Commit | Line | Data |
---|---|---|
252b5132 | 1 | \input texinfo |
f7e42eb4 NC |
2 | @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, |
3 | @c 2001 | |
4 | @c Free Software Foundation, Inc. | |
252b5132 RH |
5 | @setfilename internals.info |
6 | @node Top | |
7 | @top Assembler Internals | |
8 | @raisesections | |
9 | @cindex internals | |
10 | ||
11 | This chapter describes the internals of the assembler. It is incomplete, but | |
12 | it may help a bit. | |
13 | ||
43da67e8 | 14 | This chapter is not updated regularly, and it may be out of date. |
252b5132 RH |
15 | |
16 | @menu | |
17 | * GAS versions:: GAS versions | |
18 | * Data types:: Data types | |
19 | * GAS processing:: What GAS does when it runs | |
20 | * Porting GAS:: Porting GAS | |
21 | * Relaxation:: Relaxation | |
22 | * Broken words:: Broken words | |
23 | * Internal functions:: Internal functions | |
24 | * Test suite:: Test suite | |
25 | @end menu | |
26 | ||
27 | @node GAS versions | |
28 | @section GAS versions | |
29 | ||
30 | GAS has acquired layers of code over time. The original GAS only supported the | |
31 | a.out object file format, with three sections. Support for multiple sections | |
32 | has been added in two different ways. | |
33 | ||
34 | The preferred approach is to use the version of GAS created when the symbol | |
35 | @code{BFD_ASSEMBLER} is defined. The other versions of GAS are documented for | |
36 | historical purposes, and to help anybody who has to debug code written for | |
37 | them. | |
38 | ||
39 | The type @code{segT} is used to represent a section in code which must work | |
40 | with all versions of GAS. | |
41 | ||
42 | @menu | |
43 | * Original GAS:: Original GAS version | |
44 | * MANY_SEGMENTS:: MANY_SEGMENTS gas version | |
45 | * BFD_ASSEMBLER:: BFD_ASSEMBLER gas version | |
46 | @end menu | |
47 | ||
48 | @node Original GAS | |
49 | @subsection Original GAS | |
50 | ||
51 | The original GAS only supported the a.out object file format with three | |
52 | sections: @samp{.text}, @samp{.data}, and @samp{.bss}. This is the version of | |
53 | GAS that is compiled if neither @code{BFD_ASSEMBLER} nor @code{MANY_SEGMENTS} | |
54 | is defined. This version of GAS is still used for the m68k-aout target, and | |
55 | perhaps others. | |
56 | ||
57 | This version of GAS should not be used for any new development. | |
58 | ||
59 | There is still code that is specific to this version of GAS, notably in | |
60 | @file{write.c}. There is no way for this code to loop through all the | |
61 | sections; it simply looks at global variables like @code{text_frag_root} and | |
62 | @code{data_frag_root}. | |
63 | ||
64 | The type @code{segT} is an enum. | |
65 | ||
66 | @node MANY_SEGMENTS | |
67 | @subsection MANY_SEGMENTS gas version | |
68 | @cindex MANY_SEGMENTS | |
69 | ||
70 | The @code{MANY_SEGMENTS} version of gas is only used for COFF. It uses the BFD | |
71 | library, but it writes out all the data itself using @code{bfd_write}. This | |
72 | version of gas supports up to 40 normal sections. The section names are stored | |
73 | in the @code{seg_name} array. Other information is stored in the | |
74 | @code{segment_info} array. | |
75 | ||
76 | The type @code{segT} is an enum. Code that wants to examine all the sections | |
77 | can use a @code{segT} variable as loop index from @code{SEG_E0} up to but not | |
78 | including @code{SEG_UNKNOWN}. | |
79 | ||
80 | Most of the code specific to this version of GAS is in the file | |
81 | @file{config/obj-coff.c}, in the portion of that file that is compiled when | |
82 | @code{BFD_ASSEMBLER} is not defined. | |
83 | ||
84 | This version of GAS is still used for several COFF targets. | |
85 | ||
86 | @node BFD_ASSEMBLER | |
87 | @subsection BFD_ASSEMBLER gas version | |
88 | @cindex BFD_ASSEMBLER | |
89 | ||
90 | The preferred version of GAS is the @code{BFD_ASSEMBLER} version. In this | |
91 | version of GAS, the output file is a normal BFD, and the BFD routines are used | |
92 | to generate the output. | |
93 | ||
94 | @code{BFD_ASSEMBLER} will automatically be used for certain targets, including | |
95 | those that use the ELF, ECOFF, and SOM object file formats, and also all Alpha, | |
96 | MIPS, PowerPC, and SPARC targets. You can force the use of | |
97 | @code{BFD_ASSEMBLER} for other targets with the configure option | |
98 | @samp{--enable-bfd-assembler}; however, it has not been tested for many | |
99 | targets, and can not be assumed to work. | |
100 | ||
101 | @node Data types | |
102 | @section Data types | |
103 | @cindex internals, data types | |
104 | ||
105 | This section describes some fundamental GAS data types. | |
106 | ||
107 | @menu | |
108 | * Symbols:: The symbolS structure | |
109 | * Expressions:: The expressionS structure | |
110 | * Fixups:: The fixS structure | |
111 | * Frags:: The fragS structure | |
112 | @end menu | |
113 | ||
114 | @node Symbols | |
115 | @subsection Symbols | |
116 | @cindex internals, symbols | |
117 | @cindex symbols, internal | |
118 | @cindex symbolS structure | |
119 | ||
b4013713 ILT |
120 | The definition for the symbol structure, @code{symbolS}, is located in |
121 | @file{struc-symbol.h}. | |
122 | ||
123 | In general, the fields of this structure may not be referred to directly. | |
124 | Instead, you must use one of the accessor functions defined in @file{symbol.h}. | |
125 | These accessor functions should work for any GAS version. | |
126 | ||
127 | Symbol structures contain the following fields: | |
252b5132 RH |
128 | |
129 | @table @code | |
130 | @item sy_value | |
131 | This is an @code{expressionS} that describes the value of the symbol. It might | |
132 | refer to one or more other symbols; if so, its true value may not be known | |
133 | until @code{resolve_symbol_value} is called in @code{write_object_file}. | |
134 | ||
135 | The expression is often simply a constant. Before @code{resolve_symbol_value} | |
136 | is called, the value is the offset from the frag (@pxref{Frags}). Afterward, | |
137 | the frag address has been added in. | |
138 | ||
139 | @item sy_resolved | |
140 | This field is non-zero if the symbol's value has been completely resolved. It | |
141 | is used during the final pass over the symbol table. | |
142 | ||
143 | @item sy_resolving | |
144 | This field is used to detect loops while resolving the symbol's value. | |
145 | ||
146 | @item sy_used_in_reloc | |
147 | This field is non-zero if the symbol is used by a relocation entry. If a local | |
148 | symbol is used in a relocation entry, it must be possible to redirect those | |
149 | relocations to other symbols, or this symbol cannot be removed from the final | |
150 | symbol list. | |
151 | ||
152 | @item sy_next | |
153 | @itemx sy_previous | |
154 | These pointers to other @code{symbolS} structures describe a singly or doubly | |
155 | linked list. (If @code{SYMBOLS_NEED_BACKPOINTERS} is not defined, the | |
156 | @code{sy_previous} field will be omitted; @code{SYMBOLS_NEED_BACKPOINTERS} is | |
157 | always defined if @code{BFD_ASSEMBLER}.) These fields should be accessed with | |
158 | the @code{symbol_next} and @code{symbol_previous} macros. | |
159 | ||
160 | @item sy_frag | |
161 | This points to the frag (@pxref{Frags}) that this symbol is attached to. | |
162 | ||
163 | @item sy_used | |
164 | Whether the symbol is used as an operand or in an expression. Note: Not all of | |
165 | the backends keep this information accurate; backends which use this bit are | |
166 | responsible for setting it when a symbol is used in backend routines. | |
167 | ||
168 | @item sy_mri_common | |
169 | Whether the symbol is an MRI common symbol created by the @code{COMMON} | |
170 | pseudo-op when assembling in MRI mode. | |
171 | ||
172 | @item bsym | |
173 | If @code{BFD_ASSEMBLER} is defined, this points to the BFD @code{asymbol} that | |
174 | will be used in writing the object file. | |
175 | ||
176 | @item sy_name_offset | |
177 | (Only used if @code{BFD_ASSEMBLER} is not defined.) This is the position of | |
178 | the symbol's name in the string table of the object file. On some formats, | |
179 | this will start at position 4, with position 0 reserved for unnamed symbols. | |
180 | This field is not used until @code{write_object_file} is called. | |
181 | ||
182 | @item sy_symbol | |
183 | (Only used if @code{BFD_ASSEMBLER} is not defined.) This is the | |
184 | format-specific symbol structure, as it would be written into the object file. | |
185 | ||
186 | @item sy_number | |
187 | (Only used if @code{BFD_ASSEMBLER} is not defined.) This is a 24-bit symbol | |
188 | number, for use in constructing relocation table entries. | |
189 | ||
190 | @item sy_obj | |
191 | This format-specific data is of type @code{OBJ_SYMFIELD_TYPE}. If no macro by | |
192 | that name is defined in @file{obj-format.h}, this field is not defined. | |
193 | ||
194 | @item sy_tc | |
195 | This processor-specific data is of type @code{TC_SYMFIELD_TYPE}. If no macro | |
196 | by that name is defined in @file{targ-cpu.h}, this field is not defined. | |
197 | ||
252b5132 RH |
198 | @end table |
199 | ||
b4013713 ILT |
200 | Here is a description of the accessor functions. These should be used rather |
201 | than referring to the fields of @code{symbolS} directly. | |
252b5132 RH |
202 | |
203 | @table @code | |
204 | @item S_SET_VALUE | |
205 | @cindex S_SET_VALUE | |
206 | Set the symbol's value. | |
207 | ||
208 | @item S_GET_VALUE | |
209 | @cindex S_GET_VALUE | |
210 | Get the symbol's value. This will cause @code{resolve_symbol_value} to be | |
211 | called if necessary, so @code{S_GET_VALUE} should only be called when it is | |
212 | safe to resolve symbols (i.e., after the entire input file has been read and | |
213 | all symbols have been defined). | |
214 | ||
215 | @item S_SET_SEGMENT | |
216 | @cindex S_SET_SEGMENT | |
217 | Set the section of the symbol. | |
218 | ||
219 | @item S_GET_SEGMENT | |
220 | @cindex S_GET_SEGMENT | |
221 | Get the symbol's section. | |
222 | ||
223 | @item S_GET_NAME | |
224 | @cindex S_GET_NAME | |
225 | Get the name of the symbol. | |
226 | ||
227 | @item S_SET_NAME | |
228 | @cindex S_SET_NAME | |
229 | Set the name of the symbol. | |
230 | ||
231 | @item S_IS_EXTERNAL | |
232 | @cindex S_IS_EXTERNAL | |
233 | Return non-zero if the symbol is externally visible. | |
234 | ||
235 | @item S_IS_EXTERN | |
236 | @cindex S_IS_EXTERN | |
237 | A synonym for @code{S_IS_EXTERNAL}. Don't use it. | |
238 | ||
239 | @item S_IS_WEAK | |
240 | @cindex S_IS_WEAK | |
241 | Return non-zero if the symbol is weak. | |
242 | ||
243 | @item S_IS_COMMON | |
244 | @cindex S_IS_COMMON | |
245 | Return non-zero if this is a common symbol. Common symbols are sometimes | |
246 | represented as undefined symbols with a value, in which case this function will | |
247 | not be reliable. | |
248 | ||
249 | @item S_IS_DEFINED | |
250 | @cindex S_IS_DEFINED | |
251 | Return non-zero if this symbol is defined. This function is not reliable when | |
252 | called on a common symbol. | |
253 | ||
254 | @item S_IS_DEBUG | |
255 | @cindex S_IS_DEBUG | |
256 | Return non-zero if this is a debugging symbol. | |
257 | ||
258 | @item S_IS_LOCAL | |
