]>
Commit | Line | Data |
---|---|---|
c1f7febf RK |
1 | @c Copyright (C) 1988,89,92,93,94,96 Free Software Foundation, Inc. |
2 | @c This is part of the GCC manual. | |
3 | @c For copying conditions, see the file gcc.texi. | |
4 | ||
5 | @node C Extensions | |
6 | @chapter Extensions to the C Language Family | |
7 | @cindex extensions, C language | |
8 | @cindex C language extensions | |
9 | ||
10 | GNU C provides several language features not found in ANSI standard C. | |
11 | (The @samp{-pedantic} option directs GNU CC to print a warning message if | |
12 | any of these features is used.) To test for the availability of these | |
13 | features in conditional compilation, check for a predefined macro | |
14 | @code{__GNUC__}, which is always defined under GNU CC. | |
15 | ||
16 | These extensions are available in C and Objective C. Most of them are | |
17 | also available in C++. @xref{C++ Extensions,,Extensions to the | |
18 | C++ Language}, for extensions that apply @emph{only} to C++. | |
19 | ||
20 | @c The only difference between the two versions of this menu is that the | |
21 | @c version for clear INTERNALS has an extra node, "Constraints" (which | |
22 | @c appears in a separate chapter in the other version of the manual). | |
23 | @ifset INTERNALS | |
24 | @menu | |
25 | * Statement Exprs:: Putting statements and declarations inside expressions. | |
26 | * Local Labels:: Labels local to a statement-expression. | |
27 | * Labels as Values:: Getting pointers to labels, and computed gotos. | |
28 | * Nested Functions:: As in Algol and Pascal, lexical scoping of functions. | |
29 | * Constructing Calls:: Dispatching a call to another function. | |
30 | * Naming Types:: Giving a name to the type of some expression. | |
31 | * Typeof:: @code{typeof}: referring to the type of an expression. | |
32 | * Lvalues:: Using @samp{?:}, @samp{,} and casts in lvalues. | |
33 | * Conditionals:: Omitting the middle operand of a @samp{?:} expression. | |
34 | * Long Long:: Double-word integers---@code{long long int}. | |
35 | * Complex:: Data types for complex numbers. | |
36 | * Zero Length:: Zero-length arrays. | |
37 | * Variable Length:: Arrays whose length is computed at run time. | |
38 | * Macro Varargs:: Macros with variable number of arguments. | |
39 | * Subscripting:: Any array can be subscripted, even if not an lvalue. | |
40 | * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers. | |
41 | * Initializers:: Non-constant initializers. | |
42 | * Constructors:: Constructor expressions give structures, unions | |
43 | or arrays as values. | |
44 | * Labeled Elements:: Labeling elements of initializers. | |
45 | * Cast to Union:: Casting to union type from any member of the union. | |
46 | * Case Ranges:: `case 1 ... 9' and such. | |
47 | * Function Attributes:: Declaring that functions have no side effects, | |
48 | or that they can never return. | |
49 | * Function Prototypes:: Prototype declarations and old-style definitions. | |
50 | * C++ Comments:: C++ comments are recognized. | |
51 | * Dollar Signs:: Dollar sign is allowed in identifiers. | |
52 | * Character Escapes:: @samp{\e} stands for the character @key{ESC}. | |
53 | * Variable Attributes:: Specifying attributes of variables. | |
54 | * Type Attributes:: Specifying attributes of types. | |
55 | * Alignment:: Inquiring about the alignment of a type or variable. | |
56 | * Inline:: Defining inline functions (as fast as macros). | |
57 | * Extended Asm:: Assembler instructions with C expressions as operands. | |
58 | (With them you can define ``built-in'' functions.) | |
59 | * Asm Labels:: Specifying the assembler name to use for a C symbol. | |
60 | * Explicit Reg Vars:: Defining variables residing in specified registers. | |
61 | * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files. | |
62 | * Incomplete Enums:: @code{enum foo;}, with details to follow. | |
63 | * Function Names:: Printable strings which are the name of the current | |
64 | function. | |
65 | * Return Address:: Getting the return or frame address of a function. | |
66 | @end menu | |
67 | @end ifset | |
68 | @ifclear INTERNALS | |
69 | @menu | |
70 | * Statement Exprs:: Putting statements and declarations inside expressions. | |
71 | * Local Labels:: Labels local to a statement-expression. | |
72 | * Labels as Values:: Getting pointers to labels, and computed gotos. | |
73 | * Nested Functions:: As in Algol and Pascal, lexical scoping of functions. | |
74 | * Constructing Calls:: Dispatching a call to another function. | |
75 | * Naming Types:: Giving a name to the type of some expression. | |
76 | * Typeof:: @code{typeof}: referring to the type of an expression. | |
77 | * Lvalues:: Using @samp{?:}, @samp{,} and casts in lvalues. | |
78 | * Conditionals:: Omitting the middle operand of a @samp{?:} expression. | |
79 | * Long Long:: Double-word integers---@code{long long int}. | |
80 | * Complex:: Data types for complex numbers. | |
81 | * Zero Length:: Zero-length arrays. | |
82 | * Variable Length:: Arrays whose length is computed at run time. | |
83 | * Macro Varargs:: Macros with variable number of arguments. | |
84 | * Subscripting:: Any array can be subscripted, even if not an lvalue. | |
85 | * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers. | |
86 | * Initializers:: Non-constant initializers. | |
87 | * Constructors:: Constructor expressions give structures, unions | |
88 | or arrays as values. | |
89 | * Labeled Elements:: Labeling elements of initializers. | |
90 | * Cast to Union:: Casting to union type from any member of the union. | |
91 | * Case Ranges:: `case 1 ... 9' and such. | |
92 | * Function Attributes:: Declaring that functions have no side effects, | |
93 | or that they can never return. | |
94 | * Function Prototypes:: Prototype declarations and old-style definitions. | |
95 | * C++ Comments:: C++ comments are recognized. | |
96 | * Dollar Signs:: Dollar sign is allowed in identifiers. | |
97 | * Character Escapes:: @samp{\e} stands for the character @key{ESC}. | |
98 | * Variable Attributes:: Specifying attributes of variables. | |
99 | * Type Attributes:: Specifying attributes of types. | |
100 | * Alignment:: Inquiring about the alignment of a type or variable. | |
101 | * Inline:: Defining inline functions (as fast as macros). | |
102 | * Extended Asm:: Assembler instructions with C expressions as operands. | |
103 | (With them you can define ``built-in'' functions.) | |
104 | * Constraints:: Constraints for asm operands | |
105 | * Asm Labels:: Specifying the assembler name to use for a C symbol. | |
106 | * Explicit Reg Vars:: Defining variables residing in specified registers. | |
107 | * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files. | |
108 | * Incomplete Enums:: @code{enum foo;}, with details to follow. | |
109 | * Function Names:: Printable strings which are the name of the current | |
110 | function. | |
111 | * Return Address:: Getting the return or frame address of a function. | |
112 | @end menu | |
113 | @end ifclear | |
114 | ||
115 | @node Statement Exprs | |
116 | @section Statements and Declarations in Expressions | |
117 | @cindex statements inside expressions | |
118 | @cindex declarations inside expressions | |
119 | @cindex expressions containing statements | |
120 | @cindex macros, statements in expressions | |
121 | ||
122 | @c the above section title wrapped and causes an underfull hbox.. i | |
123 | @c changed it from "within" to "in". --mew 4feb93 | |
124 | ||
125 | A compound statement enclosed in parentheses may appear as an expression | |
126 | in GNU C. This allows you to use loops, switches, and local variables | |
127 | within an expression. | |
128 | ||
129 | Recall that a compound statement is a sequence of statements surrounded | |
130 | by braces; in this construct, parentheses go around the braces. For | |
131 | example: | |
132 | ||
133 | @example | |
134 | (@{ int y = foo (); int z; | |
135 | if (y > 0) z = y; | |
136 | else z = - y; | |
137 | z; @}) | |
138 | @end example | |
139 | ||
140 | @noindent | |
141 | is a valid (though slightly more complex than necessary) expression | |
142 | for the absolute value of @code{foo ()}. | |
143 | ||
144 | The last thing in the compound statement should be an expression | |
145 | followed by a semicolon; the value of this subexpression serves as the | |
146 | value of the entire construct. (If you use some other kind of statement | |
147 | last within the braces, the construct has type @code{void}, and thus | |
148 | effectively no value.) | |
149 | ||
150 | This feature is especially useful in making macro definitions ``safe'' (so | |
151 | that they evaluate each operand exactly once). For example, the | |
152 | ``maximum'' function is commonly defined as a macro in standard C as | |
153 | follows: | |
154 | ||
155 | @example | |
156 | #define max(a,b) ((a) > (b) ? (a) : (b)) | |
157 | @end example | |
158 | ||
159 | @noindent | |
160 | @cindex side effects, macro argument | |
161 | But this definition computes either @var{a} or @var{b} twice, with bad | |
162 | results if the operand has side effects. In GNU C, if you know the | |
163 | type of the operands (here let's assume @code{int}), you can define | |
164 | the macro safely as follows: | |
165 | ||
166 | @example | |
167 | #define maxint(a,b) \ | |
168 | (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @}) | |
169 | @end example | |
170 | ||
171 | Embedded statements are not allowed in constant expressions, such as | |
172 | the value of an enumeration constant, the width of a bit field, or | |
173 | the initial value of a static variable. | |
174 | ||
175 | If you don't know the type of the operand, you can still do this, but you | |
176 | must use @code{typeof} (@pxref{Typeof}) or type naming (@pxref{Naming | |
177 | Types}). | |
178 | ||
179 | @node Local Labels | |
180 | @section Locally Declared Labels | |
181 | @cindex local labels | |
182 | @cindex macros, local labels | |
183 | ||
184 | Each statement expression is a scope in which @dfn{local labels} can be | |
185 | declared. A local label is simply an identifier; you can jump to it | |
186 | with an ordinary @code{goto} statement, but only from within the | |
187 | statement expression it belongs to. | |
188 | ||
189 | A local label declaration looks like this: | |
190 | ||
191 | @example | |
192 | __label__ @var{label}; | |
193 | @end example | |
194 | ||
195 | @noindent | |
196 | or | |
197 | ||
198 | @example | |
199 | __label__ @var{label1}, @var{label2}, @dots{}; | |
200 | @end example | |
201 | ||
202 | Local label declarations must come at the beginning of the statement | |
203 | expression, right after the @samp{(@{}, before any ordinary | |
204 | declarations. | |
205 | ||
206 | The label declaration defines the label @emph{name}, but does not define | |
207 | the label itself. You must do this in the usual way, with | |
208 | @code{@var{label}:}, within the statements of the statement expression. | |
209 | ||
210 | The local label feature is useful because statement expressions are | |
211 | often used in macros. If the macro contains nested loops, a @code{goto} | |
212 | can be useful for breaking out of them. However, an ordinary label | |
213 | whose scope is the whole function cannot be used: if the macro can be | |
214 | expanded several times in one function, the label will be multiply | |
215 | defined in that function. A local label avoids this problem. For | |
216 | example: | |
217 | ||
218 | @example | |
219 | #define SEARCH(array, target) \ | |
220 | (@{ \ | |
221 | __label__ found; \ | |
222 | typeof (target) _SEARCH_target = (target); \ | |
223 | typeof (*(array)) *_SEARCH_array = (array); \ | |
224 | int i, j; \ | |
225 | int value; \ | |
226 | for (i = 0; i < max; i++) \ | |
227 | for (j = 0; j < max; j++) \ | |
228 | if (_SEARCH_array[i][j] == _SEARCH_target) \ | |
229 | @{ value = i; goto found; @} \ | |
230 | value = -1; \ | |
231 | found: \ | |
232 | value; \ | |
233 | @}) | |
234 | @end example | |
235 | ||
236 | @node Labels as Values | |
237 | @section Labels as Values | |
238 | @cindex labels as values | |
239 | @cindex computed gotos | |
240 | @cindex goto with computed label | |
241 | @cindex address of a label | |
242 | ||
243 | You can get the address of a label defined in the current function | |
244 | (or a containing function) with the unary operator @samp{&&}. The | |
245 | value has type @code{void *}. This value is a constant and can be used | |
246 | wherever a constant of that type is valid. For example: | |
247 | ||
248 | @example | |
249 | void *ptr; | |
250 | @dots{} | |
251 | ptr = &&foo; | |
252 | @end example | |
253 | ||
254 | To use these values, you need to be able to jump to one. This is done | |
255 | with the computed goto statement@footnote{The analogous feature in | |
256 | Fortran is called an assigned goto, but that name seems inappropriate in | |
257 | C, where one can do more than simply store label addresses in label | |
258 | variables.}, @code{goto *@var{exp};}. For example, | |
259 | ||
260 | @example | |
261 | goto *ptr; | |
262 | @end example | |
263 | ||
264 | @noindent | |
265 | Any expression of type @code{void *} is allowed. | |
266 | ||
267 | One way of using these constants is in initializing a static array that | |
268 | will serve as a jump table: | |
269 | ||
270 | @example | |
271 | static void *array[] = @{ &&foo, &&bar, &&hack @}; | |
272 | @end example | |
273 | ||
274 | Then you can select a label with indexing, like this: | |
275 | ||
276 | @example | |
277 | goto *array[i]; | |
278 | @end example | |
279 | ||
280 | @noindent | |
281 | Note that this does not check whether the subscript is in bounds---array | |
282 | indexing in C never does that. | |
283 | ||
284 | Such an array of label values serves a purpose much like that of the | |
285 | @code{switch} statement. The @code{switch} statement is cleaner, so | |
286 | use that rather than an array unless the problem does not fit a | |
287 | @code{switch} statement very well. | |
288 | ||
289 | Another use of label values is in an interpreter for threaded code. | |
290 | The labels within the interpreter function can be stored in the | |
291 | threaded code for super-fast dispatching. | |
292 | ||
293 | You can use this mechanism to jump to code in a different function. If | |
294 | you do that, totally unpredictable things will happen. The best way to | |
295 | avoid this is to store the label address only in automatic variables and | |
296 | never pass it as an argument. | |
297 | ||
298 | @node Nested Functions | |
299 | @section Nested Functions | |
300 | @cindex nested functions | |
301 | @cindex downward funargs | |
302 | @cindex thunks | |
303 | ||
304 | A @dfn{nested function} is a function defined inside another function. | |
305 | (Nested functions are not supported for GNU C++.) The nested function's | |
306 | name is local to the block where it is defined. For example, here we | |
307 | define a nested function named @code{square}, and call it twice: | |
308 | ||
309 | @example | |
310 | @group | |
311 | foo (double a, double b) | |
312 | @{ | |
313 | double square (double z) @{ return z * z; @} | |
314 | ||
315 | return square (a) + square (b); | |
316 | @} | |
317 | @end group | |
318 | @end example | |
319 | ||
320 | The nested function can access all the variables of the containing | |
321 | function that are visible at the point of its definition. This is | |
322 | called @dfn{lexical scoping}. For example, here we show a nested | |
323 | function which uses an inherited variable named @code{offset}: | |
324 | ||
325 | @example | |
326 | bar (int *array, int offset, int size) | |
327 | @{ | |
328 | int access (int *array, int index) | |
329 | @{ return array[index + offset]; @} | |
330 | int i; | |
331 | @dots{} | |
332 | for (i = 0; i < size; i++) | |
333 | @dots{} access (array, i) @dots{} | |
334 | @} | |
335 | @end example | |
336 | ||
337 | Nested function definitions are permitted within functions in the places | |
338 | where variable definitions are allowed; that is, in any block, before | |
339 | the first statement in the block. | |
340 | ||
341 | It is possible to call the nested function from outside the scope of its | |
342 | name by storing its address or passing the address to another function: | |
343 | ||
344 | @example | |
345 | hack (int *array, int size) | |
346 | @{ | |
347 | void store (int index, int value) | |
348 | @{ array[index] = value; @} | |
349 | ||
350 | intermediate (store, size); | |
351 | @} | |
352 | @end example | |
353 | ||
354 | Here, the function @code{intermediate} receives the address of | |
355 | @code{store} as an argument. If @code{intermediate} calls @code{store}, | |
356 | the arguments given to @code{store} are used to store into @code{array}. | |
357 | But this technique works only so long as the containing function | |
358 | (@code{hack}, in this example) does not exit. | |
359 | ||
360 | If you try to call the nested function through its address after the | |
361 | containing function has exited, all hell will break loose. If you try | |
362 | to call it after a containing scope level has exited, and if it refers | |
363 | to some of the variables that are no longer in scope, you may be lucky, | |
364 | but it's not wise to take the risk. If, however, the nested function | |
365 | does not refer to anything that has gone out of scope, you should be | |
366 | safe. | |
367 | ||
368 | GNU CC implements taking the address of a nested function using a | |
369 | technique called @dfn{trampolines}. A paper describing them is | |
370 | available from @samp{maya.idiap.ch} in directory @file{pub/tmb}, | |
371 | file @file{usenix88-lexic.ps.Z}. | |
372 | ||
373 | A nested function can jump to a label inherited from a containing | |
374 | function, provided the label was explicitly declared in the containing | |
375 | function (@pxref{Local Labels}). Such a jump returns instantly to the | |
376 | containing function, exiting the nested function which did the | |
377 | @code{goto} and any intermediate functions as well. Here is an example: | |
378 | ||
379 | @example | |
380 | @group | |
381 | bar (int *array, int offset, int size) | |
382 | @{ | |
383 | __label__ failure; | |
384 | int access (int *array, int index) | |
385 | @{ | |
386 | if (index > size) | |
387 | goto failure; | |
388 | return array[index + offset]; | |
389 | @} | |
390 | int i; | |
391 | @dots{} | |
392 | for (i = 0; i < size; i++) | |
393 | @dots{} access (array, i) @dots{} | |
394 | @dots{} | |
395 | return 0; | |
396 | ||
397 | /* @r{Control comes here from @code{access} | |
398 | if it detects an error.} */ | |
399 | failure: | |
400 | return -1; | |
401 | @} | |
402 | @end group | |
403 | @end example | |
404 | ||
405 | A nested function always has internal linkage. Declaring one with | |
406 | @code{extern} is erroneous. If you need to declare the nested function | |
407 | before its definition, use @code{auto} (which is otherwise meaningless | |
408 | for function declarations). | |
409 | ||
410 | @example | |
411 | bar (int *array, int offset, int size) | |
412 | @{ | |
413 | __label__ failure; | |
414 | auto int access (int *, int); | |
415 | @dots{} | |
416 | int access (int *array, int index) | |
417 | @{ | |
418 | if (index > size) | |
419 | goto failure; | |
420 | return array[index + offset]; | |
421 | @} | |
422 | @dots{} | |
423 | @} | |
424 | @end example | |
425 | ||
426 | @node Constructing Calls | |
427 | @section Constructing Function Calls | |
428 | @cindex constructing calls | |
429 | @cindex forwarding calls | |
430 | ||
431 | Using the built-in functions described below, you can record | |
432 | the arguments a function received, and call another function | |
433 | with the same arguments, without knowing the number or types | |
434 | of the arguments. | |
435 | ||
436 | You can also record the return value of that function call, | |
437 | and later return that value, without knowing what data type | |
438 | the function tried to return (as long as your caller expects | |
439 | that data type). | |
440 | ||
441 | @table @code | |
442 | @findex __builtin_apply_args | |
443 | @item __builtin_apply_args () | |
444 | This built-in function returns a pointer of type @code{void *} to data | |
445 | describing how to perform a call with the same arguments as were passed | |
446 | to the current function. | |
447 | ||
448 | The function saves the arg pointer register, structure value address, | |
449 | and all registers that might be used to pass arguments to a function | |
450 | into a block of memory allocated on the stack. Then it returns the | |
451 | address of that block. | |
452 | ||
453 | @findex __builtin_apply | |
454 | @item __builtin_apply (@var{function}, @var{arguments}, @var{size}) | |
455 | This built-in function invokes @var{function} (type @code{void (*)()}) | |
456 | with a copy of the parameters described by @var{arguments} (type | |
457 | @code{void *}) and @var{size} (type @code{int}). | |
458 | ||
459 | The value of @var{arguments} should be the value returned by | |
460 | @code{__builtin_apply_args}. The argument @var{size} specifies the size | |
461 | of the stack argument data, in bytes. | |
462 | ||
463 | This function returns a pointer of type @code{void *} to data describing | |
464 | how to return whatever value was returned by @var{function}. The data | |
