diff --git a/llvm/include/llvm/Transforms/Utils/LoopUtils.h b/llvm/include/llvm/Transforms/Utils/LoopUtils.h
@@ -372,6 +372,15 @@ RecurKind getMinMaxReductionRecurKind(Intrinsic::ID RdxID);
 /// Returns the comparison predicate used when expanding a min/max reduction.
 CmpInst::Predicate getMinMaxReductionPredicate(RecurKind RK);
 
+/// See RecurrenceDescriptor::isAnyOfPattern for a description of the pattern we
+/// are trying to match. In this pattern, we are only ever selecting between two
+/// values: 1) an initial start value \p StartVal of the reduction PHI, and 2) a
+/// loop invariant value. If any of lane value in \p Left, \p Right is not equal
+/// to \p StartVal, select the loop invariant value. This is done by selecting
+/// \p Right iff \p Left is equal to \p StartVal.
+Value *createAnyOfOp(IRBuilderBase &Builder, Value *StartVal, RecurKind RK,
+                     Value *Left, Value *Right);
+
 /// Returns a Min/Max operation corresponding to MinMaxRecurrenceKind.
 /// The Builder's fast-math-flags must be set to propagate the expected values.
 Value *createMinMaxOp(IRBuilderBase &Builder, RecurKind RK, Value *Left,

diff --git a/llvm/lib/Analysis/ValueTracking.cpp b/llvm/lib/Analysis/ValueTracking.cpp
@@ -5032,6 +5032,19 @@ void computeKnownFPClass(const Value *V, const APInt &DemandedElts,
 
       break;
     }
+    case Intrinsic::vector_reduce_fmax:
+    case Intrinsic::vector_reduce_fmin:
+    case Intrinsic::vector_reduce_fmaximum:
+    case Intrinsic::vector_reduce_fminimum: {
+      // reduce min/max will choose an element from one of the vector elements,
+      // so we can infer and class information that is common to all elements.
+      Known = computeKnownFPClass(II->getArgOperand(0), II->getFastMathFlags(),
+                                  InterestedClasses, Depth + 1, Q);
+      // Can only propagate sign if output is never NaN.
+      if (!Known.isKnownNeverNaN())
+        Known.SignBit.reset();
+      break;
+    }
     case Intrinsic::trunc:
     case Intrinsic::floor:
     case Intrinsic::ceil:

diff --git a/llvm/lib/MC/MCPseudoProbe.cpp b/llvm/lib/MC/MCPseudoProbe.cpp
@@ -343,7 +343,7 @@ template <typename T> ErrorOr<T> MCPseudoProbeDecoder::readUnencodedNumber() {
   if (Data + sizeof(T) > End) {
     return std::error_code();
   }
-  T Val = endian::readNext<T, llvm::endianness::little, unaligned>(Data);
+  T Val = endian::readNext<T, llvm::endianness::little>(Data);
   return ErrorOr<T>(Val);
 }
 

diff --git a/llvm/lib/ProfileData/InstrProf.cpp b/llvm/lib/ProfileData/InstrProf.cpp
@@ -1135,9 +1135,9 @@ static T swapToHostOrder(const unsigned char *&D, llvm::endianness Orig) {
   using namespace support;
 
   if (Orig == llvm::endianness::little)
-    return endian::readNext<T, llvm::endianness::little, unaligned>(D);
+    return endian::readNext<T, llvm::endianness::little>(D);
   else
-    return endian::readNext<T, llvm::endianness::big, unaligned>(D);
+    return endian::readNext<T, llvm::endianness::big>(D);
 }
 
 static std::unique_ptr<ValueProfData> allocValueProfData(uint32_t TotalSize) {

diff --git a/llvm/lib/ProfileData/InstrProfReader.cpp b/llvm/lib/ProfileData/InstrProfReader.cpp
@@ -115,10 +115,9 @@ readBinaryIdsInternal(const MemoryBuffer &DataBuffer,
 
     uint64_t BILen = 0;
     if (Endian == llvm::endianness::little)
-      BILen =
-          endian::readNext<uint64_t, llvm::endianness::little, unaligned>(BI);
+      BILen = endian::readNext<uint64_t, llvm::endianness::little>(BI);
     else
-      BILen = endian::readNext<uint64_t, llvm::endianness::big, unaligned>(BI);
+      BILen = endian::readNext<uint64_t, llvm::endianness::big>(BI);
 
     if (BILen == 0)
       return make_error<InstrProfError>(instrprof_error::malformed,
@@ -923,17 +922,15 @@ data_type InstrProfLookupTrait::ReadData(StringRef K, const unsigned char *D,
     // Read hash.
     if (D + sizeof(uint64_t) >= End)
       return data_type();
-    uint64_t Hash =
-        endian::readNext<uint64_t, llvm::endianness::little, unaligned>(D);
+    uint64_t Hash = endian::readNext<uint64_t, llvm::endianness::little>(D);
 
     // Initialize number of counters for GET_VERSION(FormatVersion) == 1.
     uint64_t CountsSize = N / sizeof(uint64_t) - 1;
     // If format version is different then read the number of counters.
     if (GET_VERSION(FormatVersion) != IndexedInstrProf::ProfVersion::Version1) {
       if (D + sizeof(uint64_t) > End)
         return data_type();
-      CountsSize =
-          endian::readNext<uint64_t, llvm::endianness::little, unaligned>(D);
+      CountsSize = endian::readNext<uint64_t, llvm::endianness::little>(D);
     }
     // Read counter values.
     if (D + CountsSize * sizeof(uint64_t) > End)
@@ -943,24 +940,22 @@ data_type InstrProfLookupTrait::ReadData(StringRef K, const unsigned char *D,
     CounterBuffer.reserve(CountsSize);
     for (uint64_t J = 0; J < CountsSize; ++J)
       CounterBuffer.push_back(
-          endian::readNext<uint64_t, llvm::endianness::little, unaligned>(D));
+          endian::readNext<uint64_t, llvm::endianness::little>(D));
 
     // Read bitmap bytes for GET_VERSION(FormatVersion) > 10.
     if (GET_VERSION(FormatVersion) > IndexedInstrProf::ProfVersion::Version10) {
       uint64_t BitmapBytes = 0;
       if (D + sizeof(uint64_t) > End)
         return data_type();
-      BitmapBytes =
-          endian::readNext<uint64_t, llvm::endianness::little, unaligned>(D);
+      BitmapBytes = endian::readNext<uint64_t, llvm::endianness::little>(D);
       // Read bitmap byte values.
       if (D + BitmapBytes * sizeof(uint8_t) > End)
         return data_type();
       BitmapByteBuffer.clear();
       BitmapByteBuffer.reserve(BitmapBytes);
       for (uint64_t J = 0; J < BitmapBytes; ++J)
         BitmapByteBuffer.push_back(static_cast<uint8_t>(
-            endian::readNext<uint64_t, llvm::endianness::little, unaligned>(
-                D)));
+            endian::readNext<uint64_t, llvm::endianness::little>(D)));
     }
 
     DataBuffer.emplace_back(K, Hash, std::move(CounterBuffer),
@@ -1256,8 +1251,7 @@ Error IndexedInstrProfReader::readHeader() {
     // memprof::MemProfVersion0 or the MemProf version number in
     // memprof::MemProfVersion1.
     const uint64_t FirstWord =
-        support::endian::readNext<uint64_t, llvm::endianness::little,
-                                  unaligned>(Ptr);
+        support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
 
     memprof::IndexedVersion Version = memprof::Version0;
     if (FirstWord == memprof::Version1) {
@@ -1282,17 +1276,15 @@ Error IndexedInstrProfReader::readHeader() {
     const uint64_t RecordTableOffset =
         Version == memprof::Version0
             ? FirstWord
-            : support::endian::readNext<uint64_t, llvm::endianness::little,
-                                        unaligned>(Ptr);
+            : support::endian::readNext<uint64_t, llvm::endianness::little>(
+                  Ptr);
     // The offset in the stream right before invoking
     // FrameTableGenerator.Emit.
     const uint64_t FramePayloadOffset =
-        support::endian::readNext<uint64_t, llvm::endianness::little,
-                                  unaligned>(Ptr);
+        support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     // The value returned from FrameTableGenerator.Emit.
     const uint64_t FrameTableOffset =
-        support::endian::readNext<uint64_t, llvm::endianness::little,
-                                  unaligned>(Ptr);
+        support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
 
     // Read the schema.
     auto SchemaOr = memprof::readMemProfSchema(Ptr);
@@ -1330,8 +1322,7 @@ Error IndexedInstrProfReader::readHeader() {
     const unsigned char *Ptr = Start + BinaryIdOffset;
     // Read binary ids size.
     BinaryIdsSize =
-        support::endian::readNext<uint64_t, llvm::endianness::little,
-                                  unaligned>(Ptr);
+        support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     if (BinaryIdsSize % sizeof(uint64_t))
       return error(instrprof_error::bad_header);
     // Set the binary ids start.
@@ -1348,8 +1339,7 @@ Error IndexedInstrProfReader::readHeader() {
     const unsigned char *Ptr = Start + VTableNamesOffset;
 
     CompressedVTableNamesLen =
-        support::endian::readNext<uint64_t, llvm::endianness::little,
-                                  unaligned>(Ptr);
+        support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
 
     // Writer first writes the length of compressed string, and then the actual
     // content.
@@ -1369,29 +1359,24 @@ Error IndexedInstrProfReader::readHeader() {
     if (Ptr + 2 * sizeof(uint64_t) > PtrEnd)
       return error(instrprof_error::truncated);
     const uint64_t NumTraces =
-        support::endian::readNext<uint64_t, llvm::endianness::little,
-                                  unaligned>(Ptr);
+        support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     TemporalProfTraceStreamSize =
-        support::endian::readNext<uint64_t, llvm::endianness::little,
-                                  unaligned>(Ptr);
+        support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     for (unsigned i = 0; i < NumTraces; i++) {
       // Expect at least two 64 bit fields: Weight and NumFunctions
       if (Ptr + 2 * sizeof(uint64_t) > PtrEnd)
         return error(instrprof_error::truncated);
       TemporalProfTraceTy Trace;
       Trace.Weight =
-          support::endian::readNext<uint64_t, llvm::endianness::little,
-                                    unaligned>(Ptr);
+          support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
       const uint64_t NumFunctions =
-          support::endian::readNext<uint64_t, llvm::endianness::little,
-                                    unaligned>(Ptr);
+          support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
       // Expect at least NumFunctions 64 bit fields
       if (Ptr + NumFunctions * sizeof(uint64_t) > PtrEnd)
         return error(instrprof_error::truncated);
       for (unsigned j = 0; j < NumFunctions; j++) {
         const uint64_t NameRef =
-            support::endian::readNext<uint64_t, llvm::endianness::little,
-                                      unaligned>(Ptr);
+            support::endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
         Trace.FunctionNameRefs.push_back(NameRef);
       }
       TemporalProfTraces.push_back(std::move(Trace));

diff --git a/llvm/lib/ProfileData/MemProf.cpp b/llvm/lib/ProfileData/MemProf.cpp
@@ -144,14 +144,14 @@ static IndexedMemProfRecord deserializeV0(const MemProfSchema &Schema,
 
   // Read the meminfo nodes.
   const uint64_t NumNodes =
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+      endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
   for (uint64_t I = 0; I < NumNodes; I++) {
     IndexedAllocationInfo Node;
     const uint64_t NumFrames =
-        endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     for (uint64_t J = 0; J < NumFrames; J++) {
       const FrameId Id =
-          endian::readNext<FrameId, llvm::endianness::little, unaligned>(Ptr);
+          endian::readNext<FrameId, llvm::endianness::little>(Ptr);
       Node.CallStack.push_back(Id);
     }
     Node.CSId = hashCallStack(Node.CallStack);
@@ -162,15 +162,15 @@ static IndexedMemProfRecord deserializeV0(const MemProfSchema &Schema,
 
   // Read the callsite information.
   const uint64_t NumCtxs =
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+      endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
   for (uint64_t J = 0; J < NumCtxs; J++) {
     const uint64_t NumFrames =
-        endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     llvm::SmallVector<FrameId> Frames;
     Frames.reserve(NumFrames);
     for (uint64_t K = 0; K < NumFrames; K++) {
       const FrameId Id =
-          endian::readNext<FrameId, llvm::endianness::little, unaligned>(Ptr);
+          endian::readNext<FrameId, llvm::endianness::little>(Ptr);
       Frames.push_back(Id);
     }
     Record.CallSites.push_back(Frames);
@@ -188,22 +188,21 @@ static IndexedMemProfRecord deserializeV2(const MemProfSchema &Schema,
 
   // Read the meminfo nodes.
   const uint64_t NumNodes =
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+      endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
   for (uint64_t I = 0; I < NumNodes; I++) {
     IndexedAllocationInfo Node;
-    Node.CSId =
-        endian::readNext<CallStackId, llvm::endianness::little, unaligned>(Ptr);
+    Node.CSId = endian::readNext<CallStackId, llvm::endianness::little>(Ptr);
     Node.Info.deserialize(Schema, Ptr);
     Ptr += PortableMemInfoBlock::serializedSize();
     Record.AllocSites.push_back(Node);
   }
 
   // Read the callsite information.
   const uint64_t NumCtxs =
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+      endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
   for (uint64_t J = 0; J < NumCtxs; J++) {
     CallStackId CSId =
-        endian::readNext<CallStackId, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<CallStackId, llvm::endianness::little>(Ptr);
     Record.CallSiteIds.push_back(CSId);
   }
 
