// A reloc against a global symbol.
Output_reloc(Symbol* gsym, unsigned int type, Output_data* od,
- Address address, bool is_relative);
+ Address address, bool is_relative, bool is_symbolless);
Output_reloc(Symbol* gsym, unsigned int type,
Sized_relobj<size, big_endian>* relobj,
- unsigned int shndx, Address address, bool is_relative);
+ unsigned int shndx, Address address, bool is_relative,
+ bool is_symbolless);
// A reloc against a local symbol or local section symbol.
Output_reloc(Sized_relobj<size, big_endian>* relobj,
unsigned int local_sym_index, unsigned int type,
Output_data* od, Address address, bool is_relative,
- bool is_section_symbol);
+ bool is_symbolless, bool is_section_symbol);
Output_reloc(Sized_relobj<size, big_endian>* relobj,
unsigned int local_sym_index, unsigned int type,
unsigned int shndx, Address address, bool is_relative,
- bool is_section_symbol);
+ bool is_symbolless, bool is_section_symbol);
// A reloc against the STT_SECTION symbol of an output section.
is_relative() const
{ return this->is_relative_; }
+ // Return whether this is a relocation which should not use
+ // a symbol, but which obtains its addend from a symbol.
+ bool
+ is_symbolless() const
+ { return this->is_symbolless_; }
+
// Return whether this is against a local section symbol.
bool
is_local_section_symbol() const
// input file.
unsigned int local_sym_index_;
// The reloc type--a processor specific code.
- unsigned int type_ : 30;
+ unsigned int type_ : 29;
// True if the relocation is a RELATIVE relocation.
bool is_relative_ : 1;
+ // True if the relocation is one which should not use
+ // a symbol, but which obtains its addend from a symbol.
+ bool is_symbolless_ : 1;
// True if the relocation is against a section symbol.
bool is_section_symbol_ : 1;
// If the reloc address is an input section in an object, the
// A reloc against a global symbol.
Output_reloc(Symbol* gsym, unsigned int type, Output_data* od,
- Address address, Addend addend, bool is_relative)
- : rel_(gsym, type, od, address, is_relative), addend_(addend)
+ Address address, Addend addend, bool is_relative,
+ bool is_symbolless)
+ : rel_(gsym, type, od, address, is_relative, is_symbolless),
+ addend_(addend)
{ }
Output_reloc(Symbol* gsym, unsigned int type,
Sized_relobj<size, big_endian>* relobj,
unsigned int shndx, Address address, Addend addend,
- bool is_relative)
- : rel_(gsym, type, relobj, shndx, address, is_relative), addend_(addend)
+ bool is_relative, bool is_symbolless)
+ : rel_(gsym, type, relobj, shndx, address, is_relative,
+ is_symbolless), addend_(addend)
{ }
// A reloc against a local symbol.
Output_reloc(Sized_relobj<size, big_endian>* relobj,
unsigned int local_sym_index, unsigned int type,
Output_data* od, Address address,
- Addend addend, bool is_relative, bool is_section_symbol)
+ Addend addend, bool is_relative,
+ bool is_symbolless, bool is_section_symbol)
: rel_(relobj, local_sym_index, type, od, address, is_relative,
- is_section_symbol),
+ is_symbolless, is_section_symbol),
addend_(addend)
{ }
Output_reloc(Sized_relobj<size, big_endian>* relobj,
unsigned int local_sym_index, unsigned int type,
unsigned int shndx, Address address,
- Addend addend, bool is_relative, bool is_section_symbol)
+ Addend addend, bool is_relative,
+ bool is_symbolless, bool is_section_symbol)
: rel_(relobj, local_sym_index, type, shndx, address, is_relative,
- is_section_symbol),
+ is_symbolless, is_section_symbol),
addend_(addend)
{ }
is_relative() const
{ return this->rel_.is_relative(); }
+ // Return whether this is a relocation which should not use
+ // a symbol, but which obtains its addend from a symbol.
+ bool
+ is_symbolless() const
+ { return this->rel_.is_symbolless(); }
+
// Write the reloc entry to an output view.
void
write(unsigned char* pov) const;
void
add_global(Symbol* gsym, unsigned int type, Output_data* od, Address address)
- { this->add(od, Output_reloc_type(gsym, type, od, address, false)); }
+ { this->add(od, Output_reloc_type(gsym, type, od, address, false, false)); }
void
add_global(Symbol* gsym, unsigned int type, Output_data* od,
Sized_relobj<size, big_endian>* relobj,
unsigned int shndx, Address address)
{ this->add(od, Output_reloc_type(gsym, type, relobj, shndx, address,
- false)); }
+ false, false)); }
// These are to simplify the Copy_relocs class.
void
add_global_relative(Symbol* gsym, unsigned int type, Output_data* od,
Address address)
- { this->add(od, Output_reloc_type(gsym, type, od, address, true)); }
+ { this->add(od, Output_reloc_type(gsym, type, od, address, true, true)); }
void
add_global_relative(Symbol* gsym, unsigned int type, Output_data* od,
unsigned int shndx, Address address)
{
this->add(od, Output_reloc_type(gsym, type, relobj, shndx, address,
- true));
+ true, true));
+ }
+
+ // Add a global relocation which does not use a symbol for the relocation,
+ // but which gets its addend from a symbol.
+
+ void
+ add_symbolless_global_addend(Symbol* gsym, unsigned int type,
+ Output_data* od, Address address)
+ { this->add(od, Output_reloc_type(gsym, type, od, address, false, true)); }
+
+ void
+ add_symbolless_global_addend(Symbol* gsym, unsigned int type,
+ Output_data* od,
+ Sized_relobj<size, big_endian>* relobj,
+ unsigned int shndx, Address address)
+ {
+ this->add(od, Output_reloc_type(gsym, type, relobj, shndx, address,
+ false, true));
}
// Add a reloc against a local symbol.
