[AArch64] - Return address signing dwarf support

[LukeCheeseman]

Summary

Functions that have signed return addresses need additional dwarf support:

• After signing the LR, and before authenticating it, the LR register is in a state the is unusable by a debugger or unwinder
• To account for this a new directive, .cfi_negate_ra_state, is added
• This directive says the signed state of the LR register has now changed, i.e. unsigned -> signed or signed -> unsigned
• This directive has the same CFA code as the SPARC directive GNU_window_save (0x2d), adding a macro to account for multiply defined codes
• This patch matches the gcc implementation of this support: https://patchwork.ozlabs.org/patch/800271/

[LukeCheeseman]

I would specifically like some feedback on the duplicated code. I couldn't find a way of adding vendor specific extensions and this seems like it may be a large change for a small result.

[LukeCheeseman]

• Don't generate cfi_negate_ra_state when using retaa
• The signing and return is combined so there isn't a correct point to emit it

[LukeCheeseman]

ping

[olista01]

I think this needs additional tests to cover:

• The encoding of the CFI instruction in object files
• The assembly parser and printer changes, to make sure that IR->assembly->object will work the same as IR->object.

lib/BinaryFormat/Dwarf.cpp

475

Does this get used by llvm-dwarfdump, or is it just for debugging? If it is used by anything user-facing, then we need to select the right one based on the target here.

lib/MC/MCParser/AsmParser.cpp

494 ↗

(On Diff #159525)

I think this should be done in the AArch64 assembly parser, because it doesn't apply to any other architectures.

lib/Target/AArch64/AArch64FrameLowering.cpp

897

This isn't covered by the test.

[LukeCheeseman]

• Removed the emission of negate_ra_state when emitting an AUTIASP. This causes issues when there are multiple exits from a function as the unwinder reads all the CFI directives without understanding control flow and may end up with the incorrect value in the sign state register. This isn't an issue as we currently do not support breaking on function return/tail call.
• Moved cfi_negate_ra_sign_state directive parsing into the aarch64 backend
• I haven't yet addressed the issue of printing the wrong cfi directive name with llvm-dwarfdump, I have a patch which I'm not too happy with but will upload shortly

[thegameg]

Can you please also add support for the new CFI directive in MIR? Mostly in lib/CodeGen/MIRParser/MIParser.cpp, lib/CodeGen/MIRParser/MILexer.cpp/h, lib/CodeGen/MIRPrinter.cpp and test/CodeGen/MIR/AArch64/cfi.mir.

Also, probably a MIR test with -run-pass prologepilog would be good.

[LukeCheeseman]

• Add MIR support (lexing, parsing and printing) for CFI negate_ra_sign_state
• Added MIR tests
• Added patch to print the correct CFI name given the target architecture of the object file being dumper:
  • This solution isn't pretty, the issue is that the dumping doesn't have any notion of context. I attempted to create a DwarfDumper class with vendor specific subclasses but this ended up needing significant plumbing and rework to many, far reaching places, in the LLVM code base.
  • I've opted here to pass through only the information that is required to make the dumping decision, in this case the architecture. The DW_CFA_DUP macro now takes a predicate that can be defined for selecting the vendor specific extension
  • An alternative to this is to just accept that the output is incorrect, this doesn't look to have any issues beyond printing the wrong string. Notably it is not an issue for target specific parsing/printing, only when dumping a file via llvm-dwarfdump. Even if overlapping encodings behaved differently (e.g. differing numbers of operands) the dumping would not be any more incorrect that just printing the wrong name. Whilst not a motivating point, it should be highlighted that this is the way that GNU tools handle overlapping encodings, i.e. the encoding not defined as a DUP is used.
  • I've extend a test to check for the correct string be printed

[olista01]

include/llvm/BinaryFormat/Dwarf.def

90

Maybe HANDLE_DW_CFA_TARGET or HANDLE_DW_CFA_PRED would be a better name for this?

839

Now that we have a way to represent target-specific opcodes, should we limit these to their respective targets? I think that GNU_window_save is PowerPC-specific, and GNU_args_size is X86-specific.

lib/BinaryFormat/Dwarf.cpp

461

Should this be #undef'd after these includes? The other macros are cleared at the end of Dwarf.def, but this one isn't.

461

I think this also needs to check Triple::aarch64_be.

lib/Target/AArch64/AsmParser/AArch64AsmParser.cpp

5290

The assembly parsing isn't covered by any of the tests.

5292

Do we need to check that we are at AsmToken::EndOfStatement (as is done for other directives), or is that checked somewhere else?

tools/llvm-readobj/DwarfCFIEHPrinter.h

190

Long line, please clang-format this.

[LukeCheeseman]

Added tests for assembly parsing
Added predicates for the other dwarf vendor extensions

[JDevlieghere]

include/llvm/BinaryFormat/Dwarf.def

844

It doesn't seem right that these two predicates have the same ID?

[LukeCheeseman]

include/llvm/BinaryFormat/Dwarf.def

844

The two directives are provided by two different vendors. Although it makes selecting the correct one for dumping require more information, I think this is legal based on the DWARF standard. This is also the encoding used for the GNU patch that provides the same functionality (see https://github.com/gcc-mirror/gcc/blob/f47edbb56126b7e8564f22b0b9d55565a995b7f0/include/dwarf2.def)

[thegameg]

test/CodeGen/MIR/AArch64/return-address-signing.mir

13

Are all the attributes needed here? I assume only the sign-* are.

49

A lot of these are not needed here. In order to dump the MIR file you can use -simplify-mir with -stop-before or -stop-after to skip all the default values.

[LukeCheeseman]

• Reduce MIR test case

[olista01]

LGTM, thanks!

include/llvm/BinaryFormat/Dwarf.h

 #define LLVM_BINARYFORMAT_DWARF_H
 
 #include "llvm/ADT/Optional.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/DataTypes.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/Format.h"
 #include "llvm/Support/FormatVariadicDetails.h"
+#include "llvm/ADT/Triple.h"
 
 namespace llvm {
 class StringRef;
 
 namespace dwarf {
 
 //===----------------------------------------------------------------------===//
 // DWARF constants as gleaned from the DWARF Debugging Information Format V.5

 enum RangeListEntries {
 #define HANDLE_DW_RLE(ID, NAME) DW_RLE_##NAME = ID,
 #include "llvm/BinaryFormat/Dwarf.def"
 };
 
 /// Call frame instruction encodings.
 enum CallFrameInfo {
 #define HANDLE_DW_CFA(ID, NAME) DW_CFA_##NAME = ID,
+#define HANDLE_DW_CFA_DUP(ID, NAME, ARCH) DW_CFA_##NAME = ID,
 #include "llvm/BinaryFormat/Dwarf.def"
   DW_CFA_extended = 0x00,
 
   DW_CFA_lo_user = 0x1c,
   DW_CFA_hi_user = 0x3f
 };
 
 enum Constants {

 StringRef CaseString(unsigned Case);
 StringRef ConventionString(unsigned Convention);
 StringRef InlineCodeString(unsigned Code);
 StringRef ArrayOrderString(unsigned Order);
 StringRef LNStandardString(unsigned Standard);
 StringRef LNExtendedString(unsigned Encoding);
 StringRef MacinfoString(unsigned Encoding);
 StringRef RangeListEncodingString(unsigned Encoding);
-StringRef CallFrameString(unsigned Encoding);
+StringRef CallFrameString(unsigned Encoding, Triple::ArchType Arch);
 StringRef ApplePropertyString(unsigned);
 StringRef UnitTypeString(unsigned);
 StringRef AtomTypeString(unsigned Atom);
 StringRef GDBIndexEntryKindString(GDBIndexEntryKind Kind);
 StringRef GDBIndexEntryLinkageString(GDBIndexEntryLinkage Linkage);
 StringRef IndexString(unsigned Idx);
 /// @}
 

include/llvm/BinaryFormat/Dwarf.def

 
 // TODO: Add other DW-based macros.
 #if !(                                                                         \
     defined HANDLE_DW_TAG || defined HANDLE_DW_AT || defined HANDLE_DW_FORM || \
     defined HANDLE_DW_OP || defined HANDLE_DW_LANG || defined HANDLE_DW_ATE || \
     defined HANDLE_DW_VIRTUALITY || defined HANDLE_DW_DEFAULTED ||             \
     defined HANDLE_DW_CC || defined HANDLE_DW_LNS || defined HANDLE_DW_LNE ||  \
     defined HANDLE_DW_LNCT || defined HANDLE_DW_MACRO ||                       \
-    defined HANDLE_DW_RLE || defined HANDLE_DW_CFA ||                          \
+    defined HANDLE_DW_RLE ||                                                   \
+    (defined HANDLE_DW_CFA && defined HANDLE_DW_CFA_DUP) ||                    \
     defined HANDLE_DW_APPLE_PROPERTY || defined HANDLE_DW_UT ||                \
     defined HANDLE_DWARF_SECTION || defined HANDLE_DW_IDX ||                   \
     defined HANDLE_DW_END)
 #error "Missing macro definition of HANDLE_DW*"
 #endif
 
 #ifndef HANDLE_DW_TAG
 #define HANDLE_DW_TAG(ID, NAME, VERSION, VENDOR)

 #ifndef HANDLE_DW_RLE
 #define HANDLE_DW_RLE(ID, NAME)
 #endif
 
 #ifndef HANDLE_DW_CFA
 #define HANDLE_DW_CFA(ID, NAME)
 #endif
 
+#ifndef HANDLE_DW_CFA_DUP
+#define HANDLE_DW_CFA_DUP(ID, NAME, PRED)
+#endif
+
 #ifndef HANDLE_DW_APPLE_PROPERTY
 #define HANDLE_DW_APPLE_PROPERTY(ID, NAME)
 #endif
 
 #ifndef HANDLE_DW_UT
 #define HANDLE_DW_UT(ID, NAME)
 #endif
 

 HANDLE_DW_CFA(0x10, expression)
 HANDLE_DW_CFA(0x11, offset_extended_sf)
 HANDLE_DW_CFA(0x12, def_cfa_sf)
 HANDLE_DW_CFA(0x13, def_cfa_offset_sf)
 HANDLE_DW_CFA(0x14, val_offset)
 HANDLE_DW_CFA(0x15, val_offset_sf)
 HANDLE_DW_CFA(0x16, val_expression)
 // Vendor extensions:
 HANDLE_DW_CFA(0x1d, MIPS_advance_loc8)
 HANDLE_DW_CFA(0x2d, GNU_window_save)
+HANDLE_DW_CFA_DUP(0x2d, AARCH64_negate_ra_state, SELECT_AARCH64)
 HANDLE_DW_CFA(0x2e, GNU_args_size)
 
