diff --git a/clang/lib/CodeGen/CGExprAgg.cpp b/clang/lib/CodeGen/CGExprAgg.cpp
index d0d6202974fe9..5190b22bcc162 100644
--- a/clang/lib/CodeGen/CGExprAgg.cpp
+++ b/clang/lib/CodeGen/CGExprAgg.cpp
@@ -33,6 +33,10 @@ using namespace CodeGen;
 //                        Aggregate Expression Emitter
 //===----------------------------------------------------------------------===//
 
+namespace llvm {
+extern cl::opt<bool> EnableSingleByteCoverage;
+} // namespace llvm
+
 namespace  {
 class AggExprEmitter : public StmtVisitor<AggExprEmitter> {
   CodeGenFunction &CGF;
@@ -1279,7 +1283,10 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
 
   eval.begin(CGF);
   CGF.EmitBlock(LHSBlock);
-  CGF.incrementProfileCounter(E);
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(E->getTrueExpr());
+  else
+    CGF.incrementProfileCounter(E);
   Visit(E->getTrueExpr());
   eval.end(CGF);
 
@@ -1294,6 +1301,8 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
 
   eval.begin(CGF);
   CGF.EmitBlock(RHSBlock);
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(E->getFalseExpr());
   Visit(E->getFalseExpr());
   eval.end(CGF);
 
@@ -1302,6 +1311,8 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
                     E->getType());
 
   CGF.EmitBlock(ContBlock);
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(E);
 }
 
 void AggExprEmitter::VisitChooseExpr(const ChooseExpr *CE) {
diff --git a/clang/lib/CodeGen/CGExprComplex.cpp b/clang/lib/CodeGen/CGExprComplex.cpp
index 176a7e00141f9..0266ba934da62 100644
--- a/clang/lib/CodeGen/CGExprComplex.cpp
+++ b/clang/lib/CodeGen/CGExprComplex.cpp
@@ -28,6 +28,10 @@ using namespace CodeGen;
 //                        Complex Expression Emitter
 //===----------------------------------------------------------------------===//
 
+namespace llvm {
+extern cl::opt<bool> EnableSingleByteCoverage;
+} // namespace llvm
+
 typedef CodeGenFunction::ComplexPairTy ComplexPairTy;
 
 /// Return the complex type that we are meant to emit.
@@ -1330,7 +1334,11 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
 
   eval.begin(CGF);
   CGF.EmitBlock(LHSBlock);
-  CGF.incrementProfileCounter(E);
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(E->getTrueExpr());
+  else
+    CGF.incrementProfileCounter(E);
+
   ComplexPairTy LHS = Visit(E->getTrueExpr());
   LHSBlock = Builder.GetInsertBlock();
   CGF.EmitBranch(ContBlock);
@@ -1338,9 +1346,13 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
 
   eval.begin(CGF);
   CGF.EmitBlock(RHSBlock);
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(E->getFalseExpr());
   ComplexPairTy RHS = Visit(E->getFalseExpr());
   RHSBlock = Builder.GetInsertBlock();
   CGF.EmitBlock(ContBlock);
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(E);
   eval.end(CGF);
 
   // Create a PHI node for the real part.
diff --git a/clang/lib/CodeGen/CGExprScalar.cpp b/clang/lib/CodeGen/CGExprScalar.cpp
index 10b7457522044..8536570087ad0 100644
--- a/clang/lib/CodeGen/CGExprScalar.cpp
+++ b/clang/lib/CodeGen/CGExprScalar.cpp
@@ -52,6 +52,10 @@ using llvm::Value;
 //                         Scalar Expression Emitter
 //===----------------------------------------------------------------------===//
 
+namespace llvm {
+extern cl::opt<bool> EnableSingleByteCoverage;
+} // namespace llvm
+
 namespace {
 
 /// Determine whether the given binary operation may overflow.
@@ -4925,8 +4929,13 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
 
     // If the dead side doesn't have labels we need, just emit the Live part.
     if (!CGF.ContainsLabel(dead)) {
-      if (CondExprBool)
+      if (CondExprBool) {
+        if (llvm::EnableSingleByteCoverage) {
+          CGF.incrementProfileCounter(lhsExpr);
+          CGF.incrementProfileCounter(rhsExpr);
+        }
         CGF.incrementProfileCounter(E);
+      }
       Value *Result = Visit(live);
 
       // If the live part is a throw expression, it acts like it has a void
@@ -5005,7 +5014,12 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
     llvm::Value *CondV = CGF.EvaluateExprAsBool(condExpr);
     llvm::Value *StepV = Builder.CreateZExtOrBitCast(CondV, CGF.Int64Ty);
 
-    CGF.incrementProfileCounter(E, StepV);
+    if (llvm::EnableSingleByteCoverage) {
+      CGF.incrementProfileCounter(lhsExpr);
+      CGF.incrementProfileCounter(rhsExpr);
+      CGF.incrementProfileCounter(E);
+    } else
+      CGF.incrementProfileCounter(E, StepV);
 
     llvm::Value *LHS = Visit(lhsExpr);
     llvm::Value *RHS = Visit(rhsExpr);
@@ -5037,7 +5051,11 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
   if (CGF.MCDCLogOpStack.empty())
     CGF.maybeUpdateMCDCTestVectorBitmap(condExpr);
 
-  CGF.incrementProfileCounter(E);
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(lhsExpr);
+  else
+    CGF.incrementProfileCounter(E);
+
   eval.begin(CGF);
   Value *LHS = Visit(lhsExpr);
   eval.end(CGF);
@@ -5053,6 +5071,9 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
   if (CGF.MCDCLogOpStack.empty())
     CGF.maybeUpdateMCDCTestVectorBitmap(condExpr);
 
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(rhsExpr);
+
   eval.begin(CGF);
   Value *RHS = Visit(rhsExpr);
   eval.end(CGF);
@@ -5071,6 +5092,11 @@ VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) {
   PN->addIncoming(LHS, LHSBlock);
   PN->addIncoming(RHS, RHSBlock);
 
+  // When single byte coverage mode is enabled, add a counter to continuation
+  // block.
+  if (llvm::EnableSingleByteCoverage)
+    CGF.incrementProfileCounter(E);
+
   return PN;
 }
 
diff --git a/clang/lib/CodeGen/CGStmt.cpp b/clang/lib/CodeGen/CGStmt.cpp
index af51875782c9f..d0a3a716ad75e 100644
--- a/clang/lib/CodeGen/CGStmt.cpp
+++ b/clang/lib/CodeGen/CGStmt.cpp
@@ -43,6 +43,10 @@ using namespace CodeGen;
 //                              Statement Emission
 //===----------------------------------------------------------------------===//
 
+namespace llvm {
+extern cl::opt<bool> EnableSingleByteCoverage;
+} // namespace llvm
+
 void CodeGenFunction::EmitStopPoint(const Stmt *S) {
   if (CGDebugInfo *DI = getDebugInfo()) {
     SourceLocation Loc;
@@ -856,7 +860,10 @@ void CodeGenFunction::EmitIfStmt(const IfStmt &S) {
 
   // Emit the 'then' code.
   EmitBlock(ThenBlock);
-  incrementProfileCounter(&S);
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(S.getThen());
+  else
+    incrementProfileCounter(&S);
   {
     RunCleanupsScope ThenScope(*this);
     EmitStmt(S.getThen());
@@ -870,6 +877,9 @@ void CodeGenFunction::EmitIfStmt(const IfStmt &S) {
       auto NL = ApplyDebugLocation::CreateEmpty(*this);
       EmitBlock(ElseBlock);
     }
+    // When single byte coverage mode is enabled, add a counter to else block.
+    if (llvm::EnableSingleByteCoverage)
+      incrementProfileCounter(Else);
     {
       RunCleanupsScope ElseScope(*this);
       EmitStmt(Else);
@@ -883,6 +893,11 @@ void CodeGenFunction::EmitIfStmt(const IfStmt &S) {
 