259 | @cindex S_IS_LOCAL | |
260 | Return non-zero if this is a local assembler symbol which should not be | |
261 | included in the final symbol table. Note that this is not the opposite of | |
262 | @code{S_IS_EXTERNAL}. The @samp{-L} assembler option affects the return value | |
263 | of this function. | |
264 | ||
265 | @item S_SET_EXTERNAL | |
266 | @cindex S_SET_EXTERNAL | |
267 | Mark the symbol as externally visible. | |
268 | ||
269 | @item S_CLEAR_EXTERNAL | |
270 | @cindex S_CLEAR_EXTERNAL | |
271 | Mark the symbol as not externally visible. | |
272 | ||
273 | @item S_SET_WEAK | |
274 | @cindex S_SET_WEAK | |
275 | Mark the symbol as weak. | |
276 | ||
277 | @item S_GET_TYPE | |
278 | @item S_GET_DESC | |
279 | @item S_GET_OTHER | |
280 | @cindex S_GET_TYPE | |
281 | @cindex S_GET_DESC | |
282 | @cindex S_GET_OTHER | |
283 | Get the @code{type}, @code{desc}, and @code{other} fields of the symbol. These | |
284 | are only defined for object file formats for which they make sense (primarily | |
285 | a.out). | |
286 | ||
287 | @item S_SET_TYPE | |
288 | @item S_SET_DESC | |
289 | @item S_SET_OTHER | |
290 | @cindex S_SET_TYPE | |
291 | @cindex S_SET_DESC | |
292 | @cindex S_SET_OTHER | |
293 | Set the @code{type}, @code{desc}, and @code{other} fields of the symbol. These | |
294 | are only defined for object file formats for which they make sense (primarily | |
295 | a.out). | |
296 | ||
297 | @item S_GET_SIZE | |
298 | @cindex S_GET_SIZE | |
299 | Get the size of a symbol. This is only defined for object file formats for | |
300 | which it makes sense (primarily ELF). | |
301 | ||
302 | @item S_SET_SIZE | |
303 | @cindex S_SET_SIZE | |
304 | Set the size of a symbol. This is only defined for object file formats for | |
305 | which it makes sense (primarily ELF). | |
b4013713 ILT |
306 | |
307 | @item symbol_get_value_expression | |
308 | @cindex symbol_get_value_expression | |
309 | Get a pointer to an @code{expressionS} structure which represents the value of | |
310 | the symbol as an expression. | |
311 | ||
312 | @item symbol_set_value_expression | |
313 | @cindex symbol_set_value_expression | |
314 | Set the value of a symbol to an expression. | |
315 | ||
316 | @item symbol_set_frag | |
317 | @cindex symbol_set_frag | |
318 | Set the frag where a symbol is defined. | |
319 | ||
320 | @item symbol_get_frag | |
321 | @cindex symbol_get_frag | |
322 | Get the frag where a symbol is defined. | |
323 | ||
324 | @item symbol_mark_used | |
325 | @cindex symbol_mark_used | |
326 | Mark a symbol as having been used in an expression. | |
327 | ||
328 | @item symbol_clear_used | |
329 | @cindex symbol_clear_used | |
330 | Clear the mark indicating that a symbol was used in an expression. | |
331 | ||
332 | @item symbol_used_p | |
333 | @cindex symbol_used_p | |
334 | Return whether a symbol was used in an expression. | |
335 | ||
336 | @item symbol_mark_used_in_reloc | |
337 | @cindex symbol_mark_used_in_reloc | |
338 | Mark a symbol as having been used by a relocation. | |
339 | ||
340 | @item symbol_clear_used_in_reloc | |
341 | @cindex symbol_clear_used_in_reloc | |
342 | Clear the mark indicating that a symbol was used in a relocation. | |
343 | ||
344 | @item symbol_used_in_reloc_p | |
345 | @cindex symbol_used_in_reloc_p | |
346 | Return whether a symbol was used in a relocation. | |
347 | ||
348 | @item symbol_mark_mri_common | |
349 | @cindex symbol_mark_mri_common | |
350 | Mark a symbol as an MRI common symbol. | |
351 | ||
352 | @item symbol_clear_mri_common | |
353 | @cindex symbol_clear_mri_common | |
354 | Clear the mark indicating that a symbol is an MRI common symbol. | |
355 | ||
356 | @item symbol_mri_common_p | |
357 | @cindex symbol_mri_common_p | |
358 | Return whether a symbol is an MRI common symbol. | |
359 | ||
360 | @item symbol_mark_written | |
361 | @cindex symbol_mark_written | |
362 | Mark a symbol as having been written. | |
363 | ||
364 | @item symbol_clear_written | |
365 | @cindex symbol_clear_written | |
366 | Clear the mark indicating that a symbol was written. | |
367 | ||
368 | @item symbol_written_p | |
369 | @cindex symbol_written_p | |
370 | Return whether a symbol was written. | |
371 | ||
372 | @item symbol_mark_resolved | |
373 | @cindex symbol_mark_resolved | |
374 | Mark a symbol as having been resolved. | |
375 | ||
376 | @item symbol_resolved_p | |
377 | @cindex symbol_resolved_p | |
378 | Return whether a symbol has been resolved. | |
379 | ||
380 | @item symbol_section_p | |
381 | @cindex symbol_section_p | |
382 | Return whether a symbol is a section symbol. | |
383 | ||
384 | @item symbol_equated_p | |
385 | @cindex symbol_equated_p | |
386 | Return whether a symbol is equated to another symbol. | |
387 | ||
388 | @item symbol_constant_p | |
389 | @cindex symbol_constant_p | |
390 | Return whether a symbol has a constant value, including being an offset within | |
391 | some frag. | |
392 | ||
393 | @item symbol_get_bfdsym | |
394 | @cindex symbol_get_bfdsym | |
395 | Return the BFD symbol associated with a symbol. | |
396 | ||
397 | @item symbol_set_bfdsym | |
398 | @cindex symbol_set_bfdsym | |
399 | Set the BFD symbol associated with a symbol. | |
400 | ||
401 | @item symbol_get_obj | |
402 | @cindex symbol_get_obj | |
403 | Return a pointer to the @code{OBJ_SYMFIELD_TYPE} field of a symbol. | |
404 | ||
405 | @item symbol_set_obj | |
406 | @cindex symbol_set_obj | |
407 | Set the @code{OBJ_SYMFIELD_TYPE} field of a symbol. | |
408 | ||
409 | @item symbol_get_tc | |
410 | @cindex symbol_get_tc | |
411 | Return a pointer to the @code{TC_SYMFIELD_TYPE} field of a symbol. | |
412 | ||
413 | @item symbol_set_tc | |
414 | @cindex symbol_set_tc | |
415 | Set the @code{TC_SYMFIELD_TYPE} field of a symbol. | |
416 | ||
252b5132 RH |
417 | @end table |
418 | ||
b4013713 ILT |
419 | When @code{BFD_ASSEMBLER} is defined, GAS attempts to store local |
420 | symbols--symbols which will not be written to the output file--using a | |
421 | different structure, @code{struct local_symbol}. This structure can only | |
422 | represent symbols whose value is an offset within a frag. | |
423 | ||
424 | Code outside of the symbol handler will always deal with @code{symbolS} | |
425 | structures and use the accessor functions. The accessor functions correctly | |
426 | deal with local symbols. @code{struct local_symbol} is much smaller than | |
427 | @code{symbolS} (which also automatically creates a bfd @code{asymbol} | |
428 | structure), so this saves space when assembling large files. | |
429 | ||
430 | The first field of @code{symbolS} is @code{bsym}, the pointer to the BFD | |
431 | symbol. The first field of @code{struct local_symbol} is a pointer which is | |
432 | always set to NULL. This is how the symbol accessor functions can distinguish | |
433 | local symbols from ordinary symbols. The symbol accessor functions | |
434 | automatically convert a local symbol into an ordinary symbol when necessary. | |
435 | ||
252b5132 RH |
436 | @node Expressions |
437 | @subsection Expressions | |
438 | @cindex internals, expressions | |
439 | @cindex expressions, internal | |
440 | @cindex expressionS structure | |
441 | ||
442 | Expressions are stored in an @code{expressionS} structure. The structure is | |
443 | defined in @file{expr.h}. | |
444 | ||
445 | @cindex expression | |
446 | The macro @code{expression} will create an @code{expressionS} structure based | |
447 | on the text found at the global variable @code{input_line_pointer}. | |
448 | ||
449 | @cindex make_expr_symbol | |
450 | @cindex expr_symbol_where | |
451 | A single @code{expressionS} structure can represent a single operation. | |
452 | Complex expressions are formed by creating @dfn{expression symbols} and | |
453 | combining them in @code{expressionS} structures. An expression symbol is | |
454 | created by calling @code{make_expr_symbol}. An expression symbol should | |
455 | naturally never appear in a symbol table, and the implementation of | |
456 | @code{S_IS_LOCAL} (@pxref{Symbols}) reflects that. The function | |
457 | @code{expr_symbol_where} returns non-zero if a symbol is an expression symbol, | |
458 | and also returns the file and line for the expression which caused it to be | |
459 | created. | |
460 | ||
461 | The @code{expressionS} structure has two symbol fields, a number field, an | |
462 | operator field, and a field indicating whether the number is unsigned. | |
463 | ||
464 | The operator field is of type @code{operatorT}, and describes how to interpret | |
465 | the other fields; see the definition in @file{expr.h} for the possibilities. | |
466 | ||
467 | An @code{operatorT} value of @code{O_big} indicates either a floating point | |
468 | number, stored in the global variable @code{generic_floating_point_number}, or | |
623aa224 | 469 | an integer too large to store in an @code{offsetT} type, stored in the global |
252b5132 RH |
470 | array @code{generic_bignum}. This rather inflexible approach makes it |
471 | impossible to use floating point numbers or large expressions in complex | |
472 | expressions. | |
473 | ||
474 | @node Fixups | |
475 | @subsection Fixups | |
476 | @cindex internals, fixups | |
477 | @cindex fixups | |
478 | @cindex fixS structure | |
479 | ||
480 | A @dfn{fixup} is basically anything which can not be resolved in the first | |
481 | pass. Sometimes a fixup can be resolved by the end of the assembly; if not, | |
482 | the fixup becomes a relocation entry in the object file. | |
483 | ||
484 | @cindex fix_new | |
485 | @cindex fix_new_exp | |
486 | A fixup is created by a call to @code{fix_new} or @code{fix_new_exp}. Both | |
487 | take a frag (@pxref{Frags}), a position within the frag, a size, an indication | |
488 | of whether the fixup is PC relative, and a type. In a @code{BFD_ASSEMBLER} | |
489 | GAS, the type is nominally a @code{bfd_reloc_code_real_type}, but several | |
490 | targets use other type codes to represent fixups that can not be described as | |
491 | relocations. | |
492 | ||
493 | The @code{fixS} structure has a number of fields, several of which are obsolete | |
494 | or are only used by a particular target. The important fields are: | |
495 | ||
496 | @table @code | |
497 | @item fx_frag | |
498 | The frag (@pxref{Frags}) this fixup is in. | |
499 | ||
500 | @item fx_where | |
501 | The location within the frag where the fixup occurs. | |
502 | ||
503 | @item fx_addsy | |
504 | The symbol this fixup is against. Typically, the value of this symbol is added | |
505 | into the object contents. This may be NULL. | |
506 | ||
507 | @item fx_subsy | |
508 | The value of this symbol is subtracted from the object contents. This is | |
509 | normally NULL. | |
510 | ||
511 | @item fx_offset | |
512 | A number which is added into the fixup. | |
513 | ||
514 | @item fx_addnumber | |
515 | Some CPU backends use this field to convey information between | |
516 | @code{md_apply_fix} and @code{tc_gen_reloc}. The machine independent code does | |
517 | not use it. | |
518 | ||
519 | @item fx_next | |
520 | The next fixup in the section. | |
521 | ||
522 | @item fx_r_type | |
523 | The type of the fixup. This field is only defined if @code{BFD_ASSEMBLER}, or | |
