vendor/llvm/llvm-release_700-r342383vendor/llvm/llvm-release_70-r341916

18 files changed, 152 insertions, 57 deletions

diff --git a/lib/MC/MCParser/AsmParser.cpp b/lib/MC/MCParser/AsmParser.cpp
index 501a1cccf60e..d88c6f76826f 100644
--- a/lib/MC/MCParser/AsmParser.cpp
+++ b/lib/MC/MCParser/AsmParser.cpp
@@ -3348,17 +3348,17 @@ bool AsmParser::parseDirectiveFile(SMLoc DirectiveLoc) {
     }
   }
 
-  // In case there is a -g option as well as debug info from directive .file,
-  // we turn off the -g option, directly use the existing debug info instead.
-  // Also reset any implicit ".file 0" for the assembler source.
-  if (Ctx.getGenDwarfForAssembly()) {
-    Ctx.getMCDwarfLineTable(0).resetRootFile();
-    Ctx.setGenDwarfForAssembly(false);
-  }
-
   if (FileNumber == -1)
     getStreamer().EmitFileDirective(Filename);
   else {
+    // In case there is a -g option as well as debug info from directive .file,
+    // we turn off the -g option, directly use the existing debug info instead.
+    // Also reset any implicit ".file 0" for the assembler source.
+    if (Ctx.getGenDwarfForAssembly()) {
+      Ctx.getMCDwarfLineTable(0).resetRootFile();
+      Ctx.setGenDwarfForAssembly(false);
+    }
+
     MD5::MD5Result *CKMem = nullptr;
     if (HasMD5) {
       CKMem = (MD5::MD5Result *)Ctx.allocate(sizeof(MD5::MD5Result), 1);
diff --git a/lib/Support/Unix/Path.inc b/lib/Support/Unix/Path.inc
index 7ad57d892ff1..b4279d4fcc0c 100644
--- a/lib/Support/Unix/Path.inc
+++ b/lib/Support/Unix/Path.inc
@@ -769,8 +769,10 @@ std::error_code openFile(const Twine &Name, int &ResultFD,
 
   SmallString<128> Storage;
   StringRef P = Name.toNullTerminatedStringRef(Storage);
-  if ((ResultFD = sys::RetryAfterSignal(-1, ::open, P.begin(), OpenFlags, Mode)) <
-      0)
+  // Call ::open in a lambda to avoid overload resolution in RetryAfterSignal
+  // when open is overloaded, such as in Bionic.
+  auto Open = [&]() { return ::open(P.begin(), OpenFlags, Mode); };
+  if ((ResultFD = sys::RetryAfterSignal(-1, Open)) < 0)
     return std::error_code(errno, std::generic_category());
 #ifndef O_CLOEXEC
   if (!(Flags & OF_ChildInherit)) {
diff --git a/lib/Support/Unix/Process.inc b/lib/Support/Unix/Process.inc
index fa515d44f3f2..3185f45a3a61 100644
--- a/lib/Support/Unix/Process.inc
+++ b/lib/Support/Unix/Process.inc
@@ -211,7 +211,10 @@ std::error_code Process::FixupStandardFileDescriptors() {
     assert(errno == EBADF && "expected errno to have EBADF at this point!");
 
     if (NullFD < 0) {
-      if ((NullFD = RetryAfterSignal(-1, ::open, "/dev/null", O_RDWR)) < 0)
+      // Call ::open in a lambda to avoid overload resolution in
+      // RetryAfterSignal when open is overloaded, such as in Bionic.
+      auto Open = [&]() { return ::open("/dev/null", O_RDWR); };
+      if ((NullFD = RetryAfterSignal(-1, Open)) < 0)
         return std::error_code(errno, std::generic_category());
     }
 
diff --git a/lib/Target/AMDGPU/AMDGPU.h b/lib/Target/AMDGPU/AMDGPU.h
index 796766d94622..2b49c2ea88e1 100644
--- a/lib/Target/AMDGPU/AMDGPU.h
+++ b/lib/Target/AMDGPU/AMDGPU.h
@@ -229,7 +229,7 @@ struct AMDGPUAS {
 
   enum : unsigned {
     // The maximum value for flat, generic, local, private, constant and region.
-    MAX_COMMON_ADDRESS = 5,
+    MAX_AMDGPU_ADDRESS = 6,
 