465 | is saved in a block of memory allocated on the stack. | |
466 | ||
467 | It is not always simple to compute the proper value for @var{size}. The | |
468 | value is used by @code{__builtin_apply} to compute the amount of data | |
469 | that should be pushed on the stack and copied from the incoming argument | |
470 | area. | |
471 | ||
472 | @findex __builtin_return | |
473 | @item __builtin_return (@var{result}) | |
474 | This built-in function returns the value described by @var{result} from | |
475 | the containing function. You should specify, for @var{result}, a value | |
476 | returned by @code{__builtin_apply}. | |
477 | @end table | |
478 | ||
479 | @node Naming Types | |
480 | @section Naming an Expression's Type | |
481 | @cindex naming types | |
482 | ||
483 | You can give a name to the type of an expression using a @code{typedef} | |
484 | declaration with an initializer. Here is how to define @var{name} as a | |
485 | type name for the type of @var{exp}: | |
486 | ||
487 | @example | |
488 | typedef @var{name} = @var{exp}; | |
489 | @end example | |
490 | ||
491 | This is useful in conjunction with the statements-within-expressions | |
492 | feature. Here is how the two together can be used to define a safe | |
493 | ``maximum'' macro that operates on any arithmetic type: | |
494 | ||
495 | @example | |
496 | #define max(a,b) \ | |
497 | (@{typedef _ta = (a), _tb = (b); \ | |
498 | _ta _a = (a); _tb _b = (b); \ | |
499 | _a > _b ? _a : _b; @}) | |
500 | @end example | |
501 | ||
502 | @cindex underscores in variables in macros | |
503 | @cindex @samp{_} in variables in macros | |
504 | @cindex local variables in macros | |
505 | @cindex variables, local, in macros | |
506 | @cindex macros, local variables in | |
507 | ||
508 | The reason for using names that start with underscores for the local | |
509 | variables is to avoid conflicts with variable names that occur within the | |
510 | expressions that are substituted for @code{a} and @code{b}. Eventually we | |
511 | hope to design a new form of declaration syntax that allows you to declare | |
512 | variables whose scopes start only after their initializers; this will be a | |
513 | more reliable way to prevent such conflicts. | |
514 | ||
515 | @node Typeof | |
516 | @section Referring to a Type with @code{typeof} | |
517 | @findex typeof | |
518 | @findex sizeof | |
519 | @cindex macros, types of arguments | |
520 | ||
521 | Another way to refer to the type of an expression is with @code{typeof}. | |
522 | The syntax of using of this keyword looks like @code{sizeof}, but the | |
523 | construct acts semantically like a type name defined with @code{typedef}. | |
524 | ||
525 | There are two ways of writing the argument to @code{typeof}: with an | |
526 | expression or with a type. Here is an example with an expression: | |
527 | ||
528 | @example | |
529 | typeof (x[0](1)) | |
530 | @end example | |
531 | ||
532 | @noindent | |
533 | This assumes that @code{x} is an array of functions; the type described | |
534 | is that of the values of the functions. | |
535 | ||
536 | Here is an example with a typename as the argument: | |
537 | ||
538 | @example | |
539 | typeof (int *) | |
540 | @end example | |
541 | ||
542 | @noindent | |
543 | Here the type described is that of pointers to @code{int}. | |
544 | ||
545 | If you are writing a header file that must work when included in ANSI C | |
546 | programs, write @code{__typeof__} instead of @code{typeof}. | |
547 | @xref{Alternate Keywords}. | |
548 | ||
549 | A @code{typeof}-construct can be used anywhere a typedef name could be | |
550 | used. For example, you can use it in a declaration, in a cast, or inside | |
551 | of @code{sizeof} or @code{typeof}. | |
552 | ||
553 | @itemize @bullet | |
554 | @item | |
555 | This declares @code{y} with the type of what @code{x} points to. | |
556 | ||
557 | @example | |
558 | typeof (*x) y; | |
559 | @end example | |
560 | ||
561 | @item | |
562 | This declares @code{y} as an array of such values. | |
563 | ||
564 | @example | |
565 | typeof (*x) y[4]; | |
566 | @end example | |
567 | ||
568 | @item | |
569 | This declares @code{y} as an array of pointers to characters: | |
570 | ||
571 | @example | |
572 | typeof (typeof (char *)[4]) y; | |
573 | @end example | |
574 | ||
575 | @noindent | |
576 | It is equivalent to the following traditional C declaration: | |
577 | ||
578 | @example | |
579 | char *y[4]; | |
580 | @end example | |
581 | ||
582 | To see the meaning of the declaration using @code{typeof}, and why it | |
583 | might be a useful way to write, let's rewrite it with these macros: | |
584 | ||
585 | @example | |
586 | #define pointer(T) typeof(T *) | |
587 | #define array(T, N) typeof(T [N]) | |
588 | @end example | |
589 | ||
590 | @noindent | |
591 | Now the declaration can be rewritten this way: | |
592 | ||
593 | @example | |
594 | array (pointer (char), 4) y; | |
595 | @end example | |
596 | ||
597 | @noindent | |
598 | Thus, @code{array (pointer (char), 4)} is the type of arrays of 4 | |
599 | pointers to @code{char}. | |
600 | @end itemize | |
601 | ||
602 | @node Lvalues | |
603 | @section Generalized Lvalues | |
604 | @cindex compound expressions as lvalues | |
605 | @cindex expressions, compound, as lvalues | |
606 | @cindex conditional expressions as lvalues | |
607 | @cindex expressions, conditional, as lvalues | |
608 | @cindex casts as lvalues | |
609 | @cindex generalized lvalues | |
610 | @cindex lvalues, generalized | |
611 | @cindex extensions, @code{?:} | |
612 | @cindex @code{?:} extensions | |
613 | Compound expressions, conditional expressions and casts are allowed as | |
614 | lvalues provided their operands are lvalues. This means that you can take | |
615 | their addresses or store values into them. | |
616 | ||
617 | Standard C++ allows compound expressions and conditional expressions as | |
618 | lvalues, and permits casts to reference type, so use of this extension | |
619 | is deprecated for C++ code. | |
620 | ||
621 | For example, a compound expression can be assigned, provided the last | |
622 | expression in the sequence is an lvalue. These two expressions are | |
623 | equivalent: | |
624 | ||
625 | @example | |
626 | (a, b) += 5 | |
627 | a, (b += 5) | |
628 | @end example | |
629 | ||
630 | Similarly, the address of the compound expression can be taken. These two | |
631 | expressions are equivalent: | |
632 | ||
633 | @example | |
634 | &(a, b) | |
635 | a, &b | |
636 | @end example | |
637 | ||
638 | A conditional expression is a valid lvalue if its type is not void and the | |
639 | true and false branches are both valid lvalues. For example, these two | |
640 | expressions are equivalent: | |
641 | ||
642 | @example | |
643 | (a ? b : c) = 5 | |
644 | (a ? b = 5 : (c = 5)) | |
645 | @end example | |
646 | ||
647 | A cast is a valid lvalue if its operand is an lvalue. A simple | |
648 | assignment whose left-hand side is a cast works by converting the | |
649 | right-hand side first to the specified type, then to the type of the | |
650 | inner left-hand side expression. After this is stored, the value is | |
651 | converted back to the specified type to become the value of the | |
652 | assignment. Thus, if @code{a} has type @code{char *}, the following two | |
653 | expressions are equivalent: | |
654 | ||
655 | @example | |
656 | (int)a = 5 | |
657 | (int)(a = (char *)(int)5) | |
658 | @end example | |
659 | ||
660 | An assignment-with-arithmetic operation such as @samp{+=} applied to a cast | |
661 | performs the arithmetic using the type resulting from the cast, and then | |
662 | continues as in the previous case. Therefore, these two expressions are | |
663 | equivalent: | |
664 | ||
665 | @example | |
666 | (int)a += 5 | |
667 | (int)(a = (char *)(int) ((int)a + 5)) | |
668 | @end example | |
669 | ||
670 | You cannot take the address of an lvalue cast, because the use of its | |
671 | address would not work out coherently. Suppose that @code{&(int)f} were | |
672 | permitted, where @code{f} has type @code{float}. Then the following | |
673 | statement would try to store an integer bit-pattern where a floating | |
674 | point number belongs: | |
675 | ||
676 | @example | |
677 | *&(int)f = 1; | |
678 | @end example | |
679 | ||
680 | This is quite different from what @code{(int)f = 1} would do---that | |
681 | would convert 1 to floating point and store it. Rather than cause this | |
682 | inconsistency, we think it is better to prohibit use of @samp{&} on a cast. | |
683 | ||
684 | If you really do want an @code{int *} pointer with the address of | |
685 | @code{f}, you can simply write @code{(int *)&f}. | |
686 | ||
687 | @node Conditionals | |
688 | @section Conditionals with Omitted Operands | |
689 | @cindex conditional expressions, extensions | |
690 | @cindex omitted middle-operands | |
691 | @cindex middle-operands, omitted | |
692 | @cindex extensions, @code{?:} | |
693 | @cindex @code{?:} extensions | |
694 | ||
695 | The middle operand in a conditional expression may be omitted. Then | |
696 | if the first operand is nonzero, its value is the value of the conditional | |
697 | expression. | |
698 | ||
699 | Therefore, the expression | |
700 | ||
701 | @example | |
702 | x ? : y | |
703 | @end example | |
704 | ||
705 | @noindent | |
706 | has the value of @code{x} if that is nonzero; otherwise, the value of | |
707 | @code{y}. | |
708 | ||
709 | This example is perfectly equivalent to | |
710 | ||
711 | @example | |
712 | x ? x : y | |
713 | @end example | |
714 | ||
715 | @cindex side effect in ?: | |
716 | @cindex ?: side effect | |
717 | @noindent | |
718 | In this simple case, the ability to omit the middle operand is not | |
719 | especially useful. When it becomes useful is when the first operand does, | |
720 | or may (if it is a macro argument), contain a side effect. Then repeating | |
721 | the operand in the middle would perform the side effect twice. Omitting | |
722 | the middle operand uses the value already computed without the undesirable | |
723 | effects of recomputing it. | |
724 | ||
725 | @node Long Long | |
726 | @section Double-Word Integers | |
727 | @cindex @code{long long} data types | |
728 | @cindex double-word arithmetic | |
729 | @cindex multiprecision arithmetic | |
730 | ||
731 | GNU C supports data types for integers that are twice as long as | |
732 | @code{int}. Simply write @code{long long int} for a signed integer, or | |
733 | @code{unsigned long long int} for an unsigned integer. To make an | |
734 | integer constant of type @code{long long int}, add the suffix @code{LL} | |
735 | to the integer. To make an integer constant of type @code{unsigned long | |
736 | long int}, add the suffix @code{ULL} to the integer. | |
737 | ||
738 | You can use these types in arithmetic like any other integer types. | |
739 | Addition, subtraction, and bitwise boolean operations on these types | |
740 | are open-coded on all types of machines. Multiplication is open-coded | |
741 | if the machine supports fullword-to-doubleword a widening multiply | |
742 | instruction. Division and shifts are open-coded only on machines that | |
743 | provide special support. The operations that are not open-coded use | |
744 | special library routines that come with GNU CC. | |
745 | ||
746 | There may be pitfalls when you use @code{long long} types for function | |
747 | arguments, unless you declare function prototypes. If a function | |
748 | expects type @code{int} for its argument, and you pass a value of type | |
749 | @code{long long int}, confusion will result because the caller and the | |
750 | subroutine will disagree about the number of bytes for the argument. | |
751 | Likewise, if the function expects @code{long long int} and you pass | |
752 | @code{int}. The best way to avoid such problems is to use prototypes. | |
753 | ||
754 | @node Complex | |
755 | @section Complex Numbers | |
756 | @cindex complex numbers | |
757 | ||
758 | GNU C supports complex data types. You can declare both complex integer | |
759 | types and complex floating types, using the keyword @code{__complex__}. | |
760 | ||
761 | For example, @samp{__complex__ double x;} declares @code{x} as a | |
762 | variable whose real part and imaginary part are both of type | |
763 | @code{double}. @samp{__complex__ short int y;} declares @code{y} to | |
764 | have real and imaginary parts of type @code{short int}; this is not | |
765 | likely to be useful, but it shows that the set of complex types is | |
766 | complete. | |
767 | ||
768 | To write a constant with a complex data type, use the suffix @samp{i} or | |
769 | @samp{j} (either one; they are equivalent). For example, @code{2.5fi} | |
770 | has type @code{__complex__ float} and @code{3i} has type | |
771 | @code{__complex__ int}. Such a constant always has a pure imaginary | |
772 | value, but you can form any complex value you like by adding one to a | |
773 | real constant. | |
774 | ||
775 | To extract the real part of a complex-valued expression @var{exp}, write | |
776 | @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to | |
777 | extract the imaginary part. | |
778 | ||
779 | The operator @samp{~} performs complex conjugation when used on a value | |
780 | with a complex type. | |
781 | ||
782 | GNU CC can allocate complex automatic variables in a noncontiguous | |
783 | fashion; it's even possible for the real part to be in a register while | |
784 | the imaginary part is on the stack (or vice-versa). None of the | |
785 | supported debugging info formats has a way to represent noncontiguous | |
786 | allocation like this, so GNU CC describes a noncontiguous complex | |
787 | variable as if it were two separate variables of noncomplex type. | |
788 | If the variable's actual name is @code{foo}, the two fictitious | |
789 | variables are named @code{foo$real} and @code{foo$imag}. You can | |
790 | examine and set these two fictitious variables with your debugger. | |
791 | ||
792 | A future version of GDB will know how to recognize such pairs and treat | |
793 | them as a single variable with a complex type. | |
794 | ||
795 | @node Zero Length | |
796 | @section Arrays of Length Zero | |
797 | @cindex arrays of length zero | |
798 | @cindex zero-length arrays | |
799 | @cindex length-zero arrays | |
800 | ||
801 | Zero-length arrays are allowed in GNU C. They are very useful as the last | |
802 | element of a structure which is really a header for a variable-length | |
803 | object: | |
804 | ||
805 | @example | |
806 | struct line @{ | |
807 | int length; | |
808 | char contents[0]; | |
809 | @}; | |
810 | ||
811 | @{ | |
812 | struct line *thisline = (struct line *) | |
813 | malloc (sizeof (struct line) + this_length); | |
814 | thisline->length = this_length; | |
815 | @} | |
816 | @end example | |
817 | ||
818 | In standard C, you would have to give @code{contents} a length of 1, which | |
819 | means either you waste space or complicate the argument to @code{malloc}. | |
820 | ||
821 | @node Variable Length | |
822 | @section Arrays of Variable Length | |
823 | @cindex variable-length arrays | |
824 | @cindex arrays of variable length | |
825 | ||
826 | Variable-length automatic arrays are allowed in GNU C. These arrays are | |
827 | declared like any other automatic arrays, but with a length that is not | |
828 | a constant expression. The storage is allocated at the point of | |
829 | declaration and deallocated when the brace-level is exited. For | |
830 | example: | |
831 | ||
832 | @example | |
833 | FILE * | |
834 | concat_fopen (char *s1, char *s2, char *mode) | |
835 | @{ | |
836 | char str[strlen (s1) + strlen (s2) + 1]; | |
837 | strcpy (str, s1); | |
838 | strcat (str, s2); | |
839 | return fopen (str, mode); | |
840 | @} | |
841 | @end example | |
842 | ||
843 | @cindex scope of a variable length array | |
844 | @cindex variable-length array scope | |
845 | @cindex deallocating variable length arrays | |
846 | Jumping or breaking out of the scope of the array name deallocates the | |
847 | storage. Jumping into the scope is not allowed; you get an error | |
848 | message for it. | |
849 | ||
850 | @cindex @code{alloca} vs variable-length arrays | |
851 | You can use the function @code{alloca} to get an effect much like | |
852 | variable-length arrays. The function @code{alloca} is available in | |
853 | many other C implementations (but not in all). On the other hand, | |
854 | variable-length arrays are more elegant. | |
855 | ||
856 | There are other differences between these two methods. Space allocated | |
857 | with @code{alloca} exists until the containing @emph{function} returns. | |
858 | The space for a variable-length array is deallocated as soon as the array | |
859 | name's scope ends. (If you use both variable-length arrays and | |
860 | @code{alloca} in the same function, deallocation of a variable-length array | |
861 | will also deallocate anything more recently allocated with @code{alloca}.) | |
862 | ||
863 | You can also use variable-length arrays as arguments to functions: | |
864 | ||
865 | @example | |
866 | struct entry | |
867 | tester (int len, char data[len][len]) | |
868 | @{ | |
869 | @dots{} | |
870 | @} | |
871 | @end example | |
872 | ||
873 | The length of an array is computed once when the storage is allocated | |
874 | and is remembered for the scope of the array in case you access it with | |
875 | @code{sizeof}. | |
876 | ||
877 | If you want to pass the array first and the length afterward, you can | |
878 | use a forward declaration in the parameter list---another GNU extension. | |
879 | ||
880 | @example | |
881 | struct entry | |
882 | tester (int len; char data[len][len], int len) | |
883 | @{ | |
884 | @dots{} | |
885 | @} | |
886 | @end example | |
887 | ||
888 | @cindex parameter forward declaration | |
889 | The @samp{int len} before the semicolon is a @dfn{parameter forward | |
890 | declaration}, and it serves the purpose of making the name @code{len} | |
891 | known when the declaration of @code{data} is parsed. | |
892 | ||
893 | You can write any number of such parameter forward declarations in the | |
894 | parameter list. They can be separated by commas or semicolons, but the | |
895 | last one must end with a semicolon, which is followed by the ``real'' | |
896 | parameter declarations. Each forward declaration must match a ``real'' | |
897 | declaration in parameter name and data type. | |
898 | ||
899 | @node Macro Varargs | |
900 | @section Macros with Variable Numbers of Arguments | |
901 | @cindex variable number of arguments | |
902 | @cindex macro with variable arguments | |
903 | @cindex rest argument (in macro) | |
904 | ||
905 | In GNU C, a macro can accept a variable number of arguments, much as a | |
906 | function can. The syntax for defining the macro looks much like that | |
907 | used for a function. Here is an example: | |
908 | ||
909 | @example | |
910 | #define eprintf(format, args...) \ | |
911 | fprintf (stderr, format , ## args) | |
912 | @end example | |
913 | ||
914 | Here @code{args} is a @dfn{rest argument}: it takes in zero or more | |
915 | arguments, as many as the call contains. All of them plus the commas | |
916 | between them form the value of @code{args}, which is substituted into | |
917 | the macro body where @code{args} is used. Thus, we have this expansion: | |
918 | ||
919 | @example | |
920 | eprintf ("%s:%d: ", input_file_name, line_number) | |
921 | @expansion{} | |
922 | fprintf (stderr, "%s:%d: " , input_file_name, line_number) | |
923 | @end example | |
924 | ||
925 | @noindent | |
926 | Note that the comma after the string constant comes from the definition | |
927 | of @code{eprintf}, whereas the last comma comes from the value of | |
928 | @code{args}. | |
929 | ||
930 | The reason for using @samp{##} is to handle the case when @code{args} | |
931 | matches no arguments at all. In this case, @code{args} has an empty | |
932 | value. In this case, the second comma in the definition becomes an | |
933 | embarrassment: if it got through to the expansion of the macro, we would | |
934 | get something like this: | |
935 | ||
936 | @example | |
937 | fprintf (stderr, "success!\n" , ) | |
938 | @end example | |
939 | ||
940 | @noindent | |
941 | which is invalid C syntax. @samp{##} gets rid of the comma, so we get | |
942 | the following instead: | |
943 | ||
944 | @example | |
945 | fprintf (stderr, "success!\n") | |
946 | @end example | |
947 | ||
948 | This is a special feature of the GNU C preprocessor: @samp{##} before a | |
949 | rest argument that is empty discards the preceding sequence of | |
950 | non-whitespace characters from the macro definition. (If another macro | |
951 | argument precedes, none of it is discarded.) | |
952 | ||
953 | It might be better to discard the last preprocessor token instead of the | |
954 | last preceding sequence of non-whitespace characters; in fact, we may | |
955 | someday change this feature to do so. We advise you to write the macro | |
956 | definition so that the preceding sequence of non-whitespace characters | |
957 | is just a single token, so that the meaning will not change if we change | |
958 | the definition of this feature. | |
959 | ||
960 | @node Subscripting | |
961 | @section Non-Lvalue Arrays May Have Subscripts | |
962 | @cindex subscripting | |
