[ProfileSummary] Count callsite samples when computing total samples.

[twoh]

Summary

Currently ProfileSummaryBuilder doesn't count into callsite samples when computing total samples. Considering that ProfileSummaryInfo is used to checked the hotness of not only body samples but also callsite samples (from SampleProfileLoader), I think the callsite sample counts should be considered when computing total samples.

[wmi]

That is a reasonable change to me. Thanks. However it probably needs some careful performance evaluation since some tuning may be needed for that change. I will do that in our benchmarks. How does this change on performance look like on your side?

[twoh]

@wmi Thanks for taking a look! Performance numbers were neutral from our side.

[wmi]

Did the test. No major improvement or regression was found. LGTM.

llvm/trunk/include/llvm/ProfileData/ProfileCommon.h

public:
   std::unique_ptr<ProfileSummary> getSummary();
 };
 
 class SampleProfileSummaryBuilder final : public ProfileSummaryBuilder {
 public:
   SampleProfileSummaryBuilder(std::vector<uint32_t> Cutoffs)
       : ProfileSummaryBuilder(std::move(Cutoffs)) {}
 
-  void addRecord(const sampleprof::FunctionSamples &FS);
+  void addRecord(const sampleprof::FunctionSamples &FS,
+                 bool isCallsiteSample = false);
   std::unique_ptr<ProfileSummary> getSummary();
 };
 
 /// This is called when a count is seen in the profile.
 void ProfileSummaryBuilder::addCount(uint64_t Count) {
   TotalCount += Count;
   if (Count > MaxCount)
     MaxCount = Count;
   NumCounts++;
   CountFrequencies[Count]++;
 }
 
 } // end namespace llvm
 
 #endif // LLVM_PROFILEDATA_PROFILECOMMON_H

llvm/trunk/lib/ProfileData/ProfileSummaryBuilder.cpp

void InstrProfSummaryBuilder::addRecord(const InstrProfRecord &R) {
   addEntryCount(R.Counts[0]);
   for (size_t I = 1, E = R.Counts.size(); I < E; ++I)
     addInternalCount(R.Counts[I]);
 }
 
 // To compute the detailed summary, we consider each line containing samples as
 // equivalent to a block with a count in the instrumented profile.
 void SampleProfileSummaryBuilder::addRecord(
-    const sampleprof::FunctionSamples &FS) {
+    const sampleprof::FunctionSamples &FS, bool isCallsiteSample) {
+  if (!isCallsiteSample) {
     NumFunctions++;
     if (FS.getHeadSamples() > MaxFunctionCount)
       MaxFunctionCount = FS.getHeadSamples();
+  }
   for (const auto &I : FS.getBodySamples())
     addCount(I.second.getSamples());
+  for (const auto &I : FS.getCallsiteSamples())
+    for (const auto &CS : I.second)
+      addRecord(CS.second, true);
 }
 
 // The argument to this method is a vector of cutoff percentages and the return
 // value is a vector of (Cutoff, MinCount, NumCounts) triplets.
 void ProfileSummaryBuilder::computeDetailedSummary() {
   if (DetailedSummaryCutoffs.empty())
     return;
   llvm::sort(DetailedSummaryCutoffs);

llvm/trunk/test/Transforms/SampleProfile/Inputs/summary.prof

 bar:100:3
  1: 100
 foo:200:1
  1: 200
+baz:600:1
+ 1: 0
+ 2: 300
+ 1: bar:300
+  1: 300

llvm/trunk/test/Transforms/SampleProfile/compact-binary-profile.ll

 ;
 @.str = private unnamed_addr constant [11 x i8] c"sum is %d\0A\00", align 1
 
 ; Check sample-profile phase using compactbinary format profile will annotate
 ; the IR with exactly the same result as using text format.
 ; CHECK: br i1 %cmp, label %while.body, label %while.end{{.*}} !prof ![[IDX1:[0-9]*]]
 ; CHECK: br i1 %cmp1, label %if.then, label %if.else{{.*}} !prof ![[IDX2:[0-9]*]]
 ; CHECK: call i32 (i8*, ...) @printf{{.*}} !prof ![[IDX3:[0-9]*]]
-; CHECK: = !{!"TotalCount", i64 10944}
+; CHECK: = !{!"TotalCount", i64 26781}
 ; CHECK: = !{!"MaxCount", i64 5553}
 ; CHECK: ![[IDX1]] = !{!"branch_weights", i32 5392, i32 163}
 ; CHECK: ![[IDX2]] = !{!"branch_weights", i32 5280, i32 113}
 ; CHECK: ![[IDX3]] = !{!"branch_weights", i32 1}
 
 ; Function Attrs: nounwind uwtable
 define i32 @_Z3sumii(i32 %x, i32 %y) !dbg !4 {
 entry:

llvm/trunk/test/Transforms/SampleProfile/summary.ll

 ; Test that we annotate entire program's summary to IR.
 ; RUN: opt < %s -sample-profile -sample-profile-file=%S/Inputs/summary.prof -S | FileCheck %s
 ; RUN: opt < %s -sample-profile -sample-profile-file=%S/Inputs/summary.prof -S | opt -sample-profile -sample-profile-file=%S/Inputs/summary.prof -S | FileCheck %s
 
 define i32 @bar() #0 !dbg !1 {
 entry:
   ret i32 1, !dbg !2
 }
 
+define i32 @baz() !dbg !3 {
+entry:
+    %call = call i32 @bar(), !dbg !4
+    ret i32 %call, !dbg !5
+}
+
 ; CHECK-DAG: {{![0-9]+}} = !{i32 1, !"ProfileSummary", {{![0-9]+}}}
-; CHECK-DAG: {{![0-9]+}} = !{!"NumFunctions", i64 2}
+; CHECK-DAG: {{![0-9]+}} = !{!"TotalCount", i64 900}
+; CHECK-DAG: {{![0-9]+}} = !{!"NumCounts", i64 5}
+; CHECK-DAG: {{![0-9]+}} = !{!"NumFunctions", i64 3}
 ; CHECK-DAG: {{![0-9]+}} = !{!"MaxFunctionCount", i64 3}
 
 !1 = distinct !DISubprogram(name: "bar")
 !2 = !DILocation(line: 2, scope: !2)
+!3 = distinct !DISubprogram(name: "baz")
+!4 = !DILocation(line: 1, scope: !4)
+!5 = !DILocation(line: 2, scope: !5)