@@ -263,7 +262,7 @@ Expected<MemProfSchema> readMemProfSchema(const unsigned char *&Buffer) {
 
   const unsigned char *Ptr = Buffer;
   const uint64_t NumSchemaIds =
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+      endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
   if (NumSchemaIds > static_cast<uint64_t>(Meta::Size)) {
     return make_error<InstrProfError>(instrprof_error::malformed,
                                       "memprof schema invalid");
@@ -272,7 +271,7 @@ Expected<MemProfSchema> readMemProfSchema(const unsigned char *&Buffer) {
   MemProfSchema Result;
   for (size_t I = 0; I < NumSchemaIds; I++) {
     const uint64_t Tag =
-        endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     if (Tag >= static_cast<uint64_t>(Meta::Size)) {
       return make_error<InstrProfError>(instrprof_error::malformed,
                                         "memprof schema invalid");

diff --git a/llvm/lib/ProfileData/MemProfReader.cpp b/llvm/lib/ProfileData/MemProfReader.cpp
@@ -86,7 +86,7 @@ llvm::SmallVector<SegmentEntry> readSegmentEntries(const char *Ptr) {
   using namespace support;
 
   const uint64_t NumItemsToRead =
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+      endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
   llvm::SmallVector<SegmentEntry> Items;
   for (uint64_t I = 0; I < NumItemsToRead; I++) {
     Items.push_back(*reinterpret_cast<const SegmentEntry *>(
@@ -100,11 +100,11 @@ readMemInfoBlocks(const char *Ptr) {
   using namespace support;
 
   const uint64_t NumItemsToRead =
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+      endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
   llvm::SmallVector<std::pair<uint64_t, MemInfoBlock>> Items;
   for (uint64_t I = 0; I < NumItemsToRead; I++) {
     const uint64_t Id =
-        endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     const MemInfoBlock MIB = *reinterpret_cast<const MemInfoBlock *>(Ptr);
     Items.push_back({Id, MIB});
     // Only increment by size of MIB since readNext implicitly increments.
@@ -117,20 +117,20 @@ CallStackMap readStackInfo(const char *Ptr) {
   using namespace support;
 
   const uint64_t NumItemsToRead =
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+      endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
   CallStackMap Items;
 
   for (uint64_t I = 0; I < NumItemsToRead; I++) {
     const uint64_t StackId =
-        endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
     const uint64_t NumPCs =
-        endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<uint64_t, llvm::endianness::little>(Ptr);
 
     SmallVector<uint64_t> CallStack;
     CallStack.reserve(NumPCs);
     for (uint64_t J = 0; J < NumPCs; J++) {
       CallStack.push_back(
-          endian::readNext<uint64_t, llvm::endianness::little, unaligned>(Ptr));
+          endian::readNext<uint64_t, llvm::endianness::little>(Ptr));
     }
 
     Items[StackId] = CallStack;

diff --git a/llvm/lib/ProfileData/SampleProfReader.cpp b/llvm/lib/ProfileData/SampleProfReader.cpp
@@ -503,7 +503,7 @@ ErrorOr<T> SampleProfileReaderBinary::readUnencodedNumber() {
   }
 
   using namespace support;
-  T Val = endian::readNext<T, llvm::endianness::little, unaligned>(Data);
+  T Val = endian::readNext<T, llvm::endianness::little>(Data);
   return Val;
 }
 

diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -13408,7 +13408,7 @@ static SDValue expandMul(SDNode *N, SelectionDAG &DAG,
   if (VT != Subtarget.getXLenVT())
     return SDValue();
 
-  if (!Subtarget.hasStdExtZba())
+  if (!Subtarget.hasStdExtZba() && !Subtarget.hasVendorXTHeadBa())
     return SDValue();
 
   ConstantSDNode *CNode = dyn_cast<ConstantSDNode>(N->getOperand(1));

diff --git a/llvm/lib/Target/TargetMachine.cpp b/llvm/lib/Target/TargetMachine.cpp
@@ -43,6 +43,12 @@ bool TargetMachine::isLargeGlobalValue(const GlobalValue *GVal) const {
   if (getTargetTriple().getArch() != Triple::x86_64)
     return false;
 
+  // Remaining logic below is ELF-specific. For other object file formats where
+  // the large code model is mostly used for JIT compilation, just look at the
+  // code model.
+  if (!getTargetTriple().isOSBinFormatELF())
+    return getCodeModel() == CodeModel::Large;
+
   auto *GO = GVal->getAliaseeObject();
 
   // Be conservative if we can't find an underlying GlobalObject.
@@ -51,9 +57,20 @@ bool TargetMachine::isLargeGlobalValue(const GlobalValue *GVal) const {
 
   auto *GV = dyn_cast<GlobalVariable>(GO);
 
+  auto IsPrefix = [](StringRef Name, StringRef Prefix) {
+    return Name.consume_front(Prefix) && (Name.empty() || Name[0] == '.');
+  };
+
   // Functions/GlobalIFuncs are only large under the large code model.
-  if (!GV)
+  if (!GV) {
+    // Handle explicit sections as we do for GlobalVariables with an explicit
+    // section, see comments below.
+    if (GO->hasSection()) {
+      StringRef Name = GO->getSection();
+      return IsPrefix(Name, ".ltext");
+    }
     return getCodeModel() == CodeModel::Large;
+  }
 
   if (GV->isThreadLocal())
     return false;
@@ -73,11 +90,8 @@ bool TargetMachine::isLargeGlobalValue(const GlobalValue *GVal) const {
   // data sections. The code model attribute overrides this above.
   if (GV->hasSection()) {
     StringRef Name = GV->getSection();
-    auto IsPrefix = [&](StringRef Prefix) {
-      StringRef S = Name;
-      return S.consume_front(Prefix) && (S.empty() || S[0] == '.');
-    };
-    return IsPrefix(".lbss") || IsPrefix(".ldata") || IsPrefix(".lrodata");
+    return IsPrefix(Name, ".lbss") || IsPrefix(Name, ".ldata") ||
+           IsPrefix(Name, ".lrodata");
   }
 
   // Respect large data threshold for medium and large code models.

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
@@ -1977,11 +1977,25 @@ Instruction *InstCombinerImpl::visitFPToSI(FPToSIInst &FI) {
 }
 
 Instruction *InstCombinerImpl::visitUIToFP(CastInst &CI) {
-  return commonCastTransforms(CI);
+  if (Instruction *R = commonCastTransforms(CI))
+    return R;
+  if (!CI.hasNonNeg() && isKnownNonNegative(CI.getOperand(0), SQ)) {
+    CI.setNonNeg();
+    return &CI;
+  }
+  return nullptr;
 }
 
 Instruction *InstCombinerImpl::visitSIToFP(CastInst &CI) {
-  return commonCastTransforms(CI);
+  if (Instruction *R = commonCastTransforms(CI))
+    return R;
+  if (isKnownNonNegative(CI.getOperand(0), SQ)) {
+    auto UI =
+        CastInst::Create(Instruction::UIToFP, CI.getOperand(0), CI.getType());
+    UI->setNonNeg(true);
+    return UI;
+  }
+  return nullptr;
 }
 
 Instruction *InstCombinerImpl::visitIntToPtr(IntToPtrInst &CI) {

diff --git a/llvm/lib/Transforms/Utils/LoopUtils.cpp b/llvm/lib/Transforms/Utils/LoopUtils.cpp
@@ -1034,6 +1034,15 @@ CmpInst::Predicate llvm::getMinMaxReductionPredicate(RecurKind RK) {
   }
 }
 
+Value *llvm::createAnyOfOp(IRBuilderBase &Builder, Value *StartVal,
+                           RecurKind RK, Value *Left, Value *Right) {
+  if (auto VTy = dyn_cast<VectorType>(Left->getType()))
+    StartVal = Builder.CreateVectorSplat(VTy->getElementCount(), StartVal);
+  Value *Cmp =
+      Builder.CreateCmp(CmpInst::ICMP_NE, Left, StartVal, "rdx.select.cmp");
+  return Builder.CreateSelect(Cmp, Left, Right, "rdx.select");
+}
+
 Value *llvm::createMinMaxOp(IRBuilderBase &Builder, RecurKind RK, Value *Left,
                             Value *Right) {
   Type *Ty = Left->getType();
@@ -1142,13 +1151,16 @@ Value *llvm::createAnyOfTargetReduction(IRBuilderBase &Builder, Value *Src,
     NewVal = SI->getTrueValue();
   }
 
+  // Create a splat vector with the new value and compare this to the vector
+  // we want to reduce.
+  ElementCount EC = cast<VectorType>(Src->getType())->getElementCount();
+  Value *Right = Builder.CreateVectorSplat(EC, InitVal);
+  Value *Cmp =
+      Builder.CreateCmp(CmpInst::ICMP_NE, Src, Right, "rdx.select.cmp");
+
   // If any predicate is true it means that we want to select the new value.
-  Value *AnyOf =
-      Src->getType()->isVectorTy() ? Builder.CreateOrReduce(Src) : Src;
-  // The compares in the loop may yield poison, which propagates through the
-  // bitwise ORs. Freeze it here before the condition is used.
-  AnyOf = Builder.CreateFreeze(AnyOf);
-  return Builder.CreateSelect(AnyOf, NewVal, InitVal, "rdx.select");
+  Cmp = Builder.CreateOrReduce(Cmp);
+  return Builder.CreateSelect(Cmp, NewVal, InitVal, "rdx.select");
 }
 
 Value *llvm::createSimpleTargetReduction(IRBuilderBase &Builder, Value *Src,

diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h b/llvm/lib/Transforms/Vectorize/LoopVectorizationPlanner.h
@@ -68,7 +68,9 @@ class VPBuilder {
 public:
   VPBuilder() = default;
   VPBuilder(VPBasicBlock *InsertBB) { setInsertPoint(InsertBB); }
-  VPBuilder(VPRecipeBase *InsertPt) { setInsertPoint(InsertPt); }
+  VPBuilder(VPRecipeBase *InsertPt) {
+    setInsertPoint(InsertPt->getParent(), InsertPt->getIterator());
+  }
 
   /// Clear the insertion point: created instructions will not be inserted into
   /// a block.

diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -3051,8 +3051,9 @@ PHINode *InnerLoopVectorizer::createInductionResumeValue(
   }
 
   // Create phi nodes to merge from the  backedge-taken check block.
-  PHINode *BCResumeVal = PHINode::Create(OrigPhi->getType(), 3, "bc.resume.val",
-                                         LoopScalarPreHeader->getFirstNonPHI());
+  PHINode *BCResumeVal =
+      PHINode::Create(OrigPhi->getType(), 3, "bc.resume.val",
+                      LoopScalarPreHeader->getTerminator()->getIterator());
   // Copy original phi DL over to the new one.
   BCResumeVal->setDebugLoc(OrigPhi->getDebugLoc());
 
@@ -7450,6 +7451,7 @@ static void createAndCollectMergePhiForReduction(
   auto *PhiR = cast<VPReductionPHIRecipe>(RedResult->getOperand(0));
   const RecurrenceDescriptor &RdxDesc = PhiR->getRecurrenceDescriptor();
 
+  TrackingVH<Value> ReductionStartValue = RdxDesc.getRecurrenceStartValue();
   Value *FinalValue =
       State.get(RedResult, VPIteration(State.UF - 1, VPLane::getFirstLane()));
   auto *ResumePhi =
@@ -7474,7 +7476,7 @@ static void createAndCollectMergePhiForReduction(
       BCBlockPhi->addIncoming(ResumePhi->getIncomingValueForBlock(Incoming),
                               Incoming);
     else
-      BCBlockPhi->addIncoming(RdxDesc.getRecurrenceStartValue(), Incoming);
+      BCBlockPhi->addIncoming(ReductionStartValue, Incoming);
   }
 
   auto *OrigPhi = cast<PHINode>(PhiR->getUnderlyingValue());
@@ -7767,10 +7769,11 @@ EpilogueVectorizerEpilogueLoop::createEpilogueVectorizedLoopSkeleton(
 
   // Now, compare the remaining count and if there aren't enough iterations to
   // execute the vectorized epilogue skip to the scalar part.
-  LoopVectorPreHeader->setName("vec.epilog.ph");
-  BasicBlock *VecEpilogueIterationCountCheck =
-      SplitBlock(LoopVectorPreHeader, LoopVectorPreHeader->begin(), DT, LI,
-                 nullptr, "vec.epilog.iter.check", true);
+  BasicBlock *VecEpilogueIterationCountCheck = LoopVectorPreHeader;
+  VecEpilogueIterationCountCheck->setName("vec.epilog.iter.check");
+  LoopVectorPreHeader =
+      SplitBlock(LoopVectorPreHeader, LoopVectorPreHeader->getTerminator(), DT,
+                 LI, nullptr, "vec.epilog.ph");
   emitMinimumVectorEpilogueIterCountCheck(LoopScalarPreHeader,
                                           VecEpilogueIterationCountCheck);
 
