1992-11-05 16:17:56 +08:00
|
|
|
|
/* Main header file for the bfd library -- portable access to object files.
|
1993-04-15 12:53:14 +08:00
|
|
|
|
Copyright 1990, 1991, 1992, 1993 Free Software Foundation, Inc.
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
Contributed by Cygnus Support.
|
|
|
|
|
|
1993-08-06 02:27:16 +08:00
|
|
|
|
** NOTE: bfd.h and bfd-in2.h are GENERATED files. Don't change them;
|
|
|
|
|
** instead, change bfd-in.h or the other BFD source files processed to
|
|
|
|
|
** generate these files.
|
|
|
|
|
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
This file is part of BFD, the Binary File Descriptor library.
|
|
|
|
|
|
|
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
|
|
|
it under the terms of the GNU General Public License as published by
|
|
|
|
|
the Free Software Foundation; either version 2 of the License, or
|
|
|
|
|
(at your option) any later version.
|
|
|
|
|
|
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
|
GNU General Public License for more details.
|
|
|
|
|
|
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
|
|
|
along with this program; if not, write to the Free Software
|
|
|
|
|
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
|
|
|
|
|
|
|
|
|
|
/* bfd.h -- The only header file required by users of the bfd library
|
|
|
|
|
|
1992-11-05 16:17:56 +08:00
|
|
|
|
The bfd.h file is generated from bfd-in.h and various .c files; if you
|
|
|
|
|
change it, your changes will probably be lost.
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
|
|
|
|
All the prototypes and definitions following the comment "THE FOLLOWING
|
|
|
|
|
IS EXTRACTED FROM THE SOURCE" are extracted from the source files for
|
|
|
|
|
BFD. If you change it, someone oneday will extract it from the source
|
|
|
|
|
again, and your changes will be lost. To save yourself from this bind,
|
|
|
|
|
change the definitions in the source in the bfd directory. Type "make
|
|
|
|
|
docs" and then "make headers" in that directory, and magically this file
|
|
|
|
|
will change to reflect your changes.
|
|
|
|
|
|
|
|
|
|
If you don't have the tools to perform the extraction, then you are
|
|
|
|
|
safe from someone on your system trampling over your header files.
|
|
|
|
|
You should still maintain the equivalence between the source and this
|
|
|
|
|
file though; every change you make to the .c file should be reflected
|
|
|
|
|
here. */
|
|
|
|
|
|
|
|
|
|
#ifndef __BFD_H_SEEN__
|
|
|
|
|
#define __BFD_H_SEEN__
|
|
|
|
|
|
|
|
|
|
#include "ansidecl.h"
|
|
|
|
|
#include "obstack.h"
|
|
|
|
|
|
1993-06-16 10:08:29 +08:00
|
|
|
|
#define BFD_VERSION "2.2"
|
|
|
|
|
|
|
|
|
|
#define BFD_ARCH_SIZE @WORDSIZE@
|
|
|
|
|
|
|
|
|
|
#if BFD_ARCH_SIZE >= 64
|
|
|
|
|
#define BFD64
|
|
|
|
|
#endif
|
|
|
|
|
|
1993-08-06 02:27:16 +08:00
|
|
|
|
#ifndef INLINE
|
|
|
|
|
#if __GNUC__ >= 2
|
|
|
|
|
#define INLINE __inline__
|
|
|
|
|
#else
|
|
|
|
|
#define INLINE
|
|
|
|
|
#endif
|
|
|
|
|
#endif
|
|
|
|
|
|
1993-06-21 08:30:11 +08:00
|
|
|
|
/* 64-bit type definition (if any) from bfd's sysdep.h goes here */
|
|
|
|
|
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
|
|
|
|
/* forward declaration */
|
|
|
|
|
typedef struct _bfd bfd;
|
|
|
|
|
|
1993-06-16 10:08:29 +08:00
|
|
|
|
/* To squelch erroneous compiler warnings ("illegal pointer
|
|
|
|
|
combination") from the SVR3 compiler, we would like to typedef
|
|
|
|
|
boolean to int (it doesn't like functions which return boolean.
|
|
|
|
|
Making sure they are never implicitly declared to return int
|
|
|
|
|
doesn't seem to help). But this file is not configured based on
|
|
|
|
|
the host. */
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
/* General rules: functions which are boolean return true on success
|
|
|
|
|
and false on failure (unless they're a predicate). -- bfd.doc */
|
|
|
|
|
/* I'm sure this is going to break something and someone is going to
|
|
|
|
|
force me to change it. */
|
1991-10-04 10:49:06 +08:00
|
|
|
|
/* typedef enum boolean {false, true} boolean; */
|
|
|
|
|
/* Yup, SVR4 has a "typedef enum boolean" in <sys/types.h> -fnf */
|
|
|
|
|
typedef enum bfd_boolean {false, true} boolean;
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
1992-11-05 16:17:56 +08:00
|
|
|
|
/* A pointer to a position in a file. */
|
|
|
|
|
/* FIXME: This should be using off_t from <sys/types.h>.
|
|
|
|
|
For now, try to avoid breaking stuff by not including <sys/types.h> here.
|
|
|
|
|
This will break on systems with 64-bit file offsets (e.g. 4.4BSD).
|
|
|
|
|
Probably the best long-term answer is to avoid using file_ptr AND off_t
|
|
|
|
|
in this header file, and to handle this in the BFD implementation
|
|
|
|
|
rather than in its interface. */
|
|
|
|
|
/* typedef off_t file_ptr; */
|
|
|
|
|
typedef long int file_ptr;
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
1993-06-16 10:08:29 +08:00
|
|
|
|
/* Support for different sizes of target format ints and addresses. If the
|
|
|
|
|
host implements 64-bit values, it defines HOST_64_BIT to be the appropriate
|
|
|
|
|
type. Otherwise, this code will fall back on gcc's "long long" type if gcc
|
|
|
|
|
is being used. HOST_64_BIT must be defined in such a way as to be a valid
|
|
|
|
|
type name by itself or with "unsigned" prefixed. It should be a signed
|
|
|
|
|
type by itself.
