This is Info file bfd.info, produced by Makeinfo-1.64 from the input file ./bfd.texinfo. START-INFO-DIR-ENTRY * Bfd: (bfd). The Binary File Descriptor library. END-INFO-DIR-ENTRY This file documents the BFD library. Copyright (C) 1991 Free Software Foundation, Inc. Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies. Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying, subject to the terms of the GNU General Public License, which includes the provision that the entire resulting derived work is distributed under the terms of a permission notice identical to this one. Permission is granted to copy and distribute translations of this manual into another language, under the above conditions for modified versions.  File: bfd.info, Node: section prototypes, Prev: typedef asection, Up: Sections Section prototypes ------------------ These are the functions exported by the section handling part of BFD. `bfd_get_section_by_name' ......................... *Synopsis* asection *bfd_get_section_by_name(bfd *abfd, CONST char *name); *Description* Run through ABFD and return the one of the `asection's whose name matches NAME, otherwise `NULL'. *Note Sections::, for more information. This should only be used in special cases; the normal way to process all sections of a given name is to use `bfd_map_over_sections' and `strcmp' on the name (or better yet, base it on the section flags or something else) for each section. `bfd_make_section_old_way' .......................... *Synopsis* asection *bfd_make_section_old_way(bfd *abfd, CONST char *name); *Description* Create a new empty section called NAME and attach it to the end of the chain of sections for the BFD ABFD. An attempt to create a section with a name which is already in use returns its pointer without changing the section chain. It has the funny name since this is the way it used to be before it was rewritten.... Possible errors are: * `bfd_error_invalid_operation' - If output has already started for this BFD. * `bfd_error_no_memory' - If memory allocation fails. `bfd_make_section_anyway' ......................... *Synopsis* asection *bfd_make_section_anyway(bfd *abfd, CONST char *name); *Description* Create a new empty section called NAME and attach it to the end of the chain of sections for ABFD. Create a new section even if there is already a section with that name. Return `NULL' and set `bfd_error' on error; possible errors are: * `bfd_error_invalid_operation' - If output has already started for ABFD. * `bfd_error_no_memory' - If memory allocation fails. `bfd_make_section' .................. *Synopsis* asection *bfd_make_section(bfd *, CONST char *name); *Description* Like `bfd_make_section_anyway', but return `NULL' (without calling bfd_set_error ()) without changing the section chain if there is already a section named NAME. If there is an error, return `NULL' and set `bfd_error'. `bfd_set_section_flags' ....................... *Synopsis* boolean bfd_set_section_flags(bfd *abfd, asection *sec, flagword flags); *Description* Set the attributes of the section SEC in the BFD ABFD to the value FLAGS. Return `true' on success, `false' on error. Possible error returns are: * `bfd_error_invalid_operation' - The section cannot have one or more of the attributes requested. For example, a .bss section in `a.out' may not have the `SEC_HAS_CONTENTS' field set. `bfd_map_over_sections' ....................... *Synopsis* void bfd_map_over_sections(bfd *abfd, void (*func)(bfd *abfd, asection *sect, PTR obj), PTR obj); *Description* Call the provided function FUNC for each section attached to the BFD ABFD, passing OBJ as an argument. The function will be called as if by func(abfd, the_section, obj); This is the prefered method for iterating over sections; an alternative would be to use a loop: section *p; for (p = abfd->sections; p != NULL; p = p->next) func(abfd, p, ...) `bfd_set_section_size' ...................... *Synopsis* boolean bfd_set_section_size(bfd *abfd, asection *sec, bfd_size_type val); *Description* Set SEC to the size VAL. If the operation is ok, then `true' is returned, else `false'. Possible error returns: * `bfd_error_invalid_operation' - Writing has started to the BFD, so setting the size is invalid. `bfd_set_section_contents' .......................... *Synopsis* boolean bfd_set_section_contents (bfd *abfd, asection *section, PTR data, file_ptr offset, bfd_size_type count); *Description* Sets the contents of the section SECTION in BFD ABFD to the data starting in memory at DATA. The data is written to the output section starting at offset