// ehframe.cc -- handle exception frame sections for gold

// Copyright (C) 2006-2024 Free Software Foundation, Inc.
// Written by Ian Lance Taylor <[email protected]>.

// This file is part of gold.

// This program is free software; you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation; either version 3 of the License, or
// (at your option) any later version.

// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
// GNU General Public License for more details.

// You should have received a copy of the GNU General Public License
// along with this program; if not, write to the Free Software
// Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston,
// MA 02110-1301, USA.

#include "gold.h"

#include <cstring>
#include <algorithm>

#include "elfcpp.h"
#include "dwarf.h"
#include "symtab.h"
#include "reloc.h"
#include "ehframe.h"

namespace gold
{

// This file handles generation of the exception frame header that
// gcc's runtime support libraries use to find unwind information at
// runtime.  This file also handles discarding duplicate exception
// frame information.

// The exception frame header starts with four bytes:

// 0: The version number, currently 1.

// 1: The encoding of the pointer to the exception frames.  This can
//    be any DWARF unwind encoding (DW_EH_PE_*).  It is normally a 4
//    byte PC relative offset (DW_EH_PE_pcrel | DW_EH_PE_sdata4).

// 2: The encoding of the count of the number of FDE pointers in the
//    lookup table.  This can be any DWARF unwind encoding, and in
//    particular can be DW_EH_PE_omit if the count is omitted.  It is
//    normally a 4 byte unsigned count (DW_EH_PE_udata4).

// 3: The encoding of the lookup table entries.  Currently gcc's
//    libraries will only support DW_EH_PE_datarel | DW_EH_PE_sdata4,
//    which means that the values are 4 byte offsets from the start of
//    the table.

// The exception frame header is followed by a pointer to the contents
// of the exception frame section (.eh_frame).  This pointer is
// encoded as specified in the byte at offset 1 of the header (i.e.,
// it is normally a 4 byte PC relative offset).

// If there is a lookup table, this is followed by the count of the
// number of FDE pointers, encoded as specified in the byte at offset
// 2 of the header (i.e., normally a 4 byte unsigned integer).

// This is followed by the table, which should start at an 4-byte
// aligned address in memory.  Each entry in the table is 8 bytes.
// Each entry represents an FDE.  The first four bytes of each entry
// are an offset to the starting PC for the FDE.  The last four bytes
// of each entry are an offset to the FDE data.  The offsets are from
// the start of the exception frame header information.  The entries
// are in sorted order by starting PC.

const int eh_frame_hdr_size = 4;

// Construct the exception frame header.

Eh_frame_hdr::Eh_frame_hdr(Output_section* eh_frame_section,
                          const Eh_frame* eh_frame_data)
 : Output_section_data(4),
   eh_frame_section_(eh_frame_section),
   eh_frame_data_(eh_frame_data),
   fde_offsets_(),
   any_unrecognized_eh_frame_sections_(false)
{
}

// Set the size of the exception frame header.

void
Eh_frame_hdr::set_final_data_size()
{
 unsigned int data_size = eh_frame_hdr_size + 4;
 if (!this->any_unrecognized_eh_frame_sections_)
   {
     unsigned int fde_count = this->eh_frame_data_->fde_count();
     if (fde_count != 0)
       data_size += 4 + 8 * fde_count;
     this->fde_offsets_.reserve(fde_count);
   }
 this->set_data_size(data_size);
}

// Write the data to the file.

void
Eh_frame_hdr::do_write(Output_file* of)
{
 switch (parameters->size_and_endianness())
   {
#ifdef HAVE_TARGET_32_LITTLE
   case Parameters::TARGET_32_LITTLE:
     this->do_sized_write<32, false>(of);
     break;
#endif
#ifdef HAVE_TARGET_32_BIG
   case Parameters::TARGET_32_BIG:
     this->do_sized_write<32, true>(of);
     break;
#endif
#ifdef HAVE_TARGET_64_LITTLE
   case Parameters::TARGET_64_LITTLE:
     this->do_sized_write<64, false>(of);
     break;
#endif
#ifdef HAVE_TARGET_64_BIG
   case Parameters::TARGET_64_BIG:
     this->do_sized_write<64, true>(of);
     break;
#endif
   default:
     gold_unreachable();
   }
}

// Write the data to the file with the right endianness.

template<int size, bool big_endian>
void
Eh_frame_hdr::do_sized_write(Output_file* of)
{
 const off_t off = this->offset();
 const off_t oview_size = this->data_size();
 unsigned char* const oview = of->get_output_view(off, oview_size);

 // Version number.
 oview[0] = 1;

 // Write out a 4 byte PC relative offset to the address of the
 // .eh_frame section.
 oview[1] = elfcpp::DW_EH_PE_pcrel | elfcpp::DW_EH_PE_sdata4;
 uint64_t eh_frame_address = this->eh_frame_section_->address();
 uint64_t eh_frame_hdr_address = this->address();
 uint64_t eh_frame_offset = (eh_frame_address -
                             (eh_frame_hdr_address + 4));
 elfcpp::Swap<32, big_endian>::writeval(oview + 4, eh_frame_offset);

 if (this->any_unrecognized_eh_frame_sections_
     || this->fde_offsets_.empty())
   {
     // There are no FDEs, or we didn't recognize the format of the
     // some of the .eh_frame sections, so we can't write out the
     // sorted table.
     oview[2] = elfcpp::DW_EH_PE_omit;
     oview[3] = elfcpp::DW_EH_PE_omit;

     gold_assert(oview_size == 8);
   }
 else
   {
     oview[2] = elfcpp::DW_EH_PE_udata4;
     oview[3] = elfcpp::DW_EH_PE_datarel | elfcpp::DW_EH_PE_sdata4;

     elfcpp::Swap<32, big_endian>::writeval(oview + 8,
                                            this->fde_offsets_.size());

     // We have the offsets of the FDEs in the .eh_frame section.  We
     // couldn't easily get the PC values before, as they depend on
     // relocations which are, of course, target specific.  This code
     // is run after all those relocations have been applied to the
     // output file.  Here we read the output file again to find the
     // PC values.  Then we sort the list and write it out.

