Merged master:c6f51377e12b into amd-gfx:4a16eb27d030

Local branch amd-gfx 4a16eb2 Merged master:06d100a69a08 into amd-gfx:9bd98790e775
Remote branch master c6f5137 [libcxx/variant] Add a few benchmarks for .

libcxx/benchmarks/VariantBenchmarks.h

@@ -0,0 +1,58 @@
+// -*- C++ -*-
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef BENCHMARK_VARIANT_BENCHMARKS_H
+#define BENCHMARK_VARIANT_BENCHMARKS_H
+
+#include <array>
+#include <cstddef>
+#include <tuple>
+#include <type_traits>
+#include <variant>
+
+#include "benchmark/benchmark.h"
+
+#include "GenerateInput.h"
+
+namespace VariantBenchmarks {
+
+template <std::size_t I>
+struct S {
+  static constexpr size_t v = I;
+};
+
+template <std::size_t N, std::size_t... Is>
+static auto genVariants(std::index_sequence<Is...>) {
+  using V = std::variant<S<Is>...>;
+  using F = V (*)();
+  static constexpr F fs[] = {[] { return V(std::in_place_index<Is>); }...};
+
+  std::array<V, N> result = {};
+  for (auto& v : result) {
+    v = fs[getRandomInteger(0ul, sizeof...(Is) - 1)]();
+  }
+
+  return result;
+}
+
+template <std::size_t N, std::size_t Alts>
+static void BM_Visit(benchmark::State& state) {
+  auto args = genVariants<N>(std::make_index_sequence<Alts>{});
+  for (auto _ : state) {
+    benchmark::DoNotOptimize(std::apply(
+        [](auto... vs) {
+          return std::visit([](auto... is) { return (is.v + ... + 0); }, vs...);
+        },
+        args));
+  }
+}
+
+} // end namespace VariantBenchmarks
+
+#endif // BENCHMARK_VARIANT_BENCHMARKS_H

libcxx/benchmarks/variant_visit_1.bench.cpp

@@ -0,0 +1,27 @@
+#include "benchmark/benchmark.h"
+
+#include "VariantBenchmarks.h"
+
+using namespace VariantBenchmarks;
+
+BENCHMARK_TEMPLATE(BM_Visit, 1, 1);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 2);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 3);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 4);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 5);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 6);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 7);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 8);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 9);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 10);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 20);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 30);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 40);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 50);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 60);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 70);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 80);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 90);
+BENCHMARK_TEMPLATE(BM_Visit, 1, 100);
+
+BENCHMARK_MAIN();

libcxx/benchmarks/variant_visit_2.bench.cpp

@@ -0,0 +1,22 @@
+#include "benchmark/benchmark.h"
+
+#include "VariantBenchmarks.h"
+
+using namespace VariantBenchmarks;
+
+BENCHMARK_TEMPLATE(BM_Visit, 2, 1);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 2);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 3);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 4);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 5);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 6);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 7);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 8);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 9);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 10);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 20);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 30);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 40);
+BENCHMARK_TEMPLATE(BM_Visit, 2, 50);
+
+BENCHMARK_MAIN();

libcxx/benchmarks/variant_visit_3.bench.cpp

@@ -0,0 +1,20 @@
+#include "benchmark/benchmark.h"
+
+#include "VariantBenchmarks.h"
+
+using namespace VariantBenchmarks;
+
+BENCHMARK_TEMPLATE(BM_Visit, 3, 1);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 2);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 3);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 4);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 5);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 6);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 7);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 8);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 9);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 10);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 15);
+BENCHMARK_TEMPLATE(BM_Visit, 3, 20);
+
+BENCHMARK_MAIN();

llvm/include/llvm/IR/Instruction.h

@@ -492,10 +492,6 @@ class Instruction : public User,
   /// merged DebugLoc.
   void applyMergedLocation(const DILocation *LocA, const DILocation *LocB);
 
-  /// Updates the debug location given that the instruction has been hoisted
-  /// from a block to a predecessor of that block.
-  void updateLocationAfterHoist();
-
 private:
   /// Return true if we have an entry in the on-the-side metadata hash.
   bool hasMetadataHashEntry() const {

llvm/lib/IR/DebugInfo.cpp

@@ -696,24 +696,6 @@ void Instruction::applyMergedLocation(const DILocation *LocA,
   setDebugLoc(DILocation::getMergedLocation(LocA, LocB));
 }
 