|
|
|
|
|
|
|
|
|
|
If neither is the case, then compilation will fail if 64-bit targets are
|
|
|
|
|
requested. If you don't request any 64-bit targets, you should be safe. */
|
|
|
|
|
|
|
|
|
|
#ifdef BFD64
|
|
|
|
|
|
|
|
|
|
#if defined (__GNUC__) && !defined (HOST_64_BIT)
|
|
|
|
|
#define HOST_64_BIT long long
|
|
|
|
|
typedef HOST_64_BIT int64_type;
|
|
|
|
|
typedef unsigned HOST_64_BIT uint64_type;
|
|
|
|
|
#endif
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
1993-08-06 02:27:16 +08:00
|
|
|
|
#if !defined (uint64_type) && defined (__GNUC__)
|
|
|
|
|
#define uint64_type unsigned long long
|
|
|
|
|
#define int64_type long long
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
#define uint64_typeLOW(x) ((unsigned long)(((x) & 0xffffffff)))
|
|
|
|
|
#define uint64_typeHIGH(x) ((unsigned long)(((x) >> 32) & 0xffffffff))
|
1993-08-06 02:27:16 +08:00
|
|
|
|
#endif
|
|
|
|
|
|
1993-04-17 08:39:36 +08:00
|
|
|
|
typedef unsigned HOST_64_BIT bfd_vma;
|
|
|
|
|
typedef HOST_64_BIT bfd_signed_vma;
|
|
|
|
|
typedef unsigned HOST_64_BIT bfd_size_type;
|
|
|
|
|
typedef unsigned HOST_64_BIT symvalue;
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define fprintf_vma(s,x) \
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
fprintf(s,"%08lx%08lx", uint64_typeHIGH(x), uint64_typeLOW(x))
|
1993-08-06 02:27:16 +08:00
|
|
|
|
#define sprintf_vma(s,x) \
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
sprintf(s,"%08lx%08lx", uint64_typeHIGH(x), uint64_typeLOW(x))
|
1993-06-16 10:08:29 +08:00
|
|
|
|
#else /* not BFD64 */
|
1993-04-17 08:39:36 +08:00
|
|
|
|
|
|
|
|
|
/* Represent a target address. Also used as a generic unsigned type
|
|
|
|
|
which is guaranteed to be big enough to hold any arithmetic types
|
|
|
|
|
we need to deal with. */
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
typedef unsigned long bfd_vma;
|
1993-04-17 08:39:36 +08:00
|
|
|
|
|
|
|
|
|
/* A generic signed type which is guaranteed to be big enough to hold any
|
|
|
|
|
arithmetic types we need to deal with. Can be assumed to be compatible
|
|
|
|
|
with bfd_vma in the same way that signed and unsigned ints are compatible
|
|
|
|
|
(as parameters, in assignment, etc). */
|
|
|
|
|
typedef long bfd_signed_vma;
|
|
|
|
|
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
typedef unsigned long symvalue;
|
|
|
|
|
typedef unsigned long bfd_size_type;
|
1993-04-17 08:39:36 +08:00
|
|
|
|
|
|
|
|
|
/* Print a bfd_vma x on stream s. */
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define fprintf_vma(s,x) fprintf(s, "%08lx", x)
|
1993-08-06 02:27:16 +08:00
|
|
|
|
#define sprintf_vma(s,x) sprintf(s, "%08lx", x)
|
1993-06-16 10:08:29 +08:00
|
|
|
|
#endif /* not BFD64 */
|
1992-01-25 06:44:51 +08:00
|
|
|
|
#define printf_vma(x) fprintf_vma(stdout,x)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
|
|
|
|
typedef unsigned int flagword; /* 32 bits of flags */
|
|
|
|
|
|
|
|
|
|
/** File formats */
|
|
|
|
|
|
|
|
|
|
typedef enum bfd_format {
|
|
|
|
|
bfd_unknown = 0, /* file format is unknown */
|
|
|
|
|
bfd_object, /* linker/assember/compiler output */
|
|
|
|
|
bfd_archive, /* object archive file */
|
|
|
|
|
bfd_core, /* core dump */
|
|
|
|
|
bfd_type_end} /* marks the end; don't use it! */
|
|
|
|
|
bfd_format;
|
|
|
|
|
|
1993-11-24 15:21:28 +08:00
|
|
|
|
/* Values that may appear in the flags field of a BFD. These also
|
|
|
|
|
appear in the object_flags field of the bfd_target structure, where
|
|
|
|
|
they indicate the set of flags used by that backend (not all flags
|
|
|
|
|
are meaningful for all object file formats) (FIXME: at the moment,
|
|
|
|
|
the object_flags values have mostly just been copied from backend
|
|
|
|
|
to another, and are not necessarily correct). */
|
|
|
|
|
|
|
|
|
|
/* No flags. */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define NO_FLAGS 0x00
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD contains relocation entries. */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define HAS_RELOC 0x01
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD is directly executable. */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define EXEC_P 0x02
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD has line number information (basically used for F_LNNO in a
|
|
|
|
|
COFF header). */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define HAS_LINENO 0x04
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD has debugging information. */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define HAS_DEBUG 0x08
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD has symbols. */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define HAS_SYMS 0x10
