]>
Commit | Line | Data |
---|---|---|
475b6e22 | 1 | @c Copyright (C) 1988,1989,1992,1993,1994,1996,1998,1999,2000,2001,2002,2003,2004,2005 |
f0ddce84 | 2 | @c Free Software Foundation, Inc. |
c1f7febf RK |
3 | @c This is part of the GCC manual. |
4 | @c For copying conditions, see the file gcc.texi. | |
5 | ||
6 | @node C Extensions | |
7 | @chapter Extensions to the C Language Family | |
8 | @cindex extensions, C language | |
9 | @cindex C language extensions | |
10 | ||
84330467 | 11 | @opindex pedantic |
161d7b59 | 12 | GNU C provides several language features not found in ISO standard C@. |
f0523f02 | 13 | (The @option{-pedantic} option directs GCC to print a warning message if |
c1f7febf RK |
14 | any of these features is used.) To test for the availability of these |
15 | features in conditional compilation, check for a predefined macro | |
161d7b59 | 16 | @code{__GNUC__}, which is always defined under GCC@. |
c1f7febf | 17 | |
161d7b59 | 18 | These extensions are available in C and Objective-C@. Most of them are |
c1f7febf RK |
19 | also available in C++. @xref{C++ Extensions,,Extensions to the |
20 | C++ Language}, for extensions that apply @emph{only} to C++. | |
21 | ||
4b404517 JM |
22 | Some features that are in ISO C99 but not C89 or C++ are also, as |
23 | extensions, accepted by GCC in C89 mode and in C++. | |
5490d604 | 24 | |
c1f7febf RK |
25 | @menu |
26 | * Statement Exprs:: Putting statements and declarations inside expressions. | |
14e33ee8 | 27 | * Local Labels:: Labels local to a block. |
c1f7febf RK |
28 | * Labels as Values:: Getting pointers to labels, and computed gotos. |
29 | * Nested Functions:: As in Algol and Pascal, lexical scoping of functions. | |
30 | * Constructing Calls:: Dispatching a call to another function. | |
c1f7febf | 31 | * Typeof:: @code{typeof}: referring to the type of an expression. |
c1f7febf RK |
32 | * Conditionals:: Omitting the middle operand of a @samp{?:} expression. |
33 | * Long Long:: Double-word integers---@code{long long int}. | |
34 | * Complex:: Data types for complex numbers. | |
6f4d7222 | 35 | * Hex Floats:: Hexadecimal floating-point constants. |
c1f7febf RK |
36 | * Zero Length:: Zero-length arrays. |
37 | * Variable Length:: Arrays whose length is computed at run time. | |
ba05abd3 | 38 | * Empty Structures:: Structures with no members. |
ccd96f0a NB |
39 | * Variadic Macros:: Macros with a variable number of arguments. |
40 | * Escaped Newlines:: Slightly looser rules for escaped newlines. | |
c1f7febf RK |
41 | * Subscripting:: Any array can be subscripted, even if not an lvalue. |
42 | * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers. | |
43 | * Initializers:: Non-constant initializers. | |
4b404517 | 44 | * Compound Literals:: Compound literals give structures, unions |
c1f7febf | 45 | or arrays as values. |
4b404517 | 46 | * Designated Inits:: Labeling elements of initializers. |
c1f7febf RK |
47 | * Cast to Union:: Casting to union type from any member of the union. |
48 | * Case Ranges:: `case 1 ... 9' and such. | |
4b404517 | 49 | * Mixed Declarations:: Mixing declarations and code. |
c1f7febf RK |
50 | * Function Attributes:: Declaring that functions have no side effects, |
51 | or that they can never return. | |
2c5e91d2 | 52 | * Attribute Syntax:: Formal syntax for attributes. |
c1f7febf RK |
53 | * Function Prototypes:: Prototype declarations and old-style definitions. |
54 | * C++ Comments:: C++ comments are recognized. | |
55 | * Dollar Signs:: Dollar sign is allowed in identifiers. | |
56 | * Character Escapes:: @samp{\e} stands for the character @key{ESC}. | |
57 | * Variable Attributes:: Specifying attributes of variables. | |
58 | * Type Attributes:: Specifying attributes of types. | |
59 | * Alignment:: Inquiring about the alignment of a type or variable. | |
60 | * Inline:: Defining inline functions (as fast as macros). | |
61 | * Extended Asm:: Assembler instructions with C expressions as operands. | |
62 | (With them you can define ``built-in'' functions.) | |
63 | * Constraints:: Constraints for asm operands | |
64 | * Asm Labels:: Specifying the assembler name to use for a C symbol. | |
65 | * Explicit Reg Vars:: Defining variables residing in specified registers. | |
66 | * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files. | |
67 | * Incomplete Enums:: @code{enum foo;}, with details to follow. | |
68 | * Function Names:: Printable strings which are the name of the current | |
69 | function. | |
70 | * Return Address:: Getting the return or frame address of a function. | |
1255c85c | 71 | * Vector Extensions:: Using vector instructions through built-in functions. |
7a3ea201 | 72 | * Offsetof:: Special syntax for implementing @code{offsetof}. |
c5c76735 | 73 | * Other Builtins:: Other built-in functions. |
0975678f | 74 | * Target Builtins:: Built-in functions specific to particular targets. |
a2bec818 | 75 | * Target Format Checks:: Format checks specific to particular targets. |
0168a849 | 76 | * Pragmas:: Pragmas accepted by GCC. |
b11cc610 | 77 | * Unnamed Fields:: Unnamed struct/union fields within structs/unions. |
3d78f2e9 | 78 | * Thread-Local:: Per-thread variables. |
c1f7febf | 79 | @end menu |
c1f7febf RK |
80 | |
81 | @node Statement Exprs | |
82 | @section Statements and Declarations in Expressions | |
83 | @cindex statements inside expressions | |
84 | @cindex declarations inside expressions | |
85 | @cindex expressions containing statements | |
86 | @cindex macros, statements in expressions | |
87 | ||
88 | @c the above section title wrapped and causes an underfull hbox.. i | |
89 | @c changed it from "within" to "in". --mew 4feb93 | |
c1f7febf | 90 | A compound statement enclosed in parentheses may appear as an expression |
161d7b59 | 91 | in GNU C@. This allows you to use loops, switches, and local variables |
c1f7febf RK |
92 | within an expression. |
93 | ||
94 | Recall that a compound statement is a sequence of statements surrounded | |
95 | by braces; in this construct, parentheses go around the braces. For | |
96 | example: | |
97 | ||
3ab51846 | 98 | @smallexample |
c1f7febf RK |
99 | (@{ int y = foo (); int z; |
100 | if (y > 0) z = y; | |
101 | else z = - y; | |
102 | z; @}) | |
3ab51846 | 103 | @end smallexample |
c1f7febf RK |
104 | |
105 | @noindent | |
106 | is a valid (though slightly more complex than necessary) expression | |
107 | for the absolute value of @code{foo ()}. | |
108 | ||
109 | The last thing in the compound statement should be an expression | |
110 | followed by a semicolon; the value of this subexpression serves as the | |
111 | value of the entire construct. (If you use some other kind of statement | |
112 | last within the braces, the construct has type @code{void}, and thus | |
113 | effectively no value.) | |
114 | ||
115 | This feature is especially useful in making macro definitions ``safe'' (so | |
116 | that they evaluate each operand exactly once). For example, the | |
117 | ``maximum'' function is commonly defined as a macro in standard C as | |
118 | follows: | |
119 | ||
3ab51846 | 120 | @smallexample |
c1f7febf | 121 | #define max(a,b) ((a) > (b) ? (a) : (b)) |
3ab51846 | 122 | @end smallexample |
c1f7febf RK |
123 | |
124 | @noindent | |
125 | @cindex side effects, macro argument | |
126 | But this definition computes either @var{a} or @var{b} twice, with bad | |
127 | results if the operand has side effects. In GNU C, if you know the | |
962e6e00 | 128 | type of the operands (here taken as @code{int}), you can define |
c1f7febf RK |
129 | the macro safely as follows: |
130 | ||
3ab51846 | 131 | @smallexample |
c1f7febf RK |
132 | #define maxint(a,b) \ |
133 | (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @}) | |
3ab51846 | 134 | @end smallexample |
c1f7febf RK |
135 | |
136 | Embedded statements are not allowed in constant expressions, such as | |
c771326b | 137 | the value of an enumeration constant, the width of a bit-field, or |
c1f7febf RK |
138 | the initial value of a static variable. |
139 | ||
140 | If you don't know the type of the operand, you can still do this, but you | |
95f79357 | 141 | must use @code{typeof} (@pxref{Typeof}). |
c1f7febf | 142 | |
a5bcc582 NS |
143 | In G++, the result value of a statement expression undergoes array and |
144 | function pointer decay, and is returned by value to the enclosing | |
8a36672b | 145 | expression. For instance, if @code{A} is a class, then |
b98e139b | 146 | |
a5bcc582 NS |
147 | @smallexample |
148 | A a; | |
b98e139b | 149 | |
a5bcc582 NS |
150 | (@{a;@}).Foo () |
151 | @end smallexample | |
b98e139b MM |
152 | |
153 | @noindent | |
a5bcc582 NS |
154 | will construct a temporary @code{A} object to hold the result of the |
155 | statement expression, and that will be used to invoke @code{Foo}. | |
156 | Therefore the @code{this} pointer observed by @code{Foo} will not be the | |
157 | address of @code{a}. | |
158 | ||
159 | Any temporaries created within a statement within a statement expression | |
160 | will be destroyed at the statement's end. This makes statement | |
161 | expressions inside macros slightly different from function calls. In | |
162 | the latter case temporaries introduced during argument evaluation will | |
163 | be destroyed at the end of the statement that includes the function | |
164 | call. In the statement expression case they will be destroyed during | |
165 | the statement expression. For instance, | |
b98e139b | 166 | |
a5bcc582 NS |
167 | @smallexample |
168 | #define macro(a) (@{__typeof__(a) b = (a); b + 3; @}) | |
169 | template<typename T> T function(T a) @{ T b = a; return b + 3; @} | |
170 | ||
171 | void foo () | |
172 | @{ | |
173 | macro (X ()); | |
174 | function (X ()); | |
175 | @} | |
176 | @end smallexample | |
b98e139b MM |
177 | |
178 | @noindent | |
a5bcc582 NS |
179 | will have different places where temporaries are destroyed. For the |
180 | @code{macro} case, the temporary @code{X} will be destroyed just after | |
181 | the initialization of @code{b}. In the @code{function} case that | |
182 | temporary will be destroyed when the function returns. | |
b98e139b MM |
183 | |
184 | These considerations mean that it is probably a bad idea to use | |
185 | statement-expressions of this form in header files that are designed to | |
54e1d3a6 MM |
186 | work with C++. (Note that some versions of the GNU C Library contained |
187 | header files using statement-expression that lead to precisely this | |
188 | bug.) | |
b98e139b | 189 | |
c1f7febf RK |
190 | @node Local Labels |
191 | @section Locally Declared Labels | |
192 | @cindex local labels | |
193 | @cindex macros, local labels | |
194 | ||
14e33ee8 | 195 | GCC allows you to declare @dfn{local labels} in any nested block |
8a36672b | 196 | scope. A local label is just like an ordinary label, but you can |
14e33ee8 | 197 | only reference it (with a @code{goto} statement, or by taking its |
daf2f129 | 198 | address) within the block in which it was declared. |
c1f7febf RK |
199 | |
200 | A local label declaration looks like this: | |
201 | ||
3ab51846 | 202 | @smallexample |
c1f7febf | 203 | __label__ @var{label}; |
3ab51846 | 204 | @end smallexample |
c1f7febf RK |
205 | |
206 | @noindent | |
207 | or | |
208 | ||
3ab51846 | 209 | @smallexample |
0d893a63 | 210 | __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */; |
3ab51846 | 211 | @end smallexample |
c1f7febf | 212 | |
14e33ee8 ZW |
213 | Local label declarations must come at the beginning of the block, |
214 | before any ordinary declarations or statements. | |
c1f7febf RK |
215 | |
216 | The label declaration defines the label @emph{name}, but does not define | |
217 | the label itself. You must do this in the usual way, with | |
218 | @code{@var{label}:}, within the statements of the statement expression. | |
219 | ||
14e33ee8 ZW |
220 | The local label feature is useful for complex macros. If a macro |
221 | contains nested loops, a @code{goto} can be useful for breaking out of | |
222 | them. However, an ordinary label whose scope is the whole function | |
223 | cannot be used: if the macro can be expanded several times in one | |
224 | function, the label will be multiply defined in that function. A | |
225 | local label avoids this problem. For example: | |
226 | ||
3ab51846 | 227 | @smallexample |
14e33ee8 ZW |
228 | #define SEARCH(value, array, target) \ |
229 | do @{ \ | |
230 | __label__ found; \ | |
231 | typeof (target) _SEARCH_target = (target); \ | |
232 | typeof (*(array)) *_SEARCH_array = (array); \ | |
233 | int i, j; \ | |
234 | int value; \ | |
235 | for (i = 0; i < max; i++) \ | |
236 | for (j = 0; j < max; j++) \ | |
237 | if (_SEARCH_array[i][j] == _SEARCH_target) \ | |
238 | @{ (value) = i; goto found; @} \ | |
239 | (value) = -1; \ | |
240 | found:; \ | |
241 | @} while (0) | |
3ab51846 | 242 | @end smallexample |
14e33ee8 ZW |
243 | |
244 | This could also be written using a statement-expression: | |
c1f7febf | 245 | |
3ab51846 | 246 | @smallexample |
c1f7febf | 247 | #define SEARCH(array, target) \ |
310668e8 | 248 | (@{ \ |
c1f7febf RK |
249 | __label__ found; \ |
250 | typeof (target) _SEARCH_target = (target); \ | |
251 | typeof (*(array)) *_SEARCH_array = (array); \ | |
252 | int i, j; \ | |
253 | int value; \ | |
254 | for (i = 0; i < max; i++) \ | |
255 | for (j = 0; j < max; j++) \ | |
256 | if (_SEARCH_array[i][j] == _SEARCH_target) \ | |
310668e8 | 257 | @{ value = i; goto found; @} \ |
c1f7febf RK |
258 | value = -1; \ |
259 | found: \ | |
260 | value; \ | |
261 | @}) | |
3ab51846 | 262 | @end smallexample |
c1f7febf | 263 | |
14e33ee8 ZW |
264 | Local label declarations also make the labels they declare visible to |
265 | nested functions, if there are any. @xref{Nested Functions}, for details. | |
266 | ||
c1f7febf RK |
267 | @node Labels as Values |
268 | @section Labels as Values | |
269 | @cindex labels as values | |
270 | @cindex computed gotos | |
271 | @cindex goto with computed label | |
272 | @cindex address of a label | |
273 | ||
274 | You can get the address of a label defined in the current function | |
275 | (or a containing function) with the unary operator @samp{&&}. The | |
276 | value has type @code{void *}. This value is a constant and can be used | |
277 | wherever a constant of that type is valid. For example: | |
278 | ||
3ab51846 | 279 | @smallexample |
c1f7febf | 280 | void *ptr; |
0d893a63 | 281 | /* @r{@dots{}} */ |
c1f7febf | 282 | ptr = &&foo; |
3ab51846 | 283 | @end smallexample |
c1f7febf RK |
284 | |
285 | To use these values, you need to be able to jump to one. This is done | |
286 | with the computed goto statement@footnote{The analogous feature in | |
287 | Fortran is called an assigned goto, but that name seems inappropriate in | |
288 | C, where one can do more than simply store label addresses in label | |
289 | variables.}, @code{goto *@var{exp};}. For example, | |
290 | ||
3ab51846 | 291 | @smallexample |
c1f7febf | 292 | goto *ptr; |
3ab51846 | 293 | @end smallexample |
c1f7febf RK |
294 | |
295 | @noindent | |
296 | Any expression of type @code{void *} is allowed. | |
297 | ||
298 | One way of using these constants is in initializing a static array that | |
299 | will serve as a jump table: | |
300 | ||
3ab51846 | 301 | @smallexample |
c1f7febf | 302 | static void *array[] = @{ &&foo, &&bar, &&hack @}; |
3ab51846 | 303 | @end smallexample |
c1f7febf RK |
304 | |
305 | Then you can select a label with indexing, like this: | |
306 | ||
3ab51846 | 307 | @smallexample |
c1f7febf | 308 | goto *array[i]; |
3ab51846 | 309 | @end smallexample |
c1f7febf RK |
310 | |
311 | @noindent | |
312 | Note that this does not check whether the subscript is in bounds---array | |
313 | indexing in C never does that. | |
314 | ||
315 | Such an array of label values serves a purpose much like that of the | |
316 | @code{switch} statement. The @code{switch} statement is cleaner, so | |
317 | use that rather than an array unless the problem does not fit a | |
318 | @code{switch} statement very well. | |
319 | ||
320 | Another use of label values is in an interpreter for threaded code. | |
321 | The labels within the interpreter function can be stored in the | |
322 | threaded code for super-fast dispatching. | |
323 | ||
02f52e19 | 324 | You may not use this mechanism to jump to code in a different function. |
47620e09 | 325 | If you do that, totally unpredictable things will happen. The best way to |
c1f7febf RK |
326 | avoid this is to store the label address only in automatic variables and |
327 | never pass it as an argument. | |
328 | ||
47620e09 RH |
329 | An alternate way to write the above example is |
330 | ||
3ab51846 | 331 | @smallexample |
310668e8 JM |
332 | static const int array[] = @{ &&foo - &&foo, &&bar - &&foo, |
333 | &&hack - &&foo @}; | |
47620e09 | 334 | goto *(&&foo + array[i]); |
3ab51846 | 335 | @end smallexample |
47620e09 RH |
336 | |
337 | @noindent | |
338 | This is more friendly to code living in shared libraries, as it reduces | |
339 | the number of dynamic relocations that are needed, and by consequence, | |
340 | allows the data to be read-only. | |
341 | ||
c1f7febf RK |
342 | @node Nested Functions |
343 | @section Nested Functions | |
344 | @cindex nested functions | |
345 | @cindex downward funargs | |
346 | @cindex thunks | |
347 | ||
348 | A @dfn{nested function} is a function defined inside another function. | |
349 | (Nested functions are not supported for GNU C++.) The nested function's | |
350 | name is local to the block where it is defined. For example, here we | |
351 | define a nested function named @code{square}, and call it twice: | |
352 | ||
3ab51846 | 353 | @smallexample |
c1f7febf RK |
354 | @group |
355 | foo (double a, double b) | |
356 | @{ | |
357 | double square (double z) @{ return z * z; @} | |
358 | ||
359 | return square (a) + square (b); | |
360 | @} | |
361 | @end group | |
3ab51846 | 362 | @end smallexample |
c1f7febf RK |
363 | |
364 | The nested function can access all the variables of the containing | |
365 | function that are visible at the point of its definition. This is | |
366 | called @dfn{lexical scoping}. For example, here we show a nested | |
367 | function which uses an inherited variable named @code{offset}: | |
368 | ||
3ab51846 | 369 | @smallexample |
aee96fe9 | 370 | @group |
c1f7febf RK |
371 | bar (int *array, int offset, int size) |
372 | @{ | |
373 | int access (int *array, int index) | |
374 | @{ return array[index + offset]; @} | |
375 | int i; | |
0d893a63 | 376 | /* @r{@dots{}} */ |
c1f7febf | 377 | for (i = 0; i < size; i++) |
0d893a63 | 378 | /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */ |
c1f7febf | 379 | @} |
aee96fe9 | 380 | @end group |
3ab51846 | 381 | @end smallexample |
c1f7febf RK |
382 | |
383 | Nested function definitions are permitted within functions in the places | |
475b6e22 JM |
384 | where variable definitions are allowed; that is, in any block, mixed |
385 | with the other declarations and statements in the block. | |
c1f7febf RK |
386 | |
387 | It is possible to call the nested function from outside the scope of its | |
388 | name by storing its address or passing the address to another function: | |
389 | ||
3ab51846 | 390 | @smallexample |
c1f7febf RK |
391 | hack (int *array, int size) |
392 | @{ | |
393 | void store (int index, int value) | |
394 | @{ array[index] = value; @} | |
395 | ||
396 | intermediate (store, size); | |
397 | @} | |
3ab51846 | 398 | @end smallexample |
c1f7febf RK |
399 | |
400 | Here, the function @code{intermediate} receives the address of | |
401 | @code{store} as an argument. If @code{intermediate} calls @code{store}, | |
402 | the arguments given to @code{store} are used to store into @code{array}. | |
403 | But this technique works only so long as the containing function | |
404 | (@code{hack}, in this example) does not exit. | |
405 | ||
406 | If you try to call the nested function through its address after the | |
407 | containing function has exited, all hell will break loose. If you try | |
408 | to call it after a containing scope level has exited, and if it refers | |
409 | to some of the variables that are no longer in scope, you may be lucky, | |
410 | but it's not wise to take the risk. If, however, the nested function | |
411 | does not refer to anything that has gone out of scope, you should be | |
412 | safe. | |
413 | ||
9c34dbbf ZW |
414 | GCC implements taking the address of a nested function using a technique |
415 | called @dfn{trampolines}. A paper describing them is available as | |
416 | ||
417 | @noindent | |
b73b1546 | 418 | @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}. |
c1f7febf RK |
419 | |
420 | A nested function can jump to a label inherited from a containing | |
421 | function, provided the label was explicitly declared in the containing | |
422 | function (@pxref{Local Labels}). Such a jump returns instantly to the | |
423 | containing function, exiting the nested function which did the | |
424 | @code{goto} and any intermediate functions as well. Here is an example: | |
425 | ||
3ab51846 | 426 | @smallexample |
c1f7febf RK |
427 | @group |
428 | bar (int *array, int offset, int size) | |
429 | @{ | |
430 | __label__ failure; | |
431 | int access (int *array, int index) | |
432 | @{ | |
433 | if (index > size) | |
434 | goto failure; | |
435 | return array[index + offset]; | |
436 | @} | |
437 | int i; | |
0d893a63 | 438 | /* @r{@dots{}} */ |
c1f7febf | 439 | for (i = 0; i < size; i++) |
0d893a63 MK |
440 | /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */ |
441 | /* @r{@dots{}} */ | |
c1f7febf RK |
442 | return 0; |
443 | ||
444 | /* @r{Control comes here from @code{access} | |
445 | if it detects an error.} */ | |
446 | failure: | |
447 | return -1; | |
448 | @} | |
449 | @end group | |
3ab51846 | 450 | @end smallexample |
c1f7febf | 451 | |
28697267 DJ |
452 | A nested function always has no linkage. Declaring one with |
453 | @code{extern} or @code{static} is erroneous. If you need to declare the nested function | |
c1f7febf RK |
454 | before its definition, use @code{auto} (which is otherwise meaningless |
455 | for function declarations). | |
456 | ||
3ab51846 | 457 | @smallexample |
c1f7febf RK |
458 | bar (int *array, int offset, int size) |
459 | @{ | |
460 | __label__ failure; | |
461 | auto int access (int *, int); | |
0d893a63 | 462 | /* @r{@dots{}} */ |
c1f7febf RK |
463 | int access (int *array, int index) |
464 | @{ | |
465 | if (index > size) | |
466 | goto failure; | |
467 | return array[index + offset]; | |
468 | @} | |
0d893a63 | 469 | /* @r{@dots{}} */ |
c1f7febf | 470 | @} |
3ab51846 | 471 | @end smallexample |
c1f7febf RK |
472 | |
473 | @node Constructing Calls | |
474 | @section Constructing Function Calls | |
475 | @cindex constructing calls | |
476 | @cindex forwarding calls | |
477 | ||
478 | Using the built-in functions described below, you can record | |
479 | the arguments a function received, and call another function | |
480 | with the same arguments, without knowing the number or types | |
481 | of the arguments. | |
482 | ||
483 | You can also record the return value of that function call, | |
484 | and later return that value, without knowing what data type | |
485 | the function tried to return (as long as your caller expects | |
486 | that data type). | |
487 | ||
6429bc7c EB |
488 | However, these built-in functions may interact badly with some |
489 | sophisticated features or other extensions of the language. It | |
490 | is, therefore, not recommended to use them outside very simple | |
491 | functions acting as mere forwarders for their arguments. | |
492 | ||
84330467 JM |
493 | @deftypefn {Built-in Function} {void *} __builtin_apply_args () |
494 | This built-in function returns a pointer to data | |
c1f7febf RK |
495 | describing how to perform a call with the same arguments as were passed |
496 | to the current function. | |
497 | ||
498 | The function saves the arg pointer register, structure value address, | |
499 | and all registers that might be used to pass arguments to a function | |
500 | into a block of memory allocated on the stack. Then it returns the | |
501 | address of that block. | |
84330467 | 502 | @end deftypefn |
c1f7febf | 503 | |
84330467 JM |
504 | @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size}) |
505 | This built-in function invokes @var{function} | |
506 | with a copy of the parameters described by @var{arguments} | |
507 | and @var{size}. | |
c1f7febf RK |
508 | |
509 | The value of @var{arguments} should be the value returned by | |
510 | @code{__builtin_apply_args}. The argument @var{size} specifies the size | |
511 | of the stack argument data, in bytes. | |
512 | ||
84330467 | 513 | This function returns a pointer to data describing |
c1f7febf RK |
514 | how to return whatever value was returned by @var{function}. The data |
515 | is saved in a block of memory allocated on the stack. | |
516 | ||
517 | It is not always simple to compute the proper value for @var{size}. The | |
518 | value is used by @code{__builtin_apply} to compute the amount of data | |
519 | that should be pushed on the stack and copied from the incoming argument | |
520 | area. | |
84330467 | 521 | @end deftypefn |
c1f7febf | 522 | |
84330467 | 523 | @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result}) |
c1f7febf RK |
524 | This built-in function returns the value described by @var{result} from |
525 | the containing function. You should specify, for @var{result}, a value | |
526 | returned by @code{__builtin_apply}. | |
84330467 | 527 | @end deftypefn |
c1f7febf | 528 | |
c1f7febf RK |
529 | @node Typeof |
530 | @section Referring to a Type with @code{typeof} | |
531 | @findex typeof | |
532 | @findex sizeof | |
533 | @cindex macros, types of arguments | |
534 | ||
535 | Another way to refer to the type of an expression is with @code{typeof}. | |
536 | The syntax of using of this keyword looks like @code{sizeof}, but the | |
537 | construct acts semantically like a type name defined with @code{typedef}. | |
538 | ||
539 | There are two ways of writing the argument to @code{typeof}: with an | |
540 | expression or with a type. Here is an example with an expression: | |
541 | ||
3ab51846 | 542 | @smallexample |
c1f7febf | 543 | typeof (x[0](1)) |
3ab51846 | 544 | @end smallexample |
c1f7febf RK |
545 | |
546 | @noindent | |
89aed483 JM |
547 | This assumes that @code{x} is an array of pointers to functions; |
548 | the type described is that of the values of the functions. | |
c1f7febf RK |
549 | |
550 | Here is an example with a typename as the argument: | |
551 | ||
3ab51846 | 552 | @smallexample |
c1f7febf | 553 | typeof (int *) |
3ab51846 | 554 | @end smallexample |
c1f7febf RK |
555 | |
556 | @noindent | |
557 | Here the type described is that of pointers to @code{int}. | |
558 | ||
5490d604 | 559 | If you are writing a header file that must work when included in ISO C |
c1f7febf RK |
560 | programs, write @code{__typeof__} instead of @code{typeof}. |
561 | @xref{Alternate Keywords}. | |
562 | ||
563 | A @code{typeof}-construct can be used anywhere a typedef name could be | |
564 | used. For example, you can use it in a declaration, in a cast, or inside | |
565 | of @code{sizeof} or @code{typeof}. | |
566 | ||
95f79357 ZW |
567 | @code{typeof} is often useful in conjunction with the |
568 | statements-within-expressions feature. Here is how the two together can | |
569 | be used to define a safe ``maximum'' macro that operates on any | |
570 | arithmetic type and evaluates each of its arguments exactly once: | |
571 | ||
3ab51846 | 572 | @smallexample |
95f79357 ZW |
573 | #define max(a,b) \ |
574 | (@{ typeof (a) _a = (a); \ | |
575 | typeof (b) _b = (b); \ | |
576 | _a > _b ? _a : _b; @}) | |
3ab51846 | 577 | @end smallexample |
95f79357 | 578 | |
526278c9 VR |
579 | @cindex underscores in variables in macros |
580 | @cindex @samp{_} in variables in macros | |
581 | @cindex local variables in macros | |
582 | @cindex variables, local, in macros | |
583 | @cindex macros, local variables in | |
584 | ||
585 | The reason for using names that start with underscores for the local | |
586 | variables is to avoid conflicts with variable names that occur within the | |
587 | expressions that are substituted for @code{a} and @code{b}. Eventually we | |
588 | hope to design a new form of declaration syntax that allows you to declare | |
589 | variables whose scopes start only after their initializers; this will be a | |
590 | more reliable way to prevent such conflicts. | |
591 | ||
95f79357 ZW |
592 | @noindent |
593 | Some more examples of the use of @code{typeof}: | |
594 | ||
c1f7febf RK |
595 | @itemize @bullet |
596 | @item | |
597 | This declares @code{y} with the type of what @code{x} points to. | |
598 | ||
3ab51846 | 599 | @smallexample |
c1f7febf | 600 | typeof (*x) y; |
3ab51846 | 601 | @end smallexample |
c1f7febf RK |
602 | |
603 | @item | |
604 | This declares @code{y} as an array of such values. | |
605 | ||
3ab51846 | 606 | @smallexample |
c1f7febf | 607 | typeof (*x) y[4]; |
3ab51846 | 608 | @end smallexample |
c1f7febf RK |
609 | |
610 | @item | |
611 | This declares @code{y} as an array of pointers to characters: | |
612 | ||
3ab51846 | 613 | @smallexample |
c1f7febf | 614 | typeof (typeof (char *)[4]) y; |
3ab51846 | 615 | @end smallexample |
c1f7febf RK |
616 | |
617 | @noindent | |
618 | It is equivalent to the following traditional C declaration: | |
619 | ||
3ab51846 | 620 | @smallexample |
c1f7febf | 621 | char *y[4]; |
3ab51846 | 622 | @end smallexample |
c1f7febf RK |
623 | |
624 | To see the meaning of the declaration using @code{typeof}, and why it | |
962e6e00 | 625 | might be a useful way to write, rewrite it with these macros: |
c1f7febf | 626 | |
3ab51846 | 627 | @smallexample |
c1f7febf RK |
628 | #define pointer(T) typeof(T *) |
629 | #define array(T, N) typeof(T [N]) | |
3ab51846 | 630 | @end smallexample |
c1f7febf RK |
631 | |
632 | @noindent | |
633 | Now the declaration can be rewritten this way: | |
634 | ||
3ab51846 | 635 | @smallexample |
c1f7febf | 636 | array (pointer (char), 4) y; |
3ab51846 | 637 | @end smallexample |
c1f7febf RK |
638 | |
639 | @noindent | |
640 | Thus, @code{array (pointer (char), 4)} is the type of arrays of 4 | |
641 | pointers to @code{char}. | |
642 | @end itemize | |
643 | ||
95f79357 ZW |
644 | @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported |
645 | a more limited extension which permitted one to write | |
646 | ||
3ab51846 | 647 | @smallexample |
95f79357 | 648 | typedef @var{T} = @var{expr}; |
3ab51846 | 649 | @end smallexample |
95f79357 ZW |
650 | |
651 | @noindent | |
652 | with the effect of declaring @var{T} to have the type of the expression | |
653 | @var{expr}. This extension does not work with GCC 3 (versions between | |
654 | 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which | |
655 | relies on it should be rewritten to use @code{typeof}: | |
656 | ||
3ab51846 | 657 | @smallexample |
95f79357 | 658 | typedef typeof(@var{expr}) @var{T}; |
3ab51846 | 659 | @end smallexample |
95f79357 ZW |
660 | |
661 | @noindent | |
662 | This will work with all versions of GCC@. | |
663 | ||
c1f7febf RK |
664 | @node Conditionals |
665 | @section Conditionals with Omitted Operands | |
666 | @cindex conditional expressions, extensions | |
667 | @cindex omitted middle-operands | |
668 | @cindex middle-operands, omitted | |
669 | @cindex extensions, @code{?:} | |
670 | @cindex @code{?:} extensions | |
671 | ||
672 | The middle operand in a conditional expression may be omitted. Then | |
673 | if the first operand is nonzero, its value is the value of the conditional | |
674 | expression. | |
675 | ||
676 | Therefore, the expression | |
677 | ||
3ab51846 | 678 | @smallexample |
c1f7febf | 679 | x ? : y |
3ab51846 | 680 | @end smallexample |
c1f7febf RK |
681 | |
682 | @noindent | |
683 | has the value of @code{x} if that is nonzero; otherwise, the value of | |
684 | @code{y}. | |
685 | ||
686 | This example is perfectly equivalent to | |
687 | ||
3ab51846 | 688 | @smallexample |
c1f7febf | 689 | x ? x : y |
3ab51846 | 690 | @end smallexample |
c1f7febf RK |
691 | |
692 | @cindex side effect in ?: | |
693 | @cindex ?: side effect | |
694 | @noindent | |
695 | In this simple case, the ability to omit the middle operand is not | |
696 | especially useful. When it becomes useful is when the first operand does, | |
697 | or may (if it is a macro argument), contain a side effect. Then repeating | |
698 | the operand in the middle would perform the side effect twice. Omitting | |
699 | the middle operand uses the value already computed without the undesirable | |
700 | effects of recomputing it. | |
701 | ||
702 | @node Long Long | |
703 | @section Double-Word Integers | |
704 | @cindex @code{long long} data types | |
705 | @cindex double-word arithmetic | |
706 | @cindex multiprecision arithmetic | |
4b404517 JM |
707 | @cindex @code{LL} integer suffix |
708 | @cindex @code{ULL} integer suffix | |
c1f7febf | 709 | |
4b404517 JM |
710 | ISO C99 supports data types for integers that are at least 64 bits wide, |
711 | and as an extension GCC supports them in C89 mode and in C++. | |
712 | Simply write @code{long long int} for a signed integer, or | |
c1f7febf | 713 | @code{unsigned long long int} for an unsigned integer. To make an |
84330467 | 714 | integer constant of type @code{long long int}, add the suffix @samp{LL} |
c1f7febf | 715 | to the integer. To make an integer constant of type @code{unsigned long |
84330467 | 716 | long int}, add the suffix @samp{ULL} to the integer. |
c1f7febf RK |
717 | |
718 | You can use these types in arithmetic like any other integer types. | |
719 | Addition, subtraction, and bitwise boolean operations on these types | |
720 | are open-coded on all types of machines. Multiplication is open-coded | |
721 | if the machine supports fullword-to-doubleword a widening multiply | |
722 | instruction. Division and shifts are open-coded only on machines that | |
723 | provide special support. The operations that are not open-coded use | |
161d7b59 | 724 | special library routines that come with GCC@. |
c1f7febf RK |
725 | |
726 | There may be pitfalls when you use @code{long long} types for function | |
727 | arguments, unless you declare function prototypes. If a function | |
728 | expects type @code{int} for its argument, and you pass a value of type | |
729 | @code{long long int}, confusion will result because the caller and the | |
730 | subroutine will disagree about the number of bytes for the argument. | |
731 | Likewise, if the function expects @code{long long int} and you pass | |
732 | @code{int}. The best way to avoid such problems is to use prototypes. | |
733 | ||
734 | @node Complex | |
735 | @section Complex Numbers | |
736 | @cindex complex numbers | |
4b404517 JM |
737 | @cindex @code{_Complex} keyword |
738 | @cindex @code{__complex__} keyword | |
c1f7febf | 739 | |
4b404517 JM |
740 | ISO C99 supports complex floating data types, and as an extension GCC |
741 | supports them in C89 mode and in C++, and supports complex integer data | |
742 | types which are not part of ISO C99. You can declare complex types | |
743 | using the keyword @code{_Complex}. As an extension, the older GNU | |
744 | keyword @code{__complex__} is also supported. | |
c1f7febf | 745 | |
4b404517 | 746 | For example, @samp{_Complex double x;} declares @code{x} as a |
c1f7febf | 747 | variable whose real part and imaginary part are both of type |
4b404517 | 748 | @code{double}. @samp{_Complex short int y;} declares @code{y} to |
c1f7febf RK |
749 | have real and imaginary parts of type @code{short int}; this is not |
750 | likely to be useful, but it shows that the set of complex types is | |
751 | complete. | |
752 | ||
753 | To write a constant with a complex data type, use the suffix @samp{i} or | |
754 | @samp{j} (either one; they are equivalent). For example, @code{2.5fi} | |
4b404517 JM |
755 | has type @code{_Complex float} and @code{3i} has type |
756 | @code{_Complex int}. Such a constant always has a pure imaginary | |
c1f7febf | 757 | value, but you can form any complex value you like by adding one to a |
4b404517 JM |
758 | real constant. This is a GNU extension; if you have an ISO C99 |
759 | conforming C library (such as GNU libc), and want to construct complex | |
760 | constants of floating type, you should include @code{<complex.h>} and | |
761 | use the macros @code{I} or @code{_Complex_I} instead. | |
c1f7febf | 762 | |
4b404517 JM |
763 | @cindex @code{__real__} keyword |
764 | @cindex @code{__imag__} keyword | |
c1f7febf RK |
765 | To extract the real part of a complex-valued expression @var{exp}, write |
766 | @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to | |
4b404517 JM |
767 | extract the imaginary part. This is a GNU extension; for values of |
768 | floating type, you should use the ISO C99 functions @code{crealf}, | |
769 | @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and | |
770 | @code{cimagl}, declared in @code{<complex.h>} and also provided as | |
161d7b59 | 771 | built-in functions by GCC@. |
c1f7febf | 772 | |
4b404517 | 773 | @cindex complex conjugation |
c1f7febf | 774 | The operator @samp{~} performs complex conjugation when used on a value |
4b404517 JM |
775 | with a complex type. This is a GNU extension; for values of |
776 | floating type, you should use the ISO C99 functions @code{conjf}, | |
777 | @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also | |
161d7b59 | 778 | provided as built-in functions by GCC@. |
c1f7febf | 779 | |
f0523f02 | 780 | GCC can allocate complex automatic variables in a noncontiguous |
c1f7febf | 781 | fashion; it's even possible for the real part to be in a register while |
580fb356 JW |
782 | the imaginary part is on the stack (or vice-versa). Only the DWARF2 |
783 | debug info format can represent this, so use of DWARF2 is recommended. | |
784 | If you are using the stabs debug info format, GCC describes a noncontiguous | |
785 | complex variable as if it were two separate variables of noncomplex type. | |
c1f7febf RK |
786 | If the variable's actual name is @code{foo}, the two fictitious |
787 | variables are named @code{foo$real} and @code{foo$imag}. You can | |
788 | examine and set these two fictitious variables with your debugger. | |
789 | ||
6f4d7222 | 790 | @node Hex Floats |
6b42b9ea UD |
791 | @section Hex Floats |
792 | @cindex hex floats | |
c5c76735 | 793 | |
4b404517 | 794 | ISO C99 supports floating-point numbers written not only in the usual |
6f4d7222 | 795 | decimal notation, such as @code{1.55e1}, but also numbers such as |
4b404517 JM |
796 | @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC |
797 | supports this in C89 mode (except in some cases when strictly | |
798 | conforming) and in C++. In that format the | |
84330467 | 799 | @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are |
6f4d7222 | 800 | mandatory. The exponent is a decimal number that indicates the power of |
84330467 | 801 | 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is |
aee96fe9 JM |
802 | @tex |
803 | $1 {15\over16}$, | |
804 | @end tex | |
805 | @ifnottex | |
806 | 1 15/16, | |
807 | @end ifnottex | |
808 | @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3} | |
6f4d7222 UD |
809 | is the same as @code{1.55e1}. |
810 | ||
811 | Unlike for floating-point numbers in the decimal notation the exponent | |
812 | is always required in the hexadecimal notation. Otherwise the compiler | |
813 | would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This | |
84330467 | 814 | could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the |
6f4d7222 UD |
815 | extension for floating-point constants of type @code{float}. |
816 | ||
c1f7febf RK |
817 | @node Zero Length |
818 | @section Arrays of Length Zero | |
819 | @cindex arrays of length zero | |
820 | @cindex zero-length arrays | |
821 | @cindex length-zero arrays | |
ffc5c6a9 | 822 | @cindex flexible array members |
c1f7febf | 823 | |
161d7b59 | 824 | Zero-length arrays are allowed in GNU C@. They are very useful as the |
584ef5fe | 825 | last element of a structure which is really a header for a variable-length |
c1f7febf RK |
826 | object: |
827 | ||
3ab51846 | 828 | @smallexample |
c1f7febf RK |
829 | struct line @{ |
830 | int length; | |
831 | char contents[0]; | |
832 | @}; | |
833 | ||
584ef5fe RH |
834 | struct line *thisline = (struct line *) |
835 | malloc (sizeof (struct line) + this_length); | |
836 | thisline->length = this_length; | |
3ab51846 | 837 | @end smallexample |
c1f7febf | 838 | |
3764f879 | 839 | In ISO C90, you would have to give @code{contents} a length of 1, which |
c1f7febf RK |
840 | means either you waste space or complicate the argument to @code{malloc}. |
841 | ||
02f52e19 | 842 | In ISO C99, you would use a @dfn{flexible array member}, which is |
584ef5fe RH |
843 | slightly different in syntax and semantics: |
844 | ||
845 | @itemize @bullet | |
846 | @item | |
847 | Flexible array members are written as @code{contents[]} without | |
848 | the @code{0}. | |
849 | ||
850 | @item | |
851 | Flexible array members have incomplete type, and so the @code{sizeof} | |
852 | operator may not be applied. As a quirk of the original implementation | |
853 | of zero-length arrays, @code{sizeof} evaluates to zero. | |
854 | ||
855 | @item | |
856 | Flexible array members may only appear as the last member of a | |
e7b6a0ee | 857 | @code{struct} that is otherwise non-empty. |
2984fe64 JM |
858 | |
859 | @item | |
860 | A structure containing a flexible array member, or a union containing | |
861 | such a structure (possibly recursively), may not be a member of a | |
862 | structure or an element of an array. (However, these uses are | |
863 | permitted by GCC as extensions.) | |
ffc5c6a9 | 864 | @end itemize |
a25f1211 | 865 | |
ffc5c6a9 | 866 | GCC versions before 3.0 allowed zero-length arrays to be statically |
e7b6a0ee DD |
867 | initialized, as if they were flexible arrays. In addition to those |
868 | cases that were useful, it also allowed initializations in situations | |
869 | that would corrupt later data. Non-empty initialization of zero-length | |
870 | arrays is now treated like any case where there are more initializer | |
871 | elements than the array holds, in that a suitable warning about "excess | |
872 | elements in array" is given, and the excess elements (all of them, in | |
873 | this case) are ignored. | |
ffc5c6a9 RH |
874 | |
875 | Instead GCC allows static initialization of flexible array members. | |
876 | This is equivalent to defining a new structure containing the original | |
877 | structure followed by an array of sufficient size to contain the data. | |
e979f9e8 | 878 | I.e.@: in the following, @code{f1} is constructed as if it were declared |
ffc5c6a9 | 879 | like @code{f2}. |
a25f1211 | 880 | |
3ab51846 | 881 | @smallexample |
ffc5c6a9 RH |
882 | struct f1 @{ |
883 | int x; int y[]; | |
884 | @} f1 = @{ 1, @{ 2, 3, 4 @} @}; | |
885 | ||
886 | struct f2 @{ | |
887 | struct f1 f1; int data[3]; | |
888 | @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @}; | |
3ab51846 | 889 | @end smallexample |
584ef5fe | 890 | |
ffc5c6a9 RH |
891 | @noindent |
892 | The convenience of this extension is that @code{f1} has the desired | |
893 | type, eliminating the need to consistently refer to @code{f2.f1}. | |
894 | ||
895 | This has symmetry with normal static arrays, in that an array of | |
896 | unknown size is also written with @code{[]}. | |
a25f1211 | 897 | |
ffc5c6a9 RH |
898 | Of course, this extension only makes sense if the extra data comes at |
899 | the end of a top-level object, as otherwise we would be overwriting | |
900 | data at subsequent offsets. To avoid undue complication and confusion | |
901 | with initialization of deeply nested arrays, we simply disallow any | |
902 | non-empty initialization except when the structure is the top-level | |
903 | object. For example: | |
584ef5fe | 904 | |
3ab51846 | 905 | @smallexample |
ffc5c6a9 RH |
906 | struct foo @{ int x; int y[]; @}; |
907 | struct bar @{ struct foo z; @}; | |
908 | ||
13ba36b4 JM |
909 | struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.} |
910 | struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.} | |
911 | struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.} | |
912 | struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.} | |
3ab51846 | 913 | @end smallexample |
4b606faf | 914 | |
ba05abd3 GK |
915 | @node Empty Structures |
916 | @section Structures With No Members | |
917 | @cindex empty structures | |
918 | @cindex zero-size structures | |
919 | ||
920 | GCC permits a C structure to have no members: | |
921 | ||
3ab51846 | 922 | @smallexample |
ba05abd3 GK |
923 | struct empty @{ |
924 | @}; | |
3ab51846 | 925 | @end smallexample |
ba05abd3 GK |
926 | |
927 | The structure will have size zero. In C++, empty structures are part | |
db0b376e MM |
928 | of the language. G++ treats empty structures as if they had a single |
929 | member of type @code{char}. | |
ba05abd3 | 930 | |
c1f7febf RK |
931 | @node Variable Length |
932 | @section Arrays of Variable Length | |
933 | @cindex variable-length arrays | |
934 | @cindex arrays of variable length | |
4b404517 | 935 | @cindex VLAs |
c1f7febf | 936 | |
4b404517 JM |
937 | Variable-length automatic arrays are allowed in ISO C99, and as an |
938 | extension GCC accepts them in C89 mode and in C++. (However, GCC's | |
939 | implementation of variable-length arrays does not yet conform in detail | |
940 | to the ISO C99 standard.) These arrays are | |
c1f7febf RK |
941 | declared like any other automatic arrays, but with a length that is not |
942 | a constant expression. The storage is allocated at the point of | |
943 | declaration and deallocated when the brace-level is exited. For | |
944 | example: | |
945 | ||
3ab51846 | 946 | @smallexample |
c1f7febf RK |
947 | FILE * |
948 | concat_fopen (char *s1, char *s2, char *mode) | |
949 | @{ | |
950 | char str[strlen (s1) + strlen (s2) + 1]; | |
951 | strcpy (str, s1); | |
952 | strcat (str, s2); | |
953 | return fopen (str, mode); | |
954 | @} | |
3ab51846 | 955 | @end smallexample |
c1f7febf RK |
956 | |
957 | @cindex scope of a variable length array | |
958 | @cindex variable-length array scope | |
959 | @cindex deallocating variable length arrays | |
960 | Jumping or breaking out of the scope of the array name deallocates the | |
961 | storage. Jumping into the scope is not allowed; you get an error | |
962 | message for it. | |
963 | ||
964 | @cindex @code{alloca} vs variable-length arrays | |
965 | You can use the function @code{alloca} to get an effect much like | |
966 | variable-length arrays. The function @code{alloca} is available in | |
967 | many other C implementations (but not in all). On the other hand, | |
968 | variable-length arrays are more elegant. | |
969 | ||
970 | There are other differences between these two methods. Space allocated | |
971 | with @code{alloca} exists until the containing @emph{function} returns. | |
972 | The space for a variable-length array is deallocated as soon as the array | |
973 | name's scope ends. (If you use both variable-length arrays and | |
974 | @code{alloca} in the same function, deallocation of a variable-length array | |
975 | will also deallocate anything more recently allocated with @code{alloca}.) | |
976 | ||
977 | You can also use variable-length arrays as arguments to functions: | |
978 | ||
3ab51846 | 979 | @smallexample |
c1f7febf RK |
980 | struct entry |
981 | tester (int len, char data[len][len]) | |
982 | @{ | |
0d893a63 | 983 | /* @r{@dots{}} */ |
c1f7febf | 984 | @} |
3ab51846 | 985 | @end smallexample |
c1f7febf RK |
986 | |
987 | The length of an array is computed once when the storage is allocated | |
988 | and is remembered for the scope of the array in case you access it with | |
989 | @code{sizeof}. | |
990 | ||
991 | If you want to pass the array first and the length afterward, you can | |
992 | use a forward declaration in the parameter list---another GNU extension. | |
993 | ||
3ab51846 | 994 | @smallexample |
c1f7febf RK |
995 | struct entry |
996 | tester (int len; char data[len][len], int len) | |
997 | @{ | |
0d893a63 | 998 | /* @r{@dots{}} */ |
c1f7febf | 999 | @} |
3ab51846 | 1000 | @end smallexample |
c1f7febf RK |
1001 | |
1002 | @cindex parameter forward declaration | |
1003 | The @samp{int len} before the semicolon is a @dfn{parameter forward | |
1004 | declaration}, and it serves the purpose of making the name @code{len} | |
1005 | known when the declaration of @code{data} is parsed. | |
1006 | ||
1007 | You can write any number of such parameter forward declarations in the | |
1008 | parameter list. They can be separated by commas or semicolons, but the | |
1009 | last one must end with a semicolon, which is followed by the ``real'' | |
1010 | parameter declarations. Each forward declaration must match a ``real'' | |
4b404517 JM |
1011 | declaration in parameter name and data type. ISO C99 does not support |
1012 | parameter forward declarations. | |
c1f7febf | 1013 | |
ccd96f0a NB |
1014 | @node Variadic Macros |
1015 | @section Macros with a Variable Number of Arguments. | |
c1f7febf RK |
1016 | @cindex variable number of arguments |
1017 | @cindex macro with variable arguments | |
1018 | @cindex rest argument (in macro) | |
ccd96f0a | 1019 | @cindex variadic macros |
c1f7febf | 1020 | |
ccd96f0a NB |
1021 | In the ISO C standard of 1999, a macro can be declared to accept a |
1022 | variable number of arguments much as a function can. The syntax for | |
1023 | defining the macro is similar to that of a function. Here is an | |
1024 | example: | |
c1f7febf | 1025 | |
478c9e72 | 1026 | @smallexample |
ccd96f0a | 1027 | #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__) |
478c9e72 | 1028 | @end smallexample |
c1f7febf | 1029 | |
ccd96f0a NB |
1030 | Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of |
1031 | such a macro, it represents the zero or more tokens until the closing | |
1032 | parenthesis that ends the invocation, including any commas. This set of | |
1033 | tokens replaces the identifier @code{__VA_ARGS__} in the macro body | |
1034 | wherever it appears. See the CPP manual for more information. | |
1035 | ||
1036 | GCC has long supported variadic macros, and used a different syntax that | |
1037 | allowed you to give a name to the variable arguments just like any other | |
1038 | argument. Here is an example: | |
c1f7febf | 1039 | |
3ab51846 | 1040 | @smallexample |
ccd96f0a | 1041 | #define debug(format, args...) fprintf (stderr, format, args) |
3ab51846 | 1042 | @end smallexample |
c1f7febf | 1043 | |
ccd96f0a NB |
1044 | This is in all ways equivalent to the ISO C example above, but arguably |
1045 | more readable and descriptive. | |
c1f7febf | 1046 | |
ccd96f0a NB |
1047 | GNU CPP has two further variadic macro extensions, and permits them to |
1048 | be used with either of the above forms of macro definition. | |
1049 | ||
1050 | In standard C, you are not allowed to leave the variable argument out | |
1051 | entirely; but you are allowed to pass an empty argument. For example, | |
1052 | this invocation is invalid in ISO C, because there is no comma after | |
1053 | the string: | |
c1f7febf | 1054 | |
3ab51846 | 1055 | @smallexample |
ccd96f0a | 1056 | debug ("A message") |
3ab51846 | 1057 | @end smallexample |
c1f7febf | 1058 | |
ccd96f0a NB |
1059 | GNU CPP permits you to completely omit the variable arguments in this |
1060 | way. In the above examples, the compiler would complain, though since | |
1061 | the expansion of the macro still has the extra comma after the format | |
1062 | string. | |
1063 | ||
1064 | To help solve this problem, CPP behaves specially for variable arguments | |
1065 | used with the token paste operator, @samp{##}. If instead you write | |
c1f7febf | 1066 | |
478c9e72 | 1067 | @smallexample |
ccd96f0a | 1068 | #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__) |
478c9e72 | 1069 | @end smallexample |
c1f7febf | 1070 | |
ccd96f0a NB |
1071 | and if the variable arguments are omitted or empty, the @samp{##} |
1072 | operator causes the preprocessor to remove the comma before it. If you | |
1073 | do provide some variable arguments in your macro invocation, GNU CPP | |
1074 | does not complain about the paste operation and instead places the | |
1075 | variable arguments after the comma. Just like any other pasted macro | |
1076 | argument, these arguments are not macro expanded. | |
1077 | ||
1078 | @node Escaped Newlines | |
1079 | @section Slightly Looser Rules for Escaped Newlines | |
1080 | @cindex escaped newlines | |
1081 | @cindex newlines (escaped) | |
1082 | ||
f458d1d5 ZW |
1083 | Recently, the preprocessor has relaxed its treatment of escaped |
1084 | newlines. Previously, the newline had to immediately follow a | |
e6cc3a24 ZW |
1085 | backslash. The current implementation allows whitespace in the form |
1086 | of spaces, horizontal and vertical tabs, and form feeds between the | |
ccd96f0a NB |
1087 | backslash and the subsequent newline. The preprocessor issues a |
1088 | warning, but treats it as a valid escaped newline and combines the two | |
1089 | lines to form a single logical line. This works within comments and | |
e6cc3a24 ZW |
1090 | tokens, as well as between tokens. Comments are @emph{not} treated as |
1091 | whitespace for the purposes of this relaxation, since they have not | |
1092 | yet been replaced with spaces. | |
c1f7febf RK |
1093 | |
1094 | @node Subscripting | |
1095 | @section Non-Lvalue Arrays May Have Subscripts | |
1096 | @cindex subscripting | |
1097 | @cindex arrays, non-lvalue | |
1098 | ||
1099 | @cindex subscripting and function values | |
207bf485 JM |
1100 | In ISO C99, arrays that are not lvalues still decay to pointers, and |
1101 | may be subscripted, although they may not be modified or used after | |
1102 | the next sequence point and the unary @samp{&} operator may not be | |
1103 | applied to them. As an extension, GCC allows such arrays to be | |
1104 | subscripted in C89 mode, though otherwise they do not decay to | |
1105 | pointers outside C99 mode. For example, | |
4b404517 | 1106 | this is valid in GNU C though not valid in C89: |
c1f7febf | 1107 | |
3ab51846 | 1108 | @smallexample |
c1f7febf RK |
1109 | @group |
1110 | struct foo @{int a[4];@}; | |
1111 | ||
1112 | struct foo f(); | |
1113 | ||
1114 | bar (int index) | |
1115 | @{ | |
1116 | return f().a[index]; | |
1117 | @} | |
1118 | @end group | |
3ab51846 | 1119 | @end smallexample |
c1f7febf RK |
1120 | |
1121 | @node Pointer Arith | |
1122 | @section Arithmetic on @code{void}- and Function-Pointers | |
1123 | @cindex void pointers, arithmetic | |
1124 | @cindex void, size of pointer to | |
1125 | @cindex function pointers, arithmetic | |
1126 | @cindex function, size of pointer to | |
1127 | ||
1128 | In GNU C, addition and subtraction operations are supported on pointers to | |
1129 | @code{void} and on pointers to functions. This is done by treating the | |
1130 | size of a @code{void} or of a function as 1. | |
1131 | ||
1132 | A consequence of this is that @code{sizeof} is also allowed on @code{void} | |
1133 | and on function types, and returns 1. | |
1134 | ||
84330467 JM |
1135 | @opindex Wpointer-arith |
1136 | The option @option{-Wpointer-arith} requests a warning if these extensions | |
c1f7febf RK |
1137 | are used. |
1138 | ||
1139 | @node Initializers | |
1140 | @section Non-Constant Initializers | |
1141 | @cindex initializers, non-constant | |
1142 | @cindex non-constant initializers | |
1143 | ||
4b404517 | 1144 | As in standard C++ and ISO C99, the elements of an aggregate initializer for an |
161d7b59 | 1145 | automatic variable are not required to be constant expressions in GNU C@. |
c1f7febf RK |
1146 | Here is an example of an initializer with run-time varying elements: |
1147 | ||
3ab51846 | 1148 | @smallexample |
c1f7febf RK |
1149 | foo (float f, float g) |
1150 | @{ | |
1151 | float beat_freqs[2] = @{ f-g, f+g @}; | |
0d893a63 | 1152 | /* @r{@dots{}} */ |
c1f7febf | 1153 | @} |
3ab51846 | 1154 | @end smallexample |
c1f7febf | 1155 | |
4b404517 JM |
1156 | @node Compound Literals |
1157 | @section Compound Literals | |
c1f7febf RK |
1158 | @cindex constructor expressions |
1159 | @cindex initializations in expressions | |
1160 | @cindex structures, constructor expression | |
1161 | @cindex expressions, constructor | |
4b404517 JM |
1162 | @cindex compound literals |
1163 | @c The GNU C name for what C99 calls compound literals was "constructor expressions". | |
c1f7febf | 1164 | |
4b404517 | 1165 | ISO C99 supports compound literals. A compound literal looks like |
c1f7febf RK |
1166 | a cast containing an initializer. Its value is an object of the |
1167 | type specified in the cast, containing the elements specified in | |
db3acfa5 JM |
1168 | the initializer; it is an lvalue. As an extension, GCC supports |
1169 | compound literals in C89 mode and in C++. | |
c1f7febf RK |
1170 | |
1171 | Usually, the specified type is a structure. Assume that | |
1172 | @code{struct foo} and @code{structure} are declared as shown: | |
1173 | ||
3ab51846 | 1174 | @smallexample |
c1f7febf | 1175 | struct foo @{int a; char b[2];@} structure; |
3ab51846 | 1176 | @end smallexample |
c1f7febf RK |
1177 | |
1178 | @noindent | |
4b404517 | 1179 | Here is an example of constructing a @code{struct foo} with a compound literal: |
c1f7febf | 1180 | |
3ab51846 | 1181 | @smallexample |
c1f7febf | 1182 | structure = ((struct foo) @{x + y, 'a', 0@}); |
3ab51846 | 1183 | @end smallexample |
c1f7febf RK |
1184 | |
1185 | @noindent | |
1186 | This is equivalent to writing the following: | |
1187 | ||
3ab51846 | 1188 | @smallexample |
c1f7febf RK |
1189 | @{ |
1190 | struct foo temp = @{x + y, 'a', 0@}; | |
1191 | structure = temp; | |
1192 | @} | |
3ab51846 | 1193 | @end smallexample |
c1f7febf | 1194 | |
4b404517 | 1195 | You can also construct an array. If all the elements of the compound literal |
c1f7febf | 1196 | are (made up of) simple constant expressions, suitable for use in |
db3acfa5 JM |
1197 | initializers of objects of static storage duration, then the compound |
1198 | literal can be coerced to a pointer to its first element and used in | |
1199 | such an initializer, as shown here: | |
c1f7febf | 1200 | |
3ab51846 | 1201 | @smallexample |
c1f7febf | 1202 | char **foo = (char *[]) @{ "x", "y", "z" @}; |
3ab51846 | 1203 | @end smallexample |
c1f7febf | 1204 | |
4b404517 JM |
1205 | Compound literals for scalar types and union types are is |
1206 | also allowed, but then the compound literal is equivalent | |
c1f7febf RK |
1207 | to a cast. |
1208 | ||
59c83dbf JJ |
1209 | As a GNU extension, GCC allows initialization of objects with static storage |
1210 | duration by compound literals (which is not possible in ISO C99, because | |
1211 | the initializer is not a constant). | |
1212 | It is handled as if the object was initialized only with the bracket | |
1213 | enclosed list if compound literal's and object types match. | |
1214 | The initializer list of the compound literal must be constant. | |
1215 | If the object being initialized has array type of unknown size, the size is | |
ad47f1e5 | 1216 | determined by compound literal size. |
59c83dbf | 1217 | |
3ab51846 | 1218 | @smallexample |
59c83dbf JJ |
1219 | static struct foo x = (struct foo) @{1, 'a', 'b'@}; |
1220 | static int y[] = (int []) @{1, 2, 3@}; | |
1221 | static int z[] = (int [3]) @{1@}; | |
3ab51846 | 1222 | @end smallexample |
59c83dbf JJ |
1223 | |
1224 | @noindent | |
1225 | The above lines are equivalent to the following: | |
3ab51846 | 1226 | @smallexample |
59c83dbf JJ |
1227 | static struct foo x = @{1, 'a', 'b'@}; |
1228 | static int y[] = @{1, 2, 3@}; | |
ad47f1e5 | 1229 | static int z[] = @{1, 0, 0@}; |
3ab51846 | 1230 | @end smallexample |
59c83dbf | 1231 | |
4b404517 JM |
1232 | @node Designated Inits |
1233 | @section Designated Initializers | |
c1f7febf RK |
1234 | @cindex initializers with labeled elements |
1235 | @cindex labeled elements in initializers | |
1236 | @cindex case labels in initializers | |
4b404517 | 1237 | @cindex designated initializers |
c1f7febf | 1238 | |
26d4fec7 | 1239 | Standard C89 requires the elements of an initializer to appear in a fixed |
c1f7febf RK |
1240 | order, the same as the order of the elements in the array or structure |
1241 | being initialized. | |
1242 | ||
26d4fec7 JM |
1243 | In ISO C99 you can give the elements in any order, specifying the array |
1244 | indices or structure field names they apply to, and GNU C allows this as | |
1245 | an extension in C89 mode as well. This extension is not | |
c1f7febf RK |
1246 | implemented in GNU C++. |
1247 | ||
26d4fec7 | 1248 | To specify an array index, write |
c1f7febf RK |
1249 | @samp{[@var{index}] =} before the element value. For example, |
1250 | ||
3ab51846 | 1251 | @smallexample |
26d4fec7 | 1252 | int a[6] = @{ [4] = 29, [2] = 15 @}; |
3ab51846 | 1253 | @end smallexample |
c1f7febf RK |
1254 | |
1255 | @noindent | |
1256 | is equivalent to | |
1257 | ||
3ab51846 | 1258 | @smallexample |
c1f7febf | 1259 | int a[6] = @{ 0, 0, 15, 0, 29, 0 @}; |
3ab51846 | 1260 | @end smallexample |
c1f7febf RK |
1261 | |
1262 | @noindent | |
1263 | The index values must be constant expressions, even if the array being | |
1264 | initialized is automatic. | |
1265 | ||
26d4fec7 JM |
1266 | An alternative syntax for this which has been obsolete since GCC 2.5 but |
1267 | GCC still accepts is to write @samp{[@var{index}]} before the element | |
1268 | value, with no @samp{=}. | |
1269 | ||
c1f7febf | 1270 | To initialize a range of elements to the same value, write |
26d4fec7 JM |
1271 | @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU |
1272 | extension. For example, | |
c1f7febf | 1273 | |
3ab51846 | 1274 | @smallexample |
c1f7febf | 1275 | int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @}; |
3ab51846 | 1276 | @end smallexample |
c1f7febf | 1277 | |
8b6a5902 JJ |
1278 | @noindent |
1279 | If the value in it has side-effects, the side-effects will happen only once, | |
1280 | not for each initialized field by the range initializer. | |
1281 | ||
c1f7febf RK |
1282 | @noindent |
1283 | Note that the length of the array is the highest value specified | |
1284 | plus one. | |
1285 | ||
1286 | In a structure initializer, specify the name of a field to initialize | |
26d4fec7 | 1287 | with @samp{.@var{fieldname} =} before the element value. For example, |
c1f7febf RK |
1288 | given the following structure, |
1289 | ||
3ab51846 | 1290 | @smallexample |
c1f7febf | 1291 | struct point @{ int x, y; @}; |
3ab51846 | 1292 | @end smallexample |
c1f7febf RK |
1293 | |
1294 | @noindent | |
1295 | the following initialization | |
1296 | ||
3ab51846 | 1297 | @smallexample |
26d4fec7 | 1298 | struct point p = @{ .y = yvalue, .x = xvalue @}; |
3ab51846 | 1299 | @end smallexample |
c1f7febf RK |
1300 | |
1301 | @noindent | |
1302 | is equivalent to | |
1303 | ||
3ab51846 | 1304 | @smallexample |
c1f7febf | 1305 | struct point p = @{ xvalue, yvalue @}; |
3ab51846 | 1306 | @end smallexample |
c1f7febf | 1307 | |
26d4fec7 JM |
1308 | Another syntax which has the same meaning, obsolete since GCC 2.5, is |
1309 | @samp{@var{fieldname}:}, as shown here: | |
c1f7febf | 1310 | |
3ab51846 | 1311 | @smallexample |
26d4fec7 | 1312 | struct point p = @{ y: yvalue, x: xvalue @}; |
3ab51846 | 1313 | @end smallexample |
c1f7febf | 1314 | |
4b404517 JM |
1315 | @cindex designators |
1316 | The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a | |
1317 | @dfn{designator}. You can also use a designator (or the obsolete colon | |
1318 | syntax) when initializing a union, to specify which element of the union | |
1319 | should be used. For example, | |
c1f7febf | 1320 | |
3ab51846 | 1321 | @smallexample |
c1f7febf RK |
1322 | union foo @{ int i; double d; @}; |
1323 | ||
26d4fec7 | 1324 | union foo f = @{ .d = 4 @}; |
3ab51846 | 1325 | @end smallexample |
c1f7febf RK |
1326 | |
1327 | @noindent | |
1328 | will convert 4 to a @code{double} to store it in the union using | |
