]>
Commit | Line | Data |
---|---|---|
8d8da227 | 1 | @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1996, 1998, 1999, 2000, 2001, |
66647d44 JJ |
2 | @c 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009 |
3 | @c Free Software Foundation, Inc. | |
9a8ce21f | 4 | |
c1f7febf RK |
5 | @c This is part of the GCC manual. |
6 | @c For copying conditions, see the file gcc.texi. | |
7 | ||
8 | @node C Extensions | |
9 | @chapter Extensions to the C Language Family | |
10 | @cindex extensions, C language | |
11 | @cindex C language extensions | |
12 | ||
84330467 | 13 | @opindex pedantic |
161d7b59 | 14 | GNU C provides several language features not found in ISO standard C@. |
f0523f02 | 15 | (The @option{-pedantic} option directs GCC to print a warning message if |
c1f7febf RK |
16 | any of these features is used.) To test for the availability of these |
17 | features in conditional compilation, check for a predefined macro | |
161d7b59 | 18 | @code{__GNUC__}, which is always defined under GCC@. |
c1f7febf | 19 | |
161d7b59 | 20 | These extensions are available in C and Objective-C@. Most of them are |
c1f7febf RK |
21 | also available in C++. @xref{C++ Extensions,,Extensions to the |
22 | C++ Language}, for extensions that apply @emph{only} to C++. | |
23 | ||
4b404517 JM |
24 | Some features that are in ISO C99 but not C89 or C++ are also, as |
25 | extensions, accepted by GCC in C89 mode and in C++. | |
5490d604 | 26 | |
c1f7febf RK |
27 | @menu |
28 | * Statement Exprs:: Putting statements and declarations inside expressions. | |
14e33ee8 | 29 | * Local Labels:: Labels local to a block. |
c1f7febf RK |
30 | * Labels as Values:: Getting pointers to labels, and computed gotos. |
31 | * Nested Functions:: As in Algol and Pascal, lexical scoping of functions. | |
6ccde948 | 32 | * Constructing Calls:: Dispatching a call to another function. |
c1f7febf | 33 | * Typeof:: @code{typeof}: referring to the type of an expression. |
c1f7febf | 34 | * Conditionals:: Omitting the middle operand of a @samp{?:} expression. |
6ccde948 | 35 | * Long Long:: Double-word integers---@code{long long int}. |
c1f7febf | 36 | * Complex:: Data types for complex numbers. |
c77cd3d1 | 37 | * Floating Types:: Additional Floating Types. |
85a92f7e | 38 | * Decimal Float:: Decimal Floating Types. |
6f4d7222 | 39 | * Hex Floats:: Hexadecimal floating-point constants. |
0f996086 | 40 | * Fixed-Point:: Fixed-Point Types. |
c1f7febf RK |
41 | * Zero Length:: Zero-length arrays. |
42 | * Variable Length:: Arrays whose length is computed at run time. | |
ba05abd3 | 43 | * Empty Structures:: Structures with no members. |
6ccde948 | 44 | * Variadic Macros:: Macros with a variable number of arguments. |
ccd96f0a | 45 | * Escaped Newlines:: Slightly looser rules for escaped newlines. |
c1f7febf RK |
46 | * Subscripting:: Any array can be subscripted, even if not an lvalue. |
47 | * Pointer Arith:: Arithmetic on @code{void}-pointers and function pointers. | |
48 | * Initializers:: Non-constant initializers. | |
4b404517 | 49 | * Compound Literals:: Compound literals give structures, unions |
6ccde948 RW |
50 | or arrays as values. |
51 | * Designated Inits:: Labeling elements of initializers. | |
c1f7febf | 52 | * Cast to Union:: Casting to union type from any member of the union. |
6ccde948 RW |
53 | * Case Ranges:: `case 1 ... 9' and such. |
54 | * Mixed Declarations:: Mixing declarations and code. | |
c1f7febf | 55 | * Function Attributes:: Declaring that functions have no side effects, |
6ccde948 | 56 | or that they can never return. |
2c5e91d2 | 57 | * Attribute Syntax:: Formal syntax for attributes. |
c1f7febf RK |
58 | * Function Prototypes:: Prototype declarations and old-style definitions. |
59 | * C++ Comments:: C++ comments are recognized. | |
60 | * Dollar Signs:: Dollar sign is allowed in identifiers. | |
61 | * Character Escapes:: @samp{\e} stands for the character @key{ESC}. | |
6ccde948 RW |
62 | * Variable Attributes:: Specifying attributes of variables. |
63 | * Type Attributes:: Specifying attributes of types. | |
c1f7febf RK |
64 | * Alignment:: Inquiring about the alignment of a type or variable. |
65 | * Inline:: Defining inline functions (as fast as macros). | |
66 | * Extended Asm:: Assembler instructions with C expressions as operands. | |
6ccde948 | 67 | (With them you can define ``built-in'' functions.) |
c1f7febf RK |
68 | * Constraints:: Constraints for asm operands |
69 | * Asm Labels:: Specifying the assembler name to use for a C symbol. | |
70 | * Explicit Reg Vars:: Defining variables residing in specified registers. | |
71 | * Alternate Keywords:: @code{__const__}, @code{__asm__}, etc., for header files. | |
72 | * Incomplete Enums:: @code{enum foo;}, with details to follow. | |
6ccde948 RW |
73 | * Function Names:: Printable strings which are the name of the current |
74 | function. | |
c1f7febf | 75 | * Return Address:: Getting the return or frame address of a function. |
1255c85c | 76 | * Vector Extensions:: Using vector instructions through built-in functions. |
7a3ea201 | 77 | * Offsetof:: Special syntax for implementing @code{offsetof}. |
6ccde948 | 78 | * Atomic Builtins:: Built-in functions for atomic memory access. |
10a0d495 JJ |
79 | * Object Size Checking:: Built-in functions for limited buffer overflow |
80 | checking. | |
c5c76735 | 81 | * Other Builtins:: Other built-in functions. |
0975678f | 82 | * Target Builtins:: Built-in functions specific to particular targets. |
a2bec818 | 83 | * Target Format Checks:: Format checks specific to particular targets. |
0168a849 | 84 | * Pragmas:: Pragmas accepted by GCC. |
b11cc610 | 85 | * Unnamed Fields:: Unnamed struct/union fields within structs/unions. |
3d78f2e9 | 86 | * Thread-Local:: Per-thread variables. |
f7fd775f | 87 | * Binary constants:: Binary constants using the @samp{0b} prefix. |
c1f7febf | 88 | @end menu |
c1f7febf RK |
89 | |
90 | @node Statement Exprs | |
91 | @section Statements and Declarations in Expressions | |
92 | @cindex statements inside expressions | |
93 | @cindex declarations inside expressions | |
94 | @cindex expressions containing statements | |
95 | @cindex macros, statements in expressions | |
96 | ||
97 | @c the above section title wrapped and causes an underfull hbox.. i | |
98 | @c changed it from "within" to "in". --mew 4feb93 | |
c1f7febf | 99 | A compound statement enclosed in parentheses may appear as an expression |
161d7b59 | 100 | in GNU C@. This allows you to use loops, switches, and local variables |
c1f7febf RK |
101 | within an expression. |
102 | ||
103 | Recall that a compound statement is a sequence of statements surrounded | |
104 | by braces; in this construct, parentheses go around the braces. For | |
105 | example: | |
106 | ||
3ab51846 | 107 | @smallexample |
c1f7febf RK |
108 | (@{ int y = foo (); int z; |
109 | if (y > 0) z = y; | |
110 | else z = - y; | |
111 | z; @}) | |
3ab51846 | 112 | @end smallexample |
c1f7febf RK |
113 | |
114 | @noindent | |
115 | is a valid (though slightly more complex than necessary) expression | |
116 | for the absolute value of @code{foo ()}. | |
117 | ||
118 | The last thing in the compound statement should be an expression | |
119 | followed by a semicolon; the value of this subexpression serves as the | |
120 | value of the entire construct. (If you use some other kind of statement | |
121 | last within the braces, the construct has type @code{void}, and thus | |
122 | effectively no value.) | |
123 | ||
124 | This feature is especially useful in making macro definitions ``safe'' (so | |
125 | that they evaluate each operand exactly once). For example, the | |
126 | ``maximum'' function is commonly defined as a macro in standard C as | |
127 | follows: | |
128 | ||
3ab51846 | 129 | @smallexample |
c1f7febf | 130 | #define max(a,b) ((a) > (b) ? (a) : (b)) |
3ab51846 | 131 | @end smallexample |
c1f7febf RK |
132 | |
133 | @noindent | |
134 | @cindex side effects, macro argument | |
135 | But this definition computes either @var{a} or @var{b} twice, with bad | |
136 | results if the operand has side effects. In GNU C, if you know the | |
962e6e00 | 137 | type of the operands (here taken as @code{int}), you can define |
c1f7febf RK |
138 | the macro safely as follows: |
139 | ||
3ab51846 | 140 | @smallexample |
c1f7febf RK |
141 | #define maxint(a,b) \ |
142 | (@{int _a = (a), _b = (b); _a > _b ? _a : _b; @}) | |
3ab51846 | 143 | @end smallexample |
c1f7febf RK |
144 | |
145 | Embedded statements are not allowed in constant expressions, such as | |
c771326b | 146 | the value of an enumeration constant, the width of a bit-field, or |
c1f7febf RK |
147 | the initial value of a static variable. |
148 | ||
149 | If you don't know the type of the operand, you can still do this, but you | |
95f79357 | 150 | must use @code{typeof} (@pxref{Typeof}). |
c1f7febf | 151 | |
a5bcc582 NS |
152 | In G++, the result value of a statement expression undergoes array and |
153 | function pointer decay, and is returned by value to the enclosing | |
8a36672b | 154 | expression. For instance, if @code{A} is a class, then |
b98e139b | 155 | |
a5bcc582 NS |
156 | @smallexample |
157 | A a; | |
b98e139b | 158 | |
a5bcc582 NS |
159 | (@{a;@}).Foo () |
160 | @end smallexample | |
b98e139b MM |
161 | |
162 | @noindent | |
a5bcc582 NS |
163 | will construct a temporary @code{A} object to hold the result of the |
164 | statement expression, and that will be used to invoke @code{Foo}. | |
165 | Therefore the @code{this} pointer observed by @code{Foo} will not be the | |
166 | address of @code{a}. | |
167 | ||
168 | Any temporaries created within a statement within a statement expression | |
169 | will be destroyed at the statement's end. This makes statement | |
170 | expressions inside macros slightly different from function calls. In | |
171 | the latter case temporaries introduced during argument evaluation will | |
172 | be destroyed at the end of the statement that includes the function | |
173 | call. In the statement expression case they will be destroyed during | |
174 | the statement expression. For instance, | |
b98e139b | 175 | |
a5bcc582 NS |
176 | @smallexample |
177 | #define macro(a) (@{__typeof__(a) b = (a); b + 3; @}) | |
178 | template<typename T> T function(T a) @{ T b = a; return b + 3; @} | |
179 | ||
180 | void foo () | |
181 | @{ | |
182 | macro (X ()); | |
183 | function (X ()); | |
184 | @} | |
185 | @end smallexample | |
b98e139b MM |
186 | |
187 | @noindent | |
a5bcc582 NS |
188 | will have different places where temporaries are destroyed. For the |
189 | @code{macro} case, the temporary @code{X} will be destroyed just after | |
190 | the initialization of @code{b}. In the @code{function} case that | |
191 | temporary will be destroyed when the function returns. | |
b98e139b MM |
192 | |
193 | These considerations mean that it is probably a bad idea to use | |
194 | statement-expressions of this form in header files that are designed to | |
54e1d3a6 MM |
195 | work with C++. (Note that some versions of the GNU C Library contained |
196 | header files using statement-expression that lead to precisely this | |
197 | bug.) | |
b98e139b | 198 | |
16ef3acc JM |
199 | Jumping into a statement expression with @code{goto} or using a |
200 | @code{switch} statement outside the statement expression with a | |
201 | @code{case} or @code{default} label inside the statement expression is | |
202 | not permitted. Jumping into a statement expression with a computed | |
203 | @code{goto} (@pxref{Labels as Values}) yields undefined behavior. | |
204 | Jumping out of a statement expression is permitted, but if the | |
205 | statement expression is part of a larger expression then it is | |
206 | unspecified which other subexpressions of that expression have been | |
207 | evaluated except where the language definition requires certain | |
208 | subexpressions to be evaluated before or after the statement | |
209 | expression. In any case, as with a function call the evaluation of a | |
210 | statement expression is not interleaved with the evaluation of other | |
211 | parts of the containing expression. For example, | |
212 | ||
213 | @smallexample | |
214 | foo (), ((@{ bar1 (); goto a; 0; @}) + bar2 ()), baz(); | |
215 | @end smallexample | |
216 | ||
217 | @noindent | |
218 | will call @code{foo} and @code{bar1} and will not call @code{baz} but | |
219 | may or may not call @code{bar2}. If @code{bar2} is called, it will be | |
220 | called after @code{foo} and before @code{bar1} | |
221 | ||
c1f7febf RK |
222 | @node Local Labels |
223 | @section Locally Declared Labels | |
224 | @cindex local labels | |
225 | @cindex macros, local labels | |
226 | ||
14e33ee8 | 227 | GCC allows you to declare @dfn{local labels} in any nested block |
8a36672b | 228 | scope. A local label is just like an ordinary label, but you can |
14e33ee8 | 229 | only reference it (with a @code{goto} statement, or by taking its |
daf2f129 | 230 | address) within the block in which it was declared. |
c1f7febf RK |
231 | |
232 | A local label declaration looks like this: | |
233 | ||
3ab51846 | 234 | @smallexample |
c1f7febf | 235 | __label__ @var{label}; |
3ab51846 | 236 | @end smallexample |
c1f7febf RK |
237 | |
238 | @noindent | |
239 | or | |
240 | ||
3ab51846 | 241 | @smallexample |
0d893a63 | 242 | __label__ @var{label1}, @var{label2}, /* @r{@dots{}} */; |
3ab51846 | 243 | @end smallexample |
c1f7febf | 244 | |
14e33ee8 ZW |
245 | Local label declarations must come at the beginning of the block, |
246 | before any ordinary declarations or statements. | |
c1f7febf RK |
247 | |
248 | The label declaration defines the label @emph{name}, but does not define | |
249 | the label itself. You must do this in the usual way, with | |
250 | @code{@var{label}:}, within the statements of the statement expression. | |
251 | ||
14e33ee8 ZW |
252 | The local label feature is useful for complex macros. If a macro |
253 | contains nested loops, a @code{goto} can be useful for breaking out of | |
254 | them. However, an ordinary label whose scope is the whole function | |
255 | cannot be used: if the macro can be expanded several times in one | |
256 | function, the label will be multiply defined in that function. A | |
257 | local label avoids this problem. For example: | |
258 | ||
3ab51846 | 259 | @smallexample |
14e33ee8 ZW |
260 | #define SEARCH(value, array, target) \ |
261 | do @{ \ | |
262 | __label__ found; \ | |
263 | typeof (target) _SEARCH_target = (target); \ | |
264 | typeof (*(array)) *_SEARCH_array = (array); \ | |
265 | int i, j; \ | |
266 | int value; \ | |
267 | for (i = 0; i < max; i++) \ | |
268 | for (j = 0; j < max; j++) \ | |
269 | if (_SEARCH_array[i][j] == _SEARCH_target) \ | |
270 | @{ (value) = i; goto found; @} \ | |
271 | (value) = -1; \ | |
272 | found:; \ | |
273 | @} while (0) | |
3ab51846 | 274 | @end smallexample |
14e33ee8 ZW |
275 | |
276 | This could also be written using a statement-expression: | |
c1f7febf | 277 | |
3ab51846 | 278 | @smallexample |
c1f7febf | 279 | #define SEARCH(array, target) \ |
310668e8 | 280 | (@{ \ |
c1f7febf RK |
281 | __label__ found; \ |
282 | typeof (target) _SEARCH_target = (target); \ | |
283 | typeof (*(array)) *_SEARCH_array = (array); \ | |
284 | int i, j; \ | |
285 | int value; \ | |
286 | for (i = 0; i < max; i++) \ | |
287 | for (j = 0; j < max; j++) \ | |
288 | if (_SEARCH_array[i][j] == _SEARCH_target) \ | |
310668e8 | 289 | @{ value = i; goto found; @} \ |
c1f7febf RK |
290 | value = -1; \ |
291 | found: \ | |
292 | value; \ | |
293 | @}) | |
3ab51846 | 294 | @end smallexample |
c1f7febf | 295 | |
14e33ee8 ZW |
296 | Local label declarations also make the labels they declare visible to |
297 | nested functions, if there are any. @xref{Nested Functions}, for details. | |
298 | ||
c1f7febf RK |
299 | @node Labels as Values |
300 | @section Labels as Values | |
301 | @cindex labels as values | |
302 | @cindex computed gotos | |
303 | @cindex goto with computed label | |
304 | @cindex address of a label | |
305 | ||
306 | You can get the address of a label defined in the current function | |
307 | (or a containing function) with the unary operator @samp{&&}. The | |
308 | value has type @code{void *}. This value is a constant and can be used | |
309 | wherever a constant of that type is valid. For example: | |
310 | ||
3ab51846 | 311 | @smallexample |
c1f7febf | 312 | void *ptr; |
0d893a63 | 313 | /* @r{@dots{}} */ |
c1f7febf | 314 | ptr = &&foo; |
3ab51846 | 315 | @end smallexample |
c1f7febf RK |
316 | |
317 | To use these values, you need to be able to jump to one. This is done | |
318 | with the computed goto statement@footnote{The analogous feature in | |
319 | Fortran is called an assigned goto, but that name seems inappropriate in | |
320 | C, where one can do more than simply store label addresses in label | |
321 | variables.}, @code{goto *@var{exp};}. For example, | |
322 | ||
3ab51846 | 323 | @smallexample |
c1f7febf | 324 | goto *ptr; |
3ab51846 | 325 | @end smallexample |
c1f7febf RK |
326 | |
327 | @noindent | |
328 | Any expression of type @code{void *} is allowed. | |
329 | ||
330 | One way of using these constants is in initializing a static array that | |
331 | will serve as a jump table: | |
332 | ||
3ab51846 | 333 | @smallexample |
c1f7febf | 334 | static void *array[] = @{ &&foo, &&bar, &&hack @}; |
3ab51846 | 335 | @end smallexample |
c1f7febf RK |
336 | |
337 | Then you can select a label with indexing, like this: | |
338 | ||
3ab51846 | 339 | @smallexample |
c1f7febf | 340 | goto *array[i]; |
3ab51846 | 341 | @end smallexample |
c1f7febf RK |
342 | |
343 | @noindent | |
344 | Note that this does not check whether the subscript is in bounds---array | |
345 | indexing in C never does that. | |
346 | ||
347 | Such an array of label values serves a purpose much like that of the | |
348 | @code{switch} statement. The @code{switch} statement is cleaner, so | |
349 | use that rather than an array unless the problem does not fit a | |
350 | @code{switch} statement very well. | |
351 | ||
352 | Another use of label values is in an interpreter for threaded code. | |
353 | The labels within the interpreter function can be stored in the | |
354 | threaded code for super-fast dispatching. | |
355 | ||
02f52e19 | 356 | You may not use this mechanism to jump to code in a different function. |
47620e09 | 357 | If you do that, totally unpredictable things will happen. The best way to |
c1f7febf RK |
358 | avoid this is to store the label address only in automatic variables and |
359 | never pass it as an argument. | |
360 | ||
47620e09 RH |
361 | An alternate way to write the above example is |
362 | ||
3ab51846 | 363 | @smallexample |
310668e8 JM |
364 | static const int array[] = @{ &&foo - &&foo, &&bar - &&foo, |
365 | &&hack - &&foo @}; | |
47620e09 | 366 | goto *(&&foo + array[i]); |
3ab51846 | 367 | @end smallexample |
47620e09 RH |
368 | |
369 | @noindent | |
370 | This is more friendly to code living in shared libraries, as it reduces | |
371 | the number of dynamic relocations that are needed, and by consequence, | |
372 | allows the data to be read-only. | |
373 | ||
2092ee7d JJ |
374 | The @code{&&foo} expressions for the same label might have different values |
375 | if the containing function is inlined or cloned. If a program relies on | |
376 | them being always the same, @code{__attribute__((__noinline__))} should | |
377 | be used to prevent inlining. If @code{&&foo} is used | |
378 | in a static variable initializer, inlining is forbidden. | |
379 | ||
c1f7febf RK |
380 | @node Nested Functions |
381 | @section Nested Functions | |
382 | @cindex nested functions | |
383 | @cindex downward funargs | |
384 | @cindex thunks | |
385 | ||
386 | A @dfn{nested function} is a function defined inside another function. | |
387 | (Nested functions are not supported for GNU C++.) The nested function's | |
388 | name is local to the block where it is defined. For example, here we | |
389 | define a nested function named @code{square}, and call it twice: | |
390 | ||
3ab51846 | 391 | @smallexample |
c1f7febf RK |
392 | @group |
393 | foo (double a, double b) | |
394 | @{ | |
395 | double square (double z) @{ return z * z; @} | |
396 | ||
397 | return square (a) + square (b); | |
398 | @} | |
399 | @end group | |
3ab51846 | 400 | @end smallexample |
c1f7febf RK |
401 | |
402 | The nested function can access all the variables of the containing | |
403 | function that are visible at the point of its definition. This is | |
404 | called @dfn{lexical scoping}. For example, here we show a nested | |
405 | function which uses an inherited variable named @code{offset}: | |
406 | ||
3ab51846 | 407 | @smallexample |
aee96fe9 | 408 | @group |
c1f7febf RK |
409 | bar (int *array, int offset, int size) |
410 | @{ | |
411 | int access (int *array, int index) | |
412 | @{ return array[index + offset]; @} | |
413 | int i; | |
0d893a63 | 414 | /* @r{@dots{}} */ |
c1f7febf | 415 | for (i = 0; i < size; i++) |
0d893a63 | 416 | /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */ |
c1f7febf | 417 | @} |
aee96fe9 | 418 | @end group |
3ab51846 | 419 | @end smallexample |
c1f7febf RK |
420 | |
421 | Nested function definitions are permitted within functions in the places | |
475b6e22 JM |
422 | where variable definitions are allowed; that is, in any block, mixed |
423 | with the other declarations and statements in the block. | |
c1f7febf RK |
424 | |
425 | It is possible to call the nested function from outside the scope of its | |
426 | name by storing its address or passing the address to another function: | |
427 | ||
3ab51846 | 428 | @smallexample |
c1f7febf RK |
429 | hack (int *array, int size) |
430 | @{ | |
431 | void store (int index, int value) | |
432 | @{ array[index] = value; @} | |
433 | ||
434 | intermediate (store, size); | |
435 | @} | |
3ab51846 | 436 | @end smallexample |
c1f7febf RK |
437 | |
438 | Here, the function @code{intermediate} receives the address of | |
439 | @code{store} as an argument. If @code{intermediate} calls @code{store}, | |
440 | the arguments given to @code{store} are used to store into @code{array}. | |
441 | But this technique works only so long as the containing function | |
442 | (@code{hack}, in this example) does not exit. | |
443 | ||
444 | If you try to call the nested function through its address after the | |
445 | containing function has exited, all hell will break loose. If you try | |
446 | to call it after a containing scope level has exited, and if it refers | |
447 | to some of the variables that are no longer in scope, you may be lucky, | |
448 | but it's not wise to take the risk. If, however, the nested function | |
449 | does not refer to anything that has gone out of scope, you should be | |
450 | safe. | |
451 | ||
9c34dbbf ZW |
452 | GCC implements taking the address of a nested function using a technique |
453 | called @dfn{trampolines}. A paper describing them is available as | |
454 | ||
455 | @noindent | |
b73b1546 | 456 | @uref{http://people.debian.org/~aaronl/Usenix88-lexic.pdf}. |
c1f7febf RK |
457 | |
458 | A nested function can jump to a label inherited from a containing | |
459 | function, provided the label was explicitly declared in the containing | |
460 | function (@pxref{Local Labels}). Such a jump returns instantly to the | |
461 | containing function, exiting the nested function which did the | |
462 | @code{goto} and any intermediate functions as well. Here is an example: | |
463 | ||
3ab51846 | 464 | @smallexample |
c1f7febf RK |
465 | @group |
466 | bar (int *array, int offset, int size) | |
467 | @{ | |
468 | __label__ failure; | |
469 | int access (int *array, int index) | |
470 | @{ | |
471 | if (index > size) | |
472 | goto failure; | |
473 | return array[index + offset]; | |
474 | @} | |
475 | int i; | |
0d893a63 | 476 | /* @r{@dots{}} */ |
c1f7febf | 477 | for (i = 0; i < size; i++) |
0d893a63 MK |
478 | /* @r{@dots{}} */ access (array, i) /* @r{@dots{}} */ |
479 | /* @r{@dots{}} */ | |
c1f7febf RK |
480 | return 0; |
481 | ||
482 | /* @r{Control comes here from @code{access} | |
483 | if it detects an error.} */ | |
484 | failure: | |
485 | return -1; | |
486 | @} | |
487 | @end group | |
3ab51846 | 488 | @end smallexample |
c1f7febf | 489 | |
28697267 DJ |
490 | A nested function always has no linkage. Declaring one with |
491 | @code{extern} or @code{static} is erroneous. If you need to declare the nested function | |
c1f7febf RK |
492 | before its definition, use @code{auto} (which is otherwise meaningless |
493 | for function declarations). | |
494 | ||
3ab51846 | 495 | @smallexample |
c1f7febf RK |
496 | bar (int *array, int offset, int size) |
497 | @{ | |
498 | __label__ failure; | |
499 | auto int access (int *, int); | |
0d893a63 | 500 | /* @r{@dots{}} */ |
c1f7febf RK |
501 | int access (int *array, int index) |
502 | @{ | |
503 | if (index > size) | |
504 | goto failure; | |
505 | return array[index + offset]; | |
506 | @} | |
0d893a63 | 507 | /* @r{@dots{}} */ |
c1f7febf | 508 | @} |
3ab51846 | 509 | @end smallexample |
c1f7febf RK |
510 | |
511 | @node Constructing Calls | |
512 | @section Constructing Function Calls | |
513 | @cindex constructing calls | |
514 | @cindex forwarding calls | |
515 | ||
516 | Using the built-in functions described below, you can record | |
517 | the arguments a function received, and call another function | |
518 | with the same arguments, without knowing the number or types | |
519 | of the arguments. | |
520 | ||
521 | You can also record the return value of that function call, | |
522 | and later return that value, without knowing what data type | |
523 | the function tried to return (as long as your caller expects | |
524 | that data type). | |
525 | ||
6429bc7c EB |
526 | However, these built-in functions may interact badly with some |
527 | sophisticated features or other extensions of the language. It | |
528 | is, therefore, not recommended to use them outside very simple | |
529 | functions acting as mere forwarders for their arguments. | |
530 | ||
84330467 JM |
531 | @deftypefn {Built-in Function} {void *} __builtin_apply_args () |
532 | This built-in function returns a pointer to data | |
c1f7febf RK |
533 | describing how to perform a call with the same arguments as were passed |
534 | to the current function. | |
535 | ||
536 | The function saves the arg pointer register, structure value address, | |
537 | and all registers that might be used to pass arguments to a function | |
538 | into a block of memory allocated on the stack. Then it returns the | |
539 | address of that block. | |
84330467 | 540 | @end deftypefn |
c1f7febf | 541 | |
84330467 JM |
542 | @deftypefn {Built-in Function} {void *} __builtin_apply (void (*@var{function})(), void *@var{arguments}, size_t @var{size}) |
543 | This built-in function invokes @var{function} | |
544 | with a copy of the parameters described by @var{arguments} | |
545 | and @var{size}. | |
c1f7febf RK |
546 | |
547 | The value of @var{arguments} should be the value returned by | |
548 | @code{__builtin_apply_args}. The argument @var{size} specifies the size | |
549 | of the stack argument data, in bytes. | |
550 | ||
84330467 | 551 | This function returns a pointer to data describing |
c1f7febf RK |
552 | how to return whatever value was returned by @var{function}. The data |
553 | is saved in a block of memory allocated on the stack. | |
554 | ||
555 | It is not always simple to compute the proper value for @var{size}. The | |
556 | value is used by @code{__builtin_apply} to compute the amount of data | |
557 | that should be pushed on the stack and copied from the incoming argument | |
558 | area. | |
84330467 | 559 | @end deftypefn |
c1f7febf | 560 | |
84330467 | 561 | @deftypefn {Built-in Function} {void} __builtin_return (void *@var{result}) |
c1f7febf RK |
562 | This built-in function returns the value described by @var{result} from |
563 | the containing function. You should specify, for @var{result}, a value | |
564 | returned by @code{__builtin_apply}. | |
84330467 | 565 | @end deftypefn |
c1f7febf | 566 | |
6ef5231b JJ |
567 | @deftypefn {Built-in Function} __builtin_va_arg_pack () |
568 | This built-in function represents all anonymous arguments of an inline | |
569 | function. It can be used only in inline functions which will be always | |
570 | inlined, never compiled as a separate function, such as those using | |
571 | @code{__attribute__ ((__always_inline__))} or | |
572 | @code{__attribute__ ((__gnu_inline__))} extern inline functions. | |
573 | It must be only passed as last argument to some other function | |
574 | with variable arguments. This is useful for writing small wrapper | |
575 | inlines for variable argument functions, when using preprocessor | |
576 | macros is undesirable. For example: | |
577 | @smallexample | |
578 | extern int myprintf (FILE *f, const char *format, ...); | |
579 | extern inline __attribute__ ((__gnu_inline__)) int | |
580 | myprintf (FILE *f, const char *format, ...) | |
581 | @{ | |
582 | int r = fprintf (f, "myprintf: "); | |
583 | if (r < 0) | |
584 | return r; | |
585 | int s = fprintf (f, format, __builtin_va_arg_pack ()); | |
586 | if (s < 0) | |
587 | return s; | |
588 | return r + s; | |
589 | @} | |
590 | @end smallexample | |
591 | @end deftypefn | |
592 | ||
ab0e176c JJ |
593 | @deftypefn {Built-in Function} __builtin_va_arg_pack_len () |
594 | This built-in function returns the number of anonymous arguments of | |
595 | an inline function. It can be used only in inline functions which | |
596 | will be always inlined, never compiled as a separate function, such | |
597 | as those using @code{__attribute__ ((__always_inline__))} or | |
598 | @code{__attribute__ ((__gnu_inline__))} extern inline functions. | |
599 | For example following will do link or runtime checking of open | |
600 | arguments for optimized code: | |
601 | @smallexample | |
602 | #ifdef __OPTIMIZE__ | |
603 | extern inline __attribute__((__gnu_inline__)) int | |
604 | myopen (const char *path, int oflag, ...) | |
605 | @{ | |
606 | if (__builtin_va_arg_pack_len () > 1) | |
607 | warn_open_too_many_arguments (); | |
608 | ||
609 | if (__builtin_constant_p (oflag)) | |
610 | @{ | |
611 | if ((oflag & O_CREAT) != 0 && __builtin_va_arg_pack_len () < 1) | |
612 | @{ | |
613 | warn_open_missing_mode (); | |
614 | return __open_2 (path, oflag); | |
615 | @} | |
616 | return open (path, oflag, __builtin_va_arg_pack ()); | |
617 | @} | |
618 | ||
619 | if (__builtin_va_arg_pack_len () < 1) | |
620 | return __open_2 (path, oflag); | |
621 | ||
622 | return open (path, oflag, __builtin_va_arg_pack ()); | |
623 | @} | |
624 | #endif | |
625 | @end smallexample | |
626 | @end deftypefn | |
627 | ||
c1f7febf RK |
628 | @node Typeof |
629 | @section Referring to a Type with @code{typeof} | |
630 | @findex typeof | |
631 | @findex sizeof | |
632 | @cindex macros, types of arguments | |
633 | ||
634 | Another way to refer to the type of an expression is with @code{typeof}. | |
635 | The syntax of using of this keyword looks like @code{sizeof}, but the | |
636 | construct acts semantically like a type name defined with @code{typedef}. | |
637 | ||
638 | There are two ways of writing the argument to @code{typeof}: with an | |
639 | expression or with a type. Here is an example with an expression: | |
640 | ||
3ab51846 | 641 | @smallexample |
c1f7febf | 642 | typeof (x[0](1)) |
3ab51846 | 643 | @end smallexample |
c1f7febf RK |
644 | |
645 | @noindent | |
89aed483 JM |
646 | This assumes that @code{x} is an array of pointers to functions; |
647 | the type described is that of the values of the functions. | |
c1f7febf RK |
648 | |
649 | Here is an example with a typename as the argument: | |
650 | ||
3ab51846 | 651 | @smallexample |
c1f7febf | 652 | typeof (int *) |
3ab51846 | 653 | @end smallexample |
c1f7febf RK |
654 | |
655 | @noindent | |
656 | Here the type described is that of pointers to @code{int}. | |
657 | ||
5490d604 | 658 | If you are writing a header file that must work when included in ISO C |
c1f7febf RK |
659 | programs, write @code{__typeof__} instead of @code{typeof}. |
660 | @xref{Alternate Keywords}. | |
661 | ||
662 | A @code{typeof}-construct can be used anywhere a typedef name could be | |
663 | used. For example, you can use it in a declaration, in a cast, or inside | |
664 | of @code{sizeof} or @code{typeof}. | |
665 | ||
928c19bb JM |
666 | The operand of @code{typeof} is evaluated for its side effects if and |
667 | only if it is an expression of variably modified type or the name of | |
668 | such a type. | |
669 | ||
95f79357 ZW |
670 | @code{typeof} is often useful in conjunction with the |
671 | statements-within-expressions feature. Here is how the two together can | |
672 | be used to define a safe ``maximum'' macro that operates on any | |
673 | arithmetic type and evaluates each of its arguments exactly once: | |