 // Apple Objective-C Property Attributes.
 // Keep this list in sync with clang's DeclSpec.h ObjCPropertyAttributeKind!
 HANDLE_DW_APPLE_PROPERTY(0x01, readonly)
 HANDLE_DW_APPLE_PROPERTY(0x02, getter)
 HANDLE_DW_APPLE_PROPERTY(0x04, assign)
 HANDLE_DW_APPLE_PROPERTY(0x08, readwrite)
 HANDLE_DW_APPLE_PROPERTY(0x10, retain)
 HANDLE_DW_APPLE_PROPERTY(0x20, copy)
 HANDLE_DW_APPLE_PROPERTY(0x40, nonatomic)

 #undef HANDLE_DW_DEFAULTED
 #undef HANDLE_DW_CC
 #undef HANDLE_DW_LNS
 #undef HANDLE_DW_LNE
 #undef HANDLE_DW_LNCT
 #undef HANDLE_DW_MACRO
 #undef HANDLE_DW_RLE
 #undef HANDLE_DW_CFA
+#undef HANDLE_DW_CFA_DUP
 #undef HANDLE_DW_APPLE_PROPERTY
 #undef HANDLE_DW_UT
 #undef HANDLE_DWARF_SECTION
 #undef HANDLE_DW_IDX
 #undef HANDLE_DW_END

include/llvm/DebugInfo/DWARF/DWARFContext.h

struct DWOFile {
     object::OwningBinary<object::ObjectFile> File;
     std::unique_ptr<DWARFContext> Context;
   };
   StringMap<std::weak_ptr<DWOFile>> DWOFiles;
   std::weak_ptr<DWOFile> DWP;
   bool CheckedForDWP = false;
   std::string DWPName;
 
+  Triple::ArchType Arch;
   std::unique_ptr<MCRegisterInfo> RegInfo;
 
   /// Read compile units from the debug_info section (if necessary)
   /// and type units from the debug_types sections (if necessary)
   /// and store them in NormalUnits.
   void parseNormalUnits();
 
   /// Read compile units from the debug_info.dwo section (if necessary)
public:
 
   static std::unique_ptr<DWARFContext>
   create(const StringMap<std::unique_ptr<MemoryBuffer>> &Sections,
          uint8_t AddrSize, bool isLittleEndian = sys::IsLittleEndianHost);
 
   /// Loads register info for the architecture of the provided object file.
   /// Improves readability of dumped DWARF expressions. Requires the caller to
   /// have initialized the relevant target descriptions.
-  Error loadRegisterInfo(const object::ObjectFile &Obj);
+  Error loadArchitectureInfo(const object::ObjectFile &Obj);
 
   /// Get address size from CUs.
   /// TODO: refactor compile_units() to make this const.
   uint8_t getCUAddrSize();
 
   /// Dump Error as warning message to stderr.
   static void dumpWarning(Error Warning);
 

include/llvm/DebugInfo/DWARF/DWARFDebugFrame.h

 //===- DWARFDebugFrame.h - Parsing of .debug_frame --------------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLVM_DEBUGINFO_DWARF_DWARFDEBUGFRAME_H
 #define LLVM_DEBUGINFO_DWARF_DWARFDEBUGFRAME_H
 
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/iterator.h"
 #include "llvm/ADT/SmallString.h"
+#include "llvm/ADT/Triple.h"
 #include "llvm/DebugInfo/DWARF/DWARFDataExtractor.h"
 #include "llvm/DebugInfo/DWARF/DWARFExpression.h"
 #include "llvm/Support/Error.h"
 #include <memory>
 #include <vector>
 
 namespace llvm {
 
public:
   iterator begin() { return Instructions.begin(); }
   const_iterator begin() const { return Instructions.begin(); }
   iterator end() { return Instructions.end(); }
   const_iterator end() const { return Instructions.end(); }
 
   unsigned size() const { return (unsigned)Instructions.size(); }
   bool empty() const { return Instructions.empty(); }
 
-  CFIProgram(uint64_t CodeAlignmentFactor, int64_t DataAlignmentFactor)
+  CFIProgram(uint64_t CodeAlignmentFactor, int64_t DataAlignmentFactor,
+             Triple::ArchType Arch)
       : CodeAlignmentFactor(CodeAlignmentFactor),
-        DataAlignmentFactor(DataAlignmentFactor) {}
+        DataAlignmentFactor(DataAlignmentFactor),
+        Arch(Arch) {}
 
   /// Parse and store a sequence of CFI instructions from Data,
   /// starting at *Offset and ending at EndOffset. *Offset is updated
   /// to EndOffset upon successful parsing, or indicates the offset
   /// where a problem occurred in case an error is returned.
   Error parse(DataExtractor Data, uint32_t *Offset, uint32_t EndOffset);
 
   void dump(raw_ostream &OS, const MCRegisterInfo *MRI, bool IsEH,
             unsigned IndentLevel = 1) const;
 
 private:
   std::vector<Instruction> Instructions;
   const uint64_t CodeAlignmentFactor;
   const int64_t DataAlignmentFactor;
+  Triple::ArchType Arch;
 
   /// Convenience method to add a new instruction with the given opcode.
   void addInstruction(uint8_t Opcode) {
     Instructions.push_back(Instruction(Opcode));
   }
 
   /// Add a new single-operand instruction.
   void addInstruction(uint8_t Opcode, uint64_t Operand1) {

 
 /// An entry in either debug_frame or eh_frame. This entry can be a CIE or an
 /// FDE.
 class FrameEntry {
 public:
   enum FrameKind { FK_CIE, FK_FDE };
 
   FrameEntry(FrameKind K, uint64_t Offset, uint64_t Length, uint64_t CodeAlign,
-             int64_t DataAlign)
-      : Kind(K), Offset(Offset), Length(Length), CFIs(CodeAlign, DataAlign) {}
+             int64_t DataAlign, Triple::ArchType Arch)
+      : Kind(K), Offset(Offset), Length(Length),
+        CFIs(CodeAlign, DataAlign, Arch) {}
 
   virtual ~FrameEntry() {}
 
   FrameKind getKind() const { return Kind; }
   uint64_t getOffset() const { return Offset; }
   uint64_t getLength() const { return Length; }
   const CFIProgram &cfis() const { return CFIs; }
   CFIProgram &cfis() { return CFIs; }
public:
   // CIEs (and FDEs) are simply container classes, so the only sensible way to
   // create them is by providing the full parsed contents in the constructor.
   CIE(uint64_t Offset, uint64_t Length, uint8_t Version,
       SmallString<8> Augmentation, uint8_t AddressSize,
       uint8_t SegmentDescriptorSize, uint64_t CodeAlignmentFactor,
       int64_t DataAlignmentFactor, uint64_t ReturnAddressRegister,
       SmallString<8> AugmentationData, uint32_t FDEPointerEncoding,
       uint32_t LSDAPointerEncoding, Optional<uint64_t> Personality,
-      Optional<uint32_t> PersonalityEnc)
+      Optional<uint32_t> PersonalityEnc, Triple::ArchType Arch)
       : FrameEntry(FK_CIE, Offset, Length, CodeAlignmentFactor,
-                   DataAlignmentFactor),
+                   DataAlignmentFactor, Arch),
         Version(Version), Augmentation(std::move(Augmentation)),
         AddressSize(AddressSize), SegmentDescriptorSize(SegmentDescriptorSize),
         CodeAlignmentFactor(CodeAlignmentFactor),
         DataAlignmentFactor(DataAlignmentFactor),
         ReturnAddressRegister(ReturnAddressRegister),
         AugmentationData(std::move(AugmentationData)),
         FDEPointerEncoding(FDEPointerEncoding),
         LSDAPointerEncoding(LSDAPointerEncoding), Personality(Personality),

 /// DWARF Frame Description Entry (FDE)
 class FDE : public FrameEntry {
 public:
   // Each FDE has a CIE it's "linked to". Our FDE contains is constructed with
   // an offset to the CIE (provided by parsing the FDE header). The CIE itself
   // is obtained lazily once it's actually required.
   FDE(uint64_t Offset, uint64_t Length, int64_t LinkedCIEOffset,
       uint64_t InitialLocation, uint64_t AddressRange, CIE *Cie,
-      Optional<uint64_t> LSDAAddress)
+      Optional<uint64_t> LSDAAddress, Triple::ArchType Arch)
       : FrameEntry(FK_FDE, Offset, Length,
                    Cie ? Cie->getCodeAlignmentFactor() : 0,
-                   Cie ? Cie->getDataAlignmentFactor() : 0),
+                   Cie ? Cie->getDataAlignmentFactor() : 0,
+                   Arch),
         LinkedCIEOffset(LinkedCIEOffset), InitialLocation(InitialLocation),
         AddressRange(AddressRange), LinkedCIE(Cie), LSDAAddress(LSDAAddress) {}
 
   ~FDE() override = default;
 
   const CIE *getLinkedCIE() const { return LinkedCIE; }
   uint64_t getInitialLocation() const { return InitialLocation; }
   uint64_t getAddressRange() const { return AddressRange; }
private:
   const CIE *LinkedCIE;
   const Optional<uint64_t> LSDAAddress;
 };
 
 } // end namespace dwarf
 
 /// A parsed .debug_frame or .eh_frame section
 class DWARFDebugFrame {
+  const Triple::ArchType Arch;
   // True if this is parsing an eh_frame section.
   const bool IsEH;
   // Not zero for sane pointer values coming out of eh_frame
   const uint64_t EHFrameAddress;
 
   std::vector<std::unique_ptr<dwarf::FrameEntry>> Entries;
   using iterator = pointee_iterator<decltype(Entries)::const_iterator>;
 
   /// Return the entry at the given offset or nullptr.
   dwarf::FrameEntry *getEntryAtOffset(uint64_t Offset) const;
 
 public:
   // If IsEH is true, assume it is a .eh_frame section. Otherwise,
   // it is a .debug_frame section. EHFrameAddress should be different
   // than zero for correct parsing of .eh_frame addresses when they
   // use a PC-relative encoding.
-  DWARFDebugFrame(bool IsEH = false, uint64_t EHFrameAddress = 0);
+  DWARFDebugFrame(Triple::ArchType Arch,
+                  bool IsEH = false, uint64_t EHFrameAddress = 0);
   ~DWARFDebugFrame();
 
   /// Dump the section data into the given stream.
   void dump(raw_ostream &OS, const MCRegisterInfo *MRI,
             Optional<uint64_t> Offset) const;
 
   /// Parse the section from raw data. \p Data is assumed to contain the whole
   /// frame section contents to be parsed.

include/llvm/MC/MCDwarf.h

enum OpType {
     OpDefCfa,
     OpRelOffset,
     OpAdjustCfaOffset,
     OpEscape,
     OpRestore,
     OpUndefined,
     OpRegister,
     OpWindowSave,
+    OpNegateRAState,
     OpGnuArgsSize
   };
 
 private:
   OpType Operation;
   MCSymbol *Label;
   unsigned Register;
   union {
static MCCFIInstruction createRegister(MCSymbol *L, unsigned Register1,
     return MCCFIInstruction(OpRegister, L, Register1, Register2);
   }
 