963 | @cindex arrays, non-lvalue | |
964 | ||
965 | @cindex subscripting and function values | |
966 | Subscripting is allowed on arrays that are not lvalues, even though the | |
967 | unary @samp{&} operator is not. For example, this is valid in GNU C though | |
968 | not valid in other C dialects: | |
969 | ||
970 | @example | |
971 | @group | |
972 | struct foo @{int a[4];@}; | |
973 | ||
974 | struct foo f(); | |
975 | ||
976 | bar (int index) | |
977 | @{ | |
978 | return f().a[index]; | |
979 | @} | |
980 | @end group | |
981 | @end example | |
982 | ||
983 | @node Pointer Arith | |
984 | @section Arithmetic on @code{void}- and Function-Pointers | |
985 | @cindex void pointers, arithmetic | |
986 | @cindex void, size of pointer to | |
987 | @cindex function pointers, arithmetic | |
988 | @cindex function, size of pointer to | |
989 | ||
990 | In GNU C, addition and subtraction operations are supported on pointers to | |
991 | @code{void} and on pointers to functions. This is done by treating the | |
992 | size of a @code{void} or of a function as 1. | |
993 | ||
994 | A consequence of this is that @code{sizeof} is also allowed on @code{void} | |
995 | and on function types, and returns 1. | |
996 | ||
997 | The option @samp{-Wpointer-arith} requests a warning if these extensions | |
998 | are used. | |
999 | ||
1000 | @node Initializers | |
1001 | @section Non-Constant Initializers | |
1002 | @cindex initializers, non-constant | |
1003 | @cindex non-constant initializers | |
1004 | ||
1005 | As in standard C++, the elements of an aggregate initializer for an | |
1006 | automatic variable are not required to be constant expressions in GNU C. | |
1007 | Here is an example of an initializer with run-time varying elements: | |
1008 | ||
1009 | @example | |
1010 | foo (float f, float g) | |
1011 | @{ | |
1012 | float beat_freqs[2] = @{ f-g, f+g @}; | |
1013 | @dots{} | |
1014 | @} | |
1015 | @end example | |
1016 | ||
1017 | @node Constructors | |
1018 | @section Constructor Expressions | |
1019 | @cindex constructor expressions | |
1020 | @cindex initializations in expressions | |
1021 | @cindex structures, constructor expression | |
1022 | @cindex expressions, constructor | |
1023 | ||
1024 | GNU C supports constructor expressions. A constructor looks like | |
1025 | a cast containing an initializer. Its value is an object of the | |
1026 | type specified in the cast, containing the elements specified in | |
1027 | the initializer. | |
1028 | ||
1029 | Usually, the specified type is a structure. Assume that | |
1030 | @code{struct foo} and @code{structure} are declared as shown: | |
1031 | ||
1032 | @example | |
1033 | struct foo @{int a; char b[2];@} structure; | |
1034 | @end example | |
1035 | ||
1036 | @noindent | |
1037 | Here is an example of constructing a @code{struct foo} with a constructor: | |
1038 | ||
1039 | @example | |
1040 | structure = ((struct foo) @{x + y, 'a', 0@}); | |
1041 | @end example | |
1042 | ||
1043 | @noindent | |
1044 | This is equivalent to writing the following: | |
1045 | ||
1046 | @example | |
1047 | @{ | |
1048 | struct foo temp = @{x + y, 'a', 0@}; | |
1049 | structure = temp; | |
1050 | @} | |
1051 | @end example | |
1052 | ||
1053 | You can also construct an array. If all the elements of the constructor | |
1054 | are (made up of) simple constant expressions, suitable for use in | |
1055 | initializers, then the constructor is an lvalue and can be coerced to a | |
1056 | pointer to its first element, as shown here: | |
1057 | ||
1058 | @example | |
1059 | char **foo = (char *[]) @{ "x", "y", "z" @}; | |
1060 | @end example | |
1061 | ||
1062 | Array constructors whose elements are not simple constants are | |
1063 | not very useful, because the constructor is not an lvalue. There | |
1064 | are only two valid ways to use it: to subscript it, or initialize | |
1065 | an array variable with it. The former is probably slower than a | |
1066 | @code{switch} statement, while the latter does the same thing an | |
1067 | ordinary C initializer would do. Here is an example of | |
1068 | subscripting an array constructor: | |
1069 | ||
1070 | @example | |
1071 | output = ((int[]) @{ 2, x, 28 @}) [input]; | |
1072 | @end example | |
1073 | ||
1074 | Constructor expressions for scalar types and union types are is | |
1075 | also allowed, but then the constructor expression is equivalent | |
1076 | to a cast. | |
1077 | ||
1078 | @node Labeled Elements | |
1079 | @section Labeled Elements in Initializers | |
1080 | @cindex initializers with labeled elements | |
1081 | @cindex labeled elements in initializers | |
1082 | @cindex case labels in initializers | |
1083 | ||
1084 | Standard C requires the elements of an initializer to appear in a fixed | |
1085 | order, the same as the order of the elements in the array or structure | |
1086 | being initialized. | |
1087 | ||
1088 | In GNU C you can give the elements in any order, specifying the array | |
1089 | indices or structure field names they apply to. This extension is not | |
1090 | implemented in GNU C++. | |
1091 | ||
1092 | To specify an array index, write @samp{[@var{index}]} or | |
1093 | @samp{[@var{index}] =} before the element value. For example, | |
1094 | ||
1095 | @example | |
1096 | int a[6] = @{ [4] 29, [2] = 15 @}; | |
1097 | @end example | |
1098 | ||
1099 | @noindent | |
1100 | is equivalent to | |
1101 | ||
1102 | @example | |
1103 | int a[6] = @{ 0, 0, 15, 0, 29, 0 @}; | |
1104 | @end example | |
1105 | ||
1106 | @noindent | |
1107 | The index values must be constant expressions, even if the array being | |
1108 | initialized is automatic. | |
1109 | ||
1110 | To initialize a range of elements to the same value, write | |
1111 | @samp{[@var{first} ... @var{last}] = @var{value}}. For example, | |
1112 | ||
1113 | @example | |
1114 | int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @}; | |
1115 | @end example | |
1116 | ||
1117 | @noindent | |
1118 | Note that the length of the array is the highest value specified | |
1119 | plus one. | |
1120 | ||
1121 | In a structure initializer, specify the name of a field to initialize | |
1122 | with @samp{@var{fieldname}:} before the element value. For example, | |
1123 | given the following structure, | |
1124 | ||
1125 | @example | |
1126 | struct point @{ int x, y; @}; | |
1127 | @end example | |
1128 | ||
1129 | @noindent | |
1130 | the following initialization | |
1131 | ||
1132 | @example | |
1133 | struct point p = @{ y: yvalue, x: xvalue @}; | |
1134 | @end example | |
1135 | ||
1136 | @noindent | |
1137 | is equivalent to | |
1138 | ||
1139 | @example | |
1140 | struct point p = @{ xvalue, yvalue @}; | |
1141 | @end example | |
1142 | ||
1143 | Another syntax which has the same meaning is @samp{.@var{fieldname} =}., | |
1144 | as shown here: | |
1145 | ||
1146 | @example | |
1147 | struct point p = @{ .y = yvalue, .x = xvalue @}; | |
1148 | @end example | |
1149 | ||
1150 | You can also use an element label (with either the colon syntax or the | |
1151 | period-equal syntax) when initializing a union, to specify which element | |
1152 | of the union should be used. For example, | |
1153 | ||
1154 | @example | |
1155 | union foo @{ int i; double d; @}; | |
1156 | ||
1157 | union foo f = @{ d: 4 @}; | |
1158 | @end example | |
1159 | ||
1160 | @noindent | |
1161 | will convert 4 to a @code{double} to store it in the union using | |
1162 | the second element. By contrast, casting 4 to type @code{union foo} | |
1163 | would store it into the union as the integer @code{i}, since it is | |
1164 | an integer. (@xref{Cast to Union}.) | |
1165 | ||
1166 | You can combine this technique of naming elements with ordinary C | |
1167 | initialization of successive elements. Each initializer element that | |
1168 | does not have a label applies to the next consecutive element of the | |
1169 | array or structure. For example, | |
1170 | ||
1171 | @example | |
1172 | int a[6] = @{ [1] = v1, v2, [4] = v4 @}; | |
1173 | @end example | |
1174 | ||
1175 | @noindent | |
1176 | is equivalent to | |
1177 | ||
1178 | @example | |
1179 | int a[6] = @{ 0, v1, v2, 0, v4, 0 @}; | |
1180 | @end example | |
1181 | ||
1182 | Labeling the elements of an array initializer is especially useful | |
1183 | when the indices are characters or belong to an @code{enum} type. | |
1184 | For example: | |
1185 | ||
1186 | @example | |
1187 | int whitespace[256] | |
1188 | = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1, | |
1189 | ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @}; | |
1190 | @end example | |
1191 | ||
1192 | @node Case Ranges | |
1193 | @section Case Ranges | |
1194 | @cindex case ranges | |
1195 | @cindex ranges in case statements | |
1196 | ||
1197 | You can specify a range of consecutive values in a single @code{case} label, | |
1198 | like this: | |
1199 | ||
1200 | @example | |
1201 | case @var{low} ... @var{high}: | |
1202 | @end example | |
1203 | ||
1204 | @noindent | |
1205 | This has the same effect as the proper number of individual @code{case} | |
1206 | labels, one for each integer value from @var{low} to @var{high}, inclusive. | |
1207 | ||
1208 | This feature is especially useful for ranges of ASCII character codes: | |
1209 | ||
1210 | @example | |
1211 | case 'A' ... 'Z': | |
1212 | @end example | |
1213 | ||
1214 | @strong{Be careful:} Write spaces around the @code{...}, for otherwise | |
1215 | it may be parsed wrong when you use it with integer values. For example, | |
1216 | write this: | |
1217 | ||
1218 | @example | |
1219 | case 1 ... 5: | |
1220 | @end example | |
1221 | ||
1222 | @noindent | |
1223 | rather than this: | |
1224 | ||
1225 | @example | |
1226 | case 1...5: | |
1227 | @end example | |
1228 | ||
1229 | @node Cast to Union | |
1230 | @section Cast to a Union Type | |
1231 | @cindex cast to a union | |
1232 | @cindex union, casting to a | |
1233 | ||
1234 | A cast to union type is similar to other casts, except that the type | |
1235 | specified is a union type. You can specify the type either with | |
1236 | @code{union @var{tag}} or with a typedef name. A cast to union is actually | |
1237 | a constructor though, not a cast, and hence does not yield an lvalue like | |
1238 | normal casts. (@xref{Constructors}.) | |
1239 | ||
1240 | The types that may be cast to the union type are those of the members | |
1241 | of the union. Thus, given the following union and variables: | |
1242 | ||
1243 | @example | |
1244 | union foo @{ int i; double d; @}; | |
1245 | int x; | |
1246 | double y; | |
1247 | @end example | |
1248 | ||
1249 | @noindent | |
1250 | both @code{x} and @code{y} can be cast to type @code{union} foo. | |
1251 | ||
1252 | Using the cast as the right-hand side of an assignment to a variable of | |
1253 | union type is equivalent to storing in a member of the union: | |
1254 | ||
1255 | @example | |
1256 | union foo u; | |
1257 | @dots{} | |
1258 | u = (union foo) x @equiv{} u.i = x | |
1259 | u = (union foo) y @equiv{} u.d = y | |
1260 | @end example | |
1261 | ||
1262 | You can also use the union cast as a function argument: | |
1263 | ||
1264 | @example | |
1265 | void hack (union foo); | |
1266 | @dots{} | |
1267 | hack ((union foo) x); | |
1268 | @end example | |
1269 | ||
1270 | @node Function Attributes | |
1271 | @section Declaring Attributes of Functions | |
1272 | @cindex function attributes | |
1273 | @cindex declaring attributes of functions | |
1274 | @cindex functions that never return | |
1275 | @cindex functions that have no side effects | |
1276 | @cindex functions in arbitrary sections | |
1277 | @cindex @code{volatile} applied to function | |
1278 | @cindex @code{const} applied to function | |
1279 | @cindex functions with @code{printf} or @code{scanf} style arguments | |
1280 | @cindex functions that are passed arguments in registers on the 386 | |
1281 | @cindex functions that pop the argument stack on the 386 | |
1282 | @cindex functions that do not pop the argument stack on the 386 | |
1283 | ||
1284 | In GNU C, you declare certain things about functions called in your program | |
1285 | which help the compiler optimize function calls and check your code more | |
1286 | carefully. | |
1287 | ||
1288 | The keyword @code{__attribute__} allows you to specify special | |
1289 | attributes when making a declaration. This keyword is followed by an | |
1290 | attribute specification inside double parentheses. Eight attributes, | |
1291 | @code{noreturn}, @code{const}, @code{format}, @code{section}, | |
1292 | @code{constructor}, @code{destructor}, @code{unused} and @code{weak} are | |
1293 | currently defined for functions. Other attributes, including | |
1294 | @code{section} are supported for variables declarations (@pxref{Variable | |
1295 | Attributes}) and for types (@pxref{Type Attributes}). | |
1296 | ||
1297 | You may also specify attributes with @samp{__} preceding and following | |
1298 | each keyword. This allows you to use them in header files without | |
1299 | being concerned about a possible macro of the same name. For example, | |
1300 | you may use @code{__noreturn__} instead of @code{noreturn}. | |
1301 | ||
1302 | @table @code | |
1303 | @cindex @code{noreturn} function attribute | |
1304 | @item noreturn | |
1305 | A few standard library functions, such as @code{abort} and @code{exit}, | |
1306 | cannot return. GNU CC knows this automatically. Some programs define | |
1307 | their own functions that never return. You can declare them | |
1308 | @code{noreturn} to tell the compiler this fact. For example, | |
1309 | ||
1310 | @smallexample | |
1311 | void fatal () __attribute__ ((noreturn)); | |
1312 | ||
1313 | void | |
1314 | fatal (@dots{}) | |
1315 | @{ | |
1316 | @dots{} /* @r{Print error message.} */ @dots{} | |
1317 | exit (1); | |
1318 | @} | |
1319 | @end smallexample | |
1320 | ||
1321 | The @code{noreturn} keyword tells the compiler to assume that | |
1322 | @code{fatal} cannot return. It can then optimize without regard to what | |
1323 | would happen if @code{fatal} ever did return. This makes slightly | |
1324 | better code. More importantly, it helps avoid spurious warnings of | |
1325 | uninitialized variables. | |
1326 | ||
1327 | Do not assume that registers saved by the calling function are | |
1328 | restored before calling the @code{noreturn} function. | |
1329 | ||
1330 | It does not make sense for a @code{noreturn} function to have a return | |
1331 | type other than @code{void}. | |
1332 | ||
1333 | The attribute @code{noreturn} is not implemented in GNU C versions | |
1334 | earlier than 2.5. An alternative way to declare that a function does | |
1335 | not return, which works in the current version and in some older | |
1336 | versions, is as follows: | |
1337 | ||
1338 | @smallexample | |
1339 | typedef void voidfn (); | |
1340 | ||
1341 | volatile voidfn fatal; | |
1342 | @end smallexample | |
1343 | ||
1344 | @cindex @code{const} function attribute | |
1345 | @item const | |
1346 | Many functions do not examine any values except their arguments, and | |
1347 | have no effects except the return value. Such a function can be subject | |
1348 | to common subexpression elimination and loop optimization just as an | |
1349 | arithmetic operator would be. These functions should be declared | |
1350 | with the attribute @code{const}. For example, | |
1351 | ||
1352 | @smallexample | |
1353 | int square (int) __attribute__ ((const)); | |
1354 | @end smallexample | |
1355 | ||
1356 | @noindent | |
1357 | says that the hypothetical function @code{square} is safe to call | |
1358 | fewer times than the program says. | |
1359 | ||
1360 | The attribute @code{const} is not implemented in GNU C versions earlier | |
1361 | than 2.5. An alternative way to declare that a function has no side | |
1362 | effects, which works in the current version and in some older versions, | |
1363 | is as follows: | |
1364 | ||
1365 | @smallexample | |
1366 | typedef int intfn (); | |
1367 | ||
1368 | extern const intfn square; | |
1369 | @end smallexample | |
1370 | ||
1371 | This approach does not work in GNU C++ from 2.6.0 on, since the language | |
1372 | specifies that the @samp{const} must be attached to the return value. | |
1373 | ||
1374 | @cindex pointer arguments | |
1375 | Note that a function that has pointer arguments and examines the data | |
1376 | pointed to must @emph{not} be declared @code{const}. Likewise, a | |
1377 | function that calls a non-@code{const} function usually must not be | |
1378 | @code{const}. It does not make sense for a @code{const} function to | |
1379 | return @code{void}. | |
1380 | ||
1381 | @item format (@var{archetype}, @var{string-index}, @var{first-to-check}) | |
1382 | @cindex @code{format} function attribute | |
1383 | The @code{format} attribute specifies that a function takes @code{printf} | |
1384 | or @code{scanf} style arguments which should be type-checked against a | |
1385 | format string. For example, the declaration: | |
1386 | ||
1387 | @smallexample | |
1388 | extern int | |
1389 | my_printf (void *my_object, const char *my_format, ...) | |
1390 | __attribute__ ((format (printf, 2, 3))); | |
1391 | @end smallexample | |
1392 | ||
1393 | @noindent | |
1394 | causes the compiler to check the arguments in calls to @code{my_printf} | |
1395 | for consistency with the @code{printf} style format string argument | |
1396 | @code{my_format}. | |
1397 | ||
1398 | The parameter @var{archetype} determines how the format string is | |
1399 | interpreted, and should be either @code{printf} or @code{scanf}. The | |
1400 | parameter @var{string-index} specifies which argument is the format | |
1401 | string argument (starting from 1), while @var{first-to-check} is the | |
1402 | number of the first argument to check against the format string. For | |
1403 | functions where the arguments are not available to be checked (such as | |
1404 | @code{vprintf}), specify the third parameter as zero. In this case the | |
1405 | compiler only checks the format string for consistency. | |
1406 | ||
1407 | In the example above, the format string (@code{my_format}) is the second | |
1408 | argument of the function @code{my_print}, and the arguments to check | |
1409 | start with the third argument, so the correct parameters for the format | |
1410 | attribute are 2 and 3. | |
1411 | ||
1412 | The @code{format} attribute allows you to identify your own functions | |
1413 | which take format strings as arguments, so that GNU CC can check the | |
1414 | calls to these functions for errors. The compiler always checks formats | |
1415 | for the ANSI library functions @code{printf}, @code{fprintf}, | |
1416 | @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, | |
1417 | @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such | |
1418 | warnings are requested (using @samp{-Wformat}), so there is no need to | |
1419 | modify the header file @file{stdio.h}. | |
1420 | ||
1421 | @item format_arg (@var{string-index}) | |
1422 | @cindex @code{format_arg} function attribute | |
1423 | The @code{format_arg} attribute specifies that a function takes | |
1424 | @code{printf} or @code{scanf} style arguments, modifies it (for example, | |
1425 | to translate it into another language), and passes it to a @code{printf} | |
1426 | or @code{scanf} style function. For example, the declaration: | |
1427 | ||
1428 | @smallexample | |
1429 | extern char * | |
1430 | my_dgettext (char *my_domain, const char *my_format) | |
1431 | __attribute__ ((format_arg (2))); | |
1432 | @end smallexample | |
1433 | ||
1434 | @noindent | |
1435 | causes the compiler to check the arguments in calls to | |
1436 | @code{my_dgettext} whose result is passed to a @code{printf} or | |
1437 | @code{scanf} type function for consistency with the @code{printf} style | |
1438 | format string argument @code{my_format}. | |
1439 | ||
1440 | The parameter @var{string-index} specifies which argument is the format | |
1441 | string argument (starting from 1). | |
1442 | ||
1443 | The @code{format-arg} attribute allows you to identify your own | |
1444 | functions which modify format strings, so that GNU CC can check the | |
1445 | calls to @code{printf} and @code{scanf} function whose operands are a | |
1446 | call to one of your own function. The compiler always treats | |
1447 | @code{gettext}, @code{dgettext}, and @code{dcgettext} in this manner. | |
1448 | ||
1449 | @item section ("section-name") | |
1450 | @cindex @code{section} function attribute | |
1451 | Normally, the compiler places the code it generates in the @code{text} section. | |
1452 | Sometimes, however, you need additional sections, or you need certain | |
1453 | particular functions to appear in special sections. The @code{section} | |
1454 | attribute specifies that a function lives in a particular section. | |
1455 | For example, the declaration: | |
1456 | ||
1457 | @smallexample | |
1458 | extern void foobar (void) __attribute__ ((section ("bar"))); | |
1459 | @end smallexample | |
1460 | ||
1461 | @noindent | |
1462 | puts the function @code{foobar} in the @code{bar} section. | |
1463 | ||
1464 | Some file formats do not support arbitrary sections so the @code{section} | |
1465 | attribute is not available on all platforms. | |
1466 | If you need to map the entire contents of a module to a particular | |
1467 | section, consider using the facilities of the linker instead. | |