OFFSET for COUNT bytes. Normally `true' is returned, else `false'. Possible error returns are: * `bfd_error_no_contents' - The output section does not have the `SEC_HAS_CONTENTS' attribute, so nothing can be written to it. * and some more too This routine is front end to the back end function `_bfd_set_section_contents'. `bfd_get_section_contents' .......................... *Synopsis* boolean bfd_get_section_contents (bfd *abfd, asection *section, PTR location, file_ptr offset, bfd_size_type count); *Description* Read data from SECTION in BFD ABFD into memory starting at LOCATION. The data is read at an offset of OFFSET from the start of the input section, and is read for COUNT bytes. If the contents of a constructor with the `SEC_CONSTRUCTOR' flag set are requested or if the section does not have the `SEC_HAS_CONTENTS' flag set, then the LOCATION is filled with zeroes. If no errors occur, `true' is returned, else `false'. `bfd_copy_private_section_data' ............................... *Synopsis* boolean bfd_copy_private_section_data(bfd *ibfd, asection *isec, bfd *obfd, asection *osec); *Description* Copy private section information from ISEC in the BFD IBFD to the section OSEC in the BFD OBFD. Return `true' on success, `false' on error. Possible error returns are: * `bfd_error_no_memory' - Not enough memory exists to create private data for OSEC. #define bfd_copy_private_section_data(ibfd, isection, obfd, osection) \ BFD_SEND (obfd, _bfd_copy_private_section_data, \ (ibfd, isection, obfd, osection))  File: bfd.info, Node: Symbols, Next: Archives, Prev: Sections, Up: BFD front end Symbols ======= BFD tries to maintain as much symbol information as it can when it moves information from file to file. BFD passes information to applications though the `asymbol' structure. When the application requests the symbol table, BFD reads the table in the native form and translates parts of it into the internal format. To maintain more than the information passed to applications, some targets keep some information "behind the scenes" in a structure only the particular back end knows about. For example, the coff back end keeps the original symbol table structure as well as the canonical structure when a BFD is read in. On output, the coff back end can reconstruct the output symbol table so that no information is lost, even information unique to coff which BFD doesn't know or understand. If a coff symbol table were read, but were written through an a.out back end, all the coff specific information would be lost. The symbol table of a BFD is not necessarily read in until a canonicalize request is made. Then the BFD back end fills in a table provided by the application with pointers to the canonical information. To output symbols, the application provides BFD with a table of pointers to pointers to `asymbol's. This allows applications like the linker to output a symbol as it was read, since the "behind the scenes" information will be still available. * Menu: * Reading Symbols:: * Writing Symbols:: * Mini Symbols:: * typedef asymbol:: * symbol handling functions::  File: bfd.info, Node: Reading Symbols, Next: Writing Symbols, Prev: Symbols, Up: Symbols Reading symbols --------------- There are two stages to reading a symbol table from a BFD: allocating storage, and the actual reading process. This is an excerpt from an application which reads the symbol table: long storage_needed; asymbol **symbol_table; long number_of_symbols; long i; storage_needed = bfd_get_symtab_upper_bound (abfd); if (storage_needed < 0) FAIL if (storage_needed == 0) { return ; } symbol_table = (asymbol **) xmalloc (storage_needed); ... number_of_symbols = bfd_canonicalize_symtab (abfd, symbol_table); if (number_of_symbols < 0) FAIL for (i = 0; i < number_of_symbols; i++) { process_symbol (symbol_table[i]); } All storage for the symbols themselves is in an objalloc connected to the BFD; it is freed when the BFD is closed.  