2 The raw data contained within a BFD is maintained through the
3 section abstraction. A single BFD may have any number of
4 sections. It keeps hold of them by pointing to the first;
5 each one points to the next in the list.
7 Sections are supported in BFD in @code{section.c}.
13 * section prototypes::
16 @node Section Input, Section Output, Sections, Sections
17 @subsection Section input
18 When a BFD is opened for reading, the section structures are
19 created and attached to the BFD.
21 Each section has a name which describes the section in the
22 outside world---for example, @code{a.out} would contain at least
23 three sections, called @code{.text}, @code{.data} and @code{.bss}.
25 Names need not be unique; for example a COFF file may have several
26 sections named @code{.data}.
28 Sometimes a BFD will contain more than the ``natural'' number of
29 sections. A back end may attach other sections containing
30 constructor data, or an application may add a section (using
31 @code{bfd_make_section}) to the sections attached to an already open
32 BFD. For example, the linker creates an extra section
33 @code{COMMON} for each input file's BFD to hold information about
36 The raw data is not necessarily read in when
37 the section descriptor is created. Some targets may leave the
38 data in place until a @code{bfd_get_section_contents} call is
39 made. Other back ends may read in all the data at once. For
40 example, an S-record file has to be read once to determine the
41 size of the data. An IEEE-695 file doesn't contain raw data in
42 sections, but data and relocation expressions intermixed, so
43 the data area has to be parsed to get out the data and
46 @node Section Output, typedef asection, Section Input, Sections
47 @subsection Section output
48 To write a new object style BFD, the various sections to be
49 written have to be created. They are attached to the BFD in
50 the same way as input sections; data is written to the
51 sections using @code{bfd_set_section_contents}.
53 Any program that creates or combines sections (e.g., the assembler
54 and linker) must use the @code{asection} fields @code{output_section} and
55 @code{output_offset} to indicate the file sections to which each
56 section must be written. (If the section is being created from
57 scratch, @code{output_section} should probably point to the section
58 itself and @code{output_offset} should probably be zero.)
60 The data to be written comes from input sections attached
61 (via @code{output_section} pointers) to
62 the output sections. The output section structure can be
63 considered a filter for the input section: the output section
64 determines the vma of the output data and the name, but the
65 input section determines the offset into the output section of
66 the data to be written.
68 E.g., to create a section "O", starting at 0x100, 0x123 long,
69 containing two subsections, "A" at offset 0x0 (i.e., at vma
70 0x100) and "B" at offset 0x20 (i.e., at vma 0x120) the @code{asection}
71 structures would look like:
77 output_section -----------> section name "O"
79 section name "B" | size 0x123
82 output_section --------|
85 @subsection Link orders
86 The data within a section is stored in a @dfn{link_order}.
87 These are much like the fixups in @code{gas}. The link_order
88 abstraction allows a section to grow and shrink within itself.
90 A link_order knows how big it is, and which is the next
91 link_order and where the raw data for it is; it also points to
92 a list of relocations which apply to it.
94 The link_order is used by the linker to perform relaxing on
95 final code. The compiler creates code which is as big as
96 necessary to make it work without relaxing, and the user can
97 select whether to relax. Sometimes relaxing takes a lot of
98 time. The linker runs around the relocations to see if any
99 are attached to data which can be shrunk, if so it does it on
100 a link_order by link_order basis.
103 @node typedef asection, section prototypes, Section Output, Sections
104 @subsection typedef asection
105 Here is the section structure:
110 /* This structure is used for a comdat section, as in PE. A comdat
111 section is associated with a particular symbol. When the linker
112 sees a comdat section, it keeps only one of the sections with a
113 given name and associated with a given symbol. */
115 struct bfd_comdat_info
117 /* The name of the symbol associated with a comdat section. */
120 /* The local symbol table index of the symbol associated with a
121 comdat section. This is only meaningful to the object file format
122 specific code; it is not an index into the list returned by
123 bfd_canonicalize_symtab. */
129 /* The name of the section; the name isn't a copy, the pointer is
130 the same as that passed to bfd_make_section. */
134 /* A unique sequence number. */
138 /* Which section is it; 0..nth. */
142 /* The next section in the list belonging to the BFD, or NULL. */
146 /* The field flags contains attributes of the section. Some
147 flags are read in from the object file, and some are
148 synthesized from other information. */
152 #define SEC_NO_FLAGS 0x000
154 /* Tells the OS to allocate space for this section when loading.