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD has local symbols (basically used for F_LSYMS in a COFF
|
|
|
|
|
header). */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define HAS_LOCALS 0x20
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD is a dynamic object. */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define DYNAMIC 0x40
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* Text section is write protected (if D_PAGED is not set, this is
|
|
|
|
|
like an a.out NMAGIC file) (the linker sets this by default, but
|
|
|
|
|
clears it for -r or -N). */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define WP_TEXT 0x80
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD is dynamically paged (this is like an a.out ZMAGIC file) (the
|
|
|
|
|
linker sets this by default, but clears it for -r or -n or -N). */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define D_PAGED 0x100
|
1993-11-24 15:21:28 +08:00
|
|
|
|
|
|
|
|
|
/* BFD is relaxable (this means that bfd_relax_section may be able to
|
|
|
|
|
do something). */
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define BFD_IS_RELAXABLE 0x200
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
|
|
|
|
/* symbols and relocation */
|
|
|
|
|
|
|
|
|
|
typedef unsigned long symindex;
|
|
|
|
|
|
|
|
|
|
#define BFD_NO_MORE_SYMBOLS ((symindex) ~0)
|
|
|
|
|
|
|
|
|
|
typedef enum bfd_symclass {
|
|
|
|
|
bfd_symclass_unknown = 0,
|
|
|
|
|
bfd_symclass_fcommon, /* fortran common symbols */
|
|
|
|
|
bfd_symclass_global, /* global symbol, what a surprise */
|
|
|
|
|
bfd_symclass_debugger, /* some debugger symbol */
|
|
|
|
|
bfd_symclass_undefined /* none known */
|
|
|
|
|
} symclass;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/* general purpose part of a symbol;
|
|
|
|
|
target specific parts will be found in libcoff.h, liba.out.h etc */
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#define bfd_get_section(x) ((x)->section)
|
|
|
|
|
#define bfd_get_output_section(x) ((x)->section->output_section)
|
|
|
|
|
#define bfd_set_section(x,y) ((x)->section) = (y)
|
1993-01-12 08:38:59 +08:00
|
|
|
|
#define bfd_asymbol_base(x) ((x)->section->vma)
|
1993-02-16 16:49:29 +08:00
|
|
|
|
#define bfd_asymbol_value(x) (bfd_asymbol_base(x) + (x)->value)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define bfd_asymbol_name(x) ((x)->name)
|
1993-01-12 08:38:59 +08:00
|
|
|
|
/*Perhaps future: #define bfd_asymbol_bfd(x) ((x)->section->owner)*/
|
|
|
|
|
#define bfd_asymbol_bfd(x) ((x)->the_bfd)
|
|
|
|
|
#define bfd_asymbol_flavour(x) (bfd_asymbol_bfd(x)->xvec->flavour)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
|
|
|
|
/* This is a type pun with struct ranlib on purpose! */
|
|
|
|
|
typedef struct carsym {
|
|
|
|
|
char *name;
|
|
|
|
|
file_ptr file_offset; /* look here to find the file */
|
|
|
|
|
} carsym; /* to make these you call a carsymogen */
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/* Used in generating armaps. Perhaps just a forward definition would do? */
|
|
|
|
|
struct orl { /* output ranlib */
|
|
|
|
|
char **name; /* symbol name */
|
|
|
|
|
file_ptr pos; /* bfd* or file position */
|
|
|
|
|
int namidx; /* index into string table */
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/* Linenumber stuff */
|
|
|
|
|
typedef struct lineno_cache_entry {
|
|
|
|
|
unsigned int line_number; /* Linenumber from start of function*/
|
|
|
|
|
union {
|
|
|
|
|
struct symbol_cache_entry *sym; /* Function name */
|
|
|
|
|
unsigned long offset; /* Offset into section */
|
|
|
|
|
} u;
|
|
|
|
|
} alent;
|
|
|
|
|
|
|
|
|
|
/* object and core file sections */
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#define align_power(addr, align) \
|
|
|
|
|
( ((addr) + ((1<<(align))-1)) & (-1 << (align)))
|
|
|
|
|
|
|
|
|
|
typedef struct sec *sec_ptr;
|
|
|
|
|
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define bfd_get_section_name(bfd, ptr) ((ptr)->name + 0)
|
|
|
|
|
#define bfd_get_section_vma(bfd, ptr) ((ptr)->vma + 0)
|
|
|
|
|
#define bfd_get_section_alignment(bfd, ptr) ((ptr)->alignment_power + 0)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define bfd_section_name(bfd, ptr) ((ptr)->name)
|
1992-01-25 06:44:51 +08:00
|
|
|
|
#define bfd_section_size(bfd, ptr) (bfd_get_section_size_before_reloc(ptr))
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define bfd_section_vma(bfd, ptr) ((ptr)->vma)
|
|
|
|
|
#define bfd_section_alignment(bfd, ptr) ((ptr)->alignment_power)
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define bfd_get_section_flags(bfd, ptr) ((ptr)->flags + 0)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define bfd_get_section_userdata(bfd, ptr) ((ptr)->userdata)
|
|
|
|
|
|
1993-02-16 16:49:29 +08:00