   // Emit the continuation block for code after the if.
   EmitBlock(ContBlock, true);
+
+  // When single byte coverage mode is enabled, add a counter to continuation
+  // block.
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(&S);
 }
 
 void CodeGenFunction::EmitWhileStmt(const WhileStmt &S,
@@ -927,6 +942,10 @@ void CodeGenFunction::EmitWhileStmt(const WhileStmt &S,
                  SourceLocToDebugLoc(R.getEnd()),
                  checkIfLoopMustProgress(CondIsConstInt));
 
+  // When single byte coverage mode is enabled, add a counter to loop condition.
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(S.getCond());
+
   // As long as the condition is true, go to the loop body.
   llvm::BasicBlock *LoopBody = createBasicBlock("while.body");
   if (EmitBoolCondBranch) {
@@ -959,7 +978,11 @@ void CodeGenFunction::EmitWhileStmt(const WhileStmt &S,
   {
     RunCleanupsScope BodyScope(*this);
     EmitBlock(LoopBody);
-    incrementProfileCounter(&S);
+    // When single byte coverage mode is enabled, add a counter to the body.
+    if (llvm::EnableSingleByteCoverage)
+      incrementProfileCounter(S.getBody());
+    else
+      incrementProfileCounter(&S);
     EmitStmt(S.getBody());
   }
 
@@ -981,6 +1004,11 @@ void CodeGenFunction::EmitWhileStmt(const WhileStmt &S,
   // a branch, try to erase it.
   if (!EmitBoolCondBranch)
     SimplifyForwardingBlocks(LoopHeader.getBlock());
+
+  // When single byte coverage mode is enabled, add a counter to continuation
+  // block.
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(&S);
 }
 
 void CodeGenFunction::EmitDoStmt(const DoStmt &S,
@@ -996,13 +1024,19 @@ void CodeGenFunction::EmitDoStmt(const DoStmt &S,
   // Emit the body of the loop.
   llvm::BasicBlock *LoopBody = createBasicBlock("do.body");
 
-  EmitBlockWithFallThrough(LoopBody, &S);
+  if (llvm::EnableSingleByteCoverage)
+    EmitBlockWithFallThrough(LoopBody, S.getBody());
+  else
+    EmitBlockWithFallThrough(LoopBody, &S);
   {
     RunCleanupsScope BodyScope(*this);
     EmitStmt(S.getBody());
   }
 
   EmitBlock(LoopCond.getBlock());
+  // When single byte coverage mode is enabled, add a counter to loop condition.
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(S.getCond());
 
   // C99 6.8.5.2: "The evaluation of the controlling expression takes place
   // after each execution of the loop body."
@@ -1043,6 +1077,11 @@ void CodeGenFunction::EmitDoStmt(const DoStmt &S,
   // emitting a branch, try to erase it.
   if (!EmitBoolCondBranch)
     SimplifyForwardingBlocks(LoopCond.getBlock());
+
+  // When single byte coverage mode is enabled, add a counter to continuation
+  // block.
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(&S);
 }
 
 void CodeGenFunction::EmitForStmt(const ForStmt &S,
@@ -1101,6 +1140,11 @@ void CodeGenFunction::EmitForStmt(const ForStmt &S,
       BreakContinueStack.back().ContinueBlock = Continue;
     }
 
+    // When single byte coverage mode is enabled, add a counter to loop
+    // condition.
+    if (llvm::EnableSingleByteCoverage)
+      incrementProfileCounter(S.getCond());
+
     llvm::BasicBlock *ExitBlock = LoopExit.getBlock();
     // If there are any cleanups between here and the loop-exit scope,
     // create a block to stage a loop exit along.
@@ -1131,8 +1175,12 @@ void CodeGenFunction::EmitForStmt(const ForStmt &S,
     // Treat it as a non-zero constant.  Don't even create a new block for the
     // body, just fall into it.
   }
-  incrementProfileCounter(&S);
 
+  // When single byte coverage mode is enabled, add a counter to the body.
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(S.getBody());
+  else
+    incrementProfileCounter(&S);
   {
     // Create a separate cleanup scope for the body, in case it is not
     // a compound statement.
@@ -1144,6 +1192,8 @@ void CodeGenFunction::EmitForStmt(const ForStmt &S,
   if (S.getInc()) {
     EmitBlock(Continue.getBlock());
     EmitStmt(S.getInc());
+    if (llvm::EnableSingleByteCoverage)
+      incrementProfileCounter(S.getInc());
   }
 
   BreakContinueStack.pop_back();
@@ -1159,6 +1209,11 @@ void CodeGenFunction::EmitForStmt(const ForStmt &S,
 
   // Emit the fall-through block.
   EmitBlock(LoopExit.getBlock(), true);
+
+  // When single byte coverage mode is enabled, add a counter to continuation
+  // block.
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(&S);
 }
 
 void
@@ -1211,7 +1266,10 @@ CodeGenFunction::EmitCXXForRangeStmt(const CXXForRangeStmt &S,
   }
 
   EmitBlock(ForBody);
-  incrementProfileCounter(&S);
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(S.getBody());
+  else
+    incrementProfileCounter(&S);
 
   // Create a block for the increment. In case of a 'continue', we jump there.
   JumpDest Continue = getJumpDestInCurrentScope("for.inc");
@@ -1241,6 +1299,11 @@ CodeGenFunction::EmitCXXForRangeStmt(const CXXForRangeStmt &S,
 
   // Emit the fall-through block.
   EmitBlock(LoopExit.getBlock(), true);
+
+  // When single byte coverage mode is enabled, add a counter to continuation
+  // block.
+  if (llvm::EnableSingleByteCoverage)
+    incrementProfileCounter(&S);
 }
 
 void CodeGenFunction::EmitReturnOfRValue(RValue RV, QualType Ty) {
diff --git a/clang/lib/CodeGen/CodeGenFunction.cpp b/clang/lib/CodeGen/CodeGenFunction.cpp
index 1ad905078d349..b87fc86f4e635 100644
--- a/clang/lib/CodeGen/CodeGenFunction.cpp
+++ b/clang/lib/CodeGen/CodeGenFunction.cpp
@@ -52,6 +52,10 @@
 using namespace clang;
 using namespace CodeGen;
 
+namespace llvm {
+extern cl::opt<bool> EnableSingleByteCoverage;
+} // namespace llvm
+
 /// shouldEmitLifetimeMarkers - Decide whether we need emit the life-time
 /// markers.
 static bool shouldEmitLifetimeMarkers(const CodeGenOptions &CGOpts,
@@ -1270,7 +1274,10 @@ void CodeGenFunction::EmitFunctionBody(const Stmt *Body) {
 void CodeGenFunction::EmitBlockWithFallThrough(llvm::BasicBlock *BB,
                                                const Stmt *S) {
   llvm::BasicBlock *SkipCountBB = nullptr;
-  if (HaveInsertPoint() && CGM.getCodeGenOpts().hasProfileClangInstr()) {
+  // Do not skip over the instrumentation when single byte coverage mode is
+  // enabled.
+  if (HaveInsertPoint() && CGM.getCodeGenOpts().hasProfileClangInstr() &&
+      !llvm::EnableSingleByteCoverage) {
     // When instrumenting for profiling, the fallthrough to certain
     // statements needs to skip over the instrumentation code so that we
     // get an accurate count.
diff --git a/clang/lib/CodeGen/CodeGenFunction.h b/clang/lib/CodeGen/CodeGenFunction.h
index b2800f699ff4b..06327a1847177 100644
--- a/clang/lib/CodeGen/CodeGenFunction.h
+++ b/clang/lib/CodeGen/CodeGenFunction.h
@@ -1545,7 +1545,7 @@ class CodeGenFunction : public CodeGenTypeCache {
     if (CGM.getCodeGenOpts().hasProfileClangInstr() &&
         !CurFn->hasFnAttribute(llvm::Attribute::NoProfile) &&
         !CurFn->hasFnAttribute(llvm::Attribute::SkipProfile))
-      PGO.emitCounterIncrement(Builder, S, StepV);
+      PGO.emitCounterSetOrIncrement(Builder, S, StepV);
     PGO.setCurrentStmt(S);
   }
 