     GLOBAL_ADDRESS = 1,   ///< Address space for global memory (RAT0, VTX0).
     CONSTANT_ADDRESS = 4, ///< Address space for constant memory (VTX2)
diff --git a/lib/Target/AMDGPU/AMDGPUAliasAnalysis.cpp b/lib/Target/AMDGPU/AMDGPUAliasAnalysis.cpp
index ef4b69d09d9f..974fbcb87191 100644
--- a/lib/Target/AMDGPU/AMDGPUAliasAnalysis.cpp
+++ b/lib/Target/AMDGPU/AMDGPUAliasAnalysis.cpp
@@ -50,47 +50,51 @@ void AMDGPUAAWrapperPass::getAnalysisUsage(AnalysisUsage &AU) const {
 AMDGPUAAResult::ASAliasRulesTy::ASAliasRulesTy(AMDGPUAS AS_, Triple::ArchType Arch_)
   : Arch(Arch_), AS(AS_) {
   // These arrarys are indexed by address space value
-  // enum elements 0 ... to 5
-  static const AliasResult ASAliasRulesPrivIsZero[6][6] = {
-  /*             Private    Global    Constant  Group     Flat      Region*/
-  /* Private  */ {MayAlias, NoAlias , NoAlias , NoAlias , MayAlias, NoAlias},
-  /* Global   */ {NoAlias , MayAlias, NoAlias , NoAlias , MayAlias, NoAlias},
-  /* Constant */ {NoAlias , NoAlias , MayAlias, NoAlias , MayAlias, NoAlias},
-  /* Group    */ {NoAlias , NoAlias , NoAlias , MayAlias, MayAlias, NoAlias},
-  /* Flat     */ {MayAlias, MayAlias, MayAlias, MayAlias, MayAlias, MayAlias},
-  /* Region   */ {NoAlias , NoAlias , NoAlias , NoAlias , MayAlias, MayAlias}
+  // enum elements 0 ... to 6
+  static const AliasResult ASAliasRulesPrivIsZero[7][7] = {
+  /*                    Private    Global    Constant  Group     Flat      Region    Constant 32-bit */
+  /* Private  */        {MayAlias, NoAlias , NoAlias , NoAlias , MayAlias, NoAlias , NoAlias},
+  /* Global   */        {NoAlias , MayAlias, MayAlias, NoAlias , MayAlias, NoAlias , MayAlias},
+  /* Constant */        {NoAlias , MayAlias, MayAlias, NoAlias , MayAlias, NoAlias , MayAlias},
+  /* Group    */        {NoAlias , NoAlias , NoAlias , MayAlias, MayAlias, NoAlias , NoAlias},
+  /* Flat     */        {MayAlias, MayAlias, MayAlias, MayAlias, MayAlias, MayAlias, MayAlias},
+  /* Region   */        {NoAlias , NoAlias , NoAlias , NoAlias , MayAlias, MayAlias, NoAlias},
+  /* Constant 32-bit */ {NoAlias , MayAlias, MayAlias, NoAlias , MayAlias, NoAlias , MayAlias}
   };
-  static const AliasResult ASAliasRulesGenIsZero[6][6] = {
-  /*             Flat       Global    Region    Group     Constant  Private */
-  /* Flat     */ {MayAlias, MayAlias, MayAlias, MayAlias, MayAlias, MayAlias},
-  /* Global   */ {MayAlias, MayAlias, NoAlias , NoAlias , NoAlias , NoAlias},
-  /* Constant */ {MayAlias, NoAlias , MayAlias, NoAlias , NoAlias,  NoAlias},
-  /* Group    */ {MayAlias, NoAlias , NoAlias , MayAlias, NoAlias , NoAlias},
-  /* Region   */ {MayAlias, NoAlias , NoAlias , NoAlias,  MayAlias, NoAlias},
-  /* Private  */ {MayAlias, NoAlias , NoAlias , NoAlias , NoAlias , MayAlias}
+  static const AliasResult ASAliasRulesGenIsZero[7][7] = {
+  /*                    Flat       Global    Region    Group     Constant  Private   Constant 32-bit */
+  /* Flat     */        {MayAlias, MayAlias, MayAlias, MayAlias, MayAlias, MayAlias, MayAlias},
+  /* Global   */        {MayAlias, MayAlias, NoAlias , NoAlias , MayAlias, NoAlias , MayAlias},
+  /* Region   */        {MayAlias, NoAlias , NoAlias , NoAlias,  MayAlias, NoAlias , MayAlias},
+  /* Group    */        {MayAlias, NoAlias , NoAlias , MayAlias, NoAlias , NoAlias , NoAlias},
+  /* Constant */        {MayAlias, MayAlias, MayAlias, NoAlias , NoAlias,  NoAlias , MayAlias},
+  /* Private  */        {MayAlias, NoAlias , NoAlias , NoAlias , NoAlias , MayAlias, NoAlias},
+  /* Constant 32-bit */ {MayAlias, MayAlias, MayAlias, NoAlias , MayAlias, NoAlias , NoAlias}
   };
-  assert(AS.MAX_COMMON_ADDRESS <= 5);
+  static_assert(AMDGPUAS::MAX_AMDGPU_ADDRESS <= 6, "Addr space out of range");
   if (AS.FLAT_ADDRESS == 0) {
-    assert(AS.GLOBAL_ADDRESS   == 1 &&
-           AS.REGION_ADDRESS   == 2 &&
-           AS.LOCAL_ADDRESS    == 3 &&
-           AS.CONSTANT_ADDRESS == 4 &&
-           AS.PRIVATE_ADDRESS  == 5);
+    assert(AS.GLOBAL_ADDRESS         == 1 &&
+           AS.REGION_ADDRESS         == 2 &&
+           AS.LOCAL_ADDRESS          == 3 &&
+           AS.CONSTANT_ADDRESS       == 4 &&
+           AS.PRIVATE_ADDRESS        == 5 &&
+           AS.CONSTANT_ADDRESS_32BIT == 6);
     ASAliasRules = &ASAliasRulesGenIsZero;
   } else {
-    assert(AS.PRIVATE_ADDRESS  == 0 &&
-           AS.GLOBAL_ADDRESS   == 1 &&
-           AS.CONSTANT_ADDRESS == 2 &&
-           AS.LOCAL_ADDRESS    == 3 &&
-           AS.FLAT_ADDRESS     == 4 &&
-           AS.REGION_ADDRESS   == 5);
+    assert(AS.PRIVATE_ADDRESS        == 0 &&
+           AS.GLOBAL_ADDRESS         == 1 &&
+           AS.CONSTANT_ADDRESS       == 2 &&
+           AS.LOCAL_ADDRESS          == 3 &&
+           AS.FLAT_ADDRESS           == 4 &&
+           AS.REGION_ADDRESS         == 5 &&
+           AS.CONSTANT_ADDRESS_32BIT == 6);
     ASAliasRules = &ASAliasRulesPrivIsZero;
   }
 }
 