|
|
|
|
#define bfd_is_com_section(ptr) (((ptr)->flags & SEC_IS_COMMON) != 0)
|
|
|
|
|
|
1993-04-15 12:53:14 +08:00
|
|
|
|
#define bfd_set_section_vma(bfd, ptr, val) (((ptr)->vma = (ptr)->lma= (val)), ((ptr)->user_set_vma = true), true)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define bfd_set_section_alignment(bfd, ptr, val) (((ptr)->alignment_power = (val)),true)
|
|
|
|
|
#define bfd_set_section_userdata(bfd, ptr, val) (((ptr)->userdata = (val)),true)
|
|
|
|
|
|
|
|
|
|
typedef struct stat stat_type;
|
|
|
|
|
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
/* Error handling */
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
typedef enum bfd_error
|
|
|
|
|
{
|
|
|
|
|
no_error = 0,
|
|
|
|
|
system_call_error,
|
|
|
|
|
invalid_target,
|
|
|
|
|
wrong_format,
|
|
|
|
|
invalid_operation,
|
|
|
|
|
no_memory,
|
|
|
|
|
no_symbols,
|
|
|
|
|
no_relocation_info,
|
|
|
|
|
no_more_archived_files,
|
|
|
|
|
malformed_archive,
|
|
|
|
|
symbol_not_found,
|
|
|
|
|
file_not_recognized,
|
|
|
|
|
file_ambiguously_recognized,
|
|
|
|
|
no_contents,
|
|
|
|
|
bfd_error_nonrepresentable_section,
|
|
|
|
|
no_debug_section,
|
|
|
|
|
bad_value,
|
|
|
|
|
file_truncated,
|
|
|
|
|
invalid_error_code
|
|
|
|
|
} bfd_ec;
|
1992-01-25 06:44:51 +08:00
|
|
|
|
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
extern bfd_ec bfd_error;
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
1992-11-28 20:14:14 +08:00
|
|
|
|
CONST char *bfd_errmsg PARAMS ((bfd_ec error_tag));
|
|
|
|
|
void bfd_perror PARAMS ((CONST char *message));
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
typedef enum bfd_print_symbol
|
|
|
|
|
{
|
1991-10-25 15:06:02 +08:00
|
|
|
|
bfd_print_symbol_name,
|
|
|
|
|
bfd_print_symbol_more,
|
1993-06-16 10:08:29 +08:00
|
|
|
|
bfd_print_symbol_all
|
1991-10-25 15:06:02 +08:00
|
|
|
|
} bfd_print_symbol_type;
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
|
|
|
|
|
1993-06-16 10:08:29 +08:00
|
|
|
|
/* Information about a symbol that nm needs. */
|
|
|
|
|
|
|
|
|
|
typedef struct _symbol_info
|
|
|
|
|
{
|
|
|
|
|
symvalue value;
|
|
|
|
|
char type; /* */
|
|
|
|
|
CONST char *name; /* Symbol name. */
|
|
|
|
|
char stab_other; /* Unused. */
|
|
|
|
|
short stab_desc; /* Info for N_TYPE. */
|
|
|
|
|
CONST char *stab_name;
|
|
|
|
|
} symbol_info;
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
/* Hash table routines. There is no way to free up a hash table. */
|
|
|
|
|
|
|
|
|
|
/* An element in the hash table. Most uses will actually use a larger
|
|
|
|
|
structure, and an instance of this will be the first field. */
|
|
|
|
|
|
|
|
|
|
struct bfd_hash_entry
|
|
|
|
|
{
|
|
|
|
|
/* Next entry for this hash code. */
|
|
|
|
|
struct bfd_hash_entry *next;
|
|
|
|
|
/* String being hashed. */
|
|
|
|
|
const char *string;
|
|
|
|
|
/* Hash code. This is the full hash code, not the index into the
|
|
|
|
|
table. */
|
|
|
|
|
unsigned long hash;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/* A hash table. */
|
|
|
|
|
|
|
|
|
|
struct bfd_hash_table
|
|
|
|
|
{
|
|
|
|
|
/* The hash array. */
|
|
|
|
|
struct bfd_hash_entry **table;
|
|
|
|
|
/* The number of slots in the hash table. */
|
|
|
|
|
unsigned int size;
|
|
|
|
|
/* A function used to create new elements in the hash table. The
|
|
|
|
|
first entry is itself a pointer to an element. When this
|
|
|
|
|
function is first invoked, this pointer will be NULL. However,
|
|
|
|
|
having the pointer permits a hierarchy of method functions to be
|
|
|
|
|
built each of which calls the function in the superclass. Thus
|
|
|
|
|
each function should be written to allocate a new block of memory
|
|
|
|
|
only if the argument is NULL. */
|
|
|
|
|
struct bfd_hash_entry *(*newfunc) PARAMS ((struct bfd_hash_entry *,
|
|
|
|
|
struct bfd_hash_table *,
|
|
|
|
|
const char *));
|
|
|
|
|
/* An obstack for this hash table. */
|
|
|
|
|
struct obstack memory;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/* Initialize a hash table. */
|
|
|
|
|
extern boolean bfd_hash_table_init
|
|
|
|
|
PARAMS ((struct bfd_hash_table *,
|
|
|
|
|
struct bfd_hash_entry *(*) (struct bfd_hash_entry *,
|
|
|
|
|
struct bfd_hash_table *,
|
|
|
|
|
const char *)));
|
|
|
|
|
|
|
|
|
|
/* Initialize a hash table specifying a size. */
|
|
|
|
|
extern boolean bfd_hash_table_init_n
|
|
|
|
|
PARAMS ((struct bfd_hash_table *,
|
|
|
|
|
struct bfd_hash_entry *(*) (struct bfd_hash_entry *,
|
|
|
|
|
struct bfd_hash_table *,
|
|
|
|
|
const char *),
|
|
|
|
|
unsigned int size));
|
|
|
|
|
|
|
|
|
|
/* Free up a hash table. */
|
|
|
|
|
extern void bfd_hash_table_free PARAMS ((struct bfd_hash_table *));
|
|
|
|
|
|
|
|
|
|
/* Look up a string in a hash table. If CREATE is true, a new entry
