[AMDGPU] Merge AMDGPULDSUtils into AMDGPUMemoryUtils

Differential Revision: https://reviews.llvm.org/D119502

llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp

@@ -28,7 +28,7 @@
 
 #include "AMDGPU.h"
 #include "Utils/AMDGPUBaseInfo.h"
-#include "Utils/AMDGPULDSUtils.h"
+#include "Utils/AMDGPUMemoryUtils.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/DerivedTypes.h"

llvm/lib/Target/AMDGPU/AMDGPUReplaceLDSUseWithPointer.cpp

@@ -83,7 +83,7 @@
 #include "AMDGPU.h"
 #include "GCNSubtarget.h"
 #include "Utils/AMDGPUBaseInfo.h"
-#include "Utils/AMDGPULDSUtils.h"
+#include "Utils/AMDGPUMemoryUtils.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SetOperations.h"

llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp

@@ -1443,6 +1443,10 @@ bool isModuleEntryFunctionCC(CallingConv::ID CC) {
   }
 }
 
+bool isKernelCC(const Function *Func) {
+  return AMDGPU::isModuleEntryFunctionCC(Func->getCallingConv());
+}
+
 bool hasXNACK(const MCSubtargetInfo &STI) {
   return STI.getFeatureBits()[AMDGPU::FeatureXNACK];
 }

llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.h

@@ -741,6 +741,8 @@ bool isEntryFunctionCC(CallingConv::ID CC);
 LLVM_READNONE
 bool isModuleEntryFunctionCC(CallingConv::ID CC);
 
+bool isKernelCC(const Function *Func);
+
 // FIXME: Remove this when calling conventions cleaned up
 LLVM_READNONE
 inline bool isKernel(CallingConv::ID CC) {

llvm/lib/Target/AMDGPU/Utils/AMDGPUMemoryUtils.cpp

@@ -8,12 +8,16 @@
 
 #include "AMDGPUMemoryUtils.h"
 #include "AMDGPU.h"
+#include "AMDGPUBaseInfo.h"
+#include "llvm/ADT/SetVector.h"
 #include "llvm/ADT/SmallSet.h"
 #include "llvm/Analysis/AliasAnalysis.h"
 #include "llvm/Analysis/MemorySSA.h"
+#include "llvm/IR/DataLayout.h"
 #include "llvm/IR/Instructions.h"
-#include "llvm/IR/IntrinsicsAMDGPU.h"
 #include "llvm/IR/IntrinsicInst.h"
+#include "llvm/IR/IntrinsicsAMDGPU.h"
+#include "llvm/IR/ReplaceConstant.h"
 
 #define DEBUG_TYPE "amdgpu-memory-utils"
 
