\input texinfo
@c Copyright 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1998,
@c 2000, 2001, 2002, 2003, 2004, 2006
@c Free Software Foundation, Inc.
@setfilename bfdint.info

@settitle BFD Internals
@iftex
@titlepage
@title{BFD Internals}
@author{Ian Lance Taylor}
@author{Cygnus Solutions}
@page
@end iftex

@node Top
@top BFD Internals
@raisesections
@cindex bfd internals

This document describes some BFD internal information which may be
helpful when working on BFD.  It is very incomplete.

This document is not updated regularly, and may be out of date.

The initial version of this document was written by Ian Lance Taylor
@email{ian@@cygnus.com}.

@menu
* BFD overview::		BFD overview
* BFD guidelines::		BFD programming guidelines
* BFD target vector::		BFD target vector
* BFD generated files::		BFD generated files
* BFD multiple compilations::	Files compiled multiple times in BFD
* BFD relocation handling::	BFD relocation handling
* BFD ELF support::		BFD ELF support
* BFD glossary::		Glossary
* Index::			Index
@end menu

@node BFD overview
@section BFD overview

BFD is a library which provides a single interface to read and write
object files, executables, archive files, and core files in any format.

@menu
* BFD library interfaces::	BFD library interfaces
* BFD library users::		BFD library users
* BFD view::			The BFD view of a file
* BFD blindness::		BFD loses information
@end menu

@node BFD library interfaces
@subsection BFD library interfaces

One way to look at the BFD library is to divide it into four parts by
type of interface.

The first interface is the set of generic functions which programs using
the BFD library will call.  These generic function normally translate
directly or indirectly into calls to routines which are specific to a
particular object file format.  Many of these generic functions are
actually defined as macros in @file{bfd.h}.  These functions comprise
the official BFD interface.

The second interface is the set of functions which appear in the target
vectors.  This is the bulk of the code in BFD.  A target vector is a set
of function pointers specific to a particular object file format.  The
target vector is used to implement the generic BFD functions.  These
functions are always called through the target vector, and are never
called directly.  The target vector is described in detail in @ref{BFD
target vector}.  The set of functions which appear in a particular
target vector is often referred to as a BFD backend.

The third interface is a set of oddball functions which are typically
specific to a particular object file format, are not generic functions,
and are called from outside of the BFD library.  These are used as hooks
by the linker and the assembler when a particular object file format
requires some action which the BFD generic interface does not provide.
These functions are typically declared in @file{bfd.h}, but in many
cases they are only provided when BFD is configured with support for a
particular object file format.  These functions live in a grey area, and
are not really part of the official BFD interface.

The fourth interface is the set of BFD support functions which are
called by the other BFD functions.  These manage issues like memory
allocation, error handling, file access, hash tables, swapping, and the
like.  These functions are never called from outside of the BFD library.

@node BFD library users
@subsection BFD library users

Another way to look at the BFD library is to divide it into three parts
by the manner in which it is used.

The first use is to read an object file.  The object file readers are
programs like @samp{gdb}, @samp{nm}, @samp{objdump}, and @samp{objcopy}.
These programs use BFD to view an object file in a generic form.  The
official BFD interface is normally fully adequate for these programs.

The second use is to write an object file.  The object file writers are
programs like @samp{gas} and @samp{objcopy}.  These programs use BFD to
create an object file.  The official BFD interface is normally adequate
for these programs, but for some object file formats the assembler needs
some additional hooks in order to set particular flags or other
information.  The official BFD interface includes functions to copy
private information from one object file to another, and these functions
are used by @samp{objcopy} to avoid information loss.

The third use is to link object files.  There is only one object file
linker, @samp{ld}.  Originally, @samp{ld} was an object file reader and
an object file writer, and it did the link operation using the generic
BFD structures.  However, this turned out to be too slow and too memory
intensive.

The official BFD linker functions were written to permit specific BFD
backends to perform the link without translating through the generic
structures, in the normal case where all the input files and output file
have the same object file format.  Not all of the backends currently
implement the new interface, and there are default linking functions
within BFD which use the generic structures and which work with all
backends.

For several object file formats the linker needs additional hooks which
are not provided by the official BFD interface, particularly for dynamic
linking support.  These functions are typically called from the linker
emulation template.

@node BFD view
@subsection The BFD view of a file

BFD uses generic structures to manage information.  It translates data
into the generic form when reading files, and out of the generic form
when writing files.

BFD describes a file as a pointer to the @samp{bfd} type.  A @samp{bfd}
is composed of the following elements.  The BFD information can be
displayed using the @samp{objdump} program with various options.

@table @asis
@item general information
The object file format, a few general flags, the start address.
@item architecture
The architecture, including both a general processor type (m68k, MIPS
etc.) and a specific machine number (m68000, R4000, etc.).
@item sections
A list of sections.
@item symbols
A symbol table.
@end table

BFD represents a section as a pointer to the @samp{asection} type.  Each
section has a name and a size.  Most sections also have an associated
block of data, known as the section contents.  Sections also have
associated flags, a virtual memory address, a load memory address, a
required alignment, a list of relocations, and other miscellaneous
information.

BFD represents a relocation as a pointer to the @samp{arelent} type.  A
relocation describes an action which the linker must take to modify the
section contents.  Relocations have a symbol, an address, an addend, and
a pointer to a howto structure which describes how to perform the
relocation.  For more information, see @ref{BFD relocation handling}.

BFD represents a symbol as a pointer to the @samp{asymbol} type.  A
symbol has a name, a pointer to a section, an offset within that
section, and some flags.

Archive files do not have any sections or symbols.  Instead, BFD
represents an archive file as a file which contains a list of
@samp{bfd}s.  BFD also provides access to the archive symbol map, as a
list of symbol names.  BFD provides a function to return the @samp{bfd}
within the archive which corresponds to a particular entry in the
archive symbol map.

@node BFD blindness
@subsection BFD loses information

Most object file formats have information which BFD can not represent in
its generic form, at least as currently defined.

There is often explicit information which BFD can not represent.  For
example, the COFF version stamp, or the ELF program segments.  BFD
provides special hooks to handle this information when copying,
printing, or linking an object file.  The BFD support for a particular
object file format will normally store this information in private data
and handle it using the special hooks.

In some cases there is also implicit information which BFD can not
represent.  For example, the MIPS processor distinguishes small and
large symbols, and requires that all small symbols be within 32K of the
GP register.  This means that the MIPS assembler must be able to mark
variables as either small or large, and the MIPS linker must know to put
small symbols within range of the GP register.  Since BFD can not
represent this information, this means that the assembler and linker
must have information that is specific to a particular object file
format which is outside of the BFD library.

This loss of information indicates areas where the BFD paradigm breaks
down.  It is not actually possible to represent the myriad differences
among object file formats using a single generic interface, at least not
in the manner which BFD does it today.

Nevertheless, the BFD library does greatly simplify the task of dealing
with object files, and particular problems caused by information loss
can normally be solved using some sort of relatively constrained hook
into the library.


@node BFD guidelines
@section BFD programming guidelines
@cindex bfd programming guidelines
@cindex programming guidelines for bfd
@cindex guidelines, bfd programming

There is a lot of poorly written and confusing code in BFD.  New BFD
code should be written to a higher standard.  Merely because some BFD
code is written in a particular manner does not mean that you should
emulate it.

Here are some general BFD programming guidelines:

@itemize @bullet
@item
Follow the GNU coding standards.

@item
Avoid global variables.  We ideally want BFD to be fully reentrant, so
that it can be used in multiple threads.  All uses of global or static
variables interfere with that.  Initialized constant variables are OK,
and they should be explicitly marked with @samp{const}.  Instead of global
variables, use data attached to a BFD or to a linker hash table.

@item
All externally visible functions should have names which start with
@samp{bfd_}.  All such functions should be declared in some header file,
typically @file{bfd.h}.  See, for example, the various declarations near
the end of @file{bfd-in.h}, which mostly declare functions required by
specific linker emulations.

@item
All functions which need to be visible from one file to another within
BFD, but should not be visible outside of BFD, should start with
@samp{_bfd_}.  Although external names beginning with @samp{_} are
prohibited by the ANSI standard, in practice this usage will always
work, and it is required by the GNU coding standards.

@item
Always remember that people can compile using @samp{--enable-targets} to
build several, or all, targets at once.  It must be possible to link
together the files for all targets.

@item
BFD code should compile with few or no warnings using @samp{gcc -Wall}.
Some warnings are OK, like the absence of certain function declarations
which may or may not be declared in system header files.  Warnings about
ambiguous expressions and the like should always be fixed.
@end itemize

@node BFD target vector
@section BFD target vector
@cindex bfd target vector
@cindex target vector in bfd

BFD supports multiple object file formats by using the @dfn{target
vector}.  This is simply a set of function pointers which implement
behaviour that is specific to a particular object file format.