Output_data* od, Address address)
{
this->add(od, Output_reloc_type(relobj, local_sym_index, type, od,
- address, false, false));
+ address, false, false, false));
}
void
Output_data* od, unsigned int shndx, Address address)
{
this->add(od, Output_reloc_type(relobj, local_sym_index, type, shndx,
- address, false, false));
+ address, false, false, false));
}
// Add a RELATIVE reloc against a local symbol.
Output_data* od, Address address)
{
this->add(od, Output_reloc_type(relobj, local_sym_index, type, od,
- address, true, false));
+ address, true, true, false));
}
void
Output_data* od, unsigned int shndx, Address address)
{
this->add(od, Output_reloc_type(relobj, local_sym_index, type, shndx,
- address, true, false));
+ address, true, true, false));
+ }
+
+ // Add a local relocation which does not use a symbol for the relocation,
+ // but which gets its addend from a symbol.
+
+ void
+ add_symbolless_local_addend(Sized_relobj<size, big_endian>* relobj,
+ unsigned int local_sym_index, unsigned int type,
+ Output_data* od, Address address)
+ {
+ this->add(od, Output_reloc_type(relobj, local_sym_index, type, od,
+ address, false, true, false));
+ }
+
+ void
+ add_symbolless_local_addend(Sized_relobj<size, big_endian>* relobj,
+ unsigned int local_sym_index, unsigned int type,
+ Output_data* od, unsigned int shndx,
+ Address address)
+ {
+ this->add(od, Output_reloc_type(relobj, local_sym_index, type, shndx,
+ address, false, true, false));
}
// Add a reloc against a local section symbol. This will be
Output_data* od, Address address)
{
this->add(od, Output_reloc_type(relobj, input_shndx, type, od,
- address, false, true));
+ address, false, false, true));
}
void
Output_data* od, unsigned int shndx, Address address)
{
this->add(od, Output_reloc_type(relobj, input_shndx, type, shndx,
- address, false, true));
+ address, false, false, true));
}
// A reloc against the STT_SECTION symbol of an output section.
add_global(Symbol* gsym, unsigned int type, Output_data* od,
Address address, Addend addend)
{ this->add(od, Output_reloc_type(gsym, type, od, address, addend,
- false)); }
+ false, false)); }
void
add_global(Symbol* gsym, unsigned int type, Output_data* od,
unsigned int shndx, Address address,
Addend addend)
{ this->add(od, Output_reloc_type(gsym, type, relobj, shndx, address,
- addend, false)); }
+ addend, false, false)); }
// Add a RELATIVE reloc against a global symbol. The final output
// relocation will not reference the symbol, but we must keep the symbol
void
add_global_relative(Symbol* gsym, unsigned int type, Output_data* od,
Address address, Addend addend)
- { this->add(od, Output_reloc_type(gsym, type, od, address, addend, true)); }
+ { this->add(od, Output_reloc_type(gsym, type, od, address, addend, true,
+ true)); }
void
add_global_relative(Symbol* gsym, unsigned int type, Output_data* od,
Sized_relobj<size, big_endian>* relobj,
unsigned int shndx, Address address, Addend addend)
{ this->add(od, Output_reloc_type(gsym, type, relobj, shndx, address,
- addend, true)); }
+ addend, true, true)); }
+
+ // Add a global relocation which does not use a symbol for the relocation,
+ // but which gets its addend from a symbol.
+
+ void
+ add_symbolless_global_addend(Symbol* gsym, unsigned int type, Output_data* od,
+ Address address, Addend addend)
+ { this->add(od, Output_reloc_type(gsym, type, od, address, addend,
+ false, true)); }
+
+ void
+ add_symbolless_global_addend(Symbol* gsym, unsigned int type,
+ Output_data* od,
+ Sized_relobj<size, big_endian>* relobj,
+ unsigned int shndx, Address address, Addend addend)
+ { this->add(od, Output_reloc_type(gsym, type, relobj, shndx, address,
+ addend, false, true)); }
// Add a reloc against a local symbol.
Output_data* od, Address address, Addend addend)
{
this->add(od, Output_reloc_type(relobj, local_sym_index, type, od, address,
- addend, false, false));
+ addend, false, false, false));
}
void
Addend addend)
{
this->add(od, Output_reloc_type(relobj, local_sym_index, type, shndx,
- address, addend, false, false));
+ address, addend, false, false, false));
}
// Add a RELATIVE reloc against a local symbol.
Output_data* od, Address address, Addend addend)
{
this->add(od, Output_reloc_type(relobj, local_sym_index, type, od, address,
- addend, true, false));
+ addend, true, true, false));
}
void
Addend addend)
{
this->add(od, Output_reloc_type(relobj, local_sym_index, type, shndx,
- address, addend, true, false));
+ address, addend, true, true, false));
+ }
+
+ // Add a local relocation which does not use a symbol for the relocation,
+ // but which gets it's addend from a symbol.
+
+ void
+ add_symbolless_local_addend(Sized_relobj<size, big_endian>* relobj,
+ unsigned int local_sym_index, unsigned int type,
+ Output_data* od, Address address, Addend addend)
+ {
+ this->add(od, Output_reloc_type(relobj, local_sym_index, type, od, address,
+ addend, false, true, false));
+ }
+
+ void
+ add_symbolless_local_addend(Sized_relobj<size, big_endian>* relobj,
+ unsigned int local_sym_index, unsigned int type,
+ Output_data* od, unsigned int shndx,
+ Address address, Addend addend)
+ {
+ this->add(od, Output_reloc_type(relobj, local_sym_index, type, shndx,
+ address, addend, false, true, false));
}
// Add a reloc against a local section symbol. This will be
Output_data* od, Address address, Addend addend)
{
this->add(od, Output_reloc_type(relobj, input_shndx, type, od, address,
- addend, false, true));
+ addend, false, false, true));
}
void
Addend addend)
{
this->add(od, Output_reloc_type(relobj, input_shndx, type, shndx,
- address, addend, false, true));
+ address, addend, false, false, true));
}
// A reloc against the STT_SECTION symbol of an output section.
unsigned int shndx_;
};
+// This class describes properties of merge data sections. It is used
+// as a key type for maps.