@@ -8893,10 +8896,6 @@ VPlanPtr LoopVectorizationPlanner::buildVPlan(VFRange &Range) {
 // A ComputeReductionResult recipe is added to the middle block, also for
 // in-loop reductions which compute their result in-loop, because generating
 // the subsequent bc.merge.rdx phi is driven by ComputeReductionResult recipes.
-//
-// Adjust AnyOf reductions; replace the reduction phi for the selected value
-// with a boolean reduction phi node to check if the condition is true in any
-// iteration. The final value is selected by the final ComputeReductionResult.
 void LoopVectorizationPlanner::adjustRecipesForReductions(
     VPBasicBlock *LatchVPBB, VPlanPtr &Plan, VPRecipeBuilder &RecipeBuilder,
     ElementCount MinVF) {
@@ -9071,41 +9070,6 @@ void LoopVectorizationPlanner::adjustRecipesForReductions(
       continue;
 
     const RecurrenceDescriptor &RdxDesc = PhiR->getRecurrenceDescriptor();
-    // Adjust AnyOf reductions; replace the reduction phi for the selected value
-    // with a boolean reduction phi node to check if the condition is true in
-    // any iteration. The final value is selected by the final
-    // ComputeReductionResult.
-    if (RecurrenceDescriptor::isAnyOfRecurrenceKind(
-            RdxDesc.getRecurrenceKind())) {
-      auto *Select = cast<VPRecipeBase>(*find_if(PhiR->users(), [](VPUser *U) {
-        return isa<VPWidenSelectRecipe>(U) ||
-               (isa<VPReplicateRecipe>(U) &&
-                cast<VPReplicateRecipe>(U)->getUnderlyingInstr()->getOpcode() ==
-                    Instruction::Select);
-      }));
-      VPValue *Cmp = Select->getOperand(0);
-      // If the compare is checking the reduction PHI node, adjust it to check
-      // the start value.
-      if (VPRecipeBase *CmpR = Cmp->getDefiningRecipe()) {
-        for (unsigned I = 0; I != CmpR->getNumOperands(); ++I)
-          if (CmpR->getOperand(I) == PhiR)
-            CmpR->setOperand(I, PhiR->getStartValue());
-      }
-      VPBuilder::InsertPointGuard Guard(Builder);
-      Builder.setInsertPoint(Select);
-
-      // If the true value of the select is the reduction phi, the new value is
-      // selected if the negated condition is true in any iteration.
-      if (Select->getOperand(1) == PhiR)
-        Cmp = Builder.createNot(Cmp);
-      VPValue *Or = Builder.createOr(PhiR, Cmp);
-      Select->getVPSingleValue()->replaceAllUsesWith(Or);
-
-      // Convert the reduction phi to operate on bools.
-      PhiR->setOperand(0, Plan->getOrAddLiveIn(ConstantInt::getFalse(
-                              OrigLoop->getHeader()->getContext())));
-    }
-
     // If tail is folded by masking, introduce selects between the phi
     // and the live-out instruction of each reduction, at the beginning of the
     // dedicated latch block.
@@ -9138,9 +9102,7 @@ void LoopVectorizationPlanner::adjustRecipesForReductions(
     // then extend the loop exit value to enable InstCombine to evaluate the
     // entire expression in the smaller type.
     Type *PhiTy = PhiR->getStartValue()->getLiveInIRValue()->getType();
-    if (MinVF.isVector() && PhiTy != RdxDesc.getRecurrenceType() &&
-        !RecurrenceDescriptor::isAnyOfRecurrenceKind(
-            RdxDesc.getRecurrenceKind())) {
+    if (MinVF.isVector() && PhiTy != RdxDesc.getRecurrenceType()) {
       assert(!PhiR->isInLoop() && "Unexpected truncated inloop reduction!");
       Type *RdxTy = RdxDesc.getRecurrenceType();
       auto *Trunc =
@@ -10181,19 +10143,9 @@ bool LoopVectorizePass::processLoop(Loop *L) {
           Value *ResumeV = nullptr;
           // TODO: Move setting of resume values to prepareToExecute.
           if (auto *ReductionPhi = dyn_cast<VPReductionPHIRecipe>(&R)) {
-            const RecurrenceDescriptor &RdxDesc =
-                ReductionPhi->getRecurrenceDescriptor();
-            RecurKind RK = RdxDesc.getRecurrenceKind();
-            ResumeV = ReductionResumeValues.find(&RdxDesc)->second;
-            if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) {
-              // VPReductionPHIRecipes for AnyOf reductions expect a boolean as
-              // start value; compare the final value from the main vector loop
-              // to the start value.
-              IRBuilder<> Builder(
-                  cast<Instruction>(ResumeV)->getParent()->getFirstNonPHI());
-              ResumeV = Builder.CreateICmpNE(ResumeV,
-                                             RdxDesc.getRecurrenceStartValue());
-            }
+            ResumeV = ReductionResumeValues
+                          .find(&ReductionPhi->getRecurrenceDescriptor())
+                          ->second;
           } else {
             // Create induction resume values for both widened pointer and
             // integer/fp inductions and update the start value of the induction

diff --git a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
@@ -501,8 +501,6 @@ Value *VPInstruction::generatePerPart(VPTransformState &State, unsigned Part) {
     // Reduce all of the unrolled parts into a single vector.
     Value *ReducedPartRdx = RdxParts[0];
     unsigned Op = RecurrenceDescriptor::getOpcode(RK);
-    if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK))
-      Op = Instruction::Or;
 
     if (PhiR->isOrdered()) {
       ReducedPartRdx = RdxParts[State.UF - 1];
@@ -515,16 +513,19 @@ Value *VPInstruction::generatePerPart(VPTransformState &State, unsigned Part) {
         if (Op != Instruction::ICmp && Op != Instruction::FCmp)
           ReducedPartRdx = Builder.CreateBinOp(
               (Instruction::BinaryOps)Op, RdxPart, ReducedPartRdx, "bin.rdx");
-        else
+        else if (RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) {
+          TrackingVH<Value> ReductionStartValue =
+              RdxDesc.getRecurrenceStartValue();
+          ReducedPartRdx = createAnyOfOp(Builder, ReductionStartValue, RK,
+                                         ReducedPartRdx, RdxPart);
+        } else
           ReducedPartRdx = createMinMaxOp(Builder, RK, ReducedPartRdx, RdxPart);
       }
     }
 
     // Create the reduction after the loop. Note that inloop reductions create
     // the target reduction in the loop using a Reduction recipe.
-    if ((State.VF.isVector() ||
-         RecurrenceDescriptor::isAnyOfRecurrenceKind(RK)) &&
-        !PhiR->isInLoop()) {
+    if (State.VF.isVector() && !PhiR->isInLoop()) {
       ReducedPartRdx =
           createTargetReduction(Builder, RdxDesc, ReducedPartRdx, OrigPhi);
       // If the reduction can be performed in a smaller type, we need to extend

diff --git a/llvm/lib/Transforms/Vectorize/VPlanValue.h b/llvm/lib/Transforms/Vectorize/VPlanValue.h
@@ -361,8 +361,8 @@ class VPDef {
     VPWidenMemoryInstructionSC,
     VPWidenSC,
     VPWidenSelectSC,
-    // START: Phi-like recipes. Need to be kept together.
     VPBlendSC,
+    // START: Phi-like recipes. Need to be kept together.
     VPWidenPHISC,
     VPPredInstPHISC,
     // START: SubclassID for recipes that inherit VPHeaderPHIRecipe.
@@ -376,7 +376,7 @@ class VPDef {
     VPReductionPHISC,
     // END: SubclassID for recipes that inherit VPHeaderPHIRecipe
     // END: Phi-like recipes
-    VPFirstPHISC = VPBlendSC,
+    VPFirstPHISC = VPWidenPHISC,
     VPFirstHeaderPHISC = VPCanonicalIVPHISC,
     VPLastHeaderPHISC = VPReductionPHISC,
     VPLastPHISC = VPReductionPHISC,

diff --git a/llvm/test/CodeGen/RISCV/rv64xtheadba.ll b/llvm/test/CodeGen/RISCV/rv64xtheadba.ll
@@ -268,6 +268,23 @@ define i64 @mul96(i64 %a) {
   ret i64 %c
 }
 
+define i64 @mul137(i64 %a) {
+; RV64I-LABEL: mul137:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    li a1, 137
+; RV64I-NEXT:    mul a0, a0, a1
+; RV64I-NEXT:    ret
+;
+; RV64XTHEADBA-LABEL: mul137:
+; RV64XTHEADBA:       # %bb.0:
+; RV64XTHEADBA-NEXT:    th.addsl a1, a0, a0, 3
+; RV64XTHEADBA-NEXT:    slli a0, a0, 7
+; RV64XTHEADBA-NEXT:    add a0, a0, a1
+; RV64XTHEADBA-NEXT:    ret
+  %c = mul i64 %a, 137
+  ret i64 %c
+}
+
 define i64 @mul160(i64 %a) {
 ; RV64I-LABEL: mul160:
 ; RV64I:       # %bb.0:

diff --git a/llvm/test/CodeGen/X86/code-model-elf-text-sections.ll b/llvm/test/CodeGen/X86/code-model-elf-text-sections.ll
@@ -13,13 +13,36 @@
 ; RUN: llvm-readelf -S %t | FileCheck %s --check-prefix=LARGE-DS
 
 ; SMALL: .text {{.*}} AX {{.*}}
+; SMALL: .ltext {{.*}} AXl {{.*}}
+; SMALL: .ltext.2 {{.*}} AXl {{.*}}
+; SMALL: .foo {{.*}} AX {{.*}}
 ; SMALL-DS: .text.func {{.*}} AX {{.*}}
+; SMALL-DS: .ltext {{.*}} AXl {{.*}}
+; SMALL-DS: .ltext.2 {{.*}} AXl {{.*}}
+; SMALL-DS: .foo {{.*}} AX {{.*}}
 ; LARGE: .ltext {{.*}} AXl {{.*}}
+; LARGE: .ltext.2 {{.*}} AXl {{.*}}
+; LARGE: .foo {{.*}} AX {{.*}}
 ; LARGE-DS: .ltext.func {{.*}} AXl {{.*}}
+; LARGE-DS: .ltext {{.*}} AXl {{.*}}
+; LARGE-DS: .ltext.2 {{.*}} AXl {{.*}}
+; LARGE-DS: .foo {{.*}} AX {{.*}}
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64--linux"
 
 define void @func() {
   ret void
 }
+
+define void @ltext() section ".ltext" {
+  ret void
+}
+
+define void @ltext2() section ".ltext.2" {
+  ret void
+}
+
+define void @foo() section ".foo" {
+  ret void
+}
diff --git a/llvm/test/Transforms/InstCombine/add-sitofp.ll b/llvm/test/Transforms/InstCombine/add-sitofp.ll
@@ -6,7 +6,7 @@ define double @x(i32 %a, i32 %b) {
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[A:%.*]], 24
 ; CHECK-NEXT:    [[N:%.*]] = and i32 [[M]], [[B:%.*]]
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nuw nsw i32 [[N]], 1
-; CHECK-NEXT:    [[P:%.*]] = uitofp i32 [[TMP1]] to double
+; CHECK-NEXT:    [[P:%.*]] = uitofp nneg i32 [[TMP1]] to double
 ; CHECK-NEXT:    ret double [[P]]
 ;
   %m = lshr i32 %a, 24
@@ -20,7 +20,7 @@ define double @test(i32 %a) {
 ; CHECK-LABEL: @test(
 ; CHECK-NEXT:    [[A_AND:%.*]] = and i32 [[A:%.*]], 1073741823
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nuw nsw i32 [[A_AND]], 1
-; CHECK-NEXT:    [[RES:%.*]] = uitofp i32 [[TMP1]] to double
+; CHECK-NEXT:    [[RES:%.*]] = uitofp nneg i32 [[TMP1]] to double
 ; CHECK-NEXT:    ret double [[RES]]
 ;
   ; Drop two highest bits to guarantee that %a + 1 doesn't overflow
@@ -33,7 +33,7 @@ define double @test(i32 %a) {
 define float @test_neg(i32 %a) {
 ; CHECK-LABEL: @test_neg(
 ; CHECK-NEXT:    [[A_AND:%.*]] = and i32 [[A:%.*]], 1073741823
-; CHECK-NEXT:    [[A_AND_FP:%.*]] = sitofp i32 [[A_AND]] to float
+; CHECK-NEXT:    [[A_AND_FP:%.*]] = uitofp nneg i32 [[A_AND]] to float
 ; CHECK-NEXT:    [[RES:%.*]] = fadd float [[A_AND_FP]], 1.000000e+00
 ; CHECK-NEXT:    ret float [[RES]]
 ;
@@ -49,7 +49,7 @@ define double @test_2(i32 %a, i32 %b) {
 ; CHECK-NEXT:    [[A_AND:%.*]] = and i32 [[A:%.*]], 1073741823
 ; CHECK-NEXT:    [[B_AND:%.*]] = and i32 [[B:%.*]], 1073741823
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nuw nsw i32 [[A_AND]], [[B_AND]]
-; CHECK-NEXT:    [[RES:%.*]] = uitofp i32 [[TMP1]] to double
+; CHECK-NEXT:    [[RES:%.*]] = uitofp nneg i32 [[TMP1]] to double
 ; CHECK-NEXT:    ret double [[RES]]
 ;
   ; Drop two highest bits to guarantee that %a + %b doesn't overflow
@@ -67,8 +67,8 @@ define float @test_2_neg(i32 %a, i32 %b) {
 ; CHECK-LABEL: @test_2_neg(
 ; CHECK-NEXT:    [[A_AND:%.*]] = and i32 [[A:%.*]], 1073741823
 ; CHECK-NEXT:    [[B_AND:%.*]] = and i32 [[B:%.*]], 1073741823
-; CHECK-NEXT:    [[A_AND_FP:%.*]] = sitofp i32 [[A_AND]] to float
-; CHECK-NEXT:    [[B_AND_FP:%.*]] = sitofp i32 [[B_AND]] to float
+; CHECK-NEXT:    [[A_AND_FP:%.*]] = uitofp nneg i32 [[A_AND]] to float
+; CHECK-NEXT:    [[B_AND_FP:%.*]] = uitofp nneg i32 [[B_AND]] to float
 ; CHECK-NEXT:    [[RES:%.*]] = fadd float [[A_AND_FP]], [[B_AND_FP]]
 ; CHECK-NEXT:    ret float [[RES]]
 ;
@@ -89,7 +89,7 @@ define float @test_3(i32 %a, i32 %b) {
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[A:%.*]], 24
 ; CHECK-NEXT:    [[N:%.*]] = and i32 [[M]], [[B:%.*]]
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nuw nsw i32 [[N]], 1
-; CHECK-NEXT:    [[P:%.*]] = uitofp i32 [[TMP1]] to float
+; CHECK-NEXT:    [[P:%.*]] = uitofp nneg i32 [[TMP1]] to float
 ; CHECK-NEXT:    ret float [[P]]
 ;
   %m = lshr i32 %a, 24
@@ -104,7 +104,7 @@ define <4 x double> @test_4(<4 x i32> %a, <4 x i32> %b) {
 ; CHECK-NEXT:    [[A_AND:%.*]] = and <4 x i32> [[A:%.*]], <i32 1073741823, i32 1073741823, i32 1073741823, i32 1073741823>
 ; CHECK-NEXT:    [[B_AND:%.*]] = and <4 x i32> [[B:%.*]], <i32 1073741823, i32 1073741823, i32 1073741823, i32 1073741823>
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nuw nsw <4 x i32> [[A_AND]], [[B_AND]]
-; CHECK-NEXT:    [[RES:%.*]] = uitofp <4 x i32> [[TMP1]] to <4 x double>
+; CHECK-NEXT:    [[RES:%.*]] = uitofp nneg <4 x i32> [[TMP1]] to <4 x double>
 ; CHECK-NEXT:    ret <4 x double> [[RES]]
 ;
   ; Drop two highest bits to guarantee that %a + %b doesn't overflow
@@ -122,8 +122,8 @@ define <4 x float> @test_4_neg(<4 x i32> %a, <4 x i32> %b) {
 ; CHECK-LABEL: @test_4_neg(
 ; CHECK-NEXT:    [[A_AND:%.*]] = and <4 x i32> [[A:%.*]], <i32 1073741823, i32 1073741823, i32 1073741823, i32 1073741823>
 ; CHECK-NEXT:    [[B_AND:%.*]] = and <4 x i32> [[B:%.*]], <i32 1073741823, i32 1073741823, i32 1073741823, i32 1073741823>
-; CHECK-NEXT:    [[A_AND_FP:%.*]] = sitofp <4 x i32> [[A_AND]] to <4 x float>
-; CHECK-NEXT:    [[B_AND_FP:%.*]] = sitofp <4 x i32> [[B_AND]] to <4 x float>
+; CHECK-NEXT:    [[A_AND_FP:%.*]] = uitofp nneg <4 x i32> [[A_AND]] to <4 x float>
+; CHECK-NEXT:    [[B_AND_FP:%.*]] = uitofp nneg <4 x i32> [[B_AND]] to <4 x float>
 ; CHECK-NEXT:    [[RES:%.*]] = fadd <4 x float> [[A_AND_FP]], [[B_AND_FP]]
 ; CHECK-NEXT:    ret <4 x float> [[RES]]
 ;