1468 | ||
1469 | @item constructor | |
1470 | @itemx destructor | |
1471 | @cindex @code{constructor} function attribute | |
1472 | @cindex @code{destructor} function attribute | |
1473 | The @code{constructor} attribute causes the function to be called | |
1474 | automatically before execution enters @code{main ()}. Similarly, the | |
1475 | @code{destructor} attribute causes the function to be called | |
1476 | automatically after @code{main ()} has completed or @code{exit ()} has | |
1477 | been called. Functions with these attributes are useful for | |
1478 | initializing data that will be used implicitly during the execution of | |
1479 | the program. | |
1480 | ||
1481 | These attributes are not currently implemented for Objective C. | |
1482 | ||
1483 | @item unused | |
1484 | This attribute, attached to a function, means that the function is meant | |
1485 | to be possibly unused. GNU CC will not produce a warning for this | |
1486 | function. GNU C++ does not currently support this attribute as | |
1487 | definitions without parameters are valid in C++. | |
1488 | ||
1489 | @item weak | |
1490 | @cindex @code{weak} attribute | |
1491 | The @code{weak} attribute causes the declaration to be emitted as a weak | |
1492 | symbol rather than a global. This is primarily useful in defining | |
1493 | library functions which can be overridden in user code, though it can | |
1494 | also be used with non-function declarations. Weak symbols are supported | |
1495 | for ELF targets, and also for a.out targets when using the GNU assembler | |
1496 | and linker. | |
1497 | ||
1498 | @item alias ("target") | |
1499 | @cindex @code{alias} attribute | |
1500 | The @code{alias} attribute causes the declaration to be emitted as an | |
1501 | alias for another symbol, which must be specified. For instance, | |
1502 | ||
1503 | @smallexample | |
1504 | void __f () @{ /* do something */; @} | |
1505 | void f () __attribute__ ((weak, alias ("__f"))); | |
1506 | @end smallexample | |
1507 | ||
1508 | declares @samp{f} to be a weak alias for @samp{__f}. In C++, the | |
1509 | mangled name for the target must be used. | |
1510 | ||
af3e86c2 RK |
1511 | Not all target machines support this attribute. |
1512 | ||
c1f7febf RK |
1513 | @item regparm (@var{number}) |
1514 | @cindex functions that are passed arguments in registers on the 386 | |
1515 | On the Intel 386, the @code{regparm} attribute causes the compiler to | |
1516 | pass up to @var{number} integer arguments in registers @var{EAX}, | |
1517 | @var{EDX}, and @var{ECX} instead of on the stack. Functions that take a | |
1518 | variable number of arguments will continue to be passed all of their | |
1519 | arguments on the stack. | |
1520 | ||
1521 | @item stdcall | |
1522 | @cindex functions that pop the argument stack on the 386 | |
1523 | On the Intel 386, the @code{stdcall} attribute causes the compiler to | |
1524 | assume that the called function will pop off the stack space used to | |
1525 | pass arguments, unless it takes a variable number of arguments. | |
1526 | ||
1527 | The PowerPC compiler for Windows NT currently ignores the @code{stdcall} | |
1528 | attribute. | |
1529 | ||
1530 | @item cdecl | |
1531 | @cindex functions that do pop the argument stack on the 386 | |
1532 | On the Intel 386, the @code{cdecl} attribute causes the compiler to | |
1533 | assume that the calling function will pop off the stack space used to | |
1534 | pass arguments. This is | |
1535 | useful to override the effects of the @samp{-mrtd} switch. | |
1536 | ||
1537 | The PowerPC compiler for Windows NT currently ignores the @code{cdecl} | |
1538 | attribute. | |
1539 | ||
1540 | @item longcall | |
1541 | @cindex functions called via pointer on the RS/6000 and PowerPC | |
1542 | On the RS/6000 and PowerPC, the @code{longcall} attribute causes the | |
1543 | compiler to always call the function via a pointer, so that functions | |
1544 | which reside further than 64 megabytes (67,108,864 bytes) from the | |
1545 | current location can be called. | |
1546 | ||
1547 | @item dllimport | |
1548 | @cindex functions which are imported from a dll on PowerPC Windows NT | |
1549 | On the PowerPC running Windows NT, the @code{dllimport} attribute causes | |
1550 | the compiler to call the function via a global pointer to the function | |
1551 | pointer that is set up by the Windows NT dll library. The pointer name | |
1552 | is formed by combining @code{__imp_} and the function name. | |
1553 | ||
1554 | @item dllexport | |
1555 | @cindex functions which are exported from a dll on PowerPC Windows NT | |
1556 | On the PowerPC running Windows NT, the @code{dllexport} attribute causes | |
1557 | the compiler to provide a global pointer to the function pointer, so | |
1558 | that it can be called with the @code{dllimport} attribute. The pointer | |
1559 | name is formed by combining @code{__imp_} and the function name. | |
1560 | ||
1561 | @item exception (@var{except-func} [, @var{except-arg}]) | |
1562 | @cindex functions which specify exception handling on PowerPC Windows NT | |
1563 | On the PowerPC running Windows NT, the @code{exception} attribute causes | |
1564 | the compiler to modify the structured exception table entry it emits for | |
1565 | the declared function. The string or identifier @var{except-func} is | |
1566 | placed in the third entry of the structured exception table. It | |
1567 | represents a function, which is called by the exception handling | |
1568 | mechanism if an exception occurs. If it was specified, the string or | |
1569 | identifier @var{except-arg} is placed in the fourth entry of the | |
1570 | structured exception table. | |
1571 | ||
1572 | @item function_vector | |
1573 | @cindex calling functions through the function vector on the H8/300 processors | |
1574 | Use this option on the H8/300 and H8/300H to indicate that the specified | |
1575 | function should be called through the function vector. Calling a | |
1576 | function through the function vector will reduce code size, however; | |
1577 | the function vector has a limited size (maximum 128 entries on the H8/300 | |
1578 | and 64 entries on the H8/300H) and shares space with the interrupt vector. | |
1579 | ||
1580 | You must use GAS and GLD from GNU binutils version 2.7 or later for | |
1581 | this option to work correctly. | |
1582 | ||
1583 | @item interrupt_handler | |
1584 | @cindex interrupt handler functions on the H8/300 processors | |
1585 | Use this option on the H8/300 and H8/300H to indicate that the specified | |
1586 | function is an interrupt handler. The compiler will generate function | |
1587 | entry and exit sequences suitable for use in an interrupt handler when this | |
1588 | attribute is present. | |
1589 | ||
1590 | @item eightbit_data | |
1591 | @cindex eight bit data on the H8/300 and H8/300H | |
1592 | Use this option on the H8/300 and H8/300H to indicate that the specified | |
1593 | variable should be placed into the eight bit data section. | |
1594 | The compiler will generate more efficient code for certain operations | |
1595 | on data in the eight bit data area. Note the eight bit data area is limited to | |
1596 | 256 bytes of data. | |
1597 | ||
1598 | You must use GAS and GLD from GNU binutils version 2.7 or later for | |
1599 | this option to work correctly. | |
1600 | ||
1601 | @item tiny_data | |
1602 | @cindex tiny data section on the H8/300H | |
1603 | Use this option on the H8/300H to indicate that the specified | |
1604 | variable should be placed into the tiny data section. | |
1605 | The compiler will generate more efficient code for loads and stores | |
1606 | on data in the tiny data section. Note the tiny data area is limited to | |
1607 | slightly under 32kbytes of data. | |
845da534 DE |
1608 | |
1609 | @item interrupt | |
1610 | @cindex interrupt handlers on the M32R/D | |
1611 | Use this option on the M32R/D to indicate that the specified | |
1612 | function is an interrupt handler. The compiler will generate function | |
1613 | entry and exit sequences suitable for use in an interrupt handler when this | |
1614 | attribute is present. | |
1615 | ||
1616 | @item model (@var{model-name}) | |
1617 | @cindex function addressability on the M32R/D | |
1618 | Use this attribute on the M32R/D to set the addressability of an object, | |
1619 | and the code generated for a function. | |
1620 | The identifier @var{model-name} is one of @code{small}, @code{medium}, | |
1621 | or @code{large}, representing each of the code models. | |
1622 | ||
1623 | Small model objects live in the lower 16MB of memory (so that their | |
1624 | addresses can be loaded with the @code{ld24} instruction), and are | |
1625 | callable with the @code{bl} instruction. | |
1626 | ||
1627 | Medium model objects may live anywhere in the 32 bit address space (the | |
1628 | compiler will generate @code{seth/add3} instructions to load their addresses), | |
1629 | and are callable with the @code{bl} instruction. | |
1630 | ||
1631 | Large model objects may live anywhere in the 32 bit address space (the | |
1632 | compiler will generate @code{seth/add3} instructions to load their addresses), | |
1633 | and may not be reachable with the @code{bl} instruction (the compiler will | |
1634 | generate the much slower @code{seth/add3/jl} instruction sequence). | |
1635 | ||
c1f7febf RK |
1636 | @end table |
1637 | ||
1638 | You can specify multiple attributes in a declaration by separating them | |
1639 | by commas within the double parentheses or by immediately following an | |
1640 | attribute declaration with another attribute declaration. | |
1641 | ||
1642 | @cindex @code{#pragma}, reason for not using | |
1643 | @cindex pragma, reason for not using | |
1644 | Some people object to the @code{__attribute__} feature, suggesting that ANSI C's | |
1645 | @code{#pragma} should be used instead. There are two reasons for not | |
1646 | doing this. | |
1647 | ||
1648 | @enumerate | |
1649 | @item | |
1650 | It is impossible to generate @code{#pragma} commands from a macro. | |
1651 | ||
1652 | @item | |
1653 | There is no telling what the same @code{#pragma} might mean in another | |
1654 | compiler. | |
1655 | @end enumerate | |
1656 | ||
1657 | These two reasons apply to almost any application that might be proposed | |
1658 | for @code{#pragma}. It is basically a mistake to use @code{#pragma} for | |
1659 | @emph{anything}. | |
1660 | ||
1661 | @node Function Prototypes | |
1662 | @section Prototypes and Old-Style Function Definitions | |
1663 | @cindex function prototype declarations | |
1664 | @cindex old-style function definitions | |
1665 | @cindex promotion of formal parameters | |
1666 | ||
1667 | GNU C extends ANSI C to allow a function prototype to override a later | |
1668 | old-style non-prototype definition. Consider the following example: | |
1669 | ||
1670 | @example | |
1671 | /* @r{Use prototypes unless the compiler is old-fashioned.} */ | |
d863830b | 1672 | #ifdef __STDC__ |
c1f7febf RK |
1673 | #define P(x) x |
1674 | #else | |
1675 | #define P(x) () | |
1676 | #endif | |
1677 | ||
1678 | /* @r{Prototype function declaration.} */ | |
1679 | int isroot P((uid_t)); | |
1680 | ||
1681 | /* @r{Old-style function definition.} */ | |
1682 | int | |
1683 | isroot (x) /* ??? lossage here ??? */ | |
1684 | uid_t x; | |
1685 | @{ | |
1686 | return x == 0; | |
1687 | @} | |
1688 | @end example | |
1689 | ||
1690 | Suppose the type @code{uid_t} happens to be @code{short}. ANSI C does | |
1691 | not allow this example, because subword arguments in old-style | |
1692 | non-prototype definitions are promoted. Therefore in this example the | |
1693 | function definition's argument is really an @code{int}, which does not | |
1694 | match the prototype argument type of @code{short}. | |
1695 | ||
1696 | This restriction of ANSI C makes it hard to write code that is portable | |
1697 | to traditional C compilers, because the programmer does not know | |
1698 | whether the @code{uid_t} type is @code{short}, @code{int}, or | |
1699 | @code{long}. Therefore, in cases like these GNU C allows a prototype | |
1700 | to override a later old-style definition. More precisely, in GNU C, a | |
1701 | function prototype argument type overrides the argument type specified | |
1702 | by a later old-style definition if the former type is the same as the | |
1703 | latter type before promotion. Thus in GNU C the above example is | |
1704 | equivalent to the following: | |
1705 | ||
1706 | @example | |
1707 | int isroot (uid_t); | |
1708 | ||
1709 | int | |
1710 | isroot (uid_t x) | |
1711 | @{ | |
1712 | return x == 0; | |
1713 | @} | |
1714 | @end example | |
1715 | ||
1716 | GNU C++ does not support old-style function definitions, so this | |
1717 | extension is irrelevant. | |
1718 | ||
1719 | @node C++ Comments | |
1720 | @section C++ Style Comments | |
1721 | @cindex // | |
1722 | @cindex C++ comments | |
1723 | @cindex comments, C++ style | |
1724 | ||
1725 | In GNU C, you may use C++ style comments, which start with @samp{//} and | |
1726 | continue until the end of the line. Many other C implementations allow | |
1727 | such comments, and they are likely to be in a future C standard. | |
1728 | However, C++ style comments are not recognized if you specify | |
1729 | @w{@samp{-ansi}} or @w{@samp{-traditional}}, since they are incompatible | |
1730 | with traditional constructs like @code{dividend//*comment*/divisor}. | |
1731 | ||
1732 | @node Dollar Signs | |
1733 | @section Dollar Signs in Identifier Names | |
1734 | @cindex $ | |
1735 | @cindex dollar signs in identifier names | |
1736 | @cindex identifier names, dollar signs in | |
1737 | ||
79188db9 RK |
1738 | In GNU C, you may normally use dollar signs in identifier names. |
1739 | This is because many traditional C implementations allow such identifiers. | |
1740 | However, dollar signs in identifiers are not supported on a few target | |
1741 | machines, typically because the target assembler does not allow them. | |
c1f7febf RK |
1742 | |
1743 | @node Character Escapes | |
1744 | @section The Character @key{ESC} in Constants | |
1745 | ||
1746 | You can use the sequence @samp{\e} in a string or character constant to | |
1747 | stand for the ASCII character @key{ESC}. | |
1748 | ||
1749 | @node Alignment | |
1750 | @section Inquiring on Alignment of Types or Variables | |
1751 | @cindex alignment | |
1752 | @cindex type alignment | |
1753 | @cindex variable alignment | |
1754 | ||
1755 | The keyword @code{__alignof__} allows you to inquire about how an object | |
1756 | is aligned, or the minimum alignment usually required by a type. Its | |
1757 | syntax is just like @code{sizeof}. | |
1758 | ||
1759 | For example, if the target machine requires a @code{double} value to be | |
1760 | aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8. | |
1761 | This is true on many RISC machines. On more traditional machine | |
1762 | designs, @code{__alignof__ (double)} is 4 or even 2. | |
1763 | ||
1764 | Some machines never actually require alignment; they allow reference to any | |
1765 | data type even at an odd addresses. For these machines, @code{__alignof__} | |
1766 | reports the @emph{recommended} alignment of a type. | |
1767 | ||
1768 | When the operand of @code{__alignof__} is an lvalue rather than a type, the | |
1769 | value is the largest alignment that the lvalue is known to have. It may | |
1770 | have this alignment as a result of its data type, or because it is part of | |
1771 | a structure and inherits alignment from that structure. For example, after | |
1772 | this declaration: | |
1773 | ||
1774 | @example | |
1775 | struct foo @{ int x; char y; @} foo1; | |
1776 | @end example | |
1777 | ||
1778 | @noindent | |
1779 | the value of @code{__alignof__ (foo1.y)} is probably 2 or 4, the same as | |
1780 | @code{__alignof__ (int)}, even though the data type of @code{foo1.y} | |
1781 | does not itself demand any alignment.@refill | |
1782 | ||
1783 | A related feature which lets you specify the alignment of an object is | |
1784 | @code{__attribute__ ((aligned (@var{alignment})))}; see the following | |
1785 | section. | |
1786 | ||
1787 | @node Variable Attributes | |
1788 | @section Specifying Attributes of Variables | |
1789 | @cindex attribute of variables | |
1790 | @cindex variable attributes | |
1791 | ||
1792 | The keyword @code{__attribute__} allows you to specify special | |
1793 | attributes of variables or structure fields. This keyword is followed | |
1794 | by an attribute specification inside double parentheses. Eight | |
1795 | attributes are currently defined for variables: @code{aligned}, | |
1796 | @code{mode}, @code{nocommon}, @code{packed}, @code{section}, | |
1797 | @code{transparent_union}, @code{unused}, and @code{weak}. Other | |
1798 | attributes are available for functions (@pxref{Function Attributes}) and | |
1799 | for types (@pxref{Type Attributes}). | |
1800 | ||
1801 | You may also specify attributes with @samp{__} preceding and following | |
1802 | each keyword. This allows you to use them in header files without | |
1803 | being concerned about a possible macro of the same name. For example, | |
1804 | you may use @code{__aligned__} instead of @code{aligned}. | |
1805 | ||
1806 | @table @code | |
1807 | @cindex @code{aligned} attribute | |
1808 | @item aligned (@var{alignment}) | |
1809 | This attribute specifies a minimum alignment for the variable or | |
1810 | structure field, measured in bytes. For example, the declaration: | |
1811 | ||
1812 | @smallexample | |
1813 | int x __attribute__ ((aligned (16))) = 0; | |
1814 | @end smallexample | |
1815 | ||
1816 | @noindent | |
1817 | causes the compiler to allocate the global variable @code{x} on a | |
1818 | 16-byte boundary. On a 68040, this could be used in conjunction with | |
1819 | an @code{asm} expression to access the @code{move16} instruction which | |
1820 | requires 16-byte aligned operands. | |
1821 | ||
1822 | You can also specify the alignment of structure fields. For example, to | |
1823 | create a double-word aligned @code{int} pair, you could write: | |
1824 | ||
1825 | @smallexample | |
1826 | struct foo @{ int x[2] __attribute__ ((aligned (8))); @}; | |
1827 | @end smallexample | |
1828 | ||
1829 | @noindent | |
1830 | This is an alternative to creating a union with a @code{double} member | |
1831 | that forces the union to be double-word aligned. | |
1832 | ||
1833 | It is not possible to specify the alignment of functions; the alignment | |
1834 | of functions is determined by the machine's requirements and cannot be | |
1835 | changed. You cannot specify alignment for a typedef name because such a | |
1836 | name is just an alias, not a distinct type. | |
1837 | ||
1838 | As in the preceding examples, you can explicitly specify the alignment | |
1839 | (in bytes) that you wish the compiler to use for a given variable or | |
1840 | structure field. Alternatively, you can leave out the alignment factor | |
1841 | and just ask the compiler to align a variable or field to the maximum | |
1842 | useful alignment for the target machine you are compiling for. For | |
1843 | example, you could write: | |
1844 | ||
1845 | @smallexample | |
1846 | short array[3] __attribute__ ((aligned)); | |
1847 | @end smallexample | |
1848 | ||
1849 | Whenever you leave out the alignment factor in an @code{aligned} attribute | |
1850 | specification, the compiler automatically sets the alignment for the declared | |
1851 | variable or field to the largest alignment which is ever used for any data | |
1852 | type on the target machine you are compiling for. Doing this can often make | |
1853 | copy operations more efficient, because the compiler can use whatever | |
1854 | instructions copy the biggest chunks of memory when performing copies to | |
1855 | or from the variables or fields that you have aligned this way. | |
1856 | ||
1857 | The @code{aligned} attribute can only increase the alignment; but you | |
1858 | can decrease it by specifying @code{packed} as well. See below. | |
1859 | ||
1860 | Note that the effectiveness of @code{aligned} attributes may be limited | |
1861 | by inherent limitations in your linker. On many systems, the linker is | |
1862 | only able to arrange for variables to be aligned up to a certain maximum | |
1863 | alignment. (For some linkers, the maximum supported alignment may | |
1864 | be very very small.) If your linker is only able to align variables | |
1865 | up to a maximum of 8 byte alignment, then specifying @code{aligned(16)} | |
1866 | in an @code{__attribute__} will still only provide you with 8 byte | |
1867 | alignment. See your linker documentation for further information. | |
1868 | ||
1869 | @item mode (@var{mode}) | |
1870 | @cindex @code{mode} attribute | |
1871 | This attribute specifies the data type for the declaration---whichever | |
1872 | type corresponds to the mode @var{mode}. This in effect lets you | |
1873 | request an integer or floating point type according to its width. | |
1874 | ||
1875 | You may also specify a mode of @samp{byte} or @samp{__byte__} to | |
1876 | indicate the mode corresponding to a one-byte integer, @samp{word} or | |
1877 | @samp{__word__} for the mode of a one-word integer, and @samp{pointer} | |
1878 | or @samp{__pointer__} for the mode used to represent pointers. | |
