[arm-fast-isel] Add support for ELF PIC.

This is a preliminary step towards ELF support; currently ARMFastISel hasn't been used for ELF object files yet. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164759 91177308-0d34-0410-b5e6-96231b3b80d8
hylsd · Sep 27, 2012 · 8f50647 · 8f50647
1 parent 5e92990
commit 8f50647
Show file tree

Hide file tree

Showing 6 changed files with 141 additions and 2 deletions.
diff --git a/lib/Target/ARM/ARM.h b/lib/Target/ARM/ARM.h
@@ -37,6 +37,7 @@ FunctionPass *createARMJITCodeEmitterPass(ARMBaseTargetMachine &TM,
 
 FunctionPass *createARMLoadStoreOptimizationPass(bool PreAlloc = false);
 FunctionPass *createARMExpandPseudoPass();
+FunctionPass *createARMGlobalBaseRegPass();
 FunctionPass *createARMGlobalMergePass(const TargetLowering* tli);
 FunctionPass *createARMConstantIslandPass();
 FunctionPass *createMLxExpansionPass();

diff --git a/lib/Target/ARM/ARMFastISel.cpp b/lib/Target/ARM/ARMFastISel.cpp
@@ -194,6 +194,7 @@ class ARMFastISel : public FastISel {
     unsigned ARMMoveToFPReg(EVT VT, unsigned SrcReg);
     unsigned ARMMoveToIntReg(EVT VT, unsigned SrcReg);
     unsigned ARMSelectCallOp(bool UseReg);
+    unsigned ARMLowerPICELF(const GlobalValue *GV, unsigned Align, EVT VT);
 
     // Call handling routines.
   private:
@@ -648,6 +649,9 @@ unsigned ARMFastISel::ARMMaterializeGV(const GlobalValue *GV, EVT VT) {
       Align = TD.getTypeAllocSize(GV->getType());
     }
 
+    if (Subtarget->isTargetELF() && RelocM == Reloc::PIC_)
+      return ARMLowerPICELF(GV, Align, VT);
+
     // Grab index.
     unsigned PCAdj = (RelocM != Reloc::PIC_) ? 0 :
       (Subtarget->isThumb() ? 4 : 8);
@@ -2796,6 +2800,47 @@ bool ARMFastISel::TryToFoldLoad(MachineInstr *MI, unsigned OpNo,
   return true;
 }
 
+unsigned ARMFastISel::ARMLowerPICELF(const GlobalValue *GV,
+                                     unsigned Align, EVT VT) {
+  bool UseGOTOFF = GV->hasLocalLinkage() || GV->hasHiddenVisibility();
+  ARMConstantPoolConstant *CPV =
+    ARMConstantPoolConstant::Create(GV, UseGOTOFF ? ARMCP::GOTOFF : ARMCP::GOT);
+  unsigned Idx = MCP.getConstantPoolIndex(CPV, Align);
+
+  unsigned Opc;
+  unsigned DestReg1 = createResultReg(TLI.getRegClassFor(VT));
+  // Load value.
+  if (isThumb2) {
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
+                            TII.get(ARM::t2LDRpci), DestReg1)
+                    .addConstantPoolIndex(Idx));
+    Opc = UseGOTOFF ? ARM::t2ADDrr : ARM::t2LDRs;
+  } else {
+    // The extra immediate is for addrmode2.
+    AddOptionalDefs(BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt,
+                            DL, TII.get(ARM::LDRcp), DestReg1)
+                    .addConstantPoolIndex(Idx).addImm(0));
+    Opc = UseGOTOFF ? ARM::ADDrr : ARM::LDRrs;
+  }
+
+  unsigned GlobalBaseReg = AFI->getGlobalBaseReg();
+  if (GlobalBaseReg == 0) {
+    GlobalBaseReg = MRI.createVirtualRegister(TLI.getRegClassFor(VT));
+    AFI->setGlobalBaseReg(GlobalBaseReg);
+  }
+
+  unsigned DestReg2 = createResultReg(TLI.getRegClassFor(VT));
+  MachineInstrBuilder MIB = BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt,
+                                    DL, TII.get(Opc), DestReg2)
+                            .addReg(DestReg1)
+                            .addReg(GlobalBaseReg);
+  if (!UseGOTOFF)
+    MIB.addImm(0);
+  AddOptionalDefs(MIB);
+
+  return DestReg2;
+}
+
 namespace llvm {
   FastISel *ARM::createFastISel(FunctionLoweringInfo &funcInfo,
                                 const TargetLibraryInfo *libInfo) {

diff --git a/lib/Target/ARM/ARMInstrInfo.cpp b/lib/Target/ARM/ARMInstrInfo.cpp
@@ -13,13 +13,17 @@
 