674 | ||
3ab51846 | 675 | @smallexample |
95f79357 ZW |
676 | #define max(a,b) \ |
677 | (@{ typeof (a) _a = (a); \ | |
678 | typeof (b) _b = (b); \ | |
679 | _a > _b ? _a : _b; @}) | |
3ab51846 | 680 | @end smallexample |
95f79357 | 681 | |
526278c9 VR |
682 | @cindex underscores in variables in macros |
683 | @cindex @samp{_} in variables in macros | |
684 | @cindex local variables in macros | |
685 | @cindex variables, local, in macros | |
686 | @cindex macros, local variables in | |
687 | ||
688 | The reason for using names that start with underscores for the local | |
689 | variables is to avoid conflicts with variable names that occur within the | |
690 | expressions that are substituted for @code{a} and @code{b}. Eventually we | |
691 | hope to design a new form of declaration syntax that allows you to declare | |
692 | variables whose scopes start only after their initializers; this will be a | |
693 | more reliable way to prevent such conflicts. | |
694 | ||
95f79357 ZW |
695 | @noindent |
696 | Some more examples of the use of @code{typeof}: | |
697 | ||
c1f7febf RK |
698 | @itemize @bullet |
699 | @item | |
700 | This declares @code{y} with the type of what @code{x} points to. | |
701 | ||
3ab51846 | 702 | @smallexample |
c1f7febf | 703 | typeof (*x) y; |
3ab51846 | 704 | @end smallexample |
c1f7febf RK |
705 | |
706 | @item | |
707 | This declares @code{y} as an array of such values. | |
708 | ||
3ab51846 | 709 | @smallexample |
c1f7febf | 710 | typeof (*x) y[4]; |
3ab51846 | 711 | @end smallexample |
c1f7febf RK |
712 | |
713 | @item | |
714 | This declares @code{y} as an array of pointers to characters: | |
715 | ||
3ab51846 | 716 | @smallexample |
c1f7febf | 717 | typeof (typeof (char *)[4]) y; |
3ab51846 | 718 | @end smallexample |
c1f7febf RK |
719 | |
720 | @noindent | |
721 | It is equivalent to the following traditional C declaration: | |
722 | ||
3ab51846 | 723 | @smallexample |
c1f7febf | 724 | char *y[4]; |
3ab51846 | 725 | @end smallexample |
c1f7febf RK |
726 | |
727 | To see the meaning of the declaration using @code{typeof}, and why it | |
962e6e00 | 728 | might be a useful way to write, rewrite it with these macros: |
c1f7febf | 729 | |
3ab51846 | 730 | @smallexample |
c1f7febf RK |
731 | #define pointer(T) typeof(T *) |
732 | #define array(T, N) typeof(T [N]) | |
3ab51846 | 733 | @end smallexample |
c1f7febf RK |
734 | |
735 | @noindent | |
736 | Now the declaration can be rewritten this way: | |
737 | ||
3ab51846 | 738 | @smallexample |
c1f7febf | 739 | array (pointer (char), 4) y; |
3ab51846 | 740 | @end smallexample |
c1f7febf RK |
741 | |
742 | @noindent | |
743 | Thus, @code{array (pointer (char), 4)} is the type of arrays of 4 | |
744 | pointers to @code{char}. | |
745 | @end itemize | |
746 | ||
95f79357 ZW |
747 | @emph{Compatibility Note:} In addition to @code{typeof}, GCC 2 supported |
748 | a more limited extension which permitted one to write | |
749 | ||
3ab51846 | 750 | @smallexample |
95f79357 | 751 | typedef @var{T} = @var{expr}; |
3ab51846 | 752 | @end smallexample |
95f79357 ZW |
753 | |
754 | @noindent | |
755 | with the effect of declaring @var{T} to have the type of the expression | |
756 | @var{expr}. This extension does not work with GCC 3 (versions between | |
757 | 3.0 and 3.2 will crash; 3.2.1 and later give an error). Code which | |
758 | relies on it should be rewritten to use @code{typeof}: | |
759 | ||
3ab51846 | 760 | @smallexample |
95f79357 | 761 | typedef typeof(@var{expr}) @var{T}; |
3ab51846 | 762 | @end smallexample |
95f79357 ZW |
763 | |
764 | @noindent | |
765 | This will work with all versions of GCC@. | |
766 | ||
c1f7febf RK |
767 | @node Conditionals |
768 | @section Conditionals with Omitted Operands | |
769 | @cindex conditional expressions, extensions | |
770 | @cindex omitted middle-operands | |
771 | @cindex middle-operands, omitted | |
772 | @cindex extensions, @code{?:} | |
773 | @cindex @code{?:} extensions | |
774 | ||
775 | The middle operand in a conditional expression may be omitted. Then | |
776 | if the first operand is nonzero, its value is the value of the conditional | |
777 | expression. | |
778 | ||
779 | Therefore, the expression | |
780 | ||
3ab51846 | 781 | @smallexample |
c1f7febf | 782 | x ? : y |
3ab51846 | 783 | @end smallexample |
c1f7febf RK |
784 | |
785 | @noindent | |
786 | has the value of @code{x} if that is nonzero; otherwise, the value of | |
787 | @code{y}. | |
788 | ||
789 | This example is perfectly equivalent to | |
790 | ||
3ab51846 | 791 | @smallexample |
c1f7febf | 792 | x ? x : y |
3ab51846 | 793 | @end smallexample |
c1f7febf RK |
794 | |
795 | @cindex side effect in ?: | |
796 | @cindex ?: side effect | |
797 | @noindent | |
798 | In this simple case, the ability to omit the middle operand is not | |
799 | especially useful. When it becomes useful is when the first operand does, | |
800 | or may (if it is a macro argument), contain a side effect. Then repeating | |
801 | the operand in the middle would perform the side effect twice. Omitting | |
802 | the middle operand uses the value already computed without the undesirable | |
803 | effects of recomputing it. | |
804 | ||
805 | @node Long Long | |
806 | @section Double-Word Integers | |
807 | @cindex @code{long long} data types | |
808 | @cindex double-word arithmetic | |
809 | @cindex multiprecision arithmetic | |
4b404517 JM |
810 | @cindex @code{LL} integer suffix |
811 | @cindex @code{ULL} integer suffix | |
c1f7febf | 812 | |
4b404517 JM |
813 | ISO C99 supports data types for integers that are at least 64 bits wide, |
814 | and as an extension GCC supports them in C89 mode and in C++. | |
815 | Simply write @code{long long int} for a signed integer, or | |
c1f7febf | 816 | @code{unsigned long long int} for an unsigned integer. To make an |
84330467 | 817 | integer constant of type @code{long long int}, add the suffix @samp{LL} |
c1f7febf | 818 | to the integer. To make an integer constant of type @code{unsigned long |
84330467 | 819 | long int}, add the suffix @samp{ULL} to the integer. |
c1f7febf RK |
820 | |
821 | You can use these types in arithmetic like any other integer types. | |
822 | Addition, subtraction, and bitwise boolean operations on these types | |
823 | are open-coded on all types of machines. Multiplication is open-coded | |
824 | if the machine supports fullword-to-doubleword a widening multiply | |
825 | instruction. Division and shifts are open-coded only on machines that | |
826 | provide special support. The operations that are not open-coded use | |
161d7b59 | 827 | special library routines that come with GCC@. |
c1f7febf RK |
828 | |
829 | There may be pitfalls when you use @code{long long} types for function | |
830 | arguments, unless you declare function prototypes. If a function | |
831 | expects type @code{int} for its argument, and you pass a value of type | |
832 | @code{long long int}, confusion will result because the caller and the | |
833 | subroutine will disagree about the number of bytes for the argument. | |
834 | Likewise, if the function expects @code{long long int} and you pass | |
835 | @code{int}. The best way to avoid such problems is to use prototypes. | |
836 | ||
837 | @node Complex | |
838 | @section Complex Numbers | |
839 | @cindex complex numbers | |
4b404517 JM |
840 | @cindex @code{_Complex} keyword |
841 | @cindex @code{__complex__} keyword | |
c1f7febf | 842 | |
4b404517 JM |
843 | ISO C99 supports complex floating data types, and as an extension GCC |
844 | supports them in C89 mode and in C++, and supports complex integer data | |
845 | types which are not part of ISO C99. You can declare complex types | |
846 | using the keyword @code{_Complex}. As an extension, the older GNU | |
847 | keyword @code{__complex__} is also supported. | |
c1f7febf | 848 | |
4b404517 | 849 | For example, @samp{_Complex double x;} declares @code{x} as a |
c1f7febf | 850 | variable whose real part and imaginary part are both of type |
4b404517 | 851 | @code{double}. @samp{_Complex short int y;} declares @code{y} to |
c1f7febf RK |
852 | have real and imaginary parts of type @code{short int}; this is not |
853 | likely to be useful, but it shows that the set of complex types is | |
854 | complete. | |
855 | ||
856 | To write a constant with a complex data type, use the suffix @samp{i} or | |
857 | @samp{j} (either one; they are equivalent). For example, @code{2.5fi} | |
4b404517 JM |
858 | has type @code{_Complex float} and @code{3i} has type |
859 | @code{_Complex int}. Such a constant always has a pure imaginary | |
c1f7febf | 860 | value, but you can form any complex value you like by adding one to a |
4b404517 JM |
861 | real constant. This is a GNU extension; if you have an ISO C99 |
862 | conforming C library (such as GNU libc), and want to construct complex | |
863 | constants of floating type, you should include @code{<complex.h>} and | |
864 | use the macros @code{I} or @code{_Complex_I} instead. | |
c1f7febf | 865 | |
4b404517 JM |
866 | @cindex @code{__real__} keyword |
867 | @cindex @code{__imag__} keyword | |
c1f7febf RK |
868 | To extract the real part of a complex-valued expression @var{exp}, write |
869 | @code{__real__ @var{exp}}. Likewise, use @code{__imag__} to | |
4b404517 JM |
870 | extract the imaginary part. This is a GNU extension; for values of |
871 | floating type, you should use the ISO C99 functions @code{crealf}, | |
872 | @code{creal}, @code{creall}, @code{cimagf}, @code{cimag} and | |
873 | @code{cimagl}, declared in @code{<complex.h>} and also provided as | |
161d7b59 | 874 | built-in functions by GCC@. |
c1f7febf | 875 | |
4b404517 | 876 | @cindex complex conjugation |
c1f7febf | 877 | The operator @samp{~} performs complex conjugation when used on a value |
4b404517 JM |
878 | with a complex type. This is a GNU extension; for values of |
879 | floating type, you should use the ISO C99 functions @code{conjf}, | |
880 | @code{conj} and @code{conjl}, declared in @code{<complex.h>} and also | |
161d7b59 | 881 | provided as built-in functions by GCC@. |
c1f7febf | 882 | |
f0523f02 | 883 | GCC can allocate complex automatic variables in a noncontiguous |
c1f7febf | 884 | fashion; it's even possible for the real part to be in a register while |
580fb356 JW |
885 | the imaginary part is on the stack (or vice-versa). Only the DWARF2 |
886 | debug info format can represent this, so use of DWARF2 is recommended. | |
887 | If you are using the stabs debug info format, GCC describes a noncontiguous | |
888 | complex variable as if it were two separate variables of noncomplex type. | |
c1f7febf RK |
889 | If the variable's actual name is @code{foo}, the two fictitious |
890 | variables are named @code{foo$real} and @code{foo$imag}. You can | |
891 | examine and set these two fictitious variables with your debugger. | |
892 | ||
c77cd3d1 UB |
893 | @node Floating Types |
894 | @section Additional Floating Types | |
895 | @cindex additional floating types | |
896 | @cindex @code{__float80} data type | |
897 | @cindex @code{__float128} data type | |
898 | @cindex @code{w} floating point suffix | |
899 | @cindex @code{q} floating point suffix | |
900 | @cindex @code{W} floating point suffix | |
901 | @cindex @code{Q} floating point suffix | |
902 | ||
903 | As an extension, the GNU C compiler supports additional floating | |
904 | types, @code{__float80} and @code{__float128} to support 80bit | |
905 | (@code{XFmode}) and 128 bit (@code{TFmode}) floating types. | |
906 | Support for additional types includes the arithmetic operators: | |
907 | add, subtract, multiply, divide; unary arithmetic operators; | |
908 | relational operators; equality operators; and conversions to and from | |
909 | integer and other floating types. Use a suffix @samp{w} or @samp{W} | |
910 | in a literal constant of type @code{__float80} and @samp{q} or @samp{Q} | |
911 | for @code{_float128}. You can declare complex types using the | |
912 | corresponding internal complex type, @code{XCmode} for @code{__float80} | |
913 | type and @code{TCmode} for @code{__float128} type: | |
914 | ||
915 | @smallexample | |
916 | typedef _Complex float __attribute__((mode(TC))) _Complex128; | |
917 | typedef _Complex float __attribute__((mode(XC))) _Complex80; | |
918 | @end smallexample | |
919 | ||
920 | Not all targets support additional floating point types. @code{__float80} | |
921 | is supported on i386, x86_64 and ia64 targets and target @code{__float128} | |
922 | is supported on x86_64 and ia64 targets. | |
923 | ||
9a8ce21f | 924 | @node Decimal Float |
85a92f7e JJ |
925 | @section Decimal Floating Types |
926 | @cindex decimal floating types | |
9a8ce21f JG |
927 | @cindex @code{_Decimal32} data type |
928 | @cindex @code{_Decimal64} data type | |
929 | @cindex @code{_Decimal128} data type | |
930 | @cindex @code{df} integer suffix | |
931 | @cindex @code{dd} integer suffix | |
932 | @cindex @code{dl} integer suffix | |
933 | @cindex @code{DF} integer suffix | |
934 | @cindex @code{DD} integer suffix | |
935 | @cindex @code{DL} integer suffix | |
936 | ||
85a92f7e | 937 | As an extension, the GNU C compiler supports decimal floating types as |
853eda8d | 938 | defined in the N1312 draft of ISO/IEC WDTR24732. Support for decimal |
85a92f7e JJ |
939 | floating types in GCC will evolve as the draft technical report changes. |
940 | Calling conventions for any target might also change. Not all targets | |
941 | support decimal floating types. | |
9a8ce21f | 942 | |
85a92f7e JJ |
943 | The decimal floating types are @code{_Decimal32}, @code{_Decimal64}, and |
944 | @code{_Decimal128}. They use a radix of ten, unlike the floating types | |
945 | @code{float}, @code{double}, and @code{long double} whose radix is not | |
946 | specified by the C standard but is usually two. | |
947 | ||
948 | Support for decimal floating types includes the arithmetic operators | |
9a8ce21f JG |
949 | add, subtract, multiply, divide; unary arithmetic operators; |
950 | relational operators; equality operators; and conversions to and from | |
85a92f7e | 951 | integer and other floating types. Use a suffix @samp{df} or |
9a8ce21f JG |
952 | @samp{DF} in a literal constant of type @code{_Decimal32}, @samp{dd} |
953 | or @samp{DD} for @code{_Decimal64}, and @samp{dl} or @samp{DL} for | |
954 | @code{_Decimal128}. | |
955 | ||
85a92f7e JJ |
956 | GCC support of decimal float as specified by the draft technical report |
957 | is incomplete: | |
958 | ||
959 | @itemize @bullet | |
85a92f7e JJ |
960 | @item |
961 | When the value of a decimal floating type cannot be represented in the | |
962 | integer type to which it is being converted, the result is undefined | |
963 | rather than the result value specified by the draft technical report. | |
853eda8d JJ |
964 | |
965 | @item | |
966 | GCC does not provide the C library functionality associated with | |
967 | @file{math.h}, @file{fenv.h}, @file{stdio.h}, @file{stdlib.h}, and | |
968 | @file{wchar.h}, which must come from a separate C library implementation. | |
969 | Because of this the GNU C compiler does not define macro | |
970 | @code{__STDC_DEC_FP__} to indicate that the implementation conforms to | |
971 | the technical report. | |
85a92f7e | 972 | @end itemize |
9a8ce21f JG |
973 | |
974 | Types @code{_Decimal32}, @code{_Decimal64}, and @code{_Decimal128} | |
975 | are supported by the DWARF2 debug information format. | |
976 | ||
6f4d7222 | 977 | @node Hex Floats |
6b42b9ea UD |
978 | @section Hex Floats |
979 | @cindex hex floats | |
c5c76735 | 980 | |
4b404517 | 981 | ISO C99 supports floating-point numbers written not only in the usual |
6f4d7222 | 982 | decimal notation, such as @code{1.55e1}, but also numbers such as |
4b404517 JM |
983 | @code{0x1.fp3} written in hexadecimal format. As a GNU extension, GCC |
984 | supports this in C89 mode (except in some cases when strictly | |
985 | conforming) and in C++. In that format the | |
84330467 | 986 | @samp{0x} hex introducer and the @samp{p} or @samp{P} exponent field are |
6f4d7222 | 987 | mandatory. The exponent is a decimal number that indicates the power of |
84330467 | 988 | 2 by which the significant part will be multiplied. Thus @samp{0x1.f} is |
aee96fe9 JM |
989 | @tex |
990 | $1 {15\over16}$, | |
991 | @end tex | |
992 | @ifnottex | |
993 | 1 15/16, | |
994 | @end ifnottex | |
995 | @samp{p3} multiplies it by 8, and the value of @code{0x1.fp3} | |
6f4d7222 UD |
996 | is the same as @code{1.55e1}. |
997 | ||
998 | Unlike for floating-point numbers in the decimal notation the exponent | |
999 | is always required in the hexadecimal notation. Otherwise the compiler | |
1000 | would not be able to resolve the ambiguity of, e.g., @code{0x1.f}. This | |
84330467 | 1001 | could mean @code{1.0f} or @code{1.9375} since @samp{f} is also the |
6f4d7222 UD |
1002 | extension for floating-point constants of type @code{float}. |
1003 | ||
0f996086 CF |
1004 | @node Fixed-Point |
1005 | @section Fixed-Point Types | |
1006 | @cindex fixed-point types | |
1007 | @cindex @code{_Fract} data type | |
1008 | @cindex @code{_Accum} data type | |
1009 | @cindex @code{_Sat} data type | |
1010 | @cindex @code{hr} fixed-suffix | |
1011 | @cindex @code{r} fixed-suffix | |
1012 | @cindex @code{lr} fixed-suffix | |
1013 | @cindex @code{llr} fixed-suffix | |
1014 | @cindex @code{uhr} fixed-suffix | |
1015 | @cindex @code{ur} fixed-suffix | |
1016 | @cindex @code{ulr} fixed-suffix | |
1017 | @cindex @code{ullr} fixed-suffix | |
1018 | @cindex @code{hk} fixed-suffix | |
1019 | @cindex @code{k} fixed-suffix | |
1020 | @cindex @code{lk} fixed-suffix | |
1021 | @cindex @code{llk} fixed-suffix | |
1022 | @cindex @code{uhk} fixed-suffix | |
1023 | @cindex @code{uk} fixed-suffix | |
1024 | @cindex @code{ulk} fixed-suffix | |
1025 | @cindex @code{ullk} fixed-suffix | |
1026 | @cindex @code{HR} fixed-suffix | |
1027 | @cindex @code{R} fixed-suffix | |
1028 | @cindex @code{LR} fixed-suffix | |
1029 | @cindex @code{LLR} fixed-suffix | |
1030 | @cindex @code{UHR} fixed-suffix | |
1031 | @cindex @code{UR} fixed-suffix | |
1032 | @cindex @code{ULR} fixed-suffix | |
1033 | @cindex @code{ULLR} fixed-suffix | |
1034 | @cindex @code{HK} fixed-suffix | |
1035 | @cindex @code{K} fixed-suffix | |
1036 | @cindex @code{LK} fixed-suffix | |
1037 | @cindex @code{LLK} fixed-suffix | |
1038 | @cindex @code{UHK} fixed-suffix | |
1039 | @cindex @code{UK} fixed-suffix | |
1040 | @cindex @code{ULK} fixed-suffix | |
1041 | @cindex @code{ULLK} fixed-suffix | |
1042 | ||
1043 | As an extension, the GNU C compiler supports fixed-point types as | |
1044 | defined in the N1169 draft of ISO/IEC DTR 18037. Support for fixed-point | |
1045 | types in GCC will evolve as the draft technical report changes. | |
1046 | Calling conventions for any target might also change. Not all targets | |
1047 | support fixed-point types. | |
1048 | ||
1049 | The fixed-point types are | |
1050 | @code{short _Fract}, | |
1051 | @code{_Fract}, | |
1052 | @code{long _Fract}, | |
1053 | @code{long long _Fract}, | |
1054 | @code{unsigned short _Fract}, | |
1055 | @code{unsigned _Fract}, | |
1056 | @code{unsigned long _Fract}, | |
1057 | @code{unsigned long long _Fract}, | |
1058 | @code{_Sat short _Fract}, | |
1059 | @code{_Sat _Fract}, | |
1060 | @code{_Sat long _Fract}, | |
1061 | @code{_Sat long long _Fract}, | |
1062 | @code{_Sat unsigned short _Fract}, | |
1063 | @code{_Sat unsigned _Fract}, | |
1064 | @code{_Sat unsigned long _Fract}, | |
1065 | @code{_Sat unsigned long long _Fract}, | |
1066 | @code{short _Accum}, | |
1067 | @code{_Accum}, | |
1068 | @code{long _Accum}, | |
1069 | @code{long long _Accum}, | |
1070 | @code{unsigned short _Accum}, | |
1071 | @code{unsigned _Accum}, | |
1072 | @code{unsigned long _Accum}, | |
1073 | @code{unsigned long long _Accum}, | |
1074 | @code{_Sat short _Accum}, | |
1075 | @code{_Sat _Accum}, | |
1076 | @code{_Sat long _Accum}, | |
1077 | @code{_Sat long long _Accum}, | |
1078 | @code{_Sat unsigned short _Accum}, | |
1079 | @code{_Sat unsigned _Accum}, | |
1080 | @code{_Sat unsigned long _Accum}, | |
1081 | @code{_Sat unsigned long long _Accum}. | |
8fd94bda | 1082 | |
0f996086 CF |
1083 | Fixed-point data values contain fractional and optional integral parts. |
1084 | The format of fixed-point data varies and depends on the target machine. | |
1085 | ||
8fd94bda JJ |
1086 | Support for fixed-point types includes: |
1087 | @itemize @bullet | |
1088 | @item | |
1089 | prefix and postfix increment and decrement operators (@code{++}, @code{--}) | |
1090 | @item | |
1091 | unary arithmetic operators (@code{+}, @code{-}, @code{!}) | |
1092 | @item | |
1093 | binary arithmetic operators (@code{+}, @code{-}, @code{*}, @code{/}) | |
1094 | @item | |
1095 | binary shift operators (@code{<<}, @code{>>}) | |
1096 | @item | |
1097 | relational operators (@code{<}, @code{<=}, @code{>=}, @code{>}) | |
1098 | @item | |
1099 | equality operators (@code{==}, @code{!=}) | |
1100 | @item | |
1101 | assignment operators (@code{+=}, @code{-=}, @code{*=}, @code{/=}, | |
1102 | @code{<<=}, @code{>>=}) | |
1103 | @item | |
1104 | conversions to and from integer, floating-point, or fixed-point types | |
1105 | @end itemize | |
1106 | ||
1107 | Use a suffix in a fixed-point literal constant: | |
1108 | @itemize | |
1109 | @item @samp{hr} or @samp{HR} for @code{short _Fract} and | |
1110 | @code{_Sat short _Fract} | |
1111 | @item @samp{r} or @samp{R} for @code{_Fract} and @code{_Sat _Fract} | |
1112 | @item @samp{lr} or @samp{LR} for @code{long _Fract} and | |
1113 | @code{_Sat long _Fract} | |
1114 | @item @samp{llr} or @samp{LLR} for @code{long long _Fract} and | |
1115 | @code{_Sat long long _Fract} | |
1116 | @item @samp{uhr} or @samp{UHR} for @code{unsigned short _Fract} and | |
1117 | @code{_Sat unsigned short _Fract} | |
1118 | @item @samp{ur} or @samp{UR} for @code{unsigned _Fract} and | |
1119 | @code{_Sat unsigned _Fract} | |
1120 | @item @samp{ulr} or @samp{ULR} for @code{unsigned long _Fract} and | |
1121 | @code{_Sat unsigned long _Fract} | |
1122 | @item @samp{ullr} or @samp{ULLR} for @code{unsigned long long _Fract} | |
1123 | and @code{_Sat unsigned long long _Fract} | |
1124 | @item @samp{hk} or @samp{HK} for @code{short _Accum} and | |
1125 | @code{_Sat short _Accum} | |
1126 | @item @samp{k} or @samp{K} for @code{_Accum} and @code{_Sat _Accum} | |
1127 | @item @samp{lk} or @samp{LK} for @code{long _Accum} and | |
1128 | @code{_Sat long _Accum} | |
1129 | @item @samp{llk} or @samp{LLK} for @code{long long _Accum} and | |
1130 | @code{_Sat long long _Accum} | |
1131 | @item @samp{uhk} or @samp{UHK} for @code{unsigned short _Accum} and | |
1132 | @code{_Sat unsigned short _Accum} | |
1133 | @item @samp{uk} or @samp{UK} for @code{unsigned _Accum} and | |
1134 | @code{_Sat unsigned _Accum} | |
1135 | @item @samp{ulk} or @samp{ULK} for @code{unsigned long _Accum} and | |
1136 | @code{_Sat unsigned long _Accum} | |
1137 | @item @samp{ullk} or @samp{ULLK} for @code{unsigned long long _Accum} | |
1138 | and @code{_Sat unsigned long long _Accum} | |
1139 | @end itemize | |
0f996086 CF |
1140 | |
1141 | GCC support of fixed-point types as specified by the draft technical report | |
1142 | is incomplete: | |
1143 | ||
1144 | @itemize @bullet | |
1145 | @item | |
1146 | Pragmas to control overflow and rounding behaviors are not implemented. | |
1147 | @end itemize | |
1148 | ||
1149 | Fixed-point types are supported by the DWARF2 debug information format. | |
1150 | ||
c1f7febf RK |
1151 | @node Zero Length |
1152 | @section Arrays of Length Zero | |
1153 | @cindex arrays of length zero | |
1154 | @cindex zero-length arrays | |
1155 | @cindex length-zero arrays | |
ffc5c6a9 | 1156 | @cindex flexible array members |
c1f7febf | 1157 | |
161d7b59 | 1158 | Zero-length arrays are allowed in GNU C@. They are very useful as the |
584ef5fe | 1159 | last element of a structure which is really a header for a variable-length |
c1f7febf RK |
1160 | object: |
1161 | ||
3ab51846 | 1162 | @smallexample |
c1f7febf RK |
1163 | struct line @{ |
1164 | int length; | |
1165 | char contents[0]; | |
1166 | @}; | |
1167 | ||
584ef5fe RH |
1168 | struct line *thisline = (struct line *) |
1169 | malloc (sizeof (struct line) + this_length); | |
1170 | thisline->length = this_length; | |
3ab51846 | 1171 | @end smallexample |
c1f7febf | 1172 | |
3764f879 | 1173 | In ISO C90, you would have to give @code{contents} a length of 1, which |
c1f7febf RK |
1174 | means either you waste space or complicate the argument to @code{malloc}. |
1175 | ||
02f52e19 | 1176 | In ISO C99, you would use a @dfn{flexible array member}, which is |
584ef5fe RH |
1177 | slightly different in syntax and semantics: |
1178 | ||
1179 | @itemize @bullet | |
1180 | @item | |
1181 | Flexible array members are written as @code{contents[]} without | |
1182 | the @code{0}. | |
1183 | ||
1184 | @item | |
1185 | Flexible array members have incomplete type, and so the @code{sizeof} | |
1186 | operator may not be applied. As a quirk of the original implementation | |
1187 | of zero-length arrays, @code{sizeof} evaluates to zero. | |
1188 | ||
1189 | @item | |
1190 | Flexible array members may only appear as the last member of a | |
e7b6a0ee | 1191 | @code{struct} that is otherwise non-empty. |
2984fe64 JM |
1192 | |
1193 | @item | |
1194 | A structure containing a flexible array member, or a union containing | |
1195 | such a structure (possibly recursively), may not be a member of a | |
1196 | structure or an element of an array. (However, these uses are | |
1197 | permitted by GCC as extensions.) | |
ffc5c6a9 | 1198 | @end itemize |
a25f1211 | 1199 | |
ffc5c6a9 | 1200 | GCC versions before 3.0 allowed zero-length arrays to be statically |
e7b6a0ee DD |
1201 | initialized, as if they were flexible arrays. In addition to those |
1202 | cases that were useful, it also allowed initializations in situations | |
1203 | that would corrupt later data. Non-empty initialization of zero-length | |
1204 | arrays is now treated like any case where there are more initializer | |
1205 | elements than the array holds, in that a suitable warning about "excess | |
1206 | elements in array" is given, and the excess elements (all of them, in | |
1207 | this case) are ignored. | |
ffc5c6a9 RH |
1208 | |
1209 | Instead GCC allows static initialization of flexible array members. | |
1210 | This is equivalent to defining a new structure containing the original | |
1211 | structure followed by an array of sufficient size to contain the data. | |
e979f9e8 | 1212 | I.e.@: in the following, @code{f1} is constructed as if it were declared |
ffc5c6a9 | 1213 | like @code{f2}. |
a25f1211 | 1214 | |
3ab51846 | 1215 | @smallexample |
ffc5c6a9 RH |
1216 | struct f1 @{ |
1217 | int x; int y[]; | |
1218 | @} f1 = @{ 1, @{ 2, 3, 4 @} @}; | |
1219 | ||
1220 | struct f2 @{ | |
1221 | struct f1 f1; int data[3]; | |
1222 | @} f2 = @{ @{ 1 @}, @{ 2, 3, 4 @} @}; | |
3ab51846 | 1223 | @end smallexample |
584ef5fe | 1224 | |
ffc5c6a9 RH |
1225 | @noindent |
1226 | The convenience of this extension is that @code{f1} has the desired | |
1227 | type, eliminating the need to consistently refer to @code{f2.f1}. | |
1228 | ||
1229 | This has symmetry with normal static arrays, in that an array of | |
1230 | unknown size is also written with @code{[]}. | |
a25f1211 | 1231 | |
ffc5c6a9 RH |
1232 | Of course, this extension only makes sense if the extra data comes at |
1233 | the end of a top-level object, as otherwise we would be overwriting | |
1234 | data at subsequent offsets. To avoid undue complication and confusion | |
1235 | with initialization of deeply nested arrays, we simply disallow any | |
1236 | non-empty initialization except when the structure is the top-level | |
1237 | object. For example: | |
584ef5fe | 1238 | |
3ab51846 | 1239 | @smallexample |
ffc5c6a9 RH |
1240 | struct foo @{ int x; int y[]; @}; |
1241 | struct bar @{ struct foo z; @}; | |
1242 | ||
13ba36b4 JM |
1243 | struct foo a = @{ 1, @{ 2, 3, 4 @} @}; // @r{Valid.} |
1244 | struct bar b = @{ @{ 1, @{ 2, 3, 4 @} @} @}; // @r{Invalid.} | |
1245 | struct bar c = @{ @{ 1, @{ @} @} @}; // @r{Valid.} | |
1246 | struct foo d[1] = @{ @{ 1 @{ 2, 3, 4 @} @} @}; // @r{Invalid.} | |
3ab51846 | 1247 | @end smallexample |
4b606faf | 1248 | |
ba05abd3 GK |
1249 | @node Empty Structures |
1250 | @section Structures With No Members | |
1251 | @cindex empty structures | |
1252 | @cindex zero-size structures | |
1253 | ||
1254 | GCC permits a C structure to have no members: | |
1255 | ||
3ab51846 | 1256 | @smallexample |
ba05abd3 GK |
1257 | struct empty @{ |
1258 | @}; | |
3ab51846 | 1259 | @end smallexample |
ba05abd3 GK |
1260 | |
1261 | The structure will have size zero. In C++, empty structures are part | |
db0b376e MM |
1262 | of the language. G++ treats empty structures as if they had a single |
1263 | member of type @code{char}. | |
ba05abd3 | 1264 | |
c1f7febf RK |
1265 | @node Variable Length |
1266 | @section Arrays of Variable Length | |
1267 | @cindex variable-length arrays | |
1268 | @cindex arrays of variable length | |
4b404517 | 1269 | @cindex VLAs |
c1f7febf | 1270 | |
4b404517 JM |
1271 | Variable-length automatic arrays are allowed in ISO C99, and as an |
1272 | extension GCC accepts them in C89 mode and in C++. (However, GCC's | |
1273 | implementation of variable-length arrays does not yet conform in detail | |
1274 | to the ISO C99 standard.) These arrays are | |
c1f7febf RK |
1275 | declared like any other automatic arrays, but with a length that is not |
1276 | a constant expression. The storage is allocated at the point of | |
1277 | declaration and deallocated when the brace-level is exited. For | |
1278 | example: | |
1279 | ||
3ab51846 | 1280 | @smallexample |
c1f7febf RK |
1281 | FILE * |
1282 | concat_fopen (char *s1, char *s2, char *mode) | |
1283 | @{ | |
1284 | char str[strlen (s1) + strlen (s2) + 1]; | |
1285 | strcpy (str, s1); | |
1286 | strcat (str, s2); | |
1287 | return fopen (str, mode); | |
1288 | @} | |
3ab51846 | 1289 | @end smallexample |
c1f7febf RK |
1290 | |
1291 | @cindex scope of a variable length array | |
1292 | @cindex variable-length array scope | |
1293 | @cindex deallocating variable length arrays | |
1294 | Jumping or breaking out of the scope of the array name deallocates the | |
1295 | storage. Jumping into the scope is not allowed; you get an error | |
1296 | message for it. | |
1297 | ||
1298 | @cindex @code{alloca} vs variable-length arrays | |
1299 | You can use the function @code{alloca} to get an effect much like | |
1300 | variable-length arrays. The function @code{alloca} is available in | |
1301 | many other C implementations (but not in all). On the other hand, | |
1302 | variable-length arrays are more elegant. | |
1303 | ||
1304 | There are other differences between these two methods. Space allocated | |
1305 | with @code{alloca} exists until the containing @emph{function} returns. | |
1306 | The space for a variable-length array is deallocated as soon as the array | |
1307 | name's scope ends. (If you use both variable-length arrays and | |
1308 | @code{alloca} in the same function, deallocation of a variable-length array | |
1309 | will also deallocate anything more recently allocated with @code{alloca}.) | |
1310 | ||
1311 | You can also use variable-length arrays as arguments to functions: | |
1312 | ||
3ab51846 | 1313 | @smallexample |
c1f7febf RK |
1314 | struct entry |
1315 | tester (int len, char data[len][len]) | |
1316 | @{ | |
0d893a63 | 1317 | /* @r{@dots{}} */ |
c1f7febf | 1318 | @} |
3ab51846 | 1319 | @end smallexample |
c1f7febf RK |
1320 | |
1321 | The length of an array is computed once when the storage is allocated | |
1322 | and is remembered for the scope of the array in case you access it with | |
1323 | @code{sizeof}. | |
1324 | ||
1325 | If you want to pass the array first and the length afterward, you can | |
1326 | use a forward declaration in the parameter list---another GNU extension. | |
1327 | ||
3ab51846 | 1328 | @smallexample |
c1f7febf RK |
1329 | struct entry |
1330 | tester (int len; char data[len][len], int len) | |
1331 | @{ | |
0d893a63 | 1332 | /* @r{@dots{}} */ |
c1f7febf | 1333 | @} |
3ab51846 | 1334 | @end smallexample |
c1f7febf RK |
1335 | |
1336 | @cindex parameter forward declaration | |
1337 | The @samp{int len} before the semicolon is a @dfn{parameter forward | |
1338 | declaration}, and it serves the purpose of making the name @code{len} | |
1339 | known when the declaration of @code{data} is parsed. | |
1340 | ||
1341 | You can write any number of such parameter forward declarations in the | |
1342 | parameter list. They can be separated by commas or semicolons, but the | |
1343 | last one must end with a semicolon, which is followed by the ``real'' | |
1344 | parameter declarations. Each forward declaration must match a ``real'' | |
4b404517 JM |
1345 | declaration in parameter name and data type. ISO C99 does not support |