524 | if the target defines @code{NEED_FX_R_TYPE}. | |
525 | ||
526 | @item fx_size | |
527 | The size of the fixup. This is mostly used for error checking. | |
528 | ||
529 | @item fx_pcrel | |
530 | Whether the fixup is PC relative. | |
531 | ||
532 | @item fx_done | |
533 | Non-zero if the fixup has been applied, and no relocation entry needs to be | |
534 | generated. | |
535 | ||
536 | @item fx_file | |
537 | @itemx fx_line | |
538 | The file and line where the fixup was created. | |
539 | ||
540 | @item tc_fix_data | |
541 | This has the type @code{TC_FIX_TYPE}, and is only defined if the target defines | |
542 | that macro. | |
543 | @end table | |
544 | ||
545 | @node Frags | |
546 | @subsection Frags | |
547 | @cindex internals, frags | |
548 | @cindex frags | |
549 | @cindex fragS structure. | |
550 | ||
551 | The @code{fragS} structure is defined in @file{as.h}. Each frag represents a | |
552 | portion of the final object file. As GAS reads the source file, it creates | |
553 | frags to hold the data that it reads. At the end of the assembly the frags and | |
554 | fixups are processed to produce the final contents. | |
555 | ||
556 | @table @code | |
557 | @item fr_address | |
558 | The address of the frag. This is not set until the assembler rescans the list | |
559 | of all frags after the entire input file is parsed. The function | |
560 | @code{relax_segment} fills in this field. | |
561 | ||
562 | @item fr_next | |
563 | Pointer to the next frag in this (sub)section. | |
564 | ||
565 | @item fr_fix | |
566 | Fixed number of characters we know we're going to emit to the output file. May | |
567 | be zero. | |
568 | ||
569 | @item fr_var | |
570 | Variable number of characters we may output, after the initial @code{fr_fix} | |
571 | characters. May be zero. | |
572 | ||
573 | @item fr_offset | |
574 | The interpretation of this field is controlled by @code{fr_type}. Generally, | |
575 | if @code{fr_var} is non-zero, this is a repeat count: the @code{fr_var} | |
576 | characters are output @code{fr_offset} times. | |
577 | ||
578 | @item line | |
579 | Holds line number info when an assembler listing was requested. | |
580 | ||
581 | @item fr_type | |
582 | Relaxation state. This field indicates the interpretation of @code{fr_offset}, | |
583 | @code{fr_symbol} and the variable-length tail of the frag, as well as the | |
584 | treatment it gets in various phases of processing. It does not affect the | |
585 | initial @code{fr_fix} characters; they are always supposed to be output | |
586 | verbatim (fixups aside). See below for specific values this field can have. | |
587 | ||
588 | @item fr_subtype | |
589 | Relaxation substate. If the macro @code{md_relax_frag} isn't defined, this is | |
590 | assumed to be an index into @code{TC_GENERIC_RELAX_TABLE} for the generic | |
591 | relaxation code to process (@pxref{Relaxation}). If @code{md_relax_frag} is | |
592 | defined, this field is available for any use by the CPU-specific code. | |
593 | ||
594 | @item fr_symbol | |
595 | This normally indicates the symbol to use when relaxing the frag according to | |
596 | @code{fr_type}. | |
597 | ||
598 | @item fr_opcode | |
599 | Points to the lowest-addressed byte of the opcode, for use in relaxation. | |
600 | ||
601 | @item tc_frag_data | |
602 | Target specific fragment data of type TC_FRAG_TYPE. | |
603 | Only present if @code{TC_FRAG_TYPE} is defined. | |
604 | ||
605 | @item fr_file | |
606 | @itemx fr_line | |
607 | The file and line where this frag was last modified. | |
608 | ||
609 | @item fr_literal | |
610 | Declared as a one-character array, this last field grows arbitrarily large to | |
611 | hold the actual contents of the frag. | |
612 | @end table | |
613 | ||
614 | These are the possible relaxation states, provided in the enumeration type | |
615 | @code{relax_stateT}, and the interpretations they represent for the other | |
616 | fields: | |
617 | ||
618 | @table @code | |
619 | @item rs_align | |
620 | @itemx rs_align_code | |
621 | The start of the following frag should be aligned on some boundary. In this | |
622 | frag, @code{fr_offset} is the logarithm (base 2) of the alignment in bytes. | |
623 | (For example, if alignment on an 8-byte boundary were desired, @code{fr_offset} | |
624 | would have a value of 3.) The variable characters indicate the fill pattern to | |
625 | be used. The @code{fr_subtype} field holds the maximum number of bytes to skip | |
626 | when doing this alignment. If more bytes are needed, the alignment is not | |
627 | done. An @code{fr_subtype} value of 0 means no maximum, which is the normal | |
628 | case. Target backends can use @code{rs_align_code} to handle certain types of | |
629 | alignment differently. | |
630 | ||
631 | @item rs_broken_word | |
632 | This indicates that ``broken word'' processing should be done (@pxref{Broken | |
633 | words}). If broken word processing is not necessary on the target machine, | |
634 | this enumerator value will not be defined. | |
635 | ||
636 | @item rs_cfa | |
637 | This state is used to implement exception frame optimizations. The | |
638 | @code{fr_symbol} is an expression symbol for the subtraction which may be | |
639 | relaxed. The @code{fr_opcode} field holds the frag for the preceding command | |
640 | byte. The @code{fr_offset} field holds the offset within that frag. The | |
641 | @code{fr_subtype} field is used during relaxation to hold the current size of | |
642 | the frag. | |
643 | ||
644 | @item rs_fill | |
645 | The variable characters are to be repeated @code{fr_offset} times. If | |
646 | @code{fr_offset} is 0, this frag has a length of @code{fr_fix}. Most frags | |
647 | have this type. | |
648 | ||
649 | @item rs_leb128 | |
58a77e41 | 650 | This state is used to implement the DWARF ``little endian base 128'' |
252b5132 RH |
651 | variable length number format. The @code{fr_symbol} is always an expression |
652 | symbol, as constant expressions are emitted directly. The @code{fr_offset} | |
653 | field is used during relaxation to hold the previous size of the number so | |
654 | that we can determine if the fragment changed size. | |
655 | ||
656 | @item rs_machine_dependent | |
657 | Displacement relaxation is to be done on this frag. The target is indicated by | |
658 | @code{fr_symbol} and @code{fr_offset}, and @code{fr_subtype} indicates the | |
659 | particular machine-specific addressing mode desired. @xref{Relaxation}. | |
660 | ||
661 | @item rs_org | |
662 | The start of the following frag should be pushed back to some specific offset | |
663 | within the section. (Some assemblers use the value as an absolute address; GAS | |
664 | does not handle final absolute addresses, but rather requires that the linker | |
665 | set them.) The offset is given by @code{fr_symbol} and @code{fr_offset}; one | |
666 | character from the variable-length tail is used as the fill character. | |
667 | @end table | |
668 | ||
669 | @cindex frchainS structure | |
670 | A chain of frags is built up for each subsection. The data structure | |
671 | describing a chain is called a @code{frchainS}, and contains the following | |
672 | fields: | |
673 | ||
674 | @table @code | |
675 | @item frch_root | |
676 | Points to the first frag in the chain. May be NULL if there are no frags in | |
677 | this chain. | |
678 | @item frch_last | |
679 | Points to the last frag in the chain, or NULL if there are none. | |
680 | @item frch_next | |
681 | Next in the list of @code{frchainS} structures. | |
682 | @item frch_seg | |
683 | Indicates the section this frag chain belongs to. | |
684 | @item frch_subseg | |
685 | Subsection (subsegment) number of this frag chain. | |
686 | @item fix_root, fix_tail | |
687 | (Defined only if @code{BFD_ASSEMBLER} is defined). Point to first and last | |
688 | @code{fixS} structures associated with this subsection. | |
689 | @item frch_obstack | |
690 | Not currently used. Intended to be used for frag allocation for this | |
691 | subsection. This should reduce frag generation caused by switching sections. | |
692 | @item frch_frag_now | |
693 | The current frag for this subsegment. | |
694 | @end table | |
695 | ||
696 | A @code{frchainS} corresponds to a subsection; each section has a list of | |
697 | @code{frchainS} records associated with it. In most cases, only one subsection | |
698 | of each section is used, so the list will only be one element long, but any | |
699 | processing of frag chains should be prepared to deal with multiple chains per | |
700 | section. | |
701 | ||
702 | After the input files have been completely processed, and no more frags are to | |
703 | be generated, the frag chains are joined into one per section for further | |
704 | processing. After this point, it is safe to operate on one chain per section. | |
705 | ||
706 | The assembler always has a current frag, named @code{frag_now}. More space is | |
707 | allocated for the current frag using the @code{frag_more} function; this | |
708 | returns a pointer to the amount of requested space. Relaxing is done using | |
709 | variant frags allocated by @code{frag_var} or @code{frag_variant} | |
710 | (@pxref{Relaxation}). | |
711 | ||
712 | @node GAS processing | |
713 | @section What GAS does when it runs | |
714 | @cindex internals, overview | |
715 | ||
716 | This is a quick look at what an assembler run looks like. | |
717 | ||
718 | @itemize @bullet | |
719 | @item | |
720 | The assembler initializes itself by calling various init routines. | |
721 | ||
722 | @item | |
723 | For each source file, the @code{read_a_source_file} function reads in the file | |
724 | and parses it. The global variable @code{input_line_pointer} points to the | |
725 | current text; it is guaranteed to be correct up to the end of the line, but not | |
726 | farther. | |
727 | ||
728 | @item | |
729 | For each line, the assembler passes labels to the @code{colon} function, and | |
730 | isolates the first word. If it looks like a pseudo-op, the word is looked up | |
731 | in the pseudo-op hash table @code{po_hash} and dispatched to a pseudo-op | |
732 | routine. Otherwise, the target dependent @code{md_assemble} routine is called | |
733 | to parse the instruction. | |
734 | ||
735 | @item | |
736 | When pseudo-ops or instructions output data, they add it to a frag, calling | |
737 | @code{frag_more} to get space to store it in. | |
738 | ||
739 | @item | |
740 | Pseudo-ops and instructions can also output fixups created by @code{fix_new} or | |
741 | @code{fix_new_exp}. | |
742 | ||
743 | @item | |
744 | For certain targets, instructions can create variant frags which are used to | |
745 | store relaxation information (@pxref{Relaxation}). | |
746 | ||
747 | @item | |
748 | When the input file is finished, the @code{write_object_file} routine is | |
749 | called. It assigns addresses to all the frags (@code{relax_segment}), resolves | |
750 | all the fixups (@code{fixup_segment}), resolves all the symbol values (using | |
751 | @code{resolve_symbol_value}), and finally writes out the file (in the | |
752 | @code{BFD_ASSEMBLER} case, this is done by simply calling @code{bfd_close}). | |
753 | @end itemize | |
754 | ||
755 | @node Porting GAS | |