In this section I list all of the entries in the target vector and
describe what they do.

@menu
* BFD target vector miscellaneous::	Miscellaneous constants
* BFD target vector swap::		Swapping functions
* BFD target vector format::		Format type dependent functions
* BFD_JUMP_TABLE macros::		BFD_JUMP_TABLE macros
* BFD target vector generic::		Generic functions
* BFD target vector copy::		Copy functions
* BFD target vector core::		Core file support functions
* BFD target vector archive::		Archive functions
* BFD target vector symbols::		Symbol table functions
* BFD target vector relocs::		Relocation support
* BFD target vector write::		Output functions
* BFD target vector link::		Linker functions
* BFD target vector dynamic::		Dynamic linking information functions
@end menu

@node BFD target vector miscellaneous
@subsection Miscellaneous constants

The target vector starts with a set of constants.

@table @samp
@item name
The name of the target vector.  This is an arbitrary string.  This is
how the target vector is named in command line options for tools which
use BFD, such as the @samp{--oformat} linker option.

@item flavour
A general description of the type of target.  The following flavours are
currently defined:

@table @samp
@item bfd_target_unknown_flavour
Undefined or unknown.
@item bfd_target_aout_flavour
a.out.
@item bfd_target_coff_flavour
COFF.
@item bfd_target_ecoff_flavour
ECOFF.
@item bfd_target_elf_flavour
ELF.
@item bfd_target_ieee_flavour
IEEE-695.
@item bfd_target_nlm_flavour
NLM.
@item bfd_target_oasys_flavour
OASYS.
@item bfd_target_tekhex_flavour
Tektronix hex format.
@item bfd_target_srec_flavour
Motorola S-record format.
@item bfd_target_ihex_flavour
Intel hex format.
@item bfd_target_som_flavour
SOM (used on HP/UX).
@item bfd_target_os9k_flavour
os9000.
@item bfd_target_versados_flavour
VERSAdos.
@item bfd_target_msdos_flavour
MS-DOS.
@item bfd_target_evax_flavour
openVMS.
@item bfd_target_mmo_flavour
Donald Knuth's MMIXware object format.
@end table

@item byteorder
The byte order of data in the object file.  One of
@samp{BFD_ENDIAN_BIG}, @samp{BFD_ENDIAN_LITTLE}, or
@samp{BFD_ENDIAN_UNKNOWN}.  The latter would be used for a format such
as S-records which do not record the architecture of the data.

@item header_byteorder
The byte order of header information in the object file.  Normally the
same as the @samp{byteorder} field, but there are certain cases where it
may be different.

@item object_flags
Flags which may appear in the @samp{flags} field of a BFD with this
format.

@item section_flags
Flags which may appear in the @samp{flags} field of a section within a
BFD with this format.

@item symbol_leading_char
A character which the C compiler normally puts before a symbol.  For
example, an a.out compiler will typically generate the symbol
@samp{_foo} for a function named @samp{foo} in the C source, in which
case this field would be @samp{_}.  If there is no such character, this
field will be @samp{0}.

@item ar_pad_char
The padding character to use at the end of an archive name.  Normally
@samp{/}.

@item ar_max_namelen
The maximum length of a short name in an archive.  Normally @samp{14}.

@item backend_data
A pointer to constant backend data.  This is used by backends to store
whatever additional information they need to distinguish similar target
vectors which use the same sets of functions.
@end table

@node BFD target vector swap
@subsection Swapping functions

Every target vector has function pointers used for swapping information
in and out of the target representation.  There are two sets of
functions: one for data information, and one for header information.
Each set has three sizes: 64-bit, 32-bit, and 16-bit.  Each size has
three actual functions: put, get unsigned, and get signed.

These 18 functions are used to convert data between the host and target
representations.

@node BFD target vector format
@subsection Format type dependent functions

Every target vector has three arrays of function pointers which are
indexed by the BFD format type.  The BFD format types are as follows:

@table @samp
@item bfd_unknown
Unknown format.  Not used for anything useful.
@item bfd_object
Object file.
@item bfd_archive
Archive file.
@item bfd_core
Core file.
@end table

The three arrays of function pointers are as follows:

@table @samp
@item bfd_check_format
Check whether the BFD is of a particular format (object file, archive
file, or core file) corresponding to this target vector.  This is called
by the @samp{bfd_check_format} function when examining an existing BFD.
If the BFD matches the desired format, this function will initialize any
format specific information such as the @samp{tdata} field of the BFD.
This function must be called before any other BFD target vector function
on a file opened for reading.

@item bfd_set_format
Set the format of a BFD which was created for output.  This is called by
the @samp{bfd_set_format} function after creating the BFD with a
function such as @samp{bfd_openw}.  This function will initialize format
specific information required to write out an object file or whatever of
the given format.  This function must be called before any other BFD
target vector function on a file opened for writing.

@item bfd_write_contents
Write out the contents of the BFD in the given format.  This is called
by @samp{bfd_close} function for a BFD opened for writing.  This really
should not be an array selected by format type, as the
@samp{bfd_set_format} function provides all the required information.
In fact, BFD will fail if a different format is used when calling
through the @samp{bfd_set_format} and the @samp{bfd_write_contents}
arrays; fortunately, since @samp{bfd_close} gets it right, this is a
difficult error to make.
@end table

@node BFD_JUMP_TABLE macros
@subsection @samp{BFD_JUMP_TABLE} macros
@cindex @samp{BFD_JUMP_TABLE}

Most target vectors are defined using @samp{BFD_JUMP_TABLE} macros.
These macros take a single argument, which is a prefix applied to a set
of functions.  The macros are then used to initialize the fields in the
target vector.

For example, the @samp{BFD_JUMP_TABLE_RELOCS} macro defines three
functions: @samp{_get_reloc_upper_bound}, @samp{_canonicalize_reloc},
and @samp{_bfd_reloc_type_lookup}.  A reference like
@samp{BFD_JUMP_TABLE_RELOCS (foo)} will expand into three functions
prefixed with @samp{foo}: @samp{foo_get_reloc_upper_bound}, etc.  The
@samp{BFD_JUMP_TABLE_RELOCS} macro will be placed such that those three
functions initialize the appropriate fields in the BFD target vector.

This is done because it turns out that many different target vectors can
share certain classes of functions.  For example, archives are similar
on most platforms, so most target vectors can use the same archive
functions.  Those target vectors all use @samp{BFD_JUMP_TABLE_ARCHIVE}
with the same argument, calling a set of functions which is defined in
@file{archive.c}.

Each of the @samp{BFD_JUMP_TABLE} macros is mentioned below along with
the description of the function pointers which it defines.  The function
pointers will be described using the name without the prefix which the
@samp{BFD_JUMP_TABLE} macro defines.  This name is normally the same as
the name of the field in the target vector structure.  Any differences
will be noted.

@node BFD target vector generic
@subsection Generic functions
@cindex @samp{BFD_JUMP_TABLE_GENERIC}

The @samp{BFD_JUMP_TABLE_GENERIC} macro is used for some catch all
functions which don't easily fit into other categories.

@table @samp
@item _close_and_cleanup
Free any target specific information associated with the BFD.  This is
called when any BFD is closed (the @samp{bfd_write_contents} function
mentioned earlier is only called for a BFD opened for writing).  Most
targets use @samp{bfd_alloc} to allocate all target specific
information, and therefore don't have to do anything in this function.
This function pointer is typically set to
@samp{_bfd_generic_close_and_cleanup}, which simply returns true.

@item _bfd_free_cached_info
Free any cached information associated with the BFD which can be
recreated later if necessary.  This is used to reduce the memory
consumption required by programs using BFD.  This is normally called via
the @samp{bfd_free_cached_info} macro.  It is used by the default
archive routines when computing the archive map.  Most targets do not
do anything special for this entry point, and just set it to
@samp{_bfd_generic_free_cached_info}, which simply returns true.

@item _new_section_hook
This is called from @samp{bfd_make_section_anyway} whenever a new
section is created.  Most targets use it to initialize section specific
information.  This function is called whether or not the section
corresponds to an actual section in an actual BFD.

@item _get_section_contents
Get the contents of a section.  This is called from
@samp{bfd_get_section_contents}.  Most targets set this to
@samp{_bfd_generic_get_section_contents}, which does a @samp{bfd_seek}
based on the section's @samp{filepos} field and a @samp{bfd_bread}.  The
corresponding field in the target vector is named
@samp{_bfd_get_section_contents}.

