// layout.h -- lay out output file sections for gold -*- C++ -*-
-// Copyright 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
+// Copyright (C) 2006-2016 Free Software Foundation, Inc.
// Written by Ian Lance Taylor <iant@google.com>.
// This file is part of gold.
#include <cstring>
#include <list>
+#include <map>
#include <string>
#include <utility>
#include <vector>
{
class General_options;
+class Incremental_inputs;
+class Incremental_binary;
class Input_objects;
class Mapfile;
class Symbol_table;
class Output_section_data;
class Output_section;
class Output_section_headers;
+class Output_segment_headers;
+class Output_file_header;
class Output_segment;
class Output_data;
+class Output_data_reloc_generic;
class Output_data_dynamic;
class Output_symtab_xindex;
class Output_reduced_debug_abbrev_section;
class Output_reduced_debug_info_section;
class Eh_frame;
+class Gdb_index;
class Target;
+struct Timespec;
+
+// Return TRUE if SECNAME is the name of a compressed debug section.
+extern bool
+is_compressed_debug_section(const char* secname);
+
+// Return the name of the corresponding uncompressed debug section.
+extern std::string
+corresponding_uncompressed_section_name(std::string secname);
+
+// Maintain a list of free space within a section, segment, or file.
+// Used for incremental update links.
+
+class Free_list
+{
+ public:
+ struct Free_list_node
+ {
+ Free_list_node(off_t start, off_t end)
+ : start_(start), end_(end)
+ { }
+ off_t start_;
+ off_t end_;
+ };
+ typedef std::list<Free_list_node>::const_iterator Const_iterator;
+
+ Free_list()
+ : list_(), last_remove_(list_.begin()), extend_(false), length_(0),
+ min_hole_(0)
+ { }
+
+ // Initialize the free list for a section of length LEN.
+ // If EXTEND is true, free space may be allocated past the end.
+ void
+ init(off_t len, bool extend);
+
+ // Set the minimum hole size that is allowed when allocating
+ // from the free list.
+ void
+ set_min_hole_size(off_t min_hole)
+ { this->min_hole_ = min_hole; }
+
+ // Remove a chunk from the free list.
+ void
+ remove(off_t start, off_t end);
+
+ // Allocate a chunk of space from the free list of length LEN,
+ // with alignment ALIGN, and minimum offset MINOFF.
+ off_t
+ allocate(off_t len, uint64_t align, off_t minoff);
+
+ // Return an iterator for the beginning of the free list.
+ Const_iterator
+ begin() const
+ { return this->list_.begin(); }
+
+ // Return an iterator for the end of the free list.
+ Const_iterator
+ end() const
+ { return this->list_.end(); }
+
+ // Dump the free list (for debugging).
+ void
+ dump();
+
+ // Print usage statistics.
+ static void
+ print_stats();
+
+ private:
+ typedef std::list<Free_list_node>::iterator Iterator;
+
+ // The free list.
+ std::list<Free_list_node> list_;
+
+ // The last node visited during a remove operation.
+ Iterator last_remove_;
+
+ // Whether we can extend past the original length.
+ bool extend_;
+
+ // The total length of the section, segment, or file.
+ off_t length_;
+
+ // The minimum hole size allowed. When allocating from the free list,
+ // we must not leave a hole smaller than this.
+ off_t min_hole_;
+
+ // Statistics:
+ // The total number of free lists used.
+ static unsigned int num_lists;
+ // The total number of free list nodes used.
+ static unsigned int num_nodes;
+ // The total number of calls to Free_list::remove.
+ static unsigned int num_removes;
+ // The total number of nodes visited during calls to Free_list::remove.
+ static unsigned int num_remove_visits;
+ // The total number of calls to Free_list::allocate.
+ static unsigned int num_allocates;
+ // The total number of nodes visited during calls to Free_list::allocate.
+ static unsigned int num_allocate_visits;
+};
// This task function handles mapping the input sections to output
// sections and laying them out in memory.
Layout_task_runner(const General_options& options,
const Input_objects* input_objects,
Symbol_table* symtab,
- Target* target,
+ Target* target,
Layout* layout,
Mapfile* mapfile)
: options_(options), input_objects_(input_objects), symtab_(symtab),
Mapfile* mapfile_;
};
-// This struct holds information about the comdat or .gnu.linkonce
-// that will be kept.
+// This class holds information about the comdat group or
+// .gnu.linkonce section that will be kept for a given signature.
-struct Kept_section
+class Kept_section
{
+ private:
+ // For a comdat group, we build a mapping from the name of each
+ // section in the group to the section index and the size in object.
+ // When we discard a group in some other object file, we use this
+ // map to figure out which kept section the discarded section is
+ // associated with. We then use that mapping when processing relocs
+ // against discarded sections.
+ struct Comdat_section_info
+ {
+ // The section index.
+ unsigned int shndx;
+ // The section size.
+ uint64_t size;
+
+ Comdat_section_info(unsigned int a_shndx, uint64_t a_size)
+ : shndx(a_shndx), size(a_size)
+ { }
+ };
+
+ // Most comdat groups have only one or two sections, so we use a
+ // std::map rather than an Unordered_map to optimize for that case
+ // without paying too heavily for groups with more sections.