File: bfd.info, Node: Writing Symbols, Next: Mini Symbols, Prev: Reading Symbols, Up: Symbols Writing symbols --------------- Writing of a symbol table is automatic when a BFD open for writing is closed. The application attaches a vector of pointers to pointers to symbols to the BFD being written, and fills in the symbol count. The close and cleanup code reads through the table provided and performs all the necessary operations. The BFD output code must always be provided with an "owned" symbol: one which has come from another BFD, or one which has been created using `bfd_make_empty_symbol'. Here is an example showing the creation of a symbol table with only one element: #include "bfd.h" main() { bfd *abfd; asymbol *ptrs[2]; asymbol *new; abfd = bfd_openw("foo","a.out-sunos-big"); bfd_set_format(abfd, bfd_object); new = bfd_make_empty_symbol(abfd); new->name = "dummy_symbol"; new->section = bfd_make_section_old_way(abfd, ".text"); new->flags = BSF_GLOBAL; new->value = 0x12345; ptrs[0] = new; ptrs[1] = (asymbol *)0; bfd_set_symtab(abfd, ptrs, 1); bfd_close(abfd); } ./makesym nm foo 00012345 A dummy_symbol Many formats cannot represent arbitary symbol information; for instance, the `a.out' object format does not allow an arbitary number of sections. A symbol pointing to a section which is not one of `.text', `.data' or `.bss' cannot be described.  File: bfd.info, Node: Mini Symbols, Next: typedef asymbol, Prev: Writing Symbols, Up: Symbols Mini Symbols ------------ Mini symbols provide read-only access to the symbol table. They use less memory space, but require more time to access. They can be useful for tools like nm or objdump, which may have to handle symbol tables of extremely large executables. The `bfd_read_minisymbols' function will read the symbols into memory in an internal form. It will return a `void *' pointer to a block of memory, a symbol count, and the size of each symbol. The pointer is allocated using `malloc', and should be freed by the caller when it is no longer needed. The function `bfd_minisymbol_to_symbol' will take a pointer to a minisymbol, and a pointer to a structure returned by `bfd_make_empty_symbol', and return a `asymbol' structure. The return value may or may not be the same as the value from `bfd_make_empty_symbol' which was passed in.  File: bfd.info, Node: typedef asymbol, Next: symbol handling functions, Prev: Mini Symbols, Up: Symbols typedef asymbol --------------- An `asymbol' has the form: . typedef struct symbol_cache_entry { /* A pointer to the BFD which owns the symbol. This information is necessary so that a back end can work out what additional information (invisible to the application writer) is carried with the symbol. This field is *almost* redundant, since you can use section->owner instead, except that some symbols point to the global sections bfd_{abs,com,und}_section. This could be fixed by making these globals be per-bfd (or per-target-flavor). FIXME. */ struct _bfd *the_bfd; /* Use bfd_asymbol_bfd(sym) to access this field. */ /* The text of the symbol. The name is left alone, and not copied; the application may not alter it. */ CONST char *name; /* The value of the symbol. This really should be a union of a numeric value with a pointer, since some flags indicate that a pointer to another symbol is stored here. */ symvalue value; /* Attributes of a symbol: */ #define BSF_NO_FLAGS 0x00 /* The symbol has local scope; `static' in `C'. The value is the offset into the section of the data. */ #define BSF_LOCAL 0x01 /* The symbol has global scope; initialized data in `C'. The value is the offset into the section of the data. */ #define BSF_GLOBAL 0x02 /* The symbol has global scope and is exported. The value is the offset into the section of the data. */ #define BSF_EXPORT BSF_GLOBAL /* no real difference */ /* A normal C symbol would be one of: `BSF_LOCAL', `BSF_FORT_COMM', `BSF_UNDEFINED' or `BSF_GLOBAL' */ /* The symbol is a debugging record. The value has an arbitary meaning. */ #define BSF_DEBUGGING 0x08 /* The symbol denotes a function entry point. Used in ELF, perhaps others someday. */ #define BSF_FUNCTION 0x10 /* Used by the linker. */ #define BSF_KEEP 0x20 #define BSF_KEEP_G 0x40 /* A weak global symbol, overridable without warnings by a regular global symbol of the same name. */ #define BSF_WEAK 0x80 /* This symbol was created to point to a section, e.g. ELF's STT_SECTION symbols. */ #define BSF_SECTION_SYM 0x100 /* The symbol used to be a common symbol, but now it is allocated. */ #define BSF_OLD_COMMON 0x200 /* The default value for common data. */ #define BFD_FORT_COMM_DEFAULT_VALUE 0 /* In some files the type of a symbol sometimes alters its location in an output file - ie in coff a `ISFCN' symbol which is also `C_EXT' symbol appears where it was declared and not at the end of a section. This bit is set by the target BFD part to convey this information. */ #define BSF_NOT_AT_END 0x400 /* Signal that the symbol is the label of constructor section. */ #define BSF_CONSTRUCTOR 0x800 /* Signal that the symbol is a warning symbol. The name is a warning. The name of the next symbol is the one to warn about; if a reference is made to a symbol with the same name as