     Fde_addresses<size> fde_addresses(this->fde_offsets_.size());
     this->get_fde_addresses<size, big_endian>(of, &this->fde_offsets_,
                                               &fde_addresses);

     std::sort(fde_addresses.begin(), fde_addresses.end(),
               Fde_address_compare<size>());

     typename elfcpp::Elf_types<size>::Elf_Addr output_address;
     output_address = this->address();

     unsigned char* pfde = oview + 12;
     for (typename Fde_addresses<size>::iterator p = fde_addresses.begin();
          p != fde_addresses.end();
          ++p)
       {
         elfcpp::Swap<32, big_endian>::writeval(pfde,
                                                p->first - output_address);
         elfcpp::Swap<32, big_endian>::writeval(pfde + 4,
                                                p->second - output_address);
         pfde += 8;
       }

     gold_assert(pfde - oview == oview_size);
   }

 of->write_output_view(off, oview_size, oview);
}

// Given the offset FDE_OFFSET of an FDE in the .eh_frame section, and
// the contents of the .eh_frame section EH_FRAME_CONTENTS, where the
// FDE's encoding is FDE_ENCODING, return the output address of the
// FDE's PC.

template<int size, bool big_endian>
typename elfcpp::Elf_types<size>::Elf_Addr
Eh_frame_hdr::get_fde_pc(
   typename elfcpp::Elf_types<size>::Elf_Addr eh_frame_address,
   const unsigned char* eh_frame_contents,
   section_offset_type fde_offset,
   unsigned char fde_encoding)
{
 // The FDE starts with a 4 byte length and a 4 byte offset to the
 // CIE.  The PC follows.
 const unsigned char* p = eh_frame_contents + fde_offset + 8;

 typename elfcpp::Elf_types<size>::Elf_Addr pc;
 bool is_signed = (fde_encoding & elfcpp::DW_EH_PE_signed) != 0;
 int pc_size = fde_encoding & 7;
 if (pc_size == elfcpp::DW_EH_PE_absptr)
   {
     if (size == 32)
       pc_size = elfcpp::DW_EH_PE_udata4;
     else if (size == 64)
       pc_size = elfcpp::DW_EH_PE_udata8;
     else
       gold_unreachable();
   }

 switch (pc_size)
   {
   case elfcpp::DW_EH_PE_udata2:
     pc = elfcpp::Swap<16, big_endian>::readval(p);
     if (is_signed)
       pc = (pc ^ 0x8000) - 0x8000;
     break;

   case elfcpp::DW_EH_PE_udata4:
     pc = elfcpp::Swap<32, big_endian>::readval(p);
     if (size > 32 && is_signed)
       pc = (pc ^ 0x80000000) - 0x80000000;
     break;

   case elfcpp::DW_EH_PE_udata8:
     gold_assert(size == 64);
     pc = elfcpp::Swap_unaligned<64, big_endian>::readval(p);
     break;

   default:
     // All other cases were rejected in Eh_frame::read_cie.
     gold_unreachable();
   }

 switch (fde_encoding & 0x70)
   {
   case 0:
     break;

   case elfcpp::DW_EH_PE_pcrel:
     pc += eh_frame_address + fde_offset + 8;
     break;

   case elfcpp::DW_EH_PE_datarel:
     pc += parameters->target().ehframe_datarel_base();
     break;

   default:
     // If other cases arise, then we have to handle them, or we have
     // to reject them by returning false in Eh_frame::read_cie.
     gold_unreachable();
   }

 gold_assert((fde_encoding & elfcpp::DW_EH_PE_indirect) == 0);

 return pc;
}

// Given an array of FDE offsets in the .eh_frame section, return an
// array of offsets from the exception frame header to the FDE's
// output PC and to the output address of the FDE itself.  We get the
// FDE's PC by actually looking in the .eh_frame section we just wrote
// to the output file.

template<int size, bool big_endian>
void
Eh_frame_hdr::get_fde_addresses(Output_file* of,
                               const Fde_offsets* fde_offsets,
                               Fde_addresses<size>* fde_addresses)
{
 typename elfcpp::Elf_types<size>::Elf_Addr eh_frame_address;
 eh_frame_address = this->eh_frame_section_->address();
 off_t eh_frame_offset = this->eh_frame_section_->offset();
 off_t eh_frame_size = this->eh_frame_section_->data_size();
 const unsigned char* eh_frame_contents = of->get_input_view(eh_frame_offset,
                                                             eh_frame_size);

 for (Fde_offsets::const_iterator p = fde_offsets->begin();
      p != fde_offsets->end();
      ++p)
   {
     typename elfcpp::Elf_types<size>::Elf_Addr fde_pc;
     fde_pc = this->get_fde_pc<size, big_endian>(eh_frame_address,
                                                 eh_frame_contents,
                                                 p->first, p->second);
     fde_addresses->push_back(fde_pc, eh_frame_address + p->first);
   }

 of->free_input_view(eh_frame_offset, eh_frame_size, eh_frame_contents);
}

// Class Fde.