   /// .cfi_window_save SPARC register window is saved.
   static MCCFIInstruction createWindowSave(MCSymbol *L) {
     return MCCFIInstruction(OpWindowSave, L, 0, 0, "");
   }
 
+  /// .cfi_negate_ra_state AArch64 negate RA state.
+  static MCCFIInstruction createNegateRAState(MCSymbol *L) {
+    return MCCFIInstruction(OpNegateRAState, L, 0, 0, "");
+  }
+
   /// .cfi_restore says that the rule for Register is now the same as it
   /// was at the beginning of the function, after all initial instructions added
   /// by .cfi_startproc were executed.
   static MCCFIInstruction createRestore(MCSymbol *L, unsigned Register) {
     return MCCFIInstruction(OpRestore, L, Register, 0, "");
   }
 
   /// .cfi_undefined From now on the previous value of Register can't be

include/llvm/MC/MCStreamer.h

public:
   virtual void EmitCFIAdjustCfaOffset(int64_t Adjustment);
   virtual void EmitCFIEscape(StringRef Values);
   virtual void EmitCFIReturnColumn(int64_t Register);
   virtual void EmitCFIGnuArgsSize(int64_t Size);
   virtual void EmitCFISignalFrame();
   virtual void EmitCFIUndefined(int64_t Register);
   virtual void EmitCFIRegister(int64_t Register1, int64_t Register2);
   virtual void EmitCFIWindowSave();
+  virtual void EmitCFINegateRAState();
 
   virtual void EmitWinCFIStartProc(const MCSymbol *Symbol, SMLoc Loc = SMLoc());
   virtual void EmitWinCFIEndProc(SMLoc Loc = SMLoc());
   virtual void EmitWinCFIStartChained(SMLoc Loc = SMLoc());
   virtual void EmitWinCFIEndChained(SMLoc Loc = SMLoc());
   virtual void EmitWinCFIPushReg(unsigned Register, SMLoc Loc = SMLoc());
   virtual void EmitWinCFISetFrame(unsigned Register, unsigned Offset,
                                   SMLoc Loc = SMLoc());

lib/BinaryFormat/Dwarf.cpp

 //===-- llvm/BinaryFormat/Dwarf.cpp - Dwarf Framework ------------*- C++-*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //
 // This file contains support for generic dwarf information.
 //
 //===----------------------------------------------------------------------===//
 
 #include "llvm/BinaryFormat/Dwarf.h"
 #include "llvm/ADT/StringSwitch.h"
+#include "llvm/ADT/Triple.h"
 #include "llvm/Support/ErrorHandling.h"
 
 using namespace llvm;
 using namespace dwarf;
 
 StringRef llvm::dwarf::TagString(unsigned Tag) {
   switch (Tag) {
   default:
default:
     return StringRef();
 #define HANDLE_DW_RLE(ID, NAME)                                                \
   case DW_RLE_##NAME:                                                          \
     return "DW_RLE_" #NAME;
 #include "llvm/BinaryFormat/Dwarf.def"
   }
 }
 
-StringRef llvm::dwarf::CallFrameString(unsigned Encoding) {
+StringRef llvm::dwarf::CallFrameString(unsigned Encoding,
+    Triple::ArchType Arch) {
+#define SELECT_AARCH64 Arch == llvm::Triple::aarch64
+#define HANDLE_DW_CFA(ID, NAME)
+#define HANDLE_DW_CFA_DUP(ID, NAME, PRED) \
+  if (ID == Encoding && PRED) \
+    return "DW_CFA_" #NAME;
+#include "llvm/BinaryFormat/Dwarf.def"
+
   switch (Encoding) {
   default:
     return StringRef();
+#define HANDLE_DW_CFA_DUP(ID, NAME, PRED)
 #define HANDLE_DW_CFA(ID, NAME)                                                \
   case DW_CFA_##NAME:                                                          \
     return "DW_CFA_" #NAME;
 #include "llvm/BinaryFormat/Dwarf.def"
   }
 }
 
 StringRef llvm::dwarf::ApplePropertyString(unsigned Prop) {
   switch (Prop) {
   default:
     return StringRef();
 #define HANDLE_DW_APPLE_PROPERTY(ID, NAME)                                     \

lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp

case MCCFIInstruction::OpOffset:
     OutStreamer->EmitCFIOffset(Inst.getRegister(), Inst.getOffset());
     break;
   case MCCFIInstruction::OpRegister:
     OutStreamer->EmitCFIRegister(Inst.getRegister(), Inst.getRegister2());
     break;
   case MCCFIInstruction::OpWindowSave:
     OutStreamer->EmitCFIWindowSave();
     break;
+  case MCCFIInstruction::OpNegateRAState:
+    OutStreamer->EmitCFINegateRAState();
+    break;
   case MCCFIInstruction::OpSameValue:
     OutStreamer->EmitCFISameValue(Inst.getRegister());
     break;
   case MCCFIInstruction::OpGnuArgsSize:
     OutStreamer->EmitCFIGnuArgsSize(Inst.getOffset());
     break;
   case MCCFIInstruction::OpEscape:
     OutStreamer->EmitCFIEscape(Inst.getValues());

lib/CodeGen/CFIInstrInserter.cpp

#endif
       case MCCFIInstruction::OpSameValue:
       case MCCFIInstruction::OpOffset:
       case MCCFIInstruction::OpRelOffset:
       case MCCFIInstruction::OpEscape:
       case MCCFIInstruction::OpRestore:
       case MCCFIInstruction::OpUndefined:
       case MCCFIInstruction::OpRegister:
       case MCCFIInstruction::OpWindowSave:
+      case MCCFIInstruction::OpNegateRAState:
       case MCCFIInstruction::OpGnuArgsSize:
         break;
       }
     }
   }
 
   MBBInfo.Processed = true;
 

lib/CodeGen/MIRParser/MILexer.h

enum TokenKind {
     kw_cfi_escape,
     kw_cfi_def_cfa,
     kw_cfi_register,
     kw_cfi_remember_state,
     kw_cfi_restore,
     kw_cfi_restore_state,
     kw_cfi_undefined,
     kw_cfi_window_save,
+    kw_cfi_aarch64_negate_ra_sign_state,
     kw_blockaddress,
     kw_intrinsic,
     kw_target_index,
     kw_half,
     kw_float,
     kw_double,
     kw_x86_fp80,
     kw_fp128,

lib/CodeGen/MIRParser/MILexer.cpp

return StringSwitch<MIToken::TokenKind>(Identifier)
       .Case("escape", MIToken::kw_cfi_escape)
       .Case("def_cfa", MIToken::kw_cfi_def_cfa)
       .Case("remember_state", MIToken::kw_cfi_remember_state)
       .Case("restore", MIToken::kw_cfi_restore)
       .Case("restore_state", MIToken::kw_cfi_restore_state)
       .Case("undefined", MIToken::kw_cfi_undefined)
       .Case("register", MIToken::kw_cfi_register)
       .Case("window_save", MIToken::kw_cfi_window_save)
+      .Case("negate_ra_sign_state", MIToken::kw_cfi_aarch64_negate_ra_sign_state)
       .Case("blockaddress", MIToken::kw_blockaddress)
       .Case("intrinsic", MIToken::kw_intrinsic)
       .Case("target-index", MIToken::kw_target_index)
       .Case("half", MIToken::kw_half)
       .Case("float", MIToken::kw_float)
       .Case("double", MIToken::kw_double)
       .Case("x86_fp80", MIToken::kw_x86_fp80)
       .Case("fp128", MIToken::kw_fp128)

lib/CodeGen/MIRParser/MIParser.cpp

case MIToken::kw_cfi_register: {
 
     CFIIndex =
         MF.addFrameInst(MCCFIInstruction::createRegister(nullptr, Reg, Reg2));
     break;
   }
   case MIToken::kw_cfi_window_save:
     CFIIndex = MF.addFrameInst(MCCFIInstruction::createWindowSave(nullptr));
     break;
+  case MIToken::kw_cfi_aarch64_negate_ra_sign_state:
+    CFIIndex = MF.addFrameInst(MCCFIInstruction::createNegateRAState(nullptr));
+    break;
   case MIToken::kw_cfi_escape: {
     std::string Values;
     if (parseCFIEscapeValues(Values))
       return true;
     CFIIndex = MF.addFrameInst(MCCFIInstruction::createEscape(nullptr, Values));
     break;
   }
   default:
bool MIParser::parseMachineOperand(MachineOperand &Dest,
   case MIToken::kw_cfi_escape:
   case MIToken::kw_cfi_def_cfa:
   case MIToken::kw_cfi_register:
   case MIToken::kw_cfi_remember_state:
   case MIToken::kw_cfi_restore:
   case MIToken::kw_cfi_restore_state:
   case MIToken::kw_cfi_undefined:
   case MIToken::kw_cfi_window_save:
+  case MIToken::kw_cfi_aarch64_negate_ra_sign_state:
     return parseCFIOperand(Dest);
   case MIToken::kw_blockaddress:
     return parseBlockAddressOperand(Dest);
   case MIToken::kw_intrinsic:
     return parseIntrinsicOperand(Dest);
   case MIToken::kw_target_index:
     return parseTargetIndexOperand(Dest);
   case MIToken::kw_liveout:

lib/CodeGen/MachineOperand.cpp

case MCCFIInstruction::OpRegister:
     OS << ", ";
     printCFIRegister(CFI.getRegister2(), OS, TRI);
     break;
   case MCCFIInstruction::OpWindowSave:
     OS << "window_save ";
     if (MCSymbol *Label = CFI.getLabel())
       MachineOperand::printSymbol(OS, *Label);
     break;
+  case MCCFIInstruction::OpNegateRAState:
+    OS << "negate_ra_sign_state ";
+    if (MCSymbol *Label = CFI.getLabel())
+      MachineOperand::printSymbol(OS, *Label);
+    break;
   default:
     // TODO: Print the other CFI Operations.
     OS << "<unserializable cfi directive>";
     break;
   }
 }
 
 void MachineOperand::print(raw_ostream &OS, const TargetRegisterInfo *TRI,

lib/DebugInfo/DWARF/DWARFContext.cpp

const DWARFDebugFrame *DWARFContext::getDebugFrame() {
   // which isn't specified in DWARF in general. It's only specified for CUs, but
   // .eh_frame can appear without a .debug_info section. Follow the example of
   // other tools (libdwarf) and extract this from the container (ObjectFile
   // provides this information). This problem is fixed in DWARFv4
   // See this dwarf-discuss discussion for more details:
   // http://lists.dwarfstd.org/htdig.cgi/dwarf-discuss-dwarfstd.org/2011-December/001173.html
   DWARFDataExtractor debugFrameData(DObj->getDebugFrameSection(),
                                     isLittleEndian(), DObj->getAddressSize());
-  DebugFrame.reset(new DWARFDebugFrame(false /* IsEH */));
+  DebugFrame.reset(new DWARFDebugFrame(Arch, false /* IsEH */));
   DebugFrame->parse(debugFrameData);
   return DebugFrame.get();
 }
 
 const DWARFDebugFrame *DWARFContext::getEHFrame() {
   if (EHFrame)
     return EHFrame.get();
 