|
|
|
|
|
will be created for this string if one does not already exist. The
|
|
|
|
|
COPY argument must be true if this routine should copy the string
|
|
|
|
|
into newly allocated memory when adding an entry. */
|
|
|
|
|
extern struct bfd_hash_entry *bfd_hash_lookup
|
|
|
|
|
PARAMS ((struct bfd_hash_table *, const char *, boolean create,
|
|
|
|
|
boolean copy));
|
|
|
|
|
|
|
|
|
|
/* Base method for creating a hash table entry. */
|
|
|
|
|
extern struct bfd_hash_entry *bfd_hash_newfunc
|
|
|
|
|
PARAMS ((struct bfd_hash_entry *, struct bfd_hash_table *,
|
|
|
|
|
const char *));
|
|
|
|
|
|
|
|
|
|
/* Grab some space for a hash table entry. */
|
1994-01-07 04:04:43 +08:00
|
|
|
|
extern PTR bfd_hash_allocate PARAMS ((struct bfd_hash_table *,
|
|
|
|
|
unsigned int));
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
|
|
|
|
|
/* Traverse a hash table in a random order, calling a function on each
|
|
|
|
|
element. If the function returns false, the traversal stops. The
|
|
|
|
|
INFO argument is passed to the function. */
|
|
|
|
|
extern void bfd_hash_traverse PARAMS ((struct bfd_hash_table *,
|
|
|
|
|
boolean (*) (struct bfd_hash_entry *,
|
|
|
|
|
PTR),
|
|
|
|
|
PTR info));
|
|
|
|
|
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
/* The code that implements targets can initialize a jump table with this
|
|
|
|
|
macro. It must name all its routines the same way (a prefix plus
|
|
|
|
|
the standard routine suffix), or it must #define the routines that
|
|
|
|
|
are not so named, before calling JUMP_TABLE in the initializer. */
|
|
|
|
|
|
1993-08-06 02:27:16 +08:00
|
|
|
|
/* Semi-portable string concatenation in cpp.
|
|
|
|
|
The CAT4 hack is to avoid a problem with some strict ANSI C preprocessors.
|
|
|
|
|
The problem is, "32_" is not a valid preprocessing token, and we don't
|
|
|
|
|
want extra underscores (e.g., "nlm_32_"). The XCAT2 macro will cause the
|
|
|
|
|
inner CAT macros to be evaluated first, producing still-valid pp-tokens.
|
|
|
|
|
Then the final concatenation can be done. (Sigh.) */
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#ifndef CAT
|
1993-08-06 02:27:16 +08:00
|
|
|
|
#ifdef SABER
|
|
|
|
|
#define CAT(a,b) a##b
|
|
|
|
|
#define CAT3(a,b,c) a##b##c
|
|
|
|
|
#define CAT4(a,b,c,d) a##b##c##d
|
|
|
|
|
#else
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#ifdef __STDC__
|
|
|
|
|
#define CAT(a,b) a##b
|
1993-06-16 10:08:29 +08:00
|
|
|
|
#define CAT3(a,b,c) a##b##c
|
1993-08-06 02:27:16 +08:00
|
|
|
|
#define XCAT2(a,b) CAT(a,b)
|
|
|
|
|
#define CAT4(a,b,c,d) XCAT2(CAT(a,b),CAT(c,d))
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#else
|
|
|
|
|
#define CAT(a,b) a/**/b
|
1993-06-16 10:08:29 +08:00
|
|
|
|
#define CAT3(a,b,c) a/**/b/**/c
|
1993-08-06 02:27:16 +08:00
|
|
|
|
#define CAT4(a,b,c,d) a/**/b/**/c/**/d
|
|
|
|
|
#endif
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#endif
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#define JUMP_TABLE(NAME)\
|
|
|
|
|
CAT(NAME,_core_file_failing_command),\
|
|
|
|
|
CAT(NAME,_core_file_failing_signal),\
|
|
|
|
|
CAT(NAME,_core_file_matches_executable_p),\
|
|
|
|
|
CAT(NAME,_slurp_armap),\
|
|
|
|
|
CAT(NAME,_slurp_extended_name_table),\
|
|
|
|
|
CAT(NAME,_truncate_arname),\
|
|
|
|
|
CAT(NAME,_write_armap),\
|
1993-08-06 02:27:16 +08:00
|
|
|
|
CAT(NAME,_close_and_cleanup),\
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
CAT(NAME,_set_section_contents),\
|
|
|
|
|
CAT(NAME,_get_section_contents),\
|
|
|
|
|
CAT(NAME,_new_section_hook),\
|
|
|
|
|
CAT(NAME,_get_symtab_upper_bound),\
|
|
|
|
|
CAT(NAME,_get_symtab),\
|
|
|
|
|
CAT(NAME,_get_reloc_upper_bound),\
|
|
|
|
|
CAT(NAME,_canonicalize_reloc),\
|
|
|
|
|
CAT(NAME,_make_empty_symbol),\
|
|
|
|
|
CAT(NAME,_print_symbol),\
|
1993-06-16 10:08:29 +08:00
|
|
|
|
CAT(NAME,_get_symbol_info),\
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
CAT(NAME,_get_lineno),\
|
|
|
|
|
CAT(NAME,_set_arch_mach),\
|
|
|
|
|
CAT(NAME,_openr_next_archived_file),\
|
|
|
|
|
CAT(NAME,_find_nearest_line),\
|
|
|
|
|
CAT(NAME,_generic_stat_arch_elt),\
|
|
|
|
|
CAT(NAME,_sizeof_headers),\
|
|
|
|
|
CAT(NAME,_bfd_debug_info_start),\
|
|
|
|
|
CAT(NAME,_bfd_debug_info_end),\
|
1992-01-25 06:44:51 +08:00
|
|
|
|
CAT(NAME,_bfd_debug_info_accumulate),\
|
1992-05-05 07:42:03 +08:00
|
|
|
|
CAT(NAME,_bfd_get_relocated_section_contents),\
|
1993-01-12 08:38:59 +08:00
|
|
|
|
CAT(NAME,_bfd_relax_section),\
|
1993-04-15 12:53:14 +08:00
|
|
|
|
CAT(NAME,_bfd_reloc_type_lookup),\
|
Extensive changes to move the bulk of the linker into BFD so that
more efficient backend code can be written for specific object
files. Only existing efficient backend is a.out.