// Write the FDE to OVIEW starting at OFFSET.  CIE_OFFSET is the
// offset of the CIE in OVIEW.  OUTPUT_OFFSET is the offset of the
// Eh_frame section within the output section.  FDE_ENCODING is the
// encoding, from the CIE.  ADDRALIGN is the required alignment.
// ADDRESS is the virtual address of OVIEW.  Record the FDE pc for
// EH_FRAME_HDR.  Return the new offset.

template<int size, bool big_endian>
section_offset_type
Fde::write(unsigned char* oview, section_offset_type output_offset,
          section_offset_type offset, uint64_t address, unsigned int addralign,
          section_offset_type cie_offset, unsigned char fde_encoding,
          Eh_frame_hdr* eh_frame_hdr)
{
 gold_assert((offset & (addralign - 1)) == 0);

 size_t length = this->contents_.length();

 // We add 8 when getting the aligned length to account for the
 // length word and the CIE offset.
 size_t aligned_full_length = align_address(length + 8, addralign);

 // Write the length of the FDE as a 32-bit word.  The length word
 // does not include the four bytes of the length word itself, but it
 // does include the offset to the CIE.
 elfcpp::Swap<32, big_endian>::writeval(oview + offset,
                                        aligned_full_length - 4);

 // Write the offset to the CIE as a 32-bit word.  This is the
 // difference between the address of the offset word itself and the
 // CIE address.
 elfcpp::Swap<32, big_endian>::writeval(oview + offset + 4,
                                        offset + 4 - cie_offset);

 // Copy the rest of the FDE.  Note that this is run before
 // relocation processing is done on this section, so the relocations
 // will later be applied to the FDE data.
 memcpy(oview + offset + 8, this->contents_.data(), length);

 // If this FDE is associated with a PLT, fill in the PLT's address
 // and size.
 if (this->object_ == NULL)
   {
     gold_assert(memcmp(oview + offset + 8, "\0\0\0\0\0\0\0\0", 8) == 0);
     uint64_t paddress;
     off_t psize;
     parameters->target().plt_fde_location(this->u_.from_linker.plt,
                                           oview + offset + 8,
                                           &paddress, &psize);
     uint64_t poffset = paddress - (address + offset + 8);
     int32_t spoffset = static_cast<int32_t>(poffset);
     uint32_t upsize = static_cast<uint32_t>(psize);
     if (static_cast<uint64_t>(static_cast<int64_t>(spoffset)) != poffset
         || static_cast<off_t>(upsize) != psize)
       gold_warning(_("overflow in PLT unwind data; "
                      "unwinding through PLT may fail"));
     elfcpp::Swap<32, big_endian>::writeval(oview + offset + 8, spoffset);
     elfcpp::Swap<32, big_endian>::writeval(oview + offset + 12, upsize);
   }

 if (aligned_full_length > length + 8)
   memset(oview + offset + length + 8, 0, aligned_full_length - (length + 8));

 // Tell the exception frame header about this FDE.
 if (eh_frame_hdr != NULL)
   eh_frame_hdr->record_fde(output_offset + offset, fde_encoding);

 return offset + aligned_full_length;
}

// Class Cie.

// Destructor.

Cie::~Cie()
{
 for (std::vector<Fde*>::iterator p = this->fdes_.begin();
      p != this->fdes_.end();
      ++p)
   delete *p;
}

// Set the output offset of a CIE.  Return the new output offset.

section_offset_type
Cie::set_output_offset(section_offset_type output_offset,
                      unsigned int addralign,
                      Output_section_data *output_data)
{
 size_t length = this->contents_.length();

 // Add 4 for length and 4 for zero CIE identifier tag.
 length += 8;

 if (this->object_ != NULL)
   {
     // Add a mapping so that relocations are applied correctly.
     this->object_->add_merge_mapping(output_data, this->shndx_,
                                      this->input_offset_, length,
                                      output_offset);
   }

 length = align_address(length, addralign);

 for (std::vector<Fde*>::const_iterator p = this->fdes_.begin();
      p != this->fdes_.end();
      ++p)
   {
     (*p)->add_mapping(output_offset + length, output_data);

     size_t fde_length = (*p)->length();
     fde_length = align_address(fde_length, addralign);
     length += fde_length;
   }

 return output_offset + length;
}

// Write the CIE to OVIEW starting at OFFSET.  OUTPUT_OFFSET is the
// offset of the Eh_frame section within the output section.  Round up
// the bytes to ADDRALIGN.  ADDRESS is the virtual address of OVIEW.
// EH_FRAME_HDR is the exception frame header for FDE recording.
// POST_FDES stashes FDEs created after mappings were done, for later
// writing.  Return the new offset.

template<int size, bool big_endian>
section_offset_type
Cie::write(unsigned char* oview, section_offset_type output_offset,
          section_offset_type offset, uint64_t address,
          unsigned int addralign, Eh_frame_hdr* eh_frame_hdr,
          Post_fdes* post_fdes)
{
 gold_assert((offset & (addralign - 1)) == 0);

 section_offset_type cie_offset = offset;

 size_t length = this->contents_.length();

 // We add 8 when getting the aligned length to account for the
 // length word and the CIE tag.
 size_t aligned_full_length = align_address(length + 8, addralign);