-void Instruction::updateLocationAfterHoist() {
-  const DebugLoc &DL = getDebugLoc();
-  if (!DL)
-    return;
-
-  // If we didn't hoist a call, drop the location to allow a location from a
-  // preceding instruction to propagate.
-  if (!isa<CallBase>(this)) {
-    setDebugLoc(DebugLoc());
-    return;
-  }
-
-  // Set a line 0 location for (potentially inlinable) calls. Set the scope to
-  // the parent function's scope if it's available, and drop any inlinedAt info
-  // to avoid making it look like the inlined callee was reached early.
-  setDebugLoc(DebugLoc::get(0, 0, DL.getScope()));
-}
-
 //===----------------------------------------------------------------------===//
 // LLVM C API implementations.
 //===----------------------------------------------------------------------===//

llvm/lib/Transforms/IPO/GlobalOpt.cpp

@@ -468,19 +468,16 @@ static bool CanDoGlobalSRA(GlobalVariable *GV) {
 /// Copy over the debug info for a variable to its SRA replacements.
 static void transferSRADebugInfo(GlobalVariable *GV, GlobalVariable *NGV,
                                  uint64_t FragmentOffsetInBits,
-                                 uint64_t FragmentSizeInBits) {
+                                 uint64_t FragmentSizeInBits,
+                                 uint64_t VarSize) {
   SmallVector<DIGlobalVariableExpression *, 1> GVs;
   GV->getDebugInfo(GVs);
   for (auto *GVE : GVs) {
     DIVariable *Var = GVE->getVariable();
-    Optional<uint64_t> VarSize = Var->getSizeInBits();
-
     DIExpression *Expr = GVE->getExpression();
     // If the FragmentSize is smaller than the variable,
     // emit a fragment expression.
-    // If the variable size is unknown a fragment must be
-    // emitted to be safe.
-    if (!VarSize || FragmentSizeInBits < *VarSize) {
+    if (FragmentSizeInBits < VarSize) {
       if (auto E = DIExpression::createFragmentExpression(
               Expr, FragmentOffsetInBits, FragmentSizeInBits))
         Expr = *E;
@@ -505,6 +502,7 @@ static GlobalVariable *SRAGlobal(GlobalVariable *GV, const DataLayout &DL) {
   assert(GV->hasLocalLinkage());
   Constant *Init = GV->getInitializer();
   Type *Ty = Init->getType();
+  uint64_t VarSize = DL.getTypeSizeInBits(Ty);
 
   std::map<unsigned, GlobalVariable *> NewGlobals;
 
@@ -560,7 +558,7 @@ static GlobalVariable *SRAGlobal(GlobalVariable *GV, const DataLayout &DL) {
       // Copy over the debug info for the variable.
       uint64_t Size = DL.getTypeAllocSizeInBits(NGV->getValueType());
       uint64_t FragmentOffsetInBits = Layout.getElementOffsetInBits(ElementIdx);
-      transferSRADebugInfo(GV, NGV, FragmentOffsetInBits, Size);
+      transferSRADebugInfo(GV, NGV, FragmentOffsetInBits, Size, VarSize);
     } else {
       uint64_t EltSize = DL.getTypeAllocSize(ElTy);
       Align EltAlign = DL.getABITypeAlign(ElTy);
@@ -573,7 +571,7 @@ static GlobalVariable *SRAGlobal(GlobalVariable *GV, const DataLayout &DL) {
       if (NewAlign > EltAlign)
         NGV->setAlignment(NewAlign);
       transferSRADebugInfo(GV, NGV, FragmentSizeInBits * ElementIdx,
-                           FragmentSizeInBits);
+                           FragmentSizeInBits, VarSize);
     }
   }
 