 AliasResult AMDGPUAAResult::ASAliasRulesTy::getAliasResult(unsigned AS1,
     unsigned AS2) const {
-  if (AS1 > AS.MAX_COMMON_ADDRESS || AS2 > AS.MAX_COMMON_ADDRESS) {
+  if (AS1 > AS.MAX_AMDGPU_ADDRESS || AS2 > AS.MAX_AMDGPU_ADDRESS) {
     if (Arch == Triple::amdgcn)
       report_fatal_error("Pointer address space out of range");
     return AS1 == AS2 ? MayAlias : NoAlias;
diff --git a/lib/Target/AMDGPU/AMDGPUAliasAnalysis.h b/lib/Target/AMDGPU/AMDGPUAliasAnalysis.h
index 645a38af753c..09ad51d5e42f 100644
--- a/lib/Target/AMDGPU/AMDGPUAliasAnalysis.h
+++ b/lib/Target/AMDGPU/AMDGPUAliasAnalysis.h
@@ -63,7 +63,7 @@ private:
   private:
     Triple::ArchType Arch;
     AMDGPUAS AS;
-    const AliasResult (*ASAliasRules)[6][6];
+    const AliasResult (*ASAliasRules)[7][7];
   } ASAliasRules;
 };
 
diff --git a/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp b/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
index f25f4d4693ea..7cb0e12a6809 100644
--- a/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
+++ b/lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
@@ -1451,7 +1451,11 @@ bool AMDGPUDAGToDAGISel::SelectSMRD(SDValue Addr, SDValue &SBase,
                                      SDValue &Offset, bool &Imm) const {
   SDLoc SL(Addr);
 