1346 | parameter forward declarations. | |
c1f7febf | 1347 | |
ccd96f0a NB |
1348 | @node Variadic Macros |
1349 | @section Macros with a Variable Number of Arguments. | |
c1f7febf RK |
1350 | @cindex variable number of arguments |
1351 | @cindex macro with variable arguments | |
1352 | @cindex rest argument (in macro) | |
ccd96f0a | 1353 | @cindex variadic macros |
c1f7febf | 1354 | |
ccd96f0a NB |
1355 | In the ISO C standard of 1999, a macro can be declared to accept a |
1356 | variable number of arguments much as a function can. The syntax for | |
1357 | defining the macro is similar to that of a function. Here is an | |
1358 | example: | |
c1f7febf | 1359 | |
478c9e72 | 1360 | @smallexample |
ccd96f0a | 1361 | #define debug(format, ...) fprintf (stderr, format, __VA_ARGS__) |
478c9e72 | 1362 | @end smallexample |
c1f7febf | 1363 | |
ccd96f0a NB |
1364 | Here @samp{@dots{}} is a @dfn{variable argument}. In the invocation of |
1365 | such a macro, it represents the zero or more tokens until the closing | |
1366 | parenthesis that ends the invocation, including any commas. This set of | |
1367 | tokens replaces the identifier @code{__VA_ARGS__} in the macro body | |
1368 | wherever it appears. See the CPP manual for more information. | |
1369 | ||
1370 | GCC has long supported variadic macros, and used a different syntax that | |
1371 | allowed you to give a name to the variable arguments just like any other | |
1372 | argument. Here is an example: | |
c1f7febf | 1373 | |
3ab51846 | 1374 | @smallexample |
ccd96f0a | 1375 | #define debug(format, args...) fprintf (stderr, format, args) |
3ab51846 | 1376 | @end smallexample |
c1f7febf | 1377 | |
ccd96f0a NB |
1378 | This is in all ways equivalent to the ISO C example above, but arguably |
1379 | more readable and descriptive. | |
c1f7febf | 1380 | |
ccd96f0a NB |
1381 | GNU CPP has two further variadic macro extensions, and permits them to |
1382 | be used with either of the above forms of macro definition. | |
1383 | ||
1384 | In standard C, you are not allowed to leave the variable argument out | |
1385 | entirely; but you are allowed to pass an empty argument. For example, | |
1386 | this invocation is invalid in ISO C, because there is no comma after | |
1387 | the string: | |
c1f7febf | 1388 | |
3ab51846 | 1389 | @smallexample |
ccd96f0a | 1390 | debug ("A message") |
3ab51846 | 1391 | @end smallexample |
c1f7febf | 1392 | |
ccd96f0a NB |
1393 | GNU CPP permits you to completely omit the variable arguments in this |
1394 | way. In the above examples, the compiler would complain, though since | |
1395 | the expansion of the macro still has the extra comma after the format | |
1396 | string. | |
1397 | ||
1398 | To help solve this problem, CPP behaves specially for variable arguments | |
1399 | used with the token paste operator, @samp{##}. If instead you write | |
c1f7febf | 1400 | |
478c9e72 | 1401 | @smallexample |
ccd96f0a | 1402 | #define debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__) |
478c9e72 | 1403 | @end smallexample |
c1f7febf | 1404 | |
ccd96f0a NB |
1405 | and if the variable arguments are omitted or empty, the @samp{##} |
1406 | operator causes the preprocessor to remove the comma before it. If you | |
1407 | do provide some variable arguments in your macro invocation, GNU CPP | |
1408 | does not complain about the paste operation and instead places the | |
1409 | variable arguments after the comma. Just like any other pasted macro | |
1410 | argument, these arguments are not macro expanded. | |
1411 | ||
1412 | @node Escaped Newlines | |
1413 | @section Slightly Looser Rules for Escaped Newlines | |
1414 | @cindex escaped newlines | |
1415 | @cindex newlines (escaped) | |
1416 | ||
f458d1d5 ZW |
1417 | Recently, the preprocessor has relaxed its treatment of escaped |
1418 | newlines. Previously, the newline had to immediately follow a | |
e6cc3a24 ZW |
1419 | backslash. The current implementation allows whitespace in the form |
1420 | of spaces, horizontal and vertical tabs, and form feeds between the | |
ccd96f0a NB |
1421 | backslash and the subsequent newline. The preprocessor issues a |
1422 | warning, but treats it as a valid escaped newline and combines the two | |
1423 | lines to form a single logical line. This works within comments and | |
e6cc3a24 ZW |
1424 | tokens, as well as between tokens. Comments are @emph{not} treated as |
1425 | whitespace for the purposes of this relaxation, since they have not | |
1426 | yet been replaced with spaces. | |
c1f7febf RK |
1427 | |
1428 | @node Subscripting | |
1429 | @section Non-Lvalue Arrays May Have Subscripts | |
1430 | @cindex subscripting | |
1431 | @cindex arrays, non-lvalue | |
1432 | ||
1433 | @cindex subscripting and function values | |
207bf485 JM |
1434 | In ISO C99, arrays that are not lvalues still decay to pointers, and |
1435 | may be subscripted, although they may not be modified or used after | |
1436 | the next sequence point and the unary @samp{&} operator may not be | |
1437 | applied to them. As an extension, GCC allows such arrays to be | |
1438 | subscripted in C89 mode, though otherwise they do not decay to | |
1439 | pointers outside C99 mode. For example, | |
4b404517 | 1440 | this is valid in GNU C though not valid in C89: |
c1f7febf | 1441 | |
3ab51846 | 1442 | @smallexample |
c1f7febf RK |
1443 | @group |
1444 | struct foo @{int a[4];@}; | |
1445 | ||
1446 | struct foo f(); | |
1447 | ||
1448 | bar (int index) | |
1449 | @{ | |
1450 | return f().a[index]; | |
1451 | @} | |
1452 | @end group | |
3ab51846 | 1453 | @end smallexample |
c1f7febf RK |
1454 | |
1455 | @node Pointer Arith | |
1456 | @section Arithmetic on @code{void}- and Function-Pointers | |
1457 | @cindex void pointers, arithmetic | |
1458 | @cindex void, size of pointer to | |
1459 | @cindex function pointers, arithmetic | |
1460 | @cindex function, size of pointer to | |
1461 | ||
1462 | In GNU C, addition and subtraction operations are supported on pointers to | |
1463 | @code{void} and on pointers to functions. This is done by treating the | |
1464 | size of a @code{void} or of a function as 1. | |
1465 | ||
1466 | A consequence of this is that @code{sizeof} is also allowed on @code{void} | |
1467 | and on function types, and returns 1. | |
1468 | ||
84330467 JM |
1469 | @opindex Wpointer-arith |
1470 | The option @option{-Wpointer-arith} requests a warning if these extensions | |
c1f7febf RK |
1471 | are used. |
1472 | ||
1473 | @node Initializers | |
1474 | @section Non-Constant Initializers | |
1475 | @cindex initializers, non-constant | |
1476 | @cindex non-constant initializers | |
1477 | ||
4b404517 | 1478 | As in standard C++ and ISO C99, the elements of an aggregate initializer for an |
161d7b59 | 1479 | automatic variable are not required to be constant expressions in GNU C@. |
c1f7febf RK |
1480 | Here is an example of an initializer with run-time varying elements: |
1481 | ||
3ab51846 | 1482 | @smallexample |
c1f7febf RK |
1483 | foo (float f, float g) |
1484 | @{ | |
1485 | float beat_freqs[2] = @{ f-g, f+g @}; | |
0d893a63 | 1486 | /* @r{@dots{}} */ |
c1f7febf | 1487 | @} |
3ab51846 | 1488 | @end smallexample |
c1f7febf | 1489 | |
4b404517 JM |
1490 | @node Compound Literals |
1491 | @section Compound Literals | |
c1f7febf RK |
1492 | @cindex constructor expressions |
1493 | @cindex initializations in expressions | |
1494 | @cindex structures, constructor expression | |
1495 | @cindex expressions, constructor | |
4b404517 JM |
1496 | @cindex compound literals |
1497 | @c The GNU C name for what C99 calls compound literals was "constructor expressions". | |
c1f7febf | 1498 | |
4b404517 | 1499 | ISO C99 supports compound literals. A compound literal looks like |
c1f7febf RK |
1500 | a cast containing an initializer. Its value is an object of the |
1501 | type specified in the cast, containing the elements specified in | |
db3acfa5 JM |
1502 | the initializer; it is an lvalue. As an extension, GCC supports |
1503 | compound literals in C89 mode and in C++. | |
c1f7febf RK |
1504 | |
1505 | Usually, the specified type is a structure. Assume that | |
1506 | @code{struct foo} and @code{structure} are declared as shown: | |
1507 | ||
3ab51846 | 1508 | @smallexample |
c1f7febf | 1509 | struct foo @{int a; char b[2];@} structure; |
3ab51846 | 1510 | @end smallexample |
c1f7febf RK |
1511 | |
1512 | @noindent | |
4b404517 | 1513 | Here is an example of constructing a @code{struct foo} with a compound literal: |
c1f7febf | 1514 | |
3ab51846 | 1515 | @smallexample |
c1f7febf | 1516 | structure = ((struct foo) @{x + y, 'a', 0@}); |
3ab51846 | 1517 | @end smallexample |
c1f7febf RK |
1518 | |
1519 | @noindent | |
1520 | This is equivalent to writing the following: | |
1521 | ||
3ab51846 | 1522 | @smallexample |
c1f7febf RK |
1523 | @{ |
1524 | struct foo temp = @{x + y, 'a', 0@}; | |
1525 | structure = temp; | |
1526 | @} | |
3ab51846 | 1527 | @end smallexample |
c1f7febf | 1528 | |
4b404517 | 1529 | You can also construct an array. If all the elements of the compound literal |
c1f7febf | 1530 | are (made up of) simple constant expressions, suitable for use in |
db3acfa5 JM |
1531 | initializers of objects of static storage duration, then the compound |
1532 | literal can be coerced to a pointer to its first element and used in | |
1533 | such an initializer, as shown here: | |
c1f7febf | 1534 | |
3ab51846 | 1535 | @smallexample |
c1f7febf | 1536 | char **foo = (char *[]) @{ "x", "y", "z" @}; |
3ab51846 | 1537 | @end smallexample |
c1f7febf | 1538 | |
4b404517 JM |
1539 | Compound literals for scalar types and union types are is |
1540 | also allowed, but then the compound literal is equivalent | |
c1f7febf RK |
1541 | to a cast. |
1542 | ||
59c83dbf JJ |
1543 | As a GNU extension, GCC allows initialization of objects with static storage |
1544 | duration by compound literals (which is not possible in ISO C99, because | |
1545 | the initializer is not a constant). | |
1546 | It is handled as if the object was initialized only with the bracket | |
1eaf20ec | 1547 | enclosed list if the types of the compound literal and the object match. |
59c83dbf JJ |
1548 | The initializer list of the compound literal must be constant. |
1549 | If the object being initialized has array type of unknown size, the size is | |
ad47f1e5 | 1550 | determined by compound literal size. |
59c83dbf | 1551 | |
3ab51846 | 1552 | @smallexample |
59c83dbf JJ |
1553 | static struct foo x = (struct foo) @{1, 'a', 'b'@}; |
1554 | static int y[] = (int []) @{1, 2, 3@}; | |
1555 | static int z[] = (int [3]) @{1@}; | |
3ab51846 | 1556 | @end smallexample |
59c83dbf JJ |
1557 | |
1558 | @noindent | |
1559 | The above lines are equivalent to the following: | |
3ab51846 | 1560 | @smallexample |
59c83dbf JJ |
1561 | static struct foo x = @{1, 'a', 'b'@}; |
1562 | static int y[] = @{1, 2, 3@}; | |
ad47f1e5 | 1563 | static int z[] = @{1, 0, 0@}; |
3ab51846 | 1564 | @end smallexample |
59c83dbf | 1565 | |
4b404517 JM |
1566 | @node Designated Inits |
1567 | @section Designated Initializers | |
c1f7febf RK |
1568 | @cindex initializers with labeled elements |
1569 | @cindex labeled elements in initializers | |
1570 | @cindex case labels in initializers | |
4b404517 | 1571 | @cindex designated initializers |
c1f7febf | 1572 | |
26d4fec7 | 1573 | Standard C89 requires the elements of an initializer to appear in a fixed |
c1f7febf RK |
1574 | order, the same as the order of the elements in the array or structure |
1575 | being initialized. | |
1576 | ||
26d4fec7 JM |
1577 | In ISO C99 you can give the elements in any order, specifying the array |
1578 | indices or structure field names they apply to, and GNU C allows this as | |
1579 | an extension in C89 mode as well. This extension is not | |
c1f7febf RK |
1580 | implemented in GNU C++. |
1581 | ||
26d4fec7 | 1582 | To specify an array index, write |
c1f7febf RK |
1583 | @samp{[@var{index}] =} before the element value. For example, |
1584 | ||
3ab51846 | 1585 | @smallexample |
26d4fec7 | 1586 | int a[6] = @{ [4] = 29, [2] = 15 @}; |
3ab51846 | 1587 | @end smallexample |
c1f7febf RK |
1588 | |
1589 | @noindent | |
1590 | is equivalent to | |
1591 | ||
3ab51846 | 1592 | @smallexample |
c1f7febf | 1593 | int a[6] = @{ 0, 0, 15, 0, 29, 0 @}; |
3ab51846 | 1594 | @end smallexample |
c1f7febf RK |
1595 | |
1596 | @noindent | |
1597 | The index values must be constant expressions, even if the array being | |
1598 | initialized is automatic. | |
1599 | ||
26d4fec7 JM |
1600 | An alternative syntax for this which has been obsolete since GCC 2.5 but |
1601 | GCC still accepts is to write @samp{[@var{index}]} before the element | |
1602 | value, with no @samp{=}. | |
1603 | ||
c1f7febf | 1604 | To initialize a range of elements to the same value, write |
26d4fec7 JM |
1605 | @samp{[@var{first} ... @var{last}] = @var{value}}. This is a GNU |
1606 | extension. For example, | |
c1f7febf | 1607 | |
3ab51846 | 1608 | @smallexample |
c1f7febf | 1609 | int widths[] = @{ [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 @}; |
3ab51846 | 1610 | @end smallexample |
c1f7febf | 1611 | |
8b6a5902 JJ |
1612 | @noindent |
1613 | If the value in it has side-effects, the side-effects will happen only once, | |
1614 | not for each initialized field by the range initializer. | |
1615 | ||
c1f7febf RK |
1616 | @noindent |
1617 | Note that the length of the array is the highest value specified | |
1618 | plus one. | |
1619 | ||
1620 | In a structure initializer, specify the name of a field to initialize | |
26d4fec7 | 1621 | with @samp{.@var{fieldname} =} before the element value. For example, |
c1f7febf RK |
1622 | given the following structure, |
1623 | ||
3ab51846 | 1624 | @smallexample |
c1f7febf | 1625 | struct point @{ int x, y; @}; |
3ab51846 | 1626 | @end smallexample |
c1f7febf RK |
1627 | |
1628 | @noindent | |
1629 | the following initialization | |
1630 | ||
3ab51846 | 1631 | @smallexample |
26d4fec7 | 1632 | struct point p = @{ .y = yvalue, .x = xvalue @}; |
3ab51846 | 1633 | @end smallexample |
c1f7febf RK |
1634 | |
1635 | @noindent | |
1636 | is equivalent to | |
1637 | ||
3ab51846 | 1638 | @smallexample |
c1f7febf | 1639 | struct point p = @{ xvalue, yvalue @}; |
3ab51846 | 1640 | @end smallexample |
c1f7febf | 1641 | |
26d4fec7 JM |
1642 | Another syntax which has the same meaning, obsolete since GCC 2.5, is |
1643 | @samp{@var{fieldname}:}, as shown here: | |
c1f7febf | 1644 | |
3ab51846 | 1645 | @smallexample |
26d4fec7 | 1646 | struct point p = @{ y: yvalue, x: xvalue @}; |
3ab51846 | 1647 | @end smallexample |
c1f7febf | 1648 | |
4b404517 JM |
1649 | @cindex designators |
1650 | The @samp{[@var{index}]} or @samp{.@var{fieldname}} is known as a | |
1651 | @dfn{designator}. You can also use a designator (or the obsolete colon | |
1652 | syntax) when initializing a union, to specify which element of the union | |
1653 | should be used. For example, | |
c1f7febf | 1654 | |
3ab51846 | 1655 | @smallexample |
c1f7febf RK |
1656 | union foo @{ int i; double d; @}; |
1657 | ||
26d4fec7 | 1658 | union foo f = @{ .d = 4 @}; |
3ab51846 | 1659 | @end smallexample |
c1f7febf RK |
1660 | |
1661 | @noindent | |
1662 | will convert 4 to a @code{double} to store it in the union using | |
1663 | the second element. By contrast, casting 4 to type @code{union foo} | |
1664 | would store it into the union as the integer @code{i}, since it is | |
1665 | an integer. (@xref{Cast to Union}.) | |
1666 | ||
1667 | You can combine this technique of naming elements with ordinary C | |
1668 | initialization of successive elements. Each initializer element that | |
4b404517 | 1669 | does not have a designator applies to the next consecutive element of the |
c1f7febf RK |
1670 | array or structure. For example, |
1671 | ||
3ab51846 | 1672 | @smallexample |
c1f7febf | 1673 | int a[6] = @{ [1] = v1, v2, [4] = v4 @}; |
3ab51846 | 1674 | @end smallexample |
c1f7febf RK |
1675 | |
1676 | @noindent | |
1677 | is equivalent to | |
1678 | ||
3ab51846 | 1679 | @smallexample |
c1f7febf | 1680 | int a[6] = @{ 0, v1, v2, 0, v4, 0 @}; |
3ab51846 | 1681 | @end smallexample |
c1f7febf RK |
1682 | |
1683 | Labeling the elements of an array initializer is especially useful | |
1684 | when the indices are characters or belong to an @code{enum} type. | |
1685 | For example: | |
1686 | ||
3ab51846 | 1687 | @smallexample |
c1f7febf RK |
1688 | int whitespace[256] |
1689 | = @{ [' '] = 1, ['\t'] = 1, ['\h'] = 1, | |
1690 | ['\f'] = 1, ['\n'] = 1, ['\r'] = 1 @}; | |
3ab51846 | 1691 | @end smallexample |
c1f7febf | 1692 | |
4b404517 | 1693 | @cindex designator lists |
26d4fec7 | 1694 | You can also write a series of @samp{.@var{fieldname}} and |
4b404517 | 1695 | @samp{[@var{index}]} designators before an @samp{=} to specify a |
26d4fec7 JM |
1696 | nested subobject to initialize; the list is taken relative to the |
1697 | subobject corresponding to the closest surrounding brace pair. For | |
1698 | example, with the @samp{struct point} declaration above: | |
1699 | ||
478c9e72 | 1700 | @smallexample |
26d4fec7 | 1701 | struct point ptarray[10] = @{ [2].y = yv2, [2].x = xv2, [0].x = xv0 @}; |
478c9e72 | 1702 | @end smallexample |
26d4fec7 | 1703 | |
8b6a5902 JJ |
1704 | @noindent |
1705 | If the same field is initialized multiple times, it will have value from | |
1706 | the last initialization. If any such overridden initialization has | |
1707 | side-effect, it is unspecified whether the side-effect happens or not. | |
2dd76960 | 1708 | Currently, GCC will discard them and issue a warning. |
8b6a5902 | 1709 | |
c1f7febf RK |
1710 | @node Case Ranges |
1711 | @section Case Ranges | |
1712 | @cindex case ranges | |
1713 | @cindex ranges in case statements | |
1714 | ||
1715 | You can specify a range of consecutive values in a single @code{case} label, | |
1716 | like this: | |
1717 | ||
3ab51846 | 1718 | @smallexample |
c1f7febf | 1719 | case @var{low} ... @var{high}: |
3ab51846 | 1720 | @end smallexample |
c1f7febf RK |
1721 | |
1722 | @noindent | |
1723 | This has the same effect as the proper number of individual @code{case} | |
1724 | labels, one for each integer value from @var{low} to @var{high}, inclusive. | |
1725 | ||
1726 | This feature is especially useful for ranges of ASCII character codes: | |
1727 | ||
3ab51846 | 1728 | @smallexample |
c1f7febf | 1729 | case 'A' ... 'Z': |
3ab51846 | 1730 | @end smallexample |
c1f7febf RK |
1731 | |
1732 | @strong{Be careful:} Write spaces around the @code{...}, for otherwise | |
1733 | it may be parsed wrong when you use it with integer values. For example, | |
1734 | write this: | |
1735 | ||
3ab51846 | 1736 | @smallexample |
c1f7febf | 1737 | case 1 ... 5: |
3ab51846 | 1738 | @end smallexample |
c1f7febf RK |
1739 | |
1740 | @noindent | |
1741 | rather than this: | |
1742 | ||
3ab51846 | 1743 | @smallexample |
c1f7febf | 1744 | case 1...5: |
3ab51846 | 1745 | @end smallexample |
c1f7febf RK |
1746 | |
1747 | @node Cast to Union | |
1748 | @section Cast to a Union Type | |
1749 | @cindex cast to a union | |
1750 | @cindex union, casting to a | |
1751 | ||
1752 | A cast to union type is similar to other casts, except that the type | |
1753 | specified is a union type. You can specify the type either with | |
1754 | @code{union @var{tag}} or with a typedef name. A cast to union is actually | |
1755 | a constructor though, not a cast, and hence does not yield an lvalue like | |
4b404517 | 1756 | normal casts. (@xref{Compound Literals}.) |
c1f7febf RK |
1757 | |
1758 | The types that may be cast to the union type are those of the members | |
1759 | of the union. Thus, given the following union and variables: | |
1760 | ||
3ab51846 | 1761 | @smallexample |
c1f7febf RK |
1762 | union foo @{ int i; double d; @}; |
1763 | int x; | |
1764 | double y; | |
3ab51846 | 1765 | @end smallexample |
c1f7febf RK |
1766 | |
1767 | @noindent | |
aee96fe9 | 1768 | both @code{x} and @code{y} can be cast to type @code{union foo}. |
c1f7febf RK |
1769 | |
1770 | Using the cast as the right-hand side of an assignment to a variable of | |
1771 | union type is equivalent to storing in a member of the union: | |
1772 | ||
3ab51846 | 1773 | @smallexample |
c1f7febf | 1774 | union foo u; |
0d893a63 | 1775 | /* @r{@dots{}} */ |
c1f7febf RK |
1776 | u = (union foo) x @equiv{} u.i = x |
1777 | u = (union foo) y @equiv{} u.d = y | |
3ab51846 | 1778 | @end smallexample |
c1f7febf RK |
1779 | |
1780 | You can also use the union cast as a function argument: | |
1781 | ||
3ab51846 | 1782 | @smallexample |
c1f7febf | 1783 | void hack (union foo); |
0d893a63 | 1784 | /* @r{@dots{}} */ |
c1f7febf | 1785 | hack ((union foo) x); |
3ab51846 | 1786 | @end smallexample |
c1f7febf | 1787 | |
4b404517 JM |
1788 | @node Mixed Declarations |
1789 | @section Mixed Declarations and Code | |
1790 | @cindex mixed declarations and code | |
1791 | @cindex declarations, mixed with code | |
1792 | @cindex code, mixed with declarations | |
1793 | ||
1794 | ISO C99 and ISO C++ allow declarations and code to be freely mixed | |
1795 | within compound statements. As an extension, GCC also allows this in | |
1796 | C89 mode. For example, you could do: | |
1797 | ||
3ab51846 | 1798 | @smallexample |
4b404517 | 1799 | int i; |
0d893a63 | 1800 | /* @r{@dots{}} */ |
4b404517 JM |
1801 | i++; |
1802 | int j = i + 2; | |
3ab51846 | 1803 | @end smallexample |
4b404517 JM |
1804 | |
1805 | Each identifier is visible from where it is declared until the end of | |
1806 | the enclosing block. | |
1807 | ||
c1f7febf RK |
1808 | @node Function Attributes |
1809 | @section Declaring Attributes of Functions | |
1810 | @cindex function attributes | |
1811 | @cindex declaring attributes of functions | |
1812 | @cindex functions that never return | |
6e9a3221 | 1813 | @cindex functions that return more than once |
c1f7febf RK |
1814 | @cindex functions that have no side effects |
1815 | @cindex functions in arbitrary sections | |
2a59078d | 1816 | @cindex functions that behave like malloc |
c1f7febf RK |
1817 | @cindex @code{volatile} applied to function |
1818 | @cindex @code{const} applied to function | |
26f6672d | 1819 | @cindex functions with @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style arguments |
b34c7881 | 1820 | @cindex functions with non-null pointer arguments |
c1f7febf RK |
1821 | @cindex functions that are passed arguments in registers on the 386 |
1822 | @cindex functions that pop the argument stack on the 386 | |
1823 | @cindex functions that do not pop the argument stack on the 386 | |
ab442df7 MM |
1824 | @cindex functions that have different compilation options on the 386 |
1825 | @cindex functions that have different optimization options | |
c1f7febf RK |
1826 | |
1827 | In GNU C, you declare certain things about functions called in your program | |
1828 | which help the compiler optimize function calls and check your code more | |
1829 | carefully. | |
1830 | ||
1831 | The keyword @code{__attribute__} allows you to specify special | |
1832 | attributes when making a declaration. This keyword is followed by an | |
9162542e | 1833 | attribute specification inside double parentheses. The following |
eacecf96 | 1834 | attributes are currently defined for functions on all targets: |
837edd5f GK |
1835 | @code{aligned}, @code{alloc_size}, @code{noreturn}, |
1836 | @code{returns_twice}, @code{noinline}, @code{always_inline}, | |
1837 | @code{flatten}, @code{pure}, @code{const}, @code{nothrow}, | |
1838 | @code{sentinel}, @code{format}, @code{format_arg}, | |
51bc54a6 DM |
1839 | @code{no_instrument_function}, @code{section}, @code{constructor}, |
1840 | @code{destructor}, @code{used}, @code{unused}, @code{deprecated}, | |
1841 | @code{weak}, @code{malloc}, @code{alias}, @code{warn_unused_result}, | |
d752cfdb | 1842 | @code{nonnull}, @code{gnu_inline}, @code{externally_visible}, |
d2af6a68 JJ |
1843 | @code{hot}, @code{cold}, @code{artificial}, @code{error} |
1844 | and @code{warning}. | |
837edd5f GK |
1845 | Several other attributes are defined for functions on particular |
1846 | target systems. Other attributes, including @code{section} are | |
1847 | supported for variables declarations (@pxref{Variable Attributes}) and | |
1848 | for types (@pxref{Type Attributes}). | |
c1f7febf RK |
1849 | |
1850 | You may also specify attributes with @samp{__} preceding and following | |
1851 | each keyword. This allows you to use them in header files without | |
1852 | being concerned about a possible macro of the same name. For example, | |
1853 | you may use @code{__noreturn__} instead of @code{noreturn}. | |
1854 | ||
2c5e91d2 JM |
1855 | @xref{Attribute Syntax}, for details of the exact syntax for using |
1856 | attributes. | |
1857 | ||
c1f7febf | 1858 | @table @code |
8a36672b | 1859 | @c Keep this table alphabetized by attribute name. Treat _ as space. |
c1f7febf | 1860 | |
c8619b90 NS |
1861 | @item alias ("@var{target}") |
1862 | @cindex @code{alias} attribute | |
1863 | The @code{alias} attribute causes the declaration to be emitted as an | |
1864 | alias for another symbol, which must be specified. For instance, | |
c1f7febf RK |
1865 | |
1866 | @smallexample | |
c8619b90 NS |
1867 | void __f () @{ /* @r{Do something.} */; @} |
1868 | void f () __attribute__ ((weak, alias ("__f"))); | |
c1f7febf RK |
1869 | @end smallexample |
1870 | ||
a9b0b825 | 1871 | defines @samp{f} to be a weak alias for @samp{__f}. In C++, the |
52eb57df RH |
1872 | mangled name for the target must be used. It is an error if @samp{__f} |
1873 | is not defined in the same translation unit. | |
c8619b90 NS |
1874 | |
1875 | Not all target machines support this attribute. | |
9162542e | 1876 | |
837edd5f GK |
1877 | @item aligned (@var{alignment}) |
1878 | @cindex @code{aligned} attribute | |
1879 | This attribute specifies a minimum alignment for the function, | |
1880 | measured in bytes. | |
1881 | ||
1882 | You cannot use this attribute to decrease the alignment of a function, | |
1883 | only to increase it. However, when you explicitly specify a function | |
1884 | alignment this will override the effect of the | |
1885 | @option{-falign-functions} (@pxref{Optimize Options}) option for this | |
1886 | function. | |
1887 | ||
1888 | Note that the effectiveness of @code{aligned} attributes may be | |
1889 | limited by inherent limitations in your linker. On many systems, the | |
1890 | linker is only able to arrange for functions to be aligned up to a | |
1891 | certain maximum alignment. (For some linkers, the maximum supported | |
1892 | alignment may be very very small.) See your linker documentation for | |
1893 | further information. | |
1894 | ||
1895 | The @code{aligned} attribute can also be used for variables and fields | |
1896 | (@pxref{Variable Attributes}.) | |
1897 | ||
51bc54a6 DM |
1898 | @item alloc_size |
1899 | @cindex @code{alloc_size} attribute | |
1900 | The @code{alloc_size} attribute is used to tell the compiler that the | |
1901 | function return value points to memory, where the size is given by | |
1902 | one or two of the functions parameters. GCC uses this | |
1903 | information to improve the correctness of @code{__builtin_object_size}. | |
1904 | ||
1905 | The function parameter(s) denoting the allocated size are specified by | |
1906 | one or two integer arguments supplied to the attribute. The allocated size | |
1907 | is either the value of the single function argument specified or the product | |
1908 | of the two function arguments specified. Argument numbering starts at | |
1909 | one. | |
1910 | ||
1911 | For instance, | |
1912 | ||
1913 | @smallexample | |
1914 | void* my_calloc(size_t, size_t) __attribute__((alloc_size(1,2))) | |
1c42f5c6 | 1915 | void my_realloc(void*, size_t) __attribute__((alloc_size(2))) |
51bc54a6 DM |
1916 | @end smallexample |
1917 | ||
1918 | declares that my_calloc will return memory of the size given by | |
1919 | the product of parameter 1 and 2 and that my_realloc will return memory | |
1920 | of the size given by parameter 2. | |
1921 | ||
6aa77e6c | 1922 | @item always_inline |
c8619b90 | 1923 | @cindex @code{always_inline} function attribute |
6aa77e6c AH |
1924 | Generally, functions are not inlined unless optimization is specified. |
1925 | For functions declared inline, this attribute inlines the function even | |
1926 | if no optimization level was specified. | |
1927 | ||
4eb7fd83 JJ |
1928 | @item gnu_inline |
1929 | @cindex @code{gnu_inline} function attribute | |
da1c7394 ILT |
1930 | This attribute should be used with a function which is also declared |
1931 | with the @code{inline} keyword. It directs GCC to treat the function | |
1932 | as if it were defined in gnu89 mode even when compiling in C99 or | |
1933 | gnu99 mode. | |
1934 | ||
1935 | If the function is declared @code{extern}, then this definition of the | |
1936 | function is used only for inlining. In no case is the function | |
1937 | compiled as a standalone function, not even if you take its address | |
1938 | explicitly. Such an address becomes an external reference, as if you | |
1939 | had only declared the function, and had not defined it. This has | |
1940 | almost the effect of a macro. The way to use this is to put a | |
1941 | function definition in a header file with this attribute, and put | |
1942 | another copy of the function, without @code{extern}, in a library | |
1943 | file. The definition in the header file will cause most calls to the | |
1944 | function to be inlined. If any uses of the function remain, they will | |
1945 | refer to the single copy in the library. Note that the two | |
1946 | definitions of the functions need not be precisely the same, although | |
1947 | if they do not have the same effect your program may behave oddly. | |
1948 | ||
3a47c4e4 AO |
1949 | In C, if the function is neither @code{extern} nor @code{static}, then |
1950 | the function is compiled as a standalone function, as well as being | |
da1c7394 ILT |
1951 | inlined where possible. |
1952 | ||
1953 | This is how GCC traditionally handled functions declared | |
1954 | @code{inline}. Since ISO C99 specifies a different semantics for | |
1955 | @code{inline}, this function attribute is provided as a transition | |
1956 | measure and as a useful feature in its own right. This attribute is | |
1957 | available in GCC 4.1.3 and later. It is available if either of the | |
1958 | preprocessor macros @code{__GNUC_GNU_INLINE__} or | |
1959 | @code{__GNUC_STDC_INLINE__} are defined. @xref{Inline,,An Inline | |
1960 | Function is As Fast As a Macro}. | |
4eb7fd83 | 1961 | |
3a47c4e4 AO |
1962 | In C++, this attribute does not depend on @code{extern} in any way, |
1963 | but it still requires the @code{inline} keyword to enable its special | |
1964 | behavior. | |
1965 | ||
d752cfdb | 1966 | @item artificial |
1df48f5c | 1967 | @cindex @code{artificial} function attribute |
d752cfdb JJ |
1968 | This attribute is useful for small inline wrappers which if possible |
1969 | should appear during debugging as a unit, depending on the debug | |
1970 | info format it will either mean marking the function as artificial | |
1971 | or using the caller location for all instructions within the inlined | |
1972 | body. | |
1973 | ||
0691d1d4 | 1974 | @item flatten |
1df48f5c | 1975 | @cindex @code{flatten} function attribute |
0691d1d4 RG |
1976 | Generally, inlining into a function is limited. For a function marked with |
1977 | this attribute, every call inside this function will be inlined, if possible. | |
1978 | Whether the function itself is considered for inlining depends on its size and | |
d6cc6ec9 | 1979 | the current inlining parameters. |
0691d1d4 | 1980 | |
d2af6a68 JJ |
1981 | @item error ("@var{message}") |
1982 | @cindex @code{error} function attribute | |
1983 | If this attribute is used on a function declaration and a call to such a function | |
1984 | is not eliminated through dead code elimination or other optimizations, an error | |
1985 | which will include @var{message} will be diagnosed. This is useful | |
1986 | for compile time checking, especially together with @code{__builtin_constant_p} | |
1987 | and inline functions where checking the inline function arguments is not | |
1988 | possible through @code{extern char [(condition) ? 1 : -1];} tricks. | |
1989 | While it is possible to leave the function undefined and thus invoke | |
1990 | a link failure, when using this attribute the problem will be diagnosed | |
1991 | earlier and with exact location of the call even in presence of inline | |
1992 | functions or when not emitting debugging information. | |
1993 | ||
1994 | @item warning ("@var{message}") | |
1995 | @cindex @code{warning} function attribute | |
1996 | If this attribute is used on a function declaration and a call to such a function | |
1997 | is not eliminated through dead code elimination or other optimizations, a warning | |
1998 | which will include @var{message} will be diagnosed. This is useful | |