155 This is clear for a section containing debug information only. */
156 #define SEC_ALLOC 0x001
158 /* Tells the OS to load the section from the file when loading.
159 This is clear for a .bss section. */
160 #define SEC_LOAD 0x002
162 /* The section contains data still to be relocated, so there is
163 some relocation information too. */
164 #define SEC_RELOC 0x004
166 #if 0 /* Obsolete ? */
167 #define SEC_BALIGN 0x008
170 /* A signal to the OS that the section contains read only data. */
171 #define SEC_READONLY 0x010
173 /* The section contains code only. */
174 #define SEC_CODE 0x020
176 /* The section contains data only. */
177 #define SEC_DATA 0x040
179 /* The section will reside in ROM. */
180 #define SEC_ROM 0x080
182 /* The section contains constructor information. This section
183 type is used by the linker to create lists of constructors and
184 destructors used by @code{g++}. When a back end sees a symbol
185 which should be used in a constructor list, it creates a new
186 section for the type of name (e.g., @code{__CTOR_LIST__}), attaches
187 the symbol to it, and builds a relocation. To build the lists
188 of constructors, all the linker has to do is catenate all the
189 sections called @code{__CTOR_LIST__} and relocate the data
190 contained within - exactly the operations it would peform on
192 #define SEC_CONSTRUCTOR 0x100
194 /* The section is a constructor, and should be placed at the
195 end of the text, data, or bss section(?). */
196 #define SEC_CONSTRUCTOR_TEXT 0x1100
197 #define SEC_CONSTRUCTOR_DATA 0x2100
198 #define SEC_CONSTRUCTOR_BSS 0x3100
200 /* The section has contents - a data section could be
201 @code{SEC_ALLOC} | @code{SEC_HAS_CONTENTS}; a debug section could be
202 @code{SEC_HAS_CONTENTS} */
203 #define SEC_HAS_CONTENTS 0x200
205 /* An instruction to the linker to not output the section
206 even if it has information which would normally be written. */
207 #define SEC_NEVER_LOAD 0x400
209 /* The section is a COFF shared library section. This flag is
210 only for the linker. If this type of section appears in
211 the input file, the linker must copy it to the output file
212 without changing the vma or size. FIXME: Although this
213 was originally intended to be general, it really is COFF
214 specific (and the flag was renamed to indicate this). It
215 might be cleaner to have some more general mechanism to
216 allow the back end to control what the linker does with
218 #define SEC_COFF_SHARED_LIBRARY 0x800
220 /* The section has GOT references. This flag is only for the
221 linker, and is currently only used by the elf32-hppa back end.
222 It will be set if global offset table references were detected
223 in this section, which indicate to the linker that the section
224 contains PIC code, and must be handled specially when doing a
226 #define SEC_HAS_GOT_REF 0x4000
228 /* The section contains common symbols (symbols may be defined
229 multiple times, the value of a symbol is the amount of
230 space it requires, and the largest symbol value is the one
231 used). Most targets have exactly one of these (which we
232 translate to bfd_com_section_ptr), but ECOFF has two. */
233 #define SEC_IS_COMMON 0x8000
235 /* The section contains only debugging information. For
236 example, this is set for ELF .debug and .stab sections.
