Next: , Previous: Overview, Up: Top

2 BFD Front End

2.1 typedef bfd

A BFD has type bfd; objects of this type are the cornerstone of any application using BFD. Using BFD consists of making references though the BFD and to data in the BFD.

Here is the structure that defines the type bfd. It contains the major data about the file and pointers to the rest of the data.

     struct bfd
       /* A unique identifier of the BFD  */
       unsigned int id;
       /* The filename the application opened the BFD with.  */
       const char *filename;
       /* A pointer to the target jump table.  */
       const struct bfd_target *xvec;
       /* The IOSTREAM, and corresponding IO vector that provide access
          to the file backing the BFD.  */
       void *iostream;
       const struct bfd_iovec *iovec;
       /* Is the file descriptor being cached?  That is, can it be closed as
          needed, and re-opened when accessed later?  */
       bfd_boolean cacheable;
       /* Marks whether there was a default target specified when the
          BFD was opened. This is used to select which matching algorithm
          to use to choose the back end.  */
       bfd_boolean target_defaulted;
       /* The caching routines use these to maintain a
          least-recently-used list of BFDs.  */
       struct bfd *lru_prev, *lru_next;
       /* When a file is closed by the caching routines, BFD retains
          state information on the file here...  */
       ufile_ptr where;
       /* ... and here: (``once'' means at least once).  */
       bfd_boolean opened_once;
       /* Set if we have a locally maintained mtime value, rather than
          getting it from the file each time.  */
       bfd_boolean mtime_set;
       /* File modified time, if mtime_set is TRUE.  */
       long mtime;
       /* Reserved for an unimplemented file locking extension.  */
       int ifd;
       /* The format which belongs to the BFD. (object, core, etc.)  */
       bfd_format format;
       /* The direction with which the BFD was opened.  */
       enum bfd_direction
           no_direction = 0,
           read_direction = 1,
           write_direction = 2,
           both_direction = 3
       /* Format_specific flags.  */
       flagword flags;
       /* Currently my_archive is tested before adding origin to
          anything. I believe that this can become always an add of
          origin, with origin set to 0 for non archive files.  */
       ufile_ptr origin;
       /* Remember when output has begun, to stop strange things
          from happening.  */
       bfd_boolean output_has_begun;
       /* A hash table for section names.  */
       struct bfd_hash_table section_htab;
       /* Pointer to linked list of sections.  */
       struct bfd_section *sections;
       /* The place where we add to the section list.  */
       struct bfd_section **section_tail;
       /* The number of sections.  */
       unsigned int section_count;
       /* Stuff only useful for object files:
          The start address.  */
       bfd_vma start_address;
       /* Used for input and output.  */
       unsigned int symcount;
       /* Symbol table for output BFD (with symcount entries).  */
       struct bfd_symbol  **outsymbols;
       /* Used for slurped dynamic symbol tables.  */
       unsigned int dynsymcount;
       /* Pointer to structure which contains architecture information.  */
       const struct bfd_arch_info *arch_info;
       /* Flag set if symbols from this BFD should not be exported.  */
       bfd_boolean no_export;
       /* Stuff only useful for archives.  */
       void *arelt_data;
       struct bfd *my_archive;      /* The containing archive BFD.  */
       struct bfd *next;            /* The next BFD in the archive.  */
       struct bfd *archive_head;    /* The first BFD in the archive.  */
       bfd_boolean has_armap;
       /* A chain of BFD structures involved in a link.  */
       struct bfd *link_next;
       /* A field used by _bfd_generic_link_add_archive_symbols.  This will
          be used only for archive elements.  */
       int archive_pass;
       /* Used by the back end to hold private data.  */
           struct aout_data_struct *aout_data;
           struct artdata *aout_ar_data;
           struct _oasys_data *oasys_obj_data;
           struct _oasys_ar_data *oasys_ar_data;
           struct coff_tdata *coff_obj_data;
           struct pe_tdata *pe_obj_data;
           struct xcoff_tdata *xcoff_obj_data;
           struct ecoff_tdata *ecoff_obj_data;
           struct ieee_data_struct *ieee_data;
           struct ieee_ar_data_struct *ieee_ar_data;
           struct srec_data_struct *srec_data;
           struct ihex_data_struct *ihex_data;
           struct tekhex_data_struct *tekhex_data;
           struct elf_obj_tdata *elf_obj_data;
           struct nlm_obj_tdata *nlm_obj_data;
           struct bout_data_struct *bout_data;
           struct mmo_data_struct *mmo_data;
           struct sun_core_struct *sun_core_data;
           struct sco5_core_struct *sco5_core_data;
           struct trad_core_struct *trad_core_data;
           struct som_data_struct *som_data;
           struct hpux_core_struct *hpux_core_data;
           struct hppabsd_core_struct *hppabsd_core_data;
           struct sgi_core_struct *sgi_core_data;
           struct lynx_core_struct *lynx_core_data;
           struct osf_core_struct *osf_core_data;
           struct cisco_core_struct *cisco_core_data;
           struct versados_data_struct *versados_data;
           struct netbsd_core_struct *netbsd_core_data;
           struct mach_o_data_struct *mach_o_data;
           struct mach_o_fat_data_struct *mach_o_fat_data;
           struct bfd_pef_data_struct *pef_data;
           struct bfd_pef_xlib_data_struct *pef_xlib_data;
           struct bfd_sym_data_struct *sym_data;
           void *any;
       /* Used by the application to hold private data.  */
       void *usrdata;
       /* Where all the allocated stuff under this BFD goes.  This is a
          struct objalloc *, but we use void * to avoid requiring the inclusion
          of objalloc.h.  */
       void *memory;