 #include "ARMInstrInfo.h"
 #include "ARM.h"
+#include "ARMConstantPoolValue.h"
 #include "ARMMachineFunctionInfo.h"
+#include "ARMTargetMachine.h"
 #include "MCTargetDesc/ARMAddressingModes.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/CodeGen/LiveVariables.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/CodeGen/MachineJumpTableInfo.h"
+#include "llvm/Function.h"
+#include "llvm/GlobalVariable.h"
 #include "llvm/MC/MCAsmInfo.h"
 #include "llvm/MC/MCInst.h"
 using namespace llvm;
@@ -84,3 +88,61 @@ unsigned ARMInstrInfo::getUnindexedOpcode(unsigned Opc) const {
 
   return 0;
 }
+
+namespace {
+  /// ARMCGBR - Create Global Base Reg pass. This initializes the PIC
+  /// global base register for ARM ELF.
+  struct ARMCGBR : public MachineFunctionPass {
+    static char ID;
+    ARMCGBR() : MachineFunctionPass(ID) {}
+
+    virtual bool runOnMachineFunction(MachineFunction &MF) {
+      ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
+      if (AFI->getGlobalBaseReg() == 0)
+        return false;
+
+      const ARMTargetMachine *TM =
+        static_cast<const ARMTargetMachine *>(&MF.getTarget());
+      if (TM->getRelocationModel() != Reloc::PIC_)
+        return false;
+
+      LLVMContext* Context = &MF.getFunction()->getContext();
+      GlobalValue *GV = new GlobalVariable(Type::getInt32Ty(*Context), false,
+                                           GlobalValue::ExternalLinkage, 0,
+                                           "_GLOBAL_OFFSET_TABLE_");
+      unsigned Id = AFI->createPICLabelUId();
+      ARMConstantPoolValue *CPV = ARMConstantPoolConstant::Create(GV, Id);
+      unsigned Align = TM->getTargetData()->getPrefTypeAlignment(GV->getType());
+      unsigned Idx = MF.getConstantPool()->getConstantPoolIndex(CPV, Align);
+
+      MachineBasicBlock &FirstMBB = MF.front();
+      MachineBasicBlock::iterator MBBI = FirstMBB.begin();
+      DebugLoc DL = FirstMBB.findDebugLoc(MBBI);
+      unsigned GlobalBaseReg = AFI->getGlobalBaseReg();
+      unsigned Opc = TM->getSubtarget<ARMSubtarget>().isThumb2() ?
+                     ARM::t2LDRpci : ARM::LDRcp;
+      const TargetInstrInfo &TII = *TM->getInstrInfo();
+      MachineInstrBuilder MIB = BuildMI(FirstMBB, MBBI, DL,
+                                        TII.get(Opc), GlobalBaseReg)
+                                .addConstantPoolIndex(Idx);
+      if (Opc == ARM::LDRcp)
+        MIB.addImm(0);
+      AddDefaultPred(MIB);
+
+      return true;
+    }
+
+    virtual const char *getPassName() const {
+      return "ARM PIC Global Base Reg Initialization";
+    }
+
+    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+      AU.setPreservesCFG();
+      MachineFunctionPass::getAnalysisUsage(AU);
+    }
+  };
+}
+
+char ARMCGBR::ID = 0;
+FunctionPass*
+llvm::createARMGlobalBaseRegPass() { return new ARMCGBR(); }
diff --git a/lib/Target/ARM/ARMMachineFunctionInfo.h b/lib/Target/ARM/ARMMachineFunctionInfo.h
@@ -108,6 +108,11 @@ class ARMFunctionInfo : public MachineFunctionInfo {
   /// pass.
   DenseMap<unsigned, unsigned> CPEClones;
 
+  /// GlobalBaseReg - keeps track of the virtual register initialized for
+  /// use as the global base register. This is used for PIC in some PIC
+  /// relocation models.
+  unsigned GlobalBaseReg;
+
 public:
   ARMFunctionInfo() :
     isThumb(false),
@@ -119,7 +124,7 @@ class ARMFunctionInfo : public MachineFunctionInfo {
     GPRCS1Frames(0), GPRCS2Frames(0), DPRCSFrames(0),
     NumAlignedDPRCS2Regs(0),
     JumpTableUId(0), PICLabelUId(0),
-    VarArgsFrameIndex(0), HasITBlocks(false) {}
+    VarArgsFrameIndex(0), HasITBlocks(false), GlobalBaseReg(0) {}
 
   explicit ARMFunctionInfo(MachineFunction &MF) :
     isThumb(MF.getTarget().getSubtarget<ARMSubtarget>().isThumb()),
@@ -130,7 +135,7 @@ class ARMFunctionInfo : public MachineFunctionInfo {
     GPRCS1Size(0), GPRCS2Size(0), DPRCSSize(0),
     GPRCS1Frames(32), GPRCS2Frames(32), DPRCSFrames(32),
     JumpTableUId(0), PICLabelUId(0),
-    VarArgsFrameIndex(0), HasITBlocks(false) {}
+    VarArgsFrameIndex(0), HasITBlocks(false), GlobalBaseReg(0) {}
 
   bool isThumbFunction() const { return isThumb; }
   bool isThumb1OnlyFunction() const { return isThumb && !hasThumb2; }
@@ -249,6 +254,9 @@ class ARMFunctionInfo : public MachineFunctionInfo {
   bool hasITBlocks() const { return HasITBlocks; }
   void setHasITBlocks(bool h) { HasITBlocks = h; }
 
