible out argument " << Arg
+            << " in function " << F.getName() << '\n');
+      OutArgs.push_back(&Arg);
+    }
+  }
+
+  if (OutArgs.empty())
+    return false;
+
+  typedef SmallVector<std::pair<Argument *, Value *>, 4> ReplacementVec;
+  DenseMap<ReturnInst *, ReplacementVec> Replacements;
+
+  SmallVector<ReturnInst *, 4> Returns;
+  for (BasicBlock &BB : F) {
+    if (ReturnInst *RI = dyn_cast<ReturnInst>(&BB.back()))
+      Returns.push_back(RI);
+  }
+
+  if (Returns.empty())
+    return false;
+
+  bool Changing;
+
+  do {
+    Changing = false;
+
+    // Keep retrying if we are able to successfully eliminate an argument. This
+    // helps with cases with multiple arguments which may alias, such as in a
+    // sincos implemntation. If we have 2 stores to arguments, on the first
+    // attempt the MDA query will succeed for the second store but not the
+    // first. On the second iteration we've removed that out clobbering argument
+    // (by effectively moving it into another function) and will find the second
+    // argument is OK to move.
+    for (Argument *OutArg : OutArgs) {
+      bool ThisReplaceable = true;
+      SmallVector<std::pair<ReturnInst *, StoreInst *>, 4> ReplaceableStores;
+
+      Type *ArgTy = OutArg->getType()->getPointerElementType();
+
+      // Skip this argument if converting it will push us over the register
+      // count to return limit.
+
+      // TODO: This is an approximation. When legalized this could be more. We
+      // can ask TLI for exactly how many.
+      unsigned ArgNumRegs = DL->getTypeStoreSize(ArgTy) / 4;
+      if (ArgNumRegs + ReturnNumRegs > MaxNumRetRegs)
+        continue;
+
+      // An argument is convertible only if all exit blocks are able to replace
+      // it.
+      for (ReturnInst *RI : Returns) {
+        BasicBlock *BB = RI->getParent();
+
+        MemDepResult Q = MDA->getPointerDependencyFrom(MemoryLocation(OutArg),
+                                                       true, BB->end(), BB, RI);
+        StoreInst *SI = nullptr;
+        if (Q.isDef())
+          SI = dyn_cast<StoreInst>(Q.getInst());
+
+        if (SI) {
+          DEBUG(dbgs() << "Found out argument store: " << *SI << '\n');
+          ReplaceableStores.emplace_back(RI, SI);
+        } else {
+          ThisReplaceable = false;
+          break;
+        }
+      }
+
+      if (!ThisReplaceable)
+        continue; // Try the next argument candidate.
+
+      for (std::pair<ReturnInst *, StoreInst *> Store : ReplaceableStores) {
+        Value *ReplVal = Store.second->getValueOperand();
+
+        auto &ValVec = Replacements[Store.first];
+        if (llvm::find_if(ValVec,
+              [OutArg](const std::pair<Argument *, Value *> &Entry) {
+                 return Entry.first == OutArg;}) != ValVec.end()) {
+          DEBUG(dbgs() << "Saw multiple out arg stores" << *OutArg << '\n');
+          // It is possible to see stores to the same argument multiple times,
+          // but we expect these would have been optimized out already.
+          ThisReplaceable = false;
+          break;
+        }
+
+        ValVec.emplace_back(OutArg, ReplVal);
+        Store.second->eraseFromParent();
+      }
+
+      if (ThisReplaceable) {
+        ReturnTypes.push_back(ArgTy);
+        OutArgIndexes.insert(OutArg->getArgNo());
+        ++NumOutArgumentsReplaced;
+        Changing = true;
+      }
+    }
+  } while (Changing);
+
+  if (Replacements.empty())
+    return false;
+
+  LLVMContext &Ctx = F.getParent()->getContext();
+  StructType *NewRetTy = StructType::create(Ctx, ReturnTypes, F.getName());
+
+  FunctionType *NewFuncTy = FunctionType::get(NewRetTy,
+                                              F.getFunctionType()->params(),
+                                              F.isVarArg());
+
+  DEBUG(dbgs() << "Computed new return type: " << *NewRetTy << '\n');
+
+  Function *NewFunc = Function::Create(NewFuncTy, Function::PrivateLinkage,
+                                       F.getName() + ".body");
+  F.getParent()->getFunctionList().insert(F.getIterator(), NewFunc);
+  NewFunc->copyAttributesFrom(&F);
+  NewFunc->setComdat(F.getComdat());
+
+  // We want to preserve the function and param attributes, but need to strip
+  // off any return attributes, e.g. zeroext doesn't make sense with a struct.
