[Attributor] Add interface to emit remarks in Attributor

[jhuber6]

Summary

This patch adds support for the Attributor to emit remarks on behalf of some
other pass. The attributor can now optionally take a callback function that
returns an OptimizationRemarkEmitter object when given a Function pointer. If
this is availible then a remark will be emitted for the corresponding pass
name.

Depends on D102197

[sstefan1]

Maybe we can have a test for this? Otherwise LGTM.

[thegameg]

LGTM with a test.

[jhuber6]

Adding test.

[jhuber6]

Changing after D102824 changed the Attributor constructor and adding interface for function remarks from D102784.

[jhuber6]

Updating to handle the remarks from the replace globalization pass after it was changed to use the Attributor instead.

[jhuber6]

Artefacts in the diff for some reason.

[jhuber6]

Fixing.

[jhuber6]

Comments.

[jhuber6]

Arcanist updated the wrong revision, fixing.

[jhuber6]

Changing remark if the allocation was from globalization. The message was a
little confusing since the programmer didn't explicitly allocate memory there.

[jhuber6]

Forgot to fix test.

[jhuber6]

Adding missed remark for variables that are potentially captured or freed.

llvm/include/llvm/Transforms/IPO/Attributor.h

 #include "llvm/ADT/SetVector.h"
 #include "llvm/ADT/Triple.h"
 #include "llvm/Analysis/AssumeBundleQueries.h"
 #include "llvm/Analysis/CFG.h"
 #include "llvm/Analysis/CGSCCPassManager.h"
 #include "llvm/Analysis/LazyCallGraph.h"
 #include "llvm/Analysis/LoopInfo.h"
 #include "llvm/Analysis/MustExecute.h"
+#include "llvm/Analysis/OptimizationRemarkEmitter.h"
 #include "llvm/Analysis/PostDominators.h"
 #include "llvm/Analysis/TargetLibraryInfo.h"
 #include "llvm/IR/AbstractCallSite.h"
 #include "llvm/IR/ConstantRange.h"
 #include "llvm/IR/PassManager.h"
 #include "llvm/Support/Allocator.h"
 #include "llvm/Support/Casting.h"
 #include "llvm/Support/TimeProfiler.h"

 ///       information in the IR.
 ///    b) This avoids reimplementing various kinds of lookups, e.g., to check
 ///       for existing IR attributes, in favor of a single lookups interface
 ///       provided by an abstract attribute subclass.
 ///
 /// NOTE: The mechanics of adding a new "concrete" abstract attribute are
 ///       described in the file comment.
 struct Attributor {
+
+  using OptimizationRemarkGetter =
+      function_ref<OptimizationRemarkEmitter &(Function *)>;
+
   /// Constructor
   ///
   /// \param Functions The set of functions we are deriving attributes for.
   /// \param InfoCache Cache to hold various information accessible for
   ///                  the abstract attributes.
   /// \param CGUpdater Helper to update an underlying call graph.
   /// \param Allowed If not null, a set limiting the attribute opportunities.
   /// \param DeleteFns Whether to delete functions.
   /// \param RewriteSignatures Whether to rewrite function signatures.
   Attributor(SetVector<Function *> &Functions, InformationCache &InfoCache,
              CallGraphUpdater &CGUpdater,
              DenseSet<const char *> *Allowed = nullptr, bool DeleteFns = true,
              bool RewriteSignatures = true)
       : Allocator(InfoCache.Allocator), Functions(Functions),
         InfoCache(InfoCache), CGUpdater(CGUpdater), Allowed(Allowed),
-        DeleteFns(DeleteFns), RewriteSignatures(RewriteSignatures) {}
+        DeleteFns(DeleteFns), RewriteSignatures(RewriteSignatures),
+        OREGetter(None), PassName("") {}
+
+  /// Constructor
+  ///
+  /// \param Functions The set of functions we are deriving attributes for.
+  /// \param InfoCache Cache to hold various information accessible for
+  ///                  the abstract attributes.
+  /// \param CGUpdater Helper to update an underlying call graph.
+  /// \param Allowed If not null, a set limiting the attribute opportunities.
+  /// \param DeleteFns Whether to delete functions
+  /// \param OREGetter A callback function that returns an ORE object from a
+  ///                  Function pointer.
+  /// \param PassName  The name of the pass emitting remarks.
+  Attributor(SetVector<Function *> &Functions, InformationCache &InfoCache,
+             CallGraphUpdater &CGUpdater, DenseSet<const char *> *Allowed,
+             bool DeleteFns, bool RewriteSignatures,
+             OptimizationRemarkGetter OREGetter, const char *PassName)
+      : Allocator(InfoCache.Allocator), Functions(Functions),
+        InfoCache(InfoCache), CGUpdater(CGUpdater), Allowed(Allowed),
+        DeleteFns(DeleteFns), RewriteSignatures(RewriteSignatures),
+        OREGetter(Optional<OptimizationRemarkGetter>(OREGetter)),
+        PassName(PassName) {}
 