756 | @section Porting GAS | |
757 | @cindex porting | |
758 | ||
759 | Each GAS target specifies two main things: the CPU file and the object format | |
760 | file. Two main switches in the @file{configure.in} file handle this. The | |
761 | first switches on CPU type to set the shell variable @code{cpu_type}. The | |
762 | second switches on the entire target to set the shell variable @code{fmt}. | |
763 | ||
764 | The configure script uses the value of @code{cpu_type} to select two files in | |
765 | the @file{config} directory: @file{tc-@var{CPU}.c} and @file{tc-@var{CPU}.h}. | |
766 | The configuration process will create a file named @file{targ-cpu.h} in the | |
767 | build directory which includes @file{tc-@var{CPU}.h}. | |
768 | ||
769 | The configure script also uses the value of @code{fmt} to select two files: | |
770 | @file{obj-@var{fmt}.c} and @file{obj-@var{fmt}.h}. The configuration process | |
771 | will create a file named @file{obj-format.h} in the build directory which | |
772 | includes @file{obj-@var{fmt}.h}. | |
773 | ||
774 | You can also set the emulation in the configure script by setting the @code{em} | |
775 | variable. Normally the default value of @samp{generic} is fine. The | |
776 | configuration process will create a file named @file{targ-env.h} in the build | |
777 | directory which includes @file{te-@var{em}.h}. | |
778 | ||
56385375 L |
779 | There is a special case for COFF. For historical reason, the GNU COFF |
780 | assembler doesn't follow the documented behavior on certain debug symbols for | |
781 | the compatibility with other COFF assemblers. A port can define | |
782 | @code{STRICTCOFF} in the configure script to make the GNU COFF assembler | |
783 | to follow the documented behavior. | |
784 | ||
252b5132 RH |
785 | Porting GAS to a new CPU requires writing the @file{tc-@var{CPU}} files. |
786 | Porting GAS to a new object file format requires writing the | |
787 | @file{obj-@var{fmt}} files. There is sometimes some interaction between these | |
788 | two files, but it is normally minimal. | |
789 | ||
790 | The best approach is, of course, to copy existing files. The documentation | |
791 | below assumes that you are looking at existing files to see usage details. | |
792 | ||
793 | These interfaces have grown over time, and have never been carefully thought | |
794 | out or designed. Nothing about the interfaces described here is cast in stone. | |
795 | It is possible that they will change from one version of the assembler to the | |
796 | next. Also, new macros are added all the time as they are needed. | |
797 | ||
798 | @menu | |
799 | * CPU backend:: Writing a CPU backend | |
800 | * Object format backend:: Writing an object format backend | |
801 | * Emulations:: Writing emulation files | |
802 | @end menu | |
803 | ||
804 | @node CPU backend | |
805 | @subsection Writing a CPU backend | |
806 | @cindex CPU backend | |
807 | @cindex @file{tc-@var{CPU}} | |
808 | ||
809 | The CPU backend files are the heart of the assembler. They are the only parts | |
810 | of the assembler which actually know anything about the instruction set of the | |
811 | processor. | |
812 | ||
813 | You must define a reasonably small list of macros and functions in the CPU | |
814 | backend files. You may define a large number of additional macros in the CPU | |
815 | backend files, not all of which are documented here. You must, of course, | |
816 | define macros in the @file{.h} file, which is included by every assembler | |
817 | source file. You may define the functions as macros in the @file{.h} file, or | |
818 | as functions in the @file{.c} file. | |
819 | ||
820 | @table @code | |
821 | @item TC_@var{CPU} | |
822 | @cindex TC_@var{CPU} | |
823 | By convention, you should define this macro in the @file{.h} file. For | |
824 | example, @file{tc-m68k.h} defines @code{TC_M68K}. You might have to use this | |
825 | if it is necessary to add CPU specific code to the object format file. | |
826 | ||
827 | @item TARGET_FORMAT | |
828 | This macro is the BFD target name to use when creating the output file. This | |
829 | will normally depend upon the @code{OBJ_@var{FMT}} macro. | |
830 | ||
831 | @item TARGET_ARCH | |
832 | This macro is the BFD architecture to pass to @code{bfd_set_arch_mach}. | |
833 | ||
834 | @item TARGET_MACH | |
835 | This macro is the BFD machine number to pass to @code{bfd_set_arch_mach}. If | |
836 | it is not defined, GAS will use 0. | |
837 | ||
838 | @item TARGET_BYTES_BIG_ENDIAN | |
839 | You should define this macro to be non-zero if the target is big endian, and | |
840 | zero if the target is little endian. | |
841 | ||
842 | @item md_shortopts | |
843 | @itemx md_longopts | |
844 | @itemx md_longopts_size | |
845 | @itemx md_parse_option | |
846 | @itemx md_show_usage | |
847 | @cindex md_shortopts | |
848 | @cindex md_longopts | |
849 | @cindex md_longopts_size | |
850 | @cindex md_parse_option | |
851 | @cindex md_show_usage | |
852 | GAS uses these variables and functions during option processing. | |
853 | @code{md_shortopts} is a @code{const char *} which GAS adds to the machine | |
854 | independent string passed to @code{getopt}. @code{md_longopts} is a | |
855 | @code{struct option []} which GAS adds to the machine independent long options | |
856 | passed to @code{getopt}; you may use @code{OPTION_MD_BASE}, defined in | |
857 | @file{as.h}, as the start of a set of long option indices, if necessary. | |
858 | @code{md_longopts_size} is a @code{size_t} holding the size @code{md_longopts}. | |
859 | GAS will call @code{md_parse_option} whenever @code{getopt} returns an | |
860 | unrecognized code, presumably indicating a special code value which appears in | |
861 | @code{md_longopts}. GAS will call @code{md_show_usage} when a usage message is | |
862 | printed; it should print a description of the machine specific options. | |
863 | ||
864 | @item md_begin | |
865 | @cindex md_begin | |
866 | GAS will call this function at the start of the assembly, after the command | |
867 | line arguments have been parsed and all the machine independent initializations | |
868 | have been completed. | |
869 | ||
870 | @item md_cleanup | |
871 | @cindex md_cleanup | |
872 | If you define this macro, GAS will call it at the end of each input file. | |
873 | ||
874 | @item md_assemble | |
875 | @cindex md_assemble | |
876 | GAS will call this function for each input line which does not contain a | |
877 | pseudo-op. The argument is a null terminated string. The function should | |
878 | assemble the string as an instruction with operands. Normally | |
879 | @code{md_assemble} will do this by calling @code{frag_more} and writing out | |
880 | some bytes (@pxref{Frags}). @code{md_assemble} will call @code{fix_new} to | |
881 | create fixups as needed (@pxref{Fixups}). Targets which need to do special | |
882 | purpose relaxation will call @code{frag_var}. | |
883 | ||
884 | @item md_pseudo_table | |
885 | @cindex md_pseudo_table | |
886 | This is a const array of type @code{pseudo_typeS}. It is a mapping from | |
887 | pseudo-op names to functions. You should use this table to implement | |
888 | pseudo-ops which are specific to the CPU. | |
889 | ||
890 | @item tc_conditional_pseudoop | |
891 | @cindex tc_conditional_pseudoop | |
892 | If this macro is defined, GAS will call it with a @code{pseudo_typeS} argument. | |
893 | It should return non-zero if the pseudo-op is a conditional which controls | |
894 | whether code is assembled, such as @samp{.if}. GAS knows about the normal | |
8108ad8e | 895 | conditional pseudo-ops, and you should normally not have to define this macro. |
252b5132 RH |
896 | |
897 | @item comment_chars | |
898 | @cindex comment_chars | |
899 | This is a null terminated @code{const char} array of characters which start a | |
900 | comment. | |
901 | ||
902 | @item tc_comment_chars | |
903 | @cindex tc_comment_chars | |
904 | If this macro is defined, GAS will use it instead of @code{comment_chars}. | |
905 | ||
906 | @item tc_symbol_chars | |
907 | @cindex tc_symbol_chars | |
908 | If this macro is defined, it is a pointer to a null terminated list of | |
909 | characters which may appear in an operand. GAS already assumes that all | |
910 | alphanumberic characters, and @samp{$}, @samp{.}, and @samp{_} may appear in an | |
911 | operand (see @samp{symbol_chars} in @file{app.c}). This macro may be defined | |
912 | to treat additional characters as appearing in an operand. This affects the | |
913 | way in which GAS removes whitespace before passing the string to | |
914 | @samp{md_assemble}. | |
915 | ||
916 | @item line_comment_chars | |
917 | @cindex line_comment_chars | |
918 | This is a null terminated @code{const char} array of characters which start a | |
919 | comment when they appear at the start of a line. | |
920 | ||
921 | @item line_separator_chars | |
922 | @cindex line_separator_chars | |
923 | This is a null terminated @code{const char} array of characters which separate | |
63a0b638 | 924 | lines (null and newline are such characters by default, and need not be |
65fd87bc ILT |
925 | listed in this array). Note that line_separator_chars do not separate lines |
926 | if found in a comment, such as after a character in line_comment_chars or | |
927 | comment_chars. | |
252b5132 RH |
928 | |
929 | @item EXP_CHARS | |
930 | @cindex EXP_CHARS | |
931 | This is a null terminated @code{const char} array of characters which may be | |
932 | used as the exponent character in a floating point number. This is normally | |
933 | @code{"eE"}. | |
934 | ||
935 | @item FLT_CHARS | |
936 | @cindex FLT_CHARS | |
937 | This is a null terminated @code{const char} array of characters which may be | |
938 | used to indicate a floating point constant. A zero followed by one of these | |
939 | characters is assumed to be followed by a floating point number; thus they | |
940 | operate the way that @code{0x} is used to indicate a hexadecimal constant. | |
941 | Usually this includes @samp{r} and @samp{f}. | |
942 | ||
943 | @item LEX_AT | |
944 | @cindex LEX_AT | |
65fd87bc | 945 | You may define this macro to the lexical type of the @kbd{@@} character. The |
252b5132 RH |
946 | default is zero. |
947 | ||
948 | Lexical types are a combination of @code{LEX_NAME} and @code{LEX_BEGIN_NAME}, | |
949 | both defined in @file{read.h}. @code{LEX_NAME} indicates that the character | |
950 | may appear in a name. @code{LEX_BEGIN_NAME} indicates that the character may | |
65fd87bc | 951 | appear at the beginning of a name. |
252b5132 RH |
952 | |
953 | @item LEX_BR | |
954 | @cindex LEX_BR | |
955 | You may define this macro to the lexical type of the brace characters @kbd{@{}, | |
956 | @kbd{@}}, @kbd{[}, and @kbd{]}. The default value is zero. | |
957 | ||
958 | @item LEX_PCT | |
959 | @cindex LEX_PCT | |
960 | You may define this macro to the lexical type of the @kbd{%} character. The | |
961 | default value is zero. | |
962 | ||
963 | @item LEX_QM | |
964 | @cindex LEX_QM | |
965 | You may define this macro to the lexical type of the @kbd{?} character. The | |
966 | default value it zero. | |
967 | ||
968 | @item LEX_DOLLAR | |
969 | @cindex LEX_DOLLAR | |