+ typedef std::map<std::string, Comdat_section_info> Comdat_group;
+
+ public:
Kept_section()
- : object(NULL), shndx(0), is_group(false), group_sections(NULL)
- { }
- Kept_section(Relobj* a_object, unsigned int a_shndx, bool a_is_group)
- : object(a_object), shndx(a_shndx), is_group(a_is_group),
- group_sections(NULL)
- { }
+ : object_(NULL), shndx_(0), is_comdat_(false), is_group_name_(false)
+ { this->u_.linkonce_size = 0; }
+
+ // We need to support copies for the signature map in the Layout
+ // object, but we should never copy an object after it has been
+ // marked as a comdat section.
+ Kept_section(const Kept_section& k)
+ : object_(k.object_), shndx_(k.shndx_), is_comdat_(false),
+ is_group_name_(k.is_group_name_)
+ {
+ gold_assert(!k.is_comdat_);
+ this->u_.linkonce_size = 0;
+ }
- typedef Unordered_map<std::string, unsigned int> Comdat_group;
+ ~Kept_section()
+ {
+ if (this->is_comdat_)
+ delete this->u_.group_sections;
+ }
+
+ // The object where this section lives.
+ Relobj*
+ object() const
+ { return this->object_; }
+
+ // Set the object.
+ void
+ set_object(Relobj* object)
+ {
+ gold_assert(this->object_ == NULL);
+ this->object_ = object;
+ }
- // The object containing the comdat or .gnu.linkonce.
- Relobj* object;
- // Index to the group section for comdats and the section itself for
+ // The section index.
+ unsigned int
+ shndx() const
+ { return this->shndx_; }
+
+ // Set the section index.
+ void
+ set_shndx(unsigned int shndx)
+ {
+ gold_assert(this->shndx_ == 0);
+ this->shndx_ = shndx;
+ }
+
+ // Whether this is a comdat group.
+ bool
+ is_comdat() const
+ { return this->is_comdat_; }
+
+ // Set that this is a comdat group.
+ void
+ set_is_comdat()
+ {
+ gold_assert(!this->is_comdat_);
+ this->is_comdat_ = true;
+ this->u_.group_sections = new Comdat_group();
+ }
+
+ // Whether this is associated with the name of a group or section
+ // rather than the symbol name derived from a linkonce section.
+ bool
+ is_group_name() const
+ { return this->is_group_name_; }
+
+ // Note that this represents a comdat group rather than a single
+ // linkonce section.
+ void
+ set_is_group_name()
+ { this->is_group_name_ = true; }
+
+ // Add a section to the group list.
+ void
+ add_comdat_section(const std::string& name, unsigned int shndx,
+ uint64_t size)
+ {
+ gold_assert(this->is_comdat_);
+ Comdat_section_info sinfo(shndx, size);
+ this->u_.group_sections->insert(std::make_pair(name, sinfo));
+ }
+
+ // Look for a section name in the group list, and return whether it
+ // was found. If found, returns the section index and size.
+ bool
+ find_comdat_section(const std::string& name, unsigned int* pshndx,
+ uint64_t* psize) const
+ {
+ gold_assert(this->is_comdat_);
+ Comdat_group::const_iterator p = this->u_.group_sections->find(name);
+ if (p == this->u_.group_sections->end())
+ return false;
+ *pshndx = p->second.shndx;
+ *psize = p->second.size;
+ return true;
+ }
+
+ // If there is only one section in the group list, return true, and
+ // return the section index and size.
+ bool
+ find_single_comdat_section(unsigned int* pshndx, uint64_t* psize) const
+ {
+ gold_assert(this->is_comdat_);
+ if (this->u_.group_sections->size() != 1)
+ return false;
+ Comdat_group::const_iterator p = this->u_.group_sections->begin();
+ *pshndx = p->second.shndx;
+ *psize = p->second.size;
+ return true;
+ }
+
+ // Return the size of a linkonce section.
+ uint64_t
+ linkonce_size() const
+ {
+ gold_assert(!this->is_comdat_);
+ return this->u_.linkonce_size;
+ }
+
+ // Set the size of a linkonce section.
+ void
+ set_linkonce_size(uint64_t size)
+ {
+ gold_assert(!this->is_comdat_);
+ this->u_.linkonce_size = size;
+ }
+
+ private:
+ // No assignment.
+ Kept_section& operator=(const Kept_section&);
+
+ // The object containing the comdat group or .gnu.linkonce section.
+ Relobj* object_;
+ // Index of the group section for comdats and the section itself for
// .gnu.linkonce.
- unsigned int shndx;
+ unsigned int shndx_;
+ // True if this is for a comdat group rather than a .gnu.linkonce
+ // section.
+ bool is_comdat_;
// The Kept_sections are values of a mapping, that maps names to
// them. This field is true if this struct is associated with the
// name of a comdat or .gnu.linkonce, false if it is associated with
// the name of a symbol obtained from the .gnu.linkonce.* name
// through some heuristics.