   ~Attributor();
 
   /// Run the analyses until a fixpoint is reached or enforced (timeout).
   ///
   /// The attributes registered with this Attributor can be used after as long
   /// as the Attributor is not destroyed (it owns the attributes now).
   ///
struct Attributor {
   /// Check \p Pred on all (transitive) uses of \p V.
   ///
   /// This method will evaluate \p Pred on all (transitive) uses of the
   /// associated value and return true if \p Pred holds every time.
   bool checkForAllUses(function_ref<bool(const Use &, bool &)> Pred,
                        const AbstractAttribute &QueryingAA, const Value &V,
                        DepClassTy LivenessDepClass = DepClassTy::OPTIONAL);
 
+  /// Emit a remark generically.
+  ///
+  /// This template function can be used to generically emit a remark. The
+  /// RemarkKind should be one of the following:
+  ///   - OptimizationRemark to indicate a successful optimization attempt
+  ///   - OptimizationRemarkMissed to report a failed optimization attempt
+  ///   - OptimizationRemarkAnalysis to provide additional information about an
+  ///     optimization attempt
+  ///
+  /// The remark is built using a callback function \p RemarkCB that takes a
+  /// RemarkKind as input and returns a RemarkKind.
+  template <typename RemarkKind, typename RemarkCallBack>
+  void emitRemark(Instruction *I, StringRef RemarkName,
+                  RemarkCallBack &&RemarkCB) const {
+    if (!OREGetter)
+      return;
+
+    Function *F = I->getFunction();
+    auto &ORE = OREGetter.getValue()(F);
+
+    ORE.emit([&]() { return RemarkCB(RemarkKind(PassName, RemarkName, I)); });
+  }
+
+  /// Emit a remark on a function.
+  template <typename RemarkKind, typename RemarkCallBack>
+  void emitRemark(Function *F, StringRef RemarkName,
+                  RemarkCallBack &&RemarkCB) const {
+    if (!OREGetter)
+      return;
+
+    auto &ORE = OREGetter.getValue()(F);
+
+    ORE.emit([&]() { return RemarkCB(RemarkKind(PassName, RemarkName, F)); });
+  }
+
   /// Helper struct used in the communication between an abstract attribute (AA)
   /// that wants to change the signature of a function and the Attributor which
   /// applies the changes. The struct is partially initialized with the
   /// information from the AA (see the constructor). All other members are
   /// provided by the Attributor prior to invoking any callbacks.
   struct ArgumentReplacementInfo {
     /// Callee repair callback type
     ///
private:
   /// Functions, blocks, and instructions we delete after manifest is done.
   ///
   ///{
   SmallPtrSet<Function *, 8> ToBeDeletedFunctions;
   SmallPtrSet<BasicBlock *, 8> ToBeDeletedBlocks;
   SmallDenseSet<WeakVH, 8> ToBeDeletedInsts;
   ///}
 
+  /// Callback to get an OptimizationRemarkEmitter from a Function *.
+  Optional<OptimizationRemarkGetter> OREGetter;
+
+  /// The name of the pass to emit remarks for.
+  const char *PassName = "";
+
   friend AADepGraph;
 };
 
 /// An interface to query the internal state of an abstract attribute.
 ///
 /// The abstract state is a minimal interface that allows the Attributor to
 /// communicate with the abstract attributes about their internal state without
 /// enforcing or exposing implementation details, e.g., the (existence of an)

llvm/lib/Transforms/IPO/AttributorAttributes.cpp

for (Instruction *MallocCall : MallocCalls) {
         LLVM_DEBUG(dbgs() << "H2S: Removing free call: " << *FreeCall << "\n");
         A.deleteAfterManifest(*FreeCall);
         HasChanged = ChangeStatus::CHANGED;
       }
 