970 | You may define this macro to the lexical type of the @kbd{$} character. The | |
971 | default value is @code{LEX_NAME | LEX_BEGIN_NAME}. | |
972 | ||
f805106c TW |
973 | @item NUMBERS_WITH_SUFFIX |
974 | @cindex NUMBERS_WITH_SUFFIX | |
975 | When this macro is defined to be non-zero, the parser allows the radix of a | |
58a77e41 | 976 | constant to be indicated with a suffix. Valid suffixes are binary (B), |
f805106c TW |
977 | octal (Q), and hexadecimal (H). Case is not significant. |
978 | ||
252b5132 RH |
979 | @item SINGLE_QUOTE_STRINGS |
980 | @cindex SINGLE_QUOTE_STRINGS | |
981 | If you define this macro, GAS will treat single quotes as string delimiters. | |
982 | Normally only double quotes are accepted as string delimiters. | |
983 | ||
984 | @item NO_STRING_ESCAPES | |
985 | @cindex NO_STRING_ESCAPES | |
986 | If you define this macro, GAS will not permit escape sequences in a string. | |
987 | ||
988 | @item ONLY_STANDARD_ESCAPES | |
989 | @cindex ONLY_STANDARD_ESCAPES | |
990 | If you define this macro, GAS will warn about the use of nonstandard escape | |
991 | sequences in a string. | |
992 | ||
993 | @item md_start_line_hook | |
994 | @cindex md_start_line_hook | |
995 | If you define this macro, GAS will call it at the start of each line. | |
996 | ||
997 | @item LABELS_WITHOUT_COLONS | |
998 | @cindex LABELS_WITHOUT_COLONS | |
999 | If you define this macro, GAS will assume that any text at the start of a line | |
1000 | is a label, even if it does not have a colon. | |
1001 | ||
1002 | @item TC_START_LABEL | |
39bec121 | 1003 | @itemx TC_START_LABEL_WITHOUT_COLON |
252b5132 RH |
1004 | @cindex TC_START_LABEL |
1005 | You may define this macro to control what GAS considers to be a label. The | |
1006 | default definition is to accept any name followed by a colon character. | |
1007 | ||
f28e8eb3 TW |
1008 | @item TC_START_LABEL_WITHOUT_COLON |
1009 | @cindex TC_START_LABEL_WITHOUT_COLON | |
1010 | Same as TC_START_LABEL, but should be used instead of TC_START_LABEL when | |
58a77e41 | 1011 | LABELS_WITHOUT_COLONS is defined. |
f28e8eb3 | 1012 | |
252b5132 RH |
1013 | @item NO_PSEUDO_DOT |
1014 | @cindex NO_PSEUDO_DOT | |
1015 | If you define this macro, GAS will not require pseudo-ops to start with a | |
1016 | @kbd{.} character. | |
1017 | ||
1018 | @item TC_EQUAL_IN_INSN | |
1019 | @cindex TC_EQUAL_IN_INSN | |
1020 | If you define this macro, it should return nonzero if the instruction is | |
65fd87bc ILT |
1021 | permitted to contain an @kbd{=} character. GAS will call it with two |
1022 | arguments, the character before the @kbd{=} character, and the value of | |
1023 | @code{input_line_pointer} at that point. GAS uses this macro to decide if a | |
252b5132 RH |
1024 | @kbd{=} is an assignment or an instruction. |
1025 | ||
1026 | @item TC_EOL_IN_INSN | |
1027 | @cindex TC_EOL_IN_INSN | |
1028 | If you define this macro, it should return nonzero if the current input line | |
1029 | pointer should be treated as the end of a line. | |
1030 | ||
1031 | @item md_parse_name | |
1032 | @cindex md_parse_name | |
1033 | If this macro is defined, GAS will call it for any symbol found in an | |
1034 | expression. You can define this to handle special symbols in a special way. | |
1035 | If a symbol always has a certain value, you should normally enter it in the | |
1036 | symbol table, perhaps using @code{reg_section}. | |
1037 | ||
1038 | @item md_undefined_symbol | |
1039 | @cindex md_undefined_symbol | |
1040 | GAS will call this function when a symbol table lookup fails, before it | |
1041 | creates a new symbol. Typically this would be used to supply symbols whose | |
1042 | name or value changes dynamically, possibly in a context sensitive way. | |
1043 | Predefined symbols with fixed values, such as register names or condition | |
1044 | codes, are typically entered directly into the symbol table when @code{md_begin} | |
65fd87bc | 1045 | is called. One argument is passed, a @code{char *} for the symbol. |
252b5132 RH |
1046 | |
1047 | @item md_operand | |
1048 | @cindex md_operand | |
65fd87bc ILT |
1049 | GAS will call this function with one argument, an @code{expressionS} |
1050 | pointer, for any expression that can not be recognized. When the function | |
1051 | is called, @code{input_line_pointer} will point to the start of the | |
1052 | expression. | |
252b5132 RH |
1053 | |
1054 | @item tc_unrecognized_line | |
1055 | @cindex tc_unrecognized_line | |
1056 | If you define this macro, GAS will call it when it finds a line that it can not | |
1057 | parse. | |
1058 | ||
1059 | @item md_do_align | |
1060 | @cindex md_do_align | |
1061 | You may define this macro to handle an alignment directive. GAS will call it | |
1062 | when the directive is seen in the input file. For example, the i386 backend | |
1063 | uses this to generate efficient nop instructions of varying lengths, depending | |
1064 | upon the number of bytes that the alignment will skip. | |
1065 | ||
1066 | @item HANDLE_ALIGN | |
1067 | @cindex HANDLE_ALIGN | |
1068 | You may define this macro to do special handling for an alignment directive. | |
1069 | GAS will call it at the end of the assembly. | |
1070 | ||
8684e216 HPN |
1071 | @item TC_IMPLICIT_LCOMM_ALIGNMENT (@var{size}, @var{p2var}) |
1072 | @cindex TC_IMPLICIT_LCOMM_ALIGNMENT | |
1073 | An @code{.lcomm} directive with no explicit alignment parameter will use this | |
1074 | macro to set @var{p2var} to the alignment that a request for @var{size} bytes | |
1075 | will have. The alignment is expressed as a power of two. If no alignment | |
1076 | should take place, the macro definition should do nothing. Some targets define | |
1077 | a @code{.bss} directive that is also affected by this macro. The default | |
1078 | definition will set @var{p2var} to the truncated power of two of sizes up to | |
1079 | eight bytes. | |
1080 | ||
252b5132 RH |
1081 | @item md_flush_pending_output |
1082 | @cindex md_flush_pending_output | |
1083 | If you define this macro, GAS will call it each time it skips any space because of a | |
1084 | space filling or alignment or data allocation pseudo-op. | |
1085 | ||
1086 | @item TC_PARSE_CONS_EXPRESSION | |
1087 | @cindex TC_PARSE_CONS_EXPRESSION | |
1088 | You may define this macro to parse an expression used in a data allocation | |
1089 | pseudo-op such as @code{.word}. You can use this to recognize relocation | |
1090 | directives that may appear in such directives. | |
1091 | ||
1092 | @item BITFIELD_CONS_EXPRESSION | |
1093 | @cindex BITFIELD_CONS_EXPRESSION | |
1094 | If you define this macro, GAS will recognize bitfield instructions in data | |
1095 | allocation pseudo-ops, as used on the i960. | |
1096 | ||
1097 | @item REPEAT_CONS_EXPRESSION | |
1098 | @cindex REPEAT_CONS_EXPRESSION | |
1099 | If you define this macro, GAS will recognize repeat counts in data allocation | |
1100 | pseudo-ops, as used on the MIPS. | |
1101 | ||
1102 | @item md_cons_align | |
1103 | @cindex md_cons_align | |
1104 | You may define this macro to do any special alignment before a data allocation | |
1105 | pseudo-op. | |
1106 | ||
1107 | @item TC_CONS_FIX_NEW | |
1108 | @cindex TC_CONS_FIX_NEW | |
1109 | You may define this macro to generate a fixup for a data allocation pseudo-op. | |
1110 | ||
1111 | @item TC_INIT_FIX_DATA (@var{fixp}) | |
1112 | @cindex TC_INIT_FIX_DATA | |
1113 | A C statement to initialize the target specific fields of fixup @var{fixp}. | |
1114 | These fields are defined with the @code{TC_FIX_TYPE} macro. | |
1115 | ||
1116 | @item TC_FIX_DATA_PRINT (@var{stream}, @var{fixp}) | |
1117 | @cindex TC_FIX_DATA_PRINT | |
1118 | A C statement to output target specific debugging information for | |
1119 | fixup @var{fixp} to @var{stream}. This macro is called by @code{print_fixup}. | |
1120 | ||
1121 | @item TC_FRAG_INIT (@var{fragp}) | |
1122 | @cindex TC_FRAG_INIT | |
1123 | A C statement to initialize the target specific fields of frag @var{fragp}. | |
1124 | These fields are defined with the @code{TC_FRAG_TYPE} macro. | |
1125 | ||
1126 | @item md_number_to_chars | |
1127 | @cindex md_number_to_chars | |
1128 | This should just call either @code{number_to_chars_bigendian} or | |
1129 | @code{number_to_chars_littleendian}, whichever is appropriate. On targets like | |
1130 | the MIPS which support options to change the endianness, which function to call | |
1131 | is a runtime decision. On other targets, @code{md_number_to_chars} can be a | |
1132 | simple macro. | |
1133 | ||
1134 | @item md_reloc_size | |
1135 | @cindex md_reloc_size | |
1136 | This variable is only used in the original version of gas (not | |
1137 | @code{BFD_ASSEMBLER} and not @code{MANY_SEGMENTS}). It holds the size of a | |
1138 | relocation entry. | |
1139 | ||
1140 | @item WORKING_DOT_WORD | |
1141 | @itemx md_short_jump_size | |
1142 | @itemx md_long_jump_size | |
1143 | @itemx md_create_short_jump | |
1144 | @itemx md_create_long_jump | |
e30e5a6a | 1145 | @itemx TC_CHECK_ADJUSTED_BROKEN_DOT_WORD |
252b5132 RH |
1146 | @cindex WORKING_DOT_WORD |
1147 | @cindex md_short_jump_size | |
1148 | @cindex md_long_jump_size | |
1149 | @cindex md_create_short_jump | |
1150 | @cindex md_create_long_jump | |
e30e5a6a | 1151 | @cindex TC_CHECK_ADJUSTED_BROKEN_DOT_WORD |
252b5132 RH |
1152 | If @code{WORKING_DOT_WORD} is defined, GAS will not do broken word processing |
1153 | (@pxref{Broken words}). Otherwise, you should set @code{md_short_jump_size} to | |
65fd87bc ILT |
1154 | the size of a short jump (a jump that is just long enough to jump around a |
1155 | number of long jumps) and @code{md_long_jump_size} to the size of a long jump | |
1156 | (a jump that can go anywhere in the function). You should define | |
1157 | @code{md_create_short_jump} to create a short jump around a number of long | |
1158 | jumps, and define @code{md_create_long_jump} to create a long jump. | |
e30e5a6a HPN |
1159 | If defined, the macro TC_CHECK_ADJUSTED_BROKEN_DOT_WORD will be called for each |
1160 | adjusted word just before the word is output. The macro takes two arguments, | |
1161 | an @code{addressT} with the adjusted word and a pointer to the current | |
1162 | @code{struct broken_word}. | |
252b5132 RH |
1163 | |
1164 | @item md_estimate_size_before_relax | |
1165 | @cindex md_estimate_size_before_relax | |
1166 | This function returns an estimate of the size of a @code{rs_machine_dependent} | |
1167 | frag before any relaxing is done. It may also create any necessary | |
1168 | relocations. | |
1169 | ||
1170 | @item md_relax_frag | |
1171 | @cindex md_relax_frag | |
c842b53a ILT |
1172 | This macro may be defined to relax a frag. GAS will call this with the |
1173 | segment, the frag, and the change in size of all previous frags; | |
1174 | @code{md_relax_frag} should return the change in size of the frag. | |
1175 | @xref{Relaxation}. | |
252b5132 RH |
1176 | |
1177 | @item TC_GENERIC_RELAX_TABLE | |
1178 | @cindex TC_GENERIC_RELAX_TABLE | |
1179 | If you do not define @code{md_relax_frag}, you may define | |
1180 | @code{TC_GENERIC_RELAX_TABLE} as a table of @code{relax_typeS} structures. The | |
1181 | machine independent code knows how to use such a table to relax PC relative | |