 // Write the length of the CIE as a 32-bit word.  The length word
 // does not include the four bytes of the length word itself.
 elfcpp::Swap<32, big_endian>::writeval(oview + offset,
                                        aligned_full_length - 4);

 // Write the tag which marks this as a CIE: a 32-bit zero.
 elfcpp::Swap<32, big_endian>::writeval(oview + offset + 4, 0);

 // Write out the CIE data.
 memcpy(oview + offset + 8, this->contents_.data(), length);

 if (aligned_full_length > length + 8)
   memset(oview + offset + length + 8, 0, aligned_full_length - (length + 8));

 offset += aligned_full_length;

 // Write out the associated FDEs.
 unsigned char fde_encoding = this->fde_encoding_;
 for (std::vector<Fde*>::const_iterator p = this->fdes_.begin();
      p != this->fdes_.end();
      ++p)
   {
     if ((*p)->post_map())
       post_fdes->push_back(Post_fde(*p, cie_offset, fde_encoding));
     else
       offset = (*p)->write<size, big_endian>(oview, output_offset, offset,
                                              address, addralign, cie_offset,
                                              fde_encoding, eh_frame_hdr);
   }

 return offset;
}

// We track all the CIEs we see, and merge them when possible.  This
// works because each FDE holds an offset to the relevant CIE: we
// rewrite the FDEs to point to the merged CIE.  This is worthwhile
// because in a typical C++ program many FDEs in many different object
// files will use the same CIE.

// An equality operator for Cie.

bool
operator==(const Cie& cie1, const Cie& cie2)
{
 return (cie1.personality_name_ == cie2.personality_name_
         && cie1.contents_ == cie2.contents_);
}

// A less-than operator for Cie.

bool
operator<(const Cie& cie1, const Cie& cie2)
{
 if (cie1.personality_name_ != cie2.personality_name_)
   return cie1.personality_name_ < cie2.personality_name_;
 return cie1.contents_ < cie2.contents_;
}

// Class Eh_frame.

Eh_frame::Eh_frame()
 : Output_section_data(Output_data::default_alignment()),
   eh_frame_hdr_(NULL),
   cie_offsets_(),
   unmergeable_cie_offsets_(),
   mappings_are_done_(false),
   final_data_size_(0)
{
}

// Skip an LEB128, updating *PP to point to the next character.
// Return false if we ran off the end of the string.

bool
Eh_frame::skip_leb128(const unsigned char** pp, const unsigned char* pend)
{
 const unsigned char* p;
 for (p = *pp; p < pend; ++p)
   {
     if ((*p & 0x80) == 0)
       {
         *pp = p + 1;
         return true;
       }
   }
 return false;
}

// Add input section SHNDX in OBJECT to an exception frame section.
// SYMBOLS is the contents of the symbol table section (size
// SYMBOLS_SIZE), SYMBOL_NAMES is the symbol names section (size
// SYMBOL_NAMES_SIZE).  RELOC_SHNDX is the index of a relocation
// section applying to SHNDX, or 0 if none, or -1U if more than one.
// RELOC_TYPE is the type of the reloc section if there is one, either
// SHT_REL or SHT_RELA.  We try to parse the input exception frame
// data into our data structures.  If we can't do it, we return false
// to mean that the section should be handled as a normal input
// section.

template<int size, bool big_endian>
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section(
   Sized_relobj_file<size, big_endian>* object,
   const unsigned char* symbols,
   section_size_type symbols_size,
   const unsigned char* symbol_names,
   section_size_type symbol_names_size,
   unsigned int shndx,
   unsigned int reloc_shndx,
   unsigned int reloc_type)
{
 // Get the section contents.
 section_size_type contents_len;
 const unsigned char* pcontents = object->section_contents(shndx,
                                                           &contents_len,
                                                           false);
 if (contents_len == 0)
   return EH_EMPTY_SECTION;

 // If this is the marker section for the end of the data, then
 // return false to force it to be handled as an ordinary input
 // section.  If we don't do this, we won't correctly handle the case
 // of unrecognized .eh_frame sections.
 if (contents_len == 4
     && elfcpp::Swap<32, big_endian>::readval(pcontents) == 0)
   return EH_END_MARKER_SECTION;

 New_cies new_cies;
 if (!this->do_add_ehframe_input_section(object, symbols, symbols_size,
                                         symbol_names, symbol_names_size,
                                         shndx, reloc_shndx,
                                         reloc_type, pcontents,
                                         contents_len, &new_cies))
   {
     if (this->eh_frame_hdr_ != NULL)
       this->eh_frame_hdr_->found_unrecognized_eh_frame_section();

     for (New_cies::iterator p = new_cies.begin();
          p != new_cies.end();
          ++p)
       delete p->first;

     return EH_UNRECOGNIZED_SECTION;
   }

 // Now that we know we are using this section, record any new CIEs
 // that we found.
 for (New_cies::const_iterator p = new_cies.begin();
      p != new_cies.end();
      ++p)
   {
     if (p->second)
       this->cie_offsets_.insert(p->first);
     else
       this->unmergeable_cie_offsets_.push_back(p->first);
   }

 return EH_OPTIMIZABLE_SECTION;
}

// The bulk of the implementation of add_ehframe_input_section.

