vendor/llvm/llvm-trunk-r321017

1 files changed, 47 insertions, 61 deletions

diff --git a/lib/Target/X86/X86Subtarget.cpp b/lib/Target/X86/X86Subtarget.cpp
index 24845beac22d1..8b08766b6171b 100644
--- a/lib/Target/X86/X86Subtarget.cpp
+++ b/lib/Target/X86/X86Subtarget.cpp
@@ -13,21 +13,15 @@
 
 #include "X86.h"
 
-#ifdef LLVM_BUILD_GLOBAL_ISEL
 #include "X86CallLowering.h"
 #include "X86LegalizerInfo.h"
 #include "X86RegisterBankInfo.h"
-#endif
 #include "X86Subtarget.h"
 #include "MCTargetDesc/X86BaseInfo.h"
 #include "X86TargetMachine.h"
 #include "llvm/ADT/Triple.h"
-#ifdef LLVM_BUILD_GLOBAL_ISEL
 #include "llvm/CodeGen/GlobalISel/CallLowering.h"
 #include "llvm/CodeGen/GlobalISel/InstructionSelect.h"
-#include "llvm/CodeGen/GlobalISel/Legalizer.h"
-#include "llvm/CodeGen/GlobalISel/RegBankSelect.h"
-#endif
 #include "llvm/IR/Attributes.h"
 #include "llvm/IR/ConstantRange.h"
 #include "llvm/IR/Function.h"
@@ -39,8 +33,6 @@
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetMachine.h"
-#include <cassert>
-#include <string>
 
 #if defined(_MSC_VER)
 #include <intrin.h>
@@ -151,7 +143,12 @@ X86Subtarget::classifyGlobalFunctionReference(const GlobalValue *GV,
   if (TM.shouldAssumeDSOLocal(M, GV))
     return X86II::MO_NO_FLAG;
 
-  assert(!isTargetCOFF());
+  if (isTargetCOFF()) {
+    assert(GV->hasDLLImportStorageClass() &&
+           "shouldAssumeDSOLocal gave inconsistent answer");
+    return X86II::MO_DLLIMPORT;
+  }
+
   const Function *F = dyn_cast_or_null<Function>(GV);
 
   if (isTargetELF()) {
@@ -160,6 +157,8 @@ X86Subtarget::classifyGlobalFunctionReference(const GlobalValue *GV,
       // In Regcall calling convention those registers are used for passing
       // parameters. Thus we need to prevent lazy binding in Regcall.
       return X86II::MO_GOTPCREL;
+    if (F && F->hasFnAttribute(Attribute::NonLazyBind) && is64Bit())
+      return X86II::MO_GOTPCREL;
     return X86II::MO_PLT;
   }
 
@@ -189,9 +188,12 @@ const char *X86Subtarget::getBZeroEntry() const {
 }
 
 bool X86Subtarget::hasSinCos() const {
-  return getTargetTriple().isMacOSX() &&
-    !getTargetTriple().isMacOSXVersionLT(10, 9) &&
-    is64Bit();
+  if (getTargetTriple().isMacOSX()) {
+    return !getTargetTriple().isMacOSXVersionLT(10, 9) && is64Bit();
+  } else if (getTargetTriple().isOSFuchsia()) {
+    return true;
+  }
+  return false;
 }
 
 /// Return true if the subtarget allows calls to immediate address.
@@ -263,6 +265,17 @@ void X86Subtarget::initSubtargetFeatures(StringRef CPU, StringRef FS) {
   else if (isTargetDarwin() || isTargetLinux() || isTargetSolaris() ||
            isTargetKFreeBSD() || In64BitMode)
     stackAlignment = 16;
+
+  // Some CPUs have more overhead for gather. The specified overhead is relative
+  // to the Load operation. "2" is the number provided by Intel architects. This
+  // parameter is used for cost estimation of Gather Op and comparison with
+  // other alternatives.
+  // TODO: Remove the explicit hasAVX512()?, That would mean we would only
+  // enable gather with a -march.
+  if (hasAVX512() || (hasAVX2() && hasFastGather()))
+    GatherOverhead = 2;
+  if (hasAVX512())
+    ScatterOverhead = 2;
 }
 