diff --git a/llvm/test/Transforms/InstCombine/binop-itofp.ll b/llvm/test/Transforms/InstCombine/binop-itofp.ll
@@ -21,7 +21,7 @@ define half @test_ui_ui_i8_add_fail_overflow(i8 noundef %x_in, i8 noundef %y_in)
 ; CHECK-LABEL: @test_ui_ui_i8_add_fail_overflow(
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 127
 ; CHECK-NEXT:    [[Y:%.*]] = and i8 [[Y_IN:%.*]], -127
-; CHECK-NEXT:    [[XF:%.*]] = uitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = uitofp i8 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fadd half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -49,7 +49,7 @@ define half @test_ui_ui_i8_add_C(i8 noundef %x_in) {
 define half @test_ui_ui_i8_add_C_fail_no_repr(i8 noundef %x_in) {
 ; CHECK-LABEL: @test_ui_ui_i8_add_C_fail_no_repr(
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 127
-; CHECK-NEXT:    [[XF:%.*]] = uitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fadd half [[XF]], 0xH57F8
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -62,7 +62,7 @@ define half @test_ui_ui_i8_add_C_fail_no_repr(i8 noundef %x_in) {
 define half @test_ui_ui_i8_add_C_fail_overflow(i8 noundef %x_in) {
 ; CHECK-LABEL: @test_ui_ui_i8_add_C_fail_overflow(
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 127
-; CHECK-NEXT:    [[XF:%.*]] = uitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fadd half [[XF]], 0xH5808
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -110,7 +110,7 @@ define half @test_ui_si_i8_add(i8 noundef %x_in, i8 noundef %y_in) {
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 63
 ; CHECK-NEXT:    [[Y:%.*]] = and i8 [[Y_IN:%.*]], 63
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nuw nsw i8 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = uitofp i8 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i8 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = and i8 %x_in, 63
@@ -140,7 +140,7 @@ define half @test_ui_si_i8_add_overflow(i8 noundef %x_in, i8 noundef %y_in) {
 define half @test_ui_ui_i8_sub_C(i8 noundef %x_in) {
 ; CHECK-LABEL: @test_ui_ui_i8_sub_C(
 ; CHECK-NEXT:    [[TMP1:%.*]] = and i8 [[X_IN:%.*]], 127
-; CHECK-NEXT:    [[R:%.*]] = uitofp i8 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i8 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = or i8 %x_in, 128
@@ -166,7 +166,7 @@ define half @test_si_si_i8_sub(i8 noundef %x_in, i8 noundef %y_in) {
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 63
 ; CHECK-NEXT:    [[Y:%.*]] = or i8 [[Y_IN:%.*]], -64
 ; CHECK-NEXT:    [[TMP1:%.*]] = sub nsw i8 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = sitofp i8 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i8 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = and i8 %x_in, 63
@@ -181,7 +181,7 @@ define half @test_si_si_i8_sub_fail_overflow(i8 noundef %x_in, i8 noundef %y_in)
 ; CHECK-LABEL: @test_si_si_i8_sub_fail_overflow(
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 63
 ; CHECK-NEXT:    [[Y:%.*]] = or i8 [[Y_IN:%.*]], -65
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = sitofp i8 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fsub half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -198,7 +198,7 @@ define half @test_si_si_i8_sub_C(i8 noundef %x_in) {
 ; CHECK-LABEL: @test_si_si_i8_sub_C(
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 63
 ; CHECK-NEXT:    [[TMP1:%.*]] = or disjoint i8 [[X]], 64
-; CHECK-NEXT:    [[R:%.*]] = sitofp i8 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i8 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = and i8 %x_in, 63
@@ -283,7 +283,7 @@ define half @test_ui_ui_i8_mul_C(i8 noundef %x_in) {
 define half @test_ui_ui_i8_mul_C_fail_overlow(i8 noundef %x_in) {
 ; CHECK-LABEL: @test_ui_ui_i8_mul_C_fail_overlow(
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 14
-; CHECK-NEXT:    [[XF:%.*]] = uitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], 0xH4CC0
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -315,7 +315,7 @@ define half @test_si_si_i8_mul_fail_maybe_zero(i8 noundef %x_in, i8 noundef %y_i
 ; CHECK-LABEL: @test_si_si_i8_mul_fail_maybe_zero(
 ; CHECK-NEXT:    [[X:%.*]] = and i8 [[X_IN:%.*]], 7
 ; CHECK-NEXT:    [[Y:%.*]] = or i8 [[Y_IN:%.*]], -8
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = sitofp i8 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -332,7 +332,7 @@ define half @test_si_si_i8_mul_C_fail_no_repr(i8 noundef %x_in) {
 ; CHECK-LABEL: @test_si_si_i8_mul_C_fail_no_repr(
 ; CHECK-NEXT:    [[XX:%.*]] = and i8 [[X_IN:%.*]], 6
 ; CHECK-NEXT:    [[X:%.*]] = or disjoint i8 [[XX]], 1
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], 0xHC780
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -347,7 +347,7 @@ define half @test_si_si_i8_mul_C_fail_overflow(i8 noundef %x_in) {
 ; CHECK-LABEL: @test_si_si_i8_mul_C_fail_overflow(
 ; CHECK-NEXT:    [[XX:%.*]] = and i8 [[X_IN:%.*]], 6
 ; CHECK-NEXT:    [[X:%.*]] = or disjoint i8 [[XX]], 1
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], 0xHCCC0
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -365,7 +365,7 @@ define half @test_ui_si_i8_mul(i8 noundef %x_in, i8 noundef %y_in) {
 ; CHECK-NEXT:    [[YY:%.*]] = and i8 [[Y_IN:%.*]], 7
 ; CHECK-NEXT:    [[Y:%.*]] = add nuw nsw i8 [[YY]], 1
 ; CHECK-NEXT:    [[TMP1:%.*]] = mul nuw nsw i8 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = uitofp i8 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i8 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %xx = and i8 %x_in, 6
@@ -384,7 +384,7 @@ define half @test_ui_si_i8_mul_fail_maybe_zero(i8 noundef %x_in, i8 noundef %y_i
 ; CHECK-NEXT:    [[X:%.*]] = add nuw nsw i8 [[XX]], 1
 ; CHECK-NEXT:    [[Y:%.*]] = and i8 [[Y_IN:%.*]], 7
 ; CHECK-NEXT:    [[TMP1:%.*]] = mul nuw nsw i8 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = uitofp i8 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i8 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %xx = and i8 %x_in, 7
@@ -401,7 +401,7 @@ define half @test_ui_si_i8_mul_fail_signed(i8 noundef %x_in, i8 noundef %y_in) {
 ; CHECK-NEXT:    [[XX:%.*]] = and i8 [[X_IN:%.*]], 7
 ; CHECK-NEXT:    [[X:%.*]] = add nuw nsw i8 [[XX]], 1
 ; CHECK-NEXT:    [[Y:%.*]] = or i8 [[Y_IN:%.*]], -4
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i8 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i8 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = uitofp i8 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -420,7 +420,7 @@ define half @test_ui_ui_i16_add(i16 noundef %x_in, i16 noundef %y_in) {
 ; CHECK-NEXT:    [[X:%.*]] = and i16 [[X_IN:%.*]], 2047
 ; CHECK-NEXT:    [[Y:%.*]] = and i16 [[Y_IN:%.*]], 2047
 ; CHECK-NEXT:    [[TMP1:%.*]] = add nuw nsw i16 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = uitofp i16 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i16 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = and i16 %x_in, 2047
@@ -435,8 +435,8 @@ define half @test_ui_ui_i16_add_fail_not_promotable(i16 noundef %x_in, i16 nound
 ; CHECK-LABEL: @test_ui_ui_i16_add_fail_not_promotable(
 ; CHECK-NEXT:    [[X:%.*]] = and i16 [[X_IN:%.*]], 2049
 ; CHECK-NEXT:    [[Y:%.*]] = and i16 [[Y_IN:%.*]], 2047
-; CHECK-NEXT:    [[XF:%.*]] = uitofp i16 [[X]] to half
-; CHECK-NEXT:    [[YF:%.*]] = uitofp i16 [[Y]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i16 [[X]] to half
+; CHECK-NEXT:    [[YF:%.*]] = uitofp nneg i16 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fadd half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -463,7 +463,7 @@ define half @test_ui_ui_i16_add_C(i16 noundef %x_in) {
 define half @test_ui_ui_i16_add_C_fail_overflow(i16 noundef %x_in) {
 ; CHECK-LABEL: @test_ui_ui_i16_add_C_fail_overflow(
 ; CHECK-NEXT:    [[X:%.*]] = and i16 [[X_IN:%.*]], 2047
-; CHECK-NEXT:    [[XF:%.*]] = uitofp i16 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i16 [[X]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fadd half [[XF]], 0xH7BD0
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -541,7 +541,7 @@ define half @test_si_si_i16_sub_fail_no_promotion(i16 noundef %x_in, i16 noundef
 ; CHECK-LABEL: @test_si_si_i16_sub_fail_no_promotion(
 ; CHECK-NEXT:    [[X:%.*]] = and i16 [[X_IN:%.*]], 2047
 ; CHECK-NEXT:    [[Y:%.*]] = or i16 [[Y_IN:%.*]], -2049
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i16 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i16 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = sitofp i16 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fsub half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -575,7 +575,7 @@ define half @test_ui_si_i16_sub_fail_maybe_signed(i16 noundef %x_in, i16 noundef
 ; CHECK-NEXT:    [[X:%.*]] = or i16 [[X_IN:%.*]], -2048
 ; CHECK-NEXT:    [[Y:%.*]] = and i16 [[Y_IN:%.*]], 2047
 ; CHECK-NEXT:    [[XF:%.*]] = uitofp i16 [[X]] to half
-; CHECK-NEXT:    [[YF:%.*]] = sitofp i16 [[Y]] to half
+; CHECK-NEXT:    [[YF:%.*]] = uitofp nneg i16 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fsub half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -607,8 +607,8 @@ define half @test_ui_ui_i16_mul_fail_no_promotion(i16 noundef %x_in, i16 noundef
 ; CHECK-LABEL: @test_ui_ui_i16_mul_fail_no_promotion(
 ; CHECK-NEXT:    [[X:%.*]] = and i16 [[X_IN:%.*]], 4095
 ; CHECK-NEXT:    [[Y:%.*]] = and i16 [[Y_IN:%.*]], 3
-; CHECK-NEXT:    [[XF:%.*]] = uitofp i16 [[X]] to half
-; CHECK-NEXT:    [[YF:%.*]] = uitofp i16 [[Y]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i16 [[X]] to half
+; CHECK-NEXT:    [[YF:%.*]] = uitofp nneg i16 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -643,7 +643,7 @@ define half @test_si_si_i16_mul_fail_overflow(i16 noundef %x_in, i16 noundef %y_
 ; CHECK-NEXT:    [[XX:%.*]] = and i16 [[X_IN:%.*]], 126
 ; CHECK-NEXT:    [[X:%.*]] = or disjoint i16 [[XX]], 1
 ; CHECK-NEXT:    [[Y:%.*]] = or i16 [[Y_IN:%.*]], -257
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i16 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i16 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = sitofp i16 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -690,7 +690,7 @@ define half @test_ui_si_i16_mul(i16 noundef %x_in, i16 noundef %y_in) {
 ; CHECK-NEXT:    [[YY:%.*]] = and i16 [[Y_IN:%.*]], 126
 ; CHECK-NEXT:    [[Y:%.*]] = or disjoint i16 [[YY]], 1
 ; CHECK-NEXT:    [[TMP1:%.*]] = mul nuw nsw i16 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = uitofp i16 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i16 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %xx = and i16 %x_in, 126
@@ -723,7 +723,7 @@ define half @test_ui_ui_i12_add_fail_overflow(i12 noundef %x_in, i12 noundef %y_
 ; CHECK-LABEL: @test_ui_ui_i12_add_fail_overflow(
 ; CHECK-NEXT:    [[X:%.*]] = and i12 [[X_IN:%.*]], 2047
 ; CHECK-NEXT:    [[Y:%.*]] = and i12 [[Y_IN:%.*]], -2047
-; CHECK-NEXT:    [[XF:%.*]] = uitofp i12 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i12 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = uitofp i12 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fadd half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -821,7 +821,7 @@ define half @test_si_si_i12_sub(i12 noundef %x_in, i12 noundef %y_in) {
 ; CHECK-NEXT:    [[X:%.*]] = and i12 [[X_IN:%.*]], 1023
 ; CHECK-NEXT:    [[Y:%.*]] = or i12 [[Y_IN:%.*]], -1024
 ; CHECK-NEXT:    [[TMP1:%.*]] = sub nsw i12 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = sitofp i12 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i12 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = and i12 %x_in, 1023
@@ -850,7 +850,7 @@ define half @test_ui_ui_i12_mul(i12 noundef %x_in, i12 noundef %y_in) {
 ; CHECK-NEXT:    [[X:%.*]] = and i12 [[X_IN:%.*]], 31
 ; CHECK-NEXT:    [[Y:%.*]] = and i12 [[Y_IN:%.*]], 63
 ; CHECK-NEXT:    [[TMP1:%.*]] = mul nuw nsw i12 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = uitofp i12 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i12 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = and i12 %x_in, 31
@@ -883,7 +883,7 @@ define half @test_ui_ui_i12_mul_C(i12 noundef %x_in) {
 ; CHECK-LABEL: @test_ui_ui_i12_mul_C(
 ; CHECK-NEXT:    [[X:%.*]] = shl i12 [[X_IN:%.*]], 6
 ; CHECK-NEXT:    [[TMP1:%.*]] = and i12 [[X]], 1984
-; CHECK-NEXT:    [[R:%.*]] = uitofp i12 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i12 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = and i12 %x_in, 31
@@ -915,7 +915,7 @@ define half @test_si_si_i12_mul_fail_overflow(i12 noundef %x_in, i12 noundef %y_
 ; CHECK-NEXT:    [[XX:%.*]] = and i12 [[X_IN:%.*]], 30
 ; CHECK-NEXT:    [[X:%.*]] = or disjoint i12 [[XX]], 1
 ; CHECK-NEXT:    [[Y:%.*]] = or i12 [[Y_IN:%.*]], -128
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i12 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i12 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = sitofp i12 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -933,7 +933,7 @@ define half @test_si_si_i12_mul_fail_maybe_non_zero(i12 noundef %x_in, i12 nound
 ; CHECK-LABEL: @test_si_si_i12_mul_fail_maybe_non_zero(
 ; CHECK-NEXT:    [[X:%.*]] = and i12 [[X_IN:%.*]], 30
 ; CHECK-NEXT:    [[Y:%.*]] = or i12 [[Y_IN:%.*]], -128
-; CHECK-NEXT:    [[XF:%.*]] = sitofp i12 [[X]] to half
+; CHECK-NEXT:    [[XF:%.*]] = uitofp nneg i12 [[X]] to half
 ; CHECK-NEXT:    [[YF:%.*]] = sitofp i12 [[Y]] to half
 ; CHECK-NEXT:    [[R:%.*]] = fmul half [[XF]], [[YF]]
 ; CHECK-NEXT:    ret half [[R]]
@@ -950,7 +950,7 @@ define half @test_si_si_i12_mul_C(i12 noundef %x_in) {
 ; CHECK-LABEL: @test_si_si_i12_mul_C(
 ; CHECK-NEXT:    [[X:%.*]] = or i12 [[X_IN:%.*]], -64
 ; CHECK-NEXT:    [[TMP1:%.*]] = mul nsw i12 [[X]], -16
-; CHECK-NEXT:    [[R:%.*]] = sitofp i12 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i12 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %x = or i12 %x_in, -64
@@ -979,7 +979,7 @@ define half @test_ui_si_i12_mul_nsw(i12 noundef %x_in, i12 noundef %y_in) {
 ; CHECK-NEXT:    [[YY:%.*]] = and i12 [[Y_IN:%.*]], 30
 ; CHECK-NEXT:    [[Y:%.*]] = or disjoint i12 [[YY]], 1
 ; CHECK-NEXT:    [[TMP1:%.*]] = mul nuw nsw i12 [[X]], [[Y]]
-; CHECK-NEXT:    [[R:%.*]] = uitofp i12 [[TMP1]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i12 [[TMP1]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %xx = and i12 %x_in, 31