template<int size, bool big_endian>
bool
Eh_frame::do_add_ehframe_input_section(
   Sized_relobj_file<size, big_endian>* object,
   const unsigned char* symbols,
   section_size_type symbols_size,
   const unsigned char* symbol_names,
   section_size_type symbol_names_size,
   unsigned int shndx,
   unsigned int reloc_shndx,
   unsigned int reloc_type,
   const unsigned char* pcontents,
   section_size_type contents_len,
   New_cies* new_cies)
{
 Track_relocs<size, big_endian> relocs;

 const unsigned char* p = pcontents;
 const unsigned char* pend = p + contents_len;

 // Get the contents of the reloc section if any.
 if (!relocs.initialize(object, reloc_shndx, reloc_type))
   return false;

 // Keep track of which CIEs are at which offsets.
 Offsets_to_cie cies;

 while (p < pend)
   {
     if (pend - p < 4)
       return false;

     // There shouldn't be any relocations here.
     if (relocs.advance(p + 4 - pcontents) > 0)
       return false;

     unsigned int len = elfcpp::Swap<32, big_endian>::readval(p);
     p += 4;
     if (len == 0)
       {
         // We should only find a zero-length entry at the end of the
         // section.
         if (p < pend)
           return false;
         break;
       }
     // We don't support a 64-bit .eh_frame.
     if (len == 0xffffffff)
       return false;
     if (static_cast<unsigned int>(pend - p) < len)
       return false;

     const unsigned char* const pentend = p + len;

     if (pend - p < 4)
       return false;
     if (relocs.advance(p + 4 - pcontents) > 0)
       return false;

     unsigned int id = elfcpp::Swap<32, big_endian>::readval(p);
     p += 4;

     if (id == 0)
       {
         // CIE.
         if (!this->read_cie(object, shndx, symbols, symbols_size,
                             symbol_names, symbol_names_size,
                             pcontents, p, pentend, &relocs, &cies,
                             new_cies))
           return false;
       }
     else
       {
         // FDE.
         if (!this->read_fde(object, shndx, symbols, symbols_size,
                             pcontents, id, p, pentend, &relocs, &cies))
           return false;
       }

     p = pentend;
   }

 return true;
}

// Read a CIE.  Return false if we can't parse the information.

template<int size, bool big_endian>
bool
Eh_frame::read_cie(Sized_relobj_file<size, big_endian>* object,
                  unsigned int shndx,
                  const unsigned char* symbols,
                  section_size_type symbols_size,
                  const unsigned char* symbol_names,
                  section_size_type symbol_names_size,
                  const unsigned char* pcontents,
                  const unsigned char* pcie,
                  const unsigned char* pcieend,
                  Track_relocs<size, big_endian>* relocs,
                  Offsets_to_cie* cies,
                  New_cies* new_cies)
{
 bool mergeable = true;

 // We need to find the personality routine if there is one, since we
 // can only merge CIEs which use the same routine.  We also need to
 // find the FDE encoding if there is one, so that we can read the PC
 // from the FDE.

 const unsigned char* p = pcie;

 if (pcieend - p < 1)
   return false;
 unsigned char version = *p++;
 if (version != 1 && version != 3)
   return false;

 const unsigned char* paug = p;
 const void* paugendv = memchr(p, '\0', pcieend - p);
 const unsigned char* paugend = static_cast<const unsigned char*>(paugendv);
 if (paugend == NULL)
   return false;
 p = paugend + 1;

 if (paug[0] == 'e' && paug[1] == 'h')
   {
     // This is a CIE from gcc before version 3.0.  We can't merge
     // these.  We can still read the FDEs.
     mergeable = false;
     paug += 2;
     if (*paug != '\0')
       return false;
     if (pcieend - p < size / 8)
       return false;
     p += size / 8;
   }

 // Skip the code alignment.
 if (!skip_leb128(&p, pcieend))
   return false;

 // Skip the data alignment.
 if (!skip_leb128(&p, pcieend))
   return false;

 // Skip the return column.
 if (version == 1)
   {
     if (pcieend - p < 1)
       return false;
     ++p;
   }
 else
   {
     if (!skip_leb128(&p, pcieend))
       return false;
   }

 if (*paug == 'z')
   {
     ++paug;
     // Skip the augmentation size.
     if (!skip_leb128(&p, pcieend))
       return false;
   }

 unsigned char fde_encoding = elfcpp::DW_EH_PE_absptr;
 int per_offset = -1;
 while (*paug != '\0')
   {
     switch (*paug)
       {
       case 'L': // LSDA encoding.
         if (pcieend - p < 1)
           return false;
         ++p;
         break;

       case 'R': // FDE encoding.
         if (pcieend - p < 1)
           return false;
         fde_encoding = *p;
         switch (fde_encoding & 7)
           {
           case elfcpp::DW_EH_PE_absptr:
           case elfcpp::DW_EH_PE_udata2:
           case elfcpp::DW_EH_PE_udata4:
           case elfcpp::DW_EH_PE_udata8:
             break;
           default:
             // We don't expect to see any other cases here, and
             // we're not prepared to handle them.
             return false;
           }
         ++p;
         break;

       case 'S':
         break;

       case 'P':
         // Personality encoding.
         {
           if (pcieend - p < 1)
             return false;
           unsigned char per_encoding = *p;
           ++p;

           if ((per_encoding & 0x60) == 0x60)
             return false;
           unsigned int per_width;
           switch (per_encoding & 7)
             {
             case elfcpp::DW_EH_PE_udata2:
               per_width = 2;
               break;
             case elfcpp::DW_EH_PE_udata4:
               per_width = 4;
               break;
             case elfcpp::DW_EH_PE_udata8:
               per_width = 8;
               break;
             case elfcpp::DW_EH_PE_absptr:
               per_width = size / 8;
               break;
             default:
               return false;
             }