       LLVM_DEBUG(dbgs() << "H2S: Removing malloc call: " << *MallocCall
                         << "\n");
 
+      auto Remark = [&](OptimizationRemark OR) {
+        return OR << "Moving memory allocation from the heap to the stack.";
+      };
+      A.emitRemark<OptimizationRemark>(MallocCall, "HeapToStack", Remark);
+
       Align Alignment;
       Value *Size;
       if (isCallocLikeFn(MallocCall, TLI)) {
         auto *Num = MallocCall->getOperand(0);
         auto *SizeT = MallocCall->getOperand(1);
         IRBuilder<> B(MallocCall);
         Size = B.CreateMul(Num, SizeT, "h2s.calloc.size");
       } else if (isAlignedAllocLikeFn(MallocCall, TLI)) {

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

for (CallBase *CB : MallocCalls) {
       auto *SharedMem = new GlobalVariable(
           *M, Int8ArrTy, /* IsConstant */ false, GlobalValue::InternalLinkage,
           UndefValue::get(Int8ArrTy), CB->getName(), nullptr,
           GlobalValue::NotThreadLocal,
           static_cast<unsigned>(AddressSpace::Shared));
       auto *NewBuffer =
           ConstantExpr::getPointerCast(SharedMem, Int8Ty->getPointerTo());
 
+      auto Remark = [&](OptimizationRemark OR) {
+        return OR << "Replaced globalized variable with "
+                  << ore::NV("SharedMemory", AllocSize->getZExtValue())
+                  << ((AllocSize->getZExtValue() != 1) ? " bytes " : " byte ")
+                  << "of shared memory";
+      };
+      A.emitRemark<OptimizationRemark>(CB, "OpenMPReplaceGlobalization",
+                                       Remark);
+
       SharedMem->setAlignment(MaybeAlign(8));
 
       A.changeValueAfterManifest(*CB, *NewBuffer);
       A.deleteAfterManifest(*CB);
       A.deleteAfterManifest(*FreeCalls.front());
 
       NumBytesMovedToSharedMemory += AllocSize->getZExtValue();
       Changed = ChangeStatus::CHANGED;
PreservedAnalyses OpenMPOptPass::run(Module &M, ModuleAnalysisManager &AM) {
 
   BumpPtrAllocator Allocator;
   CallGraphUpdater CGUpdater;
 
   SetVector<Function *> Functions(SCC.begin(), SCC.end());
   OMPInformationCache InfoCache(M, AG, Allocator, /*CGSCC*/ Functions,
                                 OMPInModule.getKernels());
 
-  Attributor A(Functions, InfoCache, CGUpdater, nullptr, true, false);
+  Attributor A(Functions, InfoCache, CGUpdater, nullptr, true, false, OREGetter,
+               DEBUG_TYPE);
 
   OpenMPOpt OMPOpt(SCC, CGUpdater, OREGetter, InfoCache, A);
   bool Changed = OMPOpt.run(true);
   if (Changed)
     return PreservedAnalyses::none();
 
   return PreservedAnalyses::all();
 }
PreservedAnalyses OpenMPOptCGSCCPass::run(LazyCallGraph::SCC &C,
   BumpPtrAllocator Allocator;
   CallGraphUpdater CGUpdater;
   CGUpdater.initialize(CG, C, AM, UR);
 
   SetVector<Function *> Functions(SCC.begin(), SCC.end());
   OMPInformationCache InfoCache(*(Functions.back()->getParent()), AG, Allocator,
                                 /*CGSCC*/ Functions, OMPInModule.getKernels());
 
-  Attributor A(Functions, InfoCache, CGUpdater, nullptr, false);
+  Attributor A(Functions, InfoCache, CGUpdater, nullptr, false, true, OREGetter,
+               DEBUG_TYPE);
 
   OpenMPOpt OMPOpt(SCC, CGUpdater, OREGetter, InfoCache, A);
   bool Changed = OMPOpt.run(false);
   if (Changed)
     return PreservedAnalyses::none();
 
   return PreservedAnalyses::all();
 }
bool runOnSCC(CallGraphSCC &CGSCC) override {
 
     AnalysisGetter AG;
     SetVector<Function *> Functions(SCC.begin(), SCC.end());
     BumpPtrAllocator Allocator;
     OMPInformationCache InfoCache(
         *(Functions.back()->getParent()), AG, Allocator,
         /*CGSCC*/ Functions, OMPInModule.getKernels());
 