2.2 Error reporting

Most BFD functions return nonzero on success (check their individual documentation for precise semantics). On an error, they call bfd_set_error to set an error condition that callers can check by calling bfd_get_error. If that returns bfd_error_system_call, then check errno.

The easiest way to report a BFD error to the user is to use bfd_perror.

2.2.1 Type bfd_error_type

The values returned by bfd_get_error are defined by the enumerated type bfd_error_type.

     typedef enum bfd_error
       bfd_error_no_error = 0,
     bfd_error_type; bfd_get_error


     bfd_error_type bfd_get_error (void);

Return the current BFD error condition. bfd_set_error


     void bfd_set_error (bfd_error_type error_tag);

Set the BFD error condition to be error_tag. bfd_errmsg


     const char *bfd_errmsg (bfd_error_type error_tag);

Return a string describing the error error_tag, or the system error if error_tag is bfd_error_system_call. bfd_perror


     void bfd_perror (const char *message);

Print to the standard error stream a string describing the last BFD error that occurred, or the last system error if the last BFD error was a system call failure. If message is non-NULL and non-empty, the error string printed is preceded by message, a colon, and a space. It is followed by a newline.

2.2.2 BFD error handler

Some BFD functions want to print messages describing the problem. They call a BFD error handler function. This function may be overridden by the program.

The BFD error handler acts like printf.

     typedef void (*bfd_error_handler_type) (const char *, ...); bfd_set_error_handler


     bfd_error_handler_type bfd_set_error_handler (bfd_error_handler_type);

Set the BFD error handler function. Returns the previous function. bfd_set_error_program_name


     void bfd_set_error_program_name (const char *);

Set the program name to use when printing a BFD error. This is printed before the error message followed by a colon and space. The string must not be changed after it is passed to this function. bfd_get_error_handler


     bfd_error_handler_type bfd_get_error_handler (void);

Return the BFD error handler function.

2.3 Symbols bfd_get_reloc_upper_bound


     long bfd_get_reloc_upper_bound (bfd *abfd, asection *sect);

Return the number of bytes required to store the relocation information associated with section sect attached to bfd abfd. If an error occurs, return -1. bfd_canonicalize_reloc


     long bfd_canonicalize_reloc
        (bfd *abfd, asection *sec, arelent **loc, asymbol **syms);

Call the back end associated with the open BFD abfd and translate the external form of the relocation information attached to sec into the internal canonical form. Place the table into memory at loc, which has been preallocated, usually by a call to bfd_get_reloc_upper_bound. Returns the number of relocs, or -1 on error.