1329 | the second element. By contrast, casting 4 to type @code{union foo} | |
1330 | would store it into the union as the integer @code{i}, since it is | |
1331 | an integer. (@xref{Cast to Union}.) | |
1332 | ||
1333 | You can combine this technique of naming elements with ordinary C | |
1334 | initialization of successive elements. Each initializer element that | |
4b404517 | 1335 | does not have a designator applies to the next consecutive element of the |
c1f7febf RK |
1336 | array or structure. For example, |
1337 | ||
3ab51846 | 1338 | @smallexample |
c1f7febf | 1339 | int a[6] = @{ [1] = v1, v2, [4] = v4 @}; |
3ab51846 | 1340 | @end smallexample |
c1f7febf RK |
1341 | |
1342 | @noindent | |
1343 | is equivalent to | |
1344 | ||
3ab51846 | 1345 | @smallexample |
c1f7febf | 1346 | int a[6] = @{ 0, v1, v2, 0, v4, 0 @}; |
3ab51846 | 1347 | @end smallexample |
c1f7febf RK |
1348 | |
1349 | Labeling the elements of an array initializer is especially useful | |
1350 | when the indices are characters or belong to an @code{enum} type. | |
1351 | For example: | |
1352 | ||
3ab51846 | 1353 | @smallexample |
c1f7febf RK |
1354 | int whitespace[256] |
1355 | = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1, | |
1356 | ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @}; | |
3ab51846 | 1357 | @end smallexample |
c1f7febf | 1358 | |
4b404517 | 1359 | @cindex designator lists |
26d4fec7 | 1360 | You can also write a series of @samp{.@var{fieldname}} and |
4b404517 | 1361 | @samp{[@var{index}]} designators before an @samp{=} to specify a |
26d4fec7 JM |
1362 | nested subobject to initialize; the list is taken relative to the |
1363 | subobject corresponding to the closest surrounding brace pair. For | |
1364 | example, with the @samp{struct point} declaration above: | |
1365 | ||
478c9e72 | 1366 | @smallexample |
26d4fec7 | 1367 | struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @}; |
478c9e72 | 1368 | @end smallexample |
26d4fec7 | 1369 | |
8b6a5902 JJ |
1370 | @noindent |
1371 | If the same field is initialized multiple times, it will have value from | |
1372 | the last initialization. If any such overridden initialization has | |
1373 | side-effect, it is unspecified whether the side-effect happens or not. | |
2dd76960 | 1374 | Currently, GCC will discard them and issue a warning. |
8b6a5902 | 1375 | |
c1f7febf RK |
1376 | @node Case Ranges |
1377 | @section Case Ranges | |
1378 | @cindex case ranges | |
1379 | @cindex ranges in case statements | |
1380 | ||
1381 | You can specify a range of consecutive values in a single @code{case} label, | |
1382 | like this: | |
1383 | ||
3ab51846 | 1384 | @smallexample |
c1f7febf | 1385 | case @var{low} ... @var{high}: |
3ab51846 | 1386 | @end smallexample |
c1f7febf RK |
1387 | |
1388 | @noindent | |
1389 | This has the same effect as the proper number of individual @code{case} | |
1390 | labels, one for each integer value from @var{low} to @var{high}, inclusive. | |
1391 | ||
1392 | This feature is especially useful for ranges of ASCII character codes: | |
1393 | ||
3ab51846 | 1394 | @smallexample |
c1f7febf | 1395 | case 'A' ... 'Z': |
3ab51846 | 1396 | @end smallexample |
c1f7febf RK |
1397 | |
1398 | @strong{Be careful:} Write spaces around the @code{...}, for otherwise | |
1399 | it may be parsed wrong when you use it with integer values. For example, | |
1400 | write this: | |
1401 | ||
3ab51846 | 1402 | @smallexample |
c1f7febf | 1403 | case 1 ... 5: |
3ab51846 | 1404 | @end smallexample |
c1f7febf RK |
1405 | |
1406 | @noindent | |
1407 | rather than this: | |
1408 | ||
3ab51846 | 1409 | @smallexample |
c1f7febf | 1410 | case 1...5: |
3ab51846 | 1411 | @end smallexample |
c1f7febf RK |
1412 | |
1413 | @node Cast to Union | |
1414 | @section Cast to a Union Type | |
1415 | @cindex cast to a union | |
1416 | @cindex union, casting to a | |
1417 | ||
1418 | A cast to union type is similar to other casts, except that the type | |
1419 | specified is a union type. You can specify the type either with | |
1420 | @code{union @var{tag}} or with a typedef name. A cast to union is actually | |
1421 | a constructor though, not a cast, and hence does not yield an lvalue like | |
4b404517 | 1422 | normal casts. (@xref{Compound Literals}.) |
c1f7febf RK |
1423 | |
1424 | The types that may be cast to the union type are those of the members | |
1425 | of the union. Thus, given the following union and variables: | |
1426 | ||
3ab51846 | 1427 | @smallexample |
c1f7febf RK |
1428 | union foo @{ int i; double d; @}; |
1429 | int x; | |
1430 | double y; | |
3ab51846 | 1431 | @end smallexample |
c1f7febf RK |
1432 | |
1433 | @noindent | |
aee96fe9 | 1434 | both @code{x} and @code{y} can be cast to type @code{union foo}. |
c1f7febf RK |
1435 | |
1436 | Using the cast as the right-hand side of an assignment to a variable of | |
1437 | union type is equivalent to storing in a member of the union: | |
1438 | ||
3ab51846 | 1439 | @smallexample |
c1f7febf | 1440 | union foo u; |
0d893a63 | 1441 | /* @r{@dots{}} */ |
c1f7febf RK |
1442 | u = (union foo) x @equiv{} u.i = x |
1443 | u = (union foo) y @equiv{} u.d = y | |
3ab51846 | 1444 | @end smallexample |
c1f7febf RK |
1445 | |
1446 | You can also use the union cast as a function argument: | |
1447 | ||
3ab51846 | 1448 | @smallexample |
c1f7febf | 1449 | void hack (union foo); |
0d893a63 | 1450 | /* @r{@dots{}} */ |
c1f7febf | 1451 | hack ((union foo) x); |
3ab51846 | 1452 | @end smallexample |
c1f7febf | 1453 | |
4b404517 JM |
1454 | @node Mixed Declarations |
1455 | @section Mixed Declarations and Code | |
1456 | @cindex mixed declarations and code | |
1457 | @cindex declarations, mixed with code | |
1458 | @cindex code, mixed with declarations | |
1459 | ||
1460 | ISO C99 and ISO C++ allow declarations and code to be freely mixed | |
1461 | within compound statements. As an extension, GCC also allows this in | |
1462 | C89 mode. For example, you could do: | |
1463 | ||
3ab51846 | 1464 | @smallexample |
4b404517 | 1465 | int i; |
0d893a63 | 1466 | /* @r{@dots{}} */ |
4b404517 JM |
1467 | i++; |
1468 | int j = i + 2; | |
3ab51846 | 1469 | @end smallexample |
4b404517 JM |
1470 | |
1471 | Each identifier is visible from where it is declared until the end of | |
1472 | the enclosing block. | |
1473 | ||
c1f7febf RK |
1474 | @node Function Attributes |
1475 | @section Declaring Attributes of Functions | |
1476 | @cindex function attributes | |
1477 | @cindex declaring attributes of functions | |
1478 | @cindex functions that never return | |
1479 | @cindex functions that have no side effects | |
1480 | @cindex functions in arbitrary sections | |
2a59078d | 1481 | @cindex functions that behave like malloc |
c1f7febf RK |
1482 | @cindex @code{volatile} applied to function |
1483 | @cindex @code{const} applied to function | |
26f6672d | 1484 | @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments |
b34c7881 | 1485 | @cindex functions with non-null pointer arguments |
c1f7febf RK |
1486 | @cindex functions that are passed arguments in registers on the 386 |
1487 | @cindex functions that pop the argument stack on the 386 | |
1488 | @cindex functions that do not pop the argument stack on the 386 | |
1489 | ||
1490 | In GNU C, you declare certain things about functions called in your program | |
1491 | which help the compiler optimize function calls and check your code more | |
1492 | carefully. | |
1493 | ||
1494 | The keyword @code{__attribute__} allows you to specify special | |
1495 | attributes when making a declaration. This keyword is followed by an | |
9162542e | 1496 | attribute specification inside double parentheses. The following |
eacecf96 | 1497 | attributes are currently defined for functions on all targets: |
6aa77e6c | 1498 | @code{noreturn}, @code{noinline}, @code{always_inline}, |
3d091dac | 1499 | @code{pure}, @code{const}, @code{nothrow}, @code{sentinel}, |
9162542e AO |
1500 | @code{format}, @code{format_arg}, @code{no_instrument_function}, |
1501 | @code{section}, @code{constructor}, @code{destructor}, @code{used}, | |
b34c7881 | 1502 | @code{unused}, @code{deprecated}, @code{weak}, @code{malloc}, |
72954a4f JM |
1503 | @code{alias}, @code{warn_unused_result} and @code{nonnull}. Several other |
1504 | attributes are defined for functions on particular target systems. Other | |
1505 | attributes, including @code{section} are supported for variables declarations | |
b34c7881 | 1506 | (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}). |
c1f7febf RK |
1507 | |
1508 | You may also specify attributes with @samp{__} preceding and following | |
1509 | each keyword. This allows you to use them in header files without | |
1510 | being concerned about a possible macro of the same name. For example, | |
1511 | you may use @code{__noreturn__} instead of @code{noreturn}. | |
1512 | ||
2c5e91d2 JM |
1513 | @xref{Attribute Syntax}, for details of the exact syntax for using |
1514 | attributes. | |
1515 | ||
c1f7febf | 1516 | @table @code |
8a36672b | 1517 | @c Keep this table alphabetized by attribute name. Treat _ as space. |
c1f7febf | 1518 | |
c8619b90 NS |
1519 | @item alias ("@var{target}") |
1520 | @cindex @code{alias} attribute | |
1521 | The @code{alias} attribute causes the declaration to be emitted as an | |
1522 | alias for another symbol, which must be specified. For instance, | |
c1f7febf RK |
1523 | |
1524 | @smallexample | |
c8619b90 NS |
1525 | void __f () @{ /* @r{Do something.} */; @} |
1526 | void f () __attribute__ ((weak, alias ("__f"))); | |
c1f7febf RK |
1527 | @end smallexample |
1528 | ||
c8619b90 NS |
1529 | declares @samp{f} to be a weak alias for @samp{__f}. In C++, the |
1530 | mangled name for the target must be used. | |
1531 | ||
1532 | Not all target machines support this attribute. | |
9162542e | 1533 | |
6aa77e6c | 1534 | @item always_inline |
c8619b90 | 1535 | @cindex @code{always_inline} function attribute |
6aa77e6c AH |
1536 | Generally, functions are not inlined unless optimization is specified. |
1537 | For functions declared inline, this attribute inlines the function even | |
1538 | if no optimization level was specified. | |
1539 | ||
c8619b90 NS |
1540 | @item cdecl |
1541 | @cindex functions that do pop the argument stack on the 386 | |
1542 | @opindex mrtd | |
1543 | On the Intel 386, the @code{cdecl} attribute causes the compiler to | |
1544 | assume that the calling function will pop off the stack space used to | |
1545 | pass arguments. This is | |
1546 | useful to override the effects of the @option{-mrtd} switch. | |
2a8f6b90 | 1547 | |
2a8f6b90 | 1548 | @item const |
c8619b90 | 1549 | @cindex @code{const} function attribute |
2a8f6b90 JH |
1550 | Many functions do not examine any values except their arguments, and |
1551 | have no effects except the return value. Basically this is just slightly | |
84330467 | 1552 | more strict class than the @code{pure} attribute above, since function is not |
2a59078d | 1553 | allowed to read global memory. |
2a8f6b90 JH |
1554 | |
1555 | @cindex pointer arguments | |
1556 | Note that a function that has pointer arguments and examines the data | |
1557 | pointed to must @emph{not} be declared @code{const}. Likewise, a | |
1558 | function that calls a non-@code{const} function usually must not be | |
1559 | @code{const}. It does not make sense for a @code{const} function to | |
1560 | return @code{void}. | |
1561 | ||
f0523f02 | 1562 | The attribute @code{const} is not implemented in GCC versions earlier |
c1f7febf RK |
1563 | than 2.5. An alternative way to declare that a function has no side |
1564 | effects, which works in the current version and in some older versions, | |
1565 | is as follows: | |
1566 | ||
1567 | @smallexample | |
1568 | typedef int intfn (); | |
1569 | ||
1570 | extern const intfn square; | |
1571 | @end smallexample | |
1572 | ||
1573 | This approach does not work in GNU C++ from 2.6.0 on, since the language | |
1574 | specifies that the @samp{const} must be attached to the return value. | |
1575 | ||
c8619b90 NS |
1576 | @item constructor |
1577 | @itemx destructor | |
1578 | @cindex @code{constructor} function attribute | |
1579 | @cindex @code{destructor} function attribute | |
1580 | The @code{constructor} attribute causes the function to be called | |
1581 | automatically before execution enters @code{main ()}. Similarly, the | |
1582 | @code{destructor} attribute causes the function to be called | |
1583 | automatically after @code{main ()} has completed or @code{exit ()} has | |
1584 | been called. Functions with these attributes are useful for | |
1585 | initializing data that will be used implicitly during the execution of | |
1586 | the program. | |
1587 | ||
1588 | These attributes are not currently implemented for Objective-C@. | |
1589 | ||
1590 | @item deprecated | |
1591 | @cindex @code{deprecated} attribute. | |
1592 | The @code{deprecated} attribute results in a warning if the function | |
1593 | is used anywhere in the source file. This is useful when identifying | |
1594 | functions that are expected to be removed in a future version of a | |
1595 | program. The warning also includes the location of the declaration | |
1596 | of the deprecated function, to enable users to easily find further | |
1597 | information about why the function is deprecated, or what they should | |
1598 | do instead. Note that the warnings only occurs for uses: | |
1599 | ||
1600 | @smallexample | |
1601 | int old_fn () __attribute__ ((deprecated)); | |
1602 | int old_fn (); | |
1603 | int (*fn_ptr)() = old_fn; | |
1604 | @end smallexample | |
1605 | ||
1606 | results in a warning on line 3 but not line 2. | |
1607 | ||
1608 | The @code{deprecated} attribute can also be used for variables and | |
1609 | types (@pxref{Variable Attributes}, @pxref{Type Attributes}.) | |
1610 | ||
1611 | @item dllexport | |
1612 | @cindex @code{__declspec(dllexport)} | |
b2ca3702 MM |
1613 | On Microsoft Windows targets and Symbian OS targets the |
1614 | @code{dllexport} attribute causes the compiler to provide a global | |
1615 | pointer to a pointer in a DLL, so that it can be referenced with the | |
1616 | @code{dllimport} attribute. On Microsoft Windows targets, the pointer | |
1617 | name is formed by combining @code{_imp__} and the function or variable | |
1618 | name. | |
1619 | ||
1620 | You can use @code{__declspec(dllexport)} as a synonym for | |
1621 | @code{__attribute__ ((dllexport))} for compatibility with other | |
1622 | compilers. | |
1623 | ||
1624 | On systems that support the @code{visibility} attribute, this | |
1625 | attribute also implies ``default'' visibility, unless a | |
1626 | @code{visibility} attribute is explicitly specified. You should avoid | |
1627 | the use of @code{dllexport} with ``hidden'' or ``internal'' | |
1628 | visibility; in the future GCC may issue an error for those cases. | |
c8619b90 | 1629 | |
b2ca3702 MM |
1630 | Currently, the @code{dllexport} attribute is ignored for inlined |
1631 | functions, unless the @option{-fkeep-inline-functions} flag has been | |
1632 | used. The attribute is also ignored for undefined symbols. | |
c8619b90 | 1633 | |
8a36672b JM |
1634 | When applied to C++ classes, the attribute marks defined non-inlined |
1635 | member functions and static data members as exports. Static consts | |
c8619b90 NS |
1636 | initialized in-class are not marked unless they are also defined |
1637 | out-of-class. | |
1638 | ||
b55e3aad | 1639 | For Microsoft Windows targets there are alternative methods for |
b2ca3702 | 1640 | including the symbol in the DLL's export table such as using a |
b55e3aad NC |
1641 | @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using |
1642 | the @option{--export-all} linker flag. | |
c8619b90 NS |
1643 | |
1644 | @item dllimport | |
1645 | @cindex @code{__declspec(dllimport)} | |
b2ca3702 | 1646 | On Microsoft Windows and Symbian OS targets, the @code{dllimport} |
b55e3aad | 1647 | attribute causes the compiler to reference a function or variable via |
b2ca3702 | 1648 | a global pointer to a pointer that is set up by the DLL exporting the |
8a36672b | 1649 | symbol. The attribute implies @code{extern} storage. On Microsoft |
b2ca3702 MM |
1650 | Windows targets, the pointer name is formed by combining @code{_imp__} |
1651 | and the function or variable name. | |
1652 | ||
1653 | You can use @code{__declspec(dllimport)} as a synonym for | |
1654 | @code{__attribute__ ((dllimport))} for compatibility with other | |
1655 | compilers. | |
c8619b90 | 1656 | |
8a36672b | 1657 | Currently, the attribute is ignored for inlined functions. If the |
c8619b90 NS |
1658 | attribute is applied to a symbol @emph{definition}, an error is reported. |
1659 | If a symbol previously declared @code{dllimport} is later defined, the | |
1660 | attribute is ignored in subsequent references, and a warning is emitted. | |
1661 | The attribute is also overridden by a subsequent declaration as | |
1662 | @code{dllexport}. | |
1663 | ||
1664 | When applied to C++ classes, the attribute marks non-inlined | |
1665 | member functions and static data members as imports. However, the | |
1666 | attribute is ignored for virtual methods to allow creation of vtables | |
1667 | using thunks. | |
1668 | ||
b2ca3702 | 1669 | On the SH Symbian OS target the @code{dllimport} attribute also has |
78466c0e | 1670 | another affect---it can cause the vtable and run-time type information |
b2ca3702 MM |
1671 | for a class to be exported. This happens when the class has a |
1672 | dllimport'ed constructor or a non-inline, non-pure virtual function | |
1673 | and, for either of those two conditions, the class also has a inline | |
1674 | constructor or destructor and has a key function that is defined in | |
1675 | the current translation unit. | |
b55e3aad NC |
1676 | |
1677 | For Microsoft Windows based targets the use of the @code{dllimport} | |
1678 | attribute on functions is not necessary, but provides a small | |
8a36672b | 1679 | performance benefit by eliminating a thunk in the DLL@. The use of the |
b55e3aad | 1680 | @code{dllimport} attribute on imported variables was required on older |
b2ca3702 | 1681 | versions of the GNU linker, but can now be avoided by passing the |
8a36672b | 1682 | @option{--enable-auto-import} switch to the GNU linker. As with |
b2ca3702 | 1683 | functions, using the attribute for a variable eliminates a thunk in |
8a36672b | 1684 | the DLL@. |
b2ca3702 MM |
1685 | |
1686 | One drawback to using this attribute is that a pointer to a function | |
1687 | or variable marked as @code{dllimport} cannot be used as a constant | |
1688 | address. On Microsoft Windows targets, the attribute can be disabled | |
1689 | for functions by setting the @option{-mnop-fun-dllimport} flag. | |
c8619b90 NS |
1690 | |
1691 | @item eightbit_data | |
1692 | @cindex eight bit data on the H8/300, H8/300H, and H8S | |
1693 | Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified | |
1694 | variable should be placed into the eight bit data section. | |
1695 | The compiler will generate more efficient code for certain operations | |
1696 | on data in the eight bit data area. Note the eight bit data area is limited to | |
1697 | 256 bytes of data. | |
1698 | ||
1699 | You must use GAS and GLD from GNU binutils version 2.7 or later for | |
1700 | this attribute to work correctly. | |
1701 | ||
1702 | @item far | |
1703 | @cindex functions which handle memory bank switching | |
1704 | On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to | |
1705 | use a calling convention that takes care of switching memory banks when | |
1706 | entering and leaving a function. This calling convention is also the | |
1707 | default when using the @option{-mlong-calls} option. | |
1708 | ||
1709 | On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions | |
1710 | to call and return from a function. | |
1711 | ||
1712 | On 68HC11 the compiler will generate a sequence of instructions | |
1713 | to invoke a board-specific routine to switch the memory bank and call the | |
8a36672b | 1714 | real function. The board-specific routine simulates a @code{call}. |
c8619b90 | 1715 | At the end of a function, it will jump to a board-specific routine |
8a36672b | 1716 | instead of using @code{rts}. The board-specific return routine simulates |
c8619b90 NS |
1717 | the @code{rtc}. |
1718 | ||
1719 | @item fastcall | |
1720 | @cindex functions that pop the argument stack on the 386 | |
1721 | On the Intel 386, the @code{fastcall} attribute causes the compiler to | |
8a36672b JM |
1722 | pass the first two arguments in the registers ECX and EDX@. Subsequent |
1723 | arguments are passed on the stack. The called function will pop the | |
1724 | arguments off the stack. If the number of arguments is variable all | |
c8619b90 | 1725 | arguments are pushed on the stack. |
c1f7febf RK |
1726 | |
1727 | @item format (@var{archetype}, @var{string-index}, @var{first-to-check}) | |
1728 | @cindex @code{format} function attribute | |
84330467 | 1729 | @opindex Wformat |
bb72a084 | 1730 | The @code{format} attribute specifies that a function takes @code{printf}, |
26f6672d JM |
1731 | @code{scanf}, @code{strftime} or @code{strfmon} style arguments which |
1732 | should be type-checked against a format string. For example, the | |
1733 | declaration: | |
c1f7febf RK |
1734 | |
1735 | @smallexample | |
1736 | extern int | |
1737 | my_printf (void *my_object, const char *my_format, ...) | |
1738 | __attribute__ ((format (printf, 2, 3))); | |
1739 | @end smallexample | |
1740 | ||
1741 | @noindent | |
1742 | causes the compiler to check the arguments in calls to @code{my_printf} | |
1743 | for consistency with the @code{printf} style format string argument | |
1744 | @code{my_format}. | |
1745 | ||
1746 | The parameter @var{archetype} determines how the format string is | |
26f6672d JM |
1747 | interpreted, and should be @code{printf}, @code{scanf}, @code{strftime} |
1748 | or @code{strfmon}. (You can also use @code{__printf__}, | |
1749 | @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) The | |
c1f7febf RK |
1750 | parameter @var{string-index} specifies which argument is the format |
1751 | string argument (starting from 1), while @var{first-to-check} is the | |
1752 | number of the first argument to check against the format string. For | |
1753 | functions where the arguments are not available to be checked (such as | |
1754 | @code{vprintf}), specify the third parameter as zero. In this case the | |
b722c82c JM |
1755 | compiler only checks the format string for consistency. For |
1756 | @code{strftime} formats, the third parameter is required to be zero. | |
f57a2e3a BE |
1757 | Since non-static C++ methods have an implicit @code{this} argument, the |
1758 | arguments of such methods should be counted from two, not one, when | |
1759 | giving values for @var{string-index} and @var{first-to-check}. | |
c1f7febf RK |
1760 | |
1761 | In the example above, the format string (@code{my_format}) is the second | |
1762 | argument of the function @code{my_print}, and the arguments to check | |
1763 | start with the third argument, so the correct parameters for the format | |
1764 | attribute are 2 and 3. | |
1765 | ||
84330467 | 1766 | @opindex ffreestanding |
e6e931b7 | 1767 | @opindex fno-builtin |
c1f7febf | 1768 | The @code{format} attribute allows you to identify your own functions |
f0523f02 | 1769 | which take format strings as arguments, so that GCC can check the |
b722c82c | 1770 | calls to these functions for errors. The compiler always (unless |
e6e931b7 | 1771 | @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats |
b722c82c | 1772 | for the standard library functions @code{printf}, @code{fprintf}, |
bb72a084 | 1773 | @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime}, |
c1f7febf | 1774 | @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such |
84330467 | 1775 | warnings are requested (using @option{-Wformat}), so there is no need to |
b722c82c JM |
1776 | modify the header file @file{stdio.h}. In C99 mode, the functions |
1777 | @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and | |
26f6672d | 1778 | @code{vsscanf} are also checked. Except in strictly conforming C |
b4c984fb KG |
1779 | standard modes, the X/Open function @code{strfmon} is also checked as |
1780 | are @code{printf_unlocked} and @code{fprintf_unlocked}. | |
b722c82c | 1781 | @xref{C Dialect Options,,Options Controlling C Dialect}. |
c1f7febf | 1782 | |
a2bec818 DJ |
1783 | The target may provide additional types of format checks. |
1784 | @xref{Target Format Checks,,Format Checks Specific to Particular | |
1785 | Target Machines}. | |
1786 | ||
c1f7febf RK |
1787 | @item format_arg (@var{string-index}) |
1788 | @cindex @code{format_arg} function attribute | |
84330467 | 1789 | @opindex Wformat-nonliteral |
26f6672d JM |
1790 | The @code{format_arg} attribute specifies that a function takes a format |
1791 | string for a @code{printf}, @code{scanf}, @code{strftime} or | |
1792 | @code{strfmon} style function and modifies it (for example, to translate | |
1793 | it into another language), so the result can be passed to a | |
1794 | @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style | |
1795 | function (with the remaining arguments to the format function the same | |
1796 | as they would have been for the unmodified string). For example, the | |
1797 | declaration: | |
c1f7febf RK |
1798 | |
1799 | @smallexample | |
1800 | extern char * | |
1801 | my_dgettext (char *my_domain, const char *my_format) | |
1802 | __attribute__ ((format_arg (2))); | |
1803 | @end smallexample | |
1804 | ||
1805 | @noindent | |
26f6672d JM |
1806 | causes the compiler to check the arguments in calls to a @code{printf}, |
1807 | @code{scanf}, @code{strftime} or @code{strfmon} type function, whose | |
1808 | format string argument is a call to the @code{my_dgettext} function, for | |
1809 | consistency with the format string argument @code{my_format}. If the | |
1810 | @code{format_arg} attribute had not been specified, all the compiler | |
1811 | could tell in such calls to format functions would be that the format | |
1812 | string argument is not constant; this would generate a warning when | |
84330467 | 1813 | @option{-Wformat-nonliteral} is used, but the calls could not be checked |
26f6672d | 1814 | without the attribute. |
c1f7febf RK |
1815 | |
1816 | The parameter @var{string-index} specifies which argument is the format | |
f57a2e3a BE |
1817 | string argument (starting from one). Since non-static C++ methods have |
1818 | an implicit @code{this} argument, the arguments of such methods should | |
1819 | be counted from two. | |
c1f7febf RK |
1820 | |
1821 | The @code{format-arg} attribute allows you to identify your own | |
f0523f02 | 1822 | functions which modify format strings, so that GCC can check the |
26f6672d JM |
1823 | calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} |
1824 | type function whose operands are a call to one of your own function. | |
1825 | The compiler always treats @code{gettext}, @code{dgettext}, and | |
1826 | @code{dcgettext} in this manner except when strict ISO C support is | |
84330467 | 1827 | requested by @option{-ansi} or an appropriate @option{-std} option, or |
e6e931b7 JM |
1828 | @option{-ffreestanding} or @option{-fno-builtin} |
1829 | is used. @xref{C Dialect Options,,Options | |
26f6672d | 1830 | Controlling C Dialect}. |
c1f7febf | 1831 | |
c8619b90 NS |
1832 | @item function_vector |
1833 | @cindex calling functions through the function vector on the H8/300 processors | |
1834 | Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified | |
1835 | function should be called through the function vector. Calling a | |
1836 | function through the function vector will reduce code size, however; | |
1837 | the function vector has a limited size (maximum 128 entries on the H8/300 | |
1838 | and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector. | |
b34c7881 | 1839 | |
c8619b90 NS |
1840 | You must use GAS and GLD from GNU binutils version 2.7 or later for |
1841 | this attribute to work correctly. | |
b34c7881 | 1842 | |
c8619b90 NS |
1843 | @item interrupt |
1844 | @cindex interrupt handler functions | |
1845 | Use this attribute on the ARM, AVR, C4x, M32R/D and Xstormy16 ports to indicate | |
1846 | that the specified function is an interrupt handler. The compiler will | |
1847 | generate function entry and exit sequences suitable for use in an | |
1848 | interrupt handler when this attribute is present. | |
b34c7881 | 1849 | |
c8619b90 NS |
1850 | Note, interrupt handlers for the m68k, H8/300, H8/300H, H8S, and SH processors |
1851 | can be specified via the @code{interrupt_handler} attribute. | |
b34c7881 | 1852 | |
c8619b90 | 1853 | Note, on the AVR, interrupts will be enabled inside the function. |
9162542e | 1854 | |
c8619b90 NS |
1855 | Note, for the ARM, you can specify the kind of interrupt to be handled by |
1856 | adding an optional parameter to the interrupt attribute like this: | |
e23bd218 IR |
1857 | |
1858 | @smallexample | |
c8619b90 | 1859 | void f () __attribute__ ((interrupt ("IRQ"))); |
e23bd218 IR |
1860 | @end smallexample |
1861 | ||
c8619b90 | 1862 | Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@. |
e23bd218 | 1863 | |
c8619b90 NS |
1864 | @item interrupt_handler |
1865 | @cindex interrupt handler functions on the m68k, H8/300 and SH processors | |
1866 | Use this attribute on the m68k, H8/300, H8/300H, H8S, and SH to indicate that | |
1867 | the specified function is an interrupt handler. The compiler will generate | |
1868 | function entry and exit sequences suitable for use in an interrupt | |
1869 | handler when this attribute is present. | |
72954a4f | 1870 | |
c8619b90 NS |
1871 | @item long_call/short_call |
1872 | @cindex indirect calls on ARM | |
1873 | This attribute specifies how a particular function is called on | |
1874 | ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options}) | |
1875 | command line switch and @code{#pragma long_calls} settings. The | |
1876 | @code{long_call} attribute causes the compiler to always call the | |
1877 | function by first loading its address into a register and then using the | |
1878 | contents of that register. The @code{short_call} attribute always places | |
1879 | the offset to the function from the call site into the @samp{BL} | |
1880 | instruction directly. | |
72954a4f | 1881 | |
c8619b90 NS |
1882 | @item longcall/shortcall |
1883 | @cindex functions called via pointer on the RS/6000 and PowerPC | |
1884 | On the RS/6000 and PowerPC, the @code{longcall} attribute causes the | |
1885 | compiler to always call this function via a pointer, just as it would if | |
1886 | the @option{-mlongcall} option had been specified. The @code{shortcall} | |
1887 | attribute causes the compiler not to do this. These attributes override | |
1888 | both the @option{-mlongcall} switch and the @code{#pragma longcall} | |
1889 | setting. | |
72954a4f | 1890 | |
c8619b90 NS |
1891 | @xref{RS/6000 and PowerPC Options}, for more information on whether long |
1892 | calls are necessary. | |
c1f7febf | 1893 | |
140592a0 AG |
1894 | @item malloc |
1895 | @cindex @code{malloc} attribute | |
1896 | The @code{malloc} attribute is used to tell the compiler that a function | |
928a5ba9 JM |
1897 | may be treated as if any non-@code{NULL} pointer it returns cannot |
1898 | alias any other pointer valid when the function returns. | |
140592a0 | 1899 | This will often improve optimization. |
928a5ba9 JM |
1900 | Standard functions with this property include @code{malloc} and |
1901 | @code{calloc}. @code{realloc}-like functions have this property as | |
1902 | long as the old pointer is never referred to (including comparing it | |
1903 | to the new pointer) after the function returns a non-@code{NULL} | |
1904 | value. | |
140592a0 | 1905 | |
c8619b90 NS |
1906 | @item model (@var{model-name}) |
1907 | @cindex function addressability on the M32R/D | |
1908 | @cindex variable addressability on the IA-64 | |
1909 | ||
1910 | On the M32R/D, use this attribute to set the addressability of an | |
1911 | object, and of the code generated for a function. The identifier | |
1912 | @var{model-name} is one of @code{small}, @code{medium}, or | |
1913 | @code{large}, representing each of the code models. | |
1914 | ||
1915 | Small model objects live in the lower 16MB of memory (so that their | |
1916 | addresses can be loaded with the @code{ld24} instruction), and are | |
1917 | callable with the @code{bl} instruction. | |
1918 | ||
1919 | Medium model objects may live anywhere in the 32-bit address space (the | |
1920 | compiler will generate @code{seth/add3} instructions to load their addresses), | |
1921 | and are callable with the @code{bl} instruction. | |
1922 | ||
1923 | Large model objects may live anywhere in the 32-bit address space (the | |
1924 | compiler will generate @code{seth/add3} instructions to load their addresses), | |
1925 | and may not be reachable with the @code{bl} instruction (the compiler will | |
1926 | generate the much slower @code{seth/add3/jl} instruction sequence). | |
1927 | ||
1928 | On IA-64, use this attribute to set the addressability of an object. | |
1929 | At present, the only supported identifier for @var{model-name} is | |
1930 | @code{small}, indicating addressability via ``small'' (22-bit) | |
1931 | addresses (so that their addresses can be loaded with the @code{addl} | |
1932 | instruction). Caveat: such addressing is by definition not position | |
1933 | independent and hence this attribute must not be used for objects | |
1934 | defined by shared libraries. | |
1935 | ||
1936 | @item naked | |
1937 | @cindex function without a prologue/epilogue code | |
1938 | Use this attribute on the ARM, AVR, C4x and IP2K ports to indicate that the | |
1939 | specified function does not need prologue/epilogue sequences generated by | |
1940 | the compiler. It is up to the programmer to provide these sequences. | |
1941 | ||
1942 | @item near | |
1943 | @cindex functions which do not handle memory bank switching on 68HC11/68HC12 | |
1944 | On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to | |
1945 | use the normal calling convention based on @code{jsr} and @code{rts}. | |
1946 | This attribute can be used to cancel the effect of the @option{-mlong-calls} | |
1947 | option. | |
1948 | ||
1949 | @item no_instrument_function | |
1950 | @cindex @code{no_instrument_function} function attribute | |
1951 | @opindex finstrument-functions | |
1952 | If @option{-finstrument-functions} is given, profiling function calls will | |
1953 | be generated at entry and exit of most user-compiled functions. | |
1954 | Functions with this attribute will not be so instrumented. | |
1955 | ||
1956 | @item noinline | |
1957 | @cindex @code{noinline} function attribute | |
1958 | This function attribute prevents a function from being considered for | |
1959 | inlining. | |
1960 | ||
1961 | @item nonnull (@var{arg-index}, @dots{}) | |
1962 | @cindex @code{nonnull} function attribute | |
1963 | The @code{nonnull} attribute specifies that some function parameters should | |
1964 | be non-null pointers. For instance, the declaration: | |
c1f7febf RK |
1965 | |
1966 | @smallexample | |
c8619b90 NS |
1967 | extern void * |
1968 | my_memcpy (void *dest, const void *src, size_t len) | |
1969 | __attribute__((nonnull (1, 2))); | |
c1f7febf RK |
1970 | @end smallexample |
1971 | ||
c8619b90 NS |
1972 | @noindent |
1973 | causes the compiler to check that, in calls to @code{my_memcpy}, | |
1974 | arguments @var{dest} and @var{src} are non-null. If the compiler | |
1975 | determines that a null pointer is passed in an argument slot marked | |
1976 | as non-null, and the @option{-Wnonnull} option is enabled, a warning | |
1977 | is issued. The compiler may also choose to make optimizations based | |
1978 | on the knowledge that certain function arguments will not be null. | |
af3e86c2 | 1979 | |
c8619b90 NS |
1980 | If no argument index list is given to the @code{nonnull} attribute, |
1981 | all pointer arguments are marked as non-null. To illustrate, the | |
1982 | following declaration is equivalent to the previous example: | |
47bd70b5 JJ |
1983 | |
1984 | @smallexample | |
c8619b90 NS |
1985 | extern void * |
1986 | my_memcpy (void *dest, const void *src, size_t len) | |
1987 | __attribute__((nonnull)); | |
47bd70b5 JJ |
1988 | @end smallexample |
1989 | ||
c8619b90 NS |
1990 | @item noreturn |
1991 | @cindex @code{noreturn} function attribute | |
1992 | A few standard library functions, such as @code{abort} and @code{exit}, | |
1993 | cannot return. GCC knows this automatically. Some programs define | |
1994 | their own functions that never return. You can declare them | |
1995 | @code{noreturn} to tell the compiler this fact. For example, | |
9e8aab55 | 1996 | |
c8619b90 NS |
1997 | @smallexample |
1998 | @group | |
1999 | void fatal () __attribute__ ((noreturn)); | |
d5c4db17 | 2000 | |
c8619b90 NS |
2001 | void |
2002 | fatal (/* @r{@dots{}} */) | |
2003 | @{ | |
2004 | /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */ | |
2005 | exit (1); | |
2006 | @} | |
2007 | @end group | |
2008 | @end smallexample | |
9e8aab55 | 2009 | |
c8619b90 NS |
2010 | The @code{noreturn} keyword tells the compiler to assume that |
2011 | @code{fatal} cannot return. It can then optimize without regard to what | |
2012 | would happen if @code{fatal} ever did return. This makes slightly | |
2013 | better code. More importantly, it helps avoid spurious warnings of | |
2014 | uninitialized variables. | |
9e8aab55 | 2015 | |
c8619b90 NS |
2016 | The @code{noreturn} keyword does not affect the exceptional path when that |
2017 | applies: a @code{noreturn}-marked function may still return to the caller | |
2e9522f4 | 2018 | by throwing an exception or calling @code{longjmp}. |
9e8aab55 | 2019 | |
c8619b90 NS |
2020 | Do not assume that registers saved by the calling function are |
2021 | restored before calling the @code{noreturn} function. | |
47bd70b5 | 2022 | |
c8619b90 NS |
2023 | It does not make sense for a @code{noreturn} function to have a return |
2024 | type other than @code{void}. | |
c1f7febf | 2025 | |
c8619b90 NS |
2026 | The attribute @code{noreturn} is not implemented in GCC versions |
2027 | earlier than 2.5. An alternative way to declare that a function does | |
2028 | not return, which works in the current version and in some older | |
2029 | versions, is as follows: | |
5d34c8e9 | 2030 | |
c8619b90 NS |
2031 | @smallexample |
2032 | typedef void voidfn (); | |
c1f7febf | 2033 | |
c8619b90 NS |
2034 | volatile voidfn fatal; |
2035 | @end smallexample | |
e91f04de | 2036 | |
c8619b90 NS |
2037 | @item nothrow |
2038 | @cindex @code{nothrow} function attribute | |
2039 | The @code{nothrow} attribute is used to inform the compiler that a | |
2040 | function cannot throw an exception. For example, most functions in | |
2041 | the standard C library can be guaranteed not to throw an exception | |
2042 | with the notable exceptions of @code{qsort} and @code{bsearch} that | |
2043 | take function pointer arguments. The @code{nothrow} attribute is not | |
3f3174b6 | 2044 | implemented in GCC versions earlier than 3.3. |
c1f7febf | 2045 | |
c8619b90 NS |
2046 | @item pure |
2047 | @cindex @code{pure} function attribute | |
2048 | Many functions have no effects except the return value and their | |
2049 | return value depends only on the parameters and/or global variables. | |
2050 | Such a function can be subject | |
2051 | to common subexpression elimination and loop optimization just as an | |
2052 | arithmetic operator would be. These functions should be declared | |
2053 | with the attribute @code{pure}. For example, | |
a5c76ee6 | 2054 | |
c8619b90 NS |
2055 | @smallexample |
2056 | int square (int) __attribute__ ((pure)); | |
2057 | @end smallexample | |
c1f7febf | 2058 | |
c8619b90 NS |
2059 | @noindent |
2060 | says that the hypothetical function @code{square} is safe to call | |
2061 | fewer times than the program says. | |
c27ba912 | 2062 | |
c8619b90 NS |
2063 | Some of common examples of pure functions are @code{strlen} or @code{memcmp}. |
2064 | Interesting non-pure functions are functions with infinite loops or those | |
2065 | depending on volatile memory or other system resource, that may change between | |
2066 | two consecutive calls (such as @code{feof} in a multithreading environment). | |
c1f7febf | 2067 | |
c8619b90 NS |
2068 | The attribute @code{pure} is not implemented in GCC versions earlier |
2069 | than 2.96. | |
c1f7febf | 2070 | |
c8619b90 NS |
2071 | @item regparm (@var{number}) |
2072 | @cindex @code{regparm} attribute | |
2073 | @cindex functions that are passed arguments in registers on the 386 | |
2074 | On the Intel 386, the @code{regparm} attribute causes the compiler to | |
2075 | pass up to @var{number} integer arguments in registers EAX, | |
2076 | EDX, and ECX instead of on the stack. Functions that take a | |
2077 | variable number of arguments will continue to be passed all of their | |
2078 | arguments on the stack. | |
6d3d9133 | 2079 | |
c8619b90 NS |
2080 | Beware that on some ELF systems this attribute is unsuitable for |
2081 | global functions in shared libraries with lazy binding (which is the | |
2082 | default). Lazy binding will send the first call via resolving code in | |
2083 | the loader, which might assume EAX, EDX and ECX can be clobbered, as | |
2084 | per the standard calling conventions. Solaris 8 is affected by this. | |
2085 | GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be | |
2086 | safe since the loaders there save all registers. (Lazy binding can be | |
2087 | disabled with the linker or the loader if desired, to avoid the | |
2088 | problem.) | |
6d3d9133 | 2089 | |
c8619b90 NS |
2090 | @item saveall |
2091 | @cindex save all registers on the H8/300, H8/300H, and H8S | |
2092 | Use this attribute on the H8/300, H8/300H, and H8S to indicate that | |
2093 | all registers except the stack pointer should be saved in the prologue | |
2094 | regardless of whether they are used or not. | |
6d3d9133 | 2095 | |
c8619b90 NS |
2096 | @item section ("@var{section-name}") |
2097 | @cindex @code{section} function attribute | |
2098 | Normally, the compiler places the code it generates in the @code{text} section. | |
2099 | Sometimes, however, you need additional sections, or you need certain | |
2100 | particular functions to appear in special sections. The @code{section} | |
2101 | attribute specifies that a function lives in a particular section. | |
2102 | For example, the declaration: | |
6d3d9133 NC |
2103 | |
2104 | @smallexample | |
c8619b90 | 2105 | extern void foobar (void) __attribute__ ((section ("bar"))); |
6d3d9133 NC |
2106 | @end smallexample |
2107 | ||
c8619b90 NS |
2108 | @noindent |
2109 | puts the function @code{foobar} in the @code{bar} section. | |
6d3d9133 | 2110 | |
c8619b90 NS |
2111 | Some file formats do not support arbitrary sections so the @code{section} |
2112 | attribute is not available on all platforms. | |
2113 | If you need to map the entire contents of a module to a particular | |
2114 | section, consider using the facilities of the linker instead. | |
2115 | ||
3d091dac KG |
2116 | @item sentinel |
2117 | @cindex @code{sentinel} function attribute | |
254986c7 KG |
2118 | This function attribute ensures that a parameter in a function call is |
2119 | an explicit @code{NULL}. The attribute is only valid on variadic | |
2120 | functions. By default, the sentinel is located at position zero, the | |
2121 | last parameter of the function call. If an optional integer position | |
2122 | argument P is supplied to the attribute, the sentinel must be located at | |
2123 | position P counting backwards from the end of the argument list. | |
2124 | ||
2125 | @smallexample | |
2126 | __attribute__ ((sentinel)) | |
2127 | is equivalent to | |
2128 | __attribute__ ((sentinel(0))) | |
2129 | @end smallexample | |
2130 | ||
2131 | The attribute is automatically set with a position of 0 for the built-in | |
2132 | functions @code{execl} and @code{execlp}. The built-in function | |
254ea84c | 2133 | @code{execle} has the attribute set with a position of 1. |
254986c7 KG |
2134 | |
2135 | A valid @code{NULL} in this context is defined as zero with any pointer | |
2136 | type. If your system defines the @code{NULL} macro with an integer type | |
2137 | then you need to add an explicit cast. GCC replaces @code{stddef.h} | |
2138 | with a copy that redefines NULL appropriately. | |
2139 | ||
2140 | The warnings for missing or incorrect sentinels are enabled with | |
2141 | @option{-Wformat}. | |
3d091dac | 2142 | |
c8619b90 NS |
2143 | @item short_call |
2144 | See long_call/short_call. | |
2145 | ||
2146 | @item shortcall | |
2147 | See longcall/shortcall. | |
2148 | ||
2149 | @item signal | |
2150 | @cindex signal handler functions on the AVR processors | |
2151 | Use this attribute on the AVR to indicate that the specified | |
2152 | function is a signal handler. The compiler will generate function | |
2153 | entry and exit sequences suitable for use in a signal handler when this | |
2154 | attribute is present. Interrupts will be disabled inside the function. | |
b93e3893 AO |
2155 | |
2156 | @item sp_switch | |
88ab0d1c | 2157 | Use this attribute on the SH to indicate an @code{interrupt_handler} |
b93e3893 AO |
2158 | function should switch to an alternate stack. It expects a string |
2159 | argument that names a global variable holding the address of the | |
2160 | alternate stack. | |
2161 | ||
2162 | @smallexample | |
2163 | void *alt_stack; | |
aee96fe9 JM |
2164 | void f () __attribute__ ((interrupt_handler, |
2165 | sp_switch ("alt_stack"))); | |
b93e3893 AO |
2166 | @end smallexample |
2167 | ||
c8619b90 NS |
2168 | @item stdcall |
2169 | @cindex functions that pop the argument stack on the 386 | |
2170 | On the Intel 386, the @code{stdcall} attribute causes the compiler to | |
2171 | assume that the called function will pop off the stack space used to | |
2172 | pass arguments, unless it takes a variable number of arguments. | |
c1f7febf RK |
2173 | |
2174 | @item tiny_data | |
dbacaa98 KH |
2175 | @cindex tiny data section on the H8/300H and H8S |
2176 | Use this attribute on the H8/300H and H8S to indicate that the specified | |
c1f7febf RK |
2177 | variable should be placed into the tiny data section. |
2178 | The compiler will generate more efficient code for loads and stores | |
2179 | on data in the tiny data section. Note the tiny data area is limited to | |
2180 | slightly under 32kbytes of data. | |
845da534 | 2181 | |
c8619b90 NS |
2182 | @item trap_exit |
2183 | Use this attribute on the SH for an @code{interrupt_handler} to return using | |
2184 | @code{trapa} instead of @code{rte}. This attribute expects an integer | |
2185 | argument specifying the trap number to be used. | |
845da534 | 2186 | |
c8619b90 NS |
2187 | @item unused |
2188 | @cindex @code{unused} attribute. | |
2189 | This attribute, attached to a function, means that the function is meant | |
2190 | to be possibly unused. GCC will not produce a warning for this | |
2191 | function. | |
a32767e4 | 2192 | |
c8619b90 NS |
2193 | @item used |
2194 | @cindex @code{used} attribute. | |
2195 | This attribute, attached to a function, means that code must be emitted | |
2196 | for the function even if it appears that the function is not referenced. | |
2197 | This is useful, for example, when the function is referenced only in | |
2198 | inline assembly. | |
5936c7e7 | 2199 | |
c8619b90 NS |
2200 | @item visibility ("@var{visibility_type}") |
2201 | @cindex @code{visibility} attribute | |
2202 | The @code{visibility} attribute on ELF targets causes the declaration | |
2203 | to be emitted with default, hidden, protected or internal visibility. | |
5936c7e7 | 2204 | |
c8619b90 NS |
2205 | @smallexample |
2206 | void __attribute__ ((visibility ("protected"))) | |
2207 | f () @{ /* @r{Do something.} */; @} | |
2208 | int i __attribute__ ((visibility ("hidden"))); | |
2209 | @end smallexample | |
5936c7e7 | 2210 | |
c8619b90 | 2211 | See the ELF gABI for complete details, but the short story is: |
5936c7e7 | 2212 | |
c8619b90 | 2213 | @table @dfn |
63c5b495 | 2214 | @c keep this list of visibilities in alphabetical order. |
6b6cb52e | 2215 | |
c8619b90 | 2216 | @item default |
8a36672b | 2217 | Default visibility is the normal case for ELF@. This value is |
c8619b90 NS |
2218 | available for the visibility attribute to override other options |
2219 | that may change the assumed visibility of symbols. | |
6b6cb52e | 2220 | |
c8619b90 NS |
2221 | @item hidden |
2222 | Hidden visibility indicates that the symbol will not be placed into | |
2223 | the dynamic symbol table, so no other @dfn{module} (executable or | |
2224 | shared library) can reference it directly. | |
6b6cb52e | 2225 | |
c8619b90 NS |
2226 | @item internal |
2227 | Internal visibility is like hidden visibility, but with additional | |
2228 | processor specific semantics. Unless otherwise specified by the psABI, | |
2229 | GCC defines internal visibility to mean that the function is @emph{never} | |
2230 | called from another module. Note that hidden symbols, while they cannot | |
2231 | be referenced directly by other modules, can be referenced indirectly via | |
2232 | function pointers. By indicating that a symbol cannot be called from | |
2233 | outside the module, GCC may for instance omit the load of a PIC register | |
2234 | since it is known that the calling function loaded the correct value. | |
6b6cb52e | 2235 | |
c8619b90 NS |
2236 | @item protected |
2237 | Protected visibility indicates that the symbol will be placed in the | |
2238 | dynamic symbol table, but that references within the defining module | |
2239 | will bind to the local symbol. That is, the symbol cannot be overridden | |
2240 | by another module. | |
6b6cb52e | 2241 | |
c8619b90 | 2242 | @end table |
6b6cb52e | 2243 | |
c8619b90 | 2244 | Not all ELF targets support this attribute. |
6b6cb52e | 2245 | |
c8619b90 NS |
2246 | @item warn_unused_result |
2247 | @cindex @code{warn_unused_result} attribute | |
2248 | The @code{warn_unused_result} attribute causes a warning to be emitted | |
2249 | if a caller of the function with this attribute does not use its | |
2250 | return value. This is useful for functions where not checking | |
2251 | the result is either a security problem or always a bug, such as | |
2252 | @code{realloc}. | |
6b6cb52e | 2253 | |
c8619b90 NS |
2254 | @smallexample |
2255 | int fn () __attribute__ ((warn_unused_result)); | |
2256 | int foo () | |
2257 | @{ | |
2258 | if (fn () < 0) return -1; | |
2259 | fn (); | |
2260 | return 0; | |
2261 | @} | |
2262 | @end smallexample | |
6b6cb52e | 2263 | |
c8619b90 | 2264 | results in warning on line 5. |
6b6cb52e | 2265 | |
c8619b90 NS |
2266 | @item weak |
2267 | @cindex @code{weak} attribute | |
2268 | The @code{weak} attribute causes the declaration to be emitted as a weak | |
2269 | symbol rather than a global. This is primarily useful in defining | |
2270 | library functions which can be overridden in user code, though it can | |
2271 | also be used with non-function declarations. Weak symbols are supported | |
2272 | for ELF targets, and also for a.out targets when using the GNU assembler | |
2273 | and linker. | |
6b6cb52e | 2274 | |
c1f7febf RK |
2275 | @end table |
2276 | ||
2277 | You can specify multiple attributes in a declaration by separating them | |
2278 | by commas within the double parentheses or by immediately following an | |
2279 | attribute declaration with another attribute declaration. | |
2280 | ||
2281 | @cindex @code{#pragma}, reason for not using | |
2282 | @cindex pragma, reason for not using | |
9f1bbeaa JM |
2283 | Some people object to the @code{__attribute__} feature, suggesting that |
2284 | ISO C's @code{#pragma} should be used instead. At the time | |
2285 | @code{__attribute__} was designed, there were two reasons for not doing | |
2286 | this. | |
c1f7febf RK |
2287 | |
2288 | @enumerate | |
2289 | @item | |
2290 | It is impossible to generate @code{#pragma} commands from a macro. | |
2291 | ||
2292 | @item | |
2293 | There is no telling what the same @code{#pragma} might mean in another | |
2294 | compiler. | |
2295 | @end enumerate | |
2296 | ||
9f1bbeaa JM |
2297 | These two reasons applied to almost any application that might have been |
2298 | proposed for @code{#pragma}. It was basically a mistake to use | |
2299 | @code{#pragma} for @emph{anything}. | |
2300 | ||
2301 | The ISO C99 standard includes @code{_Pragma}, which now allows pragmas | |
2302 | to be generated from macros. In addition, a @code{#pragma GCC} | |
2303 | namespace is now in use for GCC-specific pragmas. However, it has been | |
2304 | found convenient to use @code{__attribute__} to achieve a natural | |
2305 | attachment of attributes to their corresponding declarations, whereas | |
2306 | @code{#pragma GCC} is of use for constructs that do not naturally form | |
2307 | part of the grammar. @xref{Other Directives,,Miscellaneous | |
48795525 | 2308 | Preprocessing Directives, cpp, The GNU C Preprocessor}. |
c1f7febf | 2309 | |
2c5e91d2 JM |
2310 | @node Attribute Syntax |
2311 | @section Attribute Syntax | |
2312 | @cindex attribute syntax | |
2313 | ||
2314 | This section describes the syntax with which @code{__attribute__} may be | |
2315 | used, and the constructs to which attribute specifiers bind, for the C | |
161d7b59 | 2316 | language. Some details may vary for C++ and Objective-C@. Because of |
2c5e91d2 JM |
2317 | infelicities in the grammar for attributes, some forms described here |
2318 | may not be successfully parsed in all cases. | |
2319 | ||
91d231cb JM |
2320 | There are some problems with the semantics of attributes in C++. For |
2321 | example, there are no manglings for attributes, although they may affect | |
2322 | code generation, so problems may arise when attributed types are used in | |
2323 | conjunction with templates or overloading. Similarly, @code{typeid} | |
2324 | does not distinguish between types with different attributes. Support | |
2325 | for attributes in C++ may be restricted in future to attributes on | |
2326 | declarations only, but not on nested declarators. | |
2327 | ||
2c5e91d2 JM |
2328 | @xref{Function Attributes}, for details of the semantics of attributes |
2329 | applying to functions. @xref{Variable Attributes}, for details of the | |
2330 | semantics of attributes applying to variables. @xref{Type Attributes}, | |
2331 | for details of the semantics of attributes applying to structure, union | |
2332 | and enumerated types. | |
2333 | ||
2334 | An @dfn{attribute specifier} is of the form | |
2335 | @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list} | |
2336 | is a possibly empty comma-separated sequence of @dfn{attributes}, where | |
2337 | each attribute is one of the following: | |
2338 | ||
2339 | @itemize @bullet | |
2340 | @item | |
2341 | Empty. Empty attributes are ignored. | |
2342 | ||
2343 | @item | |
2344 | A word (which may be an identifier such as @code{unused}, or a reserved | |
2345 | word such as @code{const}). | |
2346 | ||
2347 | @item | |
2348 | A word, followed by, in parentheses, parameters for the attribute. | |
2349 | These parameters take one of the following forms: | |
2350 | ||
2351 | @itemize @bullet | |
2352 | @item | |
2353 | An identifier. For example, @code{mode} attributes use this form. | |
2354 | ||
2355 | @item | |
2356 | An identifier followed by a comma and a non-empty comma-separated list | |
2357 | of expressions. For example, @code{format} attributes use this form. | |
2358 | ||
2359 | @item | |
2360 | A possibly empty comma-separated list of expressions. For example, | |
2361 | @code{format_arg} attributes use this form with the list being a single | |
2362 | integer constant expression, and @code{alias} attributes use this form | |
2363 | with the list being a single string constant. | |
2364 | @end itemize | |
2365 | @end itemize | |
2366 | ||
2367 | An @dfn{attribute specifier list} is a sequence of one or more attribute | |
2368 | specifiers, not separated by any other tokens. | |
2369 | ||
50fc59e7 | 2370 | In GNU C, an attribute specifier list may appear after the colon following a |
2c5e91d2 JM |
2371 | label, other than a @code{case} or @code{default} label. The only |
2372 | attribute it makes sense to use after a label is @code{unused}. This | |
2373 | feature is intended for code generated by programs which contains labels | |
2374 | that may be unused but which is compiled with @option{-Wall}. It would | |
2375 | not normally be appropriate to use in it human-written code, though it | |
2376 | could be useful in cases where the code that jumps to the label is | |
8a36672b | 2377 | contained within an @code{#ifdef} conditional. GNU C++ does not permit |
50fc59e7 NS |
2378 | such placement of attribute lists, as it is permissible for a |
2379 | declaration, which could begin with an attribute list, to be labelled in | |
8a36672b | 2380 | C++. Declarations cannot be labelled in C90 or C99, so the ambiguity |
50fc59e7 | 2381 | does not arise there. |
2c5e91d2 JM |
2382 | |
2383 | An attribute specifier list may appear as part of a @code{struct}, | |
2384 | @code{union} or @code{enum} specifier. It may go either immediately | |
2385 | after the @code{struct}, @code{union} or @code{enum} keyword, or after | |
2386 | the closing brace. It is ignored if the content of the structure, union | |
2387 | or enumerated type is not defined in the specifier in which the | |
2388 | attribute specifier list is used---that is, in usages such as | |
2389 | @code{struct __attribute__((foo)) bar} with no following opening brace. | |
2390 | Where attribute specifiers follow the closing brace, they are considered | |
2391 | to relate to the structure, union or enumerated type defined, not to any | |
2392 | enclosing declaration the type specifier appears in, and the type | |
2393 | defined is not complete until after the attribute specifiers. | |
2394 | @c Otherwise, there would be the following problems: a shift/reduce | |
4fe9b91c | 2395 | @c conflict between attributes binding the struct/union/enum and |
2c5e91d2 JM |
2396 | @c binding to the list of specifiers/qualifiers; and "aligned" |
2397 | @c attributes could use sizeof for the structure, but the size could be | |
2398 | @c changed later by "packed" attributes. | |
2399 | ||
2400 | Otherwise, an attribute specifier appears as part of a declaration, | |
2401 | counting declarations of unnamed parameters and type names, and relates | |
2402 | to that declaration (which may be nested in another declaration, for | |
91d231cb JM |
2403 | example in the case of a parameter declaration), or to a particular declarator |
2404 | within a declaration. Where an | |
ff867905 JM |
2405 | attribute specifier is applied to a parameter declared as a function or |
2406 | an array, it should apply to the function or array rather than the | |
2407 | pointer to which the parameter is implicitly converted, but this is not | |
2408 | yet correctly implemented. | |
2c5e91d2 JM |
2409 | |
2410 | Any list of specifiers and qualifiers at the start of a declaration may | |
2411 | contain attribute specifiers, whether or not such a list may in that | |
2412 | context contain storage class specifiers. (Some attributes, however, | |
2413 | are essentially in the nature of storage class specifiers, and only make | |
2414 | sense where storage class specifiers may be used; for example, | |
2415 | @code{section}.) There is one necessary limitation to this syntax: the | |
2416 | first old-style parameter declaration in a function definition cannot | |
2417 | begin with an attribute specifier, because such an attribute applies to | |
2418 | the function instead by syntax described below (which, however, is not | |
2419 | yet implemented in this case). In some other cases, attribute | |
2420 | specifiers are permitted by this grammar but not yet supported by the | |
2421 | compiler. All attribute specifiers in this place relate to the | |
c771326b | 2422 | declaration as a whole. In the obsolescent usage where a type of |
2c5e91d2 JM |
2423 | @code{int} is implied by the absence of type specifiers, such a list of |
2424 | specifiers and qualifiers may be an attribute specifier list with no | |
2425 | other specifiers or qualifiers. | |
2426 | ||
7dcb0442 JM |
2427 | At present, the first parameter in a function prototype must have some |
2428 | type specifier which is not an attribute specifier; this resolves an | |
2429 | ambiguity in the interpretation of @code{void f(int | |
2430 | (__attribute__((foo)) x))}, but is subject to change. At present, if | |
2431 | the parentheses of a function declarator contain only attributes then | |
2432 | those attributes are ignored, rather than yielding an error or warning | |
2433 | or implying a single parameter of type int, but this is subject to | |
2434 | change. | |
2435 | ||
2c5e91d2 JM |
2436 | An attribute specifier list may appear immediately before a declarator |
2437 | (other than the first) in a comma-separated list of declarators in a | |
2438 | declaration of more than one identifier using a single list of | |
4b01f8d8 | 2439 | specifiers and qualifiers. Such attribute specifiers apply |
9c34dbbf ZW |
2440 | only to the identifier before whose declarator they appear. For |
2441 | example, in | |
2442 | ||
2443 | @smallexample | |
2444 | __attribute__((noreturn)) void d0 (void), | |
2445 | __attribute__((format(printf, 1, 2))) d1 (const char *, ...), | |
2446 | d2 (void) | |
2447 | @end smallexample | |
2448 | ||
2449 | @noindent | |
2450 | the @code{noreturn} attribute applies to all the functions | |
4b01f8d8 | 2451 | declared; the @code{format} attribute only applies to @code{d1}. |
2c5e91d2 JM |
2452 | |
2453 | An attribute specifier list may appear immediately before the comma, | |
2454 | @code{=} or semicolon terminating the declaration of an identifier other | |
2455 | than a function definition. At present, such attribute specifiers apply | |
2456 | to the declared object or function, but in future they may attach to the | |
2457 | outermost adjacent declarator. In simple cases there is no difference, | |
f282ffb3 | 2458 | but, for example, in |
9c34dbbf ZW |
2459 | |
2460 | @smallexample | |
2461 | void (****f)(void) __attribute__((noreturn)); | |
2462 | @end smallexample | |
2463 | ||
2464 | @noindent | |
2465 | at present the @code{noreturn} attribute applies to @code{f}, which | |
2466 | causes a warning since @code{f} is not a function, but in future it may | |
2467 | apply to the function @code{****f}. The precise semantics of what | |
2468 | attributes in such cases will apply to are not yet specified. Where an | |
2469 | assembler name for an object or function is specified (@pxref{Asm | |
2470 | Labels}), at present the attribute must follow the @code{asm} | |
2471 | specification; in future, attributes before the @code{asm} specification | |
2472 | may apply to the adjacent declarator, and those after it to the declared | |
2473 | object or function. | |
2c5e91d2 JM |
2474 | |
2475 | An attribute specifier list may, in future, be permitted to appear after | |
2476 | the declarator in a function definition (before any old-style parameter | |
2477 | declarations or the function body). | |
2478 | ||
0e03329a JM |
2479 | Attribute specifiers may be mixed with type qualifiers appearing inside |
2480 | the @code{[]} of a parameter array declarator, in the C99 construct by | |
2481 | which such qualifiers are applied to the pointer to which the array is | |
2482 | implicitly converted. Such attribute specifiers apply to the pointer, | |
2483 | not to the array, but at present this is not implemented and they are | |
2484 | ignored. | |
2485 | ||
2c5e91d2 JM |
2486 | An attribute specifier list may appear at the start of a nested |
2487 | declarator. At present, there are some limitations in this usage: the | |
91d231cb JM |
2488 | attributes correctly apply to the declarator, but for most individual |