diff --git a/llvm/test/Transforms/InstCombine/clamp-to-minmax.ll b/llvm/test/Transforms/InstCombine/clamp-to-minmax.ll
@@ -472,7 +472,7 @@ define float @ui32_clamp_and_cast_to_float(i32 %x) {
 ; CHECK-LABEL: @ui32_clamp_and_cast_to_float(
 ; CHECK-NEXT:    [[LO_CMP:%.*]] = icmp eq i32 [[X:%.*]], 0
 ; CHECK-NEXT:    [[MIN1:%.*]] = call i32 @llvm.umin.i32(i32 [[X]], i32 255)
-; CHECK-NEXT:    [[MIN:%.*]] = uitofp i32 [[MIN1]] to float
+; CHECK-NEXT:    [[MIN:%.*]] = uitofp nneg i32 [[MIN1]] to float
 ; CHECK-NEXT:    [[R:%.*]] = select i1 [[LO_CMP]], float 1.000000e+00, float [[MIN]]
 ; CHECK-NEXT:    ret float [[R]]
 ;
@@ -488,7 +488,7 @@ define float @ui64_clamp_and_cast_to_float(i64 %x) {
 ; CHECK-LABEL: @ui64_clamp_and_cast_to_float(
 ; CHECK-NEXT:    [[LO_CMP:%.*]] = icmp eq i64 [[X:%.*]], 0
 ; CHECK-NEXT:    [[MIN1:%.*]] = call i64 @llvm.umin.i64(i64 [[X]], i64 255)
-; CHECK-NEXT:    [[MIN:%.*]] = uitofp i64 [[MIN1]] to float
+; CHECK-NEXT:    [[MIN:%.*]] = uitofp nneg i64 [[MIN1]] to float
 ; CHECK-NEXT:    [[R:%.*]] = select i1 [[LO_CMP]], float 1.000000e+00, float [[MIN]]
 ; CHECK-NEXT:    ret float [[R]]
 ;
@@ -504,7 +504,7 @@ define float @mixed_clamp_to_float_1(i32 %x) {
 ; CHECK-LABEL: @mixed_clamp_to_float_1(
 ; CHECK-NEXT:    [[SI_MIN:%.*]] = call i32 @llvm.smin.i32(i32 [[X:%.*]], i32 255)
 ; CHECK-NEXT:    [[R1:%.*]] = call i32 @llvm.smax.i32(i32 [[SI_MIN]], i32 1)
-; CHECK-NEXT:    [[R:%.*]] = sitofp i32 [[R1]] to float
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[R1]] to float
 ; CHECK-NEXT:    ret float [[R]]
 ;
   %si_min_cmp = icmp sgt i32 %x, 255
@@ -539,7 +539,7 @@ define float @mixed_clamp_to_float_2(i32 %x) {
 ; CHECK-LABEL: @mixed_clamp_to_float_2(
 ; CHECK-NEXT:    [[SI_MIN:%.*]] = call i32 @llvm.smin.i32(i32 [[X:%.*]], i32 255)
 ; CHECK-NEXT:    [[R1:%.*]] = call i32 @llvm.smax.i32(i32 [[SI_MIN]], i32 1)
-; CHECK-NEXT:    [[R:%.*]] = sitofp i32 [[R1]] to float
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[R1]] to float
 ; CHECK-NEXT:    ret float [[R]]
 ;
   %si_min_cmp = icmp sgt i32 %x, 255
@@ -572,7 +572,7 @@ define <2 x float> @mixed_clamp_to_float_vec(<2 x i32> %x) {
 ; CHECK-LABEL: @mixed_clamp_to_float_vec(
 ; CHECK-NEXT:    [[SI_MIN:%.*]] = call <2 x i32> @llvm.smin.v2i32(<2 x i32> [[X:%.*]], <2 x i32> <i32 255, i32 255>)
 ; CHECK-NEXT:    [[R1:%.*]] = call <2 x i32> @llvm.smax.v2i32(<2 x i32> [[SI_MIN]], <2 x i32> <i32 1, i32 1>)
-; CHECK-NEXT:    [[R:%.*]] = sitofp <2 x i32> [[R1]] to <2 x float>
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg <2 x i32> [[R1]] to <2 x float>
 ; CHECK-NEXT:    ret <2 x float> [[R]]
 ;
   %si_min_cmp = icmp sgt <2 x i32> %x, <i32 255, i32 255>

diff --git a/llvm/test/Transforms/InstCombine/fpcast.ll b/llvm/test/Transforms/InstCombine/fpcast.ll
@@ -170,7 +170,7 @@ define half @sint_to_fptrunc(i32 %x) {
 define half @masked_sint_to_fptrunc1(i32 %x) {
 ; CHECK-LABEL: @masked_sint_to_fptrunc1(
 ; CHECK-NEXT:    [[M:%.*]] = and i32 [[X:%.*]], 16777215
-; CHECK-NEXT:    [[R:%.*]] = sitofp i32 [[M]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[M]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %m = and i32 %x, 16777215
@@ -182,7 +182,7 @@ define half @masked_sint_to_fptrunc1(i32 %x) {
 define half @masked_sint_to_fptrunc2(i32 %x) {
 ; CHECK-LABEL: @masked_sint_to_fptrunc2(
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[X:%.*]], 8
-; CHECK-NEXT:    [[R:%.*]] = sitofp i32 [[M]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[M]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %m = lshr i32 %x, 8
@@ -194,7 +194,7 @@ define half @masked_sint_to_fptrunc2(i32 %x) {
 define half @masked_sint_to_fptrunc3(i32 %x) {
 ; CHECK-LABEL: @masked_sint_to_fptrunc3(
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[X:%.*]], 7
-; CHECK-NEXT:    [[F:%.*]] = sitofp i32 [[M]] to float
+; CHECK-NEXT:    [[F:%.*]] = uitofp nneg i32 [[M]] to float
 ; CHECK-NEXT:    [[R:%.*]] = fptrunc float [[F]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -218,7 +218,7 @@ define double @sint_to_fpext(i32 %x) {
 define double @masked_sint_to_fpext1(i32 %x) {
 ; CHECK-LABEL: @masked_sint_to_fpext1(
 ; CHECK-NEXT:    [[M:%.*]] = and i32 [[X:%.*]], 16777215
-; CHECK-NEXT:    [[R:%.*]] = sitofp i32 [[M]] to double
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[M]] to double
 ; CHECK-NEXT:    ret double [[R]]
 ;
   %m = and i32 %x, 16777215
@@ -230,7 +230,7 @@ define double @masked_sint_to_fpext1(i32 %x) {
 define double @masked_sint_to_fpext2(i32 %x) {
 ; CHECK-LABEL: @masked_sint_to_fpext2(
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[X:%.*]], 8
-; CHECK-NEXT:    [[R:%.*]] = sitofp i32 [[M]] to double
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[M]] to double
 ; CHECK-NEXT:    ret double [[R]]
 ;
   %m = lshr i32 %x, 8
@@ -242,7 +242,7 @@ define double @masked_sint_to_fpext2(i32 %x) {
 define double @masked_sint_to_fpext3(i32 %x) {
 ; CHECK-LABEL: @masked_sint_to_fpext3(
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[X:%.*]], 7
-; CHECK-NEXT:    [[F:%.*]] = sitofp i32 [[M]] to float
+; CHECK-NEXT:    [[F:%.*]] = uitofp nneg i32 [[M]] to float
 ; CHECK-NEXT:    [[R:%.*]] = fpext float [[F]] to double
 ; CHECK-NEXT:    ret double [[R]]
 ;
@@ -266,7 +266,7 @@ define half @uint_to_fptrunc(i32 %x) {
 define half @masked_uint_to_fptrunc1(i32 %x) {
 ; CHECK-LABEL: @masked_uint_to_fptrunc1(
 ; CHECK-NEXT:    [[M:%.*]] = and i32 [[X:%.*]], 16777215
-; CHECK-NEXT:    [[R:%.*]] = uitofp i32 [[M]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[M]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %m = and i32 %x, 16777215
@@ -278,7 +278,7 @@ define half @masked_uint_to_fptrunc1(i32 %x) {
 define half @masked_uint_to_fptrunc2(i32 %x) {
 ; CHECK-LABEL: @masked_uint_to_fptrunc2(
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[X:%.*]], 8
-; CHECK-NEXT:    [[R:%.*]] = uitofp i32 [[M]] to half
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[M]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
   %m = lshr i32 %x, 8
@@ -290,7 +290,7 @@ define half @masked_uint_to_fptrunc2(i32 %x) {
 define half @masked_uint_to_fptrunc3(i32 %x) {
 ; CHECK-LABEL: @masked_uint_to_fptrunc3(
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[X:%.*]], 7
-; CHECK-NEXT:    [[F:%.*]] = uitofp i32 [[M]] to float
+; CHECK-NEXT:    [[F:%.*]] = uitofp nneg i32 [[M]] to float
 ; CHECK-NEXT:    [[R:%.*]] = fptrunc float [[F]] to half
 ; CHECK-NEXT:    ret half [[R]]
 ;
@@ -314,7 +314,7 @@ define double @uint_to_fpext(i32 %x) {
 define double @masked_uint_to_fpext1(i32 %x) {
 ; CHECK-LABEL: @masked_uint_to_fpext1(
 ; CHECK-NEXT:    [[M:%.*]] = and i32 [[X:%.*]], 16777215
-; CHECK-NEXT:    [[R:%.*]] = uitofp i32 [[M]] to double
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[M]] to double
 ; CHECK-NEXT:    ret double [[R]]
 ;
   %m = and i32 %x, 16777215
@@ -326,7 +326,7 @@ define double @masked_uint_to_fpext1(i32 %x) {
 define double @masked_uint_to_fpext2(i32 %x) {
 ; CHECK-LABEL: @masked_uint_to_fpext2(
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[X:%.*]], 8
-; CHECK-NEXT:    [[R:%.*]] = uitofp i32 [[M]] to double
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[M]] to double
 ; CHECK-NEXT:    ret double [[R]]
 ;
   %m = lshr i32 %x, 8
@@ -338,7 +338,7 @@ define double @masked_uint_to_fpext2(i32 %x) {
 define double @masked_uint_to_fpext3(i32 %x) {
 ; CHECK-LABEL: @masked_uint_to_fpext3(
 ; CHECK-NEXT:    [[M:%.*]] = lshr i32 [[X:%.*]], 7
-; CHECK-NEXT:    [[F:%.*]] = uitofp i32 [[M]] to float
+; CHECK-NEXT:    [[F:%.*]] = uitofp nneg i32 [[M]] to float
 ; CHECK-NEXT:    [[R:%.*]] = fpext float [[F]] to double
 ; CHECK-NEXT:    ret double [[R]]
 ;