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index 95e457bef28ed..1550b000a89a3 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -858,6 +858,7 @@ void CodeGenModule::Release() {
   checkAliases();
   EmitDeferredUnusedCoverageMappings();
   CodeGenPGO(*this).setValueProfilingFlag(getModule());
+  CodeGenPGO(*this).setProfileVersion(getModule());
   if (CoverageMapping)
     CoverageMapping->emit();
   if (CodeGenOpts.SanitizeCfiCrossDso) {
diff --git a/clang/lib/CodeGen/CodeGenPGO.cpp b/clang/lib/CodeGen/CodeGenPGO.cpp
index 8aebd3557690a..2619edfeb7dc7 100644
--- a/clang/lib/CodeGen/CodeGenPGO.cpp
+++ b/clang/lib/CodeGen/CodeGenPGO.cpp
@@ -23,6 +23,10 @@
 #include "llvm/Support/MD5.h"
 #include <optional>
 
+namespace llvm {
+extern cl::opt<bool> EnableSingleByteCoverage;
+} // namespace llvm
+
 static llvm::cl::opt<bool>
     EnableValueProfiling("enable-value-profiling",
                          llvm::cl::desc("Enable value profiling"),
@@ -346,6 +350,14 @@ struct MapRegionCounters : public RecursiveASTVisitor<MapRegionCounters> {
     return Base::VisitBinaryOperator(S);
   }
 
+  bool VisitConditionalOperator(ConditionalOperator *S) {
+    if (llvm::EnableSingleByteCoverage && S->getTrueExpr())
+      CounterMap[S->getTrueExpr()] = NextCounter++;
+    if (llvm::EnableSingleByteCoverage && S->getFalseExpr())
+      CounterMap[S->getFalseExpr()] = NextCounter++;
+    return Base::VisitConditionalOperator(S);
+  }
+
   /// Include \p S in the function hash.
   bool VisitStmt(Stmt *S) {
     auto Type = updateCounterMappings(S);
@@ -361,8 +373,21 @@ struct MapRegionCounters : public RecursiveASTVisitor<MapRegionCounters> {
     if (Hash.getHashVersion() == PGO_HASH_V1)
       return Base::TraverseIfStmt(If);
 
+    // When single byte coverage mode is enabled, add a counter to then and
+    // else.
+    bool NoSingleByteCoverage = !llvm::EnableSingleByteCoverage;
+    for (Stmt *CS : If->children()) {
+      if (!CS || NoSingleByteCoverage)
+        continue;
+      if (CS == If->getThen())
+        CounterMap[If->getThen()] = NextCounter++;
+      else if (CS == If->getElse())
+        CounterMap[If->getElse()] = NextCounter++;
+    }
+
     // Otherwise, keep track of which branch we're in while traversing.
     VisitStmt(If);
+
     for (Stmt *CS : If->children()) {
       if (!CS)
         continue;
@@ -376,6 +401,81 @@ struct MapRegionCounters : public RecursiveASTVisitor<MapRegionCounters> {
     return true;
   }
 
+  bool TraverseWhileStmt(WhileStmt *While) {
+    // When single byte coverage mode is enabled, add a counter to condition and
+    // body.
+    bool NoSingleByteCoverage = !llvm::EnableSingleByteCoverage;
+    for (Stmt *CS : While->children()) {
+      if (!CS || NoSingleByteCoverage)
+        continue;
+      if (CS == While->getCond())
+        CounterMap[While->getCond()] = NextCounter++;
+      else if (CS == While->getBody())
+        CounterMap[While->getBody()] = NextCounter++;
+    }
+
+    Base::TraverseWhileStmt(While);
+    if (Hash.getHashVersion() != PGO_HASH_V1)
+      Hash.combine(PGOHash::EndOfScope);
+    return true;
+  }
+
+  bool TraverseDoStmt(DoStmt *Do) {
+    // When single byte coverage mode is enabled, add a counter to condition and
+    // body.
+    bool NoSingleByteCoverage = !llvm::EnableSingleByteCoverage;
+    for (Stmt *CS : Do->children()) {
+      if (!CS || NoSingleByteCoverage)
+        continue;
+      if (CS == Do->getCond())
+        CounterMap[Do->getCond()] = NextCounter++;
+      else if (CS == Do->getBody())
+        CounterMap[Do->getBody()] = NextCounter++;
+    }
+
+    Base::TraverseDoStmt(Do);
+    if (Hash.getHashVersion() != PGO_HASH_V1)
+      Hash.combine(PGOHash::EndOfScope);
+    return true;
+  }
+
+  bool TraverseForStmt(ForStmt *For) {
+    // When single byte coverage mode is enabled, add a counter to condition,
+    // increment and body.
+    bool NoSingleByteCoverage = !llvm::EnableSingleByteCoverage;
+    for (Stmt *CS : For->children()) {
+      if (!CS || NoSingleByteCoverage)
+        continue;
+      if (CS == For->getCond())
+        CounterMap[For->getCond()] = NextCounter++;
+      else if (CS == For->getInc())
+        CounterMap[For->getInc()] = NextCounter++;
+      else if (CS == For->getBody())
+        CounterMap[For->getBody()] = NextCounter++;
+    }
+
+    Base::TraverseForStmt(For);
+    if (Hash.getHashVersion() != PGO_HASH_V1)
+      Hash.combine(PGOHash::EndOfScope);
+    return true;
+  }
+
+  bool TraverseCXXForRangeStmt(CXXForRangeStmt *ForRange) {
+    // When single byte coverage mode is enabled, add a counter to body.
+    bool NoSingleByteCoverage = !llvm::EnableSingleByteCoverage;
+    for (Stmt *CS : ForRange->children()) {
+      if (!CS || NoSingleByteCoverage)
+        continue;
+      if (CS == ForRange->getBody())
+        CounterMap[ForRange->getBody()] = NextCounter++;
+    }
+
+    Base::TraverseCXXForRangeStmt(ForRange);
+    if (Hash.getHashVersion() != PGO_HASH_V1)
+      Hash.combine(PGOHash::EndOfScope);
+    return true;
+  }
+
 // If the statement type \p N is nestable, and its nesting impacts profile
 // stability, define a custom traversal which tracks the end of the statement
 // in the hash (provided we're not using the V1 hash).
@@ -387,10 +487,6 @@ struct MapRegionCounters : public RecursiveASTVisitor<MapRegionCounters> {
     return true;                                                               \
   }
 
-  DEFINE_NESTABLE_TRAVERSAL(WhileStmt)
-  DEFINE_NESTABLE_TRAVERSAL(DoStmt)
-  DEFINE_NESTABLE_TRAVERSAL(ForStmt)
-  DEFINE_NESTABLE_TRAVERSAL(CXXForRangeStmt)
   DEFINE_NESTABLE_TRAVERSAL(ObjCForCollectionStmt)
   DEFINE_NESTABLE_TRAVERSAL(CXXTryStmt)
   DEFINE_NESTABLE_TRAVERSAL(CXXCatchStmt)
@@ -1094,8 +1190,8 @@ CodeGenPGO::applyFunctionAttributes(llvm::IndexedInstrProfReader *PGOReader,
   Fn->setEntryCount(FunctionCount);
 }
 
-void CodeGenPGO::emitCounterIncrement(CGBuilderTy &Builder, const Stmt *S,
-                                      llvm::Value *StepV) {
+void CodeGenPGO::emitCounterSetOrIncrement(CGBuilderTy &Builder, const Stmt *S,
+                                           llvm::Value *StepV) {
   if (!RegionCounterMap || !Builder.GetInsertBlock())
     return;
 
@@ -1105,13 +1201,19 @@ void CodeGenPGO::emitCounterIncrement(CGBuilderTy &Builder, const Stmt *S,
                          Builder.getInt64(FunctionHash),
                          Builder.getInt32(NumRegionCounters),
                          Builder.getInt32(Counter), StepV};
-  if (!StepV)
-    Builder.CreateCall(CGM.getIntrinsic(llvm::Intrinsic::instrprof_increment),
+
+  if (llvm::EnableSingleByteCoverage)
+    Builder.CreateCall(CGM.getIntrinsic(llvm::Intrinsic::instrprof_cover),
                        ArrayRef(Args, 4));
-  else
-    Builder.CreateCall(
-        CGM.getIntrinsic(llvm::Intrinsic::instrprof_increment_step),
-        ArrayRef(Args));
+  else {
+    if (!StepV)
+      Builder.CreateCall(CGM.getIntrinsic(llvm::Intrinsic::instrprof_increment),
+                         ArrayRef(Args, 4));
+    else
+      Builder.CreateCall(
+          CGM.getIntrinsic(llvm::Intrinsic::instrprof_increment_step),
+          ArrayRef(Args));
+  }
 }
 
 bool CodeGenPGO::canEmitMCDCCoverage(const CGBuilderTy &Builder) {
@@ -1222,6 +1324,30 @@ void CodeGenPGO::setValueProfilingFlag(llvm::Module &M) {
                     uint32_t(EnableValueProfiling));
 }
 
