xref: /netbsd-src/external/gpl3/binutils.old/dist/bfd/doc/linker.texi (revision e992f068c547fd6e84b3f104dc2340adcc955732)

@section Linker Functions
@cindex Linker
The linker uses three special entry points in the BFD target
vector.  It is not necessary to write special routines for
these entry points when creating a new BFD back end, since
generic versions are provided.  However, writing them can
speed up linking and make it use significantly less runtime
memory.

The first routine creates a hash table used by the other
routines.  The second routine adds the symbols from an object
file to the hash table.  The third routine takes all the
object files and links them together to create the output
file.  These routines are designed so that the linker proper
does not need to know anything about the symbols in the object
files that it is linking.  The linker merely arranges the
sections as directed by the linker script and lets BFD handle
the details of symbols and relocs.

The second routine and third routines are passed a pointer to
a @code{struct bfd_link_info} structure (defined in
@code{bfdlink.h}) which holds information relevant to the link,
including the linker hash table (which was created by the
first routine) and a set of callback functions to the linker
proper.

The generic linker routines are in @code{linker.c}, and use the
header file @code{genlink.h}.  As of this writing, the only back
ends which have implemented versions of these routines are
a.out (in @code{aoutx.h}) and ECOFF (in @code{ecoff.c}).  The a.out
routines are used as examples throughout this section.

@menu
* Creating a Linker Hash Table::
* Adding Symbols to the Hash Table::
* Performing the Final Link::
@end menu

@node Creating a Linker Hash Table, Adding Symbols to the Hash Table, Linker Functions, Linker Functions
@subsection Creating a linker hash table
@cindex _bfd_link_hash_table_create in target vector
@cindex target vector (_bfd_link_hash_table_create)
The linker routines must create a hash table, which must be
derived from @code{struct bfd_link_hash_table} described in
@code{bfdlink.c}.  @xref{Hash Tables}, for information on how to
create a derived hash table.  This entry point is called using
the target vector of the linker output file.

The @code{_bfd_link_hash_table_create} entry point must allocate
and initialize an instance of the desired hash table.  If the
back end does not require any additional information to be
stored with the entries in the hash table, the entry point may
simply create a @code{struct bfd_link_hash_table}.  Most likely,
however, some additional information will be needed.

For example, with each entry in the hash table the a.out
linker keeps the index the symbol has in the final output file
(this index number is used so that when doing a relocatable
link the symbol index used in the output file can be quickly
filled in when copying over a reloc).  The a.out linker code
defines the required structures and functions for a hash table
derived from @code{struct bfd_link_hash_table}.  The a.out linker
hash table is created by the function
@code{NAME(aout,link_hash_table_create)}; it simply allocates
space for the hash table, initializes it, and returns a
pointer to it.

When writing the linker routines for a new back end, you will
generally not know exactly which fields will be required until
you have finished.  You should simply create a new hash table
which defines no additional fields, and then simply add fields
as they become necessary.

@node Adding Symbols to the Hash Table, Performing the Final Link, Creating a Linker Hash Table, Linker Functions
@subsection Adding symbols to the hash table
@cindex _bfd_link_add_symbols in target vector
@cindex target vector (_bfd_link_add_symbols)
The linker proper will call the @code{_bfd_link_add_symbols}
entry point for each object file or archive which is to be
linked (typically these are the files named on the command
line, but some may also come from the linker script).  The
entry point is responsible for examining the file.  For an
object file, BFD must add any relevant symbol information to
the hash table.  For an archive, BFD must determine which
elements of the archive should be used and adding them to the
link.

The a.out version of this entry point is
@code{NAME(aout,link_add_symbols)}.

@menu
* Differing file formats::
* Adding symbols from an object file::
* Adding symbols from an archive::
@end menu

@node Differing file formats, Adding symbols from an object file, Adding Symbols to the Hash Table, Adding Symbols to the Hash Table
@subsubsection Differing file formats
Normally all the files involved in a link will be of the same
format, but it is also possible to link together different
format object files, and the back end must support that.  The
@code{_bfd_link_add_symbols} entry point is called via the target
vector of the file to be added.  This has an important
consequence: the function may not assume that the hash table
is the type created by the corresponding
@code{_bfd_link_hash_table_create} vector.  All the
@code{_bfd_link_add_symbols} function can assume about the hash
table is that it is derived from @code{struct
bfd_link_hash_table}.