-  if (CurDAG->isBaseWithConstantOffset(Addr)) {
+  // A 32-bit (address + offset) should not cause unsigned 32-bit integer
+  // wraparound, because s_load instructions perform the addition in 64 bits.
+  if ((Addr.getValueType() != MVT::i32 ||
+       Addr->getFlags().hasNoUnsignedWrap()) &&
+      CurDAG->isBaseWithConstantOffset(Addr)) {
     SDValue N0 = Addr.getOperand(0);
     SDValue N1 = Addr.getOperand(1);
 
diff --git a/lib/Target/ARM/ARMFrameLowering.cpp b/lib/Target/ARM/ARMFrameLowering.cpp
index a8c75702d7b5..56ad7a0f0446 100644
--- a/lib/Target/ARM/ARMFrameLowering.cpp
+++ b/lib/Target/ARM/ARMFrameLowering.cpp
@@ -1514,6 +1514,7 @@ static unsigned estimateRSStackSizeLimit(MachineFunction &MF,
           break;
         case ARMII::AddrMode5:
         case ARMII::AddrModeT2_i8s4:
+        case ARMII::AddrModeT2_ldrex:
           Limit = std::min(Limit, ((1U << 8) - 1) * 4);
           break;
         case ARMII::AddrModeT2_i12:
diff --git a/lib/Target/ARM/ARMInstrFormats.td b/lib/Target/ARM/ARMInstrFormats.td
index 70aded247f65..1d3b1414f090 100644
--- a/lib/Target/ARM/ARMInstrFormats.td
+++ b/lib/Target/ARM/ARMInstrFormats.td
@@ -109,6 +109,7 @@ def AddrModeT2_pc   : AddrMode<14>;
 def AddrModeT2_i8s4 : AddrMode<15>;
 def AddrMode_i12    : AddrMode<16>;
 def AddrMode5FP16   : AddrMode<17>;
+def AddrModeT2_ldrex : AddrMode<18>;
 
 // Load / store index mode.
 class IndexMode<bits<2> val> {
diff --git a/lib/Target/ARM/ARMInstrThumb2.td b/lib/Target/ARM/ARMInstrThumb2.td
index c7133b6483ef..f67075fbf9fd 100644
--- a/lib/Target/ARM/ARMInstrThumb2.td
+++ b/lib/Target/ARM/ARMInstrThumb2.td
@@ -3267,7 +3267,7 @@ def t2LDREXH : T2I_ldrex<0b0101, (outs rGPR:$Rt), (ins addr_offset_none:$addr),
                          [(set rGPR:$Rt, (ldrex_2 addr_offset_none:$addr))]>,
                Requires<[IsThumb, HasV8MBaseline]>;
 def t2LDREX  : Thumb2I<(outs rGPR:$Rt), (ins t2addrmode_imm0_1020s4:$addr),
-                       AddrModeNone, 4, NoItinerary,
+                       AddrModeT2_ldrex, 4, NoItinerary,
                        "ldrex", "\t$Rt, $addr", "",
                      [(set rGPR:$Rt, (ldrex_4 t2addrmode_imm0_1020s4:$addr))]>,
                Requires<[IsThumb, HasV8MBaseline]> {
@@ -3346,7 +3346,7 @@ def t2STREXH : T2I_strex<0b0101, (outs rGPR:$Rd),
 