+class Merge_section_properties
+{
+ public:
+ Merge_section_properties(bool is_string, uint64_t entsize,
+ uint64_t addralign)
+ : is_string_(is_string), entsize_(entsize), addralign_(addralign)
+ { }
+
+ // Whether this equals to another Merge_section_properties MSP.
+ bool
+ eq(const Merge_section_properties& msp) const
+ {
+ return ((this->is_string_ == msp.is_string_)
+ && (this->entsize_ == msp.entsize_)
+ && (this->addralign_ == msp.addralign_));
+ }
+
+ // Compute a hash value for this using 64-bit FNV-1a hash.
+ size_t
+ hash_value() const
+ {
+ uint64_t h = 14695981039346656037ULL; // FNV offset basis.
+ uint64_t prime = 1099511628211ULL;
+ h = (h ^ static_cast<uint64_t>(this->is_string_)) * prime;
+ h = (h ^ static_cast<uint64_t>(this->entsize_)) * prime;
+ h = (h ^ static_cast<uint64_t>(this->addralign_)) * prime;
+ return h;
+ }
+
+ // Functors for associative containers.
+ struct equal_to
+ {
+ bool
+ operator()(const Merge_section_properties& msp1,
+ const Merge_section_properties& msp2) const
+ { return msp1.eq(msp2); }
+ };
+
+ struct hash
+ {
+ size_t
+ operator()(const Merge_section_properties& msp) const
+ { return msp.hash_value(); }
+ };
+
+ private:
+ // Whether this merge data section is for strings.
+ bool is_string_;
+ // Entsize of this merge data section.
+ uint64_t entsize_;
+ // Address alignment.
+ uint64_t addralign_;
+};
+
+// This class is used to speed up look up of special input sections in an
+// Output_section.
+
+class Output_section_lookup_maps
+{
+ public:
+ Output_section_lookup_maps()
+ : is_valid_(true), merge_sections_by_properties_(),
+ merge_sections_by_id_(), relaxed_input_sections_by_id_()
+ { }
+
+ // Whether the maps are valid.
+ bool
+ is_valid() const
+ { return this->is_valid_; }
+
+ // Invalidate the maps.
+ void
+ invalidate()
+ { this->is_valid_ = false; }
+
+ // Clear the maps.
+ void
+ clear()
+ {
+ this->merge_sections_by_properties_.clear();
+ this->merge_sections_by_id_.clear();
+ this->relaxed_input_sections_by_id_.clear();
+ // A cleared map is valid.
+ this->is_valid_ = true;
+ }
+
+ // Find a merge section by merge section properties. Return NULL if none
+ // is found.
+ Output_merge_base*
+ find_merge_section(const Merge_section_properties& msp) const
+ {
+ gold_assert(this->is_valid_);
+ Merge_sections_by_properties::const_iterator p =
+ this->merge_sections_by_properties_.find(msp);
+ return p != this->merge_sections_by_properties_.end() ? p->second : NULL;
+ }
+
+ // Find a merge section by section ID of a merge input section. Return NULL
+ // if none is found.
+ Output_merge_base*
+ find_merge_section(const Object* object, unsigned int shndx) const
+ {
+ gold_assert(this->is_valid_);
+ Merge_sections_by_id::const_iterator p =
+ this->merge_sections_by_id_.find(Const_section_id(object, shndx));
+ return p != this->merge_sections_by_id_.end() ? p->second : NULL;
+ }
+
+ // Add a merge section pointed by POMB with properties MSP.
+ void
+ add_merge_section(const Merge_section_properties& msp,
+ Output_merge_base* pomb)
+ {
+ std::pair<Merge_section_properties, Output_merge_base*> value(msp, pomb);
+ std::pair<Merge_sections_by_properties::iterator, bool> result =
+ this->merge_sections_by_properties_.insert(value);
+ gold_assert(value.second);
+ }
+
+ // Add a mapping from a merged input section in OBJECT with index SHNDX
+ // to a merge output section pointed by POMB.
+ void
+ add_merge_input_section(const Object* object, unsigned int shndx,
+ Output_merge_base* pomb)
+ {
+ Const_section_id csid(object, shndx);
+ std::pair<Const_section_id, Output_merge_base*> value(csid, pomb);
+ std::pair<Merge_sections_by_id::iterator, bool> result =
+ this->merge_sections_by_id_.insert(value);
+ gold_assert(value.second);
+ }
+
+ // Find a relaxed input section of OBJECT with index SHNDX.
+ Output_relaxed_input_section*
+ find_relaxed_input_section(const Object* object, unsigned int shndx) const
+ {
+ gold_assert(this->is_valid_);
+ Relaxed_input_sections_by_id::const_iterator p =
+ this->relaxed_input_sections_by_id_.find(Const_section_id(object, shndx));
+ return p != this->relaxed_input_sections_by_id_.end() ? p->second : NULL;
+ }
+
+ // Add a relaxed input section pointed by POMB and whose original input
+ // section is in OBJECT with index SHNDX.
+ void
+ add_relaxed_input_section(const Relobj* relobj, unsigned int shndx,
+ Output_relaxed_input_section* poris)
+ {
+ Const_section_id csid(relobj, shndx);
+ std::pair<Const_section_id, Output_relaxed_input_section*>
+ value(csid, poris);
+ std::pair<Relaxed_input_sections_by_id::iterator, bool> result =
+ this->relaxed_input_sections_by_id_.insert(value);
+ gold_assert(value.second);
+ }
+
+ private:
+ typedef Unordered_map<Const_section_id, Output_merge_base*,
+ Const_section_id_hash>
+ Merge_sections_by_id;
+
+ typedef Unordered_map<Merge_section_properties, Output_merge_base*,
+ Merge_section_properties::hash,
+ Merge_section_properties::equal_to>
+ Merge_sections_by_properties;
+
+ typedef Unordered_map<Const_section_id, Output_relaxed_input_section*,
+ Const_section_id_hash>
+ Relaxed_input_sections_by_id;
+
+ // Whether this is valid
+ bool is_valid_;
+ // Merge sections by merge section properties.
