Write AArch64 big endian data fixup entries as BE.

[keith.walker.arm]

Summary

There was support for writing the AArch64 big endian data fixup entries in
the .eh_frame section in BE. This is changed to write all such fixup
entries in BE with no restriction on the section. This is similar to
the existing support for fixup entries for ARM.

A test is added to check the length field in the .debug_line section as
this is an example of where such a fixup occurs.

[jmolloy]

Hi Keith,

I've got some style comments, but apart from that I think this looks sensible.

James

lib/Target/AArch64/MCTargetDesc/AArch64AsmBackend.cpp

35

This line is too long.

86

I'd prefer "getFixupKindContainerSizeInBytes", as the "InBytes" suffix is similar to what we have elsewhere in LLVM and it's clearer.

[keith.walker.arm]

Fixed line length issue.
Renamed getFixupKindContainereSizeBytes() to getFixupKindContainereSizeInBytes()

[• rafael]

This fixes PR23020?

[keith.walker.arm]

In D16064#327293, @rafael wrote:

This fixes PR23020?

Unfortunately not. This issue fixes processing fixups in sections other than .eh_frame (such as the .debug_line section). PR23020 is about actually creating required fixup entries in .eh_frame.

lib/Target/AArch64/MCTargetDesc/AArch64AsmBackend.cpp

 #include "llvm/Support/MachO.h"
 using namespace llvm;
 
 namespace {
 
 class AArch64AsmBackend : public MCAsmBackend {
   static const unsigned PCRelFlagVal =
       MCFixupKindInfo::FKF_IsAlignedDownTo32Bits | MCFixupKindInfo::FKF_IsPCRel;
+public:
+  bool IsLittleEndian;
 
 public:
-  AArch64AsmBackend(const Target &T) : MCAsmBackend() {}
+  AArch64AsmBackend(const Target &T, bool IsLittleEndian) : MCAsmBackend(), IsLittleEndian(IsLittleEndian) {}
 
   unsigned getNumFixupKinds() const override {
     return AArch64::NumTargetFixupKinds;
   }
 
   const MCFixupKindInfo &getFixupKindInfo(MCFixupKind Kind) const override {
     const static MCFixupKindInfo Infos[AArch64::NumTargetFixupKinds] = {
       // This table *must* be in the order that the fixup_* kinds are defined in
bool fixupNeedsRelaxation(const MCFixup &Fixup, uint64_t Value,
                             const MCRelaxableFragment *DF,
                             const MCAsmLayout &Layout) const override;
   void relaxInstruction(const MCInst &Inst, MCInst &Res) const override;
   bool writeNopData(uint64_t Count, MCObjectWriter *OW) const override;
 
   void HandleAssemblerFlag(MCAssemblerFlag Flag) {}
 
   unsigned getPointerSize() const { return 8; }
+
+  unsigned getFixupKindContainerSizeBytes(unsigned Kind) const;
 };
 
 } // end anonymous namespace
 
 /// \brief The number of bytes the fixup may change.
 static unsigned getFixupKindNumBytes(unsigned Kind) {
   switch (Kind) {
   default:
static uint64_t adjustFixupValue(unsigned Kind, uint64_t Value) {
   case FK_Data_1:
   case FK_Data_2:
   case FK_Data_4:
   case FK_Data_8:
     return Value;
   }
 }
 