 def t2STREX  : Thumb2I<(outs rGPR:$Rd), (ins rGPR:$Rt,
                              t2addrmode_imm0_1020s4:$addr),
-                  AddrModeNone, 4, NoItinerary,
+                  AddrModeT2_ldrex, 4, NoItinerary,
                   "strex", "\t$Rd, $Rt, $addr", "",
                   [(set rGPR:$Rd,
                         (strex_4 rGPR:$Rt, t2addrmode_imm0_1020s4:$addr))]>,
diff --git a/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h b/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h
index b918006fe9e3..beeb5dec4baf 100644
--- a/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h
+++ b/lib/Target/ARM/MCTargetDesc/ARMBaseInfo.h
@@ -201,7 +201,8 @@ namespace ARMII {
     AddrModeT2_pc   = 14, // +/- i12 for pc relative data
     AddrModeT2_i8s4 = 15, // i8 * 4
     AddrMode_i12    = 16,
-    AddrMode5FP16   = 17  // i8 * 2
+    AddrMode5FP16   = 17,  // i8 * 2
+    AddrModeT2_ldrex = 18, // i8 * 4, with unscaled offset in MCInst
   };
 
   inline static const char *AddrModeToString(AddrMode addrmode) {
@@ -224,6 +225,7 @@ namespace ARMII {
     case AddrModeT2_pc:   return "AddrModeT2_pc";
     case AddrModeT2_i8s4: return "AddrModeT2_i8s4";
     case AddrMode_i12:    return "AddrMode_i12";
+    case AddrModeT2_ldrex:return "AddrModeT2_ldrex";
     }
   }
 
diff --git a/lib/Target/ARM/Thumb2InstrInfo.cpp b/lib/Target/ARM/Thumb2InstrInfo.cpp
index d5f0ba9ee485..1a91a7030657 100644
--- a/lib/Target/ARM/Thumb2InstrInfo.cpp
+++ b/lib/Target/ARM/Thumb2InstrInfo.cpp
@@ -621,6 +621,11 @@ bool llvm::rewriteT2FrameIndex(MachineInstr &MI, unsigned FrameRegIdx,
       // MCInst operand expects already scaled value.
       Scale = 1;
       assert((Offset & 3) == 0 && "Can't encode this offset!");
+    } else if (AddrMode == ARMII::AddrModeT2_ldrex) {
+      Offset += MI.getOperand(FrameRegIdx + 1).getImm() * 4;
+      NumBits = 8; // 8 bits scaled by 4
+      Scale = 4;
+      assert((Offset & 3) == 0 && "Can't encode this offset!");
     } else {
       llvm_unreachable("Unsupported addressing mode!");
     }
diff --git a/lib/Target/BPF/MCTargetDesc/BPFAsmBackend.cpp b/lib/Target/BPF/MCTargetDesc/BPFAsmBackend.cpp
index 6c255e9ef780..1822d8688fa2 100644
--- a/lib/Target/BPF/MCTargetDesc/BPFAsmBackend.cpp
+++ b/lib/Target/BPF/MCTargetDesc/BPFAsmBackend.cpp
@@ -10,6 +10,8 @@
 #include "MCTargetDesc/BPFMCTargetDesc.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/MC/MCAsmBackend.h"
+#include "llvm/MC/MCAssembler.h"
+#include "llvm/MC/MCContext.h"
 #include "llvm/MC/MCFixup.h"
 #include "llvm/MC/MCObjectWriter.h"
 #include "llvm/Support/EndianStream.h"
@@ -71,7 +73,12 @@ void BPFAsmBackend::applyFixup(const MCAssembler &Asm, const MCFixup &Fixup,
                                bool IsResolved,
                                const MCSubtargetInfo *STI) const {
   if (Fixup.getKind() == FK_SecRel_4 || Fixup.getKind() == FK_SecRel_8) {
-    assert(Value == 0);
+    if (Value) {
+      MCContext &Ctx = Asm.getContext();
+      Ctx.reportError(Fixup.getLoc(),
+                      "Unsupported relocation: try to compile with -O2 or above, "
+                      "or check your static variable usage");
+    }
   } else if (Fixup.getKind() == FK_Data_4) {
     support::endian::write<uint32_t>(&Data[Fixup.getOffset()], Value, Endian);
   } else if (Fixup.getKind() == FK_Data_8) {
diff --git a/lib/Target/X86/AsmParser/X86AsmParser.cpp b/lib/Target/X86/AsmParser/X86AsmParser.cpp
index b02e4d80fbba..8b7b250e1a09 100644
--- a/lib/Target/X86/AsmParser/X86AsmParser.cpp
+++ b/lib/Target/X86/AsmParser/X86AsmParser.cpp
@@ -1054,7 +1054,7 @@ static bool CheckBaseRegAndIndexRegAndScale(unsigned BaseReg, unsigned IndexReg,
   // RIP/EIP-relative addressing is only supported in 64-bit mode.
   if (!Is64BitMode && BaseReg != 0 &&
       (BaseReg == X86::RIP || BaseReg == X86::EIP)) {
-    ErrMsg = "RIP-relative addressing requires 64-bit mode";
+    ErrMsg = "IP-relative addressing requires 64-bit mode";
     return true;
   }
 