* seclet.c, seclet.h: Removed.
* hash.c, linker.c, genlink.h: New files.
* bfd-in.h: Removed bfd_error_vector. Declared hash table
structures and functions.
(JUMP_TABLE): Removed bfd_seclet_link, added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* All backends: Changed accordingly.
* bfd-in2.h: Rebuilt.
* bfd.c (struct _bfd): Added link_next and archive_pass fields.
Removed ld_symbols field.
(bfd_nonrepresentable_section, bfd_undefined_symbol,
bfd_reloc_value_truncated, bfd_reloc_is_dangerous,
bfd_error_vector): Removed.
(bfd_default_error_trap, bfd_error_trap,
bfd_error_nonrepresentabltrap): Removed.
(bfd_get_relocated_section_contents): Pass link_info. Pass
link_order instead of seclet. Pass symbols.
(bfd_relax_section): Pass link_info.
(bfd_seclet_link): Removed.
(bfd_link_hash_table_create, bfd_link_add_symbols,
bfd_final_link): New macros.
* libbfd-in.h: If __GNUC__ is defined and alloca is not, define
alloca as __builtin_alloca. Declare internal linking functions.
* libbfd.h: Rebuilt.
* libbfd.c (bfd_seek): Comment out fseek assertion. It's worked
for months.
* reloc.c (reloc_howto_type): Added error_message argument to
special_function field. Changed all callers and all definitions.
(bfd_get_reloc_size): Make argument a const pointer.
(bfd_perform_relocation): Add error_message argument to hold
string set if return value if bfd_reloc_dangerous. Changed all
callers.
(_bfd_final_link_relocate, _bfd_relocate_contents): New functions.
* section.c (asection): Renamed seclets_head and seclets_tail to
link_order_head and link_order_tail.
* targets.c (bfd_target): Replaced seclet argument with link_info
and link_order and symbols arguments in
bfd_get_relocated_section_contents. Added symbols argument to
bfd_relax_section. Removed bfd_seclet_link. Added
bfd_link_hash_table_create, bfd_link_add_symbols and
bfd_final_link.
* libaout.h (struct aoutdata): Added external_syms,
external_sym_count, external_strings, sym_hashes fields.
(obj_aout_external_syms, obj_aout_external_sym_count,
obj_aout_external_strings, obj_aout_sym_hashes): New accessor
macros.
(WRITE_HEADERS): Only output symbols if outsymbols is not NULL.
* aoutx.h: Wrote new back end linker routines.
(translate_to_native_sym_flags): Return boolean value. Don't use
bfd_error_vector.
(NAME(aout,write_syms)): Return boolean value. Check return value
of translate_to_native_sym_flags and bfd_write.
* aout-target.h (final_link_callback): New function.
(MY_bfd_final_link): New function.
* aout-adobe.c (aout_adobe_write_object_contents): Check return
value of aout_32_write_syms.
* hp300hpux.c (MY(write_object_contents)): Likewise.
* i386lynx.c (WRITE_HEADERS): Likewise.
* libaout.h (WRITE_HEADERS): Likewise.
* bout.c: Changed functions to use link_info->callbacks rather
than bfd_error_vector, and link_orders rather than seclets.
* coff-alpha.c: Likewise.
* coff-h8300.c: Likewise.
* coff-h8500.c: Likewise.
* coff-sh.c: Likewise.
* coff-z8k.c: Likewise.
* elf32-hppa.c: Likewise.
* reloc16.c: Likewise.
* coff-alpha.c (alpha_ecoff_get_relocated_section_contents): Look
up _gp in the hash table rather than in outsymbols.
* coff-a29k.c (a29k_reloc): Pass errors back in new error_message
argument rather than printing them.
* coffcode.h (bfd_coff_reloc16_extra_cases): Take link_info and
link_order arguments rather than seclet. Changed all uses and
definitions.
(bfd_coff_reloc16_estimate): Pass link_info arguments. Changed
all uses and definitions.
* libcoff.h: Rebuilt.
* ecoff.c (ecoff_get_extr): If symbol is defined by linker, but
not by ECOFF, make it scAbs.
(ecoff_bfd_final_link): Renamed from ecoff_bfd_seclet_link and
rewritten.
* elf32-mips.c (mips_elf_final_link): Renamed from
mips_elf_seclet_link and rewritten.
* elf32-hppa.c (elf32_hppa_stub_description): Added link_info
field.
(new_stub, add_stub_by_name, hppa_elf_build_arg_reloc_stub,
hppa_elf_build_long_branch_stub, hppa_look_for_stubs_in_section):
Added link_info arguments. Changed all callers.
* elfcode.h (elf_slurp_symbol_table): Don't quit if outsymbols is
not NULL.
* oasys.c (oasys_write_sections): Return boolean value rather than
using bfd_error_vector.
(oasys_write_object_contents): Check return value of
oasys_write_sections.
* hosts/std-host.h: Don't declare qsort or strtol.
* Makefile.in: Rebuild dependencies.
(BFD_LIBS): Removed seclet.o. Added hash.o and linker.o.