1999 | for compile time checking, especially together with @code{__builtin_constant_p} | |
2000 | and inline functions. While it is possible to define the function with | |
2001 | a message in @code{.gnu.warning*} section, when using this attribute the problem | |
2002 | will be diagnosed earlier and with exact location of the call even in presence | |
2003 | of inline functions or when not emitting debugging information. | |
2004 | ||
c8619b90 NS |
2005 | @item cdecl |
2006 | @cindex functions that do pop the argument stack on the 386 | |
2007 | @opindex mrtd | |
2008 | On the Intel 386, the @code{cdecl} attribute causes the compiler to | |
2009 | assume that the calling function will pop off the stack space used to | |
2010 | pass arguments. This is | |
2011 | useful to override the effects of the @option{-mrtd} switch. | |
2a8f6b90 | 2012 | |
2a8f6b90 | 2013 | @item const |
c8619b90 | 2014 | @cindex @code{const} function attribute |
2a8f6b90 JH |
2015 | Many functions do not examine any values except their arguments, and |
2016 | have no effects except the return value. Basically this is just slightly | |
50c177f7 | 2017 | more strict class than the @code{pure} attribute below, since function is not |
2a59078d | 2018 | allowed to read global memory. |
2a8f6b90 JH |
2019 | |
2020 | @cindex pointer arguments | |
2021 | Note that a function that has pointer arguments and examines the data | |
2022 | pointed to must @emph{not} be declared @code{const}. Likewise, a | |
2023 | function that calls a non-@code{const} function usually must not be | |
2024 | @code{const}. It does not make sense for a @code{const} function to | |
2025 | return @code{void}. | |
2026 | ||
f0523f02 | 2027 | The attribute @code{const} is not implemented in GCC versions earlier |
c1f7febf RK |
2028 | than 2.5. An alternative way to declare that a function has no side |
2029 | effects, which works in the current version and in some older versions, | |
2030 | is as follows: | |
2031 | ||
2032 | @smallexample | |
2033 | typedef int intfn (); | |
2034 | ||
2035 | extern const intfn square; | |
2036 | @end smallexample | |
2037 | ||
2038 | This approach does not work in GNU C++ from 2.6.0 on, since the language | |
2039 | specifies that the @samp{const} must be attached to the return value. | |
2040 | ||
c8619b90 NS |
2041 | @item constructor |
2042 | @itemx destructor | |
fc8600f9 MM |
2043 | @itemx constructor (@var{priority}) |
2044 | @itemx destructor (@var{priority}) | |
c8619b90 NS |
2045 | @cindex @code{constructor} function attribute |
2046 | @cindex @code{destructor} function attribute | |
2047 | The @code{constructor} attribute causes the function to be called | |
2048 | automatically before execution enters @code{main ()}. Similarly, the | |
2049 | @code{destructor} attribute causes the function to be called | |
2050 | automatically after @code{main ()} has completed or @code{exit ()} has | |
2051 | been called. Functions with these attributes are useful for | |
2052 | initializing data that will be used implicitly during the execution of | |
2053 | the program. | |
2054 | ||
fc8600f9 MM |
2055 | You may provide an optional integer priority to control the order in |
2056 | which constructor and destructor functions are run. A constructor | |
2057 | with a smaller priority number runs before a constructor with a larger | |
2058 | priority number; the opposite relationship holds for destructors. So, | |
2059 | if you have a constructor that allocates a resource and a destructor | |
2060 | that deallocates the same resource, both functions typically have the | |
2061 | same priority. The priorities for constructor and destructor | |
2062 | functions are the same as those specified for namespace-scope C++ | |
2063 | objects (@pxref{C++ Attributes}). | |
2064 | ||
c8619b90 NS |
2065 | These attributes are not currently implemented for Objective-C@. |
2066 | ||
2067 | @item deprecated | |
9b86d6bb | 2068 | @itemx deprecated (@var{msg}) |
c8619b90 NS |
2069 | @cindex @code{deprecated} attribute. |
2070 | The @code{deprecated} attribute results in a warning if the function | |
2071 | is used anywhere in the source file. This is useful when identifying | |
2072 | functions that are expected to be removed in a future version of a | |
2073 | program. The warning also includes the location of the declaration | |
2074 | of the deprecated function, to enable users to easily find further | |
2075 | information about why the function is deprecated, or what they should | |
2076 | do instead. Note that the warnings only occurs for uses: | |
2077 | ||
2078 | @smallexample | |
2079 | int old_fn () __attribute__ ((deprecated)); | |
2080 | int old_fn (); | |
2081 | int (*fn_ptr)() = old_fn; | |
2082 | @end smallexample | |
2083 | ||
9b86d6bb L |
2084 | results in a warning on line 3 but not line 2. The optional msg |
2085 | argument, which must be a string, will be printed in the warning if | |
2086 | present. | |
c8619b90 NS |
2087 | |
2088 | The @code{deprecated} attribute can also be used for variables and | |
2089 | types (@pxref{Variable Attributes}, @pxref{Type Attributes}.) | |
2090 | ||
2091 | @item dllexport | |
2092 | @cindex @code{__declspec(dllexport)} | |
b2ca3702 MM |
2093 | On Microsoft Windows targets and Symbian OS targets the |
2094 | @code{dllexport} attribute causes the compiler to provide a global | |
2095 | pointer to a pointer in a DLL, so that it can be referenced with the | |
2096 | @code{dllimport} attribute. On Microsoft Windows targets, the pointer | |
2097 | name is formed by combining @code{_imp__} and the function or variable | |
2098 | name. | |
2099 | ||
2100 | You can use @code{__declspec(dllexport)} as a synonym for | |
2101 | @code{__attribute__ ((dllexport))} for compatibility with other | |
2102 | compilers. | |
2103 | ||
2104 | On systems that support the @code{visibility} attribute, this | |
3a687f8b MM |
2105 | attribute also implies ``default'' visibility. It is an error to |
2106 | explicitly specify any other visibility. | |
c8619b90 | 2107 | |
b2ca3702 MM |
2108 | Currently, the @code{dllexport} attribute is ignored for inlined |
2109 | functions, unless the @option{-fkeep-inline-functions} flag has been | |
2110 | used. The attribute is also ignored for undefined symbols. | |
c8619b90 | 2111 | |
8a36672b JM |
2112 | When applied to C++ classes, the attribute marks defined non-inlined |
2113 | member functions and static data members as exports. Static consts | |
c8619b90 NS |
2114 | initialized in-class are not marked unless they are also defined |
2115 | out-of-class. | |
2116 | ||
b55e3aad | 2117 | For Microsoft Windows targets there are alternative methods for |
b2ca3702 | 2118 | including the symbol in the DLL's export table such as using a |
b55e3aad NC |
2119 | @file{.def} file with an @code{EXPORTS} section or, with GNU ld, using |
2120 | the @option{--export-all} linker flag. | |
c8619b90 NS |
2121 | |
2122 | @item dllimport | |
2123 | @cindex @code{__declspec(dllimport)} | |
b2ca3702 | 2124 | On Microsoft Windows and Symbian OS targets, the @code{dllimport} |
b55e3aad | 2125 | attribute causes the compiler to reference a function or variable via |
b2ca3702 | 2126 | a global pointer to a pointer that is set up by the DLL exporting the |
3a687f8b MM |
2127 | symbol. The attribute implies @code{extern}. On Microsoft Windows |
2128 | targets, the pointer name is formed by combining @code{_imp__} and the | |
2129 | function or variable name. | |
b2ca3702 MM |
2130 | |
2131 | You can use @code{__declspec(dllimport)} as a synonym for | |
2132 | @code{__attribute__ ((dllimport))} for compatibility with other | |
2133 | compilers. | |
c8619b90 | 2134 | |
3a687f8b MM |
2135 | On systems that support the @code{visibility} attribute, this |
2136 | attribute also implies ``default'' visibility. It is an error to | |
2137 | explicitly specify any other visibility. | |
2138 | ||
8a36672b | 2139 | Currently, the attribute is ignored for inlined functions. If the |
c8619b90 NS |
2140 | attribute is applied to a symbol @emph{definition}, an error is reported. |
2141 | If a symbol previously declared @code{dllimport} is later defined, the | |
2142 | attribute is ignored in subsequent references, and a warning is emitted. | |
2143 | The attribute is also overridden by a subsequent declaration as | |
2144 | @code{dllexport}. | |
2145 | ||
2146 | When applied to C++ classes, the attribute marks non-inlined | |
2147 | member functions and static data members as imports. However, the | |
2148 | attribute is ignored for virtual methods to allow creation of vtables | |
2149 | using thunks. | |
2150 | ||
b2ca3702 | 2151 | On the SH Symbian OS target the @code{dllimport} attribute also has |
78466c0e | 2152 | another affect---it can cause the vtable and run-time type information |
b2ca3702 MM |
2153 | for a class to be exported. This happens when the class has a |
2154 | dllimport'ed constructor or a non-inline, non-pure virtual function | |
e4ae5e77 | 2155 | and, for either of those two conditions, the class also has an inline |
b2ca3702 MM |
2156 | constructor or destructor and has a key function that is defined in |
2157 | the current translation unit. | |
b55e3aad NC |
2158 | |
2159 | For Microsoft Windows based targets the use of the @code{dllimport} | |
2160 | attribute on functions is not necessary, but provides a small | |
8a36672b | 2161 | performance benefit by eliminating a thunk in the DLL@. The use of the |
b55e3aad | 2162 | @code{dllimport} attribute on imported variables was required on older |
b2ca3702 | 2163 | versions of the GNU linker, but can now be avoided by passing the |
8a36672b | 2164 | @option{--enable-auto-import} switch to the GNU linker. As with |
b2ca3702 | 2165 | functions, using the attribute for a variable eliminates a thunk in |
8a36672b | 2166 | the DLL@. |
b2ca3702 | 2167 | |
d32034a7 DS |
2168 | One drawback to using this attribute is that a pointer to a |
2169 | @emph{variable} marked as @code{dllimport} cannot be used as a constant | |
2170 | address. However, a pointer to a @emph{function} with the | |
2171 | @code{dllimport} attribute can be used as a constant initializer; in | |
2172 | this case, the address of a stub function in the import lib is | |
2173 | referenced. On Microsoft Windows targets, the attribute can be disabled | |
b2ca3702 | 2174 | for functions by setting the @option{-mnop-fun-dllimport} flag. |
c8619b90 NS |
2175 | |
2176 | @item eightbit_data | |
2177 | @cindex eight bit data on the H8/300, H8/300H, and H8S | |
2178 | Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified | |
2179 | variable should be placed into the eight bit data section. | |
2180 | The compiler will generate more efficient code for certain operations | |
2181 | on data in the eight bit data area. Note the eight bit data area is limited to | |
2182 | 256 bytes of data. | |
2183 | ||
2184 | You must use GAS and GLD from GNU binutils version 2.7 or later for | |
2185 | this attribute to work correctly. | |
2186 | ||
0d4a78eb BS |
2187 | @item exception_handler |
2188 | @cindex exception handler functions on the Blackfin processor | |
2189 | Use this attribute on the Blackfin to indicate that the specified function | |
2190 | is an exception handler. The compiler will generate function entry and | |
2191 | exit sequences suitable for use in an exception handler when this | |
2192 | attribute is present. | |
2193 | ||
1df48f5c JW |
2194 | @item externally_visible |
2195 | @cindex @code{externally_visible} attribute. | |
2196 | This attribute, attached to a global variable or function, nullifies | |
2197 | the effect of the @option{-fwhole-program} command-line option, so the | |
2198 | object remains visible outside the current compilation unit. | |
2199 | ||
c8619b90 NS |
2200 | @item far |
2201 | @cindex functions which handle memory bank switching | |
2202 | On 68HC11 and 68HC12 the @code{far} attribute causes the compiler to | |
2203 | use a calling convention that takes care of switching memory banks when | |
2204 | entering and leaving a function. This calling convention is also the | |
2205 | default when using the @option{-mlong-calls} option. | |
2206 | ||
2207 | On 68HC12 the compiler will use the @code{call} and @code{rtc} instructions | |
2208 | to call and return from a function. | |
2209 | ||
2210 | On 68HC11 the compiler will generate a sequence of instructions | |
2211 | to invoke a board-specific routine to switch the memory bank and call the | |
8a36672b | 2212 | real function. The board-specific routine simulates a @code{call}. |
c8619b90 | 2213 | At the end of a function, it will jump to a board-specific routine |
8a36672b | 2214 | instead of using @code{rts}. The board-specific return routine simulates |
c8619b90 NS |
2215 | the @code{rtc}. |
2216 | ||
2217 | @item fastcall | |
2218 | @cindex functions that pop the argument stack on the 386 | |
2219 | On the Intel 386, the @code{fastcall} attribute causes the compiler to | |
2f84b963 RG |
2220 | pass the first argument (if of integral type) in the register ECX and |
2221 | the second argument (if of integral type) in the register EDX@. Subsequent | |
2222 | and other typed arguments are passed on the stack. The called function will | |
2223 | pop the arguments off the stack. If the number of arguments is variable all | |
c8619b90 | 2224 | arguments are pushed on the stack. |
c1f7febf RK |
2225 | |
2226 | @item format (@var{archetype}, @var{string-index}, @var{first-to-check}) | |
2227 | @cindex @code{format} function attribute | |
84330467 | 2228 | @opindex Wformat |
bb72a084 | 2229 | The @code{format} attribute specifies that a function takes @code{printf}, |
26f6672d JM |
2230 | @code{scanf}, @code{strftime} or @code{strfmon} style arguments which |
2231 | should be type-checked against a format string. For example, the | |
2232 | declaration: | |
c1f7febf RK |
2233 | |
2234 | @smallexample | |
2235 | extern int | |
2236 | my_printf (void *my_object, const char *my_format, ...) | |
2237 | __attribute__ ((format (printf, 2, 3))); | |
2238 | @end smallexample | |
2239 | ||
2240 | @noindent | |
2241 | causes the compiler to check the arguments in calls to @code{my_printf} | |
2242 | for consistency with the @code{printf} style format string argument | |
2243 | @code{my_format}. | |
2244 | ||
2245 | The parameter @var{archetype} determines how the format string is | |
6590fc9f KT |
2246 | interpreted, and should be @code{printf}, @code{scanf}, @code{strftime}, |
2247 | @code{gnu_printf}, @code{gnu_scanf}, @code{gnu_strftime} or | |
2248 | @code{strfmon}. (You can also use @code{__printf__}, | |
2249 | @code{__scanf__}, @code{__strftime__} or @code{__strfmon__}.) On | |
2250 | MinGW targets, @code{ms_printf}, @code{ms_scanf}, and | |
2251 | @code{ms_strftime} are also present. | |
2252 | @var{archtype} values such as @code{printf} refer to the formats accepted | |
2253 | by the system's C run-time library, while @code{gnu_} values always refer | |
2254 | to the formats accepted by the GNU C Library. On Microsoft Windows | |
2255 | targets, @code{ms_} values refer to the formats accepted by the | |
2256 | @file{msvcrt.dll} library. | |
2257 | The parameter @var{string-index} | |
2258 | specifies which argument is the format string argument (starting | |
2259 | from 1), while @var{first-to-check} is the number of the first | |
2260 | argument to check against the format string. For functions | |
2261 | where the arguments are not available to be checked (such as | |
c1f7febf | 2262 | @code{vprintf}), specify the third parameter as zero. In this case the |
b722c82c JM |
2263 | compiler only checks the format string for consistency. For |
2264 | @code{strftime} formats, the third parameter is required to be zero. | |
f57a2e3a BE |
2265 | Since non-static C++ methods have an implicit @code{this} argument, the |
2266 | arguments of such methods should be counted from two, not one, when | |
2267 | giving values for @var{string-index} and @var{first-to-check}. | |
c1f7febf RK |
2268 | |
2269 | In the example above, the format string (@code{my_format}) is the second | |
2270 | argument of the function @code{my_print}, and the arguments to check | |
2271 | start with the third argument, so the correct parameters for the format | |
2272 | attribute are 2 and 3. | |
2273 | ||
84330467 | 2274 | @opindex ffreestanding |
e6e931b7 | 2275 | @opindex fno-builtin |
c1f7febf | 2276 | The @code{format} attribute allows you to identify your own functions |
f0523f02 | 2277 | which take format strings as arguments, so that GCC can check the |
b722c82c | 2278 | calls to these functions for errors. The compiler always (unless |
e6e931b7 | 2279 | @option{-ffreestanding} or @option{-fno-builtin} is used) checks formats |
b722c82c | 2280 | for the standard library functions @code{printf}, @code{fprintf}, |
bb72a084 | 2281 | @code{sprintf}, @code{scanf}, @code{fscanf}, @code{sscanf}, @code{strftime}, |
c1f7febf | 2282 | @code{vprintf}, @code{vfprintf} and @code{vsprintf} whenever such |
84330467 | 2283 | warnings are requested (using @option{-Wformat}), so there is no need to |
b722c82c JM |
2284 | modify the header file @file{stdio.h}. In C99 mode, the functions |
2285 | @code{snprintf}, @code{vsnprintf}, @code{vscanf}, @code{vfscanf} and | |
26f6672d | 2286 | @code{vsscanf} are also checked. Except in strictly conforming C |
b4c984fb KG |
2287 | standard modes, the X/Open function @code{strfmon} is also checked as |
2288 | are @code{printf_unlocked} and @code{fprintf_unlocked}. | |
b722c82c | 2289 | @xref{C Dialect Options,,Options Controlling C Dialect}. |
c1f7febf | 2290 | |
a2bec818 DJ |
2291 | The target may provide additional types of format checks. |
2292 | @xref{Target Format Checks,,Format Checks Specific to Particular | |
2293 | Target Machines}. | |
2294 | ||
c1f7febf RK |
2295 | @item format_arg (@var{string-index}) |
2296 | @cindex @code{format_arg} function attribute | |
84330467 | 2297 | @opindex Wformat-nonliteral |
26f6672d JM |
2298 | The @code{format_arg} attribute specifies that a function takes a format |
2299 | string for a @code{printf}, @code{scanf}, @code{strftime} or | |
2300 | @code{strfmon} style function and modifies it (for example, to translate | |
2301 | it into another language), so the result can be passed to a | |
2302 | @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} style | |
2303 | function (with the remaining arguments to the format function the same | |
2304 | as they would have been for the unmodified string). For example, the | |
2305 | declaration: | |
c1f7febf RK |
2306 | |
2307 | @smallexample | |
2308 | extern char * | |
2309 | my_dgettext (char *my_domain, const char *my_format) | |
2310 | __attribute__ ((format_arg (2))); | |
2311 | @end smallexample | |
2312 | ||
2313 | @noindent | |
26f6672d JM |
2314 | causes the compiler to check the arguments in calls to a @code{printf}, |
2315 | @code{scanf}, @code{strftime} or @code{strfmon} type function, whose | |
2316 | format string argument is a call to the @code{my_dgettext} function, for | |
2317 | consistency with the format string argument @code{my_format}. If the | |
2318 | @code{format_arg} attribute had not been specified, all the compiler | |
2319 | could tell in such calls to format functions would be that the format | |
2320 | string argument is not constant; this would generate a warning when | |
84330467 | 2321 | @option{-Wformat-nonliteral} is used, but the calls could not be checked |
26f6672d | 2322 | without the attribute. |
c1f7febf RK |
2323 | |
2324 | The parameter @var{string-index} specifies which argument is the format | |
f57a2e3a BE |
2325 | string argument (starting from one). Since non-static C++ methods have |
2326 | an implicit @code{this} argument, the arguments of such methods should | |
2327 | be counted from two. | |
c1f7febf RK |
2328 | |
2329 | The @code{format-arg} attribute allows you to identify your own | |
f0523f02 | 2330 | functions which modify format strings, so that GCC can check the |
26f6672d JM |
2331 | calls to @code{printf}, @code{scanf}, @code{strftime} or @code{strfmon} |
2332 | type function whose operands are a call to one of your own function. | |
2333 | The compiler always treats @code{gettext}, @code{dgettext}, and | |
2334 | @code{dcgettext} in this manner except when strict ISO C support is | |
84330467 | 2335 | requested by @option{-ansi} or an appropriate @option{-std} option, or |
e6e931b7 JM |
2336 | @option{-ffreestanding} or @option{-fno-builtin} |
2337 | is used. @xref{C Dialect Options,,Options | |
26f6672d | 2338 | Controlling C Dialect}. |
c1f7febf | 2339 | |
c8619b90 | 2340 | @item function_vector |
561642fa | 2341 | @cindex calling functions through the function vector on H8/300, M16C, M32C and SH2A processors |
c8619b90 NS |
2342 | Use this attribute on the H8/300, H8/300H, and H8S to indicate that the specified |
2343 | function should be called through the function vector. Calling a | |
2344 | function through the function vector will reduce code size, however; | |
2345 | the function vector has a limited size (maximum 128 entries on the H8/300 | |
2346 | and 64 entries on the H8/300H and H8S) and shares space with the interrupt vector. | |
b34c7881 | 2347 | |
561642fa AP |
2348 | In SH2A target, this attribute declares a function to be called using the |
2349 | TBR relative addressing mode. The argument to this attribute is the entry | |
2350 | number of the same function in a vector table containing all the TBR | |
2351 | relative addressable functions. For the successful jump, register TBR | |
2352 | should contain the start address of this TBR relative vector table. | |
2353 | In the startup routine of the user application, user needs to care of this | |
2354 | TBR register initialization. The TBR relative vector table can have at | |
2355 | max 256 function entries. The jumps to these functions will be generated | |
2356 | using a SH2A specific, non delayed branch instruction JSR/N @@(disp8,TBR). | |
c8619b90 NS |
2357 | You must use GAS and GLD from GNU binutils version 2.7 or later for |
2358 | this attribute to work correctly. | |
b34c7881 | 2359 | |
561642fa AP |
2360 | Please refer the example of M16C target, to see the use of this |
2361 | attribute while declaring a function, | |
2362 | ||
2363 | In an application, for a function being called once, this attribute will | |
2364 | save at least 8 bytes of code; and if other successive calls are being | |
2365 | made to the same function, it will save 2 bytes of code per each of these | |
2366 | calls. | |
2367 | ||
5abd2125 JS |
2368 | On M16C/M32C targets, the @code{function_vector} attribute declares a |
2369 | special page subroutine call function. Use of this attribute reduces | |
2370 | the code size by 2 bytes for each call generated to the | |
2371 | subroutine. The argument to the attribute is the vector number entry | |
2372 | from the special page vector table which contains the 16 low-order | |
2373 | bits of the subroutine's entry address. Each vector table has special | |
2374 | page number (18 to 255) which are used in @code{jsrs} instruction. | |
2375 | Jump addresses of the routines are generated by adding 0x0F0000 (in | |
2376 | case of M16C targets) or 0xFF0000 (in case of M32C targets), to the 2 | |
2377 | byte addresses set in the vector table. Therefore you need to ensure | |
2378 | that all the special page vector routines should get mapped within the | |
2379 | address range 0x0F0000 to 0x0FFFFF (for M16C) and 0xFF0000 to 0xFFFFFF | |
2380 | (for M32C). | |
2381 | ||
2382 | In the following example 2 bytes will be saved for each call to | |
2383 | function @code{foo}. | |
2384 | ||
2385 | @smallexample | |
2386 | void foo (void) __attribute__((function_vector(0x18))); | |
2387 | void foo (void) | |
2388 | @{ | |
2389 | @} | |
2390 | ||
2391 | void bar (void) | |
2392 | @{ | |
2393 | foo(); | |
2394 | @} | |
2395 | @end smallexample | |
2396 | ||
2397 | If functions are defined in one file and are called in another file, | |
2398 | then be sure to write this declaration in both files. | |
2399 | ||
2400 | This attribute is ignored for R8C target. | |
2401 | ||
c8619b90 NS |
2402 | @item interrupt |
2403 | @cindex interrupt handler functions | |
e19da24c | 2404 | Use this attribute on the ARM, AVR, CRX, M32C, M32R/D, m68k, MIPS |
2bccb817 KH |
2405 | and Xstormy16 ports to indicate that the specified function is an |
2406 | interrupt handler. The compiler will generate function entry and exit | |
2407 | sequences suitable for use in an interrupt handler when this attribute | |
2408 | is present. | |
b34c7881 | 2409 | |
2bccb817 | 2410 | Note, interrupt handlers for the Blackfin, H8/300, H8/300H, H8S, and |
0d4a78eb | 2411 | SH processors can be specified via the @code{interrupt_handler} attribute. |
b34c7881 | 2412 | |
c8619b90 | 2413 | Note, on the AVR, interrupts will be enabled inside the function. |
9162542e | 2414 | |
c8619b90 NS |
2415 | Note, for the ARM, you can specify the kind of interrupt to be handled by |
2416 | adding an optional parameter to the interrupt attribute like this: | |
e23bd218 IR |
2417 | |
2418 | @smallexample | |
c8619b90 | 2419 | void f () __attribute__ ((interrupt ("IRQ"))); |
e23bd218 IR |
2420 | @end smallexample |
2421 | ||
c8619b90 | 2422 | Permissible values for this parameter are: IRQ, FIQ, SWI, ABORT and UNDEF@. |
e23bd218 | 2423 | |
7a085dce | 2424 | On ARMv7-M the interrupt type is ignored, and the attribute means the function |
5b3e6663 PB |
2425 | may be called with a word aligned stack pointer. |
2426 | ||
e19da24c CF |
2427 | On MIPS targets, you can use the following attributes to modify the behavior |
2428 | of an interrupt handler: | |
2429 | @table @code | |
2430 | @item use_shadow_register_set | |
2431 | @cindex @code{use_shadow_register_set} attribute | |
2432 | Assume that the handler uses a shadow register set, instead of | |
2433 | the main general-purpose registers. | |
2434 | ||
2435 | @item keep_interrupts_masked | |
2436 | @cindex @code{keep_interrupts_masked} attribute | |
2437 | Keep interrupts masked for the whole function. Without this attribute, | |
2438 | GCC tries to reenable interrupts for as much of the function as it can. | |
2439 | ||
2440 | @item use_debug_exception_return | |
2441 | @cindex @code{use_debug_exception_return} attribute | |
2442 | Return using the @code{deret} instruction. Interrupt handlers that don't | |
2443 | have this attribute return using @code{eret} instead. | |
2444 | @end table | |
2445 | ||
2446 | You can use any combination of these attributes, as shown below: | |
2447 | @smallexample | |
2448 | void __attribute__ ((interrupt)) v0 (); | |
2449 | void __attribute__ ((interrupt, use_shadow_register_set)) v1 (); | |
2450 | void __attribute__ ((interrupt, keep_interrupts_masked)) v2 (); | |
2451 | void __attribute__ ((interrupt, use_debug_exception_return)) v3 (); | |
2452 | void __attribute__ ((interrupt, use_shadow_register_set, | |
2453 | keep_interrupts_masked)) v4 (); | |
2454 | void __attribute__ ((interrupt, use_shadow_register_set, | |
2455 | use_debug_exception_return)) v5 (); | |
2456 | void __attribute__ ((interrupt, keep_interrupts_masked, | |
2457 | use_debug_exception_return)) v6 (); | |
2458 | void __attribute__ ((interrupt, use_shadow_register_set, | |
2459 | keep_interrupts_masked, | |
2460 | use_debug_exception_return)) v7 (); | |
2461 | @end smallexample | |
2462 | ||
c8619b90 | 2463 | @item interrupt_handler |
0d4a78eb BS |
2464 | @cindex interrupt handler functions on the Blackfin, m68k, H8/300 and SH processors |
2465 | Use this attribute on the Blackfin, m68k, H8/300, H8/300H, H8S, and SH to | |
2466 | indicate that the specified function is an interrupt handler. The compiler | |
2467 | will generate function entry and exit sequences suitable for use in an | |
2468 | interrupt handler when this attribute is present. | |
2469 | ||
a4242737 KH |
2470 | @item interrupt_thread |
2471 | @cindex interrupt thread functions on fido | |
2472 | Use this attribute on fido, a subarchitecture of the m68k, to indicate | |
2473 | that the specified function is an interrupt handler that is designed | |
2474 | to run as a thread. The compiler omits generate prologue/epilogue | |
2475 | sequences and replaces the return instruction with a @code{sleep} | |
2476 | instruction. This attribute is available only on fido. | |
2477 | ||
d8f8ca80 RR |
2478 | @item isr |
2479 | @cindex interrupt service routines on ARM | |
2480 | Use this attribute on ARM to write Interrupt Service Routines. This is an | |
2481 | alias to the @code{interrupt} attribute above. | |
2482 | ||
0d4a78eb BS |
2483 | @item kspisusp |
2484 | @cindex User stack pointer in interrupts on the Blackfin | |
2485 | When used together with @code{interrupt_handler}, @code{exception_handler} | |
2486 | or @code{nmi_handler}, code will be generated to load the stack pointer | |
2487 | from the USP register in the function prologue. | |
72954a4f | 2488 | |
4af797b5 JZ |
2489 | @item l1_text |
2490 | @cindex @code{l1_text} function attribute | |
2491 | This attribute specifies a function to be placed into L1 Instruction | |
0ee2ea09 | 2492 | SRAM@. The function will be put into a specific section named @code{.l1.text}. |
4af797b5 JZ |
2493 | With @option{-mfdpic}, function calls with a such function as the callee |
2494 | or caller will use inlined PLT. | |
2495 | ||
c8619b90 NS |
2496 | @item long_call/short_call |
2497 | @cindex indirect calls on ARM | |
2498 | This attribute specifies how a particular function is called on | |
2499 | ARM@. Both attributes override the @option{-mlong-calls} (@pxref{ARM Options}) | |
2500 | command line switch and @code{#pragma long_calls} settings. The | |
87c365a4 NS |
2501 | @code{long_call} attribute indicates that the function might be far |
2502 | away from the call site and require a different (more expensive) | |
2503 | calling sequence. The @code{short_call} attribute always places | |
c8619b90 NS |
2504 | the offset to the function from the call site into the @samp{BL} |
2505 | instruction directly. | |
72954a4f | 2506 | |
c8619b90 NS |
2507 | @item longcall/shortcall |
2508 | @cindex functions called via pointer on the RS/6000 and PowerPC | |
87c365a4 NS |
2509 | On the Blackfin, RS/6000 and PowerPC, the @code{longcall} attribute |
2510 | indicates that the function might be far away from the call site and | |
2511 | require a different (more expensive) calling sequence. The | |
2512 | @code{shortcall} attribute indicates that the function is always close | |
2513 | enough for the shorter calling sequence to be used. These attributes | |
2514 | override both the @option{-mlongcall} switch and, on the RS/6000 and | |
2515 | PowerPC, the @code{#pragma longcall} setting. | |
72954a4f | 2516 | |
c8619b90 NS |
2517 | @xref{RS/6000 and PowerPC Options}, for more information on whether long |
2518 | calls are necessary. | |
c1f7febf | 2519 | |
cd3a59b3 | 2520 | @item long_call/near/far |
4dbdb061 | 2521 | @cindex indirect calls on MIPS |
cd3a59b3 SL |
2522 | These attributes specify how a particular function is called on MIPS@. |
2523 | The attributes override the @option{-mlong-calls} (@pxref{MIPS Options}) | |
2524 | command-line switch. The @code{long_call} and @code{far} attributes are | |
2525 | synonyms, and cause the compiler to always call | |
4dbdb061 | 2526 | the function by first loading its address into a register, and then using |
cd3a59b3 SL |
2527 | the contents of that register. The @code{near} attribute has the opposite |
2528 | effect; it specifies that non-PIC calls should be made using the more | |
2529 | efficient @code{jal} instruction. | |
4dbdb061 | 2530 | |
140592a0 AG |
2531 | @item malloc |
2532 | @cindex @code{malloc} attribute | |
2533 | The @code{malloc} attribute is used to tell the compiler that a function | |
928a5ba9 JM |
2534 | may be treated as if any non-@code{NULL} pointer it returns cannot |
2535 | alias any other pointer valid when the function returns. | |
140592a0 | 2536 | This will often improve optimization. |
928a5ba9 JM |
2537 | Standard functions with this property include @code{malloc} and |
2538 | @code{calloc}. @code{realloc}-like functions have this property as | |
2539 | long as the old pointer is never referred to (including comparing it | |
2540 | to the new pointer) after the function returns a non-@code{NULL} | |
2541 | value. | |
140592a0 | 2542 | |
f9e4a411 SL |
2543 | @item mips16/nomips16 |
2544 | @cindex @code{mips16} attribute | |
2545 | @cindex @code{nomips16} attribute | |
2546 | ||
2547 | On MIPS targets, you can use the @code{mips16} and @code{nomips16} | |
2548 | function attributes to locally select or turn off MIPS16 code generation. | |
2549 | A function with the @code{mips16} attribute is emitted as MIPS16 code, | |
2550 | while MIPS16 code generation is disabled for functions with the | |