237 strip tests this flag to see if a section can be
239 #define SEC_DEBUGGING 0x10000
241 /* The contents of this section are held in memory pointed to
242 by the contents field. This is checked by bfd_get_section_contents,
243 and the data is retrieved from memory if appropriate. */
244 #define SEC_IN_MEMORY 0x20000
246 /* The contents of this section are to be excluded by the
247 linker for executable and shared objects unless those
248 objects are to be further relocated. */
249 #define SEC_EXCLUDE 0x40000
251 /* The contents of this section are to be sorted by the
252 based on the address specified in the associated symbol
254 #define SEC_SORT_ENTRIES 0x80000
256 /* When linking, duplicate sections of the same name should be
257 discarded, rather than being combined into a single section as
258 is usually done. This is similar to how common symbols are
259 handled. See SEC_LINK_DUPLICATES below. */
260 #define SEC_LINK_ONCE 0x100000
262 /* If SEC_LINK_ONCE is set, this bitfield describes how the linker
263 should handle duplicate sections. */
264 #define SEC_LINK_DUPLICATES 0x600000
266 /* This value for SEC_LINK_DUPLICATES means that duplicate
267 sections with the same name should simply be discarded. */
268 #define SEC_LINK_DUPLICATES_DISCARD 0x0
270 /* This value for SEC_LINK_DUPLICATES means that the linker
271 should warn if there are any duplicate sections, although
272 it should still only link one copy. */
273 #define SEC_LINK_DUPLICATES_ONE_ONLY 0x200000
275 /* This value for SEC_LINK_DUPLICATES means that the linker
276 should warn if any duplicate sections are a different size. */
277 #define SEC_LINK_DUPLICATES_SAME_SIZE 0x400000
279 /* This value for SEC_LINK_DUPLICATES means that the linker
280 should warn if any duplicate sections contain different
282 #define SEC_LINK_DUPLICATES_SAME_CONTENTS 0x600000
284 /* This section was created by the linker as part of dynamic
285 relocation or other arcane processing. It is skipped when
286 going through the first-pass output, trusting that someone
287 else up the line will take care of it later. */
288 #define SEC_LINKER_CREATED 0x800000
290 /* This section should not be subject to garbage collection. */
291 #define SEC_KEEP 0x1000000
293 /* This section contains "short" data, and should be placed
295 #define SEC_SMALL_DATA 0x2000000
297 /* This section contains data which may be shared with other
298 executables or shared objects. */
299 #define SEC_SHARED 0x4000000
301 /* When a section with this flag is being linked, then if the size of
302 the input section is less than a page, it should not cross a page
303 boundary. If the size of the input section is one page or more, it
304 should be aligned on a page boundary. */
305 #define SEC_BLOCK 0x8000000
307 /* Conditionally link this section; do not link if there are no
308 references found to any symbol in the section. */
309 #define SEC_CLINK 0x10000000
311 /* End of section flags. */
313 /* Some internal packed boolean fields. */
315 /* See the vma field. */
316 unsigned int user_set_vma : 1;
318 /* Whether relocations have been processed. */
319 unsigned int reloc_done : 1;
321 /* A mark flag used by some of the linker backends. */
322 unsigned int linker_mark : 1;
324 /* Another mark flag used by some of the linker backends. Set for
325 output sections that have a input section. */
326 unsigned int linker_has_input : 1;
328 /* A mark flag used by some linker backends for garbage collection. */
329 unsigned int gc_mark : 1;
331 /* Used by the ELF code to mark sections which have been allocated to segments. */
332 unsigned int segment_mark : 1;
334 /* End of internal packed boolean fields. */
336 /* The virtual memory address of the section - where it will be
337 at run time. The symbols are relocated against this. The
338 user_set_vma flag is maintained by bfd; if it's not set, the
339 backend can assign addresses (for example, in @code{a.out}, where
340 the default address for @code{.data} is dependent on the specific
341 target and various flags). */
345 /* The load address of the section - where it would be in a
346 rom image; really only used for writing section header
351 /* The size of the section in octets, as it will be output.