   DWARFDataExtractor debugFrameData(DObj->getEHFrameSection(), isLittleEndian(),
                                     DObj->getAddressSize());
-  DebugFrame.reset(new DWARFDebugFrame(true /* IsEH */));
+  DebugFrame.reset(new DWARFDebugFrame(Arch, true /* IsEH */));
   DebugFrame->parse(debugFrameData);
   return DebugFrame.get();
 }
 
 const DWARFDebugMacro *DWARFContext::getDebugMacro() {
   if (Macro)
     return Macro.get();
 

 std::unique_ptr<DWARFContext>
 DWARFContext::create(const StringMap<std::unique_ptr<MemoryBuffer>> &Sections,
                      uint8_t AddrSize, bool isLittleEndian) {
   auto DObj =
       llvm::make_unique<DWARFObjInMemory>(Sections, AddrSize, isLittleEndian);
   return llvm::make_unique<DWARFContext>(std::move(DObj), "");
 }
 
-Error DWARFContext::loadRegisterInfo(const object::ObjectFile &Obj) {
+Error DWARFContext::loadArchitectureInfo(const object::ObjectFile &Obj) {
   // Detect the architecture from the object file. We usually don't need OS
   // info to lookup a target and create register info.
+  Arch = Triple::ArchType(Obj.getArch());
+
   Triple TT;
   TT.setArch(Triple::ArchType(Obj.getArch()));
   TT.setVendor(Triple::UnknownVendor);
   TT.setOS(Triple::UnknownOS);
   std::string TargetLookupError;
   const Target *TheTarget =
       TargetRegistry::lookupTarget(TT.str(), TargetLookupError);
   if (!TargetLookupError.empty())

lib/DebugInfo/DWARF/DWARFDebugFrame.cpp

void CFIProgram::printOperand(raw_ostream &OS, const MCRegisterInfo *MRI,
                               unsigned OperandIdx, uint64_t Operand) const {
   assert(OperandIdx < 2);
   uint8_t Opcode = Instr.Opcode;
   OperandType Type = getOperandTypes()[Opcode][OperandIdx];
 
   switch (Type) {
   case OT_Unset: {
     OS << " Unsupported " << (OperandIdx ? "second" : "first") << " operand to";
-    auto OpcodeName = CallFrameString(Opcode);
+    auto OpcodeName = CallFrameString(Opcode, Arch);
     if (!OpcodeName.empty())
       OS << " " << OpcodeName;
     else
       OS << format(" Opcode %x",  Opcode);
     break;
   }
   case OT_None:
     break;

 
 void CFIProgram::dump(raw_ostream &OS, const MCRegisterInfo *MRI, bool IsEH,
                       unsigned IndentLevel) const {
   for (const auto &Instr : Instructions) {
     uint8_t Opcode = Instr.Opcode;
     if (Opcode & DWARF_CFI_PRIMARY_OPCODE_MASK)
       Opcode &= DWARF_CFI_PRIMARY_OPCODE_MASK;
     OS.indent(2 * IndentLevel);
-    OS << CallFrameString(Opcode) << ":";
+    OS << CallFrameString(Opcode, Arch) << ":";
     for (unsigned i = 0; i < Instr.Ops.size(); ++i)
       printOperand(OS, MRI, IsEH, Instr, i, Instr.Ops[i]);
     OS << '\n';
   }
 }
 
 void CIE::dump(raw_ostream &OS, const MCRegisterInfo *MRI, bool IsEH) const {
   OS << format("%08x %08x %08x CIE", (uint32_t)Offset, (uint32_t)Length,
OS << format("cie=%08x pc=%08x...%08x\n", (int32_t)LinkedCIEOffset,
                (uint32_t)InitialLocation,
                (uint32_t)InitialLocation + (uint32_t)AddressRange);
   if (LSDAAddress)
     OS << format("  LSDA Address: %08x\n", *LSDAAddress);
   CFIs.dump(OS, MRI, IsEH);
   OS << "\n";
 }
 
-DWARFDebugFrame::DWARFDebugFrame(bool IsEH, uint64_t EHFrameAddress)
-    : IsEH(IsEH), EHFrameAddress(EHFrameAddress) {}
+DWARFDebugFrame::DWARFDebugFrame(Triple::ArchType Arch,
+    bool IsEH, uint64_t EHFrameAddress)
+    : Arch(Arch), IsEH(IsEH), EHFrameAddress(EHFrameAddress) {}
 
 DWARFDebugFrame::~DWARFDebugFrame() = default;
 
 static void LLVM_ATTRIBUTE_UNUSED dumpDataAux(DataExtractor Data,
                                               uint32_t Offset, int Length) {
   errs() << "DUMP: ";
   for (int i = 0; i < Length; ++i) {
     uint8_t c = Data.getU8(&Offset);
if (IsCIE) {
                                                   EndAugmentationOffset);
         }
       }
 
       auto Cie = llvm::make_unique<CIE>(
           StartOffset, Length, Version, AugmentationString, AddressSize,
           SegmentDescriptorSize, CodeAlignmentFactor, DataAlignmentFactor,
           ReturnAddressRegister, AugmentationData, FDEPointerEncoding,
-          LSDAPointerEncoding, Personality, PersonalityEncoding);
+          LSDAPointerEncoding, Personality, PersonalityEncoding, Arch);
       CIEs[StartOffset] = Cie.get();
       Entries.emplace_back(std::move(Cie));
     } else {
       // FDE
       uint64_t CIEPointer = Id;
       uint64_t InitialLocation = 0;
       uint64_t AddressRange = 0;
       Optional<uint64_t> LSDAAddress;
if (IsCIE) {
         }
       } else {
         InitialLocation = Data.getAddress(&Offset);
         AddressRange = Data.getAddress(&Offset);
       }
 
       Entries.emplace_back(new FDE(StartOffset, Length, CIEPointer,
                                    InitialLocation, AddressRange,
-                                   Cie, LSDAAddress));
+                                   Cie, LSDAAddress, Arch));
     }
 
     if (Error E =
             Entries.back()->cfis().parse(Data, &Offset, EndStructureOffset)) {
       report_fatal_error(toString(std::move(E)));
     }
 
     if (Offset != EndStructureOffset)

lib/MC/MCAsmStreamer.cpp

public:
   void EmitCFIRelOffset(int64_t Register, int64_t Offset) override;
   void EmitCFIAdjustCfaOffset(int64_t Adjustment) override;
   void EmitCFIEscape(StringRef Values) override;
   void EmitCFIGnuArgsSize(int64_t Size) override;
   void EmitCFISignalFrame() override;
   void EmitCFIUndefined(int64_t Register) override;
   void EmitCFIRegister(int64_t Register1, int64_t Register2) override;
   void EmitCFIWindowSave() override;
+  void EmitCFINegateRAState() override;
   void EmitCFIReturnColumn(int64_t Register) override;
 
   void EmitWinCFIStartProc(const MCSymbol *Symbol, SMLoc Loc) override;
   void EmitWinCFIEndProc(SMLoc Loc) override;
   void EmitWinCFIStartChained(SMLoc Loc) override;
   void EmitWinCFIEndChained(SMLoc Loc) override;
   void EmitWinCFIPushReg(unsigned Register, SMLoc Loc) override;
   void EmitWinCFISetFrame(unsigned Register, unsigned Offset,

 }
 
 void MCAsmStreamer::EmitCFIWindowSave() {
   MCStreamer::EmitCFIWindowSave();
   OS << "\t.cfi_window_save";
   EmitEOL();
 }
 
+void MCAsmStreamer::EmitCFINegateRAState() {
+  MCStreamer::EmitCFINegateRAState();
+  OS << "\t.cfi_negate_ra_state";
+  EmitEOL();
+}
+
 void MCAsmStreamer::EmitCFIReturnColumn(int64_t Register) {
   MCStreamer::EmitCFIReturnColumn(Register);
   OS << "\t.cfi_return_column " << Register;
   EmitEOL();
 }
 
 void MCAsmStreamer::EmitWinCFIStartProc(const MCSymbol *Symbol, SMLoc Loc) {
   MCStreamer::EmitWinCFIStartProc(Symbol, Loc);

lib/MC/MCDwarf.cpp

case MCCFIInstruction::OpRegister: {
     Streamer.EmitULEB128IntValue(Reg1);
     Streamer.EmitULEB128IntValue(Reg2);
     return;
   }
   case MCCFIInstruction::OpWindowSave:
     Streamer.EmitIntValue(dwarf::DW_CFA_GNU_window_save, 1);
     return;
 
+  case MCCFIInstruction::OpNegateRAState:
+    Streamer.EmitIntValue(dwarf::DW_CFA_AARCH64_negate_ra_state, 1);
+    return;
+
   case MCCFIInstruction::OpUndefined: {
     unsigned Reg = Instr.getRegister();
     Streamer.EmitIntValue(dwarf::DW_CFA_undefined, 1);
     Streamer.EmitULEB128IntValue(Reg);
     return;
   }
   case MCCFIInstruction::OpAdjustCfaOffset:
   case MCCFIInstruction::OpDefCfaOffset: {

lib/MC/MCStreamer.cpp

void MCStreamer::EmitCFIWindowSave() {
   MCCFIInstruction Instruction =
     MCCFIInstruction::createWindowSave(Label);
   MCDwarfFrameInfo *CurFrame = getCurrentDwarfFrameInfo();
   if (!CurFrame)
     return;
   CurFrame->Instructions.push_back(Instruction);
 }
 
+void MCStreamer::EmitCFINegateRAState() {
+  MCSymbol *Label = EmitCFILabel();
+  MCCFIInstruction Instruction = MCCFIInstruction::createNegateRAState(Label);
+  MCDwarfFrameInfo *CurFrame = getCurrentDwarfFrameInfo();
+  if (!CurFrame)
+    return;
+  CurFrame->Instructions.push_back(Instruction);
+}
+
 void MCStreamer::EmitCFIReturnColumn(int64_t Register) {
   MCDwarfFrameInfo *CurFrame = getCurrentDwarfFrameInfo();
   if (!CurFrame)
     return;
   CurFrame->RAReg = Register;
 }
 