@item _get_section_contents_in_window
Set a @samp{bfd_window} to hold the contents of a section.  This is
called from @samp{bfd_get_section_contents_in_window}.  The
@samp{bfd_window} idea never really caught on, and I don't think this is
ever called.  Pretty much all targets implement this as
@samp{bfd_generic_get_section_contents_in_window}, which uses
@samp{bfd_get_section_contents} to do the right thing.  The
corresponding field in the target vector is named
@samp{_bfd_get_section_contents_in_window}.
@end table

@node BFD target vector copy
@subsection Copy functions
@cindex @samp{BFD_JUMP_TABLE_COPY}

The @samp{BFD_JUMP_TABLE_COPY} macro is used for functions which are
called when copying BFDs, and for a couple of functions which deal with
internal BFD information.

@table @samp
@item _bfd_copy_private_bfd_data
This is called when copying a BFD, via @samp{bfd_copy_private_bfd_data}.
If the input and output BFDs have the same format, this will copy any
private information over.  This is called after all the section contents
have been written to the output file.  Only a few targets do anything in
this function.

@item _bfd_merge_private_bfd_data
This is called when linking, via @samp{bfd_merge_private_bfd_data}.  It
gives the backend linker code a chance to set any special flags in the
output file based on the contents of the input file.  Only a few targets
do anything in this function.

@item _bfd_copy_private_section_data
This is similar to @samp{_bfd_copy_private_bfd_data}, but it is called
for each section, via @samp{bfd_copy_private_section_data}.  This
function is called before any section contents have been written.  Only
a few targets do anything in this function.

@item _bfd_copy_private_symbol_data
This is called via @samp{bfd_copy_private_symbol_data}, but I don't
think anything actually calls it.  If it were defined, it could be used
to copy private symbol data from one BFD to another.  However, most BFDs
store extra symbol information by allocating space which is larger than
the @samp{asymbol} structure and storing private information in the
extra space.  Since @samp{objcopy} and other programs copy symbol
information by copying pointers to @samp{asymbol} structures, the
private symbol information is automatically copied as well.  Most
targets do not do anything in this function.

@item _bfd_set_private_flags
This is called via @samp{bfd_set_private_flags}.  It is basically a hook
for the assembler to set magic information.  For example, the PowerPC
ELF assembler uses it to set flags which appear in the e_flags field of
the ELF header.  Most targets do not do anything in this function.

@item _bfd_print_private_bfd_data
This is called by @samp{objdump} when the @samp{-p} option is used.  It
is called via @samp{bfd_print_private_data}.  It prints any interesting
information about the BFD which can not be otherwise represented by BFD
and thus can not be printed by @samp{objdump}.  Most targets do not do
anything in this function.
@end table

@node BFD target vector core
@subsection Core file support functions
@cindex @samp{BFD_JUMP_TABLE_CORE}

The @samp{BFD_JUMP_TABLE_CORE} macro is used for functions which deal
with core files.  Obviously, these functions only do something
interesting for targets which have core file support.

@table @samp
@item _core_file_failing_command
Given a core file, this returns the command which was run to produce the
core file.

@item _core_file_failing_signal
Given a core file, this returns the signal number which produced the
core file.

@item _core_file_matches_executable_p
Given a core file and a BFD for an executable, this returns whether the
core file was generated by the executable.
@end table

@node BFD target vector archive
@subsection Archive functions
@cindex @samp{BFD_JUMP_TABLE_ARCHIVE}

The @samp{BFD_JUMP_TABLE_ARCHIVE} macro is used for functions which deal
with archive files.  Most targets use COFF style archive files
(including ELF targets), and these use @samp{_bfd_archive_coff} as the
argument to @samp{BFD_JUMP_TABLE_ARCHIVE}.  Some targets use BSD/a.out
style archives, and these use @samp{_bfd_archive_bsd}.  (The main
difference between BSD and COFF archives is the format of the archive
symbol table).  Targets with no archive support use
@samp{_bfd_noarchive}.  Finally, a few targets have unusual archive
handling.

@table @samp
@item _slurp_armap
Read in the archive symbol table, storing it in private BFD data.  This
is normally called from the archive @samp{check_format} routine.  The
corresponding field in the target vector is named
@samp{_bfd_slurp_armap}.

@item _slurp_extended_name_table
Read in the extended name table from the archive, if there is one,
storing it in private BFD data.  This is normally called from the
archive @samp{check_format} routine.  The corresponding field in the
target vector is named @samp{_bfd_slurp_extended_name_table}.

@item construct_extended_name_table
Build and return an extended name table if one is needed to write out
the archive.  This also adjusts the archive headers to refer to the
extended name table appropriately.  This is normally called from the
archive @samp{write_contents} routine.  The corresponding field in the
target vector is named @samp{_bfd_construct_extended_name_table}.

@item _truncate_arname
This copies a file name into an archive header, truncating it as
required.  It is normally called from the archive @samp{write_contents}
routine.  This function is more interesting in targets which do not
support extended name tables, but I think the GNU @samp{ar} program
always uses extended name tables anyhow.  The corresponding field in the
target vector is named @samp{_bfd_truncate_arname}.

@item _write_armap
Write out the archive symbol table using calls to @samp{bfd_bwrite}.
This is normally called from the archive @samp{write_contents} routine.
The corresponding field in the target vector is named @samp{write_armap}
(no leading underscore).

@item _read_ar_hdr
Read and parse an archive header.  This handles expanding the archive
header name into the real file name using the extended name table.  This
is called by routines which read the archive symbol table or the archive
itself.  The corresponding field in the target vector is named
@samp{_bfd_read_ar_hdr_fn}.

@item _openr_next_archived_file
Given an archive and a BFD representing a file stored within the
archive, return a BFD for the next file in the archive.  This is called
via @samp{bfd_openr_next_archived_file}.  The corresponding field in the
target vector is named @samp{openr_next_archived_file} (no leading
underscore).

@item _get_elt_at_index
Given an archive and an index, return a BFD for the file in the archive
corresponding to that entry in the archive symbol table.  This is called
via @samp{bfd_get_elt_at_index}.  The corresponding field in the target
vector is named @samp{_bfd_get_elt_at_index}.

@item _generic_stat_arch_elt
Do a stat on an element of an archive, returning information read from
the archive header (modification time, uid, gid, file mode, size).  This
is called via @samp{bfd_stat_arch_elt}.  The corresponding field in the
target vector is named @samp{_bfd_stat_arch_elt}.

@item _update_armap_timestamp
After the entire contents of an archive have been written out, update
the timestamp of the archive symbol table to be newer than that of the
file.  This is required for a.out style archives.  This is normally
called by the archive @samp{write_contents} routine.  The corresponding
field in the target vector is named @samp{_bfd_update_armap_timestamp}.
@end table

@node BFD target vector symbols
@subsection Symbol table functions
@cindex @samp{BFD_JUMP_TABLE_SYMBOLS}

The @samp{BFD_JUMP_TABLE_SYMBOLS} macro is used for functions which deal
with symbols.

@table @samp
@item _get_symtab_upper_bound
Return a sensible upper bound on the amount of memory which will be
required to read the symbol table.  In practice most targets return the
amount of memory required to hold @samp{asymbol} pointers for all the
symbols plus a trailing @samp{NULL} entry, and store the actual symbol
information in BFD private data.  This is called via
@samp{bfd_get_symtab_upper_bound}.  The corresponding field in the
target vector is named @samp{_bfd_get_symtab_upper_bound}.

@item _canonicalize_symtab
Read in the symbol table.  This is called via
@samp{bfd_canonicalize_symtab}.  The corresponding field in the target
vector is named @samp{_bfd_canonicalize_symtab}.

@item _make_empty_symbol
Create an empty symbol for the BFD.  This is needed because most targets
store extra information with each symbol by allocating a structure
larger than an @samp{asymbol} and storing the extra information at the
end.  This function will allocate the right amount of memory, and return
what looks like a pointer to an empty @samp{asymbol}.  This is called
via @samp{bfd_make_empty_symbol}.  The corresponding field in the target
vector is named @samp{_bfd_make_empty_symbol}.

@item _print_symbol
Print information about the symbol.  This is called via
@samp{bfd_print_symbol}.  One of the arguments indicates what sort of
information should be printed:

@table @samp
@item bfd_print_symbol_name
Just print the symbol name.
@item bfd_print_symbol_more
Print the symbol name and some interesting flags.  I don't think
anything actually uses this.
@item bfd_print_symbol_all
Print all information about the symbol.  This is used by @samp{objdump}
when run with the @samp{-t} option.
@end table
The corresponding field in the target vector is named
@samp{_bfd_print_symbol}.

@item _get_symbol_info
Return a standard set of information about the symbol.  This is called
via @samp{bfd_symbol_info}.  The corresponding field in the target
vector is named @samp{_bfd_get_symbol_info}.