1879 | ||
1880 | @item nocommon | |
1881 | @cindex @code{nocommon} attribute | |
1882 | This attribute specifies requests GNU CC not to place a variable | |
1883 | ``common'' but instead to allocate space for it directly. If you | |
1884 | specify the @samp{-fno-common} flag, GNU CC will do this for all | |
1885 | variables. | |
1886 | ||
1887 | Specifying the @code{nocommon} attribute for a variable provides an | |
1888 | initialization of zeros. A variable may only be initialized in one | |
1889 | source file. | |
1890 | ||
1891 | @item packed | |
1892 | @cindex @code{packed} attribute | |
1893 | The @code{packed} attribute specifies that a variable or structure field | |
1894 | should have the smallest possible alignment---one byte for a variable, | |
1895 | and one bit for a field, unless you specify a larger value with the | |
1896 | @code{aligned} attribute. | |
1897 | ||
1898 | Here is a structure in which the field @code{x} is packed, so that it | |
1899 | immediately follows @code{a}: | |
1900 | ||
1901 | @example | |
1902 | struct foo | |
1903 | @{ | |
1904 | char a; | |
1905 | int x[2] __attribute__ ((packed)); | |
1906 | @}; | |
1907 | @end example | |
1908 | ||
1909 | @item section ("section-name") | |
1910 | @cindex @code{section} variable attribute | |
1911 | Normally, the compiler places the objects it generates in sections like | |
1912 | @code{data} and @code{bss}. Sometimes, however, you need additional sections, | |
1913 | or you need certain particular variables to appear in special sections, | |
1914 | for example to map to special hardware. The @code{section} | |
1915 | attribute specifies that a variable (or function) lives in a particular | |
1916 | section. For example, this small program uses several specific section names: | |
1917 | ||
1918 | @smallexample | |
1919 | struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @}; | |
1920 | struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @}; | |
1921 | char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @}; | |
1922 | int init_data __attribute__ ((section ("INITDATA"))) = 0; | |
1923 | ||
1924 | main() | |
1925 | @{ | |
1926 | /* Initialize stack pointer */ | |
1927 | init_sp (stack + sizeof (stack)); | |
1928 | ||
1929 | /* Initialize initialized data */ | |
1930 | memcpy (&init_data, &data, &edata - &data); | |
1931 | ||
1932 | /* Turn on the serial ports */ | |
1933 | init_duart (&a); | |
1934 | init_duart (&b); | |
1935 | @} | |
1936 | @end smallexample | |
1937 | ||
1938 | @noindent | |
1939 | Use the @code{section} attribute with an @emph{initialized} definition | |
1940 | of a @emph{global} variable, as shown in the example. GNU CC issues | |
1941 | a warning and otherwise ignores the @code{section} attribute in | |
1942 | uninitialized variable declarations. | |
1943 | ||
1944 | You may only use the @code{section} attribute with a fully initialized | |
1945 | global definition because of the way linkers work. The linker requires | |
1946 | each object be defined once, with the exception that uninitialized | |
1947 | variables tentatively go in the @code{common} (or @code{bss}) section | |
1948 | and can be multiply "defined". You can force a variable to be | |
1949 | initialized with the @samp{-fno-common} flag or the @code{nocommon} | |
1950 | attribute. | |
1951 | ||
1952 | Some file formats do not support arbitrary sections so the @code{section} | |
1953 | attribute is not available on all platforms. | |
1954 | If you need to map the entire contents of a module to a particular | |
1955 | section, consider using the facilities of the linker instead. | |
1956 | ||
1957 | @item transparent_union | |
1958 | This attribute, attached to a function parameter which is a union, means | |
1959 | that the corresponding argument may have the type of any union member, | |
1960 | but the argument is passed as if its type were that of the first union | |
1961 | member. For more details see @xref{Type Attributes}. You can also use | |
1962 | this attribute on a @code{typedef} for a union data type; then it | |
1963 | applies to all function parameters with that type. | |
1964 | ||
1965 | @item unused | |
1966 | This attribute, attached to a variable, means that the variable is meant | |
1967 | to be possibly unused. GNU CC will not produce a warning for this | |
1968 | variable. | |
1969 | ||
1970 | @item weak | |
1971 | The @code{weak} attribute is described in @xref{Function Attributes}. | |
845da534 DE |
1972 | |
1973 | @item model (@var{model-name}) | |
1974 | @cindex variable addressability on the M32R/D | |
1975 | Use this attribute on the M32R/D to set the addressability of an object. | |
1976 | The identifier @var{model-name} is one of @code{small}, @code{medium}, | |
1977 | or @code{large}, representing each of the code models. | |
1978 | ||
1979 | Small model objects live in the lower 16MB of memory (so that their | |
1980 | addresses can be loaded with the @code{ld24} instruction). | |
1981 | ||
1982 | Medium and large model objects may live anywhere in the 32 bit address space | |
1983 | (the compiler will generate @code{seth/add3} instructions to load their | |
1984 | addresses). | |
1985 | ||
c1f7febf RK |
1986 | @end table |
1987 | ||
1988 | To specify multiple attributes, separate them by commas within the | |
1989 | double parentheses: for example, @samp{__attribute__ ((aligned (16), | |
1990 | packed))}. | |
1991 | ||
1992 | @node Type Attributes | |
1993 | @section Specifying Attributes of Types | |
1994 | @cindex attribute of types | |
1995 | @cindex type attributes | |
1996 | ||
1997 | The keyword @code{__attribute__} allows you to specify special | |
1998 | attributes of @code{struct} and @code{union} types when you define such | |
1999 | types. This keyword is followed by an attribute specification inside | |
2000 | double parentheses. Three attributes are currently defined for types: | |
2001 | @code{aligned}, @code{packed}, and @code{transparent_union}. Other | |
2002 | attributes are defined for functions (@pxref{Function Attributes}) and | |
2003 | for variables (@pxref{Variable Attributes}). | |
2004 | ||
2005 | You may also specify any one of these attributes with @samp{__} | |
2006 | preceding and following its keyword. This allows you to use these | |
2007 | attributes in header files without being concerned about a possible | |
2008 | macro of the same name. For example, you may use @code{__aligned__} | |
2009 | instead of @code{aligned}. | |
2010 | ||
2011 | You may specify the @code{aligned} and @code{transparent_union} | |
2012 | attributes either in a @code{typedef} declaration or just past the | |
2013 | closing curly brace of a complete enum, struct or union type | |
2014 | @emph{definition} and the @code{packed} attribute only past the closing | |
2015 | brace of a definition. | |
2016 | ||
2017 | @table @code | |
2018 | @cindex @code{aligned} attribute | |
2019 | @item aligned (@var{alignment}) | |
2020 | This attribute specifies a minimum alignment (in bytes) for variables | |
2021 | of the specified type. For example, the declarations: | |
2022 | ||
2023 | @smallexample | |
2024 | struct S @{ short f[3]; @} __attribute__ ((aligned (8)); | |
2025 | typedef int more_aligned_int __attribute__ ((aligned (8)); | |
2026 | @end smallexample | |
2027 | ||
2028 | @noindent | |
d863830b | 2029 | force the compiler to insure (as far as it can) that each variable whose |
c1f7febf RK |
2030 | type is @code{struct S} or @code{more_aligned_int} will be allocated and |
2031 | aligned @emph{at least} on a 8-byte boundary. On a Sparc, having all | |
2032 | variables of type @code{struct S} aligned to 8-byte boundaries allows | |
2033 | the compiler to use the @code{ldd} and @code{std} (doubleword load and | |
2034 | store) instructions when copying one variable of type @code{struct S} to | |
2035 | another, thus improving run-time efficiency. | |
2036 | ||
2037 | Note that the alignment of any given @code{struct} or @code{union} type | |
2038 | is required by the ANSI C standard to be at least a perfect multiple of | |
2039 | the lowest common multiple of the alignments of all of the members of | |
2040 | the @code{struct} or @code{union} in question. This means that you @emph{can} | |
2041 | effectively adjust the alignment of a @code{struct} or @code{union} | |
2042 | type by attaching an @code{aligned} attribute to any one of the members | |
2043 | of such a type, but the notation illustrated in the example above is a | |
2044 | more obvious, intuitive, and readable way to request the compiler to | |
2045 | adjust the alignment of an entire @code{struct} or @code{union} type. | |
2046 | ||
2047 | As in the preceding example, you can explicitly specify the alignment | |
2048 | (in bytes) that you wish the compiler to use for a given @code{struct} | |
2049 | or @code{union} type. Alternatively, you can leave out the alignment factor | |
2050 | and just ask the compiler to align a type to the maximum | |
2051 | useful alignment for the target machine you are compiling for. For | |
2052 | example, you could write: | |
2053 | ||
2054 | @smallexample | |
2055 | struct S @{ short f[3]; @} __attribute__ ((aligned)); | |
2056 | @end smallexample | |
2057 | ||
2058 | Whenever you leave out the alignment factor in an @code{aligned} | |
2059 | attribute specification, the compiler automatically sets the alignment | |
2060 | for the type to the largest alignment which is ever used for any data | |
2061 | type on the target machine you are compiling for. Doing this can often | |
2062 | make copy operations more efficient, because the compiler can use | |
2063 | whatever instructions copy the biggest chunks of memory when performing | |
2064 | copies to or from the variables which have types that you have aligned | |
2065 | this way. | |
2066 | ||
2067 | In the example above, if the size of each @code{short} is 2 bytes, then | |
2068 | the size of the entire @code{struct S} type is 6 bytes. The smallest | |
2069 | power of two which is greater than or equal to that is 8, so the | |
2070 | compiler sets the alignment for the entire @code{struct S} type to 8 | |
2071 | bytes. | |
2072 | ||
2073 | Note that although you can ask the compiler to select a time-efficient | |
2074 | alignment for a given type and then declare only individual stand-alone | |
2075 | objects of that type, the compiler's ability to select a time-efficient | |
2076 | alignment is primarily useful only when you plan to create arrays of | |
2077 | variables having the relevant (efficiently aligned) type. If you | |
2078 | declare or use arrays of variables of an efficiently-aligned type, then | |
2079 | it is likely that your program will also be doing pointer arithmetic (or | |
2080 | subscripting, which amounts to the same thing) on pointers to the | |
2081 | relevant type, and the code that the compiler generates for these | |
2082 | pointer arithmetic operations will often be more efficient for | |
2083 | efficiently-aligned types than for other types. | |
2084 | ||
2085 | The @code{aligned} attribute can only increase the alignment; but you | |
2086 | can decrease it by specifying @code{packed} as well. See below. | |
2087 | ||
2088 | Note that the effectiveness of @code{aligned} attributes may be limited | |
2089 | by inherent limitations in your linker. On many systems, the linker is | |
2090 | only able to arrange for variables to be aligned up to a certain maximum | |
2091 | alignment. (For some linkers, the maximum supported alignment may | |
2092 | be very very small.) If your linker is only able to align variables | |
2093 | up to a maximum of 8 byte alignment, then specifying @code{aligned(16)} | |
2094 | in an @code{__attribute__} will still only provide you with 8 byte | |
2095 | alignment. See your linker documentation for further information. | |
2096 | ||
2097 | @item packed | |
2098 | This attribute, attached to an @code{enum}, @code{struct}, or | |
2099 | @code{union} type definition, specified that the minimum required memory | |
2100 | be used to represent the type. | |
2101 | ||
2102 | Specifying this attribute for @code{struct} and @code{union} types is | |
2103 | equivalent to specifying the @code{packed} attribute on each of the | |
2104 | structure or union members. Specifying the @samp{-fshort-enums} | |
2105 | flag on the line is equivalent to specifying the @code{packed} | |
2106 | attribute on all @code{enum} definitions. | |
2107 | ||
2108 | You may only specify this attribute after a closing curly brace on an | |
1cd4bca9 BK |
2109 | @code{enum} definition, not in a @code{typedef} declaration, unless that |
2110 | declaration also contains the definition of the @code{enum}. | |
c1f7febf RK |
2111 | |
2112 | @item transparent_union | |
2113 | This attribute, attached to a @code{union} type definition, indicates | |
2114 | that any function parameter having that union type causes calls to that | |
2115 | function to be treated in a special way. | |
2116 | ||
2117 | First, the argument corresponding to a transparent union type can be of | |
2118 | any type in the union; no cast is required. Also, if the union contains | |
2119 | a pointer type, the corresponding argument can be a null pointer | |
2120 | constant or a void pointer expression; and if the union contains a void | |
2121 | pointer type, the corresponding argument can be any pointer expression. | |
2122 | If the union member type is a pointer, qualifiers like @code{const} on | |
2123 | the referenced type must be respected, just as with normal pointer | |
2124 | conversions. | |
2125 | ||
2126 | Second, the argument is passed to the function using the calling | |
2127 | conventions of first member of the transparent union, not the calling | |
2128 | conventions of the union itself. All members of the union must have the | |
2129 | same machine representation; this is necessary for this argument passing | |
2130 | to work properly. | |
2131 | ||
2132 | Transparent unions are designed for library functions that have multiple | |
2133 | interfaces for compatibility reasons. For example, suppose the | |
2134 | @code{wait} function must accept either a value of type @code{int *} to | |
2135 | comply with Posix, or a value of type @code{union wait *} to comply with | |
2136 | the 4.1BSD interface. If @code{wait}'s parameter were @code{void *}, | |
2137 | @code{wait} would accept both kinds of arguments, but it would also | |
2138 | accept any other pointer type and this would make argument type checking | |
2139 | less useful. Instead, @code{<sys/wait.h>} might define the interface | |
2140 | as follows: | |
2141 | ||
2142 | @smallexample | |
2143 | typedef union | |
2144 | @{ | |
2145 | int *__ip; | |
2146 | union wait *__up; | |
2147 | @} wait_status_ptr_t __attribute__ ((__transparent_union__)); | |
2148 | ||
2149 | pid_t wait (wait_status_ptr_t); | |
2150 | @end smallexample | |
2151 | ||
2152 | This interface allows either @code{int *} or @code{union wait *} | |
2153 | arguments to be passed, using the @code{int *} calling convention. | |
2154 | The program can call @code{wait} with arguments of either type: | |
2155 | ||
2156 | @example | |
2157 | int w1 () @{ int w; return wait (&w); @} | |
2158 | int w2 () @{ union wait w; return wait (&w); @} | |
2159 | @end example | |
2160 | ||
2161 | With this interface, @code{wait}'s implementation might look like this: | |
2162 | ||
2163 | @example | |
2164 | pid_t wait (wait_status_ptr_t p) | |
2165 | @{ | |
2166 | return waitpid (-1, p.__ip, 0); | |
2167 | @} | |
2168 | @end example | |
d863830b JL |
2169 | |
2170 | @item unused | |
2171 | When attached to a type (including a @code{union} or a @code{struct}), | |
2172 | this attribute means that variables of that type are meant to appear | |
2173 | possibly unused. GNU CC will not produce a warning for any variables of | |
2174 | that type, even if the variable appears to do nothing. This is often | |
2175 | the case with lock or thread classes, which are usually defined and then | |
2176 | not referenced, but contain constructors and destructors that have | |
956d6950 | 2177 | nontrivial bookkeeping functions. |
d863830b | 2178 | |
c1f7febf RK |
2179 | @end table |
2180 | ||
2181 | To specify multiple attributes, separate them by commas within the | |
2182 | double parentheses: for example, @samp{__attribute__ ((aligned (16), | |
2183 | packed))}. | |
2184 | ||
2185 | @node Inline | |
2186 | @section An Inline Function is As Fast As a Macro | |
2187 | @cindex inline functions | |
2188 | @cindex integrating function code | |
2189 | @cindex open coding | |
2190 | @cindex macros, inline alternative | |
2191 | ||
2192 | By declaring a function @code{inline}, you can direct GNU CC to | |
2193 | integrate that function's code into the code for its callers. This | |
2194 | makes execution faster by eliminating the function-call overhead; in | |
2195 | addition, if any of the actual argument values are constant, their known | |
2196 | values may permit simplifications at compile time so that not all of the | |
2197 | inline function's code needs to be included. The effect on code size is | |
2198 | less predictable; object code may be larger or smaller with function | |
2199 | inlining, depending on the particular case. Inlining of functions is an | |
2200 | optimization and it really ``works'' only in optimizing compilation. If | |
2201 | you don't use @samp{-O}, no function is really inline. | |
2202 | ||
2203 | To declare a function inline, use the @code{inline} keyword in its | |
2204 | declaration, like this: | |
2205 | ||
2206 | @example | |
2207 | inline int | |
2208 | inc (int *a) | |
2209 | @{ | |
2210 | (*a)++; | |
2211 | @} | |
2212 | @end example | |
2213 | ||
2214 | (If you are writing a header file to be included in ANSI C programs, write | |
2215 | @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.) | |
2216 | ||
2217 | You can also make all ``simple enough'' functions inline with the option | |
2218 | @samp{-finline-functions}. Note that certain usages in a function | |
2219 | definition can make it unsuitable for inline substitution. | |
2220 | ||
2221 | Note that in C and Objective C, unlike C++, the @code{inline} keyword | |
2222 | does not affect the linkage of the function. | |
2223 | ||
2224 | @cindex automatic @code{inline} for C++ member fns | |
2225 | @cindex @code{inline} automatic for C++ member fns | |
2226 | @cindex member fns, automatically @code{inline} | |
2227 | @cindex C++ member fns, automatically @code{inline} | |
2228 | GNU CC automatically inlines member functions defined within the class | |
2229 | body of C++ programs even if they are not explicitly declared | |
2230 | @code{inline}. (You can override this with @samp{-fno-default-inline}; | |
2231 | @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.) | |
2232 | ||
2233 | @cindex inline functions, omission of | |
2234 | When a function is both inline and @code{static}, if all calls to the | |
2235 | function are integrated into the caller, and the function's address is | |
2236 | never used, then the function's own assembler code is never referenced. | |
2237 | In this case, GNU CC does not actually output assembler code for the | |
2238 | function, unless you specify the option @samp{-fkeep-inline-functions}. | |
2239 | Some calls cannot be integrated for various reasons (in particular, | |
2240 | calls that precede the function's definition cannot be integrated, and | |
2241 | neither can recursive calls within the definition). If there is a | |
2242 | nonintegrated call, then the function is compiled to assembler code as | |
2243 | usual. The function must also be compiled as usual if the program | |
2244 | refers to its address, because that can't be inlined. | |
2245 | ||
2246 | @cindex non-static inline function | |
2247 | When an inline function is not @code{static}, then the compiler must assume | |
2248 | that there may be calls from other source files; since a global symbol can | |
2249 | be defined only once in any program, the function must not be defined in | |
2250 | the other source files, so the calls therein cannot be integrated. | |
2251 | Therefore, a non-@code{static} inline function is always compiled on its | |
2252 | own in the usual fashion. | |
2253 | ||
2254 | If you specify both @code{inline} and @code{extern} in the function | |
2255 | definition, then the definition is used only for inlining. In no case | |
2256 | is the function compiled on its own, not even if you refer to its | |
2257 | address explicitly. Such an address becomes an external reference, as | |
2258 | if you had only declared the function, and had not defined it. | |
2259 | ||
2260 | This combination of @code{inline} and @code{extern} has almost the | |
2261 | effect of a macro. The way to use it is to put a function definition in | |
2262 | a header file with these keywords, and put another copy of the | |
2263 | definition (lacking @code{inline} and @code{extern}) in a library file. | |
2264 | The definition in the header file will cause most calls to the function | |
2265 | to be inlined. If any uses of the function remain, they will refer to | |
2266 | the single copy in the library. | |
2267 | ||
2268 | GNU C does not inline any functions when not optimizing. It is not | |
2269 | clear whether it is better to inline or not, in this case, but we found | |