 WinEH::FrameInfo *MCStreamer::EnsureValidWinFrameInfo(SMLoc Loc) {

lib/Target/AArch64/AArch64FrameLowering.cpp

void AArch64FrameLowering::emitPrologue(MachineFunction &MF,
 
   // Debug location must be unknown since the first debug location is used
   // to determine the end of the prologue.
   DebugLoc DL;
 
   if (ShouldSignReturnAddress(MF)) {
     BuildMI(MBB, MBBI, DL, TII->get(AArch64::PACIASP))
         .setMIFlag(MachineInstr::FrameSetup);
+
+    unsigned CFIIndex =
+        MF.addFrameInst(MCCFIInstruction::createNegateRAState(nullptr));
+    BuildMI(MBB, MBBI, DL, TII->get(TargetOpcode::CFI_INSTRUCTION))
+        .addCFIIndex(CFIIndex)
+        .setMIFlags(MachineInstr::FrameSetup);
   }
 
   // All calls are tail calls in GHC calling conv, and functions have no
   // prologue/epilogue.
   if (MF.getFunction().getCallingConv() == CallingConv::GHC)
     return;
 
   int NumBytes = (int)MFI.getStackSize();
static void InsertReturnAddressAuth(MachineFunction &MF,
   if (Subtarget.hasV8_3aOps() && MBBI != MBB.end() &&
       MBBI->getOpcode() == AArch64::RET_ReallyLR) {
     BuildMI(MBB, MBBI, DL, TII->get(AArch64::RETAA)).copyImplicitOps(*MBBI);
     MBB.erase(MBBI);
   } else {
     BuildMI(MBB, MBBI, DL, TII->get(AArch64::AUTIASP))
         .setMIFlag(MachineInstr::FrameDestroy);
   }
 }
 
 void AArch64FrameLowering::emitEpilogue(MachineFunction &MF,
                                         MachineBasicBlock &MBB) const {
   MachineBasicBlock::iterator MBBI = MBB.getLastNonDebugInstr();
   MachineFrameInfo &MFI = MF.getFrameInfo();
   const AArch64Subtarget &Subtarget = MF.getSubtarget<AArch64Subtarget>();
   const TargetInstrInfo *TII = Subtarget.getInstrInfo();
   DebugLoc DL;

lib/Target/AArch64/AsmParser/AArch64AsmParser.cpp

private:
 
   bool parseDirectiveTLSDescCall(SMLoc L);
 
   bool parseDirectiveLOH(StringRef LOH, SMLoc L);
   bool parseDirectiveLtorg(SMLoc L);
 
   bool parseDirectiveReq(StringRef Name, SMLoc L);
   bool parseDirectiveUnreq(SMLoc L);
+  bool parseDirectiveCFINegateRAState();
 
   bool validateInstruction(MCInst &Inst, SMLoc &IDLoc,
                            SmallVectorImpl<SMLoc> &Loc);
   bool MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode,
                                OperandVector &Operands, MCStreamer &Out,
                                uint64_t &ErrorInfo,
                                bool MatchingInlineAsm) override;
 /// @name Auto-generated Match Functions
bool AArch64AsmParser::ParseDirective(AsmToken DirectiveID) {
   else if (IDVal == ".tlsdesccall")
     parseDirectiveTLSDescCall(Loc);
   else if (IDVal == ".ltorg" || IDVal == ".pool")
     parseDirectiveLtorg(Loc);
   else if (IDVal == ".unreq")
     parseDirectiveUnreq(Loc);
   else if (IDVal == ".inst")
     parseDirectiveInst(Loc);
+  else if (IDVal == ".cfi_negate_ra_state")
+    parseDirectiveCFINegateRAState();
   else if (IsMachO) {
     if (IDVal == MCLOHDirectiveName())
       parseDirectiveLOH(IDVal, Loc);
     else
       return true;
   } else
     return true;
   return false;
if (getTok().isNot(AsmToken::Identifier))
     return TokError("unexpected input in .unreq directive.");
   RegisterReqs.erase(Parser.getTok().getIdentifier().lower());
   Parser.Lex(); // Eat the identifier.
   if (parseToken(AsmToken::EndOfStatement))
     return addErrorSuffix("in '.unreq' directive");
   return false;
 }
 
+bool AArch64AsmParser::parseDirectiveCFINegateRAState() {
+  getStreamer().EmitCFINegateRAState();
+  return false;
+}
+
 bool
 AArch64AsmParser::classifySymbolRef(const MCExpr *Expr,
                                     AArch64MCExpr::VariantKind &ELFRefKind,
                                     MCSymbolRefExpr::VariantKind &DarwinRefKind,
                                     int64_t &Addend) {
   ELFRefKind = AArch64MCExpr::VK_INVALID;
   DarwinRefKind = MCSymbolRefExpr::VK_None;
   Addend = 0;

test/CodeGen/AArch64/sign-return-address-cfi-negate-ra-state.ll

+; RUN: llc -mtriple=aarch64-none-eabi < %s | FileCheck %s
+; RUN: llc -mtriple=aarch64-none-eabi -filetype=obj -o - <%s | llvm-dwarfdump -v - | FileCheck --check-prefix=CHECK-DUMP %s
+
+@.str = private unnamed_addr constant [15 x i8] c"some exception\00", align 1
+@_ZTIPKc = external dso_local constant i8*
+
+; CHECK: @_Z3fooi
+; CHECK: paciasp
+; CHECK-NEXT: .cfi_negate_ra_state
+; CHECK-NOT: .cfi_negate_ra_state
+define dso_local i32 @_Z3fooi(i32 %x) #0 {
+entry:
+  %retval = alloca i32, align 4
+  %x.addr = alloca i32, align 4
+  store i32 %x, i32* %x.addr, align 4
+  %exception = call i8* @__cxa_allocate_exception(i64 8) #1
+  %0 = bitcast i8* %exception to i8**
+  store i8* getelementptr inbounds ([15 x i8], [15 x i8]* @.str, i32 0, i32 0), i8** %0, align 16
+  call void @__cxa_throw(i8* %exception, i8* bitcast (i8** @_ZTIPKc to i8*), i8* null) #2
+  unreachable
+
+return:                                           ; No predecessors!
+  %1 = load i32, i32* %retval, align 4
+  ret i32 %1
+}
+
+declare dso_local i8* @__cxa_allocate_exception(i64)
+
+declare dso_local void @__cxa_throw(i8*, i8*, i8*)
+
+attributes #0 = { "sign-return-address"="all" }
+
+;CHECK-DUMP: DW_CFA_AARCH64_negate_ra_state

test/CodeGen/AArch64/sign-return-address.ll

 ; CHECK-NOT: auti{{[a,b]}}sp
 define i32 @leaf_sign_non_leaf(i32 %x) "sign-return-address"="non-leaf"  {
   ret i32 %x
 }
 
 ; CHECK-LABEL: @leaf_sign_all
 ; CHECK: paciasp
 ; CHECK: autiasp
-; CHECK-NEXT: ret
+; CHECK: ret
 define i32 @leaf_sign_all(i32 %x) "sign-return-address"="all" {
   ret i32 %x
 }
 
 ; CHECK: @leaf_clobbers_lr
 ; CHECK: paciasp
-; CHECK-NEXT: str x30, [sp, #-16]!
+; CHECK: str x30, [sp, #-16]!
 ; CHECK: ldr  x30, [sp], #16
 ; CHECK-NEXT: autiasp
-; CHECK-NEXT: ret
+; CHECK: ret
 define i64 @leaf_clobbers_lr(i64 %x) "sign-return-address"="non-leaf"  {
   call void asm sideeffect "mov x30, $0", "r,~{lr}"(i64 %x) #1
   ret i64 %x
 }
 
 declare i32 @foo(i32)
 
 ; CHECK: @non_leaf_sign_all
 ; CHECK: paciasp
 ; CHECK: autiasp
-; CHECK-NEXT: ret
+; CHECK: ret
 define i32 @non_leaf_sign_all(i32 %x) "sign-return-address"="all" {
   %call = call i32 @foo(i32 %x)
   ret i32 %call
 }
 
 ; CHECK: @non_leaf_sign_non_leaf
 ; CHECK: paciasp
-; CHECK-NEXT: str x30, [sp, #-16]!
+; CHECK: str x30, [sp, #-16]!
 ; CHECK: ldr  x30, [sp], #16
-; CHECK-NEXT: autiasp
-; CHECK-NEXT: ret
+; CHECK: autiasp
+; CHECK: ret
 define i32 @non_leaf_sign_non_leaf(i32 %x) "sign-return-address"="non-leaf"  {
   %call = call i32 @foo(i32 %x)
   ret i32 %call
 }
 
 ; CHECK-LABEL: @leaf_sign_all_v83
 ; CHECK: paciasp
 ; CHECK-NOT: ret
-; CHECK-NEXT: retaa
+; CHECK: retaa
 ; CHECK-NOT: ret
 define i32 @leaf_sign_all_v83(i32 %x) "sign-return-address"="all" "target-features"="+v8.3a" {
   ret i32 %x
 }
 
 declare fastcc i64 @bar(i64)
 
 ; CHECK-LABEL: @spill_lr_and_tail_call
 ; CHECK: paciasp
-; CHECK-NEXT: str x30, [sp, #-16]!
+; CHECK: str x30, [sp, #-16]!
 ; CHECK: ldr  x30, [sp], #16
-; CHECK-NEXT: autiasp
-; CHECK-NEXT: b  bar
+; CHECK: autiasp
+; CHECK: b  bar
 define fastcc void @spill_lr_and_tail_call(i64 %x) "sign-return-address"="all" {
   call void asm sideeffect "mov x30, $0", "r,~{lr}"(i64 %x) #1
   tail call fastcc i64 @bar(i64 %x)
   ret void
 }

test/CodeGen/MIR/AArch64/cfi.mir

bb.0.entry:
     CFI_INSTRUCTION remember_state
     ; CHECK: CFI_INSTRUCTION remember_state
     CFI_INSTRUCTION restore_state
     ; CHECK: CFI_INSTRUCTION restore_state
     CFI_INSTRUCTION escape 0x61, 0x62, 0x63
     ; CHECK: CFI_INSTRUCTION escape 0x61, 0x62, 0x63
     CFI_INSTRUCTION window_save
     ; CHECK: CFI_INSTRUCTION window_save
+    CFI_INSTRUCTION negate_ra_sign_state
+    ; CHECK: CFI_INSTRUCTION negate_ra_sign_state
     RET_ReallyLR