352 Contains a value even if the section has no contents (e.g., the
353 size of @code{.bss}). This will be filled in after relocation. */
355 bfd_size_type _cooked_size;
357 /* The original size on disk of the section, in octets. Normally this
358 value is the same as the size, but if some relaxing has
359 been done, then this value will be bigger. */
361 bfd_size_type _raw_size;
363 /* If this section is going to be output, then this value is the
364 offset in *bytes* into the output section of the first byte in the
365 input section (byte ==> smallest addressable unit on the
366 target). In most cases, if this was going to start at the
367 100th octet (8-bit quantity) in the output section, this value
368 would be 100. However, if the target byte size is 16 bits
369 (bfd_octets_per_byte is "2"), this value would be 50. */
371 bfd_vma output_offset;
373 /* The output section through which to map on output. */
375 struct sec *output_section;
377 /* The alignment requirement of the section, as an exponent of 2 -
378 e.g., 3 aligns to 2^3 (or 8). */
380 unsigned int alignment_power;
382 /* If an input section, a pointer to a vector of relocation
383 records for the data in this section. */
385 struct reloc_cache_entry *relocation;
387 /* If an output section, a pointer to a vector of pointers to
388 relocation records for the data in this section. */
390 struct reloc_cache_entry **orelocation;
392 /* The number of relocation records in one of the above */
394 unsigned reloc_count;
396 /* Information below is back end specific - and not always used
399 /* File position of section data. */
403 /* File position of relocation info. */
405 file_ptr rel_filepos;
407 /* File position of line data. */
409 file_ptr line_filepos;
411 /* Pointer to data for applications. */
415 /* If the SEC_IN_MEMORY flag is set, this points to the actual
417 unsigned char *contents;
419 /* Attached line number information. */
423 /* Number of line number records. */
425 unsigned int lineno_count;
427 /* Optional information about a COMDAT entry; NULL if not COMDAT. */
429 struct bfd_comdat_info *comdat;
431 /* Points to the kept section if this section is a link-once section,
433 struct sec *kept_section;
435 /* When a section is being output, this value changes as more
436 linenumbers are written out. */
438 file_ptr moving_line_filepos;
440 /* What the section number is in the target world. */
446 /* If this is a constructor section then here is a list of the
447 relocations created to relocate items within it. */
449 struct relent_chain *constructor_chain;
451 /* The BFD which owns the section. */
455 /* A symbol which points at this section only */
456 struct symbol_cache_entry *symbol;
457 struct symbol_cache_entry **symbol_ptr_ptr;
459 struct bfd_link_order *link_order_head;
460 struct bfd_link_order *link_order_tail;
463 /* These sections are global, and are managed by BFD. The application
464 and target back end are not permitted to change the values in
465 these sections. New code should use the section_ptr macros rather
466 than referring directly to the const sections. The const sections
467 may eventually vanish. */
468 #define BFD_ABS_SECTION_NAME "*ABS*"
469 #define BFD_UND_SECTION_NAME "*UND*"
470 #define BFD_COM_SECTION_NAME "*COM*"
471 #define BFD_IND_SECTION_NAME "*IND*"
473 /* the absolute section */
474 extern const asection bfd_abs_section;
475 #define bfd_abs_section_ptr ((asection *) &bfd_abs_section)
476 #define bfd_is_abs_section(sec) ((sec) == bfd_abs_section_ptr)
477 /* Pointer to the undefined section */
478 extern const asection bfd_und_section;
479 #define bfd_und_section_ptr ((asection *) &bfd_und_section)
480 #define bfd_is_und_section(sec) ((sec) == bfd_und_section_ptr)
481 /* Pointer to the common section */
482 extern const asection bfd_com_section;
483 #define bfd_com_section_ptr ((asection *) &bfd_com_section)
484 /* Pointer to the indirect section */
485 extern const asection bfd_ind_section;
486 #define bfd_ind_section_ptr ((asection *) &bfd_ind_section)
487 #define bfd_is_ind_section(sec) ((sec) == bfd_ind_section_ptr)
489 extern const struct symbol_cache_entry * const bfd_abs_symbol;
490 extern const struct symbol_cache_entry * const bfd_com_symbol;
491 extern const struct symbol_cache_entry * const bfd_und_symbol;
492 extern const struct symbol_cache_entry * const bfd_ind_symbol;
493 #define bfd_get_section_size_before_reloc(section) \
494 ((section)->reloc_done ? (abort (), (bfd_size_type) 1) \
495 : (section)->_raw_size)
496 #define bfd_get_section_size_after_reloc(section) \
497 ((section)->reloc_done ? (section)->_cooked_size \
498 : (abort (), (bfd_size_type) 1))