 void X86Subtarget::initializeEnvironment() {
@@ -274,12 +287,15 @@ void X86Subtarget::initializeEnvironment() {
   HasPOPCNT = false;
   HasSSE4A = false;
   HasAES = false;
+  HasVAES = false;
   HasFXSR = false;
   HasXSAVE = false;
   HasXSAVEOPT = false;
   HasXSAVEC = false;
   HasXSAVES = false;
   HasPCLMUL = false;
+  HasVPCLMULQDQ = false;
+  HasGFNI = false;
   HasFMA = false;
   HasFMA4 = false;
   HasXOP = false;
@@ -293,6 +309,7 @@ void X86Subtarget::initializeEnvironment() {
   HasBMI = false;
   HasBMI2 = false;
   HasVBMI = false;
+  HasVBMI2 = false;
   HasIFMA = false;
   HasRTM = false;
   HasERI = false;
@@ -304,6 +321,8 @@ void X86Subtarget::initializeEnvironment() {
   HasVLX = false;
   HasADX = false;
   HasPKU = false;
+  HasVNNI = false;
+  HasBITALG = false;
   HasSHA = false;
   HasPRFCHW = false;
   HasRDSEED = false;
@@ -311,10 +330,11 @@ void X86Subtarget::initializeEnvironment() {
   HasMWAITX = false;
   HasCLZERO = false;
   HasMPX = false;
+  HasSHSTK = false;
+  HasIBT = false;
   HasSGX = false;
   HasCLFLUSHOPT = false;
   HasCLWB = false;
-  IsBTMemSlow = false;
   IsPMULLDSlow = false;
   IsSHLDSlow = false;
   IsUAMem16Slow = false;
@@ -323,15 +343,17 @@ void X86Subtarget::initializeEnvironment() {
   HasCmpxchg16b = false;
   UseLeaForSP = false;
   HasFastPartialYMMorZMMWrite = false;
+  HasFastGather = false;
   HasFastScalarFSQRT = false;
   HasFastVectorFSQRT = false;
   HasFastLZCNT = false;
   HasFastSHLDRotate = false;
+  HasMacroFusion = false;
   HasERMSB = false;
   HasSlowDivide32 = false;
   HasSlowDivide64 = false;
   PadShortFunctions = false;
-  CallRegIndirect = false;
+  SlowTwoMemOps = false;
   LEAUsesAG = false;
   SlowLEA = false;
   Slow3OpsLEA = false;
@@ -340,6 +362,9 @@ void X86Subtarget::initializeEnvironment() {
   // FIXME: this is a known good value for Yonah. How about others?
   MaxInlineSizeThreshold = 128;
   UseSoftFloat = false;
+  X86ProcFamily = Others;
+  GatherOverhead = 1024;
+  ScatterOverhead = 1024;
 }
 
 X86Subtarget &X86Subtarget::initializeSubtargetDependencies(StringRef CPU,
@@ -349,35 +374,6 @@ X86Subtarget &X86Subtarget::initializeSubtargetDependencies(StringRef CPU,
   return *this;
 }
 
-#ifdef LLVM_BUILD_GLOBAL_ISEL
-namespace {
-
-struct X86GISelActualAccessor : public GISelAccessor {
-  std::unique_ptr<CallLowering> CallLoweringInfo;
-  std::unique_ptr<LegalizerInfo> Legalizer;
-  std::unique_ptr<RegisterBankInfo> RegBankInfo;
-  std::unique_ptr<InstructionSelector> InstSelector;
-
-  const CallLowering *getCallLowering() const override {
-    return CallLoweringInfo.get();
-  }
-
-  const InstructionSelector *getInstructionSelector() const override {
-    return InstSelector.get();
-  }
-
-  const LegalizerInfo *getLegalizerInfo() const override {
-    return Legalizer.get();
-  }
-
-  const RegisterBankInfo *getRegBankInfo() const override {
-    return RegBankInfo.get();
-  }
-};
-
-} // end anonymous namespace
-#endif
-
 X86Subtarget::X86Subtarget(const Triple &TT, StringRef CPU, StringRef FS,
                            const X86TargetMachine &TM,
                            unsigned StackAlignOverride)
@@ -402,39 +398,29 @@ X86Subtarget::X86Subtarget(const Triple &TT, StringRef CPU, StringRef FS,
     setPICStyle(PICStyles::StubPIC);
   else if (isTargetELF())
     setPICStyle(PICStyles::GOT);
-#ifndef LLVM_BUILD_GLOBAL_ISEL
-  GISelAccessor *GISel = new GISelAccessor();
-#else
-  X86GISelActualAccessor *GISel = new X86GISelActualAccessor();
 
-  GISel->CallLoweringInfo.reset(new X86CallLowering(*getTargetLowering()));
-  GISel->Legalizer.reset(new X86LegalizerInfo(*this, TM));
+  CallLoweringInfo.reset(new X86CallLowering(*getTargetLowering()));
+  Legalizer.reset(new X86LegalizerInfo(*this, TM));
 
   auto *RBI = new X86RegisterBankInfo(*getRegisterInfo());
-  GISel->RegBankInfo.reset(RBI);
-  GISel->InstSelector.reset(createX86InstructionSelector(TM, *this, *RBI));
-#endif
-  setGISelAccessor(*GISel);
+  RegBankInfo.reset(RBI);
+  InstSelector.reset(createX86InstructionSelector(TM, *this, *RBI));
 }
 
 const CallLowering *X86Subtarget::getCallLowering() const {
-  assert(GISel && "Access to GlobalISel APIs not set");
-  return GISel->getCallLowering();
+  return CallLoweringInfo.get();
 }
 
 const InstructionSelector *X86Subtarget::getInstructionSelector() const {
-  assert(GISel && "Access to GlobalISel APIs not set");
-  return GISel->getInstructionSelector();
+  return InstSelector.get();
 }
 
 const LegalizerInfo *X86Subtarget::getLegalizerInfo() const {
-  assert(GISel && "Access to GlobalISel APIs not set");
-  return GISel->getLegalizerInfo();
+  return Legalizer.get();
 }
 
 const RegisterBankInfo *X86Subtarget::getRegBankInfo() const {
-  assert(GISel && "Access to GlobalISel APIs not set");
-  return GISel->getRegBankInfo();
+  return RegBankInfo.get();
 }
 
 bool X86Subtarget::enableEarlyIfConversion() const {