Sometimes the @code{_bfd_link_add_symbols} function must store
some information in the hash table entry to be used by the
@code{_bfd_final_link} function.  In such a case the output bfd
xvec must be checked to make sure that the hash table was
created by an object file of the same format.

The @code{_bfd_final_link} routine must be prepared to handle a
hash entry without any extra information added by the
@code{_bfd_link_add_symbols} function.  A hash entry without
extra information will also occur when the linker script
directs the linker to create a symbol.  Note that, regardless
of how a hash table entry is added, all the fields will be
initialized to some sort of null value by the hash table entry
initialization function.

See @code{ecoff_link_add_externals} for an example of how to
check the output bfd before saving information (in this
case, the ECOFF external symbol debugging information) in a
hash table entry.

@node Adding symbols from an object file, Adding symbols from an archive, Differing file formats, Adding Symbols to the Hash Table
@subsubsection Adding symbols from an object file
When the @code{_bfd_link_add_symbols} routine is passed an object
file, it must add all externally visible symbols in that
object file to the hash table.  The actual work of adding the
symbol to the hash table is normally handled by the function
@code{_bfd_generic_link_add_one_symbol}.  The
@code{_bfd_link_add_symbols} routine is responsible for reading
all the symbols from the object file and passing the correct
information to @code{_bfd_generic_link_add_one_symbol}.

The @code{_bfd_link_add_symbols} routine should not use
@code{bfd_canonicalize_symtab} to read the symbols.  The point of
providing this routine is to avoid the overhead of converting
the symbols into generic @code{asymbol} structures.

@findex _bfd_generic_link_add_one_symbol
@code{_bfd_generic_link_add_one_symbol} handles the details of
combining common symbols, warning about multiple definitions,
and so forth.  It takes arguments which describe the symbol to
add, notably symbol flags, a section, and an offset.  The
symbol flags include such things as @code{BSF_WEAK} or
@code{BSF_INDIRECT}.  The section is a section in the object
file, or something like @code{bfd_und_section_ptr} for an undefined
symbol or @code{bfd_com_section_ptr} for a common symbol.

If the @code{_bfd_final_link} routine is also going to need to
read the symbol information, the @code{_bfd_link_add_symbols}
routine should save it somewhere attached to the object file
BFD.  However, the information should only be saved if the
@code{keep_memory} field of the @code{info} argument is TRUE, so
that the @code{-no-keep-memory} linker switch is effective.

The a.out function which adds symbols from an object file is
@code{aout_link_add_object_symbols}, and most of the interesting
work is in @code{aout_link_add_symbols}.  The latter saves
pointers to the hash tables entries created by
@code{_bfd_generic_link_add_one_symbol} indexed by symbol number,
so that the @code{_bfd_final_link} routine does not have to call
the hash table lookup routine to locate the entry.

@node Adding symbols from an archive, , Adding symbols from an object file, Adding Symbols to the Hash Table
@subsubsection Adding symbols from an archive
When the @code{_bfd_link_add_symbols} routine is passed an
archive, it must look through the symbols defined by the
archive and decide which elements of the archive should be
included in the link.  For each such element it must call the
@code{add_archive_element} linker callback, and it must add the
symbols from the object file to the linker hash table.  (The
callback may in fact indicate that a replacement BFD should be
used, in which case the symbols from that BFD should be added
to the linker hash table instead.)

@findex _bfd_generic_link_add_archive_symbols
In most cases the work of looking through the symbols in the
archive should be done by the
@code{_bfd_generic_link_add_archive_symbols} function.
@code{_bfd_generic_link_add_archive_symbols} is passed a function
to call to make the final decision about adding an archive
element to the link and to do the actual work of adding the
symbols to the linker hash table.  If the element is to
be included, the @code{add_archive_element} linker callback
routine must be called with the element as an argument, and
the element's symbols must be added to the linker hash table
just as though the element had itself been passed to the
@code{_bfd_link_add_symbols} function.