+/// getFixupKindContainerSizeBytes - The number of bytes of the
+/// container involved in big endian or 0 if the item is little endian
+unsigned AArch64AsmBackend::getFixupKindContainerSizeBytes(unsigned Kind) const {
+  if (IsLittleEndian)
+    return 0;
+
+  switch (Kind) {
+  default:
+    llvm_unreachable("Unknown fixup kind!");
+
+  case FK_Data_1:
+    return 1;
+  case FK_Data_2:
+    return 2;
+  case FK_Data_4:
+    return 4;
+  case FK_Data_8:
+    return 8;
+
+  case AArch64::fixup_aarch64_tlsdesc_call:
+  case AArch64::fixup_aarch64_movw:
+  case AArch64::fixup_aarch64_pcrel_branch14:
+  case AArch64::fixup_aarch64_add_imm12:
+  case AArch64::fixup_aarch64_ldst_imm12_scale1:
+  case AArch64::fixup_aarch64_ldst_imm12_scale2:
+  case AArch64::fixup_aarch64_ldst_imm12_scale4:
+  case AArch64::fixup_aarch64_ldst_imm12_scale8:
+  case AArch64::fixup_aarch64_ldst_imm12_scale16:
+  case AArch64::fixup_aarch64_ldr_pcrel_imm19:
+  case AArch64::fixup_aarch64_pcrel_branch19:
+  case AArch64::fixup_aarch64_pcrel_adr_imm21:
+  case AArch64::fixup_aarch64_pcrel_adrp_imm21:
+  case AArch64::fixup_aarch64_pcrel_branch26:
+  case AArch64::fixup_aarch64_pcrel_call26:
+    // Instructions are always little endian
+    return 0;
+  }
+}
+
 void AArch64AsmBackend::applyFixup(const MCFixup &Fixup, char *Data,
                                    unsigned DataSize, uint64_t Value,
                                    bool IsPCRel) const {
   unsigned NumBytes = getFixupKindNumBytes(Fixup.getKind());
   if (!Value)
     return; // Doesn't change encoding.
   MCFixupKindInfo Info = getFixupKindInfo(Fixup.getKind());
   // Apply any target-specific value adjustments.
   Value = adjustFixupValue(Fixup.getKind(), Value);
 
   // Shift the value into position.
   Value <<= Info.TargetOffset;
 
   unsigned Offset = Fixup.getOffset();
   assert(Offset + NumBytes <= DataSize && "Invalid fixup offset!");
 
+  // Used to point to big endian bytes.
+  unsigned FullSizeBytes = getFixupKindContainerSizeBytes(Fixup.getKind());
+
   // For each byte of the fragment that the fixup touches, mask in the
   // bits from the fixup value.
-  for (unsigned i = 0; i != NumBytes; ++i)
+  if (FullSizeBytes == 0) {
+    // Handle as little-endian
+    for (unsigned i = 0; i != NumBytes; ++i) {
       Data[Offset + i] |= uint8_t((Value >> (i * 8)) & 0xff);
     }
+  } else {
+    // Handle as big-endian
+    assert((Offset + FullSizeBytes) <= DataSize && "Invalid fixup size!");
+    assert(NumBytes <= FullSizeBytes && "Invalid fixup size!");
+    for (unsigned i = 0; i != NumBytes; ++i) {
+      unsigned Idx = FullSizeBytes - 1 - i;
+      Data[Offset + Idx] |= uint8_t((Value >> (i * 8)) & 0xff);
+    }
+  }
+}
 
 bool AArch64AsmBackend::mayNeedRelaxation(const MCInst &Inst) const {
   return false;
 }
 
 bool AArch64AsmBackend::fixupNeedsRelaxation(const MCFixup &Fixup,
                                              uint64_t Value,
                                              const MCRelaxableFragment *DF,
class DarwinAArch64AsmBackend : public AArch64AsmBackend {
   /// See UNWIND_AArch64_FRAMELESS_STACK_SIZE_MASK in compact_unwind_encoding.h.
   /// The stack size always needs to be 16 byte aligned.
   uint32_t encodeStackAdjustment(uint32_t StackSize) const {
     return (StackSize / 16) << 12;
   }
 
 public:
   DarwinAArch64AsmBackend(const Target &T, const MCRegisterInfo &MRI)
-      : AArch64AsmBackend(T), MRI(MRI) {}
+      : AArch64AsmBackend(T, /*IsLittleEndian*/true), MRI(MRI) {}
 
   MCObjectWriter *createObjectWriter(raw_pwrite_stream &OS) const override {
     return createAArch64MachObjectWriter(OS, MachO::CPU_TYPE_ARM64,
                                          MachO::CPU_SUBTYPE_ARM64_ALL);
   }
 
   /// \brief Generate the compact unwind encoding from the CFI directives.
   uint32_t generateCompactUnwindEncoding(

 
 } // end anonymous namespace
 
 namespace {
 
 class ELFAArch64AsmBackend : public AArch64AsmBackend {
 public:
   uint8_t OSABI;
-  bool IsLittleEndian;
 