- bool is_group;
- // For comdats, a map from names of the sections in the group to
- // indexes in OBJECT_. NULL for .gnu.linkonce.
- Comdat_group* group_sections;
+ bool is_group_name_;
+ union
+ {
+ // If the is_comdat_ field is true, this holds a map from names of
+ // the sections in the group to section indexes in object_ and to
+ // section sizes.
+ Comdat_group* group_sections;
+ // If the is_comdat_ field is false, this holds the size of the
+ // single section.
+ uint64_t linkonce_size;
+ } u_;
+};
+
+// The ordering for output sections. This controls how output
+// sections are ordered within a PT_LOAD output segment.
+
+enum Output_section_order
+{
+ // Unspecified. Used for non-load segments. Also used for the file
+ // and segment headers.
+ ORDER_INVALID,
+
+ // The PT_INTERP section should come first, so that the dynamic
+ // linker can pick it up quickly.
+ ORDER_INTERP,
+
+ // Loadable read-only note sections come next so that the PT_NOTE
+ // segment is on the first page of the executable.
+ ORDER_RO_NOTE,
+
+ // Put read-only sections used by the dynamic linker early in the
+ // executable to minimize paging.
+ ORDER_DYNAMIC_LINKER,
+
+ // Put reloc sections used by the dynamic linker after other
+ // sections used by the dynamic linker; otherwise, objcopy and strip
+ // get confused.
+ ORDER_DYNAMIC_RELOCS,
+
+ // Put the PLT reloc section after the other dynamic relocs;
+ // otherwise, prelink gets confused.
+ ORDER_DYNAMIC_PLT_RELOCS,
+
+ // The .init section.
+ ORDER_INIT,
+
+ // The PLT.
+ ORDER_PLT,
+
+ // The regular text sections.
+ ORDER_TEXT,
+
+ // The .fini section.
+ ORDER_FINI,
+
+ // The read-only sections.
+ ORDER_READONLY,
+
+ // The exception frame sections.
+ ORDER_EHFRAME,
+
+ // The TLS sections come first in the data section.
+ ORDER_TLS_DATA,
+ ORDER_TLS_BSS,
+
+ // Local RELRO (read-only after relocation) sections come before
+ // non-local RELRO sections. This data will be fully resolved by
+ // the prelinker.
+ ORDER_RELRO_LOCAL,
+
+ // Non-local RELRO sections are grouped together after local RELRO
+ // sections. All RELRO sections must be adjacent so that they can
+ // all be put into a PT_GNU_RELRO segment.
+ ORDER_RELRO,
+
+ // We permit marking exactly one output section as the last RELRO
+ // section. We do this so that the read-only GOT can be adjacent to
+ // the writable GOT.
+ ORDER_RELRO_LAST,
+
+ // Similarly, we permit marking exactly one output section as the
+ // first non-RELRO section.
+ ORDER_NON_RELRO_FIRST,
+
+ // The regular data sections come after the RELRO sections.
+ ORDER_DATA,
+
+ // Large data sections normally go in large data segments.
+ ORDER_LARGE_DATA,
+
+ // Group writable notes so that we can have a single PT_NOTE
+ // segment.
+ ORDER_RW_NOTE,
+
+ // The small data sections must be at the end of the data sections,
+ // so that they can be adjacent to the small BSS sections.
+ ORDER_SMALL_DATA,
+
+ // The BSS sections start here.
+
+ // The small BSS sections must be at the start of the BSS sections,
+ // so that they can be adjacent to the small data sections.
+ ORDER_SMALL_BSS,
+
+ // The regular BSS sections.
+ ORDER_BSS,
+
+ // The large BSS sections come after the other BSS sections.
+ ORDER_LARGE_BSS,
+
+ // Maximum value.
+ ORDER_MAX
};
// This class handles the details of laying out input sections.
public:
Layout(int number_of_input_files, Script_options*);
+ ~Layout()
+ {
+ delete this->relaxation_debug_check_;
+ delete this->segment_states_;
+ }
+
+ // For incremental links, record the base file to be modified.
+ void
+ set_incremental_base(Incremental_binary* base);
+
+ Incremental_binary*
+ incremental_base()
+ { return this->incremental_base_; }
+
+ // For incremental links, record the initial fixed layout of a section
+ // from the base file, and return a pointer to the Output_section.
+ template<int size, bool big_endian>
+ Output_section*
+ init_fixed_output_section(const char*, elfcpp::Shdr<size, big_endian>&);
+
// Given an input section SHNDX, named NAME, with data in SHDR, from
// the object file OBJECT, return the output section where this
// input section should go. RELOC_SHNDX is the index of a
// within the output section.
template<int size, bool big_endian>
Output_section*
- layout(Sized_relobj<size, big_endian> *object, unsigned int shndx,
+ layout(Sized_relobj_file<size, big_endian> *object, unsigned int shndx,
const char* name, const elfcpp::Shdr<size, big_endian>& shdr,
unsigned int reloc_shndx, unsigned int reloc_type, off_t* offset);
+ std::map<Section_id, unsigned int>*
+ get_section_order_map()
+ { return &this->section_order_map_; }
+
+ // Struct to store segment info when mapping some input sections to
+ // unique segments using linker plugins. Mapping an input section to
+ // a unique segment is done by first placing such input sections in
+ // unique output sections and then mapping the output section to a
+ // unique segment. NAME is the name of the output section. FLAGS
+ // and ALIGN are the extra flags and alignment of the segment.
