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43da67e8 | 11 | This chapter is not updated regularly, and it may be out of date.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