   ELFAArch64AsmBackend(const Target &T, uint8_t OSABI, bool IsLittleEndian)
-    : AArch64AsmBackend(T), OSABI(OSABI), IsLittleEndian(IsLittleEndian) {}
+    : AArch64AsmBackend(T, IsLittleEndian), OSABI(OSABI) {}
 
   MCObjectWriter *createObjectWriter(raw_pwrite_stream &OS) const override {
     return createAArch64ELFObjectWriter(OS, OSABI, IsLittleEndian);
   }
 
   void processFixupValue(const MCAssembler &Asm, const MCAsmLayout &Layout,
                          const MCFixup &Fixup, const MCFragment *DF,
                          const MCValue &Target, uint64_t &Value,
                          bool &IsResolved) override;
-
-  void applyFixup(const MCFixup &Fixup, char *Data, unsigned DataSize,
-                  uint64_t Value, bool IsPCRel) const override;
 };
 
 void ELFAArch64AsmBackend::processFixupValue(
     const MCAssembler &Asm, const MCAsmLayout &Layout, const MCFixup &Fixup,
     const MCFragment *DF, const MCValue &Target, uint64_t &Value,
     bool &IsResolved) {
   // The ADRP instruction adds some multiple of 0x1000 to the current PC &
   // ~0xfff. This means that the required offset to reach a symbol can vary by
   // up to one step depending on where the ADRP is in memory. For example:
   //
   //     ADRP x0, there
   //  there:
   //
   // If the ADRP occurs at address 0xffc then "there" will be at 0x1000 and
   // we'll need that as an offset. At any other address "there" will be in the
   // same page as the ADRP and the instruction should encode 0x0. Assuming the
   // section isn't 0x1000-aligned, we therefore need to delegate this decision
   // to the linker -- a relocation!
   if ((uint32_t)Fixup.getKind() == AArch64::fixup_aarch64_pcrel_adrp_imm21)
     IsResolved = false;
 }
 
-// Returns whether this fixup is based on an address in the .eh_frame section,
-// and therefore should be byte swapped.
-// FIXME: Should be replaced with something more principled.
-static bool isByteSwappedFixup(const MCExpr *E) {
-  MCValue Val;
-  if (!E->evaluateAsRelocatable(Val, nullptr, nullptr))
-    return false;
-
-  if (!Val.getSymA() || Val.getSymA()->getSymbol().isUndefined())
-    return false;
-
-  const MCSectionELF *SecELF =
-      dyn_cast<MCSectionELF>(&Val.getSymA()->getSymbol().getSection());
-  return SecELF->getSectionName() == ".eh_frame";
-}
-
-void ELFAArch64AsmBackend::applyFixup(const MCFixup &Fixup, char *Data,
-                                      unsigned DataSize, uint64_t Value,
-                                      bool IsPCRel) const {
-  // store fixups in .eh_frame section in big endian order
-  if (!IsLittleEndian && Fixup.getKind() == FK_Data_4) {
-    if (isByteSwappedFixup(Fixup.getValue()))
-      Value = ByteSwap_32(unsigned(Value));
-  }
-  AArch64AsmBackend::applyFixup (Fixup, Data, DataSize, Value, IsPCRel);
-}
 }
 
 MCAsmBackend *llvm::createAArch64leAsmBackend(const Target &T,
                                               const MCRegisterInfo &MRI,
                                               const Triple &TheTriple,
                                               StringRef CPU) {
   if (TheTriple.isOSBinFormatMachO())
     return new DarwinAArch64AsmBackend(T, MRI);

test/DebugInfo/AArch64/line-header.ll

+; RUN: llc -mtriple=aarch64-none-linux -O0 -filetype=obj - < %S/../Inputs/line.ll | llvm-dwarfdump - | FileCheck %s
+; RUN: llc -mtriple=aarch64_be-none-linux -O0 -filetype=obj - < %S/../Inputs/line.ll | llvm-dwarfdump - | FileCheck %s
+
+; check line table length is correctly calculated for both big and little endian
+CHECK_LABEL: .debug_line contents:
+CHECK: total_length: 0x0000003c