@item _bfd_is_local_label_name
Return whether the given string would normally represent the name of a
local label.  This is called via @samp{bfd_is_local_label} and
@samp{bfd_is_local_label_name}.  Local labels are normally discarded by
the assembler.  In the linker, this defines the difference between the
@samp{-x} and @samp{-X} options.

@item _get_lineno
Return line number information for a symbol.  This is only meaningful
for a COFF target.  This is called when writing out COFF line numbers.

@item _find_nearest_line
Given an address within a section, use the debugging information to find
the matching file name, function name, and line number, if any.  This is
called via @samp{bfd_find_nearest_line}.  The corresponding field in the
target vector is named @samp{_bfd_find_nearest_line}.

@item _bfd_make_debug_symbol
Make a debugging symbol.  This is only meaningful for a COFF target,
where it simply returns a symbol which will be placed in the
@samp{N_DEBUG} section when it is written out.  This is called via
@samp{bfd_make_debug_symbol}.

@item _read_minisymbols
Minisymbols are used to reduce the memory requirements of programs like
@samp{nm}.  A minisymbol is a cookie pointing to internal symbol
information which the caller can use to extract complete symbol
information.  This permits BFD to not convert all the symbols into
generic form, but to instead convert them one at a time.  This is called
via @samp{bfd_read_minisymbols}.  Most targets do not implement this,
and just use generic support which is based on using standard
@samp{asymbol} structures.

@item _minisymbol_to_symbol
Convert a minisymbol to a standard @samp{asymbol}.  This is called via
@samp{bfd_minisymbol_to_symbol}.
@end table

@node BFD target vector relocs
@subsection Relocation support
@cindex @samp{BFD_JUMP_TABLE_RELOCS}

The @samp{BFD_JUMP_TABLE_RELOCS} macro is used for functions which deal
with relocations.

@table @samp
@item _get_reloc_upper_bound
Return a sensible upper bound on the amount of memory which will be
required to read the relocations for a section.  In practice most
targets return the amount of memory required to hold @samp{arelent}
pointers for all the relocations plus a trailing @samp{NULL} entry, and
store the actual relocation information in BFD private data.  This is
called via @samp{bfd_get_reloc_upper_bound}.

@item _canonicalize_reloc
Return the relocation information for a section.  This is called via
@samp{bfd_canonicalize_reloc}.  The corresponding field in the target
vector is named @samp{_bfd_canonicalize_reloc}.

@item _bfd_reloc_type_lookup
Given a relocation code, return the corresponding howto structure
(@pxref{BFD relocation codes}).  This is called via
@samp{bfd_reloc_type_lookup}.  The corresponding field in the target
vector is named @samp{reloc_type_lookup}.
@end table

@node BFD target vector write
@subsection Output functions
@cindex @samp{BFD_JUMP_TABLE_WRITE}

The @samp{BFD_JUMP_TABLE_WRITE} macro is used for functions which deal
with writing out a BFD.

@table @samp
@item _set_arch_mach
Set the architecture and machine number for a BFD.  This is called via
@samp{bfd_set_arch_mach}.  Most targets implement this by calling
@samp{bfd_default_set_arch_mach}.  The corresponding field in the target
vector is named @samp{_bfd_set_arch_mach}.

@item _set_section_contents
Write out the contents of a section.  This is called via
@samp{bfd_set_section_contents}.  The corresponding field in the target
vector is named @samp{_bfd_set_section_contents}.
@end table

@node BFD target vector link
@subsection Linker functions
@cindex @samp{BFD_JUMP_TABLE_LINK}

The @samp{BFD_JUMP_TABLE_LINK} macro is used for functions called by the
linker.

@table @samp
@item _sizeof_headers
Return the size of the header information required for a BFD.  This is
used to implement the @samp{SIZEOF_HEADERS} linker script function.  It
is normally used to align the first section at an efficient position on
the page.  This is called via @samp{bfd_sizeof_headers}.  The
corresponding field in the target vector is named
@samp{_bfd_sizeof_headers}.

@item _bfd_get_relocated_section_contents
Read the contents of a section and apply the relocation information.
This handles both a final link and a relocatable link; in the latter
case, it adjust the relocation information as well.  This is called via
@samp{bfd_get_relocated_section_contents}.  Most targets implement it by
calling @samp{bfd_generic_get_relocated_section_contents}.

@item _bfd_relax_section
Try to use relaxation to shrink the size of a section.  This is called
by the linker when the @samp{-relax} option is used.  This is called via
@samp{bfd_relax_section}.  Most targets do not support any sort of
relaxation.

@item _bfd_link_hash_table_create
Create the symbol hash table to use for the linker.  This linker hook
permits the backend to control the size and information of the elements
in the linker symbol hash table.  This is called via
@samp{bfd_link_hash_table_create}.

@item _bfd_link_add_symbols
Given an object file or an archive, add all symbols into the linker
symbol hash table.  Use callbacks to the linker to include archive
elements in the link.  This is called via @samp{bfd_link_add_symbols}.

@item _bfd_final_link
Finish the linking process.  The linker calls this hook after all of the
input files have been read, when it is ready to finish the link and
generate the output file.  This is called via @samp{bfd_final_link}.

@item _bfd_link_split_section
I don't know what this is for.  Nothing seems to call it.  The only
non-trivial definition is in @file{som.c}.
@end table

@node BFD target vector dynamic
@subsection Dynamic linking information functions
@cindex @samp{BFD_JUMP_TABLE_DYNAMIC}

The @samp{BFD_JUMP_TABLE_DYNAMIC} macro is used for functions which read
dynamic linking information.

@table @samp
@item _get_dynamic_symtab_upper_bound
Return a sensible upper bound on the amount of memory which will be
required to read the dynamic symbol table.  In practice most targets
return the amount of memory required to hold @samp{asymbol} pointers for
all the symbols plus a trailing @samp{NULL} entry, and store the actual
symbol information in BFD private data.  This is called via
@samp{bfd_get_dynamic_symtab_upper_bound}.  The corresponding field in
the target vector is named @samp{_bfd_get_dynamic_symtab_upper_bound}.

@item _canonicalize_dynamic_symtab
Read the dynamic symbol table.  This is called via
@samp{bfd_canonicalize_dynamic_symtab}.  The corresponding field in the
target vector is named @samp{_bfd_canonicalize_dynamic_symtab}.

@item _get_dynamic_reloc_upper_bound
Return a sensible upper bound on the amount of memory which will be
required to read the dynamic relocations.  In practice most targets
return the amount of memory required to hold @samp{arelent} pointers for
all the relocations plus a trailing @samp{NULL} entry, and store the
actual relocation information in BFD private data.  This is called via
@samp{bfd_get_dynamic_reloc_upper_bound}.  The corresponding field in
the target vector is named @samp{_bfd_get_dynamic_reloc_upper_bound}.

@item _canonicalize_dynamic_reloc
Read the dynamic relocations.  This is called via
@samp{bfd_canonicalize_dynamic_reloc}.  The corresponding field in the
target vector is named @samp{_bfd_canonicalize_dynamic_reloc}.
@end table

@node BFD generated files
@section BFD generated files
@cindex generated files in bfd
@cindex bfd generated files

BFD contains several automatically generated files.  This section
describes them.  Some files are created at configure time, when you
configure BFD.  Some files are created at make time, when you build
BFD.  Some files are automatically rebuilt at make time, but only if
you configure with the @samp{--enable-maintainer-mode} option.  Some
files live in the object directory---the directory from which you run
configure---and some live in the source directory.  All files that live
in the source directory are checked into the CVS repository.

@table @file
@item bfd.h
@cindex @file{bfd.h}
@cindex @file{bfd-in3.h}
Lives in the object directory.  Created at make time from
@file{bfd-in2.h} via @file{bfd-in3.h}.  @file{bfd-in3.h} is created at
configure time from @file{bfd-in2.h}.  There are automatic dependencies
to rebuild @file{bfd-in3.h} and hence @file{bfd.h} if @file{bfd-in2.h}
changes, so you can normally ignore @file{bfd-in3.h}, and just think
about @file{bfd-in2.h} and @file{bfd.h}.

@file{bfd.h} is built by replacing a few strings in @file{bfd-in2.h}.
To see them, search for @samp{@@} in @file{bfd-in2.h}.  They mainly
control whether BFD is built for a 32 bit target or a 64 bit target.

@item bfd-in2.h
@cindex @file{bfd-in2.h}
Lives in the source directory.  Created from @file{bfd-in.h} and several
other BFD source files.  If you configure with the
@samp{--enable-maintainer-mode} option, @file{bfd-in2.h} is rebuilt
automatically when a source file changes.