+ Merge_sections_by_properties merge_sections_by_properties_;
+ // Merge sections by section IDs.
+ Merge_sections_by_id merge_sections_by_id_;
+ // Relaxed sections by section IDs.
+ Relaxed_input_sections_by_id relaxed_input_sections_by_id_;
+};
+
// An output section. We don't expect to have too many output
// sections, so we don't bother to do a template on the size.
// within the output section.
template<int size, bool big_endian>
off_t
- add_input_section(Sized_relobj<size, big_endian>* object, unsigned int shndx,
- const char *name,
+ add_input_section(Layout* layout, Sized_relobj<size, big_endian>* object,
+ unsigned int shndx, const char *name,
const elfcpp::Shdr<size, big_endian>& shdr,
unsigned int reloc_shndx, bool have_sections_script);
set_may_sort_attached_input_sections()
{ this->may_sort_attached_input_sections_ = true; }
+ // Returns true if input sections must be sorted according to the
+ // order in which their name appear in the --section-ordering-file.
+ bool
+ input_section_order_specified()
+ { return this->input_section_order_specified_; }
+
+ // Record that input sections must be sorted as some of their names
+ // match the patterns specified through --section-ordering-file.
+ void
+ set_input_section_order_specified()
+ { this->input_section_order_specified_ = true; }
+
// Return whether the input sections attached to this output section
// require sorting. This is used to handle constructor priorities
// compatibly with GNU ld.
// The next few calls are for linker script support.
- // We need to export the input sections to linker scripts. Previously
- // we export a pair of Relobj pointer and section index. We now need to
- // handle relaxed input sections as well. So we use this class.
- class Simple_input_section
+ // In some cases we need to keep a list of the input sections
+ // associated with this output section. We only need the list if we
+ // might have to change the offsets of the input section within the
+ // output section after we add the input section. The ordinary
+ // input sections will be written out when we process the object
+ // file, and as such we don't need to track them here. We do need
+ // to track Output_section_data objects here. We store instances of
+ // this structure in a std::vector, so it must be a POD. There can
+ // be many instances of this structure, so we use a union to save
+ // some space.
+ class Input_section
{
- private:
- static const unsigned int invalid_shndx = static_cast<unsigned int>(-1);
-
public:
- Simple_input_section(Relobj *relobj, unsigned int shndx)
- : shndx_(shndx)
+ Input_section()
+ : shndx_(0), p2align_(0)
{
- gold_assert(shndx != invalid_shndx);
- this->u_.relobj = relobj;
+ this->u1_.data_size = 0;
+ this->u2_.object = NULL;
}
-
- Simple_input_section(Output_relaxed_input_section* section)
- : shndx_(invalid_shndx)
- { this->u_.relaxed_input_section = section; }
- // Whether this is a relaxed section.
- bool
- is_relaxed_input_section() const
- { return this->shndx_ == invalid_shndx; }
+ // For an ordinary input section.
+ Input_section(Relobj* object, unsigned int shndx, off_t data_size,
+ uint64_t addralign)
+ : shndx_(shndx),
+ p2align_(ffsll(static_cast<long long>(addralign))),
+ section_order_index_(0)
+ {
+ gold_assert(shndx != OUTPUT_SECTION_CODE
+ && shndx != MERGE_DATA_SECTION_CODE
+ && shndx != MERGE_STRING_SECTION_CODE
+ && shndx != RELAXED_INPUT_SECTION_CODE);
+ this->u1_.data_size = data_size;
+ this->u2_.object = object;
+ }
- // Return object of an input section.
- Relobj*
- relobj() const
+ // For a non-merge output section.
+ Input_section(Output_section_data* posd)
+ : shndx_(OUTPUT_SECTION_CODE), p2align_(0),
+ section_order_index_(0)
+ {
+ this->u1_.data_size = 0;
+ this->u2_.posd = posd;
+ }
+
+ // For a merge section.
+ Input_section(Output_section_data* posd, bool is_string, uint64_t entsize)
+ : shndx_(is_string
+ ? MERGE_STRING_SECTION_CODE
+ : MERGE_DATA_SECTION_CODE),
+ p2align_(0),
+ section_order_index_(0)
+ {
+ this->u1_.entsize = entsize;
+ this->u2_.posd = posd;
+ }
+
+ // For a relaxed input section.
+ Input_section(Output_relaxed_input_section *psection)
+ : shndx_(RELAXED_INPUT_SECTION_CODE), p2align_(0),
+ section_order_index_(0)
{
- return ((this->shndx_ != invalid_shndx)
- ? this->u_.relobj
- : this->u_.relaxed_input_section->relobj());
+ this->u1_.data_size = 0;
+ this->u2_.poris = psection;
}
- // Return index of an input section.
unsigned int
- shndx() const
+ section_order_index() const
{
- return ((this->shndx_ != invalid_shndx)
- ? this->shndx_
- : this->u_.relaxed_input_section->shndx());
+ return this->section_order_index_;
}
- // Return the Output_relaxed_input_section object of a relaxed section.
- Output_relaxed_input_section*
- relaxed_input_section() const
+ void
+ set_section_order_index(unsigned int number)
{
- gold_assert(this->shndx_ == invalid_shndx);
- return this->u_.relaxed_input_section;
+ this->section_order_index_ = number;
}
- private:
- // Pointer to either an Relobj or an Output_relaxed_input_section.
- union
+ // The required alignment.
+ uint64_t
+ addralign() const
{
- Relobj* relobj;
- Output_relaxed_input_section* relaxed_input_section;
- } u_;
- // Section index for an non-relaxed section or invalid_shndx for
- // a relaxed section.
- unsigned int shndx_;
- };
+ if (this->p2align_ != 0)
+ return static_cast<uint64_t>(1) << (this->p2align_ - 1);
+ else if (!this->is_input_section())
+ return this->u2_.posd->addralign();
+ else
+ return 0;
+ }
+
+ // Set the required alignment, which must be either 0 or a power of 2.
+ // For input sections that are sub-classes of Output_section_data, a
+ // alignment of zero means asking the underlying object for alignment.