+ struct Unique_segment_info
+ {
+ // Identifier for the segment. ELF segments dont have names. This
+ // is used as the name of the output section mapped to the segment.
+ const char* name;
+ // Additional segment flags.
+ uint64_t flags;
+ // Segment alignment.
+ uint64_t align;
+ };
+
+ // Mapping from input section to segment.
+ typedef std::map<Const_section_id, Unique_segment_info*>
+ Section_segment_map;
+
+ // Maps section SECN to SEGMENT s.
+ void
+ insert_section_segment_map(Const_section_id secn, Unique_segment_info *s);
+
+ // Some input sections require special ordering, for compatibility
+ // with GNU ld. Given the name of an input section, return -1 if it
+ // does not require special ordering. Otherwise, return the index
+ // by which it should be ordered compared to other input sections
+ // that require special ordering.
+ static int
+ special_ordering_of_input_section(const char* name);
+
+ bool
+ is_section_ordering_specified()
+ { return this->section_ordering_specified_; }
+
+ void
+ set_section_ordering_specified()
+ { this->section_ordering_specified_ = true; }
+
+ bool
+ is_unique_segment_for_sections_specified() const
+ { return this->unique_segment_for_sections_specified_; }
+
+ void
+ set_unique_segment_for_sections_specified()
+ { this->unique_segment_for_sections_specified_ = true; }
+
+ // For incremental updates, allocate a block of memory from the
+ // free list. Find a block starting at or after MINOFF.
+ off_t
+ allocate(off_t len, uint64_t align, off_t minoff)
+ { return this->free_list_.allocate(len, align, minoff); }
+
+ unsigned int
+ find_section_order_index(const std::string&);
+
+ // Read the sequence of input sections from the file specified with
+ // linker option --section-ordering-file.
+ void
+ read_layout_from_file();
+
// Layout an input reloc section when doing a relocatable link. The
// section is RELOC_SHNDX in OBJECT, with data in SHDR.
// DATA_SECTION is the reloc section to which it refers. RR is the
// relocatable information.
template<int size, bool big_endian>
Output_section*
- layout_reloc(Sized_relobj<size, big_endian>* object,
+ layout_reloc(Sized_relobj_file<size, big_endian>* object,
unsigned int reloc_shndx,
const elfcpp::Shdr<size, big_endian>& shdr,
Output_section* data_section,
template<int size, bool big_endian>
void
layout_group(Symbol_table* symtab,
- Sized_relobj<size, big_endian>* object,
+ Sized_relobj_file<size, big_endian>* object,
unsigned int group_shndx,
const char* group_section_name,
const char* signature,
// returns the output section, and sets *OFFSET to the offset.
template<int size, bool big_endian>
Output_section*
- layout_eh_frame(Sized_relobj<size, big_endian>* object,
+ layout_eh_frame(Sized_relobj_file<size, big_endian>* object,
const unsigned char* symbols,
off_t symbols_size,
const unsigned char* symbol_names,
unsigned int reloc_shndx, unsigned int reloc_type,
off_t* offset);
+ // After processing all input files, we call this to make sure that
+ // the optimized .eh_frame sections have been added to the output
+ // section.
+ void
+ finalize_eh_frame_section();
+
+ // Add .eh_frame information for a PLT. The FDE must start with a
+ // 4-byte PC-relative reference to the start of the PLT, followed by
+ // a 4-byte size of PLT.
+ void
+ add_eh_frame_for_plt(Output_data* plt, const unsigned char* cie_data,
+ size_t cie_length, const unsigned char* fde_data,
+ size_t fde_length);
+
+ // Scan a .debug_info or .debug_types section, and add summary
+ // information to the .gdb_index section.
+ template<int size, bool big_endian>
+ void
+ add_to_gdb_index(bool is_type_unit,
+ Sized_relobj<size, big_endian>* object,
+ const unsigned char* symbols,
+ off_t symbols_size,
+ unsigned int shndx,
+ unsigned int reloc_shndx,
+ unsigned int reloc_type);
+
// Handle a GNU stack note. This is called once per input object
// file. SEEN_GNU_STACK is true if the object file has a
// .note.GNU-stack section. GNU_STACK_FLAGS is the section flags
// from that section if there was one.
void
- layout_gnu_stack(bool seen_gnu_stack, uint64_t gnu_stack_flags);
+ layout_gnu_stack(bool seen_gnu_stack, uint64_t gnu_stack_flags,
+ const Object*);
// Add an Output_section_data to the layout. This is used for
- // special sections like the GOT section.
+ // special sections like the GOT section. ORDER is where the
+ // section should wind up in the output segment. IS_RELRO is true
+ // for relro sections.