+  unsigned getGlobalBaseReg() const { return GlobalBaseReg; }
+  void setGlobalBaseReg(unsigned Reg) { GlobalBaseReg = Reg; }
+
   void recordCPEClone(unsigned CPIdx, unsigned CPCloneIdx) {
     if (!CPEClones.insert(std::make_pair(CPCloneIdx, CPIdx)).second)
       assert(0 && "Duplicate entries!");

diff --git a/lib/Target/ARM/ARMTargetMachine.cpp b/lib/Target/ARM/ARMTargetMachine.cpp
@@ -143,6 +143,11 @@ bool ARMPassConfig::addPreISel() {
 
 bool ARMPassConfig::addInstSelector() {
   addPass(createARMISelDag(getARMTargetMachine(), getOptLevel()));
+
+  const ARMSubtarget *Subtarget = &getARMSubtarget();
+  if (Subtarget->isTargetELF() && !Subtarget->isThumb1Only() &&
+      TM->Options.EnableFastISel)
+    addPass(createARMGlobalBaseRegPass());
   return false;
 }
 

diff --git a/test/CodeGen/ARM/fast-isel-pic.ll b/test/CodeGen/ARM/fast-isel-pic.ll
@@ -1,6 +1,8 @@
 ; RUN: llc < %s -O0 -fast-isel-abort -relocation-model=pic -mtriple=thumbv7-apple-ios | FileCheck %s --check-prefix=THUMB
 ; RUN: llc < %s -O0 -fast-isel-abort -relocation-model=pic -mtriple=arm-apple-ios | FileCheck %s --check-prefix=ARM
 ; RUN: llc < %s -O0 -fast-isel-abort -relocation-model=pic -mtriple=armv7-apple-ios | FileCheck %s --check-prefix=ARMv7
+; RUN: llc < %s -O0 -fast-isel-abort -relocation-model=pic -mtriple=thumbv7-none-linux-gnueabi | FileCheck %s --check-prefix=THUMB-ELF
+; RUN: llc < %s -O0 -fast-isel-abort -relocation-model=pic -mtriple=armv7-none-linux-gnueabi | FileCheck %s --check-prefix=ARMv7-ELF
 
 @g = global i32 0, align 4
 
@@ -10,13 +12,21 @@ entry:
 ; THUMB: movw [[reg0:r[0-9]+]],
 ; THUMB: movt [[reg0]],
 ; THUMB: add  [[reg0]], pc
+; THUMB-ELF: LoadGV
+; THUMB-ELF: ldr.n r[[reg0:[0-9]+]],
+; THUMB-ELF: ldr.n r[[reg1:[0-9]+]],
+; THUMB-ELF: ldr r[[reg0]], [r[[reg1]], r[[reg0]]]
 ; ARM: LoadGV
 ; ARM: ldr [[reg1:r[0-9]+]],
 ; ARM: add [[reg1]], pc, [[reg1]]
 ; ARMv7: LoadGV
 ; ARMv7: movw [[reg2:r[0-9]+]],
 ; ARMv7: movt [[reg2]],
 ; ARMv7: add  [[reg2]], pc, [[reg2]]
+; ARMv7-ELF: LoadGV
+; ARMv7-ELF: ldr r[[reg2:[0-9]+]],
+; ARMv7-ELF: ldr r[[reg3:[0-9]+]],
+; ARMv7-ELF: ldr r[[reg2]], [r[[reg3]], r[[reg2]]]
   %tmp = load i32* @g
   ret i32 %tmp
 }
@@ -30,6 +40,10 @@ entry:
 ; THUMB: movt r[[reg3]],
 ; THUMB: add  r[[reg3]], pc
 ; THUMB: ldr  r[[reg3]], [r[[reg3]]]
+; THUMB-ELF: LoadIndirectSymbol
+; THUMB-ELF: ldr.n r[[reg3:[0-9]+]],
+; THUMB-ELF: ldr.n r[[reg4:[0-9]+]],
+; THUMB-ELF: ldr r[[reg3]], [r[[reg4]], r[[reg3]]]
 ; ARM: LoadIndirectSymbol
 ; ARM: ldr [[reg4:r[0-9]+]],
 ; ARM: ldr [[reg4]], [pc, [[reg4]]]
@@ -38,6 +52,10 @@ entry:
 ; ARMv7: movt r[[reg5]],
 ; ARMv7: add  r[[reg5]], pc, r[[reg5]]
 ; ARMv7: ldr  r[[reg5]], [r[[reg5]]]
+; ARMv7-ELF: LoadIndirectSymbol
+; ARMv7-ELF: ldr r[[reg5:[0-9]+]],
+; ARMv7-ELF: ldr r[[reg6:[0-9]+]],
+; ARMv7-ELF: ldr r[[reg5]], [r[[reg6]], r[[reg5]]]
   %tmp = load i32* @i
   ret i32 %tmp
 }