llvm/lib/Transforms/Scalar/GVN.cpp

@@ -46,6 +46,7 @@
 #include "llvm/IR/Constant.h"
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/DataLayout.h"
+#include "llvm/IR/DebugInfoMetadata.h"
 #include "llvm/IR/DebugLoc.h"
 #include "llvm/IR/Dominators.h"
 #include "llvm/IR/Function.h"
@@ -1313,7 +1314,8 @@ bool GVN::PerformLoadPRE(LoadInst *LI, AvailValInBlkVect &ValuesPerBlock,
     // Instructions that have been inserted in predecessor(s) to materialize
     // the load address do not retain their original debug locations. Doing
     // so could lead to confusing (but correct) source attributions.
-    I->updateLocationAfterHoist();
+    if (const DebugLoc &DL = I->getDebugLoc())
+      I->setDebugLoc(DebugLoc::get(0, 0, DL.getScope(), DL.getInlinedAt()));
 
     // FIXME: We really _ought_ to insert these value numbers into their
     // parent's availability map.  However, in doing so, we risk getting into

llvm/lib/Transforms/Scalar/LICM.cpp

@@ -1658,7 +1658,10 @@ static void hoist(Instruction &I, const DominatorTree *DT, const Loop *CurLoop,
     // Move the new node to the destination block, before its terminator.
     moveInstructionBefore(I, *Dest->getTerminator(), *SafetyInfo, MSSAU, SE);
 
-  I.updateLocationAfterHoist();
+  // Apply line 0 debug locations when we are moving instructions to different
+  // basic blocks because we want to avoid jumpy line tables.
+  if (const DebugLoc &DL = I.getDebugLoc())
+    I.setDebugLoc(DebugLoc::get(0, 0, DL.getScope(), DL.getInlinedAt()));
 
   if (isa<LoadInst>(I))
     ++NumMovedLoads;

llvm/test/DebugInfo/Generic/global-sra-struct-fwd-decl.ll

@@ -0,0 +1,63 @@
+; RUN: opt -S -globalopt < %s | FileCheck %s
+; Generated at -O2 -g from:
+; typedef struct {} a;
+; static struct {
+;     long b;
+;     a c;
+; } d;
+; e() {
+;     long f = d.b + 1;
+;     d.b = f;
+; }
+; (IR is modified so that d's struct type is forward declared.)
+
+; Check that the global variable "d" is not
+; emitted as a fragment if its struct type is
+; forward declared but d.c has zero length, so
+; a fragment shouldn't be emitted.
+
+source_filename = "t.c"
+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-linux-gnu"
+
+%struct.anon = type { i64, %struct.a }
+%struct.a = type {}
+
+; CHECK: @d.0 = internal unnamed_addr global i64 0, align 8, !dbg ![[GVE:.*]]
+@d = internal global %struct.anon zeroinitializer, align 8, !dbg !0
+
+; Function Attrs: noinline nounwind uwtable
+define dso_local i32 @e() #0 !dbg !18 {
+entry:
+  %0 = load i64, i64* getelementptr inbounds (%struct.anon, %struct.anon* @d, i32 0, i32 0), align 8
+  %add = add nsw i64 %0, 1
+  call void @llvm.dbg.value(metadata i64 %add, metadata !24, metadata !DIExpression()), !dbg !25
+  store i64 %add, i64* getelementptr inbounds (%struct.anon, %struct.anon* @d, i32 0, i32 0), align 8
+  ret i32 undef
+}
+
+; Function Attrs: nounwind readnone speculatable willreturn
+declare void @llvm.dbg.declare(metadata, metadata, metadata)
+
+; Function Attrs: nounwind readnone speculatable willreturn
+declare void @llvm.dbg.value(metadata, metadata, metadata)
+
+!llvm.dbg.cu = !{!2}
+!llvm.module.flags = !{!14, !15}
+
+; CHECK: ![[GVE]] = !DIGlobalVariableExpression(var: !1, expr: !DIExpression())
+!0 = !DIGlobalVariableExpression(var: !1, expr: !DIExpression())
+!1 = distinct !DIGlobalVariable(name: "d", scope: !2, file: !3, line: 6, type: !7, isLocal: true, isDefinition: true)
+!2 = distinct !DICompileUnit(language: DW_LANG_C99, file: !3, isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !{}, globals: !{!0}, splitDebugInlining: false, nameTableKind: None)
+!3 = !DIFile(filename: "t.c", directory: "/")
+!7 = distinct !DICompositeType(tag: DW_TAG_structure_type, file: !3, line: 3, flags: DIFlagFwdDecl)
+!10 = !DIBasicType(name: "long int", size: 64, encoding: DW_ATE_signed)
+!12 = !DIDerivedType(tag: DW_TAG_typedef, name: "a", file: !3, line: 2, baseType: !13)
+!13 = distinct !DICompositeType(tag: DW_TAG_structure_type, file: !3, line: 1, elements: !{})
+!14 = !{i32 7, !"Dwarf Version", i32 4}
+!15 = !{i32 2, !"Debug Info Version", i32 3}
+!18 = distinct !DISubprogram(name: "e", scope: !3, file: !3, line: 7, type: !19, scopeLine: 7, spFlags: DISPFlagDefinition, unit: !2, retainedNodes: !{})
+!19 = !DISubroutineType(types: !{!21})
+!21 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
+!24 = !DILocalVariable(name: "f", scope: !18, file: !3, line: 8, type: !10)
+!25 = !DILocation(line: 0, scope: !18)