1182 | references. See @file{tc-m68k.c} for an example. @xref{Relaxation}. | |
1183 | ||
1184 | @item md_prepare_relax_scan | |
1185 | @cindex md_prepare_relax_scan | |
1186 | If defined, it is a C statement that is invoked prior to scanning | |
1187 | the relax table. | |
1188 | ||
1189 | @item LINKER_RELAXING_SHRINKS_ONLY | |
1190 | @cindex LINKER_RELAXING_SHRINKS_ONLY | |
1191 | If you define this macro, and the global variable @samp{linkrelax} is set | |
1192 | (because of a command line option, or unconditionally in @code{md_begin}), a | |
1193 | @samp{.align} directive will cause extra space to be allocated. The linker can | |
1194 | then discard this space when relaxing the section. | |
1195 | ||
8108ad8e | 1196 | @item TC_LINKRELAX_FIXUP (@var{segT}) |
58a77e41 EC |
1197 | @cindex TC_LINKRELAX_FIXUP |
1198 | If defined, this macro allows control over whether fixups for a | |
1199 | given section will be processed when the @var{linkrelax} variable is | |
1200 | set. The macro is given the N_TYPE bits for the section in its | |
1201 | @var{segT} argument. If the macro evaluates to a non-zero value | |
1202 | then the fixups will be converted into relocs, otherwise they will | |
1203 | be passed to @var{md_apply_fix3} as normal. | |
1204 | ||
252b5132 RH |
1205 | @item md_convert_frag |
1206 | @cindex md_convert_frag | |
1207 | GAS will call this for each rs_machine_dependent fragment. | |
1208 | The instruction is completed using the data from the relaxation pass. | |
1209 | It may also create any necessary relocations. | |
1210 | @xref{Relaxation}. | |
1211 | ||
1212 | @item md_apply_fix | |
1213 | @cindex md_apply_fix | |
1214 | GAS will call this for each fixup. It should store the correct value in the | |
fa67f437 AM |
1215 | object file. @code{fixup_segment} performs a generic overflow check on the |
1216 | @code{valueT *val} argument after @code{md_apply_fix} returns. If the overflow | |
1217 | check is relevant for the target machine, then @code{md_apply_fix} should | |
1218 | modify @code{valueT *val}, typically to the value stored in the object file. | |
252b5132 RH |
1219 | |
1220 | @item TC_HANDLES_FX_DONE | |
1221 | @cindex TC_HANDLES_FX_DONE | |
1222 | If this macro is defined, it means that @code{md_apply_fix} correctly sets the | |
1223 | @code{fx_done} field in the fixup. | |
1224 | ||
1225 | @item tc_gen_reloc | |
1226 | @cindex tc_gen_reloc | |
1227 | A @code{BFD_ASSEMBLER} GAS will call this to generate a reloc. GAS will pass | |
1228 | the resulting reloc to @code{bfd_install_relocation}. This currently works | |
1229 | poorly, as @code{bfd_install_relocation} often does the wrong thing, and | |
1230 | instances of @code{tc_gen_reloc} have been written to work around the problems, | |
1231 | which in turns makes it difficult to fix @code{bfd_install_relocation}. | |
1232 | ||
1233 | @item RELOC_EXPANSION_POSSIBLE | |
1234 | @cindex RELOC_EXPANSION_POSSIBLE | |
1235 | If you define this macro, it means that @code{tc_gen_reloc} may return multiple | |
1236 | relocation entries for a single fixup. In this case, the return value of | |
1237 | @code{tc_gen_reloc} is a pointer to a null terminated array. | |
1238 | ||
1239 | @item MAX_RELOC_EXPANSION | |
1240 | @cindex MAX_RELOC_EXPANSION | |
1241 | You must define this if @code{RELOC_EXPANSION_POSSIBLE} is defined; it | |
1242 | indicates the largest number of relocs which @code{tc_gen_reloc} may return for | |
1243 | a single fixup. | |
1244 | ||
1245 | @item tc_fix_adjustable | |
1246 | @cindex tc_fix_adjustable | |
1247 | You may define this macro to indicate whether a fixup against a locally defined | |
1248 | symbol should be adjusted to be against the section symbol. It should return a | |
1249 | non-zero value if the adjustment is acceptable. | |
1250 | ||
1251 | @item MD_PCREL_FROM_SECTION | |
1252 | @cindex MD_PCREL_FROM_SECTION | |
1253 | If you define this macro, it should return the offset between the address of a | |
1254 | PC relative fixup and the position from which the PC relative adjustment should | |
1255 | be made. On many processors, the base of a PC relative instruction is the next | |
1256 | instruction, so this macro would return the length of an instruction. | |
1257 | ||
1258 | @item md_pcrel_from | |
1259 | @cindex md_pcrel_from | |
1260 | This is the default value of @code{MD_PCREL_FROM_SECTION}. The difference is | |
1261 | that @code{md_pcrel_from} does not take a section argument. | |
1262 | ||
1263 | @item tc_frob_label | |
1264 | @cindex tc_frob_label | |
1265 | If you define this macro, GAS will call it each time a label is defined. | |
1266 | ||
1267 | @item md_section_align | |
1268 | @cindex md_section_align | |
1269 | GAS will call this function for each section at the end of the assembly, to | |
65fd87bc ILT |
1270 | permit the CPU backend to adjust the alignment of a section. The function |
1271 | must take two arguments, a @code{segT} for the section and a @code{valueT} | |
1272 | for the size of the section, and return a @code{valueT} for the rounded | |
1273 | size. | |
252b5132 | 1274 | |
9f10757c TW |
1275 | @item md_macro_start |
1276 | @cindex md_macro_start | |
1277 | If defined, GAS will call this macro when it starts to include a macro | |
1278 | expansion. @code{macro_nest} indicates the current macro nesting level, which | |
58a77e41 | 1279 | includes the one being expanded. |
9f10757c TW |
1280 | |
1281 | @item md_macro_info | |
1282 | @cindex md_macro_info | |
1283 | If defined, GAS will call this macro after the macro expansion has been | |
1284 | included in the input and after parsing the macro arguments. The single | |
1285 | argument is a pointer to the macro processing's internal representation of the | |
1286 | macro (macro_entry *), which includes expansion of the formal arguments. | |
1287 | ||
1288 | @item md_macro_end | |
1289 | @cindex md_macro_end | |
1290 | Complement to md_macro_start. If defined, it is called when finished | |
58a77e41 | 1291 | processing an inserted macro expansion, just before decrementing macro_nest. |
9f10757c | 1292 | |
f28e8eb3 TW |
1293 | @item DOUBLEBAR_PARALLEL |
1294 | @cindex DOUBLEBAR_PARALLEL | |
1295 | Affects the preprocessor so that lines containing '||' don't have their | |
1296 | whitespace stripped following the double bar. This is useful for targets that | |
1297 | implement parallel instructions. | |
1298 | ||
1299 | @item KEEP_WHITE_AROUND_COLON | |
1300 | @cindex KEEP_WHITE_AROUND_COLON | |
1301 | Normally, whitespace is compressed and removed when, in the presence of the | |
1302 | colon, the adjoining tokens can be distinguished. This option affects the | |
1303 | preprocessor so that whitespace around colons is preserved. This is useful | |
1304 | when colons might be removed from the input after preprocessing but before | |
1305 | assembling, so that adjoining tokens can still be distinguished if there is | |
1306 | whitespace, or concatentated if there is not. | |
1307 | ||
252b5132 RH |
1308 | @item tc_frob_section |
1309 | @cindex tc_frob_section | |
1310 | If you define this macro, a @code{BFD_ASSEMBLER} GAS will call it for each | |
1311 | section at the end of the assembly. | |
1312 | ||
1313 | @item tc_frob_file_before_adjust | |
1314 | @cindex tc_frob_file_before_adjust | |
1315 | If you define this macro, GAS will call it after the symbol values are | |
1316 | resolved, but before the fixups have been changed from local symbols to section | |
1317 | symbols. | |
1318 | ||
1319 | @item tc_frob_symbol | |
1320 | @cindex tc_frob_symbol | |
1321 | If you define this macro, GAS will call it for each symbol. You can indicate | |
1322 | that the symbol should not be included in the object file by definining this | |
1323 | macro to set its second argument to a non-zero value. | |
1324 | ||
1325 | @item tc_frob_file | |
1326 | @cindex tc_frob_file | |
1327 | If you define this macro, GAS will call it after the symbol table has been | |
1328 | completed, but before the relocations have been generated. | |
1329 | ||
1330 | @item tc_frob_file_after_relocs | |
1331 | If you define this macro, GAS will call it after the relocs have been | |
1332 | generated. | |
1333 | ||
1334 | @item LISTING_HEADER | |
1335 | A string to use on the header line of a listing. The default value is simply | |
1336 | @code{"GAS LISTING"}. | |
1337 | ||
1338 | @item LISTING_WORD_SIZE | |
1339 | The number of bytes to put into a word in a listing. This affects the way the | |
1340 | bytes are clumped together in the listing. For example, a value of 2 might | |
1341 | print @samp{1234 5678} where a value of 1 would print @samp{12 34 56 78}. The | |
1342 | default value is 4. | |
1343 | ||
1344 | @item LISTING_LHS_WIDTH | |
1345 | The number of words of data to print on the first line of a listing for a | |
1346 | particular source line, where each word is @code{LISTING_WORD_SIZE} bytes. The | |
1347 | default value is 1. | |
1348 | ||
1349 | @item LISTING_LHS_WIDTH_SECOND | |
1350 | Like @code{LISTING_LHS_WIDTH}, but applying to the second and subsequent line | |
1351 | of the data printed for a particular source line. The default value is 1. | |
1352 | ||
1353 | @item LISTING_LHS_CONT_LINES | |
1354 | The maximum number of continuation lines to print in a listing for a particular | |
1355 | source line. The default value is 4. | |
1356 | ||
1357 | @item LISTING_RHS_WIDTH | |
1358 | The maximum number of characters to print from one line of the input file. The | |
1359 | default value is 100. | |
b8a9dcab NC |
1360 | |
1361 | @item TC_COFF_SECTION_DEFAULT_ATTRIBUTES | |
1362 | @cindex TC_COFF_SECTION_DEFAULT_ATTRIBUTES | |
1363 | The COFF @code{.section} directive will use the value of this macro to set | |
1364 | a new section's attributes when a directive has no valid flags or when the | |
1365 | flag is @code{w}. The default value of the macro is @code{SEC_LOAD | SEC_DATA}. | |
1366 | ||
252b5132 RH |
1367 | @end table |
1368 | ||
1369 | @node Object format backend | |
1370 | @subsection Writing an object format backend | |
1371 | @cindex object format backend | |
1372 | @cindex @file{obj-@var{fmt}} | |
1373 | ||
1374 | As with the CPU backend, the object format backend must define a few things, | |
1375 | and may define some other things. The interface to the object format backend | |
1376 | is generally simpler; most of the support for an object file format consists of | |
1377 | defining a number of pseudo-ops. | |
1378 | ||
1379 | The object format @file{.h} file must include @file{targ-cpu.h}. | |
1380 | ||
1381 | This section will only define the @code{BFD_ASSEMBLER} version of GAS. It is | |
1382 | impossible to support a new object file format using any other version anyhow, | |
1383 | as the original GAS version only supports a.out, and the @code{MANY_SEGMENTS} | |
1384 | GAS version only supports COFF. | |
1385 | ||
1386 | @table @code | |
1387 | @item OBJ_@var{format} | |
1388 | @cindex OBJ_@var{format} | |
1389 | By convention, you should define this macro in the @file{.h} file. For | |
1390 | example, @file{obj-elf.h} defines @code{OBJ_ELF}. You might have to use this | |
1391 | if it is necessary to add object file format specific code to the CPU file. | |
1392 | ||
1393 | @item obj_begin | |