+ void
+ set_addralign(uint64_t addralign)
+ {
+ if (addralign == 0)
+ this->p2align_ = 0;
+ else
+ {
+ gold_assert((addralign & (addralign - 1)) == 0);
+ this->p2align_ = ffsll(static_cast<long long>(addralign));
+ }
+ }
- // Store the list of input sections for this Output_section into the
- // list passed in. This removes the input sections, leaving only
- // any Output_section_data elements. This returns the size of those
- // Output_section_data elements. ADDRESS is the address of this
- // output section. FILL is the fill value to use, in case there are
- // any spaces between the remaining Output_section_data elements.
- uint64_t
- get_input_sections(uint64_t address, const std::string& fill,
- std::list<Simple_input_section>*);
+ // Return the required size.
+ off_t
+ data_size() const;
- // Add a simple input section.
- void
- add_simple_input_section(const Simple_input_section& input_section,
- off_t data_size, uint64_t addralign);
+ // Whether this is an input section.
+ bool
+ is_input_section() const
+ {
+ return (this->shndx_ != OUTPUT_SECTION_CODE
+ && this->shndx_ != MERGE_DATA_SECTION_CODE
+ && this->shndx_ != MERGE_STRING_SECTION_CODE
+ && this->shndx_ != RELAXED_INPUT_SECTION_CODE);
+ }
- // Set the current size of the output section.
- void
- set_current_data_size(off_t size)
- { this->set_current_data_size_for_child(size); }
+ // Return whether this is a merge section which matches the
+ // parameters.
+ bool
+ is_merge_section(bool is_string, uint64_t entsize,
+ uint64_t addralign) const
+ {
+ return (this->shndx_ == (is_string
+ ? MERGE_STRING_SECTION_CODE
+ : MERGE_DATA_SECTION_CODE)
+ && this->u1_.entsize == entsize
+ && this->addralign() == addralign);
+ }
- // Get the current size of the output section.
- off_t
- current_data_size() const
- { return this->current_data_size_for_child(); }
+ // Return whether this is a merge section for some input section.
+ bool
+ is_merge_section() const
+ {
+ return (this->shndx_ == MERGE_DATA_SECTION_CODE
+ || this->shndx_ == MERGE_STRING_SECTION_CODE);
+ }
- // End of linker script support.
+ // Return whether this is a relaxed input section.
+ bool
+ is_relaxed_input_section() const
+ { return this->shndx_ == RELAXED_INPUT_SECTION_CODE; }
- // Save states before doing section layout.
- // This is used for relaxation.
- void
- save_states();
+ // Return whether this is a generic Output_section_data.
+ bool
+ is_output_section_data() const
+ {
+ return this->shndx_ == OUTPUT_SECTION_CODE;
+ }
- // Restore states prior to section layout.
- void
- restore_states();
+ // Return the object for an input section.
+ Relobj*
+ relobj() const;
- // Discard states.
+ // Return the input section index for an input section.
+ unsigned int
+ shndx() const;
+
+ // For non-input-sections, return the associated Output_section_data
+ // object.
+ Output_section_data*
+ output_section_data() const
+ {
+ gold_assert(!this->is_input_section());
+ return this->u2_.posd;
+ }
+
+ // For a merge section, return the Output_merge_base pointer.
+ Output_merge_base*
+ output_merge_base() const
+ {
+ gold_assert(this->is_merge_section());
+ return this->u2_.pomb;
+ }
+
+ // Return the Output_relaxed_input_section object.
+ Output_relaxed_input_section*
+ relaxed_input_section() const
+ {
+ gold_assert(this->is_relaxed_input_section());
+ return this->u2_.poris;
+ }
+
+ // Set the output section.
+ void
+ set_output_section(Output_section* os)
+ {
+ gold_assert(!this->is_input_section());
+ Output_section_data *posd =
+ this->is_relaxed_input_section() ? this->u2_.poris : this->u2_.posd;
+ posd->set_output_section(os);
+ }
+
+ // Set the address and file offset. This is called during
+ // Layout::finalize. SECTION_FILE_OFFSET is the file offset of
+ // the enclosing section.
+ void
+ set_address_and_file_offset(uint64_t address, off_t file_offset,
+ off_t section_file_offset);
+
+ // Reset the address and file offset.
+ void
+ reset_address_and_file_offset();
+
+ // Finalize the data size.
+ void
+ finalize_data_size();
+
+ // Add an input section, for SHF_MERGE sections.
+ bool
+ add_input_section(Relobj* object, unsigned int shndx)
+ {
+ gold_assert(this->shndx_ == MERGE_DATA_SECTION_CODE
+ || this->shndx_ == MERGE_STRING_SECTION_CODE);
+ return this->u2_.posd->add_input_section(object, shndx);
+ }
+
+ // Given an input OBJECT, an input section index SHNDX within that
+ // object, and an OFFSET relative to the start of that input
+ // section, return whether or not the output offset is known. If
+ // this function returns true, it sets *POUTPUT to the offset in
+ // the output section, relative to the start of the input section
+ // in the output section. *POUTPUT may be different from OFFSET
+ // for a merged section.
+ bool
+ output_offset(const Relobj* object, unsigned int shndx,
+ section_offset_type offset,
+ section_offset_type *poutput) const;
+
+ // Return whether this is the merge section for the input section
+ // SHNDX in OBJECT.
+ bool
+ is_merge_section_for(const Relobj* object, unsigned int shndx) const;
+
+ // Write out the data. This does nothing for an input section.
+ void
+ write(Output_file*);
+
+ // Write the data to a buffer. This does nothing for an input
+ // section.
+ void
+ write_to_buffer(unsigned char*);
+
+ // Print to a map file.
+ void
+ print_to_mapfile(Mapfile*) const;
+
+ // Print statistics about merge sections to stderr.
+ void
+ print_merge_stats(const char* section_name)
+ {
+ if (this->shndx_ == MERGE_DATA_SECTION_CODE
+ || this->shndx_ == MERGE_STRING_SECTION_CODE)
+ this->u2_.posd->print_merge_stats(section_name);
+ }
+
+ private:
+ // Code values which appear in shndx_. If the value is not one of
+ // these codes, it is the input section index in the object file.