+  NewFunc->stealArgumentListFrom(F);
+
+  AttrBuilder RetAttrs;
+  RetAttrs.addAttribute(Attribute::SExt);
+  RetAttrs.addAttribute(Attribute::ZExt);
+  RetAttrs.addAttribute(Attribute::NoAlias);
+  NewFunc->removeAttributes(AttributeList::ReturnIndex, RetAttrs);
+  // TODO: How to preserve metadata?
+
+  // Move the body of the function into the new rewritten function, and replace
+  // this function with a stub.
+  NewFunc->getBasicBlockList().splice(NewFunc->begin(), F.getBasicBlockList());
+
+  for (std::pair<ReturnInst *, ReplacementVec> &Replacement : Replacements) {
+    ReturnInst *RI = Replacement.first;
+    IRBuilder<> B(RI);
+    B.SetCurrentDebugLocation(RI->getDebugLoc());
+
+    int RetIdx = 0;
+    Value *NewRetVal = UndefValue::get(NewRetTy);
+
+    Value *RetVal = RI->getReturnValue();
+    if (RetVal)
+      NewRetVal = B.CreateInsertValue(NewRetVal, RetVal, RetIdx++);
+
+    for (std::pair<Argument *, Value *> ReturnPoint : Replacement.second) {
+      NewRetVal = B.CreateInsertValue(NewRetVal, ReturnPoint.second, RetIdx++);
+    }
+
+    if (RetVal)
+      RI->setOperand(0, NewRetVal);
+    else {
+      B.CreateRet(NewRetVal);
+      RI->eraseFromParent();
+    }
+  }
+
+  SmallVector<Value *, 16> StubCallArgs;
+  for (Argument &Arg : F.args()) {
+    if (OutArgIndexes.count(Arg.getArgNo())) {
+      // It's easier to preserve the type of the argument list. We rely on
+      // DeadArgumentElimination to take care of these.
+      StubCallArgs.push_back(UndefValue::get(Arg.getType()));
+    } else {
+      StubCallArgs.push_back(&Arg);
+    }
+  }
+
+  BasicBlock *StubBB = BasicBlock::Create(Ctx, "", &F);
+  IRBuilder<> B(StubBB);
+  CallInst *StubCall = B.CreateCall(NewFunc, StubCallArgs);
+
+  int RetIdx = RetTy->isVoidTy() ? 0 : 1;
+  for (Argument &Arg : F.args()) {
+    if (!OutArgIndexes.count(Arg.getArgNo()))
+      continue;
+
+    auto *EltTy = Arg.getType()->getPointerElementType();
+    unsigned Align = Arg.getParamAlignment();
+    if (Align == 0)
+      Align = DL->getABITypeAlignment(EltTy);
+
+    Value *Val = B.CreateExtractValue(StubCall, RetIdx++);
+    B.CreateAlignedStore(Val, &Arg, Align);
+  }
+
+  if (!RetTy->isVoidTy()) {
+    B.CreateRet(B.CreateExtractValue(StubCall, 0));
+  } else {
+    B.CreateRetVoid();
+  }
+
+  // The function is now a stub we want to inline.
+  F.addFnAttr(Attribute::AlwaysInline);
+
+  ++NumOutArgumentFunctionsReplaced;
+  return true;
+}
+
+FunctionPass *llvm::createAMDGPURewriteOutArgumentsPass() {
+  return new AMDGPURewriteOutArguments();
+}