diff --git a/llvm/lib/Transforms/Instrumentation/CFGMST.h b/llvm/lib/Transforms/Instrumentation/CFGMST.h
--- a/llvm/lib/Transforms/Instrumentation/CFGMST.h
+++ b/llvm/lib/Transforms/Instrumentation/CFGMST.h
@@ -20,6 +20,7 @@
 #include "llvm/Analysis/BranchProbabilityInfo.h"
 #include "llvm/Analysis/CFG.h"
 #include "llvm/Support/BranchProbability.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
@@ -28,6 +29,11 @@
 
 #define DEBUG_TYPE "cfgmst"
 
+using namespace llvm;
+static cl::opt<bool> PGOInstrumentEntry(
+    "pgo-instrument-entry", cl::init(false), cl::Hidden,
+    cl::desc("Force to instrument function entry basicblock."));
+
 namespace llvm {
 
 /// An union-find based Minimum Spanning Tree for CFG
@@ -100,8 +106,11 @@
 
     const BasicBlock *Entry = &(F.getEntryBlock());
     uint64_t EntryWeight = (BFI != nullptr ? BFI->getEntryFreq() : 2);
+    // If we want to instrument the entry count, lower the weight to 0.
+    if (PGOInstrumentEntry)
+      EntryWeight = 0;
     Edge *EntryIncoming = nullptr, *EntryOutgoing = nullptr,
-        *ExitOutgoing = nullptr, *ExitIncoming = nullptr;
+         *ExitOutgoing = nullptr, *ExitIncoming = nullptr;
     uint64_t MaxEntryOutWeight = 0, MaxExitOutWeight = 0, MaxExitInWeight = 0;
 
     // Add a fake edge to the entry.
@@ -135,6 +144,8 @@
           }
           if (BPI != nullptr)
             Weight = BPI->getEdgeProbability(&*BB, TargetBB).scale(scaleFactor);
+          if (Weight == 0)
+            Weight++;
           auto *E = &addEdge(&*BB, TargetBB, Weight);
           E->IsCritical = Critical;
           LLVM_DEBUG(dbgs() << "  Edge: from " << BB->getName() << " to "
@@ -278,6 +289,9 @@
     buildEdges();
     sortEdgesByWeight();
     computeMinimumSpanningTree();
+    if (PGOInstrumentEntry && (AllEdges.size() > 1))
+      std::iter_swap(std::move(AllEdges.begin()),
+                     std::move(AllEdges.begin() + AllEdges.size() - 1));
   }
 };
 
diff --git a/llvm/lib/Transforms/Instrumentation/InstrProfiling.cpp b/llvm/lib/Transforms/Instrumentation/InstrProfiling.cpp
--- a/llvm/lib/Transforms/Instrumentation/InstrProfiling.cpp
+++ b/llvm/lib/Transforms/Instrumentation/InstrProfiling.cpp
@@ -110,6 +110,12 @@
              " for promoted counters only"),
     cl::init(false));
 
+cl::opt<bool> AtomicFirstCounter(
+    "atomic-first-counter", cl::ZeroOrMore,
+    cl::desc("Use atomic fetch add for first counter in a function (usually "
+             "the entry counter)"),
+    cl::init(false));
+
 // If the option is not specified, the default behavior about whether
 // counter promotion is done depends on how instrumentaiton lowering
 // pipeline is setup, i.e., the default value of true of this option
@@ -696,7 +702,8 @@
     Addr = Builder.CreateIntToPtr(Add, Int64PtrTy);
   }
 
-  if (Options.Atomic || AtomicCounterUpdateAll) {
+  if (Options.Atomic || AtomicCounterUpdateAll ||
+      (Index == 0 && AtomicFirstCounter)) {
     Builder.CreateAtomicRMW(AtomicRMWInst::Add, Addr, Inc->getStep(),
                             AtomicOrdering::Monotonic);
   } else {
diff --git a/llvm/test/Transforms/PGOProfile/instr_entry_bb.ll b/llvm/test/Transforms/PGOProfile/instr_entry_bb.ll
new file mode 100644
--- /dev/null
+++ b/llvm/test/Transforms/PGOProfile/instr_entry_bb.ll
@@ -0,0 +1,50 @@
+; RUN: opt < %s -pgo-instr-gen -pgo-instrument-entry -S | FileCheck %s --check-prefix=GEN
+; RUN: opt < %s -passes=pgo-instr-gen -pgo-instrument-entry -S | FileCheck %s --check-prefix=GEN
+; RUN: opt < %s -pgo-instr-gen -pgo-instrument-entry -instrprof -atomic-first-counter -S | FileCheck %s --check-prefix=GENA
+; RUN: opt < %s -passes=pgo-instr-gen,instrprof -pgo-instrument-entry -atomic-first-counter -S | FileCheck %s --check-prefix=GENA
+
+; RUN: llvm-profdata merge %S/Inputs/branch2.proftext -o %t.profdata
+; RUN: opt < %s -pgo-instr-use -pgo-test-profile-file=%t.profdata -pgo-instrument-entry -S | FileCheck %s --check-prefix=USE
+; RUN: opt < %s -passes=pgo-instr-use -pgo-test-profile-file=%t.profdata -pgo-instrument-entry -S | FileCheck %s --check-prefix=USE
+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"
+
+; GEN: $__llvm_profile_raw_version = comdat any
+; GEN: @__llvm_profile_raw_version = constant i64 {{[0-9]+}}, comdat
+; GEN: @__profn_test_br_2 = private constant [9 x i8] c"test_br_2"
+
+define i32 @test_br_2(i32 %i) {
+entry:
+; GEN: entry:
+; GEN: call void @llvm.instrprof.increment(i8* getelementptr inbounds ([9 x i8], [9 x i8]* @__profn_test_br_2, i32 0, i32 0), i64 29667547796, i32 2, i32 0)
+; GENA: entry:
+; GENA: %{{[0-9+]}} = atomicrmw add i64* getelementptr inbounds ([2 x i64], [2 x i64]* @__profc_test_br_2, i64 0, i64 0), i64 1 monotonic
+; USE: br i1 %cmp, label %if.then, label %if.else
+; USE-SAME: !prof ![[BW_ENTRY:[0-9]+]]
+; USE: ![[BW_ENTRY]] = !{!"branch_weights", i32 0, i32 1}
+  %cmp = icmp sgt i32 %i, 0
+  br i1 %cmp, label %if.then, label %if.else
+
+if.then:
+; GEN: if.then:
+; GEN-NOT: llvm.instrprof.increment
+  %add = add nsw i32 %i, 2
+  br label %if.end
+
+if.else:
+; GEN: if.else:
+; GEN: call void @llvm.instrprof.increment(i8* getelementptr inbounds ([9 x i8], [9 x i8]* @__profn_test_br_2, i32 0, i32 0), i64 29667547796, i32 2, i32 1)
+; GENA: if.else:
+; GENA:  %pgocount = load i64, i64* getelementptr inbounds ([2 x i64], [2 x i64]* @__profc_test_br_2, i64 0, i64 1), align 8
+; GENA:  [[V:%[0-9]*]] = add i64 %pgocount, 1
+; GENA:  store i64 [[V]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @__profc_test_br_2, i64 0, i64 1), align 8
+  %sub = sub nsw i32 %i, 2
+  br label %if.end
+
+if.end:
+; GEN: if.end:
+; GEN-NOT: llvm.instrprof.increment
+  %retv = phi i32 [ %add, %if.then ], [ %sub, %if.else ]
+  ret i32 %retv
+; GEN: ret
+}