+void CodeGenPGO::setProfileVersion(llvm::Module &M) {
+  if (CGM.getCodeGenOpts().hasProfileClangInstr() &&
+      llvm::EnableSingleByteCoverage) {
+    const StringRef VarName(INSTR_PROF_QUOTE(INSTR_PROF_RAW_VERSION_VAR));
+    llvm::Type *IntTy64 = llvm::Type::getInt64Ty(M.getContext());
+    uint64_t ProfileVersion =
+        (INSTR_PROF_RAW_VERSION | VARIANT_MASK_BYTE_COVERAGE);
+
+    auto IRLevelVersionVariable = new llvm::GlobalVariable(
+        M, IntTy64, true, llvm::GlobalValue::WeakAnyLinkage,
+        llvm::Constant::getIntegerValue(IntTy64,
+                                        llvm::APInt(64, ProfileVersion)),
+        VarName);
+
+    IRLevelVersionVariable->setVisibility(llvm::GlobalValue::DefaultVisibility);
+    llvm::Triple TT(M.getTargetTriple());
+    if (TT.supportsCOMDAT()) {
+      IRLevelVersionVariable->setLinkage(llvm::GlobalValue::ExternalLinkage);
+      IRLevelVersionVariable->setComdat(M.getOrInsertComdat(VarName));
+    }
+    IRLevelVersionVariable->setDSOLocal(true);
+  }
+}
+
 // This method either inserts a call to the profile run-time during
 // instrumentation or puts profile data into metadata for PGO use.
 void CodeGenPGO::valueProfile(CGBuilderTy &Builder, uint32_t ValueKind,
diff --git a/clang/lib/CodeGen/CodeGenPGO.h b/clang/lib/CodeGen/CodeGenPGO.h
index d3c2b277238fc..036fbf6815a49 100644
--- a/clang/lib/CodeGen/CodeGenPGO.h
+++ b/clang/lib/CodeGen/CodeGenPGO.h
@@ -94,6 +94,8 @@ class CodeGenPGO {
   // Set a module flag indicating if value profiling is enabled.
   void setValueProfilingFlag(llvm::Module &M);
 
+  void setProfileVersion(llvm::Module &M);
+
 private:
   void setFuncName(llvm::Function *Fn);
   void setFuncName(StringRef Name, llvm::GlobalValue::LinkageTypes Linkage);
@@ -108,8 +110,8 @@ class CodeGenPGO {
   bool canEmitMCDCCoverage(const CGBuilderTy &Builder);
 
 public:
-  void emitCounterIncrement(CGBuilderTy &Builder, const Stmt *S,
-                            llvm::Value *StepV);
+  void emitCounterSetOrIncrement(CGBuilderTy &Builder, const Stmt *S,
+                                 llvm::Value *StepV);
   void emitMCDCTestVectorBitmapUpdate(CGBuilderTy &Builder, const Expr *S,
                                       Address MCDCCondBitmapAddr);
   void emitMCDCParameters(CGBuilderTy &Builder);
diff --git a/clang/lib/CodeGen/CoverageMappingGen.cpp b/clang/lib/CodeGen/CoverageMappingGen.cpp
index e25a92758f32b..71215da362d3d 100644
--- a/clang/lib/CodeGen/CoverageMappingGen.cpp
+++ b/clang/lib/CodeGen/CoverageMappingGen.cpp
@@ -31,6 +31,14 @@
 // is textually included.
 #define COVMAP_V3
 
+namespace llvm {
+cl::opt<bool>
+    EnableSingleByteCoverage("enable-single-byte-coverage",
+                             llvm::cl::ZeroOrMore,
+                             llvm::cl::desc("Enable single byte coverage"),
+                             llvm::cl::Hidden, llvm::cl::init(false));
+} // namespace llvm
+
 static llvm::cl::opt<bool> EmptyLineCommentCoverage(
     "emptyline-comment-coverage",
     llvm::cl::desc("Emit emptylines and comment lines as skipped regions (only "
@@ -832,16 +840,22 @@ struct CounterCoverageMappingBuilder
 
   /// Return a counter for the subtraction of \c RHS from \c LHS
   Counter subtractCounters(Counter LHS, Counter RHS, bool Simplify = true) {
+    assert(!llvm::EnableSingleByteCoverage &&
+           "cannot add counters when single byte coverage mode is enabled");
     return Builder.subtract(LHS, RHS, Simplify);
   }
 
   /// Return a counter for the sum of \c LHS and \c RHS.
   Counter addCounters(Counter LHS, Counter RHS, bool Simplify = true) {
+    assert(!llvm::EnableSingleByteCoverage &&
+           "cannot add counters when single byte coverage mode is enabled");
     return Builder.add(LHS, RHS, Simplify);
   }
 
   Counter addCounters(Counter C1, Counter C2, Counter C3,
                       bool Simplify = true) {
+    assert(!llvm::EnableSingleByteCoverage &&
+           "cannot add counters when single byte coverage mode is enabled");
     return addCounters(addCounters(C1, C2, Simplify), C3, Simplify);
   }
 
@@ -1443,8 +1457,9 @@ struct CounterCoverageMappingBuilder
 
   void VisitBreakStmt(const BreakStmt *S) {
     assert(!BreakContinueStack.empty() && "break not in a loop or switch!");
-    BreakContinueStack.back().BreakCount = addCounters(
-        BreakContinueStack.back().BreakCount, getRegion().getCounter());
+    if (!llvm::EnableSingleByteCoverage)
+      BreakContinueStack.back().BreakCount = addCounters(
+          BreakContinueStack.back().BreakCount, getRegion().getCounter());
     // FIXME: a break in a switch should terminate regions for all preceding
     // case statements, not just the most recent one.
     terminateRegion(S);
@@ -1452,8 +1467,9 @@ struct CounterCoverageMappingBuilder
 
   void VisitContinueStmt(const ContinueStmt *S) {
     assert(!BreakContinueStack.empty() && "continue stmt not in a loop!");
-    BreakContinueStack.back().ContinueCount = addCounters(
-        BreakContinueStack.back().ContinueCount, getRegion().getCounter());
+    if (!llvm::EnableSingleByteCoverage)
+      BreakContinueStack.back().ContinueCount = addCounters(
+          BreakContinueStack.back().ContinueCount, getRegion().getCounter());
     terminateRegion(S);
   }
 
@@ -1471,7 +1487,9 @@ struct CounterCoverageMappingBuilder
     extendRegion(S);
 
     Counter ParentCount = getRegion().getCounter();
-    Counter BodyCount = getRegionCounter(S);
+    Counter BodyCount = llvm::EnableSingleByteCoverage
+                            ? getRegionCounter(S->getBody())
+                            : getRegionCounter(S);
 
     // Handle the body first so that we can get the backedge count.
     BreakContinueStack.push_back(BreakContinue());
@@ -1484,7 +1502,9 @@ struct CounterCoverageMappingBuilder
 
     // Go back to handle the condition.
     Counter CondCount =
-        addCounters(ParentCount, BackedgeCount, BC.ContinueCount);
+        llvm::EnableSingleByteCoverage
+            ? getRegionCounter(S->getCond())
+            : addCounters(ParentCount, BackedgeCount, BC.ContinueCount);
     propagateCounts(CondCount, S->getCond());
     adjustForOutOfOrderTraversal(getEnd(S));
 
@@ -1494,7 +1514,11 @@ struct CounterCoverageMappingBuilder
       fillGapAreaWithCount(Gap->getBegin(), Gap->getEnd(), BodyCount);
 