the next symbol, a warning is issued by the linker. */ #define BSF_WARNING 0x1000 /* Signal that the symbol is indirect. This symbol is an indirect pointer to the symbol with the same name as the next symbol. */ #define BSF_INDIRECT 0x2000 /* BSF_FILE marks symbols that contain a file name. This is used for ELF STT_FILE symbols. */ #define BSF_FILE 0x4000 /* Symbol is from dynamic linking information. */ #define BSF_DYNAMIC 0x8000 /* The symbol denotes a data object. Used in ELF, and perhaps others someday. */ #define BSF_OBJECT 0x10000 flagword flags; /* A pointer to the section to which this symbol is relative. This will always be non NULL, there are special sections for undefined and absolute symbols. */ struct sec *section; /* Back end special data. */ union { PTR p; bfd_vma i; } udata; } asymbol;  File: bfd.info, Node: symbol handling functions, Prev: typedef asymbol, Up: Symbols Symbol handling functions ------------------------- `bfd_get_symtab_upper_bound' ............................ *Description* Return the number of bytes required to store a vector of pointers to `asymbols' for all the symbols in the BFD ABFD, including a terminal NULL pointer. If there are no symbols in the BFD, then return 0. If an error occurs, return -1. #define bfd_get_symtab_upper_bound(abfd) \ BFD_SEND (abfd, _bfd_get_symtab_upper_bound, (abfd)) `bfd_is_local_label' .................... *Synopsis* boolean bfd_is_local_label(bfd *abfd, asymbol *sym); *Description* Return true if the given symbol SYM in the BFD ABFD is a compiler generated local label, else return false. `bfd_is_local_label_name' ......................... *Synopsis* boolean bfd_is_local_label_name(bfd *abfd, const char *name); *Description* Return true if a symbol with the name NAME in the BFD ABFD is a compiler generated local label, else return false. This just checks whether the name has the form of a local label. #define bfd_is_local_label_name(abfd, name) \ BFD_SEND (abfd, _bfd_is_local_label_name, (abfd, name)) `bfd_canonicalize_symtab' ......................... *Description* Read the symbols from the BFD ABFD, and fills in the vector LOCATION with pointers to the symbols and a trailing NULL. Return the actual number of symbol pointers, not including the NULL. #define bfd_canonicalize_symtab(abfd, location) \ BFD_SEND (abfd, _bfd_canonicalize_symtab,\ (abfd, location)) `bfd_set_symtab' ................ *Synopsis* boolean bfd_set_symtab (bfd *abfd, asymbol **location, unsigned int count); *Description* Arrange that when the output BFD ABFD is closed, the table LOCATION of COUNT pointers to symbols will be written. `bfd_print_symbol_vandf' ........................ *Synopsis* void bfd_print_symbol_vandf(PTR file, asymbol *symbol); *Description* Print the value and flags of the SYMBOL supplied to the stream FILE. `bfd_make_empty_symbol' ....................... *Description* Create a new `asymbol' structure for the BFD ABFD and return a pointer to it. This routine is necessary because each back end has private information surrounding the `asymbol'. Building your own `asymbol' and pointing to it will not create the private information, and will cause problems later on. #define bfd_make_empty_symbol(abfd) \ BFD_SEND (abfd, _bfd_make_empty_symbol, (abfd)) `bfd_make_debug_symbol' ....................... *Description* Create a new `asymbol' structure for the BFD ABFD, to be used as a debugging symbol. Further details of its use have yet to be worked out. #define bfd_make_debug_symbol(abfd,ptr,size) \ BFD_SEND (abfd, _bfd_make_debug_symbol, (abfd, ptr, size)) `bfd_decode_symclass' ..................... *Description* Return a character corresponding to the symbol class of SYMBOL, or '?' for an unknown class. *Synopsis* int bfd_decode_symclass(asymbol *symbol); `bfd_symbol_info' ................. *Description* Fill in the basic info about symbol that nm needs. Additional info may be added by the back-ends after calling this function. *Synopsis* void bfd_symbol_info(asymbol *symbol, symbol_info *ret); `bfd_copy_private_symbol_data' .............................. *Synopsis* boolean bfd_copy_private_symbol_data(bfd *ibfd, asymbol *isym, bfd *obfd, asymbol *osym); *Description* Copy private symbol information from ISYM in the BFD IBFD to the symbol OSYM in the BFD OBFD. Return `true' on success, `false' on error. Possible error returns are: * `bfd_error_no_memory' - Not enough memory exists to create private data for OSEC. #define bfd_copy_private_symbol_data(ibfd, isymbol, obfd, osymbol) \ BFD_SEND (obfd, _bfd_copy_private_symbol_data, \ (ibfd, isymbol, obfd, osymbol))  