The syms table is also needed for horrible internal magic reasons. bfd_set_reloc


     void bfd_set_reloc
        (bfd *abfd, asection *sec, arelent **rel, unsigned int count);

Set the relocation pointer and count within section sec to the values rel and count. The argument abfd is ignored. bfd_set_file_flags


     bfd_boolean bfd_set_file_flags (bfd *abfd, flagword flags);

Set the flag word in the BFD abfd to the value flags.

Possible errors are: bfd_get_arch_size


     int bfd_get_arch_size (bfd *abfd);

Returns the architecture address size, in bits, as determined by the object file's format. For ELF, this information is included in the header.

Returns the arch size in bits if known, -1 otherwise. bfd_get_sign_extend_vma


     int bfd_get_sign_extend_vma (bfd *abfd);

Indicates if the target architecture "naturally" sign extends an address. Some architectures implicitly sign extend address values when they are converted to types larger than the size of an address. For instance, bfd_get_start_address() will return an address sign extended to fill a bfd_vma when this is the case.

Returns 1 if the target architecture is known to sign extend addresses, 0 if the target architecture is known to not sign extend addresses, and -1 otherwise. bfd_set_start_address


     bfd_boolean bfd_set_start_address (bfd *abfd, bfd_vma vma);

Make vma the entry point of output BFD abfd.

Returns TRUE on success, FALSE otherwise. bfd_get_gp_size


     unsigned int bfd_get_gp_size (bfd *abfd);

Return the maximum size of objects to be optimized using the GP register under MIPS ECOFF. This is typically set by the -G argument to the compiler, assembler or linker. bfd_set_gp_size


     void bfd_set_gp_size (bfd *abfd, unsigned int i);

Set the maximum size of objects to be optimized using the GP register under ECOFF or MIPS ELF. This is typically set by the -G argument to the compiler, assembler or linker. bfd_scan_vma


     bfd_vma bfd_scan_vma (const char *string, const char **end, int base);

Convert, like strtoul, a numerical expression string into a bfd_vma integer, and return that integer. (Though without as many bells and whistles as strtoul.) The expression is assumed to be unsigned (i.e., positive). If given a base, it is used as the base for conversion. A base of 0 causes the function to interpret the string in hex if a leading "0x" or "0X" is found, otherwise in octal if a leading zero is found, otherwise in decimal.

If the value would overflow, the maximum bfd_vma value is returned. bfd_copy_private_header_data


     bfd_boolean bfd_copy_private_header_data (bfd *ibfd, bfd *obfd);

Copy private BFD header information from the BFD ibfd to the the BFD obfd. This copies information that may require sections to exist, but does not require symbol tables. Return true on success, false on error. Possible error returns are:

     #define bfd_copy_private_header_data(ibfd, obfd) \
          BFD_SEND (obfd, _bfd_copy_private_header_data, \
                    (ibfd, obfd)) bfd_copy_private_bfd_data


     bfd_boolean bfd_copy_private_bfd_data (bfd *ibfd, bfd *obfd);

Copy private BFD information from the BFD ibfd to the the BFD obfd. Return TRUE on success, FALSE on error. Possible error returns are:

     #define bfd_copy_private_bfd_data(ibfd, obfd) \
          BFD_SEND (obfd, _bfd_copy_private_bfd_data, \
                    (ibfd, obfd)) bfd_merge_private_bfd_data


     bfd_boolean bfd_merge_private_bfd_data (bfd *ibfd, bfd *obfd);

Merge private BFD information from the BFD ibfd to the the output file BFD obfd when linking. Return TRUE on success, FALSE on error. Possible error returns are:

     #define bfd_merge_private_bfd_data(ibfd, obfd) \
          BFD_SEND (obfd, _bfd_merge_private_bfd_data, \
                    (ibfd, obfd)) bfd_set_private_flags


     bfd_boolean bfd_set_private_flags (bfd *abfd, flagword flags);

Set private BFD flag information in the BFD abfd. Return TRUE on success, FALSE on error. Possible error returns are:

     #define bfd_set_private_flags(abfd, flags) \
          BFD_SEND (abfd, _bfd_set_private_flags, (abfd, flags)) Other functions

The following functions exist but have not yet been documented.