           if ((per_encoding & 0xf0) == elfcpp::DW_EH_PE_aligned)
             {
               unsigned int len = p - pcie;
               len += per_width - 1;
               len &= ~ (per_width - 1);
               if (static_cast<unsigned int>(pcieend - p) < len)
                 return false;
               p += len;
             }

           per_offset = p - pcontents;

           if (static_cast<unsigned int>(pcieend - p) < per_width)
             return false;
           p += per_width;
         }
         break;

       default:
         return false;
       }

     ++paug;
   }

 const char* personality_name = "";
 if (per_offset != -1)
   {
     if (relocs->advance(per_offset) > 0)
       return false;
     if (relocs->next_offset() != per_offset)
       return false;

     unsigned int personality_symndx = relocs->next_symndx();
     if (personality_symndx == -1U)
       return false;

     if (personality_symndx < object->local_symbol_count())
       {
         // We can only merge this CIE if the personality routine is
         // a global symbol.  We can still read the FDEs.
         mergeable = false;
       }
     else
       {
         const int sym_size = elfcpp::Elf_sizes<size>::sym_size;
         if (personality_symndx >= symbols_size / sym_size)
           return false;
         elfcpp::Sym<size, big_endian> sym(symbols
                                           + (personality_symndx * sym_size));
         unsigned int name_offset = sym.get_st_name();
         if (name_offset >= symbol_names_size)
           return false;
         personality_name = (reinterpret_cast<const char*>(symbol_names)
                             + name_offset);
       }

     int r = relocs->advance(per_offset + 1);
     gold_assert(r == 1);
   }

 if (relocs->advance(pcieend - pcontents) > 0)
   return false;

 Cie cie(object, shndx, (pcie - 8) - pcontents, fde_encoding,
         personality_name, pcie, pcieend - pcie);
 Cie* cie_pointer = NULL;
 if (mergeable)
   {
     Cie_offsets::iterator find_cie = this->cie_offsets_.find(&cie);
     if (find_cie != this->cie_offsets_.end())
       cie_pointer = *find_cie;
     else
       {
         // See if we already saw this CIE in this object file.
         for (New_cies::const_iterator pc = new_cies->begin();
              pc != new_cies->end();
              ++pc)
           {
             if (*(pc->first) == cie)
               {
                 cie_pointer = pc->first;
                 break;
               }
           }
       }
   }

 if (cie_pointer == NULL)
   {
     cie_pointer = new Cie(cie);
     new_cies->push_back(std::make_pair(cie_pointer, mergeable));
   }
 else
   {
     // We are deleting this CIE.  Record that in our mapping from
     // input sections to the output section.  At this point we don't
     // know for sure that we are doing a special mapping for this
     // input section, but that's OK--if we don't do a special
     // mapping, nobody will ever ask for the mapping we add here.
     object->add_merge_mapping(this, shndx, (pcie - 8) - pcontents,
                               pcieend - (pcie - 8), -1);
   }

 // Record this CIE plus the offset in the input section.
 cies->insert(std::make_pair(pcie - pcontents, cie_pointer));

 return true;
}

// Read an FDE.  Return false if we can't parse the information.

template<int size, bool big_endian>
bool
Eh_frame::read_fde(Sized_relobj_file<size, big_endian>* object,
                  unsigned int shndx,
                  const unsigned char* symbols,
                  section_size_type symbols_size,
                  const unsigned char* pcontents,
                  unsigned int offset,
                  const unsigned char* pfde,
                  const unsigned char* pfdeend,
                  Track_relocs<size, big_endian>* relocs,
                  Offsets_to_cie* cies)
{
 // OFFSET is the distance between the 4 bytes before PFDE to the
 // start of the CIE.  The offset we recorded for the CIE is 8 bytes
 // after the start of the CIE--after the length and the zero tag.
 unsigned int cie_offset = (pfde - 4 - pcontents) - offset + 8;
 Offsets_to_cie::const_iterator pcie = cies->find(cie_offset);
 if (pcie == cies->end())
   return false;
 Cie* cie = pcie->second;

 int pc_size = 0;
 switch (cie->fde_encoding() & 7)
   {
   case elfcpp::DW_EH_PE_udata2:
     pc_size = 2;
     break;
   case elfcpp::DW_EH_PE_udata4:
     pc_size = 4;
     break;
   case elfcpp::DW_EH_PE_udata8:
     gold_assert(size == 64);
     pc_size = 8;
     break;
   case elfcpp::DW_EH_PE_absptr:
     pc_size = size == 32 ? 4 : 8;
     break;
   default:
     // All other cases were rejected in Eh_frame::read_cie.
     gold_unreachable();
   }

 // The FDE should start with a reloc to the start of the code which
 // it describes.
 if (relocs->advance(pfde - pcontents) > 0)
   return false;
 if (relocs->next_offset() != pfde - pcontents)
   {
     // In an object produced by a relocatable link, gold may have
     // discarded a COMDAT group in the previous link, but not the
     // corresponding FDEs. In that case, gold will have discarded
     // the relocations, so the FDE will have a non-relocatable zero
     // (regardless of whether the PC encoding is absolute, pc-relative,
     // or data-relative) instead of a pointer to the start of the code.