Output_section*
add_output_section_data(const char* name, elfcpp::Elf_Word type,
elfcpp::Elf_Xword flags,
- Output_section_data*);
+ Output_section_data*, Output_section_order order,
+ bool is_relro);
+
+ // Increase the size of the relro segment by this much.
+ void
+ increase_relro(unsigned int s)
+ { this->increase_relro_ += s; }
// Create dynamic sections if necessary.
void
void
define_section_symbols(Symbol_table*);
+ // Create automatic note sections.
+ void
+ create_notes();
+
// Create sections for linker scripts.
void
create_script_sections()
dynpool() const
{ return &this->dynpool_; }
+ // Return the .dynamic output section. This is only valid after the
+ // layout has been finalized.
+ Output_section*
+ dynamic_section() const
+ { return this->dynamic_section_; }
+
// Return the symtab_xindex section used to hold large section
// indexes for the normal symbol table.
Output_symtab_xindex*
is_linkonce(const char* name)
{ return strncmp(name, ".gnu.linkonce", sizeof(".gnu.linkonce") - 1) == 0; }
+ // Whether we have added an input section.
+ bool
+ have_added_input_section() const
+ { return this->have_added_input_section_; }
+
// Return true if a section is a debugging section.
static inline bool
is_debug_info_section(const char* name)
{
// Debugging sections can only be recognized by name.
return (strncmp(name, ".debug", sizeof(".debug") - 1) == 0
- || strncmp(name, ".gnu.linkonce.wi.",
- sizeof(".gnu.linkonce.wi.") - 1) == 0
- || strncmp(name, ".line", sizeof(".line") - 1) == 0
- || strncmp(name, ".stab", sizeof(".stab") - 1) == 0);
+ || strncmp(name, ".zdebug", sizeof(".zdebug") - 1) == 0
+ || strncmp(name, ".gnu.linkonce.wi.",
+ sizeof(".gnu.linkonce.wi.") - 1) == 0
+ || strncmp(name, ".line", sizeof(".line") - 1) == 0
+ || strncmp(name, ".stab", sizeof(".stab") - 1) == 0
+ || strncmp(name, ".pdr", sizeof(".pdr") - 1) == 0);
}
+ // Return true if RELOBJ is an input file whose base name matches
+ // FILE_NAME. The base name must have an extension of ".o", and
+ // must be exactly FILE_NAME.o or FILE_NAME, one character, ".o".
+ static bool
+ match_file_name(const Relobj* relobj, const char* file_name);
+
+ // Return whether section SHNDX in RELOBJ is a .ctors/.dtors section
+ // with more than one word being mapped to a .init_array/.fini_array
+ // section.
+ bool
+ is_ctors_in_init_array(Relobj* relobj, unsigned int shndx) const;
+
// Check if a comdat group or .gnu.linkonce section with the given
// NAME is selected for the link. If there is already a section,
// *KEPT_SECTION is set to point to the signature and the function
- // returns false. Otherwise, the CANDIDATE signature is recorded
- // for this NAME in the layout object, *KEPT_SECTION is set to the
- // internal copy and the function return false. In some cases, with
- // CANDIDATE->GROUP_ being false, KEPT_SECTION can point back to
- // CANDIDATE.
+ // returns false. Otherwise, OBJECT, SHNDX,IS_COMDAT, and
+ // IS_GROUP_NAME are recorded for this NAME in the layout object,
+ // *KEPT_SECTION is set to the internal copy and the function return
+ // false.
bool
- find_or_add_kept_section(const std::string& name,
- Kept_section* candidate,
- Kept_section** kept_section);
-
- // Find the given comdat signature, and return the object and section
- // index of the kept group.
- Relobj*
- find_kept_object(const std::string&, unsigned int*) const;
+ find_or_add_kept_section(const std::string& name, Relobj* object,
+ unsigned int shndx, bool is_comdat,
+ bool is_group_name, Kept_section** kept_section);
// Finalize the layout after all the input sections have been added.
off_t
return this->symtab_section_;
}
+ // Return the file offset of the normal symbol table.
+ off_t
+ symtab_section_offset() const;
+
+ // Return the section index of the normal symbol tabl.e
+ unsigned int
+ symtab_section_shndx() const;
+
// Return the dynamic symbol table.
Output_section*
dynsym_section() const
script_options() const
{ return this->script_options_; }
+ // Return the object managing inputs in incremental build. NULL in
+ // non-incremental builds.
+ Incremental_inputs*
+ incremental_inputs() const
+ { return this->incremental_inputs_; }
+
+ // For the target-specific code to add dynamic tags which are common
+ // to most targets.
+ void
+ add_target_dynamic_tags(bool use_rel, const Output_data* plt_got,
+ const Output_data* plt_rel,
+ const Output_data_reloc_generic* dyn_rel,
+ bool add_debug, bool dynrel_includes_plt);
+
+ // Add a target-specific dynamic tag with constant value.
