old-cross-binutils/bfd/bfd-in.h

517 lines
19 KiB
C
Raw Normal View History

/* Main header file for the bfd library -- portable access to object files.
Copyright 1990, 1991, 1992, 1993, 1994 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 03:55:40 +00:00
Contributed by Cygnus Support.
** 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 03:55:40 +00:00
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
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 03:55:40 +00:00
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"
/* These two lines get substitutions done by commands in Makefile.in. */
#define BFD_VERSION "@VERSION@"
#define BFD_ARCH_SIZE @WORDSIZE@
#if BFD_ARCH_SIZE >= 64
#define BFD64
#endif
#ifndef INLINE
#if __GNUC__ >= 2
#define INLINE __inline__
#else
#define INLINE
#endif
#endif
/* 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 03:55:40 +00:00
/* forward declaration */
typedef struct _bfd bfd;
/* 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 03:55:40 +00:00
/* 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. */
/* typedef enum boolean {false, true} boolean; */
/* Yup, SVR4 has a "typedef enum boolean" in <sys/types.h> -fnf */
/* It gets worse if the host also defines a true/false enum... -sts */
#ifndef TRUE_FALSE_ALREADY_DEFINED
typedef enum bfd_boolean {false, true} boolean;
#define BFD_TRUE_FALSE
#else
typedef enum bfd_boolean {bfd_false, bfd_true} boolean;
#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 03:55:40 +00:00
/* 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 03:55:40 +00:00
/* Support for different sizes of target format ints and addresses. If the
host implements 64-bit values, it defines BFD_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. BFD_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 (BFD_HOST_64_BIT)
#define BFD_HOST_64_BIT long long
typedef BFD_HOST_64_BIT int64_type;
typedef unsigned BFD_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 03:55:40 +00:00
#if !defined (uint64_type) && defined (__GNUC__)
#define uint64_type unsigned long long
#define int64_type long long
#endif
#ifndef uint64_typeLOW
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.
1993-12-30 19:56:50 +00:00
#define uint64_typeLOW(x) ((unsigned long)(((x) & 0xffffffff)))
#define uint64_typeHIGH(x) ((unsigned long)(((x) >> 32) & 0xffffffff))
#endif
typedef unsigned BFD_HOST_64_BIT bfd_vma;
typedef BFD_HOST_64_BIT bfd_signed_vma;
typedef unsigned BFD_HOST_64_BIT bfd_size_type;
typedef unsigned BFD_HOST_64_BIT symvalue;
#ifndef fprintf_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 03:55:40 +00:00
#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.
1993-12-30 19:56:50 +00:00
fprintf(s,"%08lx%08lx", uint64_typeHIGH(x), uint64_typeLOW(x))
#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.
1993-12-30 19:56:50 +00:00
sprintf(s,"%08lx%08lx", uint64_typeHIGH(x), uint64_typeLOW(x))
#endif
#else /* not BFD64 */
/* 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 03:55:40 +00:00
typedef unsigned long bfd_vma;
/* 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 03:55:40 +00:00
typedef unsigned long symvalue;
typedef unsigned long bfd_size_type;
/* 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 03:55:40 +00:00
#define fprintf_vma(s,x) fprintf(s, "%08lx", x)
#define sprintf_vma(s,x) sprintf(s, "%08lx", x)
#endif /* not BFD64 */
#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 03:55:40 +00:00
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;
/* 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. */
#define NO_FLAGS 0x00
/* BFD contains relocation entries. */
#define HAS_RELOC 0x01
/* BFD is directly executable. */
#define EXEC_P 0x02
/* BFD has line number information (basically used for F_LNNO in a
COFF header). */
#define HAS_LINENO 0x04
/* BFD has debugging information. */
#define HAS_DEBUG 0x08
/* BFD has symbols. */
#define HAS_SYMS 0x10
/* BFD has local symbols (basically used for F_LSYMS in a COFF
header). */
#define HAS_LOCALS 0x20
/* BFD is a dynamic object. */
#define DYNAMIC 0x40
/* 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). */
#define WP_TEXT 0x80
/* 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). */
#define D_PAGED 0x100
/* BFD is relaxable (this means that bfd_relax_section may be able to
do something). */
#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 03:55:40 +00:00
/* symbols and relocation */
/* A count of carsyms (canonical archive symbols). */
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 03:55:40 +00:00
typedef unsigned long symindex;
#define BFD_NO_MORE_SYMBOLS ((symindex) ~0)
/* General purpose part of a symbol X;
target specific parts are in libcoff.h, libaout.h, etc. */
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 03:55:40 +00:00
#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)
#define bfd_asymbol_base(x) ((x)->section->vma)
1993-02-16 08:49:29 +00:00
#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 03:55:40 +00:00
#define bfd_asymbol_name(x) ((x)->name)
/*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 03:55:40 +00:00
/* A canonical archive symbol. */
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 03:55:40 +00:00
/* 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 (archive tables of contents).