2270 | that a correct implementation when not optimizing was difficult. So we | |
2271 | did the easy thing, and turned it off. | |
2272 | ||
2273 | @node Extended Asm | |
2274 | @section Assembler Instructions with C Expression Operands | |
2275 | @cindex extended @code{asm} | |
2276 | @cindex @code{asm} expressions | |
2277 | @cindex assembler instructions | |
2278 | @cindex registers | |
2279 | ||
2280 | In an assembler instruction using @code{asm}, you can now specify the | |
2281 | operands of the instruction using C expressions. This means no more | |
2282 | guessing which registers or memory locations will contain the data you want | |
2283 | to use. | |
2284 | ||
2285 | You must specify an assembler instruction template much like what appears | |
2286 | in a machine description, plus an operand constraint string for each | |
2287 | operand. | |
2288 | ||
2289 | For example, here is how to use the 68881's @code{fsinx} instruction: | |
2290 | ||
2291 | @example | |
2292 | asm ("fsinx %1,%0" : "=f" (result) : "f" (angle)); | |
2293 | @end example | |
2294 | ||
2295 | @noindent | |
2296 | Here @code{angle} is the C expression for the input operand while | |
2297 | @code{result} is that of the output operand. Each has @samp{"f"} as its | |
2298 | operand constraint, saying that a floating point register is required. The | |
2299 | @samp{=} in @samp{=f} indicates that the operand is an output; all output | |
2300 | operands' constraints must use @samp{=}. The constraints use the same | |
2301 | language used in the machine description (@pxref{Constraints}). | |
2302 | ||
2303 | Each operand is described by an operand-constraint string followed by the C | |
2304 | expression in parentheses. A colon separates the assembler template from | |
2305 | the first output operand, and another separates the last output operand | |
2306 | from the first input, if any. Commas separate output operands and separate | |
2307 | inputs. The total number of operands is limited to ten or to the maximum | |
2308 | number of operands in any instruction pattern in the machine description, | |
2309 | whichever is greater. | |
2310 | ||
2311 | If there are no output operands, and there are input operands, then there | |
2312 | must be two consecutive colons surrounding the place where the output | |
2313 | operands would go. | |
2314 | ||
2315 | Output operand expressions must be lvalues; the compiler can check this. | |
2316 | The input operands need not be lvalues. The compiler cannot check whether | |
2317 | the operands have data types that are reasonable for the instruction being | |
2318 | executed. It does not parse the assembler instruction template and does | |
2319 | not know what it means, or whether it is valid assembler input. The | |
2320 | extended @code{asm} feature is most often used for machine instructions | |
2321 | that the compiler itself does not know exist. If the output expression | |
2322 | cannot be directly addressed (for example, it is a bit field), your | |
2323 | constraint must allow a register. In that case, GNU CC will use | |
2324 | the register as the output of the @code{asm}, and then store that | |
2325 | register into the output. | |
2326 | ||
2327 | The ordinary output operands must be write-only; GNU CC will assume | |
2328 | that the values in these operands before the instruction are dead and | |
2329 | need not be generated. Extended asm supports input-output or | |
2330 | read-write operands. Use the constraint character @samp{+} to indicate | |
2331 | such an operand and list it with the output operands. | |
2332 | ||
2333 | When the constraints for the read-write operand | |
2334 | (or the operand in which only some of the bits are to be changed) | |
2335 | allows a register, you may, as an alternative, logically | |
2336 | split its function into two separate operands, one input operand and one | |
2337 | write-only output operand. The connection between them is expressed by | |
2338 | constraints which say they need to be in the same location when the | |
2339 | instruction executes. You can use the same C expression for both | |
2340 | operands, or different expressions. For example, here we write the | |
2341 | (fictitious) @samp{combine} instruction with @code{bar} as its read-only | |
2342 | source operand and @code{foo} as its read-write destination: | |
2343 | ||
2344 | @example | |
2345 | asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar)); | |
2346 | @end example | |
2347 | ||
2348 | @noindent | |
2349 | The constraint @samp{"0"} for operand 1 says that it must occupy the same | |
2350 | location as operand 0. A digit in constraint is allowed only in an input | |
2351 | operand, and it must refer to an output operand. | |
2352 | ||
2353 | Only a digit in the constraint can guarantee that one operand will be in | |
2354 | the same place as another. The mere fact that @code{foo} is the value of | |
2355 | both operands is not enough to guarantee that they will be in the same | |
2356 | place in the generated assembler code. The following would not work: | |
2357 | ||
2358 | @example | |
2359 | asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar)); | |
2360 | @end example | |
2361 | ||
2362 | Various optimizations or reloading could cause operands 0 and 1 to be in | |
2363 | different registers; GNU CC knows no reason not to do so. For example, the | |
2364 | compiler might find a copy of the value of @code{foo} in one register and | |
2365 | use it for operand 1, but generate the output operand 0 in a different | |
2366 | register (copying it afterward to @code{foo}'s own address). Of course, | |
2367 | since the register for operand 1 is not even mentioned in the assembler | |
2368 | code, the result will not work, but GNU CC can't tell that. | |
2369 | ||
2370 | Some instructions clobber specific hard registers. To describe this, write | |
2371 | a third colon after the input operands, followed by the names of the | |
2372 | clobbered hard registers (given as strings). Here is a realistic example | |
2373 | for the Vax: | |
2374 | ||
2375 | @example | |
2376 | asm volatile ("movc3 %0,%1,%2" | |
2377 | : /* no outputs */ | |
2378 | : "g" (from), "g" (to), "g" (count) | |
2379 | : "r0", "r1", "r2", "r3", "r4", "r5"); | |
2380 | @end example | |
2381 | ||
2382 | If you refer to a particular hardware register from the assembler code, | |
2383 | then you will probably have to list the register after the third colon | |
2384 | to tell the compiler that the register's value is modified. In many | |
2385 | assemblers, the register names begin with @samp{%}; to produce one | |
2386 | @samp{%} in the assembler code, you must write @samp{%%} in the input. | |
2387 | ||
2388 | If your assembler instruction can alter the condition code register, | |
2389 | add @samp{cc} to the list of clobbered registers. GNU CC on some | |
2390 | machines represents the condition codes as a specific hardware | |
2391 | register; @samp{cc} serves to name this register. On other machines, | |
2392 | the condition code is handled differently, and specifying @samp{cc} | |
2393 | has no effect. But it is valid no matter what the machine. | |
2394 | ||
2395 | If your assembler instruction modifies memory in an unpredictable | |
2396 | fashion, add @samp{memory} to the list of clobbered registers. | |
2397 | This will cause GNU CC to not keep memory values cached in | |
2398 | registers across the assembler instruction. | |
2399 | ||
2400 | You can put multiple assembler instructions together in a single @code{asm} | |
2401 | template, separated either with newlines (written as @samp{\n}) or with | |
2402 | semicolons if the assembler allows such semicolons. The GNU assembler | |
2403 | allows semicolons and all Unix assemblers seem to do so. The input | |
2404 | operands are guaranteed not to use any of the clobbered registers, and | |
2405 | neither will the output operands' addresses, so you can read and write the | |
2406 | clobbered registers as many times as you like. Here is an example of | |
2407 | multiple instructions in a template; it assumes that the subroutine | |
2408 | @code{_foo} accepts arguments in registers 9 and 10: | |
2409 | ||
2410 | @example | |
2411 | asm ("movl %0,r9;movl %1,r10;call _foo" | |
2412 | : /* no outputs */ | |
2413 | : "g" (from), "g" (to) | |
2414 | : "r9", "r10"); | |
2415 | @end example | |
2416 | ||
2417 | Unless an output operand has the @samp{&} constraint modifier, GNU CC may | |
2418 | allocate it in the same register as an unrelated input operand, on the | |
2419 | assumption that the inputs are consumed before the outputs are produced. | |
2420 | This assumption may be false if the assembler code actually consists of | |
2421 | more than one instruction. In such a case, use @samp{&} for each output | |
2422 | operand that may not overlap an input. | |
2423 | @xref{Modifiers}. | |
2424 | ||
2425 | If you want to test the condition code produced by an assembler instruction, | |
2426 | you must include a branch and a label in the @code{asm} construct, as follows: | |
2427 | ||
2428 | @example | |
2429 | asm ("clr %0;frob %1;beq 0f;mov #1,%0;0:" | |
2430 | : "g" (result) | |
2431 | : "g" (input)); | |
2432 | @end example | |
2433 | ||
2434 | @noindent | |
2435 | This assumes your assembler supports local labels, as the GNU assembler | |
2436 | and most Unix assemblers do. | |
2437 | ||
2438 | Speaking of labels, jumps from one @code{asm} to another are not | |
2439 | supported. The compiler's optimizers do not know about these jumps, | |
2440 | and therefore they cannot take account of them when deciding how to | |
2441 | optimize. | |
2442 | ||
2443 | @cindex macros containing @code{asm} | |
2444 | Usually the most convenient way to use these @code{asm} instructions is to | |
2445 | encapsulate them in macros that look like functions. For example, | |
2446 | ||
2447 | @example | |
2448 | #define sin(x) \ | |
2449 | (@{ double __value, __arg = (x); \ | |
2450 | asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \ | |
2451 | __value; @}) | |
2452 | @end example | |
2453 | ||
2454 | @noindent | |
2455 | Here the variable @code{__arg} is used to make sure that the instruction | |
2456 | operates on a proper @code{double} value, and to accept only those | |
2457 | arguments @code{x} which can convert automatically to a @code{double}. | |
2458 | ||
2459 | Another way to make sure the instruction operates on the correct data type | |
2460 | is to use a cast in the @code{asm}. This is different from using a | |
2461 | variable @code{__arg} in that it converts more different types. For | |
2462 | example, if the desired type were @code{int}, casting the argument to | |
2463 | @code{int} would accept a pointer with no complaint, while assigning the | |
2464 | argument to an @code{int} variable named @code{__arg} would warn about | |
2465 | using a pointer unless the caller explicitly casts it. | |
2466 | ||
2467 | If an @code{asm} has output operands, GNU CC assumes for optimization | |
2468 | purposes that the instruction has no side effects except to change the | |
2469 | output operands. This does not mean that instructions with a side effect | |
2470 | cannot be used, but you must be careful, because the compiler may eliminate | |
2471 | them if the output operands aren't used, or move them out of loops, or | |
2472 | replace two with one if they constitute a common subexpression. Also, if | |
2473 | your instruction does have a side effect on a variable that otherwise | |
2474 | appears not to change, the old value of the variable may be reused later if | |
2475 | it happens to be found in a register. | |
2476 | ||
2477 | You can prevent an @code{asm} instruction from being deleted, moved | |
2478 | significantly, or combined, by writing the keyword @code{volatile} after | |
2479 | the @code{asm}. For example: | |
2480 | ||
2481 | @example | |
2482 | #define set_priority(x) \ | |
2483 | asm volatile ("set_priority %0": /* no outputs */ : "g" (x)) | |
2484 | @end example | |
2485 | ||
2486 | @noindent | |
2487 | An instruction without output operands will not be deleted or moved | |
2488 | significantly, regardless, unless it is unreachable. | |
2489 | ||
2490 | Note that even a volatile @code{asm} instruction can be moved in ways | |
2491 | that appear insignificant to the compiler, such as across jump | |
2492 | instructions. You can't expect a sequence of volatile @code{asm} | |
2493 | instructions to remain perfectly consecutive. If you want consecutive | |
2494 | output, use a single @code{asm}. | |
2495 | ||
2496 | It is a natural idea to look for a way to give access to the condition | |
2497 | code left by the assembler instruction. However, when we attempted to | |
2498 | implement this, we found no way to make it work reliably. The problem | |
2499 | is that output operands might need reloading, which would result in | |
2500 | additional following ``store'' instructions. On most machines, these | |
2501 | instructions would alter the condition code before there was time to | |
2502 | test it. This problem doesn't arise for ordinary ``test'' and | |
2503 | ``compare'' instructions because they don't have any output operands. | |
2504 | ||
2505 | If you are writing a header file that should be includable in ANSI C | |
2506 | programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate | |
2507 | Keywords}. | |
2508 | ||
2509 | @ifclear INTERNALS | |
2510 | @c Show the details on constraints if they do not appear elsewhere in | |
2511 | @c the manual | |
2512 | @include md.texi | |
2513 | @end ifclear | |
2514 | ||
2515 | @node Asm Labels | |
2516 | @section Controlling Names Used in Assembler Code | |
2517 | @cindex assembler names for identifiers | |
2518 | @cindex names used in assembler code | |
2519 | @cindex identifiers, names in assembler code | |
2520 | ||
2521 | You can specify the name to be used in the assembler code for a C | |
2522 | function or variable by writing the @code{asm} (or @code{__asm__}) | |
2523 | keyword after the declarator as follows: | |
2524 | ||
2525 | @example | |
2526 | int foo asm ("myfoo") = 2; | |
2527 | @end example | |
2528 | ||
2529 | @noindent | |
2530 | This specifies that the name to be used for the variable @code{foo} in | |
2531 | the assembler code should be @samp{myfoo} rather than the usual | |
2532 | @samp{_foo}. | |
2533 | ||
2534 | On systems where an underscore is normally prepended to the name of a C | |
2535 | function or variable, this feature allows you to define names for the | |
2536 | linker that do not start with an underscore. | |
2537 | ||
2538 | You cannot use @code{asm} in this way in a function @emph{definition}; but | |
2539 | you can get the same effect by writing a declaration for the function | |
2540 | before its definition and putting @code{asm} there, like this: | |
2541 | ||
2542 | @example | |
2543 | extern func () asm ("FUNC"); | |
2544 | ||
2545 | func (x, y) | |
2546 | int x, y; | |
2547 | @dots{} | |
2548 | @end example | |
2549 | ||
2550 | It is up to you to make sure that the assembler names you choose do not | |
2551 | conflict with any other assembler symbols. Also, you must not use a | |
2552 | register name; that would produce completely invalid assembler code. GNU | |
2553 | CC does not as yet have the ability to store static variables in registers. | |
2554 | Perhaps that will be added. | |
2555 | ||
2556 | @node Explicit Reg Vars | |
2557 | @section Variables in Specified Registers | |
2558 | @cindex explicit register variables | |
2559 | @cindex variables in specified registers | |
2560 | @cindex specified registers | |
2561 | @cindex registers, global allocation | |
2562 | ||
2563 | GNU C allows you to put a few global variables into specified hardware | |
2564 | registers. You can also specify the register in which an ordinary | |
2565 | register variable should be allocated. | |
2566 | ||
2567 | @itemize @bullet | |
2568 | @item | |
2569 | Global register variables reserve registers throughout the program. | |
2570 | This may be useful in programs such as programming language | |
2571 | interpreters which have a couple of global variables that are accessed | |
2572 | very often. | |
2573 | ||
2574 | @item | |
2575 | Local register variables in specific registers do not reserve the | |
2576 | registers. The compiler's data flow analysis is capable of determining | |
2577 | where the specified registers contain live values, and where they are | |
2578 | available for other uses. | |
2579 | ||
2580 | These local variables are sometimes convenient for use with the extended | |
2581 | @code{asm} feature (@pxref{Extended Asm}), if you want to write one | |
2582 | output of the assembler instruction directly into a particular register. | |
2583 | (This will work provided the register you specify fits the constraints | |
2584 | specified for that operand in the @code{asm}.) | |
2585 | @end itemize | |
2586 | ||
2587 | @menu | |
2588 | * Global Reg Vars:: | |
2589 | * Local Reg Vars:: | |
2590 | @end menu | |
2591 | ||
2592 | @node Global Reg Vars | |
2593 | @subsection Defining Global Register Variables | |
2594 | @cindex global register variables | |
2595 | @cindex registers, global variables in | |
2596 | ||
2597 | You can define a global register variable in GNU C like this: | |
2598 | ||
2599 | @example | |
2600 | register int *foo asm ("a5"); | |
2601 | @end example | |
2602 | ||
2603 | @noindent | |
2604 | Here @code{a5} is the name of the register which should be used. Choose a | |
2605 | register which is normally saved and restored by function calls on your | |
2606 | machine, so that library routines will not clobber it. | |
2607 | ||
2608 | Naturally the register name is cpu-dependent, so you would need to | |
2609 | conditionalize your program according to cpu type. The register | |
2610 | @code{a5} would be a good choice on a 68000 for a variable of pointer | |
2611 | type. On machines with register windows, be sure to choose a ``global'' | |
2612 | register that is not affected magically by the function call mechanism. | |
2613 | ||
2614 | In addition, operating systems on one type of cpu may differ in how they | |
2615 | name the registers; then you would need additional conditionals. For | |
2616 | example, some 68000 operating systems call this register @code{%a5}. | |
2617 | ||
2618 | Eventually there may be a way of asking the compiler to choose a register | |
2619 | automatically, but first we need to figure out how it should choose and | |
2620 | how to enable you to guide the choice. No solution is evident. | |
2621 | ||
2622 | Defining a global register variable in a certain register reserves that | |
2623 | register entirely for this use, at least within the current compilation. | |
2624 | The register will not be allocated for any other purpose in the functions | |
2625 | in the current compilation. The register will not be saved and restored by | |
2626 | these functions. Stores into this register are never deleted even if they | |
2627 | would appear to be dead, but references may be deleted or moved or | |
2628 | simplified. | |
2629 | ||
2630 | It is not safe to access the global register variables from signal | |
2631 | handlers, or from more than one thread of control, because the system | |
2632 | library routines may temporarily use the register for other things (unless | |
2633 | you recompile them specially for the task at hand). | |
2634 | ||
2635 | @cindex @code{qsort}, and global register variables | |
2636 | It is not safe for one function that uses a global register variable to | |
2637 | call another such function @code{foo} by way of a third function | |
2638 | @code{lose} that was compiled without knowledge of this variable (i.e. in a | |
2639 | different source file in which the variable wasn't declared). This is | |
2640 | because @code{lose} might save the register and put some other value there. | |
2641 | For example, you can't expect a global register variable to be available in | |
2642 | the comparison-function that you pass to @code{qsort}, since @code{qsort} | |
2643 | might have put something else in that register. (If you are prepared to | |
2644 | recompile @code{qsort} with the same global register variable, you can | |
2645 | solve this problem.) | |
2646 | ||
2647 | If you want to recompile @code{qsort} or other source files which do not | |
2648 | actually use your global register variable, so that they will not use that | |
2649 | register for any other purpose, then it suffices to specify the compiler | |
2650 | option @samp{-ffixed-@var{reg}}. You need not actually add a global | |
2651 | register declaration to their source code. | |
2652 | ||
2653 | A function which can alter the value of a global register variable cannot | |
2654 | safely be called from a function compiled without this variable, because it | |
2655 | could clobber the value the caller expects to find there on return. | |
2656 | Therefore, the function which is the entry point into the part of the | |
2657 | program that uses the global register variable must explicitly save and | |
2658 | restore the value which belongs to its caller. | |
2659 | ||
2660 | @cindex register variable after @code{longjmp} | |
2661 | @cindex global register after @code{longjmp} | |
2662 | @cindex value after @code{longjmp} | |
2663 | @findex longjmp | |
2664 | @findex setjmp | |