File: bfd.info, Node: Archives, Next: Formats, Prev: Symbols, Up: BFD front end Archives ======== *Description* An archive (or library) is just another BFD. It has a symbol table, although there's not much a user program will do with it. The big difference between an archive BFD and an ordinary BFD is that the archive doesn't have sections. Instead it has a chain of BFDs that are considered its contents. These BFDs can be manipulated like any other. The BFDs contained in an archive opened for reading will all be opened for reading. You may put either input or output BFDs into an archive opened for output; they will be handled correctly when the archive is closed. Use `bfd_openr_next_archived_file' to step through the contents of an archive opened for input. You don't have to read the entire archive if you don't want to! Read it until you find what you want. Archive contents of output BFDs are chained through the `next' pointer in a BFD. The first one is findable through the `archive_head' slot of the archive. Set it with `bfd_set_archive_head' (q.v.). A given BFD may be in only one open output archive at a time. As expected, the BFD archive code is more general than the archive code of any given environment. BFD archives may contain files of different formats (e.g., a.out and coff) and even different architectures. You may even place archives recursively into archives! This can cause unexpected confusion, since some archive formats are more expressive than others. For instance, Intel COFF archives can preserve long filenames; SunOS a.out archives cannot. If you move a file from the first to the second format and back again, the filename may be truncated. Likewise, different a.out environments have different conventions as to how they truncate filenames, whether they preserve directory names in filenames, etc. When interoperating with native tools, be sure your files are homogeneous. Beware: most of these formats do not react well to the presence of spaces in filenames. We do the best we can, but can't always handle this case due to restrictions in the format of archives. Many Unix utilities are braindead in regards to spaces and such in filenames anyway, so this shouldn't be much of a restriction. Archives are supported in BFD in `archive.c'. `bfd_get_next_mapent' ..................... *Synopsis* symindex bfd_get_next_mapent(bfd *abfd, symindex previous, carsym **sym); *Description* Step through archive ABFD's symbol table (if it has one). Successively update SYM with the next symbol's information, returning that symbol's (internal) index into the symbol table. Supply `BFD_NO_MORE_SYMBOLS' as the PREVIOUS entry to get the first one; returns `BFD_NO_MORE_SYMBOLS' when you've already got the last one. A `carsym' is a canonical archive symbol. The only user-visible element is its name, a null-terminated string. `bfd_set_archive_head' ...................... *Synopsis* boolean bfd_set_archive_head(bfd *output, bfd *new_head); *Description* Set the head of the chain of BFDs contained in the archive OUTPUT to NEW_HEAD. `bfd_openr_next_archived_file' .............................. *Synopsis* bfd *bfd_openr_next_archived_file(bfd *archive, bfd *previous); *Description* Provided a BFD, ARCHIVE, containing an archive and NULL, open an input BFD on the first contained element and returns that. Subsequent calls should pass the archive and the previous return value to return a created BFD to the next contained element. NULL is returned when there are no more.  File: bfd.info, Node: Formats, Next: Relocations, Prev: Archives, Up: BFD front end File formats ============ A format is a BFD concept of high level file contents type. The formats supported by BFD are: * `bfd_object' The BFD may contain data, symbols, relocations and debug info. * `bfd_archive' The BFD contains other BFDs and an optional index. * `bfd_core' The BFD contains the result of an executable core dump. `bfd_check_format' .................. *Synopsis* boolean bfd_check_format(bfd *abfd, bfd_format format); *Description* Verify if the file attached to the BFD ABFD is compatible with the format FORMAT (i.e., one of `bfd_object', `bfd_archive' or `bfd_core'). If the BFD has been set to a specific target before the call, only the named target and format combination is checked. If the target has not been set, or has been set to `default', then all the known target backends is interrogated to determine a match. If the default target matches, it is used. If not, exactly one target must recognize the file, or an error results. The function