2489 | attributes the semantics this implies are not implemented. | |
2490 | When attribute specifiers follow the @code{*} of a pointer | |
4b01f8d8 | 2491 | declarator, they may be mixed with any type qualifiers present. |
91d231cb | 2492 | The following describes the formal semantics of this syntax. It will make the |
2c5e91d2 JM |
2493 | most sense if you are familiar with the formal specification of |
2494 | declarators in the ISO C standard. | |
2495 | ||
2496 | Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T | |
2497 | D1}, where @code{T} contains declaration specifiers that specify a type | |
2498 | @var{Type} (such as @code{int}) and @code{D1} is a declarator that | |
2499 | contains an identifier @var{ident}. The type specified for @var{ident} | |
2500 | for derived declarators whose type does not include an attribute | |
2501 | specifier is as in the ISO C standard. | |
2502 | ||
2503 | If @code{D1} has the form @code{( @var{attribute-specifier-list} D )}, | |
2504 | and the declaration @code{T D} specifies the type | |
2505 | ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then | |
2506 | @code{T D1} specifies the type ``@var{derived-declarator-type-list} | |
2507 | @var{attribute-specifier-list} @var{Type}'' for @var{ident}. | |
2508 | ||
2509 | If @code{D1} has the form @code{* | |
2510 | @var{type-qualifier-and-attribute-specifier-list} D}, and the | |
2511 | declaration @code{T D} specifies the type | |
2512 | ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then | |
2513 | @code{T D1} specifies the type ``@var{derived-declarator-type-list} | |
2514 | @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for | |
2515 | @var{ident}. | |
2516 | ||
f282ffb3 | 2517 | For example, |
9c34dbbf ZW |
2518 | |
2519 | @smallexample | |
2520 | void (__attribute__((noreturn)) ****f) (void); | |
2521 | @end smallexample | |
2522 | ||
2523 | @noindent | |
2524 | specifies the type ``pointer to pointer to pointer to pointer to | |
2525 | non-returning function returning @code{void}''. As another example, | |
2526 | ||
2527 | @smallexample | |
2528 | char *__attribute__((aligned(8))) *f; | |
2529 | @end smallexample | |
2530 | ||
2531 | @noindent | |
2532 | specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''. | |
91d231cb JM |
2533 | Note again that this does not work with most attributes; for example, |
2534 | the usage of @samp{aligned} and @samp{noreturn} attributes given above | |
2535 | is not yet supported. | |
2536 | ||
2537 | For compatibility with existing code written for compiler versions that | |
2538 | did not implement attributes on nested declarators, some laxity is | |
2539 | allowed in the placing of attributes. If an attribute that only applies | |
2540 | to types is applied to a declaration, it will be treated as applying to | |
2541 | the type of that declaration. If an attribute that only applies to | |
2542 | declarations is applied to the type of a declaration, it will be treated | |
2543 | as applying to that declaration; and, for compatibility with code | |
2544 | placing the attributes immediately before the identifier declared, such | |
2545 | an attribute applied to a function return type will be treated as | |
2546 | applying to the function type, and such an attribute applied to an array | |
2547 | element type will be treated as applying to the array type. If an | |
2548 | attribute that only applies to function types is applied to a | |
2549 | pointer-to-function type, it will be treated as applying to the pointer | |
2550 | target type; if such an attribute is applied to a function return type | |
2551 | that is not a pointer-to-function type, it will be treated as applying | |
2552 | to the function type. | |
2c5e91d2 | 2553 | |
c1f7febf RK |
2554 | @node Function Prototypes |
2555 | @section Prototypes and Old-Style Function Definitions | |
2556 | @cindex function prototype declarations | |
2557 | @cindex old-style function definitions | |
2558 | @cindex promotion of formal parameters | |
2559 | ||
5490d604 | 2560 | GNU C extends ISO C to allow a function prototype to override a later |
c1f7febf RK |
2561 | old-style non-prototype definition. Consider the following example: |
2562 | ||
3ab51846 | 2563 | @smallexample |
c1f7febf | 2564 | /* @r{Use prototypes unless the compiler is old-fashioned.} */ |
d863830b | 2565 | #ifdef __STDC__ |
c1f7febf RK |
2566 | #define P(x) x |
2567 | #else | |
2568 | #define P(x) () | |
2569 | #endif | |
2570 | ||
2571 | /* @r{Prototype function declaration.} */ | |
2572 | int isroot P((uid_t)); | |
2573 | ||
2574 | /* @r{Old-style function definition.} */ | |
2575 | int | |
2576 | isroot (x) /* ??? lossage here ??? */ | |
2577 | uid_t x; | |
2578 | @{ | |
2579 | return x == 0; | |
2580 | @} | |
3ab51846 | 2581 | @end smallexample |
c1f7febf | 2582 | |
5490d604 | 2583 | Suppose the type @code{uid_t} happens to be @code{short}. ISO C does |
c1f7febf RK |
2584 | not allow this example, because subword arguments in old-style |
2585 | non-prototype definitions are promoted. Therefore in this example the | |
2586 | function definition's argument is really an @code{int}, which does not | |
2587 | match the prototype argument type of @code{short}. | |
2588 | ||
5490d604 | 2589 | This restriction of ISO C makes it hard to write code that is portable |
c1f7febf RK |
2590 | to traditional C compilers, because the programmer does not know |
2591 | whether the @code{uid_t} type is @code{short}, @code{int}, or | |
2592 | @code{long}. Therefore, in cases like these GNU C allows a prototype | |
2593 | to override a later old-style definition. More precisely, in GNU C, a | |
2594 | function prototype argument type overrides the argument type specified | |
2595 | by a later old-style definition if the former type is the same as the | |
2596 | latter type before promotion. Thus in GNU C the above example is | |
2597 | equivalent to the following: | |
2598 | ||
3ab51846 | 2599 | @smallexample |
c1f7febf RK |
2600 | int isroot (uid_t); |
2601 | ||
2602 | int | |
2603 | isroot (uid_t x) | |
2604 | @{ | |
2605 | return x == 0; | |
2606 | @} | |
3ab51846 | 2607 | @end smallexample |
c1f7febf | 2608 | |
9c34dbbf | 2609 | @noindent |
c1f7febf RK |
2610 | GNU C++ does not support old-style function definitions, so this |
2611 | extension is irrelevant. | |
2612 | ||
2613 | @node C++ Comments | |
2614 | @section C++ Style Comments | |
2615 | @cindex // | |
2616 | @cindex C++ comments | |
2617 | @cindex comments, C++ style | |
2618 | ||
2619 | In GNU C, you may use C++ style comments, which start with @samp{//} and | |
2620 | continue until the end of the line. Many other C implementations allow | |
f458d1d5 ZW |
2621 | such comments, and they are included in the 1999 C standard. However, |
2622 | C++ style comments are not recognized if you specify an @option{-std} | |
2623 | option specifying a version of ISO C before C99, or @option{-ansi} | |
2624 | (equivalent to @option{-std=c89}). | |
c1f7febf RK |
2625 | |
2626 | @node Dollar Signs | |
2627 | @section Dollar Signs in Identifier Names | |
2628 | @cindex $ | |
2629 | @cindex dollar signs in identifier names | |
2630 | @cindex identifier names, dollar signs in | |
2631 | ||
79188db9 RK |
2632 | In GNU C, you may normally use dollar signs in identifier names. |
2633 | This is because many traditional C implementations allow such identifiers. | |
2634 | However, dollar signs in identifiers are not supported on a few target | |
2635 | machines, typically because the target assembler does not allow them. | |
c1f7febf RK |
2636 | |
2637 | @node Character Escapes | |
2638 | @section The Character @key{ESC} in Constants | |
2639 | ||
2640 | You can use the sequence @samp{\e} in a string or character constant to | |
2641 | stand for the ASCII character @key{ESC}. | |
2642 | ||
2643 | @node Alignment | |
2644 | @section Inquiring on Alignment of Types or Variables | |
2645 | @cindex alignment | |
2646 | @cindex type alignment | |
2647 | @cindex variable alignment | |
2648 | ||
2649 | The keyword @code{__alignof__} allows you to inquire about how an object | |
2650 | is aligned, or the minimum alignment usually required by a type. Its | |
2651 | syntax is just like @code{sizeof}. | |
2652 | ||
2653 | For example, if the target machine requires a @code{double} value to be | |
2654 | aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8. | |
2655 | This is true on many RISC machines. On more traditional machine | |
2656 | designs, @code{__alignof__ (double)} is 4 or even 2. | |
2657 | ||
2658 | Some machines never actually require alignment; they allow reference to any | |
64c18e57 | 2659 | data type even at an odd address. For these machines, @code{__alignof__} |
c1f7febf RK |
2660 | reports the @emph{recommended} alignment of a type. |
2661 | ||
5372b3fb NB |
2662 | If the operand of @code{__alignof__} is an lvalue rather than a type, |
2663 | its value is the required alignment for its type, taking into account | |
2664 | any minimum alignment specified with GCC's @code{__attribute__} | |
2665 | extension (@pxref{Variable Attributes}). For example, after this | |
2666 | declaration: | |
c1f7febf | 2667 | |
3ab51846 | 2668 | @smallexample |
c1f7febf | 2669 | struct foo @{ int x; char y; @} foo1; |
3ab51846 | 2670 | @end smallexample |
c1f7febf RK |
2671 | |
2672 | @noindent | |
5372b3fb NB |
2673 | the value of @code{__alignof__ (foo1.y)} is 1, even though its actual |
2674 | alignment is probably 2 or 4, the same as @code{__alignof__ (int)}. | |
c1f7febf | 2675 | |
9d27bffe SS |
2676 | It is an error to ask for the alignment of an incomplete type. |
2677 | ||
c1f7febf RK |
2678 | @node Variable Attributes |
2679 | @section Specifying Attributes of Variables | |
2680 | @cindex attribute of variables | |
2681 | @cindex variable attributes | |
2682 | ||
2683 | The keyword @code{__attribute__} allows you to specify special | |
2684 | attributes of variables or structure fields. This keyword is followed | |
905e8651 RH |
2685 | by an attribute specification inside double parentheses. Some |
2686 | attributes are currently defined generically for variables. | |
2687 | Other attributes are defined for variables on particular target | |
2688 | systems. Other attributes are available for functions | |
2689 | (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}). | |
2690 | Other front ends might define more attributes | |
2691 | (@pxref{C++ Extensions,,Extensions to the C++ Language}). | |
c1f7febf RK |
2692 | |
2693 | You may also specify attributes with @samp{__} preceding and following | |
2694 | each keyword. This allows you to use them in header files without | |
2695 | being concerned about a possible macro of the same name. For example, | |
2696 | you may use @code{__aligned__} instead of @code{aligned}. | |
2697 | ||
2c5e91d2 JM |
2698 | @xref{Attribute Syntax}, for details of the exact syntax for using |
2699 | attributes. | |
2700 | ||
c1f7febf RK |
2701 | @table @code |
2702 | @cindex @code{aligned} attribute | |
2703 | @item aligned (@var{alignment}) | |
2704 | This attribute specifies a minimum alignment for the variable or | |
2705 | structure field, measured in bytes. For example, the declaration: | |
2706 | ||
2707 | @smallexample | |
2708 | int x __attribute__ ((aligned (16))) = 0; | |
2709 | @end smallexample | |
2710 | ||
2711 | @noindent | |
2712 | causes the compiler to allocate the global variable @code{x} on a | |
2713 | 16-byte boundary. On a 68040, this could be used in conjunction with | |
2714 | an @code{asm} expression to access the @code{move16} instruction which | |
2715 | requires 16-byte aligned operands. | |
2716 | ||
2717 | You can also specify the alignment of structure fields. For example, to | |
2718 | create a double-word aligned @code{int} pair, you could write: | |
2719 | ||
2720 | @smallexample | |
2721 | struct foo @{ int x[2] __attribute__ ((aligned (8))); @}; | |
2722 | @end smallexample | |
2723 | ||
2724 | @noindent | |
2725 | This is an alternative to creating a union with a @code{double} member | |
2726 | that forces the union to be double-word aligned. | |
2727 | ||
c1f7febf RK |
2728 | As in the preceding examples, you can explicitly specify the alignment |
2729 | (in bytes) that you wish the compiler to use for a given variable or | |
2730 | structure field. Alternatively, you can leave out the alignment factor | |
2731 | and just ask the compiler to align a variable or field to the maximum | |
2732 | useful alignment for the target machine you are compiling for. For | |
2733 | example, you could write: | |
2734 | ||
2735 | @smallexample | |
2736 | short array[3] __attribute__ ((aligned)); | |
2737 | @end smallexample | |
2738 | ||
2739 | Whenever you leave out the alignment factor in an @code{aligned} attribute | |
2740 | specification, the compiler automatically sets the alignment for the declared | |
2741 | variable or field to the largest alignment which is ever used for any data | |
2742 | type on the target machine you are compiling for. Doing this can often make | |
2743 | copy operations more efficient, because the compiler can use whatever | |
2744 | instructions copy the biggest chunks of memory when performing copies to | |
2745 | or from the variables or fields that you have aligned this way. | |
2746 | ||
2747 | The @code{aligned} attribute can only increase the alignment; but you | |
2748 | can decrease it by specifying @code{packed} as well. See below. | |
2749 | ||
2750 | Note that the effectiveness of @code{aligned} attributes may be limited | |
2751 | by inherent limitations in your linker. On many systems, the linker is | |
2752 | only able to arrange for variables to be aligned up to a certain maximum | |
2753 | alignment. (For some linkers, the maximum supported alignment may | |
2754 | be very very small.) If your linker is only able to align variables | |
2755 | up to a maximum of 8 byte alignment, then specifying @code{aligned(16)} | |
2756 | in an @code{__attribute__} will still only provide you with 8 byte | |
2757 | alignment. See your linker documentation for further information. | |
2758 | ||
0bfa5f65 RH |
2759 | @item cleanup (@var{cleanup_function}) |
2760 | @cindex @code{cleanup} attribute | |
2761 | The @code{cleanup} attribute runs a function when the variable goes | |
2762 | out of scope. This attribute can only be applied to auto function | |
2763 | scope variables; it may not be applied to parameters or variables | |
2764 | with static storage duration. The function must take one parameter, | |
2765 | a pointer to a type compatible with the variable. The return value | |
2766 | of the function (if any) is ignored. | |
2767 | ||
2768 | If @option{-fexceptions} is enabled, then @var{cleanup_function} | |
2769 | will be run during the stack unwinding that happens during the | |
2770 | processing of the exception. Note that the @code{cleanup} attribute | |
2771 | does not allow the exception to be caught, only to perform an action. | |
2772 | It is undefined what happens if @var{cleanup_function} does not | |
2773 | return normally. | |
2774 | ||
905e8651 RH |
2775 | @item common |
2776 | @itemx nocommon | |
2777 | @cindex @code{common} attribute | |
2778 | @cindex @code{nocommon} attribute | |
2779 | @opindex fcommon | |
2780 | @opindex fno-common | |
2781 | The @code{common} attribute requests GCC to place a variable in | |
2782 | ``common'' storage. The @code{nocommon} attribute requests the | |
78466c0e | 2783 | opposite---to allocate space for it directly. |
905e8651 | 2784 | |
daf2f129 | 2785 | These attributes override the default chosen by the |
905e8651 RH |
2786 | @option{-fno-common} and @option{-fcommon} flags respectively. |
2787 | ||
2788 | @item deprecated | |
2789 | @cindex @code{deprecated} attribute | |
2790 | The @code{deprecated} attribute results in a warning if the variable | |
2791 | is used anywhere in the source file. This is useful when identifying | |
2792 | variables that are expected to be removed in a future version of a | |
2793 | program. The warning also includes the location of the declaration | |
2794 | of the deprecated variable, to enable users to easily find further | |
2795 | information about why the variable is deprecated, or what they should | |
64c18e57 | 2796 | do instead. Note that the warning only occurs for uses: |
905e8651 RH |
2797 | |
2798 | @smallexample | |
2799 | extern int old_var __attribute__ ((deprecated)); | |
2800 | extern int old_var; | |
2801 | int new_fn () @{ return old_var; @} | |
2802 | @end smallexample | |
2803 | ||
2804 | results in a warning on line 3 but not line 2. | |
2805 | ||
2806 | The @code{deprecated} attribute can also be used for functions and | |
2807 | types (@pxref{Function Attributes}, @pxref{Type Attributes}.) | |
2808 | ||
c1f7febf RK |
2809 | @item mode (@var{mode}) |
2810 | @cindex @code{mode} attribute | |
2811 | This attribute specifies the data type for the declaration---whichever | |
2812 | type corresponds to the mode @var{mode}. This in effect lets you | |
2813 | request an integer or floating point type according to its width. | |
2814 | ||
2815 | You may also specify a mode of @samp{byte} or @samp{__byte__} to | |
2816 | indicate the mode corresponding to a one-byte integer, @samp{word} or | |
2817 | @samp{__word__} for the mode of a one-word integer, and @samp{pointer} | |
2818 | or @samp{__pointer__} for the mode used to represent pointers. | |
2819 | ||
c1f7febf RK |
2820 | @item packed |
2821 | @cindex @code{packed} attribute | |
2822 | The @code{packed} attribute specifies that a variable or structure field | |
2823 | should have the smallest possible alignment---one byte for a variable, | |
2824 | and one bit for a field, unless you specify a larger value with the | |
2825 | @code{aligned} attribute. | |
2826 | ||
2827 | Here is a structure in which the field @code{x} is packed, so that it | |
2828 | immediately follows @code{a}: | |
2829 | ||
3ab51846 | 2830 | @smallexample |
c1f7febf RK |
2831 | struct foo |
2832 | @{ | |
2833 | char a; | |
2834 | int x[2] __attribute__ ((packed)); | |
2835 | @}; | |
3ab51846 | 2836 | @end smallexample |
c1f7febf | 2837 | |
84330467 | 2838 | @item section ("@var{section-name}") |
c1f7febf RK |
2839 | @cindex @code{section} variable attribute |
2840 | Normally, the compiler places the objects it generates in sections like | |
2841 | @code{data} and @code{bss}. Sometimes, however, you need additional sections, | |
2842 | or you need certain particular variables to appear in special sections, | |
2843 | for example to map to special hardware. The @code{section} | |
2844 | attribute specifies that a variable (or function) lives in a particular | |
2845 | section. For example, this small program uses several specific section names: | |
2846 | ||
2847 | @smallexample | |
2848 | struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @}; | |
2849 | struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @}; | |
2850 | char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @}; | |
2851 | int init_data __attribute__ ((section ("INITDATA"))) = 0; | |
2852 | ||
2853 | main() | |
2854 | @{ | |
2855 | /* Initialize stack pointer */ | |
2856 | init_sp (stack + sizeof (stack)); | |
2857 | ||
2858 | /* Initialize initialized data */ | |
2859 | memcpy (&init_data, &data, &edata - &data); | |
2860 | ||
2861 | /* Turn on the serial ports */ | |
2862 | init_duart (&a); | |
2863 | init_duart (&b); | |
2864 | @} | |
2865 | @end smallexample | |
2866 | ||
2867 | @noindent | |
2868 | Use the @code{section} attribute with an @emph{initialized} definition | |
f0523f02 | 2869 | of a @emph{global} variable, as shown in the example. GCC issues |
c1f7febf RK |
2870 | a warning and otherwise ignores the @code{section} attribute in |
2871 | uninitialized variable declarations. | |
2872 | ||
2873 | You may only use the @code{section} attribute with a fully initialized | |
2874 | global definition because of the way linkers work. The linker requires | |
2875 | each object be defined once, with the exception that uninitialized | |
2876 | variables tentatively go in the @code{common} (or @code{bss}) section | |
84330467 JM |
2877 | and can be multiply ``defined''. You can force a variable to be |
2878 | initialized with the @option{-fno-common} flag or the @code{nocommon} | |
c1f7febf RK |
2879 | attribute. |
2880 | ||
2881 | Some file formats do not support arbitrary sections so the @code{section} | |
2882 | attribute is not available on all platforms. | |
2883 | If you need to map the entire contents of a module to a particular | |
2884 | section, consider using the facilities of the linker instead. | |
2885 | ||
593d3a34 MK |
2886 | @item shared |
2887 | @cindex @code{shared} variable attribute | |
95fef11f | 2888 | On Microsoft Windows, in addition to putting variable definitions in a named |
02f52e19 | 2889 | section, the section can also be shared among all running copies of an |
161d7b59 | 2890 | executable or DLL@. For example, this small program defines shared data |
84330467 | 2891 | by putting it in a named section @code{shared} and marking the section |
593d3a34 MK |
2892 | shareable: |
2893 | ||
2894 | @smallexample | |
2895 | int foo __attribute__((section ("shared"), shared)) = 0; | |
2896 | ||
2897 | int | |
2898 | main() | |
2899 | @{ | |
310668e8 JM |
2900 | /* Read and write foo. All running |
2901 | copies see the same value. */ | |
593d3a34 MK |
2902 | return 0; |
2903 | @} | |
2904 | @end smallexample | |
2905 | ||
2906 | @noindent | |
2907 | You may only use the @code{shared} attribute along with @code{section} | |
02f52e19 | 2908 | attribute with a fully initialized global definition because of the way |
593d3a34 MK |
2909 | linkers work. See @code{section} attribute for more information. |
2910 | ||
95fef11f | 2911 | The @code{shared} attribute is only available on Microsoft Windows@. |
593d3a34 | 2912 | |
905e8651 RH |
2913 | @item tls_model ("@var{tls_model}") |
2914 | @cindex @code{tls_model} attribute | |
2915 | The @code{tls_model} attribute sets thread-local storage model | |
2916 | (@pxref{Thread-Local}) of a particular @code{__thread} variable, | |
4ec7afd7 | 2917 | overriding @option{-ftls-model=} command line switch on a per-variable |
905e8651 RH |
2918 | basis. |
2919 | The @var{tls_model} argument should be one of @code{global-dynamic}, | |
2920 | @code{local-dynamic}, @code{initial-exec} or @code{local-exec}. | |
2921 | ||
2922 | Not all targets support this attribute. | |
2923 | ||
c1f7febf RK |
2924 | @item transparent_union |
2925 | This attribute, attached to a function parameter which is a union, means | |
2926 | that the corresponding argument may have the type of any union member, | |
2927 | but the argument is passed as if its type were that of the first union | |
2928 | member. For more details see @xref{Type Attributes}. You can also use | |
2929 | this attribute on a @code{typedef} for a union data type; then it | |
2930 | applies to all function parameters with that type. | |
2931 | ||
2932 | @item unused | |
2933 | This attribute, attached to a variable, means that the variable is meant | |
f0523f02 | 2934 | to be possibly unused. GCC will not produce a warning for this |
c1f7febf RK |
2935 | variable. |
2936 | ||
1b9191d2 AH |
2937 | @item vector_size (@var{bytes}) |
2938 | This attribute specifies the vector size for the variable, measured in | |
2939 | bytes. For example, the declaration: | |
2940 | ||
2941 | @smallexample | |
2942 | int foo __attribute__ ((vector_size (16))); | |
2943 | @end smallexample | |
2944 | ||
2945 | @noindent | |
2946 | causes the compiler to set the mode for @code{foo}, to be 16 bytes, | |
2947 | divided into @code{int} sized units. Assuming a 32-bit int (a vector of | |
2948 | 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@. | |
2949 | ||
2950 | This attribute is only applicable to integral and float scalars, | |
2951 | although arrays, pointers, and function return values are allowed in | |
2952 | conjunction with this construct. | |
2953 | ||
2954 | Aggregates with this attribute are invalid, even if they are of the same | |
2955 | size as a corresponding scalar. For example, the declaration: | |
2956 | ||
2957 | @smallexample | |
ad706f54 | 2958 | struct S @{ int a; @}; |
1b9191d2 AH |
2959 | struct S __attribute__ ((vector_size (16))) foo; |
2960 | @end smallexample | |
2961 | ||
2962 | @noindent | |
2963 | is invalid even if the size of the structure is the same as the size of | |
2964 | the @code{int}. | |
2965 | ||
c1f7febf RK |
2966 | @item weak |
2967 | The @code{weak} attribute is described in @xref{Function Attributes}. | |
6b6cb52e DS |
2968 | |
2969 | @item dllimport | |
2970 | The @code{dllimport} attribute is described in @xref{Function Attributes}. | |
2971 | ||
2972 | @item dlexport | |
2973 | The @code{dllexport} attribute is described in @xref{Function Attributes}. | |
2974 | ||
905e8651 RH |
2975 | @end table |
2976 | ||
2977 | @subsection M32R/D Variable Attributes | |
845da534 | 2978 | |
8a36672b | 2979 | One attribute is currently defined for the M32R/D@. |
905e8651 RH |
2980 | |
2981 | @table @code | |
845da534 DE |
2982 | @item model (@var{model-name}) |
2983 | @cindex variable addressability on the M32R/D | |
2984 | Use this attribute on the M32R/D to set the addressability of an object. | |
2985 | The identifier @var{model-name} is one of @code{small}, @code{medium}, | |
2986 | or @code{large}, representing each of the code models. | |
2987 | ||
2988 | Small model objects live in the lower 16MB of memory (so that their | |
2989 | addresses can be loaded with the @code{ld24} instruction). | |
2990 | ||
02f52e19 | 2991 | Medium and large model objects may live anywhere in the 32-bit address space |
845da534 DE |
2992 | (the compiler will generate @code{seth/add3} instructions to load their |
2993 | addresses). | |
905e8651 | 2994 | @end table |
845da534 | 2995 | |
fe77449a DR |
2996 | @subsection i386 Variable Attributes |
2997 | ||
2998 | Two attributes are currently defined for i386 configurations: | |
2999 | @code{ms_struct} and @code{gcc_struct} | |
3000 | ||
905e8651 | 3001 | @table @code |
fe77449a DR |
3002 | @item ms_struct |
3003 | @itemx gcc_struct | |
905e8651 RH |
3004 | @cindex @code{ms_struct} attribute |
3005 | @cindex @code{gcc_struct} attribute | |
fe77449a DR |
3006 | |
3007 | If @code{packed} is used on a structure, or if bit-fields are used | |
3008 | it may be that the Microsoft ABI packs them differently | |
3009 | than GCC would normally pack them. Particularly when moving packed | |
3010 | data between functions compiled with GCC and the native Microsoft compiler | |
3011 | (either via function call or as data in a file), it may be necessary to access | |
3012 | either format. | |
3013 | ||
95fef11f | 3014 | Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86 |
fe77449a | 3015 | compilers to match the native Microsoft compiler. |
c1f7febf RK |
3016 | @end table |
3017 | ||
54e9a19d DD |
3018 | @subsection Xstormy16 Variable Attributes |
3019 | ||
3020 | One attribute is currently defined for xstormy16 configurations: | |
3021 | @code{below100} | |
3022 | ||
3023 | @table @code | |
3024 | @item below100 | |
3025 | @cindex @code{below100} attribute | |
3026 | ||
3027 | If a variable has the @code{below100} attribute (@code{BELOW100} is | |
3028 | allowed also), GCC will place the variable in the first 0x100 bytes of | |
3029 | memory and use special opcodes to access it. Such variables will be | |
3030 | placed in either the @code{.bss_below100} section or the | |
3031 | @code{.data_below100} section. | |
3032 | ||
3033 | @end table | |
3034 | ||
c1f7febf RK |
3035 | @node Type Attributes |
3036 | @section Specifying Attributes of Types | |
3037 | @cindex attribute of types | |
3038 | @cindex type attributes | |
3039 | ||
3040 | The keyword @code{__attribute__} allows you to specify special | |
3041 | attributes of @code{struct} and @code{union} types when you define such | |
3042 | types. This keyword is followed by an attribute specification inside | |
d18b1ed8 | 3043 | double parentheses. Six attributes are currently defined for types: |
e23bd218 | 3044 | @code{aligned}, @code{packed}, @code{transparent_union}, @code{unused}, |
d18b1ed8 OS |
3045 | @code{deprecated} and @code{may_alias}. Other attributes are defined for |
3046 | functions (@pxref{Function Attributes}) and for variables | |
3047 | (@pxref{Variable Attributes}). | |
c1f7febf RK |
3048 | |
3049 | You may also specify any one of these attributes with @samp{__} | |
3050 | preceding and following its keyword. This allows you to use these | |
3051 | attributes in header files without being concerned about a possible | |
3052 | macro of the same name. For example, you may use @code{__aligned__} | |
3053 | instead of @code{aligned}. | |
3054 | ||
3055 | You may specify the @code{aligned} and @code{transparent_union} | |
3056 | attributes either in a @code{typedef} declaration or just past the | |
3057 | closing curly brace of a complete enum, struct or union type | |
3058 | @emph{definition} and the @code{packed} attribute only past the closing | |
3059 | brace of a definition. | |
3060 | ||
4051959b JM |
3061 | You may also specify attributes between the enum, struct or union |
3062 | tag and the name of the type rather than after the closing brace. | |
3063 | ||
2c5e91d2 JM |
3064 | @xref{Attribute Syntax}, for details of the exact syntax for using |
3065 | attributes. | |
3066 | ||
c1f7febf RK |
3067 | @table @code |
3068 | @cindex @code{aligned} attribute | |
3069 | @item aligned (@var{alignment}) | |
3070 | This attribute specifies a minimum alignment (in bytes) for variables | |
3071 | of the specified type. For example, the declarations: | |
3072 | ||
3073 | @smallexample | |
f69eecfb JL |
3074 | struct S @{ short f[3]; @} __attribute__ ((aligned (8))); |
3075 | typedef int more_aligned_int __attribute__ ((aligned (8))); | |
c1f7febf RK |
3076 | @end smallexample |
3077 | ||
3078 | @noindent | |
d863830b | 3079 | force the compiler to insure (as far as it can) that each variable whose |
c1f7febf | 3080 | type is @code{struct S} or @code{more_aligned_int} will be allocated and |
981f6289 | 3081 | aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all |
c1f7febf RK |
3082 | variables of type @code{struct S} aligned to 8-byte boundaries allows |
3083 | the compiler to use the @code{ldd} and @code{std} (doubleword load and | |
3084 | store) instructions when copying one variable of type @code{struct S} to | |
3085 | another, thus improving run-time efficiency. | |
3086 | ||
3087 | Note that the alignment of any given @code{struct} or @code{union} type | |
5490d604 | 3088 | is required by the ISO C standard to be at least a perfect multiple of |
c1f7febf RK |
3089 | the lowest common multiple of the alignments of all of the members of |
3090 | the @code{struct} or @code{union} in question. This means that you @emph{can} | |
3091 | effectively adjust the alignment of a @code{struct} or @code{union} | |
3092 | type by attaching an @code{aligned} attribute to any one of the members | |
3093 | of such a type, but the notation illustrated in the example above is a | |
3094 | more obvious, intuitive, and readable way to request the compiler to | |
3095 | adjust the alignment of an entire @code{struct} or @code{union} type. | |
3096 | ||
3097 | As in the preceding example, you can explicitly specify the alignment | |
3098 | (in bytes) that you wish the compiler to use for a given @code{struct} | |
3099 | or @code{union} type. Alternatively, you can leave out the alignment factor | |
3100 | and just ask the compiler to align a type to the maximum | |
3101 | useful alignment for the target machine you are compiling for. For | |
3102 | example, you could write: | |
3103 | ||
3104 | @smallexample | |
3105 | struct S @{ short f[3]; @} __attribute__ ((aligned)); | |
3106 | @end smallexample | |
3107 | ||
3108 | Whenever you leave out the alignment factor in an @code{aligned} | |
3109 | attribute specification, the compiler automatically sets the alignment | |
3110 | for the type to the largest alignment which is ever used for any data | |
3111 | type on the target machine you are compiling for. Doing this can often | |
3112 | make copy operations more efficient, because the compiler can use | |
3113 | whatever instructions copy the biggest chunks of memory when performing | |
3114 | copies to or from the variables which have types that you have aligned | |
3115 | this way. | |
3116 | ||
3117 | In the example above, if the size of each @code{short} is 2 bytes, then | |
3118 | the size of the entire @code{struct S} type is 6 bytes. The smallest | |
3119 | power of two which is greater than or equal to that is 8, so the | |
3120 | compiler sets the alignment for the entire @code{struct S} type to 8 | |
3121 | bytes. | |
3122 | ||
3123 | Note that although you can ask the compiler to select a time-efficient | |
3124 | alignment for a given type and then declare only individual stand-alone | |
3125 | objects of that type, the compiler's ability to select a time-efficient | |
3126 | alignment is primarily useful only when you plan to create arrays of | |
3127 | variables having the relevant (efficiently aligned) type. If you | |
3128 | declare or use arrays of variables of an efficiently-aligned type, then | |
3129 | it is likely that your program will also be doing pointer arithmetic (or | |
3130 | subscripting, which amounts to the same thing) on pointers to the | |
3131 | relevant type, and the code that the compiler generates for these | |
3132 | pointer arithmetic operations will often be more efficient for | |
3133 | efficiently-aligned types than for other types. | |
3134 | ||
3135 | The @code{aligned} attribute can only increase the alignment; but you | |
3136 | can decrease it by specifying @code{packed} as well. See below. | |
3137 | ||
3138 | Note that the effectiveness of @code{aligned} attributes may be limited | |
3139 | by inherent limitations in your linker. On many systems, the linker is | |
3140 | only able to arrange for variables to be aligned up to a certain maximum | |
3141 | alignment. (For some linkers, the maximum supported alignment may | |
3142 | be very very small.) If your linker is only able to align variables | |
3143 | up to a maximum of 8 byte alignment, then specifying @code{aligned(16)} | |
3144 | in an @code{__attribute__} will still only provide you with 8 byte | |
3145 | alignment. See your linker documentation for further information. | |
3146 | ||
3147 | @item packed | |
a5bcc582 NS |
3148 | This attribute, attached to @code{struct} or @code{union} type |
3149 | definition, specifies that each member of the structure or union is | |
8a36672b | 3150 | placed to minimize the memory required. When attached to an @code{enum} |
a5bcc582 | 3151 | definition, it indicates that the smallest integral type should be used. |
c1f7febf | 3152 | |
84330467 | 3153 | @opindex fshort-enums |
c1f7febf RK |
3154 | Specifying this attribute for @code{struct} and @code{union} types is |
3155 | equivalent to specifying the @code{packed} attribute on each of the | |
84330467 | 3156 | structure or union members. Specifying the @option{-fshort-enums} |
c1f7febf RK |
3157 | flag on the line is equivalent to specifying the @code{packed} |
3158 | attribute on all @code{enum} definitions. | |
3159 | ||
a5bcc582 NS |
3160 | In the following example @code{struct my_packed_struct}'s members are |
3161 | packed closely together, but the internal layout of its @code{s} member | |
78466c0e | 3162 | is not packed---to do that, @code{struct my_unpacked_struct} would need to |
a5bcc582 NS |
3163 | be packed too. |
3164 | ||
3165 | @smallexample | |
3166 | struct my_unpacked_struct | |
3167 | @{ | |
3168 | char c; | |
3169 | int i; | |
3170 | @}; | |
3171 | ||
3172 | struct my_packed_struct __attribute__ ((__packed__)) | |
3173 | @{ | |
3174 | char c; | |
3175 | int i; | |
3176 | struct my_unpacked_struct s; | |
3177 | @}; | |
3178 | @end smallexample | |
3179 | ||
3180 | You may only specify this attribute on the definition of a @code{enum}, | |
3181 | @code{struct} or @code{union}, not on a @code{typedef} which does not | |
3182 | also define the enumerated type, structure or union. | |
c1f7febf RK |
3183 | |
3184 | @item transparent_union | |
3185 | This attribute, attached to a @code{union} type definition, indicates | |
3186 | that any function parameter having that union type causes calls to that | |
3187 | function to be treated in a special way. | |
3188 | ||
3189 | First, the argument corresponding to a transparent union type can be of | |
3190 | any type in the union; no cast is required. Also, if the union contains | |
3191 | a pointer type, the corresponding argument can be a null pointer | |
3192 | constant or a void pointer expression; and if the union contains a void | |
3193 | pointer type, the corresponding argument can be any pointer expression. | |
3194 | If the union member type is a pointer, qualifiers like @code{const} on | |
3195 | the referenced type must be respected, just as with normal pointer | |
3196 | conversions. | |
3197 | ||
3198 | Second, the argument is passed to the function using the calling | |
64c18e57 | 3199 | conventions of the first member of the transparent union, not the calling |
c1f7febf RK |
3200 | conventions of the union itself. All members of the union must have the |
3201 | same machine representation; this is necessary for this argument passing | |
3202 | to work properly. | |
3203 | ||
3204 | Transparent unions are designed for library functions that have multiple | |
3205 | interfaces for compatibility reasons. For example, suppose the | |
3206 | @code{wait} function must accept either a value of type @code{int *} to | |
3207 | comply with Posix, or a value of type @code{union wait *} to comply with | |
3208 | the 4.1BSD interface. If @code{wait}'s parameter were @code{void *}, | |
3209 | @code{wait} would accept both kinds of arguments, but it would also | |
3210 | accept any other pointer type and this would make argument type checking | |
3211 | less useful. Instead, @code{<sys/wait.h>} might define the interface | |
3212 | as follows: | |
3213 | ||
3214 | @smallexample | |
3215 | typedef union | |
3216 | @{ | |
3217 | int *__ip; | |
3218 | union wait *__up; | |
3219 | @} wait_status_ptr_t __attribute__ ((__transparent_union__)); | |
3220 | ||
3221 | pid_t wait (wait_status_ptr_t); | |
3222 | @end smallexample | |
3223 | ||
3224 | This interface allows either @code{int *} or @code{union wait *} | |
3225 | arguments to be passed, using the @code{int *} calling convention. | |
3226 | The program can call @code{wait} with arguments of either type: | |
3227 | ||
3ab51846 | 3228 | @smallexample |
c1f7febf RK |
3229 | int w1 () @{ int w; return wait (&w); @} |
3230 | int w2 () @{ union wait w; return wait (&w); @} | |
3ab51846 | 3231 | @end smallexample |
c1f7febf RK |
3232 | |
3233 | With this interface, @code{wait}'s implementation might look like this: | |
3234 | ||
3ab51846 | 3235 | @smallexample |
c1f7febf RK |
3236 | pid_t wait (wait_status_ptr_t p) |
3237 | @{ | |
3238 | return waitpid (-1, p.__ip, 0); | |
3239 | @} | |
3ab51846 | 3240 | @end smallexample |
d863830b JL |
3241 | |
3242 | @item unused | |
3243 | When attached to a type (including a @code{union} or a @code{struct}), | |
3244 | this attribute means that variables of that type are meant to appear | |
f0523f02 | 3245 | possibly unused. GCC will not produce a warning for any variables of |
d863830b JL |
3246 | that type, even if the variable appears to do nothing. This is often |
3247 | the case with lock or thread classes, which are usually defined and then | |
3248 | not referenced, but contain constructors and destructors that have | |
956d6950 | 3249 | nontrivial bookkeeping functions. |
d863830b | 3250 | |
e23bd218 IR |
3251 | @item deprecated |
3252 | The @code{deprecated} attribute results in a warning if the type | |
3253 | is used anywhere in the source file. This is useful when identifying | |
3254 | types that are expected to be removed in a future version of a program. | |
3255 | If possible, the warning also includes the location of the declaration | |
3256 | of the deprecated type, to enable users to easily find further | |
3257 | information about why the type is deprecated, or what they should do | |
3258 | instead. Note that the warnings only occur for uses and then only | |
adc9fe67 | 3259 | if the type is being applied to an identifier that itself is not being |
e23bd218 IR |
3260 | declared as deprecated. |
3261 | ||
3262 | @smallexample | |
3263 | typedef int T1 __attribute__ ((deprecated)); | |
3264 | T1 x; | |
3265 | typedef T1 T2; | |
3266 | T2 y; | |
3267 | typedef T1 T3 __attribute__ ((deprecated)); | |
3268 | T3 z __attribute__ ((deprecated)); | |
3269 | @end smallexample | |
3270 | ||
3271 | results in a warning on line 2 and 3 but not lines 4, 5, or 6. No | |
3272 | warning is issued for line 4 because T2 is not explicitly | |
3273 | deprecated. Line 5 has no warning because T3 is explicitly | |
3274 | deprecated. Similarly for line 6. | |
3275 | ||
3276 | The @code{deprecated} attribute can also be used for functions and | |
3277 | variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.) | |
3278 | ||
d18b1ed8 OS |
3279 | @item may_alias |
3280 | Accesses to objects with types with this attribute are not subjected to | |
3281 | type-based alias analysis, but are instead assumed to be able to alias | |
3282 | any other type of objects, just like the @code{char} type. See | |
3283 | @option{-fstrict-aliasing} for more information on aliasing issues. | |
3284 | ||
3285 | Example of use: | |
3286 | ||
478c9e72 | 3287 | @smallexample |
d18b1ed8 OS |
3288 | typedef short __attribute__((__may_alias__)) short_a; |
3289 | ||
3290 | int | |
3291 | main (void) | |
3292 | @{ | |
3293 | int a = 0x12345678; | |
3294 | short_a *b = (short_a *) &a; | |
3295 | ||
3296 | b[1] = 0; | |
3297 | ||
3298 | if (a == 0x12345678) | |
3299 | abort(); | |
3300 | ||
3301 | exit(0); | |
3302 | @} | |
478c9e72 | 3303 | @end smallexample |
d18b1ed8 OS |
3304 | |
3305 | If you replaced @code{short_a} with @code{short} in the variable | |
3306 | declaration, the above program would abort when compiled with | |
3307 | @option{-fstrict-aliasing}, which is on by default at @option{-O2} or | |
3308 | above in recent GCC versions. | |
fe77449a | 3309 | |
04fb56d5 MM |
3310 | @subsection ARM Type Attributes |
3311 | ||
3312 | On those ARM targets that support @code{dllimport} (such as Symbian | |
f0eb93a8 | 3313 | OS), you can use the @code{notshared} attribute to indicate that the |
04fb56d5 | 3314 | virtual table and other similar data for a class should not be |
8a36672b | 3315 | exported from a DLL@. For example: |
04fb56d5 MM |
3316 | |
3317 | @smallexample | |
3318 | class __declspec(notshared) C @{ | |
3319 | public: | |
f0eb93a8 | 3320 | __declspec(dllimport) C(); |
04fb56d5 MM |
3321 | virtual void f(); |
3322 | @} | |
3323 | ||
3324 | __declspec(dllexport) | |
3325 | C::C() @{@} | |
3326 | @end smallexample | |
3327 | ||
3328 | In this code, @code{C::C} is exported from the current DLL, but the | |
3329 | virtual table for @code{C} is not exported. (You can use | |
3330 | @code{__attribute__} instead of @code{__declspec} if you prefer, but | |
3331 | most Symbian OS code uses @code{__declspec}.) | |
3332 | ||
fe77449a DR |
3333 | @subsection i386 Type Attributes |
3334 | ||
3335 | Two attributes are currently defined for i386 configurations: | |
3336 | @code{ms_struct} and @code{gcc_struct} | |
3337 | ||
3338 | @item ms_struct | |
3339 | @itemx gcc_struct | |
3340 | @cindex @code{ms_struct} | |
3341 | @cindex @code{gcc_struct} | |
3342 | ||
3343 | If @code{packed} is used on a structure, or if bit-fields are used | |
3344 | it may be that the Microsoft ABI packs them differently | |
3345 | than GCC would normally pack them. Particularly when moving packed | |
3346 | data between functions compiled with GCC and the native Microsoft compiler | |
3347 | (either via function call or as data in a file), it may be necessary to access | |
3348 | either format. | |
3349 | ||
95fef11f | 3350 | Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86 |
fe77449a | 3351 | compilers to match the native Microsoft compiler. |
c1f7febf RK |
3352 | @end table |
3353 | ||
3354 | To specify multiple attributes, separate them by commas within the | |
3355 | double parentheses: for example, @samp{__attribute__ ((aligned (16), | |
3356 | packed))}. | |
3357 | ||
3358 | @node Inline | |
3359 | @section An Inline Function is As Fast As a Macro | |
3360 | @cindex inline functions | |
3361 | @cindex integrating function code | |
3362 | @cindex open coding | |
3363 | @cindex macros, inline alternative | |
3364 | ||
f0523f02 | 3365 | By declaring a function @code{inline}, you can direct GCC to |
c1f7febf RK |
3366 | integrate that function's code into the code for its callers. This |
3367 | makes execution faster by eliminating the function-call overhead; in | |
3368 | addition, if any of the actual argument values are constant, their known | |
3369 | values may permit simplifications at compile time so that not all of the | |
3370 | inline function's code needs to be included. The effect on code size is | |
3371 | less predictable; object code may be larger or smaller with function | |
3372 | inlining, depending on the particular case. Inlining of functions is an | |
3373 | optimization and it really ``works'' only in optimizing compilation. If | |
84330467 | 3374 | you don't use @option{-O}, no function is really inline. |
c1f7febf | 3375 | |
4b404517 JM |
3376 | Inline functions are included in the ISO C99 standard, but there are |
3377 | currently substantial differences between what GCC implements and what | |
3378 | the ISO C99 standard requires. | |
3379 | ||
c1f7febf RK |
3380 | To declare a function inline, use the @code{inline} keyword in its |
3381 | declaration, like this: | |
3382 | ||
3ab51846 | 3383 | @smallexample |
c1f7febf RK |
3384 | inline int |
3385 | inc (int *a) | |
3386 | @{ | |
3387 | (*a)++; | |
3388 | @} | |
3ab51846 | 3389 | @end smallexample |
c1f7febf | 3390 | |
5490d604 | 3391 | (If you are writing a header file to be included in ISO C programs, write |
c1f7febf | 3392 | @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}.) |
c1f7febf | 3393 | You can also make all ``simple enough'' functions inline with the option |
84330467 | 3394 | @option{-finline-functions}. |
247b14bd | 3395 | |
84330467 | 3396 | @opindex Winline |
247b14bd RH |
3397 | Note that certain usages in a function definition can make it unsuitable |
3398 | for inline substitution. Among these usages are: use of varargs, use of | |
3399 | alloca, use of variable sized data types (@pxref{Variable Length}), | |
3400 | use of computed goto (@pxref{Labels as Values}), use of nonlocal goto, | |
84330467 | 3401 | and nested functions (@pxref{Nested Functions}). Using @option{-Winline} |
247b14bd RH |
3402 | will warn when a function marked @code{inline} could not be substituted, |
3403 | and will give the reason for the failure. | |
c1f7febf | 3404 | |
2147b154 | 3405 | Note that in C and Objective-C, unlike C++, the @code{inline} keyword |
c1f7febf RK |
3406 | does not affect the linkage of the function. |
3407 | ||
3408 | @cindex automatic @code{inline} for C++ member fns | |
3409 | @cindex @code{inline} automatic for C++ member fns | |
3410 | @cindex member fns, automatically @code{inline} | |
3411 | @cindex C++ member fns, automatically @code{inline} | |
84330467 | 3412 | @opindex fno-default-inline |
f0523f02 | 3413 | GCC automatically inlines member functions defined within the class |
c1f7febf | 3414 | body of C++ programs even if they are not explicitly declared |
84330467 | 3415 | @code{inline}. (You can override this with @option{-fno-default-inline}; |
c1f7febf RK |
3416 | @pxref{C++ Dialect Options,,Options Controlling C++ Dialect}.) |
3417 | ||
3418 | @cindex inline functions, omission of | |
84330467 | 3419 | @opindex fkeep-inline-functions |
c1f7febf RK |
3420 | When a function is both inline and @code{static}, if all calls to the |
3421 | function are integrated into the caller, and the function's address is | |
3422 | never used, then the function's own assembler code is never referenced. | |
f0523f02 | 3423 | In this case, GCC does not actually output assembler code for the |
84330467 | 3424 | function, unless you specify the option @option{-fkeep-inline-functions}. |
c1f7febf RK |
3425 | Some calls cannot be integrated for various reasons (in particular, |
3426 | calls that precede the function's definition cannot be integrated, and | |
3427 | neither can recursive calls within the definition). If there is a | |
3428 | nonintegrated call, then the function is compiled to assembler code as | |
3429 | usual. The function must also be compiled as usual if the program | |
3430 | refers to its address, because that can't be inlined. | |
3431 | ||
3432 | @cindex non-static inline function | |
3433 | When an inline function is not @code{static}, then the compiler must assume | |
3434 | that there may be calls from other source files; since a global symbol can | |
3435 | be defined only once in any program, the function must not be defined in | |
3436 | the other source files, so the calls therein cannot be integrated. | |
3437 | Therefore, a non-@code{static} inline function is always compiled on its | |
3438 | own in the usual fashion. | |
3439 | ||
3440 | If you specify both @code{inline} and @code{extern} in the function | |
3441 | definition, then the definition is used only for inlining. In no case | |
3442 | is the function compiled on its own, not even if you refer to its | |
3443 | address explicitly. Such an address becomes an external reference, as | |
3444 | if you had only declared the function, and had not defined it. | |
3445 | ||
3446 | This combination of @code{inline} and @code{extern} has almost the | |
3447 | effect of a macro. The way to use it is to put a function definition in | |
3448 | a header file with these keywords, and put another copy of the | |
3449 | definition (lacking @code{inline} and @code{extern}) in a library file. | |
3450 | The definition in the header file will cause most calls to the function | |
3451 | to be inlined. If any uses of the function remain, they will refer to | |
3452 | the single copy in the library. | |
3453 | ||
64c18e57 DF |
3454 | Since GCC eventually will implement ISO C99 semantics for |
3455 | inline functions, it is best to use @code{static inline} only | |
9c2d4260 | 3456 | to guarantee compatibility. (The |
4b404517 JM |
3457 | existing semantics will remain available when @option{-std=gnu89} is |
3458 | specified, but eventually the default will be @option{-std=gnu99} and | |
3459 | that will implement the C99 semantics, though it does not do so yet.) | |
3460 | ||
6aa77e6c AH |
3461 | GCC does not inline any functions when not optimizing unless you specify |
3462 | the @samp{always_inline} attribute for the function, like this: | |
3463 | ||
3ab51846 | 3464 | @smallexample |
6aa77e6c AH |
3465 | /* Prototype. */ |
3466 | inline void foo (const char) __attribute__((always_inline)); | |
3ab51846 | 3467 | @end smallexample |
c1f7febf RK |
3468 | |
3469 | @node Extended Asm | |
3470 | @section Assembler Instructions with C Expression Operands | |
3471 | @cindex extended @code{asm} | |
3472 | @cindex @code{asm} expressions | |
3473 | @cindex assembler instructions | |
3474 | @cindex registers | |
3475 | ||
c85f7c16 JL |
3476 | In an assembler instruction using @code{asm}, you can specify the |
3477 | operands of the instruction using C expressions. This means you need not | |
3478 | guess which registers or memory locations will contain the data you want | |
c1f7febf RK |
3479 | to use. |
3480 | ||
c85f7c16 JL |
3481 | You must specify an assembler instruction template much like what |
3482 | appears in a machine description, plus an operand constraint string for | |
3483 | each operand. | |
c1f7febf RK |
3484 | |
3485 | For example, here is how to use the 68881's @code{fsinx} instruction: | |
3486 | ||
3ab51846 | 3487 | @smallexample |
c1f7febf | 3488 | asm ("fsinx %1,%0" : "=f" (result) : "f" (angle)); |
3ab51846 | 3489 | @end smallexample |
c1f7febf RK |
3490 | |
3491 | @noindent | |
3492 | Here @code{angle} is the C expression for the input operand while | |
3493 | @code{result} is that of the output operand. Each has @samp{"f"} as its | |
c85f7c16 JL |
3494 | operand constraint, saying that a floating point register is required. |
3495 | The @samp{=} in @samp{=f} indicates that the operand is an output; all | |
3496 | output operands' constraints must use @samp{=}. The constraints use the | |
3497 | same language used in the machine description (@pxref{Constraints}). | |
3498 | ||
3499 | Each operand is described by an operand-constraint string followed by | |
3500 | the C expression in parentheses. A colon separates the assembler | |
3501 | template from the first output operand and another separates the last | |
3502 | output operand from the first input, if any. Commas separate the | |
84b72302 RH |
3503 | operands within each group. The total number of operands is currently |
3504 | limited to 30; this limitation may be lifted in some future version of | |
8a36672b | 3505 | GCC@. |
c85f7c16 JL |
3506 | |
3507 | If there are no output operands but there are input operands, you must | |
3508 | place two consecutive colons surrounding the place where the output | |
c1f7febf RK |
3509 | operands would go. |
3510 | ||
84b72302 RH |
3511 | As of GCC version 3.1, it is also possible to specify input and output |
3512 | operands using symbolic names which can be referenced within the | |
3513 | assembler code. These names are specified inside square brackets | |
3514 | preceding the constraint string, and can be referenced inside the | |
3515 | assembler code using @code{%[@var{name}]} instead of a percentage sign | |
3516 | followed by the operand number. Using named operands the above example | |
3517 | could look like: | |
3518 | ||
3ab51846 | 3519 | @smallexample |
84b72302 RH |
3520 | asm ("fsinx %[angle],%[output]" |
3521 | : [output] "=f" (result) | |
3522 | : [angle] "f" (angle)); | |
3ab51846 | 3523 | @end smallexample |
84b72302 RH |
3524 | |
3525 | @noindent | |
3526 | Note that the symbolic operand names have no relation whatsoever to | |
3527 | other C identifiers. You may use any name you like, even those of | |
64c18e57 | 3528 | existing C symbols, but you must ensure that no two operands within the same |
84b72302 RH |
3529 | assembler construct use the same symbolic name. |
3530 | ||
c1f7febf | 3531 | Output operand expressions must be lvalues; the compiler can check this. |
c85f7c16 JL |
3532 | The input operands need not be lvalues. The compiler cannot check |
3533 | whether the operands have data types that are reasonable for the | |
3534 | instruction being executed. It does not parse the assembler instruction | |
3535 | template and does not know what it means or even whether it is valid | |
3536 | assembler input. The extended @code{asm} feature is most often used for | |
3537 | machine instructions the compiler itself does not know exist. If | |
3538 | the output expression cannot be directly addressed (for example, it is a | |
f0523f02 | 3539 | bit-field), your constraint must allow a register. In that case, GCC |
c85f7c16 JL |
3540 | will use the register as the output of the @code{asm}, and then store |
3541 | that register into the output. | |
3542 | ||
f0523f02 | 3543 | The ordinary output operands must be write-only; GCC will assume that |
c85f7c16 JL |
3544 | the values in these operands before the instruction are dead and need |
3545 | not be generated. Extended asm supports input-output or read-write | |
3546 | operands. Use the constraint character @samp{+} to indicate such an | |
373a04f1 JM |
3547 | operand and list it with the output operands. You should only use |
3548 | read-write operands when the constraints for the operand (or the | |
3549 | operand in which only some of the bits are to be changed) allow a | |
3550 | register. | |
3551 | ||
3552 | You may, as an alternative, logically split its function into two | |
3553 | separate operands, one input operand and one write-only output | |
3554 | operand. The connection between them is expressed by constraints | |
3555 | which say they need to be in the same location when the instruction | |
3556 | executes. You can use the same C expression for both operands, or | |
3557 | different expressions. For example, here we write the (fictitious) | |
3558 | @samp{combine} instruction with @code{bar} as its read-only source | |
3559 | operand and @code{foo} as its read-write destination: | |
c1f7febf | 3560 | |
3ab51846 | 3561 | @smallexample |
c1f7febf | 3562 | asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar)); |
3ab51846 | 3563 | @end smallexample |
c1f7febf RK |
3564 | |
3565 | @noindent | |
c85f7c16 | 3566 | The constraint @samp{"0"} for operand 1 says that it must occupy the |
84b72302 RH |
3567 | same location as operand 0. A number in constraint is allowed only in |
3568 | an input operand and it must refer to an output operand. | |
c1f7febf | 3569 | |
84b72302 | 3570 | Only a number in the constraint can guarantee that one operand will be in |
c85f7c16 JL |
3571 | the same place as another. The mere fact that @code{foo} is the value |
3572 | of both operands is not enough to guarantee that they will be in the | |
3573 | same place in the generated assembler code. The following would not | |
3574 | work reliably: | |
c1f7febf | 3575 | |
3ab51846 | 3576 | @smallexample |
c1f7febf | 3577 | asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar)); |
3ab51846 | 3578 | @end smallexample |
c1f7febf RK |
3579 | |
3580 | Various optimizations or reloading could cause operands 0 and 1 to be in | |
f0523f02 | 3581 | different registers; GCC knows no reason not to do so. For example, the |
c1f7febf RK |
3582 | compiler might find a copy of the value of @code{foo} in one register and |
3583 | use it for operand 1, but generate the output operand 0 in a different | |
3584 | register (copying it afterward to @code{foo}'s own address). Of course, | |
3585 | since the register for operand 1 is not even mentioned in the assembler | |
f0523f02 | 3586 | code, the result will not work, but GCC can't tell that. |
c1f7febf | 3587 | |
84b72302 RH |
3588 | As of GCC version 3.1, one may write @code{[@var{name}]} instead of |
3589 | the operand number for a matching constraint. For example: | |
3590 | ||
3ab51846 | 3591 | @smallexample |
84b72302 RH |
3592 | asm ("cmoveq %1,%2,%[result]" |
3593 | : [result] "=r"(result) | |
3594 | : "r" (test), "r"(new), "[result]"(old)); | |
3ab51846 | 3595 | @end smallexample |
84b72302 | 3596 | |
805c33df HPN |
3597 | Sometimes you need to make an @code{asm} operand be a specific register, |
3598 | but there's no matching constraint letter for that register @emph{by | |
3599 | itself}. To force the operand into that register, use a local variable | |
3600 | for the operand and specify the register in the variable declaration. | |
3601 | @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any | |
3602 | register constraint letter that matches the register: | |
3603 | ||
3604 | @smallexample | |
3605 | register int *p1 asm ("r0") = @dots{}; | |
3606 | register int *p2 asm ("r1") = @dots{}; | |
3607 | register int *result asm ("r0"); | |
3608 | asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2)); | |
3609 | @end smallexample | |
3610 | ||
b55d5746 HPN |
3611 | @anchor{Example of asm with clobbered asm reg} |
3612 | In the above example, beware that a register that is call-clobbered by | |
3613 | the target ABI will be overwritten by any function call in the | |
3614 | assignment, including library calls for arithmetic operators. | |
3615 | Assuming it is a call-clobbered register, this may happen to @code{r0} | |
3616 | above by the assignment to @code{p2}. If you have to use such a | |
3617 | register, use temporary variables for expressions between the register | |
3618 | assignment and use: | |
3619 | ||
3620 | @smallexample | |
3621 | int t1 = @dots{}; | |
3622 | register int *p1 asm ("r0") = @dots{}; | |
3623 | register int *p2 asm ("r1") = t1; | |
3624 | register int *result asm ("r0"); | |
3625 | asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2)); | |
3626 | @end smallexample | |
3627 | ||
c85f7c16 JL |
3628 | Some instructions clobber specific hard registers. To describe this, |
3629 | write a third colon after the input operands, followed by the names of | |
3630 | the clobbered hard registers (given as strings). Here is a realistic | |
3631 | example for the VAX: | |
c1f7febf | 3632 | |
3ab51846 | 3633 | @smallexample |
c1f7febf RK |
3634 | asm volatile ("movc3 %0,%1,%2" |
3635 | : /* no outputs */ | |
3636 | : "g" (from), "g" (to), "g" (count) | |
3637 | : "r0", "r1", "r2", "r3", "r4", "r5"); | |
3ab51846 | 3638 | @end smallexample |
c1f7febf | 3639 | |
c5c76735 JL |
3640 | You may not write a clobber description in a way that overlaps with an |
3641 | input or output operand. For example, you may not have an operand | |
3642 | describing a register class with one member if you mention that register | |
acb5d088 HPN |
3643 | in the clobber list. Variables declared to live in specific registers |
3644 | (@pxref{Explicit Reg Vars}), and used as asm input or output operands must | |
3645 | have no part mentioned in the clobber description. | |
3646 | There is no way for you to specify that an input | |
c5c76735 JL |
3647 | operand is modified without also specifying it as an output |
3648 | operand. Note that if all the output operands you specify are for this | |
3649 | purpose (and hence unused), you will then also need to specify | |
3650 | @code{volatile} for the @code{asm} construct, as described below, to | |
f0523f02 | 3651 | prevent GCC from deleting the @code{asm} statement as unused. |
8fe1938e | 3652 | |
c1f7febf | 3653 | If you refer to a particular hardware register from the assembler code, |
c85f7c16 JL |
3654 | you will probably have to list the register after the third colon to |
3655 | tell the compiler the register's value is modified. In some assemblers, | |
3656 | the register names begin with @samp{%}; to produce one @samp{%} in the | |
3657 | assembler code, you must write @samp{%%} in the input. | |
3658 | ||
3659 | If your assembler instruction can alter the condition code register, add | |
f0523f02 | 3660 | @samp{cc} to the list of clobbered registers. GCC on some machines |
c85f7c16 JL |
3661 | represents the condition codes as a specific hardware register; |
3662 | @samp{cc} serves to name this register. On other machines, the | |
3663 | condition code is handled differently, and specifying @samp{cc} has no | |
3664 | effect. But it is valid no matter what the machine. | |
c1f7febf | 3665 | |
bbf5a54d | 3666 | If your assembler instructions access memory in an unpredictable |
c85f7c16 | 3667 | fashion, add @samp{memory} to the list of clobbered registers. This |
bbf5a54d AJ |
3668 | will cause GCC to not keep memory values cached in registers across the |
3669 | assembler instruction and not optimize stores or loads to that memory. | |
3670 | You will also want to add the @code{volatile} keyword if the memory | |
3671 | affected is not listed in the inputs or outputs of the @code{asm}, as | |
3672 | the @samp{memory} clobber does not count as a side-effect of the | |
3673 | @code{asm}. If you know how large the accessed memory is, you can add | |
3674 | it as input or output but if this is not known, you should add | |
3675 | @samp{memory}. As an example, if you access ten bytes of a string, you | |
3676 | can use a memory input like: | |
3677 | ||
cd1a8088 | 3678 | @smallexample |
bbf5a54d | 3679 | @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}. |
cd1a8088 | 3680 | @end smallexample |
bbf5a54d AJ |
3681 | |
3682 | Note that in the following example the memory input is necessary, | |
3683 | otherwise GCC might optimize the store to @code{x} away: | |
cd1a8088 | 3684 | @smallexample |