@item elf32-target.h
@itemx elf64-target.h
@cindex @file{elf32-target.h}
@cindex @file{elf64-target.h}
Live in the object directory.  Created from @file{elfxx-target.h}.
These files are versions of @file{elfxx-target.h} customized for either
a 32 bit ELF target or a 64 bit ELF target.

@item libbfd.h
@cindex @file{libbfd.h}
Lives in the source directory.  Created from @file{libbfd-in.h} and
several other BFD source files.  If you configure with the
@samp{--enable-maintainer-mode} option, @file{libbfd.h} is rebuilt
automatically when a source file changes.

@item libcoff.h
@cindex @file{libcoff.h}
Lives in the source directory.  Created from @file{libcoff-in.h} and
@file{coffcode.h}.  If you configure with the
@samp{--enable-maintainer-mode} option, @file{libcoff.h} is rebuilt
automatically when a source file changes.

@item targmatch.h
@cindex @file{targmatch.h}
Lives in the object directory.  Created at make time from
@file{config.bfd}.  This file is used to map configuration triplets into
BFD target vector variable names at run time.
@end table

@node BFD multiple compilations
@section Files compiled multiple times in BFD
Several files in BFD are compiled multiple times.  By this I mean that
there are header files which contain function definitions.  These header
files are included by other files, and thus the functions are compiled
once per file which includes them.

Preprocessor macros are used to control the compilation, so that each
time the files are compiled the resulting functions are slightly
different.  Naturally, if they weren't different, there would be no
reason to compile them multiple times.

This is a not a particularly good programming technique, and future BFD
work should avoid it.

@itemize @bullet
@item
Since this technique is rarely used, even experienced C programmers find
it confusing.

@item
It is difficult to debug programs which use BFD, since there is no way
to describe which version of a particular function you are looking at.

@item
Programs which use BFD wind up incorporating two or more slightly
different versions of the same function, which wastes space in the
executable.

@item
This technique is never required nor is it especially efficient.  It is
always possible to use statically initialized structures holding
function pointers and magic constants instead.
@end itemize

The following is a list of the files which are compiled multiple times.

@table @file
@item aout-target.h
@cindex @file{aout-target.h}
Describes a few functions and the target vector for a.out targets.  This
is used by individual a.out targets with different definitions of
@samp{N_TXTADDR} and similar a.out macros.

@item aoutf1.h
@cindex @file{aoutf1.h}
Implements standard SunOS a.out files.  In principle it supports 64 bit
a.out targets based on the preprocessor macro @samp{ARCH_SIZE}, but
since all known a.out targets are 32 bits, this code may or may not
work.  This file is only included by a few other files, and it is
difficult to justify its existence.

@item aoutx.h
@cindex @file{aoutx.h}
Implements basic a.out support routines.  This file can be compiled for
either 32 or 64 bit support.  Since all known a.out targets are 32 bits,
the 64 bit support may or may not work.  I believe the original
intention was that this file would only be included by @samp{aout32.c}
and @samp{aout64.c}, and that other a.out targets would simply refer to
the functions it defined.  Unfortunately, some other a.out targets
started including it directly, leading to a somewhat confused state of
affairs.

@item coffcode.h
@cindex @file{coffcode.h}
Implements basic COFF support routines.  This file is included by every
COFF target.  It implements code which handles COFF magic numbers as
well as various hook functions called by the generic COFF functions in
@file{coffgen.c}.  This file is controlled by a number of different
macros, and more are added regularly.

@item coffswap.h
@cindex @file{coffswap.h}
Implements COFF swapping routines.  This file is included by
@file{coffcode.h}, and thus by every COFF target.  It implements the
routines which swap COFF structures between internal and external
format.  The main control for this file is the external structure
definitions in the files in the @file{include/coff} directory.  A COFF
target file will include one of those files before including
@file{coffcode.h} and thus @file{coffswap.h}.  There are a few other
macros which affect @file{coffswap.h} as well, mostly describing whether
certain fields are present in the external structures.

@item ecoffswap.h
@cindex @file{ecoffswap.h}
Implements ECOFF swapping routines.  This is like @file{coffswap.h}, but
for ECOFF.  It is included by the ECOFF target files (of which there are
only two).  The control is the preprocessor macro @samp{ECOFF_32} or
@samp{ECOFF_64}.

@item elfcode.h
@cindex @file{elfcode.h}
Implements ELF functions that use external structure definitions.  This
file is included by two other files: @file{elf32.c} and @file{elf64.c}.
It is controlled by the @samp{ARCH_SIZE} macro which is defined to be
@samp{32} or @samp{64} before including it.  The @samp{NAME} macro is
used internally to give the functions different names for the two target
sizes.

@item elfcore.h
@cindex @file{elfcore.h}
Like @file{elfcode.h}, but for functions that are specific to ELF core
files.  This is included only by @file{elfcode.h}.

@item elfxx-target.h
@cindex @file{elfxx-target.h}
This file is the source for the generated files @file{elf32-target.h}
and @file{elf64-target.h}, one of which is included by every ELF target.
It defines the ELF target vector.

@item freebsd.h
@cindex @file{freebsd.h}
Presumably intended to be included by all FreeBSD targets, but in fact
there is only one such target, @samp{i386-freebsd}.  This defines a
function used to set the right magic number for FreeBSD, as well as
various macros, and includes @file{aout-target.h}.

@item netbsd.h
@cindex @file{netbsd.h}
Like @file{freebsd.h}, except that there are several files which include
it.

@item nlm-target.h
@cindex @file{nlm-target.h}
Defines the target vector for a standard NLM target.

@item nlmcode.h
@cindex @file{nlmcode.h}
Like @file{elfcode.h}, but for NLM targets.  This is only included by
@file{nlm32.c} and @file{nlm64.c}, both of which define the macro
@samp{ARCH_SIZE} to an appropriate value.  There are no 64 bit NLM
targets anyhow, so this is sort of useless.

@item nlmswap.h
@cindex @file{nlmswap.h}
Like @file{coffswap.h}, but for NLM targets.  This is included by each
NLM target, but I think it winds up compiling to the exact same code for
every target, and as such is fairly useless.

@item peicode.h
@cindex @file{peicode.h}
Provides swapping routines and other hooks for PE targets.
@file{coffcode.h} will include this rather than @file{coffswap.h} for a
PE target.  This defines PE specific versions of the COFF swapping
routines, and also defines some macros which control @file{coffcode.h}
itself.
@end table

@node BFD relocation handling
@section BFD relocation handling
@cindex bfd relocation handling
@cindex relocations in bfd

The handling of relocations is one of the more confusing aspects of BFD.
Relocation handling has been implemented in various different ways, all
somewhat incompatible, none perfect.

@menu
* BFD relocation concepts::	BFD relocation concepts
* BFD relocation functions::	BFD relocation functions
* BFD relocation codes::	BFD relocation codes
* BFD relocation future::	BFD relocation future
@end menu

@node BFD relocation concepts
@subsection BFD relocation concepts

A relocation is an action which the linker must take when linking.  It
describes a change to the contents of a section.  The change is normally
based on the final value of one or more symbols.  Relocations are
created by the assembler when it creates an object file.

Most relocations are simple.  A typical simple relocation is to set 32
bits at a given offset in a section to the value of a symbol.  This type
of relocation would be generated for code like @code{int *p = &i;} where
@samp{p} and @samp{i} are global variables.  A relocation for the symbol
@samp{i} would be generated such that the linker would initialize the
area of memory which holds the value of @samp{p} to the value of the
symbol @samp{i}.

Slightly more complex relocations may include an addend, which is a
constant to add to the symbol value before using it.  In some cases a
relocation will require adding the symbol value to the existing contents
of the section in the object file.  In others the relocation will simply
replace the contents of the section with the symbol value.  Some
relocations are PC relative, so that the value to be stored in the
section is the difference between the value of a symbol and the final
address of the section contents.

In general, relocations can be arbitrarily complex.  For example,
relocations used in dynamic linking systems often require the linker to
allocate space in a different section and use the offset within that
section as the value to store.  In the IEEE object file format,
relocations may involve arbitrary expressions.

When doing a relocatable link, the linker may or may not have to do
anything with a relocation, depending upon the definition of the
relocation.  Simple relocations generally do not require any special
action.

@node BFD relocation functions
@subsection BFD relocation functions

In BFD, each section has an array of @samp{arelent} structures.  Each
structure has a pointer to a symbol, an address within the section, an
addend, and a pointer to a @samp{reloc_howto_struct} structure.  The
howto structure has a bunch of fields describing the reloc, including a
type field.  The type field is specific to the object file format
backend; none of the generic code in BFD examines it.

Originally, the function @samp{bfd_perform_relocation} was supposed to
handle all relocations.  In theory, many relocations would be simple
enough to be described by the fields in the howto structure.  For those
that weren't, the howto structure included a @samp{special_function}
field to use as an escape.