When the a.out @code{_bfd_link_add_symbols} function receives an
archive, it calls @code{_bfd_generic_link_add_archive_symbols}
passing @code{aout_link_check_archive_element} as the function
argument. @code{aout_link_check_archive_element} calls
@code{aout_link_check_ar_symbols}.  If the latter decides to add
the element (an element is only added if it provides a real,
non-common, definition for a previously undefined or common
symbol) it calls the @code{add_archive_element} callback and then
@code{aout_link_check_archive_element} calls
@code{aout_link_add_symbols} to actually add the symbols to the
linker hash table - possibly those of a substitute BFD, if the
@code{add_archive_element} callback avails itself of that option.

The ECOFF back end is unusual in that it does not normally
call @code{_bfd_generic_link_add_archive_symbols}, because ECOFF
archives already contain a hash table of symbols.  The ECOFF
back end searches the archive itself to avoid the overhead of
creating a new hash table.

@node Performing the Final Link, , Adding Symbols to the Hash Table, Linker Functions
@subsection Performing the final link
@cindex _bfd_link_final_link in target vector
@cindex target vector (_bfd_final_link)
When all the input files have been processed, the linker calls
the @code{_bfd_final_link} entry point of the output BFD.  This
routine is responsible for producing the final output file,
which has several aspects.  It must relocate the contents of
the input sections and copy the data into the output sections.
It must build an output symbol table including any local
symbols from the input files and the global symbols from the
hash table.  When producing relocatable output, it must
modify the input relocs and write them into the output file.
There may also be object format dependent work to be done.

The linker will also call the @code{write_object_contents} entry
point when the BFD is closed.  The two entry points must work
together in order to produce the correct output file.

The details of how this works are inevitably dependent upon
the specific object file format.  The a.out
@code{_bfd_final_link} routine is @code{NAME(aout,final_link)}.

@menu
* Information provided by the linker::
* Relocating the section contents::
* Writing the symbol table::
@end menu

@node Information provided by the linker, Relocating the section contents, Performing the Final Link, Performing the Final Link
@subsubsection Information provided by the linker
Before the linker calls the @code{_bfd_final_link} entry point,
it sets up some data structures for the function to use.

The @code{input_bfds} field of the @code{bfd_link_info} structure
will point to a list of all the input files included in the
link.  These files are linked through the @code{link.next} field
of the @code{bfd} structure.

Each section in the output file will have a list of
@code{link_order} structures attached to the @code{map_head.link_order}
field (the @code{link_order} structure is defined in
@code{bfdlink.h}).  These structures describe how to create the
contents of the output section in terms of the contents of
various input sections, fill constants, and, eventually, other
types of information.  They also describe relocs that must be
created by the BFD backend, but do not correspond to any input
file; this is used to support -Ur, which builds constructors
while generating a relocatable object file.

@node Relocating the section contents, Writing the symbol table, Information provided by the linker, Performing the Final Link
@subsubsection Relocating the section contents
The @code{_bfd_final_link} function should look through the
@code{link_order} structures attached to each section of the
output file.  Each @code{link_order} structure should either be
handled specially, or it should be passed to the function
@code{_bfd_default_link_order} which will do the right thing
(@code{_bfd_default_link_order} is defined in @code{linker.c}).

For efficiency, a @code{link_order} of type
@code{bfd_indirect_link_order} whose associated section belongs
to a BFD of the same format as the output BFD must be handled
specially.  This type of @code{link_order} describes part of an
output section in terms of a section belonging to one of the
input files.  The @code{_bfd_final_link} function should read the
contents of the section and any associated relocs, apply the
relocs to the section contents, and write out the modified
section contents.  If performing a relocatable link, the
relocs themselves must also be modified and written out.