bbf5a54d AJ |
3685 | int foo () |
3686 | @{ | |
3687 | int x = 42; | |
3688 | int *y = &x; | |
3689 | int result; | |
3690 | asm ("magic stuff accessing an 'int' pointed to by '%1'" | |
3691 | "=&d" (r) : "a" (y), "m" (*y)); | |
f0eb93a8 | 3692 | return result; |
bbf5a54d | 3693 | @} |
cd1a8088 | 3694 | @end smallexample |
c1f7febf | 3695 | |
c85f7c16 | 3696 | You can put multiple assembler instructions together in a single |
8720914b HPN |
3697 | @code{asm} template, separated by the characters normally used in assembly |
3698 | code for the system. A combination that works in most places is a newline | |
3699 | to break the line, plus a tab character to move to the instruction field | |
3700 | (written as @samp{\n\t}). Sometimes semicolons can be used, if the | |
3701 | assembler allows semicolons as a line-breaking character. Note that some | |
3702 | assembler dialects use semicolons to start a comment. | |
3703 | The input operands are guaranteed not to use any of the clobbered | |
c85f7c16 JL |
3704 | registers, and neither will the output operands' addresses, so you can |
3705 | read and write the clobbered registers as many times as you like. Here | |
3706 | is an example of multiple instructions in a template; it assumes the | |
3707 | subroutine @code{_foo} accepts arguments in registers 9 and 10: | |
c1f7febf | 3708 | |
3ab51846 | 3709 | @smallexample |
8720914b | 3710 | asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo" |
c1f7febf RK |
3711 | : /* no outputs */ |
3712 | : "g" (from), "g" (to) | |
3713 | : "r9", "r10"); | |
3ab51846 | 3714 | @end smallexample |
c1f7febf | 3715 | |
f0523f02 | 3716 | Unless an output operand has the @samp{&} constraint modifier, GCC |
c85f7c16 JL |
3717 | may allocate it in the same register as an unrelated input operand, on |
3718 | the assumption the inputs are consumed before the outputs are produced. | |
c1f7febf RK |
3719 | This assumption may be false if the assembler code actually consists of |
3720 | more than one instruction. In such a case, use @samp{&} for each output | |
c85f7c16 | 3721 | operand that may not overlap an input. @xref{Modifiers}. |
c1f7febf | 3722 | |
c85f7c16 JL |
3723 | If you want to test the condition code produced by an assembler |
3724 | instruction, you must include a branch and a label in the @code{asm} | |
3725 | construct, as follows: | |
c1f7febf | 3726 | |
3ab51846 | 3727 | @smallexample |
8720914b | 3728 | asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:" |
c1f7febf RK |
3729 | : "g" (result) |
3730 | : "g" (input)); | |
3ab51846 | 3731 | @end smallexample |
c1f7febf RK |
3732 | |
3733 | @noindent | |
3734 | This assumes your assembler supports local labels, as the GNU assembler | |
3735 | and most Unix assemblers do. | |
3736 | ||
3737 | Speaking of labels, jumps from one @code{asm} to another are not | |
c85f7c16 JL |
3738 | supported. The compiler's optimizers do not know about these jumps, and |
3739 | therefore they cannot take account of them when deciding how to | |
c1f7febf RK |
3740 | optimize. |
3741 | ||
3742 | @cindex macros containing @code{asm} | |
3743 | Usually the most convenient way to use these @code{asm} instructions is to | |
3744 | encapsulate them in macros that look like functions. For example, | |
3745 | ||
3ab51846 | 3746 | @smallexample |
c1f7febf RK |
3747 | #define sin(x) \ |
3748 | (@{ double __value, __arg = (x); \ | |
3749 | asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \ | |
3750 | __value; @}) | |
3ab51846 | 3751 | @end smallexample |
c1f7febf RK |
3752 | |
3753 | @noindent | |
3754 | Here the variable @code{__arg} is used to make sure that the instruction | |
3755 | operates on a proper @code{double} value, and to accept only those | |
3756 | arguments @code{x} which can convert automatically to a @code{double}. | |
3757 | ||
c85f7c16 JL |
3758 | Another way to make sure the instruction operates on the correct data |
3759 | type is to use a cast in the @code{asm}. This is different from using a | |
c1f7febf RK |
3760 | variable @code{__arg} in that it converts more different types. For |
3761 | example, if the desired type were @code{int}, casting the argument to | |
3762 | @code{int} would accept a pointer with no complaint, while assigning the | |
3763 | argument to an @code{int} variable named @code{__arg} would warn about | |
3764 | using a pointer unless the caller explicitly casts it. | |
3765 | ||
f0523f02 | 3766 | If an @code{asm} has output operands, GCC assumes for optimization |
c85f7c16 JL |
3767 | purposes the instruction has no side effects except to change the output |
3768 | operands. This does not mean instructions with a side effect cannot be | |
3769 | used, but you must be careful, because the compiler may eliminate them | |
3770 | if the output operands aren't used, or move them out of loops, or | |
3771 | replace two with one if they constitute a common subexpression. Also, | |
3772 | if your instruction does have a side effect on a variable that otherwise | |
3773 | appears not to change, the old value of the variable may be reused later | |
3774 | if it happens to be found in a register. | |
c1f7febf | 3775 | |
2f59e40e DJ |
3776 | You can prevent an @code{asm} instruction from being deleted |
3777 | by writing the keyword @code{volatile} after | |
c1f7febf RK |
3778 | the @code{asm}. For example: |
3779 | ||
3ab51846 | 3780 | @smallexample |
310668e8 JM |
3781 | #define get_and_set_priority(new) \ |
3782 | (@{ int __old; \ | |
3783 | asm volatile ("get_and_set_priority %0, %1" \ | |
3784 | : "=g" (__old) : "g" (new)); \ | |
c85f7c16 | 3785 | __old; @}) |
3ab51846 | 3786 | @end smallexample |
c1f7febf RK |
3787 | |
3788 | @noindent | |
e71b34aa MM |
3789 | The @code{volatile} keyword indicates that the instruction has |
3790 | important side-effects. GCC will not delete a volatile @code{asm} if | |
3791 | it is reachable. (The instruction can still be deleted if GCC can | |
3792 | prove that control-flow will never reach the location of the | |
f0eb93a8 | 3793 | instruction.) Note that even a volatile @code{asm} instruction |
2f59e40e | 3794 | can be moved relative to other code, including across jump |
f0eb93a8 JM |
3795 | instructions. For example, on many targets there is a system |
3796 | register which can be set to control the rounding mode of | |
2f59e40e DJ |
3797 | floating point operations. You might try |
3798 | setting it with a volatile @code{asm}, like this PowerPC example: | |
e71b34aa | 3799 | |
3ab51846 | 3800 | @smallexample |
2f59e40e DJ |
3801 | asm volatile("mtfsf 255,%0" : : "f" (fpenv)); |
3802 | sum = x + y; | |
3ab51846 | 3803 | @end smallexample |
e71b34aa | 3804 | |
ebb48a4d | 3805 | @noindent |
2f59e40e DJ |
3806 | This will not work reliably, as the compiler may move the addition back |
3807 | before the volatile @code{asm}. To make it work you need to add an | |
3808 | artificial dependency to the @code{asm} referencing a variable in the code | |
3809 | you don't want moved, for example: | |
3810 | ||
3811 | @smallexample | |
3812 | asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv)); | |
3813 | sum = x + y; | |
3814 | @end smallexample | |
3815 | ||
3816 | Similarly, you can't expect a | |
3817 | sequence of volatile @code{asm} instructions to remain perfectly | |
3818 | consecutive. If you want consecutive output, use a single @code{asm}. | |
3819 | Also, GCC will perform some optimizations across a volatile @code{asm} | |
3820 | instruction; GCC does not ``forget everything'' when it encounters | |
3821 | a volatile @code{asm} instruction the way some other compilers do. | |
3822 | ||
3823 | An @code{asm} instruction without any output operands will be treated | |
3824 | identically to a volatile @code{asm} instruction. | |
c1f7febf RK |
3825 | |
3826 | It is a natural idea to look for a way to give access to the condition | |
3827 | code left by the assembler instruction. However, when we attempted to | |
3828 | implement this, we found no way to make it work reliably. The problem | |
3829 | is that output operands might need reloading, which would result in | |
3830 | additional following ``store'' instructions. On most machines, these | |
3831 | instructions would alter the condition code before there was time to | |
3832 | test it. This problem doesn't arise for ordinary ``test'' and | |
3833 | ``compare'' instructions because they don't have any output operands. | |
3834 | ||
eda3fbbe GB |
3835 | For reasons similar to those described above, it is not possible to give |
3836 | an assembler instruction access to the condition code left by previous | |
3837 | instructions. | |
3838 | ||
5490d604 | 3839 | If you are writing a header file that should be includable in ISO C |
c1f7febf RK |
3840 | programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate |
3841 | Keywords}. | |
3842 | ||
ece7fc1c RE |
3843 | @subsection Size of an @code{asm} |
3844 | ||
3845 | Some targets require that GCC track the size of each instruction used in | |
3846 | order to generate correct code. Because the final length of an | |
3847 | @code{asm} is only known by the assembler, GCC must make an estimate as | |
3848 | to how big it will be. The estimate is formed by counting the number of | |
3849 | statements in the pattern of the @code{asm} and multiplying that by the | |
3850 | length of the longest instruction on that processor. Statements in the | |
3851 | @code{asm} are identified by newline characters and whatever statement | |
3852 | separator characters are supported by the assembler; on most processors | |
3853 | this is the `@code{;}' character. | |
3854 | ||
3855 | Normally, GCC's estimate is perfectly adequate to ensure that correct | |
3856 | code is generated, but it is possible to confuse the compiler if you use | |
3857 | pseudo instructions or assembler macros that expand into multiple real | |
3858 | instructions or if you use assembler directives that expand to more | |
3859 | space in the object file than would be needed for a single instruction. | |
3860 | If this happens then the assembler will produce a diagnostic saying that | |
3861 | a label is unreachable. | |
3862 | ||
fe0ce426 JH |
3863 | @subsection i386 floating point asm operands |
3864 | ||
3865 | There are several rules on the usage of stack-like regs in | |
3866 | asm_operands insns. These rules apply only to the operands that are | |
3867 | stack-like regs: | |
3868 | ||
3869 | @enumerate | |
3870 | @item | |
3871 | Given a set of input regs that die in an asm_operands, it is | |
3872 | necessary to know which are implicitly popped by the asm, and | |
3873 | which must be explicitly popped by gcc. | |
3874 | ||
3875 | An input reg that is implicitly popped by the asm must be | |
3876 | explicitly clobbered, unless it is constrained to match an | |
3877 | output operand. | |
3878 | ||
3879 | @item | |
3880 | For any input reg that is implicitly popped by an asm, it is | |
3881 | necessary to know how to adjust the stack to compensate for the pop. | |
3882 | If any non-popped input is closer to the top of the reg-stack than | |
3883 | the implicitly popped reg, it would not be possible to know what the | |
84330467 | 3884 | stack looked like---it's not clear how the rest of the stack ``slides |
fe0ce426 JH |
3885 | up''. |
3886 | ||
3887 | All implicitly popped input regs must be closer to the top of | |
3888 | the reg-stack than any input that is not implicitly popped. | |
3889 | ||
3890 | It is possible that if an input dies in an insn, reload might | |
3891 | use the input reg for an output reload. Consider this example: | |
3892 | ||
3ab51846 | 3893 | @smallexample |
fe0ce426 | 3894 | asm ("foo" : "=t" (a) : "f" (b)); |
3ab51846 | 3895 | @end smallexample |
fe0ce426 JH |
3896 | |
3897 | This asm says that input B is not popped by the asm, and that | |
c771326b | 3898 | the asm pushes a result onto the reg-stack, i.e., the stack is one |
fe0ce426 JH |
3899 | deeper after the asm than it was before. But, it is possible that |
3900 | reload will think that it can use the same reg for both the input and | |
3901 | the output, if input B dies in this insn. | |
3902 | ||
3903 | If any input operand uses the @code{f} constraint, all output reg | |
3904 | constraints must use the @code{&} earlyclobber. | |
3905 | ||
3906 | The asm above would be written as | |
3907 | ||
3ab51846 | 3908 | @smallexample |
fe0ce426 | 3909 | asm ("foo" : "=&t" (a) : "f" (b)); |
3ab51846 | 3910 | @end smallexample |
fe0ce426 JH |
3911 | |
3912 | @item | |
3913 | Some operands need to be in particular places on the stack. All | |
84330467 | 3914 | output operands fall in this category---there is no other way to |
fe0ce426 JH |
3915 | know which regs the outputs appear in unless the user indicates |
3916 | this in the constraints. | |
3917 | ||
3918 | Output operands must specifically indicate which reg an output | |
3919 | appears in after an asm. @code{=f} is not allowed: the operand | |
3920 | constraints must select a class with a single reg. | |
3921 | ||
3922 | @item | |
3923 | Output operands may not be ``inserted'' between existing stack regs. | |
3924 | Since no 387 opcode uses a read/write operand, all output operands | |
3925 | are dead before the asm_operands, and are pushed by the asm_operands. | |
3926 | It makes no sense to push anywhere but the top of the reg-stack. | |
3927 | ||
3928 | Output operands must start at the top of the reg-stack: output | |
3929 | operands may not ``skip'' a reg. | |
3930 | ||
3931 | @item | |
3932 | Some asm statements may need extra stack space for internal | |
3933 | calculations. This can be guaranteed by clobbering stack registers | |
3934 | unrelated to the inputs and outputs. | |
3935 | ||
3936 | @end enumerate | |
3937 | ||
3938 | Here are a couple of reasonable asms to want to write. This asm | |
3939 | takes one input, which is internally popped, and produces two outputs. | |
3940 | ||
3ab51846 | 3941 | @smallexample |
fe0ce426 | 3942 | asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp)); |
3ab51846 | 3943 | @end smallexample |
fe0ce426 JH |
3944 | |
3945 | This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode, | |
3946 | and replaces them with one output. The user must code the @code{st(1)} | |
3947 | clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs. | |
3948 | ||
3ab51846 | 3949 | @smallexample |
fe0ce426 | 3950 | asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)"); |
3ab51846 | 3951 | @end smallexample |
fe0ce426 | 3952 | |
c1f7febf | 3953 | @include md.texi |
c1f7febf RK |
3954 | |
3955 | @node Asm Labels | |
3956 | @section Controlling Names Used in Assembler Code | |
3957 | @cindex assembler names for identifiers | |
3958 | @cindex names used in assembler code | |
3959 | @cindex identifiers, names in assembler code | |
3960 | ||
3961 | You can specify the name to be used in the assembler code for a C | |
3962 | function or variable by writing the @code{asm} (or @code{__asm__}) | |
3963 | keyword after the declarator as follows: | |
3964 | ||
3ab51846 | 3965 | @smallexample |
c1f7febf | 3966 | int foo asm ("myfoo") = 2; |
3ab51846 | 3967 | @end smallexample |
c1f7febf RK |
3968 | |
3969 | @noindent | |
3970 | This specifies that the name to be used for the variable @code{foo} in | |
3971 | the assembler code should be @samp{myfoo} rather than the usual | |
3972 | @samp{_foo}. | |
3973 | ||
3974 | On systems where an underscore is normally prepended to the name of a C | |
3975 | function or variable, this feature allows you to define names for the | |
3976 | linker that do not start with an underscore. | |
3977 | ||
0adc3c19 MM |
3978 | It does not make sense to use this feature with a non-static local |
3979 | variable since such variables do not have assembler names. If you are | |
3980 | trying to put the variable in a particular register, see @ref{Explicit | |
3981 | Reg Vars}. GCC presently accepts such code with a warning, but will | |
3982 | probably be changed to issue an error, rather than a warning, in the | |
3983 | future. | |
3984 | ||
c1f7febf RK |
3985 | You cannot use @code{asm} in this way in a function @emph{definition}; but |
3986 | you can get the same effect by writing a declaration for the function | |
3987 | before its definition and putting @code{asm} there, like this: | |
3988 | ||
3ab51846 | 3989 | @smallexample |
c1f7febf RK |
3990 | extern func () asm ("FUNC"); |
3991 | ||
3992 | func (x, y) | |
3993 | int x, y; | |
0d893a63 | 3994 | /* @r{@dots{}} */ |
3ab51846 | 3995 | @end smallexample |
c1f7febf RK |
3996 | |
3997 | It is up to you to make sure that the assembler names you choose do not | |
3998 | conflict with any other assembler symbols. Also, you must not use a | |
f0523f02 JM |
3999 | register name; that would produce completely invalid assembler code. GCC |
4000 | does not as yet have the ability to store static variables in registers. | |
c1f7febf RK |
4001 | Perhaps that will be added. |
4002 | ||
4003 | @node Explicit Reg Vars | |
4004 | @section Variables in Specified Registers | |
4005 | @cindex explicit register variables | |
4006 | @cindex variables in specified registers | |
4007 | @cindex specified registers | |
4008 | @cindex registers, global allocation | |
4009 | ||
4010 | GNU C allows you to put a few global variables into specified hardware | |
4011 | registers. You can also specify the register in which an ordinary | |
4012 | register variable should be allocated. | |
4013 | ||
4014 | @itemize @bullet | |
4015 | @item | |
4016 | Global register variables reserve registers throughout the program. | |
4017 | This may be useful in programs such as programming language | |
4018 | interpreters which have a couple of global variables that are accessed | |
4019 | very often. | |
4020 | ||
4021 | @item | |
4022 | Local register variables in specific registers do not reserve the | |
805c33df HPN |
4023 | registers, except at the point where they are used as input or output |
4024 | operands in an @code{asm} statement and the @code{asm} statement itself is | |
4025 | not deleted. The compiler's data flow analysis is capable of determining | |
c1f7febf | 4026 | where the specified registers contain live values, and where they are |
8d344fbc | 4027 | available for other uses. Stores into local register variables may be deleted |
0deaf590 JL |
4028 | when they appear to be dead according to dataflow analysis. References |
4029 | to local register variables may be deleted or moved or simplified. | |
c1f7febf RK |
4030 | |
4031 | These local variables are sometimes convenient for use with the extended | |
4032 | @code{asm} feature (@pxref{Extended Asm}), if you want to write one | |
4033 | output of the assembler instruction directly into a particular register. | |
4034 | (This will work provided the register you specify fits the constraints | |
4035 | specified for that operand in the @code{asm}.) | |
4036 | @end itemize | |
4037 | ||
4038 | @menu | |
4039 | * Global Reg Vars:: | |
4040 | * Local Reg Vars:: | |
4041 | @end menu | |
4042 | ||
4043 | @node Global Reg Vars | |
4044 | @subsection Defining Global Register Variables | |
4045 | @cindex global register variables | |
4046 | @cindex registers, global variables in | |
4047 | ||
4048 | You can define a global register variable in GNU C like this: | |
4049 | ||
3ab51846 | 4050 | @smallexample |
c1f7febf | 4051 | register int *foo asm ("a5"); |
3ab51846 | 4052 | @end smallexample |
c1f7febf RK |
4053 | |
4054 | @noindent | |
4055 | Here @code{a5} is the name of the register which should be used. Choose a | |
4056 | register which is normally saved and restored by function calls on your | |
4057 | machine, so that library routines will not clobber it. | |
4058 | ||
4059 | Naturally the register name is cpu-dependent, so you would need to | |
4060 | conditionalize your program according to cpu type. The register | |
4061 | @code{a5} would be a good choice on a 68000 for a variable of pointer | |
4062 | type. On machines with register windows, be sure to choose a ``global'' | |
4063 | register that is not affected magically by the function call mechanism. | |
4064 | ||
4065 | In addition, operating systems on one type of cpu may differ in how they | |
4066 | name the registers; then you would need additional conditionals. For | |
4067 | example, some 68000 operating systems call this register @code{%a5}. | |
4068 | ||
4069 | Eventually there may be a way of asking the compiler to choose a register | |
4070 | automatically, but first we need to figure out how it should choose and | |
4071 | how to enable you to guide the choice. No solution is evident. | |
4072 | ||
4073 | Defining a global register variable in a certain register reserves that | |
4074 | register entirely for this use, at least within the current compilation. | |
4075 | The register will not be allocated for any other purpose in the functions | |
4076 | in the current compilation. The register will not be saved and restored by | |
4077 | these functions. Stores into this register are never deleted even if they | |
4078 | would appear to be dead, but references may be deleted or moved or | |
4079 | simplified. | |
4080 | ||
4081 | It is not safe to access the global register variables from signal | |
4082 | handlers, or from more than one thread of control, because the system | |
4083 | library routines may temporarily use the register for other things (unless | |
4084 | you recompile them specially for the task at hand). | |
4085 | ||
4086 | @cindex @code{qsort}, and global register variables | |
4087 | It is not safe for one function that uses a global register variable to | |
4088 | call another such function @code{foo} by way of a third function | |
e979f9e8 | 4089 | @code{lose} that was compiled without knowledge of this variable (i.e.@: in a |
c1f7febf RK |
4090 | different source file in which the variable wasn't declared). This is |
4091 | because @code{lose} might save the register and put some other value there. | |
4092 | For example, you can't expect a global register variable to be available in | |
4093 | the comparison-function that you pass to @code{qsort}, since @code{qsort} | |
4094 | might have put something else in that register. (If you are prepared to | |
4095 | recompile @code{qsort} with the same global register variable, you can | |
4096 | solve this problem.) | |
4097 | ||
4098 | If you want to recompile @code{qsort} or other source files which do not | |
4099 | actually use your global register variable, so that they will not use that | |
4100 | register for any other purpose, then it suffices to specify the compiler | |
84330467 | 4101 | option @option{-ffixed-@var{reg}}. You need not actually add a global |
c1f7febf RK |
4102 | register declaration to their source code. |
4103 | ||
4104 | A function which can alter the value of a global register variable cannot | |
4105 | safely be called from a function compiled without this variable, because it | |
4106 | could clobber the value the caller expects to find there on return. | |
4107 | Therefore, the function which is the entry point into the part of the | |
4108 | program that uses the global register variable must explicitly save and | |
4109 | restore the value which belongs to its caller. | |
4110 | ||
4111 | @cindex register variable after @code{longjmp} | |
4112 | @cindex global register after @code{longjmp} | |
4113 | @cindex value after @code{longjmp} | |
4114 | @findex longjmp | |
4115 | @findex setjmp | |
4116 | On most machines, @code{longjmp} will restore to each global register | |
4117 | variable the value it had at the time of the @code{setjmp}. On some | |
4118 | machines, however, @code{longjmp} will not change the value of global | |
4119 | register variables. To be portable, the function that called @code{setjmp} | |
4120 | should make other arrangements to save the values of the global register | |
4121 | variables, and to restore them in a @code{longjmp}. This way, the same | |
4122 | thing will happen regardless of what @code{longjmp} does. | |
4123 | ||
4124 | All global register variable declarations must precede all function | |
4125 | definitions. If such a declaration could appear after function | |
4126 | definitions, the declaration would be too late to prevent the register from | |
4127 | being used for other purposes in the preceding functions. | |
4128 | ||
4129 | Global register variables may not have initial values, because an | |
4130 | executable file has no means to supply initial contents for a register. | |
4131 | ||
981f6289 | 4132 | On the SPARC, there are reports that g3 @dots{} g7 are suitable |
c1f7febf RK |
4133 | registers, but certain library functions, such as @code{getwd}, as well |
4134 | as the subroutines for division and remainder, modify g3 and g4. g1 and | |
4135 | g2 are local temporaries. | |
4136 | ||
4137 | On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7. | |
4138 | Of course, it will not do to use more than a few of those. | |
4139 | ||
4140 | @node Local Reg Vars | |
4141 | @subsection Specifying Registers for Local Variables | |
4142 | @cindex local variables, specifying registers | |
4143 | @cindex specifying registers for local variables | |
4144 | @cindex registers for local variables | |
4145 | ||
4146 | You can define a local register variable with a specified register | |
4147 | like this: | |
4148 | ||
3ab51846 | 4149 | @smallexample |
c1f7febf | 4150 | register int *foo asm ("a5"); |
3ab51846 | 4151 | @end smallexample |
c1f7febf RK |
4152 | |
4153 | @noindent | |
4154 | Here @code{a5} is the name of the register which should be used. Note | |
4155 | that this is the same syntax used for defining global register | |
4156 | variables, but for a local variable it would appear within a function. | |
4157 | ||
4158 | Naturally the register name is cpu-dependent, but this is not a | |
4159 | problem, since specific registers are most often useful with explicit | |
4160 | assembler instructions (@pxref{Extended Asm}). Both of these things | |
4161 | generally require that you conditionalize your program according to | |
4162 | cpu type. | |
4163 | ||
4164 | In addition, operating systems on one type of cpu may differ in how they | |
4165 | name the registers; then you would need additional conditionals. For | |
4166 | example, some 68000 operating systems call this register @code{%a5}. | |
4167 | ||
c1f7febf RK |
4168 | Defining such a register variable does not reserve the register; it |
4169 | remains available for other uses in places where flow control determines | |
d754127f | 4170 | the variable's value is not live. |
e5e809f4 | 4171 | |
f0523f02 | 4172 | This option does not guarantee that GCC will generate code that has |
e5e809f4 | 4173 | this variable in the register you specify at all times. You may not |
805c33df HPN |
4174 | code an explicit reference to this register in the @emph{assembler |
4175 | instruction template} part of an @code{asm} statement and assume it will | |
4176 | always refer to this variable. However, using the variable as an | |
4177 | @code{asm} @emph{operand} guarantees that the specified register is used | |
4178 | for the operand. | |
c1f7febf | 4179 | |
8d344fbc | 4180 | Stores into local register variables may be deleted when they appear to be dead |
0deaf590 JL |
4181 | according to dataflow analysis. References to local register variables may |
4182 | be deleted or moved or simplified. | |
4183 | ||
b55d5746 HPN |
4184 | As for global register variables, it's recommended that you choose a |
4185 | register which is normally saved and restored by function calls on | |
4186 | your machine, so that library routines will not clobber it. A common | |
4187 | pitfall is to initialize multiple call-clobbered registers with | |
4188 | arbitrary expressions, where a function call or library call for an | |
4189 | arithmetic operator will overwrite a register value from a previous | |
4190 | assignment, for example @code{r0} below: | |
4191 | @smallexample | |
4192 | register int *p1 asm ("r0") = @dots{}; | |
4193 | register int *p2 asm ("r1") = @dots{}; | |
4194 | @end smallexample | |
4195 | In those cases, a solution is to use a temporary variable for | |
4196 | each arbitrary expression. @xref{Example of asm with clobbered asm reg}. | |
4197 | ||
c1f7febf RK |
4198 | @node Alternate Keywords |
4199 | @section Alternate Keywords | |
4200 | @cindex alternate keywords | |
4201 | @cindex keywords, alternate | |
4202 | ||
5490d604 | 4203 | @option{-ansi} and the various @option{-std} options disable certain |
f458d1d5 ZW |
4204 | keywords. This causes trouble when you want to use GNU C extensions, or |
4205 | a general-purpose header file that should be usable by all programs, | |
4206 | including ISO C programs. The keywords @code{asm}, @code{typeof} and | |
4207 | @code{inline} are not available in programs compiled with | |
4208 | @option{-ansi} or @option{-std} (although @code{inline} can be used in a | |
4209 | program compiled with @option{-std=c99}). The ISO C99 keyword | |
5490d604 JM |
4210 | @code{restrict} is only available when @option{-std=gnu99} (which will |
4211 | eventually be the default) or @option{-std=c99} (or the equivalent | |
bd819a4a | 4212 | @option{-std=iso9899:1999}) is used. |
c1f7febf RK |
4213 | |
4214 | The way to solve these problems is to put @samp{__} at the beginning and | |
4215 | end of each problematical keyword. For example, use @code{__asm__} | |
f458d1d5 | 4216 | instead of @code{asm}, and @code{__inline__} instead of @code{inline}. |
c1f7febf RK |
4217 | |
4218 | Other C compilers won't accept these alternative keywords; if you want to | |
4219 | compile with another compiler, you can define the alternate keywords as | |
4220 | macros to replace them with the customary keywords. It looks like this: | |
4221 | ||
3ab51846 | 4222 | @smallexample |
c1f7febf RK |
4223 | #ifndef __GNUC__ |
4224 | #define __asm__ asm | |
4225 | #endif | |
3ab51846 | 4226 | @end smallexample |
c1f7febf | 4227 | |
6e6b0525 | 4228 | @findex __extension__ |
84330467 JM |
4229 | @opindex pedantic |
4230 | @option{-pedantic} and other options cause warnings for many GNU C extensions. | |
dbe519e0 | 4231 | You can |
c1f7febf RK |
4232 | prevent such warnings within one expression by writing |
4233 | @code{__extension__} before the expression. @code{__extension__} has no | |
4234 | effect aside from this. | |
4235 | ||
4236 | @node Incomplete Enums | |
4237 | @section Incomplete @code{enum} Types | |
4238 | ||
4239 | You can define an @code{enum} tag without specifying its possible values. | |
4240 | This results in an incomplete type, much like what you get if you write | |
4241 | @code{struct foo} without describing the elements. A later declaration | |
4242 | which does specify the possible values completes the type. | |
4243 | ||
4244 | You can't allocate variables or storage using the type while it is | |
4245 | incomplete. However, you can work with pointers to that type. | |
4246 | ||
4247 | This extension may not be very useful, but it makes the handling of | |
4248 | @code{enum} more consistent with the way @code{struct} and @code{union} | |
4249 | are handled. | |
4250 | ||
4251 | This extension is not supported by GNU C++. | |
4252 | ||
4253 | @node Function Names | |
4254 | @section Function Names as Strings | |
e6cc3a24 | 4255 | @cindex @code{__func__} identifier |
4b404517 JM |
4256 | @cindex @code{__FUNCTION__} identifier |
4257 | @cindex @code{__PRETTY_FUNCTION__} identifier | |
c1f7febf | 4258 | |
e6cc3a24 ZW |
4259 | GCC provides three magic variables which hold the name of the current |
4260 | function, as a string. The first of these is @code{__func__}, which | |
4261 | is part of the C99 standard: | |
4262 | ||
4263 | @display | |
4264 | The identifier @code{__func__} is implicitly declared by the translator | |
4265 | as if, immediately following the opening brace of each function | |
4266 | definition, the declaration | |
4267 | ||
4268 | @smallexample | |
4269 | static const char __func__[] = "function-name"; | |
4270 | @end smallexample | |
c1f7febf | 4271 | |
e6cc3a24 ZW |
4272 | appeared, where function-name is the name of the lexically-enclosing |
4273 | function. This name is the unadorned name of the function. | |
4274 | @end display | |
4275 | ||
4276 | @code{__FUNCTION__} is another name for @code{__func__}. Older | |
4277 | versions of GCC recognize only this name. However, it is not | |
4278 | standardized. For maximum portability, we recommend you use | |
4279 | @code{__func__}, but provide a fallback definition with the | |
4280 | preprocessor: | |
4281 | ||
4282 | @smallexample | |
4283 | #if __STDC_VERSION__ < 199901L | |
4284 | # if __GNUC__ >= 2 | |
4285 | # define __func__ __FUNCTION__ | |
4286 | # else | |
4287 | # define __func__ "<unknown>" | |
4288 | # endif | |
4289 | #endif | |
4290 | @end smallexample | |
4291 | ||
4292 | In C, @code{__PRETTY_FUNCTION__} is yet another name for | |
4293 | @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains | |
4294 | the type signature of the function as well as its bare name. For | |
4295 | example, this program: | |
c1f7febf RK |
4296 | |
4297 | @smallexample | |
4298 | extern "C" @{ | |
4299 | extern int printf (char *, ...); | |
4300 | @} | |
4301 | ||
4302 | class a @{ | |
4303 | public: | |
a721a601 | 4304 | void sub (int i) |
c1f7febf RK |
4305 | @{ |
4306 | printf ("__FUNCTION__ = %s\n", __FUNCTION__); | |
4307 | printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__); | |
4308 | @} | |
4309 | @}; | |
4310 | ||
4311 | int | |
4312 | main (void) | |
4313 | @{ | |
4314 | a ax; | |
4315 | ax.sub (0); | |
4316 | return 0; | |
4317 | @} | |
4318 | @end smallexample | |
4319 | ||
4320 | @noindent | |
4321 | gives this output: | |
4322 | ||
4323 | @smallexample | |
4324 | __FUNCTION__ = sub | |
e6cc3a24 | 4325 | __PRETTY_FUNCTION__ = void a::sub(int) |
22acfb79 NM |
4326 | @end smallexample |
4327 | ||
e6cc3a24 ZW |
4328 | These identifiers are not preprocessor macros. In GCC 3.3 and |
4329 | earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__} | |
4330 | were treated as string literals; they could be used to initialize | |
4331 | @code{char} arrays, and they could be concatenated with other string | |
4332 | literals. GCC 3.4 and later treat them as variables, like | |
4333 | @code{__func__}. In C++, @code{__FUNCTION__} and | |
4334 | @code{__PRETTY_FUNCTION__} have always been variables. | |
22acfb79 | 4335 | |
c1f7febf RK |
4336 | @node Return Address |
4337 | @section Getting the Return or Frame Address of a Function | |
4338 | ||
4339 | These functions may be used to get information about the callers of a | |
4340 | function. | |
4341 | ||
84330467 | 4342 | @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level}) |
c1f7febf RK |
4343 | This function returns the return address of the current function, or of |
4344 | one of its callers. The @var{level} argument is number of frames to | |
4345 | scan up the call stack. A value of @code{0} yields the return address | |
4346 | of the current function, a value of @code{1} yields the return address | |
8a36672b | 4347 | of the caller of the current function, and so forth. When inlining |
95b1627e EC |
4348 | the expected behavior is that the function will return the address of |
4349 | the function that will be returned to. To work around this behavior use | |
4350 | the @code{noinline} function attribute. | |
c1f7febf RK |
4351 | |
4352 | The @var{level} argument must be a constant integer. | |
4353 | ||
4354 | On some machines it may be impossible to determine the return address of | |
4355 | any function other than the current one; in such cases, or when the top | |
dd96fbc5 | 4356 | of the stack has been reached, this function will return @code{0} or a |
8a36672b | 4357 | random value. In addition, @code{__builtin_frame_address} may be used |
dd96fbc5 | 4358 | to determine if the top of the stack has been reached. |
c1f7febf | 4359 | |
df2a54e9 | 4360 | This function should only be used with a nonzero argument for debugging |
c1f7febf | 4361 | purposes. |
84330467 | 4362 | @end deftypefn |
c1f7febf | 4363 | |
84330467 | 4364 | @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level}) |
c1f7febf RK |
4365 | This function is similar to @code{__builtin_return_address}, but it |
4366 | returns the address of the function frame rather than the return address | |
4367 | of the function. Calling @code{__builtin_frame_address} with a value of | |
4368 | @code{0} yields the frame address of the current function, a value of | |
4369 | @code{1} yields the frame address of the caller of the current function, | |
4370 | and so forth. | |
4371 | ||
4372 | The frame is the area on the stack which holds local variables and saved | |
4373 | registers. The frame address is normally the address of the first word | |
4374 | pushed on to the stack by the function. However, the exact definition | |
4375 | depends upon the processor and the calling convention. If the processor | |
4376 | has a dedicated frame pointer register, and the function has a frame, | |
4377 | then @code{__builtin_frame_address} will return the value of the frame | |
4378 | pointer register. | |
4379 | ||
dd96fbc5 L |
4380 | On some machines it may be impossible to determine the frame address of |
4381 | any function other than the current one; in such cases, or when the top | |
4382 | of the stack has been reached, this function will return @code{0} if | |
4383 | the first frame pointer is properly initialized by the startup code. | |
4384 | ||
df2a54e9 | 4385 | This function should only be used with a nonzero argument for debugging |
dd96fbc5 | 4386 | purposes. |
84330467 | 4387 | @end deftypefn |
c1f7febf | 4388 | |
1255c85c BS |
4389 | @node Vector Extensions |
4390 | @section Using vector instructions through built-in functions | |
4391 | ||
4392 | On some targets, the instruction set contains SIMD vector instructions that | |
4393 | operate on multiple values contained in one large register at the same time. | |
4394 | For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used | |
4395 | this way. | |
4396 | ||
4397 | The first step in using these extensions is to provide the necessary data | |
4398 | types. This should be done using an appropriate @code{typedef}: | |
4399 | ||
3ab51846 | 4400 | @smallexample |
4a5eab38 | 4401 | typedef int v4si __attribute__ ((vector_size (16))); |
3ab51846 | 4402 | @end smallexample |
1255c85c | 4403 | |
4a5eab38 PB |
4404 | The @code{int} type specifies the base type, while the attribute specifies |
4405 | the vector size for the variable, measured in bytes. For example, the | |
4406 | declaration above causes the compiler to set the mode for the @code{v4si} | |
4407 | type to be 16 bytes wide and divided into @code{int} sized units. For | |
4408 | a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the | |
4409 | corresponding mode of @code{foo} will be @acronym{V4SI}. | |
1255c85c | 4410 | |
4a5eab38 PB |
4411 | The @code{vector_size} attribute is only applicable to integral and |
4412 | float scalars, although arrays, pointers, and function return values | |
4413 | are allowed in conjunction with this construct. | |
4414 | ||
4415 | All the basic integer types can be used as base types, both as signed | |
4416 | and as unsigned: @code{char}, @code{short}, @code{int}, @code{long}, | |
4417 | @code{long long}. In addition, @code{float} and @code{double} can be | |
4418 | used to build floating-point vector types. | |
1255c85c | 4419 | |
cb2a532e | 4420 | Specifying a combination that is not valid for the current architecture |
2dd76960 | 4421 | will cause GCC to synthesize the instructions using a narrower mode. |
cb2a532e | 4422 | For example, if you specify a variable of type @code{V4SI} and your |
2dd76960 | 4423 | architecture does not allow for this specific SIMD type, GCC will |
cb2a532e AH |
4424 | produce code that uses 4 @code{SIs}. |
4425 | ||
4426 | The types defined in this manner can be used with a subset of normal C | |
2dd76960 | 4427 | operations. Currently, GCC will allow using the following operators |
3a3e1600 | 4428 | on these types: @code{+, -, *, /, unary minus, ^, |, &, ~}@. |
cb2a532e AH |
4429 | |
4430 | The operations behave like C++ @code{valarrays}. Addition is defined as | |
4431 | the addition of the corresponding elements of the operands. For | |
4432 | example, in the code below, each of the 4 elements in @var{a} will be | |
4433 | added to the corresponding 4 elements in @var{b} and the resulting | |
4434 | vector will be stored in @var{c}. | |
4435 | ||
3ab51846 | 4436 | @smallexample |
4a5eab38 | 4437 | typedef int v4si __attribute__ ((vector_size (16))); |
cb2a532e AH |
4438 | |
4439 | v4si a, b, c; | |
4440 | ||
4441 | c = a + b; | |
3ab51846 | 4442 | @end smallexample |
cb2a532e | 4443 | |
3a3e1600 GK |
4444 | Subtraction, multiplication, division, and the logical operations |
4445 | operate in a similar manner. Likewise, the result of using the unary | |
4446 | minus or complement operators on a vector type is a vector whose | |
4447 | elements are the negative or complemented values of the corresponding | |
cb2a532e AH |
4448 | elements in the operand. |
4449 | ||
4450 | You can declare variables and use them in function calls and returns, as | |
4451 | well as in assignments and some casts. You can specify a vector type as | |
4452 | a return type for a function. Vector types can also be used as function | |
4453 | arguments. It is possible to cast from one vector type to another, | |
4454 | provided they are of the same size (in fact, you can also cast vectors | |
4455 | to and from other datatypes of the same size). | |
4456 | ||
4457 | You cannot operate between vectors of different lengths or different | |
90a21764 | 4458 | signedness without a cast. |
cb2a532e AH |
4459 | |
4460 | A port that supports hardware vector operations, usually provides a set | |
4461 | of built-in functions that can be used to operate on vectors. For | |
4462 | example, a function to add two vectors and multiply the result by a | |
4463 | third could look like this: | |
1255c85c | 4464 | |
3ab51846 | 4465 | @smallexample |
1255c85c BS |
4466 | v4si f (v4si a, v4si b, v4si c) |
4467 | @{ | |
4468 | v4si tmp = __builtin_addv4si (a, b); | |
4469 | return __builtin_mulv4si (tmp, c); | |
4470 | @} | |
4471 | ||
3ab51846 | 4472 | @end smallexample |
1255c85c | 4473 | |
7a3ea201 RH |
4474 | @node Offsetof |
4475 | @section Offsetof | |
4476 | @findex __builtin_offsetof | |
4477 | ||
4478 | GCC implements for both C and C++ a syntactic extension to implement | |
4479 | the @code{offsetof} macro. | |
4480 | ||
4481 | @smallexample | |
4482 | primary: | |
4483 | "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")" | |
4484 | ||
4485 | offsetof_member_designator: | |
4486 | @code{identifier} | |
4487 | | offsetof_member_designator "." @code{identifier} | |
4488 | | offsetof_member_designator "[" @code{expr} "]" | |
4489 | @end smallexample | |
4490 | ||
4491 | This extension is sufficient such that | |
4492 | ||
4493 | @smallexample | |
4494 | #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member}) | |
4495 | @end smallexample | |
4496 | ||
4497 | is a suitable definition of the @code{offsetof} macro. In C++, @var{type} | |
4498 | may be dependent. In either case, @var{member} may consist of a single | |
4499 | identifier, or a sequence of member accesses and array references. | |
4500 | ||
185ebd6c | 4501 | @node Other Builtins |
f0523f02 | 4502 | @section Other built-in functions provided by GCC |
c771326b | 4503 | @cindex built-in functions |
01702459 JM |
4504 | @findex __builtin_isgreater |
4505 | @findex __builtin_isgreaterequal | |
4506 | @findex __builtin_isless | |
4507 | @findex __builtin_islessequal | |
4508 | @findex __builtin_islessgreater | |
4509 | @findex __builtin_isunordered | |
17684d46 RG |
4510 | @findex __builtin_powi |
4511 | @findex __builtin_powif | |
4512 | @findex __builtin_powil | |
98ff7c4d KG |
4513 | @findex _Exit |
4514 | @findex _exit | |
01702459 JM |
4515 | @findex abort |
4516 | @findex abs | |
98ff7c4d KG |
4517 | @findex acos |
4518 | @findex acosf | |
4519 | @findex acosh | |
4520 | @findex acoshf | |
4521 | @findex acoshl | |
4522 | @findex acosl | |
01702459 | 4523 | @findex alloca |
98ff7c4d KG |
4524 | @findex asin |
4525 | @findex asinf | |
4526 | @findex asinh | |
4527 | @findex asinhf | |
4528 | @findex asinhl | |
4529 | @findex asinl | |
29f523be | 4530 | @findex atan |
46847aa6 RS |
4531 | @findex atan2 |
4532 | @findex atan2f | |
4533 | @findex atan2l | |
29f523be | 4534 | @findex atanf |
98ff7c4d KG |
4535 | @findex atanh |
4536 | @findex atanhf | |
4537 | @findex atanhl | |
29f523be | 4538 | @findex atanl |
01702459 JM |
4539 | @findex bcmp |
4540 | @findex bzero | |
075ec276 RS |
4541 | @findex cabs |
4542 | @findex cabsf | |
4543 | @findex cabsl | |
11bf0eb0 KG |
4544 | @findex cacos |
4545 | @findex cacosf | |
4546 | @findex cacosh | |
4547 | @findex cacoshf | |
4548 | @findex cacoshl | |
4549 | @findex cacosl | |
1331d16f | 4550 | @findex calloc |
11bf0eb0 KG |
4551 | @findex carg |
4552 | @findex cargf | |
4553 | @findex cargl | |
4554 | @findex casin | |
4555 | @findex casinf | |
4556 | @findex casinh | |
4557 | @findex casinhf | |
4558 | @findex casinhl | |
4559 | @findex casinl | |
4560 | @findex catan | |
4561 | @findex catanf | |
4562 | @findex catanh | |
4563 | @findex catanhf | |
4564 | @findex catanhl | |
4565 | @findex catanl | |
98ff7c4d KG |
4566 | @findex cbrt |
4567 | @findex cbrtf | |
4568 | @findex cbrtl | |
11bf0eb0 KG |
4569 | @findex ccos |
4570 | @findex ccosf | |
4571 | @findex ccosh | |
4572 | @findex ccoshf | |
4573 | @findex ccoshl | |
4574 | @findex ccosl | |
b052d8ee RS |
4575 | @findex ceil |
4576 | @findex ceilf | |
4577 | @findex ceill | |
11bf0eb0 KG |
4578 | @findex cexp |
4579 | @findex cexpf | |
4580 | @findex cexpl | |
341e3d11 JM |
4581 | @findex cimag |
4582 | @findex cimagf | |
4583 | @findex cimagl | |
4584 | @findex conj | |
4585 | @findex conjf | |
4586 | @findex conjl | |
98ff7c4d KG |
4587 | @findex copysign |
4588 | @findex copysignf | |
4589 | @findex copysignl | |
01702459 JM |
4590 | @findex cos |
4591 | @findex cosf | |
98ff7c4d KG |
4592 | @findex cosh |
4593 | @findex coshf | |
4594 | @findex coshl | |
01702459 | 4595 | @findex cosl |
11bf0eb0 KG |
4596 | @findex cpow |
4597 | @findex cpowf | |
4598 | @findex cpowl | |
4599 | @findex cproj | |
4600 | @findex cprojf | |
4601 | @findex cprojl | |
341e3d11 JM |
4602 | @findex creal |
4603 | @findex crealf | |
4604 | @findex creall | |
11bf0eb0 KG |
4605 | @findex csin |
4606 | @findex csinf | |
4607 | @findex csinh | |
4608 | @findex csinhf | |
4609 | @findex csinhl | |
4610 | @findex csinl | |
4611 | @findex csqrt | |
4612 | @findex csqrtf | |
4613 | @findex csqrtl | |
4614 | @findex ctan | |
4615 | @findex ctanf | |
4616 | @findex ctanh | |
4617 | @findex ctanhf | |
4618 | @findex ctanhl | |
4619 | @findex ctanl | |
178b2b9f RS |
4620 | @findex dcgettext |
4621 | @findex dgettext | |
98ff7c4d KG |
4622 | @findex drem |
4623 | @findex dremf | |
4624 | @findex dreml | |
488f17e1 KG |
4625 | @findex erf |
4626 | @findex erfc | |
4627 | @findex erfcf | |
4628 | @findex erfcl | |
4629 | @findex erff | |
4630 | @findex erfl | |
01702459 | 4631 | @findex exit |
e7b489c8 | 4632 | @findex exp |
98ff7c4d KG |
4633 | @findex exp10 |
4634 | @findex exp10f | |
4635 | @findex exp10l | |
4636 | @findex exp2 | |
4637 | @findex exp2f | |
4638 | @findex exp2l | |
e7b489c8 RS |
4639 | @findex expf |
4640 | @findex expl | |
98ff7c4d KG |
4641 | @findex expm1 |
4642 | @findex expm1f | |
4643 | @findex expm1l | |
01702459 JM |
4644 | @findex fabs |
4645 | @findex fabsf | |
4646 | @findex fabsl | |
98ff7c4d KG |
4647 | @findex fdim |
4648 | @findex fdimf | |
4649 | @findex fdiml | |
01702459 | 4650 | @findex ffs |
b052d8ee RS |
4651 | @findex floor |
4652 | @findex floorf | |
4653 | @findex floorl | |
98ff7c4d KG |
4654 | @findex fma |
4655 | @findex fmaf | |
4656 | @findex fmal | |
4657 | @findex fmax | |
4658 | @findex fmaxf | |
4659 | @findex fmaxl | |
4660 | @findex fmin | |
4661 | @findex fminf | |
4662 | @findex fminl | |
b052d8ee RS |
4663 | @findex fmod |
4664 | @findex fmodf | |
4665 | @findex fmodl | |
18f988a0 | 4666 | @findex fprintf |
b4c984fb | 4667 | @findex fprintf_unlocked |
01702459 | 4668 | @findex fputs |
b4c984fb | 4669 | @findex fputs_unlocked |
a2a919aa KG |
4670 | @findex frexp |
4671 | @findex frexpf | |
4672 | @findex frexpl | |
178b2b9f | 4673 | @findex fscanf |
488f17e1 KG |
4674 | @findex gamma |
4675 | @findex gammaf | |
4676 | @findex gammal | |
178b2b9f | 4677 | @findex gettext |
98ff7c4d KG |
4678 | @findex hypot |
4679 | @findex hypotf | |
4680 | @findex hypotl | |
4681 | @findex ilogb | |
4682 | @findex ilogbf | |
4683 | @findex ilogbl | |
e78f4a97 | 4684 | @findex imaxabs |
c7b6c6cd | 4685 | @findex index |
740e5b6f KG |
4686 | @findex isalnum |
4687 | @findex isalpha | |
4688 | @findex isascii | |
4689 | @findex isblank | |
4690 | @findex iscntrl | |
4691 | @findex isdigit | |
4692 | @findex isgraph | |
4693 | @findex islower | |
4694 | @findex isprint | |
4695 | @findex ispunct | |
4696 | @findex isspace | |
4697 | @findex isupper | |
ca4944e1 KG |
4698 | @findex iswalnum |
4699 | @findex iswalpha | |
4700 | @findex iswblank | |
4701 | @findex iswcntrl | |
4702 | @findex iswdigit | |
4703 | @findex iswgraph | |
4704 | @findex iswlower | |
4705 | @findex iswprint | |
4706 | @findex iswpunct | |
4707 | @findex iswspace | |
4708 | @findex iswupper | |
4709 | @findex iswxdigit | |
740e5b6f | 4710 | @findex isxdigit |
488f17e1 KG |
4711 | @findex j0 |
4712 | @findex j0f | |
4713 | @findex j0l | |
4714 | @findex j1 | |
4715 | @findex j1f | |
4716 | @findex j1l | |
4717 | @findex jn | |
4718 | @findex jnf | |
4719 | @findex jnl | |
01702459 | 4720 | @findex labs |
98ff7c4d KG |
4721 | @findex ldexp |
4722 | @findex ldexpf | |
4723 | @findex ldexpl | |
488f17e1 KG |
4724 | @findex lgamma |
4725 | @findex lgammaf | |
4726 | @findex lgammal | |
01702459 | 4727 | @findex llabs |
98ff7c4d KG |
4728 | @findex llrint |
4729 | @findex llrintf | |
4730 | @findex llrintl | |
4731 | @findex llround | |
4732 | @findex llroundf | |
4733 | @findex llroundl | |
e7b489c8 | 4734 | @findex log |
98ff7c4d KG |
4735 | @findex log10 |
4736 | @findex log10f | |
4737 | @findex log10l | |
4738 | @findex log1p | |
4739 | @findex log1pf | |
4740 | @findex log1pl | |
4741 | @findex log2 | |
4742 | @findex log2f | |
4743 | @findex log2l | |
4744 | @findex logb | |
4745 | @findex logbf | |
4746 | @findex logbl | |
e7b489c8 RS |
4747 | @findex logf |
4748 | @findex logl | |
98ff7c4d KG |
4749 | @findex lrint |
4750 | @findex lrintf | |
4751 | @findex lrintl | |
4752 | @findex lround | |
4753 | @findex lroundf | |
4754 | @findex lroundl | |
1331d16f | 4755 | @findex malloc |
01702459 JM |
4756 | @findex memcmp |
4757 | @findex memcpy | |
9cb65f92 | 4758 | @findex mempcpy |
01702459 | 4759 | @findex memset |
a2a919aa KG |
4760 | @findex modf |
4761 | @findex modff | |
4762 | @findex modfl | |
b052d8ee RS |
4763 | @findex nearbyint |
4764 | @findex nearbyintf | |
4765 | @findex nearbyintl | |
98ff7c4d KG |
4766 | @findex nextafter |
4767 | @findex nextafterf | |
4768 | @findex nextafterl | |
4769 | @findex nexttoward | |
4770 | @findex nexttowardf | |
4771 | @findex nexttowardl | |
46847aa6 | 4772 | @findex pow |
98ff7c4d KG |
4773 | @findex pow10 |
4774 | @findex pow10f | |
4775 | @findex pow10l | |
46847aa6 RS |
4776 | @findex powf |
4777 | @findex powl | |
01702459 | 4778 | @findex printf |
b4c984fb | 4779 | @findex printf_unlocked |
08291658 RS |
4780 | @findex putchar |
4781 | @findex puts | |
98ff7c4d KG |
4782 | @findex remainder |
4783 | @findex remainderf | |
4784 | @findex remainderl | |
a2a919aa KG |
4785 | @findex remquo |
4786 | @findex remquof | |
4787 | @findex remquol | |
c7b6c6cd | 4788 | @findex rindex |
98ff7c4d KG |
4789 | @findex rint |
4790 | @findex rintf | |
4791 | @findex rintl | |
b052d8ee RS |
4792 | @findex round |
4793 | @findex roundf | |
4794 | @findex roundl | |
98ff7c4d KG |
4795 | @findex scalb |
4796 | @findex scalbf | |
4797 | @findex scalbl | |
4798 | @findex scalbln | |
4799 | @findex scalblnf | |
4800 | @findex scalblnf | |
4801 | @findex scalbn | |
4802 | @findex scalbnf | |
4803 | @findex scanfnl | |
ef79730c RS |
4804 | @findex signbit |
4805 | @findex signbitf | |
4806 | @findex signbitl | |
488f17e1 KG |
4807 | @findex significand |
4808 | @findex significandf | |
4809 | @findex significandl | |
01702459 | 4810 | @findex sin |
a2a919aa KG |
4811 | @findex sincos |
4812 | @findex sincosf | |
4813 | @findex sincosl | |
01702459 | 4814 | @findex sinf |
98ff7c4d KG |
4815 | @findex sinh |
4816 | @findex sinhf | |
4817 | @findex sinhl | |
01702459 | 4818 | @findex sinl |
08291658 RS |
4819 | @findex snprintf |
4820 | @findex sprintf | |
01702459 JM |
4821 | @findex sqrt |
4822 | @findex sqrtf | |
4823 | @findex sqrtl | |
08291658 | 4824 | @findex sscanf |
9cb65f92 | 4825 | @findex stpcpy |
d118937d | 4826 | @findex strcat |
01702459 JM |
4827 | @findex strchr |
4828 | @findex strcmp | |
4829 | @findex strcpy | |
d118937d | 4830 | @findex strcspn |
1331d16f | 4831 | @findex strdup |
178b2b9f RS |
4832 | @findex strfmon |
4833 | @findex strftime | |
01702459 | 4834 | @findex strlen |
d118937d | 4835 | @findex strncat |
da9e9f08 KG |
4836 | @findex strncmp |
4837 | @findex strncpy | |
01702459 JM |
4838 | @findex strpbrk |
4839 | @findex strrchr | |
d118937d | 4840 | @findex strspn |
01702459 | 4841 | @findex strstr |
29f523be RS |
4842 | @findex tan |
4843 | @findex tanf | |
98ff7c4d KG |
4844 | @findex tanh |
4845 | @findex tanhf | |
4846 | @findex tanhl | |
29f523be | 4847 | @findex tanl |
488f17e1 KG |
4848 | @findex tgamma |
4849 | @findex tgammaf | |
4850 | @findex tgammal | |
740e5b6f KG |
4851 | @findex toascii |
4852 | @findex tolower | |
4853 | @findex toupper | |
ca4944e1 KG |
4854 | @findex towlower |
4855 | @findex towupper | |
4977bab6 ZW |
4856 | @findex trunc |
4857 | @findex truncf | |
4858 | @findex truncl | |
178b2b9f RS |
4859 | @findex vfprintf |
4860 | @findex vfscanf | |
08291658 RS |
4861 | @findex vprintf |
4862 | @findex vscanf | |
4863 | @findex vsnprintf | |
4864 | @findex vsprintf | |
4865 | @findex vsscanf | |
488f17e1 KG |
4866 | @findex y0 |
4867 | @findex y0f | |
4868 | @findex y0l | |
4869 | @findex y1 | |
4870 | @findex y1f | |
4871 | @findex y1l | |
4872 | @findex yn | |
4873 | @findex ynf | |
4874 | @findex ynl | |
185ebd6c | 4875 | |
f0523f02 | 4876 | GCC provides a large number of built-in functions other than the ones |
185ebd6c RH |
4877 | mentioned above. Some of these are for internal use in the processing |
4878 | of exceptions or variable-length argument lists and will not be | |
4879 | documented here because they may change from time to time; we do not | |
4880 | recommend general use of these functions. | |
4881 | ||
4882 | The remaining functions are provided for optimization purposes. | |
4883 | ||
84330467 | 4884 | @opindex fno-builtin |
9c34dbbf ZW |
4885 | GCC includes built-in versions of many of the functions in the standard |
4886 | C library. The versions prefixed with @code{__builtin_} will always be | |
4887 | treated as having the same meaning as the C library function even if you | |
8a36672b | 4888 | specify the @option{-fno-builtin} option. (@pxref{C Dialect Options}) |
9c34dbbf | 4889 | Many of these functions are only optimized in certain cases; if they are |
01702459 JM |
4890 | not optimized in a particular case, a call to the library function will |
4891 | be emitted. | |
4892 | ||
84330467 JM |
4893 | @opindex ansi |
4894 | @opindex std | |
b052d8ee | 4895 | Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or |
98ff7c4d KG |
4896 | @option{-std=c99}), the functions |
4897 | @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero}, | |
4898 | @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml}, | |
4899 | @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll}, | |
4900 | @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, @code{fputs_unlocked}, | |
488f17e1 | 4901 | @code{gammaf}, @code{gammal}, @code{gamma}, @code{gettext}, |
740e5b6f KG |
4902 | @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0}, |
4903 | @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn}, | |
4904 | @code{mempcpy}, @code{pow10f}, @code{pow10l}, @code{pow10}, | |
4905 | @code{printf_unlocked}, @code{rindex}, @code{scalbf}, @code{scalbl}, | |
4906 | @code{scalb}, @code{signbit}, @code{signbitf}, @code{signbitl}, | |
488f17e1 | 4907 | @code{significandf}, @code{significandl}, @code{significand}, |
a2a919aa | 4908 | @code{sincosf}, @code{sincosl}, @code{sincos}, @code{stpcpy}, |
740e5b6f KG |
4909 | @code{strdup}, @code{strfmon}, @code{toascii}, @code{y0f}, @code{y0l}, |
4910 | @code{y0}, @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and | |
4911 | @code{yn} | |
1331d16f | 4912 | may be handled as built-in functions. |
b052d8ee | 4913 | All these functions have corresponding versions |
9c34dbbf ZW |
4914 | prefixed with @code{__builtin_}, which may be used even in strict C89 |
4915 | mode. | |
01702459 | 4916 | |
075ec276 | 4917 | The ISO C99 functions |
98ff7c4d KG |
4918 | @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf}, |
4919 | @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh}, | |
11bf0eb0 KG |
4920 | @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf}, |
4921 | @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos}, | |
4922 | @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf}, | |
4923 | @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin}, | |
4924 | @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh}, | |
4925 | @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt}, | |
4926 | @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl}, | |
4927 | @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf}, | |
740e5b6f KG |
4928 | @code{cimagl}, @code{cimag}, @code{conjf}, @code{conjl}, @code{conj}, |
4929 | @code{copysignf}, @code{copysignl}, @code{copysign}, @code{cpowf}, | |
4930 | @code{cpowl}, @code{cpow}, @code{cprojf}, @code{cprojl}, @code{cproj}, | |
4931 | @code{crealf}, @code{creall}, @code{creal}, @code{csinf}, @code{csinhf}, | |
4932 | @code{csinhl}, @code{csinh}, @code{csinl}, @code{csin}, @code{csqrtf}, | |
4933 | @code{csqrtl}, @code{csqrt}, @code{ctanf}, @code{ctanhf}, @code{ctanhl}, | |
4934 | @code{ctanh}, @code{ctanl}, @code{ctan}, @code{erfcf}, @code{erfcl}, | |
4935 | @code{erfc}, @code{erff}, @code{erfl}, @code{erf}, @code{exp2f}, | |
4936 | @code{exp2l}, @code{exp2}, @code{expm1f}, @code{expm1l}, @code{expm1}, | |
4937 | @code{fdimf}, @code{fdiml}, @code{fdim}, @code{fmaf}, @code{fmal}, | |
4938 | @code{fmaxf}, @code{fmaxl}, @code{fmax}, @code{fma}, @code{fminf}, | |
4939 | @code{fminl}, @code{fmin}, @code{hypotf}, @code{hypotl}, @code{hypot}, | |
4940 | @code{ilogbf}, @code{ilogbl}, @code{ilogb}, @code{imaxabs}, | |
ca4944e1 KG |
4941 | @code{isblank}, @code{iswblank}, @code{lgammaf}, @code{lgammal}, |
4942 | @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl}, | |
4943 | @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround}, | |
4944 | @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l}, | |
4945 | @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf}, | |
4946 | @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl}, | |
4947 | @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint}, | |
740e5b6f KG |
4948 | @code{nextafterf}, @code{nextafterl}, @code{nextafter}, |
4949 | @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward}, | |
4950 | @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof}, | |
4951 | @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint}, | |
4952 | @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf}, | |
4953 | @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl}, | |
4954 | @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal}, | |
4955 | @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc}, | |
4956 | @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf} | |
08291658 | 4957 | are handled as built-in functions |
b052d8ee | 4958 | except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}). |
46847aa6 | 4959 | |
98ff7c4d KG |
4960 | There are also built-in versions of the ISO C99 functions |
4961 | @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f}, | |
29f523be | 4962 | @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill}, |
98ff7c4d KG |
4963 | @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf}, |
4964 | @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl}, | |
a2a919aa KG |
4965 | @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf}, |
4966 | @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl}, | |
4967 | @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf}, | |
4968 | @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl}, | |
4969 | @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl} | |
46847aa6 RS |
4970 | that are recognized in any mode since ISO C90 reserves these names for |
4971 | the purpose to which ISO C99 puts them. All these functions have | |
4972 | corresponding versions prefixed with @code{__builtin_}. | |
4973 | ||
ca4944e1 KG |
4974 | The ISO C94 functions |
4975 | @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit}, | |
4976 | @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct}, | |
4977 | @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and | |
4978 | @code{towupper} | |
4979 | are handled as built-in functions | |
4980 | except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}). | |
4981 | ||
98ff7c4d KG |
4982 | The ISO C90 functions |
4983 | @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2}, | |
4984 | @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos}, | |
4985 | @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod}, | |
740e5b6f KG |
4986 | @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf}, |
4987 | @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit}, | |
4988 | @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct}, | |
4989 | @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower}, | |
4990 | @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log}, | |
4991 | @code{malloc}, @code{memcmp}, @code{memcpy}, @code{memset}, @code{modf}, | |
4992 | @code{pow}, @code{printf}, @code{putchar}, @code{puts}, @code{scanf}, | |
4993 | @code{sinh}, @code{sin}, @code{snprintf}, @code{sprintf}, @code{sqrt}, | |
4994 | @code{sscanf}, @code{strcat}, @code{strchr}, @code{strcmp}, | |
4995 | @code{strcpy}, @code{strcspn}, @code{strlen}, @code{strncat}, | |
4996 | @code{strncmp}, @code{strncpy}, @code{strpbrk}, @code{strrchr}, | |
4997 | @code{strspn}, @code{strstr}, @code{tanh}, @code{tan}, @code{vfprintf}, | |
4998 | @code{vprintf} and @code{vsprintf} | |
08291658 | 4999 | are all recognized as built-in functions unless |
46847aa6 RS |
5000 | @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}} |
5001 | is specified for an individual function). All of these functions have | |
4977bab6 | 5002 | corresponding versions prefixed with @code{__builtin_}. |
9c34dbbf ZW |
5003 | |
5004 | GCC provides built-in versions of the ISO C99 floating point comparison | |
5005 | macros that avoid raising exceptions for unordered operands. They have | |
5006 | the same names as the standard macros ( @code{isgreater}, | |
5007 | @code{isgreaterequal}, @code{isless}, @code{islessequal}, | |
5008 | @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_} | |
5009 | prefixed. We intend for a library implementor to be able to simply | |
5010 | @code{#define} each standard macro to its built-in equivalent. | |
185ebd6c | 5011 | |
ecbcf7b3 AH |
5012 | @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2}) |
5013 | ||
5014 | You can use the built-in function @code{__builtin_types_compatible_p} to | |
5015 | determine whether two types are the same. | |
5016 | ||
5017 | This built-in function returns 1 if the unqualified versions of the | |
5018 | types @var{type1} and @var{type2} (which are types, not expressions) are | |