Perhaps just a forward definition would do? */
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 03:55:40 +00:00
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 */
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 03:55:40 +00:00
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;
#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 03:55:40 +00:00
#define bfd_section_name(bfd, ptr) ((ptr)->name)
#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 03:55:40 +00:00
#define bfd_section_vma(bfd, ptr) ((ptr)->vma)
#define bfd_section_alignment(bfd, ptr) ((ptr)->alignment_power)
#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 03:55:40 +00:00
#define bfd_get_section_userdata(bfd, ptr) ((ptr)->userdata)
1993-02-16 08:49:29 +00:00
#define bfd_is_com_section(ptr) (((ptr)->flags & SEC_IS_COMMON) != 0)
#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 03:55:40 +00:00
#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;
typedef enum bfd_print_symbol
{
bfd_print_symbol_name,
bfd_print_symbol_more,
bfd_print_symbol_all
} 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 03:55:40 +00:00
/* 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 03:55:40 +00:00
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.
1993-12-30 19:56:50 +00:00
/* 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. */
* ecoff.c: First cut at new style of linker backend for ECOFF--added a bunch of functions. Also: (ecoff_sec_to_styp_flags): Set flags for .pdata and .xdata. (ecoff_slurp_symbolic_header): New function. (ecoff_slurp_symbolic_info): Call ecoff_slurp_symbolic_header. (ecoff_compute_reloc_file_positions): New function. (ecoff_set_section_contents): Get out quickly if count is zero. Check errors better. (ecoff_write_object_contents): Put .xdata section in data segment. Call ecoff_compute_reloc_file_positions. Don't output relocs or external symbols if outsymbols is NULL. (ecoff_bfd_final_link): Completely rewritten. * libecoff.h: Include bfdlink.h. (struct ecoff_backend_data): Add relocate_section field. (ecoff_data_type): Add sym_hashes and symndx_to_section fields. (struct ecoff_link_hash_entry): Define. (struct ecoff_link_hash_table): Define. (ecoff_bfd_link_add_symbols): Declare as function, not macro. (ecoff_bfd_link_hash_table_create): Likewise. * ecofflink.c (bfd_ecoff_debug_one_external): New function. (bfd_ecoff_debug_externals): Call bfd_ecoff_debug_one_external. * bfd-in.h (bfd_ecoff_debug_one_external): Declare. * bfd-in2.h: Rebuilt. * coff-alpha.c (alpha_howto_table): Mark BRADDR as partial_inplace, and set the src_mask to 0x1fffff. (alpha_ecoff_get_relocated_section_contents): Remove unused variable gp_warned. (alpha_convert_external_reloc): New static function. (alpha_relocate_section): New static function. (alpha_ecoff_backend_data): Initialize relocate_section field. * coff-mips.c (mips_relocate_refhi): New static function. (mips_relocate_section): New static function. (mips_ecoff_backend_data): Initialize relocate_section field.
1994-01-06 20:04:43 +00:00
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.