1394 | If you define this macro, GAS will call it at the start of the assembly, after | |
1395 | the command line arguments have been parsed and all the machine independent | |
1396 | initializations have been completed. | |
1397 | ||
1398 | @item obj_app_file | |
1399 | @cindex obj_app_file | |
1400 | If you define this macro, GAS will invoke it when it sees a @code{.file} | |
1401 | pseudo-op or a @samp{#} line as used by the C preprocessor. | |
1402 | ||
1403 | @item OBJ_COPY_SYMBOL_ATTRIBUTES | |
1404 | @cindex OBJ_COPY_SYMBOL_ATTRIBUTES | |
1405 | You should define this macro to copy object format specific information from | |
1406 | one symbol to another. GAS will call it when one symbol is equated to | |
1407 | another. | |
1408 | ||
1409 | @item obj_fix_adjustable | |
1410 | @cindex obj_fix_adjustable | |
1411 | You may define this macro to indicate whether a fixup against a locally defined | |
1412 | symbol should be adjusted to be against the section symbol. It should return a | |
1413 | non-zero value if the adjustment is acceptable. | |
1414 | ||
1415 | @item obj_sec_sym_ok_for_reloc | |
1416 | @cindex obj_sec_sym_ok_for_reloc | |
1417 | You may define this macro to indicate that it is OK to use a section symbol in | |
1418 | a relocateion entry. If it is not, GAS will define a new symbol at the start | |
1419 | of a section. | |
1420 | ||
1421 | @item EMIT_SECTION_SYMBOLS | |
1422 | @cindex EMIT_SECTION_SYMBOLS | |
1423 | You should define this macro with a zero value if you do not want to include | |
1424 | section symbols in the output symbol table. The default value for this macro | |
1425 | is one. | |
1426 | ||
1427 | @item obj_adjust_symtab | |
1428 | @cindex obj_adjust_symtab | |
1429 | If you define this macro, GAS will invoke it just before setting the symbol | |
1430 | table of the output BFD. For example, the COFF support uses this macro to | |
1431 | generate a @code{.file} symbol if none was generated previously. | |
1432 | ||
1433 | @item SEPARATE_STAB_SECTIONS | |
1434 | @cindex SEPARATE_STAB_SECTIONS | |
0aa5d426 HPN |
1435 | You may define this macro to a nonzero value to indicate that stabs should be |
1436 | placed in separate sections, as in ELF. | |
252b5132 RH |
1437 | |
1438 | @item INIT_STAB_SECTION | |
1439 | @cindex INIT_STAB_SECTION | |
1440 | You may define this macro to initialize the stabs section in the output file. | |
1441 | ||
1442 | @item OBJ_PROCESS_STAB | |
1443 | @cindex OBJ_PROCESS_STAB | |
1444 | You may define this macro to do specific processing on a stabs entry. | |
1445 | ||
1446 | @item obj_frob_section | |
1447 | @cindex obj_frob_section | |
1448 | If you define this macro, GAS will call it for each section at the end of the | |
1449 | assembly. | |
1450 | ||
1451 | @item obj_frob_file_before_adjust | |
1452 | @cindex obj_frob_file_before_adjust | |
1453 | If you define this macro, GAS will call it after the symbol values are | |
1454 | resolved, but before the fixups have been changed from local symbols to section | |
1455 | symbols. | |
1456 | ||
1457 | @item obj_frob_symbol | |
1458 | @cindex obj_frob_symbol | |
1459 | If you define this macro, GAS will call it for each symbol. You can indicate | |
1460 | that the symbol should not be included in the object file by definining this | |
1461 | macro to set its second argument to a non-zero value. | |
1462 | ||
1463 | @item obj_frob_file | |
1464 | @cindex obj_frob_file | |
1465 | If you define this macro, GAS will call it after the symbol table has been | |
1466 | completed, but before the relocations have been generated. | |
1467 | ||
1468 | @item obj_frob_file_after_relocs | |
1469 | If you define this macro, GAS will call it after the relocs have been | |
1470 | generated. | |
945a1a6b ILT |
1471 | |
1472 | @item SET_SECTION_RELOCS (@var{sec}, @var{relocs}, @var{n}) | |
1473 | @cindex SET_SECTION_RELOCS | |
1474 | If you define this, it will be called after the relocations have been set for | |
1475 | the section @var{sec}. The list of relocations is in @var{relocs}, and the | |
1476 | number of relocations is in @var{n}. This is only used with | |
1477 | @code{BFD_ASSEMBLER}. | |
252b5132 RH |
1478 | @end table |
1479 | ||
1480 | @node Emulations | |
1481 | @subsection Writing emulation files | |
1482 | ||
1483 | Normally you do not have to write an emulation file. You can just use | |
1484 | @file{te-generic.h}. | |
1485 | ||
1486 | If you do write your own emulation file, it must include @file{obj-format.h}. | |
1487 | ||
1488 | An emulation file will often define @code{TE_@var{EM}}; this may then be used | |
1489 | in other files to change the output. | |
1490 | ||
1491 | @node Relaxation | |
1492 | @section Relaxation | |
1493 | @cindex relaxation | |
1494 | ||
1495 | @dfn{Relaxation} is a generic term used when the size of some instruction or | |
1496 | data depends upon the value of some symbol or other data. | |
1497 | ||
1498 | GAS knows to relax a particular type of PC relative relocation using a table. | |
1499 | You can also define arbitrarily complex forms of relaxation yourself. | |
1500 | ||
1501 | @menu | |
1502 | * Relaxing with a table:: Relaxing with a table | |
1503 | * General relaxing:: General relaxing | |
1504 | @end menu | |
1505 | ||
1506 | @node Relaxing with a table | |
1507 | @subsection Relaxing with a table | |
1508 | ||
1509 | If you do not define @code{md_relax_frag}, and you do define | |
1510 | @code{TC_GENERIC_RELAX_TABLE}, GAS will relax @code{rs_machine_dependent} frags | |
1511 | based on the frag subtype and the displacement to some specified target | |
1512 | address. The basic idea is that several machines have different addressing | |
1513 | modes for instructions that can specify different ranges of values, with | |
1514 | successive modes able to access wider ranges, including the entirety of the | |
1515 | previous range. Smaller ranges are assumed to be more desirable (perhaps the | |
1516 | instruction requires one word instead of two or three); if this is not the | |
1517 | case, don't describe the smaller-range, inferior mode. | |
1518 | ||
1519 | The @code{fr_subtype} field of a frag is an index into a CPU-specific | |
1520 | relaxation table. That table entry indicates the range of values that can be | |
1521 | stored, the number of bytes that will have to be added to the frag to | |
1522 | accomodate the addressing mode, and the index of the next entry to examine if | |
1523 | the value to be stored is outside the range accessible by the current | |
1524 | addressing mode. The @code{fr_symbol} field of the frag indicates what symbol | |
1525 | is to be accessed; the @code{fr_offset} field is added in. | |
1526 | ||
1527 | If the @code{TC_PCREL_ADJUST} macro is defined, which currently should only happen | |
1528 | for the NS32k family, the @code{TC_PCREL_ADJUST} macro is called on the frag to | |
1529 | compute an adjustment to be made to the displacement. | |
1530 | ||
1531 | The value fitted by the relaxation code is always assumed to be a displacement | |
1532 | from the current frag. (More specifically, from @code{fr_fix} bytes into the | |
1533 | frag.) | |
1534 | @ignore | |
1535 | This seems kinda silly. What about fitting small absolute values? I suppose | |
1536 | @code{md_assemble} is supposed to take care of that, but if the operand is a | |
1537 | difference between symbols, it might not be able to, if the difference was not | |
1538 | computable yet. | |
1539 | @end ignore | |
1540 | ||
1541 | The end of the relaxation sequence is indicated by a ``next'' value of 0. This | |
1542 | means that the first entry in the table can't be used. | |
1543 | ||
1544 | For some configurations, the linker can do relaxing within a section of an | |
1545 | object file. If call instructions of various sizes exist, the linker can | |
1546 | determine which should be used in each instance, when a symbol's value is | |
1547 | resolved. In order for the linker to avoid wasting space and having to insert | |
1548 | no-op instructions, it must be able to expand or shrink the section contents | |
1549 | while still preserving intra-section references and meeting alignment | |
1550 | requirements. | |
1551 | ||
1552 | For the i960 using b.out format, no expansion is done; instead, each | |
1553 | @samp{.align} directive causes extra space to be allocated, enough that when | |
1554 | the linker is relaxing a section and removing unneeded space, it can discard | |
1555 | some or all of this extra padding and cause the following data to be correctly | |
1556 | aligned. | |
1557 | ||
1558 | For the H8/300, I think the linker expands calls that can't reach, and doesn't | |
1559 | worry about alignment issues; the cpu probably never needs any significant | |
1560 | alignment beyond the instruction size. | |
1561 | ||
1562 | The relaxation table type contains these fields: | |
1563 | ||
1564 | @table @code | |
1565 | @item long rlx_forward | |
1566 | Forward reach, must be non-negative. | |
1567 | @item long rlx_backward | |
1568 | Backward reach, must be zero or negative. | |
1569 | @item rlx_length | |
1570 | Length in bytes of this addressing mode. | |
1571 | @item rlx_more | |
1572 | Index of the next-longer relax state, or zero if there is no next relax state. | |
1573 | @end table | |
1574 | ||
1575 | The relaxation is done in @code{relax_segment} in @file{write.c}. The | |
1576 | difference in the length fields between the original mode and the one finally | |
1577 | chosen by the relaxing code is taken as the size by which the current frag will | |
1578 | be increased in size. For example, if the initial relaxing mode has a length | |
1579 | of 2 bytes, and because of the size of the displacement, it gets upgraded to a | |
1580 | mode with a size of 6 bytes, it is assumed that the frag will grow by 4 bytes. | |
1581 | (The initial two bytes should have been part of the fixed portion of the frag, | |
1582 | since it is already known that they will be output.) This growth must be | |
1583 | effected by @code{md_convert_frag}; it should increase the @code{fr_fix} field | |
1584 | by the appropriate size, and fill in the appropriate bytes of the frag. | |
1585 | (Enough space for the maximum growth should have been allocated in the call to | |
1586 | frag_var as the second argument.) | |
1587 | ||
1588 | If relocation records are needed, they should be emitted by | |
1589 | @code{md_estimate_size_before_relax}. This function should examine the target | |
1590 | symbol of the supplied frag and correct the @code{fr_subtype} of the frag if | |
1591 | needed. When this function is called, if the symbol has not yet been defined, | |
1592 | it will not become defined later; however, its value may still change if the | |
1593 | section it is in gets relaxed. | |
1594 | ||
1595 | Usually, if the symbol is in the same section as the frag (given by the | |
1596 | @var{sec} argument), the narrowest likely relaxation mode is stored in | |
1597 | @code{fr_subtype}, and that's that. | |
1598 | ||
1599 | If the symbol is undefined, or in a different section (and therefore moveable | |
1600 | to an arbitrarily large distance), the largest available relaxation mode is | |