returns `true' on success, otherwise `false' with one of the following error codes: * `bfd_error_invalid_operation' - if `format' is not one of `bfd_object', `bfd_archive' or `bfd_core'. * `bfd_error_system_call' - if an error occured during a read - even some file mismatches can cause bfd_error_system_calls. * `file_not_recognised' - none of the backends recognised the file format. * `bfd_error_file_ambiguously_recognized' - more than one backend recognised the file format. `bfd_check_format_matches' .......................... *Synopsis* boolean bfd_check_format_matches(bfd *abfd, bfd_format format, char ***matching); *Description* Like `bfd_check_format', except when it returns false with `bfd_errno' set to `bfd_error_file_ambiguously_recognized'. In that case, if MATCHING is not NULL, it will be filled in with a NULL-terminated list of the names of the formats that matched, allocated with `malloc'. Then the user may choose a format and try again. When done with the list that MATCHING points to, the caller should free it. `bfd_set_format' ................ *Synopsis* boolean bfd_set_format(bfd *abfd, bfd_format format); *Description* This function sets the file format of the BFD ABFD to the format FORMAT. If the target set in the BFD does not support the format requested, the format is invalid, or the BFD is not open for writing, then an error occurs. `bfd_format_string' ................... *Synopsis* CONST char *bfd_format_string(bfd_format format); *Description* Return a pointer to a const string `invalid', `object', `archive', `core', or `unknown', depending upon the value of FORMAT.  File: bfd.info, Node: Relocations, Next: Core Files, Prev: Formats, Up: BFD front end Relocations =========== BFD maintains relocations in much the same way it maintains symbols: they are left alone until required, then read in en-mass and translated into an internal form. A common routine `bfd_perform_relocation' acts upon the canonical form to do the fixup. Relocations are maintained on a per section basis, while symbols are maintained on a per BFD basis. All that a back end has to do to fit the BFD interface is to create a `struct reloc_cache_entry' for each relocation in a particular section, and fill in the right bits of the structures. * Menu: * typedef arelent:: * howto manager::  File: bfd.info, Node: typedef arelent, Next: howto manager, Prev: Relocations, Up: Relocations typedef arelent --------------- This is the structure of a relocation entry: . typedef enum bfd_reloc_status { /* No errors detected */ bfd_reloc_ok, /* The relocation was performed, but there was an overflow. */ bfd_reloc_overflow, /* The address to relocate was not within the section supplied. */ bfd_reloc_outofrange, /* Used by special functions */ bfd_reloc_continue, /* Unsupported relocation size requested. */ bfd_reloc_notsupported, /* Unused */ bfd_reloc_other, /* The symbol to relocate against was undefined. */ bfd_reloc_undefined, /* The relocation was performed, but may not be ok - presently generated only when linking i960 coff files with i960 b.out symbols. If this type is returned, the error_message argument to bfd_perform_relocation will be set. */ bfd_reloc_dangerous } bfd_reloc_status_type; typedef struct reloc_cache_entry { /* A pointer into the canonical table of pointers */ struct symbol_cache_entry **sym_ptr_ptr; /* offset in section */ bfd_size_type address; /* addend for relocation value */ bfd_vma addend; /* Pointer to how to perform the required relocation */ reloc_howto_type *howto; } arelent; *Description* Here is a description of each of the fields within an `arelent': * `sym_ptr_ptr' The symbol table pointer points to a pointer to the symbol associated with the relocation request. It is the pointer into the table returned by the back end's `get_symtab' action. *Note Symbols::. The symbol is referenced through a pointer to a pointer so that tools like the linker can fix up all the symbols of the same name by modifying only one pointer. The relocation routine looks in the symbol and uses the base of the section the symbol is attached to and the value of the symbol as the initial relocation offset. If the symbol pointer is zero, then the section provided is looked up. * `address' The `address' field gives the offset in bytes from the base of the section data which owns the relocation record to the first byte of relocatable information. The actual data relocated will be relative to this point; for example, a relocation type which modifies the bottom two bytes of a four byte word would not touch the first byte pointed