@@ -23,6 +27,117 @@ namespace llvm {
 
 namespace AMDGPU {
 
+Align getAlign(DataLayout const &DL, const GlobalVariable *GV) {
+  return DL.getValueOrABITypeAlignment(GV->getPointerAlignment(DL),
+                                       GV->getValueType());
+}
+
+static void collectFunctionUses(User *U, const Function *F,
+                                SetVector<Instruction *> &InstUsers) {
+  SmallVector<User *> Stack{U};
+
+  while (!Stack.empty()) {
+    U = Stack.pop_back_val();
+
+    if (auto *I = dyn_cast<Instruction>(U)) {
+      if (I->getFunction() == F)
+        InstUsers.insert(I);
+      continue;
+    }
+
+    if (!isa<ConstantExpr>(U))
+      continue;
+
+    append_range(Stack, U->users());
+  }
+}
+
+void replaceConstantUsesInFunction(ConstantExpr *C, const Function *F) {
+  SetVector<Instruction *> InstUsers;
+
+  collectFunctionUses(C, F, InstUsers);
+  for (Instruction *I : InstUsers) {
+    convertConstantExprsToInstructions(I, C);
+  }
+}
+
+static bool shouldLowerLDSToStruct(const GlobalVariable &GV,
+                                   const Function *F) {
+  // We are not interested in kernel LDS lowering for module LDS itself.
+  if (F && GV.getName() == "llvm.amdgcn.module.lds")
+    return false;
+
+  bool Ret = false;
+  SmallPtrSet<const User *, 8> Visited;
+  SmallVector<const User *, 16> Stack(GV.users());
+
+  assert(!F || isKernelCC(F));
+
+  while (!Stack.empty()) {
+    const User *V = Stack.pop_back_val();
+    Visited.insert(V);
+
+    if (isa<GlobalValue>(V)) {
+      // This use of the LDS variable is the initializer of a global variable.
+      // This is ill formed. The address of an LDS variable is kernel dependent
+      // and unknown until runtime. It can't be written to a global variable.
+      continue;
+    }
+
+    if (auto *I = dyn_cast<Instruction>(V)) {
+      const Function *UF = I->getFunction();
+      if (UF == F) {
+        // Used from this kernel, we want to put it into the structure.
+        Ret = true;
+      } else if (!F) {
+        // For module LDS lowering, lowering is required if the user instruction
+        // is from non-kernel function.
+        Ret |= !isKernelCC(UF);
+      }
+      continue;
+    }
+
+    // User V should be a constant, recursively visit users of V.
+    assert(isa<Constant>(V) && "Expected a constant.");
+    append_range(Stack, V->users());
+  }
+
+  return Ret;
+}
+
+std::vector<GlobalVariable *> findVariablesToLower(Module &M,
+                                                   const Function *F) {
+  std::vector<llvm::GlobalVariable *> LocalVars;
+  for (auto &GV : M.globals()) {
+    if (GV.getType()->getPointerAddressSpace() != AMDGPUAS::LOCAL_ADDRESS) {
+      continue;
+    }
+    if (!GV.hasInitializer()) {
+      // addrspace(3) without initializer implies cuda/hip extern __shared__
+      // the semantics for such a variable appears to be that all extern
+      // __shared__ variables alias one another, in which case this transform
+      // is not required
+      continue;
+    }
+    if (!isa<UndefValue>(GV.getInitializer())) {
+      // Initializers are unimplemented for LDS address space.
+      // Leave such variables in place for consistent error reporting.
+      continue;
+    }
+    if (GV.isConstant()) {
+      // A constant undef variable can't be written to, and any load is
+      // undef, so it should be eliminated by the optimizer. It could be
+      // dropped by the back end if not. This pass skips over it.
+      continue;
+    }
+    if (!shouldLowerLDSToStruct(GV, F)) {
+      continue;
+    }
+    LocalVars.push_back(&GV);
+  }
+  return LocalVars;
+}
+
 bool isReallyAClobber(const Value *Ptr, MemoryDef *Def, AAResults *AA) {
   Instruction *DefInst = Def->getMemoryInst();
 

llvm/lib/Target/AMDGPU/Utils/AMDGPUMemoryUtils.h

@@ -9,16 +9,32 @@
 #ifndef LLVM_LIB_TARGET_AMDGPU_UTILS_AMDGPUMEMORYUTILS_H
 #define LLVM_LIB_TARGET_AMDGPU_UTILS_AMDGPUMEMORYUTILS_H
 
+#include <vector>
+
 namespace llvm {
 
+struct Align;
 class AAResults;
+class ConstantExpr;
+class DataLayout;
+class Function;
+class GlobalVariable;
 class LoadInst;
 class MemoryDef;
 class MemorySSA;
+class Module;
 class Value;
 
 namespace AMDGPU {
 
+Align getAlign(DataLayout const &DL, const GlobalVariable *GV);
+
+std::vector<GlobalVariable *> findVariablesToLower(Module &M,
+                                                   const Function *F = nullptr);
+
+/// Replace all uses of constant \p C with instructions in \p F.
+void replaceConstantUsesInFunction(ConstantExpr *C, const Function *F);
+
 /// Given a \p Def clobbering a load from \p Ptr accroding to the MSSA check
 /// if this is actually a memory update or an artifical clobber to facilitate
 /// ordering constraints.

llvm/lib/Target/AMDGPU/Utils/CMakeLists.txt

@@ -1,7 +1,6 @@
 add_llvm_component_library(LLVMAMDGPUUtils
   AMDGPUAsmUtils.cpp
   AMDGPUBaseInfo.cpp
-  AMDGPULDSUtils.cpp
   AMDGPUMemoryUtils.cpp
   AMDGPUPALMetadata.cpp
   AMDKernelCodeTUtils.cpp