+ enum
+ {
+ // An Output_section_data.
+ OUTPUT_SECTION_CODE = -1U,
+ // An Output_section_data for an SHF_MERGE section with
+ // SHF_STRINGS not set.
+ MERGE_DATA_SECTION_CODE = -2U,
+ // An Output_section_data for an SHF_MERGE section with
+ // SHF_STRINGS set.
+ MERGE_STRING_SECTION_CODE = -3U,
+ // An Output_section_data for a relaxed input section.
+ RELAXED_INPUT_SECTION_CODE = -4U
+ };
+
+ // For an ordinary input section, this is the section index in the
+ // input file. For an Output_section_data, this is
+ // OUTPUT_SECTION_CODE or MERGE_DATA_SECTION_CODE or
+ // MERGE_STRING_SECTION_CODE.
+ unsigned int shndx_;
+ // The required alignment, stored as a power of 2.
+ unsigned int p2align_;
+ union
+ {
+ // For an ordinary input section, the section size.
+ off_t data_size;
+ // For OUTPUT_SECTION_CODE or RELAXED_INPUT_SECTION_CODE, this is not
+ // used. For MERGE_DATA_SECTION_CODE or MERGE_STRING_SECTION_CODE, the
+ // entity size.
+ uint64_t entsize;
+ } u1_;
+ union
+ {
+ // For an ordinary input section, the object which holds the
+ // input section.
+ Relobj* object;
+ // For OUTPUT_SECTION_CODE or MERGE_DATA_SECTION_CODE or
+ // MERGE_STRING_SECTION_CODE, the data.
+ Output_section_data* posd;
+ Output_merge_base* pomb;
+ // For RELAXED_INPUT_SECTION_CODE, the data.
+ Output_relaxed_input_section* poris;
+ } u2_;
+ // The line number of the pattern it matches in the --section-ordering-file
+ // file. It is 0 if does not match any pattern.
+ unsigned int section_order_index_;
+ };
+
+ // Store the list of input sections for this Output_section into the
+ // list passed in. This removes the input sections, leaving only
+ // any Output_section_data elements. This returns the size of those
+ // Output_section_data elements. ADDRESS is the address of this
+ // output section. FILL is the fill value to use, in case there are
+ // any spaces between the remaining Output_section_data elements.
+ uint64_t
+ get_input_sections(uint64_t address, const std::string& fill,
+ std::list<Input_section>*);
+
+ // Add a script input section. A script input section can either be
+ // a plain input section or a sub-class of Output_section_data.
+ void
+ add_script_input_section(const Input_section& input_section);
+
+ // Set the current size of the output section.
+ void
+ set_current_data_size(off_t size)
+ { this->set_current_data_size_for_child(size); }
+
+ // Get the current size of the output section.
+ off_t
+ current_data_size() const
+ { return this->current_data_size_for_child(); }
+
+ // End of linker script support.
+
+ // Save states before doing section layout.
+ // This is used for relaxation.
+ void
+ save_states();
+
+ // Restore states prior to section layout.
+ void
+ restore_states();
+
+ // Discard states.
void
discard_states();
void
adjust_section_offsets();
+ // Whether this is a NOLOAD section.
+ bool
+ is_noload() const
+ { return this->is_noload_; }
+
+ // Set NOLOAD flag.
+ void
+ set_is_noload()
+ { this->is_noload_ = true; }
+
// Print merge statistics to stderr.
void
print_merge_stats();
void
write_to_postprocessing_buffer();
- // In some cases we need to keep a list of the input sections
- // associated with this output section. We only need the list if we
- // might have to change the offsets of the input section within the
- // output section after we add the input section. The ordinary
- // input sections will be written out when we process the object
- // file, and as such we don't need to track them here. We do need
- // to track Output_section_data objects here. We store instances of
- // this structure in a std::vector, so it must be a POD. There can
- // be many instances of this structure, so we use a union to save
- // some space.
- class Input_section
- {
- public:
- Input_section()
- : shndx_(0), p2align_(0)
- {
- this->u1_.data_size = 0;
- this->u2_.object = NULL;
- }
-
- // For an ordinary input section.
- Input_section(Relobj* object, unsigned int shndx, off_t data_size,
- uint64_t addralign)
- : shndx_(shndx),
- p2align_(ffsll(static_cast<long long>(addralign)))
- {
- gold_assert(shndx != OUTPUT_SECTION_CODE
- && shndx != MERGE_DATA_SECTION_CODE
- && shndx != MERGE_STRING_SECTION_CODE
- && shndx != RELAXED_INPUT_SECTION_CODE);
- this->u1_.data_size = data_size;
- this->u2_.object = object;
- }
-
- // For a non-merge output section.
- Input_section(Output_section_data* posd)
- : shndx_(OUTPUT_SECTION_CODE), p2align_(0)
- {
- this->u1_.data_size = 0;
- this->u2_.posd = posd;
- }
-
- // For a merge section.
- Input_section(Output_section_data* posd, bool is_string, uint64_t entsize)
- : shndx_(is_string
- ? MERGE_STRING_SECTION_CODE
- : MERGE_DATA_SECTION_CODE),
- p2align_(0)
- {
- this->u1_.entsize = entsize;
- this->u2_.posd = posd;
- }
-
- // For a relaxed input section.
- Input_section(Output_relaxed_input_section *psection)
- : shndx_(RELAXED_INPUT_SECTION_CODE), p2align_(0)
- {
- this->u1_.data_size = 0;
- this->u2_.poris = psection;
- }
-
- // The required alignment.
- uint64_t
- addralign() const
- {
- if (!this->is_input_section())
- return this->u2_.posd->addralign();
- return (this->p2align_ == 0
- ? 0
- : static_cast<uint64_t>(1) << (this->p2align_ - 1));
- }
-
- // Return the required size.
- off_t
- data_size() const;
-
- // Whether this is an input section.