     #define bfd_sizeof_headers(abfd, reloc) \
            BFD_SEND (abfd, _bfd_sizeof_headers, (abfd, reloc))
     #define bfd_find_nearest_line(abfd, sec, syms, off, file, func, line) \
            BFD_SEND (abfd, _bfd_find_nearest_line, \
                      (abfd, sec, syms, off, file, func, line))
     #define bfd_debug_info_start(abfd) \
            BFD_SEND (abfd, _bfd_debug_info_start, (abfd))
     #define bfd_debug_info_end(abfd) \
            BFD_SEND (abfd, _bfd_debug_info_end, (abfd))
     #define bfd_debug_info_accumulate(abfd, section) \
            BFD_SEND (abfd, _bfd_debug_info_accumulate, (abfd, section))
     #define bfd_stat_arch_elt(abfd, stat) \
            BFD_SEND (abfd, _bfd_stat_arch_elt,(abfd, stat))
     #define bfd_update_armap_timestamp(abfd) \
            BFD_SEND (abfd, _bfd_update_armap_timestamp, (abfd))
     #define bfd_set_arch_mach(abfd, arch, mach)\
            BFD_SEND ( abfd, _bfd_set_arch_mach, (abfd, arch, mach))
     #define bfd_relax_section(abfd, section, link_info, again) \
            BFD_SEND (abfd, _bfd_relax_section, (abfd, section, link_info, again))
     #define bfd_gc_sections(abfd, link_info) \
            BFD_SEND (abfd, _bfd_gc_sections, (abfd, link_info))
     #define bfd_merge_sections(abfd, link_info) \
            BFD_SEND (abfd, _bfd_merge_sections, (abfd, link_info))
     #define bfd_is_group_section(abfd, sec) \
            BFD_SEND (abfd, _bfd_is_group_section, (abfd, sec))
     #define bfd_discard_group(abfd, sec) \
            BFD_SEND (abfd, _bfd_discard_group, (abfd, sec))
     #define bfd_link_hash_table_create(abfd) \
            BFD_SEND (abfd, _bfd_link_hash_table_create, (abfd))
     #define bfd_link_hash_table_free(abfd, hash) \
            BFD_SEND (abfd, _bfd_link_hash_table_free, (hash))
     #define bfd_link_add_symbols(abfd, info) \
            BFD_SEND (abfd, _bfd_link_add_symbols, (abfd, info))
     #define bfd_link_just_syms(abfd, sec, info) \
            BFD_SEND (abfd, _bfd_link_just_syms, (sec, info))
     #define bfd_final_link(abfd, info) \
            BFD_SEND (abfd, _bfd_final_link, (abfd, info))
     #define bfd_free_cached_info(abfd) \
            BFD_SEND (abfd, _bfd_free_cached_info, (abfd))
     #define bfd_get_dynamic_symtab_upper_bound(abfd) \
            BFD_SEND (abfd, _bfd_get_dynamic_symtab_upper_bound, (abfd))
     #define bfd_print_private_bfd_data(abfd, file)\
            BFD_SEND (abfd, _bfd_print_private_bfd_data, (abfd, file))
     #define bfd_canonicalize_dynamic_symtab(abfd, asymbols) \
            BFD_SEND (abfd, _bfd_canonicalize_dynamic_symtab, (abfd, asymbols))
     #define bfd_get_synthetic_symtab(abfd, count, syms, dyncount, dynsyms, ret) \
            BFD_SEND (abfd, _bfd_get_synthetic_symtab, (abfd, count, syms, \
                                                        dyncount, dynsyms, ret))
     #define bfd_get_dynamic_reloc_upper_bound(abfd) \
            BFD_SEND (abfd, _bfd_get_dynamic_reloc_upper_bound, (abfd))
     #define bfd_canonicalize_dynamic_reloc(abfd, arels, asyms) \
            BFD_SEND (abfd, _bfd_canonicalize_dynamic_reloc, (abfd, arels, asyms))
     extern bfd_byte *bfd_get_relocated_section_contents
       (bfd *, struct bfd_link_info *, struct bfd_link_order *, bfd_byte *,
        bfd_boolean, asymbol **); bfd_alt_mach_code


     bfd_boolean bfd_alt_mach_code (bfd *abfd, int alternative);