     uint64_t pc_value = 0;
     switch (pc_size)
       {
       case 2:
         pc_value = elfcpp::Swap<16, big_endian>::readval(pfde);
         break;
       case 4:
         pc_value = elfcpp::Swap<32, big_endian>::readval(pfde);
         break;
       case 8:
         pc_value = elfcpp::Swap_unaligned<64, big_endian>::readval(pfde);
         break;
       default:
         gold_unreachable();
       }

     if (pc_value == 0)
       {
         // This FDE applies to a discarded function.  We
         // can discard this FDE.
         object->add_merge_mapping(this, shndx, (pfde - 8) - pcontents,
                                   pfdeend - (pfde - 8), -1);
         return true;
       }

     // Otherwise, reject the FDE.
     return false;
   }

 unsigned int symndx = relocs->next_symndx();
 if (symndx == -1U)
   return false;

 // There can be another reloc in the FDE, if the CIE specifies an
 // LSDA (language specific data area).  We currently don't care.  We
 // will care later if we want to optimize the LSDA from an absolute
 // pointer to a PC relative offset when generating a shared library.
 relocs->advance(pfdeend - pcontents);

 // Find the section index for code that this FDE describes.
 // If we have discarded the section, we can also discard the FDE.
 unsigned int fde_shndx;
 const int sym_size = elfcpp::Elf_sizes<size>::sym_size;
 if (symndx >= symbols_size / sym_size)
   return false;
 elfcpp::Sym<size, big_endian> sym(symbols + symndx * sym_size);
 bool is_ordinary;
 fde_shndx = object->adjust_sym_shndx(symndx, sym.get_st_shndx(),
                                      &is_ordinary);
 bool is_discarded = (is_ordinary
                      && fde_shndx != elfcpp::SHN_UNDEF
                      && fde_shndx < object->shnum()
                      && !object->is_section_included(fde_shndx));

 // Fetch the address range field from the FDE. The offset and size
 // of the field depends on the PC encoding given in the CIE, but
 // it is always an absolute value. If the address range is 0, this
 // FDE corresponds to a function that was discarded during optimization
 // (too late to discard the corresponding FDE).
 uint64_t address_range = 0;
 switch (pc_size)
   {
   case 2:
     address_range = elfcpp::Swap<16, big_endian>::readval(pfde + 2);
     break;
   case 4:
     address_range = elfcpp::Swap<32, big_endian>::readval(pfde + 4);
     break;
   case 8:
     address_range = elfcpp::Swap_unaligned<64, big_endian>::readval(pfde + 8);
     break;
   default:
     gold_unreachable();
   }

 if (is_discarded || address_range == 0)
   {
     // This FDE applies to a discarded function.  We
     // can discard this FDE.
     object->add_merge_mapping(this, shndx, (pfde - 8) - pcontents,
                               pfdeend - (pfde - 8), -1);
     return true;
   }

 cie->add_fde(new Fde(object, shndx, (pfde - 8) - pcontents,
                      pfde, pfdeend - pfde));

 return true;
}

// Add unwind information for a PLT.

void
Eh_frame::add_ehframe_for_plt(Output_data* plt, const unsigned char* cie_data,
                             size_t cie_length, const unsigned char* fde_data,
                             size_t fde_length)
{
 Cie cie(NULL, 0, 0, elfcpp::DW_EH_PE_pcrel | elfcpp::DW_EH_PE_sdata4, "",
         cie_data, cie_length);
 Cie_offsets::iterator find_cie = this->cie_offsets_.find(&cie);
 Cie* pcie;
 if (find_cie != this->cie_offsets_.end())
   pcie = *find_cie;
 else
   {
     gold_assert(!this->mappings_are_done_);
     pcie = new Cie(cie);
     this->cie_offsets_.insert(pcie);
   }

 Fde* fde = new Fde(plt, fde_data, fde_length, this->mappings_are_done_);
 pcie->add_fde(fde);

 if (this->mappings_are_done_)
   this->final_data_size_ += align_address(fde_length + 8, this->addralign());
}

// Remove all post-map unwind information for a PLT.

void
Eh_frame::remove_ehframe_for_plt(Output_data* plt,
                                const unsigned char* cie_data,
                                size_t cie_length)
{
 if (!this->mappings_are_done_)
   return;

 Cie cie(NULL, 0, 0, elfcpp::DW_EH_PE_pcrel | elfcpp::DW_EH_PE_sdata4, "",
         cie_data, cie_length);
 Cie_offsets::iterator find_cie = this->cie_offsets_.find(&cie);
 gold_assert (find_cie != this->cie_offsets_.end());
 Cie* pcie = *find_cie;

 while (pcie->fde_count() != 0)
   {
     const Fde* fde = pcie->last_fde();
     if (!fde->post_map(plt))
       break;
     size_t length = fde->length();
     this->final_data_size_ -= align_address(length + 8, this->addralign());
     pcie->remove_fde();
   }
}

// Return the number of FDEs.

unsigned int
Eh_frame::fde_count() const
{
 unsigned int ret = 0;
 for (Unmergeable_cie_offsets::const_iterator p =
        this->unmergeable_cie_offsets_.begin();
      p != this->unmergeable_cie_offsets_.end();
      ++p)
   ret += (*p)->fde_count();
 for (Cie_offsets::const_iterator p = this->cie_offsets_.begin();
      p != this->cie_offsets_.end();
      ++p)
   ret += (*p)->fde_count();
 return ret;
}

// Set the final data size.