+ void
+ add_target_specific_dynamic_tag(elfcpp::DT tag, unsigned int val);
+
// Compute and write out the build ID if needed.
void
- write_build_id(Output_file*) const;
+ write_build_id(Output_file*, unsigned char*, size_t) const;
// Rewrite output file in binary format.
void
void
get_allocated_sections(Section_list*) const;
+ // Store the executable sections into the section list.
+ void
+ get_executable_sections(Section_list*) const;
+
// Make a section for a linker script to hold data.
Output_section*
- make_output_section_for_script(const char* name);
+ make_output_section_for_script(const char* name,
+ Script_sections::Section_type section_type);
// Make a segment. This is used by the linker script code.
Output_segment*
// Attach sections to segments.
void
- attach_sections_to_segments();
+ attach_sections_to_segments(const Target*);
+
+ // For relaxation clean up, we need to know output section data created
+ // from a linker script.
+ void
+ new_output_section_data_from_script(Output_section_data* posd)
+ {
+ if (this->record_output_section_data_from_script_)
+ this->script_output_section_data_list_.push_back(posd);
+ }
+
+ // Return section list.
+ const Section_list&
+ section_list() const
+ { return this->section_list_; }
+
+ // Returns TRUE iff NAME (an input section from RELOBJ) will
+ // be mapped to an output section that should be KEPT.
+ bool
+ keep_input_section(const Relobj*, const char*);
+
+ // Add a special output object that will be recreated afresh
+ // if there is another relaxation iteration.
+ void
+ add_relax_output(Output_data* data)
+ { this->relax_output_list_.push_back(data); }
+
+ // Clear out (and free) everything added by add_relax_output.
+ void
+ reset_relax_output();
private:
Layout(const Layout&);
Layout& operator=(const Layout&);
- // Mapping from .gnu.linkonce section names to output section names.
- struct Linkonce_mapping
+ // Mapping from input section names to output section names.
+ struct Section_name_mapping
{
const char* from;
int fromlen;
const char* to;
int tolen;
};
- static const Linkonce_mapping linkonce_mapping[];
- static const int linkonce_mapping_count;
+ static const Section_name_mapping section_name_mapping[];
+ static const int section_name_mapping_count;
// During a relocatable link, a list of group sections and
// signatures.
// Create a note section, filling in the header.
Output_section*
- create_note(const char* name, int note_type, const char *section_name,
+ create_note(const char* name, int note_type, const char* section_name,
size_t descsz, bool allocate, size_t* trailing_padding);
// Create a note section for gold version.
// Record whether the stack must be executable.
void
- create_executable_stack_info(const Target*);
+ create_executable_stack_info();
// Create a build ID note if needed.
void
create_build_id();
+ // Link .stab and .stabstr sections.
+ void
+ link_stabs_sections();
+
+ // Create .gnu_incremental_inputs and .gnu_incremental_strtab sections needed
+ // for the next run of incremental linking to check what has changed.
+ void
+ create_incremental_info_sections(Symbol_table*);
+
// Find the first read-only PT_LOAD segment, creating one if
// necessary.
Output_segment*
- find_first_load_seg();
+ find_first_load_seg(const Target*);
// Count the local symbols in the regular symbol table and the dynamic
// symbol table, and build the respective string pools.
void
finish_dynamic_section(const Input_objects*, const Symbol_table*);
+ // Set the size of the _DYNAMIC symbol.
+ void
+ set_dynamic_symbol_size(const Symbol_table*);
+
// Create the .interp section and PT_INTERP segment.
void
create_interp(const Target* target);
// Return whether to include this section in the link.
template<int size, bool big_endian>
bool
- include_section(Sized_relobj<size, big_endian>* object, const char* name,
+ include_section(Sized_relobj_file<size, big_endian>* object, const char* name,
const elfcpp::Shdr<size, big_endian>&);
// Return the output section name to use given an input section
// name. Set *PLEN to the length of the name. *PLEN must be
// initialized to the length of NAME.
static const char*
- output_section_name(const char* name, size_t* plen);
-
- // Return the output section name to use for a linkonce section
- // name. PLEN is as for output_section_name.
- static const char*
- linkonce_output_name(const char* name, size_t* plen);
+ output_section_name(const Relobj*, const char* name, size_t* plen);
// Return the number of allocated output sections.
size_t
// Return the output section for NAME, TYPE and FLAGS.
Output_section*
get_output_section(const char* name, Stringpool::Key name_key,
- elfcpp::Elf_Word type, elfcpp::Elf_Xword flags);
+ elfcpp::Elf_Word type, elfcpp::Elf_Xword flags,
+ Output_section_order order, bool is_relro);
+
+ // Clear the input section flags that should not be copied to the
+ // output section.
+ elfcpp::Elf_Xword
+ get_output_section_flags (elfcpp::Elf_Xword input_section_flags);
// Choose the output section for NAME in RELOBJ.
Output_section*
choose_output_section(const Relobj* relobj, const char* name,
elfcpp::Elf_Word type, elfcpp::Elf_Xword flags,
- bool is_input_section);
+ bool is_input_section, Output_section_order order,
+ bool is_relro);
// Create a new Output_section.