(CFILES): Removed seclet.c. Added hash.c and linker.c.
(HFILES): Removed seclet.h. Added genlink.h.
1993-12-31 03:56:50 +08:00
|
|
|
|
CAT(NAME,_bfd_make_debug_symbol),\
|
|
|
|
|
CAT(NAME,_bfd_link_hash_table_create),\
|
|
|
|
|
CAT(NAME,_bfd_link_add_symbols),\
|
|
|
|
|
CAT(NAME,_bfd_final_link)
|
1991-10-25 15:06:02 +08:00
|
|
|
|
|
1993-01-12 08:38:59 +08:00
|
|
|
|
#define COFF_SWAP_TABLE (PTR) &bfd_coff_std_swap_table
|
1991-10-25 15:06:02 +08:00
|
|
|
|
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
|
|
|
|
/* User program access to BFD facilities */
|
|
|
|
|
|
|
|
|
|
/* Cast from const char * to char * so that caller can assign to
|
|
|
|
|
a char * without a warning. */
|
|
|
|
|
#define bfd_get_filename(abfd) ((char *) (abfd)->filename)
|
1993-11-24 15:21:28 +08:00
|
|
|
|
#define bfd_get_cacheable(abfd) ((abfd)->cacheable)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define bfd_get_format(abfd) ((abfd)->format)
|
|
|
|
|
#define bfd_get_target(abfd) ((abfd)->xvec->name)
|
1993-11-24 15:21:28 +08:00
|
|
|
|
#define bfd_get_flavour(abfd) ((abfd)->xvec->flavour)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define bfd_get_file_flags(abfd) ((abfd)->flags)
|
|
|
|
|
#define bfd_applicable_file_flags(abfd) ((abfd)->xvec->object_flags)
|
|
|
|
|
#define bfd_applicable_section_flags(abfd) ((abfd)->xvec->section_flags)
|
1992-01-25 06:44:51 +08:00
|
|
|
|
#define bfd_my_archive(abfd) ((abfd)->my_archive)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
#define bfd_has_map(abfd) ((abfd)->has_armap)
|
|
|
|
|
|
|
|
|
|
#define bfd_valid_reloc_types(abfd) ((abfd)->xvec->valid_reloc_types)
|
|
|
|
|
#define bfd_usrdata(abfd) ((abfd)->usrdata)
|
|
|
|
|
|
|
|
|
|
#define bfd_get_start_address(abfd) ((abfd)->start_address)
|
|
|
|
|
#define bfd_get_symcount(abfd) ((abfd)->symcount)
|
|
|
|
|
#define bfd_get_outsymbols(abfd) ((abfd)->outsymbols)
|
|
|
|
|
#define bfd_count_sections(abfd) ((abfd)->section_count)
|
|
|
|
|
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
#define bfd_get_symbol_leading_char(abfd) ((abfd)->xvec->symbol_leading_char)
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
1993-11-24 15:21:28 +08:00
|
|
|
|
#define bfd_set_cacheable(abfd,bool) (((abfd)->cacheable = (bool)), true)
|
|
|
|
|
|
|
|
|
|
/* Byte swapping routines. */
|
|
|
|
|
|
* bfd-in.h (bfd_get{b,l}[_signed_]{16,32,64}): Declare argument to
be a const pointer.
* bfd-in2.h: Rebuilt.
* libbfd.c (bfd_get{b,l}[_signed_]{16,32,64}): Declare argument to
be a const pointer.
* targets.c (bfd_target): Change swap function pointers
accordingly.
* archive.c (do_slurp_coff_armap): Change swap accordingly.
* aix386-core.c: Change NO_GET and NO_GETS accordingly.
* hppabsd-core.c, hpux-core.c, irix-core.c, osf-core.c,
ptrace-core.c, trad-core.c: Change NO_GET and NO_SIGNED_GET
accordingly.
1994-01-04 05:28:43 +08:00
|
|
|
|
bfd_vma bfd_getb64 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_vma bfd_getl64 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_signed_vma bfd_getb_signed_64 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_signed_vma bfd_getl_signed_64 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_vma bfd_getb32 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_vma bfd_getl32 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_signed_vma bfd_getb_signed_32 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_signed_vma bfd_getl_signed_32 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_vma bfd_getb16 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_vma bfd_getl16 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_signed_vma bfd_getb_signed_16 PARAMS ((const unsigned char *));
|
|
|
|
|
bfd_signed_vma bfd_getl_signed_16 PARAMS ((const unsigned char *));
|
1993-11-24 15:21:28 +08:00
|
|
|
|
void bfd_putb64 PARAMS ((bfd_vma, unsigned char *));
|
|
|
|
|
void bfd_putl64 PARAMS ((bfd_vma, unsigned char *));
|
|
|
|
|
void bfd_putb32 PARAMS ((bfd_vma, unsigned char *));
|
|
|
|
|
void bfd_putl32 PARAMS ((bfd_vma, unsigned char *));
|
|
|
|
|
void bfd_putb16 PARAMS ((bfd_vma, unsigned char *));
|
|
|
|
|
void bfd_putl16 PARAMS ((bfd_vma, unsigned char *));
|
|
|
|
|
|
|
|
|
|
/* ECOFF linking routines. */
|
|
|
|
|
#ifdef __STDC__
|
|
|
|
|
struct ecoff_debug_info;
|
|
|
|
|
struct ecoff_debug_swap;
|
|
|
|
|
struct ecoff_extr;
|
|
|
|
|
struct symbol_cache_entry;
|
|
|
|
|
#endif
|
|
|
|
|
extern boolean bfd_ecoff_debug_accumulate
|
|
|
|
|
PARAMS ((bfd *output_bfd, struct ecoff_debug_info *output_debug,
|
|
|
|
|
const struct ecoff_debug_swap *output_swap,
|
|
|
|
|
bfd *input_bfd, struct ecoff_debug_info *input_debug,