test/CodeGen/MIR/AArch64/return-address-signing.mir

+# RUN: llc -mtriple=aarch64-arm-none-eabi -run-pass=prologepilog -o - %s 2>&1 | FileCheck %s
+--- |
+  ; ModuleID = '/work/scratch/test.ll'
+  source_filename = "/work/scratch/test.c"
+  target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+  target triple = "aarch64-arm-none-eabi"
+  
+  ; Function Attrs: noinline nounwind optnone
+  define dso_local i32 @foo() #0 {
+  entry:
+    ret i32 2
+  }
+  
+  attributes #0 = { noinline nounwind optnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "sign-return-address"="all" "sign-return-address-key"="a_key" "stack-protector-buffer-size"="8" "target-cpu"="generic" "target-features"="+neon,+v8.3a" "unsafe-fp-math"="false" "use-soft-float"="false" }
+  
+  !llvm.module.flags = !{!0}
+  !llvm.ident = !{!1}
+  
+  !0 = !{i32 1, !"wchar_size", i32 4}
+  !1 = !{!"clang version 8.0.0 (https://git.llvm.org/git/clang.git/ f9833ff4fea02c1a3d9ce1d8432be3af9ffc7de4) (https://git.llvm.org/git/llvm.git/ 176f75d48ef7f2d9b3234b5e067e9410cd6cb800)"}
+
+...
+---
+#CHECK: foo
+name:            foo
+alignment:       2
+exposesReturnsTwice: false
+legalized:       false
+regBankSelected: false
+selected:        false
+failedISel:      false
+tracksRegLiveness: true
+registers:       
+liveins:         
+frameInfo:       
+  isFrameAddressTaken: false
+  isReturnAddressTaken: false
+  hasStackMap:     false
+  hasPatchPoint:   false
+  stackSize:       0
+  offsetAdjustment: 0
+  maxAlignment:    0
+  adjustsStack:    false
+  hasCalls:        false
+  stackProtector:  ''
+  maxCallFrameSize: 0
+  hasOpaqueSPAdjustment: false
+  hasVAStart:      false
+  hasMustTailInVarArgFunc: false
+  localFrameSize:  0
+  savePoint:       ''
+  restorePoint:    ''
+fixedStack:      
+stack:           
+constants:       
+#CHECK:    frame-setup PACIASP implicit-def $lr, implicit $lr, implicit $sp
+#CHECK:    frame-setup CFI_INSTRUCTION negate_ra_sign_state 
+body:             |
+  bb.0.entry:
+    $w0 = MOVi32imm 2
+    RET_ReallyLR implicit killed $w0
+
+...

tools/llvm-dwarfdump/llvm-dwarfdump.cpp

static bool lookup(DWARFContext &DICtx, uint64_t Address, raw_ostream &OS) {
   return true;
 }
 
 bool collectStatsForObjectFile(ObjectFile &Obj, DWARFContext &DICtx,
                                Twine Filename, raw_ostream &OS);
 
 static bool dumpObjectFile(ObjectFile &Obj, DWARFContext &DICtx, Twine Filename,
                            raw_ostream &OS) {
-  logAllUnhandledErrors(DICtx.loadRegisterInfo(Obj), errs(),
+  logAllUnhandledErrors(DICtx.loadArchitectureInfo(Obj), errs(),
                         Filename.str() + ": ");
   // The UUID dump already contains all the same information.
   if (!(DumpType & DIDT_UUID) || DumpType == DIDT_All)
     OS << Filename << ":\tfile format " << Obj.getFileFormatName() << '\n';
 
   // Handle the --lookup option.
   if (Lookup)
     return lookup(DICtx, Lookup, OS);

tools/llvm-readobj/DwarfCFIEHPrinter.h

 #define LLVM_TOOLS_LLVM_READOBJ_DWARFCFIEHPRINTER_H
 
 #include "Error.h"
 #include "llvm-readobj.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/BinaryFormat/Dwarf.h"
 #include "llvm/Object/ELF.h"
 #include "llvm/Object/ELFTypes.h"
+#include "llvm/Object/ELFObjectFile.h"
 #include "llvm/Support/Casting.h"
 #include "llvm/Support/ScopedPrinter.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/DebugInfo/DWARF/DWARFDataExtractor.h"
 #include "llvm/DebugInfo/DWARF/DWARFDebugFrame.h"
 #include "llvm/Support/Endian.h"
 #include "llvm/Support/Format.h"
 #include "llvm/Support/type_traits.h"
 
 namespace llvm {
 namespace DwarfCFIEH {
 
 template <typename ELFT>
 class PrinterContext {
   ScopedPrinter &W;
-  const object::ELFFile<ELFT> *Obj;
+  const object::ELFObjectFile<ELFT> *ObjF;
 
   void printEHFrameHdr(uint64_t Offset, uint64_t Address, uint64_t Size) const;
 
   void printEHFrame(const typename ELFT::Shdr *EHFrameShdr) const;
 
 public:
-  PrinterContext(ScopedPrinter &W, const object::ELFFile<ELFT> *Obj)
-      : W(W), Obj(Obj) {}
+  PrinterContext(ScopedPrinter &W, const object::ELFObjectFile<ELFT> *ObjF)
+      : W(W), ObjF(ObjF) {}
 
   void printUnwindInformation() const;
 };
 
 template <class ELFO>
 static const typename ELFO::Elf_Shdr *findSectionByAddress(const ELFO *Obj,
                                                            uint64_t Addr) {
   auto Sections = Obj->sections();
   if (Error E = Sections.takeError())
     reportError(toString(std::move(E)));
 
   for (const auto &Shdr : *Sections)
     if (Shdr.sh_addr == Addr)
       return &Shdr;
   return nullptr;
 }
 
 template <typename ELFT>
 void PrinterContext<ELFT>::printUnwindInformation() const {
+  const object::ELFFile<ELFT> *Obj = ObjF->getELFFile();
   const typename ELFT::Phdr *EHFramePhdr = nullptr;
 
   auto PHs = Obj->program_headers();
   if (Error E = PHs.takeError())
     reportError(toString(std::move(E)));
 
   for (const auto &Phdr : *PHs) {
     if (Phdr.p_type == ELF::PT_GNU_EH_FRAME) {

 void PrinterContext<ELFT>::printEHFrameHdr(uint64_t EHFrameHdrOffset,
                                            uint64_t EHFrameHdrAddress,
                                            uint64_t EHFrameHdrSize) const {
   ListScope L(W, "EH_FRAME Header");
   W.startLine() << format("Address: 0x%" PRIx64 "\n", EHFrameHdrAddress);
   W.startLine() << format("Offset: 0x%" PRIx64 "\n", EHFrameHdrOffset);
   W.startLine() << format("Size: 0x%" PRIx64 "\n", EHFrameHdrSize);
 
+  const object::ELFFile<ELFT> *Obj = ObjF->getELFFile();
   const auto *EHFrameHdrShdr = findSectionByAddress(Obj, EHFrameHdrAddress);
   if (EHFrameHdrShdr) {
     auto SectionName = Obj->getSectionName(EHFrameHdrShdr);
     if (Error E = SectionName.takeError())
       reportError(toString(std::move(E)));
 
     W.printString("Corresponding Section", *SectionName);
   }
void PrinterContext<ELFT>::printEHFrame(
     const typename ELFT::Shdr *EHFrameShdr) const {
   uint64_t Address = EHFrameShdr->sh_addr;
   uint64_t ShOffset = EHFrameShdr->sh_offset;
   W.startLine() << format(".eh_frame section at offset 0x%" PRIx64
                           " address 0x%" PRIx64 ":\n",
                           ShOffset, Address);
   W.indent();
 
+  const object::ELFFile<ELFT> *Obj = ObjF->getELFFile();
   auto Result = Obj->getSectionContents(EHFrameShdr);
   if (Error E = Result.takeError())
     reportError(toString(std::move(E)));
 
   auto Contents = Result.get();
   DWARFDataExtractor DE(
       StringRef(reinterpret_cast<const char *>(Contents.data()),
                 Contents.size()),
       ELFT::TargetEndianness == support::endianness::little,
       ELFT::Is64Bits ? 8 : 4);
-  DWARFDebugFrame EHFrame(/*IsEH=*/true, /*EHFrameAddress=*/Address);
+  DWARFDebugFrame EHFrame(Triple::ArchType(ObjF->getArch()), /*IsEH=*/true, /*EHFrameAddress=*/Address);
   EHFrame.parse(DE);
 
   for (const auto &Entry : EHFrame) {
     if (const auto *CIE = dyn_cast<dwarf::CIE>(&Entry)) {
       W.startLine() << format("[0x%" PRIx64 "] CIE length=%" PRIu64 "\n",
                               Address + CIE->getOffset(),
                               CIE->getLength());
       W.indent();

tools/llvm-readobj/ELFDumper.cpp

if (EntSize != sizeof(Type) || Size % EntSize)
       reportError("Invalid entity size");
     return {Start, Start + (Size / EntSize)};
   }
 };
 
 template<typename ELFT>
 class ELFDumper : public ObjDumper {
 public:
-  ELFDumper(const ELFFile<ELFT> *Obj, ScopedPrinter &Writer);
+  ELFDumper(const object::ELFObjectFile<ELFT> *ObjF, ScopedPrinter &Writer);
 
   void printFileHeaders() override;
   void printSections() override;
   void printRelocations() override;
   void printDynamicRelocations() override;
   void printSymbols() override;
   void printDynamicSymbols() override;
   void printUnwindInfo() override;
public:
   void printELFLinkerOptions() override;
 
 private:
   std::unique_ptr<DumpStyle<ELFT>> ELFDumperStyle;
 
   TYPEDEF_ELF_TYPES(ELFT)
 
   DynRegionInfo checkDRI(DynRegionInfo DRI) {
+    const ELFFile<ELFT> *Obj = ObjF->getELFFile();
     if (DRI.Addr < Obj->base() ||
         (const uint8_t *)DRI.Addr + DRI.Size > Obj->base() + Obj->getBufSize())
       error(llvm::object::object_error::parse_failed);
     return DRI;
   }
 
   DynRegionInfo createDRIFrom(const Elf_Phdr *P, uintX_t EntSize) {
-    return checkDRI({Obj->base() + P->p_offset, P->p_filesz, EntSize});
+    return checkDRI({ObjF->getELFFile()->base() + P->p_offset, P->p_filesz, EntSize});
   }
 
   DynRegionInfo createDRIFrom(const Elf_Shdr *S) {
-    return checkDRI({Obj->base() + S->sh_offset, S->sh_size, S->sh_entsize});
+    return checkDRI({ObjF->getELFFile()->base() + S->sh_offset, S->sh_size, S->sh_entsize});
   }
 
   void parseDynamicTable(ArrayRef<const Elf_Phdr *> LoadSegments);
 
   void printValue(uint64_t Type, uint64_t Value);
 
   StringRef getDynamicString(uint64_t Offset) const;
   StringRef getSymbolVersion(StringRef StrTab, const Elf_Sym *symb,
                              bool &IsDefault) const;
   void LoadVersionMap() const;
   void LoadVersionNeeds(const Elf_Shdr *ec) const;
   void LoadVersionDefs(const Elf_Shdr *sec) const;
 