While this seems plausible, a look at @samp{bfd_perform_relocation}
shows that it failed.  The function has odd special cases.  Some of the
fields in the howto structure, such as @samp{pcrel_offset}, were not
adequately documented.

The linker uses @samp{bfd_perform_relocation} to do all relocations when
the input and output file have different formats (e.g., when generating
S-records).  The generic linker code, which is used by all targets which
do not define their own special purpose linker, uses
@samp{bfd_get_relocated_section_contents}, which for most targets turns
into a call to @samp{bfd_generic_get_relocated_section_contents}, which
calls @samp{bfd_perform_relocation}.  So @samp{bfd_perform_relocation}
is still widely used, which makes it difficult to change, since it is
difficult to test all possible cases.

The assembler used @samp{bfd_perform_relocation} for a while.  This
turned out to be the wrong thing to do, since
@samp{bfd_perform_relocation} was written to handle relocations on an
existing object file, while the assembler needed to create relocations
in a new object file.  The assembler was changed to use the new function
@samp{bfd_install_relocation} instead, and @samp{bfd_install_relocation}
was created as a copy of @samp{bfd_perform_relocation}.

Unfortunately, the work did not progress any farther, so
@samp{bfd_install_relocation} remains a simple copy of
@samp{bfd_perform_relocation}, with all the odd special cases and
confusing code.  This again is difficult to change, because again any
change can affect any assembler target, and so is difficult to test.

The new linker, when using the same object file format for all input
files and the output file, does not convert relocations into
@samp{arelent} structures, so it can not use
@samp{bfd_perform_relocation} at all.  Instead, users of the new linker
are expected to write a @samp{relocate_section} function which will
handle relocations in a target specific fashion.

There are two helper functions for target specific relocation:
@samp{_bfd_final_link_relocate} and @samp{_bfd_relocate_contents}.
These functions use a howto structure, but they @emph{do not} use the
@samp{special_function} field.  Since the functions are normally called
from target specific code, the @samp{special_function} field adds
little; any relocations which require special handling can be handled
without calling those functions.

So, if you want to add a new target, or add a new relocation to an
existing target, you need to do the following:

@itemize @bullet
@item
Make sure you clearly understand what the contents of the section should
look like after assembly, after a relocatable link, and after a final
link.  Make sure you clearly understand the operations the linker must
perform during a relocatable link and during a final link.

@item
Write a howto structure for the relocation.  The howto structure is
flexible enough to represent any relocation which should be handled by
setting a contiguous bitfield in the destination to the value of a
symbol, possibly with an addend, possibly adding the symbol value to the
value already present in the destination.

@item
Change the assembler to generate your relocation.  The assembler will
call @samp{bfd_install_relocation}, so your howto structure has to be
able to handle that.  You may need to set the @samp{special_function}
field to handle assembly correctly.  Be careful to ensure that any code
you write to handle the assembler will also work correctly when doing a
relocatable link.  For example, see @samp{bfd_elf_generic_reloc}.

@item
Test the assembler.  Consider the cases of relocation against an
undefined symbol, a common symbol, a symbol defined in the object file
in the same section, and a symbol defined in the object file in a
different section.  These cases may not all be applicable for your
reloc.

@item
If your target uses the new linker, which is recommended, add any
required handling to the target specific relocation function.  In simple
cases this will just involve a call to @samp{_bfd_final_link_relocate}
or @samp{_bfd_relocate_contents}, depending upon the definition of the
relocation and whether the link is relocatable or not.

@item
Test the linker.  Test the case of a final link.  If the relocation can
overflow, use a linker script to force an overflow and make sure the
error is reported correctly.  Test a relocatable link, whether the
symbol is defined or undefined in the relocatable output.  For both the
final and relocatable link, test the case when the symbol is a common
symbol, when the symbol looked like a common symbol but became a defined
symbol, when the symbol is defined in a different object file, and when
the symbol is defined in the same object file.

@item
In order for linking to another object file format, such as S-records,
to work correctly, @samp{bfd_perform_relocation} has to do the right
thing for the relocation.  You may need to set the
@samp{special_function} field to handle this correctly.  Test this by
doing a link in which the output object file format is S-records.

@item
Using the linker to generate relocatable output in a different object
file format is impossible in the general case, so you generally don't
have to worry about that.  The GNU linker makes sure to stop that from
happening when an input file in a different format has relocations.

Linking input files of different object file formats together is quite
unusual, but if you're really dedicated you may want to consider testing
this case, both when the output object file format is the same as your
format, and when it is different.
@end itemize

@node BFD relocation codes
@subsection BFD relocation codes

BFD has another way of describing relocations besides the howto
structures described above: the enum @samp{bfd_reloc_code_real_type}.

Every known relocation type can be described as a value in this
enumeration.  The enumeration contains many target specific relocations,
but where two or more targets have the same relocation, a single code is
used.  For example, the single value @samp{BFD_RELOC_32} is used for all
simple 32 bit relocation types.

The main purpose of this relocation code is to give the assembler some
mechanism to create @samp{arelent} structures.  In order for the
assembler to create an @samp{arelent} structure, it has to be able to
obtain a howto structure.  The function @samp{bfd_reloc_type_lookup},
which simply calls the target vector entry point
@samp{reloc_type_lookup}, takes a relocation code and returns a howto
structure.

The function @samp{bfd_get_reloc_code_name} returns the name of a
relocation code.  This is mainly used in error messages.

Using both howto structures and relocation codes can be somewhat
confusing.  There are many processor specific relocation codes.
However, the relocation is only fully defined by the howto structure.
The same relocation code will map to different howto structures in
different object file formats.  For example, the addend handling may be
different.

Most of the relocation codes are not really general.  The assembler can
not use them without already understanding what sorts of relocations can
be used for a particular target.  It might be possible to replace the
relocation codes with something simpler.

@node BFD relocation future
@subsection BFD relocation future

Clearly the current BFD relocation support is in bad shape.  A
wholescale rewrite would be very difficult, because it would require
thorough testing of every BFD target.  So some sort of incremental
change is required.

My vague thoughts on this would involve defining a new, clearly defined,
howto structure.  Some mechanism would be used to determine which type
of howto structure was being used by a particular format.

The new howto structure would clearly define the relocation behaviour in
the case of an assembly, a relocatable link, and a final link.  At
least one special function would be defined as an escape, and it might
make sense to define more.

One or more generic functions similar to @samp{bfd_perform_relocation}
would be written to handle the new howto structure.

This should make it possible to write a generic version of the relocate
section functions used by the new linker.  The target specific code
would provide some mechanism (a function pointer or an initial
conversion) to convert target specific relocations into howto
structures.

Ideally it would be possible to use this generic relocate section
function for the generic linker as well.  That is, it would replace the
@samp{bfd_generic_get_relocated_section_contents} function which is
currently normally used.

For the special case of ELF dynamic linking, more consideration needs to
be given to writing ELF specific but ELF target generic code to handle
special relocation types such as GOT and PLT.

@node BFD ELF support
@section BFD ELF support
@cindex elf support in bfd
@cindex bfd elf support

The ELF object file format is defined in two parts: a generic ABI and a
processor specific supplement.  The ELF support in BFD is split in a
similar fashion.  The processor specific support is largely kept within
a single file.  The generic support is provided by several other files.
The processor specific support provides a set of function pointers and
constants used by the generic support.

@menu
* BFD ELF sections and segments::	ELF sections and segments
* BFD ELF generic support::		BFD ELF generic support
* BFD ELF processor specific support::	BFD ELF processor specific support
* BFD ELF core files::			BFD ELF core files
* BFD ELF future::			BFD ELF future
@end menu

@node BFD ELF sections and segments
@subsection ELF sections and segments

The ELF ABI permits a file to have either sections or segments or both.
Relocatable object files conventionally have only sections.
Executables conventionally have both.  Core files conventionally have
only program segments.

ELF sections are similar to sections in other object file formats: they
have a name, a VMA, file contents, flags, and other miscellaneous
information.  ELF relocations are stored in sections of a particular
type; BFD automatically converts these sections into internal relocation
information.

ELF program segments are intended for fast interpretation by a system
loader.  They have a type, a VMA, an LMA, file contents, and a couple of
other fields.  When an ELF executable is run on a Unix system, the
system loader will examine the program segments to decide how to load
it.  The loader will ignore the section information.  Loadable program
segments (type @samp{PT_LOAD}) are directly loaded into memory.  Other
program segments are interpreted by the loader, and generally provide
dynamic linking information.