2551 | @code{nomips16} attribute. These attributes override the | |
2552 | @option{-mips16} and @option{-mno-mips16} options on the command line | |
2553 | (@pxref{MIPS Options}). | |
2554 | ||
2555 | When compiling files containing mixed MIPS16 and non-MIPS16 code, the | |
2556 | preprocessor symbol @code{__mips16} reflects the setting on the command line, | |
2557 | not that within individual functions. Mixed MIPS16 and non-MIPS16 code | |
2558 | may interact badly with some GCC extensions such as @code{__builtin_apply} | |
2559 | (@pxref{Constructing Calls}). | |
2560 | ||
c8619b90 NS |
2561 | @item model (@var{model-name}) |
2562 | @cindex function addressability on the M32R/D | |
2563 | @cindex variable addressability on the IA-64 | |
2564 | ||
2565 | On the M32R/D, use this attribute to set the addressability of an | |
2566 | object, and of the code generated for a function. The identifier | |
2567 | @var{model-name} is one of @code{small}, @code{medium}, or | |
2568 | @code{large}, representing each of the code models. | |
2569 | ||
2570 | Small model objects live in the lower 16MB of memory (so that their | |
2571 | addresses can be loaded with the @code{ld24} instruction), and are | |
2572 | callable with the @code{bl} instruction. | |
2573 | ||
2574 | Medium model objects may live anywhere in the 32-bit address space (the | |
2575 | compiler will generate @code{seth/add3} instructions to load their addresses), | |
2576 | and are callable with the @code{bl} instruction. | |
2577 | ||
2578 | Large model objects may live anywhere in the 32-bit address space (the | |
2579 | compiler will generate @code{seth/add3} instructions to load their addresses), | |
2580 | and may not be reachable with the @code{bl} instruction (the compiler will | |
2581 | generate the much slower @code{seth/add3/jl} instruction sequence). | |
2582 | ||
2583 | On IA-64, use this attribute to set the addressability of an object. | |
2584 | At present, the only supported identifier for @var{model-name} is | |
2585 | @code{small}, indicating addressability via ``small'' (22-bit) | |
2586 | addresses (so that their addresses can be loaded with the @code{addl} | |
2587 | instruction). Caveat: such addressing is by definition not position | |
2588 | independent and hence this attribute must not be used for objects | |
2589 | defined by shared libraries. | |
2590 | ||
7c800926 | 2591 | @item ms_abi/sysv_abi |
f9ac6b1e | 2592 | @cindex @code{ms_abi} attribute |
7c800926 KT |
2593 | @cindex @code{sysv_abi} attribute |
2594 | ||
cdcabece | 2595 | On 64-bit x86_64-*-* targets, you can use an ABI attribute to indicate |
7c800926 KT |
2596 | which calling convention should be used for a function. The @code{ms_abi} |
2597 | attribute tells the compiler to use the Microsoft ABI, while the | |
2598 | @code{sysv_abi} attribute tells the compiler to use the ABI used on | |
2599 | GNU/Linux and other systems. The default is to use the Microsoft ABI | |
2600 | when targeting Windows. On all other systems, the default is the AMD ABI. | |
2601 | ||
2602 | Note, This feature is currently sorried out for Windows targets trying to | |
2603 | ||
c8619b90 NS |
2604 | @item naked |
2605 | @cindex function without a prologue/epilogue code | |
8d8da227 | 2606 | Use this attribute on the ARM, AVR, IP2K and SPU ports to indicate that |
85d9c13c | 2607 | the specified function does not need prologue/epilogue sequences generated by |
007e61c2 PB |
2608 | the compiler. It is up to the programmer to provide these sequences. The |
2609 | only statements that can be safely included in naked functions are | |
2610 | @code{asm} statements that do not have operands. All other statements, | |
2611 | including declarations of local variables, @code{if} statements, and so | |
2612 | forth, should be avoided. Naked functions should be used to implement the | |
2613 | body of an assembly function, while allowing the compiler to construct | |
2614 | the requisite function declaration for the assembler. | |
c8619b90 NS |
2615 | |
2616 | @item near | |
2617 | @cindex functions which do not handle memory bank switching on 68HC11/68HC12 | |
2618 | On 68HC11 and 68HC12 the @code{near} attribute causes the compiler to | |
2619 | use the normal calling convention based on @code{jsr} and @code{rts}. | |
2620 | This attribute can be used to cancel the effect of the @option{-mlong-calls} | |
2621 | option. | |
2622 | ||
0d4a78eb BS |
2623 | @item nesting |
2624 | @cindex Allow nesting in an interrupt handler on the Blackfin processor. | |
2625 | Use this attribute together with @code{interrupt_handler}, | |
2626 | @code{exception_handler} or @code{nmi_handler} to indicate that the function | |
2627 | entry code should enable nested interrupts or exceptions. | |
2628 | ||
2629 | @item nmi_handler | |
2630 | @cindex NMI handler functions on the Blackfin processor | |
2631 | Use this attribute on the Blackfin to indicate that the specified function | |
2632 | is an NMI handler. The compiler will generate function entry and | |
2633 | exit sequences suitable for use in an NMI handler when this | |
2634 | attribute is present. | |
2635 | ||
c8619b90 NS |
2636 | @item no_instrument_function |
2637 | @cindex @code{no_instrument_function} function attribute | |
2638 | @opindex finstrument-functions | |
2639 | If @option{-finstrument-functions} is given, profiling function calls will | |
2640 | be generated at entry and exit of most user-compiled functions. | |
2641 | Functions with this attribute will not be so instrumented. | |
2642 | ||
2643 | @item noinline | |
2644 | @cindex @code{noinline} function attribute | |
2645 | This function attribute prevents a function from being considered for | |
2646 | inlining. | |
ccd2a21e HPN |
2647 | @c Don't enumerate the optimizations by name here; we try to be |
2648 | @c future-compatible with this mechanism. | |
2649 | If the function does not have side-effects, there are optimizations | |
2650 | other than inlining that causes function calls to be optimized away, | |
2651 | although the function call is live. To keep such calls from being | |
2652 | optimized away, put | |
2653 | @smallexample | |
2654 | asm (""); | |
2655 | @end smallexample | |
2656 | (@pxref{Extended Asm}) in the called function, to serve as a special | |
2657 | side-effect. | |
c8619b90 NS |
2658 | |
2659 | @item nonnull (@var{arg-index}, @dots{}) | |
2660 | @cindex @code{nonnull} function attribute | |
2661 | The @code{nonnull} attribute specifies that some function parameters should | |
2662 | be non-null pointers. For instance, the declaration: | |
c1f7febf RK |
2663 | |
2664 | @smallexample | |
c8619b90 NS |
2665 | extern void * |
2666 | my_memcpy (void *dest, const void *src, size_t len) | |
6ccde948 | 2667 | __attribute__((nonnull (1, 2))); |
c1f7febf RK |
2668 | @end smallexample |
2669 | ||
c8619b90 NS |
2670 | @noindent |
2671 | causes the compiler to check that, in calls to @code{my_memcpy}, | |
2672 | arguments @var{dest} and @var{src} are non-null. If the compiler | |
2673 | determines that a null pointer is passed in an argument slot marked | |
2674 | as non-null, and the @option{-Wnonnull} option is enabled, a warning | |
2675 | is issued. The compiler may also choose to make optimizations based | |
2676 | on the knowledge that certain function arguments will not be null. | |
af3e86c2 | 2677 | |
c8619b90 NS |
2678 | If no argument index list is given to the @code{nonnull} attribute, |
2679 | all pointer arguments are marked as non-null. To illustrate, the | |
2680 | following declaration is equivalent to the previous example: | |
47bd70b5 JJ |
2681 | |
2682 | @smallexample | |
c8619b90 NS |
2683 | extern void * |
2684 | my_memcpy (void *dest, const void *src, size_t len) | |
6ccde948 | 2685 | __attribute__((nonnull)); |
47bd70b5 JJ |
2686 | @end smallexample |
2687 | ||
c8619b90 NS |
2688 | @item noreturn |
2689 | @cindex @code{noreturn} function attribute | |
2690 | A few standard library functions, such as @code{abort} and @code{exit}, | |
2691 | cannot return. GCC knows this automatically. Some programs define | |
2692 | their own functions that never return. You can declare them | |
2693 | @code{noreturn} to tell the compiler this fact. For example, | |
9e8aab55 | 2694 | |
c8619b90 NS |
2695 | @smallexample |
2696 | @group | |
2697 | void fatal () __attribute__ ((noreturn)); | |
d5c4db17 | 2698 | |
c8619b90 NS |
2699 | void |
2700 | fatal (/* @r{@dots{}} */) | |
2701 | @{ | |
2702 | /* @r{@dots{}} */ /* @r{Print error message.} */ /* @r{@dots{}} */ | |
2703 | exit (1); | |
2704 | @} | |
2705 | @end group | |
2706 | @end smallexample | |
9e8aab55 | 2707 | |
c8619b90 NS |
2708 | The @code{noreturn} keyword tells the compiler to assume that |
2709 | @code{fatal} cannot return. It can then optimize without regard to what | |
2710 | would happen if @code{fatal} ever did return. This makes slightly | |
2711 | better code. More importantly, it helps avoid spurious warnings of | |
2712 | uninitialized variables. | |
9e8aab55 | 2713 | |
c8619b90 NS |
2714 | The @code{noreturn} keyword does not affect the exceptional path when that |
2715 | applies: a @code{noreturn}-marked function may still return to the caller | |
2e9522f4 | 2716 | by throwing an exception or calling @code{longjmp}. |
9e8aab55 | 2717 | |
c8619b90 NS |
2718 | Do not assume that registers saved by the calling function are |
2719 | restored before calling the @code{noreturn} function. | |
47bd70b5 | 2720 | |
c8619b90 NS |
2721 | It does not make sense for a @code{noreturn} function to have a return |
2722 | type other than @code{void}. | |
c1f7febf | 2723 | |
c8619b90 NS |
2724 | The attribute @code{noreturn} is not implemented in GCC versions |
2725 | earlier than 2.5. An alternative way to declare that a function does | |
2726 | not return, which works in the current version and in some older | |
2727 | versions, is as follows: | |
5d34c8e9 | 2728 | |
c8619b90 NS |
2729 | @smallexample |
2730 | typedef void voidfn (); | |
c1f7febf | 2731 | |
c8619b90 NS |
2732 | volatile voidfn fatal; |
2733 | @end smallexample | |
e91f04de | 2734 | |
a1e73046 PC |
2735 | This approach does not work in GNU C++. |
2736 | ||
c8619b90 NS |
2737 | @item nothrow |
2738 | @cindex @code{nothrow} function attribute | |
2739 | The @code{nothrow} attribute is used to inform the compiler that a | |
2740 | function cannot throw an exception. For example, most functions in | |
2741 | the standard C library can be guaranteed not to throw an exception | |
2742 | with the notable exceptions of @code{qsort} and @code{bsearch} that | |
2743 | take function pointer arguments. The @code{nothrow} attribute is not | |
3f3174b6 | 2744 | implemented in GCC versions earlier than 3.3. |
c1f7febf | 2745 | |
ab442df7 MM |
2746 | @item optimize |
2747 | @cindex @code{optimize} function attribute | |
2748 | The @code{optimize} attribute is used to specify that a function is to | |
2749 | be compiled with different optimization options than specified on the | |
2750 | command line. Arguments can either be numbers or strings. Numbers | |
2751 | are assumed to be an optimization level. Strings that begin with | |
2752 | @code{O} are assumed to be an optimization option, while other options | |
2753 | are assumed to be used with a @code{-f} prefix. You can also use the | |
2754 | @samp{#pragma GCC optimize} pragma to set the optimization options | |
2755 | that affect more than one function. | |
2756 | @xref{Function Specific Option Pragmas}, for details about the | |
e3606f3b | 2757 | @samp{#pragma GCC optimize} pragma. |
ab442df7 MM |
2758 | |
2759 | This can be used for instance to have frequently executed functions | |
2760 | compiled with more aggressive optimization options that produce faster | |
2761 | and larger code, while other functions can be called with less | |
5779e713 | 2762 | aggressive options. |
ab442df7 | 2763 | |
c8619b90 NS |
2764 | @item pure |
2765 | @cindex @code{pure} function attribute | |
2766 | Many functions have no effects except the return value and their | |
2767 | return value depends only on the parameters and/or global variables. | |
2768 | Such a function can be subject | |
2769 | to common subexpression elimination and loop optimization just as an | |
2770 | arithmetic operator would be. These functions should be declared | |
2771 | with the attribute @code{pure}. For example, | |
a5c76ee6 | 2772 | |
c8619b90 NS |
2773 | @smallexample |
2774 | int square (int) __attribute__ ((pure)); | |
2775 | @end smallexample | |
c1f7febf | 2776 | |
c8619b90 NS |
2777 | @noindent |
2778 | says that the hypothetical function @code{square} is safe to call | |
2779 | fewer times than the program says. | |
c27ba912 | 2780 | |
c8619b90 NS |
2781 | Some of common examples of pure functions are @code{strlen} or @code{memcmp}. |
2782 | Interesting non-pure functions are functions with infinite loops or those | |
2783 | depending on volatile memory or other system resource, that may change between | |
2784 | two consecutive calls (such as @code{feof} in a multithreading environment). | |
c1f7febf | 2785 | |
c8619b90 NS |
2786 | The attribute @code{pure} is not implemented in GCC versions earlier |
2787 | than 2.96. | |
c1f7febf | 2788 | |
52bf96d2 JH |
2789 | @item hot |
2790 | @cindex @code{hot} function attribute | |
2791 | The @code{hot} attribute is used to inform the compiler that a function is a | |
2792 | hot spot of the compiled program. The function is optimized more aggressively | |
2793 | and on many target it is placed into special subsection of the text section so | |
2794 | all hot functions appears close together improving locality. | |
2795 | ||
2796 | When profile feedback is available, via @option{-fprofile-use}, hot functions | |
2797 | are automatically detected and this attribute is ignored. | |
2798 | ||
ab442df7 MM |
2799 | The @code{hot} attribute is not implemented in GCC versions earlier |
2800 | than 4.3. | |
2801 | ||
52bf96d2 JH |
2802 | @item cold |
2803 | @cindex @code{cold} function attribute | |
2804 | The @code{cold} attribute is used to inform the compiler that a function is | |
2805 | unlikely executed. The function is optimized for size rather than speed and on | |
2806 | many targets it is placed into special subsection of the text section so all | |
2807 | cold functions appears close together improving code locality of non-cold parts | |
2808 | of program. The paths leading to call of cold functions within code are marked | |
44c7bd63 | 2809 | as unlikely by the branch prediction mechanism. It is thus useful to mark |
52bf96d2 JH |
2810 | functions used to handle unlikely conditions, such as @code{perror}, as cold to |
2811 | improve optimization of hot functions that do call marked functions in rare | |
2812 | occasions. | |
2813 | ||
2814 | When profile feedback is available, via @option{-fprofile-use}, hot functions | |
2815 | are automatically detected and this attribute is ignored. | |
2816 | ||
ab442df7 MM |
2817 | The @code{cold} attribute is not implemented in GCC versions earlier than 4.3. |
2818 | ||
c8619b90 NS |
2819 | @item regparm (@var{number}) |
2820 | @cindex @code{regparm} attribute | |
2821 | @cindex functions that are passed arguments in registers on the 386 | |
2822 | On the Intel 386, the @code{regparm} attribute causes the compiler to | |
2f84b963 RG |
2823 | pass arguments number one to @var{number} if they are of integral type |
2824 | in registers EAX, EDX, and ECX instead of on the stack. Functions that | |
2825 | take a variable number of arguments will continue to be passed all of their | |
c8619b90 | 2826 | arguments on the stack. |
6d3d9133 | 2827 | |
c8619b90 NS |
2828 | Beware that on some ELF systems this attribute is unsuitable for |
2829 | global functions in shared libraries with lazy binding (which is the | |
2830 | default). Lazy binding will send the first call via resolving code in | |
2831 | the loader, which might assume EAX, EDX and ECX can be clobbered, as | |
2832 | per the standard calling conventions. Solaris 8 is affected by this. | |
2833 | GNU systems with GLIBC 2.1 or higher, and FreeBSD, are believed to be | |
21440b2d | 2834 | safe since the loaders there save EAX, EDX and ECX. (Lazy binding can be |
c8619b90 NS |
2835 | disabled with the linker or the loader if desired, to avoid the |
2836 | problem.) | |
6d3d9133 | 2837 | |
2f84b963 RG |
2838 | @item sseregparm |
2839 | @cindex @code{sseregparm} attribute | |
2840 | On the Intel 386 with SSE support, the @code{sseregparm} attribute | |
56829cae | 2841 | causes the compiler to pass up to 3 floating point arguments in |
2f84b963 RG |
2842 | SSE registers instead of on the stack. Functions that take a |
2843 | variable number of arguments will continue to pass all of their | |
2844 | floating point arguments on the stack. | |
2845 | ||
33932946 SH |
2846 | @item force_align_arg_pointer |
2847 | @cindex @code{force_align_arg_pointer} attribute | |
2848 | On the Intel x86, the @code{force_align_arg_pointer} attribute may be | |
2849 | applied to individual function definitions, generating an alternate | |
2e3f842f L |
2850 | prologue and epilogue that realigns the runtime stack if necessary. |
2851 | This supports mixing legacy codes that run with a 4-byte aligned stack | |
2852 | with modern codes that keep a 16-byte stack for SSE compatibility. | |
33932946 | 2853 | |
561642fa AP |
2854 | @item resbank |
2855 | @cindex @code{resbank} attribute | |
2856 | On the SH2A target, this attribute enables the high-speed register | |
2857 | saving and restoration using a register bank for @code{interrupt_handler} | |
a640c13b | 2858 | routines. Saving to the bank is performed automatically after the CPU |
561642fa AP |
2859 | accepts an interrupt that uses a register bank. |
2860 | ||
2861 | The nineteen 32-bit registers comprising general register R0 to R14, | |
2862 | control register GBR, and system registers MACH, MACL, and PR and the | |
2863 | vector table address offset are saved into a register bank. Register | |
2864 | banks are stacked in first-in last-out (FILO) sequence. Restoration | |
2865 | from the bank is executed by issuing a RESBANK instruction. | |
2866 | ||
6e9a3221 AN |
2867 | @item returns_twice |
2868 | @cindex @code{returns_twice} attribute | |
2869 | The @code{returns_twice} attribute tells the compiler that a function may | |
2870 | return more than one time. The compiler will ensure that all registers | |
2871 | are dead before calling such a function and will emit a warning about | |
2872 | the variables that may be clobbered after the second return from the | |
2873 | function. Examples of such functions are @code{setjmp} and @code{vfork}. | |
2874 | The @code{longjmp}-like counterpart of such function, if any, might need | |
2875 | to be marked with the @code{noreturn} attribute. | |
2876 | ||
c8619b90 | 2877 | @item saveall |
0d4a78eb BS |
2878 | @cindex save all registers on the Blackfin, H8/300, H8/300H, and H8S |
2879 | Use this attribute on the Blackfin, H8/300, H8/300H, and H8S to indicate that | |
c8619b90 NS |
2880 | all registers except the stack pointer should be saved in the prologue |
2881 | regardless of whether they are used or not. | |
6d3d9133 | 2882 | |
c8619b90 NS |
2883 | @item section ("@var{section-name}") |
2884 | @cindex @code{section} function attribute | |
2885 | Normally, the compiler places the code it generates in the @code{text} section. | |
2886 | Sometimes, however, you need additional sections, or you need certain | |
2887 | particular functions to appear in special sections. The @code{section} | |
2888 | attribute specifies that a function lives in a particular section. | |
2889 | For example, the declaration: | |
6d3d9133 NC |
2890 | |
2891 | @smallexample | |
c8619b90 | 2892 | extern void foobar (void) __attribute__ ((section ("bar"))); |
6d3d9133 NC |
2893 | @end smallexample |
2894 | ||
c8619b90 NS |
2895 | @noindent |
2896 | puts the function @code{foobar} in the @code{bar} section. | |
6d3d9133 | 2897 | |
c8619b90 NS |
2898 | Some file formats do not support arbitrary sections so the @code{section} |
2899 | attribute is not available on all platforms. | |
2900 | If you need to map the entire contents of a module to a particular | |
2901 | section, consider using the facilities of the linker instead. | |
2902 | ||
3d091dac KG |
2903 | @item sentinel |
2904 | @cindex @code{sentinel} function attribute | |
254986c7 KG |
2905 | This function attribute ensures that a parameter in a function call is |
2906 | an explicit @code{NULL}. The attribute is only valid on variadic | |
2907 | functions. By default, the sentinel is located at position zero, the | |
2908 | last parameter of the function call. If an optional integer position | |
2909 | argument P is supplied to the attribute, the sentinel must be located at | |
2910 | position P counting backwards from the end of the argument list. | |
2911 | ||
2912 | @smallexample | |
2913 | __attribute__ ((sentinel)) | |
2914 | is equivalent to | |
2915 | __attribute__ ((sentinel(0))) | |
2916 | @end smallexample | |
2917 | ||
2918 | The attribute is automatically set with a position of 0 for the built-in | |
2919 | functions @code{execl} and @code{execlp}. The built-in function | |
254ea84c | 2920 | @code{execle} has the attribute set with a position of 1. |
254986c7 KG |
2921 | |
2922 | A valid @code{NULL} in this context is defined as zero with any pointer | |
2923 | type. If your system defines the @code{NULL} macro with an integer type | |
2924 | then you need to add an explicit cast. GCC replaces @code{stddef.h} | |
2925 | with a copy that redefines NULL appropriately. | |
2926 | ||
2927 | The warnings for missing or incorrect sentinels are enabled with | |
2928 | @option{-Wformat}. | |
3d091dac | 2929 | |
c8619b90 NS |
2930 | @item short_call |
2931 | See long_call/short_call. | |
2932 | ||
2933 | @item shortcall | |
2934 | See longcall/shortcall. | |
2935 | ||
2936 | @item signal | |
2937 | @cindex signal handler functions on the AVR processors | |
2938 | Use this attribute on the AVR to indicate that the specified | |
2939 | function is a signal handler. The compiler will generate function | |
2940 | entry and exit sequences suitable for use in a signal handler when this | |
2941 | attribute is present. Interrupts will be disabled inside the function. | |
b93e3893 AO |
2942 | |
2943 | @item sp_switch | |
88ab0d1c | 2944 | Use this attribute on the SH to indicate an @code{interrupt_handler} |
b93e3893 AO |
2945 | function should switch to an alternate stack. It expects a string |
2946 | argument that names a global variable holding the address of the | |
2947 | alternate stack. | |
2948 | ||
2949 | @smallexample | |
2950 | void *alt_stack; | |
aee96fe9 JM |
2951 | void f () __attribute__ ((interrupt_handler, |
2952 | sp_switch ("alt_stack"))); | |
b93e3893 AO |
2953 | @end smallexample |
2954 | ||
c8619b90 NS |
2955 | @item stdcall |
2956 | @cindex functions that pop the argument stack on the 386 | |
2957 | On the Intel 386, the @code{stdcall} attribute causes the compiler to | |
2958 | assume that the called function will pop off the stack space used to | |
2959 | pass arguments, unless it takes a variable number of arguments. | |
c1f7febf | 2960 | |
4b84f3de SE |
2961 | @item syscall_linkage |
2962 | @cindex @code{syscall_linkage} attribute | |
2963 | This attribute is used to modify the IA64 calling convention by marking | |
2964 | all input registers as live at all function exits. This makes it possible | |
2965 | to restart a system call after an interrupt without having to save/restore | |
2966 | the input registers. This also prevents kernel data from leaking into | |
2967 | application code. | |
2968 | ||
1df48f5c JW |
2969 | @item target |
2970 | @cindex @code{target} function attribute | |
2971 | The @code{target} attribute is used to specify that a function is to | |
2972 | be compiled with different target options than specified on the | |
2973 | command line. This can be used for instance to have functions | |
2974 | compiled with a different ISA (instruction set architecture) than the | |
2975 | default. You can also use the @samp{#pragma GCC target} pragma to set | |
2976 | more than one function to be compiled with specific target options. | |
2977 | @xref{Function Specific Option Pragmas}, for details about the | |
2978 | @samp{#pragma GCC target} pragma. | |
2979 | ||
2980 | For instance on a 386, you could compile one function with | |
2981 | @code{target("sse4.1,arch=core2")} and another with | |
2982 | @code{target("sse4a,arch=amdfam10")} that would be equivalent to | |
2983 | compiling the first function with @option{-msse4.1} and | |
2984 | @option{-march=core2} options, and the second function with | |
2985 | @option{-msse4a} and @option{-march=amdfam10} options. It is up to the | |
2986 | user to make sure that a function is only invoked on a machine that | |
2987 | supports the particular ISA it was compiled for (for example by using | |
2988 | @code{cpuid} on 386 to determine what feature bits and architecture | |
2989 | family are used). | |
2990 | ||
2991 | @smallexample | |
2992 | int core2_func (void) __attribute__ ((__target__ ("arch=core2"))); | |
2993 | int sse3_func (void) __attribute__ ((__target__ ("sse3"))); | |
2994 | @end smallexample | |
2995 | ||
2996 | On the 386, the following options are allowed: | |
2997 | ||
2998 | @table @samp | |
2999 | @item abm | |
3000 | @itemx no-abm | |
3001 | @cindex @code{target("abm")} attribute | |
3002 | Enable/disable the generation of the advanced bit instructions. | |
3003 | ||
3004 | @item aes | |
3005 | @itemx no-aes | |
3006 | @cindex @code{target("aes")} attribute | |
3007 | Enable/disable the generation of the AES instructions. | |
3008 | ||
3009 | @item mmx | |
3010 | @itemx no-mmx | |
3011 | @cindex @code{target("mmx")} attribute | |
3012 | Enable/disable the generation of the MMX instructions. | |
3013 | ||
3014 | @item pclmul | |
3015 | @itemx no-pclmul | |
3016 | @cindex @code{target("pclmul")} attribute | |
3017 | Enable/disable the generation of the PCLMUL instructions. | |
3018 | ||
3019 | @item popcnt | |
3020 | @itemx no-popcnt | |
3021 | @cindex @code{target("popcnt")} attribute | |
3022 | Enable/disable the generation of the POPCNT instruction. | |
3023 | ||
3024 | @item sse | |
3025 | @itemx no-sse | |
3026 | @cindex @code{target("sse")} attribute | |
3027 | Enable/disable the generation of the SSE instructions. | |
3028 | ||
3029 | @item sse2 | |
3030 | @itemx no-sse2 | |
3031 | @cindex @code{target("sse2")} attribute | |
3032 | Enable/disable the generation of the SSE2 instructions. | |
3033 | ||
3034 | @item sse3 | |
3035 | @itemx no-sse3 | |
3036 | @cindex @code{target("sse3")} attribute | |
3037 | Enable/disable the generation of the SSE3 instructions. | |
3038 | ||
3039 | @item sse4 | |
3040 | @itemx no-sse4 | |
3041 | @cindex @code{target("sse4")} attribute | |
3042 | Enable/disable the generation of the SSE4 instructions (both SSE4.1 | |
3043 | and SSE4.2). | |
3044 | ||
3045 | @item sse4.1 | |
3046 | @itemx no-sse4.1 | |
3047 | @cindex @code{target("sse4.1")} attribute | |
3048 | Enable/disable the generation of the sse4.1 instructions. | |
3049 | ||
3050 | @item sse4.2 | |
3051 | @itemx no-sse4.2 | |
3052 | @cindex @code{target("sse4.2")} attribute | |
3053 | Enable/disable the generation of the sse4.2 instructions. | |
3054 | ||
3055 | @item sse4a | |
3056 | @itemx no-sse4a | |
3057 | @cindex @code{target("sse4a")} attribute | |
3058 | Enable/disable the generation of the SSE4A instructions. | |
3059 | ||
3060 | @item sse5 | |
3061 | @itemx no-sse5 | |
3062 | @cindex @code{target("sse5")} attribute | |
3063 | Enable/disable the generation of the SSE5 instructions. | |
3064 | ||
3065 | @item ssse3 | |
3066 | @itemx no-ssse3 | |
3067 | @cindex @code{target("ssse3")} attribute | |
3068 | Enable/disable the generation of the SSSE3 instructions. | |
3069 | ||
3070 | @item cld | |
3071 | @itemx no-cld | |
3072 | @cindex @code{target("cld")} attribute | |
3073 | Enable/disable the generation of the CLD before string moves. | |
3074 | ||
3075 | @item fancy-math-387 | |
3076 | @itemx no-fancy-math-387 | |
3077 | @cindex @code{target("fancy-math-387")} attribute | |
3078 | Enable/disable the generation of the @code{sin}, @code{cos}, and | |
3079 | @code{sqrt} instructions on the 387 floating point unit. | |
3080 | ||
3081 | @item fused-madd | |
3082 | @itemx no-fused-madd | |
3083 | @cindex @code{target("fused-madd")} attribute | |
3084 | Enable/disable the generation of the fused multiply/add instructions. | |
3085 | ||
3086 | @item ieee-fp | |
3087 | @itemx no-ieee-fp | |
3088 | @cindex @code{target("ieee-fp")} attribute | |
3089 | Enable/disable the generation of floating point that depends on IEEE arithmetic. | |
3090 | ||
3091 | @item inline-all-stringops | |
3092 | @itemx no-inline-all-stringops | |
3093 | @cindex @code{target("inline-all-stringops")} attribute | |
3094 | Enable/disable inlining of string operations. | |
3095 | ||
3096 | @item inline-stringops-dynamically | |
3097 | @itemx no-inline-stringops-dynamically | |
3098 | @cindex @code{target("inline-stringops-dynamically")} attribute | |
3099 | Enable/disable the generation of the inline code to do small string | |
3100 | operations and calling the library routines for large operations. | |
3101 | ||
3102 | @item align-stringops | |
3103 | @itemx no-align-stringops | |
3104 | @cindex @code{target("align-stringops")} attribute | |
3105 | Do/do not align destination of inlined string operations. | |
3106 | ||
3107 | @item recip | |
3108 | @itemx no-recip | |
3109 | @cindex @code{target("recip")} attribute | |
3110 | Enable/disable the generation of RCPSS, RCPPS, RSQRTSS and RSQRTPS | |
048fd785 | 3111 | instructions followed an additional Newton-Raphson step instead of |
1df48f5c JW |
3112 | doing a floating point division. |
3113 | ||
3114 | @item arch=@var{ARCH} | |
3115 | @cindex @code{target("arch=@var{ARCH}")} attribute | |
3116 | Specify the architecture to generate code for in compiling the function. | |
3117 | ||
3118 | @item tune=@var{TUNE} | |
3119 | @cindex @code{target("tune=@var{TUNE}")} attribute | |
3120 | Specify the architecture to tune for in compiling the function. | |
3121 | ||
3122 | @item fpmath=@var{FPMATH} | |
3123 | @cindex @code{target("fpmath=@var{FPMATH}")} attribute | |
3124 | Specify which floating point unit to use. The | |
3125 | @code{target("fpmath=sse,387")} option must be specified as | |
3126 | @code{target("fpmath=sse+387")} because the comma would separate | |
3127 | different options. | |
3128 | @end table | |
3129 | ||
3130 | On the 386, you can use either multiple strings to specify multiple | |
3131 | options, or you can separate the option with a comma (@code{,}). | |
3132 | ||
3133 | On the 386, the inliner will not inline a function that has different | |
3134 | target options than the caller, unless the callee has a subset of the | |
3135 | target options of the caller. For example a function declared with | |
3136 | @code{target("sse5")} can inline a function with | |
3137 | @code{target("sse2")}, since @code{-msse5} implies @code{-msse2}. | |
3138 | ||
3139 | The @code{target} attribute is not implemented in GCC versions earlier | |
3140 | than 4.4, and at present only the 386 uses it. | |
3141 | ||
c1f7febf | 3142 | @item tiny_data |
dbacaa98 KH |
3143 | @cindex tiny data section on the H8/300H and H8S |
3144 | Use this attribute on the H8/300H and H8S to indicate that the specified | |
c1f7febf RK |
3145 | variable should be placed into the tiny data section. |
3146 | The compiler will generate more efficient code for loads and stores | |
3147 | on data in the tiny data section. Note the tiny data area is limited to | |
3148 | slightly under 32kbytes of data. | |
845da534 | 3149 | |
c8619b90 NS |
3150 | @item trap_exit |
3151 | Use this attribute on the SH for an @code{interrupt_handler} to return using | |
3152 | @code{trapa} instead of @code{rte}. This attribute expects an integer | |
3153 | argument specifying the trap number to be used. | |
845da534 | 3154 | |
c8619b90 NS |
3155 | @item unused |
3156 | @cindex @code{unused} attribute. | |
3157 | This attribute, attached to a function, means that the function is meant | |
3158 | to be possibly unused. GCC will not produce a warning for this | |
3159 | function. | |
a32767e4 | 3160 | |
c8619b90 NS |
3161 | @item used |
3162 | @cindex @code{used} attribute. | |
3163 | This attribute, attached to a function, means that code must be emitted | |
3164 | for the function even if it appears that the function is not referenced. | |
3165 | This is useful, for example, when the function is referenced only in | |
3166 | inline assembly. | |
5936c7e7 | 3167 | |
812b587e | 3168 | @item version_id |
4b84f3de SE |
3169 | @cindex @code{version_id} attribute |
3170 | This IA64 HP-UX attribute, attached to a global variable or function, renames a | |
812b587e SE |