to in a big endian world. * `addend' The `addend' is a value provided by the back end to be added (!) to the relocation offset. Its interpretation is dependent upon the howto. For example, on the 68k the code: char foo[]; main() { return foo[0x12345678]; } Could be compiled into: linkw fp,#-4 moveb @#12345678,d0 extbl d0 unlk fp rts This could create a reloc pointing to `foo', but leave the offset in the data, something like: RELOCATION RECORDS FOR [.text]: offset type value 00000006 32 _foo 00000000 4e56 fffc ; linkw fp,#-4 00000004 1039 1234 5678 ; moveb @#12345678,d0 0000000a 49c0 ; extbl d0 0000000c 4e5e ; unlk fp 0000000e 4e75 ; rts Using coff and an 88k, some instructions don't have enough space in them to represent the full address range, and pointers have to be loaded in two parts. So you'd get something like: or.u r13,r0,hi16(_foo+0x12345678) ld.b r2,r13,lo16(_foo+0x12345678) jmp r1 This should create two relocs, both pointing to `_foo', and with 0x12340000 in their addend field. The data would consist of: RELOCATION RECORDS FOR [.text]: offset type value 00000002 HVRT16 _foo+0x12340000 00000006 LVRT16 _foo+0x12340000 00000000 5da05678 ; or.u r13,r0,0x5678 00000004 1c4d5678 ; ld.b r2,r13,0x5678 00000008 f400c001 ; jmp r1 The relocation routine digs out the value from the data, adds it to the addend to get the original offset, and then adds the value of `_foo'. Note that all 32 bits have to be kept around somewhere, to cope with carry from bit 15 to bit 16. One further example is the sparc and the a.out format. The sparc has a similar problem to the 88k, in that some instructions don't have room for an entire offset, but on the sparc the parts are created in odd sized lumps. The designers of the a.out format chose to not use the data within the section for storing part of the offset; all the offset is kept within the reloc. Anything in the data should be ignored. save %sp,-112,%sp sethi %hi(_foo+0x12345678),%g2 ldsb [%g2+%lo(_foo+0x12345678)],%i0 ret restore Both relocs contain a pointer to `foo', and the offsets contain junk. RELOCATION RECORDS FOR [.text]: offset type value 00000004 HI22 _foo+0x12345678 00000008 LO10 _foo+0x12345678 00000000 9de3bf90 ; save %sp,-112,%sp 00000004 05000000 ; sethi %hi(_foo+0),%g2 00000008 f048a000 ; ldsb [%g2+%lo(_foo+0)],%i0 0000000c 81c7e008 ; ret 00000010 81e80000 ; restore * `howto' The `howto' field can be imagined as a relocation instruction. It is a pointer to a structure which contains information on what to do with all of the other information in the reloc record and data section. A back end would normally have a relocation instruction set and turn relocations into pointers to the correct structure on input - but it would be possible to create each howto field on demand. `enum complain_overflow' ........................ Indicates what sort of overflow checking should be done when performing a relocation. . enum complain_overflow { /* Do not complain on overflow. */ complain_overflow_dont, /* Complain if the bitfield overflows, whether it is considered as signed or unsigned. */ complain_overflow_bitfield, /* Complain if the value overflows when considered as signed number. */ complain_overflow_signed, /* Complain if the value overflows when considered as an unsigned number. */ complain_overflow_unsigned }; `reloc_howto_type' .................. The `reloc_howto_type' is a structure which contains all the information that libbfd needs to know to tie up a back end's data. .struct symbol_cache_entry; /* Forward declaration */ struct reloc_howto_struct { /* The type field has mainly a documentary use - the back end can do what it wants with it, though normally the back end's external idea of what a reloc number is stored in this field. For example, a PC relative word relocation in a coff environment has the type 023 - because that's what the outside world calls a R_PCRWORD reloc. */ unsigned int type; /* The value the final relocation is shifted right by. This drops unwanted data from the relocation. */ unsigned int rightshift; /* The size of the item to be relocated. This is *not* a power-of-two measure. To get the number of bytes operated on by a type of relocation, use bfd_get_reloc_size. */ int size; /* The number of bits in the item to be relocated. This is used when doing overflow checking. */ unsigned int bitsize; /* Notes that the relocation is relative to the location in the data section of the addend. The relocation function will subtract from the relocation value the address of the location being relocated. */ boolean pc_relative; /* The bit