     Counter OutCount =
-        addCounters(BC.BreakCount, subtractCounters(CondCount, BodyCount));
+        llvm::EnableSingleByteCoverage
+            ? getRegionCounter(S)
+            : addCounters(BC.BreakCount,
+                          subtractCounters(CondCount, BodyCount));
+
     if (OutCount != ParentCount) {
       pushRegion(OutCount);
       GapRegionCounter = OutCount;
@@ -1503,38 +1527,53 @@ struct CounterCoverageMappingBuilder
     }
 
     // Create Branch Region around condition.
-    createBranchRegion(S->getCond(), BodyCount,
-                       subtractCounters(CondCount, BodyCount));
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(S->getCond(), BodyCount,
+                         subtractCounters(CondCount, BodyCount));
   }
 
   void VisitDoStmt(const DoStmt *S) {
     extendRegion(S);
 
     Counter ParentCount = getRegion().getCounter();
-    Counter BodyCount = getRegionCounter(S);
+    Counter BodyCount = llvm::EnableSingleByteCoverage
+                            ? getRegionCounter(S->getBody())
+                            : getRegionCounter(S);
 
     BreakContinueStack.push_back(BreakContinue());
     extendRegion(S->getBody());
-    Counter BackedgeCount =
-        propagateCounts(addCounters(ParentCount, BodyCount), S->getBody());
+
+    Counter BackedgeCount;
+    if (llvm::EnableSingleByteCoverage)
+      propagateCounts(BodyCount, S->getBody());
+    else
+      BackedgeCount =
+          propagateCounts(addCounters(ParentCount, BodyCount), S->getBody());
+
     BreakContinue BC = BreakContinueStack.pop_back_val();
 
     bool BodyHasTerminateStmt = HasTerminateStmt;
     HasTerminateStmt = false;
 
-    Counter CondCount = addCounters(BackedgeCount, BC.ContinueCount);
+    Counter CondCount = llvm::EnableSingleByteCoverage
+                            ? getRegionCounter(S->getCond())
+                            : addCounters(BackedgeCount, BC.ContinueCount);
     propagateCounts(CondCount, S->getCond());
 
     Counter OutCount =
-        addCounters(BC.BreakCount, subtractCounters(CondCount, BodyCount));
+        llvm::EnableSingleByteCoverage
+            ? getRegionCounter(S)
+            : addCounters(BC.BreakCount,
+                          subtractCounters(CondCount, BodyCount));
     if (OutCount != ParentCount) {
       pushRegion(OutCount);
       GapRegionCounter = OutCount;
     }
 
     // Create Branch Region around condition.
-    createBranchRegion(S->getCond(), BodyCount,
-                       subtractCounters(CondCount, BodyCount));
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(S->getCond(), BodyCount,
+                         subtractCounters(CondCount, BodyCount));
 
     if (BodyHasTerminateStmt)
       HasTerminateStmt = true;
@@ -1546,7 +1585,9 @@ struct CounterCoverageMappingBuilder
       Visit(S->getInit());
 
     Counter ParentCount = getRegion().getCounter();
-    Counter BodyCount = getRegionCounter(S);
+    Counter BodyCount = llvm::EnableSingleByteCoverage
+                            ? getRegionCounter(S->getBody())
+                            : getRegionCounter(S);
 
     // The loop increment may contain a break or continue.
     if (S->getInc())
@@ -1565,14 +1606,23 @@ struct CounterCoverageMappingBuilder
     // the count for all the continue statements.
     BreakContinue IncrementBC;
     if (const Stmt *Inc = S->getInc()) {
-      propagateCounts(addCounters(BackedgeCount, BodyBC.ContinueCount), Inc);
+      Counter IncCount;
+      if (llvm::EnableSingleByteCoverage)
+        IncCount = getRegionCounter(S->getInc());
+      else
+        IncCount = addCounters(BackedgeCount, BodyBC.ContinueCount);
+      propagateCounts(IncCount, Inc);
       IncrementBC = BreakContinueStack.pop_back_val();
     }
 
     // Go back to handle the condition.
-    Counter CondCount = addCounters(
-        addCounters(ParentCount, BackedgeCount, BodyBC.ContinueCount),
-        IncrementBC.ContinueCount);
+    Counter CondCount =
+        llvm::EnableSingleByteCoverage
+            ? getRegionCounter(S->getCond())
+            : addCounters(
+                  addCounters(ParentCount, BackedgeCount, BodyBC.ContinueCount),
+                  IncrementBC.ContinueCount);
+
     if (const Expr *Cond = S->getCond()) {
       propagateCounts(CondCount, Cond);
       adjustForOutOfOrderTraversal(getEnd(S));
@@ -1583,8 +1633,11 @@ struct CounterCoverageMappingBuilder
     if (Gap)
       fillGapAreaWithCount(Gap->getBegin(), Gap->getEnd(), BodyCount);
 
-    Counter OutCount = addCounters(BodyBC.BreakCount, IncrementBC.BreakCount,
-                                   subtractCounters(CondCount, BodyCount));
+    Counter OutCount =
+        llvm::EnableSingleByteCoverage
+            ? getRegionCounter(S)
+            : addCounters(BodyBC.BreakCount, IncrementBC.BreakCount,
+                          subtractCounters(CondCount, BodyCount));
     if (OutCount != ParentCount) {
       pushRegion(OutCount);
       GapRegionCounter = OutCount;
@@ -1593,8 +1646,9 @@ struct CounterCoverageMappingBuilder
     }
 
     // Create Branch Region around condition.
-    createBranchRegion(S->getCond(), BodyCount,
-                       subtractCounters(CondCount, BodyCount));
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(S->getCond(), BodyCount,
+                         subtractCounters(CondCount, BodyCount));
   }
 
   void VisitCXXForRangeStmt(const CXXForRangeStmt *S) {
@@ -1605,7 +1659,9 @@ struct CounterCoverageMappingBuilder
     Visit(S->getRangeStmt());
 
     Counter ParentCount = getRegion().getCounter();
-    Counter BodyCount = getRegionCounter(S);
+    Counter BodyCount = llvm::EnableSingleByteCoverage
+                            ? getRegionCounter(S->getBody())
+                            : getRegionCounter(S);
 
     BreakContinueStack.push_back(BreakContinue());
     extendRegion(S->getBody());
@@ -1620,10 +1676,15 @@ struct CounterCoverageMappingBuilder
     if (Gap)
       fillGapAreaWithCount(Gap->getBegin(), Gap->getEnd(), BodyCount);
 
-    Counter LoopCount =
-        addCounters(ParentCount, BackedgeCount, BC.ContinueCount);
-    Counter OutCount =
-        addCounters(BC.BreakCount, subtractCounters(LoopCount, BodyCount));
+    Counter OutCount;
+    Counter LoopCount;
+    if (llvm::EnableSingleByteCoverage)
+      OutCount = getRegionCounter(S);
+    else {
+      LoopCount = addCounters(ParentCount, BackedgeCount, BC.ContinueCount);
+      OutCount =
+          addCounters(BC.BreakCount, subtractCounters(LoopCount, BodyCount));
+    }
     if (OutCount != ParentCount) {
       pushRegion(OutCount);
       GapRegionCounter = OutCount;
@@ -1632,8 +1693,9 @@ struct CounterCoverageMappingBuilder
     }
 
     // Create Branch Region around condition.
-    createBranchRegion(S->getCond(), BodyCount,
-                       subtractCounters(LoopCount, BodyCount));
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(S->getCond(), BodyCount,
+                         subtractCounters(LoopCount, BodyCount));
   }
 
   void VisitObjCForCollectionStmt(const ObjCForCollectionStmt *S) {
@@ -1694,7 +1756,7 @@ struct CounterCoverageMappingBuilder
       propagateCounts(Counter::getZero(), Body);
     BreakContinue BC = BreakContinueStack.pop_back_val();
 
-    if (!BreakContinueStack.empty())
+    if (!BreakContinueStack.empty() && !llvm::EnableSingleByteCoverage)
       BreakContinueStack.back().ContinueCount = addCounters(
           BreakContinueStack.back().ContinueCount, BC.ContinueCount);
 