3171 | symbol to contain a version string, thus allowing for function level |
3172 | versioning. HP-UX system header files may use version level functioning | |
3173 | for some system calls. | |
3174 | ||
3175 | @smallexample | |
3176 | extern int foo () __attribute__((version_id ("20040821"))); | |
3177 | @end smallexample | |
3178 | ||
3179 | Calls to @var{foo} will be mapped to calls to @var{foo@{20040821@}}. | |
3180 | ||
c8619b90 NS |
3181 | @item visibility ("@var{visibility_type}") |
3182 | @cindex @code{visibility} attribute | |
46bdbc00 GK |
3183 | This attribute affects the linkage of the declaration to which it is attached. |
3184 | There are four supported @var{visibility_type} values: default, | |
3185 | hidden, protected or internal visibility. | |
5936c7e7 | 3186 | |
c8619b90 NS |
3187 | @smallexample |
3188 | void __attribute__ ((visibility ("protected"))) | |
3189 | f () @{ /* @r{Do something.} */; @} | |
3190 | int i __attribute__ ((visibility ("hidden"))); | |
3191 | @end smallexample | |
5936c7e7 | 3192 | |
46bdbc00 GK |
3193 | The possible values of @var{visibility_type} correspond to the |
3194 | visibility settings in the ELF gABI. | |
5936c7e7 | 3195 | |
c8619b90 | 3196 | @table @dfn |
63c5b495 | 3197 | @c keep this list of visibilities in alphabetical order. |
6b6cb52e | 3198 | |
c8619b90 | 3199 | @item default |
46bdbc00 GK |
3200 | Default visibility is the normal case for the object file format. |
3201 | This value is available for the visibility attribute to override other | |
3202 | options that may change the assumed visibility of entities. | |
3203 | ||
3204 | On ELF, default visibility means that the declaration is visible to other | |
3205 | modules and, in shared libraries, means that the declared entity may be | |
3206 | overridden. | |
3207 | ||
3208 | On Darwin, default visibility means that the declaration is visible to | |
3209 | other modules. | |
3210 | ||
3211 | Default visibility corresponds to ``external linkage'' in the language. | |
6b6cb52e | 3212 | |
c8619b90 | 3213 | @item hidden |
46bdbc00 GK |
3214 | Hidden visibility indicates that the entity declared will have a new |
3215 | form of linkage, which we'll call ``hidden linkage''. Two | |
3216 | declarations of an object with hidden linkage refer to the same object | |
3217 | if they are in the same shared object. | |
6b6cb52e | 3218 | |
c8619b90 NS |
3219 | @item internal |
3220 | Internal visibility is like hidden visibility, but with additional | |
46bdbc00 GK |
3221 | processor specific semantics. Unless otherwise specified by the |
3222 | psABI, GCC defines internal visibility to mean that a function is | |
3223 | @emph{never} called from another module. Compare this with hidden | |
3224 | functions which, while they cannot be referenced directly by other | |
3225 | modules, can be referenced indirectly via function pointers. By | |
3226 | indicating that a function cannot be called from outside the module, | |
3227 | GCC may for instance omit the load of a PIC register since it is known | |
3228 | that the calling function loaded the correct value. | |
6b6cb52e | 3229 | |
c8619b90 | 3230 | @item protected |
46bdbc00 GK |
3231 | Protected visibility is like default visibility except that it |
3232 | indicates that references within the defining module will bind to the | |
3233 | definition in that module. That is, the declared entity cannot be | |
3234 | overridden by another module. | |
6b6cb52e | 3235 | |
c8619b90 | 3236 | @end table |
6b6cb52e | 3237 | |
46bdbc00 GK |
3238 | All visibilities are supported on many, but not all, ELF targets |
3239 | (supported when the assembler supports the @samp{.visibility} | |
3240 | pseudo-op). Default visibility is supported everywhere. Hidden | |
3241 | visibility is supported on Darwin targets. | |
3242 | ||
3243 | The visibility attribute should be applied only to declarations which | |
3244 | would otherwise have external linkage. The attribute should be applied | |
3245 | consistently, so that the same entity should not be declared with | |
3246 | different settings of the attribute. | |
3247 | ||
3248 | In C++, the visibility attribute applies to types as well as functions | |
b9e75696 JM |
3249 | and objects, because in C++ types have linkage. A class must not have |
3250 | greater visibility than its non-static data member types and bases, | |
3251 | and class members default to the visibility of their class. Also, a | |
b70f0f48 JM |
3252 | declaration without explicit visibility is limited to the visibility |
3253 | of its type. | |
46bdbc00 GK |
3254 | |
3255 | In C++, you can mark member functions and static member variables of a | |
d1facce0 | 3256 | class with the visibility attribute. This is useful if you know a |
46bdbc00 GK |
3257 | particular method or static member variable should only be used from |
3258 | one shared object; then you can mark it hidden while the rest of the | |
3259 | class has default visibility. Care must be taken to avoid breaking | |
b70f0f48 JM |
3260 | the One Definition Rule; for example, it is usually not useful to mark |
3261 | an inline method as hidden without marking the whole class as hidden. | |
6b6cb52e | 3262 | |
b9e75696 JM |
3263 | A C++ namespace declaration can also have the visibility attribute. |
3264 | This attribute applies only to the particular namespace body, not to | |
3265 | other definitions of the same namespace; it is equivalent to using | |
3266 | @samp{#pragma GCC visibility} before and after the namespace | |
3267 | definition (@pxref{Visibility Pragmas}). | |
3268 | ||
3269 | In C++, if a template argument has limited visibility, this | |
3270 | restriction is implicitly propagated to the template instantiation. | |
3271 | Otherwise, template instantiations and specializations default to the | |
3272 | visibility of their template. | |
3273 | ||
b70f0f48 JM |
3274 | If both the template and enclosing class have explicit visibility, the |
3275 | visibility from the template is used. | |
3276 | ||
c8619b90 NS |
3277 | @item warn_unused_result |
3278 | @cindex @code{warn_unused_result} attribute | |
3279 | The @code{warn_unused_result} attribute causes a warning to be emitted | |
3280 | if a caller of the function with this attribute does not use its | |
3281 | return value. This is useful for functions where not checking | |
3282 | the result is either a security problem or always a bug, such as | |
3283 | @code{realloc}. | |
6b6cb52e | 3284 | |
c8619b90 NS |
3285 | @smallexample |
3286 | int fn () __attribute__ ((warn_unused_result)); | |
3287 | int foo () | |
3288 | @{ | |
3289 | if (fn () < 0) return -1; | |
3290 | fn (); | |
3291 | return 0; | |
3292 | @} | |
3293 | @end smallexample | |
6b6cb52e | 3294 | |
c8619b90 | 3295 | results in warning on line 5. |
6b6cb52e | 3296 | |
c8619b90 NS |
3297 | @item weak |
3298 | @cindex @code{weak} attribute | |
3299 | The @code{weak} attribute causes the declaration to be emitted as a weak | |
3300 | symbol rather than a global. This is primarily useful in defining | |
3301 | library functions which can be overridden in user code, though it can | |
3302 | also be used with non-function declarations. Weak symbols are supported | |
3303 | for ELF targets, and also for a.out targets when using the GNU assembler | |
3304 | and linker. | |
6b6cb52e | 3305 | |
a0203ca7 AO |
3306 | @item weakref |
3307 | @itemx weakref ("@var{target}") | |
3308 | @cindex @code{weakref} attribute | |
3309 | The @code{weakref} attribute marks a declaration as a weak reference. | |
3310 | Without arguments, it should be accompanied by an @code{alias} attribute | |
3311 | naming the target symbol. Optionally, the @var{target} may be given as | |
3312 | an argument to @code{weakref} itself. In either case, @code{weakref} | |
3313 | implicitly marks the declaration as @code{weak}. Without a | |
3314 | @var{target}, given as an argument to @code{weakref} or to @code{alias}, | |
3315 | @code{weakref} is equivalent to @code{weak}. | |
3316 | ||
3317 | @smallexample | |
a9b0b825 | 3318 | static int x() __attribute__ ((weakref ("y"))); |
a0203ca7 | 3319 | /* is equivalent to... */ |
a9b0b825 | 3320 | static int x() __attribute__ ((weak, weakref, alias ("y"))); |
a0203ca7 | 3321 | /* and to... */ |
a9b0b825 GK |
3322 | static int x() __attribute__ ((weakref)); |
3323 | static int x() __attribute__ ((alias ("y"))); | |
a0203ca7 AO |
3324 | @end smallexample |
3325 | ||
3326 | A weak reference is an alias that does not by itself require a | |
3327 | definition to be given for the target symbol. If the target symbol is | |
3328 | only referenced through weak references, then the becomes a @code{weak} | |
3329 | undefined symbol. If it is directly referenced, however, then such | |
3330 | strong references prevail, and a definition will be required for the | |
3331 | symbol, not necessarily in the same translation unit. | |
3332 | ||
3333 | The effect is equivalent to moving all references to the alias to a | |
3334 | separate translation unit, renaming the alias to the aliased symbol, | |
3335 | declaring it as weak, compiling the two separate translation units and | |
3336 | performing a reloadable link on them. | |
3337 | ||
a9b0b825 GK |
3338 | At present, a declaration to which @code{weakref} is attached can |
3339 | only be @code{static}. | |
3340 | ||
c1f7febf RK |
3341 | @end table |
3342 | ||
3343 | You can specify multiple attributes in a declaration by separating them | |
3344 | by commas within the double parentheses or by immediately following an | |
3345 | attribute declaration with another attribute declaration. | |
3346 | ||
3347 | @cindex @code{#pragma}, reason for not using | |
3348 | @cindex pragma, reason for not using | |
9f1bbeaa JM |
3349 | Some people object to the @code{__attribute__} feature, suggesting that |
3350 | ISO C's @code{#pragma} should be used instead. At the time | |
3351 | @code{__attribute__} was designed, there were two reasons for not doing | |
3352 | this. | |
c1f7febf RK |
3353 | |
3354 | @enumerate | |
3355 | @item | |
3356 | It is impossible to generate @code{#pragma} commands from a macro. | |
3357 | ||
3358 | @item | |
3359 | There is no telling what the same @code{#pragma} might mean in another | |
3360 | compiler. | |
3361 | @end enumerate | |
3362 | ||
9f1bbeaa JM |
3363 | These two reasons applied to almost any application that might have been |
3364 | proposed for @code{#pragma}. It was basically a mistake to use | |
3365 | @code{#pragma} for @emph{anything}. | |
3366 | ||
3367 | The ISO C99 standard includes @code{_Pragma}, which now allows pragmas | |
3368 | to be generated from macros. In addition, a @code{#pragma GCC} | |
3369 | namespace is now in use for GCC-specific pragmas. However, it has been | |
3370 | found convenient to use @code{__attribute__} to achieve a natural | |
3371 | attachment of attributes to their corresponding declarations, whereas | |
3372 | @code{#pragma GCC} is of use for constructs that do not naturally form | |
3373 | part of the grammar. @xref{Other Directives,,Miscellaneous | |
48795525 | 3374 | Preprocessing Directives, cpp, The GNU C Preprocessor}. |
c1f7febf | 3375 | |
2c5e91d2 JM |
3376 | @node Attribute Syntax |
3377 | @section Attribute Syntax | |
3378 | @cindex attribute syntax | |
3379 | ||
3380 | This section describes the syntax with which @code{__attribute__} may be | |
3381 | used, and the constructs to which attribute specifiers bind, for the C | |
161d7b59 | 3382 | language. Some details may vary for C++ and Objective-C@. Because of |
2c5e91d2 JM |
3383 | infelicities in the grammar for attributes, some forms described here |
3384 | may not be successfully parsed in all cases. | |
3385 | ||
91d231cb JM |
3386 | There are some problems with the semantics of attributes in C++. For |
3387 | example, there are no manglings for attributes, although they may affect | |
3388 | code generation, so problems may arise when attributed types are used in | |
3389 | conjunction with templates or overloading. Similarly, @code{typeid} | |
3390 | does not distinguish between types with different attributes. Support | |
3391 | for attributes in C++ may be restricted in future to attributes on | |
3392 | declarations only, but not on nested declarators. | |
3393 | ||
2c5e91d2 JM |
3394 | @xref{Function Attributes}, for details of the semantics of attributes |
3395 | applying to functions. @xref{Variable Attributes}, for details of the | |
3396 | semantics of attributes applying to variables. @xref{Type Attributes}, | |
3397 | for details of the semantics of attributes applying to structure, union | |
3398 | and enumerated types. | |
3399 | ||
3400 | An @dfn{attribute specifier} is of the form | |
3401 | @code{__attribute__ ((@var{attribute-list}))}. An @dfn{attribute list} | |
3402 | is a possibly empty comma-separated sequence of @dfn{attributes}, where | |
3403 | each attribute is one of the following: | |
3404 | ||
3405 | @itemize @bullet | |
3406 | @item | |
3407 | Empty. Empty attributes are ignored. | |
3408 | ||
3409 | @item | |
3410 | A word (which may be an identifier such as @code{unused}, or a reserved | |
3411 | word such as @code{const}). | |
3412 | ||
3413 | @item | |
3414 | A word, followed by, in parentheses, parameters for the attribute. | |
3415 | These parameters take one of the following forms: | |
3416 | ||
3417 | @itemize @bullet | |
3418 | @item | |
3419 | An identifier. For example, @code{mode} attributes use this form. | |
3420 | ||
3421 | @item | |
3422 | An identifier followed by a comma and a non-empty comma-separated list | |
3423 | of expressions. For example, @code{format} attributes use this form. | |
3424 | ||
3425 | @item | |
3426 | A possibly empty comma-separated list of expressions. For example, | |
3427 | @code{format_arg} attributes use this form with the list being a single | |
3428 | integer constant expression, and @code{alias} attributes use this form | |
3429 | with the list being a single string constant. | |
3430 | @end itemize | |
3431 | @end itemize | |
3432 | ||
3433 | An @dfn{attribute specifier list} is a sequence of one or more attribute | |
3434 | specifiers, not separated by any other tokens. | |
3435 | ||
50fc59e7 | 3436 | In GNU C, an attribute specifier list may appear after the colon following a |
2c5e91d2 JM |
3437 | label, other than a @code{case} or @code{default} label. The only |
3438 | attribute it makes sense to use after a label is @code{unused}. This | |
3439 | feature is intended for code generated by programs which contains labels | |
3440 | that may be unused but which is compiled with @option{-Wall}. It would | |
3441 | not normally be appropriate to use in it human-written code, though it | |
3442 | could be useful in cases where the code that jumps to the label is | |
5bca4e80 ILT |
3443 | contained within an @code{#ifdef} conditional. GNU C++ only permits |
3444 | attributes on labels if the attribute specifier is immediately | |
3445 | followed by a semicolon (i.e., the label applies to an empty | |
3446 | statement). If the semicolon is missing, C++ label attributes are | |
3447 | ambiguous, as it is permissible for a declaration, which could begin | |
3448 | with an attribute list, to be labelled in C++. Declarations cannot be | |
3449 | labelled in C90 or C99, so the ambiguity does not arise there. | |
2c5e91d2 JM |
3450 | |
3451 | An attribute specifier list may appear as part of a @code{struct}, | |
3452 | @code{union} or @code{enum} specifier. It may go either immediately | |
3453 | after the @code{struct}, @code{union} or @code{enum} keyword, or after | |
b9e75696 | 3454 | the closing brace. The former syntax is preferred. |
2c5e91d2 JM |
3455 | Where attribute specifiers follow the closing brace, they are considered |
3456 | to relate to the structure, union or enumerated type defined, not to any | |
3457 | enclosing declaration the type specifier appears in, and the type | |
3458 | defined is not complete until after the attribute specifiers. | |
3459 | @c Otherwise, there would be the following problems: a shift/reduce | |
4fe9b91c | 3460 | @c conflict between attributes binding the struct/union/enum and |
2c5e91d2 JM |
3461 | @c binding to the list of specifiers/qualifiers; and "aligned" |
3462 | @c attributes could use sizeof for the structure, but the size could be | |
3463 | @c changed later by "packed" attributes. | |
3464 | ||
3465 | Otherwise, an attribute specifier appears as part of a declaration, | |
3466 | counting declarations of unnamed parameters and type names, and relates | |
3467 | to that declaration (which may be nested in another declaration, for | |
91d231cb JM |
3468 | example in the case of a parameter declaration), or to a particular declarator |
3469 | within a declaration. Where an | |
ff867905 JM |
3470 | attribute specifier is applied to a parameter declared as a function or |
3471 | an array, it should apply to the function or array rather than the | |
3472 | pointer to which the parameter is implicitly converted, but this is not | |
3473 | yet correctly implemented. | |
2c5e91d2 JM |
3474 | |
3475 | Any list of specifiers and qualifiers at the start of a declaration may | |
3476 | contain attribute specifiers, whether or not such a list may in that | |
3477 | context contain storage class specifiers. (Some attributes, however, | |
3478 | are essentially in the nature of storage class specifiers, and only make | |
3479 | sense where storage class specifiers may be used; for example, | |
3480 | @code{section}.) There is one necessary limitation to this syntax: the | |
3481 | first old-style parameter declaration in a function definition cannot | |
3482 | begin with an attribute specifier, because such an attribute applies to | |
3483 | the function instead by syntax described below (which, however, is not | |
3484 | yet implemented in this case). In some other cases, attribute | |
3485 | specifiers are permitted by this grammar but not yet supported by the | |
3486 | compiler. All attribute specifiers in this place relate to the | |
c771326b | 3487 | declaration as a whole. In the obsolescent usage where a type of |
2c5e91d2 JM |
3488 | @code{int} is implied by the absence of type specifiers, such a list of |
3489 | specifiers and qualifiers may be an attribute specifier list with no | |
3490 | other specifiers or qualifiers. | |
3491 | ||
7dcb0442 JM |
3492 | At present, the first parameter in a function prototype must have some |
3493 | type specifier which is not an attribute specifier; this resolves an | |
3494 | ambiguity in the interpretation of @code{void f(int | |
3495 | (__attribute__((foo)) x))}, but is subject to change. At present, if | |
3496 | the parentheses of a function declarator contain only attributes then | |
3497 | those attributes are ignored, rather than yielding an error or warning | |
3498 | or implying a single parameter of type int, but this is subject to | |
3499 | change. | |
3500 | ||
2c5e91d2 JM |
3501 | An attribute specifier list may appear immediately before a declarator |
3502 | (other than the first) in a comma-separated list of declarators in a | |
3503 | declaration of more than one identifier using a single list of | |
4b01f8d8 | 3504 | specifiers and qualifiers. Such attribute specifiers apply |
9c34dbbf ZW |
3505 | only to the identifier before whose declarator they appear. For |
3506 | example, in | |
3507 | ||
3508 | @smallexample | |
3509 | __attribute__((noreturn)) void d0 (void), | |
3510 | __attribute__((format(printf, 1, 2))) d1 (const char *, ...), | |
3511 | d2 (void) | |
3512 | @end smallexample | |
3513 | ||
3514 | @noindent | |
3515 | the @code{noreturn} attribute applies to all the functions | |
4b01f8d8 | 3516 | declared; the @code{format} attribute only applies to @code{d1}. |
2c5e91d2 JM |
3517 | |
3518 | An attribute specifier list may appear immediately before the comma, | |
3519 | @code{=} or semicolon terminating the declaration of an identifier other | |
770a9950 JM |
3520 | than a function definition. Such attribute specifiers apply |
3521 | to the declared object or function. Where an | |
9c34dbbf | 3522 | assembler name for an object or function is specified (@pxref{Asm |
770a9950 JM |
3523 | Labels}), the attribute must follow the @code{asm} |
3524 | specification. | |
2c5e91d2 JM |
3525 | |
3526 | An attribute specifier list may, in future, be permitted to appear after | |
3527 | the declarator in a function definition (before any old-style parameter | |
3528 | declarations or the function body). | |
3529 | ||
0e03329a JM |
3530 | Attribute specifiers may be mixed with type qualifiers appearing inside |
3531 | the @code{[]} of a parameter array declarator, in the C99 construct by | |
3532 | which such qualifiers are applied to the pointer to which the array is | |
3533 | implicitly converted. Such attribute specifiers apply to the pointer, | |
3534 | not to the array, but at present this is not implemented and they are | |
3535 | ignored. | |
3536 | ||
2c5e91d2 JM |
3537 | An attribute specifier list may appear at the start of a nested |
3538 | declarator. At present, there are some limitations in this usage: the | |
91d231cb JM |
3539 | attributes correctly apply to the declarator, but for most individual |
3540 | attributes the semantics this implies are not implemented. | |
3541 | When attribute specifiers follow the @code{*} of a pointer | |
4b01f8d8 | 3542 | declarator, they may be mixed with any type qualifiers present. |
91d231cb | 3543 | The following describes the formal semantics of this syntax. It will make the |
2c5e91d2 JM |
3544 | most sense if you are familiar with the formal specification of |
3545 | declarators in the ISO C standard. | |
3546 | ||
3547 | Consider (as in C99 subclause 6.7.5 paragraph 4) a declaration @code{T | |
3548 | D1}, where @code{T} contains declaration specifiers that specify a type | |
3549 | @var{Type} (such as @code{int}) and @code{D1} is a declarator that | |
3550 | contains an identifier @var{ident}. The type specified for @var{ident} | |
3551 | for derived declarators whose type does not include an attribute | |
3552 | specifier is as in the ISO C standard. | |
3553 | ||
3554 | If @code{D1} has the form @code{( @var{attribute-specifier-list} D )}, | |
3555 | and the declaration @code{T D} specifies the type | |
3556 | ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then | |
3557 | @code{T D1} specifies the type ``@var{derived-declarator-type-list} | |
3558 | @var{attribute-specifier-list} @var{Type}'' for @var{ident}. | |
3559 | ||
3560 | If @code{D1} has the form @code{* | |
3561 | @var{type-qualifier-and-attribute-specifier-list} D}, and the | |
3562 | declaration @code{T D} specifies the type | |
3563 | ``@var{derived-declarator-type-list} @var{Type}'' for @var{ident}, then | |
3564 | @code{T D1} specifies the type ``@var{derived-declarator-type-list} | |
3565 | @var{type-qualifier-and-attribute-specifier-list} @var{Type}'' for | |
3566 | @var{ident}. | |
3567 | ||
f282ffb3 | 3568 | For example, |
9c34dbbf ZW |
3569 | |
3570 | @smallexample | |
3571 | void (__attribute__((noreturn)) ****f) (void); | |
3572 | @end smallexample | |
3573 | ||
3574 | @noindent | |
3575 | specifies the type ``pointer to pointer to pointer to pointer to | |
3576 | non-returning function returning @code{void}''. As another example, | |
3577 | ||
3578 | @smallexample | |
3579 | char *__attribute__((aligned(8))) *f; | |
3580 | @end smallexample | |
3581 | ||
3582 | @noindent | |
3583 | specifies the type ``pointer to 8-byte-aligned pointer to @code{char}''. | |
91d231cb JM |
3584 | Note again that this does not work with most attributes; for example, |
3585 | the usage of @samp{aligned} and @samp{noreturn} attributes given above | |
3586 | is not yet supported. | |
3587 | ||
3588 | For compatibility with existing code written for compiler versions that | |
3589 | did not implement attributes on nested declarators, some laxity is | |
3590 | allowed in the placing of attributes. If an attribute that only applies | |
3591 | to types is applied to a declaration, it will be treated as applying to | |
3592 | the type of that declaration. If an attribute that only applies to | |
3593 | declarations is applied to the type of a declaration, it will be treated | |
3594 | as applying to that declaration; and, for compatibility with code | |
3595 | placing the attributes immediately before the identifier declared, such | |
3596 | an attribute applied to a function return type will be treated as | |
3597 | applying to the function type, and such an attribute applied to an array | |
3598 | element type will be treated as applying to the array type. If an | |
3599 | attribute that only applies to function types is applied to a | |
3600 | pointer-to-function type, it will be treated as applying to the pointer | |
3601 | target type; if such an attribute is applied to a function return type | |
3602 | that is not a pointer-to-function type, it will be treated as applying | |
3603 | to the function type. | |
2c5e91d2 | 3604 | |
c1f7febf RK |
3605 | @node Function Prototypes |
3606 | @section Prototypes and Old-Style Function Definitions | |
3607 | @cindex function prototype declarations | |
3608 | @cindex old-style function definitions | |
3609 | @cindex promotion of formal parameters | |
3610 | ||
5490d604 | 3611 | GNU C extends ISO C to allow a function prototype to override a later |
c1f7febf RK |
3612 | old-style non-prototype definition. Consider the following example: |
3613 | ||
3ab51846 | 3614 | @smallexample |
c1f7febf | 3615 | /* @r{Use prototypes unless the compiler is old-fashioned.} */ |
d863830b | 3616 | #ifdef __STDC__ |
c1f7febf RK |
3617 | #define P(x) x |
3618 | #else | |
3619 | #define P(x) () | |
3620 | #endif | |
3621 | ||
3622 | /* @r{Prototype function declaration.} */ | |
3623 | int isroot P((uid_t)); | |
3624 | ||
3625 | /* @r{Old-style function definition.} */ | |
3626 | int | |
12bcfaa1 | 3627 | isroot (x) /* @r{??? lossage here ???} */ |
c1f7febf RK |
3628 | uid_t x; |
3629 | @{ | |
3630 | return x == 0; | |
3631 | @} | |
3ab51846 | 3632 | @end smallexample |
c1f7febf | 3633 | |
5490d604 | 3634 | Suppose the type @code{uid_t} happens to be @code{short}. ISO C does |
c1f7febf RK |
3635 | not allow this example, because subword arguments in old-style |
3636 | non-prototype definitions are promoted. Therefore in this example the | |
3637 | function definition's argument is really an @code{int}, which does not | |
3638 | match the prototype argument type of @code{short}. | |
3639 | ||
5490d604 | 3640 | This restriction of ISO C makes it hard to write code that is portable |
c1f7febf RK |
3641 | to traditional C compilers, because the programmer does not know |
3642 | whether the @code{uid_t} type is @code{short}, @code{int}, or | |
3643 | @code{long}. Therefore, in cases like these GNU C allows a prototype | |
3644 | to override a later old-style definition. More precisely, in GNU C, a | |
3645 | function prototype argument type overrides the argument type specified | |
3646 | by a later old-style definition if the former type is the same as the | |
3647 | latter type before promotion. Thus in GNU C the above example is | |
3648 | equivalent to the following: | |
3649 | ||
3ab51846 | 3650 | @smallexample |
c1f7febf RK |
3651 | int isroot (uid_t); |
3652 | ||
3653 | int | |
3654 | isroot (uid_t x) | |
3655 | @{ | |
3656 | return x == 0; | |
3657 | @} | |
3ab51846 | 3658 | @end smallexample |
c1f7febf | 3659 | |
9c34dbbf | 3660 | @noindent |
c1f7febf RK |
3661 | GNU C++ does not support old-style function definitions, so this |
3662 | extension is irrelevant. | |
3663 | ||
3664 | @node C++ Comments | |
3665 | @section C++ Style Comments | |
3666 | @cindex // | |
3667 | @cindex C++ comments | |
3668 | @cindex comments, C++ style | |
3669 | ||
3670 | In GNU C, you may use C++ style comments, which start with @samp{//} and | |
3671 | continue until the end of the line. Many other C implementations allow | |
f458d1d5 ZW |
3672 | such comments, and they are included in the 1999 C standard. However, |
3673 | C++ style comments are not recognized if you specify an @option{-std} | |
3674 | option specifying a version of ISO C before C99, or @option{-ansi} | |
3675 | (equivalent to @option{-std=c89}). | |
c1f7febf RK |
3676 | |
3677 | @node Dollar Signs | |
3678 | @section Dollar Signs in Identifier Names | |
3679 | @cindex $ | |
3680 | @cindex dollar signs in identifier names | |
3681 | @cindex identifier names, dollar signs in | |
3682 | ||
79188db9 RK |
3683 | In GNU C, you may normally use dollar signs in identifier names. |
3684 | This is because many traditional C implementations allow such identifiers. | |
3685 | However, dollar signs in identifiers are not supported on a few target | |
3686 | machines, typically because the target assembler does not allow them. | |
c1f7febf RK |
3687 | |
3688 | @node Character Escapes | |
3689 | @section The Character @key{ESC} in Constants | |
3690 | ||
3691 | You can use the sequence @samp{\e} in a string or character constant to | |
3692 | stand for the ASCII character @key{ESC}. | |
3693 | ||
3694 | @node Alignment | |
3695 | @section Inquiring on Alignment of Types or Variables | |
3696 | @cindex alignment | |
3697 | @cindex type alignment | |
3698 | @cindex variable alignment | |
3699 | ||
3700 | The keyword @code{__alignof__} allows you to inquire about how an object | |
3701 | is aligned, or the minimum alignment usually required by a type. Its | |
3702 | syntax is just like @code{sizeof}. | |
3703 | ||
3704 | For example, if the target machine requires a @code{double} value to be | |
3705 | aligned on an 8-byte boundary, then @code{__alignof__ (double)} is 8. | |
3706 | This is true on many RISC machines. On more traditional machine | |
3707 | designs, @code{__alignof__ (double)} is 4 or even 2. | |
3708 | ||
3709 | Some machines never actually require alignment; they allow reference to any | |
64c18e57 | 3710 | data type even at an odd address. For these machines, @code{__alignof__} |
2a31c49c HPN |
3711 | reports the smallest alignment that GCC will give the data type, usually as |
3712 | mandated by the target ABI. | |
c1f7febf | 3713 | |
5372b3fb NB |
3714 | If the operand of @code{__alignof__} is an lvalue rather than a type, |
3715 | its value is the required alignment for its type, taking into account | |
3716 | any minimum alignment specified with GCC's @code{__attribute__} | |
3717 | extension (@pxref{Variable Attributes}). For example, after this | |
3718 | declaration: | |
c1f7febf | 3719 | |
3ab51846 | 3720 | @smallexample |
c1f7febf | 3721 | struct foo @{ int x; char y; @} foo1; |
3ab51846 | 3722 | @end smallexample |
c1f7febf RK |
3723 | |
3724 | @noindent | |
5372b3fb NB |
3725 | the value of @code{__alignof__ (foo1.y)} is 1, even though its actual |
3726 | alignment is probably 2 or 4, the same as @code{__alignof__ (int)}. | |
c1f7febf | 3727 | |
9d27bffe SS |
3728 | It is an error to ask for the alignment of an incomplete type. |
3729 | ||
c1f7febf RK |
3730 | @node Variable Attributes |
3731 | @section Specifying Attributes of Variables | |
3732 | @cindex attribute of variables | |
3733 | @cindex variable attributes | |
3734 | ||
3735 | The keyword @code{__attribute__} allows you to specify special | |
3736 | attributes of variables or structure fields. This keyword is followed | |
905e8651 RH |
3737 | by an attribute specification inside double parentheses. Some |
3738 | attributes are currently defined generically for variables. | |
3739 | Other attributes are defined for variables on particular target | |
3740 | systems. Other attributes are available for functions | |
3741 | (@pxref{Function Attributes}) and for types (@pxref{Type Attributes}). | |
3742 | Other front ends might define more attributes | |
3743 | (@pxref{C++ Extensions,,Extensions to the C++ Language}). | |
c1f7febf RK |
3744 | |
3745 | You may also specify attributes with @samp{__} preceding and following | |
3746 | each keyword. This allows you to use them in header files without | |
3747 | being concerned about a possible macro of the same name. For example, | |
3748 | you may use @code{__aligned__} instead of @code{aligned}. | |
3749 | ||
2c5e91d2 JM |
3750 | @xref{Attribute Syntax}, for details of the exact syntax for using |
3751 | attributes. | |
3752 | ||
c1f7febf RK |
3753 | @table @code |
3754 | @cindex @code{aligned} attribute | |
3755 | @item aligned (@var{alignment}) | |
3756 | This attribute specifies a minimum alignment for the variable or | |
3757 | structure field, measured in bytes. For example, the declaration: | |
3758 | ||
3759 | @smallexample | |
3760 | int x __attribute__ ((aligned (16))) = 0; | |
3761 | @end smallexample | |
3762 | ||
3763 | @noindent | |
3764 | causes the compiler to allocate the global variable @code{x} on a | |
3765 | 16-byte boundary. On a 68040, this could be used in conjunction with | |