@findex _bfd_relocate_contents
@findex _bfd_final_link_relocate
The functions @code{_bfd_relocate_contents} and
@code{_bfd_final_link_relocate} provide some general support for
performing the actual relocations, notably overflow checking.
Their arguments include information about the symbol the
relocation is against and a @code{reloc_howto_type} argument
which describes the relocation to perform.  These functions
are defined in @code{reloc.c}.

The a.out function which handles reading, relocating, and
writing section contents is @code{aout_link_input_section}.  The
actual relocation is done in @code{aout_link_input_section_std}
and @code{aout_link_input_section_ext}.

@node Writing the symbol table, , Relocating the section contents, Performing the Final Link
@subsubsection Writing the symbol table
The @code{_bfd_final_link} function must gather all the symbols
in the input files and write them out.  It must also write out
all the symbols in the global hash table.  This must be
controlled by the @code{strip} and @code{discard} fields of the
@code{bfd_link_info} structure.

The local symbols of the input files will not have been
entered into the linker hash table.  The @code{_bfd_final_link}
routine must consider each input file and include the symbols
in the output file.  It may be convenient to do this when
looking through the @code{link_order} structures, or it may be
done by stepping through the @code{input_bfds} list.

The @code{_bfd_final_link} routine must also traverse the global
hash table to gather all the externally visible symbols.  It
is possible that most of the externally visible symbols may be
written out when considering the symbols of each input file,
but it is still necessary to traverse the hash table since the
linker script may have defined some symbols that are not in
any of the input files.

The @code{strip} field of the @code{bfd_link_info} structure
controls which symbols are written out.  The possible values
are listed in @code{bfdlink.h}.  If the value is @code{strip_some},
then the @code{keep_hash} field of the @code{bfd_link_info}
structure is a hash table of symbols to keep; each symbol
should be looked up in this hash table, and only symbols which
are present should be included in the output file.

If the @code{strip} field of the @code{bfd_link_info} structure
permits local symbols to be written out, the @code{discard} field
is used to further controls which local symbols are included
in the output file.  If the value is @code{discard_l}, then all
local symbols which begin with a certain prefix are discarded;
this is controlled by the @code{bfd_is_local_label_name} entry point.

The a.out backend handles symbols by calling
@code{aout_link_write_symbols} on each input BFD and then
traversing the global hash table with the function
@code{aout_link_write_other_symbol}.  It builds a string table
while writing out the symbols, which is written to the output
file at the end of @code{NAME(aout,final_link)}.

@findex bfd_link_split_section
@subsubsection @code{bfd_link_split_section}
@strong{Synopsis}
@example
bool bfd_link_split_section (bfd *abfd, asection *sec);
@end example
@strong{Description}@*
Return nonzero if @var{sec} should be split during a
reloceatable or final link.
@example
#define bfd_link_split_section(abfd, sec) \
       BFD_SEND (abfd, _bfd_link_split_section, (abfd, sec))

@end example

@findex bfd_section_already_linked
@subsubsection @code{bfd_section_already_linked}
@strong{Synopsis}
@example
bool bfd_section_already_linked (bfd *abfd,
    asection *sec,
    struct bfd_link_info *info);
@end example
@strong{Description}@*
Check if @var{data} has been already linked during a reloceatable
or final link.  Return TRUE if it has.
@example
#define bfd_section_already_linked(abfd, sec, info) \
       BFD_SEND (abfd, _section_already_linked, (abfd, sec, info))

@end example

@findex bfd_generic_define_common_symbol
@subsubsection @code{bfd_generic_define_common_symbol}
@strong{Synopsis}
@example
bool bfd_generic_define_common_symbol
   (bfd *output_bfd, struct bfd_link_info *info,
    struct bfd_link_hash_entry *h);
@end example
@strong{Description}@*
Convert common symbol @var{h} into a defined symbol.
Return TRUE on success and FALSE on failure.
@example
#define bfd_define_common_symbol(output_bfd, info, h) \
       BFD_SEND (output_bfd, _bfd_define_common_symbol, (output_bfd, info, h))