@@ -1709,6 +1771,11 @@ struct CounterCoverageMappingBuilder
     MostRecentLocation = getStart(S);
     handleFileExit(ExitLoc);
 
+    // When single byte coverage mode is enabled, do not create branch region by
+    // early returning.
+    if (llvm::EnableSingleByteCoverage)
+      return;
+
     // Create a Branch Region around each Case. Subtract the case's
     // counter from the Parent counter to track the "False" branch count.
     Counter CaseCountSum;
@@ -1741,8 +1808,10 @@ struct CounterCoverageMappingBuilder
     extendRegion(S);
 
     SourceMappingRegion &Parent = getRegion();
+    Counter Count = llvm::EnableSingleByteCoverage
+                        ? getRegionCounter(S)
+                        : addCounters(Parent.getCounter(), getRegionCounter(S));
 
-    Counter Count = addCounters(Parent.getCounter(), getRegionCounter(S));
     // Reuse the existing region if it starts at our label. This is typical of
     // the first case in a switch.
     if (Parent.hasStartLoc() && Parent.getBeginLoc() == getStart(S))
@@ -1860,7 +1929,9 @@ struct CounterCoverageMappingBuilder
     extendRegion(S->getCond());
 
     Counter ParentCount = getRegion().getCounter();
-    Counter ThenCount = getRegionCounter(S);
+    Counter ThenCount = llvm::EnableSingleByteCoverage
+                            ? getRegionCounter(S->getThen())
+                            : getRegionCounter(S);
 
     // Emitting a counter for the condition makes it easier to interpret the
     // counter for the body when looking at the coverage.
@@ -1874,7 +1945,12 @@ struct CounterCoverageMappingBuilder
 
     extendRegion(S->getThen());
     Counter OutCount = propagateCounts(ThenCount, S->getThen());
-    Counter ElseCount = subtractCounters(ParentCount, ThenCount);
+
+    Counter ElseCount;
+    if (!llvm::EnableSingleByteCoverage)
+      ElseCount = subtractCounters(ParentCount, ThenCount);
+    else if (S->getElse())
+      ElseCount = getRegionCounter(S->getElse());
 
     if (const Stmt *Else = S->getElse()) {
       bool ThenHasTerminateStmt = HasTerminateStmt;
@@ -1885,21 +1961,28 @@ struct CounterCoverageMappingBuilder
       if (Gap)
         fillGapAreaWithCount(Gap->getBegin(), Gap->getEnd(), ElseCount);
       extendRegion(Else);
-      OutCount = addCounters(OutCount, propagateCounts(ElseCount, Else));
+
+      Counter ElseOutCount = propagateCounts(ElseCount, Else);
+      if (!llvm::EnableSingleByteCoverage)
+        OutCount = addCounters(OutCount, ElseOutCount);
 
       if (ThenHasTerminateStmt)
         HasTerminateStmt = true;
-    } else
+    } else if (!llvm::EnableSingleByteCoverage)
       OutCount = addCounters(OutCount, ElseCount);
 
+    if (llvm::EnableSingleByteCoverage)
+      OutCount = getRegionCounter(S);
+
     if (OutCount != ParentCount) {
       pushRegion(OutCount);
       GapRegionCounter = OutCount;
     }
 
-    // Create Branch Region around condition.
-    createBranchRegion(S->getCond(), ThenCount,
-                       subtractCounters(ParentCount, ThenCount));
+    if (!S->isConsteval() && !llvm::EnableSingleByteCoverage)
+      // Create Branch Region around condition.
+      createBranchRegion(S->getCond(), ThenCount,
+                         subtractCounters(ParentCount, ThenCount));
   }
 
   void VisitCXXTryStmt(const CXXTryStmt *S) {
@@ -1925,7 +2008,9 @@ struct CounterCoverageMappingBuilder
     extendRegion(E);
 
     Counter ParentCount = getRegion().getCounter();
-    Counter TrueCount = getRegionCounter(E);
+    Counter TrueCount = llvm::EnableSingleByteCoverage
+                            ? getRegionCounter(E->getTrueExpr())
+                            : getRegionCounter(E);
 
     propagateCounts(ParentCount, E->getCond());
     Counter OutCount;
@@ -1944,9 +2029,15 @@ struct CounterCoverageMappingBuilder
     }
 
     extendRegion(E->getFalseExpr());
-    OutCount = addCounters(
-        OutCount, propagateCounts(subtractCounters(ParentCount, TrueCount),
-                                  E->getFalseExpr()));
+    Counter FalseCount = llvm::EnableSingleByteCoverage
+                             ? getRegionCounter(E->getFalseExpr())
+                             : subtractCounters(ParentCount, TrueCount);
+
+    Counter FalseOutCount = propagateCounts(FalseCount, E->getFalseExpr());
+    if (llvm::EnableSingleByteCoverage)
+      OutCount = getRegionCounter(E);
+    else
+      OutCount = addCounters(OutCount, FalseOutCount);
 
     if (OutCount != ParentCount) {
       pushRegion(OutCount);
@@ -1954,8 +2045,9 @@ struct CounterCoverageMappingBuilder
     }
 
     // Create Branch Region around condition.
-    createBranchRegion(E->getCond(), TrueCount,
-                       subtractCounters(ParentCount, TrueCount));
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(E->getCond(), TrueCount,
+                         subtractCounters(ParentCount, TrueCount));
   }
 
   void createDecision(const BinaryOperator *E) {
@@ -2002,12 +2094,14 @@ struct CounterCoverageMappingBuilder
     Counter ParentCnt = getRegion().getCounter();
 
     // Create Branch Region around LHS condition.
-    createBranchRegion(E->getLHS(), RHSExecCnt,
-                       subtractCounters(ParentCnt, RHSExecCnt), DecisionLHS);
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(E->getLHS(), RHSExecCnt,
+                         subtractCounters(ParentCnt, RHSExecCnt), DecisionLHS);
 
     // Create Branch Region around RHS condition.
-    createBranchRegion(E->getRHS(), RHSTrueCnt,
-                       subtractCounters(RHSExecCnt, RHSTrueCnt), DecisionRHS);
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(E->getRHS(), RHSTrueCnt,
+                         subtractCounters(RHSExecCnt, RHSTrueCnt), DecisionRHS);
 
     // Create MCDC Decision Region if at top-level (root).
     if (IsRootNode)
@@ -2058,12 +2152,14 @@ struct CounterCoverageMappingBuilder
     Counter ParentCnt = getRegion().getCounter();
 
     // Create Branch Region around LHS condition.
-    createBranchRegion(E->getLHS(), subtractCounters(ParentCnt, RHSExecCnt),
-                       RHSExecCnt, DecisionLHS);
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(E->getLHS(), subtractCounters(ParentCnt, RHSExecCnt),
+                         RHSExecCnt, DecisionLHS);
 
     // Create Branch Region around RHS condition.
-    createBranchRegion(E->getRHS(), subtractCounters(RHSExecCnt, RHSFalseCnt),
-                       RHSFalseCnt, DecisionRHS);
+    if (!llvm::EnableSingleByteCoverage)
+      createBranchRegion(E->getRHS(), subtractCounters(RHSExecCnt, RHSFalseCnt),
+                         RHSFalseCnt, DecisionRHS);
 