5019 | compatible, 0 otherwise. The result of this built-in function can be | |
5020 | used in integer constant expressions. | |
5021 | ||
5022 | This built-in function ignores top level qualifiers (e.g., @code{const}, | |
5023 | @code{volatile}). For example, @code{int} is equivalent to @code{const | |
5024 | int}. | |
5025 | ||
5026 | The type @code{int[]} and @code{int[5]} are compatible. On the other | |
5027 | hand, @code{int} and @code{char *} are not compatible, even if the size | |
5028 | of their types, on the particular architecture are the same. Also, the | |
5029 | amount of pointer indirection is taken into account when determining | |
5030 | similarity. Consequently, @code{short *} is not similar to | |
5031 | @code{short **}. Furthermore, two types that are typedefed are | |
5032 | considered compatible if their underlying types are compatible. | |
5033 | ||
bca63328 JM |
5034 | An @code{enum} type is not considered to be compatible with another |
5035 | @code{enum} type even if both are compatible with the same integer | |
5036 | type; this is what the C standard specifies. | |
5037 | For example, @code{enum @{foo, bar@}} is not similar to | |
ecbcf7b3 AH |
5038 | @code{enum @{hot, dog@}}. |
5039 | ||
5040 | You would typically use this function in code whose execution varies | |
5041 | depending on the arguments' types. For example: | |
5042 | ||
5043 | @smallexample | |
6e5bb5ad JM |
5044 | #define foo(x) \ |
5045 | (@{ \ | |
5046 | typeof (x) tmp; \ | |
5047 | if (__builtin_types_compatible_p (typeof (x), long double)) \ | |
5048 | tmp = foo_long_double (tmp); \ | |
5049 | else if (__builtin_types_compatible_p (typeof (x), double)) \ | |
5050 | tmp = foo_double (tmp); \ | |
5051 | else if (__builtin_types_compatible_p (typeof (x), float)) \ | |
5052 | tmp = foo_float (tmp); \ | |
5053 | else \ | |
5054 | abort (); \ | |
5055 | tmp; \ | |
ecbcf7b3 AH |
5056 | @}) |
5057 | @end smallexample | |
5058 | ||
8a36672b | 5059 | @emph{Note:} This construct is only available for C@. |
ecbcf7b3 AH |
5060 | |
5061 | @end deftypefn | |
5062 | ||
5063 | @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2}) | |
5064 | ||
5065 | You can use the built-in function @code{__builtin_choose_expr} to | |
5066 | evaluate code depending on the value of a constant expression. This | |
5067 | built-in function returns @var{exp1} if @var{const_exp}, which is a | |
5068 | constant expression that must be able to be determined at compile time, | |
5069 | is nonzero. Otherwise it returns 0. | |
5070 | ||
5071 | This built-in function is analogous to the @samp{? :} operator in C, | |
5072 | except that the expression returned has its type unaltered by promotion | |
5073 | rules. Also, the built-in function does not evaluate the expression | |
5074 | that was not chosen. For example, if @var{const_exp} evaluates to true, | |
5075 | @var{exp2} is not evaluated even if it has side-effects. | |
5076 | ||
5077 | This built-in function can return an lvalue if the chosen argument is an | |
5078 | lvalue. | |
5079 | ||
5080 | If @var{exp1} is returned, the return type is the same as @var{exp1}'s | |
5081 | type. Similarly, if @var{exp2} is returned, its return type is the same | |
5082 | as @var{exp2}. | |
5083 | ||
5084 | Example: | |
5085 | ||
5086 | @smallexample | |
478c9e72 JJ |
5087 | #define foo(x) \ |
5088 | __builtin_choose_expr ( \ | |
5089 | __builtin_types_compatible_p (typeof (x), double), \ | |
5090 | foo_double (x), \ | |
5091 | __builtin_choose_expr ( \ | |
5092 | __builtin_types_compatible_p (typeof (x), float), \ | |
5093 | foo_float (x), \ | |
5094 | /* @r{The void expression results in a compile-time error} \ | |
5095 | @r{when assigning the result to something.} */ \ | |
ecbcf7b3 AH |
5096 | (void)0)) |
5097 | @end smallexample | |
5098 | ||
8a36672b | 5099 | @emph{Note:} This construct is only available for C@. Furthermore, the |
ecbcf7b3 AH |
5100 | unused expression (@var{exp1} or @var{exp2} depending on the value of |
5101 | @var{const_exp}) may still generate syntax errors. This may change in | |
5102 | future revisions. | |
5103 | ||
5104 | @end deftypefn | |
5105 | ||
84330467 JM |
5106 | @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp}) |
5107 | You can use the built-in function @code{__builtin_constant_p} to | |
185ebd6c | 5108 | determine if a value is known to be constant at compile-time and hence |
f0523f02 | 5109 | that GCC can perform constant-folding on expressions involving that |
185ebd6c RH |
5110 | value. The argument of the function is the value to test. The function |
5111 | returns the integer 1 if the argument is known to be a compile-time | |
5112 | constant and 0 if it is not known to be a compile-time constant. A | |
5113 | return of 0 does not indicate that the value is @emph{not} a constant, | |
f0523f02 | 5114 | but merely that GCC cannot prove it is a constant with the specified |
84330467 | 5115 | value of the @option{-O} option. |
185ebd6c RH |
5116 | |
5117 | You would typically use this function in an embedded application where | |
5118 | memory was a critical resource. If you have some complex calculation, | |
5119 | you may want it to be folded if it involves constants, but need to call | |
5120 | a function if it does not. For example: | |
5121 | ||
4d390518 | 5122 | @smallexample |
310668e8 JM |
5123 | #define Scale_Value(X) \ |
5124 | (__builtin_constant_p (X) \ | |
5125 | ? ((X) * SCALE + OFFSET) : Scale (X)) | |
185ebd6c RH |
5126 | @end smallexample |
5127 | ||
84330467 | 5128 | You may use this built-in function in either a macro or an inline |
185ebd6c | 5129 | function. However, if you use it in an inlined function and pass an |
f0523f02 | 5130 | argument of the function as the argument to the built-in, GCC will |
185ebd6c | 5131 | never return 1 when you call the inline function with a string constant |
4b404517 | 5132 | or compound literal (@pxref{Compound Literals}) and will not return 1 |
185ebd6c | 5133 | when you pass a constant numeric value to the inline function unless you |
84330467 | 5134 | specify the @option{-O} option. |
13104975 ZW |
5135 | |
5136 | You may also use @code{__builtin_constant_p} in initializers for static | |
5137 | data. For instance, you can write | |
5138 | ||
5139 | @smallexample | |
79323c50 | 5140 | static const int table[] = @{ |
13104975 | 5141 | __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1, |
0d893a63 | 5142 | /* @r{@dots{}} */ |
79323c50 | 5143 | @}; |
13104975 ZW |
5144 | @end smallexample |
5145 | ||
5146 | @noindent | |
5147 | This is an acceptable initializer even if @var{EXPRESSION} is not a | |
5148 | constant expression. GCC must be more conservative about evaluating the | |
5149 | built-in in this case, because it has no opportunity to perform | |
5150 | optimization. | |
5151 | ||
5152 | Previous versions of GCC did not accept this built-in in data | |
5153 | initializers. The earliest version where it is completely safe is | |
5154 | 3.0.1. | |
84330467 | 5155 | @end deftypefn |
185ebd6c | 5156 | |
84330467 JM |
5157 | @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c}) |
5158 | @opindex fprofile-arcs | |
02f52e19 | 5159 | You may use @code{__builtin_expect} to provide the compiler with |
994a57cd | 5160 | branch prediction information. In general, you should prefer to |
84330467 | 5161 | use actual profile feedback for this (@option{-fprofile-arcs}), as |
994a57cd | 5162 | programmers are notoriously bad at predicting how their programs |
60b6e1f5 | 5163 | actually perform. However, there are applications in which this |
994a57cd RH |
5164 | data is hard to collect. |
5165 | ||
5166 | The return value is the value of @var{exp}, which should be an | |
5167 | integral expression. The value of @var{c} must be a compile-time | |
84330467 | 5168 | constant. The semantics of the built-in are that it is expected |
994a57cd RH |
5169 | that @var{exp} == @var{c}. For example: |
5170 | ||
5171 | @smallexample | |
5172 | if (__builtin_expect (x, 0)) | |
5173 | foo (); | |
5174 | @end smallexample | |
5175 | ||
5176 | @noindent | |
5177 | would indicate that we do not expect to call @code{foo}, since | |
5178 | we expect @code{x} to be zero. Since you are limited to integral | |
5179 | expressions for @var{exp}, you should use constructions such as | |
5180 | ||
5181 | @smallexample | |
5182 | if (__builtin_expect (ptr != NULL, 1)) | |
5183 | error (); | |
5184 | @end smallexample | |
5185 | ||
5186 | @noindent | |
5187 | when testing pointer or floating-point values. | |
84330467 | 5188 | @end deftypefn |
994a57cd | 5189 | |
3bca17dd | 5190 | @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...) |
a9ccbb60 JJ |
5191 | This function is used to minimize cache-miss latency by moving data into |
5192 | a cache before it is accessed. | |
5193 | You can insert calls to @code{__builtin_prefetch} into code for which | |
5194 | you know addresses of data in memory that is likely to be accessed soon. | |
5195 | If the target supports them, data prefetch instructions will be generated. | |
5196 | If the prefetch is done early enough before the access then the data will | |
5197 | be in the cache by the time it is accessed. | |
5198 | ||
5199 | The value of @var{addr} is the address of the memory to prefetch. | |
e83d297b | 5200 | There are two optional arguments, @var{rw} and @var{locality}. |
a9ccbb60 | 5201 | The value of @var{rw} is a compile-time constant one or zero; one |
e83d297b JJ |
5202 | means that the prefetch is preparing for a write to the memory address |
5203 | and zero, the default, means that the prefetch is preparing for a read. | |
a9ccbb60 JJ |
5204 | The value @var{locality} must be a compile-time constant integer between |
5205 | zero and three. A value of zero means that the data has no temporal | |
5206 | locality, so it need not be left in the cache after the access. A value | |
5207 | of three means that the data has a high degree of temporal locality and | |
5208 | should be left in all levels of cache possible. Values of one and two | |
e83d297b JJ |
5209 | mean, respectively, a low or moderate degree of temporal locality. The |
5210 | default is three. | |
a9ccbb60 JJ |
5211 | |
5212 | @smallexample | |
5213 | for (i = 0; i < n; i++) | |
5214 | @{ | |
5215 | a[i] = a[i] + b[i]; | |
5216 | __builtin_prefetch (&a[i+j], 1, 1); | |
5217 | __builtin_prefetch (&b[i+j], 0, 1); | |
0d893a63 | 5218 | /* @r{@dots{}} */ |
a9ccbb60 JJ |
5219 | @} |
5220 | @end smallexample | |
5221 | ||
f282ffb3 | 5222 | Data prefetch does not generate faults if @var{addr} is invalid, but |
a9ccbb60 JJ |
5223 | the address expression itself must be valid. For example, a prefetch |
5224 | of @code{p->next} will not fault if @code{p->next} is not a valid | |
5225 | address, but evaluation will fault if @code{p} is not a valid address. | |
5226 | ||
5227 | If the target does not support data prefetch, the address expression | |
5228 | is evaluated if it includes side effects but no other code is generated | |
5229 | and GCC does not issue a warning. | |
5230 | @end deftypefn | |
5231 | ||
ab5e2615 RH |
5232 | @deftypefn {Built-in Function} double __builtin_huge_val (void) |
5233 | Returns a positive infinity, if supported by the floating-point format, | |
5234 | else @code{DBL_MAX}. This function is suitable for implementing the | |
5235 | ISO C macro @code{HUGE_VAL}. | |
5236 | @end deftypefn | |
5237 | ||
5238 | @deftypefn {Built-in Function} float __builtin_huge_valf (void) | |
5239 | Similar to @code{__builtin_huge_val}, except the return type is @code{float}. | |
5240 | @end deftypefn | |
5241 | ||
dad78426 | 5242 | @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void) |
ab5e2615 RH |
5243 | Similar to @code{__builtin_huge_val}, except the return |
5244 | type is @code{long double}. | |
5245 | @end deftypefn | |
5246 | ||
5247 | @deftypefn {Built-in Function} double __builtin_inf (void) | |
5248 | Similar to @code{__builtin_huge_val}, except a warning is generated | |
5249 | if the target floating-point format does not support infinities. | |
5250 | This function is suitable for implementing the ISO C99 macro @code{INFINITY}. | |
5251 | @end deftypefn | |
5252 | ||
5253 | @deftypefn {Built-in Function} float __builtin_inff (void) | |
5254 | Similar to @code{__builtin_inf}, except the return type is @code{float}. | |
5255 | @end deftypefn | |
5256 | ||
dad78426 | 5257 | @deftypefn {Built-in Function} {long double} __builtin_infl (void) |
ab5e2615 RH |
5258 | Similar to @code{__builtin_inf}, except the return |
5259 | type is @code{long double}. | |
5260 | @end deftypefn | |
5261 | ||
1472e41c RH |
5262 | @deftypefn {Built-in Function} double __builtin_nan (const char *str) |
5263 | This is an implementation of the ISO C99 function @code{nan}. | |
5264 | ||
5265 | Since ISO C99 defines this function in terms of @code{strtod}, which we | |
c0478a66 | 5266 | do not implement, a description of the parsing is in order. The string |
1472e41c RH |
5267 | is parsed as by @code{strtol}; that is, the base is recognized by |
5268 | leading @samp{0} or @samp{0x} prefixes. The number parsed is placed | |
5269 | in the significand such that the least significant bit of the number | |
daf2f129 | 5270 | is at the least significant bit of the significand. The number is |
1472e41c | 5271 | truncated to fit the significand field provided. The significand is |
8a36672b | 5272 | forced to be a quiet NaN@. |
1472e41c RH |
5273 | |
5274 | This function, if given a string literal, is evaluated early enough | |
5275 | that it is considered a compile-time constant. | |
5276 | @end deftypefn | |
5277 | ||
5278 | @deftypefn {Built-in Function} float __builtin_nanf (const char *str) | |
5279 | Similar to @code{__builtin_nan}, except the return type is @code{float}. | |
5280 | @end deftypefn | |
5281 | ||
dad78426 | 5282 | @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str) |
1472e41c RH |
5283 | Similar to @code{__builtin_nan}, except the return type is @code{long double}. |
5284 | @end deftypefn | |
5285 | ||
5286 | @deftypefn {Built-in Function} double __builtin_nans (const char *str) | |
daf2f129 | 5287 | Similar to @code{__builtin_nan}, except the significand is forced |
8a36672b | 5288 | to be a signaling NaN@. The @code{nans} function is proposed by |
aaa67502 | 5289 | @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}. |
1472e41c RH |
5290 | @end deftypefn |
5291 | ||
5292 | @deftypefn {Built-in Function} float __builtin_nansf (const char *str) | |
5293 | Similar to @code{__builtin_nans}, except the return type is @code{float}. | |
5294 | @end deftypefn | |
5295 | ||
dad78426 | 5296 | @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str) |
1472e41c RH |
5297 | Similar to @code{__builtin_nans}, except the return type is @code{long double}. |
5298 | @end deftypefn | |
5299 | ||
2928cd7a RH |
5300 | @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x) |
5301 | Returns one plus the index of the least significant 1-bit of @var{x}, or | |
5302 | if @var{x} is zero, returns zero. | |
5303 | @end deftypefn | |
5304 | ||
5305 | @deftypefn {Built-in Function} int __builtin_clz (unsigned int x) | |
5306 | Returns the number of leading 0-bits in @var{x}, starting at the most | |
5307 | significant bit position. If @var{x} is 0, the result is undefined. | |
5308 | @end deftypefn | |
5309 | ||
5310 | @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x) | |
5311 | Returns the number of trailing 0-bits in @var{x}, starting at the least | |
5312 | significant bit position. If @var{x} is 0, the result is undefined. | |
5313 | @end deftypefn | |
5314 | ||
5315 | @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x) | |
5316 | Returns the number of 1-bits in @var{x}. | |
5317 | @end deftypefn | |
5318 | ||
5319 | @deftypefn {Built-in Function} int __builtin_parity (unsigned int x) | |
8a36672b | 5320 | Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x} |
2928cd7a RH |
5321 | modulo 2. |
5322 | @end deftypefn | |
5323 | ||
5324 | @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long) | |
5325 | Similar to @code{__builtin_ffs}, except the argument type is | |
5326 | @code{unsigned long}. | |
5327 | @end deftypefn | |
5328 | ||
5329 | @deftypefn {Built-in Function} int __builtin_clzl (unsigned long) | |
5330 | Similar to @code{__builtin_clz}, except the argument type is | |
5331 | @code{unsigned long}. | |
5332 | @end deftypefn | |
5333 | ||
5334 | @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long) | |
5335 | Similar to @code{__builtin_ctz}, except the argument type is | |
5336 | @code{unsigned long}. | |
5337 | @end deftypefn | |
5338 | ||
5339 | @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long) | |
5340 | Similar to @code{__builtin_popcount}, except the argument type is | |
5341 | @code{unsigned long}. | |
5342 | @end deftypefn | |
5343 | ||
5344 | @deftypefn {Built-in Function} int __builtin_parityl (unsigned long) | |
5345 | Similar to @code{__builtin_parity}, except the argument type is | |
5346 | @code{unsigned long}. | |
5347 | @end deftypefn | |
5348 | ||
5349 | @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long) | |
5350 | Similar to @code{__builtin_ffs}, except the argument type is | |
5351 | @code{unsigned long long}. | |
5352 | @end deftypefn | |
5353 | ||
5354 | @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long) | |
5355 | Similar to @code{__builtin_clz}, except the argument type is | |
5356 | @code{unsigned long long}. | |
5357 | @end deftypefn | |
5358 | ||
5359 | @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long) | |
5360 | Similar to @code{__builtin_ctz}, except the argument type is | |
5361 | @code{unsigned long long}. | |
5362 | @end deftypefn | |
5363 | ||
5364 | @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long) | |
5365 | Similar to @code{__builtin_popcount}, except the argument type is | |
5366 | @code{unsigned long long}. | |
5367 | @end deftypefn | |
5368 | ||
5369 | @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long) | |
5370 | Similar to @code{__builtin_parity}, except the argument type is | |
5371 | @code{unsigned long long}. | |
5372 | @end deftypefn | |
5373 | ||
17684d46 RG |
5374 | @deftypefn {Built-in Function} double __builtin_powi (double, int) |
5375 | Returns the first argument raised to the power of the second. Unlike the | |
5376 | @code{pow} function no guarantees about precision and rounding are made. | |
5377 | @end deftypefn | |
5378 | ||
5379 | @deftypefn {Built-in Function} float __builtin_powif (float, int) | |
5380 | Similar to @code{__builtin_powi}, except the argument and return types | |
5381 | are @code{float}. | |
5382 | @end deftypefn | |
5383 | ||
5384 | @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int) | |
5385 | Similar to @code{__builtin_powi}, except the argument and return types | |
5386 | are @code{long double}. | |
5387 | @end deftypefn | |
5388 | ||
2928cd7a | 5389 | |
0975678f JM |
5390 | @node Target Builtins |
5391 | @section Built-in Functions Specific to Particular Target Machines | |
5392 | ||
5393 | On some target machines, GCC supports many built-in functions specific | |
5394 | to those machines. Generally these generate calls to specific machine | |
5395 | instructions, but allow the compiler to schedule those calls. | |
5396 | ||
5397 | @menu | |
6d8fd7bb | 5398 | * Alpha Built-in Functions:: |
4bc73018 | 5399 | * ARM Built-in Functions:: |
c3ee0579 | 5400 | * FR-V Built-in Functions:: |
0975678f | 5401 | * X86 Built-in Functions:: |
d840bfd3 | 5402 | * MIPS Paired-Single Support:: |
333c8841 | 5403 | * PowerPC AltiVec Built-in Functions:: |
c5145ceb | 5404 | * SPARC VIS Built-in Functions:: |
0975678f JM |
5405 | @end menu |
5406 | ||
6d8fd7bb RH |
5407 | @node Alpha Built-in Functions |
5408 | @subsection Alpha Built-in Functions | |
5409 | ||
5410 | These built-in functions are available for the Alpha family of | |
5411 | processors, depending on the command-line switches used. | |
5412 | ||
95b1627e | 5413 | The following built-in functions are always available. They |
6d8fd7bb RH |
5414 | all generate the machine instruction that is part of the name. |
5415 | ||
3ab51846 | 5416 | @smallexample |
6d8fd7bb RH |
5417 | long __builtin_alpha_implver (void) |
5418 | long __builtin_alpha_rpcc (void) | |
5419 | long __builtin_alpha_amask (long) | |
5420 | long __builtin_alpha_cmpbge (long, long) | |
c4b50f1a RH |
5421 | long __builtin_alpha_extbl (long, long) |
5422 | long __builtin_alpha_extwl (long, long) | |
5423 | long __builtin_alpha_extll (long, long) | |
6d8fd7bb | 5424 | long __builtin_alpha_extql (long, long) |
c4b50f1a RH |
5425 | long __builtin_alpha_extwh (long, long) |
5426 | long __builtin_alpha_extlh (long, long) | |
6d8fd7bb | 5427 | long __builtin_alpha_extqh (long, long) |
c4b50f1a RH |
5428 | long __builtin_alpha_insbl (long, long) |
5429 | long __builtin_alpha_inswl (long, long) | |
5430 | long __builtin_alpha_insll (long, long) | |
5431 | long __builtin_alpha_insql (long, long) | |
5432 | long __builtin_alpha_inswh (long, long) | |
5433 | long __builtin_alpha_inslh (long, long) | |
5434 | long __builtin_alpha_insqh (long, long) | |
5435 | long __builtin_alpha_mskbl (long, long) | |
5436 | long __builtin_alpha_mskwl (long, long) | |
5437 | long __builtin_alpha_mskll (long, long) | |
5438 | long __builtin_alpha_mskql (long, long) | |
5439 | long __builtin_alpha_mskwh (long, long) | |
5440 | long __builtin_alpha_msklh (long, long) | |
5441 | long __builtin_alpha_mskqh (long, long) | |
5442 | long __builtin_alpha_umulh (long, long) | |
6d8fd7bb RH |
5443 | long __builtin_alpha_zap (long, long) |
5444 | long __builtin_alpha_zapnot (long, long) | |
3ab51846 | 5445 | @end smallexample |
6d8fd7bb RH |
5446 | |
5447 | The following built-in functions are always with @option{-mmax} | |
5448 | or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or | |
5449 | later. They all generate the machine instruction that is part | |
5450 | of the name. | |
5451 | ||
3ab51846 | 5452 | @smallexample |
6d8fd7bb RH |
5453 | long __builtin_alpha_pklb (long) |
5454 | long __builtin_alpha_pkwb (long) | |
5455 | long __builtin_alpha_unpkbl (long) | |
5456 | long __builtin_alpha_unpkbw (long) | |
5457 | long __builtin_alpha_minub8 (long, long) | |
5458 | long __builtin_alpha_minsb8 (long, long) | |
5459 | long __builtin_alpha_minuw4 (long, long) | |
5460 | long __builtin_alpha_minsw4 (long, long) | |
5461 | long __builtin_alpha_maxub8 (long, long) | |
5462 | long __builtin_alpha_maxsb8 (long, long) | |
5463 | long __builtin_alpha_maxuw4 (long, long) | |
5464 | long __builtin_alpha_maxsw4 (long, long) | |
5465 | long __builtin_alpha_perr (long, long) | |
3ab51846 | 5466 | @end smallexample |
6d8fd7bb | 5467 | |
c4b50f1a RH |
5468 | The following built-in functions are always with @option{-mcix} |
5469 | or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or | |
5470 | later. They all generate the machine instruction that is part | |
5471 | of the name. | |
5472 | ||
3ab51846 | 5473 | @smallexample |
c4b50f1a RH |
5474 | long __builtin_alpha_cttz (long) |
5475 | long __builtin_alpha_ctlz (long) | |
5476 | long __builtin_alpha_ctpop (long) | |
3ab51846 | 5477 | @end smallexample |
c4b50f1a | 5478 | |
116b7a5e RH |
5479 | The following builtins are available on systems that use the OSF/1 |
5480 | PALcode. Normally they invoke the @code{rduniq} and @code{wruniq} | |
5481 | PAL calls, but when invoked with @option{-mtls-kernel}, they invoke | |
5482 | @code{rdval} and @code{wrval}. | |
5483 | ||
3ab51846 | 5484 | @smallexample |
116b7a5e RH |
5485 | void *__builtin_thread_pointer (void) |
5486 | void __builtin_set_thread_pointer (void *) | |
3ab51846 | 5487 | @end smallexample |
116b7a5e | 5488 | |
4bc73018 NC |
5489 | @node ARM Built-in Functions |
5490 | @subsection ARM Built-in Functions | |
5491 | ||
5492 | These built-in functions are available for the ARM family of | |
5493 | processors, when the @option{-mcpu=iwmmxt} switch is used: | |
5494 | ||
3ab51846 | 5495 | @smallexample |
d63851eb ILT |
5496 | typedef int v2si __attribute__ ((vector_size (8))); |
5497 | typedef short v4hi __attribute__ ((vector_size (8))); | |
5498 | typedef char v8qi __attribute__ ((vector_size (8))); | |
5499 | ||
5500 | int __builtin_arm_getwcx (int) | |
5501 | void __builtin_arm_setwcx (int, int) | |
5502 | int __builtin_arm_textrmsb (v8qi, int) | |
5503 | int __builtin_arm_textrmsh (v4hi, int) | |
5504 | int __builtin_arm_textrmsw (v2si, int) | |
5505 | int __builtin_arm_textrmub (v8qi, int) | |
5506 | int __builtin_arm_textrmuh (v4hi, int) | |
5507 | int __builtin_arm_textrmuw (v2si, int) | |
5508 | v8qi __builtin_arm_tinsrb (v8qi, int) | |
5509 | v4hi __builtin_arm_tinsrh (v4hi, int) | |
5510 | v2si __builtin_arm_tinsrw (v2si, int) | |
5511 | long long __builtin_arm_tmia (long long, int, int) | |
5512 | long long __builtin_arm_tmiabb (long long, int, int) | |
5513 | long long __builtin_arm_tmiabt (long long, int, int) | |
5514 | long long __builtin_arm_tmiaph (long long, int, int) | |
5515 | long long __builtin_arm_tmiatb (long long, int, int) | |
5516 | long long __builtin_arm_tmiatt (long long, int, int) | |
5517 | int __builtin_arm_tmovmskb (v8qi) | |
5518 | int __builtin_arm_tmovmskh (v4hi) | |
5519 | int __builtin_arm_tmovmskw (v2si) | |
5520 | long long __builtin_arm_waccb (v8qi) | |
5521 | long long __builtin_arm_wacch (v4hi) | |
5522 | long long __builtin_arm_waccw (v2si) | |
5523 | v8qi __builtin_arm_waddb (v8qi, v8qi) | |
5524 | v8qi __builtin_arm_waddbss (v8qi, v8qi) | |
5525 | v8qi __builtin_arm_waddbus (v8qi, v8qi) | |
5526 | v4hi __builtin_arm_waddh (v4hi, v4hi) | |
5527 | v4hi __builtin_arm_waddhss (v4hi, v4hi) | |
5528 | v4hi __builtin_arm_waddhus (v4hi, v4hi) | |
4bc73018 | 5529 | v2si __builtin_arm_waddw (v2si, v2si) |
4bc73018 | 5530 | v2si __builtin_arm_waddwss (v2si, v2si) |
4bc73018 | 5531 | v2si __builtin_arm_waddwus (v2si, v2si) |
d63851eb ILT |
5532 | v8qi __builtin_arm_walign (v8qi, v8qi, int) |
5533 | long long __builtin_arm_wand(long long, long long) | |
5534 | long long __builtin_arm_wandn (long long, long long) | |
5535 | v8qi __builtin_arm_wavg2b (v8qi, v8qi) | |
5536 | v8qi __builtin_arm_wavg2br (v8qi, v8qi) | |
5537 | v4hi __builtin_arm_wavg2h (v4hi, v4hi) | |
5538 | v4hi __builtin_arm_wavg2hr (v4hi, v4hi) | |
5539 | v8qi __builtin_arm_wcmpeqb (v8qi, v8qi) | |
5540 | v4hi __builtin_arm_wcmpeqh (v4hi, v4hi) | |
4bc73018 | 5541 | v2si __builtin_arm_wcmpeqw (v2si, v2si) |
d63851eb ILT |
5542 | v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi) |
5543 | v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi) | |
4bc73018 | 5544 | v2si __builtin_arm_wcmpgtsw (v2si, v2si) |
d63851eb ILT |
5545 | v8qi __builtin_arm_wcmpgtub (v8qi, v8qi) |
5546 | v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi) | |
5547 | v2si __builtin_arm_wcmpgtuw (v2si, v2si) | |
5548 | long long __builtin_arm_wmacs (long long, v4hi, v4hi) | |
5549 | long long __builtin_arm_wmacsz (v4hi, v4hi) | |
5550 | long long __builtin_arm_wmacu (long long, v4hi, v4hi) | |
5551 | long long __builtin_arm_wmacuz (v4hi, v4hi) | |
5552 | v4hi __builtin_arm_wmadds (v4hi, v4hi) | |
5553 | v4hi __builtin_arm_wmaddu (v4hi, v4hi) | |
5554 | v8qi __builtin_arm_wmaxsb (v8qi, v8qi) | |
5555 | v4hi __builtin_arm_wmaxsh (v4hi, v4hi) | |
4bc73018 | 5556 | v2si __builtin_arm_wmaxsw (v2si, v2si) |
d63851eb ILT |
5557 | v8qi __builtin_arm_wmaxub (v8qi, v8qi) |
5558 | v4hi __builtin_arm_wmaxuh (v4hi, v4hi) | |
4bc73018 | 5559 | v2si __builtin_arm_wmaxuw (v2si, v2si) |
d63851eb ILT |
5560 | v8qi __builtin_arm_wminsb (v8qi, v8qi) |
5561 | v4hi __builtin_arm_wminsh (v4hi, v4hi) | |
4bc73018 | 5562 | v2si __builtin_arm_wminsw (v2si, v2si) |
d63851eb ILT |
5563 | v8qi __builtin_arm_wminub (v8qi, v8qi) |
5564 | v4hi __builtin_arm_wminuh (v4hi, v4hi) | |
4bc73018 | 5565 | v2si __builtin_arm_wminuw (v2si, v2si) |
d63851eb ILT |
5566 | v4hi __builtin_arm_wmulsm (v4hi, v4hi) |
5567 | v4hi __builtin_arm_wmulul (v4hi, v4hi) | |
5568 | v4hi __builtin_arm_wmulum (v4hi, v4hi) | |
5569 | long long __builtin_arm_wor (long long, long long) | |
5570 | v2si __builtin_arm_wpackdss (long long, long long) | |
5571 | v2si __builtin_arm_wpackdus (long long, long long) | |
5572 | v8qi __builtin_arm_wpackhss (v4hi, v4hi) | |
5573 | v8qi __builtin_arm_wpackhus (v4hi, v4hi) | |
5574 | v4hi __builtin_arm_wpackwss (v2si, v2si) | |
5575 | v4hi __builtin_arm_wpackwus (v2si, v2si) | |
5576 | long long __builtin_arm_wrord (long long, long long) | |
5577 | long long __builtin_arm_wrordi (long long, int) | |
5578 | v4hi __builtin_arm_wrorh (v4hi, long long) | |
5579 | v4hi __builtin_arm_wrorhi (v4hi, int) | |
5580 | v2si __builtin_arm_wrorw (v2si, long long) | |
5581 | v2si __builtin_arm_wrorwi (v2si, int) | |
5582 | v2si __builtin_arm_wsadb (v8qi, v8qi) | |
5583 | v2si __builtin_arm_wsadbz (v8qi, v8qi) | |
5584 | v2si __builtin_arm_wsadh (v4hi, v4hi) | |
5585 | v2si __builtin_arm_wsadhz (v4hi, v4hi) | |
5586 | v4hi __builtin_arm_wshufh (v4hi, int) | |
5587 | long long __builtin_arm_wslld (long long, long long) | |
5588 | long long __builtin_arm_wslldi (long long, int) | |
5589 | v4hi __builtin_arm_wsllh (v4hi, long long) | |
5590 | v4hi __builtin_arm_wsllhi (v4hi, int) | |
5591 | v2si __builtin_arm_wsllw (v2si, long long) | |
4bc73018 | 5592 | v2si __builtin_arm_wsllwi (v2si, int) |
d63851eb ILT |
5593 | long long __builtin_arm_wsrad (long long, long long) |
5594 | long long __builtin_arm_wsradi (long long, int) | |
5595 | v4hi __builtin_arm_wsrah (v4hi, long long) | |
5596 | v4hi __builtin_arm_wsrahi (v4hi, int) | |
5597 | v2si __builtin_arm_wsraw (v2si, long long) | |
4bc73018 | 5598 | v2si __builtin_arm_wsrawi (v2si, int) |
d63851eb ILT |
5599 | long long __builtin_arm_wsrld (long long, long long) |
5600 | long long __builtin_arm_wsrldi (long long, int) | |
5601 | v4hi __builtin_arm_wsrlh (v4hi, long long) | |
5602 | v4hi __builtin_arm_wsrlhi (v4hi, int) | |
5603 | v2si __builtin_arm_wsrlw (v2si, long long) | |
4bc73018 | 5604 | v2si __builtin_arm_wsrlwi (v2si, int) |
d63851eb ILT |
5605 | v8qi __builtin_arm_wsubb (v8qi, v8qi) |
5606 | v8qi __builtin_arm_wsubbss (v8qi, v8qi) | |
5607 | v8qi __builtin_arm_wsubbus (v8qi, v8qi) | |
5608 | v4hi __builtin_arm_wsubh (v4hi, v4hi) | |
5609 | v4hi __builtin_arm_wsubhss (v4hi, v4hi) | |
5610 | v4hi __builtin_arm_wsubhus (v4hi, v4hi) | |
5611 | v2si __builtin_arm_wsubw (v2si, v2si) | |
5612 | v2si __builtin_arm_wsubwss (v2si, v2si) | |
5613 | v2si __builtin_arm_wsubwus (v2si, v2si) | |
5614 | v4hi __builtin_arm_wunpckehsb (v8qi) | |
5615 | v2si __builtin_arm_wunpckehsh (v4hi) | |
5616 | long long __builtin_arm_wunpckehsw (v2si) | |
5617 | v4hi __builtin_arm_wunpckehub (v8qi) | |
5618 | v2si __builtin_arm_wunpckehuh (v4hi) | |
5619 | long long __builtin_arm_wunpckehuw (v2si) | |
5620 | v4hi __builtin_arm_wunpckelsb (v8qi) | |
5621 | v2si __builtin_arm_wunpckelsh (v4hi) | |
5622 | long long __builtin_arm_wunpckelsw (v2si) | |
5623 | v4hi __builtin_arm_wunpckelub (v8qi) | |
5624 | v2si __builtin_arm_wunpckeluh (v4hi) | |
5625 | long long __builtin_arm_wunpckeluw (v2si) | |
5626 | v8qi __builtin_arm_wunpckihb (v8qi, v8qi) | |
5627 | v4hi __builtin_arm_wunpckihh (v4hi, v4hi) | |
4bc73018 | 5628 | v2si __builtin_arm_wunpckihw (v2si, v2si) |
d63851eb ILT |
5629 | v8qi __builtin_arm_wunpckilb (v8qi, v8qi) |
5630 | v4hi __builtin_arm_wunpckilh (v4hi, v4hi) | |
4bc73018 | 5631 | v2si __builtin_arm_wunpckilw (v2si, v2si) |
d63851eb ILT |
5632 | long long __builtin_arm_wxor (long long, long long) |
5633 | long long __builtin_arm_wzero () | |
3ab51846 | 5634 | @end smallexample |
4bc73018 | 5635 | |
c3ee0579 RS |
5636 | @node FR-V Built-in Functions |
5637 | @subsection FR-V Built-in Functions | |
5638 | ||
5639 | GCC provides many FR-V-specific built-in functions. In general, | |
5640 | these functions are intended to be compatible with those described | |
5641 | by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu | |
5642 | Semiconductor}. The two exceptions are @code{__MDUNPACKH} and | |
5643 | @code{__MBTOHE}, the gcc forms of which pass 128-bit values by | |
5644 | pointer rather than by value. | |
5645 | ||
5646 | Most of the functions are named after specific FR-V instructions. | |
27ef2cdd | 5647 | Such functions are said to be ``directly mapped'' and are summarized |
c3ee0579 RS |
5648 | here in tabular form. |
5649 | ||
5650 | @menu | |
5651 | * Argument Types:: | |
5652 | * Directly-mapped Integer Functions:: | |
5653 | * Directly-mapped Media Functions:: | |
5654 | * Other Built-in Functions:: | |
5655 | @end menu | |
5656 | ||
5657 | @node Argument Types | |
5658 | @subsubsection Argument Types | |
5659 | ||
5660 | The arguments to the built-in functions can be divided into three groups: | |
5661 | register numbers, compile-time constants and run-time values. In order | |
5662 | to make this classification clear at a glance, the arguments and return | |
5663 | values are given the following pseudo types: | |
5664 | ||
5665 | @multitable @columnfractions .20 .30 .15 .35 | |
5666 | @item Pseudo type @tab Real C type @tab Constant? @tab Description | |
5667 | @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword | |
5668 | @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word | |
5669 | @item @code{sw1} @tab @code{int} @tab No @tab a signed word | |
5670 | @item @code{uw2} @tab @code{unsigned long long} @tab No | |
5671 | @tab an unsigned doubleword | |
5672 | @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword | |
5673 | @item @code{const} @tab @code{int} @tab Yes @tab an integer constant | |
5674 | @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number | |
5675 | @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number | |
5676 | @end multitable | |
5677 | ||
5678 | These pseudo types are not defined by GCC, they are simply a notational | |
5679 | convenience used in this manual. | |
5680 | ||
5681 | Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2} | |
5682 | and @code{sw2} are evaluated at run time. They correspond to | |
5683 | register operands in the underlying FR-V instructions. | |
5684 | ||
5685 | @code{const} arguments represent immediate operands in the underlying | |
5686 | FR-V instructions. They must be compile-time constants. | |
5687 | ||
5688 | @code{acc} arguments are evaluated at compile time and specify the number | |
5689 | of an accumulator register. For example, an @code{acc} argument of 2 | |
5690 | will select the ACC2 register. | |
5691 | ||
5692 | @code{iacc} arguments are similar to @code{acc} arguments but specify the | |
5693 | number of an IACC register. See @pxref{Other Built-in Functions} | |
5694 | for more details. | |
5695 | ||
5696 | @node Directly-mapped Integer Functions | |
5697 | @subsubsection Directly-mapped Integer Functions | |
5698 | ||
5699 | The functions listed below map directly to FR-V I-type instructions. | |
5700 | ||
5701 | @multitable @columnfractions .45 .32 .23 | |
5702 | @item Function prototype @tab Example usage @tab Assembly output | |
5703 | @item @code{sw1 __ADDSS (sw1, sw1)} | |
5704 | @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})} | |
5705 | @tab @code{ADDSS @var{a},@var{b},@var{c}} | |
5706 | @item @code{sw1 __SCAN (sw1, sw1)} | |
5707 | @tab @code{@var{c} = __SCAN (@var{a}, @var{b})} | |
5708 | @tab @code{SCAN @var{a},@var{b},@var{c}} | |
5709 | @item @code{sw1 __SCUTSS (sw1)} | |
5710 | @tab @code{@var{b} = __SCUTSS (@var{a})} | |
5711 | @tab @code{SCUTSS @var{a},@var{b}} | |
5712 | @item @code{sw1 __SLASS (sw1, sw1)} | |
5713 | @tab @code{@var{c} = __SLASS (@var{a}, @var{b})} | |
5714 | @tab @code{SLASS @var{a},@var{b},@var{c}} | |
5715 | @item @code{void __SMASS (sw1, sw1)} | |
5716 | @tab @code{__SMASS (@var{a}, @var{b})} | |
5717 | @tab @code{SMASS @var{a},@var{b}} | |
5718 | @item @code{void __SMSSS (sw1, sw1)} | |
5719 | @tab @code{__SMSSS (@var{a}, @var{b})} | |
5720 | @tab @code{SMSSS @var{a},@var{b}} | |
5721 | @item @code{void __SMU (sw1, sw1)} | |
5722 | @tab @code{__SMU (@var{a}, @var{b})} | |
5723 | @tab @code{SMU @var{a},@var{b}} | |
5724 | @item @code{sw2 __SMUL (sw1, sw1)} | |
5725 | @tab @code{@var{c} = __SMUL (@var{a}, @var{b})} | |
5726 | @tab @code{SMUL @var{a},@var{b},@var{c}} | |
5727 | @item @code{sw1 __SUBSS (sw1, sw1)} | |
5728 | @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})} | |
5729 | @tab @code{SUBSS @var{a},@var{b},@var{c}} | |
5730 | @item @code{uw2 __UMUL (uw1, uw1)} | |
5731 | @tab @code{@var{c} = __UMUL (@var{a}, @var{b})} | |
5732 | @tab @code{UMUL @var{a},@var{b},@var{c}} | |
5733 | @end multitable | |
5734 | ||
5735 | @node Directly-mapped Media Functions | |
5736 | @subsubsection Directly-mapped Media Functions | |
5737 | ||
5738 | The functions listed below map directly to FR-V M-type instructions. | |
5739 | ||
5740 | @multitable @columnfractions .45 .32 .23 | |
5741 | @item Function prototype @tab Example usage @tab Assembly output | |
5742 | @item @code{uw1 __MABSHS (sw1)} | |
5743 | @tab @code{@var{b} = __MABSHS (@var{a})} | |
5744 | @tab @code{MABSHS @var{a},@var{b}} | |
5745 | @item @code{void __MADDACCS (acc, acc)} | |
5746 | @tab @code{__MADDACCS (@var{b}, @var{a})} | |
5747 | @tab @code{MADDACCS @var{a},@var{b}} | |
5748 | @item @code{sw1 __MADDHSS (sw1, sw1)} | |
5749 | @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})} | |
5750 | @tab @code{MADDHSS @var{a},@var{b},@var{c}} | |
5751 | @item @code{uw1 __MADDHUS (uw1, uw1)} | |
5752 | @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})} | |
5753 | @tab @code{MADDHUS @var{a},@var{b},@var{c}} | |
5754 | @item @code{uw1 __MAND (uw1, uw1)} | |
5755 | @tab @code{@var{c} = __MAND (@var{a}, @var{b})} | |
5756 | @tab @code{MAND @var{a},@var{b},@var{c}} | |
5757 | @item @code{void __MASACCS (acc, acc)} | |
5758 | @tab @code{__MASACCS (@var{b}, @var{a})} | |
5759 | @tab @code{MASACCS @var{a},@var{b}} | |
5760 | @item @code{uw1 __MAVEH (uw1, uw1)} | |
5761 | @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})} | |
5762 | @tab @code{MAVEH @var{a},@var{b},@var{c}} | |
5763 | @item @code{uw2 __MBTOH (uw1)} | |
5764 | @tab @code{@var{b} = __MBTOH (@var{a})} | |
5765 | @tab @code{MBTOH @var{a},@var{b}} | |
5766 | @item @code{void __MBTOHE (uw1 *, uw1)} | |
5767 | @tab @code{__MBTOHE (&@var{b}, @var{a})} | |
5768 | @tab @code{MBTOHE @var{a},@var{b}} | |
5769 | @item @code{void __MCLRACC (acc)} | |
5770 | @tab @code{__MCLRACC (@var{a})} | |
5771 | @tab @code{MCLRACC @var{a}} | |
5772 | @item @code{void __MCLRACCA (void)} | |
5773 | @tab @code{__MCLRACCA ()} | |
5774 | @tab @code{MCLRACCA} | |
5775 | @item @code{uw1 __Mcop1 (uw1, uw1)} | |
5776 | @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})} | |
5777 | @tab @code{Mcop1 @var{a},@var{b},@var{c}} | |
5778 | @item @code{uw1 __Mcop2 (uw1, uw1)} | |
5779 | @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})} | |
5780 | @tab @code{Mcop2 @var{a},@var{b},@var{c}} | |
5781 | @item @code{uw1 __MCPLHI (uw2, const)} | |
5782 | @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})} | |
5783 | @tab @code{MCPLHI @var{a},#@var{b},@var{c}} | |
5784 | @item @code{uw1 __MCPLI (uw2, const)} | |
5785 | @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})} | |
5786 | @tab @code{MCPLI @var{a},#@var{b},@var{c}} | |
5787 | @item @code{void __MCPXIS (acc, sw1, sw1)} | |
5788 | @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})} | |
5789 | @tab @code{MCPXIS @var{a},@var{b},@var{c}} | |
5790 | @item @code{void __MCPXIU (acc, uw1, uw1)} | |
5791 | @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})} | |
5792 | @tab @code{MCPXIU @var{a},@var{b},@var{c}} | |
5793 | @item @code{void __MCPXRS (acc, sw1, sw1)} | |
5794 | @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})} | |
5795 | @tab @code{MCPXRS @var{a},@var{b},@var{c}} | |
5796 | @item @code{void __MCPXRU (acc, uw1, uw1)} | |
5797 | @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})} | |
5798 | @tab @code{MCPXRU @var{a},@var{b},@var{c}} | |
5799 | @item @code{uw1 __MCUT (acc, uw1)} | |
5800 | @tab @code{@var{c} = __MCUT (@var{a}, @var{b})} | |
5801 | @tab @code{MCUT @var{a},@var{b},@var{c}} | |
5802 | @item @code{uw1 __MCUTSS (acc, sw1)} | |
5803 | @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})} | |
5804 | @tab @code{MCUTSS @var{a},@var{b},@var{c}} | |
5805 | @item @code{void __MDADDACCS (acc, acc)} | |
5806 | @tab @code{__MDADDACCS (@var{b}, @var{a})} | |
5807 | @tab @code{MDADDACCS @var{a},@var{b}} | |
5808 | @item @code{void __MDASACCS (acc, acc)} | |
5809 | @tab @code{__MDASACCS (@var{b}, @var{a})} | |
5810 | @tab @code{MDASACCS @var{a},@var{b}} | |
5811 | @item @code{uw2 __MDCUTSSI (acc, const)} | |
5812 | @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})} | |
5813 | @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}} | |
5814 | @item @code{uw2 __MDPACKH (uw2, uw2)} | |
5815 | @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})} | |
5816 | @tab @code{MDPACKH @var{a},@var{b},@var{c}} | |
5817 | @item @code{uw2 __MDROTLI (uw2, const)} | |
5818 | @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})} | |
5819 | @tab @code{MDROTLI @var{a},#@var{b},@var{c}} | |
5820 | @item @code{void __MDSUBACCS (acc, acc)} | |
5821 | @tab @code{__MDSUBACCS (@var{b}, @var{a})} | |
5822 | @tab @code{MDSUBACCS @var{a},@var{b}} | |
5823 | @item @code{void __MDUNPACKH (uw1 *, uw2)} | |
5824 | @tab @code{__MDUNPACKH (&@var{b}, @var{a})} | |
5825 | @tab @code{MDUNPACKH @var{a},@var{b}} | |
5826 | @item @code{uw2 __MEXPDHD (uw1, const)} | |
5827 | @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})} | |
5828 | @tab @code{MEXPDHD @var{a},#@var{b},@var{c}} | |
5829 | @item @code{uw1 __MEXPDHW (uw1, const)} | |
5830 | @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})} | |
5831 | @tab @code{MEXPDHW @var{a},#@var{b},@var{c}} | |
5832 | @item @code{uw1 __MHDSETH (uw1, const)} | |
5833 | @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})} | |
5834 | @tab @code{MHDSETH @var{a},#@var{b},@var{c}} | |
5835 | @item @code{sw1 __MHDSETS (const)} | |
5836 | @tab @code{@var{b} = __MHDSETS (@var{a})} | |
5837 | @tab @code{MHDSETS #@var{a},@var{b}} | |
5838 | @item @code{uw1 __MHSETHIH (uw1, const)} | |
5839 | @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})} | |
5840 | @tab @code{MHSETHIH #@var{a},@var{b}} | |
5841 | @item @code{sw1 __MHSETHIS (sw1, const)} | |
5842 | @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})} | |
5843 | @tab @code{MHSETHIS #@var{a},@var{b}} | |
5844 | @item @code{uw1 __MHSETLOH (uw1, const)} | |
5845 | @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})} | |
5846 | @tab @code{MHSETLOH #@var{a},@var{b}} | |
5847 | @item @code{sw1 __MHSETLOS (sw1, const)} | |
5848 | @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})} | |
5849 | @tab @code{MHSETLOS #@var{a},@var{b}} | |
5850 | @item @code{uw1 __MHTOB (uw2)} | |
5851 | @tab @code{@var{b} = __MHTOB (@var{a})} | |
5852 | @tab @code{MHTOB @var{a},@var{b}} | |
5853 | @item @code{void __MMACHS (acc, sw1, sw1)} | |
5854 | @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})} | |
5855 | @tab @code{MMACHS @var{a},@var{b},@var{c}} | |
5856 | @item @code{void __MMACHU (acc, uw1, uw1)} | |
5857 | @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})} | |
5858 | @tab @code{MMACHU @var{a},@var{b},@var{c}} | |
5859 | @item @code{void __MMRDHS (acc, sw1, sw1)} | |
5860 | @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})} | |
5861 | @tab @code{MMRDHS @var{a},@var{b},@var{c}} | |
5862 | @item @code{void __MMRDHU (acc, uw1, uw1)} | |
5863 | @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})} | |
5864 | @tab @code{MMRDHU @var{a},@var{b},@var{c}} | |
5865 | @item @code{void __MMULHS (acc, sw1, sw1)} | |
5866 | @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})} | |
5867 | @tab @code{MMULHS @var{a},@var{b},@var{c}} | |
5868 | @item @code{void __MMULHU (acc, uw1, uw1)} | |
5869 | @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})} | |
5870 | @tab @code{MMULHU @var{a},@var{b},@var{c}} | |
5871 | @item @code{void __MMULXHS (acc, sw1, sw1)} | |
5872 | @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})} | |
5873 | @tab @code{MMULXHS @var{a},@var{b},@var{c}} | |
5874 | @item @code{void __MMULXHU (acc, uw1, uw1)} | |
5875 | @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})} | |
5876 | @tab @code{MMULXHU @var{a},@var{b},@var{c}} | |
5877 | @item @code{uw1 __MNOT (uw1)} | |
5878 | @tab @code{@var{b} = __MNOT (@var{a})} | |
5879 | @tab @code{MNOT @var{a},@var{b}} | |
5880 | @item @code{uw1 __MOR (uw1, uw1)} | |
5881 | @tab @code{@var{c} = __MOR (@var{a}, @var{b})} | |
5882 | @tab @code{MOR @var{a},@var{b},@var{c}} | |
5883 | @item @code{uw1 __MPACKH (uh, uh)} | |
5884 | @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})} | |
5885 | @tab @code{MPACKH @var{a},@var{b},@var{c}} | |
5886 | @item @code{sw2 __MQADDHSS (sw2, sw2)} | |
5887 | @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})} | |
5888 | @tab @code{MQADDHSS @var{a},@var{b},@var{c}} | |
5889 | @item @code{uw2 __MQADDHUS (uw2, uw2)} | |
5890 | @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})} | |
5891 | @tab @code{MQADDHUS @var{a},@var{b},@var{c}} | |
5892 | @item @code{void __MQCPXIS (acc, sw2, sw2)} | |
5893 | @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})} | |
5894 | @tab @code{MQCPXIS @var{a},@var{b},@var{c}} | |
5895 | @item @code{void __MQCPXIU (acc, uw2, uw2)} | |
5896 | @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})} | |
5897 | @tab @code{MQCPXIU @var{a},@var{b},@var{c}} | |
5898 | @item @code{void __MQCPXRS (acc, sw2, sw2)} | |
5899 | @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})} | |
5900 | @tab @code{MQCPXRS @var{a},@var{b},@var{c}} | |
5901 | @item @code{void __MQCPXRU (acc, uw2, uw2)} | |
5902 | @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})} | |
5903 | @tab @code{MQCPXRU @var{a},@var{b},@var{c}} | |
5904 | @item @code{sw2 __MQLCLRHS (sw2, sw2)} | |
5905 | @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})} | |
5906 | @tab @code{MQLCLRHS @var{a},@var{b},@var{c}} | |
5907 | @item @code{sw2 __MQLMTHS (sw2, sw2)} | |
5908 | @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})} | |
5909 | @tab @code{MQLMTHS @var{a},@var{b},@var{c}} | |
5910 | @item @code{void __MQMACHS (acc, sw2, sw2)} | |
5911 | @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})} | |
5912 | @tab @code{MQMACHS @var{a},@var{b},@var{c}} | |
5913 | @item @code{void __MQMACHU (acc, uw2, uw2)} | |
5914 | @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})} | |
5915 | @tab @code{MQMACHU @var{a},@var{b},@var{c}} | |
5916 | @item @code{void __MQMACXHS (acc, sw2, sw2)} | |
5917 | @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})} | |
5918 | @tab @code{MQMACXHS @var{a},@var{b},@var{c}} | |
5919 | @item @code{void __MQMULHS (acc, sw2, sw2)} | |
5920 | @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})} | |
5921 | @tab @code{MQMULHS @var{a},@var{b},@var{c}} | |
5922 | @item @code{void __MQMULHU (acc, uw2, uw2)} | |
5923 | @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})} | |
5924 | @tab @code{MQMULHU @var{a},@var{b},@var{c}} | |
5925 | @item @code{void __MQMULXHS (acc, sw2, sw2)} | |
5926 | @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})} | |
5927 | @tab @code{MQMULXHS @var{a},@var{b},@var{c}} | |
5928 | @item @code{void __MQMULXHU (acc, uw2, uw2)} | |
5929 | @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})} | |
5930 | @tab @code{MQMULXHU @var{a},@var{b},@var{c}} | |
5931 | @item @code{sw2 __MQSATHS (sw2, sw2)} | |
5932 | @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})} | |
5933 | @tab @code{MQSATHS @var{a},@var{b},@var{c}} | |
5934 | @item @code{uw2 __MQSLLHI (uw2, int)} | |
5935 | @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})} | |
5936 | @tab @code{MQSLLHI @var{a},@var{b},@var{c}} | |
5937 | @item @code{sw2 __MQSRAHI (sw2, int)} | |
5938 | @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})} | |
5939 | @tab @code{MQSRAHI @var{a},@var{b},@var{c}} | |
5940 | @item @code{sw2 __MQSUBHSS (sw2, sw2)} | |
5941 | @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})} | |
5942 | @tab @code{MQSUBHSS @var{a},@var{b},@var{c}} | |
5943 | @item @code{uw2 __MQSUBHUS (uw2, uw2)} | |
5944 | @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})} | |
5945 | @tab @code{MQSUBHUS @var{a},@var{b},@var{c}} | |
5946 | @item @code{void __MQXMACHS (acc, sw2, sw2)} | |
5947 | @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})} | |
5948 | @tab @code{MQXMACHS @var{a},@var{b},@var{c}} | |
5949 | @item @code{void __MQXMACXHS (acc, sw2, sw2)} | |
5950 | @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})} | |
5951 | @tab @code{MQXMACXHS @var{a},@var{b},@var{c}} | |
5952 | @item @code{uw1 __MRDACC (acc)} | |
5953 | @tab @code{@var{b} = __MRDACC (@var{a})} | |
5954 | @tab @code{MRDACC @var{a},@var{b}} | |
5955 | @item @code{uw1 __MRDACCG (acc)} | |
5956 | @tab @code{@var{b} = __MRDACCG (@var{a})} | |
5957 | @tab @code{MRDACCG @var{a},@var{b}} | |
5958 | @item @code{uw1 __MROTLI (uw1, const)} | |
5959 | @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})} | |
5960 | @tab @code{MROTLI @var{a},#@var{b},@var{c}} | |
5961 | @item @code{uw1 __MROTRI (uw1, const)} | |
5962 | @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})} | |
5963 | @tab @code{MROTRI @var{a},#@var{b},@var{c}} | |
5964 | @item @code{sw1 __MSATHS (sw1, sw1)} | |
5965 | @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})} | |
5966 | @tab @code{MSATHS @var{a},@var{b},@var{c}} | |
5967 | @item @code{uw1 __MSATHU (uw1, uw1)} | |
5968 | @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})} | |
5969 | @tab @code{MSATHU @var{a},@var{b},@var{c}} | |
5970 | @item @code{uw1 __MSLLHI (uw1, const)} | |
5971 | @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})} | |
5972 | @tab @code{MSLLHI @var{a},#@var{b},@var{c}} | |
5973 | @item @code{sw1 __MSRAHI (sw1, const)} | |
5974 | @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})} | |
5975 | @tab @code{MSRAHI @var{a},#@var{b},@var{c}} | |
5976 | @item @code{uw1 __MSRLHI (uw1, const)} | |
5977 | @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})} | |
5978 | @tab @code{MSRLHI @var{a},#@var{b},@var{c}} | |
5979 | @item @code{void __MSUBACCS (acc, acc)} | |
5980 | @tab @code{__MSUBACCS (@var{b}, @var{a})} | |
5981 | @tab @code{MSUBACCS @var{a},@var{b}} | |
5982 | @item @code{sw1 __MSUBHSS (sw1, sw1)} | |
5983 | @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})} | |
5984 | @tab @code{MSUBHSS @var{a},@var{b},@var{c}} | |
5985 | @item @code{uw1 __MSUBHUS (uw1, uw1)} | |
5986 | @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})} | |
5987 | @tab @code{MSUBHUS @var{a},@var{b},@var{c}} | |
5988 | @item @code{void __MTRAP (void)} | |
5989 | @tab @code{__MTRAP ()} | |
5990 | @tab @code{MTRAP} | |
5991 | @item @code{uw2 __MUNPACKH (uw1)} | |
5992 | @tab @code{@var{b} = __MUNPACKH (@var{a})} | |
5993 | @tab @code{MUNPACKH @var{a},@var{b}} | |
5994 | @item @code{uw1 __MWCUT (uw2, uw1)} | |
5995 | @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})} | |
5996 | @tab @code{MWCUT @var{a},@var{b},@var{c}} | |
5997 | @item @code{void __MWTACC (acc, uw1)} | |
5998 | @tab @code{__MWTACC (@var{b}, @var{a})} | |
5999 | @tab @code{MWTACC @var{a},@var{b}} | |
6000 | @item @code{void __MWTACCG (acc, uw1)} | |
6001 | @tab @code{__MWTACCG (@var{b}, @var{a})} | |
6002 | @tab @code{MWTACCG @var{a},@var{b}} | |
6003 | @item @code{uw1 __MXOR (uw1, uw1)} | |
6004 | @tab @code{@var{c} = __MXOR (@var{a}, @var{b})} | |
6005 | @tab @code{MXOR @var{a},@var{b},@var{c}} | |
6006 | @end multitable | |
6007 | ||
6008 | @node Other Built-in Functions | |
6009 | @subsubsection Other Built-in Functions | |
6010 | ||
6011 | This section describes built-in functions that are not named after | |
6012 | a specific FR-V instruction. | |
6013 | ||
6014 | @table @code | |
6015 | @item sw2 __IACCreadll (iacc @var{reg}) | |
6016 | Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved | |
6017 | for future expansion and must be 0. | |
6018 | ||
6019 | @item sw1 __IACCreadl (iacc @var{reg}) | |
6020 | Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1. | |
6021 | Other values of @var{reg} are rejected as invalid. | |
6022 | ||
6023 | @item void __IACCsetll (iacc @var{reg}, sw2 @var{x}) | |
6024 | Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument | |
6025 | is reserved for future expansion and must be 0. | |
6026 | ||
6027 | @item void __IACCsetl (iacc @var{reg}, sw1 @var{x}) | |
6028 | Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg} | |
6029 | is 1. Other values of @var{reg} are rejected as invalid. | |
6030 | ||
6031 | @item void __data_prefetch0 (const void *@var{x}) | |
6032 | Use the @code{dcpl} instruction to load the contents of address @var{x} | |
6033 | into the data cache. | |
6034 | ||
6035 | @item void __data_prefetch (const void *@var{x}) | |
6036 | Use the @code{nldub} instruction to load the contents of address @var{x} | |
6037 | into the data cache. The instruction will be issued in slot I1@. | |
6038 | @end table | |
6039 | ||
0975678f JM |
6040 | @node X86 Built-in Functions |
6041 | @subsection X86 Built-in Functions | |
6042 | ||
6043 | These built-in functions are available for the i386 and x86-64 family | |
6044 | of computers, depending on the command-line switches used. | |
6045 | ||
6046 | The following machine modes are available for use with MMX built-in functions | |
333c8841 AH |
6047 | (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers, |
6048 | @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a | |
6049 | vector of eight 8-bit integers. Some of the built-in functions operate on | |
6050 | MMX registers as a whole 64-bit entity, these use @code{DI} as their mode. | |
0975678f JM |
6051 | |
6052 | If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector | |
333c8841 | 6053 | of two 32-bit floating point values. |
0975678f | 6054 | |
333c8841 AH |
6055 | If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit |
6056 | floating point values. Some instructions use a vector of four 32-bit | |
0975678f | 6057 | integers, these use @code{V4SI}. Finally, some instructions operate on an |
333c8841 | 6058 | entire vector register, interpreting it as a 128-bit integer, these use mode |
0975678f JM |
6059 | @code{TI}. |
6060 | ||
6061 | The following built-in functions are made available by @option{-mmmx}. | |
6062 | All of them generate the machine instruction that is part of the name. | |
6063 | ||
3ab51846 | 6064 | @smallexample |
0975678f JM |
6065 | v8qi __builtin_ia32_paddb (v8qi, v8qi) |
6066 | v4hi __builtin_ia32_paddw (v4hi, v4hi) | |
6067 | v2si __builtin_ia32_paddd (v2si, v2si) | |
6068 | v8qi __builtin_ia32_psubb (v8qi, v8qi) | |
6069 | v4hi __builtin_ia32_psubw (v4hi, v4hi) | |
6070 | v2si __builtin_ia32_psubd (v2si, v2si) | |
6071 | v8qi __builtin_ia32_paddsb (v8qi, v8qi) | |
6072 | v4hi __builtin_ia32_paddsw (v4hi, v4hi) | |
6073 | v8qi __builtin_ia32_psubsb (v8qi, v8qi) | |
6074 | v4hi __builtin_ia32_psubsw (v4hi, v4hi) | |
6075 | v8qi __builtin_ia32_paddusb (v8qi, v8qi) | |
6076 | v4hi __builtin_ia32_paddusw (v4hi, v4hi) | |
6077 | v8qi __builtin_ia32_psubusb (v8qi, v8qi) | |
6078 | v4hi __builtin_ia32_psubusw (v4hi, v4hi) | |
6079 | v4hi __builtin_ia32_pmullw (v4hi, v4hi) | |
6080 | v4hi __builtin_ia32_pmulhw (v4hi, v4hi) | |
6081 | di __builtin_ia32_pand (di, di) | |
6082 | di __builtin_ia32_pandn (di,di) | |
6083 | di __builtin_ia32_por (di, di) | |
6084 | di __builtin_ia32_pxor (di, di) | |
6085 | v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi) | |
6086 | v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi) | |
6087 | v2si __builtin_ia32_pcmpeqd (v2si, v2si) | |
6088 | v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi) | |
6089 | v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi) | |
6090 | v2si __builtin_ia32_pcmpgtd (v2si, v2si) | |
6091 | v8qi __builtin_ia32_punpckhbw (v8qi, v8qi) | |
6092 | v4hi __builtin_ia32_punpckhwd (v4hi, v4hi) | |
6093 | v2si __builtin_ia32_punpckhdq (v2si, v2si) | |
6094 | v8qi __builtin_ia32_punpcklbw (v8qi, v8qi) | |
6095 | v4hi __builtin_ia32_punpcklwd (v4hi, v4hi) | |
6096 | v2si __builtin_ia32_punpckldq (v2si, v2si) | |
6097 | v8qi __builtin_ia32_packsswb (v4hi, v4hi) | |
6098 | v4hi __builtin_ia32_packssdw (v2si, v2si) | |
6099 | v8qi __builtin_ia32_packuswb (v4hi, v4hi) | |
3ab51846 | 6100 | @end smallexample |
0975678f JM |
6101 | |
6102 | The following built-in functions are made available either with | |
6103 | @option{-msse}, or with a combination of @option{-m3dnow} and | |
6104 | @option{-march=athlon}. All of them generate the machine | |
6105 | instruction that is part of the name. | |
6106 | ||
3ab51846 | 6107 | @smallexample |
0975678f JM |
6108 | v4hi __builtin_ia32_pmulhuw (v4hi, v4hi) |
6109 | v8qi __builtin_ia32_pavgb (v8qi, v8qi) | |
6110 | v4hi __builtin_ia32_pavgw (v4hi, v4hi) | |
6111 | v4hi __builtin_ia32_psadbw (v8qi, v8qi) | |
6112 | v8qi __builtin_ia32_pmaxub (v8qi, v8qi) | |
6113 | v4hi __builtin_ia32_pmaxsw (v4hi, v4hi) | |
6114 | v8qi __builtin_ia32_pminub (v8qi, v8qi) | |
6115 | v4hi __builtin_ia32_pminsw (v4hi, v4hi) | |
6116 | int __builtin_ia32_pextrw (v4hi, int) | |
6117 | v4hi __builtin_ia32_pinsrw (v4hi, int, int) | |
6118 | int __builtin_ia32_pmovmskb (v8qi) | |
6119 | void __builtin_ia32_maskmovq (v8qi, v8qi, char *) | |
6120 | void __builtin_ia32_movntq (di *, di) | |
6121 | void __builtin_ia32_sfence (void) | |
3ab51846 | 6122 | @end smallexample |
0975678f JM |
6123 | |
6124 | The following built-in functions are available when @option{-msse} is used. | |
6125 | All of them generate the machine instruction that is part of the name. | |
6126 | ||
3ab51846 | 6127 | @smallexample |
0975678f JM |
6128 | int __builtin_ia32_comieq (v4sf, v4sf) |
6129 | int __builtin_ia32_comineq (v4sf, v4sf) | |
6130 | int __builtin_ia32_comilt (v4sf, v4sf) | |
6131 | int __builtin_ia32_comile (v4sf, v4sf) | |
6132 | int __builtin_ia32_comigt (v4sf, v4sf) | |
6133 | int __builtin_ia32_comige (v4sf, v4sf) | |
6134 | int __builtin_ia32_ucomieq (v4sf, v4sf) | |
6135 | int __builtin_ia32_ucomineq (v4sf, v4sf) | |
6136 | int __builtin_ia32_ucomilt (v4sf, v4sf) | |
6137 | int __builtin_ia32_ucomile (v4sf, v4sf) | |
6138 | int __builtin_ia32_ucomigt (v4sf, v4sf) | |
6139 | int __builtin_ia32_ucomige (v4sf, v4sf) | |
6140 | v4sf __builtin_ia32_addps (v4sf, v4sf) | |
6141 | v4sf __builtin_ia32_subps (v4sf, v4sf) | |
6142 | v4sf __builtin_ia32_mulps (v4sf, v4sf) | |
6143 | v4sf __builtin_ia32_divps (v4sf, v4sf) | |
6144 | v4sf __builtin_ia32_addss (v4sf, v4sf) | |
6145 | v4sf __builtin_ia32_subss (v4sf, v4sf) | |
6146 | v4sf __builtin_ia32_mulss (v4sf, v4sf) | |
6147 | v4sf __builtin_ia32_divss (v4sf, v4sf) | |
6148 | v4si __builtin_ia32_cmpeqps (v4sf, v4sf) | |
6149 | v4si __builtin_ia32_cmpltps (v4sf, v4sf) | |
6150 | v4si __builtin_ia32_cmpleps (v4sf, v4sf) | |
6151 | v4si __builtin_ia32_cmpgtps (v4sf, v4sf) | |
6152 | v4si __builtin_ia32_cmpgeps (v4sf, v4sf) | |
6153 | v4si __builtin_ia32_cmpunordps (v4sf, v4sf) | |
6154 | v4si __builtin_ia32_cmpneqps (v4sf, v4sf) | |
6155 | v4si __builtin_ia32_cmpnltps (v4sf, v4sf) | |
6156 | v4si __builtin_ia32_cmpnleps (v4sf, v4sf) | |
6157 | v4si __builtin_ia32_cmpngtps (v4sf, v4sf) | |
6158 | v4si __builtin_ia32_cmpngeps (v4sf, v4sf) | |
6159 | v4si __builtin_ia32_cmpordps (v4sf, v4sf) | |
6160 | v4si __builtin_ia32_cmpeqss (v4sf, v4sf) | |
6161 | v4si __builtin_ia32_cmpltss (v4sf, v4sf) | |
6162 | v4si __builtin_ia32_cmpless (v4sf, v4sf) | |
0975678f JM |
6163 | v4si __builtin_ia32_cmpunordss (v4sf, v4sf) |
6164 | v4si __builtin_ia32_cmpneqss (v4sf, v4sf) | |
6165 | v4si __builtin_ia32_cmpnlts (v4sf, v4sf) | |
6166 | v4si __builtin_ia32_cmpnless (v4sf, v4sf) | |
0975678f JM |
6167 | v4si __builtin_ia32_cmpordss (v4sf, v4sf) |
6168 | v4sf __builtin_ia32_maxps (v4sf, v4sf) | |
6169 | v4sf __builtin_ia32_maxss (v4sf, v4sf) | |
6170 | v4sf __builtin_ia32_minps (v4sf, v4sf) | |
6171 | v4sf __builtin_ia32_minss (v4sf, v4sf) | |
6172 | v4sf __builtin_ia32_andps (v4sf, v4sf) | |
6173 | v4sf __builtin_ia32_andnps (v4sf, v4sf) | |
6174 | v4sf __builtin_ia32_orps (v4sf, v4sf) | |
6175 | v4sf __builtin_ia32_xorps (v4sf, v4sf) | |
6176 | v4sf __builtin_ia32_movss (v4sf, v4sf) | |
6177 | v4sf __builtin_ia32_movhlps (v4sf, v4sf) | |
6178 | v4sf __builtin_ia32_movlhps (v4sf, v4sf) | |
6179 | v4sf __builtin_ia32_unpckhps (v4sf, v4sf) | |
6180 | v4sf __builtin_ia32_unpcklps (v4sf, v4sf) | |
6181 | v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si) | |
6182 | v4sf __builtin_ia32_cvtsi2ss (v4sf, int) | |
6183 | v2si __builtin_ia32_cvtps2pi (v4sf) | |
6184 | int __builtin_ia32_cvtss2si (v4sf) | |
6185 | v2si __builtin_ia32_cvttps2pi (v4sf) | |
6186 | int __builtin_ia32_cvttss2si (v4sf) | |
6187 | v4sf __builtin_ia32_rcpps (v4sf) | |
6188 | v4sf __builtin_ia32_rsqrtps (v4sf) | |
6189 | v4sf __builtin_ia32_sqrtps (v4sf) | |
6190 | v4sf __builtin_ia32_rcpss (v4sf) | |
6191 | v4sf __builtin_ia32_rsqrtss (v4sf) | |
6192 | v4sf __builtin_ia32_sqrtss (v4sf) | |
6193 | v4sf __builtin_ia32_shufps (v4sf, v4sf, int) | |
6194 | void __builtin_ia32_movntps (float *, v4sf) | |
6195 | int __builtin_ia32_movmskps (v4sf) | |
3ab51846 | 6196 | @end smallexample |
0975678f JM |
6197 | |
6198 | The following built-in functions are available when @option{-msse} is used. | |
6199 | ||
6200 | @table @code | |
6201 | @item v4sf __builtin_ia32_loadaps (float *) | |
6202 | Generates the @code{movaps} machine instruction as a load from memory. | |
6203 | @item void __builtin_ia32_storeaps (float *, v4sf) | |
6204 | Generates the @code{movaps} machine instruction as a store to memory. | |
6205 | @item v4sf __builtin_ia32_loadups (float *) | |
6206 | Generates the @code{movups} machine instruction as a load from memory. | |
6207 | @item void __builtin_ia32_storeups (float *, v4sf) | |
6208 | Generates the @code{movups} machine instruction as a store to memory. | |
6209 | @item v4sf __builtin_ia32_loadsss (float *) | |
6210 | Generates the @code{movss} machine instruction as a load from memory. | |
6211 | @item void __builtin_ia32_storess (float *, v4sf) | |
6212 | Generates the @code{movss} machine instruction as a store to memory. | |
6213 | @item v4sf __builtin_ia32_loadhps (v4sf, v2si *) | |
6214 | Generates the @code{movhps} machine instruction as a load from memory. | |
6215 | @item v4sf __builtin_ia32_loadlps (v4sf, v2si *) | |
6216 | Generates the @code{movlps} machine instruction as a load from memory | |
6217 | @item void __builtin_ia32_storehps (v4sf, v2si *) | |
6218 | Generates the @code{movhps} machine instruction as a store to memory. | |
6219 | @item void __builtin_ia32_storelps (v4sf, v2si *) | |
6220 | Generates the @code{movlps} machine instruction as a store to memory. | |
6221 | @end table | |
6222 | ||
9e200aaf | 6223 | The following built-in functions are available when @option{-msse3} is used. |
22c7c85e L |
6224 | All of them generate the machine instruction that is part of the name. |
6225 | ||
3ab51846 | 6226 | @smallexample |
22c7c85e L |
6227 | v2df __builtin_ia32_addsubpd (v2df, v2df) |
6228 | v2df __builtin_ia32_addsubps (v2df, v2df) | |
6229 | v2df __builtin_ia32_haddpd (v2df, v2df) | |
6230 | v2df __builtin_ia32_haddps (v2df, v2df) | |
6231 | v2df __builtin_ia32_hsubpd (v2df, v2df) | |
6232 | v2df __builtin_ia32_hsubps (v2df, v2df) | |
6233 | v16qi __builtin_ia32_lddqu (char const *) | |
6234 | void __builtin_ia32_monitor (void *, unsigned int, unsigned int) | |
6235 | v2df __builtin_ia32_movddup (v2df) | |
6236 | v4sf __builtin_ia32_movshdup (v4sf) | |
6237 | v4sf __builtin_ia32_movsldup (v4sf) | |
6238 | void __builtin_ia32_mwait (unsigned int, unsigned int) | |
3ab51846 | 6239 | @end smallexample |
22c7c85e | 6240 | |
9e200aaf | 6241 | The following built-in functions are available when @option{-msse3} is used. |
22c7c85e L |
6242 | |
6243 | @table @code | |
6244 | @item v2df __builtin_ia32_loadddup (double const *) | |
6245 | Generates the @code{movddup} machine instruction as a load from memory. | |
6246 | @end table | |
6247 | ||
0975678f JM |
6248 | The following built-in functions are available when @option{-m3dnow} is used. |
6249 | All of them generate the machine instruction that is part of the name. | |
6250 | ||
3ab51846 | 6251 | @smallexample |