- bool
- is_input_section() const
- {
- return (this->shndx_ != OUTPUT_SECTION_CODE
- && this->shndx_ != MERGE_DATA_SECTION_CODE
- && this->shndx_ != MERGE_STRING_SECTION_CODE
- && this->shndx_ != RELAXED_INPUT_SECTION_CODE);
- }
-
- // Return whether this is a merge section which matches the
- // parameters.
- bool
- is_merge_section(bool is_string, uint64_t entsize,
- uint64_t addralign) const
- {
- return (this->shndx_ == (is_string
- ? MERGE_STRING_SECTION_CODE
- : MERGE_DATA_SECTION_CODE)
- && this->u1_.entsize == entsize
- && this->addralign() == addralign);
- }
-
- // Return whether this is a relaxed input section.
- bool
- is_relaxed_input_section() const
- { return this->shndx_ == RELAXED_INPUT_SECTION_CODE; }
-
- // Return whether this is a generic Output_section_data.
- bool
- is_output_section_data() const
- {
- return this->shndx_ == OUTPUT_SECTION_CODE;
- }
-
- // Return the object for an input section.
- Relobj*
- relobj() const
- {
- if (this->is_input_section())
- return this->u2_.object;
- else if (this->is_relaxed_input_section())
- return this->u2_.poris->relobj();
- else
- gold_unreachable();
- }
-
- // Return the input section index for an input section.
- unsigned int
- shndx() const
- {
- if (this->is_input_section())
- return this->shndx_;
- else if (this->is_relaxed_input_section())
- return this->u2_.poris->shndx();
- else
- gold_unreachable();
- }
-
- // For non-input-sections, return the associated Output_section_data
- // object.
- Output_section_data*
- output_section_data() const
- {
- gold_assert(!this->is_input_section());
- return this->u2_.posd;
- }
-
- // Return the Output_relaxed_input_section object.
- Output_relaxed_input_section*
- relaxed_input_section() const
- {
- gold_assert(this->is_relaxed_input_section());
- return this->u2_.poris;
- }
-
- // Set the output section.
- void
- set_output_section(Output_section* os)
- {
- gold_assert(!this->is_input_section());
- Output_section_data *posd =
- this->is_relaxed_input_section() ? this->u2_.poris : this->u2_.posd;
- posd->set_output_section(os);
- }
-
- // Set the address and file offset. This is called during
- // Layout::finalize. SECTION_FILE_OFFSET is the file offset of
- // the enclosing section.
- void
- set_address_and_file_offset(uint64_t address, off_t file_offset,
- off_t section_file_offset);
-
- // Reset the address and file offset.
- void
- reset_address_and_file_offset();
-
- // Finalize the data size.
- void
- finalize_data_size();
-
- // Add an input section, for SHF_MERGE sections.
- bool
- add_input_section(Relobj* object, unsigned int shndx)
- {
- gold_assert(this->shndx_ == MERGE_DATA_SECTION_CODE
- || this->shndx_ == MERGE_STRING_SECTION_CODE);
- return this->u2_.posd->add_input_section(object, shndx);
- }
-
- // Given an input OBJECT, an input section index SHNDX within that
- // object, and an OFFSET relative to the start of that input
- // section, return whether or not the output offset is known. If
- // this function returns true, it sets *POUTPUT to the offset in
- // the output section, relative to the start of the input section
- // in the output section. *POUTPUT may be different from OFFSET
- // for a merged section.
- bool
- output_offset(const Relobj* object, unsigned int shndx,
- section_offset_type offset,
- section_offset_type *poutput) const;
-
- // Return whether this is the merge section for the input section
- // SHNDX in OBJECT.
- bool
- is_merge_section_for(const Relobj* object, unsigned int shndx) const;
-
- // Write out the data. This does nothing for an input section.
- void
- write(Output_file*);
-
- // Write the data to a buffer. This does nothing for an input
- // section.
- void
- write_to_buffer(unsigned char*);
-
- // Print to a map file.
- void
- print_to_mapfile(Mapfile*) const;
-
- // Print statistics about merge sections to stderr.
- void
- print_merge_stats(const char* section_name)
- {
- if (this->shndx_ == MERGE_DATA_SECTION_CODE
- || this->shndx_ == MERGE_STRING_SECTION_CODE)
- this->u2_.posd->print_merge_stats(section_name);
- }
-
- private:
- // Code values which appear in shndx_. If the value is not one of
- // these codes, it is the input section index in the object file.
- enum
- {
- // An Output_section_data.
- OUTPUT_SECTION_CODE = -1U,
- // An Output_section_data for an SHF_MERGE section with
- // SHF_STRINGS not set.
- MERGE_DATA_SECTION_CODE = -2U,
- // An Output_section_data for an SHF_MERGE section with
- // SHF_STRINGS set.
- MERGE_STRING_SECTION_CODE = -3U,
- // An Output_section_data for a relaxed input section.
- RELAXED_INPUT_SECTION_CODE = -4U
- };
-
- // For an ordinary input section, this is the section index in the
- // input file. For an Output_section_data, this is
- // OUTPUT_SECTION_CODE or MERGE_DATA_SECTION_CODE or
- // MERGE_STRING_SECTION_CODE.
- unsigned int shndx_;
- // The required alignment, stored as a power of 2.
- unsigned int p2align_;
- union
- {
- // For an ordinary input section, the section size.
- off_t data_size;
- // For OUTPUT_SECTION_CODE or RELAXED_INPUT_SECTION_CODE, this is not
- // used. For MERGE_DATA_SECTION_CODE or MERGE_STRING_SECTION_CODE, the
- // entity size.
- uint64_t entsize;
- } u1_;
- union
- {
- // For an ordinary input section, the object which holds the
- // input section.
- Relobj* object;
- // For OUTPUT_SECTION_CODE or MERGE_DATA_SECTION_CODE or
- // MERGE_STRING_SECTION_CODE, the data.
- Output_section_data* posd;
- // For RELAXED_INPUT_SECTION_CODE, the data.
- Output_relaxed_input_section* poris;
- } u2_;
- };
-
typedef std::vector<Input_section> Input_section_list;
// Allow a child class to access the input sections.