position of the reloc value in the destination. The relocated value is left shifted by this amount. */ unsigned int bitpos; /* What type of overflow error should be checked for when relocating. */ enum complain_overflow complain_on_overflow; /* If this field is non null, then the supplied function is called rather than the normal function. This allows really strange relocation methods to be accomodated (e.g., i960 callj instructions). */ bfd_reloc_status_type (*special_function) PARAMS ((bfd *abfd, arelent *reloc_entry, struct symbol_cache_entry *symbol, PTR data, asection *input_section, bfd *output_bfd, char **error_message)); /* The textual name of the relocation type. */ char *name; /* When performing a partial link, some formats must modify the relocations rather than the data - this flag signals this.*/ boolean partial_inplace; /* The src_mask selects which parts of the read in data are to be used in the relocation sum. E.g., if this was an 8 bit bit of data which we read and relocated, this would be 0x000000ff. When we have relocs which have an addend, such as sun4 extended relocs, the value in the offset part of a relocating field is garbage so we never use it. In this case the mask would be 0x00000000. */ bfd_vma src_mask; /* The dst_mask selects which parts of the instruction are replaced into the instruction. In most cases src_mask == dst_mask, except in the above special case, where dst_mask would be 0x000000ff, and src_mask would be 0x00000000. */ bfd_vma dst_mask; /* When some formats create PC relative instructions, they leave the value of the pc of the place being relocated in the offset slot of the instruction, so that a PC relative relocation can be made just by adding in an ordinary offset (e.g., sun3 a.out). Some formats leave the displacement part of an instruction empty (e.g., m88k bcs); this flag signals the fact.*/ boolean pcrel_offset; }; `The HOWTO Macro' ................. *Description* The HOWTO define is horrible and will go away. #define HOWTO(C, R,S,B, P, BI, O, SF, NAME, INPLACE, MASKSRC, MASKDST, PC) \ {(unsigned)C,R,S,B, P, BI, O,SF,NAME,INPLACE,MASKSRC,MASKDST,PC} *Description* And will be replaced with the totally magic way. But for the moment, we are compatible, so do it this way. #define NEWHOWTO( FUNCTION, NAME,SIZE,REL,IN) HOWTO(0,0,SIZE,0,REL,0,complain_overflow_dont,FUNCTION, NAME,false,0,0,IN) *Description* Helper routine to turn a symbol into a relocation value. #define HOWTO_PREPARE(relocation, symbol) \ { \ if (symbol != (asymbol *)NULL) { \ if (bfd_is_com_section (symbol->section)) { \ relocation = 0; \ } \ else { \ relocation = symbol->value; \ } \ } \ } `bfd_get_reloc_size' .................... *Synopsis* int bfd_get_reloc_size (reloc_howto_type *); *Description* For a reloc_howto_type that operates on a fixed number of bytes, this returns the number of bytes operated on. `arelent_chain' ............... *Description* How relocs are tied together in an `asection': typedef struct relent_chain { arelent relent; struct relent_chain *next; } arelent_chain; `bfd_perform_relocation' ........................ *Synopsis* bfd_reloc_status_type bfd_perform_relocation (bfd *abfd, arelent *reloc_entry, PTR data, asection *input_section, bfd *output_bfd, char **error_message); *Description* If OUTPUT_BFD is supplied to this function, the generated image will be relocatable; the relocations are copied to the output file after they have been changed to reflect the new state of the world. There are two ways of reflecting the results of partial linkage in an output file: by modifying the output data in place, and by modifying the relocation record. Some native formats (e.g., basic a.out and basic coff) have no way of specifying an addend in the relocation type, so the addend has to go in the output data. This is no big deal since in these formats the output data slot will always be big enough for the addend. Complex reloc types with addends were invented to solve just this problem. The ERROR_MESSAGE argument is set to an error message if this return `bfd_reloc_dangerous'. `bfd_install_relocation' ........................ *Synopsis* bfd_reloc_status_type bfd_install_relocation (bfd *abfd, arelent *reloc_entry, PTR data, bfd_vma data_start, asection *input_section, char **error_message); *Description* This looks remarkably like `bfd_perform_relocation', except it does not expect that the section contents have been filled in. I.e., it's suitable for use when creating, rather than applying a relocation. For now, this function should be considered reserved for the assembler.