When an ELF file has both program segments and sections, an ELF program
segment may encompass one or more ELF sections, in the sense that the
portion of the file which corresponds to the program segment may include
the portions of the file corresponding to one or more sections.  When
there is more than one section in a loadable program segment, the
relative positions of the section contents in the file must correspond
to the relative positions they should hold when the program segment is
loaded.  This requirement should be obvious if you consider that the
system loader will load an entire program segment at a time.

On a system which supports dynamic paging, such as any native Unix
system, the contents of a loadable program segment must be at the same
offset in the file as in memory, modulo the memory page size used on the
system.  This is because the system loader will map the file into memory
starting at the start of a page.  The system loader can easily remap
entire pages to the correct load address.  However, if the contents of
the file were not correctly aligned within the page, the system loader
would have to shift the contents around within the page, which is too
expensive.  For example, if the LMA of a loadable program segment is
@samp{0x40080} and the page size is @samp{0x1000}, then the position of
the segment contents within the file must equal @samp{0x80} modulo
@samp{0x1000}.

BFD has only a single set of sections.  It does not provide any generic
way to examine both sections and segments.  When BFD is used to open an
object file or executable, the BFD sections will represent ELF sections.
When BFD is used to open a core file, the BFD sections will represent
ELF program segments.

When BFD is used to examine an object file or executable, any program
segments will be read to set the LMA of the sections.  This is because
ELF sections only have a VMA, while ELF program segments have both a VMA
and an LMA.  Any program segments will be copied by the
@samp{copy_private} entry points.  They will be printed by the
@samp{print_private} entry point.  Otherwise, the program segments are
ignored.  In particular, programs which use BFD currently have no direct
access to the program segments.

When BFD is used to create an executable, the program segments will be
created automatically based on the section information.  This is done in
the function @samp{assign_file_positions_for_segments} in @file{elf.c}.
This function has been tweaked many times, and probably still has
problems that arise in particular cases.

There is a hook which may be used to explicitly define the program
segments when creating an executable: the @samp{bfd_record_phdr}
function in @file{bfd.c}.  If this function is called, BFD will not
create program segments itself, but will only create the program
segments specified by the caller.  The linker uses this function to
implement the @samp{PHDRS} linker script command.

@node BFD ELF generic support
@subsection BFD ELF generic support

In general, functions which do not read external data from the ELF file
are found in @file{elf.c}.  They operate on the internal forms of the
ELF structures, which are defined in @file{include/elf/internal.h}.  The
internal structures are defined in terms of @samp{bfd_vma}, and so may
be used for both 32 bit and 64 bit ELF targets.

The file @file{elfcode.h} contains functions which operate on the
external data.  @file{elfcode.h} is compiled twice, once via
@file{elf32.c} with @samp{ARCH_SIZE} defined as @samp{32}, and once via
@file{elf64.c} with @samp{ARCH_SIZE} defined as @samp{64}.
@file{elfcode.h} includes functions to swap the ELF structures in and
out of external form, as well as a few more complex functions.

Linker support is found in @file{elflink.c}.  The
linker support is only used if the processor specific file defines
@samp{elf_backend_relocate_section}, which is required to relocate the
section contents.  If that macro is not defined, the generic linker code
is used, and relocations are handled via @samp{bfd_perform_relocation}.

The core file support is in @file{elfcore.h}, which is compiled twice,
for both 32 and 64 bit support.  The more interesting cases of core file
support only work on a native system which has the @file{sys/procfs.h}
header file.  Without that file, the core file support does little more
than read the ELF program segments as BFD sections.

The BFD internal header file @file{elf-bfd.h} is used for communication
among these files and the processor specific files.

The default entries for the BFD ELF target vector are found mainly in
@file{elf.c}.  Some functions are found in @file{elfcode.h}.

The processor specific files may override particular entries in the
target vector, but most do not, with one exception: the
@samp{bfd_reloc_type_lookup} entry point is always processor specific.

@node BFD ELF processor specific support
@subsection BFD ELF processor specific support

By convention, the processor specific support for a particular processor
will be found in @file{elf@var{nn}-@var{cpu}.c}, where @var{nn} is
either 32 or 64, and @var{cpu} is the name of the processor.

@menu
* BFD ELF processor required::	Required processor specific support
* BFD ELF processor linker::	Processor specific linker support
* BFD ELF processor other::	Other processor specific support options
@end menu

@node BFD ELF processor required
@subsubsection Required processor specific support

When writing a @file{elf@var{nn}-@var{cpu}.c} file, you must do the
following:

@itemize @bullet
@item
Define either @samp{TARGET_BIG_SYM} or @samp{TARGET_LITTLE_SYM}, or
both, to a unique C name to use for the target vector.  This name should
appear in the list of target vectors in @file{targets.c}, and will also
have to appear in @file{config.bfd} and @file{configure.in}.  Define
@samp{TARGET_BIG_SYM} for a big-endian processor,
@samp{TARGET_LITTLE_SYM} for a little-endian processor, and define both
for a bi-endian processor.
@item
Define either @samp{TARGET_BIG_NAME} or @samp{TARGET_LITTLE_NAME}, or
both, to a string used as the name of the target vector.  This is the
name which a user of the BFD tool would use to specify the object file
format.  It would normally appear in a linker emulation parameters
file.
@item
Define @samp{ELF_ARCH} to the BFD architecture (an element of the
@samp{bfd_architecture} enum, typically @samp{bfd_arch_@var{cpu}}).
@item
Define @samp{ELF_MACHINE_CODE} to the magic number which should appear
in the @samp{e_machine} field of the ELF header.  As of this writing,
these magic numbers are assigned by Caldera; if you want to get a magic
number for a particular processor, try sending a note to
@email{registry@@caldera.com}.  In the BFD sources, the magic numbers are
found in @file{include/elf/common.h}; they have names beginning with
@samp{EM_}.
@item
Define @samp{ELF_MAXPAGESIZE} to the maximum size of a virtual page in
memory.  This can normally be found at the start of chapter 5 in the
processor specific supplement.  For a processor which will only be used
in an embedded system, or which has no memory management hardware, this
can simply be @samp{1}.
@item
If the format should use @samp{Rel} rather than @samp{Rela} relocations,
define @samp{USE_REL}.  This is normally defined in chapter 4 of the
processor specific supplement.

In the absence of a supplement, it's easier to work with @samp{Rela}
relocations.  @samp{Rela} relocations will require more space in object
files (but not in executables, except when using dynamic linking).
However, this is outweighed by the simplicity of addend handling when
using @samp{Rela} relocations.  With @samp{Rel} relocations, the addend
must be stored in the section contents, which makes relocatable links
more complex.

For example, consider C code like @code{i = a[1000];} where @samp{a} is
a global array.  The instructions which load the value of @samp{a[1000]}
will most likely use a relocation which refers to the symbol
representing @samp{a}, with an addend that gives the offset from the
start of @samp{a} to element @samp{1000}.  When using @samp{Rel}
relocations, that addend must be stored in the instructions themselves.
If you are adding support for a RISC chip which uses two or more
instructions to load an address, then the addend may not fit in a single
instruction, and will have to be somehow split among the instructions.
This makes linking awkward, particularly when doing a relocatable link
in which the addend may have to be updated.  It can be done---the MIPS
ELF support does it---but it should be avoided when possible.

It is possible, though somewhat awkward, to support both @samp{Rel} and
@samp{Rela} relocations for a single target; @file{elf64-mips.c} does it
by overriding the relocation reading and writing routines.
@item
Define howto structures for all the relocation types.
@item
Define a @samp{bfd_reloc_type_lookup} routine.  This must be named
@samp{bfd_elf@var{nn}_bfd_reloc_type_lookup}, and may be either a
function or a macro.  It must translate a BFD relocation code into a
howto structure.  This is normally a table lookup or a simple switch.
@item
If using @samp{Rel} relocations, define @samp{elf_info_to_howto_rel}.
If using @samp{Rela} relocations, define @samp{elf_info_to_howto}.
Either way, this is a macro defined as the name of a function which
takes an @samp{arelent} and a @samp{Rel} or @samp{Rela} structure, and
sets the @samp{howto} field of the @samp{arelent} based on the
@samp{Rel} or @samp{Rela} structure.  This is normally uses
@samp{ELF@var{nn}_R_TYPE} to get the ELF relocation type and uses it as
an index into a table of howto structures.
@end itemize

You must also add the magic number for this processor to the
@samp{prep_headers} function in @file{elf.c}.

You must also create a header file in the @file{include/elf} directory
called @file{@var{cpu}.h}.  This file should define any target specific 
information which may be needed outside of the BFD code.  In particular
it should use the @samp{START_RELOC_NUMBERS}, @samp{RELOC_NUMBER},
@samp{FAKE_RELOC}, @samp{EMPTY_RELOC} and @samp{END_RELOC_NUMBERS}
macros to create a table mapping the number used to identify a
relocation to a name describing that relocation.