// This class is used to sort the input sections.
class Input_section_sort_entry;
- // This is the sort comparison function.
+ // This is the sort comparison function for ctors and dtors.
struct Input_section_sort_compare
{
bool
const Input_section_sort_entry&) const;
};
+ // This is the sort comparison function for .init_array and .fini_array.
+ struct Input_section_sort_init_fini_compare
+ {
+ bool
+ operator()(const Input_section_sort_entry&,
+ const Input_section_sort_entry&) const;
+ };
+
+ // This is the sort comparison function when a section order is specified
+ // from an input file.
+ struct Input_section_sort_section_order_index_compare
+ {
+ bool
+ operator()(const Input_section_sort_entry&,
+ const Input_section_sort_entry&) const;
+ };
+
// Fill data. This is used to fill in data between input sections.
// It is also used for data statements (BYTE, WORD, etc.) in linker
// scripts. When we have to keep track of the input sections, we
typedef std::vector<Fill> Fill_list;
- // This class describes properties of merge data sections. It is used
- // as a key type for maps.
- class Merge_section_properties
- {
- public:
- Merge_section_properties(bool is_string, uint64_t entsize,
- uint64_t addralign)
- : is_string_(is_string), entsize_(entsize), addralign_(addralign)
- { }
-
- // Whether this equals to another Merge_section_properties MSP.
- bool
- eq(const Merge_section_properties& msp) const
- {
- return ((this->is_string_ == msp.is_string_)
- && (this->entsize_ == msp.entsize_)
- && (this->addralign_ == msp.addralign_));
- }
-
- // Compute a hash value for this using 64-bit FNV-1a hash.
- size_t
- hash_value() const
- {
- uint64_t h = 14695981039346656037ULL; // FNV offset basis.
- uint64_t prime = 1099511628211ULL;
- h = (h ^ static_cast<uint64_t>(this->is_string_)) * prime;
- h = (h ^ static_cast<uint64_t>(this->entsize_)) * prime;
- h = (h ^ static_cast<uint64_t>(this->addralign_)) * prime;
- return h;
- }
-
- // Functors for associative containers.
- struct equal_to
- {
- bool
- operator()(const Merge_section_properties& msp1,
- const Merge_section_properties& msp2) const
- { return msp1.eq(msp2); }
- };
-
- struct hash
- {
- size_t
- operator()(const Merge_section_properties& msp) const
- { return msp.hash_value(); }
- };
-
- private:
- // Whether this merge data section is for strings.
- bool is_string_;
- // Entsize of this merge data section.
- uint64_t entsize_;
- // Address alignment.
- uint64_t addralign_;
- };
-
- // Map that link Merge_section_properties to Output_merge_base.
- typedef Unordered_map<Merge_section_properties, Output_merge_base*,
- Merge_section_properties::hash,
- Merge_section_properties::equal_to>
- Merge_section_by_properties_map;
-
- // Map that link Const_section_id to Output_section_data.
- typedef Unordered_map<Const_section_id, Output_section_data*,
- Const_section_id_hash>
- Output_section_data_by_input_section_map;
-
- // Map that link Const_section_id to Output_relaxed_input_section.
- typedef Unordered_map<Const_section_id, Output_relaxed_input_section*,
- Const_section_id_hash>
- Output_relaxed_input_section_by_input_section_map;
-
// Map used during relaxation of existing sections. This map
// a section id an input section list index. We assume that
// Input_section_list is a vector.
add_output_section_data(Input_section*);
// Add an SHF_MERGE input section. Returns true if the section was
- // handled.
+ // handled. If KEEPS_INPUT_SECTIONS is true, the output merge section
+ // stores information about the merged input sections.
bool
add_merge_input_section(Relobj* object, unsigned int shndx, uint64_t flags,
- uint64_t entsize, uint64_t addralign);
+ uint64_t entsize, uint64_t addralign,
+ bool keeps_input_sections);
// Add an output SHF_MERGE section POSD to this output section.
// IS_STRING indicates whether it is a SHF_STRINGS section, and
const Relaxation_map& map,
Input_section_list* input_sections);
+ // Build the lookup maps for merge and relaxed input sections.
+ void
+ build_lookup_maps() const;
+
// Most of these fields are only valid after layout.
// The name of the section. This will point into a Stringpool.
// section, false if it means the symbol index of the corresponding
// section symbol.
bool info_uses_section_index_ : 1;
+ // True if input sections attached to this output section have to be
+ // sorted according to a specified order.
+ bool input_section_order_specified_ : 1;
// True if the input sections attached to this output section may
// need sorting.
bool may_sort_attached_input_sections_ : 1;
bool is_entsize_zero_ : 1;
// Whether section offsets need adjustment due to relaxation.
bool section_offsets_need_adjustment_ : 1;
+ // Whether this is a NOLOAD section.
+ bool is_noload_ : 1;
// For SHT_TLS sections, the offset of this section relative to the base
// of the TLS segment.
uint64_t tls_offset_;
// Saved checkpoint.
Checkpoint_output_section* checkpoint_;
- // Map from input sections to merge sections.
- Output_section_data_by_input_section_map merge_section_map_;
- // Map from merge section properties to merge_sections;
- Merge_section_by_properties_map merge_section_by_properties_map_;
- // Map from input sections to relaxed input sections. This is mutable
- // because it is updated lazily. We may need to update it in a
- // const qualified method.
- mutable Output_relaxed_input_section_by_input_section_map
- relaxed_input_section_map_;
- // Whether relaxed_input_section_map_ is valid.
- mutable bool is_relaxed_input_section_map_valid_;
+ // Fast lookup maps for merged and relaxed input sections.
+ Output_section_lookup_maps* lookup_maps_;
};
// An output segment. PT_LOAD segments are built from collections of
// upward based on the section alignments.
void
set_minimum_p_align(uint64_t align)
- { this->min_p_align_ = align; }
+ {
+ if (align > this->min_p_align_)
+ this->min_p_align_ = align;
+ }
// Set the offset of this segment based on the section. This should
// only be called for a non-PT_LOAD segment.