0975678f JM |
6252 | void __builtin_ia32_femms (void) |
6253 | v8qi __builtin_ia32_pavgusb (v8qi, v8qi) | |
6254 | v2si __builtin_ia32_pf2id (v2sf) | |
6255 | v2sf __builtin_ia32_pfacc (v2sf, v2sf) | |
6256 | v2sf __builtin_ia32_pfadd (v2sf, v2sf) | |
6257 | v2si __builtin_ia32_pfcmpeq (v2sf, v2sf) | |
6258 | v2si __builtin_ia32_pfcmpge (v2sf, v2sf) | |
6259 | v2si __builtin_ia32_pfcmpgt (v2sf, v2sf) | |
6260 | v2sf __builtin_ia32_pfmax (v2sf, v2sf) | |
6261 | v2sf __builtin_ia32_pfmin (v2sf, v2sf) | |
6262 | v2sf __builtin_ia32_pfmul (v2sf, v2sf) | |
6263 | v2sf __builtin_ia32_pfrcp (v2sf) | |
6264 | v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf) | |
6265 | v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf) | |
6266 | v2sf __builtin_ia32_pfrsqrt (v2sf) | |
6267 | v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf) | |
6268 | v2sf __builtin_ia32_pfsub (v2sf, v2sf) | |
6269 | v2sf __builtin_ia32_pfsubr (v2sf, v2sf) | |
6270 | v2sf __builtin_ia32_pi2fd (v2si) | |
6271 | v4hi __builtin_ia32_pmulhrw (v4hi, v4hi) | |
3ab51846 | 6272 | @end smallexample |
0975678f JM |
6273 | |
6274 | The following built-in functions are available when both @option{-m3dnow} | |
6275 | and @option{-march=athlon} are used. All of them generate the machine | |
6276 | instruction that is part of the name. | |
6277 | ||
3ab51846 | 6278 | @smallexample |
0975678f JM |
6279 | v2si __builtin_ia32_pf2iw (v2sf) |
6280 | v2sf __builtin_ia32_pfnacc (v2sf, v2sf) | |
6281 | v2sf __builtin_ia32_pfpnacc (v2sf, v2sf) | |
6282 | v2sf __builtin_ia32_pi2fw (v2si) | |
6283 | v2sf __builtin_ia32_pswapdsf (v2sf) | |
6284 | v2si __builtin_ia32_pswapdsi (v2si) | |
3ab51846 | 6285 | @end smallexample |
0975678f | 6286 | |
d840bfd3 CF |
6287 | @node MIPS Paired-Single Support |
6288 | @subsection MIPS Paired-Single Support | |
6289 | ||
6290 | The MIPS64 architecture includes a number of instructions that | |
6291 | operate on pairs of single-precision floating-point values. | |
6292 | Each pair is packed into a 64-bit floating-point register, | |
6293 | with one element being designated the ``upper half'' and | |
6294 | the other being designated the ``lower half''. | |
6295 | ||
6296 | GCC supports paired-single operations using both the generic | |
6297 | vector extensions (@pxref{Vector Extensions}) and a collection of | |
6298 | MIPS-specific built-in functions. Both kinds of support are | |
6299 | enabled by the @option{-mpaired-single} command-line option. | |
6300 | ||
6301 | The vector type associated with paired-single values is usually | |
6302 | called @code{v2sf}. It can be defined in C as follows: | |
6303 | ||
6304 | @smallexample | |
6305 | typedef float v2sf __attribute__ ((vector_size (8))); | |
6306 | @end smallexample | |
6307 | ||
6308 | @code{v2sf} values are initialized in the same way as aggregates. | |
6309 | For example: | |
6310 | ||
6311 | @smallexample | |
6312 | v2sf a = @{1.5, 9.1@}; | |
6313 | v2sf b; | |
6314 | float e, f; | |
6315 | b = (v2sf) @{e, f@}; | |
6316 | @end smallexample | |
6317 | ||
6318 | @emph{Note:} The CPU's endianness determines which value is stored in | |
6319 | the upper half of a register and which value is stored in the lower half. | |
6320 | On little-endian targets, the first value is the lower one and the second | |
6321 | value is the upper one. The opposite order applies to big-endian targets. | |
6322 | For example, the code above will set the lower half of @code{a} to | |
6323 | @code{1.5} on little-endian targets and @code{9.1} on big-endian targets. | |
6324 | ||
6325 | @menu | |
6326 | * Paired-Single Arithmetic:: | |
6327 | * Paired-Single Built-in Functions:: | |
6328 | * MIPS-3D Built-in Functions:: | |
6329 | @end menu | |
6330 | ||
6331 | @node Paired-Single Arithmetic | |
6332 | @subsubsection Paired-Single Arithmetic | |
6333 | ||
6334 | The table below lists the @code{v2sf} operations for which hardware | |
6335 | support exists. @code{a}, @code{b} and @code{c} are @code{v2sf} | |
6336 | values and @code{x} is an integral value. | |
6337 | ||
6338 | @multitable @columnfractions .50 .50 | |
6339 | @item C code @tab MIPS instruction | |
6340 | @item @code{a + b} @tab @code{add.ps} | |
6341 | @item @code{a - b} @tab @code{sub.ps} | |
6342 | @item @code{-a} @tab @code{neg.ps} | |
6343 | @item @code{a * b} @tab @code{mul.ps} | |
6344 | @item @code{a * b + c} @tab @code{madd.ps} | |
6345 | @item @code{a * b - c} @tab @code{msub.ps} | |
6346 | @item @code{-(a * b + c)} @tab @code{nmadd.ps} | |
6347 | @item @code{-(a * b - c)} @tab @code{nmsub.ps} | |
6348 | @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps} | |
6349 | @end multitable | |
6350 | ||
6351 | Note that the multiply-accumulate instructions can be disabled | |
6352 | using the command-line option @code{-mno-fused-madd}. | |
6353 | ||
6354 | @node Paired-Single Built-in Functions | |
6355 | @subsubsection Paired-Single Built-in Functions | |
6356 | ||
6357 | The following paired-single functions map directly to a particular | |
6358 | MIPS instruction. Please refer to the architecture specification | |
6359 | for details on what each instruction does. | |
6360 | ||
6361 | @table @code | |
6362 | @item v2sf __builtin_mips_pll_ps (v2sf, v2sf) | |
6363 | Pair lower lower (@code{pll.ps}). | |
6364 | ||
6365 | @item v2sf __builtin_mips_pul_ps (v2sf, v2sf) | |
6366 | Pair upper lower (@code{pul.ps}). | |
6367 | ||
6368 | @item v2sf __builtin_mips_plu_ps (v2sf, v2sf) | |
6369 | Pair lower upper (@code{plu.ps}). | |
6370 | ||
6371 | @item v2sf __builtin_mips_puu_ps (v2sf, v2sf) | |
6372 | Pair upper upper (@code{puu.ps}). | |
6373 | ||
6374 | @item v2sf __builtin_mips_cvt_ps_s (float, float) | |
6375 | Convert pair to paired single (@code{cvt.ps.s}). | |
6376 | ||
6377 | @item float __builtin_mips_cvt_s_pl (v2sf) | |
6378 | Convert pair lower to single (@code{cvt.s.pl}). | |
6379 | ||
6380 | @item float __builtin_mips_cvt_s_pu (v2sf) | |
6381 | Convert pair upper to single (@code{cvt.s.pu}). | |
6382 | ||
6383 | @item v2sf __builtin_mips_abs_ps (v2sf) | |
6384 | Absolute value (@code{abs.ps}). | |
6385 | ||
6386 | @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int) | |
6387 | Align variable (@code{alnv.ps}). | |
6388 | ||
6389 | @emph{Note:} The value of the third parameter must be 0 or 4 | |
6390 | modulo 8, otherwise the result will be unpredictable. Please read the | |
6391 | instruction description for details. | |
6392 | @end table | |
6393 | ||
6394 | The following multi-instruction functions are also available. | |
6395 | In each case, @var{cond} can be any of the 16 floating-point conditions: | |
6396 | @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult}, | |
6397 | @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl}, | |
6398 | @code{lt}, @code{nge}, @code{le} or @code{ngt}. | |
6399 | ||
6400 | @table @code | |
6401 | @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
6402 | @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
6403 | Conditional move based on floating point comparison (@code{c.@var{cond}.ps}, | |
6404 | @code{movt.ps}/@code{movf.ps}). | |
6405 | ||
6406 | The @code{movt} functions return the value @var{x} computed by: | |
6407 | ||
6408 | @smallexample | |
6409 | c.@var{cond}.ps @var{cc},@var{a},@var{b} | |
6410 | mov.ps @var{x},@var{c} | |
6411 | movt.ps @var{x},@var{d},@var{cc} | |
6412 | @end smallexample | |
6413 | ||
6414 | The @code{movf} functions are similar but use @code{movf.ps} instead | |
6415 | of @code{movt.ps}. | |
6416 | ||
6417 | @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
6418 | @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
6419 | Comparison of two paired-single values (@code{c.@var{cond}.ps}, | |
6420 | @code{bc1t}/@code{bc1f}). | |
6421 | ||
6422 | These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps} | |
6423 | and return either the upper or lower half of the result. For example: | |
6424 | ||
6425 | @smallexample | |
6426 | v2sf a, b; | |
6427 | if (__builtin_mips_upper_c_eq_ps (a, b)) | |
6428 | upper_halves_are_equal (); | |
6429 | else | |
6430 | upper_halves_are_unequal (); | |
6431 | ||
6432 | if (__builtin_mips_lower_c_eq_ps (a, b)) | |
6433 | lower_halves_are_equal (); | |
6434 | else | |
6435 | lower_halves_are_unequal (); | |
6436 | @end smallexample | |
6437 | @end table | |
6438 | ||
6439 | @node MIPS-3D Built-in Functions | |
6440 | @subsubsection MIPS-3D Built-in Functions | |
6441 | ||
6442 | The MIPS-3D Application-Specific Extension (ASE) includes additional | |
6443 | paired-single instructions that are designed to improve the performance | |
6444 | of 3D graphics operations. Support for these instructions is controlled | |
6445 | by the @option{-mips3d} command-line option. | |
6446 | ||
6447 | The functions listed below map directly to a particular MIPS-3D | |
6448 | instruction. Please refer to the architecture specification for | |
6449 | more details on what each instruction does. | |
6450 | ||
6451 | @table @code | |
6452 | @item v2sf __builtin_mips_addr_ps (v2sf, v2sf) | |
6453 | Reduction add (@code{addr.ps}). | |
6454 | ||
6455 | @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf) | |
6456 | Reduction multiply (@code{mulr.ps}). | |
6457 | ||
6458 | @item v2sf __builtin_mips_cvt_pw_ps (v2sf) | |
6459 | Convert paired single to paired word (@code{cvt.pw.ps}). | |
6460 | ||
6461 | @item v2sf __builtin_mips_cvt_ps_pw (v2sf) | |
6462 | Convert paired word to paired single (@code{cvt.ps.pw}). | |
6463 | ||
6464 | @item float __builtin_mips_recip1_s (float) | |
6465 | @itemx double __builtin_mips_recip1_d (double) | |
6466 | @itemx v2sf __builtin_mips_recip1_ps (v2sf) | |
6467 | Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}). | |
6468 | ||
6469 | @item float __builtin_mips_recip2_s (float, float) | |
6470 | @itemx double __builtin_mips_recip2_d (double, double) | |
6471 | @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf) | |
6472 | Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}). | |
6473 | ||
6474 | @item float __builtin_mips_rsqrt1_s (float) | |
6475 | @itemx double __builtin_mips_rsqrt1_d (double) | |
6476 | @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf) | |
6477 | Reduced precision reciprocal square root (sequence step 1) | |
6478 | (@code{rsqrt1.@var{fmt}}). | |
6479 | ||
6480 | @item float __builtin_mips_rsqrt2_s (float, float) | |
6481 | @itemx double __builtin_mips_rsqrt2_d (double, double) | |
6482 | @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf) | |
6483 | Reduced precision reciprocal square root (sequence step 2) | |
6484 | (@code{rsqrt2.@var{fmt}}). | |
6485 | @end table | |
6486 | ||
6487 | The following multi-instruction functions are also available. | |
6488 | In each case, @var{cond} can be any of the 16 floating-point conditions: | |
6489 | @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult}, | |
6490 | @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, | |
6491 | @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}. | |
6492 | ||
6493 | @table @code | |
6494 | @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b}) | |
6495 | @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b}) | |
6496 | Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}}, | |
6497 | @code{bc1t}/@code{bc1f}). | |
6498 | ||
6499 | These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s} | |
6500 | or @code{cabs.@var{cond}.d} and return the result as a boolean value. | |
6501 | For example: | |
6502 | ||
6503 | @smallexample | |
6504 | float a, b; | |
6505 | if (__builtin_mips_cabs_eq_s (a, b)) | |
6506 | true (); | |
6507 | else | |
6508 | false (); | |
6509 | @end smallexample | |
6510 | ||
6511 | @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
6512 | @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
6513 | Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps}, | |
6514 | @code{bc1t}/@code{bc1f}). | |
6515 | ||
6516 | These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps} | |
6517 | and return either the upper or lower half of the result. For example: | |
6518 | ||
6519 | @smallexample | |
6520 | v2sf a, b; | |
6521 | if (__builtin_mips_upper_cabs_eq_ps (a, b)) | |
6522 | upper_halves_are_equal (); | |
6523 | else | |
6524 | upper_halves_are_unequal (); | |
6525 | ||
6526 | if (__builtin_mips_lower_cabs_eq_ps (a, b)) | |
6527 | lower_halves_are_equal (); | |
6528 | else | |
6529 | lower_halves_are_unequal (); | |
6530 | @end smallexample | |
6531 | ||
6532 | @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
6533 | @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
6534 | Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps}, | |
6535 | @code{movt.ps}/@code{movf.ps}). | |
6536 | ||
6537 | The @code{movt} functions return the value @var{x} computed by: | |
6538 | ||
6539 | @smallexample | |
6540 | cabs.@var{cond}.ps @var{cc},@var{a},@var{b} | |
6541 | mov.ps @var{x},@var{c} | |
6542 | movt.ps @var{x},@var{d},@var{cc} | |
6543 | @end smallexample | |
6544 | ||
6545 | The @code{movf} functions are similar but use @code{movf.ps} instead | |
6546 | of @code{movt.ps}. | |
6547 | ||
6548 | @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
6549 | @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
6550 | @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
6551 | @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
6552 | Comparison of two paired-single values | |
6553 | (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps}, | |
6554 | @code{bc1any2t}/@code{bc1any2f}). | |
6555 | ||
6556 | These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps} | |
6557 | or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either | |
6558 | result is true and the @code{all} forms return true if both results are true. | |
6559 | For example: | |
6560 | ||
6561 | @smallexample | |
6562 | v2sf a, b; | |
6563 | if (__builtin_mips_any_c_eq_ps (a, b)) | |
6564 | one_is_true (); | |
6565 | else | |
6566 | both_are_false (); | |
6567 | ||
6568 | if (__builtin_mips_all_c_eq_ps (a, b)) | |
6569 | both_are_true (); | |
6570 | else | |
6571 | one_is_false (); | |
6572 | @end smallexample | |
6573 | ||
6574 | @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
6575 | @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
6576 | @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
6577 | @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
6578 | Comparison of four paired-single values | |
6579 | (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps}, | |
6580 | @code{bc1any4t}/@code{bc1any4f}). | |
6581 | ||
6582 | These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps} | |
6583 | to compare @var{a} with @var{b} and to compare @var{c} with @var{d}. | |
6584 | The @code{any} forms return true if any of the four results are true | |
6585 | and the @code{all} forms return true if all four results are true. | |
6586 | For example: | |
6587 | ||
6588 | @smallexample | |
6589 | v2sf a, b, c, d; | |
6590 | if (__builtin_mips_any_c_eq_4s (a, b, c, d)) | |
6591 | some_are_true (); | |
6592 | else | |
6593 | all_are_false (); | |
6594 | ||
6595 | if (__builtin_mips_all_c_eq_4s (a, b, c, d)) | |
6596 | all_are_true (); | |
6597 | else | |
6598 | some_are_false (); | |
6599 | @end smallexample | |
6600 | @end table | |
6601 | ||
333c8841 AH |
6602 | @node PowerPC AltiVec Built-in Functions |
6603 | @subsection PowerPC AltiVec Built-in Functions | |
6604 | ||
b0b343db JJ |
6605 | GCC provides an interface for the PowerPC family of processors to access |
6606 | the AltiVec operations described in Motorola's AltiVec Programming | |
6607 | Interface Manual. The interface is made available by including | |
6608 | @code{<altivec.h>} and using @option{-maltivec} and | |
6609 | @option{-mabi=altivec}. The interface supports the following vector | |
6610 | types. | |
333c8841 | 6611 | |
b0b343db JJ |
6612 | @smallexample |
6613 | vector unsigned char | |
6614 | vector signed char | |
6615 | vector bool char | |
333c8841 | 6616 | |
b0b343db JJ |
6617 | vector unsigned short |
6618 | vector signed short | |
6619 | vector bool short | |
6620 | vector pixel | |
6621 | ||
6622 | vector unsigned int | |
6623 | vector signed int | |
6624 | vector bool int | |
6625 | vector float | |
6626 | @end smallexample | |
6627 | ||
6628 | GCC's implementation of the high-level language interface available from | |
6629 | C and C++ code differs from Motorola's documentation in several ways. | |
6630 | ||
6631 | @itemize @bullet | |
6632 | ||
6633 | @item | |
6634 | A vector constant is a list of constant expressions within curly braces. | |
6635 | ||
6636 | @item | |
6637 | A vector initializer requires no cast if the vector constant is of the | |
6638 | same type as the variable it is initializing. | |
333c8841 | 6639 | |
b0b343db | 6640 | @item |
5edea4c6 JJ |
6641 | If @code{signed} or @code{unsigned} is omitted, the signedness of the |
6642 | vector type is the default signedness of the base type. The default | |
6643 | varies depending on the operating system, so a portable program should | |
6644 | always specify the signedness. | |
4e6e4e4c JJ |
6645 | |
6646 | @item | |
6647 | Compiling with @option{-maltivec} adds keywords @code{__vector}, | |
6648 | @code{__pixel}, and @code{__bool}. Macros @option{vector}, | |
6649 | @code{pixel}, and @code{bool} are defined in @code{<altivec.h>} and can | |
6650 | be undefined. | |
6651 | ||
6652 | @item | |
6653 | GCC allows using a @code{typedef} name as the type specifier for a | |
6654 | vector type. | |
b0b343db JJ |
6655 | |
6656 | @item | |
6657 | For C, overloaded functions are implemented with macros so the following | |
6658 | does not work: | |
90989b26 AH |
6659 | |
6660 | @smallexample | |
8254cb45 | 6661 | vec_add ((vector signed int)@{1, 2, 3, 4@}, foo); |
90989b26 AH |
6662 | @end smallexample |
6663 | ||
b0b343db JJ |
6664 | Since @code{vec_add} is a macro, the vector constant in the example |
6665 | is treated as four separate arguments. Wrap the entire argument in | |
6666 | parentheses for this to work. | |
6667 | @end itemize | |
90989b26 | 6668 | |
ae4b4a02 AH |
6669 | @emph{Note:} Only the @code{<altivec.h>} interface is supported. |
6670 | Internally, GCC uses built-in functions to achieve the functionality in | |
6671 | the aforementioned header file, but they are not supported and are | |
6672 | subject to change without notice. | |
6673 | ||
b0b343db JJ |
6674 | The following interfaces are supported for the generic and specific |
6675 | AltiVec operations and the AltiVec predicates. In cases where there | |
6676 | is a direct mapping between generic and specific operations, only the | |
6677 | generic names are shown here, although the specific operations can also | |
6678 | be used. | |
333c8841 | 6679 | |
b0b343db JJ |
6680 | Arguments that are documented as @code{const int} require literal |
6681 | integral values within the range required for that operation. | |
333c8841 | 6682 | |
b0b343db JJ |
6683 | @smallexample |
6684 | vector signed char vec_abs (vector signed char); | |
6685 | vector signed short vec_abs (vector signed short); | |
6686 | vector signed int vec_abs (vector signed int); | |
6687 | vector float vec_abs (vector float); | |
333c8841 | 6688 | |
b0b343db JJ |
6689 | vector signed char vec_abss (vector signed char); |
6690 | vector signed short vec_abss (vector signed short); | |
6691 | vector signed int vec_abss (vector signed int); | |
333c8841 | 6692 | |
b0b343db JJ |
6693 | vector signed char vec_add (vector bool char, vector signed char); |
6694 | vector signed char vec_add (vector signed char, vector bool char); | |
6695 | vector signed char vec_add (vector signed char, vector signed char); | |
6696 | vector unsigned char vec_add (vector bool char, vector unsigned char); | |
6697 | vector unsigned char vec_add (vector unsigned char, vector bool char); | |
924fcc4e JM |
6698 | vector unsigned char vec_add (vector unsigned char, |
6699 | vector unsigned char); | |
b0b343db JJ |
6700 | vector signed short vec_add (vector bool short, vector signed short); |
6701 | vector signed short vec_add (vector signed short, vector bool short); | |
333c8841 | 6702 | vector signed short vec_add (vector signed short, vector signed short); |
b0b343db | 6703 | vector unsigned short vec_add (vector bool short, |
924fcc4e JM |
6704 | vector unsigned short); |
6705 | vector unsigned short vec_add (vector unsigned short, | |
b0b343db | 6706 | vector bool short); |
6e5bb5ad JM |
6707 | vector unsigned short vec_add (vector unsigned short, |
6708 | vector unsigned short); | |
b0b343db JJ |
6709 | vector signed int vec_add (vector bool int, vector signed int); |
6710 | vector signed int vec_add (vector signed int, vector bool int); | |
333c8841 | 6711 | vector signed int vec_add (vector signed int, vector signed int); |
b0b343db JJ |
6712 | vector unsigned int vec_add (vector bool int, vector unsigned int); |
6713 | vector unsigned int vec_add (vector unsigned int, vector bool int); | |
333c8841 AH |
6714 | vector unsigned int vec_add (vector unsigned int, vector unsigned int); |
6715 | vector float vec_add (vector float, vector float); | |
6716 | ||
b0b343db JJ |
6717 | vector float vec_vaddfp (vector float, vector float); |
6718 | ||
6719 | vector signed int vec_vadduwm (vector bool int, vector signed int); | |
6720 | vector signed int vec_vadduwm (vector signed int, vector bool int); | |
6721 | vector signed int vec_vadduwm (vector signed int, vector signed int); | |
6722 | vector unsigned int vec_vadduwm (vector bool int, vector unsigned int); | |
6723 | vector unsigned int vec_vadduwm (vector unsigned int, vector bool int); | |
6724 | vector unsigned int vec_vadduwm (vector unsigned int, | |
6725 | vector unsigned int); | |
6726 | ||
6727 | vector signed short vec_vadduhm (vector bool short, | |
6728 | vector signed short); | |
6729 | vector signed short vec_vadduhm (vector signed short, | |
6730 | vector bool short); | |
6731 | vector signed short vec_vadduhm (vector signed short, | |
6732 | vector signed short); | |
6733 | vector unsigned short vec_vadduhm (vector bool short, | |
6734 | vector unsigned short); | |
6735 | vector unsigned short vec_vadduhm (vector unsigned short, | |
6736 | vector bool short); | |
6737 | vector unsigned short vec_vadduhm (vector unsigned short, | |
6738 | vector unsigned short); | |
6739 | ||
6740 | vector signed char vec_vaddubm (vector bool char, vector signed char); | |
6741 | vector signed char vec_vaddubm (vector signed char, vector bool char); | |
6742 | vector signed char vec_vaddubm (vector signed char, vector signed char); | |
6743 | vector unsigned char vec_vaddubm (vector bool char, | |
6744 | vector unsigned char); | |
6745 | vector unsigned char vec_vaddubm (vector unsigned char, | |
6746 | vector bool char); | |
6747 | vector unsigned char vec_vaddubm (vector unsigned char, | |
6748 | vector unsigned char); | |
6749 | ||
333c8841 AH |
6750 | vector unsigned int vec_addc (vector unsigned int, vector unsigned int); |
6751 | ||
b0b343db JJ |
6752 | vector unsigned char vec_adds (vector bool char, vector unsigned char); |
6753 | vector unsigned char vec_adds (vector unsigned char, vector bool char); | |
924fcc4e JM |
6754 | vector unsigned char vec_adds (vector unsigned char, |
6755 | vector unsigned char); | |
b0b343db JJ |
6756 | vector signed char vec_adds (vector bool char, vector signed char); |
6757 | vector signed char vec_adds (vector signed char, vector bool char); | |
333c8841 | 6758 | vector signed char vec_adds (vector signed char, vector signed char); |
b0b343db | 6759 | vector unsigned short vec_adds (vector bool short, |
924fcc4e JM |
6760 | vector unsigned short); |
6761 | vector unsigned short vec_adds (vector unsigned short, | |
b0b343db | 6762 | vector bool short); |
6e5bb5ad JM |
6763 | vector unsigned short vec_adds (vector unsigned short, |
6764 | vector unsigned short); | |
b0b343db JJ |
6765 | vector signed short vec_adds (vector bool short, vector signed short); |
6766 | vector signed short vec_adds (vector signed short, vector bool short); | |
333c8841 | 6767 | vector signed short vec_adds (vector signed short, vector signed short); |
b0b343db JJ |
6768 | vector unsigned int vec_adds (vector bool int, vector unsigned int); |
6769 | vector unsigned int vec_adds (vector unsigned int, vector bool int); | |
333c8841 | 6770 | vector unsigned int vec_adds (vector unsigned int, vector unsigned int); |
b0b343db JJ |
6771 | vector signed int vec_adds (vector bool int, vector signed int); |
6772 | vector signed int vec_adds (vector signed int, vector bool int); | |
333c8841 AH |
6773 | vector signed int vec_adds (vector signed int, vector signed int); |
6774 | ||
b0b343db JJ |
6775 | vector signed int vec_vaddsws (vector bool int, vector signed int); |
6776 | vector signed int vec_vaddsws (vector signed int, vector bool int); | |
6777 | vector signed int vec_vaddsws (vector signed int, vector signed int); | |
6778 | ||
6779 | vector unsigned int vec_vadduws (vector bool int, vector unsigned int); | |
6780 | vector unsigned int vec_vadduws (vector unsigned int, vector bool int); | |
6781 | vector unsigned int vec_vadduws (vector unsigned int, | |
6782 | vector unsigned int); | |
6783 | ||
6784 | vector signed short vec_vaddshs (vector bool short, | |
6785 | vector signed short); | |
6786 | vector signed short vec_vaddshs (vector signed short, | |
6787 | vector bool short); | |
6788 | vector signed short vec_vaddshs (vector signed short, | |
6789 | vector signed short); | |
6790 | ||
6791 | vector unsigned short vec_vadduhs (vector bool short, | |
6792 | vector unsigned short); | |
6793 | vector unsigned short vec_vadduhs (vector unsigned short, | |
6794 | vector bool short); | |
6795 | vector unsigned short vec_vadduhs (vector unsigned short, | |
6796 | vector unsigned short); | |
6797 | ||
6798 | vector signed char vec_vaddsbs (vector bool char, vector signed char); | |
6799 | vector signed char vec_vaddsbs (vector signed char, vector bool char); | |
6800 | vector signed char vec_vaddsbs (vector signed char, vector signed char); | |
6801 | ||
6802 | vector unsigned char vec_vaddubs (vector bool char, | |
6803 | vector unsigned char); | |
6804 | vector unsigned char vec_vaddubs (vector unsigned char, | |
6805 | vector bool char); | |
6806 | vector unsigned char vec_vaddubs (vector unsigned char, | |
6807 | vector unsigned char); | |
6808 | ||
333c8841 | 6809 | vector float vec_and (vector float, vector float); |
b0b343db JJ |
6810 | vector float vec_and (vector float, vector bool int); |
6811 | vector float vec_and (vector bool int, vector float); | |
6812 | vector bool int vec_and (vector bool int, vector bool int); | |
6813 | vector signed int vec_and (vector bool int, vector signed int); | |
6814 | vector signed int vec_and (vector signed int, vector bool int); | |
333c8841 | 6815 | vector signed int vec_and (vector signed int, vector signed int); |
b0b343db JJ |
6816 | vector unsigned int vec_and (vector bool int, vector unsigned int); |
6817 | vector unsigned int vec_and (vector unsigned int, vector bool int); | |
333c8841 | 6818 | vector unsigned int vec_and (vector unsigned int, vector unsigned int); |
b0b343db JJ |
6819 | vector bool short vec_and (vector bool short, vector bool short); |
6820 | vector signed short vec_and (vector bool short, vector signed short); | |
6821 | vector signed short vec_and (vector signed short, vector bool short); | |
333c8841 | 6822 | vector signed short vec_and (vector signed short, vector signed short); |
b0b343db | 6823 | vector unsigned short vec_and (vector bool short, |
924fcc4e JM |
6824 | vector unsigned short); |
6825 | vector unsigned short vec_and (vector unsigned short, | |
b0b343db | 6826 | vector bool short); |
6e5bb5ad JM |
6827 | vector unsigned short vec_and (vector unsigned short, |
6828 | vector unsigned short); | |
b0b343db JJ |
6829 | vector signed char vec_and (vector bool char, vector signed char); |
6830 | vector bool char vec_and (vector bool char, vector bool char); | |
6831 | vector signed char vec_and (vector signed char, vector bool char); | |
333c8841 | 6832 | vector signed char vec_and (vector signed char, vector signed char); |
b0b343db JJ |
6833 | vector unsigned char vec_and (vector bool char, vector unsigned char); |
6834 | vector unsigned char vec_and (vector unsigned char, vector bool char); | |
924fcc4e JM |
6835 | vector unsigned char vec_and (vector unsigned char, |
6836 | vector unsigned char); | |
333c8841 AH |
6837 | |
6838 | vector float vec_andc (vector float, vector float); | |
b0b343db JJ |
6839 | vector float vec_andc (vector float, vector bool int); |
6840 | vector float vec_andc (vector bool int, vector float); | |
6841 | vector bool int vec_andc (vector bool int, vector bool int); | |
6842 | vector signed int vec_andc (vector bool int, vector signed int); | |
6843 | vector signed int vec_andc (vector signed int, vector bool int); | |
333c8841 | 6844 | vector signed int vec_andc (vector signed int, vector signed int); |
b0b343db JJ |
6845 | vector unsigned int vec_andc (vector bool int, vector unsigned int); |
6846 | vector unsigned int vec_andc (vector unsigned int, vector bool int); | |
333c8841 | 6847 | vector unsigned int vec_andc (vector unsigned int, vector unsigned int); |
b0b343db JJ |
6848 | vector bool short vec_andc (vector bool short, vector bool short); |
6849 | vector signed short vec_andc (vector bool short, vector signed short); | |
6850 | vector signed short vec_andc (vector signed short, vector bool short); | |
333c8841 | 6851 | vector signed short vec_andc (vector signed short, vector signed short); |
b0b343db | 6852 | vector unsigned short vec_andc (vector bool short, |
924fcc4e JM |
6853 | vector unsigned short); |
6854 | vector unsigned short vec_andc (vector unsigned short, | |
b0b343db | 6855 | vector bool short); |
6e5bb5ad JM |
6856 | vector unsigned short vec_andc (vector unsigned short, |
6857 | vector unsigned short); | |
b0b343db JJ |
6858 | vector signed char vec_andc (vector bool char, vector signed char); |
6859 | vector bool char vec_andc (vector bool char, vector bool char); | |
6860 | vector signed char vec_andc (vector signed char, vector bool char); | |
333c8841 | 6861 | vector signed char vec_andc (vector signed char, vector signed char); |
b0b343db JJ |
6862 | vector unsigned char vec_andc (vector bool char, vector unsigned char); |
6863 | vector unsigned char vec_andc (vector unsigned char, vector bool char); | |
924fcc4e JM |
6864 | vector unsigned char vec_andc (vector unsigned char, |
6865 | vector unsigned char); | |
333c8841 | 6866 | |
924fcc4e JM |
6867 | vector unsigned char vec_avg (vector unsigned char, |
6868 | vector unsigned char); | |
333c8841 | 6869 | vector signed char vec_avg (vector signed char, vector signed char); |
6e5bb5ad JM |
6870 | vector unsigned short vec_avg (vector unsigned short, |
6871 | vector unsigned short); | |
333c8841 AH |
6872 | vector signed short vec_avg (vector signed short, vector signed short); |
6873 | vector unsigned int vec_avg (vector unsigned int, vector unsigned int); | |
6874 | vector signed int vec_avg (vector signed int, vector signed int); | |
6875 | ||
b0b343db JJ |
6876 | vector signed int vec_vavgsw (vector signed int, vector signed int); |
6877 | ||
6878 | vector unsigned int vec_vavguw (vector unsigned int, | |
6879 | vector unsigned int); | |
6880 | ||
6881 | vector signed short vec_vavgsh (vector signed short, | |
6882 | vector signed short); | |
6883 | ||
6884 | vector unsigned short vec_vavguh (vector unsigned short, | |
6885 | vector unsigned short); | |
6886 | ||
6887 | vector signed char vec_vavgsb (vector signed char, vector signed char); | |
6888 | ||
6889 | vector unsigned char vec_vavgub (vector unsigned char, | |
6890 | vector unsigned char); | |
6891 | ||
333c8841 AH |
6892 | vector float vec_ceil (vector float); |
6893 | ||
6894 | vector signed int vec_cmpb (vector float, vector float); | |
6895 | ||
b0b343db JJ |
6896 | vector bool char vec_cmpeq (vector signed char, vector signed char); |
6897 | vector bool char vec_cmpeq (vector unsigned char, vector unsigned char); | |
6898 | vector bool short vec_cmpeq (vector signed short, vector signed short); | |
6899 | vector bool short vec_cmpeq (vector unsigned short, | |
6900 | vector unsigned short); | |
6901 | vector bool int vec_cmpeq (vector signed int, vector signed int); | |
6902 | vector bool int vec_cmpeq (vector unsigned int, vector unsigned int); | |
6903 | vector bool int vec_cmpeq (vector float, vector float); | |
333c8841 | 6904 | |
b0b343db | 6905 | vector bool int vec_vcmpeqfp (vector float, vector float); |
333c8841 | 6906 | |
b0b343db JJ |
6907 | vector bool int vec_vcmpequw (vector signed int, vector signed int); |
6908 | vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int); | |
6909 | ||
6910 | vector bool short vec_vcmpequh (vector signed short, | |
6911 | vector signed short); | |
6912 | vector bool short vec_vcmpequh (vector unsigned short, | |
6913 | vector unsigned short); | |
333c8841 | 6914 | |
b0b343db JJ |
6915 | vector bool char vec_vcmpequb (vector signed char, vector signed char); |
6916 | vector bool char vec_vcmpequb (vector unsigned char, | |
6917 | vector unsigned char); | |
333c8841 | 6918 | |
b0b343db | 6919 | vector bool int vec_cmpge (vector float, vector float); |
333c8841 | 6920 | |
b0b343db JJ |
6921 | vector bool char vec_cmpgt (vector unsigned char, vector unsigned char); |
6922 | vector bool char vec_cmpgt (vector signed char, vector signed char); | |
6923 | vector bool short vec_cmpgt (vector unsigned short, | |
6924 | vector unsigned short); | |
6925 | vector bool short vec_cmpgt (vector signed short, vector signed short); | |
6926 | vector bool int vec_cmpgt (vector unsigned int, vector unsigned int); | |
6927 | vector bool int vec_cmpgt (vector signed int, vector signed int); | |
6928 | vector bool int vec_cmpgt (vector float, vector float); | |
6929 | ||
6930 | vector bool int vec_vcmpgtfp (vector float, vector float); | |
6931 | ||
6932 | vector bool int vec_vcmpgtsw (vector signed int, vector signed int); | |
6933 | ||
6934 | vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int); | |
333c8841 | 6935 | |
b0b343db JJ |
6936 | vector bool short vec_vcmpgtsh (vector signed short, |
6937 | vector signed short); | |
6938 | ||
6939 | vector bool short vec_vcmpgtuh (vector unsigned short, | |
6940 | vector unsigned short); | |
6941 | ||
6942 | vector bool char vec_vcmpgtsb (vector signed char, vector signed char); | |
6943 | ||
6944 | vector bool char vec_vcmpgtub (vector unsigned char, | |
6945 | vector unsigned char); | |
6946 | ||
6947 | vector bool int vec_cmple (vector float, vector float); | |
6948 | ||
6949 | vector bool char vec_cmplt (vector unsigned char, vector unsigned char); | |
6950 | vector bool char vec_cmplt (vector signed char, vector signed char); | |
6951 | vector bool short vec_cmplt (vector unsigned short, | |
6952 | vector unsigned short); | |
6953 | vector bool short vec_cmplt (vector signed short, vector signed short); | |
6954 | vector bool int vec_cmplt (vector unsigned int, vector unsigned int); | |
6955 | vector bool int vec_cmplt (vector signed int, vector signed int); | |
6956 | vector bool int vec_cmplt (vector float, vector float); | |
333c8841 | 6957 | |
b0b343db JJ |
6958 | vector float vec_ctf (vector unsigned int, const int); |
6959 | vector float vec_ctf (vector signed int, const int); | |
333c8841 | 6960 | |
b0b343db JJ |
6961 | vector float vec_vcfsx (vector signed int, const int); |
6962 | ||
6963 | vector float vec_vcfux (vector unsigned int, const int); | |
6964 | ||
6965 | vector signed int vec_cts (vector float, const int); | |
6966 | ||
6967 | vector unsigned int vec_ctu (vector float, const int); | |
6968 | ||
6969 | void vec_dss (const int); | |
333c8841 AH |
6970 | |
6971 | void vec_dssall (void); | |
6972 | ||
b0b343db JJ |
6973 | void vec_dst (const vector unsigned char *, int, const int); |
6974 | void vec_dst (const vector signed char *, int, const int); | |
6975 | void vec_dst (const vector bool char *, int, const int); | |
6976 | void vec_dst (const vector unsigned short *, int, const int); | |
6977 | void vec_dst (const vector signed short *, int, const int); | |
6978 | void vec_dst (const vector bool short *, int, const int); | |
6979 | void vec_dst (const vector pixel *, int, const int); | |
6980 | void vec_dst (const vector unsigned int *, int, const int); | |
6981 | void vec_dst (const vector signed int *, int, const int); | |
6982 | void vec_dst (const vector bool int *, int, const int); | |
6983 | void vec_dst (const vector float *, int, const int); | |
6984 | void vec_dst (const unsigned char *, int, const int); | |
6985 | void vec_dst (const signed char *, int, const int); | |
6986 | void vec_dst (const unsigned short *, int, const int); | |
6987 | void vec_dst (const short *, int, const int); | |
6988 | void vec_dst (const unsigned int *, int, const int); | |
6989 | void vec_dst (const int *, int, const int); | |
6990 | void vec_dst (const unsigned long *, int, const int); | |
6991 | void vec_dst (const long *, int, const int); | |
6992 | void vec_dst (const float *, int, const int); | |
6993 | ||
6994 | void vec_dstst (const vector unsigned char *, int, const int); | |
6995 | void vec_dstst (const vector signed char *, int, const int); | |
6996 | void vec_dstst (const vector bool char *, int, const int); | |
6997 | void vec_dstst (const vector unsigned short *, int, const int); | |
6998 | void vec_dstst (const vector signed short *, int, const int); | |
6999 | void vec_dstst (const vector bool short *, int, const int); | |
7000 | void vec_dstst (const vector pixel *, int, const int); | |
7001 | void vec_dstst (const vector unsigned int *, int, const int); | |
7002 | void vec_dstst (const vector signed int *, int, const int); | |
7003 | void vec_dstst (const vector bool int *, int, const int); | |
7004 | void vec_dstst (const vector float *, int, const int); | |
7005 | void vec_dstst (const unsigned char *, int, const int); | |
7006 | void vec_dstst (const signed char *, int, const int); | |
7007 | void vec_dstst (const unsigned short *, int, const int); | |
7008 | void vec_dstst (const short *, int, const int); | |
7009 | void vec_dstst (const unsigned int *, int, const int); | |
7010 | void vec_dstst (const int *, int, const int); | |
7011 | void vec_dstst (const unsigned long *, int, const int); | |
7012 | void vec_dstst (const long *, int, const int); | |
7013 | void vec_dstst (const float *, int, const int); | |
7014 | ||
7015 | void vec_dststt (const vector unsigned char *, int, const int); | |
7016 | void vec_dststt (const vector signed char *, int, const int); | |
7017 | void vec_dststt (const vector bool char *, int, const int); | |
7018 | void vec_dststt (const vector unsigned short *, int, const int); | |
7019 | void vec_dststt (const vector signed short *, int, const int); | |
7020 | void vec_dststt (const vector bool short *, int, const int); | |
7021 | void vec_dststt (const vector pixel *, int, const int); | |
7022 | void vec_dststt (const vector unsigned int *, int, const int); | |
7023 | void vec_dststt (const vector signed int *, int, const int); | |
7024 | void vec_dststt (const vector bool int *, int, const int); | |
7025 | void vec_dststt (const vector float *, int, const int); | |
7026 | void vec_dststt (const unsigned char *, int, const int); | |
7027 | void vec_dststt (const signed char *, int, const int); | |
7028 | void vec_dststt (const unsigned short *, int, const int); | |
7029 | void vec_dststt (const short *, int, const int); | |
7030 | void vec_dststt (const unsigned int *, int, const int); | |
7031 | void vec_dststt (const int *, int, const int); | |
7032 | void vec_dststt (const unsigned long *, int, const int); | |
7033 | void vec_dststt (const long *, int, const int); | |
7034 | void vec_dststt (const float *, int, const int); | |
7035 | ||
7036 | void vec_dstt (const vector unsigned char *, int, const int); | |
7037 | void vec_dstt (const vector signed char *, int, const int); | |
7038 | void vec_dstt (const vector bool char *, int, const int); | |
7039 | void vec_dstt (const vector unsigned short *, int, const int); | |
7040 | void vec_dstt (const vector signed short *, int, const int); | |
7041 | void vec_dstt (const vector bool short *, int, const int); | |
7042 | void vec_dstt (const vector pixel *, int, const int); | |
7043 | void vec_dstt (const vector unsigned int *, int, const int); | |
7044 | void vec_dstt (const vector signed int *, int, const int); | |
7045 | void vec_dstt (const vector bool int *, int, const int); | |
7046 | void vec_dstt (const vector float *, int, const int); | |
7047 | void vec_dstt (const unsigned char *, int, const int); | |
7048 | void vec_dstt (const signed char *, int, const int); | |
7049 | void vec_dstt (const unsigned short *, int, const int); | |
7050 | void vec_dstt (const short *, int, const int); | |
7051 | void vec_dstt (const unsigned int *, int, const int); | |
7052 | void vec_dstt (const int *, int, const int); | |
7053 | void vec_dstt (const unsigned long *, int, const int); | |
7054 | void vec_dstt (const long *, int, const int); | |
7055 | void vec_dstt (const float *, int, const int); | |
7056 | ||
7057 | vector float vec_expte (vector float); | |
7058 | ||
7059 | vector float vec_floor (vector float); | |
7060 | ||
7061 | vector float vec_ld (int, const vector float *); | |
7062 | vector float vec_ld (int, const float *); | |
7063 | vector bool int vec_ld (int, const vector bool int *); | |
7064 | vector signed int vec_ld (int, const vector signed int *); | |
7065 | vector signed int vec_ld (int, const int *); | |
7066 | vector signed int vec_ld (int, const long *); | |
7067 | vector unsigned int vec_ld (int, const vector unsigned int *); | |
7068 | vector unsigned int vec_ld (int, const unsigned int *); | |
7069 | vector unsigned int vec_ld (int, const unsigned long *); | |
7070 | vector bool short vec_ld (int, const vector bool short *); | |
7071 | vector pixel vec_ld (int, const vector pixel *); | |
7072 | vector signed short vec_ld (int, const vector signed short *); | |
7073 | vector signed short vec_ld (int, const short *); | |
7074 | vector unsigned short vec_ld (int, const vector unsigned short *); | |
7075 | vector unsigned short vec_ld (int, const unsigned short *); | |
7076 | vector bool char vec_ld (int, const vector bool char *); | |
7077 | vector signed char vec_ld (int, const vector signed char *); | |
7078 | vector signed char vec_ld (int, const signed char *); | |
7079 | vector unsigned char vec_ld (int, const vector unsigned char *); | |
7080 | vector unsigned char vec_ld (int, const unsigned char *); | |
7081 | ||
7082 | vector signed char vec_lde (int, const signed char *); | |
7083 | vector unsigned char vec_lde (int, const unsigned char *); | |
7084 | vector signed short vec_lde (int, const short *); | |
7085 | vector unsigned short vec_lde (int, const unsigned short *); | |
7086 | vector float vec_lde (int, const float *); | |
7087 | vector signed int vec_lde (int, const int *); | |
7088 | vector unsigned int vec_lde (int, const unsigned int *); | |
7089 | vector signed int vec_lde (int, const long *); | |
7090 | vector unsigned int vec_lde (int, const unsigned long *); | |
7091 | ||
7092 | vector float vec_lvewx (int, float *); | |
7093 | vector signed int vec_lvewx (int, int *); | |
7094 | vector unsigned int vec_lvewx (int, unsigned int *); | |
7095 | vector signed int vec_lvewx (int, long *); | |
7096 | vector unsigned int vec_lvewx (int, unsigned long *); | |
7097 | ||
7098 | vector signed short vec_lvehx (int, short *); | |
7099 | vector unsigned short vec_lvehx (int, unsigned short *); | |
7100 | ||
7101 | vector signed char vec_lvebx (int, char *); | |
7102 | vector unsigned char vec_lvebx (int, unsigned char *); | |
7103 | ||
7104 | vector float vec_ldl (int, const vector float *); | |
7105 | vector float vec_ldl (int, const float *); | |
7106 | vector bool int vec_ldl (int, const vector bool int *); | |
7107 | vector signed int vec_ldl (int, const vector signed int *); | |
7108 | vector signed int vec_ldl (int, const int *); | |
7109 | vector signed int vec_ldl (int, const long *); | |
7110 | vector unsigned int vec_ldl (int, const vector unsigned int *); | |
7111 | vector unsigned int vec_ldl (int, const unsigned int *); | |
7112 | vector unsigned int vec_ldl (int, const unsigned long *); | |
7113 | vector bool short vec_ldl (int, const vector bool short *); | |
7114 | vector pixel vec_ldl (int, const vector pixel *); | |
7115 | vector signed short vec_ldl (int, const vector signed short *); | |
7116 | vector signed short vec_ldl (int, const short *); | |
7117 | vector unsigned short vec_ldl (int, const vector unsigned short *); | |
7118 | vector unsigned short vec_ldl (int, const unsigned short *); | |
7119 | vector bool char vec_ldl (int, const vector bool char *); | |
7120 | vector signed char vec_ldl (int, const vector signed char *); | |
7121 | vector signed char vec_ldl (int, const signed char *); | |
7122 | vector unsigned char vec_ldl (int, const vector unsigned char *); | |
7123 | vector unsigned char vec_ldl (int, const unsigned char *); | |
333c8841 AH |
7124 | |
7125 | vector float vec_loge (vector float); | |
7126 | ||
b0b343db JJ |
7127 | vector unsigned char vec_lvsl (int, const volatile unsigned char *); |
7128 | vector unsigned char vec_lvsl (int, const volatile signed char *); | |
7129 | vector unsigned char vec_lvsl (int, const volatile unsigned short *); | |
7130 | vector unsigned char vec_lvsl (int, const volatile short *); | |
7131 | vector unsigned char vec_lvsl (int, const volatile unsigned int *); | |
7132 | vector unsigned char vec_lvsl (int, const volatile int *); | |
7133 | vector unsigned char vec_lvsl (int, const volatile unsigned long *); | |
7134 | vector unsigned char vec_lvsl (int, const volatile long *); | |
7135 | vector unsigned char vec_lvsl (int, const volatile float *); | |
7136 | ||
7137 | vector unsigned char vec_lvsr (int, const volatile unsigned char *); | |
7138 | vector unsigned char vec_lvsr (int, const volatile signed char *); | |
7139 | vector unsigned char vec_lvsr (int, const volatile unsigned short *); | |
7140 | vector unsigned char vec_lvsr (int, const volatile short *); | |
7141 | vector unsigned char vec_lvsr (int, const volatile unsigned int *); | |
7142 | vector unsigned char vec_lvsr (int, const volatile int *); | |
7143 | vector unsigned char vec_lvsr (int, const volatile unsigned long *); | |
7144 | vector unsigned char vec_lvsr (int, const volatile long *); | |
7145 | vector unsigned char vec_lvsr (int, const volatile float *); | |
333c8841 AH |
7146 | |
7147 | vector float vec_madd (vector float, vector float, vector float); | |
7148 | ||
b0b343db JJ |
7149 | vector signed short vec_madds (vector signed short, |
7150 | vector signed short, | |
6e5bb5ad | 7151 | vector signed short); |
333c8841 | 7152 | |
b0b343db JJ |
7153 | vector unsigned char vec_max (vector bool char, vector unsigned char); |
7154 | vector unsigned char vec_max (vector unsigned char, vector bool char); | |
924fcc4e JM |
7155 | vector unsigned char vec_max (vector unsigned char, |
7156 | vector unsigned char); | |
b0b343db JJ |
7157 | vector signed char vec_max (vector bool char, vector signed char); |
7158 | vector signed char vec_max (vector signed char, vector bool char); | |
333c8841 | 7159 | vector signed char vec_max (vector signed char, vector signed char); |
b0b343db | 7160 | vector unsigned short vec_max (vector bool short, |
924fcc4e JM |
7161 | vector unsigned short); |
7162 | vector unsigned short vec_max (vector unsigned short, | |
b0b343db | 7163 | vector bool short); |
6e5bb5ad JM |
7164 | vector unsigned short vec_max (vector unsigned short, |
7165 | vector unsigned short); | |
b0b343db JJ |
7166 | vector signed short vec_max (vector bool short, vector signed short); |
7167 | vector signed short vec_max (vector signed short, vector bool short); | |
333c8841 | 7168 | vector signed short vec_max (vector signed short, vector signed short); |
b0b343db JJ |
7169 | vector unsigned int vec_max (vector bool int, vector unsigned int); |
7170 | vector unsigned int vec_max (vector unsigned int, vector bool int); | |
333c8841 | 7171 | vector unsigned int vec_max (vector unsigned int, vector unsigned int); |
b0b343db JJ |
7172 | vector signed int vec_max (vector bool int, vector signed int); |
7173 | vector signed int vec_max (vector signed int, vector bool int); | |
333c8841 AH |
7174 | vector signed int vec_max (vector signed int, vector signed int); |
7175 | vector float vec_max (vector float, vector float); | |
7176 | ||
b0b343db JJ |
7177 | vector float vec_vmaxfp (vector float, vector float); |
7178 | ||
7179 | vector signed int vec_vmaxsw (vector bool int, vector signed int); | |
7180 | vector signed int vec_vmaxsw (vector signed int, vector bool int); | |
7181 | vector signed int vec_vmaxsw (vector signed int, vector signed int); | |
7182 | ||
7183 | vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int); | |
7184 | vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int); | |
7185 | vector unsigned int vec_vmaxuw (vector unsigned int, | |
7186 | vector unsigned int); | |
7187 | ||
7188 | vector signed short vec_vmaxsh (vector bool short, vector signed short); | |
7189 | vector signed short vec_vmaxsh (vector signed short, vector bool short); | |
7190 | vector signed short vec_vmaxsh (vector signed short, | |
7191 | vector signed short); | |
7192 | ||
7193 | vector unsigned short vec_vmaxuh (vector bool short, | |
7194 | vector unsigned short); | |
7195 | vector unsigned short vec_vmaxuh (vector unsigned short, | |
7196 | vector bool short); | |
7197 | vector unsigned short vec_vmaxuh (vector unsigned short, | |
7198 | vector unsigned short); | |
7199 | ||
7200 | vector signed char vec_vmaxsb (vector bool char, vector signed char); | |
7201 | vector signed char vec_vmaxsb (vector signed char, vector bool char); | |
7202 | vector signed char vec_vmaxsb (vector signed char, vector signed char); | |
7203 | ||
7204 | vector unsigned char vec_vmaxub (vector bool char, | |
7205 | vector unsigned char); | |
7206 | vector unsigned char vec_vmaxub (vector unsigned char, | |
7207 | vector bool char); | |
7208 | vector unsigned char vec_vmaxub (vector unsigned char, | |
7209 | vector unsigned char); | |
7210 | ||
7211 | vector bool char vec_mergeh (vector bool char, vector bool char); | |
333c8841 | 7212 | vector signed char vec_mergeh (vector signed char, vector signed char); |
6e5bb5ad JM |
7213 | vector unsigned char vec_mergeh (vector unsigned char, |
7214 | vector unsigned char); | |
b0b343db JJ |
7215 | vector bool short vec_mergeh (vector bool short, vector bool short); |
7216 | vector pixel vec_mergeh (vector pixel, vector pixel); | |
924fcc4e JM |
7217 | vector signed short vec_mergeh (vector signed short, |
7218 | vector signed short); | |
6e5bb5ad JM |
7219 | vector unsigned short vec_mergeh (vector unsigned short, |
7220 | vector unsigned short); | |
333c8841 | 7221 | vector float vec_mergeh (vector float, vector float); |
b0b343db | 7222 | vector bool int vec_mergeh (vector bool int, vector bool int); |
333c8841 | 7223 | vector signed int vec_mergeh (vector signed int, vector signed int); |
924fcc4e JM |
7224 | vector unsigned int vec_mergeh (vector unsigned int, |
7225 | vector unsigned int); | |
333c8841 | 7226 | |
b0b343db JJ |
7227 | vector float vec_vmrghw (vector float, vector float); |
7228 | vector bool int vec_vmrghw (vector bool int, vector bool int); | |
7229 | vector signed int vec_vmrghw (vector signed int, vector signed int); | |
7230 | vector unsigned int vec_vmrghw (vector unsigned int, | |
7231 | vector unsigned int); | |
7232 | ||
7233 | vector bool short vec_vmrghh (vector bool short, vector bool short); | |
7234 | vector signed short vec_vmrghh (vector signed short, | |
7235 | vector signed short); | |
7236 | vector unsigned short vec_vmrghh (vector unsigned short, | |
7237 | vector unsigned short); | |
7238 | vector pixel vec_vmrghh (vector pixel, vector pixel); | |
7239 | ||
7240 | vector bool char vec_vmrghb (vector bool char, vector bool char); | |
7241 | vector signed char vec_vmrghb (vector signed char, vector signed char); | |
7242 | vector unsigned char vec_vmrghb (vector unsigned char, | |
7243 | vector unsigned char); | |
7244 | ||
7245 | vector bool char vec_mergel (vector bool char, vector bool char); | |
333c8841 | 7246 | vector signed char vec_mergel (vector signed char, vector signed char); |
6e5bb5ad JM |
7247 | vector unsigned char vec_mergel (vector unsigned char, |
7248 | vector unsigned char); | |
b0b343db JJ |
7249 | vector bool short vec_mergel (vector bool short, vector bool short); |
7250 | vector pixel vec_mergel (vector pixel, vector pixel); | |
924fcc4e JM |
7251 | vector signed short vec_mergel (vector signed short, |
7252 | vector signed short); | |
6e5bb5ad JM |
7253 | vector unsigned short vec_mergel (vector unsigned short, |
7254 | vector unsigned short); | |
333c8841 | 7255 | vector float vec_mergel (vector float, vector float); |
b0b343db | 7256 | vector bool int vec_mergel (vector bool int, vector bool int); |
333c8841 | 7257 | vector signed int vec_mergel (vector signed int, vector signed int); |
924fcc4e JM |
7258 | vector unsigned int vec_mergel (vector unsigned int, |
7259 | vector unsigned int); | |
333c8841 | 7260 | |
b0b343db JJ |
7261 | vector float vec_vmrglw (vector float, vector float); |
7262 | vector signed int vec_vmrglw (vector signed int, vector signed int); | |
7263 | vector unsigned int vec_vmrglw (vector unsigned int, | |
7264 | vector unsigned int); | |
7265 | vector bool int vec_vmrglw (vector bool int, vector bool int); | |
333c8841 | 7266 | |
b0b343db JJ |
7267 | vector bool short vec_vmrglh (vector bool short, vector bool short); |
7268 | vector signed short vec_vmrglh (vector signed short, | |
7269 | vector signed short); | |
7270 | vector unsigned short vec_vmrglh (vector unsigned short, | |
7271 | vector unsigned short); | |
7272 | vector pixel vec_vmrglh (vector pixel, vector pixel); | |
7273 | ||
7274 | vector bool char vec_vmrglb (vector bool char, vector bool char); | |
7275 | vector signed char vec_vmrglb (vector signed char, vector signed char); | |
7276 | vector unsigned char vec_vmrglb (vector unsigned char, | |
7277 | vector unsigned char); | |
333c8841 | 7278 | |
b0b343db | 7279 | vector unsigned short vec_mfvscr (void); |
333c8841 | 7280 | |
b0b343db JJ |
7281 | vector unsigned char vec_min (vector bool char, vector unsigned char); |
7282 | vector unsigned char vec_min (vector unsigned char, vector bool char); | |
924fcc4e JM |
7283 | vector unsigned char vec_min (vector unsigned char, |
7284 | vector unsigned char); | |
b0b343db JJ |
7285 | vector signed char vec_min (vector bool char, vector signed char); |
7286 | vector signed char vec_min (vector signed char, vector bool char); | |
333c8841 | 7287 | vector signed char vec_min (vector signed char, vector signed char); |
b0b343db | 7288 | vector unsigned short vec_min (vector bool short, |
924fcc4e JM |
7289 | vector unsigned short); |
7290 | vector unsigned short vec_min (vector unsigned short, | |
b0b343db | 7291 | vector bool short); |
6e5bb5ad JM |
7292 | vector unsigned short vec_min (vector unsigned short, |
7293 | vector unsigned short); | |
b0b343db JJ |
7294 | vector signed short vec_min (vector bool short, vector signed short); |
7295 | vector signed short vec_min (vector signed short, vector bool short); | |
333c8841 | 7296 | vector signed short vec_min (vector signed short, vector signed short); |
b0b343db JJ |
7297 | vector unsigned int vec_min (vector bool int, vector unsigned int); |
7298 | vector unsigned int vec_min (vector unsigned int, vector bool int); | |
333c8841 | 7299 | vector unsigned int vec_min (vector unsigned int, vector unsigned int); |
b0b343db JJ |
7300 | vector signed int vec_min (vector bool int, vector signed int); |
7301 | vector signed int vec_min (vector signed int, vector bool int); | |
333c8841 AH |
7302 | vector signed int vec_min (vector signed int, vector signed int); |
7303 | vector float vec_min (vector float, vector float); | |
7304 | ||
b0b343db JJ |
7305 | vector float vec_vminfp (vector float, vector float); |
7306 | ||
7307 | vector signed int vec_vminsw (vector bool int, vector signed int); | |
7308 | vector signed int vec_vminsw (vector signed int, vector bool int); | |
7309 | vector signed int vec_vminsw (vector signed int, vector signed int); | |
7310 | ||
7311 | vector unsigned int vec_vminuw (vector bool int, vector unsigned int); | |
7312 | vector unsigned int vec_vminuw (vector unsigned int, vector bool int); | |
7313 | vector unsigned int vec_vminuw (vector unsigned int, | |
7314 | vector unsigned int); | |
7315 | ||
7316 | vector signed short vec_vminsh (vector bool short, vector signed short); | |
7317 | vector signed short vec_vminsh (vector signed short, vector bool short); | |
7318 | vector signed short vec_vminsh (vector signed short, | |
7319 | vector signed short); | |
7320 | ||
7321 | vector unsigned short vec_vminuh (vector bool short, | |
7322 | vector unsigned short); | |
7323 | vector unsigned short vec_vminuh (vector unsigned short, | |
7324 | vector bool short); | |
7325 | vector unsigned short vec_vminuh (vector unsigned short, | |
7326 | vector unsigned short); | |
7327 | ||
7328 | vector signed char vec_vminsb (vector bool char, vector signed char); | |
7329 | vector signed char vec_vminsb (vector signed char, vector bool char); | |
7330 | vector signed char vec_vminsb (vector signed char, vector signed char); | |
7331 | ||
7332 | vector unsigned char vec_vminub (vector bool char, | |
7333 | vector unsigned char); | |
7334 | vector unsigned char vec_vminub (vector unsigned char, | |
7335 | vector bool char); | |
7336 | vector unsigned char vec_vminub (vector unsigned char, | |
7337 | vector unsigned char); | |
7338 | ||
7339 | vector signed short vec_mladd (vector signed short, | |
7340 | vector signed short, | |
6e5bb5ad | 7341 | vector signed short); |
924fcc4e JM |
7342 | vector signed short vec_mladd (vector signed short, |
7343 | vector unsigned short, | |
6e5bb5ad | 7344 | vector unsigned short); |
924fcc4e JM |
7345 | vector signed short vec_mladd (vector unsigned short, |
7346 | vector signed short, | |
6e5bb5ad JM |
7347 | vector signed short); |
7348 | vector unsigned short vec_mladd (vector unsigned short, | |
7349 | vector unsigned short, | |
7350 | vector unsigned short); | |
7351 | ||
924fcc4e JM |
7352 | vector signed short vec_mradds (vector signed short, |
7353 | vector signed short, | |
6e5bb5ad JM |
7354 | vector signed short); |
7355 | ||
924fcc4e JM |
7356 | vector unsigned int vec_msum (vector unsigned char, |
7357 | vector unsigned char, | |
6e5bb5ad | 7358 | vector unsigned int); |
b0b343db JJ |
7359 | vector signed int vec_msum (vector signed char, |
7360 | vector unsigned char, | |
6e5bb5ad | 7361 | vector signed int); |
924fcc4e JM |
7362 | vector unsigned int vec_msum (vector unsigned short, |
7363 | vector unsigned short, | |
6e5bb5ad | 7364 | vector unsigned int); |
b0b343db JJ |
7365 | vector signed int vec_msum (vector signed short, |
7366 | vector signed short, | |
6e5bb5ad JM |
7367 | vector signed int); |
7368 | ||
b0b343db JJ |
7369 | vector signed int vec_vmsumshm (vector signed short, |
7370 | vector signed short, | |
7371 | vector signed int); | |
7372 | ||
7373 | vector unsigned int vec_vmsumuhm (vector unsigned short, | |
7374 | vector unsigned short, | |
7375 | vector unsigned int); | |
7376 | ||
7377 | vector signed int vec_vmsummbm (vector signed char, | |
7378 | vector unsigned char, | |
7379 | vector signed int); | |
7380 | ||
7381 | vector unsigned int vec_vmsumubm (vector unsigned char, | |
7382 | vector unsigned char, | |
7383 | vector unsigned int); | |
7384 | ||
6e5bb5ad | 7385 | vector unsigned int vec_msums (vector unsigned short, |
924fcc4e JM |
7386 | vector unsigned short, |
7387 | vector unsigned int); | |
b0b343db JJ |
7388 | vector signed int vec_msums (vector signed short, |
7389 | vector signed short, | |
6e5bb5ad | 7390 | vector signed int); |
333c8841 | 7391 | |
b0b343db JJ |
7392 | vector signed int vec_vmsumshs (vector signed short, |
7393 | vector signed short, | |
7394 | vector signed int); | |
7395 | ||
7396 | vector unsigned int vec_vmsumuhs (vector unsigned short, | |
7397 | vector unsigned short, | |
7398 | vector unsigned int); | |
7399 | ||
333c8841 AH |
7400 | void vec_mtvscr (vector signed int); |
7401 | void vec_mtvscr (vector unsigned int); | |
b0b343db | 7402 | void vec_mtvscr (vector bool int); |
333c8841 AH |
7403 | void vec_mtvscr (vector signed short); |
7404 | void vec_mtvscr (vector unsigned short); | |
b0b343db JJ |
7405 | void vec_mtvscr (vector bool short); |
7406 | void vec_mtvscr (vector pixel); | |
333c8841 AH |
7407 | void vec_mtvscr (vector signed char); |
7408 | void vec_mtvscr (vector unsigned char); | |
b0b343db | 7409 | void vec_mtvscr (vector bool char); |
333c8841 | 7410 | |
924fcc4e JM |
7411 | vector unsigned short vec_mule (vector unsigned char, |
7412 | vector unsigned char); | |
b0b343db JJ |
7413 | vector signed short vec_mule (vector signed char, |
7414 | vector signed char); | |
924fcc4e JM |
7415 | vector unsigned int vec_mule (vector unsigned short, |
7416 | vector unsigned short); | |
333c8841 AH |
7417 | vector signed int vec_mule (vector signed short, vector signed short); |
7418 | ||
b0b343db JJ |
7419 | vector signed int vec_vmulesh (vector signed short, |
7420 | vector signed short); | |
7421 | ||
7422 | vector unsigned int vec_vmuleuh (vector unsigned short, | |
7423 | vector unsigned short); | |
7424 | ||
7425 | vector signed short vec_vmulesb (vector signed char, | |
7426 | vector signed char); | |
7427 | ||
7428 | vector unsigned short vec_vmuleub (vector unsigned char, | |
7429 | vector unsigned char); | |
7430 | ||
924fcc4e JM |
7431 | vector unsigned short vec_mulo (vector unsigned char, |
7432 | vector unsigned char); | |
333c8841 | 7433 | vector signed short vec_mulo (vector signed char, vector signed char); |
924fcc4e JM |
7434 | vector unsigned int vec_mulo (vector unsigned short, |
7435 | vector unsigned short); | |
333c8841 AH |
7436 | vector signed int vec_mulo (vector signed short, vector signed short); |
7437 | ||
b0b343db JJ |
7438 | vector signed int vec_vmulosh (vector signed short, |
7439 | vector signed short); | |
7440 | ||
7441 | vector unsigned int vec_vmulouh (vector unsigned short, | |
7442 | vector unsigned short); | |
7443 | ||
7444 | vector signed short vec_vmulosb (vector signed char, | |
7445 | vector signed char); | |
7446 | ||
7447 | vector unsigned short vec_vmuloub (vector unsigned char, | |
7448 | vector unsigned char); | |
7449 | ||
333c8841 AH |
7450 | vector float vec_nmsub (vector float, vector float, vector float); |
7451 | ||
7452 | vector float vec_nor (vector float, vector float); | |
7453 | vector signed int vec_nor (vector signed int, vector signed int); | |
7454 | vector unsigned int vec_nor (vector unsigned int, vector unsigned int); | |
b0b343db | 7455 | vector bool int vec_nor (vector bool int, vector bool int); |
333c8841 | 7456 | vector signed short vec_nor (vector signed short, vector signed short); |
6e5bb5ad JM |
7457 | vector unsigned short vec_nor (vector unsigned short, |
7458 | vector unsigned short); | |
b0b343db | 7459 | vector bool short vec_nor (vector bool short, vector bool short); |
333c8841 | 7460 | vector signed char vec_nor (vector signed char, vector signed char); |
924fcc4e JM |
7461 | vector unsigned char vec_nor (vector unsigned char, |
7462 | vector unsigned char); | |
b0b343db | 7463 | vector bool char vec_nor (vector bool char, vector bool char); |
333c8841 AH |
7464 | |
7465 | vector float vec_or (vector float, vector float); | |
b0b343db JJ |
7466 | vector float vec_or (vector float, vector bool int); |
7467 | vector float vec_or (vector bool int, vector float); | |
7468 | vector bool int vec_or (vector bool int, vector bool int); | |
7469 | vector signed int vec_or (vector bool int, vector signed int); | |
7470 | vector signed int vec_or (vector signed int, vector bool int); | |