     // Create MCDC Decision Region if at top-level (root).
     if (IsRootNode)
diff --git a/clang/test/CoverageMapping/single-byte-counters.cpp b/clang/test/CoverageMapping/single-byte-counters.cpp
new file mode 100644
index 0000000000000..8e9b613dcc68f
--- /dev/null
+++ b/clang/test/CoverageMapping/single-byte-counters.cpp
@@ -0,0 +1,169 @@
+// RUN: %clang_cc1 -mllvm -emptyline-comment-coverage=false -mllvm -enable-single-byte-coverage=true -fprofile-instrument=clang -fcoverage-mapping -dump-coverage-mapping -emit-llvm-only -main-file-name single-byte-counters.cpp %s | FileCheck %s
+
+// CHECK: testIf
+int testIf(int x) { // CHECK-NEXT: File 0, [[@LINE]]:19 -> [[@LINE+10]]:2 = #0
+                    // CHECK-NEXT: File 0, [[@LINE+5]]:7 -> [[@LINE+5]]:13 = #0
+                    // CHECK-NEXT: Gap,File 0, [[@LINE+4]]:14 -> [[@LINE+5]]:5 = #1
+                    // CHECK-NEXT: File 0, [[@LINE+4]]:5 -> [[@LINE+4]]:16 = #1
+                    // CHECK-NEXT: File 0, [[@LINE+5]]:3 -> [[@LINE+5]]:16 = #2
+  int result = 0;
+  if (x == 0)
+    result = -1;
+
+  return result;
+}
+
+// CHECK-NEXT: testIfElse
+int testIfElse(int x) { // CHECK-NEXT: File 0, [[@LINE]]:23 -> [[@LINE+13]]:2 = #0
+                        // CHECK-NEXT: File 0, [[@LINE+7]]:7 -> [[@LINE+7]]:12 = #0
+                        // CHECK-NEXT: Gap,File 0, [[@LINE+6]]:13 -> [[@LINE+7]]:5 = #1
+                        // CHECK-NEXT: File 0, [[@LINE+6]]:5 -> [[@LINE+6]]:15 = #1
+                        // CHECK-NEXT: Gap,File 0, [[@LINE+5]]:16 -> [[@LINE+7]]:5 = #2
+                        // CHECK-NEXT: File 0, [[@LINE+6]]:5 -> [[@LINE+6]]:19 = #2
+                        // CHECK-NEXT: File 0, [[@LINE+6]]:3 -> [[@LINE+6]]:16 = #3
+  int result = 0;
+  if (x < 0)
+    result = 0;
+  else
+    result = x * x;
+  return result;
+}
+
+// CHECK-NEXT: testIfElseReturn
+int testIfElseReturn(int x) { // CHECK-NEXT: File 0, [[@LINE]]:29 -> [[@LINE+14]]:2 = #0
+                              // CHECK-NEXT: File 0, [[@LINE+8]]:7 -> [[@LINE+8]]:12 = #0
+                              // CHECK-NEXT: Gap,File 0, [[@LINE+7]]:13 -> [[@LINE+8]]:5 = #1
+                              // CHECK-NEXT: File 0, [[@LINE+7]]:5 -> [[@LINE+7]]:19 = #1
+                              // CHECK-NEXT: Gap,File 0, [[@LINE+6]]:20 -> [[@LINE+8]]:5 = #2
+                              // CHECK-NEXT: File 0, [[@LINE+7]]:5 -> [[@LINE+7]]:13 = #2
+                              // CHECK-NEXT: Gap,File 0, [[@LINE+6]]:14 -> [[@LINE+7]]:3 = #3
+                              // CHECK-NEXT: File 0, [[@LINE+6]]:3 -> [[@LINE+6]]:16 = #3
+  int result = 0;
+  if (x > 0)
+    result = x * x;
+  else
+    return 0;
+  return result;
+}
+
+// CHECK-NEXT: testSwitch
+int testSwitch(int x) { // CHECK-NEXT: File 0, [[@LINE]]:23 -> [[@LINE+22]]:2 = #0
+                        // CHECK-NEXT: Gap,File 0, [[@LINE+9]]:14 -> [[@LINE+17]]:15 = 0
+                        // CHECK-NEXT: File 0, [[@LINE+9]]:3 -> [[@LINE+11]]:10 = #2
+                        // CHECK-NEXT: Gap,File 0, [[@LINE+10]]:11 -> [[@LINE+11]]:3 = 0
+                        // CHECK-NEXT: File 0, [[@LINE+10]]:3 -> [[@LINE+12]]:10 = #3
+                        // CHECK-NEXT: Gap,File 0, [[@LINE+11]]:11 -> [[@LINE+12]]:3 = 0
+                        // CHECK-NEXT: File 0, [[@LINE+11]]:3 -> [[@LINE+12]]:15 = #4
+                        // CHECK-NEXT: Gap,File 0, [[@LINE+12]]:4 -> [[@LINE+14]]:3 = #1
+                        // CHECK-NEXT: File 0, [[@LINE+13]]:3 -> [[@LINE+13]]:16 = #1
+  int result;
+  switch (x) {
+  case 1:
+    result = 1;
+    break;
+  case 2:
+    result = 2;
+    break;
+  default:
+    result = 0;
+  }
+
+  return result;
+}
+
+// CHECK-NEXT: testWhile
+int testWhile() { // CHECK-NEXT: File 0, [[@LINE]]:17 -> [[@LINE+13]]:2 = #0
+                  // CHECK-NEXT: File 0, [[@LINE+6]]:10 -> [[@LINE+6]]:16 = #1
+                  // CHECK-NEXT: Gap,File 0, [[@LINE+5]]:17 -> [[@LINE+5]]:18 = #2
+                  // CHECK-NEXT: File 0, [[@LINE+4]]:18 -> [[@LINE+7]]:4 = #2
+                  // CHECK-NEXT: File 0, [[@LINE+8]]:3 -> [[@LINE+8]]:13 = #3
+  int i = 0;
+  int sum = 0;
+  while (i < 10) {
+    sum += i;
+    i++;
+  }
+
+  return sum;
+}
+
+// CHECK-NEXT: testContinue
+int testContinue() { // CHECK-NEXT: File 0, [[@LINE]]:20 -> [[@LINE+21]]:2 = #0
+                     // CHECK-NEXT: File 0, [[@LINE+12]]:10 -> [[@LINE+12]]:16 = #1
+                     // CHECK-NEXT: Gap,File 0, [[@LINE+11]]:17 -> [[@LINE+11]]:18 = #2
+                     // CHECK-NEXT: File 0, [[@LINE+10]]:18 -> [[@LINE+15]]:4 = #2
+                     // CHECK-NEXT: File 0, [[@LINE+10]]:9 -> [[@LINE+10]]:15 = #2
+                     // CHECK-NEXT: Gap,File 0, [[@LINE+9]]:16 -> [[@LINE+10]]:7 = #4
+                     // CHECK-NEXT: File 0, [[@LINE+9]]:7 -> [[@LINE+9]]:15 = #4
+                     // CHECK-NEXT: Gap,File 0, [[@LINE+8]]:16 -> [[@LINE+9]]:5 = #5
+                     // CHECK-NEXT: File 0, [[@LINE+8]]:5 -> [[@LINE+10]]:4 = #5
+                     // CHECK-NEXT: Gap,File 0, [[@LINE+9]]:4 -> [[@LINE+11]]:3 = #3
+                     // CHECK-NEXT: File 0, [[@LINE+10]]:3 -> [[@LINE+10]]:13 = #3
+  int i = 0;
+  int sum = 0;
+  while (i < 10) {
+    if (i == 4)
+      continue;
+    sum += i;
+    i++;
+  }
+
+  return sum;
+}
+
+// CHECK-NEXT: testFor
+int testFor() { // CHECK-NEXT: File 0, [[@LINE]]:15 -> [[@LINE+13]]:2 = #0
+                // CHECK-NEXT: File 0, [[@LINE+7]]:19 -> [[@LINE+7]]:25 = #1
+                // CHECK-NEXT: File 0, [[@LINE+6]]:27 -> [[@LINE+6]]:30 = #2
+                // CHECK-NEXT: Gap,File 0, [[@LINE+5]]:31 -> [[@LINE+5]]:32 = #3
+                // CHECK-NEXT: File 0, [[@LINE+4]]:32 -> [[@LINE+6]]:4 = #3
+                // CHECK-NEXT: File 0, [[@LINE+7]]:3 -> [[@LINE+7]]:13 = #4
+  int i;
+  int sum = 0;
+  for (int i = 0; i < 10; i++) {
+    sum += i;
+  }
+
+  return sum;
+}
+
+// CHECK-NEXT: testForRange
+int testForRange() { // CHECK-NEXT: File 0, [[@LINE]]:20 -> [[@LINE+12]]:2 = #0
+                     // CHECK-NEXT: Gap,File 0, [[@LINE+6]]:28 -> [[@LINE+6]]:29 = #1
+                     // CHECK-NEXT: File 0, [[@LINE+5]]:29 -> [[@LINE+7]]:4 = #1
+                     // CHECK-NEXT: File 0, [[@LINE+8]]:3 -> [[@LINE+8]]:13 = #2
+  int sum = 0;
+  int array[] = {1, 2, 3, 4, 5};
+
+  for (int element : array) {
+      sum += element;
+  }
+
+  return sum;
+}
+
+// CHECK-NEXT: testDo
+int testDo() { // CHECK-NEXT: File 0, [[@LINE]]:14 -> [[@LINE+12]]:2 = #0
+               // CHECK-NEXT: File 0, [[@LINE+5]]:6 -> [[@LINE+8]]:4 = #1
+               // CHECK-NEXT: File 0, [[@LINE+7]]:12 -> [[@LINE+7]]:17 = #2
+               // CHECK-NEXT: File 0, [[@LINE+8]]:3 -> [[@LINE+8]]:13 = #3
+  int i = 0;
+  int sum = 0;
+  do {
+    sum += i;
+    i++;
+  } while (i < 5);
+
+  return sum;
+}
+
+// CHECK-NEXT: testConditional
+int testConditional(int x) { // CHECK-NEXT: File 0, [[@LINE]]:28 -> [[@LINE+8]]:2 = #0
+                             // CHECK-NEXT: File 0, [[@LINE+5]]:15 -> [[@LINE+5]]:22 = #0
+                             // CHECK-NEXT: Gap,File 0, [[@LINE+4]]:24 -> [[@LINE+4]]:25 = #2
+                             // CHECK-NEXT: File 0, [[@LINE+3]]:25 -> [[@LINE+3]]:26 = #2
+                             // CHECK-NEXT: File 0, [[@LINE+2]]:29 -> [[@LINE+2]]:31 = #3
+                             // CHECK-NEXT: File 0, [[@LINE+2]]:2 -> [[@LINE+2]]:15 = #1
+ int result = (x > 0) ? 1 : -1;
+ return result;
+}
diff --git a/compiler-rt/lib/profile/InstrProfiling.h b/compiler-rt/lib/profile/InstrProfiling.h
index 0123908336918..705b54fddcd7b 100644
--- a/compiler-rt/lib/profile/InstrProfiling.h
+++ b/compiler-rt/lib/profile/InstrProfiling.h
@@ -294,7 +294,8 @@ COMPILER_RT_VISIBILITY extern int INSTR_PROF_PROFILE_RUNTIME_VAR;
  * variable is defined as weak so that compiler can emit an overriding
  * definition depending on user option.
  */
-extern uint64_t INSTR_PROF_RAW_VERSION_VAR; /* __llvm_profile_raw_version */
+COMPILER_RT_VISIBILITY extern uint64_t
+    INSTR_PROF_RAW_VERSION_VAR; /* __llvm_profile_raw_version */
 
