From fb4bcc452c0cfba85b7e5b177727eb8c8f966c58 Mon Sep 17 00:00:00 2001
From: Rong Xu <xur@google.com>
Date: Wed, 7 Nov 2018 23:51:20 +0000
Subject: [PATCH] [PGO] Exit early if all count values are zero

If all the edge counts for a function are zero, skip count population and
annotation, as nothing will happen. This can save some compile time.

Differential Revision: https://reviews.llvm.org/D54212

llvm-svn: 346370
---
 .../Instrumentation/PGOInstrumentation.cpp    | 15 ++++++++--
 .../PGOProfile/Inputs/func_entry.proftext     | 17 +++++++++++
 llvm/test/Transforms/PGOProfile/func_entry.ll | 29 +++++++++++++++++++
 3 files changed, 58 insertions(+), 3 deletions(-)
 create mode 100644 llvm/test/Transforms/PGOProfile/Inputs/func_entry.proftext
 create mode 100644 llvm/test/Transforms/PGOProfile/func_entry.ll

diff --git a/llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp b/llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
index 4790c9e5cfed..876ae23dfd29 100644
--- a/llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
+++ b/llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
@@ -859,7 +859,7 @@ public:
         FreqAttr(FFA_Normal) {}
 
   // Read counts for the instrumented BB from profile.
-  bool readCounters(IndexedInstrProfReader *PGOReader);
+  bool readCounters(IndexedInstrProfReader *PGOReader, bool &AllZeros);
 
   // Populate the counts for all BBs.
   void populateCounters();
@@ -904,6 +904,7 @@ public:
     FuncInfo.dumpInfo(Str);
   }
 
+  uint64_t getProgramMaxCount() const { return ProgramMaxCount; }
 private:
   Function &F;
   Module *M;
@@ -1013,7 +1014,7 @@ void PGOUseFunc::setEdgeCount(DirectEdges &Edges, uint64_t Value) {
 // Read the profile from ProfileFileName and assign the value to the
 // instrumented BB and the edges. This function also updates ProgramMaxCount.
 // Return true if the profile are successfully read, and false on errors.
-bool PGOUseFunc::readCounters(IndexedInstrProfReader *PGOReader) {
+bool PGOUseFunc::readCounters(IndexedInstrProfReader *PGOReader, bool &AllZeros) {
   auto &Ctx = M->getContext();
   Expected<InstrProfRecord> Result =
       PGOReader->getInstrProfRecord(FuncInfo.FuncName, FuncInfo.FunctionHash);
@@ -1053,6 +1054,7 @@ bool PGOUseFunc::readCounters(IndexedInstrProfReader *PGOReader) {
     LLVM_DEBUG(dbgs() << "  " << I << ": " << CountFromProfile[I] << "\n");
     ValueSum += CountFromProfile[I];
   }
+  AllZeros = (ValueSum == 0);
 
   LLVM_DEBUG(dbgs() << "SUM =  " << ValueSum << "\n");
 
@@ -1477,8 +1479,15 @@ static bool annotateAllFunctions(
     // later in getInstrBB() to avoid invalidating it.
     SplitIndirectBrCriticalEdges(F, BPI, BFI);
     PGOUseFunc Func(F, &M, ComdatMembers, BPI, BFI);
-    if (!Func.readCounters(PGOReader.get()))
+    bool AllZeros = false;
+    if (!Func.readCounters(PGOReader.get(), AllZeros))
       continue;
+    if (AllZeros) {
+      F.setEntryCount(ProfileCount(0, Function::PCT_Real));
+      if (Func.getProgramMaxCount() != 0)
+        ColdFunctions.push_back(&F);
+      continue;
+    }
     Func.populateCounters();
     Func.setBranchWeights();
     Func.annotateValueSites();
diff --git a/llvm/test/Transforms/PGOProfile/Inputs/func_entry.proftext b/llvm/test/Transforms/PGOProfile/Inputs/func_entry.proftext
new file mode 100644
index 000000000000..2dc2c2ec9f36
--- /dev/null
+++ b/llvm/test/Transforms/PGOProfile/Inputs/func_entry.proftext
@@ -0,0 +1,17 @@
+# IR level Instrumentation Flag
+:ir
+foo
+# Func Hash:
+12884901887
+# Num Counters:
+1
+# Counter Values:
+9999
+
+bar
+# Func Hash:
+12884901887
+# Num Counters:
+1
+# Counter Values:
+0
diff --git a/llvm/test/Transforms/PGOProfile/func_entry.ll b/llvm/test/Transforms/PGOProfile/func_entry.ll
new file mode 100644
index 000000000000..dac996e35cb3
--- /dev/null
+++ b/llvm/test/Transforms/PGOProfile/func_entry.ll
@@ -0,0 +1,29 @@
+; RUN: llvm-profdata merge %S/Inputs/func_entry.proftext -o %t.profdata
+; RUN: opt < %s -passes=pgo-instr-use -pgo-test-profile-file=%t.profdata -S | FileCheck %s
+
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+@s = common dso_local local_unnamed_addr global i32 0, align 4
+
+define void @bar() {
+; CHECK-LABEL: @bar
+; CHECK-SAME: !prof ![[FUNC_ENTRY_COUNT_ZERO:[0-9]+]]
+
+entry:
+  store i32 1, i32* @s, align 4
+  ret void
+}
+
+define void @foo() {
+; CHECK-LABEL: @foo
+; CHECK-SAME: !prof ![[FUNC_ENTRY_COUNT_NON_ZERO:[0-9]+]]
+entry:
+  %0 = load i32, i32* @s, align 4
+  %add = add nsw i32 %0, 4
+  store i32 %add, i32* @s, align 4
+  ret void
+}
+
+; CHECK-DAG: ![[FUNC_ENTRY_COUNT_ZERO]] = !{!"function_entry_count", i64 0}
+; CHECK-DAG: ![[FUNC_ENTRY_COUNT_NON_ZERO]] = !{!"function_entry_count", i64 9999}