@@ -1099,7 +1099,7 @@ bool X86AsmParser::ParseRegister(unsigned &RegNo,
     // checked.
     // FIXME: Check AH, CH, DH, BH cannot be used in an instruction requiring a
     // REX prefix.
-    if (RegNo == X86::RIZ || RegNo == X86::RIP || RegNo == X86::EIP ||
+    if (RegNo == X86::RIZ || RegNo == X86::RIP ||
         X86MCRegisterClasses[X86::GR64RegClassID].contains(RegNo) ||
         X86II::isX86_64NonExtLowByteReg(RegNo) ||
         X86II::isX86_64ExtendedReg(RegNo))
diff --git a/lib/Transforms/Scalar/LoopSink.cpp b/lib/Transforms/Scalar/LoopSink.cpp
index 760177c9c5e9..7d62349d4719 100644
--- a/lib/Transforms/Scalar/LoopSink.cpp
+++ b/lib/Transforms/Scalar/LoopSink.cpp
@@ -152,6 +152,14 @@ findBBsToSinkInto(const Loop &L, const SmallPtrSetImpl<BasicBlock *> &UseBBs,
     }
   }
 
+  // Can't sink into blocks that have no valid insertion point.
+  for (BasicBlock *BB : BBsToSinkInto) {
+    if (BB->getFirstInsertionPt() == BB->end()) {
+      BBsToSinkInto.clear();
+      break;
+    }
+  }
+
   // If the total frequency of BBsToSinkInto is larger than preheader frequency,
   // do not sink.
   if (adjustedSumFreq(BBsToSinkInto, BFI) >
diff --git a/lib/Transforms/Scalar/SROA.cpp b/lib/Transforms/Scalar/SROA.cpp
index de16b608f752..bf482bf5272e 100644
--- a/lib/Transforms/Scalar/SROA.cpp
+++ b/lib/Transforms/Scalar/SROA.cpp
@@ -3046,6 +3046,42 @@ private:
     return true;
   }
 
+  void fixLoadStoreAlign(Instruction &Root) {
+    // This algorithm implements the same visitor loop as
+    // hasUnsafePHIOrSelectUse, and fixes the alignment of each load
+    // or store found.
+    SmallPtrSet<Instruction *, 4> Visited;
+    SmallVector<Instruction *, 4> Uses;
+    Visited.insert(&Root);
+    Uses.push_back(&Root);
+    do {
+      Instruction *I = Uses.pop_back_val();
+
+      if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
+        unsigned LoadAlign = LI->getAlignment();
+        if (!LoadAlign)
+          LoadAlign = DL.getABITypeAlignment(LI->getType());
+        LI->setAlignment(std::min(LoadAlign, getSliceAlign()));
+        continue;
+      }
+      if (StoreInst *SI = dyn_cast<StoreInst>(I)) {
+        unsigned StoreAlign = SI->getAlignment();
+        if (!StoreAlign) {
+          Value *Op = SI->getOperand(0);
+          StoreAlign = DL.getABITypeAlignment(Op->getType());
+        }
+        SI->setAlignment(std::min(StoreAlign, getSliceAlign()));
+        continue;
+      }
+
+      assert(isa<BitCastInst>(I) || isa<PHINode>(I) ||
+             isa<SelectInst>(I) || isa<GetElementPtrInst>(I));
+      for (User *U : I->users())
+        if (Visited.insert(cast<Instruction>(U)).second)
+          Uses.push_back(cast<Instruction>(U));
+    } while (!Uses.empty());
+  }
+
   bool visitPHINode(PHINode &PN) {
     LLVM_DEBUG(dbgs() << "    original: " << PN << "\n");
     assert(BeginOffset >= NewAllocaBeginOffset && "PHIs are unsplittable");
@@ -3069,6 +3105,9 @@ private:
     LLVM_DEBUG(dbgs() << "          to: " << PN << "\n");
     deleteIfTriviallyDead(OldPtr);
 