Output_section*
make_output_section(const char* name, elfcpp::Elf_Word type,
- elfcpp::Elf_Xword flags);
+ elfcpp::Elf_Xword flags, Output_section_order order,
+ bool is_relro);
// Attach a section to a segment.
void
- attach_section_to_segment(Output_section*);
+ attach_section_to_segment(const Target*, Output_section*);
+
+ // Get section order.
+ Output_section_order
+ default_section_order(Output_section*, bool is_relro_local);
// Attach an allocated section to a segment.
void
- attach_allocated_section_to_segment(Output_section*);
+ attach_allocated_section_to_segment(const Target*, Output_section*);
+
+ // Make the .eh_frame section.
+ Output_section*
+ make_eh_frame_section(const Relobj*);
// Set the final file offsets of all the segments.
off_t
Output_segment*
set_section_addresses_from_script(Symbol_table*);
+ // Find appropriate places or orphan sections in a script.
+ void
+ place_orphan_sections_in_script();
+
// Return whether SEG1 comes before SEG2 in the output file.
- static bool
+ bool
segment_precedes(const Output_segment* seg1, const Output_segment* seg2);
+ // Use to save and restore segments during relaxation.
+ typedef Unordered_map<const Output_segment*, const Output_segment*>
+ Segment_states;
+
+ // Save states of current output segments.
+ void
+ save_segments(Segment_states*);
+
+ // Restore output segment states.
+ void
+ restore_segments(const Segment_states*);
+
+ // Clean up after relaxation so that it is possible to lay out the
+ // sections and segments again.
+ void
+ clean_up_after_relaxation();
+
+ // Doing preparation work for relaxation. This is factored out to make
+ // Layout::finalized a bit smaller and easier to read.
+ void
+ prepare_for_relaxation();
+
+ // Main body of the relaxation loop, which lays out the section.
+ off_t
+ relaxation_loop_body(int, Target*, Symbol_table*, Output_segment**,
+ Output_segment*, Output_segment_headers*,
+ Output_file_header*, unsigned int*);
+
// A mapping used for kept comdats/.gnu.linkonce group signatures.
typedef Unordered_map<std::string, Kept_section> Signatures;
// A comparison class for segments.
- struct Compare_segments
+ class Compare_segments
{
+ public:
+ Compare_segments(Layout* layout)
+ : layout_(layout)
+ { }
+
bool
operator()(const Output_segment* seg1, const Output_segment* seg2)
- { return Layout::segment_precedes(seg1, seg2); }
+ { return this->layout_->segment_precedes(seg1, seg2); }
+
+ private:
+ Layout* layout_;
+ };
+
+ typedef std::vector<Output_section_data*> Output_section_data_list;
+
+ // Debug checker class.
+ class Relaxation_debug_check
+ {
+ public:
+ Relaxation_debug_check()
+ : section_infos_()
+ { }
+
+ // Check that sections and special data are in reset states.
+ void
+ check_output_data_for_reset_values(const Layout::Section_list&,
+ const Layout::Data_list& special_outputs,
+ const Layout::Data_list& relax_outputs);
+
+ // Record information of a section list.
+ void
+ read_sections(const Layout::Section_list&);
+
+ // Verify a section list with recorded information.
+ void
+ verify_sections(const Layout::Section_list&);
+
+ private:
+ // Information we care about a section.
+ struct Section_info
+ {
+ // Output section described by this.
+ Output_section* output_section;
+ // Load address.
+ uint64_t address;
+ // Data size.
+ off_t data_size;
+ // File offset.
+ off_t offset;
+ };
+
+ // Section information.
+ std::vector<Section_info> section_infos_;
};
// The number of input files, for sizing tables.
// The list of unattached Output_data objects which require special
// handling because they are not Output_sections.
Data_list special_output_list_;
+ // Like special_output_list_, but cleared and recreated on each
+ // iteration of relaxation.
+ Data_list relax_output_list_;
// The section headers.
Output_section_headers* section_headers_;
// A pointer to the PT_TLS segment if there is one.
Output_segment* tls_segment_;
// A pointer to the PT_GNU_RELRO segment if there is one.
Output_segment* relro_segment_;
+ // A pointer to the PT_INTERP segment if there is one.
+ Output_segment* interp_segment_;
+ // A backend may increase the size of the PT_GNU_RELRO segment if
+ // there is one. This is the amount to increase it by.
+ unsigned int increase_relro_;
// The SHT_SYMTAB output section.
Output_section* symtab_section_;
// The SHT_SYMTAB_SHNDX for the regular symbol table if there is one.
Output_symtab_xindex* dynsym_xindex_;
// The SHT_DYNAMIC output section if there is one.
Output_section* dynamic_section_;
+ // The _DYNAMIC symbol if there is one.
+ Symbol* dynamic_symbol_;
// The dynamic data which goes into dynamic_section_.
Output_data_dynamic* dynamic_data_;
// The exception frame output section if there is one.
bool added_eh_frame_data_;
// The exception frame header output section if there is one.