llvm/test/DebugInfo/Generic/licm-hoist-debug-loc.ll

@@ -18,8 +18,9 @@
 ; We make sure that the instruction that is hoisted into the preheader
 ; does not have a debug location.
 ; CHECK: for.body.lr.ph:
-; CHECK: getelementptr{{.*}}%p.addr, i64 4{{$}}
+; CHECK: getelementptr{{.*}}%p.addr, i64 4{{.*}} !dbg [[zero:![0-9]+]]
 ; CHECK: for.body:
+; CHECK: [[zero]] = !DILocation(line: 0
 ;
 ; ModuleID = 't.ll'
 source_filename = "test.c"

llvm/test/Transforms/GVN/PRE/phi-translate.ll

@@ -4,8 +4,8 @@ target datalayout = "e-p:64:64:64"
 
 ; CHECK-LABEL: @foo(
 ; CHECK: entry.end_crit_edge:
-; CHECK: %[[INDEX:[a-z0-9.]+]] = sext i32 %x to i64{{$}}
-; CHECK: %[[ADDRESS:[a-z0-9.]+]] = getelementptr [100 x i32], [100 x i32]* @G, i64 0, i64 %[[INDEX]]{{$}}
+; CHECK: %[[INDEX:[a-z0-9.]+]] = sext i32 %x to i64{{.*}} !dbg [[ZERO_LOC:![0-9]+]]
+; CHECK: %[[ADDRESS:[a-z0-9.]+]] = getelementptr [100 x i32], [100 x i32]* @G, i64 0, i64 %[[INDEX]]{{.*}} !dbg [[ZERO_LOC]]
 ; CHECK:   %n.pre = load i32, i32* %[[ADDRESS]], align 4, !dbg [[N_LOC:![0-9]+]]
 ; CHECK: br label %end
 ; CHECK: then:
@@ -14,7 +14,8 @@ target datalayout = "e-p:64:64:64"
 ; CHECK:   %n = phi i32 [ %n.pre, %entry.end_crit_edge ], [ %z, %then ], !dbg [[N_LOC]]
 ; CHECK:   ret i32 %n
 
-; CHECK: [[N_LOC]] = !DILocation(line: 47, column: 1, scope: !{{.*}})
+; CHECK-DAG: [[N_LOC]] = !DILocation(line: 47, column: 1, scope: !{{.*}})
+; CHECK-DAG: [[ZERO_LOC]] = !DILocation(line: 0
 
 @G = external global [100 x i32]
 define i32 @foo(i32 %x, i32 %z) !dbg !6 {

llvm/test/Transforms/LICM/hoisting-preheader-debugloc.ll

@@ -1,6 +1,6 @@
 ; RUN: opt -S -licm < %s | FileCheck %s
 
-; CHECK: %arrayidx4.promoted = load i32, i32* %arrayidx4, align 1, !tbaa !{{[0-9]+$}}
+; CHECK: %arrayidx4.promoted = load i32, i32* %arrayidx4, align 1, !tbaa !59
 
 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-linux-gnu"