When more than one machine code number is available for the same machine type, this function can be used to switch between the preferred one (alternative == 0) and any others. Currently, only ELF supports this feature, with up to two alternate machine codes.

     struct bfd_preserve
       void *marker;
       void *tdata;
       flagword flags;
       const struct bfd_arch_info *arch_info;
       struct bfd_section *sections;
       struct bfd_section **section_tail;
       unsigned int section_count;
       struct bfd_hash_table section_htab;
     }; bfd_preserve_save


     bfd_boolean bfd_preserve_save (bfd *, struct bfd_preserve *);

When testing an object for compatibility with a particular target back-end, the back-end object_p function needs to set up certain fields in the bfd on successfully recognizing the object. This typically happens in a piecemeal fashion, with failures possible at many points. On failure, the bfd is supposed to be restored to its initial state, which is virtually impossible. However, restoring a subset of the bfd state works in practice. This function stores the subset and reinitializes the bfd. bfd_preserve_restore


     void bfd_preserve_restore (bfd *, struct bfd_preserve *);

This function restores bfd state saved by bfd_preserve_save. If MARKER is non-NULL in struct bfd_preserve then that block and all subsequently bfd_alloc'd memory is freed. bfd_preserve_finish


     void bfd_preserve_finish (bfd *, struct bfd_preserve *);

This function should be called when the bfd state saved by bfd_preserve_save is no longer needed. ie. when the back-end object_p function returns with success. struct bfd_iovec

The struct bfd_iovec contains the internal file I/O class. Each BFD has an instance of this class and all file I/O is routed through it (it is assumed that the instance implements all methods listed below).

     struct bfd_iovec
       /* To avoid problems with macros, a "b" rather than "f"
          prefix is prepended to each method name.  */
       /* Attempt to read/write NBYTES on ABFD's IOSTREAM storing/fetching
          bytes starting at PTR.  Return the number of bytes actually
          transfered (a read past end-of-file returns less than NBYTES),
          or -1 (setting bfd_error) if an error occurs.  */
       file_ptr (*bread) (struct bfd *abfd, void *ptr, file_ptr nbytes);
       file_ptr (*bwrite) (struct bfd *abfd, const void *ptr,
                           file_ptr nbytes);
       /* Return the current IOSTREAM file offset, or -1 (setting bfd_error
          if an error occurs.  */
       file_ptr (*btell) (struct bfd *abfd);
       /* For the following, on successful completion a value of 0 is returned.
          Otherwise, a value of -1 is returned (and  bfd_error is set).  */
       int (*bseek) (struct bfd *abfd, file_ptr offset, int whence);
       int (*bclose) (struct bfd *abfd);
       int (*bflush) (struct bfd *abfd);
       int (*bstat) (struct bfd *abfd, struct stat *sb);
     }; bfd_get_mtime


     long bfd_get_mtime (bfd *abfd);

Return the file modification time (as read from the file system, or from the archive header for archive members). bfd_get_size


     long bfd_get_size (bfd *abfd);

Return the file size (as read from file system) for the file associated with BFD abfd.

The initial motivation for, and use of, this routine is not so we can get the exact size of the object the BFD applies to, since that might not be generally possible (archive members for example). It would be ideal if someone could eventually modify it so that such results were guaranteed.

Instead, we want to ask questions like "is this NNN byte sized object I'm about to try read from file offset YYY reasonable?" As as example of where we might do this, some object formats use string tables for which the first sizeof (long) bytes of the table contain the size of the table itself, including the size bytes. If an application tries to read what it thinks is one of these string tables, without some way to validate the size, and for some reason the size is wrong (byte swapping error, wrong location for the string table, etc.), the only clue is likely to be a read error when it tries to read the table, or a "virtual memory exhausted" error when it tries to allocate 15 bazillon bytes of space for the 15 bazillon byte table it is about to read. This function at least allows us to answer the question, "is the size reasonable?".