-    Attributor A(Functions, InfoCache, CGUpdater, nullptr, false);
+    Attributor A(Functions, InfoCache, CGUpdater, nullptr, false, true,
+                 OREGetter, DEBUG_TYPE);
 
     OpenMPOpt OMPOpt(SCC, CGUpdater, OREGetter, InfoCache, A);
     return OMPOpt.run(false);
   }
 
   bool doFinalization(CallGraph &CG) override { return CGUpdater.finalize(); }
 };
 

llvm/test/Transforms/OpenMP/remove_globalization.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature
 ; RUN: opt -S -passes=openmp-opt < %s | FileCheck %s
+; RUN: opt -passes=openmp-opt -pass-remarks=openmp-opt -disable-output < %s 2>&1 | FileCheck %s -check-prefix=CHECK-REMARKS
 target datalayout = "e-i64:64-i128:128-v16:16-v32:32-n16:32:64"
 target triple = "nvptx64"
 
+; CHECK-REMARKS: remark: remove_globalization.c:2:2: Moving memory allocation from the heap to the stack.
+
 @S = external local_unnamed_addr global i8*
 
 define void @kernel() {
 ; CHECK-LABEL: define {{[^@]+}}@kernel() {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    call void @foo() #[[ATTR0:[0-9]+]]
 ; CHECK-NEXT:    call void @bar() #[[ATTR0]]
 ; CHECK-NEXT:    ret void
 ;
 entry:
   call void @foo()
   call void @bar()
   ret void
 }
 
 define internal void @foo() {
 ; CHECK-LABEL: define {{[^@]+}}@foo
 ; CHECK-SAME: () #[[ATTR0]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = alloca i8, i64 4, align 1
 ; CHECK-NEXT:    ret void
 ;
 entry:
-  %0 = call i8* @__kmpc_alloc_shared(i64 4)
+  %0 = call i8* @__kmpc_alloc_shared(i64 4), !dbg !8
   call void @use(i8* %0)
   call void @__kmpc_free_shared(i8* %0)
   ret void
 }
 
 define internal void @bar() {
 ; CHECK-LABEL: define {{[^@]+}}@bar
 ; CHECK-SAME: () #[[ATTR0]] {

 !nvvm.annotations = !{!5}
 
 !0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 13.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None)
 !1 = !DIFile(filename: "remove_globalization.c", directory: "/tmp/remove_globalization.c")
 !2 = !{}
 !3 = !{i32 2, !"Debug Info Version", i32 3}
 !4 = !{i32 1, !"wchar_size", i32 4}
 !5 = !{void ()* @kernel, !"kernel", i32 1}
+!6 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 1, type: !7, scopeLine: 1, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2)
+!7 = !DISubroutineType(types: !2)
+!8 = !DILocation(line: 2, column: 2, scope: !6)

llvm/test/Transforms/OpenMP/replace_globalization.ll

 ; RUN: opt -S -passes='openmp-opt' < %s | FileCheck %s
+; RUN: opt -passes=openmp-opt -pass-remarks=openmp-opt -disable-output < %s 2>&1 | FileCheck %s -check-prefix=CHECK-REMARKS
 target datalayout = "e-i64:64-i128:128-v16:16-v32:32-n16:32:64"
 target triple = "nvptx64"
 
 @S = external local_unnamed_addr global i8*
 
+; CHECK-REMARKS: remark: replace_globalization.c:5:7: Replaced globalized variable with 16 bytes of shared memory
+; CHECK-REMARKS: remark: replace_globalization.c:5:14: Replaced globalized variable with 4 bytes of shared memory
 ; CHECK: [[SHARED_X:@.+]] = internal addrspace(3) global [16 x i8] undef
 ; CHECK: [[SHARED_Y:@.+]] = internal addrspace(3) global [4 x i8] undef
 
 ; CHECK: %{{.*}} = call i8* @__kmpc_alloc_shared({{.*}})
 ; CHECK: call void @__kmpc_free_shared({{.*}})
 define dso_local void @foo() {
 entry:
   %x = call i8* @__kmpc_alloc_shared(i64 4)
master:
   br label %exit
 exit:
   ret void
 }
 
 
 define void @use(i8* %x) {
 entry:
-  %addr = alloca i8*
   store i8* %x, i8** @S
   ret void
 }
 
 declare i8* @__kmpc_alloc_shared(i64)
 
 declare void @__kmpc_free_shared(i8*)