 /*!
  * This variable is a weak symbol defined in InstrProfiling.c. It allows
diff --git a/llvm/include/llvm/ProfileData/Coverage/CoverageMapping.h b/llvm/include/llvm/ProfileData/Coverage/CoverageMapping.h
index b39f262c3d976..f8f549eea2e09 100644
--- a/llvm/include/llvm/ProfileData/Coverage/CoverageMapping.h
+++ b/llvm/include/llvm/ProfileData/Coverage/CoverageMapping.h
@@ -359,15 +359,19 @@ struct CountedRegion : public CounterMappingRegion {
   uint64_t ExecutionCount;
   uint64_t FalseExecutionCount;
   bool Folded;
+  bool HasSingleByteCoverage;
 
-  CountedRegion(const CounterMappingRegion &R, uint64_t ExecutionCount)
+  CountedRegion(const CounterMappingRegion &R, uint64_t ExecutionCount,
+                bool HasSingleByteCoverage)
       : CounterMappingRegion(R), ExecutionCount(ExecutionCount),
-        FalseExecutionCount(0), Folded(false) {}
+        FalseExecutionCount(0), Folded(false),
+        HasSingleByteCoverage(HasSingleByteCoverage) {}
 
   CountedRegion(const CounterMappingRegion &R, uint64_t ExecutionCount,
-                uint64_t FalseExecutionCount)
+                uint64_t FalseExecutionCount, bool HasSingleByteCoverage)
       : CounterMappingRegion(R), ExecutionCount(ExecutionCount),
-        FalseExecutionCount(FalseExecutionCount), Folded(false) {}
+        FalseExecutionCount(FalseExecutionCount), Folded(false),
+        HasSingleByteCoverage(HasSingleByteCoverage) {}
 };
 
 /// MCDC Record grouping all information together.
@@ -661,10 +665,11 @@ struct FunctionRecord {
   }
 
   void pushRegion(CounterMappingRegion Region, uint64_t Count,
-                  uint64_t FalseCount) {
+                  uint64_t FalseCount, bool HasSingleByteCoverage) {
     if (Region.Kind == CounterMappingRegion::BranchRegion ||
         Region.Kind == CounterMappingRegion::MCDCBranchRegion) {
-      CountedBranchRegions.emplace_back(Region, Count, FalseCount);
+      CountedBranchRegions.emplace_back(Region, Count, FalseCount,
+                                        HasSingleByteCoverage);
       // If both counters are hard-coded to zero, then this region represents a
       // constant-folded branch.
       if (Region.Count.isZero() && Region.FalseCount.isZero())
@@ -673,7 +678,8 @@ struct FunctionRecord {
     }
     if (CountedRegions.empty())
       ExecutionCount = Count;
-    CountedRegions.emplace_back(Region, Count, FalseCount);
+    CountedRegions.emplace_back(Region, Count, FalseCount,
+                                HasSingleByteCoverage);
   }
 };
 
diff --git a/llvm/include/llvm/ProfileData/InstrProfWriter.h b/llvm/include/llvm/ProfileData/InstrProfWriter.h
index 047b14f223bd9..7a806fd7fcf34 100644
--- a/llvm/include/llvm/ProfileData/InstrProfWriter.h
+++ b/llvm/include/llvm/ProfileData/InstrProfWriter.h
@@ -168,6 +168,10 @@ class InstrProfWriter {
 
   InstrProfKind getProfileKind() const { return ProfileKind; }
 
+  bool hasSingleByteCoverage() const {
+    return static_cast<bool>(ProfileKind & InstrProfKind::SingleByteCoverage);
+  }
+
   // Internal interface for testing purpose only.
   void setValueProfDataEndianness(llvm::endianness Endianness);
   void setOutputSparse(bool Sparse);
diff --git a/llvm/lib/ProfileData/Coverage/CoverageMapping.cpp b/llvm/lib/ProfileData/Coverage/CoverageMapping.cpp
index 334f5dce879e5..3f892d7aff2f4 100644
--- a/llvm/lib/ProfileData/Coverage/CoverageMapping.cpp
+++ b/llvm/lib/ProfileData/Coverage/CoverageMapping.cpp
@@ -860,7 +860,8 @@ Error CoverageMapping::loadFunctionRecord(
       consumeError(std::move(E));
       return Error::success();
     }
-    Function.pushRegion(Region, *ExecutionCount, *AltExecutionCount);
+    Function.pushRegion(Region, *ExecutionCount, *AltExecutionCount,
+                        ProfileReader.hasSingleByteCoverage());
 
     // Record ExpansionRegion.
     if (Region.Kind == CounterMappingRegion::ExpansionRegion) {
@@ -1282,8 +1283,14 @@ class SegmentBuilder {
       // value for that area.
       // We add counts of the regions of the same kind as the active region
       // to handle the both situations.
-      if (I->Kind == Active->Kind)
-        Active->ExecutionCount += I->ExecutionCount;
+      if (I->Kind == Active->Kind) {
+        assert(I->HasSingleByteCoverage == Active->HasSingleByteCoverage &&
+               "Regions are generated in different coverage modes");
+        if (I->HasSingleByteCoverage)
+          Active->ExecutionCount = Active->ExecutionCount || I->ExecutionCount;
+        else
+          Active->ExecutionCount += I->ExecutionCount;
+      }
     }
     return Regions.drop_back(std::distance(++Active, End));
   }