diff --git a/llvm/test/Transforms/InstCombine/known-fpclass-reduce-signbit.ll b/llvm/test/Transforms/InstCombine/known-fpclass-reduce-signbit.ll
@@ -0,0 +1,119 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
+; RUN: opt < %s -S -passes=instcombine | FileCheck %s
+
+define i1 @vector_reduce_maximum_signbit(<4 x double> nofpclass(nan nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_maximum_signbit
+; CHECK-SAME: (<4 x double> nofpclass(nan nzero) [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call double @llvm.vector.reduce.fmaximum.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+define i1 @vector_reduce_maximum_signbit_fail_maybe_nan(<4 x double> nofpclass(nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_maximum_signbit_fail_maybe_nan
+; CHECK-SAME: (<4 x double> nofpclass(nzero) [[X:%.*]]) {
+; CHECK-NEXT:    [[X_ABS:%.*]] = call <4 x double> @llvm.fabs.v4f64(<4 x double> [[X]])
+; CHECK-NEXT:    [[OP:%.*]] = call double @llvm.vector.reduce.fmaximum.v4f64(<4 x double> [[X_ABS]])
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp oge double [[OP]], 0.000000e+00
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call double @llvm.vector.reduce.fmaximum.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+
+define i1 @vector_reduce_minimum_signbit(<4 x double> nofpclass(nan nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_minimum_signbit
+; CHECK-SAME: (<4 x double> nofpclass(nan nzero) [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call double @llvm.vector.reduce.fminimum.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+define i1 @vector_reduce_minimum_signbit_fail_maybe_nan(<4 x double> nofpclass(nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_minimum_signbit_fail_maybe_nan
+; CHECK-SAME: (<4 x double> nofpclass(nzero) [[X:%.*]]) {
+; CHECK-NEXT:    [[X_ABS:%.*]] = call <4 x double> @llvm.fabs.v4f64(<4 x double> [[X]])
+; CHECK-NEXT:    [[OP:%.*]] = call double @llvm.vector.reduce.fminimum.v4f64(<4 x double> [[X_ABS]])
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp oge double [[OP]], 0.000000e+00
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call double @llvm.vector.reduce.fminimum.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+define i1 @vector_reduce_max_signbit(<4 x double> nofpclass(nan nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_max_signbit
+; CHECK-SAME: (<4 x double> nofpclass(nan nzero) [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+define i1 @vector_reduce_max_signbit_fail_maybe_nan(<4 x double> nofpclass(nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_max_signbit_fail_maybe_nan
+; CHECK-SAME: (<4 x double> nofpclass(nzero) [[X:%.*]]) {
+; CHECK-NEXT:    [[X_ABS:%.*]] = call <4 x double> @llvm.fabs.v4f64(<4 x double> [[X]])
+; CHECK-NEXT:    [[OP:%.*]] = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> [[X_ABS]])
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp oge double [[OP]], 0.000000e+00
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+
+define i1 @vector_reduce_min_signbit(<4 x double> nofpclass(nan nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_min_signbit
+; CHECK-SAME: (<4 x double> nofpclass(nan nzero) [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+define i1 @vector_reduce_min_signbit_fail_maybe_nan(<4 x double> nofpclass(nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_min_signbit_fail_maybe_nan
+; CHECK-SAME: (<4 x double> nofpclass(nzero) [[X:%.*]]) {
+; CHECK-NEXT:    [[X_ABS:%.*]] = call <4 x double> @llvm.fabs.v4f64(<4 x double> [[X]])
+; CHECK-NEXT:    [[OP:%.*]] = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> [[X_ABS]])
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp oge double [[OP]], 0.000000e+00
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+
+
+define i1 @vector_reduce_min_signbit_nnan_from_fmf(<4 x double> nofpclass(nzero) %x) {
+; CHECK-LABEL: define i1 @vector_reduce_min_signbit_nnan_from_fmf
+; CHECK-SAME: (<4 x double> nofpclass(nzero) [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %x.abs = call <4 x double> @llvm.fabs.v4f64(<4 x double> %x)
+  %op = call nnan double @llvm.vector.reduce.fmin.v4f64(<4 x double> %x.abs)
+  %cmp = fcmp oge double %op, 0.0
+  ret i1 %cmp
+}
+
+
diff --git a/llvm/test/Transforms/InstCombine/minmax-fold.ll b/llvm/test/Transforms/InstCombine/minmax-fold.ll
@@ -131,7 +131,7 @@ define i64 @t9(i32 %a) {
 define float @t10(i32 %x) {
 ; CHECK-LABEL: @t10(
 ; CHECK-NEXT:    [[R1:%.*]] = call i32 @llvm.smax.i32(i32 [[X:%.*]], i32 255)
-; CHECK-NEXT:    [[R:%.*]] = sitofp i32 [[R1]] to float
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i32 [[R1]] to float
 ; CHECK-NEXT:    ret float [[R]]
 ;
   %f_x = sitofp i32 %x to float
@@ -143,7 +143,7 @@ define float @t10(i32 %x) {
 define float @t11(i64 %x) {
 ; CHECK-LABEL: @t11(
 ; CHECK-NEXT:    [[R1:%.*]] = call i64 @llvm.smax.i64(i64 [[X:%.*]], i64 255)
-; CHECK-NEXT:    [[R:%.*]] = sitofp i64 [[R1]] to float
+; CHECK-NEXT:    [[R:%.*]] = uitofp nneg i64 [[R1]] to float
 ; CHECK-NEXT:    ret float [[R]]
 ;
   %f_x = sitofp i64 %x to float
@@ -526,7 +526,7 @@ falselabel:
 define double @PR31751_umin1(i32 %x) {
 ; CHECK-LABEL: @PR31751_umin1(
 ; CHECK-NEXT:    [[SEL:%.*]] = call i32 @llvm.umin.i32(i32 [[X:%.*]], i32 2147483647)
-; CHECK-NEXT:    [[CONV:%.*]] = sitofp i32 [[SEL]] to double
+; CHECK-NEXT:    [[CONV:%.*]] = uitofp nneg i32 [[SEL]] to double
 ; CHECK-NEXT:    ret double [[CONV]]
 ;
   %cmp = icmp slt i32 %x, 0
@@ -538,7 +538,7 @@ define double @PR31751_umin1(i32 %x) {
 define double @PR31751_umin2(i32 %x) {
 ; CHECK-LABEL: @PR31751_umin2(
 ; CHECK-NEXT:    [[SEL:%.*]] = call i32 @llvm.umin.i32(i32 [[X:%.*]], i32 2147483647)
-; CHECK-NEXT:    [[CONV:%.*]] = sitofp i32 [[SEL]] to double
+; CHECK-NEXT:    [[CONV:%.*]] = uitofp nneg i32 [[SEL]] to double
 ; CHECK-NEXT:    ret double [[CONV]]
 ;
   %cmp = icmp ult i32 %x, 2147483647
@@ -550,7 +550,7 @@ define double @PR31751_umin2(i32 %x) {
 define double @PR31751_umin3(i32 %x) {
 ; CHECK-LABEL: @PR31751_umin3(
 ; CHECK-NEXT:    [[SEL:%.*]] = call i32 @llvm.umin.i32(i32 [[X:%.*]], i32 2147483647)
-; CHECK-NEXT:    [[CONV:%.*]] = sitofp i32 [[SEL]] to double
+; CHECK-NEXT:    [[CONV:%.*]] = uitofp nneg i32 [[SEL]] to double
 ; CHECK-NEXT:    ret double [[CONV]]
 ;
   %cmp = icmp ugt i32 %x, 2147483647

diff --git a/llvm/test/Transforms/InstCombine/minmax-fp.ll b/llvm/test/Transforms/InstCombine/minmax-fp.ll
@@ -257,7 +257,7 @@ define double @t16(i32 %x) {
 define double @t17(i32 %x) {
 ; CHECK-LABEL: @t17(
 ; CHECK-NEXT:    [[SEL1:%.*]] = call i32 @llvm.smax.i32(i32 [[X:%.*]], i32 2)
-; CHECK-NEXT:    [[SEL:%.*]] = sitofp i32 [[SEL1]] to double
+; CHECK-NEXT:    [[SEL:%.*]] = uitofp nneg i32 [[SEL1]] to double
 ; CHECK-NEXT:    ret double [[SEL]]
 ;
   %cmp = icmp sgt i32 %x, 2

diff --git a/llvm/test/Transforms/InstCombine/pr27236.ll b/llvm/test/Transforms/InstCombine/pr27236.ll
@@ -4,7 +4,7 @@
 define float @test1(i32 %scale) {
 ; CHECK-LABEL: @test1(
 ; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @llvm.smax.i32(i32 [[SCALE:%.*]], i32 1)
-; CHECK-NEXT:    [[TMP2:%.*]] = sitofp i32 [[TMP1]] to float
+; CHECK-NEXT:    [[TMP2:%.*]] = uitofp nneg i32 [[TMP1]] to float
 ; CHECK-NEXT:    ret float [[TMP2]]
 ;
   %1 = icmp sgt i32 1, %scale

diff --git a/llvm/test/Transforms/InstCombine/sitofp.ll b/llvm/test/Transforms/InstCombine/sitofp.ll
@@ -256,7 +256,7 @@ define i25 @consider_lowbits_masked_input(i25 %A) {
 define i32 @overflow_masked_input(i32 %A) {
 ; CHECK-LABEL: @overflow_masked_input(
 ; CHECK-NEXT:    [[M:%.*]] = and i32 [[A:%.*]], 16777217
-; CHECK-NEXT:    [[B:%.*]] = uitofp i32 [[M]] to float
+; CHECK-NEXT:    [[B:%.*]] = uitofp nneg i32 [[M]] to float
 ; CHECK-NEXT:    [[C:%.*]] = fptoui float [[B]] to i32
 ; CHECK-NEXT:    ret i32 [[C]]
 ;

diff --git a/llvm/test/Transforms/InstSimplify/known-never-infinity.ll b/llvm/test/Transforms/InstSimplify/known-never-infinity.ll
@@ -1109,6 +1109,106 @@ define float @fcmp_ult_neginf_implies_class_assert(float %arg) {
   ret float %mul_by_zero
 }
 
+define i1 @isKnownNeverInfinity_vector_reduce_maximum(<4 x double> %x) {
+; CHECK-LABEL: define i1 @isKnownNeverInfinity_vector_reduce_maximum
+; CHECK-SAME: (<4 x double> [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %ninf.x = fadd ninf <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
+  %op = call double @llvm.vector.reduce.fmaximum.v4f64(<4 x double> %ninf.x)
+  %cmp = fcmp une double %op, 0x7ff0000000000000
+  ret i1 %cmp
+}
+
+define i1 @isKnownNeverInfinity_vector_reduce_maximum_fail(<4 x double> %x) {
+; CHECK-LABEL: define i1 @isKnownNeverInfinity_vector_reduce_maximum_fail
+; CHECK-SAME: (<4 x double> [[X:%.*]]) {
+; CHECK-NEXT:    [[NINF_X:%.*]] = fadd <4 x double> [[X]], <double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00>
+; CHECK-NEXT:    [[OP:%.*]] = call double @llvm.vector.reduce.fmaximum.v4f64(<4 x double> [[NINF_X]])
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp une double [[OP]], 0x7FF0000000000000
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %ninf.x = fadd <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
+  %op = call double @llvm.vector.reduce.fmaximum.v4f64(<4 x double> %ninf.x)
+  %cmp = fcmp une double %op, 0x7ff0000000000000
+  ret i1 %cmp
+}
+
+define i1 @isKnownNeverInfinity_vector_reduce_minimum(<4 x double> %x) {
+; CHECK-LABEL: define i1 @isKnownNeverInfinity_vector_reduce_minimum
+; CHECK-SAME: (<4 x double> [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %ninf.x = fadd ninf <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
+  %op = call double @llvm.vector.reduce.fminimum.v4f64(<4 x double> %ninf.x)
+  %cmp = fcmp une double %op, 0x7ff0000000000000
+  ret i1 %cmp
+}
+
+define i1 @isKnownNeverInfinity_vector_reduce_minimum_fail(<4 x double> %x) {
+; CHECK-LABEL: define i1 @isKnownNeverInfinity_vector_reduce_minimum_fail
+; CHECK-SAME: (<4 x double> [[X:%.*]]) {
+; CHECK-NEXT:    [[NINF_X:%.*]] = fadd <4 x double> [[X]], <double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00>
+; CHECK-NEXT:    [[OP:%.*]] = call double @llvm.vector.reduce.fminimum.v4f64(<4 x double> [[NINF_X]])
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp une double [[OP]], 0x7FF0000000000000
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %ninf.x = fadd <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
+  %op = call double @llvm.vector.reduce.fminimum.v4f64(<4 x double> %ninf.x)
+  %cmp = fcmp une double %op, 0x7ff0000000000000
+  ret i1 %cmp
+}
+
+define i1 @isKnownNeverInfinity_vector_reduce_fmax(<4 x double> %x) {
+; CHECK-LABEL: define i1 @isKnownNeverInfinity_vector_reduce_fmax
+; CHECK-SAME: (<4 x double> [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %ninf.x = fadd ninf <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
+  %op = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> %ninf.x)
+  %cmp = fcmp une double %op, 0x7ff0000000000000
+  ret i1 %cmp
+}
+
+define i1 @isKnownNeverInfinity_vector_reduce_fmax_fail(<4 x double> %x) {
+; CHECK-LABEL: define i1 @isKnownNeverInfinity_vector_reduce_fmax_fail
+; CHECK-SAME: (<4 x double> [[X:%.*]]) {
+; CHECK-NEXT:    [[NINF_X:%.*]] = fadd <4 x double> [[X]], <double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00>
+; CHECK-NEXT:    [[OP:%.*]] = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> [[NINF_X]])
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp une double [[OP]], 0x7FF0000000000000
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %ninf.x = fadd <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
+  %op = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> %ninf.x)
+  %cmp = fcmp une double %op, 0x7ff0000000000000
+  ret i1 %cmp
+}
+
+define i1 @isKnownNeverInfinity_vector_reduce_fmin(<4 x double> %x) {
+; CHECK-LABEL: define i1 @isKnownNeverInfinity_vector_reduce_fmin
+; CHECK-SAME: (<4 x double> [[X:%.*]]) {
+; CHECK-NEXT:    ret i1 true
+;
+  %ninf.x = fadd ninf <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
+  %op = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> %ninf.x)
+  %cmp = fcmp une double %op, 0x7ff0000000000000
+  ret i1 %cmp
+}
+
+define i1 @isKnownNeverInfinity_vector_reduce_fmin_fail(<4 x double> %x) {
+; CHECK-LABEL: define i1 @isKnownNeverInfinity_vector_reduce_fmin_fail
+; CHECK-SAME: (<4 x double> [[X:%.*]]) {
+; CHECK-NEXT:    [[NINF_X:%.*]] = fadd <4 x double> [[X]], <double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00>
+; CHECK-NEXT:    [[OP:%.*]] = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> [[NINF_X]])
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp une double [[OP]], 0x7FF0000000000000
+; CHECK-NEXT:    ret i1 [[CMP]]
+;
+  %ninf.x = fadd <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
+  %op = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> %ninf.x)
+  %cmp = fcmp une double %op, 0x7ff0000000000000
+  ret i1 %cmp
+}
+
 declare double @llvm.arithmetic.fence.f64(double)
 declare double @llvm.canonicalize.f64(double)
 declare double @llvm.ceil.f64(double)

diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-select-cmp.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-select-cmp.ll
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/select-cmp-reduction.ll b/llvm/test/Transforms/LoopVectorize/RISCV/select-cmp-reduction.ll
diff --git a/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll b/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
@@ -25,7 +25,7 @@ define void @fp_iv_loop1(ptr noalias nocapture %A, i32 %N) #0 {
 ; AUTO_VEC-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY:%.*]], label [[VECTOR_PH:%.*]]
 ; AUTO_VEC:       vector.ph:
 ; AUTO_VEC-NEXT:    [[N_VEC:%.*]] = and i64 [[ZEXT]], 2147483616
-; AUTO_VEC-NEXT:    [[DOTCAST:%.*]] = sitofp i64 [[N_VEC]] to float
+; AUTO_VEC-NEXT:    [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
 ; AUTO_VEC-NEXT:    [[TMP0:%.*]] = fmul fast float [[DOTCAST]], 5.000000e-01
 ; AUTO_VEC-NEXT:    [[IND_END:%.*]] = fadd fast float [[TMP0]], 1.000000e+00
 ; AUTO_VEC-NEXT:    br label [[VECTOR_BODY:%.*]]
@@ -201,7 +201,7 @@ define double @external_use_with_fast_math(ptr %a, i64 %n) {
 ; AUTO_VEC-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY:%.*]], label [[VECTOR_PH:%.*]]
 ; AUTO_VEC:       vector.ph:
 ; AUTO_VEC-NEXT:    [[N_VEC:%.*]] = and i64 [[SMAX]], 9223372036854775792
-; AUTO_VEC-NEXT:    [[DOTCAST:%.*]] = sitofp i64 [[N_VEC]] to double
+; AUTO_VEC-NEXT:    [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to double
 ; AUTO_VEC-NEXT:    [[TMP0:%.*]] = fmul fast double [[DOTCAST]], 3.000000e+00
 ; AUTO_VEC-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; AUTO_VEC:       vector.body:
@@ -366,7 +366,7 @@ define void @fadd_reassoc_FMF(ptr nocapture %p, i32 %N) {
 ; AUTO_VEC-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY:%.*]], label [[VECTOR_PH:%.*]]
 ; AUTO_VEC:       vector.ph:
 ; AUTO_VEC-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP0]], 4294967264
-; AUTO_VEC-NEXT:    [[DOTCAST:%.*]] = sitofp i64 [[N_VEC]] to float
+; AUTO_VEC-NEXT:    [[DOTCAST:%.*]] = uitofp nneg i64 [[N_VEC]] to float
 ; AUTO_VEC-NEXT:    [[TMP1:%.*]] = fmul reassoc float [[DOTCAST]], 4.200000e+01
 ; AUTO_VEC-NEXT:    [[IND_END:%.*]] = fadd reassoc float [[TMP1]], 1.000000e+00
 ; AUTO_VEC-NEXT:    br label [[VECTOR_BODY:%.*]]

diff --git a/llvm/test/Transforms/LoopVectorize/blend-in-header.ll b/llvm/test/Transforms/LoopVectorize/blend-in-header.ll
@@ -0,0 +1,233 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
+; RUN: opt -p loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -S %s | FileCheck %s
+
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128-ni:1-p2:32:8:8:32-ni:2"
+
+; Test with blend recipe in header VPBB, from
+; https://github.com/llvm/llvm-project/issues/88297.
+define i64 @pr88297() {
+; CHECK-LABEL: define i64 @pr88297() {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp eq i32 [[INDEX_NEXT]], 1000
+; CHECK-NEXT:    br i1 [[TMP0]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK:       middle.block:
+; CHECK-NEXT:    br i1 false, label [[EXIT:%.*]], label [[SCALAR_PH]]
+; CHECK:       scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
+; CHECK:       loop.header:
+; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
+; CHECK-NEXT:    br i1 false, label [[LOOP_LATCH]], label [[THEN:%.*]]
+; CHECK:       then:
+; CHECK-NEXT:    br label [[LOOP_LATCH]]
+; CHECK:       loop.latch:
+; CHECK-NEXT:    [[R:%.*]] = phi i64 [ 1, [[THEN]] ], [ 0, [[LOOP_HEADER]] ]
+; CHECK-NEXT:    [[IV_NEXT]] = add i32 [[IV]], 1
+; CHECK-NEXT:    [[ICMP:%.*]] = icmp sgt i32 [[IV]], 1000
+; CHECK-NEXT:    br i1 [[ICMP]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[R_LCSSA:%.*]] = phi i64 [ [[R]], [[LOOP_LATCH]] ], [ 1, [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    ret i64 [[R_LCSSA]]
+;
+entry:
+  br label %loop.header
+
+loop.header:
+  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  br i1 false, label %loop.latch, label %then
+
+then:
+  br label %loop.latch
+
+loop.latch:
+  %r = phi i64 [ 1, %then ], [ 0, %loop.header ]
+  %iv.next = add i32 %iv, 1
+  %icmp = icmp sgt i32 %iv, 1000
+  br i1 %icmp, label %exit, label %loop.header
+
+exit:
+  %r.lcssa = phi i64 [ %r, %loop.latch ]
+  ret i64 %r.lcssa
+}
+
+define i64 @pr88297_incoming_ops_reordered() {
+; CHECK-LABEL: define i64 @pr88297_incoming_ops_reordered() {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp eq i32 [[INDEX_NEXT]], 1000
+; CHECK-NEXT:    br i1 [[TMP0]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK:       middle.block:
+; CHECK-NEXT:    br i1 false, label [[EXIT:%.*]], label [[SCALAR_PH]]
+; CHECK:       scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
+; CHECK:       loop.header:
+; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
+; CHECK-NEXT:    br i1 false, label [[LOOP_LATCH]], label [[THEN:%.*]]
+; CHECK:       then:
+; CHECK-NEXT:    br label [[LOOP_LATCH]]
+; CHECK:       loop.latch:
+; CHECK-NEXT:    [[R:%.*]] = phi i64 [ 0, [[LOOP_HEADER]] ], [ 1, [[THEN]] ]
+; CHECK-NEXT:    [[IV_NEXT]] = add i32 [[IV]], 1
+; CHECK-NEXT:    [[ICMP:%.*]] = icmp sgt i32 [[IV]], 1000
+; CHECK-NEXT:    br i1 [[ICMP]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[R_LCSSA:%.*]] = phi i64 [ [[R]], [[LOOP_LATCH]] ], [ 1, [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    ret i64 [[R_LCSSA]]
+;
+entry:
+  br label %loop.header
+
+loop.header:
+  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  br i1 false, label %loop.latch, label %then
+
+then:
+  br label %loop.latch
+
+loop.latch:
+  %r = phi i64 [ 0, %loop.header ], [ 1, %then ]
+  %iv.next = add i32 %iv, 1
+  %icmp = icmp sgt i32 %iv, 1000
+  br i1 %icmp, label %exit, label %loop.header
+
+exit:
+  %r.lcssa = phi i64 [ %r, %loop.latch ]
+  ret i64 %r.lcssa
+}
+
+define i64 @invar_cond(i1 %c) {
+; CHECK-LABEL: define i64 @invar_cond(
+; CHECK-SAME: i1 [[C:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[C]], i64 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[PREDPHI:%.*]] = select <4 x i1> [[BROADCAST_SPLAT]], <4 x i64> zeroinitializer, <4 x i64> <i64 1, i64 1, i64 1, i64 1>
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp eq i32 [[INDEX_NEXT]], 1000
+; CHECK-NEXT:    br i1 [[TMP0]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK:       middle.block:
+; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <4 x i64> [[PREDPHI]], i32 3
+; CHECK-NEXT:    br i1 false, label [[EXIT:%.*]], label [[SCALAR_PH]]
+; CHECK:       scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
+; CHECK:       loop.header:
+; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
+; CHECK-NEXT:    br i1 [[C]], label [[LOOP_LATCH]], label [[THEN:%.*]]
+; CHECK:       then:
+; CHECK-NEXT:    br label [[LOOP_LATCH]]
+; CHECK:       loop.latch:
+; CHECK-NEXT:    [[R:%.*]] = phi i64 [ 1, [[THEN]] ], [ 0, [[LOOP_HEADER]] ]
+; CHECK-NEXT:    [[IV_NEXT]] = add i32 [[IV]], 1
+; CHECK-NEXT:    [[ICMP:%.*]] = icmp sgt i32 [[IV]], 1000
+; CHECK-NEXT:    br i1 [[ICMP]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP7:![0-9]+]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[R_LCSSA:%.*]] = phi i64 [ [[R]], [[LOOP_LATCH]] ], [ [[TMP1]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    ret i64 [[R_LCSSA]]
+;
+entry:
+  br label %loop.header
+
+loop.header:
+  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  br i1 %c, label %loop.latch, label %then
+
+then:
+  br label %loop.latch
+
+loop.latch:
+  %r = phi i64 [ 1, %then ], [ 0, %loop.header ]
+  %iv.next = add i32 %iv, 1
+  %icmp = icmp sgt i32 %iv, 1000
+  br i1 %icmp, label %exit, label %loop.header
+
+exit:
+  %r.lcssa = phi i64 [ %r, %loop.latch ]
+  ret i64 %r.lcssa
+}
+
+define i64 @invar_cond_incoming_ops_reordered(i1 %c) {
+; CHECK-LABEL: define i64 @invar_cond_incoming_ops_reordered(
+; CHECK-SAME: i1 [[C:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[C]], i64 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP1:%.*]] = xor <4 x i1> [[BROADCAST_SPLAT]], <i1 true, i1 true, i1 true, i1 true>
+; CHECK-NEXT:    [[PREDPHI:%.*]] = select <4 x i1> [[TMP1]], <4 x i64> <i64 1, i64 1, i64 1, i64 1>, <4 x i64> zeroinitializer
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP0:%.*]] = icmp eq i32 [[INDEX_NEXT]], 1000
+; CHECK-NEXT:    br i1 [[TMP0]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; CHECK:       middle.block:
+; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <4 x i64> [[PREDPHI]], i32 3
+; CHECK-NEXT:    br i1 false, label [[EXIT:%.*]], label [[SCALAR_PH]]
+; CHECK:       scalar.ph:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
+; CHECK:       loop.header:
+; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
+; CHECK-NEXT:    br i1 [[C]], label [[LOOP_LATCH]], label [[THEN:%.*]]
+; CHECK:       then:
+; CHECK-NEXT:    br label [[LOOP_LATCH]]
+; CHECK:       loop.latch:
+; CHECK-NEXT:    [[R:%.*]] = phi i64 [ 0, [[LOOP_HEADER]] ], [ 1, [[THEN]] ]
+; CHECK-NEXT:    [[IV_NEXT]] = add i32 [[IV]], 1
+; CHECK-NEXT:    [[ICMP:%.*]] = icmp sgt i32 [[IV]], 1000
+; CHECK-NEXT:    br i1 [[ICMP]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP9:![0-9]+]]
+; CHECK:       exit:
+; CHECK-NEXT:    [[R_LCSSA:%.*]] = phi i64 [ [[R]], [[LOOP_LATCH]] ], [ [[TMP2]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    ret i64 [[R_LCSSA]]
+;
+entry:
+  br label %loop.header
+
+loop.header:
+  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  br i1 %c, label %loop.latch, label %then
+
+then:
+  br label %loop.latch
+
+loop.latch:
+  %r = phi i64 [ 0, %loop.header ], [ 1, %then ]
+  %iv.next = add i32 %iv, 1
+  %icmp = icmp sgt i32 %iv, 1000
+  br i1 %icmp, label %exit, label %loop.header
+
+exit:
+  %r.lcssa = phi i64 [ %r, %loop.latch ]
+  ret i64 %r.lcssa
+}
+;.
+; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
+; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
+; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
+; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META2]], [[META1]]}
+; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META1]], [[META2]]}
+; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META1]]}
+; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META1]], [[META2]]}
+; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META2]], [[META1]]}
+; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META1]], [[META2]]}
+; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META2]], [[META1]]}
+;.
diff --git a/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll b/llvm/test/Transforms/LoopVectorize/epilog-vectorization-any-of-reductions.ll
diff --git a/llvm/test/Transforms/LoopVectorize/float-induction.ll b/llvm/test/Transforms/LoopVectorize/float-induction.ll
diff --git a/llvm/test/Transforms/LoopVectorize/select-cmp-predicated.ll b/llvm/test/Transforms/LoopVectorize/select-cmp-predicated.ll
diff --git a/llvm/test/Transforms/LoopVectorize/select-cmp.ll b/llvm/test/Transforms/LoopVectorize/select-cmp.ll
diff --git a/llvm/test/Transforms/LoopVectorize/select-reduction-start-value-may-be-undef-or-poison.ll b/llvm/test/Transforms/LoopVectorize/select-reduction-start-value-may-be-undef-or-poison.ll
@@ -8,25 +8,26 @@ define i64 @pr62565_incoming_value_known_undef(i64 %a, ptr %src) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[A]], i64 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i64> [ undef, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i32 1, [[INDEX]]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[OFFSET_IDX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i32 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq <2 x i32> [[WIDE_LOAD]], <i32 1, i32 1>
-; CHECK-NEXT:    [[NOT:%*]] = xor <2 x i1> [[TMP3]], <i1 true, i1 true>
-; CHECK-NEXT:    [[TMP4]] = or <2 x i1> [[VEC_PHI]], [[NOT]]
+; CHECK-NEXT:    [[TMP4]] = select <2 x i1> [[TMP3]], <2 x i64> [[VEC_PHI]], <2 x i64> [[BROADCAST_SPLAT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], 32
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[TMP4]])
-; CHECK-NEXT:    [[FR_TMP6:%.*]] = freeze i1 [[TMP6]]
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR_TMP6]], i64 [[A]], i64 undef
+; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <2 x i64> [[TMP4]], undef
+; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[RDX_SELECT_CMP]])
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP6]], i64 [[A]], i64 undef
 ; CHECK-NEXT:    br i1 true, label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 33, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
@@ -71,25 +72,26 @@ define i64 @pr62565_incoming_value_known_poison(i64 %a, ptr %src) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[A]], i64 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i64> [ poison, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i32 1, [[INDEX]]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[OFFSET_IDX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i32 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq <2 x i32> [[WIDE_LOAD]], <i32 1, i32 1>
-; CHECK-NEXT:    [[NOT:%.*]] = xor <2 x i1> [[TMP3]], <i1 true, i1 true>
-; CHECK-NEXT:    [[TMP4]] = or <2 x i1> [[VEC_PHI]], [[NOT]]
+; CHECK-NEXT:    [[TMP4]] = select <2 x i1> [[TMP3]], <2 x i64> [[VEC_PHI]], <2 x i64> [[BROADCAST_SPLAT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], 32
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[TMP4]])
-; CHECK-NEXT:    [[FR_TMP6:%.*]] = freeze i1 [[TMP6]]
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR_TMP6]], i64 [[A]], i64 poison
+; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <2 x i64> [[TMP4]], poison
+; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[RDX_SELECT_CMP]])
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP6]], i64 [[A]], i64 poison
 ; CHECK-NEXT:    br i1 true, label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 33, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
@@ -134,25 +136,30 @@ define i64 @pr62565_incoming_value_may_be_poison(i64 %a, ptr %src, i64 %start) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
+; CHECK-NEXT:    [[MINMAX_IDENT_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[START]], i64 0
+; CHECK-NEXT:    [[MINMAX_IDENT_SPLAT:%.*]] = shufflevector <2 x i64> [[MINMAX_IDENT_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[A]], i64 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i1> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <2 x i64> [ [[MINMAX_IDENT_SPLAT]], [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i32 1, [[INDEX]]
 ; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[OFFSET_IDX]], 0
 ; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i32 [[TMP0]]
 ; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
 ; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
 ; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq <2 x i32> [[WIDE_LOAD]], <i32 1, i32 1>
-; CHECK-NEXT:    [[NOT:%.*]] = xor <2 x i1> [[TMP3]], <i1 true, i1 true>
-; CHECK-NEXT:    [[TMP4]] = or <2 x i1> [[VEC_PHI]], [[NOT]]
+; CHECK-NEXT:    [[TMP4]] = select <2 x i1> [[TMP3]], <2 x i64> [[VEC_PHI]], <2 x i64> [[BROADCAST_SPLAT]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
 ; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], 32
 ; CHECK-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
 ; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[TMP4]])
-; CHECK-NEXT:    [[FR_TMP6:%.*]] = freeze i1 [[TMP6]]
-; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[FR_TMP6]], i64 [[A]], i64 [[START]]
+; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[START]], i64 0
+; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <2 x i64> [[DOTSPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[RDX_SELECT_CMP:%.*]] = icmp ne <2 x i64> [[TMP4]], [[DOTSPLAT]]
+; CHECK-NEXT:    [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> [[RDX_SELECT_CMP]])
+; CHECK-NEXT:    [[RDX_SELECT:%.*]] = select i1 [[TMP6]], i64 [[A]], i64 [[START]]
 ; CHECK-NEXT:    br i1 true, label [[EXIT:%.*]], label [[SCALAR_PH]]
 ; CHECK:       scalar.ph:
 ; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i32 [ 33, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]

diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/phi-labels.ll b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/phi-labels.ll
@@ -0,0 +1,39 @@
+; RUN: opt < %s -S | FileCheck %s
+
+define i32 @phi_after_label(i1 %cc) {
+entry:
+  br i1 %cc, label %then, label %end
+
+then:
+  br label %end
+
+end:
+  %r = phi i32 [ 0, %entry ], [ 1, %then ]
+  ret i32 %r
+}
+
+define void @phi_before_label(i32 %bound) {
+entry:
+  br label %loop
+
+loop:
+  %ctr = phi i32 [ 0, %entry ], [ %ctr.next, %loop ]
+  %ctr.next = add i32 %ctr, 1
+  %cc = icmp ult i32 %ctr.next, %bound
+  br i1 %cc, label %loop, label %end
+
+end:
+  ret void
+}
+
+define i32 @phi_after_label_unnamed(i1 %cc) {
+0:
+  br i1 %cc, label %1, label %2
+
+1:
+  br label %2
+
+2:
+  %r = phi i32 [ 0, %0 ], [ 1, %1 ]
+  ret i32 %r
+}
diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/phi-labels.ll.expected b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/phi-labels.ll.expected
@@ -0,0 +1,71 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
+; RUN: opt < %s -S | FileCheck %s
+
+define i32 @phi_after_label(i1 %cc) {
+; CHECK-LABEL: define i32 @phi_after_label(
+; CHECK-SAME: i1 [[CC:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br i1 [[CC]], label [[THEN:%.*]], label [[END:%.*]]
+; CHECK:       then:
+; CHECK-NEXT:    br label [[END]]
+; CHECK:       end:
+; CHECK-NEXT:    [[R:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ 1, [[THEN]] ]
+; CHECK-NEXT:    ret i32 [[R]]
+;
+entry:
+  br i1 %cc, label %then, label %end
+
+then:
+  br label %end
+
+end:
+  %r = phi i32 [ 0, %entry ], [ 1, %then ]
+  ret i32 %r
+}
+
+define void @phi_before_label(i32 %bound) {
+; CHECK-LABEL: define void @phi_before_label(
+; CHECK-SAME: i32 [[BOUND:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    br label [[LOOP:%.*]]
+; CHECK:       loop:
+; CHECK-NEXT:    [[CTR:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[CTR_NEXT:%.*]], [[LOOP]] ]
+; CHECK-NEXT:    [[CTR_NEXT]] = add i32 [[CTR]], 1
+; CHECK-NEXT:    [[CC:%.*]] = icmp ult i32 [[CTR_NEXT]], [[BOUND]]
+; CHECK-NEXT:    br i1 [[CC]], label [[LOOP]], label [[END:%.*]]
+; CHECK:       end:
+; CHECK-NEXT:    ret void
+;
+entry:
+  br label %loop
+
+loop:
+  %ctr = phi i32 [ 0, %entry ], [ %ctr.next, %loop ]
+  %ctr.next = add i32 %ctr, 1
+  %cc = icmp ult i32 %ctr.next, %bound
+  br i1 %cc, label %loop, label %end
+
+end:
+  ret void
+}
+
+define i32 @phi_after_label_unnamed(i1 %cc) {
+; CHECK-LABEL: define i32 @phi_after_label_unnamed(
+; CHECK-SAME: i1 [[CC:%.*]]) {
+; CHECK-NEXT:    br i1 [[CC]], label [[TMP1:%.*]], label [[TMP2:%.*]]
+; CHECK:       1:
+; CHECK-NEXT:    br label [[TMP2]]
+; CHECK:       2:
+; CHECK-NEXT:    [[R:%.*]] = phi i32 [ 0, [[TMP0:%.*]] ], [ 1, [[TMP1]] ]
+; CHECK-NEXT:    ret i32 [[R]]
+;
+0:
+  br i1 %cc, label %1, label %2
+
+1:
+  br label %2
+
+2:
+  %r = phi i32 [ 0, %0 ], [ 1, %1 ]
+  ret i32 %r
+}
diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/stable_ir_values_funcs.ll b/llvm/test/tools/UpdateTestChecks/update_test_checks/Inputs/stable_ir_values_funcs.ll
@@ -0,0 +1,23 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --include-generated-funcs --version 3
+; RUN: opt < %s -S | FileCheck %s
+
+; The assumption underlying this test is that there are pre-existing check lines
+; but something has changed, and we would like to avoid needless changes of
+; meta variable names so that diffs end up being easier to read, e.g. avoid
+; changing X_I33 into X_I34 or renumbering the various TMP variables.
+
+define i32 @func({i32, i32} %x, i32 %y) {
+  %x.i34 = extractvalue {i32, i32} %x, 0
+  %1 = add i32 %y, 1
+  %2 = add i32 %x.i34, %1
+  %3 = mul i32 %2, 3
+  ret i32 %3
+}
+
+; CHECK-LABEL: define i32 @func(
+; CHECK-SAME: { i32, i32 } [[X:%.*]], i32 [[Y:%.*]]) {
+; CHECK-NEXT:    [[X_I33:%.*]] = extractvalue { i32, i32 } [[X]], 0
+; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[X_I33]], [[Y]]
+; CHECK-NEXT:    [[TMP2:%.*]] = mul i32 [[TMP1]], 3
+; CHECK-NEXT:    ret i32 [[TMP2]]
+;
diff --git a/.../test/tools/UpdateTestChecks/update_test_checks/Inputs/stable_ir_values_funcs.ll.expected b/.../test/tools/UpdateTestChecks/update_test_checks/Inputs/stable_ir_values_funcs.ll.expected
@@ -0,0 +1,24 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --include-generated-funcs --version 3
+; RUN: opt < %s -S | FileCheck %s
+
+; The assumption underlying this test is that there are pre-existing check lines
+; but something has changed, and we would like to avoid needless changes of
+; meta variable names so that diffs end up being easier to read, e.g. avoid
+; changing X_I33 into X_I34 or renumbering the various TMP variables.
+
+define i32 @func({i32, i32} %x, i32 %y) {
+  %x.i34 = extractvalue {i32, i32} %x, 0
+  %1 = add i32 %y, 1
+  %2 = add i32 %x.i34, %1
+  %3 = mul i32 %2, 3
+  ret i32 %3
+}
+
+; CHECK-LABEL: define i32 @func(
+; CHECK-SAME: { i32, i32 } [[X:%.*]], i32 [[Y:%.*]]) {
+; CHECK-NEXT:    [[X_I34:%.*]] = extractvalue { i32, i32 } [[X]], 0
+; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[Y]], 1
+; CHECK-NEXT:    [[TMP2:%.*]] = add i32 [[X_I34]], [[TMP1]]
+; CHECK-NEXT:    [[TMP3:%.*]] = mul i32 [[TMP2]], 3
+; CHECK-NEXT:    ret i32 [[TMP3]]
+;
diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/phi-labels.test b/llvm/test/tools/UpdateTestChecks/update_test_checks/phi-labels.test
@@ -0,0 +1,5 @@
+# RUN: cp -f %S/Inputs/phi-labels.ll %t.ll && %update_test_checks --version 4 %t.ll
+# RUN: diff -u %t.ll %S/Inputs/phi-labels.ll.expected
+## Check that running the script again does not change the result:
+# RUN: %update_test_checks %t.ll
+# RUN: diff -u %t.ll %S/Inputs/phi-labels.ll.expected
diff --git a/llvm/test/tools/UpdateTestChecks/update_test_checks/stable_ir_values_funcs.test b/llvm/test/tools/UpdateTestChecks/update_test_checks/stable_ir_values_funcs.test
@@ -0,0 +1,2 @@
+# RUN: cp -f %S/Inputs/stable_ir_values_funcs.ll %t.ll && %update_test_checks %t.ll
+# RUN: diff -u %t.ll %S/Inputs/stable_ir_values_funcs.ll.expected
diff --git a/llvm/utils/UpdateTestChecks/common.py b/llvm/utils/UpdateTestChecks/common.py
@@ -980,10 +980,6 @@ def __init__(
     def is_local_def_ir_value(self):
         return self.ir_prefix == "%"
 
-    # Return true if this kind of IR value is "global", basically if it matches '#{{.*}}'.
-    def is_global_scope_ir_value_match(self, match):
-        return self.global_ir_rhs_regexp is not None
-
     # Return the IR prefix and check prefix we use for this kind or IR value,
     # e.g., (%, TMP) for locals. If the IR prefix is a regex, return the prefix
     # used in the IR output
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		# RUN: cp -f %S/Inputs/stable_ir_values_funcs.ll %t.ll && %update_test_checks %t.ll
		# RUN: diff -u %t.ll %S/Inputs/stable_ir_values_funcs.ll.expected