3766 | an @code{asm} expression to access the @code{move16} instruction which | |
3767 | requires 16-byte aligned operands. | |
3768 | ||
3769 | You can also specify the alignment of structure fields. For example, to | |
3770 | create a double-word aligned @code{int} pair, you could write: | |
3771 | ||
3772 | @smallexample | |
3773 | struct foo @{ int x[2] __attribute__ ((aligned (8))); @}; | |
3774 | @end smallexample | |
3775 | ||
3776 | @noindent | |
3777 | This is an alternative to creating a union with a @code{double} member | |
3778 | that forces the union to be double-word aligned. | |
3779 | ||
c1f7febf RK |
3780 | As in the preceding examples, you can explicitly specify the alignment |
3781 | (in bytes) that you wish the compiler to use for a given variable or | |
3782 | structure field. Alternatively, you can leave out the alignment factor | |
6e4f1168 L |
3783 | and just ask the compiler to align a variable or field to the |
3784 | default alignment for the target architecture you are compiling for. | |
3785 | The default alignment is sufficient for all scalar types, but may not be | |
3786 | enough for all vector types on a target which supports vector operations. | |
3787 | The default alignment is fixed for a particular target ABI. | |
3788 | ||
3789 | Gcc also provides a target specific macro @code{__BIGGEST_ALIGNMENT__}, | |
3790 | which is the largest alignment ever used for any data type on the | |
3791 | target machine you are compiling for. For example, you could write: | |
c1f7febf RK |
3792 | |
3793 | @smallexample | |
6e4f1168 | 3794 | short array[3] __attribute__ ((aligned (__BIGGEST_ALIGNMENT__))); |
c1f7febf RK |
3795 | @end smallexample |
3796 | ||
6e4f1168 L |
3797 | The compiler automatically sets the alignment for the declared |
3798 | variable or field to @code{__BIGGEST_ALIGNMENT__}. Doing this can | |
3799 | often make copy operations more efficient, because the compiler can | |
3800 | use whatever instructions copy the biggest chunks of memory when | |
3801 | performing copies to or from the variables or fields that you have | |
3802 | aligned this way. Note that the value of @code{__BIGGEST_ALIGNMENT__} | |
3803 | may change depending on command line options. | |
c1f7febf | 3804 | |
e9f9692b MW |
3805 | When used on a struct, or struct member, the @code{aligned} attribute can |
3806 | only increase the alignment; in order to decrease it, the @code{packed} | |
3807 | attribute must be specified as well. When used as part of a typedef, the | |
3808 | @code{aligned} attribute can both increase and decrease alignment, and | |
3809 | specifying the @code{packed} attribute will generate a warning. | |
c1f7febf RK |
3810 | |
3811 | Note that the effectiveness of @code{aligned} attributes may be limited | |
3812 | by inherent limitations in your linker. On many systems, the linker is | |
3813 | only able to arrange for variables to be aligned up to a certain maximum | |
3814 | alignment. (For some linkers, the maximum supported alignment may | |
3815 | be very very small.) If your linker is only able to align variables | |
3816 | up to a maximum of 8 byte alignment, then specifying @code{aligned(16)} | |
3817 | in an @code{__attribute__} will still only provide you with 8 byte | |
3818 | alignment. See your linker documentation for further information. | |
3819 | ||
837edd5f GK |
3820 | The @code{aligned} attribute can also be used for functions |
3821 | (@pxref{Function Attributes}.) | |
3822 | ||
0bfa5f65 RH |
3823 | @item cleanup (@var{cleanup_function}) |
3824 | @cindex @code{cleanup} attribute | |
3825 | The @code{cleanup} attribute runs a function when the variable goes | |
3826 | out of scope. This attribute can only be applied to auto function | |
3827 | scope variables; it may not be applied to parameters or variables | |
3828 | with static storage duration. The function must take one parameter, | |
3829 | a pointer to a type compatible with the variable. The return value | |
3830 | of the function (if any) is ignored. | |
3831 | ||
3832 | If @option{-fexceptions} is enabled, then @var{cleanup_function} | |
3833 | will be run during the stack unwinding that happens during the | |
3834 | processing of the exception. Note that the @code{cleanup} attribute | |
3835 | does not allow the exception to be caught, only to perform an action. | |
3836 | It is undefined what happens if @var{cleanup_function} does not | |
3837 | return normally. | |
3838 | ||
905e8651 RH |
3839 | @item common |
3840 | @itemx nocommon | |
3841 | @cindex @code{common} attribute | |
3842 | @cindex @code{nocommon} attribute | |
3843 | @opindex fcommon | |
3844 | @opindex fno-common | |
3845 | The @code{common} attribute requests GCC to place a variable in | |
3846 | ``common'' storage. The @code{nocommon} attribute requests the | |
78466c0e | 3847 | opposite---to allocate space for it directly. |
905e8651 | 3848 | |
daf2f129 | 3849 | These attributes override the default chosen by the |
905e8651 RH |
3850 | @option{-fno-common} and @option{-fcommon} flags respectively. |
3851 | ||
3852 | @item deprecated | |
9b86d6bb | 3853 | @itemx deprecated (@var{msg}) |
905e8651 RH |
3854 | @cindex @code{deprecated} attribute |
3855 | The @code{deprecated} attribute results in a warning if the variable | |
3856 | is used anywhere in the source file. This is useful when identifying | |
3857 | variables that are expected to be removed in a future version of a | |
3858 | program. The warning also includes the location of the declaration | |
3859 | of the deprecated variable, to enable users to easily find further | |
3860 | information about why the variable is deprecated, or what they should | |
64c18e57 | 3861 | do instead. Note that the warning only occurs for uses: |
905e8651 RH |
3862 | |
3863 | @smallexample | |
3864 | extern int old_var __attribute__ ((deprecated)); | |
3865 | extern int old_var; | |
3866 | int new_fn () @{ return old_var; @} | |
3867 | @end smallexample | |
3868 | ||
9b86d6bb L |
3869 | results in a warning on line 3 but not line 2. The optional msg |
3870 | argument, which must be a string, will be printed in the warning if | |
3871 | present. | |
905e8651 RH |
3872 | |
3873 | The @code{deprecated} attribute can also be used for functions and | |
3874 | types (@pxref{Function Attributes}, @pxref{Type Attributes}.) | |
3875 | ||
c1f7febf RK |
3876 | @item mode (@var{mode}) |
3877 | @cindex @code{mode} attribute | |
3878 | This attribute specifies the data type for the declaration---whichever | |
3879 | type corresponds to the mode @var{mode}. This in effect lets you | |
3880 | request an integer or floating point type according to its width. | |
3881 | ||
3882 | You may also specify a mode of @samp{byte} or @samp{__byte__} to | |
3883 | indicate the mode corresponding to a one-byte integer, @samp{word} or | |
3884 | @samp{__word__} for the mode of a one-word integer, and @samp{pointer} | |
3885 | or @samp{__pointer__} for the mode used to represent pointers. | |
3886 | ||
c1f7febf RK |
3887 | @item packed |
3888 | @cindex @code{packed} attribute | |
3889 | The @code{packed} attribute specifies that a variable or structure field | |
3890 | should have the smallest possible alignment---one byte for a variable, | |
3891 | and one bit for a field, unless you specify a larger value with the | |
3892 | @code{aligned} attribute. | |
3893 | ||
3894 | Here is a structure in which the field @code{x} is packed, so that it | |
3895 | immediately follows @code{a}: | |
3896 | ||
3ab51846 | 3897 | @smallexample |
c1f7febf RK |
3898 | struct foo |
3899 | @{ | |
3900 | char a; | |
3901 | int x[2] __attribute__ ((packed)); | |
3902 | @}; | |
3ab51846 | 3903 | @end smallexample |
c1f7febf | 3904 | |
2cd36c22 AN |
3905 | @emph{Note:} The 4.1, 4.2 and 4.3 series of GCC ignore the |
3906 | @code{packed} attribute on bit-fields of type @code{char}. This has | |
3907 | been fixed in GCC 4.4 but the change can lead to differences in the | |
048fd785 | 3908 | structure layout. See the documentation of |
2cd36c22 AN |
3909 | @option{-Wpacked-bitfield-compat} for more information. |
3910 | ||
84330467 | 3911 | @item section ("@var{section-name}") |
c1f7febf RK |
3912 | @cindex @code{section} variable attribute |
3913 | Normally, the compiler places the objects it generates in sections like | |
3914 | @code{data} and @code{bss}. Sometimes, however, you need additional sections, | |
3915 | or you need certain particular variables to appear in special sections, | |
3916 | for example to map to special hardware. The @code{section} | |
3917 | attribute specifies that a variable (or function) lives in a particular | |
3918 | section. For example, this small program uses several specific section names: | |
3919 | ||
3920 | @smallexample | |
3921 | struct duart a __attribute__ ((section ("DUART_A"))) = @{ 0 @}; | |
3922 | struct duart b __attribute__ ((section ("DUART_B"))) = @{ 0 @}; | |
3923 | char stack[10000] __attribute__ ((section ("STACK"))) = @{ 0 @}; | |
8b9d598f | 3924 | int init_data __attribute__ ((section ("INITDATA"))); |
c1f7febf RK |
3925 | |
3926 | main() | |
3927 | @{ | |
12bcfaa1 | 3928 | /* @r{Initialize stack pointer} */ |
c1f7febf RK |
3929 | init_sp (stack + sizeof (stack)); |
3930 | ||
12bcfaa1 | 3931 | /* @r{Initialize initialized data} */ |
c1f7febf RK |
3932 | memcpy (&init_data, &data, &edata - &data); |
3933 | ||
12bcfaa1 | 3934 | /* @r{Turn on the serial ports} */ |
c1f7febf RK |
3935 | init_duart (&a); |
3936 | init_duart (&b); | |
3937 | @} | |
3938 | @end smallexample | |
3939 | ||
3940 | @noindent | |
8b9d598f SE |
3941 | Use the @code{section} attribute with |
3942 | @emph{global} variables and not @emph{local} variables, | |
3943 | as shown in the example. | |
c1f7febf | 3944 | |
8b9d598f SE |
3945 | You may use the @code{section} attribute with initialized or |
3946 | uninitialized global variables but the linker requires | |
c1f7febf RK |
3947 | each object be defined once, with the exception that uninitialized |
3948 | variables tentatively go in the @code{common} (or @code{bss}) section | |
8b9d598f SE |
3949 | and can be multiply ``defined''. Using the @code{section} attribute |
3950 | will change what section the variable goes into and may cause the | |
3951 | linker to issue an error if an uninitialized variable has multiple | |
3952 | definitions. You can force a variable to be initialized with the | |
3953 | @option{-fno-common} flag or the @code{nocommon} attribute. | |
c1f7febf RK |
3954 | |
3955 | Some file formats do not support arbitrary sections so the @code{section} | |
3956 | attribute is not available on all platforms. | |
3957 | If you need to map the entire contents of a module to a particular | |
3958 | section, consider using the facilities of the linker instead. | |
3959 | ||
593d3a34 MK |
3960 | @item shared |
3961 | @cindex @code{shared} variable attribute | |
95fef11f | 3962 | On Microsoft Windows, in addition to putting variable definitions in a named |
02f52e19 | 3963 | section, the section can also be shared among all running copies of an |
161d7b59 | 3964 | executable or DLL@. For example, this small program defines shared data |
84330467 | 3965 | by putting it in a named section @code{shared} and marking the section |
593d3a34 MK |
3966 | shareable: |
3967 | ||
3968 | @smallexample | |
3969 | int foo __attribute__((section ("shared"), shared)) = 0; | |
3970 | ||
3971 | int | |
3972 | main() | |
3973 | @{ | |
12bcfaa1 JM |
3974 | /* @r{Read and write foo. All running |
3975 | copies see the same value.} */ | |
593d3a34 MK |
3976 | return 0; |
3977 | @} | |
3978 | @end smallexample | |
3979 | ||
3980 | @noindent | |
3981 | You may only use the @code{shared} attribute along with @code{section} | |
02f52e19 | 3982 | attribute with a fully initialized global definition because of the way |
593d3a34 MK |
3983 | linkers work. See @code{section} attribute for more information. |
3984 | ||
95fef11f | 3985 | The @code{shared} attribute is only available on Microsoft Windows@. |
593d3a34 | 3986 | |
905e8651 RH |
3987 | @item tls_model ("@var{tls_model}") |
3988 | @cindex @code{tls_model} attribute | |
3989 | The @code{tls_model} attribute sets thread-local storage model | |
3990 | (@pxref{Thread-Local}) of a particular @code{__thread} variable, | |
4ec7afd7 | 3991 | overriding @option{-ftls-model=} command line switch on a per-variable |
905e8651 RH |
3992 | basis. |
3993 | The @var{tls_model} argument should be one of @code{global-dynamic}, | |
3994 | @code{local-dynamic}, @code{initial-exec} or @code{local-exec}. | |
3995 | ||
3996 | Not all targets support this attribute. | |
3997 | ||
c1f7febf RK |
3998 | @item unused |
3999 | This attribute, attached to a variable, means that the variable is meant | |
f0523f02 | 4000 | to be possibly unused. GCC will not produce a warning for this |
c1f7febf RK |
4001 | variable. |
4002 | ||
5f79d643 RM |
4003 | @item used |
4004 | This attribute, attached to a variable, means that the variable must be | |
4005 | emitted even if it appears that the variable is not referenced. | |
4006 | ||
1b9191d2 AH |
4007 | @item vector_size (@var{bytes}) |
4008 | This attribute specifies the vector size for the variable, measured in | |
4009 | bytes. For example, the declaration: | |
4010 | ||
4011 | @smallexample | |
4012 | int foo __attribute__ ((vector_size (16))); | |
4013 | @end smallexample | |
4014 | ||
4015 | @noindent | |
4016 | causes the compiler to set the mode for @code{foo}, to be 16 bytes, | |
4017 | divided into @code{int} sized units. Assuming a 32-bit int (a vector of | |
4018 | 4 units of 4 bytes), the corresponding mode of @code{foo} will be V4SI@. | |
4019 | ||
4020 | This attribute is only applicable to integral and float scalars, | |
4021 | although arrays, pointers, and function return values are allowed in | |
4022 | conjunction with this construct. | |
4023 | ||
4024 | Aggregates with this attribute are invalid, even if they are of the same | |
4025 | size as a corresponding scalar. For example, the declaration: | |
4026 | ||
4027 | @smallexample | |
ad706f54 | 4028 | struct S @{ int a; @}; |
1b9191d2 AH |
4029 | struct S __attribute__ ((vector_size (16))) foo; |
4030 | @end smallexample | |
4031 | ||
4032 | @noindent | |
4033 | is invalid even if the size of the structure is the same as the size of | |
4034 | the @code{int}. | |
4035 | ||
a20f6f00 DS |
4036 | @item selectany |
4037 | The @code{selectany} attribute causes an initialized global variable to | |
4038 | have link-once semantics. When multiple definitions of the variable are | |
4039 | encountered by the linker, the first is selected and the remainder are | |
4040 | discarded. Following usage by the Microsoft compiler, the linker is told | |
4041 | @emph{not} to warn about size or content differences of the multiple | |
4042 | definitions. | |
4043 | ||
4044 | Although the primary usage of this attribute is for POD types, the | |
4045 | attribute can also be applied to global C++ objects that are initialized | |
4046 | by a constructor. In this case, the static initialization and destruction | |
4047 | code for the object is emitted in each translation defining the object, | |
4048 | but the calls to the constructor and destructor are protected by a | |
0ac11108 | 4049 | link-once guard variable. |
a20f6f00 DS |
4050 | |
4051 | The @code{selectany} attribute is only available on Microsoft Windows | |
4052 | targets. You can use @code{__declspec (selectany)} as a synonym for | |
4053 | @code{__attribute__ ((selectany))} for compatibility with other | |
4054 | compilers. | |
4055 | ||
c1f7febf | 4056 | @item weak |
38bb2b65 | 4057 | The @code{weak} attribute is described in @ref{Function Attributes}. |
6b6cb52e DS |
4058 | |
4059 | @item dllimport | |
38bb2b65 | 4060 | The @code{dllimport} attribute is described in @ref{Function Attributes}. |
6b6cb52e | 4061 | |
9baf8aea | 4062 | @item dllexport |
38bb2b65 | 4063 | The @code{dllexport} attribute is described in @ref{Function Attributes}. |
6b6cb52e | 4064 | |
905e8651 RH |
4065 | @end table |
4066 | ||
4af797b5 JZ |
4067 | @subsection Blackfin Variable Attributes |
4068 | ||
4069 | Three attributes are currently defined for the Blackfin. | |
4070 | ||
4071 | @table @code | |
4072 | @item l1_data | |
4073 | @item l1_data_A | |
4074 | @item l1_data_B | |
4075 | @cindex @code{l1_data} variable attribute | |
4076 | @cindex @code{l1_data_A} variable attribute | |
4077 | @cindex @code{l1_data_B} variable attribute | |
4078 | Use these attributes on the Blackfin to place the variable into L1 Data SRAM. | |
4079 | Variables with @code{l1_data} attribute will be put into the specific section | |
4080 | named @code{.l1.data}. Those with @code{l1_data_A} attribute will be put into | |
4081 | the specific section named @code{.l1.data.A}. Those with @code{l1_data_B} | |
4082 | attribute will be put into the specific section named @code{.l1.data.B}. | |
4083 | @end table | |
4084 | ||
905e8651 | 4085 | @subsection M32R/D Variable Attributes |
845da534 | 4086 | |
8a36672b | 4087 | One attribute is currently defined for the M32R/D@. |
905e8651 RH |
4088 | |
4089 | @table @code | |
845da534 DE |
4090 | @item model (@var{model-name}) |
4091 | @cindex variable addressability on the M32R/D | |
4092 | Use this attribute on the M32R/D to set the addressability of an object. | |
4093 | The identifier @var{model-name} is one of @code{small}, @code{medium}, | |
4094 | or @code{large}, representing each of the code models. | |
4095 | ||
4096 | Small model objects live in the lower 16MB of memory (so that their | |
4097 | addresses can be loaded with the @code{ld24} instruction). | |
4098 | ||
02f52e19 | 4099 | Medium and large model objects may live anywhere in the 32-bit address space |
845da534 DE |
4100 | (the compiler will generate @code{seth/add3} instructions to load their |
4101 | addresses). | |
905e8651 | 4102 | @end table |
845da534 | 4103 | |
1ccbef77 | 4104 | @anchor{i386 Variable Attributes} |
fe77449a DR |
4105 | @subsection i386 Variable Attributes |
4106 | ||
4107 | Two attributes are currently defined for i386 configurations: | |
4108 | @code{ms_struct} and @code{gcc_struct} | |
4109 | ||
905e8651 | 4110 | @table @code |
fe77449a DR |
4111 | @item ms_struct |
4112 | @itemx gcc_struct | |
905e8651 RH |
4113 | @cindex @code{ms_struct} attribute |
4114 | @cindex @code{gcc_struct} attribute | |
fe77449a DR |
4115 | |
4116 | If @code{packed} is used on a structure, or if bit-fields are used | |
4117 | it may be that the Microsoft ABI packs them differently | |
4118 | than GCC would normally pack them. Particularly when moving packed | |
4119 | data between functions compiled with GCC and the native Microsoft compiler | |
4120 | (either via function call or as data in a file), it may be necessary to access | |
4121 | either format. | |
4122 | ||
95fef11f | 4123 | Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86 |
fe77449a | 4124 | compilers to match the native Microsoft compiler. |
0ac11108 EC |
4125 | |
4126 | The Microsoft structure layout algorithm is fairly simple with the exception | |
4127 | of the bitfield packing: | |
4128 | ||
4129 | The padding and alignment of members of structures and whether a bit field | |
4130 | can straddle a storage-unit boundary | |
4131 | ||
4132 | @enumerate | |
4133 | @item Structure members are stored sequentially in the order in which they are | |
4134 | declared: the first member has the lowest memory address and the last member | |
4135 | the highest. | |
4136 | ||
4137 | @item Every data object has an alignment-requirement. The alignment-requirement | |
4138 | for all data except structures, unions, and arrays is either the size of the | |
4139 | object or the current packing size (specified with either the aligned attribute | |
4140 | or the pack pragma), whichever is less. For structures, unions, and arrays, | |
4141 | the alignment-requirement is the largest alignment-requirement of its members. | |
4142 | Every object is allocated an offset so that: | |
4143 | ||
4144 | offset % alignment-requirement == 0 | |
4145 | ||
4146 | @item Adjacent bit fields are packed into the same 1-, 2-, or 4-byte allocation | |
4147 | unit if the integral types are the same size and if the next bit field fits | |
4148 | into the current allocation unit without crossing the boundary imposed by the | |
4149 | common alignment requirements of the bit fields. | |
4150 | @end enumerate | |
4151 | ||
4152 | Handling of zero-length bitfields: | |
4153 | ||
4154 | MSVC interprets zero-length bitfields in the following ways: | |
4155 | ||
4156 | @enumerate | |
4157 | @item If a zero-length bitfield is inserted between two bitfields that would | |
4158 | normally be coalesced, the bitfields will not be coalesced. | |
4159 | ||
4160 | For example: | |
4161 | ||
4162 | @smallexample | |
4163 | struct | |
4164 | @{ | |
4165 | unsigned long bf_1 : 12; | |
4166 | unsigned long : 0; | |
4167 | unsigned long bf_2 : 12; | |
4168 | @} t1; | |
4169 | @end smallexample | |
4170 | ||
4171 | The size of @code{t1} would be 8 bytes with the zero-length bitfield. If the | |
4172 | zero-length bitfield were removed, @code{t1}'s size would be 4 bytes. | |
4173 | ||
4174 | @item If a zero-length bitfield is inserted after a bitfield, @code{foo}, and the | |
4175 | alignment of the zero-length bitfield is greater than the member that follows it, | |
4176 | @code{bar}, @code{bar} will be aligned as the type of the zero-length bitfield. | |
4177 | ||
4178 | For example: | |
4179 | ||
4180 | @smallexample | |
4181 | struct | |
4182 | @{ | |
4183 | char foo : 4; | |
4184 | short : 0; | |
4185 | char bar; | |
4186 | @} t2; | |
4187 | ||
4188 | struct | |
4189 | @{ | |
4190 | char foo : 4; | |
4191 | short : 0; | |
4192 | double bar; | |
4193 | @} t3; | |
4194 | @end smallexample | |
4195 | ||
4196 | For @code{t2}, @code{bar} will be placed at offset 2, rather than offset 1. | |
4197 | Accordingly, the size of @code{t2} will be 4. For @code{t3}, the zero-length | |
4198 | bitfield will not affect the alignment of @code{bar} or, as a result, the size | |
4199 | of the structure. | |
4200 | ||
4201 | Taking this into account, it is important to note the following: | |
4202 | ||
4203 | @enumerate | |
4204 | @item If a zero-length bitfield follows a normal bitfield, the type of the | |
4205 | zero-length bitfield may affect the alignment of the structure as whole. For | |
4206 | example, @code{t2} has a size of 4 bytes, since the zero-length bitfield follows a | |
4207 | normal bitfield, and is of type short. | |
4208 | ||
4209 | @item Even if a zero-length bitfield is not followed by a normal bitfield, it may | |
4210 | still affect the alignment of the structure: | |
4211 | ||
4212 | @smallexample | |
4213 | struct | |
4214 | @{ | |
4215 | char foo : 6; | |
4216 | long : 0; | |
4217 | @} t4; | |
4218 | @end smallexample | |
4219 | ||
4220 | Here, @code{t4} will take up 4 bytes. | |
4221 | @end enumerate | |
4222 | ||
4223 | @item Zero-length bitfields following non-bitfield members are ignored: | |
4224 | ||
4225 | @smallexample | |
4226 | struct | |
4227 | @{ | |
4228 | char foo; | |
4229 | long : 0; | |
4230 | char bar; | |
4231 | @} t5; | |
4232 | @end smallexample | |
4233 | ||
4234 | Here, @code{t5} will take up 2 bytes. | |
4235 | @end enumerate | |
c1f7febf RK |
4236 | @end table |
4237 | ||
1ccbef77 EC |
4238 | @subsection PowerPC Variable Attributes |
4239 | ||
63d0dca4 DE |
4240 | Three attributes currently are defined for PowerPC configurations: |
4241 | @code{altivec}, @code{ms_struct} and @code{gcc_struct}. | |
1ccbef77 | 4242 | |
63d0dca4 | 4243 | For full documentation of the struct attributes please see the |
38bb2b65 | 4244 | documentation in @ref{i386 Variable Attributes}. |
63d0dca4 DE |
4245 | |
4246 | For documentation of @code{altivec} attribute please see the | |
38bb2b65 | 4247 | documentation in @ref{PowerPC Type Attributes}. |
1ccbef77 | 4248 | |
85d9c13c TS |
4249 | @subsection SPU Variable Attributes |
4250 | ||
4251 | The SPU supports the @code{spu_vector} attribute for variables. For | |
38bb2b65 SL |
4252 | documentation of this attribute please see the documentation in |
4253 | @ref{SPU Type Attributes}. | |
85d9c13c | 4254 | |
54e9a19d DD |
4255 | @subsection Xstormy16 Variable Attributes |
4256 | ||
4257 | One attribute is currently defined for xstormy16 configurations: | |
38bb2b65 | 4258 | @code{below100}. |
54e9a19d DD |
4259 | |
4260 | @table @code | |
4261 | @item below100 | |
4262 | @cindex @code{below100} attribute | |
4263 | ||
4264 | If a variable has the @code{below100} attribute (@code{BELOW100} is | |
4265 | allowed also), GCC will place the variable in the first 0x100 bytes of | |
4266 | memory and use special opcodes to access it. Such variables will be | |
4267 | placed in either the @code{.bss_below100} section or the | |
4268 | @code{.data_below100} section. | |
4269 | ||
4270 | @end table | |
4271 | ||
79532d34 EW |
4272 | @subsection AVR Variable Attributes |
4273 | ||
4274 | @table @code | |
4275 | @item progmem | |
4276 | @cindex @code{progmem} variable attribute | |
4277 | The @code{progmem} attribute is used on the AVR to place data in the Program | |
4278 | Memory address space. The AVR is a Harvard Architecture processor and data | |
4279 | normally resides in the Data Memory address space. | |
4280 | @end table | |
4281 | ||
c1f7febf RK |
4282 | @node Type Attributes |
4283 | @section Specifying Attributes of Types | |
4284 | @cindex attribute of types | |
4285 | @cindex type attributes | |
4286 | ||
4287 | The keyword @code{__attribute__} allows you to specify special | |
b9e75696 JM |
4288 | attributes of @code{struct} and @code{union} types when you define |
4289 | such types. This keyword is followed by an attribute specification | |
4290 | inside double parentheses. Seven attributes are currently defined for | |
4291 | types: @code{aligned}, @code{packed}, @code{transparent_union}, | |
4292 | @code{unused}, @code{deprecated}, @code{visibility}, and | |
4293 | @code{may_alias}. Other attributes are defined for functions | |
4294 | (@pxref{Function Attributes}) and for variables (@pxref{Variable | |
4295 | Attributes}). | |
c1f7febf RK |
4296 | |
4297 | You may also specify any one of these attributes with @samp{__} | |
4298 | preceding and following its keyword. This allows you to use these | |
4299 | attributes in header files without being concerned about a possible | |
4300 | macro of the same name. For example, you may use @code{__aligned__} | |
4301 | instead of @code{aligned}. | |
4302 | ||
4009f2e7 JM |
4303 | You may specify type attributes in an enum, struct or union type |
4304 | declaration or definition, or for other types in a @code{typedef} | |
4305 | declaration. | |
c1f7febf | 4306 | |
b9e75696 JM |
4307 | For an enum, struct or union type, you may specify attributes either |
4308 | between the enum, struct or union tag and the name of the type, or | |
4309 | just past the closing curly brace of the @emph{definition}. The | |
4310 | former syntax is preferred. | |
4051959b | 4311 | |
2c5e91d2 JM |
4312 | @xref{Attribute Syntax}, for details of the exact syntax for using |
4313 | attributes. | |
4314 | ||
c1f7febf RK |
4315 | @table @code |
4316 | @cindex @code{aligned} attribute | |
4317 | @item aligned (@var{alignment}) | |
4318 | This attribute specifies a minimum alignment (in bytes) for variables | |
4319 | of the specified type. For example, the declarations: | |
4320 | ||
4321 | @smallexample | |
f69eecfb JL |
4322 | struct S @{ short f[3]; @} __attribute__ ((aligned (8))); |
4323 | typedef int more_aligned_int __attribute__ ((aligned (8))); | |
c1f7febf RK |
4324 | @end smallexample |
4325 | ||
4326 | @noindent | |
d863830b | 4327 | force the compiler to insure (as far as it can) that each variable whose |
c1f7febf | 4328 | type is @code{struct S} or @code{more_aligned_int} will be allocated and |
981f6289 | 4329 | aligned @emph{at least} on a 8-byte boundary. On a SPARC, having all |
c1f7febf RK |
4330 | variables of type @code{struct S} aligned to 8-byte boundaries allows |
4331 | the compiler to use the @code{ldd} and @code{std} (doubleword load and | |
4332 | store) instructions when copying one variable of type @code{struct S} to | |
4333 | another, thus improving run-time efficiency. | |
4334 | ||
4335 | Note that the alignment of any given @code{struct} or @code{union} type | |
5490d604 | 4336 | is required by the ISO C standard to be at least a perfect multiple of |
c1f7febf RK |
4337 | the lowest common multiple of the alignments of all of the members of |
4338 | the @code{struct} or @code{union} in question. This means that you @emph{can} | |
4339 | effectively adjust the alignment of a @code{struct} or @code{union} | |
4340 | type by attaching an @code{aligned} attribute to any one of the members | |
4341 | of such a type, but the notation illustrated in the example above is a | |
4342 | more obvious, intuitive, and readable way to request the compiler to | |
4343 | adjust the alignment of an entire @code{struct} or @code{union} type. | |
4344 | ||
4345 | As in the preceding example, you can explicitly specify the alignment | |
4346 | (in bytes) that you wish the compiler to use for a given @code{struct} | |
4347 | or @code{union} type. Alternatively, you can leave out the alignment factor | |
4348 | and just ask the compiler to align a type to the maximum | |
4349 | useful alignment for the target machine you are compiling for. For | |
4350 | example, you could write: | |
4351 | ||
4352 | @smallexample | |
4353 | struct S @{ short f[3]; @} __attribute__ ((aligned)); | |
4354 | @end smallexample | |
4355 | ||
4356 | Whenever you leave out the alignment factor in an @code{aligned} | |
4357 | attribute specification, the compiler automatically sets the alignment | |
4358 | for the type to the largest alignment which is ever used for any data | |
4359 | type on the target machine you are compiling for. Doing this can often | |
4360 | make copy operations more efficient, because the compiler can use | |
4361 | whatever instructions copy the biggest chunks of memory when performing | |
4362 | copies to or from the variables which have types that you have aligned | |
4363 | this way. | |
4364 | ||
4365 | In the example above, if the size of each @code{short} is 2 bytes, then | |
4366 | the size of the entire @code{struct S} type is 6 bytes. The smallest | |
4367 | power of two which is greater than or equal to that is 8, so the | |
4368 | compiler sets the alignment for the entire @code{struct S} type to 8 | |
4369 | bytes. | |
4370 | ||
4371 | Note that although you can ask the compiler to select a time-efficient | |
4372 | alignment for a given type and then declare only individual stand-alone | |
4373 | objects of that type, the compiler's ability to select a time-efficient | |
4374 | alignment is primarily useful only when you plan to create arrays of | |
4375 | variables having the relevant (efficiently aligned) type. If you | |
4376 | declare or use arrays of variables of an efficiently-aligned type, then | |
4377 | it is likely that your program will also be doing pointer arithmetic (or | |
4378 | subscripting, which amounts to the same thing) on pointers to the | |
4379 | relevant type, and the code that the compiler generates for these | |
4380 | pointer arithmetic operations will often be more efficient for | |
4381 | efficiently-aligned types than for other types. | |
4382 | ||
4383 | The @code{aligned} attribute can only increase the alignment; but you | |
4384 | can decrease it by specifying @code{packed} as well. See below. | |
4385 | ||
4386 | Note that the effectiveness of @code{aligned} attributes may be limited | |
4387 | by inherent limitations in your linker. On many systems, the linker is | |
4388 | only able to arrange for variables to be aligned up to a certain maximum | |
4389 | alignment. (For some linkers, the maximum supported alignment may | |
4390 | be very very small.) If your linker is only able to align variables | |
4391 | up to a maximum of 8 byte alignment, then specifying @code{aligned(16)} | |
4392 | in an @code{__attribute__} will still only provide you with 8 byte | |
4393 | alignment. See your linker documentation for further information. | |
4394 | ||
4395 | @item packed | |
a5bcc582 | 4396 | This attribute, attached to @code{struct} or @code{union} type |
d1a701eb MM |
4397 | definition, specifies that each member (other than zero-width bitfields) |
4398 | of the structure or union is placed to minimize the memory required. When | |
4399 | attached to an @code{enum} definition, it indicates that the smallest | |
4400 | integral type should be used. | |
c1f7febf | 4401 | |
84330467 | 4402 | @opindex fshort-enums |
c1f7febf RK |
4403 | Specifying this attribute for @code{struct} and @code{union} types is |
4404 | equivalent to specifying the @code{packed} attribute on each of the | |