|
|
|
|
|
const struct ecoff_debug_swap *input_swap,
|
|
|
|
|
boolean relocateable));
|
|
|
|
|
extern boolean bfd_ecoff_debug_link_other
|
|
|
|
|
PARAMS ((bfd *output_bfd, struct ecoff_debug_info *output_debug,
|
|
|
|
|
const struct ecoff_debug_swap *output_swap, bfd *input_bfd));
|
|
|
|
|
extern boolean bfd_ecoff_debug_externals
|
|
|
|
|
PARAMS ((bfd *abfd, struct ecoff_debug_info *debug,
|
|
|
|
|
const struct ecoff_debug_swap *swap,
|
|
|
|
|
boolean relocateable,
|
|
|
|
|
boolean (*get_extr) (struct symbol_cache_entry *,
|
|
|
|
|
struct ecoff_extr *),
|
|
|
|
|
void (*set_index) (struct symbol_cache_entry *,
|
|
|
|
|
bfd_size_type)));
|
1994-01-07 04:04:43 +08:00
|
|
|
|
extern boolean bfd_ecoff_debug_one_external
|
|
|
|
|
PARAMS ((bfd *abfd, struct ecoff_debug_info *debug,
|
|
|
|
|
const struct ecoff_debug_swap *swap,
|
|
|
|
|
const char *name, struct ecoff_extr *esym));
|
1993-11-24 15:21:28 +08:00
|
|
|
|
extern bfd_size_type bfd_ecoff_debug_size
|
|
|
|
|
PARAMS ((bfd *abfd, struct ecoff_debug_info *debug,
|
|
|
|
|
const struct ecoff_debug_swap *swap));
|
|
|
|
|
extern boolean bfd_ecoff_write_debug
|
|
|
|
|
PARAMS ((bfd *abfd, struct ecoff_debug_info *debug,
|
|
|
|
|
const struct ecoff_debug_swap *swap, file_ptr where));
|
Insert the incidentals and things I forgot to ci last time, same log message:
Mon Sep 30 15:13:46 1991 Steve Chamberlain (steve at cygnus.com)
* cpu-a29k.c, cpu-i386.c, cpu-m68k.c, cpu-mips.c, cpu-vax.c,
cpu-h8300.c, cpu-i960.c, cpu-m88k.c, cpu-sparc.c: added. These
files will eventually contain processor specific bits for bfd,
like strange relocation information and dis/assembly. So far only
the H8 has been even partially done. This work also ties in with
the change in handling architectures.
* amdcoff.c: (a29k_reloc) fix error message.
* aout-f1.h: (choose_reloc_size) now calls bfd_get_arch to
discover the architecture of the bfd. (sunos4_callback) calls the
function bfd_set_arch_mach rather than stuffing stuff directly
into the bfd. (sunos4_write_object_contents), changed names of
accessor functions.
* aoutx.h: (set_arch_mach) now calls bfd_default_set_arch_mach to
setup the environment.
* archive.c: (bfd_slurp_coff_armap) coff archives always have the
headers in big endian format, regardless of the endianess of the
host or target.
* archures.c: totally changed. Now an architecture is represented
with a pointer to an info structure rather than an enumerated type
and a long. The old info is available as two elements in the
structure. Future enhancements to architecure support will
involve pointers to methods being placed into the info structure.
* bfd.c: changed the definition of the bfd structure for the new
architecture stuff.
* bout.c: (b_out_set_arch_mach) changed to use the new
architecture mechanism.
* coffcode.h: (coff_set_arch_mach, coff_set_flags) changed to use
the new architecture mechanism.
* configure.in: added h8 stuff.
* ieee.c: too many changes to note. Now ieee files written with
bfd gas and ld can be read by gld and ieee only linkers and
simulators.
* libbfd.c, libbfd.h: changed prototype of bfd_write.
* newsos3.c: (newos3_callback) now calls bfd_set_arch_mach rather
than fixing the structure directly.
* oasys.c: (oasys_object_p) now calls bfd_default_set_arch_mach rather
than fixing the structure directly.
* opncls.c: (new_bfd) makes sure that bfd_init has been called
before opening a bfd.
* srec.c: (srec_set_arch_mach) now calls bfd_default_set_arch_mach
rather than fixing the structure directly.
* targets.c: (target_vector) now by defining SELECT_VECS (perhaps
in the t/hmake file) a user can select which backends they want
linked with bfd without changing the source.
* init.c: new, looks after initializing modules.
* howto.c: for future use, will allow an application to work out
what cookie to use as a handle on a relcoatio howto.
1991-10-01 11:55:40 +08:00
|
|
|
|
|
Some of these are guesses, if you know different, just yell.
Thu Oct 8 08:52:48 1992 Steve Chamberlain (sac@thepub.cygnus.com)
Now a bfd knows whether underscores are normally prepended
to symbols in its file format. Helps with error messages.
* aout-adobe.c, aout-target.h, bout.c, coff-a29k.c, coff-h8300.c,
coff-z8k.c: targets set so they have leading underscore
* coff-i386.c, coff-i960.c, coff-m68k.c, coff-mips.c, coff-m88k.c,
coff-rs6000.c, coff-we32k.c, elf.c, ieee.c, srec.c: targets set
without leading underscore flag
* targets.c: add symbol leading char to xvec description
* bfd-in.h (bfd_get_symbol_leading_char): new macro.
1992-10-09 00:02:09 +08:00
|
|
|
|
/* And more from the source. */
|