2665 | On most machines, @code{longjmp} will restore to each global register | |
2666 | variable the value it had at the time of the @code{setjmp}. On some | |
2667 | machines, however, @code{longjmp} will not change the value of global | |
2668 | register variables. To be portable, the function that called @code{setjmp} | |
2669 | should make other arrangements to save the values of the global register | |
2670 | variables, and to restore them in a @code{longjmp}. This way, the same | |
2671 | thing will happen regardless of what @code{longjmp} does. | |
2672 | ||
2673 | All global register variable declarations must precede all function | |
2674 | definitions. If such a declaration could appear after function | |
2675 | definitions, the declaration would be too late to prevent the register from | |
2676 | being used for other purposes in the preceding functions. | |
2677 | ||
2678 | Global register variables may not have initial values, because an | |
2679 | executable file has no means to supply initial contents for a register. | |
2680 | ||
2681 | On the Sparc, there are reports that g3 @dots{} g7 are suitable | |
2682 | registers, but certain library functions, such as @code{getwd}, as well | |
2683 | as the subroutines for division and remainder, modify g3 and g4. g1 and | |
2684 | g2 are local temporaries. | |
2685 | ||
2686 | On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7. | |
2687 | Of course, it will not do to use more than a few of those. | |
2688 | ||
2689 | @node Local Reg Vars | |
2690 | @subsection Specifying Registers for Local Variables | |
2691 | @cindex local variables, specifying registers | |
2692 | @cindex specifying registers for local variables | |
2693 | @cindex registers for local variables | |
2694 | ||
2695 | You can define a local register variable with a specified register | |
2696 | like this: | |
2697 | ||
2698 | @example | |
2699 | register int *foo asm ("a5"); | |
2700 | @end example | |
2701 | ||
2702 | @noindent | |
2703 | Here @code{a5} is the name of the register which should be used. Note | |
2704 | that this is the same syntax used for defining global register | |
2705 | variables, but for a local variable it would appear within a function. | |
2706 | ||
2707 | Naturally the register name is cpu-dependent, but this is not a | |
2708 | problem, since specific registers are most often useful with explicit | |
2709 | assembler instructions (@pxref{Extended Asm}). Both of these things | |
2710 | generally require that you conditionalize your program according to | |
2711 | cpu type. | |
2712 | ||
2713 | In addition, operating systems on one type of cpu may differ in how they | |
2714 | name the registers; then you would need additional conditionals. For | |
2715 | example, some 68000 operating systems call this register @code{%a5}. | |
2716 | ||
2717 | Eventually there may be a way of asking the compiler to choose a register | |
2718 | automatically, but first we need to figure out how it should choose and | |
2719 | how to enable you to guide the choice. No solution is evident. | |
2720 | ||
2721 | Defining such a register variable does not reserve the register; it | |
2722 | remains available for other uses in places where flow control determines | |
2723 | the variable's value is not live. However, these registers are made | |
2724 | unavailable for use in the reload pass. I would not be surprised if | |
2725 | excessive use of this feature leaves the compiler too few available | |
2726 | registers to compile certain functions. | |
2727 | ||
2728 | @node Alternate Keywords | |
2729 | @section Alternate Keywords | |
2730 | @cindex alternate keywords | |
2731 | @cindex keywords, alternate | |
2732 | ||
2733 | The option @samp{-traditional} disables certain keywords; @samp{-ansi} | |
2734 | disables certain others. This causes trouble when you want to use GNU C | |
2735 | extensions, or ANSI C features, in a general-purpose header file that | |
2736 | should be usable by all programs, including ANSI C programs and traditional | |
2737 | ones. The keywords @code{asm}, @code{typeof} and @code{inline} cannot be | |
2738 | used since they won't work in a program compiled with @samp{-ansi}, while | |
2739 | the keywords @code{const}, @code{volatile}, @code{signed}, @code{typeof} | |
2740 | and @code{inline} won't work in a program compiled with | |
2741 | @samp{-traditional}.@refill | |
2742 | ||
2743 | The way to solve these problems is to put @samp{__} at the beginning and | |
2744 | end of each problematical keyword. For example, use @code{__asm__} | |
2745 | instead of @code{asm}, @code{__const__} instead of @code{const}, and | |
2746 | @code{__inline__} instead of @code{inline}. | |
2747 | ||
2748 | Other C compilers won't accept these alternative keywords; if you want to | |
2749 | compile with another compiler, you can define the alternate keywords as | |
2750 | macros to replace them with the customary keywords. It looks like this: | |
2751 | ||
2752 | @example | |
2753 | #ifndef __GNUC__ | |
2754 | #define __asm__ asm | |
2755 | #endif | |
2756 | @end example | |
2757 | ||
2758 | @samp{-pedantic} causes warnings for many GNU C extensions. You can | |
2759 | prevent such warnings within one expression by writing | |
2760 | @code{__extension__} before the expression. @code{__extension__} has no | |
2761 | effect aside from this. | |
2762 | ||
2763 | @node Incomplete Enums | |
2764 | @section Incomplete @code{enum} Types | |
2765 | ||
2766 | You can define an @code{enum} tag without specifying its possible values. | |
2767 | This results in an incomplete type, much like what you get if you write | |
2768 | @code{struct foo} without describing the elements. A later declaration | |
2769 | which does specify the possible values completes the type. | |
2770 | ||
2771 | You can't allocate variables or storage using the type while it is | |
2772 | incomplete. However, you can work with pointers to that type. | |
2773 | ||
2774 | This extension may not be very useful, but it makes the handling of | |
2775 | @code{enum} more consistent with the way @code{struct} and @code{union} | |
2776 | are handled. | |
2777 | ||
2778 | This extension is not supported by GNU C++. | |
2779 | ||
2780 | @node Function Names | |
2781 | @section Function Names as Strings | |
2782 | ||
2783 | GNU CC predefines two string variables to be the name of the current function. | |
2784 | The variable @code{__FUNCTION__} is the name of the function as it appears | |
2785 | in the source. The variable @code{__PRETTY_FUNCTION__} is the name of | |
2786 | the function pretty printed in a language specific fashion. | |
2787 | ||
2788 | These names are always the same in a C function, but in a C++ function | |
2789 | they may be different. For example, this program: | |
2790 | ||
2791 | @smallexample | |
2792 | extern "C" @{ | |
2793 | extern int printf (char *, ...); | |
2794 | @} | |
2795 | ||
2796 | class a @{ | |
2797 | public: | |
2798 | sub (int i) | |
2799 | @{ | |
2800 | printf ("__FUNCTION__ = %s\n", __FUNCTION__); | |
2801 | printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__); | |
2802 | @} | |
2803 | @}; | |
2804 | ||
2805 | int | |
2806 | main (void) | |
2807 | @{ | |
2808 | a ax; | |
2809 | ax.sub (0); | |
2810 | return 0; | |
2811 | @} | |
2812 | @end smallexample | |
2813 | ||
2814 | @noindent | |
2815 | gives this output: | |
2816 | ||
2817 | @smallexample | |
2818 | __FUNCTION__ = sub | |
2819 | __PRETTY_FUNCTION__ = int a::sub (int) | |
2820 | @end smallexample | |
2821 | ||
2822 | These names are not macros: they are predefined string variables. | |
2823 | For example, @samp{#ifdef __FUNCTION__} does not have any special | |
2824 | meaning inside a function, since the preprocessor does not do anything | |
2825 | special with the identifier @code{__FUNCTION__}. | |
2826 | ||
2827 | @node Return Address | |
2828 | @section Getting the Return or Frame Address of a Function | |
2829 | ||
2830 | These functions may be used to get information about the callers of a | |
2831 | function. | |
2832 | ||
2833 | @table @code | |
2834 | @item __builtin_return_address (@var{level}) | |
2835 | This function returns the return address of the current function, or of | |
2836 | one of its callers. The @var{level} argument is number of frames to | |
2837 | scan up the call stack. A value of @code{0} yields the return address | |
2838 | of the current function, a value of @code{1} yields the return address | |
2839 | of the caller of the current function, and so forth. | |
2840 | ||
2841 | The @var{level} argument must be a constant integer. | |
2842 | ||
2843 | On some machines it may be impossible to determine the return address of | |
2844 | any function other than the current one; in such cases, or when the top | |
2845 | of the stack has been reached, this function will return @code{0}. | |
2846 | ||
2847 | This function should only be used with a non-zero argument for debugging | |
2848 | purposes. | |
2849 | ||
2850 | @item __builtin_frame_address (@var{level}) | |
2851 | This function is similar to @code{__builtin_return_address}, but it | |
2852 | returns the address of the function frame rather than the return address | |
2853 | of the function. Calling @code{__builtin_frame_address} with a value of | |
2854 | @code{0} yields the frame address of the current function, a value of | |
2855 | @code{1} yields the frame address of the caller of the current function, | |
2856 | and so forth. | |
2857 | ||
2858 | The frame is the area on the stack which holds local variables and saved | |
2859 | registers. The frame address is normally the address of the first word | |
2860 | pushed on to the stack by the function. However, the exact definition | |
2861 | depends upon the processor and the calling convention. If the processor | |
2862 | has a dedicated frame pointer register, and the function has a frame, | |
2863 | then @code{__builtin_frame_address} will return the value of the frame | |
2864 | pointer register. | |
2865 | ||
2866 | The caveats that apply to @code{__builtin_return_address} apply to this | |
2867 | function as well. | |
2868 | @end table | |
2869 | ||
2870 | @node C++ Extensions | |
2871 | @chapter Extensions to the C++ Language | |
2872 | @cindex extensions, C++ language | |
2873 | @cindex C++ language extensions | |
2874 | ||
2875 | The GNU compiler provides these extensions to the C++ language (and you | |
2876 | can also use most of the C language extensions in your C++ programs). If you | |
2877 | want to write code that checks whether these features are available, you can | |
2878 | test for the GNU compiler the same way as for C programs: check for a | |
2879 | predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to | |
2880 | test specifically for GNU C++ (@pxref{Standard Predefined,,Standard | |
2881 | Predefined Macros,cpp.info,The C Preprocessor}). | |
2882 | ||
2883 | @menu | |
2884 | * Naming Results:: Giving a name to C++ function return values. | |
2885 | * Min and Max:: C++ Minimum and maximum operators. | |
2886 | * Destructors and Goto:: Goto is safe to use in C++ even when destructors | |
2887 | are needed. | |
2888 | * C++ Interface:: You can use a single C++ header file for both | |
2889 | declarations and definitions. | |
2890 | * Template Instantiation:: Methods for ensuring that exactly one copy of | |
2891 | each needed template instantiation is emitted. | |
2892 | * C++ Signatures:: You can specify abstract types to get subtype | |
2893 | polymorphism independent from inheritance. | |
2894 | @end menu | |
2895 | ||
2896 | @node Naming Results | |
2897 | @section Named Return Values in C++ | |
2898 | ||
2899 | @cindex @code{return}, in C++ function header | |
2900 | @cindex return value, named, in C++ | |
2901 | @cindex named return value in C++ | |
2902 | @cindex C++ named return value | |
2903 | GNU C++ extends the function-definition syntax to allow you to specify a | |
2904 | name for the result of a function outside the body of the definition, in | |
2905 | C++ programs: | |
2906 | ||
2907 | @example | |
2908 | @group | |
2909 | @var{type} | |
2910 | @var{functionname} (@var{args}) return @var{resultname}; | |
2911 | @{ | |
2912 | @dots{} | |
2913 | @var{body} | |
2914 | @dots{} | |
2915 | @} | |
2916 | @end group | |
2917 | @end example | |
2918 | ||
2919 | You can use this feature to avoid an extra constructor call when | |
2920 | a function result has a class type. For example, consider a function | |
2921 | @code{m}, declared as @w{@samp{X v = m ();}}, whose result is of class | |
2922 | @code{X}: | |
2923 | ||
2924 | @example | |
2925 | X | |
2926 | m () | |
2927 | @{ | |
2928 | X b; | |
2929 | b.a = 23; | |
2930 | return b; | |
2931 | @} | |
2932 | @end example | |
2933 | ||
2934 | @cindex implicit argument: return value | |
2935 | Although @code{m} appears to have no arguments, in fact it has one implicit | |
2936 | argument: the address of the return value. At invocation, the address | |
2937 | of enough space to hold @code{v} is sent in as the implicit argument. | |
2938 | Then @code{b} is constructed and its @code{a} field is set to the value | |
2939 | 23. Finally, a copy constructor (a constructor of the form @samp{X(X&)}) | |
2940 | is applied to @code{b}, with the (implicit) return value location as the | |
2941 | target, so that @code{v} is now bound to the return value. | |
2942 | ||
2943 | But this is wasteful. The local @code{b} is declared just to hold | |
2944 | something that will be copied right out. While a compiler that | |
2945 | combined an ``elision'' algorithm with interprocedural data flow | |
2946 | analysis could conceivably eliminate all of this, it is much more | |
2947 | practical to allow you to assist the compiler in generating | |
2948 | efficient code by manipulating the return value explicitly, | |
2949 | thus avoiding the local variable and copy constructor altogether. | |
2950 | ||
2951 | Using the extended GNU C++ function-definition syntax, you can avoid the | |
2952 | temporary allocation and copying by naming @code{r} as your return value | |
2953 | at the outset, and assigning to its @code{a} field directly: | |
2954 | ||
2955 | @example | |
2956 | X | |
2957 | m () return r; | |
2958 | @{ | |
2959 | r.a = 23; | |
2960 | @} | |
2961 | @end example | |
2962 | ||
2963 | @noindent | |
2964 | The declaration of @code{r} is a standard, proper declaration, whose effects | |
2965 | are executed @strong{before} any of the body of @code{m}. | |
2966 | ||
2967 | Functions of this type impose no additional restrictions; in particular, | |
2968 | you can execute @code{return} statements, or return implicitly by | |
2969 | reaching the end of the function body (``falling off the edge''). | |
2970 | Cases like | |
2971 | ||
2972 | @example | |
2973 | X | |
2974 | m () return r (23); | |
2975 | @{ | |
2976 | return; | |
2977 | @} | |
2978 | @end example | |
2979 | ||
2980 | @noindent | |
2981 | (or even @w{@samp{X m () return r (23); @{ @}}}) are unambiguous, since | |
2982 | the return value @code{r} has been initialized in either case. The | |
2983 | following code may be hard to read, but also works predictably: | |
2984 | ||
2985 | @example | |
2986 | X | |
2987 | m () return r; | |
2988 | @{ | |
2989 | X b; | |
2990 | return b; | |
2991 | @} | |
2992 | @end example | |
2993 | ||
2994 | The return value slot denoted by @code{r} is initialized at the outset, | |
2995 | but the statement @samp{return b;} overrides this value. The compiler | |
2996 | deals with this by destroying @code{r} (calling the destructor if there | |
2997 | is one, or doing nothing if there is not), and then reinitializing | |
2998 | @code{r} with @code{b}. | |
2999 | ||
3000 | This extension is provided primarily to help people who use overloaded | |
3001 | operators, where there is a great need to control not just the | |
3002 | arguments, but the return values of functions. For classes where the | |
3003 | copy constructor incurs a heavy performance penalty (especially in the | |
3004 | common case where there is a quick default constructor), this is a major | |
3005 | savings. The disadvantage of this extension is that you do not control | |
3006 | when the default constructor for the return value is called: it is | |
3007 | always called at the beginning. | |
3008 | ||
3009 | @node Min and Max | |
3010 | @section Minimum and Maximum Operators in C++ | |
3011 | ||
3012 | It is very convenient to have operators which return the ``minimum'' or the | |
3013 | ``maximum'' of two arguments. In GNU C++ (but not in GNU C), | |
3014 | ||
3015 | @table @code | |
3016 | @item @var{a} <? @var{b} | |
3017 | @findex <? | |
3018 | @cindex minimum operator | |
3019 | is the @dfn{minimum}, returning the smaller of the numeric values | |
3020 | @var{a} and @var{b}; | |
3021 | ||
3022 | @item @var{a} >? @var{b} | |
3023 | @findex >? | |
3024 | @cindex maximum operator | |
3025 | is the @dfn{maximum}, returning the larger of the numeric values @var{a} | |
3026 | and @var{b}. | |
3027 | @end table | |
3028 | ||
3029 | These operations are not primitive in ordinary C++, since you can | |
3030 | use a macro to return the minimum of two things in C++, as in the | |
3031 | following example. | |
3032 | ||
3033 | @example | |
3034 | #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y)) | |
3035 | @end example | |
3036 | ||
3037 | @noindent | |
3038 | You might then use @w{@samp{int min = MIN (i, j);}} to set @var{min} to | |
3039 | the minimum value of variables @var{i} and @var{j}. | |
3040 | ||
3041 | However, side effects in @code{X} or @code{Y} may cause unintended | |
3042 | behavior. For example, @code{MIN (i++, j++)} will fail, incrementing | |
3043 | the smaller counter twice. A GNU C extension allows you to write safe | |
3044 | macros that avoid this kind of problem (@pxref{Naming Types,,Naming an | |
3045 | Expression's Type}). However, writing @code{MIN} and @code{MAX} as | |
3046 | macros also forces you to use function-call notation for a | |
3047 | fundamental arithmetic operation. Using GNU C++ extensions, you can | |
3048 | write @w{@samp{int min = i <? j;}} instead. | |
3049 | ||
3050 | Since @code{<?} and @code{>?} are built into the compiler, they properly | |
3051 | handle expressions with side-effects; @w{@samp{int min = i++ <? j++;}} | |
3052 | works correctly. | |
3053 | ||
3054 | @node Destructors and Goto | |
3055 | @section @code{goto} and Destructors in GNU C++ | |
3056 | ||
3057 | @cindex @code{goto} in C++ | |
3058 | @cindex destructors vs @code{goto} | |
3059 | In C++ programs, you can safely use the @code{goto} statement. When you | |
3060 | use it to exit a block which contains aggregates requiring destructors, | |
3061 | the destructors will run before the @code{goto} transfers control. | |
3062 | ||
3063 | @cindex constructors vs @code{goto} | |
3064 | The compiler still forbids using @code{goto} to @emph{enter} a scope | |
3065 | that requires constructors. | |
3066 | ||
3067 | @node C++ Interface | |
3068 | @section Declarations and Definitions in One Header | |
3069 | ||
3070 | @cindex interface and implementation headers, C++ | |
3071 | @cindex C++ interface and implementation headers | |
3072 | C++ object definitions can be quite complex. In principle, your source | |
3073 | code will need two kinds of things for each object that you use across | |
3074 | more than one source file. First, you need an @dfn{interface} | |
3075 | specification, describing its structure with type declarations and | |
3076 | function prototypes. Second, you need the @dfn{implementation} itself. | |
3077 | It can be tedious to maintain a separate interface description in a | |
3078 | header file, in parallel to the actual implementation. It is also | |
3079 | dangerous, since separate interface and implementation definitions may | |
3080 | not remain parallel. | |
3081 | ||
3082 | @cindex pragmas, interface and implementation | |
3083 | With GNU C++, you can use a single header file for both purposes. | |
3084 | ||
3085 | @quotation | |
3086 | @emph{Warning:} The mechanism to specify this is in transition. For the | |
3087 | nonce, you must use one of two @code{#pragma} commands; in a future | |
3088 | release of GNU C++, an alternative mechanism will make these | |
3089 | @code{#pragma} commands unnecessary. | |
3090 | @end quotation | |
3091 | ||
3092 | The header file contains the full definitions, but is marked with | |
3093 | @samp{#pragma interface} in the source code. This allows the compiler | |
3094 | to use the header file only as an interface specification when ordinary | |
3095 | source files incorporate it with @code{#include}. In the single source | |
3096 | file where the full implementation belongs, you can use either a naming | |
3097 | convention or @samp{#pragma implementation} to indicate this alternate | |
3098 | use of the header file. | |
3099 | ||
3100 | @table @code | |
3101 | @item #pragma interface | |
3102 | @itemx #pragma interface "@var{subdir}/@var{objects}.h" | |
3103 | @kindex #pragma interface | |
3104 | Use this directive in @emph{header files} that define object classes, to save | |
3105 | space in most of the object files that use those classes. Normally, | |
3106 | local copies of certain information (backup copies of inline member | |
3107 | functions, debugging information, and the internal tables that implement | |