+    // Fix the alignment of any loads or stores using this PHI node.
+    fixLoadStoreAlign(PN);
+
     // PHIs can't be promoted on their own, but often can be speculated. We
     // check the speculation outside of the rewriter so that we see the
     // fully-rewritten alloca.
@@ -3093,6 +3132,9 @@ private:
     LLVM_DEBUG(dbgs() << "          to: " << SI << "\n");
     deleteIfTriviallyDead(OldPtr);
 
+    // Fix the alignment of any loads or stores using this select.
+    fixLoadStoreAlign(SI);
+
     // Selects can't be promoted on their own, but often can be speculated. We
     // check the speculation outside of the rewriter so that we see the
     // fully-rewritten alloca.
diff --git a/lib/Transforms/Utils/CloneFunction.cpp b/lib/Transforms/Utils/CloneFunction.cpp
index 807360340055..9ae60962a631 100644
--- a/lib/Transforms/Utils/CloneFunction.cpp
+++ b/lib/Transforms/Utils/CloneFunction.cpp
@@ -636,6 +636,22 @@ void llvm::CloneAndPruneIntoFromInst(Function *NewFunc, const Function *OldFunc,
   Function::iterator Begin = cast<BasicBlock>(VMap[StartingBB])->getIterator();
   Function::iterator I = Begin;
   while (I != NewFunc->end()) {
+    // We need to simplify conditional branches and switches with a constant
+    // operand. We try to prune these out when cloning, but if the
+    // simplification required looking through PHI nodes, those are only
+    // available after forming the full basic block. That may leave some here,
+    // and we still want to prune the dead code as early as possible.
+    //
+    // Do the folding before we check if the block is dead since we want code
+    // like
+    //  bb:
+    //    br i1 undef, label %bb, label %bb
+    // to be simplified to
+    //  bb:
+    //    br label %bb
+    // before we call I->getSinglePredecessor().
+    ConstantFoldTerminator(&*I);
+
     // Check if this block has become dead during inlining or other
     // simplifications. Note that the first block will appear dead, as it has
     // not yet been wired up properly.
@@ -646,13 +662,6 @@ void llvm::CloneAndPruneIntoFromInst(Function *NewFunc, const Function *OldFunc,
       continue;
     }
 
-    // We need to simplify conditional branches and switches with a constant
-    // operand. We try to prune these out when cloning, but if the
-    // simplification required looking through PHI nodes, those are only
-    // available after forming the full basic block. That may leave some here,
-    // and we still want to prune the dead code as early as possible.
-    ConstantFoldTerminator(&*I);
-
     BranchInst *BI = dyn_cast<BranchInst>(I->getTerminator());
     if (!BI || BI->isConditional()) { ++I; continue; }
 
diff --git a/lib/Transforms/Vectorize/LoopVectorize.cpp b/lib/Transforms/Vectorize/LoopVectorize.cpp
index 859d0c92ca5a..1c7d0a63a5ca 100644
--- a/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -4510,6 +4510,13 @@ void LoopVectorizationCostModel::collectLoopUniforms(unsigned VF) {
     for (auto OV : I->operand_values()) {
       if (isOutOfScope(OV))
         continue;
+      // First order recurrence Phi's should typically be considered
+      // non-uniform.
+      auto *OP = dyn_cast<PHINode>(OV);
+      if (OP && Legal->isFirstOrderRecurrence(OP))
+        continue;
+      // If all the users of the operand are uniform, then add the
+      // operand into the uniform worklist.
       auto *OI = cast<Instruction>(OV);
       if (llvm::all_of(OI->users(), [&](User *U) -> bool {
             auto *J = cast<Instruction>(U);