333c8841 | 7471 | vector signed int vec_or (vector signed int, vector signed int); |
b0b343db JJ |
7472 | vector unsigned int vec_or (vector bool int, vector unsigned int); |
7473 | vector unsigned int vec_or (vector unsigned int, vector bool int); | |
333c8841 | 7474 | vector unsigned int vec_or (vector unsigned int, vector unsigned int); |
b0b343db JJ |
7475 | vector bool short vec_or (vector bool short, vector bool short); |
7476 | vector signed short vec_or (vector bool short, vector signed short); | |
7477 | vector signed short vec_or (vector signed short, vector bool short); | |
333c8841 | 7478 | vector signed short vec_or (vector signed short, vector signed short); |
b0b343db JJ |
7479 | vector unsigned short vec_or (vector bool short, vector unsigned short); |
7480 | vector unsigned short vec_or (vector unsigned short, vector bool short); | |
924fcc4e JM |
7481 | vector unsigned short vec_or (vector unsigned short, |
7482 | vector unsigned short); | |
b0b343db JJ |
7483 | vector signed char vec_or (vector bool char, vector signed char); |
7484 | vector bool char vec_or (vector bool char, vector bool char); | |
7485 | vector signed char vec_or (vector signed char, vector bool char); | |
333c8841 | 7486 | vector signed char vec_or (vector signed char, vector signed char); |
b0b343db JJ |
7487 | vector unsigned char vec_or (vector bool char, vector unsigned char); |
7488 | vector unsigned char vec_or (vector unsigned char, vector bool char); | |
924fcc4e JM |
7489 | vector unsigned char vec_or (vector unsigned char, |
7490 | vector unsigned char); | |
333c8841 AH |
7491 | |
7492 | vector signed char vec_pack (vector signed short, vector signed short); | |
6e5bb5ad JM |
7493 | vector unsigned char vec_pack (vector unsigned short, |
7494 | vector unsigned short); | |
b0b343db | 7495 | vector bool char vec_pack (vector bool short, vector bool short); |
333c8841 | 7496 | vector signed short vec_pack (vector signed int, vector signed int); |
924fcc4e JM |
7497 | vector unsigned short vec_pack (vector unsigned int, |
7498 | vector unsigned int); | |
b0b343db | 7499 | vector bool short vec_pack (vector bool int, vector bool int); |
333c8841 | 7500 | |
b0b343db JJ |
7501 | vector bool short vec_vpkuwum (vector bool int, vector bool int); |
7502 | vector signed short vec_vpkuwum (vector signed int, vector signed int); | |
7503 | vector unsigned short vec_vpkuwum (vector unsigned int, | |
7504 | vector unsigned int); | |
7505 | ||
7506 | vector bool char vec_vpkuhum (vector bool short, vector bool short); | |
7507 | vector signed char vec_vpkuhum (vector signed short, | |
7508 | vector signed short); | |
7509 | vector unsigned char vec_vpkuhum (vector unsigned short, | |
7510 | vector unsigned short); | |
7511 | ||
7512 | vector pixel vec_packpx (vector unsigned int, vector unsigned int); | |
333c8841 | 7513 | |
6e5bb5ad JM |
7514 | vector unsigned char vec_packs (vector unsigned short, |
7515 | vector unsigned short); | |
333c8841 | 7516 | vector signed char vec_packs (vector signed short, vector signed short); |
924fcc4e JM |
7517 | vector unsigned short vec_packs (vector unsigned int, |
7518 | vector unsigned int); | |
333c8841 AH |
7519 | vector signed short vec_packs (vector signed int, vector signed int); |
7520 | ||
b0b343db JJ |
7521 | vector signed short vec_vpkswss (vector signed int, vector signed int); |
7522 | ||
7523 | vector unsigned short vec_vpkuwus (vector unsigned int, | |
7524 | vector unsigned int); | |
7525 | ||
7526 | vector signed char vec_vpkshss (vector signed short, | |
7527 | vector signed short); | |
7528 | ||
7529 | vector unsigned char vec_vpkuhus (vector unsigned short, | |
7530 | vector unsigned short); | |
7531 | ||
6e5bb5ad JM |
7532 | vector unsigned char vec_packsu (vector unsigned short, |
7533 | vector unsigned short); | |
924fcc4e JM |
7534 | vector unsigned char vec_packsu (vector signed short, |
7535 | vector signed short); | |
7536 | vector unsigned short vec_packsu (vector unsigned int, | |
7537 | vector unsigned int); | |
333c8841 AH |
7538 | vector unsigned short vec_packsu (vector signed int, vector signed int); |
7539 | ||
b0b343db JJ |
7540 | vector unsigned short vec_vpkswus (vector signed int, |
7541 | vector signed int); | |
7542 | ||
7543 | vector unsigned char vec_vpkshus (vector signed short, | |
7544 | vector signed short); | |
7545 | ||
7546 | vector float vec_perm (vector float, | |
7547 | vector float, | |
924fcc4e | 7548 | vector unsigned char); |
b0b343db JJ |
7549 | vector signed int vec_perm (vector signed int, |
7550 | vector signed int, | |
6e5bb5ad | 7551 | vector unsigned char); |
b0b343db JJ |
7552 | vector unsigned int vec_perm (vector unsigned int, |
7553 | vector unsigned int, | |
6e5bb5ad | 7554 | vector unsigned char); |
b0b343db JJ |
7555 | vector bool int vec_perm (vector bool int, |
7556 | vector bool int, | |
7557 | vector unsigned char); | |
7558 | vector signed short vec_perm (vector signed short, | |
7559 | vector signed short, | |
6e5bb5ad JM |
7560 | vector unsigned char); |
7561 | vector unsigned short vec_perm (vector unsigned short, | |
7562 | vector unsigned short, | |
7563 | vector unsigned char); | |
b0b343db JJ |
7564 | vector bool short vec_perm (vector bool short, |
7565 | vector bool short, | |
7566 | vector unsigned char); | |
7567 | vector pixel vec_perm (vector pixel, | |
7568 | vector pixel, | |
7569 | vector unsigned char); | |
7570 | vector signed char vec_perm (vector signed char, | |
7571 | vector signed char, | |
6e5bb5ad | 7572 | vector unsigned char); |
924fcc4e JM |
7573 | vector unsigned char vec_perm (vector unsigned char, |
7574 | vector unsigned char, | |
6e5bb5ad | 7575 | vector unsigned char); |
b0b343db JJ |
7576 | vector bool char vec_perm (vector bool char, |
7577 | vector bool char, | |
7578 | vector unsigned char); | |
333c8841 AH |
7579 | |
7580 | vector float vec_re (vector float); | |
7581 | ||
b0b343db JJ |
7582 | vector signed char vec_rl (vector signed char, |
7583 | vector unsigned char); | |
924fcc4e JM |
7584 | vector unsigned char vec_rl (vector unsigned char, |
7585 | vector unsigned char); | |
333c8841 | 7586 | vector signed short vec_rl (vector signed short, vector unsigned short); |
924fcc4e JM |
7587 | vector unsigned short vec_rl (vector unsigned short, |
7588 | vector unsigned short); | |
333c8841 AH |
7589 | vector signed int vec_rl (vector signed int, vector unsigned int); |
7590 | vector unsigned int vec_rl (vector unsigned int, vector unsigned int); | |
7591 | ||
b0b343db JJ |
7592 | vector signed int vec_vrlw (vector signed int, vector unsigned int); |
7593 | vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int); | |
7594 | ||
7595 | vector signed short vec_vrlh (vector signed short, | |
7596 | vector unsigned short); | |
7597 | vector unsigned short vec_vrlh (vector unsigned short, | |
7598 | vector unsigned short); | |
7599 | ||
7600 | vector signed char vec_vrlb (vector signed char, vector unsigned char); | |
7601 | vector unsigned char vec_vrlb (vector unsigned char, | |
7602 | vector unsigned char); | |
7603 | ||
333c8841 AH |
7604 | vector float vec_round (vector float); |
7605 | ||
7606 | vector float vec_rsqrte (vector float); | |
7607 | ||
b0b343db | 7608 | vector float vec_sel (vector float, vector float, vector bool int); |
333c8841 | 7609 | vector float vec_sel (vector float, vector float, vector unsigned int); |
b0b343db JJ |
7610 | vector signed int vec_sel (vector signed int, |
7611 | vector signed int, | |
7612 | vector bool int); | |
7613 | vector signed int vec_sel (vector signed int, | |
7614 | vector signed int, | |
6e5bb5ad | 7615 | vector unsigned int); |
b0b343db JJ |
7616 | vector unsigned int vec_sel (vector unsigned int, |
7617 | vector unsigned int, | |
7618 | vector bool int); | |
7619 | vector unsigned int vec_sel (vector unsigned int, | |
7620 | vector unsigned int, | |
6e5bb5ad | 7621 | vector unsigned int); |
b0b343db JJ |
7622 | vector bool int vec_sel (vector bool int, |
7623 | vector bool int, | |
7624 | vector bool int); | |
7625 | vector bool int vec_sel (vector bool int, | |
7626 | vector bool int, | |
7627 | vector unsigned int); | |
7628 | vector signed short vec_sel (vector signed short, | |
7629 | vector signed short, | |
7630 | vector bool short); | |
7631 | vector signed short vec_sel (vector signed short, | |
7632 | vector signed short, | |
6e5bb5ad JM |
7633 | vector unsigned short); |
7634 | vector unsigned short vec_sel (vector unsigned short, | |
924fcc4e | 7635 | vector unsigned short, |
b0b343db | 7636 | vector bool short); |
6e5bb5ad JM |
7637 | vector unsigned short vec_sel (vector unsigned short, |
7638 | vector unsigned short, | |
7639 | vector unsigned short); | |
b0b343db JJ |
7640 | vector bool short vec_sel (vector bool short, |
7641 | vector bool short, | |
7642 | vector bool short); | |
7643 | vector bool short vec_sel (vector bool short, | |
7644 | vector bool short, | |
7645 | vector unsigned short); | |
7646 | vector signed char vec_sel (vector signed char, | |
7647 | vector signed char, | |
7648 | vector bool char); | |
7649 | vector signed char vec_sel (vector signed char, | |
7650 | vector signed char, | |
6e5bb5ad | 7651 | vector unsigned char); |
924fcc4e JM |
7652 | vector unsigned char vec_sel (vector unsigned char, |
7653 | vector unsigned char, | |
b0b343db | 7654 | vector bool char); |
924fcc4e JM |
7655 | vector unsigned char vec_sel (vector unsigned char, |
7656 | vector unsigned char, | |
6e5bb5ad | 7657 | vector unsigned char); |
b0b343db JJ |
7658 | vector bool char vec_sel (vector bool char, |
7659 | vector bool char, | |
7660 | vector bool char); | |
7661 | vector bool char vec_sel (vector bool char, | |
7662 | vector bool char, | |
7663 | vector unsigned char); | |
7664 | ||
7665 | vector signed char vec_sl (vector signed char, | |
7666 | vector unsigned char); | |
924fcc4e JM |
7667 | vector unsigned char vec_sl (vector unsigned char, |
7668 | vector unsigned char); | |
333c8841 | 7669 | vector signed short vec_sl (vector signed short, vector unsigned short); |
924fcc4e JM |
7670 | vector unsigned short vec_sl (vector unsigned short, |
7671 | vector unsigned short); | |
333c8841 AH |
7672 | vector signed int vec_sl (vector signed int, vector unsigned int); |
7673 | vector unsigned int vec_sl (vector unsigned int, vector unsigned int); | |
7674 | ||
b0b343db JJ |
7675 | vector signed int vec_vslw (vector signed int, vector unsigned int); |
7676 | vector unsigned int vec_vslw (vector unsigned int, vector unsigned int); | |
7677 | ||
7678 | vector signed short vec_vslh (vector signed short, | |
7679 | vector unsigned short); | |
7680 | vector unsigned short vec_vslh (vector unsigned short, | |
7681 | vector unsigned short); | |
7682 | ||
7683 | vector signed char vec_vslb (vector signed char, vector unsigned char); | |
7684 | vector unsigned char vec_vslb (vector unsigned char, | |
7685 | vector unsigned char); | |
7686 | ||
7687 | vector float vec_sld (vector float, vector float, const int); | |
7688 | vector signed int vec_sld (vector signed int, | |
7689 | vector signed int, | |
7690 | const int); | |
7691 | vector unsigned int vec_sld (vector unsigned int, | |
7692 | vector unsigned int, | |
7693 | const int); | |
7694 | vector bool int vec_sld (vector bool int, | |
7695 | vector bool int, | |
7696 | const int); | |
7697 | vector signed short vec_sld (vector signed short, | |
7698 | vector signed short, | |
7699 | const int); | |
6e5bb5ad | 7700 | vector unsigned short vec_sld (vector unsigned short, |
b0b343db JJ |
7701 | vector unsigned short, |
7702 | const int); | |
7703 | vector bool short vec_sld (vector bool short, | |
7704 | vector bool short, | |
7705 | const int); | |
7706 | vector pixel vec_sld (vector pixel, | |
7707 | vector pixel, | |
7708 | const int); | |
7709 | vector signed char vec_sld (vector signed char, | |
7710 | vector signed char, | |
7711 | const int); | |
924fcc4e JM |
7712 | vector unsigned char vec_sld (vector unsigned char, |
7713 | vector unsigned char, | |
b0b343db JJ |
7714 | const int); |
7715 | vector bool char vec_sld (vector bool char, | |
7716 | vector bool char, | |
7717 | const int); | |
333c8841 | 7718 | |
b0b343db JJ |
7719 | vector signed int vec_sll (vector signed int, |
7720 | vector unsigned int); | |
7721 | vector signed int vec_sll (vector signed int, | |
7722 | vector unsigned short); | |
7723 | vector signed int vec_sll (vector signed int, | |
7724 | vector unsigned char); | |
7725 | vector unsigned int vec_sll (vector unsigned int, | |
7726 | vector unsigned int); | |
924fcc4e JM |
7727 | vector unsigned int vec_sll (vector unsigned int, |
7728 | vector unsigned short); | |
b0b343db JJ |
7729 | vector unsigned int vec_sll (vector unsigned int, |
7730 | vector unsigned char); | |
7731 | vector bool int vec_sll (vector bool int, | |
7732 | vector unsigned int); | |
7733 | vector bool int vec_sll (vector bool int, | |
7734 | vector unsigned short); | |
7735 | vector bool int vec_sll (vector bool int, | |
7736 | vector unsigned char); | |
7737 | vector signed short vec_sll (vector signed short, | |
7738 | vector unsigned int); | |
924fcc4e JM |
7739 | vector signed short vec_sll (vector signed short, |
7740 | vector unsigned short); | |
b0b343db JJ |
7741 | vector signed short vec_sll (vector signed short, |
7742 | vector unsigned char); | |
924fcc4e JM |
7743 | vector unsigned short vec_sll (vector unsigned short, |
7744 | vector unsigned int); | |
6e5bb5ad JM |
7745 | vector unsigned short vec_sll (vector unsigned short, |
7746 | vector unsigned short); | |
924fcc4e JM |
7747 | vector unsigned short vec_sll (vector unsigned short, |
7748 | vector unsigned char); | |
b0b343db JJ |
7749 | vector bool short vec_sll (vector bool short, vector unsigned int); |
7750 | vector bool short vec_sll (vector bool short, vector unsigned short); | |
7751 | vector bool short vec_sll (vector bool short, vector unsigned char); | |
7752 | vector pixel vec_sll (vector pixel, vector unsigned int); | |
7753 | vector pixel vec_sll (vector pixel, vector unsigned short); | |
7754 | vector pixel vec_sll (vector pixel, vector unsigned char); | |
333c8841 AH |
7755 | vector signed char vec_sll (vector signed char, vector unsigned int); |
7756 | vector signed char vec_sll (vector signed char, vector unsigned short); | |
7757 | vector signed char vec_sll (vector signed char, vector unsigned char); | |
924fcc4e JM |
7758 | vector unsigned char vec_sll (vector unsigned char, |
7759 | vector unsigned int); | |
7760 | vector unsigned char vec_sll (vector unsigned char, | |
7761 | vector unsigned short); | |
7762 | vector unsigned char vec_sll (vector unsigned char, | |
7763 | vector unsigned char); | |
b0b343db JJ |
7764 | vector bool char vec_sll (vector bool char, vector unsigned int); |
7765 | vector bool char vec_sll (vector bool char, vector unsigned short); | |
7766 | vector bool char vec_sll (vector bool char, vector unsigned char); | |
333c8841 AH |
7767 | |
7768 | vector float vec_slo (vector float, vector signed char); | |
7769 | vector float vec_slo (vector float, vector unsigned char); | |
7770 | vector signed int vec_slo (vector signed int, vector signed char); | |
7771 | vector signed int vec_slo (vector signed int, vector unsigned char); | |
7772 | vector unsigned int vec_slo (vector unsigned int, vector signed char); | |
7773 | vector unsigned int vec_slo (vector unsigned int, vector unsigned char); | |
333c8841 AH |
7774 | vector signed short vec_slo (vector signed short, vector signed char); |
7775 | vector signed short vec_slo (vector signed short, vector unsigned char); | |
924fcc4e JM |
7776 | vector unsigned short vec_slo (vector unsigned short, |
7777 | vector signed char); | |
7778 | vector unsigned short vec_slo (vector unsigned short, | |
7779 | vector unsigned char); | |
b0b343db JJ |
7780 | vector pixel vec_slo (vector pixel, vector signed char); |
7781 | vector pixel vec_slo (vector pixel, vector unsigned char); | |
333c8841 AH |
7782 | vector signed char vec_slo (vector signed char, vector signed char); |
7783 | vector signed char vec_slo (vector signed char, vector unsigned char); | |
7784 | vector unsigned char vec_slo (vector unsigned char, vector signed char); | |
924fcc4e JM |
7785 | vector unsigned char vec_slo (vector unsigned char, |
7786 | vector unsigned char); | |
333c8841 | 7787 | |
b0b343db JJ |
7788 | vector signed char vec_splat (vector signed char, const int); |
7789 | vector unsigned char vec_splat (vector unsigned char, const int); | |
7790 | vector bool char vec_splat (vector bool char, const int); | |
7791 | vector signed short vec_splat (vector signed short, const int); | |
7792 | vector unsigned short vec_splat (vector unsigned short, const int); | |
7793 | vector bool short vec_splat (vector bool short, const int); | |
7794 | vector pixel vec_splat (vector pixel, const int); | |
7795 | vector float vec_splat (vector float, const int); | |
7796 | vector signed int vec_splat (vector signed int, const int); | |
7797 | vector unsigned int vec_splat (vector unsigned int, const int); | |
7798 | vector bool int vec_splat (vector bool int, const int); | |
7799 | ||
7800 | vector float vec_vspltw (vector float, const int); | |
7801 | vector signed int vec_vspltw (vector signed int, const int); | |
7802 | vector unsigned int vec_vspltw (vector unsigned int, const int); | |
7803 | vector bool int vec_vspltw (vector bool int, const int); | |
7804 | ||
7805 | vector bool short vec_vsplth (vector bool short, const int); | |
7806 | vector signed short vec_vsplth (vector signed short, const int); | |
7807 | vector unsigned short vec_vsplth (vector unsigned short, const int); | |
7808 | vector pixel vec_vsplth (vector pixel, const int); | |
7809 | ||
7810 | vector signed char vec_vspltb (vector signed char, const int); | |
7811 | vector unsigned char vec_vspltb (vector unsigned char, const int); | |
7812 | vector bool char vec_vspltb (vector bool char, const int); | |
333c8841 | 7813 | |
b0b343db | 7814 | vector signed char vec_splat_s8 (const int); |
333c8841 | 7815 | |
b0b343db | 7816 | vector signed short vec_splat_s16 (const int); |
333c8841 | 7817 | |
b0b343db | 7818 | vector signed int vec_splat_s32 (const int); |
333c8841 | 7819 | |
b0b343db | 7820 | vector unsigned char vec_splat_u8 (const int); |
333c8841 | 7821 | |
b0b343db | 7822 | vector unsigned short vec_splat_u16 (const int); |
333c8841 | 7823 | |
b0b343db | 7824 | vector unsigned int vec_splat_u32 (const int); |
333c8841 AH |
7825 | |
7826 | vector signed char vec_sr (vector signed char, vector unsigned char); | |
924fcc4e JM |
7827 | vector unsigned char vec_sr (vector unsigned char, |
7828 | vector unsigned char); | |
b0b343db JJ |
7829 | vector signed short vec_sr (vector signed short, |
7830 | vector unsigned short); | |
924fcc4e JM |
7831 | vector unsigned short vec_sr (vector unsigned short, |
7832 | vector unsigned short); | |
333c8841 AH |
7833 | vector signed int vec_sr (vector signed int, vector unsigned int); |
7834 | vector unsigned int vec_sr (vector unsigned int, vector unsigned int); | |
7835 | ||
b0b343db JJ |
7836 | vector signed int vec_vsrw (vector signed int, vector unsigned int); |
7837 | vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int); | |
7838 | ||
7839 | vector signed short vec_vsrh (vector signed short, | |
7840 | vector unsigned short); | |
7841 | vector unsigned short vec_vsrh (vector unsigned short, | |
7842 | vector unsigned short); | |
7843 | ||
7844 | vector signed char vec_vsrb (vector signed char, vector unsigned char); | |
7845 | vector unsigned char vec_vsrb (vector unsigned char, | |
7846 | vector unsigned char); | |
7847 | ||
333c8841 | 7848 | vector signed char vec_sra (vector signed char, vector unsigned char); |
924fcc4e JM |
7849 | vector unsigned char vec_sra (vector unsigned char, |
7850 | vector unsigned char); | |
7851 | vector signed short vec_sra (vector signed short, | |
7852 | vector unsigned short); | |
6e5bb5ad JM |
7853 | vector unsigned short vec_sra (vector unsigned short, |
7854 | vector unsigned short); | |
333c8841 AH |
7855 | vector signed int vec_sra (vector signed int, vector unsigned int); |
7856 | vector unsigned int vec_sra (vector unsigned int, vector unsigned int); | |
7857 | ||
b0b343db JJ |
7858 | vector signed int vec_vsraw (vector signed int, vector unsigned int); |
7859 | vector unsigned int vec_vsraw (vector unsigned int, | |
7860 | vector unsigned int); | |
7861 | ||
7862 | vector signed short vec_vsrah (vector signed short, | |
7863 | vector unsigned short); | |
7864 | vector unsigned short vec_vsrah (vector unsigned short, | |
7865 | vector unsigned short); | |
7866 | ||
7867 | vector signed char vec_vsrab (vector signed char, vector unsigned char); | |
7868 | vector unsigned char vec_vsrab (vector unsigned char, | |
7869 | vector unsigned char); | |
7870 | ||
333c8841 AH |
7871 | vector signed int vec_srl (vector signed int, vector unsigned int); |
7872 | vector signed int vec_srl (vector signed int, vector unsigned short); | |
7873 | vector signed int vec_srl (vector signed int, vector unsigned char); | |
7874 | vector unsigned int vec_srl (vector unsigned int, vector unsigned int); | |
924fcc4e JM |
7875 | vector unsigned int vec_srl (vector unsigned int, |
7876 | vector unsigned short); | |
333c8841 | 7877 | vector unsigned int vec_srl (vector unsigned int, vector unsigned char); |
b0b343db JJ |
7878 | vector bool int vec_srl (vector bool int, vector unsigned int); |
7879 | vector bool int vec_srl (vector bool int, vector unsigned short); | |
7880 | vector bool int vec_srl (vector bool int, vector unsigned char); | |
333c8841 | 7881 | vector signed short vec_srl (vector signed short, vector unsigned int); |
924fcc4e JM |
7882 | vector signed short vec_srl (vector signed short, |
7883 | vector unsigned short); | |
333c8841 | 7884 | vector signed short vec_srl (vector signed short, vector unsigned char); |
924fcc4e JM |
7885 | vector unsigned short vec_srl (vector unsigned short, |
7886 | vector unsigned int); | |
6e5bb5ad JM |
7887 | vector unsigned short vec_srl (vector unsigned short, |
7888 | vector unsigned short); | |
924fcc4e JM |
7889 | vector unsigned short vec_srl (vector unsigned short, |
7890 | vector unsigned char); | |
b0b343db JJ |
7891 | vector bool short vec_srl (vector bool short, vector unsigned int); |
7892 | vector bool short vec_srl (vector bool short, vector unsigned short); | |
7893 | vector bool short vec_srl (vector bool short, vector unsigned char); | |
7894 | vector pixel vec_srl (vector pixel, vector unsigned int); | |
7895 | vector pixel vec_srl (vector pixel, vector unsigned short); | |
7896 | vector pixel vec_srl (vector pixel, vector unsigned char); | |
333c8841 AH |
7897 | vector signed char vec_srl (vector signed char, vector unsigned int); |
7898 | vector signed char vec_srl (vector signed char, vector unsigned short); | |
7899 | vector signed char vec_srl (vector signed char, vector unsigned char); | |
924fcc4e JM |
7900 | vector unsigned char vec_srl (vector unsigned char, |
7901 | vector unsigned int); | |
7902 | vector unsigned char vec_srl (vector unsigned char, | |
7903 | vector unsigned short); | |
7904 | vector unsigned char vec_srl (vector unsigned char, | |
7905 | vector unsigned char); | |
b0b343db JJ |
7906 | vector bool char vec_srl (vector bool char, vector unsigned int); |
7907 | vector bool char vec_srl (vector bool char, vector unsigned short); | |
7908 | vector bool char vec_srl (vector bool char, vector unsigned char); | |
333c8841 AH |
7909 | |
7910 | vector float vec_sro (vector float, vector signed char); | |
7911 | vector float vec_sro (vector float, vector unsigned char); | |
7912 | vector signed int vec_sro (vector signed int, vector signed char); | |
7913 | vector signed int vec_sro (vector signed int, vector unsigned char); | |
7914 | vector unsigned int vec_sro (vector unsigned int, vector signed char); | |
7915 | vector unsigned int vec_sro (vector unsigned int, vector unsigned char); | |
333c8841 AH |
7916 | vector signed short vec_sro (vector signed short, vector signed char); |
7917 | vector signed short vec_sro (vector signed short, vector unsigned char); | |
924fcc4e JM |
7918 | vector unsigned short vec_sro (vector unsigned short, |
7919 | vector signed char); | |
7920 | vector unsigned short vec_sro (vector unsigned short, | |
7921 | vector unsigned char); | |
b0b343db JJ |
7922 | vector pixel vec_sro (vector pixel, vector signed char); |
7923 | vector pixel vec_sro (vector pixel, vector unsigned char); | |
333c8841 AH |
7924 | vector signed char vec_sro (vector signed char, vector signed char); |
7925 | vector signed char vec_sro (vector signed char, vector unsigned char); | |
7926 | vector unsigned char vec_sro (vector unsigned char, vector signed char); | |
924fcc4e JM |
7927 | vector unsigned char vec_sro (vector unsigned char, |
7928 | vector unsigned char); | |
333c8841 | 7929 | |
333c8841 | 7930 | void vec_st (vector float, int, vector float *); |
b0b343db JJ |
7931 | void vec_st (vector float, int, float *); |
7932 | void vec_st (vector signed int, int, vector signed int *); | |
333c8841 | 7933 | void vec_st (vector signed int, int, int *); |
333c8841 | 7934 | void vec_st (vector unsigned int, int, vector unsigned int *); |
b0b343db JJ |
7935 | void vec_st (vector unsigned int, int, unsigned int *); |
7936 | void vec_st (vector bool int, int, vector bool int *); | |
7937 | void vec_st (vector bool int, int, unsigned int *); | |
7938 | void vec_st (vector bool int, int, int *); | |
333c8841 | 7939 | void vec_st (vector signed short, int, vector signed short *); |
b0b343db | 7940 | void vec_st (vector signed short, int, short *); |
333c8841 | 7941 | void vec_st (vector unsigned short, int, vector unsigned short *); |
b0b343db JJ |
7942 | void vec_st (vector unsigned short, int, unsigned short *); |
7943 | void vec_st (vector bool short, int, vector bool short *); | |
7944 | void vec_st (vector bool short, int, unsigned short *); | |
7945 | void vec_st (vector pixel, int, vector pixel *); | |
7946 | void vec_st (vector pixel, int, unsigned short *); | |
7947 | void vec_st (vector pixel, int, short *); | |
7948 | void vec_st (vector bool short, int, short *); | |
333c8841 | 7949 | void vec_st (vector signed char, int, vector signed char *); |
b0b343db | 7950 | void vec_st (vector signed char, int, signed char *); |
333c8841 | 7951 | void vec_st (vector unsigned char, int, vector unsigned char *); |
b0b343db JJ |
7952 | void vec_st (vector unsigned char, int, unsigned char *); |
7953 | void vec_st (vector bool char, int, vector bool char *); | |
7954 | void vec_st (vector bool char, int, unsigned char *); | |
7955 | void vec_st (vector bool char, int, signed char *); | |
333c8841 | 7956 | |
333c8841 AH |
7957 | void vec_ste (vector signed char, int, signed char *); |
7958 | void vec_ste (vector unsigned char, int, unsigned char *); | |
b0b343db JJ |
7959 | void vec_ste (vector bool char, int, signed char *); |
7960 | void vec_ste (vector bool char, int, unsigned char *); | |
333c8841 | 7961 | void vec_ste (vector signed short, int, short *); |
b0b343db JJ |
7962 | void vec_ste (vector unsigned short, int, unsigned short *); |
7963 | void vec_ste (vector bool short, int, short *); | |
7964 | void vec_ste (vector bool short, int, unsigned short *); | |
7965 | void vec_ste (vector pixel, int, short *); | |
7966 | void vec_ste (vector pixel, int, unsigned short *); | |
7967 | void vec_ste (vector float, int, float *); | |
333c8841 AH |
7968 | void vec_ste (vector signed int, int, int *); |
7969 | void vec_ste (vector unsigned int, int, unsigned int *); | |
b0b343db JJ |
7970 | void vec_ste (vector bool int, int, int *); |
7971 | void vec_ste (vector bool int, int, unsigned int *); | |
7972 | ||
7973 | void vec_stvewx (vector float, int, float *); | |
7974 | void vec_stvewx (vector signed int, int, int *); | |
7975 | void vec_stvewx (vector unsigned int, int, unsigned int *); | |
7976 | void vec_stvewx (vector bool int, int, int *); | |
7977 | void vec_stvewx (vector bool int, int, unsigned int *); | |
7978 | ||
7979 | void vec_stvehx (vector signed short, int, short *); | |
7980 | void vec_stvehx (vector unsigned short, int, unsigned short *); | |
7981 | void vec_stvehx (vector bool short, int, short *); | |
7982 | void vec_stvehx (vector bool short, int, unsigned short *); | |
7983 | void vec_stvehx (vector pixel, int, short *); | |
7984 | void vec_stvehx (vector pixel, int, unsigned short *); | |
7985 | ||
7986 | void vec_stvebx (vector signed char, int, signed char *); | |
7987 | void vec_stvebx (vector unsigned char, int, unsigned char *); | |
7988 | void vec_stvebx (vector bool char, int, signed char *); | |
7989 | void vec_stvebx (vector bool char, int, unsigned char *); | |
333c8841 AH |
7990 | |
7991 | void vec_stl (vector float, int, vector float *); | |
7992 | void vec_stl (vector float, int, float *); | |
7993 | void vec_stl (vector signed int, int, vector signed int *); | |
7994 | void vec_stl (vector signed int, int, int *); | |
333c8841 AH |
7995 | void vec_stl (vector unsigned int, int, vector unsigned int *); |
7996 | void vec_stl (vector unsigned int, int, unsigned int *); | |
b0b343db JJ |
7997 | void vec_stl (vector bool int, int, vector bool int *); |
7998 | void vec_stl (vector bool int, int, unsigned int *); | |
7999 | void vec_stl (vector bool int, int, int *); | |
333c8841 | 8000 | void vec_stl (vector signed short, int, vector signed short *); |
b0b343db JJ |
8001 | void vec_stl (vector signed short, int, short *); |
8002 | void vec_stl (vector unsigned short, int, vector unsigned short *); | |
333c8841 | 8003 | void vec_stl (vector unsigned short, int, unsigned short *); |
b0b343db JJ |
8004 | void vec_stl (vector bool short, int, vector bool short *); |
8005 | void vec_stl (vector bool short, int, unsigned short *); | |
8006 | void vec_stl (vector bool short, int, short *); | |
8007 | void vec_stl (vector pixel, int, vector pixel *); | |
8008 | void vec_stl (vector pixel, int, unsigned short *); | |
8009 | void vec_stl (vector pixel, int, short *); | |
333c8841 | 8010 | void vec_stl (vector signed char, int, vector signed char *); |
b0b343db | 8011 | void vec_stl (vector signed char, int, signed char *); |
333c8841 | 8012 | void vec_stl (vector unsigned char, int, vector unsigned char *); |
b0b343db JJ |
8013 | void vec_stl (vector unsigned char, int, unsigned char *); |
8014 | void vec_stl (vector bool char, int, vector bool char *); | |
8015 | void vec_stl (vector bool char, int, unsigned char *); | |
8016 | void vec_stl (vector bool char, int, signed char *); | |
333c8841 | 8017 | |
b0b343db JJ |
8018 | vector signed char vec_sub (vector bool char, vector signed char); |
8019 | vector signed char vec_sub (vector signed char, vector bool char); | |
333c8841 | 8020 | vector signed char vec_sub (vector signed char, vector signed char); |
b0b343db JJ |
8021 | vector unsigned char vec_sub (vector bool char, vector unsigned char); |
8022 | vector unsigned char vec_sub (vector unsigned char, vector bool char); | |
924fcc4e JM |
8023 | vector unsigned char vec_sub (vector unsigned char, |
8024 | vector unsigned char); | |
b0b343db JJ |
8025 | vector signed short vec_sub (vector bool short, vector signed short); |
8026 | vector signed short vec_sub (vector signed short, vector bool short); | |
333c8841 | 8027 | vector signed short vec_sub (vector signed short, vector signed short); |
b0b343db | 8028 | vector unsigned short vec_sub (vector bool short, |
924fcc4e JM |
8029 | vector unsigned short); |
8030 | vector unsigned short vec_sub (vector unsigned short, | |
b0b343db | 8031 | vector bool short); |
6e5bb5ad JM |
8032 | vector unsigned short vec_sub (vector unsigned short, |
8033 | vector unsigned short); | |
b0b343db JJ |
8034 | vector signed int vec_sub (vector bool int, vector signed int); |
8035 | vector signed int vec_sub (vector signed int, vector bool int); | |
333c8841 | 8036 | vector signed int vec_sub (vector signed int, vector signed int); |
b0b343db JJ |
8037 | vector unsigned int vec_sub (vector bool int, vector unsigned int); |
8038 | vector unsigned int vec_sub (vector unsigned int, vector bool int); | |
333c8841 AH |
8039 | vector unsigned int vec_sub (vector unsigned int, vector unsigned int); |
8040 | vector float vec_sub (vector float, vector float); | |
8041 | ||
b0b343db JJ |
8042 | vector float vec_vsubfp (vector float, vector float); |
8043 | ||
8044 | vector signed int vec_vsubuwm (vector bool int, vector signed int); | |
8045 | vector signed int vec_vsubuwm (vector signed int, vector bool int); | |
8046 | vector signed int vec_vsubuwm (vector signed int, vector signed int); | |
8047 | vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int); | |
8048 | vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int); | |
8049 | vector unsigned int vec_vsubuwm (vector unsigned int, | |
8050 | vector unsigned int); | |
8051 | ||
8052 | vector signed short vec_vsubuhm (vector bool short, | |
8053 | vector signed short); | |
8054 | vector signed short vec_vsubuhm (vector signed short, | |
8055 | vector bool short); | |
8056 | vector signed short vec_vsubuhm (vector signed short, | |
8057 | vector signed short); | |
8058 | vector unsigned short vec_vsubuhm (vector bool short, | |
8059 | vector unsigned short); | |
8060 | vector unsigned short vec_vsubuhm (vector unsigned short, | |
8061 | vector bool short); | |
8062 | vector unsigned short vec_vsubuhm (vector unsigned short, | |
8063 | vector unsigned short); | |
8064 | ||
8065 | vector signed char vec_vsububm (vector bool char, vector signed char); | |
8066 | vector signed char vec_vsububm (vector signed char, vector bool char); | |
8067 | vector signed char vec_vsububm (vector signed char, vector signed char); | |
8068 | vector unsigned char vec_vsububm (vector bool char, | |
8069 | vector unsigned char); | |
8070 | vector unsigned char vec_vsububm (vector unsigned char, | |
8071 | vector bool char); | |
8072 | vector unsigned char vec_vsububm (vector unsigned char, | |
8073 | vector unsigned char); | |
8074 | ||
333c8841 AH |
8075 | vector unsigned int vec_subc (vector unsigned int, vector unsigned int); |
8076 | ||
b0b343db JJ |
8077 | vector unsigned char vec_subs (vector bool char, vector unsigned char); |
8078 | vector unsigned char vec_subs (vector unsigned char, vector bool char); | |
924fcc4e JM |
8079 | vector unsigned char vec_subs (vector unsigned char, |
8080 | vector unsigned char); | |
b0b343db JJ |
8081 | vector signed char vec_subs (vector bool char, vector signed char); |
8082 | vector signed char vec_subs (vector signed char, vector bool char); | |
333c8841 | 8083 | vector signed char vec_subs (vector signed char, vector signed char); |
b0b343db | 8084 | vector unsigned short vec_subs (vector bool short, |
924fcc4e JM |
8085 | vector unsigned short); |
8086 | vector unsigned short vec_subs (vector unsigned short, | |
b0b343db | 8087 | vector bool short); |
6e5bb5ad JM |
8088 | vector unsigned short vec_subs (vector unsigned short, |
8089 | vector unsigned short); | |
b0b343db JJ |
8090 | vector signed short vec_subs (vector bool short, vector signed short); |
8091 | vector signed short vec_subs (vector signed short, vector bool short); | |
333c8841 | 8092 | vector signed short vec_subs (vector signed short, vector signed short); |
b0b343db JJ |
8093 | vector unsigned int vec_subs (vector bool int, vector unsigned int); |
8094 | vector unsigned int vec_subs (vector unsigned int, vector bool int); | |
333c8841 | 8095 | vector unsigned int vec_subs (vector unsigned int, vector unsigned int); |
b0b343db JJ |
8096 | vector signed int vec_subs (vector bool int, vector signed int); |
8097 | vector signed int vec_subs (vector signed int, vector bool int); | |
333c8841 AH |
8098 | vector signed int vec_subs (vector signed int, vector signed int); |
8099 | ||
b0b343db JJ |
8100 | vector signed int vec_vsubsws (vector bool int, vector signed int); |
8101 | vector signed int vec_vsubsws (vector signed int, vector bool int); | |
8102 | vector signed int vec_vsubsws (vector signed int, vector signed int); | |
8103 | ||
8104 | vector unsigned int vec_vsubuws (vector bool int, vector unsigned int); | |
8105 | vector unsigned int vec_vsubuws (vector unsigned int, vector bool int); | |
8106 | vector unsigned int vec_vsubuws (vector unsigned int, | |
8107 | vector unsigned int); | |
8108 | ||
8109 | vector signed short vec_vsubshs (vector bool short, | |
8110 | vector signed short); | |
8111 | vector signed short vec_vsubshs (vector signed short, | |
8112 | vector bool short); | |
8113 | vector signed short vec_vsubshs (vector signed short, | |
8114 | vector signed short); | |
8115 | ||
8116 | vector unsigned short vec_vsubuhs (vector bool short, | |
8117 | vector unsigned short); | |
8118 | vector unsigned short vec_vsubuhs (vector unsigned short, | |
8119 | vector bool short); | |
8120 | vector unsigned short vec_vsubuhs (vector unsigned short, | |
8121 | vector unsigned short); | |
8122 | ||
8123 | vector signed char vec_vsubsbs (vector bool char, vector signed char); | |
8124 | vector signed char vec_vsubsbs (vector signed char, vector bool char); | |
8125 | vector signed char vec_vsubsbs (vector signed char, vector signed char); | |
8126 | ||
8127 | vector unsigned char vec_vsububs (vector bool char, | |
8128 | vector unsigned char); | |
8129 | vector unsigned char vec_vsububs (vector unsigned char, | |
8130 | vector bool char); | |
8131 | vector unsigned char vec_vsububs (vector unsigned char, | |
8132 | vector unsigned char); | |
8133 | ||
924fcc4e JM |
8134 | vector unsigned int vec_sum4s (vector unsigned char, |
8135 | vector unsigned int); | |
333c8841 AH |
8136 | vector signed int vec_sum4s (vector signed char, vector signed int); |
8137 | vector signed int vec_sum4s (vector signed short, vector signed int); | |
8138 | ||
b0b343db JJ |
8139 | vector signed int vec_vsum4shs (vector signed short, vector signed int); |
8140 | ||
8141 | vector signed int vec_vsum4sbs (vector signed char, vector signed int); | |
8142 | ||
8143 | vector unsigned int vec_vsum4ubs (vector unsigned char, | |
8144 | vector unsigned int); | |
8145 | ||
333c8841 AH |
8146 | vector signed int vec_sum2s (vector signed int, vector signed int); |
8147 | ||
8148 | vector signed int vec_sums (vector signed int, vector signed int); | |
8149 | ||
8150 | vector float vec_trunc (vector float); | |
8151 | ||
8152 | vector signed short vec_unpackh (vector signed char); | |
b0b343db | 8153 | vector bool short vec_unpackh (vector bool char); |
333c8841 | 8154 | vector signed int vec_unpackh (vector signed short); |
b0b343db JJ |
8155 | vector bool int vec_unpackh (vector bool short); |
8156 | vector unsigned int vec_unpackh (vector pixel); | |
8157 | ||
8158 | vector bool int vec_vupkhsh (vector bool short); | |
8159 | vector signed int vec_vupkhsh (vector signed short); | |
8160 | ||
8161 | vector unsigned int vec_vupkhpx (vector pixel); | |
8162 | ||
8163 | vector bool short vec_vupkhsb (vector bool char); | |
8164 | vector signed short vec_vupkhsb (vector signed char); | |
333c8841 AH |
8165 | |
8166 | vector signed short vec_unpackl (vector signed char); | |
b0b343db JJ |
8167 | vector bool short vec_unpackl (vector bool char); |
8168 | vector unsigned int vec_unpackl (vector pixel); | |
333c8841 | 8169 | vector signed int vec_unpackl (vector signed short); |
b0b343db JJ |
8170 | vector bool int vec_unpackl (vector bool short); |
8171 | ||
8172 | vector unsigned int vec_vupklpx (vector pixel); | |
8173 | ||
8174 | vector bool int vec_vupklsh (vector bool short); | |
8175 | vector signed int vec_vupklsh (vector signed short); | |
8176 | ||
8177 | vector bool short vec_vupklsb (vector bool char); | |
8178 | vector signed short vec_vupklsb (vector signed char); | |
333c8841 AH |
8179 | |
8180 | vector float vec_xor (vector float, vector float); | |
b0b343db JJ |
8181 | vector float vec_xor (vector float, vector bool int); |
8182 | vector float vec_xor (vector bool int, vector float); | |
8183 | vector bool int vec_xor (vector bool int, vector bool int); | |
8184 | vector signed int vec_xor (vector bool int, vector signed int); | |
8185 | vector signed int vec_xor (vector signed int, vector bool int); | |
333c8841 | 8186 | vector signed int vec_xor (vector signed int, vector signed int); |
b0b343db JJ |
8187 | vector unsigned int vec_xor (vector bool int, vector unsigned int); |
8188 | vector unsigned int vec_xor (vector unsigned int, vector bool int); | |
333c8841 | 8189 | vector unsigned int vec_xor (vector unsigned int, vector unsigned int); |
b0b343db JJ |
8190 | vector bool short vec_xor (vector bool short, vector bool short); |
8191 | vector signed short vec_xor (vector bool short, vector signed short); | |
8192 | vector signed short vec_xor (vector signed short, vector bool short); | |
333c8841 | 8193 | vector signed short vec_xor (vector signed short, vector signed short); |
b0b343db | 8194 | vector unsigned short vec_xor (vector bool short, |
924fcc4e JM |
8195 | vector unsigned short); |
8196 | vector unsigned short vec_xor (vector unsigned short, | |
b0b343db | 8197 | vector bool short); |
6e5bb5ad JM |
8198 | vector unsigned short vec_xor (vector unsigned short, |
8199 | vector unsigned short); | |
b0b343db JJ |
8200 | vector signed char vec_xor (vector bool char, vector signed char); |
8201 | vector bool char vec_xor (vector bool char, vector bool char); | |
8202 | vector signed char vec_xor (vector signed char, vector bool char); | |
333c8841 | 8203 | vector signed char vec_xor (vector signed char, vector signed char); |
b0b343db JJ |
8204 | vector unsigned char vec_xor (vector bool char, vector unsigned char); |
8205 | vector unsigned char vec_xor (vector unsigned char, vector bool char); | |
924fcc4e JM |
8206 | vector unsigned char vec_xor (vector unsigned char, |
8207 | vector unsigned char); | |
333c8841 | 8208 | |
b0b343db JJ |
8209 | int vec_all_eq (vector signed char, vector bool char); |
8210 | int vec_all_eq (vector signed char, vector signed char); | |
8211 | int vec_all_eq (vector unsigned char, vector bool char); | |
8212 | int vec_all_eq (vector unsigned char, vector unsigned char); | |
8213 | int vec_all_eq (vector bool char, vector bool char); | |
8214 | int vec_all_eq (vector bool char, vector unsigned char); | |
8215 | int vec_all_eq (vector bool char, vector signed char); | |
8216 | int vec_all_eq (vector signed short, vector bool short); | |
8217 | int vec_all_eq (vector signed short, vector signed short); | |
8218 | int vec_all_eq (vector unsigned short, vector bool short); | |
8219 | int vec_all_eq (vector unsigned short, vector unsigned short); | |
8220 | int vec_all_eq (vector bool short, vector bool short); | |
8221 | int vec_all_eq (vector bool short, vector unsigned short); | |
8222 | int vec_all_eq (vector bool short, vector signed short); | |
8223 | int vec_all_eq (vector pixel, vector pixel); | |
8224 | int vec_all_eq (vector signed int, vector bool int); | |
8225 | int vec_all_eq (vector signed int, vector signed int); | |
8226 | int vec_all_eq (vector unsigned int, vector bool int); | |
8227 | int vec_all_eq (vector unsigned int, vector unsigned int); | |
8228 | int vec_all_eq (vector bool int, vector bool int); | |
8229 | int vec_all_eq (vector bool int, vector unsigned int); | |
8230 | int vec_all_eq (vector bool int, vector signed int); | |
8231 | int vec_all_eq (vector float, vector float); | |
8232 | ||
8233 | int vec_all_ge (vector bool char, vector unsigned char); | |
8234 | int vec_all_ge (vector unsigned char, vector bool char); | |
8235 | int vec_all_ge (vector unsigned char, vector unsigned char); | |
8236 | int vec_all_ge (vector bool char, vector signed char); | |
8237 | int vec_all_ge (vector signed char, vector bool char); | |
8238 | int vec_all_ge (vector signed char, vector signed char); | |
8239 | int vec_all_ge (vector bool short, vector unsigned short); | |
8240 | int vec_all_ge (vector unsigned short, vector bool short); | |
8241 | int vec_all_ge (vector unsigned short, vector unsigned short); | |
8242 | int vec_all_ge (vector signed short, vector signed short); | |
8243 | int vec_all_ge (vector bool short, vector signed short); | |
8244 | int vec_all_ge (vector signed short, vector bool short); | |
8245 | int vec_all_ge (vector bool int, vector unsigned int); | |
8246 | int vec_all_ge (vector unsigned int, vector bool int); | |
8247 | int vec_all_ge (vector unsigned int, vector unsigned int); | |
8248 | int vec_all_ge (vector bool int, vector signed int); | |
8249 | int vec_all_ge (vector signed int, vector bool int); | |
8250 | int vec_all_ge (vector signed int, vector signed int); | |
8251 | int vec_all_ge (vector float, vector float); | |
8252 | ||
8253 | int vec_all_gt (vector bool char, vector unsigned char); | |
8254 | int vec_all_gt (vector unsigned char, vector bool char); | |
8255 | int vec_all_gt (vector unsigned char, vector unsigned char); | |
8256 | int vec_all_gt (vector bool char, vector signed char); | |
8257 | int vec_all_gt (vector signed char, vector bool char); | |
8258 | int vec_all_gt (vector signed char, vector signed char); | |
8259 | int vec_all_gt (vector bool short, vector unsigned short); | |
8260 | int vec_all_gt (vector unsigned short, vector bool short); | |
8261 | int vec_all_gt (vector unsigned short, vector unsigned short); | |
8262 | int vec_all_gt (vector bool short, vector signed short); | |
8263 | int vec_all_gt (vector signed short, vector bool short); | |
8264 | int vec_all_gt (vector signed short, vector signed short); | |
8265 | int vec_all_gt (vector bool int, vector unsigned int); | |
8266 | int vec_all_gt (vector unsigned int, vector bool int); | |
8267 | int vec_all_gt (vector unsigned int, vector unsigned int); | |
8268 | int vec_all_gt (vector bool int, vector signed int); | |
8269 | int vec_all_gt (vector signed int, vector bool int); | |
8270 | int vec_all_gt (vector signed int, vector signed int); | |
8271 | int vec_all_gt (vector float, vector float); | |
8272 | ||
8273 | int vec_all_in (vector float, vector float); | |
8274 | ||
8275 | int vec_all_le (vector bool char, vector unsigned char); | |
8276 | int vec_all_le (vector unsigned char, vector bool char); | |
8277 | int vec_all_le (vector unsigned char, vector unsigned char); | |
8278 | int vec_all_le (vector bool char, vector signed char); | |
8279 | int vec_all_le (vector signed char, vector bool char); | |
8280 | int vec_all_le (vector signed char, vector signed char); | |
8281 | int vec_all_le (vector bool short, vector unsigned short); | |
8282 | int vec_all_le (vector unsigned short, vector bool short); | |
8283 | int vec_all_le (vector unsigned short, vector unsigned short); | |
8284 | int vec_all_le (vector bool short, vector signed short); | |
8285 | int vec_all_le (vector signed short, vector bool short); | |
8286 | int vec_all_le (vector signed short, vector signed short); | |
8287 | int vec_all_le (vector bool int, vector unsigned int); | |
8288 | int vec_all_le (vector unsigned int, vector bool int); | |
8289 | int vec_all_le (vector unsigned int, vector unsigned int); | |
8290 | int vec_all_le (vector bool int, vector signed int); | |
8291 | int vec_all_le (vector signed int, vector bool int); | |
8292 | int vec_all_le (vector signed int, vector signed int); | |
8293 | int vec_all_le (vector float, vector float); | |
8294 | ||
8295 | int vec_all_lt (vector bool char, vector unsigned char); | |
8296 | int vec_all_lt (vector unsigned char, vector bool char); | |
8297 | int vec_all_lt (vector unsigned char, vector unsigned char); | |
8298 | int vec_all_lt (vector bool char, vector signed char); | |
8299 | int vec_all_lt (vector signed char, vector bool char); | |
8300 | int vec_all_lt (vector signed char, vector signed char); | |
8301 | int vec_all_lt (vector bool short, vector unsigned short); | |
8302 | int vec_all_lt (vector unsigned short, vector bool short); | |
8303 | int vec_all_lt (vector unsigned short, vector unsigned short); | |
8304 | int vec_all_lt (vector bool short, vector signed short); | |
8305 | int vec_all_lt (vector signed short, vector bool short); | |
8306 | int vec_all_lt (vector signed short, vector signed short); | |
8307 | int vec_all_lt (vector bool int, vector unsigned int); | |
8308 | int vec_all_lt (vector unsigned int, vector bool int); | |
8309 | int vec_all_lt (vector unsigned int, vector unsigned int); | |
8310 | int vec_all_lt (vector bool int, vector signed int); | |
8311 | int vec_all_lt (vector signed int, vector bool int); | |
8312 | int vec_all_lt (vector signed int, vector signed int); | |
8313 | int vec_all_lt (vector float, vector float); | |
8314 | ||
8315 | int vec_all_nan (vector float); | |
8316 | ||
8317 | int vec_all_ne (vector signed char, vector bool char); | |
8318 | int vec_all_ne (vector signed char, vector signed char); | |
8319 | int vec_all_ne (vector unsigned char, vector bool char); | |
8320 | int vec_all_ne (vector unsigned char, vector unsigned char); | |
8321 | int vec_all_ne (vector bool char, vector bool char); | |
8322 | int vec_all_ne (vector bool char, vector unsigned char); | |
8323 | int vec_all_ne (vector bool char, vector signed char); | |
8324 | int vec_all_ne (vector signed short, vector bool short); | |
8325 | int vec_all_ne (vector signed short, vector signed short); | |
8326 | int vec_all_ne (vector unsigned short, vector bool short); | |
8327 | int vec_all_ne (vector unsigned short, vector unsigned short); | |
8328 | int vec_all_ne (vector bool short, vector bool short); | |
8329 | int vec_all_ne (vector bool short, vector unsigned short); | |
8330 | int vec_all_ne (vector bool short, vector signed short); | |
8331 | int vec_all_ne (vector pixel, vector pixel); | |
8332 | int vec_all_ne (vector signed int, vector bool int); | |
8333 | int vec_all_ne (vector signed int, vector signed int); | |
8334 | int vec_all_ne (vector unsigned int, vector bool int); | |
8335 | int vec_all_ne (vector unsigned int, vector unsigned int); | |
8336 | int vec_all_ne (vector bool int, vector bool int); | |
8337 | int vec_all_ne (vector bool int, vector unsigned int); | |
8338 | int vec_all_ne (vector bool int, vector signed int); | |
8339 | int vec_all_ne (vector float, vector float); | |
8340 | ||
8341 | int vec_all_nge (vector float, vector float); | |
8342 | ||
8343 | int vec_all_ngt (vector float, vector float); | |
8344 | ||
8345 | int vec_all_nle (vector float, vector float); | |
8346 | ||
8347 | int vec_all_nlt (vector float, vector float); | |
8348 | ||
8349 | int vec_all_numeric (vector float); | |
8350 | ||
8351 | int vec_any_eq (vector signed char, vector bool char); | |
8352 | int vec_any_eq (vector signed char, vector signed char); | |
8353 | int vec_any_eq (vector unsigned char, vector bool char); | |
8354 | int vec_any_eq (vector unsigned char, vector unsigned char); | |
8355 | int vec_any_eq (vector bool char, vector bool char); | |
8356 | int vec_any_eq (vector bool char, vector unsigned char); | |
8357 | int vec_any_eq (vector bool char, vector signed char); | |
8358 | int vec_any_eq (vector signed short, vector bool short); | |
8359 | int vec_any_eq (vector signed short, vector signed short); | |
8360 | int vec_any_eq (vector unsigned short, vector bool short); | |
8361 | int vec_any_eq (vector unsigned short, vector unsigned short); | |
8362 | int vec_any_eq (vector bool short, vector bool short); | |
8363 | int vec_any_eq (vector bool short, vector unsigned short); | |
8364 | int vec_any_eq (vector bool short, vector signed short); | |
8365 | int vec_any_eq (vector pixel, vector pixel); | |
8366 | int vec_any_eq (vector signed int, vector bool int); | |
8367 | int vec_any_eq (vector signed int, vector signed int); | |
8368 | int vec_any_eq (vector unsigned int, vector bool int); | |
8369 | int vec_any_eq (vector unsigned int, vector unsigned int); | |
8370 | int vec_any_eq (vector bool int, vector bool int); | |
8371 | int vec_any_eq (vector bool int, vector unsigned int); | |
8372 | int vec_any_eq (vector bool int, vector signed int); | |
8373 | int vec_any_eq (vector float, vector float); | |
8374 | ||
8375 | int vec_any_ge (vector signed char, vector bool char); | |
8376 | int vec_any_ge (vector unsigned char, vector bool char); | |
8377 | int vec_any_ge (vector unsigned char, vector unsigned char); | |
8378 | int vec_any_ge (vector signed char, vector signed char); | |
8379 | int vec_any_ge (vector bool char, vector unsigned char); | |
8380 | int vec_any_ge (vector bool char, vector signed char); | |
8381 | int vec_any_ge (vector unsigned short, vector bool short); | |
8382 | int vec_any_ge (vector unsigned short, vector unsigned short); | |
8383 | int vec_any_ge (vector signed short, vector signed short); | |
8384 | int vec_any_ge (vector signed short, vector bool short); | |
8385 | int vec_any_ge (vector bool short, vector unsigned short); | |
8386 | int vec_any_ge (vector bool short, vector signed short); | |
8387 | int vec_any_ge (vector signed int, vector bool int); | |
8388 | int vec_any_ge (vector unsigned int, vector bool int); | |
8389 | int vec_any_ge (vector unsigned int, vector unsigned int); | |
8390 | int vec_any_ge (vector signed int, vector signed int); | |
8391 | int vec_any_ge (vector bool int, vector unsigned int); | |
8392 | int vec_any_ge (vector bool int, vector signed int); | |
8393 | int vec_any_ge (vector float, vector float); | |
8394 | ||
8395 | int vec_any_gt (vector bool char, vector unsigned char); | |
8396 | int vec_any_gt (vector unsigned char, vector bool char); | |
8397 | int vec_any_gt (vector unsigned char, vector unsigned char); | |
8398 | int vec_any_gt (vector bool char, vector signed char); | |
8399 | int vec_any_gt (vector signed char, vector bool char); | |
8400 | int vec_any_gt (vector signed char, vector signed char); | |
8401 | int vec_any_gt (vector bool short, vector unsigned short); | |
8402 | int vec_any_gt (vector unsigned short, vector bool short); | |
8403 | int vec_any_gt (vector unsigned short, vector unsigned short); | |
8404 | int vec_any_gt (vector bool short, vector signed short); | |
8405 | int vec_any_gt (vector signed short, vector bool short); | |
8406 | int vec_any_gt (vector signed short, vector signed short); | |
8407 | int vec_any_gt (vector bool int, vector unsigned int); | |
8408 | int vec_any_gt (vector unsigned int, vector bool int); | |
8409 | int vec_any_gt (vector unsigned int, vector unsigned int); | |
8410 | int vec_any_gt (vector bool int, vector signed int); | |
8411 | int vec_any_gt (vector signed int, vector bool int); | |
8412 | int vec_any_gt (vector signed int, vector signed int); | |
8413 | int vec_any_gt (vector float, vector float); | |
8414 | ||
8415 | int vec_any_le (vector bool char, vector unsigned char); | |
8416 | int vec_any_le (vector unsigned char, vector bool char); | |
8417 | int vec_any_le (vector unsigned char, vector unsigned char); | |
8418 | int vec_any_le (vector bool char, vector signed char); | |
8419 | int vec_any_le (vector signed char, vector bool char); | |
8420 | int vec_any_le (vector signed char, vector signed char); | |
8421 | int vec_any_le (vector bool short, vector unsigned short); | |
8422 | int vec_any_le (vector unsigned short, vector bool short); | |
8423 | int vec_any_le (vector unsigned short, vector unsigned short); | |
8424 | int vec_any_le (vector bool short, vector signed short); | |
8425 | int vec_any_le (vector signed short, vector bool short); | |
8426 | int vec_any_le (vector signed short, vector signed short); | |
8427 | int vec_any_le (vector bool int, vector unsigned int); | |
8428 | int vec_any_le (vector unsigned int, vector bool int); | |
8429 | int vec_any_le (vector unsigned int, vector unsigned int); | |
8430 | int vec_any_le (vector bool int, vector signed int); | |
8431 | int vec_any_le (vector signed int, vector bool int); | |
8432 | int vec_any_le (vector signed int, vector signed int); | |
8433 | int vec_any_le (vector float, vector float); | |
8434 | ||
8435 | int vec_any_lt (vector bool char, vector unsigned char); | |
8436 | int vec_any_lt (vector unsigned char, vector bool char); | |
8437 | int vec_any_lt (vector unsigned char, vector unsigned char); | |
8438 | int vec_any_lt (vector bool char, vector signed char); | |
8439 | int vec_any_lt (vector signed char, vector bool char); | |
8440 | int vec_any_lt (vector signed char, vector signed char); | |
8441 | int vec_any_lt (vector bool short, vector unsigned short); | |
8442 | int vec_any_lt (vector unsigned short, vector bool short); | |
8443 | int vec_any_lt (vector unsigned short, vector unsigned short); | |
8444 | int vec_any_lt (vector bool short, vector signed short); | |
8445 | int vec_any_lt (vector signed short, vector bool short); | |
8446 | int vec_any_lt (vector signed short, vector signed short); | |
8447 | int vec_any_lt (vector bool int, vector unsigned int); | |
8448 | int vec_any_lt (vector unsigned int, vector bool int); | |
8449 | int vec_any_lt (vector unsigned int, vector unsigned int); | |
8450 | int vec_any_lt (vector bool int, vector signed int); | |
8451 | int vec_any_lt (vector signed int, vector bool int); | |
8452 | int vec_any_lt (vector signed int, vector signed int); | |
8453 | int vec_any_lt (vector float, vector float); | |
8454 | ||
8455 | int vec_any_nan (vector float); | |
8456 | ||
8457 | int vec_any_ne (vector signed char, vector bool char); | |
8458 | int vec_any_ne (vector signed char, vector signed char); | |
8459 | int vec_any_ne (vector unsigned char, vector bool char); | |
8460 | int vec_any_ne (vector unsigned char, vector unsigned char); | |
8461 | int vec_any_ne (vector bool char, vector bool char); | |
8462 | int vec_any_ne (vector bool char, vector unsigned char); | |
8463 | int vec_any_ne (vector bool char, vector signed char); | |
8464 | int vec_any_ne (vector signed short, vector bool short); | |
8465 | int vec_any_ne (vector signed short, vector signed short); | |
8466 | int vec_any_ne (vector unsigned short, vector bool short); | |
8467 | int vec_any_ne (vector unsigned short, vector unsigned short); | |
8468 | int vec_any_ne (vector bool short, vector bool short); | |
8469 | int vec_any_ne (vector bool short, vector unsigned short); | |
8470 | int vec_any_ne (vector bool short, vector signed short); | |
8471 | int vec_any_ne (vector pixel, vector pixel); | |
8472 | int vec_any_ne (vector signed int, vector bool int); | |
8473 | int vec_any_ne (vector signed int, vector signed int); | |
8474 | int vec_any_ne (vector unsigned int, vector bool int); | |
8475 | int vec_any_ne (vector unsigned int, vector unsigned int); | |
8476 | int vec_any_ne (vector bool int, vector bool int); | |
8477 | int vec_any_ne (vector bool int, vector unsigned int); | |
8478 | int vec_any_ne (vector bool int, vector signed int); | |
8479 | int vec_any_ne (vector float, vector float); | |
8480 | ||
8481 | int vec_any_nge (vector float, vector float); | |
8482 | ||
8483 | int vec_any_ngt (vector float, vector float); | |
8484 | ||
8485 | int vec_any_nle (vector float, vector float); | |
8486 | ||
8487 | int vec_any_nlt (vector float, vector float); | |
8488 | ||
8489 | int vec_any_numeric (vector float); | |
8490 | ||
8491 | int vec_any_out (vector float, vector float); | |
333c8841 AH |
8492 | @end smallexample |
8493 | ||
c5145ceb JM |
8494 | @node SPARC VIS Built-in Functions |
8495 | @subsection SPARC VIS Built-in Functions | |
8496 | ||
8497 | GCC supports SIMD operations on the SPARC using both the generic vector | |
2fd13506 | 8498 | extensions (@pxref{Vector Extensions}) as well as built-in functions for |
c5145ceb JM |
8499 | the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis} |
8500 | switch, the VIS extension is exposed as the following built-in functions: | |
8501 | ||
8502 | @smallexample | |
8503 | typedef int v2si __attribute__ ((vector_size (8))); | |
8504 | typedef short v4hi __attribute__ ((vector_size (8))); | |
8505 | typedef short v2hi __attribute__ ((vector_size (4))); | |
8506 | typedef char v8qi __attribute__ ((vector_size (8))); | |
8507 | typedef char v4qi __attribute__ ((vector_size (4))); | |
8508 | ||
8509 | void * __builtin_vis_alignaddr (void *, long); | |
8510 | int64_t __builtin_vis_faligndatadi (int64_t, int64_t); | |
8511 | v2si __builtin_vis_faligndatav2si (v2si, v2si); | |
8512 | v4hi __builtin_vis_faligndatav4hi (v4si, v4si); | |
8513 | v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi); | |
8514 | ||
8515 | v4hi __builtin_vis_fexpand (v4qi); | |
8516 | ||
8517 | v4hi __builtin_vis_fmul8x16 (v4qi, v4hi); | |
8518 | v4hi __builtin_vis_fmul8x16au (v4qi, v4hi); | |
8519 | v4hi __builtin_vis_fmul8x16al (v4qi, v4hi); | |
8520 | v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi); | |
8521 | v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi); | |
8522 | v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi); | |
8523 | v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi); | |
8524 | ||
8525 | v4qi __builtin_vis_fpack16 (v4hi); | |
8526 | v8qi __builtin_vis_fpack32 (v2si, v2si); | |
8527 | v2hi __builtin_vis_fpackfix (v2si); | |
8528 | v8qi __builtin_vis_fpmerge (v4qi, v4qi); | |
8529 | ||
8530 | int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t); | |
8531 | @end smallexample | |
8532 | ||
a2bec818 DJ |
8533 | @node Target Format Checks |
8534 | @section Format Checks Specific to Particular Target Machines | |
8535 | ||
8536 | For some target machines, GCC supports additional options to the | |
8537 | format attribute | |
8538 | (@pxref{Function Attributes,,Declaring Attributes of Functions}). | |
8539 | ||
8540 | @menu | |
8541 | * Solaris Format Checks:: | |
8542 | @end menu | |
8543 | ||
8544 | @node Solaris Format Checks | |
8545 | @subsection Solaris Format Checks | |
8546 | ||
8547 | Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format | |
8548 | check. @code{cmn_err} accepts a subset of the standard @code{printf} | |
8549 | conversions, and the two-argument @code{%b} conversion for displaying | |
8550 | bit-fields. See the Solaris man page for @code{cmn_err} for more information. | |
8551 | ||
0168a849 SS |
8552 | @node Pragmas |
8553 | @section Pragmas Accepted by GCC | |
8554 | @cindex pragmas | |
8555 | @cindex #pragma | |
8556 | ||
8557 | GCC supports several types of pragmas, primarily in order to compile | |
8558 | code originally written for other compilers. Note that in general | |
8559 | we do not recommend the use of pragmas; @xref{Function Attributes}, | |
8560 | for further explanation. | |
8561 | ||
8562 | @menu | |
8563 | * ARM Pragmas:: | |
a5c76ee6 | 8564 | * RS/6000 and PowerPC Pragmas:: |
0168a849 | 8565 | * Darwin Pragmas:: |
07a43492 | 8566 | * Solaris Pragmas:: |
84b8b0e0 | 8567 | * Symbol-Renaming Pragmas:: |
467cecf3 | 8568 | * Structure-Packing Pragmas:: |
0168a849 SS |
8569 | @end menu |
8570 | ||
8571 | @node ARM Pragmas | |
8572 | @subsection ARM Pragmas | |
8573 | ||
8574 | The ARM target defines pragmas for controlling the default addition of | |
8575 | @code{long_call} and @code{short_call} attributes to functions. | |
8576 | @xref{Function Attributes}, for information about the effects of these | |
8577 | attributes. | |
8578 | ||
8579 | @table @code | |
8580 | @item long_calls | |
8581 | @cindex pragma, long_calls | |
8582 | Set all subsequent functions to have the @code{long_call} attribute. | |
8583 | ||
8584 | @item no_long_calls | |
8585 | @cindex pragma, no_long_calls | |
8586 | Set all subsequent functions to have the @code{short_call} attribute. | |
8587 | ||
8588 | @item long_calls_off | |
8589 | @cindex pragma, long_calls_off | |
8590 | Do not affect the @code{long_call} or @code{short_call} attributes of | |
8591 | subsequent functions. | |
8592 | @end table | |
8593 | ||
a5c76ee6 ZW |
8594 | @node RS/6000 and PowerPC Pragmas |
8595 | @subsection RS/6000 and PowerPC Pragmas | |
8596 | ||
8597 | The RS/6000 and PowerPC targets define one pragma for controlling | |
8598 | whether or not the @code{longcall} attribute is added to function | |
8599 | declarations by default. This pragma overrides the @option{-mlongcall} | |
95b1627e | 8600 | option, but not the @code{longcall} and @code{shortcall} attributes. |
a5c76ee6 ZW |
8601 | @xref{RS/6000 and PowerPC Options}, for more information about when long |
8602 | calls are and are not necessary. | |
8603 | ||