void
Eh_frame::set_final_data_size()
{
 // We can be called more than once if Layout::set_segment_offsets
 // finds a better mapping.  We don't want to add all the mappings
 // again.
 if (this->mappings_are_done_)
   {
     this->set_data_size(this->final_data_size_);
     return;
   }

 section_offset_type output_start = 0;
 if (this->is_offset_valid())
   output_start = this->offset() - this->output_section()->offset();
 section_offset_type output_offset = output_start;

 for (Unmergeable_cie_offsets::iterator p =
        this->unmergeable_cie_offsets_.begin();
      p != this->unmergeable_cie_offsets_.end();
      ++p)
   output_offset = (*p)->set_output_offset(output_offset,
                                           this->addralign(),
                                           this);

 for (Cie_offsets::iterator p = this->cie_offsets_.begin();
      p != this->cie_offsets_.end();
      ++p)
   output_offset = (*p)->set_output_offset(output_offset,
                                           this->addralign(),
                                           this);

 this->mappings_are_done_ = true;
 this->final_data_size_ = output_offset - output_start;

 gold_assert((output_offset & (this->addralign() - 1)) == 0);
 this->set_data_size(this->final_data_size_);
}

// Return an output offset for an input offset.

bool
Eh_frame::do_output_offset(const Relobj* object, unsigned int shndx,
                          section_offset_type offset,
                          section_offset_type* poutput) const
{
 return object->merge_output_offset(shndx, offset, poutput);
}

// Write the data to the output file.

void
Eh_frame::do_write(Output_file* of)
{
 const off_t offset = this->offset();
 const off_t oview_size = this->data_size();
 unsigned char* const oview = of->get_output_view(offset, oview_size);

 switch (parameters->size_and_endianness())
   {
#ifdef HAVE_TARGET_32_LITTLE
   case Parameters::TARGET_32_LITTLE:
     this->do_sized_write<32, false>(oview);
     break;
#endif
#ifdef HAVE_TARGET_32_BIG
   case Parameters::TARGET_32_BIG:
     this->do_sized_write<32, true>(oview);
     break;
#endif
#ifdef HAVE_TARGET_64_LITTLE
   case Parameters::TARGET_64_LITTLE:
     this->do_sized_write<64, false>(oview);
     break;
#endif
#ifdef HAVE_TARGET_64_BIG
   case Parameters::TARGET_64_BIG:
     this->do_sized_write<64, true>(oview);
     break;
#endif
   default:
     gold_unreachable();
   }

 of->write_output_view(offset, oview_size, oview);
}

// Write the data to the output file--template version.

template<int size, bool big_endian>
void
Eh_frame::do_sized_write(unsigned char* oview)
{
 uint64_t address = this->address();
 unsigned int addralign = this->addralign();
 section_offset_type o = 0;
 const off_t output_offset = this->offset() - this->output_section()->offset();
 Post_fdes post_fdes;
 for (Unmergeable_cie_offsets::iterator p =
        this->unmergeable_cie_offsets_.begin();
      p != this->unmergeable_cie_offsets_.end();
      ++p)
   o = (*p)->write<size, big_endian>(oview, output_offset, o, address,
                                     addralign, this->eh_frame_hdr_,
                                     &post_fdes);
 for (Cie_offsets::iterator p = this->cie_offsets_.begin();
      p != this->cie_offsets_.end();
      ++p)
   o = (*p)->write<size, big_endian>(oview, output_offset, o, address,
                                     addralign, this->eh_frame_hdr_,
                                     &post_fdes);
 for (Post_fdes::iterator p = post_fdes.begin();
      p != post_fdes.end();
      ++p)
   o = (*p).fde->write<size, big_endian>(oview, output_offset, o, address,
                                         addralign, (*p).cie_offset,
                                         (*p).fde_encoding,
                                         this->eh_frame_hdr_);
}

#ifdef HAVE_TARGET_32_LITTLE
template
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section<32, false>(
   Sized_relobj_file<32, false>* object,
   const unsigned char* symbols,
   section_size_type symbols_size,
   const unsigned char* symbol_names,
   section_size_type symbol_names_size,
   unsigned int shndx,
   unsigned int reloc_shndx,
   unsigned int reloc_type);
#endif

#ifdef HAVE_TARGET_32_BIG
template
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section<32, true>(
   Sized_relobj_file<32, true>* object,
   const unsigned char* symbols,
   section_size_type symbols_size,
   const unsigned char* symbol_names,
   section_size_type symbol_names_size,
   unsigned int shndx,
   unsigned int reloc_shndx,
   unsigned int reloc_type);
#endif

#ifdef HAVE_TARGET_64_LITTLE
template
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section<64, false>(
   Sized_relobj_file<64, false>* object,
   const unsigned char* symbols,
   section_size_type symbols_size,
   const unsigned char* symbol_names,
   section_size_type symbol_names_size,
   unsigned int shndx,
   unsigned int reloc_shndx,
   unsigned int reloc_type);
#endif

#ifdef HAVE_TARGET_64_BIG
template
Eh_frame::Eh_frame_section_disposition
Eh_frame::add_ehframe_input_section<64, true>(
   Sized_relobj_file<64, true>* object,
   const unsigned char* symbols,
   section_size_type symbols_size,
   const unsigned char* symbol_names,
   section_size_type symbol_names_size,
   unsigned int shndx,
   unsigned int reloc_shndx,
   unsigned int reloc_type);
#endif

} // End namespace gold.