@end example

@findex _bfd_generic_link_hide_symbol
@subsubsection @code{_bfd_generic_link_hide_symbol}
@strong{Synopsis}
@example
void _bfd_generic_link_hide_symbol
   (bfd *output_bfd, struct bfd_link_info *info,
    struct bfd_link_hash_entry *h);
@end example
@strong{Description}@*
Hide symbol @var{h}.
This is an internal function.  It should not be called from
outside the BFD library.
@example
#define bfd_link_hide_symbol(output_bfd, info, h) \
       BFD_SEND (output_bfd, _bfd_link_hide_symbol, (output_bfd, info, h))

@end example

@findex bfd_generic_define_start_stop
@subsubsection @code{bfd_generic_define_start_stop}
@strong{Synopsis}
@example
struct bfd_link_hash_entry *bfd_generic_define_start_stop
   (struct bfd_link_info *info,
    const char *symbol, asection *sec);
@end example
@strong{Description}@*
Define a __start, __stop, .startof. or .sizeof. symbol.
Return the symbol or NULL if no such undefined symbol exists.
@example
#define bfd_define_start_stop(output_bfd, info, symbol, sec) \
       BFD_SEND (output_bfd, _bfd_define_start_stop, (info, symbol, sec))

@end example

@findex bfd_find_version_for_sym
@subsubsection @code{bfd_find_version_for_sym}
@strong{Synopsis}
@example
struct bfd_elf_version_tree * bfd_find_version_for_sym
   (struct bfd_elf_version_tree *verdefs,
    const char *sym_name, bool *hide);
@end example
@strong{Description}@*
Search an elf version script tree for symbol versioning
info and export / don't-export status for a given symbol.
Return non-NULL on success and NULL on failure; also sets
the output @samp{hide} boolean parameter.

@findex bfd_hide_sym_by_version
@subsubsection @code{bfd_hide_sym_by_version}
@strong{Synopsis}
@example
bool bfd_hide_sym_by_version
   (struct bfd_elf_version_tree *verdefs, const char *sym_name);
@end example
@strong{Description}@*
Search an elf version script tree for symbol versioning
info for a given symbol.  Return TRUE if the symbol is hidden.

@findex bfd_link_check_relocs
@subsubsection @code{bfd_link_check_relocs}
@strong{Synopsis}
@example
bool bfd_link_check_relocs
   (bfd *abfd, struct bfd_link_info *info);
@end example
@strong{Description}@*
Checks the relocs in ABFD for validity.
Does not execute the relocs.
Return TRUE if everything is OK, FALSE otherwise.
This is the external entry point to this code.

@findex _bfd_generic_link_check_relocs
@subsubsection @code{_bfd_generic_link_check_relocs}
@strong{Synopsis}
@example
bool _bfd_generic_link_check_relocs
   (bfd *abfd, struct bfd_link_info *info);
@end example
@strong{Description}@*
Stub function for targets that do not implement reloc checking.
Return TRUE.
This is an internal function.  It should not be called from
outside the BFD library.

@findex bfd_merge_private_bfd_data
@subsubsection @code{bfd_merge_private_bfd_data}
@strong{Synopsis}
@example
bool bfd_merge_private_bfd_data
   (bfd *ibfd, struct bfd_link_info *info);
@end example
@strong{Description}@*
Merge private BFD information from the BFD @var{ibfd} to the
the output file BFD when linking.  Return @code{TRUE} on success,
@code{FALSE} on error.  Possible error returns are:

@itemize @bullet

@item
@code{bfd_error_no_memory} -
Not enough memory exists to create private data for @var{obfd}.
@end itemize
@example
#define bfd_merge_private_bfd_data(ibfd, info) \
       BFD_SEND ((info)->output_bfd, _bfd_merge_private_bfd_data, \
                 (ibfd, info))
@end example

@findex _bfd_generic_verify_endian_match
@subsubsection @code{_bfd_generic_verify_endian_match}
@strong{Synopsis}
@example
bool _bfd_generic_verify_endian_match
   (bfd *ibfd, struct bfd_link_info *info);
@end example
@strong{Description}@*
Can be used from / for bfd_merge_private_bfd_data to check that
endianness matches between input and output file.  Returns
TRUE for a match, otherwise returns FALSE and emits an error.