501 @node section prototypes, , typedef asection, Sections
502 @subsection Section prototypes
503 These are the functions exported by the section handling part of BFD.
505 @findex bfd_get_section_by_name
506 @subsubsection @code{bfd_get_section_by_name}
509 asection *bfd_get_section_by_name(bfd *abfd, const char *name);
511 @strong{Description}@*
512 Run through @var{abfd} and return the one of the
513 @code{asection}s whose name matches @var{name}, otherwise @code{NULL}.
514 @xref{Sections}, for more information.
516 This should only be used in special cases; the normal way to process
517 all sections of a given name is to use @code{bfd_map_over_sections} and
518 @code{strcmp} on the name (or better yet, base it on the section flags
519 or something else) for each section.
521 @findex bfd_get_unique_section_name
522 @subsubsection @code{bfd_get_unique_section_name}
525 char *bfd_get_unique_section_name(bfd *abfd,
529 @strong{Description}@*
530 Invent a section name that is unique in @var{abfd} by tacking
531 a dot and a digit suffix onto the original @var{templat}. If
532 @var{count} is non-NULL, then it specifies the first number
533 tried as a suffix to generate a unique name. The value
534 pointed to by @var{count} will be incremented in this case.
536 @findex bfd_make_section_old_way
537 @subsubsection @code{bfd_make_section_old_way}
540 asection *bfd_make_section_old_way(bfd *abfd, const char *name);
542 @strong{Description}@*
543 Create a new empty section called @var{name}
544 and attach it to the end of the chain of sections for the
545 BFD @var{abfd}. An attempt to create a section with a name which
546 is already in use returns its pointer without changing the
549 It has the funny name since this is the way it used to be
550 before it was rewritten....
556 @code{bfd_error_invalid_operation} -
557 If output has already started for this BFD.
559 @code{bfd_error_no_memory} -
560 If memory allocation fails.
563 @findex bfd_make_section_anyway
564 @subsubsection @code{bfd_make_section_anyway}
567 asection *bfd_make_section_anyway(bfd *abfd, const char *name);
569 @strong{Description}@*
570 Create a new empty section called @var{name} and attach it to the end of
571 the chain of sections for @var{abfd}. Create a new section even if there
572 is already a section with that name.
574 Return @code{NULL} and set @code{bfd_error} on error; possible errors are:
578 @code{bfd_error_invalid_operation} - If output has already started for @var{abfd}.
580 @code{bfd_error_no_memory} - If memory allocation fails.
583 @findex bfd_make_section
584 @subsubsection @code{bfd_make_section}
587 asection *bfd_make_section(bfd *, const char *name);
589 @strong{Description}@*
590 Like @code{bfd_make_section_anyway}, but return @code{NULL} (without calling
591 bfd_set_error ()) without changing the section chain if there is already a
592 section named @var{name}. If there is an error, return @code{NULL} and set
595 @findex bfd_set_section_flags
596 @subsubsection @code{bfd_set_section_flags}
599 boolean bfd_set_section_flags(bfd *abfd, asection *sec, flagword flags);
601 @strong{Description}@*
602 Set the attributes of the section @var{sec} in the BFD
603 @var{abfd} to the value @var{flags}. Return @code{true} on success,
604 @code{false} on error. Possible error returns are:
609 @code{bfd_error_invalid_operation} -
610 The section cannot have one or more of the attributes
611 requested. For example, a .bss section in @code{a.out} may not
612 have the @code{SEC_HAS_CONTENTS} field set.