3108 | virtual functions) must be kept in each object file that includes class | |
3109 | definitions. You can use this pragma to avoid such duplication. When a | |
3110 | header file containing @samp{#pragma interface} is included in a | |
3111 | compilation, this auxiliary information will not be generated (unless | |
3112 | the main input source file itself uses @samp{#pragma implementation}). | |
3113 | Instead, the object files will contain references to be resolved at link | |
3114 | time. | |
3115 | ||
3116 | The second form of this directive is useful for the case where you have | |
3117 | multiple headers with the same name in different directories. If you | |
3118 | use this form, you must specify the same string to @samp{#pragma | |
3119 | implementation}. | |
3120 | ||
3121 | @item #pragma implementation | |
3122 | @itemx #pragma implementation "@var{objects}.h" | |
3123 | @kindex #pragma implementation | |
3124 | Use this pragma in a @emph{main input file}, when you want full output from | |
3125 | included header files to be generated (and made globally visible). The | |
3126 | included header file, in turn, should use @samp{#pragma interface}. | |
3127 | Backup copies of inline member functions, debugging information, and the | |
3128 | internal tables used to implement virtual functions are all generated in | |
3129 | implementation files. | |
3130 | ||
3131 | @cindex implied @code{#pragma implementation} | |
3132 | @cindex @code{#pragma implementation}, implied | |
3133 | @cindex naming convention, implementation headers | |
3134 | If you use @samp{#pragma implementation} with no argument, it applies to | |
3135 | an include file with the same basename@footnote{A file's @dfn{basename} | |
3136 | was the name stripped of all leading path information and of trailing | |
3137 | suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source | |
3138 | file. For example, in @file{allclass.cc}, giving just | |
3139 | @samp{#pragma implementation} | |
3140 | by itself is equivalent to @samp{#pragma implementation "allclass.h"}. | |
3141 | ||
3142 | In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as | |
3143 | an implementation file whenever you would include it from | |
3144 | @file{allclass.cc} even if you never specified @samp{#pragma | |
3145 | implementation}. This was deemed to be more trouble than it was worth, | |
3146 | however, and disabled. | |
3147 | ||
3148 | If you use an explicit @samp{#pragma implementation}, it must appear in | |
3149 | your source file @emph{before} you include the affected header files. | |
3150 | ||
3151 | Use the string argument if you want a single implementation file to | |
3152 | include code from multiple header files. (You must also use | |
3153 | @samp{#include} to include the header file; @samp{#pragma | |
3154 | implementation} only specifies how to use the file---it doesn't actually | |
3155 | include it.) | |
3156 | ||
3157 | There is no way to split up the contents of a single header file into | |
3158 | multiple implementation files. | |
3159 | @end table | |
3160 | ||
3161 | @cindex inlining and C++ pragmas | |
3162 | @cindex C++ pragmas, effect on inlining | |
3163 | @cindex pragmas in C++, effect on inlining | |
3164 | @samp{#pragma implementation} and @samp{#pragma interface} also have an | |
3165 | effect on function inlining. | |
3166 | ||
3167 | If you define a class in a header file marked with @samp{#pragma | |
3168 | interface}, the effect on a function defined in that class is similar to | |
3169 | an explicit @code{extern} declaration---the compiler emits no code at | |
3170 | all to define an independent version of the function. Its definition | |
3171 | is used only for inlining with its callers. | |
3172 | ||
3173 | Conversely, when you include the same header file in a main source file | |
3174 | that declares it as @samp{#pragma implementation}, the compiler emits | |
3175 | code for the function itself; this defines a version of the function | |
3176 | that can be found via pointers (or by callers compiled without | |
3177 | inlining). If all calls to the function can be inlined, you can avoid | |
3178 | emitting the function by compiling with @samp{-fno-implement-inlines}. | |
3179 | If any calls were not inlined, you will get linker errors. | |
3180 | ||
3181 | @node Template Instantiation | |
3182 | @section Where's the Template? | |
3183 | ||
3184 | @cindex template instantiation | |
3185 | ||
3186 | C++ templates are the first language feature to require more | |
3187 | intelligence from the environment than one usually finds on a UNIX | |
3188 | system. Somehow the compiler and linker have to make sure that each | |
3189 | template instance occurs exactly once in the executable if it is needed, | |
3190 | and not at all otherwise. There are two basic approaches to this | |
3191 | problem, which I will refer to as the Borland model and the Cfront model. | |
3192 | ||
3193 | @table @asis | |
3194 | @item Borland model | |
3195 | Borland C++ solved the template instantiation problem by adding the code | |
469b759e JM |
3196 | equivalent of common blocks to their linker; the compiler emits template |
3197 | instances in each translation unit that uses them, and the linker | |
3198 | collapses them together. The advantage of this model is that the linker | |
3199 | only has to consider the object files themselves; there is no external | |
3200 | complexity to worry about. This disadvantage is that compilation time | |
3201 | is increased because the template code is being compiled repeatedly. | |
3202 | Code written for this model tends to include definitions of all | |
3203 | templates in the header file, since they must be seen to be | |
3204 | instantiated. | |
c1f7febf RK |
3205 | |
3206 | @item Cfront model | |
3207 | The AT&T C++ translator, Cfront, solved the template instantiation | |
3208 | problem by creating the notion of a template repository, an | |
469b759e JM |
3209 | automatically maintained place where template instances are stored. A |
3210 | more modern version of the repository works as follows: As individual | |
3211 | object files are built, the compiler places any template definitions and | |
3212 | instantiations encountered in the repository. At link time, the link | |
3213 | wrapper adds in the objects in the repository and compiles any needed | |
3214 | instances that were not previously emitted. The advantages of this | |
3215 | model are more optimal compilation speed and the ability to use the | |
3216 | system linker; to implement the Borland model a compiler vendor also | |
c1f7febf | 3217 | needs to replace the linker. The disadvantages are vastly increased |
469b759e JM |
3218 | complexity, and thus potential for error; for some code this can be |
3219 | just as transparent, but in practice it can been very difficult to build | |
c1f7febf | 3220 | multiple programs in one directory and one program in multiple |
469b759e JM |
3221 | directories. Code written for this model tends to separate definitions |
3222 | of non-inline member templates into a separate file, which should be | |
3223 | compiled separately. | |
c1f7febf RK |
3224 | @end table |
3225 | ||
469b759e | 3226 | When used with GNU ld version 2.8 or later on an ELF system such as |
a4b3b54a JM |
3227 | Linux/GNU or Solaris 2, or on Microsoft Windows, g++ supports the |
3228 | Borland model. On other systems, g++ implements neither automatic | |
3229 | model. | |
469b759e JM |
3230 | |
3231 | A future version of g++ will support a hybrid model whereby the compiler | |
3232 | will emit any instantiations for which the template definition is | |
3233 | included in the compile, and store template definitions and | |
3234 | instantiation context information into the object file for the rest. | |
3235 | The link wrapper will extract that information as necessary and invoke | |
3236 | the compiler to produce the remaining instantiations. The linker will | |
3237 | then combine duplicate instantiations. | |
3238 | ||
3239 | In the mean time, you have the following options for dealing with | |
3240 | template instantiations: | |
c1f7febf RK |
3241 | |
3242 | @enumerate | |
d863830b JL |
3243 | @item |
3244 | Compile your template-using code with @samp{-frepo}. The compiler will | |
3245 | generate files with the extension @samp{.rpo} listing all of the | |
3246 | template instantiations used in the corresponding object files which | |
3247 | could be instantiated there; the link wrapper, @samp{collect2}, will | |
3248 | then update the @samp{.rpo} files to tell the compiler where to place | |
3249 | those instantiations and rebuild any affected object files. The | |
3250 | link-time overhead is negligible after the first pass, as the compiler | |
3251 | will continue to place the instantiations in the same files. | |
3252 | ||
3253 | This is your best option for application code written for the Borland | |
3254 | model, as it will just work. Code written for the Cfront model will | |
3255 | need to be modified so that the template definitions are available at | |
3256 | one or more points of instantiation; usually this is as simple as adding | |
3257 | @code{#include <tmethods.cc>} to the end of each template header. | |
3258 | ||
3259 | For library code, if you want the library to provide all of the template | |
3260 | instantiations it needs, just try to link all of its object files | |
3261 | together; the link will fail, but cause the instantiations to be | |
3262 | generated as a side effect. Be warned, however, that this may cause | |
3263 | conflicts if multiple libraries try to provide the same instantiations. | |
3264 | For greater control, use explicit instantiation as described in the next | |
3265 | option. | |
3266 | ||
c1f7febf RK |
3267 | @item |
3268 | Compile your code with @samp{-fno-implicit-templates} to disable the | |
3269 | implicit generation of template instances, and explicitly instantiate | |
3270 | all the ones you use. This approach requires more knowledge of exactly | |
3271 | which instances you need than do the others, but it's less | |
3272 | mysterious and allows greater control. You can scatter the explicit | |
3273 | instantiations throughout your program, perhaps putting them in the | |
3274 | translation units where the instances are used or the translation units | |
3275 | that define the templates themselves; you can put all of the explicit | |
3276 | instantiations you need into one big file; or you can create small files | |
3277 | like | |
3278 | ||
3279 | @example | |
3280 | #include "Foo.h" | |
3281 | #include "Foo.cc" | |
3282 | ||
3283 | template class Foo<int>; | |
3284 | template ostream& operator << | |
3285 | (ostream&, const Foo<int>&); | |
3286 | @end example | |
3287 | ||
3288 | for each of the instances you need, and create a template instantiation | |
3289 | library from those. | |
3290 | ||
3291 | If you are using Cfront-model code, you can probably get away with not | |
3292 | using @samp{-fno-implicit-templates} when compiling files that don't | |
3293 | @samp{#include} the member template definitions. | |
3294 | ||
3295 | If you use one big file to do the instantiations, you may want to | |
3296 | compile it without @samp{-fno-implicit-templates} so you get all of the | |
3297 | instances required by your explicit instantiations (but not by any | |
3298 | other files) without having to specify them as well. | |
3299 | ||
3300 | g++ has extended the template instantiation syntax outlined in the | |
3301 | Working Paper to allow forward declaration of explicit instantiations, | |
3302 | explicit instantiation of members of template classes and instantiation | |
3303 | of the compiler support data for a template class (i.e. the vtable) | |
3304 | without instantiating any of its members: | |
3305 | ||
3306 | @example | |
3307 | extern template int max (int, int); | |
3308 | template void Foo<int>::f (); | |
3309 | inline template class Foo<int>; | |
3310 | @end example | |
3311 | ||
3312 | @item | |
3313 | Do nothing. Pretend g++ does implement automatic instantiation | |
3314 | management. Code written for the Borland model will work fine, but | |
3315 | each translation unit will contain instances of each of the templates it | |
3316 | uses. In a large program, this can lead to an unacceptable amount of code | |
3317 | duplication. | |
3318 | ||
3319 | @item | |
3320 | Add @samp{#pragma interface} to all files containing template | |
3321 | definitions. For each of these files, add @samp{#pragma implementation | |
3322 | "@var{filename}"} to the top of some @samp{.C} file which | |
3323 | @samp{#include}s it. Then compile everything with | |
3324 | @samp{-fexternal-templates}. The templates will then only be expanded | |
3325 | in the translation unit which implements them (i.e. has a @samp{#pragma | |
3326 | implementation} line for the file where they live); all other files will | |
3327 | use external references. If you're lucky, everything should work | |
3328 | properly. If you get undefined symbol errors, you need to make sure | |
3329 | that each template instance which is used in the program is used in the | |
3330 | file which implements that template. If you don't have any use for a | |
3331 | particular instance in that file, you can just instantiate it | |
3332 | explicitly, using the syntax from the latest C++ working paper: | |
3333 | ||
3334 | @example | |
3335 | template class A<int>; | |
3336 | template ostream& operator << (ostream&, const A<int>&); | |
3337 | @end example | |
3338 | ||
3339 | This strategy will work with code written for either model. If you are | |
3340 | using code written for the Cfront model, the file containing a class | |
3341 | template and the file containing its member templates should be | |
3342 | implemented in the same translation unit. | |
3343 | ||
3344 | A slight variation on this approach is to instead use the flag | |
3345 | @samp{-falt-external-templates}; this flag causes template | |
3346 | instances to be emitted in the translation unit that implements the | |
3347 | header where they are first instantiated, rather than the one which | |
3348 | implements the file where the templates are defined. This header must | |
3349 | be the same in all translation units, or things are likely to break. | |
3350 | ||
3351 | @xref{C++ Interface,,Declarations and Definitions in One Header}, for | |
3352 | more discussion of these pragmas. | |
3353 | @end enumerate | |
3354 | ||
3355 | @node C++ Signatures | |
3356 | @section Type Abstraction using Signatures | |
3357 | ||
3358 | @findex signature | |
3359 | @cindex type abstraction, C++ | |
3360 | @cindex C++ type abstraction | |
3361 | @cindex subtype polymorphism, C++ | |
3362 | @cindex C++ subtype polymorphism | |
3363 | @cindex signatures, C++ | |
3364 | @cindex C++ signatures | |
3365 | ||
3366 | In GNU C++, you can use the keyword @code{signature} to define a | |
3367 | completely abstract class interface as a datatype. You can connect this | |
3368 | abstraction with actual classes using signature pointers. If you want | |
3369 | to use signatures, run the GNU compiler with the | |
3370 | @samp{-fhandle-signatures} command-line option. (With this option, the | |
3371 | compiler reserves a second keyword @code{sigof} as well, for a future | |
3372 | extension.) | |
3373 | ||
3374 | Roughly, signatures are type abstractions or interfaces of classes. | |
3375 | Some other languages have similar facilities. C++ signatures are | |
3376 | related to ML's signatures, Haskell's type classes, definition modules | |
3377 | in Modula-2, interface modules in Modula-3, abstract types in Emerald, | |
3378 | type modules in Trellis/Owl, categories in Scratchpad II, and types in | |
3379 | POOL-I. For a more detailed discussion of signatures, see | |
3380 | @cite{Signatures: A Language Extension for Improving Type Abstraction and | |
3381 | Subtype Polymorphism in C++} | |
3382 | by @w{Gerald} Baumgartner and Vincent F. Russo (Tech report | |
3383 | CSD--TR--95--051, Dept. of Computer Sciences, Purdue University, | |
3384 | August 1995, a slightly improved version appeared in | |
3385 | @emph{Software---Practice & Experience}, @b{25}(8), pp. 863--889, | |
3386 | August 1995). You can get the tech report by anonymous FTP from | |
3387 | @code{ftp.cs.purdue.edu} in @file{pub/gb/Signature-design.ps.gz}. | |
3388 | ||
3389 | Syntactically, a signature declaration is a collection of | |
3390 | member function declarations and nested type declarations. | |
3391 | For example, this signature declaration defines a new abstract type | |
3392 | @code{S} with member functions @samp{int foo ()} and @samp{int bar (int)}: | |
3393 | ||
3394 | @example | |
3395 | signature S | |
3396 | @{ | |
3397 | int foo (); | |
3398 | int bar (int); | |
3399 | @}; | |
3400 | @end example | |
3401 | ||
3402 | Since signature types do not include implementation definitions, you | |
3403 | cannot write an instance of a signature directly. Instead, you can | |
3404 | define a pointer to any class that contains the required interfaces as a | |
3405 | @dfn{signature pointer}. Such a class @dfn{implements} the signature | |
3406 | type. | |
3407 | @c Eventually signature references should work too. | |
3408 | ||
3409 | To use a class as an implementation of @code{S}, you must ensure that | |
3410 | the class has public member functions @samp{int foo ()} and @samp{int | |
3411 | bar (int)}. The class can have other member functions as well, public | |
3412 | or not; as long as it offers what's declared in the signature, it is | |
3413 | suitable as an implementation of that signature type. | |
3414 | ||
3415 | For example, suppose that @code{C} is a class that meets the | |
3416 | requirements of signature @code{S} (@code{C} @dfn{conforms to} | |
3417 | @code{S}). Then | |
3418 | ||
3419 | @example | |
3420 | C obj; | |
3421 | S * p = &obj; | |
3422 | @end example | |
3423 | ||
3424 | @noindent | |
3425 | defines a signature pointer @code{p} and initializes it to point to an | |
3426 | object of type @code{C}. | |
3427 | The member function call @w{@samp{int i = p->foo ();}} | |
3428 | executes @samp{obj.foo ()}. | |
3429 | ||
3430 | @cindex @code{signature} in C++, advantages | |
3431 | Abstract virtual classes provide somewhat similar facilities in standard | |
3432 | C++. There are two main advantages to using signatures instead: | |
3433 | ||
3434 | @enumerate | |
3435 | @item | |
3436 | Subtyping becomes independent from inheritance. A class or signature | |
3437 | type @code{T} is a subtype of a signature type @code{S} independent of | |
3438 | any inheritance hierarchy as long as all the member functions declared | |
3439 | in @code{S} are also found in @code{T}. So you can define a subtype | |
3440 | hierarchy that is completely independent from any inheritance | |
3441 | (implementation) hierarchy, instead of being forced to use types that | |
3442 | mirror the class inheritance hierarchy. | |
3443 | ||
3444 | @item | |
3445 | Signatures allow you to work with existing class hierarchies as | |
3446 | implementations of a signature type. If those class hierarchies are | |
3447 | only available in compiled form, you're out of luck with abstract virtual | |
3448 | classes, since an abstract virtual class cannot be retrofitted on top of | |
3449 | existing class hierarchies. So you would be required to write interface | |
3450 | classes as subtypes of the abstract virtual class. | |
3451 | @end enumerate | |
3452 | ||
3453 | @cindex default implementation, signature member function | |
3454 | @cindex signature member function default implementation | |
3455 | There is one more detail about signatures. A signature declaration can | |
3456 | contain member function @emph{definitions} as well as member function | |
3457 | declarations. A signature member function with a full definition is | |
3458 | called a @emph{default implementation}; classes need not contain that | |
3459 | particular interface in order to conform. For example, a | |
3460 | class @code{C} can conform to the signature | |
3461 | ||
3462 | @example | |
3463 | signature T | |
3464 | @{ | |
3465 | int f (int); | |
3466 | int f0 () @{ return f (0); @}; | |
3467 | @}; | |
3468 | @end example | |
3469 | ||
3470 | @noindent | |
3471 | whether or not @code{C} implements the member function @samp{int f0 ()}. | |
3472 | If you define @code{C::f0}, that definition takes precedence; | |
3473 | otherwise, the default implementation @code{S::f0} applies. | |
3474 | ||
3475 | @ignore | |
3476 | There will be more support for signatures in the future. | |
3477 | Add to this doc as the implementation grows. | |
3478 | In particular, the following features are planned but not yet | |
3479 | implemented: | |
3480 | @itemize @bullet | |
3481 | @item signature references, | |
3482 | @item signature inheritance, | |
3483 | @item the @code{sigof} construct for extracting the signature information | |
3484 | of a class, | |
3485 | @item views for renaming member functions when matching a class type | |
3486 | with a signature type, | |
3487 | @item specifying exceptions with signature member functions, and | |
3488 | @item signature templates. | |
3489 | @end itemize | |
3490 | This list is roughly in the order in which we intend to implement | |
3491 | them. Watch this space for updates. | |
3492 | @end ignore |