While not a BFD component, you probably also want to make the binutils
program @samp{readelf} parse your ELF objects.  For this, you need to add
code for @code{EM_@var{cpu}} as appropriate in @file{binutils/readelf.c}.

@node BFD ELF processor linker
@subsubsection Processor specific linker support

The linker will be much more efficient if you define a relocate section
function.  This will permit BFD to use the ELF specific linker support.

If you do not define a relocate section function, BFD must use the
generic linker support, which requires converting all symbols and
relocations into BFD @samp{asymbol} and @samp{arelent} structures.  In
this case, relocations will be handled by calling
@samp{bfd_perform_relocation}, which will use the howto structures you
have defined.  @xref{BFD relocation handling}.

In order to support linking into a different object file format, such as
S-records, @samp{bfd_perform_relocation} must work correctly with your
howto structures, so you can't skip that step.  However, if you define
the relocate section function, then in the normal case of linking into
an ELF file the linker will not need to convert symbols and relocations,
and will be much more efficient.

To use a relocation section function, define the macro
@samp{elf_backend_relocate_section} as the name of a function which will
take the contents of a section, as well as relocation, symbol, and other
information, and modify the section contents according to the relocation
information.  In simple cases, this is little more than a loop over the
relocations which computes the value of each relocation and calls
@samp{_bfd_final_link_relocate}.  The function must check for a
relocatable link, and in that case normally needs to do nothing other
than adjust the addend for relocations against a section symbol.

The complex cases generally have to do with dynamic linker support.  GOT
and PLT relocations must be handled specially, and the linker normally
arranges to set up the GOT and PLT sections while handling relocations.
When generating a shared library, random relocations must normally be
copied into the shared library, or converted to RELATIVE relocations
when possible.

@node BFD ELF processor other
@subsubsection Other processor specific support options

There are many other macros which may be defined in
@file{elf@var{nn}-@var{cpu}.c}.  These macros may be found in
@file{elfxx-target.h}.

Macros may be used to override some of the generic ELF target vector
functions.

Several processor specific hook functions which may be defined as
macros.  These functions are found as function pointers in the
@samp{elf_backend_data} structure defined in @file{elf-bfd.h}.  In
general, a hook function is set by defining a macro
@samp{elf_backend_@var{name}}.

There are a few processor specific constants which may also be defined.
These are again found in the @samp{elf_backend_data} structure.

I will not define the various functions and constants here; see the
comments in @file{elf-bfd.h}.

Normally any odd characteristic of a particular ELF processor is handled
via a hook function.  For example, the special @samp{SHN_MIPS_SCOMMON}
section number found in MIPS ELF is handled via the hooks
@samp{section_from_bfd_section}, @samp{symbol_processing},
@samp{add_symbol_hook}, and @samp{output_symbol_hook}.

Dynamic linking support, which involves processor specific relocations
requiring special handling, is also implemented via hook functions.

@node BFD ELF core files
@subsection BFD ELF core files
@cindex elf core files

On native ELF Unix systems, core files are generated without any
sections.  Instead, they only have program segments.

When BFD is used to read an ELF core file, the BFD sections will
actually represent program segments.  Since ELF program segments do not
have names, BFD will invent names like @samp{segment@var{n}} where
@var{n} is a number.

A single ELF program segment may include both an initialized part and an
uninitialized part.  The size of the initialized part is given by the
@samp{p_filesz} field.  The total size of the segment is given by the
@samp{p_memsz} field.  If @samp{p_memsz} is larger than @samp{p_filesz},
then the extra space is uninitialized, or, more precisely, initialized
to zero.

BFD will represent such a program segment as two different sections.
The first, named @samp{segment@var{n}a}, will represent the initialized
part of the program segment.  The second, named @samp{segment@var{n}b},
will represent the uninitialized part.

ELF core files store special information such as register values in
program segments with the type @samp{PT_NOTE}.  BFD will attempt to
interpret the information in these segments, and will create additional
sections holding the information.  Some of this interpretation requires
information found in the host header file @file{sys/procfs.h}, and so
will only work when BFD is built on a native system.

BFD does not currently provide any way to create an ELF core file.  In
general, BFD does not provide a way to create core files.  The way to
implement this would be to write @samp{bfd_set_format} and
@samp{bfd_write_contents} routines for the @samp{bfd_core} type; see
@ref{BFD target vector format}.

@node BFD ELF future
@subsection BFD ELF future

The current dynamic linking support has too much code duplication.
While each processor has particular differences, much of the dynamic
linking support is quite similar for each processor.  The GOT and PLT
are handled in fairly similar ways, the details of -Bsymbolic linking
are generally similar, etc.  This code should be reworked to use more
generic functions, eliminating the duplication.

Similarly, the relocation handling has too much duplication.  Many of
the @samp{reloc_type_lookup} and @samp{info_to_howto} functions are
quite similar.  The relocate section functions are also often quite
similar, both in the standard linker handling and the dynamic linker
handling.  Many of the COFF processor specific backends share a single
relocate section function (@samp{_bfd_coff_generic_relocate_section}),
and it should be possible to do something like this for the ELF targets
as well.

The appearance of the processor specific magic number in
@samp{prep_headers} in @file{elf.c} is somewhat bogus.  It should be
possible to add support for a new processor without changing the generic
support.

The processor function hooks and constants are ad hoc and need better
documentation.

@node BFD glossary
@section BFD glossary
@cindex glossary for bfd
@cindex bfd glossary

This is a short glossary of some BFD terms.

@table @asis
@item a.out
The a.out object file format.  The original Unix object file format.
Still used on SunOS, though not Solaris.  Supports only three sections.

@item archive
A collection of object files produced and manipulated by the @samp{ar}
program.

@item backend
The implementation within BFD of a particular object file format.  The
set of functions which appear in a particular target vector.

@item BFD
The BFD library itself.  Also, each object file, archive, or executable
opened by the BFD library has the type @samp{bfd *}, and is sometimes
referred to as a bfd.

@item COFF
The Common Object File Format.  Used on Unix SVR3.  Used by some
embedded targets, although ELF is normally better.

@item DLL
A shared library on Windows.

@item dynamic linker
When a program linked against a shared library is run, the dynamic
linker will locate the appropriate shared library and arrange to somehow
include it in the running image.

@item dynamic object
Another name for an ELF shared library.

@item ECOFF
The Extended Common Object File Format.  Used on Alpha Digital Unix
(formerly OSF/1), as well as Ultrix and Irix 4.  A variant of COFF.

@item ELF
The Executable and Linking Format.  The object file format used on most
modern Unix systems, including GNU/Linux, Solaris, Irix, and SVR4.  Also
used on many embedded systems.

@item executable
A program, with instructions and symbols, and perhaps dynamic linking
information.  Normally produced by a linker.

@item LMA
Load Memory Address.  This is the address at which a section will be
loaded.  Compare with VMA, below.

@item NLM
NetWare Loadable Module.  Used to describe the format of an object which
be loaded into NetWare, which is some kind of PC based network server
program.

@item object file
A binary file including machine instructions, symbols, and relocation
information.  Normally produced by an assembler.

@item object file format
The format of an object file.  Typically object files and executables
for a particular system are in the same format, although executables
will not contain any relocation information.

@item PE
The Portable Executable format.  This is the object file format used for
Windows (specifically, Win32) object files.  It is based closely on
COFF, but has a few significant differences.

@item PEI
The Portable Executable Image format.  This is the object file format
used for Windows (specifically, Win32) executables.  It is very similar
to PE, but includes some additional header information.

@item relocations
Information used by the linker to adjust section contents.  Also called
relocs.

@item section
Object files and executable are composed of sections.  Sections have
optional data and optional relocation information.

@item shared library
A library of functions which may be used by many executables without
actually being linked into each executable.  There are several different
implementations of shared libraries, each having slightly different
features.

@item symbol
Each object file and executable may have a list of symbols, often
referred to as the symbol table.  A symbol is basically a name and an
address.  There may also be some additional information like the type of
symbol, although the type of a symbol is normally something simple like
function or object, and should be confused with the more complex C
notion of type.  Typically every global function and variable in a C
program will have an associated symbol.

@item target vector
A set of functions which implement support for a particular object file
format.  The @samp{bfd_target} structure.

@item Win32
The current Windows API, implemented by Windows 95 and later and Windows
NT 3.51 and later, but not by Windows 3.1.

@item XCOFF
The eXtended Common Object File Format.  Used on AIX.  A variant of
COFF, with a completely different symbol table implementation.

@item VMA
Virtual Memory Address.  This is the address a section will have when
an executable is run.  Compare with LMA, above.
@end table

@node Index
@unnumberedsec Index
@printindex cp

@contents
@bye