8604 | @table @code | |
8605 | @item longcall (1) | |
8606 | @cindex pragma, longcall | |
8607 | Apply the @code{longcall} attribute to all subsequent function | |
8608 | declarations. | |
8609 | ||
8610 | @item longcall (0) | |
8611 | Do not apply the @code{longcall} attribute to subsequent function | |
8612 | declarations. | |
8613 | @end table | |
8614 | ||
0168a849 SS |
8615 | @c Describe c4x pragmas here. |
8616 | @c Describe h8300 pragmas here. | |
0168a849 SS |
8617 | @c Describe sh pragmas here. |
8618 | @c Describe v850 pragmas here. | |
8619 | ||
8620 | @node Darwin Pragmas | |
8621 | @subsection Darwin Pragmas | |
8622 | ||
8623 | The following pragmas are available for all architectures running the | |
8624 | Darwin operating system. These are useful for compatibility with other | |
85ebf0c6 | 8625 | Mac OS compilers. |
0168a849 SS |
8626 | |
8627 | @table @code | |
8628 | @item mark @var{tokens}@dots{} | |
8629 | @cindex pragma, mark | |
8630 | This pragma is accepted, but has no effect. | |
8631 | ||
8632 | @item options align=@var{alignment} | |
8633 | @cindex pragma, options align | |
8634 | This pragma sets the alignment of fields in structures. The values of | |
8635 | @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or | |
8636 | @code{power}, to emulate PowerPC alignment. Uses of this pragma nest | |
8637 | properly; to restore the previous setting, use @code{reset} for the | |
8638 | @var{alignment}. | |
8639 | ||
8640 | @item segment @var{tokens}@dots{} | |
8641 | @cindex pragma, segment | |
8642 | This pragma is accepted, but has no effect. | |
8643 | ||
8644 | @item unused (@var{var} [, @var{var}]@dots{}) | |
8645 | @cindex pragma, unused | |
8646 | This pragma declares variables to be possibly unused. GCC will not | |
8647 | produce warnings for the listed variables. The effect is similar to | |
8648 | that of the @code{unused} attribute, except that this pragma may appear | |
8649 | anywhere within the variables' scopes. | |
8650 | @end table | |
8651 | ||
07a43492 DJ |
8652 | @node Solaris Pragmas |
8653 | @subsection Solaris Pragmas | |
8654 | ||
8655 | The Solaris target supports @code{#pragma redefine_extname} | |
8656 | (@pxref{Symbol-Renaming Pragmas}). It also supports additional | |
8657 | @code{#pragma} directives for compatibility with the system compiler. | |
8658 | ||
8659 | @table @code | |
8660 | @item align @var{alignment} (@var{variable} [, @var{variable}]...) | |
8661 | @cindex pragma, align | |
8662 | ||
8663 | Increase the minimum alignment of each @var{variable} to @var{alignment}. | |
8664 | This is the same as GCC's @code{aligned} attribute @pxref{Variable | |
b5b3e36a DJ |
8665 | Attributes}). Macro expansion occurs on the arguments to this pragma |
8666 | when compiling C and Objective-C. It does not currently occur when | |
8667 | compiling C++, but this is a bug which may be fixed in a future | |
8668 | release. | |
07a43492 DJ |
8669 | |
8670 | @item fini (@var{function} [, @var{function}]...) | |
8671 | @cindex pragma, fini | |
8672 | ||
8673 | This pragma causes each listed @var{function} to be called after | |
8674 | main, or during shared module unloading, by adding a call to the | |
8675 | @code{.fini} section. | |
8676 | ||
8677 | @item init (@var{function} [, @var{function}]...) | |
8678 | @cindex pragma, init | |
8679 | ||
8680 | This pragma causes each listed @var{function} to be called during | |
8681 | initialization (before @code{main}) or during shared module loading, by | |
8682 | adding a call to the @code{.init} section. | |
8683 | ||
8684 | @end table | |
8685 | ||
84b8b0e0 ZW |
8686 | @node Symbol-Renaming Pragmas |
8687 | @subsection Symbol-Renaming Pragmas | |
41c64394 | 8688 | |
84b8b0e0 ZW |
8689 | For compatibility with the Solaris and Tru64 UNIX system headers, GCC |
8690 | supports two @code{#pragma} directives which change the name used in | |
8691 | assembly for a given declaration. These pragmas are only available on | |
8692 | platforms whose system headers need them. To get this effect on all | |
8693 | platforms supported by GCC, use the asm labels extension (@pxref{Asm | |
8694 | Labels}). | |
41c64394 RH |
8695 | |
8696 | @table @code | |
8697 | @item redefine_extname @var{oldname} @var{newname} | |
8698 | @cindex pragma, redefine_extname | |
8699 | ||
84b8b0e0 ZW |
8700 | This pragma gives the C function @var{oldname} the assembly symbol |
8701 | @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME} | |
8702 | will be defined if this pragma is available (currently only on | |
8703 | Solaris). | |
41c64394 | 8704 | |
41c64394 RH |
8705 | @item extern_prefix @var{string} |
8706 | @cindex pragma, extern_prefix | |
8707 | ||
84b8b0e0 ZW |
8708 | This pragma causes all subsequent external function and variable |
8709 | declarations to have @var{string} prepended to their assembly symbols. | |
8710 | This effect may be terminated with another @code{extern_prefix} pragma | |
8711 | whose argument is an empty string. The preprocessor macro | |
8712 | @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is | |
8a36672b | 8713 | available (currently only on Tru64 UNIX)@. |
41c64394 RH |
8714 | @end table |
8715 | ||
84b8b0e0 ZW |
8716 | These pragmas and the asm labels extension interact in a complicated |
8717 | manner. Here are some corner cases you may want to be aware of. | |
8718 | ||
8719 | @enumerate | |
8720 | @item Both pragmas silently apply only to declarations with external | |
8721 | linkage. Asm labels do not have this restriction. | |
8722 | ||
8723 | @item In C++, both pragmas silently apply only to declarations with | |
8724 | ``C'' linkage. Again, asm labels do not have this restriction. | |
8725 | ||
8726 | @item If any of the three ways of changing the assembly name of a | |
8727 | declaration is applied to a declaration whose assembly name has | |
8728 | already been determined (either by a previous use of one of these | |
8729 | features, or because the compiler needed the assembly name in order to | |
8730 | generate code), and the new name is different, a warning issues and | |
8731 | the name does not change. | |
8732 | ||
8733 | @item The @var{oldname} used by @code{#pragma redefine_extname} is | |
8734 | always the C-language name. | |
8735 | ||
8736 | @item If @code{#pragma extern_prefix} is in effect, and a declaration | |
8737 | occurs with an asm label attached, the prefix is silently ignored for | |
8738 | that declaration. | |
8739 | ||
8740 | @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname} | |
8741 | apply to the same declaration, whichever triggered first wins, and a | |
8742 | warning issues if they contradict each other. (We would like to have | |
8743 | @code{#pragma redefine_extname} always win, for consistency with asm | |
8744 | labels, but if @code{#pragma extern_prefix} triggers first we have no | |
8745 | way of knowing that that happened.) | |
8746 | @end enumerate | |
8747 | ||
467cecf3 JB |
8748 | @node Structure-Packing Pragmas |
8749 | @subsection Structure-Packing Pragmas | |
8750 | ||
8751 | For compatibility with Win32, GCC supports as set of @code{#pragma} | |
8752 | directives which change the maximum alignment of members of structures, | |
8753 | unions, and classes subsequently defined. The @var{n} value below always | |
8754 | is required to be a small power of two and specifies the new alignment | |
8755 | in bytes. | |
8756 | ||
8757 | @enumerate | |
8758 | @item @code{#pragma pack(@var{n})} simply sets the new alignment. | |
8759 | @item @code{#pragma pack()} sets the alignment to the one that was in | |
8760 | effect when compilation started (see also command line option | |
8761 | @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}). | |
8762 | @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment | |
8763 | setting on an internal stack and then optionally sets the new alignment. | |
8764 | @item @code{#pragma pack(pop)} restores the alignment setting to the one | |
8765 | saved at the top of the internal stack (and removes that stack entry). | |
8766 | Note that @code{#pragma pack([@var{n}])} does not influence this internal | |
8767 | stack; thus it is possible to have @code{#pragma pack(push)} followed by | |
8768 | multiple @code{#pragma pack(@var{n})} instances and finalized by a single | |
8769 | @code{#pragma pack(pop)}. | |
8770 | @end enumerate | |
8771 | ||
3e96a2fd | 8772 | @node Unnamed Fields |
2fbebc71 | 8773 | @section Unnamed struct/union fields within structs/unions |
3e96a2fd DD |
8774 | @cindex struct |
8775 | @cindex union | |
8776 | ||
8777 | For compatibility with other compilers, GCC allows you to define | |
8778 | a structure or union that contains, as fields, structures and unions | |
8779 | without names. For example: | |
8780 | ||
3ab51846 | 8781 | @smallexample |
3e96a2fd DD |
8782 | struct @{ |
8783 | int a; | |
8784 | union @{ | |
8785 | int b; | |
8786 | float c; | |
8787 | @}; | |
8788 | int d; | |
8789 | @} foo; | |
3ab51846 | 8790 | @end smallexample |
3e96a2fd DD |
8791 | |
8792 | In this example, the user would be able to access members of the unnamed | |
8793 | union with code like @samp{foo.b}. Note that only unnamed structs and | |
8794 | unions are allowed, you may not have, for example, an unnamed | |
8795 | @code{int}. | |
8796 | ||
8797 | You must never create such structures that cause ambiguous field definitions. | |
8798 | For example, this structure: | |
8799 | ||
3ab51846 | 8800 | @smallexample |
3e96a2fd DD |
8801 | struct @{ |
8802 | int a; | |
8803 | struct @{ | |
8804 | int a; | |
8805 | @}; | |
8806 | @} foo; | |
3ab51846 | 8807 | @end smallexample |
3e96a2fd DD |
8808 | |
8809 | It is ambiguous which @code{a} is being referred to with @samp{foo.a}. | |
8810 | Such constructs are not supported and must be avoided. In the future, | |
8811 | such constructs may be detected and treated as compilation errors. | |
8812 | ||
2fbebc71 JM |
8813 | @opindex fms-extensions |
8814 | Unless @option{-fms-extensions} is used, the unnamed field must be a | |
8815 | structure or union definition without a tag (for example, @samp{struct | |
8816 | @{ int a; @};}). If @option{-fms-extensions} is used, the field may | |
8817 | also be a definition with a tag such as @samp{struct foo @{ int a; | |
8818 | @};}, a reference to a previously defined structure or union such as | |
8819 | @samp{struct foo;}, or a reference to a @code{typedef} name for a | |
8820 | previously defined structure or union type. | |
8821 | ||
3d78f2e9 RH |
8822 | @node Thread-Local |
8823 | @section Thread-Local Storage | |
8824 | @cindex Thread-Local Storage | |
9217ef40 | 8825 | @cindex @acronym{TLS} |
3d78f2e9 RH |
8826 | @cindex __thread |
8827 | ||
9217ef40 RH |
8828 | Thread-local storage (@acronym{TLS}) is a mechanism by which variables |
8829 | are allocated such that there is one instance of the variable per extant | |
3d78f2e9 RH |
8830 | thread. The run-time model GCC uses to implement this originates |
8831 | in the IA-64 processor-specific ABI, but has since been migrated | |
8832 | to other processors as well. It requires significant support from | |
8833 | the linker (@command{ld}), dynamic linker (@command{ld.so}), and | |
8834 | system libraries (@file{libc.so} and @file{libpthread.so}), so it | |
9217ef40 | 8835 | is not available everywhere. |
3d78f2e9 RH |
8836 | |
8837 | At the user level, the extension is visible with a new storage | |
8838 | class keyword: @code{__thread}. For example: | |
8839 | ||
3ab51846 | 8840 | @smallexample |
3d78f2e9 RH |
8841 | __thread int i; |
8842 | extern __thread struct state s; | |
8843 | static __thread char *p; | |
3ab51846 | 8844 | @end smallexample |
3d78f2e9 RH |
8845 | |
8846 | The @code{__thread} specifier may be used alone, with the @code{extern} | |
8847 | or @code{static} specifiers, but with no other storage class specifier. | |
8848 | When used with @code{extern} or @code{static}, @code{__thread} must appear | |
8849 | immediately after the other storage class specifier. | |
8850 | ||
8851 | The @code{__thread} specifier may be applied to any global, file-scoped | |
244c2241 RH |
8852 | static, function-scoped static, or static data member of a class. It may |
8853 | not be applied to block-scoped automatic or non-static data member. | |
3d78f2e9 RH |
8854 | |
8855 | When the address-of operator is applied to a thread-local variable, it is | |
8856 | evaluated at run-time and returns the address of the current thread's | |
8857 | instance of that variable. An address so obtained may be used by any | |
8858 | thread. When a thread terminates, any pointers to thread-local variables | |
8859 | in that thread become invalid. | |
8860 | ||
8861 | No static initialization may refer to the address of a thread-local variable. | |
8862 | ||
244c2241 RH |
8863 | In C++, if an initializer is present for a thread-local variable, it must |
8864 | be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++ | |
8865 | standard. | |
3d78f2e9 RH |
8866 | |
8867 | See @uref{http://people.redhat.com/drepper/tls.pdf, | |
8868 | ELF Handling For Thread-Local Storage} for a detailed explanation of | |
8869 | the four thread-local storage addressing models, and how the run-time | |
8870 | is expected to function. | |
8871 | ||
9217ef40 RH |
8872 | @menu |
8873 | * C99 Thread-Local Edits:: | |
8874 | * C++98 Thread-Local Edits:: | |
8875 | @end menu | |
8876 | ||
8877 | @node C99 Thread-Local Edits | |
8878 | @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage | |
8879 | ||
8880 | The following are a set of changes to ISO/IEC 9899:1999 (aka C99) | |
8881 | that document the exact semantics of the language extension. | |
8882 | ||
8883 | @itemize @bullet | |
8884 | @item | |
8885 | @cite{5.1.2 Execution environments} | |
8886 | ||
8887 | Add new text after paragraph 1 | |
8888 | ||
8889 | @quotation | |
8890 | Within either execution environment, a @dfn{thread} is a flow of | |
8891 | control within a program. It is implementation defined whether | |
8892 | or not there may be more than one thread associated with a program. | |
8893 | It is implementation defined how threads beyond the first are | |
8894 | created, the name and type of the function called at thread | |
8895 | startup, and how threads may be terminated. However, objects | |
8896 | with thread storage duration shall be initialized before thread | |
8897 | startup. | |
8898 | @end quotation | |
8899 | ||
8900 | @item | |
8901 | @cite{6.2.4 Storage durations of objects} | |
8902 | ||
8903 | Add new text before paragraph 3 | |
8904 | ||
8905 | @quotation | |
8906 | An object whose identifier is declared with the storage-class | |
8907 | specifier @w{@code{__thread}} has @dfn{thread storage duration}. | |
8908 | Its lifetime is the entire execution of the thread, and its | |
8909 | stored value is initialized only once, prior to thread startup. | |
8910 | @end quotation | |
8911 | ||
8912 | @item | |
8913 | @cite{6.4.1 Keywords} | |
8914 | ||
8915 | Add @code{__thread}. | |
8916 | ||
8917 | @item | |
8918 | @cite{6.7.1 Storage-class specifiers} | |
8919 | ||
8920 | Add @code{__thread} to the list of storage class specifiers in | |
8921 | paragraph 1. | |
8922 | ||
8923 | Change paragraph 2 to | |
8924 | ||
8925 | @quotation | |
8926 | With the exception of @code{__thread}, at most one storage-class | |
8927 | specifier may be given [@dots{}]. The @code{__thread} specifier may | |
8928 | be used alone, or immediately following @code{extern} or | |
8929 | @code{static}. | |
8930 | @end quotation | |
8931 | ||
8932 | Add new text after paragraph 6 | |
8933 | ||
8934 | @quotation | |
8935 | The declaration of an identifier for a variable that has | |
8936 | block scope that specifies @code{__thread} shall also | |
8937 | specify either @code{extern} or @code{static}. | |
8938 | ||
8939 | The @code{__thread} specifier shall be used only with | |
8940 | variables. | |
8941 | @end quotation | |
8942 | @end itemize | |
8943 | ||
8944 | @node C++98 Thread-Local Edits | |
8945 | @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage | |
8946 | ||
8947 | The following are a set of changes to ISO/IEC 14882:1998 (aka C++98) | |
8948 | that document the exact semantics of the language extension. | |
8949 | ||
8950 | @itemize @bullet | |
8d23a2c8 | 8951 | @item |
9217ef40 RH |
8952 | @b{[intro.execution]} |
8953 | ||
8954 | New text after paragraph 4 | |
8955 | ||
8956 | @quotation | |
8957 | A @dfn{thread} is a flow of control within the abstract machine. | |
8958 | It is implementation defined whether or not there may be more than | |
8959 | one thread. | |
8960 | @end quotation | |
8961 | ||
8962 | New text after paragraph 7 | |
8963 | ||
8964 | @quotation | |
95b1627e | 8965 | It is unspecified whether additional action must be taken to |
9217ef40 RH |
8966 | ensure when and whether side effects are visible to other threads. |
8967 | @end quotation | |
8968 | ||
8969 | @item | |
8970 | @b{[lex.key]} | |
8971 | ||
8972 | Add @code{__thread}. | |
8973 | ||
8974 | @item | |
8975 | @b{[basic.start.main]} | |
8976 | ||
8977 | Add after paragraph 5 | |
8978 | ||
8979 | @quotation | |
8980 | The thread that begins execution at the @code{main} function is called | |
95b1627e | 8981 | the @dfn{main thread}. It is implementation defined how functions |
9217ef40 RH |
8982 | beginning threads other than the main thread are designated or typed. |
8983 | A function so designated, as well as the @code{main} function, is called | |
8984 | a @dfn{thread startup function}. It is implementation defined what | |
8985 | happens if a thread startup function returns. It is implementation | |
8986 | defined what happens to other threads when any thread calls @code{exit}. | |
8987 | @end quotation | |
8988 | ||
8989 | @item | |
8990 | @b{[basic.start.init]} | |
8991 | ||
8992 | Add after paragraph 4 | |
8993 | ||
8994 | @quotation | |
8995 | The storage for an object of thread storage duration shall be | |
c0478a66 | 8996 | statically initialized before the first statement of the thread startup |
9217ef40 RH |
8997 | function. An object of thread storage duration shall not require |
8998 | dynamic initialization. | |
8999 | @end quotation | |
9000 | ||
9001 | @item | |
9002 | @b{[basic.start.term]} | |
9003 | ||
9004 | Add after paragraph 3 | |
9005 | ||
9006 | @quotation | |
244c2241 RH |
9007 | The type of an object with thread storage duration shall not have a |
9008 | non-trivial destructor, nor shall it be an array type whose elements | |
9009 | (directly or indirectly) have non-trivial destructors. | |
9217ef40 RH |
9010 | @end quotation |
9011 | ||
9012 | @item | |
9013 | @b{[basic.stc]} | |
9014 | ||
9015 | Add ``thread storage duration'' to the list in paragraph 1. | |
9016 | ||
9017 | Change paragraph 2 | |
9018 | ||
9019 | @quotation | |
9020 | Thread, static, and automatic storage durations are associated with | |
9021 | objects introduced by declarations [@dots{}]. | |
9022 | @end quotation | |
9023 | ||
9024 | Add @code{__thread} to the list of specifiers in paragraph 3. | |
9025 | ||
9026 | @item | |
9027 | @b{[basic.stc.thread]} | |
9028 | ||
9029 | New section before @b{[basic.stc.static]} | |
9030 | ||
9031 | @quotation | |
63519d23 | 9032 | The keyword @code{__thread} applied to a non-local object gives the |
9217ef40 RH |
9033 | object thread storage duration. |
9034 | ||
9035 | A local variable or class data member declared both @code{static} | |
9036 | and @code{__thread} gives the variable or member thread storage | |
9037 | duration. | |
9038 | @end quotation | |
9039 | ||
9040 | @item | |
9041 | @b{[basic.stc.static]} | |
9042 | ||
9043 | Change paragraph 1 | |
9044 | ||
9045 | @quotation | |
9046 | All objects which have neither thread storage duration, dynamic | |
9047 | storage duration nor are local [@dots{}]. | |
9048 | @end quotation | |
9049 | ||
9050 | @item | |
9051 | @b{[dcl.stc]} | |
9052 | ||
9053 | Add @code{__thread} to the list in paragraph 1. | |
9054 | ||
9055 | Change paragraph 1 | |
9056 | ||
9057 | @quotation | |
9058 | With the exception of @code{__thread}, at most one | |
9059 | @var{storage-class-specifier} shall appear in a given | |
9060 | @var{decl-specifier-seq}. The @code{__thread} specifier may | |
9061 | be used alone, or immediately following the @code{extern} or | |
9062 | @code{static} specifiers. [@dots{}] | |
9063 | @end quotation | |
9064 | ||
9065 | Add after paragraph 5 | |
9066 | ||
9067 | @quotation | |
9068 | The @code{__thread} specifier can be applied only to the names of objects | |
9069 | and to anonymous unions. | |
9070 | @end quotation | |
9071 | ||
9072 | @item | |
9073 | @b{[class.mem]} | |
9074 | ||
9075 | Add after paragraph 6 | |
9076 | ||
9077 | @quotation | |
9078 | Non-@code{static} members shall not be @code{__thread}. | |
9079 | @end quotation | |
9080 | @end itemize | |
9081 | ||
c1f7febf RK |
9082 | @node C++ Extensions |
9083 | @chapter Extensions to the C++ Language | |
9084 | @cindex extensions, C++ language | |
9085 | @cindex C++ language extensions | |
9086 | ||
9087 | The GNU compiler provides these extensions to the C++ language (and you | |
9088 | can also use most of the C language extensions in your C++ programs). If you | |
9089 | want to write code that checks whether these features are available, you can | |
9090 | test for the GNU compiler the same way as for C programs: check for a | |
9091 | predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to | |
48795525 GP |
9092 | test specifically for GNU C++ (@pxref{Common Predefined Macros,, |
9093 | Predefined Macros,cpp,The GNU C Preprocessor}). | |
c1f7febf RK |
9094 | |
9095 | @menu | |
c1f7febf | 9096 | * Min and Max:: C++ Minimum and maximum operators. |
02cac427 | 9097 | * Volatiles:: What constitutes an access to a volatile object. |
49419c8f | 9098 | * Restricted Pointers:: C99 restricted pointers and references. |
7a81cf7f | 9099 | * Vague Linkage:: Where G++ puts inlines, vtables and such. |
c1f7febf | 9100 | * C++ Interface:: You can use a single C++ header file for both |
e6f3b89d | 9101 | declarations and definitions. |
c1f7febf | 9102 | * Template Instantiation:: Methods for ensuring that exactly one copy of |
e6f3b89d | 9103 | each needed template instantiation is emitted. |
0ded1f18 JM |
9104 | * Bound member functions:: You can extract a function pointer to the |
9105 | method denoted by a @samp{->*} or @samp{.*} expression. | |
e6f3b89d | 9106 | * C++ Attributes:: Variable, function, and type attributes for C++ only. |
86098eb8 | 9107 | * Strong Using:: Strong using-directives for namespace composition. |
1f730ff7 | 9108 | * Java Exceptions:: Tweaking exception handling to work with Java. |
90ea7324 | 9109 | * Deprecated Features:: Things will disappear from g++. |
e6f3b89d | 9110 | * Backwards Compatibility:: Compatibilities with earlier definitions of C++. |
c1f7febf RK |
9111 | @end menu |
9112 | ||
c1f7febf RK |
9113 | @node Min and Max |
9114 | @section Minimum and Maximum Operators in C++ | |
9115 | ||
9116 | It is very convenient to have operators which return the ``minimum'' or the | |
9117 | ``maximum'' of two arguments. In GNU C++ (but not in GNU C), | |
9118 | ||
9119 | @table @code | |
9120 | @item @var{a} <? @var{b} | |
9121 | @findex <? | |
9122 | @cindex minimum operator | |
9123 | is the @dfn{minimum}, returning the smaller of the numeric values | |
9124 | @var{a} and @var{b}; | |
9125 | ||
9126 | @item @var{a} >? @var{b} | |
9127 | @findex >? | |
9128 | @cindex maximum operator | |
9129 | is the @dfn{maximum}, returning the larger of the numeric values @var{a} | |
9130 | and @var{b}. | |
9131 | @end table | |
9132 | ||
9133 | These operations are not primitive in ordinary C++, since you can | |
9134 | use a macro to return the minimum of two things in C++, as in the | |
9135 | following example. | |
9136 | ||
3ab51846 | 9137 | @smallexample |
c1f7febf | 9138 | #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y)) |
3ab51846 | 9139 | @end smallexample |
c1f7febf RK |
9140 | |
9141 | @noindent | |
9142 | You might then use @w{@samp{int min = MIN (i, j);}} to set @var{min} to | |
9143 | the minimum value of variables @var{i} and @var{j}. | |
9144 | ||
9145 | However, side effects in @code{X} or @code{Y} may cause unintended | |
9146 | behavior. For example, @code{MIN (i++, j++)} will fail, incrementing | |
95f79357 ZW |
9147 | the smaller counter twice. The GNU C @code{typeof} extension allows you |
9148 | to write safe macros that avoid this kind of problem (@pxref{Typeof}). | |
9149 | However, writing @code{MIN} and @code{MAX} as macros also forces you to | |
9150 | use function-call notation for a fundamental arithmetic operation. | |
9151 | Using GNU C++ extensions, you can write @w{@samp{int min = i <? j;}} | |
9152 | instead. | |
c1f7febf RK |
9153 | |
9154 | Since @code{<?} and @code{>?} are built into the compiler, they properly | |
9155 | handle expressions with side-effects; @w{@samp{int min = i++ <? j++;}} | |
9156 | works correctly. | |
9157 | ||
02cac427 NS |
9158 | @node Volatiles |
9159 | @section When is a Volatile Object Accessed? | |
9160 | @cindex accessing volatiles | |
9161 | @cindex volatile read | |
9162 | @cindex volatile write | |
9163 | @cindex volatile access | |
9164 | ||
767094dd JM |
9165 | Both the C and C++ standard have the concept of volatile objects. These |
9166 | are normally accessed by pointers and used for accessing hardware. The | |
8117da65 | 9167 | standards encourage compilers to refrain from optimizations |
02cac427 | 9168 | concerning accesses to volatile objects that it might perform on |
767094dd JM |
9169 | non-volatile objects. The C standard leaves it implementation defined |
9170 | as to what constitutes a volatile access. The C++ standard omits to | |
02cac427 | 9171 | specify this, except to say that C++ should behave in a similar manner |
767094dd | 9172 | to C with respect to volatiles, where possible. The minimum either |
8117da65 | 9173 | standard specifies is that at a sequence point all previous accesses to |
02cac427 | 9174 | volatile objects have stabilized and no subsequent accesses have |
767094dd | 9175 | occurred. Thus an implementation is free to reorder and combine |
02cac427 | 9176 | volatile accesses which occur between sequence points, but cannot do so |
767094dd | 9177 | for accesses across a sequence point. The use of volatiles does not |
02cac427 NS |
9178 | allow you to violate the restriction on updating objects multiple times |
9179 | within a sequence point. | |
9180 | ||
9181 | In most expressions, it is intuitively obvious what is a read and what is | |
767094dd | 9182 | a write. For instance |
02cac427 | 9183 | |
3ab51846 | 9184 | @smallexample |
c771326b JM |
9185 | volatile int *dst = @var{somevalue}; |
9186 | volatile int *src = @var{someothervalue}; | |
02cac427 | 9187 | *dst = *src; |
3ab51846 | 9188 | @end smallexample |
02cac427 NS |
9189 | |
9190 | @noindent | |
9191 | will cause a read of the volatile object pointed to by @var{src} and stores the | |
767094dd | 9192 | value into the volatile object pointed to by @var{dst}. There is no |
02cac427 NS |
9193 | guarantee that these reads and writes are atomic, especially for objects |
9194 | larger than @code{int}. | |
9195 | ||
9196 | Less obvious expressions are where something which looks like an access | |
767094dd | 9197 | is used in a void context. An example would be, |
02cac427 | 9198 | |
3ab51846 | 9199 | @smallexample |
c771326b | 9200 | volatile int *src = @var{somevalue}; |
02cac427 | 9201 | *src; |
3ab51846 | 9202 | @end smallexample |
02cac427 NS |
9203 | |
9204 | With C, such expressions are rvalues, and as rvalues cause a read of | |
f0523f02 | 9205 | the object, GCC interprets this as a read of the volatile being pointed |
767094dd | 9206 | to. The C++ standard specifies that such expressions do not undergo |
02cac427 | 9207 | lvalue to rvalue conversion, and that the type of the dereferenced |
767094dd | 9208 | object may be incomplete. The C++ standard does not specify explicitly |
02cac427 | 9209 | that it is this lvalue to rvalue conversion which is responsible for |
767094dd JM |
9210 | causing an access. However, there is reason to believe that it is, |
9211 | because otherwise certain simple expressions become undefined. However, | |
f0523f02 | 9212 | because it would surprise most programmers, G++ treats dereferencing a |
02cac427 | 9213 | pointer to volatile object of complete type in a void context as a read |
767094dd | 9214 | of the object. When the object has incomplete type, G++ issues a |
02cac427 NS |
9215 | warning. |
9216 | ||
3ab51846 | 9217 | @smallexample |
02cac427 NS |
9218 | struct S; |
9219 | struct T @{int m;@}; | |
c771326b JM |
9220 | volatile S *ptr1 = @var{somevalue}; |
9221 | volatile T *ptr2 = @var{somevalue}; | |
02cac427 NS |
9222 | *ptr1; |
9223 | *ptr2; | |
3ab51846 | 9224 | @end smallexample |
02cac427 NS |
9225 | |
9226 | In this example, a warning is issued for @code{*ptr1}, and @code{*ptr2} | |
767094dd | 9227 | causes a read of the object pointed to. If you wish to force an error on |
02cac427 NS |
9228 | the first case, you must force a conversion to rvalue with, for instance |
9229 | a static cast, @code{static_cast<S>(*ptr1)}. | |
9230 | ||
f0523f02 | 9231 | When using a reference to volatile, G++ does not treat equivalent |
02cac427 | 9232 | expressions as accesses to volatiles, but instead issues a warning that |
767094dd | 9233 | no volatile is accessed. The rationale for this is that otherwise it |
02cac427 NS |
9234 | becomes difficult to determine where volatile access occur, and not |
9235 | possible to ignore the return value from functions returning volatile | |
767094dd | 9236 | references. Again, if you wish to force a read, cast the reference to |
02cac427 NS |
9237 | an rvalue. |
9238 | ||
535233a8 NS |
9239 | @node Restricted Pointers |
9240 | @section Restricting Pointer Aliasing | |
9241 | @cindex restricted pointers | |
9242 | @cindex restricted references | |
9243 | @cindex restricted this pointer | |
9244 | ||
2dd76960 | 9245 | As with the C front end, G++ understands the C99 feature of restricted pointers, |
535233a8 | 9246 | specified with the @code{__restrict__}, or @code{__restrict} type |
767094dd | 9247 | qualifier. Because you cannot compile C++ by specifying the @option{-std=c99} |
535233a8 NS |
9248 | language flag, @code{restrict} is not a keyword in C++. |
9249 | ||
9250 | In addition to allowing restricted pointers, you can specify restricted | |
9251 | references, which indicate that the reference is not aliased in the local | |
9252 | context. | |
9253 | ||
3ab51846 | 9254 | @smallexample |
535233a8 NS |
9255 | void fn (int *__restrict__ rptr, int &__restrict__ rref) |
9256 | @{ | |
0d893a63 | 9257 | /* @r{@dots{}} */ |
535233a8 | 9258 | @} |
3ab51846 | 9259 | @end smallexample |
535233a8 NS |
9260 | |
9261 | @noindent | |
9262 | In the body of @code{fn}, @var{rptr} points to an unaliased integer and | |
9263 | @var{rref} refers to a (different) unaliased integer. | |
9264 | ||
9265 | You may also specify whether a member function's @var{this} pointer is | |
9266 | unaliased by using @code{__restrict__} as a member function qualifier. | |
9267 | ||
3ab51846 | 9268 | @smallexample |
535233a8 NS |
9269 | void T::fn () __restrict__ |
9270 | @{ | |
0d893a63 | 9271 | /* @r{@dots{}} */ |
535233a8 | 9272 | @} |
3ab51846 | 9273 | @end smallexample |
535233a8 NS |
9274 | |
9275 | @noindent | |
9276 | Within the body of @code{T::fn}, @var{this} will have the effective | |
767094dd | 9277 | definition @code{T *__restrict__ const this}. Notice that the |
535233a8 NS |
9278 | interpretation of a @code{__restrict__} member function qualifier is |
9279 | different to that of @code{const} or @code{volatile} qualifier, in that it | |
767094dd | 9280 | is applied to the pointer rather than the object. This is consistent with |
535233a8 NS |
9281 | other compilers which implement restricted pointers. |
9282 | ||
9283 | As with all outermost parameter qualifiers, @code{__restrict__} is | |
767094dd | 9284 | ignored in function definition matching. This means you only need to |
535233a8 NS |
9285 | specify @code{__restrict__} in a function definition, rather than |
9286 | in a function prototype as well. | |
9287 | ||
7a81cf7f JM |
9288 | @node Vague Linkage |
9289 | @section Vague Linkage | |
9290 | @cindex vague linkage | |
9291 | ||
9292 | There are several constructs in C++ which require space in the object | |
9293 | file but are not clearly tied to a single translation unit. We say that | |
9294 | these constructs have ``vague linkage''. Typically such constructs are | |
9295 | emitted wherever they are needed, though sometimes we can be more | |
9296 | clever. | |
9297 | ||
9298 | @table @asis | |
9299 | @item Inline Functions | |
9300 | Inline functions are typically defined in a header file which can be | |
9301 | included in many different compilations. Hopefully they can usually be | |
9302 | inlined, but sometimes an out-of-line copy is necessary, if the address | |
9303 | of the function is taken or if inlining fails. In general, we emit an | |
9304 | out-of-line copy in all translation units where one is needed. As an | |
9305 | exception, we only emit inline virtual functions with the vtable, since | |
9306 | it will always require a copy. | |
9307 | ||
9308 | Local static variables and string constants used in an inline function | |
9309 | are also considered to have vague linkage, since they must be shared | |
9310 | between all inlined and out-of-line instances of the function. | |
9311 | ||
9312 | @item VTables | |
9313 | @cindex vtable | |
9314 | C++ virtual functions are implemented in most compilers using a lookup | |
9315 | table, known as a vtable. The vtable contains pointers to the virtual | |
9316 | functions provided by a class, and each object of the class contains a | |
9317 | pointer to its vtable (or vtables, in some multiple-inheritance | |
9318 | situations). If the class declares any non-inline, non-pure virtual | |
9319 | functions, the first one is chosen as the ``key method'' for the class, | |
9320 | and the vtable is only emitted in the translation unit where the key | |
9321 | method is defined. | |
9322 | ||
9323 | @emph{Note:} If the chosen key method is later defined as inline, the | |
9324 | vtable will still be emitted in every translation unit which defines it. | |
9325 | Make sure that any inline virtuals are declared inline in the class | |
9326 | body, even if they are not defined there. | |
9327 | ||
9328 | @item type_info objects | |
9329 | @cindex type_info | |
9330 | @cindex RTTI | |
9331 | C++ requires information about types to be written out in order to | |
9332 | implement @samp{dynamic_cast}, @samp{typeid} and exception handling. | |
9333 | For polymorphic classes (classes with virtual functions), the type_info | |
9334 | object is written out along with the vtable so that @samp{dynamic_cast} | |
9335 | can determine the dynamic type of a class object at runtime. For all | |
9336 | other types, we write out the type_info object when it is used: when | |
9337 | applying @samp{typeid} to an expression, throwing an object, or | |
9338 | referring to a type in a catch clause or exception specification. | |
9339 | ||
9340 | @item Template Instantiations | |
9341 | Most everything in this section also applies to template instantiations, | |
9342 | but there are other options as well. | |
9343 | @xref{Template Instantiation,,Where's the Template?}. | |
9344 | ||
9345 | @end table | |
9346 | ||
9347 | When used with GNU ld version 2.8 or later on an ELF system such as | |
95fef11f | 9348 | GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of |
7a81cf7f JM |
9349 | these constructs will be discarded at link time. This is known as |
9350 | COMDAT support. | |
9351 | ||
9352 | On targets that don't support COMDAT, but do support weak symbols, GCC | |
9353 | will use them. This way one copy will override all the others, but | |
9354 | the unused copies will still take up space in the executable. | |
9355 | ||
9356 | For targets which do not support either COMDAT or weak symbols, | |
9357 | most entities with vague linkage will be emitted as local symbols to | |
9358 | avoid duplicate definition errors from the linker. This will not happen | |
9359 | for local statics in inlines, however, as having multiple copies will | |
9360 | almost certainly break things. | |
9361 | ||
9362 | @xref{C++ Interface,,Declarations and Definitions in One Header}, for | |
9363 | another way to control placement of these constructs. | |
9364 | ||
c1f7febf | 9365 | @node C++ Interface |
fc72b380 | 9366 | @section #pragma interface and implementation |
c1f7febf RK |
9367 | |
9368 | @cindex interface and implementation headers, C++ | |
9369 | @cindex C++ interface and implementation headers | |
c1f7febf | 9370 | @cindex pragmas, interface and implementation |
c1f7febf | 9371 | |
fc72b380 JM |
9372 | @code{#pragma interface} and @code{#pragma implementation} provide the |
9373 | user with a way of explicitly directing the compiler to emit entities | |
9374 | with vague linkage (and debugging information) in a particular | |
9375 | translation unit. | |
c1f7febf | 9376 | |
fc72b380 JM |
9377 | @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in |
9378 | most cases, because of COMDAT support and the ``key method'' heuristic | |
9379 | mentioned in @ref{Vague Linkage}. Using them can actually cause your | |
27ef2cdd | 9380 | program to grow due to unnecessary out-of-line copies of inline |
fc72b380 JM |
9381 | functions. Currently (3.4) the only benefit of these |
9382 | @code{#pragma}s is reduced duplication of debugging information, and | |
9383 | that should be addressed soon on DWARF 2 targets with the use of | |
9384 | COMDAT groups. | |
c1f7febf RK |
9385 | |
9386 | @table @code | |
9387 | @item #pragma interface | |
9388 | @itemx #pragma interface "@var{subdir}/@var{objects}.h" | |
9389 | @kindex #pragma interface | |
9390 | Use this directive in @emph{header files} that define object classes, to save | |
9391 | space in most of the object files that use those classes. Normally, | |
9392 | local copies of certain information (backup copies of inline member | |
9393 | functions, debugging information, and the internal tables that implement | |
9394 | virtual functions) must be kept in each object file that includes class | |
9395 | definitions. You can use this pragma to avoid such duplication. When a | |
9396 | header file containing @samp{#pragma interface} is included in a | |
9397 | compilation, this auxiliary information will not be generated (unless | |
9398 | the main input source file itself uses @samp{#pragma implementation}). | |
9399 | Instead, the object files will contain references to be resolved at link | |
9400 | time. | |
9401 | ||
9402 | The second form of this directive is useful for the case where you have | |
9403 | multiple headers with the same name in different directories. If you | |
9404 | use this form, you must specify the same string to @samp{#pragma | |
9405 | implementation}. | |
9406 | ||
9407 | @item #pragma implementation | |
9408 | @itemx #pragma implementation "@var{objects}.h" | |
9409 | @kindex #pragma implementation | |
9410 | Use this pragma in a @emph{main input file}, when you want full output from | |
9411 | included header files to be generated (and made globally visible). The | |
9412 | included header file, in turn, should use @samp{#pragma interface}. | |
9413 | Backup copies of inline member functions, debugging information, and the | |
9414 | internal tables used to implement virtual functions are all generated in | |
9415 | implementation files. | |
9416 | ||
9417 | @cindex implied @code{#pragma implementation} | |
9418 | @cindex @code{#pragma implementation}, implied | |
9419 | @cindex naming convention, implementation headers | |
9420 | If you use @samp{#pragma implementation} with no argument, it applies to | |
9421 | an include file with the same basename@footnote{A file's @dfn{basename} | |
9422 | was the name stripped of all leading path information and of trailing | |
9423 | suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source | |
9424 | file. For example, in @file{allclass.cc}, giving just | |
9425 | @samp{#pragma implementation} | |
9426 | by itself is equivalent to @samp{#pragma implementation "allclass.h"}. | |
9427 | ||
9428 | In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as | |
9429 | an implementation file whenever you would include it from | |
9430 | @file{allclass.cc} even if you never specified @samp{#pragma | |
9431 | implementation}. This was deemed to be more trouble than it was worth, | |
9432 | however, and disabled. | |
9433 | ||
c1f7febf RK |
9434 | Use the string argument if you want a single implementation file to |
9435 | include code from multiple header files. (You must also use | |
9436 | @samp{#include} to include the header file; @samp{#pragma | |
9437 | implementation} only specifies how to use the file---it doesn't actually | |
9438 | include it.) | |
9439 | ||
9440 | There is no way to split up the contents of a single header file into | |
9441 | multiple implementation files. | |
9442 | @end table | |
9443 | ||
9444 | @cindex inlining and C++ pragmas | |
9445 | @cindex C++ pragmas, effect on inlining | |
9446 | @cindex pragmas in C++, effect on inlining | |
9447 | @samp{#pragma implementation} and @samp{#pragma interface} also have an | |
9448 | effect on function inlining. | |
9449 | ||
9450 | If you define a class in a header file marked with @samp{#pragma | |
fc72b380 JM |
9451 | interface}, the effect on an inline function defined in that class is |
9452 | similar to an explicit @code{extern} declaration---the compiler emits | |
9453 | no code at all to define an independent version of the function. Its | |
9454 | definition is used only for inlining with its callers. | |
c1f7febf | 9455 | |
84330467 | 9456 | @opindex fno-implement-inlines |
c1f7febf RK |
9457 | Conversely, when you include the same header file in a main source file |
9458 | that declares it as @samp{#pragma implementation}, the compiler emits | |
9459 | code for the function itself; this defines a version of the function | |
9460 | that can be found via pointers (or by callers compiled without | |
9461 | inlining). If all calls to the function can be inlined, you can avoid | |
84330467 | 9462 | emitting the function by compiling with @option{-fno-implement-inlines}. |
c1f7febf RK |
9463 | If any calls were not inlined, you will get linker errors. |
9464 | ||
9465 | @node Template Instantiation | |
9466 | @section Where's the Template? | |
c1f7febf RK |
9467 | @cindex template instantiation |
9468 | ||
9469 | C++ templates are the first language feature to require more | |
9470 | intelligence from the environment than one usually finds on a UNIX | |
9471 | system. Somehow the compiler and linker have to make sure that each | |
9472 | template instance occurs exactly once in the executable if it is needed, | |
9473 | and not at all otherwise. There are two basic approaches to this | |
962e6e00 | 9474 | problem, which are referred to as the Borland model and the Cfront model. |
c1f7febf RK |
9475 | |
9476 | @table @asis | |
9477 | @item Borland model | |
9478 | Borland C++ solved the template instantiation problem by adding the code | |
469b759e JM |
9479 | equivalent of common blocks to their linker; the compiler emits template |
9480 | instances in each translation unit that uses them, and the linker | |
9481 | collapses them together. The advantage of this model is that the linker | |
9482 | only has to consider the object files themselves; there is no external | |
9483 | complexity to worry about. This disadvantage is that compilation time | |
9484 | is increased because the template code is being compiled repeatedly. | |
9485 | Code written for this model tends to include definitions of all | |
9486 | templates in the header file, since they must be seen to be | |
9487 | instantiated. | |
c1f7febf RK |
9488 | |
9489 | @item Cfront model | |
9490 | The AT&T C++ translator, Cfront, solved the template instantiation | |
9491 | problem by creating the notion of a template repository, an | |
469b759e JM |
9492 | automatically maintained place where template instances are stored. A |
9493 | more modern version of the repository works as follows: As individual | |
9494 | object files are built, the compiler places any template definitions and | |
9495 | instantiations encountered in the repository. At link time, the link | |
9496 | wrapper adds in the objects in the repository and compiles any needed | |
9497 | instances that were not previously emitted. The advantages of this | |
9498 | model are more optimal compilation speed and the ability to use the | |
9499 | system linker; to implement the Borland model a compiler vendor also | |
c1f7febf | 9500 | needs to replace the linker. The disadvantages are vastly increased |
469b759e JM |
9501 | complexity, and thus potential for error; for some code this can be |
9502 | just as transparent, but in practice it can been very difficult to build | |
c1f7febf | 9503 | multiple programs in one directory and one program in multiple |
469b759e JM |
9504 | directories. Code written for this model tends to separate definitions |
9505 | of non-inline member templates into a separate file, which should be | |
9506 | compiled separately. | |
c1f7febf RK |
9507 | @end table |
9508 | ||
469b759e | 9509 | When used with GNU ld version 2.8 or later on an ELF system such as |
2dd76960 JM |
9510 | GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the |
9511 | Borland model. On other systems, G++ implements neither automatic | |
a4b3b54a | 9512 | model. |
469b759e | 9513 | |
2dd76960 | 9514 | A future version of G++ will support a hybrid model whereby the compiler |
469b759e JM |
9515 | will emit any instantiations for which the template definition is |
9516 | included in the compile, and store template definitions and | |
9517 | instantiation context information into the object file for the rest. | |
9518 | The link wrapper will extract that information as necessary and invoke | |
9519 | the compiler to produce the remaining instantiations. The linker will | |
9520 | then combine duplicate instantiations. | |
9521 | ||
9522 | In the mean time, you have the following options for dealing with | |
9523 | template instantiations: | |
c1f7febf RK |
9524 | |
9525 | @enumerate | |
d863830b | 9526 | @item |
84330467 JM |
9527 | @opindex frepo |
9528 | Compile your template-using code with @option{-frepo}. The compiler will | |
d863830b JL |
9529 | generate files with the extension @samp{.rpo} listing all of the |
9530 | template instantiations used in the corresponding object files which | |
9531 | could be instantiated there; the link wrapper, @samp{collect2}, will | |
9532 | then update the @samp{.rpo} files to tell the compiler where to place | |
9533 | those instantiations and rebuild any affected object files. The | |
9534 | link-time overhead is negligible after the first pass, as the compiler | |
9535 | will continue to place the instantiations in the same files. | |
9536 | ||
9537 | This is your best option for application code written for the Borland | |
9538 | model, as it will just work. Code written for the Cfront model will | |
9539 | need to be modified so that the template definitions are available at | |
9540 | one or more points of instantiation; usually this is as simple as adding | |
9541 | @code{#include <tmethods.cc>} to the end of each template header. | |
9542 | ||
9543 | For library code, if you want the library to provide all of the template | |
9544 | instantiations it needs, just try to link all of its object files | |
9545 | together; the link will fail, but cause the instantiations to be | |
9546 | generated as a side effect. Be warned, however, that this may cause | |
9547 | conflicts if multiple libraries try to provide the same instantiations. | |
9548 | For greater control, use explicit instantiation as described in the next | |
9549 | option. | |
9550 | ||
c1f7febf | 9551 | @item |
84330467 JM |
9552 | @opindex fno-implicit-templates |
9553 | Compile your code with @option{-fno-implicit-templates} to disable the | |
c1f7febf RK |
9554 | implicit generation of template instances, and explicitly instantiate |
9555 | all the ones you use. This approach requires more knowledge of exactly | |
9556 | which instances you need than do the others, but it's less | |
9557 | mysterious and allows greater control. You can scatter the explicit | |
9558 | instantiations throughout your program, perhaps putting them in the | |
9559 | translation units where the instances are used or the translation units | |
9560 | that define the templates themselves; you can put all of the explicit | |
9561 | instantiations you need into one big file; or you can create small files | |
9562 | like | |
9563 | ||
3ab51846 | 9564 | @smallexample |
c1f7febf RK |
9565 | #include "Foo.h" |
9566 | #include "Foo.cc" | |
9567 | ||
9568 | template class Foo<int>; | |
9569 | template ostream& operator << | |
9570 | (ostream&, const Foo<int>&); | |
3ab51846 | 9571 | @end smallexample |
c1f7febf RK |
9572 | |
9573 | for each of the instances you need, and create a template instantiation | |
9574 | library from those. | |
9575 | ||
9576 | If you are using Cfront-model code, you can probably get away with not | |
84330467 | 9577 | using @option{-fno-implicit-templates} when compiling files that don't |
c1f7febf RK |
9578 | @samp{#include} the member template definitions. |
9579 | ||
9580 | If you use one big file to do the instantiations, you may want to | |
84330467 | 9581 | compile it without @option{-fno-implicit-templates} so you get all of the |
c1f7febf RK |
9582 | instances required by your explicit instantiations (but not by any |
9583 | other files) without having to specify them as well. | |
9584 | ||
2dd76960 | 9585 | G++ has extended the template instantiation syntax given in the ISO |
6d9c4c83 | 9586 | standard to allow forward declaration of explicit instantiations |
4003d7f9 | 9587 | (with @code{extern}), instantiation of the compiler support data for a |
e979f9e8 | 9588 | template class (i.e.@: the vtable) without instantiating any of its |
4003d7f9 JM |
9589 | members (with @code{inline}), and instantiation of only the static data |
9590 | members of a template class, without the support data or member | |
9591 | functions (with (@code{static}): | |
c1f7febf | 9592 | |
3ab51846 | 9593 | @smallexample |
c1f7febf | 9594 | extern template int max (int, int); |
c1f7febf | 9595 | inline template class Foo<int>; |
4003d7f9 | 9596 | static template class Foo<int>; |
3ab51846 | 9597 | @end smallexample |
c1f7febf RK |
9598 | |
9599 | @item | |
2dd76960 | 9600 | Do nothing. Pretend G++ does implement automatic instantiation |
c1f7febf RK |
9601 | management. Code written for the Borland model will work fine, but |
9602 | each translation unit will contain instances of each of the templates it | |
9603 | uses. In a large program, this can lead to an unacceptable amount of code | |
9604 | duplication. | |
c1f7febf RK |
9605 | @end enumerate |
9606 | ||
0ded1f18 JM |
9607 | @node Bound member functions |
9608 | @section Extracting the function pointer from a bound pointer to member function | |
0ded1f18 JM |
9609 | @cindex pmf |
9610 | @cindex pointer to member function | |
9611 | @cindex bound pointer to member function | |
9612 | ||
9613 | In C++, pointer to member functions (PMFs) are implemented using a wide | |
9614 | pointer of sorts to handle all the possible call mechanisms; the PMF | |
9615 | needs to store information about how to adjust the @samp{this} pointer, | |
9616 | and if the function pointed to is virtual, where to find the vtable, and | |
9617 | where in the vtable to look for the member function. If you are using | |
9618 | PMFs in an inner loop, you should really reconsider that decision. If | |
9619 | that is not an option, you can extract the pointer to the function that | |
9620 | would be called for a given object/PMF pair and call it directly inside | |
9621 | the inner loop, to save a bit of time. | |
9622 | ||
9623 | Note that you will still be paying the penalty for the call through a | |
9624 | function pointer; on most modern architectures, such a call defeats the | |
161d7b59 | 9625 | branch prediction features of the CPU@. This is also true of normal |
0ded1f18 JM |
9626 | virtual function calls. |
9627 | ||
9628 | The syntax for this extension is | |
9629 | ||
3ab51846 | 9630 | @smallexample |
0ded1f18 JM |
9631 | extern A a; |
9632 | extern int (A::*fp)(); | |
9633 | typedef int (*fptr)(A *); | |
9634 | ||
9635 | fptr p = (fptr)(a.*fp); | |
3ab51846 | 9636 | @end smallexample |
0ded1f18 | 9637 | |
e979f9e8 | 9638 | For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}), |
767094dd | 9639 | no object is needed to obtain the address of the function. They can be |
0fb6bbf5 ML |
9640 | converted to function pointers directly: |
9641 | ||
3ab51846 | 9642 | @smallexample |
0fb6bbf5 | 9643 | fptr p1 = (fptr)(&A::foo); |
3ab51846 | 9644 | @end smallexample |
0fb6bbf5 | 9645 | |
84330467 JM |
9646 | @opindex Wno-pmf-conversions |
9647 | You must specify @option{-Wno-pmf-conversions} to use this extension. | |
0ded1f18 | 9648 | |
5c25e11d PE |
9649 | @node C++ Attributes |
9650 | @section C++-Specific Variable, Function, and Type Attributes | |
9651 | ||
9652 | Some attributes only make sense for C++ programs. | |
9653 | ||
9654 | @table @code | |
9655 | @item init_priority (@var{priority}) | |
9656 | @cindex init_priority attribute | |
9657 | ||
9658 | ||
9659 | In Standard C++, objects defined at namespace scope are guaranteed to be | |
9660 | initialized in an order in strict accordance with that of their definitions | |
9661 | @emph{in a given translation unit}. No guarantee is made for initializations | |
9662 | across translation units. However, GNU C++ allows users to control the | |
3844cd2e | 9663 | order of initialization of objects defined at namespace scope with the |
5c25e11d PE |
9664 | @code{init_priority} attribute by specifying a relative @var{priority}, |
9665 | a constant integral expression currently bounded between 101 and 65535 | |
9666 | inclusive. Lower numbers indicate a higher priority. | |
9667 | ||
9668 | In the following example, @code{A} would normally be created before | |
9669 | @code{B}, but the @code{init_priority} attribute has reversed that order: | |
9670 | ||
478c9e72 | 9671 | @smallexample |
5c25e11d PE |
9672 | Some_Class A __attribute__ ((init_priority (2000))); |
9673 | Some_Class B __attribute__ ((init_priority (543))); | |
478c9e72 | 9674 | @end smallexample |
5c25e11d PE |
9675 | |
9676 | @noindent | |
9677 | Note that the particular values of @var{priority} do not matter; only their | |
9678 | relative ordering. | |
9679 | ||
60c87482 BM |
9680 | @item java_interface |
9681 | @cindex java_interface attribute | |
9682 | ||
02f52e19 | 9683 | This type attribute informs C++ that the class is a Java interface. It may |
60c87482 | 9684 | only be applied to classes declared within an @code{extern "Java"} block. |
02f52e19 AJ |
9685 | Calls to methods declared in this interface will be dispatched using GCJ's |
9686 | interface table mechanism, instead of regular virtual table dispatch. | |
60c87482 | 9687 | |
5c25e11d PE |
9688 | @end table |
9689 | ||
86098eb8 JM |
9690 | See also @xref{Strong Using}. |
9691 | ||
9692 | @node Strong Using | |
9693 | @section Strong Using | |
9694 | ||
fea77ed9 MM |
9695 | @strong{Caution:} The semantics of this extension are not fully |
9696 | defined. Users should refrain from using this extension as its | |
9697 | semantics may change subtly over time. It is possible that this | |
9698 | extension wil be removed in future versions of G++. | |
9699 | ||
86098eb8 JM |
9700 | A using-directive with @code{__attribute ((strong))} is stronger |
9701 | than a normal using-directive in two ways: | |
9702 | ||
9703 | @itemize @bullet | |
9704 | @item | |
9705 | Templates from the used namespace can be specialized as though they were members of the using namespace. | |
9706 | ||
9707 | @item | |
9708 | The using namespace is considered an associated namespace of all | |
9709 | templates in the used namespace for purposes of argument-dependent | |
9710 | name lookup. | |
9711 | @end itemize | |
9712 | ||
9713 | This is useful for composing a namespace transparently from | |
9714 | implementation namespaces. For example: | |
9715 | ||
9716 | @smallexample | |
9717 | namespace std @{ | |
9718 | namespace debug @{ | |
9719 | template <class T> struct A @{ @}; | |
9720 | @} | |
9721 | using namespace debug __attribute ((__strong__)); | |
cd1a8088 | 9722 | template <> struct A<int> @{ @}; // @r{ok to specialize} |
86098eb8 JM |
9723 | |
9724 | template <class T> void f (A<T>); | |
9725 | @} | |
9726 | ||
9727 | int main() | |
9728 | @{ | |
cd1a8088 | 9729 | f (std::A<float>()); // @r{lookup finds} std::f |
86098eb8 JM |
9730 | f (std::A<int>()); |
9731 | @} | |
9732 | @end smallexample | |
9733 | ||
1f730ff7 ZW |
9734 | @node Java Exceptions |
9735 | @section Java Exceptions | |
9736 | ||
9737 | The Java language uses a slightly different exception handling model | |
9738 | from C++. Normally, GNU C++ will automatically detect when you are | |
9739 | writing C++ code that uses Java exceptions, and handle them | |
9740 | appropriately. However, if C++ code only needs to execute destructors | |
9741 | when Java exceptions are thrown through it, GCC will guess incorrectly. | |
9c34dbbf | 9742 | Sample problematic code is: |
1f730ff7 | 9743 | |
478c9e72 | 9744 | @smallexample |
1f730ff7 | 9745 | struct S @{ ~S(); @}; |
cd1a8088 | 9746 | extern void bar(); // @r{is written in Java, and may throw exceptions} |
1f730ff7 ZW |
9747 | void foo() |
9748 | @{ | |
9749 | S s; | |
9750 | bar(); | |
9751 | @} | |
478c9e72 | 9752 | @end smallexample |
1f730ff7 ZW |
9753 | |
9754 | @noindent | |
9755 | The usual effect of an incorrect guess is a link failure, complaining of | |
9756 | a missing routine called @samp{__gxx_personality_v0}. | |
9757 | ||
9758 | You can inform the compiler that Java exceptions are to be used in a | |
9759 | translation unit, irrespective of what it might think, by writing | |
9760 | @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This | |
9761 | @samp{#pragma} must appear before any functions that throw or catch | |
9762 | exceptions, or run destructors when exceptions are thrown through them. | |
9763 | ||
9764 | You cannot mix Java and C++ exceptions in the same translation unit. It | |
9765 | is believed to be safe to throw a C++ exception from one file through | |
9c34dbbf ZW |
9766 | another file compiled for the Java exception model, or vice versa, but |
9767 | there may be bugs in this area. | |
1f730ff7 | 9768 | |
e6f3b89d PE |
9769 | @node Deprecated Features |
9770 | @section Deprecated Features | |
9771 | ||
9772 | In the past, the GNU C++ compiler was extended to experiment with new | |
767094dd | 9773 | features, at a time when the C++ language was still evolving. Now that |
e6f3b89d | 9774 | the C++ standard is complete, some of those features are superseded by |
767094dd JM |
9775 | superior alternatives. Using the old features might cause a warning in |
9776 | some cases that the feature will be dropped in the future. In other | |
e6f3b89d PE |
9777 | cases, the feature might be gone already. |
9778 | ||
9779 | While the list below is not exhaustive, it documents some of the options | |
9780 | that are now deprecated: | |
9781 | ||
9782 | @table @code | |
9783 | @item -fexternal-templates | |
9784 | @itemx -falt-external-templates | |
2dd76960 | 9785 | These are two of the many ways for G++ to implement template |
767094dd | 9786 | instantiation. @xref{Template Instantiation}. The C++ standard clearly |
e6f3b89d | 9787 | defines how template definitions have to be organized across |
2dd76960 | 9788 | implementation units. G++ has an implicit instantiation mechanism that |
e6f3b89d PE |
9789 | should work just fine for standard-conforming code. |
9790 | ||
9791 | @item -fstrict-prototype | |
9792 | @itemx -fno-strict-prototype | |
9793 | Previously it was possible to use an empty prototype parameter list to | |
9794 | indicate an unspecified number of parameters (like C), rather than no | |
767094dd | 9795 | parameters, as C++ demands. This feature has been removed, except where |
e6f3b89d PE |
9796 | it is required for backwards compatibility @xref{Backwards Compatibility}. |
9797 | @end table | |
9798 | ||
ae209f28 NS |
9799 | G++ allows a virtual function returning @samp{void *} to be overridden |
9800 | by one returning a different pointer type. This extension to the | |
9801 | covariant return type rules is now deprecated and will be removed from a | |
9802 | future version. | |
9803 | ||
ad1a6d45 | 9804 | The named return value extension has been deprecated, and is now |
2dd76960 | 9805 | removed from G++. |
e6f3b89d | 9806 | |
82c18d5c | 9807 | The use of initializer lists with new expressions has been deprecated, |
2dd76960 | 9808 | and is now removed from G++. |
ad1a6d45 NS |
9809 | |
9810 | Floating and complex non-type template parameters have been deprecated, | |
2dd76960 | 9811 | and are now removed from G++. |
ad1a6d45 | 9812 | |
90ea7324 | 9813 | The implicit typename extension has been deprecated and is now |
2dd76960 | 9814 | removed from G++. |
90ea7324 NS |
9815 | |
9816 | The use of default arguments in function pointers, function typedefs and | |
9817 | and other places where they are not permitted by the standard is | |
2dd76960 | 9818 | deprecated and will be removed from a future version of G++. |
82c18d5c | 9819 | |
6871294a JW |
9820 | G++ allows floating-point literals to appear in integral constant expressions, |
9821 | e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} } | |
9822 | This extension is deprecated and will be removed from a future version. | |
9823 | ||
9824 | G++ allows static data members of const floating-point type to be declared | |
9825 | with an initializer in a class definition. The standard only allows | |
9826 | initializers for static members of const integral types and const | |
9827 | enumeration types so this extension has been deprecated and will be removed | |
9828 | from a future version. | |
9829 | ||
e6f3b89d PE |
9830 | @node Backwards Compatibility |
9831 | @section Backwards Compatibility | |
9832 | @cindex Backwards Compatibility | |
9833 | @cindex ARM [Annotated C++ Reference Manual] | |
9834 | ||
aee96fe9 | 9835 | Now that there is a definitive ISO standard C++, G++ has a specification |
767094dd | 9836 | to adhere to. The C++ language evolved over time, and features that |
e6f3b89d | 9837 | used to be acceptable in previous drafts of the standard, such as the ARM |
767094dd | 9838 | [Annotated C++ Reference Manual], are no longer accepted. In order to allow |
aee96fe9 | 9839 | compilation of C++ written to such drafts, G++ contains some backwards |
767094dd | 9840 | compatibilities. @emph{All such backwards compatibility features are |
aee96fe9 | 9841 | liable to disappear in future versions of G++.} They should be considered |
e6f3b89d PE |
9842 | deprecated @xref{Deprecated Features}. |
9843 | ||
9844 | @table @code | |
9845 | @item For scope | |
9846 | If a variable is declared at for scope, it used to remain in scope until | |
9847 | the end of the scope which contained the for statement (rather than just | |
aee96fe9 | 9848 | within the for scope). G++ retains this, but issues a warning, if such a |
e6f3b89d PE |
9849 | variable is accessed outside the for scope. |
9850 | ||
ad1a6d45 | 9851 | @item Implicit C language |
630d3d5a | 9852 | Old C system header files did not contain an @code{extern "C" @{@dots{}@}} |
767094dd JM |
9853 | scope to set the language. On such systems, all header files are |
9854 | implicitly scoped inside a C language scope. Also, an empty prototype | |
e6f3b89d PE |
9855 | @code{()} will be treated as an unspecified number of arguments, rather |
9856 | than no arguments, as C++ demands. | |
9857 | @end table |