-  const ELFO *Obj;
+  const object::ELFObjectFile<ELFT> *ObjF;
   DynRegionInfo DynRelRegion;
   DynRegionInfo DynRelaRegion;
   DynRegionInfo DynRelrRegion;
   DynRegionInfo DynPLTRelRegion;
   DynRegionInfo DynSymRegion;
   DynRegionInfo DynamicTable;
   StringRef DynamicStringTable;
   StringRef SOName;
public:
   const Elf_GnuHash *getGnuHashTable() const { return GnuHashTable; }
 };
 
 template <class ELFT>
 void ELFDumper<ELFT>::printSymbolsHelper(bool IsDynamic) const {
   StringRef StrTable, SymtabName;
   size_t Entries = 0;
   Elf_Sym_Range Syms(nullptr, nullptr);
+  const ELFFile<ELFT> *Obj = ObjF->getELFFile();
   if (IsDynamic) {
     StrTable = DynamicStringTable;
     Syms = dynamic_symbols();
     SymtabName = DynSymtabName;
     if (DynSymRegion.Addr)
       Entries = DynSymRegion.Size / DynSymRegion.EntSize;
   } else {
     if (!DotSymtabSec)
private:
   ScopedPrinter &W;
 };
 
 } // end anonymous namespace
 
 namespace llvm {
 
 template <class ELFT>
-static std::error_code createELFDumper(const ELFFile<ELFT> *Obj,
+static std::error_code createELFDumper(const ELFObjectFile<ELFT> *Obj,
                                        ScopedPrinter &Writer,
                                        std::unique_ptr<ObjDumper> &Result) {
   Result.reset(new ELFDumper<ELFT>(Obj, Writer));
   return readobj_error::success;
 }
 
 std::error_code createELFDumper(const object::ObjectFile *Obj,
                                 ScopedPrinter &Writer,
                                 std::unique_ptr<ObjDumper> &Result) {
   // Little-endian 32-bit
   if (const ELF32LEObjectFile *ELFObj = dyn_cast<ELF32LEObjectFile>(Obj))
-    return createELFDumper(ELFObj->getELFFile(), Writer, Result);
+    return createELFDumper(ELFObj, Writer, Result);
 
   // Big-endian 32-bit
   if (const ELF32BEObjectFile *ELFObj = dyn_cast<ELF32BEObjectFile>(Obj))
-    return createELFDumper(ELFObj->getELFFile(), Writer, Result);
+    return createELFDumper(ELFObj, Writer, Result);
 
   // Little-endian 64-bit
   if (const ELF64LEObjectFile *ELFObj = dyn_cast<ELF64LEObjectFile>(Obj))
-    return createELFDumper(ELFObj->getELFFile(), Writer, Result);
+    return createELFDumper(ELFObj, Writer, Result);
 
   // Big-endian 64-bit
   if (const ELF64BEObjectFile *ELFObj = dyn_cast<ELF64BEObjectFile>(Obj))
-    return createELFDumper(ELFObj->getELFFile(), Writer, Result);
+    return createELFDumper(ELFObj, Writer, Result);
 
   return readobj_error::unsupported_obj_file_format;
 }
 
 } // end namespace llvm
 
 // Iterate through the versions needed section, and place each Elf_Vernaux
 // in the VersionMap according to its index.
 template <class ELFT>
 void ELFDumper<ELFT>::LoadVersionNeeds(const Elf_Shdr *sec) const {
   unsigned vn_size = sec->sh_size;  // Size of section in bytes
   unsigned vn_count = sec->sh_info; // Number of Verneed entries
-  const char *sec_start = (const char *)Obj->base() + sec->sh_offset;
+  const char *sec_start = (const char *)ObjF->getELFFile()->base() + sec->sh_offset;
   const char *sec_end = sec_start + vn_size;
   // The first Verneed entry is at the start of the section.
   const char *p = sec_start;
   for (unsigned i = 0; i < vn_count; i++) {
     if (p + sizeof(Elf_Verneed) > sec_end)
       report_fatal_error("Section ended unexpectedly while scanning "
                          "version needed records.");
     const Elf_Verneed *vn = reinterpret_cast<const Elf_Verneed *>(p);

 }
 
 // Iterate through the version definitions, and place each Elf_Verdef
 // in the VersionMap according to its index.
 template <class ELFT>
 void ELFDumper<ELFT>::LoadVersionDefs(const Elf_Shdr *sec) const {
   unsigned vd_size = sec->sh_size;  // Size of section in bytes
   unsigned vd_count = sec->sh_info; // Number of Verdef entries
-  const char *sec_start = (const char *)Obj->base() + sec->sh_offset;
+  const char *sec_start = (const char *)ObjF->getELFFile()->base() + sec->sh_offset;
   const char *sec_end = sec_start + vd_size;
   // The first Verdef entry is at the start of the section.
   const char *p = sec_start;
   for (unsigned i = 0; i < vd_count; i++) {
     if (p + sizeof(Elf_Verdef) > sec_end)
       report_fatal_error("Section ended unexpectedly while scanning "
                          "version definitions.");
     const Elf_Verdef *vd = reinterpret_cast<const Elf_Verdef *>(p);
for (unsigned J = 0; J < Need->vn_cnt; ++J) {
       PAux += Aux->vna_next;
     }
     P += Need->vn_next;
   }
 }
 
 template <typename ELFT> void ELFDumper<ELFT>::printVersionInfo() {
   // Dump version symbol section.
-  printVersionSymbolSection(this, Obj, dot_gnu_version_sec, W);
+  printVersionSymbolSection(this, ObjF->getELFFile(), dot_gnu_version_sec, W);
 
   // Dump version definition section.
-  printVersionDefinitionSection(this, Obj, dot_gnu_version_d_sec, W);
+  printVersionDefinitionSection(this, ObjF->getELFFile(), dot_gnu_version_d_sec, W);
 
   // Dump version dependency section.
-  printVersionDependencySection(this, Obj, dot_gnu_version_r_sec, W);
+  printVersionDependencySection(this, ObjF->getELFFile(), dot_gnu_version_r_sec, W);
 }
 
 template <typename ELFT>
 StringRef ELFDumper<ELFT>::getSymbolVersion(StringRef StrTab,
                                             const Elf_Sym *symb,
                                             bool &IsDefault) const {
   // This is a dynamic symbol. Look in the GNU symbol version table.
   if (!dot_gnu_version_sec) {
     // No version table.
     IsDefault = false;
     return StringRef("");
   }
 
   // Determine the position in the symbol table of this entry.
   size_t entry_index = (reinterpret_cast<uintptr_t>(symb) -
                         reinterpret_cast<uintptr_t>(DynSymRegion.Addr)) /
                        sizeof(Elf_Sym);
 
   // Get the corresponding version index entry
   const Elf_Versym *vs = unwrapOrError(
-      Obj->template getEntry<Elf_Versym>(dot_gnu_version_sec, entry_index));
+      ObjF->getELFFile()->template getEntry<Elf_Versym>(dot_gnu_version_sec, entry_index));
   size_t version_index = vs->vs_index & ELF::VERSYM_VERSION;
 
   // Special markers for unversioned symbols.
   if (version_index == ELF::VER_NDX_LOCAL ||
       version_index == ELF::VER_NDX_GLOBAL) {
     IsDefault = false;
     return StringRef("");
   }
StringRef ELFDumper<ELFT>::getSymbolVersion(StringRef StrTab,
   }
   if (name_offset >= StrTab.size())
     reportError("Invalid string offset");
   return StringRef(StrTab.data() + name_offset);
 }
 
 template <typename ELFT>
 StringRef ELFDumper<ELFT>::getStaticSymbolName(uint32_t Index) const {
+  const ELFFile<ELFT> *Obj = ObjF->getELFFile();
   StringRef StrTable = unwrapOrError(Obj->getStringTableForSymtab(*DotSymtabSec));
   Elf_Sym_Range Syms = unwrapOrError(Obj->symbols(DotSymtabSec));
   if (Index >= Syms.size())
     reportError("Invalid symbol index");
   const Elf_Sym *Sym = &Syms[Index];
   return unwrapOrError(Sym->getName(StrTable));
 }
 
void ELFDumper<ELFT>::getSectionNameIndex(const Elf_Sym *Symbol,
   else if (Symbol->isCommon())
     SectionName = "Common";
   else if (Symbol->isReserved() && SectionIndex != SHN_XINDEX)
     SectionName = "Reserved";
   else {
     if (SectionIndex == SHN_XINDEX)
       SectionIndex = unwrapOrError(object::getExtendedSymbolTableIndex<ELFT>(
           Symbol, FirstSym, ShndxTable));
+    const ELFFile<ELFT> *Obj = ObjF->getELFFile();
     const typename ELFT::Shdr *Sec =
         unwrapOrError(Obj->getSection(SectionIndex));
     SectionName = unwrapOrError(Obj->getSectionName(Sec));
   }
 }
 
 template <class ELFO>
 static const typename ELFO::Elf_Shdr *
static const char *getElfMipsOptionsOdkType(unsigned Odk) {
   LLVM_READOBJ_ENUM_CASE(ELF, ODK_IDENT);
   LLVM_READOBJ_ENUM_CASE(ELF, ODK_PAGESIZE);
   default:
     return "Unknown";
   }
 }
 
 template <typename ELFT>
-ELFDumper<ELFT>::ELFDumper(const ELFFile<ELFT> *Obj, ScopedPrinter &Writer)
-    : ObjDumper(Writer), Obj(Obj) {
+ELFDumper<ELFT>::ELFDumper(const object::ELFObjectFile<ELFT> *ObjF,
+    ScopedPrinter &Writer)
+    : ObjDumper(Writer), ObjF(ObjF) {
   SmallVector<const Elf_Phdr *, 4> LoadSegments;
+  const ELFFile<ELFT> *Obj = ObjF->getELFFile();
   for (const Elf_Phdr &Phdr : unwrapOrError(Obj->program_headers())) {
     if (Phdr.p_type == ELF::PT_DYNAMIC) {
       DynamicTable = createDRIFrom(&Phdr, sizeof(Elf_Dyn));
       continue;
     }
     if (Phdr.p_type != ELF::PT_LOAD || Phdr.p_filesz == 0)
       continue;
     LoadSegments.push_back(&Phdr);
const Elf_Phdr *const *I =
                          });
     if (I == LoadSegments.begin())
       report_fatal_error("Virtual address is not in any segment");
     --I;
     const Elf_Phdr &Phdr = **I;
     uint64_t Delta = VAddr - Phdr.p_vaddr;
     if (Delta >= Phdr.p_filesz)
       report_fatal_error("Virtual address is not in any segment");
-    return Obj->base() + Phdr.p_offset + Delta;
+    return ObjF->getELFFile()->base() + Phdr.p_offset + Delta;
   };
 