1993-12-30 19:56:50 +00:00
/* 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));
/* 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 03:55:40 +00:00
#ifndef CAT
#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
#if defined(__STDC__) || defined(ALMOST_STDC)
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 03:55:40 +00:00
#define CAT(a,b) a##b
#define CAT3(a,b,c) a##b##c
#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 03:55:40 +00:00
#else
#define CAT(a,b) a/**/b
#define CAT3(a,b,c) a/**/b/**/c
#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 03:55:40 +00:00
#endif
#endif
#define COFF_SWAP_TABLE (PTR) &bfd_coff_std_swap_table
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 03:55:40 +00:00
/* 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)
#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 03:55:40 +00:00
#define bfd_get_format(abfd) ((abfd)->format)
#define bfd_get_target(abfd) ((abfd)->xvec->name)
#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 03:55:40 +00:00
#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)
#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 03:55:40 +00:00
#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)
#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 03:55:40 +00:00
#define bfd_set_cacheable(abfd,bool) (((abfd)->cacheable = (bool)), true)
/* Byte swapping routines. */
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 *));
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. */
#if defined(__STDC__) || defined(ALMOST_STDC)
struct ecoff_debug_info;
struct ecoff_debug_swap;
struct ecoff_extr;
struct symbol_cache_entry;
struct bfd_link_info;
#endif
extern PTR bfd_ecoff_debug_init
PARAMS ((bfd *output_bfd, struct ecoff_debug_info *output_debug,
const struct ecoff_debug_swap *output_swap,
struct bfd_link_info *));
extern void bfd_ecoff_debug_free
PARAMS ((PTR handle, bfd *output_bfd, struct ecoff_debug_info *output_debug,
const struct ecoff_debug_swap *output_swap,
struct bfd_link_info *));
extern boolean bfd_ecoff_debug_accumulate
PARAMS ((PTR handle, 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,
struct bfd_link_info *));
extern boolean bfd_ecoff_debug_accumulate_other
PARAMS ((PTR handle, bfd *output_bfd, struct ecoff_debug_info *output_debug,
const struct ecoff_debug_swap *output_swap, bfd *input_bfd,
struct bfd_link_info *));
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)));
* ecoff.c: First cut at new style of linker backend for ECOFF--added a bunch of functions. Also: (ecoff_sec_to_styp_flags): Set flags for .pdata and .xdata. (ecoff_slurp_symbolic_header): New function. (ecoff_slurp_symbolic_info): Call ecoff_slurp_symbolic_header. (ecoff_compute_reloc_file_positions): New function. (ecoff_set_section_contents): Get out quickly if count is zero. Check errors better. (ecoff_write_object_contents): Put .xdata section in data segment. Call ecoff_compute_reloc_file_positions. Don't output relocs or external symbols if outsymbols is NULL. (ecoff_bfd_final_link): Completely rewritten. * libecoff.h: Include bfdlink.h. (struct ecoff_backend_data): Add relocate_section field. (ecoff_data_type): Add sym_hashes and symndx_to_section fields. (struct ecoff_link_hash_entry): Define. (struct ecoff_link_hash_table): Define. (ecoff_bfd_link_add_symbols): Declare as function, not macro. (ecoff_bfd_link_hash_table_create): Likewise. * ecofflink.c (bfd_ecoff_debug_one_external): New function. (bfd_ecoff_debug_externals): Call bfd_ecoff_debug_one_external. * bfd-in.h (bfd_ecoff_debug_one_external): Declare. * bfd-in2.h: Rebuilt. * coff-alpha.c (alpha_howto_table): Mark BRADDR as partial_inplace, and set the src_mask to 0x1fffff. (alpha_ecoff_get_relocated_section_contents): Remove unused variable gp_warned. (alpha_convert_external_reloc): New static function. (alpha_relocate_section): New static function. (alpha_ecoff_backend_data): Initialize relocate_section field. * coff-mips.c (mips_relocate_refhi): New static function. (mips_relocate_section): New static function. (mips_ecoff_backend_data): Initialize relocate_section field.
1994-01-06 20:04:43 +00:00
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));
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));
extern boolean bfd_ecoff_write_accumulated_debug
PARAMS ((PTR handle, bfd *abfd, struct ecoff_debug_info *debug,
const struct ecoff_debug_swap *swap,
struct bfd_link_info *info, 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 03:55:40 +00:00
/* And more from the source. */