615 @findex bfd_map_over_sections
616 @subsubsection @code{bfd_map_over_sections}
619 void bfd_map_over_sections(bfd *abfd,
620 void (*func) (bfd *abfd,
625 @strong{Description}@*
626 Call the provided function @var{func} for each section
627 attached to the BFD @var{abfd}, passing @var{obj} as an
628 argument. The function will be called as if by
631 func(abfd, the_section, obj);
634 This is the prefered method for iterating over sections; an
635 alternative would be to use a loop:
639 for (p = abfd->sections; p != NULL; p = p->next)
643 @findex bfd_set_section_size
644 @subsubsection @code{bfd_set_section_size}
647 boolean bfd_set_section_size(bfd *abfd, asection *sec, bfd_size_type val);
649 @strong{Description}@*
650 Set @var{sec} to the size @var{val}. If the operation is
651 ok, then @code{true} is returned, else @code{false}.
653 Possible error returns:
657 @code{bfd_error_invalid_operation} -
658 Writing has started to the BFD, so setting the size is invalid.
661 @findex bfd_set_section_contents
662 @subsubsection @code{bfd_set_section_contents}
665 boolean bfd_set_section_contents
670 bfd_size_type count);
672 @strong{Description}@*
673 Sets the contents of the section @var{section} in BFD
674 @var{abfd} to the data starting in memory at @var{data}. The
675 data is written to the output section starting at offset
676 @var{offset} for @var{count} octets.
678 Normally @code{true} is returned, else @code{false}. Possible error
683 @code{bfd_error_no_contents} -
684 The output section does not have the @code{SEC_HAS_CONTENTS}
685 attribute, so nothing can be written to it.
689 This routine is front end to the back end function
690 @code{_bfd_set_section_contents}.
692 @findex bfd_get_section_contents
693 @subsubsection @code{bfd_get_section_contents}
696 boolean bfd_get_section_contents
697 (bfd *abfd, asection *section, PTR location,
698 file_ptr offset, bfd_size_type count);
700 @strong{Description}@*
701 Read data from @var{section} in BFD @var{abfd}
702 into memory starting at @var{location}. The data is read at an
703 offset of @var{offset} from the start of the input section,
704 and is read for @var{count} bytes.
706 If the contents of a constructor with the @code{SEC_CONSTRUCTOR}
707 flag set are requested or if the section does not have the
708 @code{SEC_HAS_CONTENTS} flag set, then the @var{location} is filled
709 with zeroes. If no errors occur, @code{true} is returned, else
712 @findex bfd_copy_private_section_data
713 @subsubsection @code{bfd_copy_private_section_data}
716 boolean bfd_copy_private_section_data(bfd *ibfd, asection *isec, bfd *obfd, asection *osec);
718 @strong{Description}@*
719 Copy private section information from @var{isec} in the BFD
720 @var{ibfd} to the section @var{osec} in the BFD @var{obfd}.
721 Return @code{true} on success, @code{false} on error. Possible error
727 @code{bfd_error_no_memory} -
728 Not enough memory exists to create private data for @var{osec}.
731 #define bfd_copy_private_section_data(ibfd, isection, obfd, osection) \
732 BFD_SEND (obfd, _bfd_copy_private_section_data, \
733 (ibfd, isection, obfd, osection))
736 @findex _bfd_strip_section_from_output
737 @subsubsection @code{_bfd_strip_section_from_output}
740 void _bfd_strip_section_from_output
741 (struct bfd_link_info *info, asection *section);
743 @strong{Description}@*
744 Remove @var{section} from the output. If the output section
745 becomes empty, remove it from the output bfd. @var{info} may
746 be NULL; if it is not, it is used to decide whether the output