   uint64_t SONameOffset = 0;
   const char *StringTableBegin = nullptr;
   uint64_t StringTableSize = 0;
   for (const Elf_Dyn &Dyn : dynamic_table()) {
     switch (Dyn.d_tag) {
     case ELF::DT_HASH:

 
 template <typename ELFT>
 typename ELFDumper<ELFT>::Elf_Relr_Range ELFDumper<ELFT>::dyn_relrs() const {
   return DynRelrRegion.getAsArrayRef<Elf_Relr>();
 }
 
 template<class ELFT>
 void ELFDumper<ELFT>::printFileHeaders() {
-  ELFDumperStyle->printFileHeaders(Obj);
+  ELFDumperStyle->printFileHeaders(ObjF->getELFFile());
 }
 
 template<class ELFT>
 void ELFDumper<ELFT>::printSections() {
-  ELFDumperStyle->printSections(Obj);
+  ELFDumperStyle->printSections(ObjF->getELFFile());
 }
 
 template<class ELFT>
 void ELFDumper<ELFT>::printRelocations() {
-  ELFDumperStyle->printRelocations(Obj);
+  ELFDumperStyle->printRelocations(ObjF->getELFFile());
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printProgramHeaders() {
-  ELFDumperStyle->printProgramHeaders(Obj);
+  ELFDumperStyle->printProgramHeaders(ObjF->getELFFile());
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printDynamicRelocations() {
-  ELFDumperStyle->printDynamicRelocations(Obj);
+  ELFDumperStyle->printDynamicRelocations(ObjF->getELFFile());
 }
 
 template<class ELFT>
 void ELFDumper<ELFT>::printSymbols() {
-  ELFDumperStyle->printSymbols(Obj);
+  ELFDumperStyle->printSymbols(ObjF->getELFFile());
 }
 
 template<class ELFT>
 void ELFDumper<ELFT>::printDynamicSymbols() {
-  ELFDumperStyle->printDynamicSymbols(Obj);
+  ELFDumperStyle->printDynamicSymbols(ObjF->getELFFile());
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printHashHistogram() {
-  ELFDumperStyle->printHashHistogram(Obj);
+  ELFDumperStyle->printHashHistogram(ObjF->getELFFile());
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printCGProfile() {
-  ELFDumperStyle->printCGProfile(Obj);
+  ELFDumperStyle->printCGProfile(ObjF->getELFFile());
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printNotes() {
-  ELFDumperStyle->printNotes(Obj);
+  ELFDumperStyle->printNotes(ObjF->getELFFile());
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printELFLinkerOptions() {
-  ELFDumperStyle->printELFLinkerOptions(Obj);
+  ELFDumperStyle->printELFLinkerOptions(ObjF->getELFFile());
 }
 
 static const char *getTypeString(unsigned Arch, uint64_t Type) {
 #define DYNAMIC_TAG(n, v)
   switch (Arch) {
   case EM_HEXAGON:
     switch (Type) {
 #define HEXAGON_DYNAMIC_TAG(name, value)                                       \
void ELFDumper<ELFT>::printValue(uint64_t Type, uint64_t Value) {
   default:
     OS << format(ConvChar, Value);
     break;
   }
 }
 
 template<class ELFT>
 void ELFDumper<ELFT>::printUnwindInfo() {
-  const unsigned Machine = Obj->getHeader()->e_machine;
+  const unsigned Machine = ObjF->getELFFile()->getHeader()->e_machine;
   if (Machine == EM_386 || Machine == EM_X86_64) {
-    DwarfCFIEH::PrinterContext<ELFT> Ctx(W, Obj);
+    DwarfCFIEH::PrinterContext<ELFT> Ctx(W, ObjF);
     return Ctx.printUnwindInformation();
   }
   W.startLine() << "UnwindInfo not implemented.\n";
 }
 
 namespace {
 
 template <> void ELFDumper<ELF32LE>::printUnwindInfo() {
+  const ELFFile<ELF32LE> *Obj = ObjF->getELFFile();
   const unsigned Machine = Obj->getHeader()->e_machine;
   if (Machine == EM_ARM) {
     ARM::EHABI::PrinterContext<ELF32LE> Ctx(W, Obj, DotSymtabSec);
     return Ctx.PrintUnwindInformation();
   }
   W.startLine() << "UnwindInfo not implemented.\n";
 }
 
void ELFDumper<ELFT>::printDynamicTable() {
   W.startLine()
      << "  Tag" << (Is64 ? "                " : "        ") << "Type"
      << "                 " << "Name/Value\n";
   while (I != E) {
     const Elf_Dyn &Entry = *I;
     uintX_t Tag = Entry.getTag();
     ++I;
     W.startLine() << "  " << format_hex(Tag, Is64 ? 18 : 10, opts::Output != opts::GNU) << " "
-                  << format("%-21s", getTypeString(Obj->getHeader()->e_machine, Tag));
+                  << format("%-21s", getTypeString(ObjF->getELFFile()->getHeader()->e_machine, Tag));
     printValue(Tag, Entry.getVal());
     OS << "\n";
   }
 
   W.startLine() << "]\n";
 }
 
 template<class ELFT>

 template <class ELFT>
 void ELFDumper<ELFT>::printAttributes() {
   W.startLine() << "Attributes not implemented.\n";
 }
 
 namespace {
 
 template <> void ELFDumper<ELF32LE>::printAttributes() {
+  const ELFFile<ELF32LE> *Obj = ObjF->getELFFile();
   if (Obj->getHeader()->e_machine != EM_ARM) {
     W.startLine() << "Attributes not implemented.\n";
     return;
   }
 
   DictScope BA(W, "BuildAttributes");
   for (const ELFO::Elf_Shdr &Sec : unwrapOrError(Obj->sections())) {
     if (Sec.sh_type != ELF::SHT_ARM_ATTRIBUTES)
if (PltRelSec->sh_type == ELF::SHT_REL) {
     return unwrapOrError(Obj->getRelocationSymbol(&Rels[Offset], PltSymTable));
   } else {
     Elf_Rela_Range Rels = unwrapOrError(Obj->relas(PltRelSec));
     return unwrapOrError(Obj->getRelocationSymbol(&Rels[Offset], PltSymTable));
   }
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printMipsPLTGOT() {
+  const ELFFile<ELFT> *Obj = ObjF->getELFFile();
   if (Obj->getHeader()->e_machine != EM_MIPS)
     reportError("MIPS PLT GOT is available for MIPS targets only");
 
   MipsGOTParser<ELFT> Parser(Obj, dynamic_table(), dynamic_symbols());
   if (Parser.hasGot())
     ELFDumperStyle->printMipsGOT(Parser);
   if (Parser.hasPlt())
     ELFDumperStyle->printMipsPLT(Parser);
static int getMipsRegisterSize(uint8_t Flag) {
   case Mips::AFL_REG_128:
     return 128;
   default:
     return -1;
   }
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printMipsABIFlags() {
+  const ELFFile<ELFT> *Obj = ObjF->getELFFile();
   const Elf_Shdr *Shdr = findSectionByName(*Obj, ".MIPS.abiflags");
   if (!Shdr) {
     W.startLine() << "There is no .MIPS.abiflags section in the file.\n";
     return;
   }
   ArrayRef<uint8_t> Sec = unwrapOrError(Obj->getSectionContents(Shdr));
   if (Sec.size() != sizeof(Elf_Mips_ABIFlags<ELFT>)) {
     W.startLine() << "The .MIPS.abiflags section has a wrong size.\n";
static void printMipsReginfoData(ScopedPrinter &W,
   W.printHex("General Mask", Reginfo.ri_gprmask);
   W.printHex("Co-Proc Mask0", Reginfo.ri_cprmask[0]);
   W.printHex("Co-Proc Mask1", Reginfo.ri_cprmask[1]);
   W.printHex("Co-Proc Mask2", Reginfo.ri_cprmask[2]);
   W.printHex("Co-Proc Mask3", Reginfo.ri_cprmask[3]);
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printMipsReginfo() {
+  const ELFFile<ELFT> *Obj = ObjF->getELFFile();
   const Elf_Shdr *Shdr = findSectionByName(*Obj, ".reginfo");
   if (!Shdr) {
     W.startLine() << "There is no .reginfo section in the file.\n";
     return;
   }
   ArrayRef<uint8_t> Sec = unwrapOrError(Obj->getSectionContents(Shdr));
   if (Sec.size() != sizeof(Elf_Mips_RegInfo<ELFT>)) {
     W.startLine() << "The .reginfo section has a wrong size.\n";
     return;
   }
 
   DictScope GS(W, "MIPS RegInfo");
   auto *Reginfo = reinterpret_cast<const Elf_Mips_RegInfo<ELFT> *>(Sec.data());
   printMipsReginfoData(W, *Reginfo);
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printMipsOptions() {
+  const ELFFile<ELFT> *Obj = ObjF->getELFFile();
   const Elf_Shdr *Shdr = findSectionByName(*Obj, ".MIPS.options");
   if (!Shdr) {
     W.startLine() << "There is no .MIPS.options section in the file.\n";
     return;
   }
 
   DictScope GS(W, "MIPS Options");
 
default:
       W.startLine() << "Unsupported MIPS options tag.\n";
       break;
     }
     Sec = Sec.slice(O->size);
   }
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printStackMap() const {
+  const ELFFile<ELFT> *Obj = ObjF->getELFFile();
   const Elf_Shdr *StackMapSection = nullptr;
   for (const auto &Sec : unwrapOrError(Obj->sections())) {
     StringRef Name = unwrapOrError(Obj->getSectionName(&Sec));
     if (Name == ".llvm_stackmaps") {
       StackMapSection = &Sec;
       break;
     }
   }
 
   if (!StackMapSection)
     return;
 
   ArrayRef<uint8_t> StackMapContentsArray =
       unwrapOrError(Obj->getSectionContents(StackMapSection));
 
   prettyPrintStackMap(
       W, StackMapV2Parser<ELFT::TargetEndianness>(StackMapContentsArray));
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printGroupSections() {
-  ELFDumperStyle->printGroupSections(Obj);
+  ELFDumperStyle->printGroupSections(ObjF->getELFFile());
 }
 
 template <class ELFT> void ELFDumper<ELFT>::printAddrsig() {
-  ELFDumperStyle->printAddrsig(Obj);
+  ELFDumperStyle->printAddrsig(ObjF->getELFFile());
 }
 
 static inline void printFields(formatted_raw_ostream &OS, StringRef Str1,
                                StringRef Str2) {
   OS.PadToColumn(2u);
   OS << Str1;
   OS.PadToColumn(37u);
   OS << Str2 << "\n";