Output_section* eh_frame_hdr_section_;
+ // The data for the .gdb_index section.
+ Gdb_index* gdb_index_data_;
// The space for the build ID checksum if there is one.
Output_section_data* build_id_note_;
// The output section containing dwarf abbreviations
Group_signatures group_signatures_;
// The size of the output file.
off_t output_file_size_;
+ // Whether we have added an input section to an output section.
+ bool have_added_input_section_;
// Whether we have attached the sections to the segments.
bool sections_are_attached_;
// Whether we have seen an object file marked to require an
bool any_postprocessing_sections_;
// Whether we have resized the signatures_ hash table.
bool resized_signatures_;
+ // Whether we have created a .stab*str output section.
+ bool have_stabstr_section_;
+ // True if the input sections in the output sections should be sorted
+ // as specified in a section ordering file.
+ bool section_ordering_specified_;
+ // True if some input sections need to be mapped to a unique segment,
+ // after being mapped to a unique Output_section.
+ bool unique_segment_for_sections_specified_;
+ // In incremental build, holds information check the inputs and build the
+ // .gnu_incremental_inputs section.
+ Incremental_inputs* incremental_inputs_;
+ // Whether we record output section data created in script
+ bool record_output_section_data_from_script_;
+ // List of output data that needs to be removed at relaxation clean up.
+ Output_section_data_list script_output_section_data_list_;
+ // Structure to save segment states before entering the relaxation loop.
+ Segment_states* segment_states_;
+ // A relaxation debug checker. We only create one when in debugging mode.
+ Relaxation_debug_check* relaxation_debug_check_;
+ // Plugins specify section_ordering using this map. This is set in
+ // update_section_order in plugin.cc
+ std::map<Section_id, unsigned int> section_order_map_;
+ // This maps an input section to a unique segment. This is done by first
+ // placing such input sections in unique output sections and then mapping
+ // the output section to a unique segment. Unique_segment_info stores
+ // any additional flags and alignment of the new segment.
+ Section_segment_map section_segment_map_;
+ // Hash a pattern to its position in the section ordering file.
+ Unordered_map<std::string, unsigned int> input_section_position_;
+ // Vector of glob only patterns in the section_ordering file.
+ std::vector<std::string> input_section_glob_;
+ // For incremental links, the base file to be modified.
+ Incremental_binary* incremental_base_;
+ // For incremental links, a list of free space within the file.
+ Free_list free_list_;
};
// This task handles writing out data in output sections which is not
public:
Write_sections_task(const Layout* layout, Output_file* of,
Task_token* output_sections_blocker,
+ Task_token* input_sections_blocker,
Task_token* final_blocker)
: layout_(layout), of_(of),
output_sections_blocker_(output_sections_blocker),
+ input_sections_blocker_(input_sections_blocker),
final_blocker_(final_blocker)
{ }
const Layout* layout_;
Output_file* of_;
Task_token* output_sections_blocker_;
+ Task_token* input_sections_blocker_;
Task_token* final_blocker_;
};
{
public:
Write_symbols_task(const Layout* layout, const Symbol_table* symtab,
- const Input_objects* input_objects,
+ const Input_objects* /*input_objects*/,
const Stringpool* sympool, const Stringpool* dynpool,
Output_file* of, Task_token* final_blocker)
- : layout_(layout), symtab_(symtab), input_objects_(input_objects),
+ : layout_(layout), symtab_(symtab),
sympool_(sympool), dynpool_(dynpool), of_(of),
final_blocker_(final_blocker)
{ }
private:
const Layout* layout_;
const Symbol_table* symtab_;
- const Input_objects* input_objects_;
const Stringpool* sympool_;
const Stringpool* dynpool_;
Output_file* of_;
Task_token* final_blocker_;
};
+// This task function handles computation of the build id.
+// When using --build-id=tree, it schedules the tasks that
+// compute the hashes for each chunk of the file. This task
+// cannot run until we have finalized the size of the output
+// file, after the completion of Write_after_input_sections_task.
+
+class Build_id_task_runner : public Task_function_runner
+{
+ public:
+ Build_id_task_runner(const General_options* options, const Layout* layout,
+ Output_file* of)
+ : options_(options), layout_(layout), of_(of)
+ { }
+
+ // Run the operation.
+ void
+ run(Workqueue*, const Task*);
+
+ private:
+ const General_options* options_;
+ const Layout* layout_;
+ Output_file* of_;
+};
+
// This task function handles closing the file.
class Close_task_runner : public Task_function_runner
{
public:
Close_task_runner(const General_options* options, const Layout* layout,
- Output_file* of)
- : options_(options), layout_(layout), of_(of)
+ Output_file* of, unsigned char* array_of_hashes,
+ size_t size_of_hashes)
+ : options_(options), layout_(layout), of_(of),
+ array_of_hashes_(array_of_hashes), size_of_hashes_(size_of_hashes)
{ }
// Run the operation.
const General_options* options_;
const Layout* layout_;
Output_file* of_;
+ unsigned char* const array_of_hashes_;
+ const size_t size_of_hashes_;
};
// A small helper function to align an address.