1601 | specified, @code{fix_new} is called to produce the relocation record, | |
1602 | @code{fr_fix} is increased to include the relocated field (remember, this | |
1603 | storage was allocated when @code{frag_var} was called), and @code{frag_wane} is | |
1604 | called to convert the frag to an @code{rs_fill} frag with no variant part. | |
1605 | Sometimes changing addressing modes may also require rewriting the instruction. | |
1606 | It can be accessed via @code{fr_opcode} or @code{fr_fix}. | |
1607 | ||
67db5ab4 HPN |
1608 | If you generate frags separately for the basic insn opcode and any relaxable |
1609 | operands, do not call @code{fix_new} thinking you can emit fixups for the | |
1610 | opcode field from the relaxable frag. It is not garanteed to be the same frag. | |
1611 | If you need to emit fixups for the opcode field from inspection of the | |
1612 | relaxable frag, then you need to generate a common frag for both the basic | |
1613 | opcode and relaxable fields, or you need to provide the frag for the opcode to | |
1614 | pass to @code{fix_new}. The latter can be done for example by defining | |
1615 | @code{TC_FRAG_TYPE} to include a pointer to it and defining @code{TC_FRAG_INIT} | |
1616 | to set the pointer. | |
1617 | ||
252b5132 RH |
1618 | Sometimes @code{fr_var} is increased instead, and @code{frag_wane} is not |
1619 | called. I'm not sure, but I think this is to keep @code{fr_fix} referring to | |
1620 | an earlier byte, and @code{fr_subtype} set to @code{rs_machine_dependent} so | |
1621 | that @code{md_convert_frag} will get called. | |
1622 | ||
1623 | @node General relaxing | |
1624 | @subsection General relaxing | |
1625 | ||
1626 | If using a simple table is not suitable, you may implement arbitrarily complex | |
1627 | relaxation semantics yourself. For example, the MIPS backend uses this to emit | |
1628 | different instruction sequences depending upon the size of the symbol being | |
1629 | accessed. | |
1630 | ||
1631 | When you assemble an instruction that may need relaxation, you should allocate | |
1632 | a frag using @code{frag_var} or @code{frag_variant} with a type of | |
1633 | @code{rs_machine_dependent}. You should store some sort of information in the | |
1634 | @code{fr_subtype} field so that you can figure out what to do with the frag | |
1635 | later. | |
1636 | ||
1637 | When GAS reaches the end of the input file, it will look through the frags and | |
1638 | work out their final sizes. | |
1639 | ||
1640 | GAS will first call @code{md_estimate_size_before_relax} on each | |
1641 | @code{rs_machine_dependent} frag. This function must return an estimated size | |
1642 | for the frag. | |
1643 | ||
1644 | GAS will then loop over the frags, calling @code{md_relax_frag} on each | |
1645 | @code{rs_machine_dependent} frag. This function should return the change in | |
1646 | size of the frag. GAS will keep looping over the frags until none of the frags | |
1647 | changes size. | |
1648 | ||
1649 | @node Broken words | |
1650 | @section Broken words | |
1651 | @cindex internals, broken words | |
1652 | @cindex broken words | |
1653 | ||
1654 | Some compilers, including GCC, will sometimes emit switch tables specifying | |
1655 | 16-bit @code{.word} displacements to branch targets, and branch instructions | |
1656 | that load entries from that table to compute the target address. If this is | |
1657 | done on a 32-bit machine, there is a chance (at least with really large | |
1658 | functions) that the displacement will not fit in 16 bits. The assembler | |
1659 | handles this using a concept called @dfn{broken words}. This idea is well | |
1660 | named, since there is an implied promise that the 16-bit field will in fact | |
1661 | hold the specified displacement. | |
1662 | ||
1663 | If broken word processing is enabled, and a situation like this is encountered, | |
1664 | the assembler will insert a jump instruction into the instruction stream, close | |
1665 | enough to be reached with the 16-bit displacement. This jump instruction will | |
1666 | transfer to the real desired target address. Thus, as long as the @code{.word} | |
1667 | value really is used as a displacement to compute an address to jump to, the | |
1668 | net effect will be correct (minus a very small efficiency cost). If | |
1669 | @code{.word} directives with label differences for values are used for other | |
1670 | purposes, however, things may not work properly. For targets which use broken | |
1671 | words, the @samp{-K} option will warn when a broken word is discovered. | |
1672 | ||
1673 | The broken word code is turned off by the @code{WORKING_DOT_WORD} macro. It | |
1674 | isn't needed if @code{.word} emits a value large enough to contain an address | |
1675 | (or, more correctly, any possible difference between two addresses). | |
1676 | ||
1677 | @node Internal functions | |
1678 | @section Internal functions | |
1679 | ||
1680 | This section describes basic internal functions used by GAS. | |
1681 | ||
1682 | @menu | |
1683 | * Warning and error messages:: Warning and error messages | |
1684 | * Hash tables:: Hash tables | |
1685 | @end menu | |
1686 | ||
1687 | @node Warning and error messages | |
1688 | @subsection Warning and error messages | |
1689 | ||
1690 | @deftypefun @{@} int had_warnings (void) | |
1691 | @deftypefunx @{@} int had_errors (void) | |
1692 | Returns non-zero if any warnings or errors, respectively, have been printed | |
1693 | during this invocation. | |
1694 | @end deftypefun | |
1695 | ||
1696 | @deftypefun @{@} void as_perror (const char *@var{gripe}, const char *@var{filename}) | |
1697 | Displays a BFD or system error, then clears the error status. | |
1698 | @end deftypefun | |
1699 | ||
1700 | @deftypefun @{@} void as_tsktsk (const char *@var{format}, ...) | |
1701 | @deftypefunx @{@} void as_warn (const char *@var{format}, ...) | |
1702 | @deftypefunx @{@} void as_bad (const char *@var{format}, ...) | |
1703 | @deftypefunx @{@} void as_fatal (const char *@var{format}, ...) | |
1704 | These functions display messages about something amiss with the input file, or | |
1705 | internal problems in the assembler itself. The current file name and line | |
1706 | number are printed, followed by the supplied message, formatted using | |
1707 | @code{vfprintf}, and a final newline. | |
1708 | ||
1709 | An error indicated by @code{as_bad} will result in a non-zero exit status when | |
1710 | the assembler has finished. Calling @code{as_fatal} will result in immediate | |
1711 | termination of the assembler process. | |
1712 | @end deftypefun | |
1713 | ||
1714 | @deftypefun @{@} void as_warn_where (char *@var{file}, unsigned int @var{line}, const char *@var{format}, ...) | |
1715 | @deftypefunx @{@} void as_bad_where (char *@var{file}, unsigned int @var{line}, const char *@var{format}, ...) | |
1716 | These variants permit specification of the file name and line number, and are | |
1717 | used when problems are detected when reprocessing information saved away when | |
1718 | processing some earlier part of the file. For example, fixups are processed | |
1719 | after all input has been read, but messages about fixups should refer to the | |
1720 | original filename and line number that they are applicable to. | |
1721 | @end deftypefun | |
1722 | ||
1723 | @deftypefun @{@} void fprint_value (FILE *@var{file}, valueT @var{val}) | |
1724 | @deftypefunx @{@} void sprint_value (char *@var{buf}, valueT @var{val}) | |
1725 | These functions are helpful for converting a @code{valueT} value into printable | |
1726 | format, in case it's wider than modes that @code{*printf} can handle. If the | |
1727 | type is narrow enough, a decimal number will be produced; otherwise, it will be | |
1728 | in hexadecimal. The value itself is not examined to make this determination. | |
1729 | @end deftypefun | |
1730 | ||
1731 | @node Hash tables | |
1732 | @subsection Hash tables | |
1733 | @cindex hash tables | |
1734 | ||
1735 | @deftypefun @{@} @{struct hash_control *@} hash_new (void) | |
1736 | Creates the hash table control structure. | |
1737 | @end deftypefun | |
1738 | ||
1739 | @deftypefun @{@} void hash_die (struct hash_control *) | |
1740 | Destroy a hash table. | |
1741 | @end deftypefun | |
1742 | ||
1743 | @deftypefun @{@} PTR hash_delete (struct hash_control *, const char *) | |
1744 | Deletes entry from the hash table, returns the value it had. | |
1745 | @end deftypefun | |
1746 | ||
1747 | @deftypefun @{@} PTR hash_replace (struct hash_control *, const char *, PTR) | |
1748 | Updates the value for an entry already in the table, returning the old value. | |
1749 | If no entry was found, just returns NULL. | |
1750 | @end deftypefun | |
1751 | ||
1752 | @deftypefun @{@} @{const char *@} hash_insert (struct hash_control *, const char *, PTR) | |
1753 | Inserting a value already in the table is an error. | |
1754 | Returns an error message or NULL. | |
1755 | @end deftypefun | |
1756 | ||
1757 | @deftypefun @{@} @{const char *@} hash_jam (struct hash_control *, const char *, PTR) | |
1758 | Inserts if the value isn't already present, updates it if it is. | |
1759 | @end deftypefun | |
1760 | ||
1761 | @node Test suite | |
1762 | @section Test suite | |
1763 | @cindex test suite | |
1764 | ||
1765 | The test suite is kind of lame for most processors. Often it only checks to | |
1766 | see if a couple of files can be assembled without the assembler reporting any | |
1767 | errors. For more complete testing, write a test which either examines the | |
1768 | assembler listing, or runs @code{objdump} and examines its output. For the | |
1769 | latter, the TCL procedure @code{run_dump_test} may come in handy. It takes the | |
1770 | base name of a file, and looks for @file{@var{file}.d}. This file should | |
1771 | contain as its initial lines a set of variable settings in @samp{#} comments, | |
1772 | in the form: | |
1773 | ||
1774 | @example | |
1775 | #@var{varname}: @var{value} | |
1776 | @end example | |
1777 | ||
1778 | The @var{varname} may be @code{objdump}, @code{nm}, or @code{as}, in which case | |
1779 | it specifies the options to be passed to the specified programs. Exactly one | |
1780 | of @code{objdump} or @code{nm} must be specified, as that also specifies which | |
1781 | program to run after the assembler has finished. If @var{varname} is | |
1782 | @code{source}, it specifies the name of the source file; otherwise, | |
1783 | @file{@var{file}.s} is used. If @var{varname} is @code{name}, it specifies the | |
1784 | name of the test to be used in the @code{pass} or @code{fail} messages. | |
1785 | ||
1786 | The non-commented parts of the file are interpreted as regular expressions, one | |
1787 | per line. Blank lines in the @code{objdump} or @code{nm} output are skipped, | |
1788 | as are blank lines in the @code{.d} file; the other lines are tested to see if | |
1789 | the regular expression matches the program output. If it does not, the test | |
1790 | fails. | |
1791 | ||
1792 | Note that this means the tests must be modified if the @code{objdump} output | |
1793 | style is changed. | |
1794 | ||
1795 | @bye | |
1796 | @c Local Variables: | |
1797 | @c fill-column: 79 | |
1798 | @c End: |