84330467 | 4405 | structure or union members. Specifying the @option{-fshort-enums} |
c1f7febf RK |
4406 | flag on the line is equivalent to specifying the @code{packed} |
4407 | attribute on all @code{enum} definitions. | |
4408 | ||
a5bcc582 NS |
4409 | In the following example @code{struct my_packed_struct}'s members are |
4410 | packed closely together, but the internal layout of its @code{s} member | |
78466c0e | 4411 | is not packed---to do that, @code{struct my_unpacked_struct} would need to |
a5bcc582 NS |
4412 | be packed too. |
4413 | ||
4414 | @smallexample | |
4415 | struct my_unpacked_struct | |
4416 | @{ | |
4417 | char c; | |
4418 | int i; | |
4419 | @}; | |
4420 | ||
75b66a16 | 4421 | struct __attribute__ ((__packed__)) my_packed_struct |
a5bcc582 NS |
4422 | @{ |
4423 | char c; | |
4424 | int i; | |
4425 | struct my_unpacked_struct s; | |
4426 | @}; | |
4427 | @end smallexample | |
4428 | ||
e4ae5e77 | 4429 | You may only specify this attribute on the definition of an @code{enum}, |
a5bcc582 NS |
4430 | @code{struct} or @code{union}, not on a @code{typedef} which does not |
4431 | also define the enumerated type, structure or union. | |
c1f7febf RK |
4432 | |
4433 | @item transparent_union | |
4434 | This attribute, attached to a @code{union} type definition, indicates | |
4435 | that any function parameter having that union type causes calls to that | |
4436 | function to be treated in a special way. | |
4437 | ||
4438 | First, the argument corresponding to a transparent union type can be of | |
4439 | any type in the union; no cast is required. Also, if the union contains | |
4440 | a pointer type, the corresponding argument can be a null pointer | |
4441 | constant or a void pointer expression; and if the union contains a void | |
4442 | pointer type, the corresponding argument can be any pointer expression. | |
4443 | If the union member type is a pointer, qualifiers like @code{const} on | |
4444 | the referenced type must be respected, just as with normal pointer | |
4445 | conversions. | |
4446 | ||
4447 | Second, the argument is passed to the function using the calling | |
64c18e57 | 4448 | conventions of the first member of the transparent union, not the calling |
c1f7febf RK |
4449 | conventions of the union itself. All members of the union must have the |
4450 | same machine representation; this is necessary for this argument passing | |
4451 | to work properly. | |
4452 | ||
4453 | Transparent unions are designed for library functions that have multiple | |
4454 | interfaces for compatibility reasons. For example, suppose the | |
4455 | @code{wait} function must accept either a value of type @code{int *} to | |
4456 | comply with Posix, or a value of type @code{union wait *} to comply with | |
4457 | the 4.1BSD interface. If @code{wait}'s parameter were @code{void *}, | |
4458 | @code{wait} would accept both kinds of arguments, but it would also | |
4459 | accept any other pointer type and this would make argument type checking | |
4460 | less useful. Instead, @code{<sys/wait.h>} might define the interface | |
4461 | as follows: | |
4462 | ||
4463 | @smallexample | |
4009f2e7 | 4464 | typedef union __attribute__ ((__transparent_union__)) |
c1f7febf RK |
4465 | @{ |
4466 | int *__ip; | |
4467 | union wait *__up; | |
4009f2e7 | 4468 | @} wait_status_ptr_t; |
c1f7febf RK |
4469 | |
4470 | pid_t wait (wait_status_ptr_t); | |
4471 | @end smallexample | |
4472 | ||
4473 | This interface allows either @code{int *} or @code{union wait *} | |
4474 | arguments to be passed, using the @code{int *} calling convention. | |
4475 | The program can call @code{wait} with arguments of either type: | |
4476 | ||
3ab51846 | 4477 | @smallexample |
c1f7febf RK |
4478 | int w1 () @{ int w; return wait (&w); @} |
4479 | int w2 () @{ union wait w; return wait (&w); @} | |
3ab51846 | 4480 | @end smallexample |
c1f7febf RK |
4481 | |
4482 | With this interface, @code{wait}'s implementation might look like this: | |
4483 | ||
3ab51846 | 4484 | @smallexample |
c1f7febf RK |
4485 | pid_t wait (wait_status_ptr_t p) |
4486 | @{ | |
4487 | return waitpid (-1, p.__ip, 0); | |
4488 | @} | |
3ab51846 | 4489 | @end smallexample |
d863830b JL |
4490 | |
4491 | @item unused | |
4492 | When attached to a type (including a @code{union} or a @code{struct}), | |
4493 | this attribute means that variables of that type are meant to appear | |
f0523f02 | 4494 | possibly unused. GCC will not produce a warning for any variables of |
d863830b JL |
4495 | that type, even if the variable appears to do nothing. This is often |
4496 | the case with lock or thread classes, which are usually defined and then | |
4497 | not referenced, but contain constructors and destructors that have | |
956d6950 | 4498 | nontrivial bookkeeping functions. |
d863830b | 4499 | |
e23bd218 | 4500 | @item deprecated |
9b86d6bb | 4501 | @itemx deprecated (@var{msg}) |
e23bd218 IR |
4502 | The @code{deprecated} attribute results in a warning if the type |
4503 | is used anywhere in the source file. This is useful when identifying | |
4504 | types that are expected to be removed in a future version of a program. | |
4505 | If possible, the warning also includes the location of the declaration | |
4506 | of the deprecated type, to enable users to easily find further | |
4507 | information about why the type is deprecated, or what they should do | |
4508 | instead. Note that the warnings only occur for uses and then only | |
adc9fe67 | 4509 | if the type is being applied to an identifier that itself is not being |
e23bd218 IR |
4510 | declared as deprecated. |
4511 | ||
4512 | @smallexample | |
4513 | typedef int T1 __attribute__ ((deprecated)); | |
4514 | T1 x; | |
4515 | typedef T1 T2; | |
4516 | T2 y; | |
4517 | typedef T1 T3 __attribute__ ((deprecated)); | |
4518 | T3 z __attribute__ ((deprecated)); | |
4519 | @end smallexample | |
4520 | ||
4521 | results in a warning on line 2 and 3 but not lines 4, 5, or 6. No | |
4522 | warning is issued for line 4 because T2 is not explicitly | |
4523 | deprecated. Line 5 has no warning because T3 is explicitly | |
9b86d6bb L |
4524 | deprecated. Similarly for line 6. The optional msg |
4525 | argument, which must be a string, will be printed in the warning if | |
4526 | present. | |
e23bd218 IR |
4527 | |
4528 | The @code{deprecated} attribute can also be used for functions and | |
4529 | variables (@pxref{Function Attributes}, @pxref{Variable Attributes}.) | |
4530 | ||
d18b1ed8 | 4531 | @item may_alias |
ac7ee6ad RG |
4532 | Accesses through pointers to types with this attribute are not subject |
4533 | to type-based alias analysis, but are instead assumed to be able to alias | |
4534 | any other type of objects. In the context of 6.5/7 an lvalue expression | |
4535 | dereferencing such a pointer is treated like having a character type. | |
4536 | See @option{-fstrict-aliasing} for more information on aliasing issues. | |
4537 | This extension exists to support some vector APIs, in which pointers to | |
4538 | one vector type are permitted to alias pointers to a different vector type. | |
4539 | ||
4540 | Note that an object of a type with this attribute does not have any | |
4541 | special semantics. | |
d18b1ed8 OS |
4542 | |
4543 | Example of use: | |
4544 | ||
478c9e72 | 4545 | @smallexample |
d18b1ed8 OS |
4546 | typedef short __attribute__((__may_alias__)) short_a; |
4547 | ||
4548 | int | |
4549 | main (void) | |
4550 | @{ | |
4551 | int a = 0x12345678; | |
4552 | short_a *b = (short_a *) &a; | |
4553 | ||
4554 | b[1] = 0; | |
4555 | ||
4556 | if (a == 0x12345678) | |
4557 | abort(); | |
4558 | ||
4559 | exit(0); | |
4560 | @} | |
478c9e72 | 4561 | @end smallexample |
d18b1ed8 OS |
4562 | |
4563 | If you replaced @code{short_a} with @code{short} in the variable | |
4564 | declaration, the above program would abort when compiled with | |
4565 | @option{-fstrict-aliasing}, which is on by default at @option{-O2} or | |
4566 | above in recent GCC versions. | |
fe77449a | 4567 | |
b9e75696 | 4568 | @item visibility |
b9e75696 JM |
4569 | In C++, attribute visibility (@pxref{Function Attributes}) can also be |
4570 | applied to class, struct, union and enum types. Unlike other type | |
4571 | attributes, the attribute must appear between the initial keyword and | |
4572 | the name of the type; it cannot appear after the body of the type. | |
4573 | ||
b70f0f48 JM |
4574 | Note that the type visibility is applied to vague linkage entities |
4575 | associated with the class (vtable, typeinfo node, etc.). In | |
4576 | particular, if a class is thrown as an exception in one shared object | |
4577 | and caught in another, the class must have default visibility. | |
4578 | Otherwise the two shared objects will be unable to use the same | |
4579 | typeinfo node and exception handling will break. | |
4580 | ||
38bb2b65 SL |
4581 | @end table |
4582 | ||
04fb56d5 MM |
4583 | @subsection ARM Type Attributes |
4584 | ||
4585 | On those ARM targets that support @code{dllimport} (such as Symbian | |
f0eb93a8 | 4586 | OS), you can use the @code{notshared} attribute to indicate that the |
04fb56d5 | 4587 | virtual table and other similar data for a class should not be |
8a36672b | 4588 | exported from a DLL@. For example: |
04fb56d5 MM |
4589 | |
4590 | @smallexample | |
4591 | class __declspec(notshared) C @{ | |
4592 | public: | |
f0eb93a8 | 4593 | __declspec(dllimport) C(); |
04fb56d5 MM |
4594 | virtual void f(); |
4595 | @} | |
4596 | ||
4597 | __declspec(dllexport) | |
4598 | C::C() @{@} | |
4599 | @end smallexample | |
4600 | ||
4601 | In this code, @code{C::C} is exported from the current DLL, but the | |
4602 | virtual table for @code{C} is not exported. (You can use | |
4603 | @code{__attribute__} instead of @code{__declspec} if you prefer, but | |
4604 | most Symbian OS code uses @code{__declspec}.) | |
4605 | ||
63d0dca4 | 4606 | @anchor{i386 Type Attributes} |
fe77449a DR |
4607 | @subsection i386 Type Attributes |
4608 | ||
4609 | Two attributes are currently defined for i386 configurations: | |
38bb2b65 SL |
4610 | @code{ms_struct} and @code{gcc_struct}. |
4611 | ||
4612 | @table @code | |
fe77449a DR |
4613 | |
4614 | @item ms_struct | |
4615 | @itemx gcc_struct | |
4616 | @cindex @code{ms_struct} | |
4617 | @cindex @code{gcc_struct} | |
4618 | ||
4619 | If @code{packed} is used on a structure, or if bit-fields are used | |
4620 | it may be that the Microsoft ABI packs them differently | |
4621 | than GCC would normally pack them. Particularly when moving packed | |
4622 | data between functions compiled with GCC and the native Microsoft compiler | |
4623 | (either via function call or as data in a file), it may be necessary to access | |
4624 | either format. | |
4625 | ||
95fef11f | 4626 | Currently @option{-m[no-]ms-bitfields} is provided for the Microsoft Windows X86 |
fe77449a | 4627 | compilers to match the native Microsoft compiler. |
c1f7febf RK |
4628 | @end table |
4629 | ||
4630 | To specify multiple attributes, separate them by commas within the | |
4631 | double parentheses: for example, @samp{__attribute__ ((aligned (16), | |
4632 | packed))}. | |
4633 | ||
63d0dca4 DE |
4634 | @anchor{PowerPC Type Attributes} |
4635 | @subsection PowerPC Type Attributes | |
4636 | ||
4637 | Three attributes currently are defined for PowerPC configurations: | |
4638 | @code{altivec}, @code{ms_struct} and @code{gcc_struct}. | |
4639 | ||
38bb2b65 SL |
4640 | For full documentation of the @code{ms_struct} and @code{gcc_struct} |
4641 | attributes please see the documentation in @ref{i386 Type Attributes}. | |
63d0dca4 DE |
4642 | |
4643 | The @code{altivec} attribute allows one to declare AltiVec vector data | |
4644 | types supported by the AltiVec Programming Interface Manual. The | |
4645 | attribute requires an argument to specify one of three vector types: | |
4646 | @code{vector__}, @code{pixel__} (always followed by unsigned short), | |
4647 | and @code{bool__} (always followed by unsigned). | |
4648 | ||
4649 | @smallexample | |
4650 | __attribute__((altivec(vector__))) | |
4651 | __attribute__((altivec(pixel__))) unsigned short | |
4652 | __attribute__((altivec(bool__))) unsigned | |
4653 | @end smallexample | |
4654 | ||
4655 | These attributes mainly are intended to support the @code{__vector}, | |
4656 | @code{__pixel}, and @code{__bool} AltiVec keywords. | |
4657 | ||
85d9c13c TS |
4658 | @anchor{SPU Type Attributes} |
4659 | @subsection SPU Type Attributes | |
4660 | ||
4661 | The SPU supports the @code{spu_vector} attribute for types. This attribute | |
4662 | allows one to declare vector data types supported by the Sony/Toshiba/IBM SPU | |
4663 | Language Extensions Specification. It is intended to support the | |
4664 | @code{__vector} keyword. | |
4665 | ||
4666 | ||
c1f7febf RK |
4667 | @node Inline |
4668 | @section An Inline Function is As Fast As a Macro | |
4669 | @cindex inline functions | |
4670 | @cindex integrating function code | |
4671 | @cindex open coding | |
4672 | @cindex macros, inline alternative | |
4673 | ||
0a052b16 GK |
4674 | By declaring a function inline, you can direct GCC to make |
4675 | calls to that function faster. One way GCC can achieve this is to | |
c1f7febf RK |
4676 | integrate that function's code into the code for its callers. This |
4677 | makes execution faster by eliminating the function-call overhead; in | |
0a052b16 GK |
4678 | addition, if any of the actual argument values are constant, their |
4679 | known values may permit simplifications at compile time so that not | |
4680 | all of the inline function's code needs to be included. The effect on | |
4681 | code size is less predictable; object code may be larger or smaller | |
4682 | with function inlining, depending on the particular case. You can | |
4683 | also direct GCC to try to integrate all ``simple enough'' functions | |
4684 | into their callers with the option @option{-finline-functions}. | |
4685 | ||
4686 | GCC implements three different semantics of declaring a function | |
da1c7394 ILT |
4687 | inline. One is available with @option{-std=gnu89} or |
4688 | @option{-fgnu89-inline} or when @code{gnu_inline} attribute is present | |
4689 | on all inline declarations, another when @option{-std=c99} or | |
4690 | @option{-std=gnu99} (without @option{-fgnu89-inline}), and the third | |
4691 | is used when compiling C++. | |
4b404517 | 4692 | |
c1f7febf RK |
4693 | To declare a function inline, use the @code{inline} keyword in its |
4694 | declaration, like this: | |
4695 | ||
3ab51846 | 4696 | @smallexample |
0a052b16 | 4697 | static inline int |
c1f7febf RK |
4698 | inc (int *a) |
4699 | @{ | |
4700 | (*a)++; | |
4701 | @} | |
3ab51846 | 4702 | @end smallexample |
c1f7febf | 4703 | |
0a052b16 GK |
4704 | If you are writing a header file to be included in ISO C89 programs, write |
4705 | @code{__inline__} instead of @code{inline}. @xref{Alternate Keywords}. | |
247b14bd | 4706 | |
0a052b16 GK |
4707 | The three types of inlining behave similarly in two important cases: |
4708 | when the @code{inline} keyword is used on a @code{static} function, | |
4709 | like the example above, and when a function is first declared without | |
4710 | using the @code{inline} keyword and then is defined with | |
4711 | @code{inline}, like this: | |
c1f7febf | 4712 | |
0a052b16 GK |
4713 | @smallexample |
4714 | extern int inc (int *a); | |
4715 | inline int | |
4716 | inc (int *a) | |
4717 | @{ | |
4718 | (*a)++; | |
4719 | @} | |
4720 | @end smallexample | |
c1f7febf | 4721 | |
0a052b16 GK |
4722 | In both of these common cases, the program behaves the same as if you |
4723 | had not used the @code{inline} keyword, except for its speed. | |
c1f7febf RK |
4724 | |
4725 | @cindex inline functions, omission of | |
84330467 | 4726 | @opindex fkeep-inline-functions |
c1f7febf RK |
4727 | When a function is both inline and @code{static}, if all calls to the |
4728 | function are integrated into the caller, and the function's address is | |
4729 | never used, then the function's own assembler code is never referenced. | |
f0523f02 | 4730 | In this case, GCC does not actually output assembler code for the |
84330467 | 4731 | function, unless you specify the option @option{-fkeep-inline-functions}. |
c1f7febf RK |
4732 | Some calls cannot be integrated for various reasons (in particular, |
4733 | calls that precede the function's definition cannot be integrated, and | |
4734 | neither can recursive calls within the definition). If there is a | |
4735 | nonintegrated call, then the function is compiled to assembler code as | |
4736 | usual. The function must also be compiled as usual if the program | |
4737 | refers to its address, because that can't be inlined. | |
4738 | ||
0a052b16 GK |
4739 | @opindex Winline |
4740 | Note that certain usages in a function definition can make it unsuitable | |
4741 | for inline substitution. Among these usages are: use of varargs, use of | |
4742 | alloca, use of variable sized data types (@pxref{Variable Length}), | |
4743 | use of computed goto (@pxref{Labels as Values}), use of nonlocal goto, | |
4744 | and nested functions (@pxref{Nested Functions}). Using @option{-Winline} | |
4745 | will warn when a function marked @code{inline} could not be substituted, | |
4746 | and will give the reason for the failure. | |
4747 | ||
4748 | @cindex automatic @code{inline} for C++ member fns | |
4749 | @cindex @code{inline} automatic for C++ member fns | |
4750 | @cindex member fns, automatically @code{inline} | |
4751 | @cindex C++ member fns, automatically @code{inline} | |
4752 | @opindex fno-default-inline | |
4753 | As required by ISO C++, GCC considers member functions defined within | |
4754 | the body of a class to be marked inline even if they are | |
4755 | not explicitly declared with the @code{inline} keyword. You can | |
4756 | override this with @option{-fno-default-inline}; @pxref{C++ Dialect | |
4757 | Options,,Options Controlling C++ Dialect}. | |
4758 | ||
4759 | GCC does not inline any functions when not optimizing unless you specify | |
4760 | the @samp{always_inline} attribute for the function, like this: | |
4761 | ||
4762 | @smallexample | |
4763 | /* @r{Prototype.} */ | |
4764 | inline void foo (const char) __attribute__((always_inline)); | |
4765 | @end smallexample | |
4766 | ||
4767 | The remainder of this section is specific to GNU C89 inlining. | |
4768 | ||
c1f7febf RK |
4769 | @cindex non-static inline function |
4770 | When an inline function is not @code{static}, then the compiler must assume | |
4771 | that there may be calls from other source files; since a global symbol can | |
4772 | be defined only once in any program, the function must not be defined in | |
4773 | the other source files, so the calls therein cannot be integrated. | |
4774 | Therefore, a non-@code{static} inline function is always compiled on its | |
4775 | own in the usual fashion. | |
4776 | ||
4777 | If you specify both @code{inline} and @code{extern} in the function | |
4778 | definition, then the definition is used only for inlining. In no case | |
4779 | is the function compiled on its own, not even if you refer to its | |
4780 | address explicitly. Such an address becomes an external reference, as | |
4781 | if you had only declared the function, and had not defined it. | |
4782 | ||
4783 | This combination of @code{inline} and @code{extern} has almost the | |
4784 | effect of a macro. The way to use it is to put a function definition in | |
4785 | a header file with these keywords, and put another copy of the | |
4786 | definition (lacking @code{inline} and @code{extern}) in a library file. | |
4787 | The definition in the header file will cause most calls to the function | |
4788 | to be inlined. If any uses of the function remain, they will refer to | |
4789 | the single copy in the library. | |
4790 | ||
c1f7febf RK |
4791 | @node Extended Asm |
4792 | @section Assembler Instructions with C Expression Operands | |
4793 | @cindex extended @code{asm} | |
4794 | @cindex @code{asm} expressions | |
4795 | @cindex assembler instructions | |
4796 | @cindex registers | |
4797 | ||
c85f7c16 JL |
4798 | In an assembler instruction using @code{asm}, you can specify the |
4799 | operands of the instruction using C expressions. This means you need not | |
4800 | guess which registers or memory locations will contain the data you want | |
c1f7febf RK |
4801 | to use. |
4802 | ||
c85f7c16 JL |
4803 | You must specify an assembler instruction template much like what |
4804 | appears in a machine description, plus an operand constraint string for | |
4805 | each operand. | |
c1f7febf RK |
4806 | |
4807 | For example, here is how to use the 68881's @code{fsinx} instruction: | |
4808 | ||
3ab51846 | 4809 | @smallexample |
c1f7febf | 4810 | asm ("fsinx %1,%0" : "=f" (result) : "f" (angle)); |
3ab51846 | 4811 | @end smallexample |
c1f7febf RK |
4812 | |
4813 | @noindent | |
4814 | Here @code{angle} is the C expression for the input operand while | |
4815 | @code{result} is that of the output operand. Each has @samp{"f"} as its | |
c85f7c16 JL |
4816 | operand constraint, saying that a floating point register is required. |
4817 | The @samp{=} in @samp{=f} indicates that the operand is an output; all | |
4818 | output operands' constraints must use @samp{=}. The constraints use the | |
4819 | same language used in the machine description (@pxref{Constraints}). | |
4820 | ||
4821 | Each operand is described by an operand-constraint string followed by | |
4822 | the C expression in parentheses. A colon separates the assembler | |
4823 | template from the first output operand and another separates the last | |
4824 | output operand from the first input, if any. Commas separate the | |
84b72302 RH |
4825 | operands within each group. The total number of operands is currently |
4826 | limited to 30; this limitation may be lifted in some future version of | |
8a36672b | 4827 | GCC@. |
c85f7c16 JL |
4828 | |
4829 | If there are no output operands but there are input operands, you must | |
4830 | place two consecutive colons surrounding the place where the output | |
c1f7febf RK |
4831 | operands would go. |
4832 | ||
84b72302 RH |
4833 | As of GCC version 3.1, it is also possible to specify input and output |
4834 | operands using symbolic names which can be referenced within the | |
4835 | assembler code. These names are specified inside square brackets | |
4836 | preceding the constraint string, and can be referenced inside the | |
4837 | assembler code using @code{%[@var{name}]} instead of a percentage sign | |
4838 | followed by the operand number. Using named operands the above example | |
4839 | could look like: | |
4840 | ||
3ab51846 | 4841 | @smallexample |
84b72302 RH |
4842 | asm ("fsinx %[angle],%[output]" |
4843 | : [output] "=f" (result) | |
4844 | : [angle] "f" (angle)); | |
3ab51846 | 4845 | @end smallexample |
84b72302 RH |
4846 | |
4847 | @noindent | |
4848 | Note that the symbolic operand names have no relation whatsoever to | |
4849 | other C identifiers. You may use any name you like, even those of | |
64c18e57 | 4850 | existing C symbols, but you must ensure that no two operands within the same |
84b72302 RH |
4851 | assembler construct use the same symbolic name. |
4852 | ||
c1f7febf | 4853 | Output operand expressions must be lvalues; the compiler can check this. |
c85f7c16 JL |
4854 | The input operands need not be lvalues. The compiler cannot check |
4855 | whether the operands have data types that are reasonable for the | |
4856 | instruction being executed. It does not parse the assembler instruction | |
4857 | template and does not know what it means or even whether it is valid | |
4858 | assembler input. The extended @code{asm} feature is most often used for | |
4859 | machine instructions the compiler itself does not know exist. If | |
4860 | the output expression cannot be directly addressed (for example, it is a | |
f0523f02 | 4861 | bit-field), your constraint must allow a register. In that case, GCC |
c85f7c16 JL |
4862 | will use the register as the output of the @code{asm}, and then store |
4863 | that register into the output. | |
4864 | ||
f0523f02 | 4865 | The ordinary output operands must be write-only; GCC will assume that |
c85f7c16 JL |
4866 | the values in these operands before the instruction are dead and need |
4867 | not be generated. Extended asm supports input-output or read-write | |
4868 | operands. Use the constraint character @samp{+} to indicate such an | |
373a04f1 JM |
4869 | operand and list it with the output operands. You should only use |
4870 | read-write operands when the constraints for the operand (or the | |
4871 | operand in which only some of the bits are to be changed) allow a | |
4872 | register. | |
4873 | ||
4874 | You may, as an alternative, logically split its function into two | |
4875 | separate operands, one input operand and one write-only output | |
4876 | operand. The connection between them is expressed by constraints | |
4877 | which say they need to be in the same location when the instruction | |
4878 | executes. You can use the same C expression for both operands, or | |
4879 | different expressions. For example, here we write the (fictitious) | |
4880 | @samp{combine} instruction with @code{bar} as its read-only source | |
4881 | operand and @code{foo} as its read-write destination: | |
c1f7febf | 4882 | |
3ab51846 | 4883 | @smallexample |
c1f7febf | 4884 | asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar)); |
3ab51846 | 4885 | @end smallexample |
c1f7febf RK |
4886 | |
4887 | @noindent | |
c85f7c16 | 4888 | The constraint @samp{"0"} for operand 1 says that it must occupy the |
84b72302 RH |
4889 | same location as operand 0. A number in constraint is allowed only in |
4890 | an input operand and it must refer to an output operand. | |
c1f7febf | 4891 | |
84b72302 | 4892 | Only a number in the constraint can guarantee that one operand will be in |
c85f7c16 JL |
4893 | the same place as another. The mere fact that @code{foo} is the value |
4894 | of both operands is not enough to guarantee that they will be in the | |
4895 | same place in the generated assembler code. The following would not | |
4896 | work reliably: | |
c1f7febf | 4897 | |
3ab51846 | 4898 | @smallexample |
c1f7febf | 4899 | asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar)); |
3ab51846 | 4900 | @end smallexample |
c1f7febf RK |
4901 | |
4902 | Various optimizations or reloading could cause operands 0 and 1 to be in | |
f0523f02 | 4903 | different registers; GCC knows no reason not to do so. For example, the |
c1f7febf RK |
4904 | compiler might find a copy of the value of @code{foo} in one register and |
4905 | use it for operand 1, but generate the output operand 0 in a different | |
4906 | register (copying it afterward to @code{foo}'s own address). Of course, | |
4907 | since the register for operand 1 is not even mentioned in the assembler | |
f0523f02 | 4908 | code, the result will not work, but GCC can't tell that. |
c1f7febf | 4909 | |
84b72302 RH |
4910 | As of GCC version 3.1, one may write @code{[@var{name}]} instead of |
4911 | the operand number for a matching constraint. For example: | |
4912 | ||
3ab51846 | 4913 | @smallexample |
84b72302 RH |
4914 | asm ("cmoveq %1,%2,%[result]" |
4915 | : [result] "=r"(result) | |
4916 | : "r" (test), "r"(new), "[result]"(old)); | |
3ab51846 | 4917 | @end smallexample |
84b72302 | 4918 | |
805c33df HPN |
4919 | Sometimes you need to make an @code{asm} operand be a specific register, |
4920 | but there's no matching constraint letter for that register @emph{by | |
4921 | itself}. To force the operand into that register, use a local variable | |
4922 | for the operand and specify the register in the variable declaration. | |
4923 | @xref{Explicit Reg Vars}. Then for the @code{asm} operand, use any | |
4924 | register constraint letter that matches the register: | |
4925 | ||
4926 | @smallexample | |
4927 | register int *p1 asm ("r0") = @dots{}; | |
4928 | register int *p2 asm ("r1") = @dots{}; | |
4929 | register int *result asm ("r0"); | |
4930 | asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2)); | |
4931 | @end smallexample | |
4932 | ||
b55d5746 HPN |
4933 | @anchor{Example of asm with clobbered asm reg} |
4934 | In the above example, beware that a register that is call-clobbered by | |
4935 | the target ABI will be overwritten by any function call in the | |
4936 | assignment, including library calls for arithmetic operators. | |
0c6390fa L |
4937 | Also a register may be clobbered when generating some operations, |
4938 | like variable shift, memory copy or memory move on x86. | |
b55d5746 HPN |
4939 | Assuming it is a call-clobbered register, this may happen to @code{r0} |
4940 | above by the assignment to @code{p2}. If you have to use such a | |
4941 | register, use temporary variables for expressions between the register | |
4942 | assignment and use: | |
4943 | ||
4944 | @smallexample | |
4945 | int t1 = @dots{}; | |
4946 | register int *p1 asm ("r0") = @dots{}; | |
4947 | register int *p2 asm ("r1") = t1; | |
4948 | register int *result asm ("r0"); | |
4949 | asm ("sysint" : "=r" (result) : "0" (p1), "r" (p2)); | |
4950 | @end smallexample | |
4951 | ||
c85f7c16 JL |
4952 | Some instructions clobber specific hard registers. To describe this, |
4953 | write a third colon after the input operands, followed by the names of | |
4954 | the clobbered hard registers (given as strings). Here is a realistic | |
4955 | example for the VAX: | |
c1f7febf | 4956 | |
3ab51846 | 4957 | @smallexample |
c1f7febf | 4958 | asm volatile ("movc3 %0,%1,%2" |
12bcfaa1 | 4959 | : /* @r{no outputs} */ |
c1f7febf RK |
4960 | : "g" (from), "g" (to), "g" (count) |
4961 | : "r0", "r1", "r2", "r3", "r4", "r5"); | |
3ab51846 | 4962 | @end smallexample |
c1f7febf | 4963 | |
c5c76735 JL |
4964 | You may not write a clobber description in a way that overlaps with an |
4965 | input or output operand. For example, you may not have an operand | |
4966 | describing a register class with one member if you mention that register | |
acb5d088 HPN |
4967 | in the clobber list. Variables declared to live in specific registers |
4968 | (@pxref{Explicit Reg Vars}), and used as asm input or output operands must | |
4969 | have no part mentioned in the clobber description. | |
4970 | There is no way for you to specify that an input | |
c5c76735 JL |
4971 | operand is modified without also specifying it as an output |
4972 | operand. Note that if all the output operands you specify are for this | |
4973 | purpose (and hence unused), you will then also need to specify | |
4974 | @code{volatile} for the @code{asm} construct, as described below, to | |
f0523f02 | 4975 | prevent GCC from deleting the @code{asm} statement as unused. |
8fe1938e | 4976 | |
c1f7febf | 4977 | If you refer to a particular hardware register from the assembler code, |
c85f7c16 JL |
4978 | you will probably have to list the register after the third colon to |
4979 | tell the compiler the register's value is modified. In some assemblers, | |
4980 | the register names begin with @samp{%}; to produce one @samp{%} in the | |
4981 | assembler code, you must write @samp{%%} in the input. | |
4982 | ||
4983 | If your assembler instruction can alter the condition code register, add | |
f0523f02 | 4984 | @samp{cc} to the list of clobbered registers. GCC on some machines |
c85f7c16 JL |
4985 | represents the condition codes as a specific hardware register; |
4986 | @samp{cc} serves to name this register. On other machines, the | |
4987 | condition code is handled differently, and specifying @samp{cc} has no | |
4988 | effect. But it is valid no matter what the machine. | |
c1f7febf | 4989 | |
bbf5a54d | 4990 | If your assembler instructions access memory in an unpredictable |
c85f7c16 | 4991 | fashion, add @samp{memory} to the list of clobbered registers. This |
bbf5a54d AJ |
4992 | will cause GCC to not keep memory values cached in registers across the |
4993 | assembler instruction and not optimize stores or loads to that memory. | |
4994 | You will also want to add the @code{volatile} keyword if the memory | |
4995 | affected is not listed in the inputs or outputs of the @code{asm}, as | |
4996 | the @samp{memory} clobber does not count as a side-effect of the | |
4997 | @code{asm}. If you know how large the accessed memory is, you can add | |
4998 | it as input or output but if this is not known, you should add | |
4999 | @samp{memory}. As an example, if you access ten bytes of a string, you | |
5000 | can use a memory input like: | |
5001 | ||
cd1a8088 | 5002 | @smallexample |
bbf5a54d | 5003 | @{"m"( (@{ struct @{ char x[10]; @} *p = (void *)ptr ; *p; @}) )@}. |
cd1a8088 | 5004 | @end smallexample |
bbf5a54d AJ |
5005 | |
5006 | Note that in the following example the memory input is necessary, | |
5007 | otherwise GCC might optimize the store to @code{x} away: | |
cd1a8088 | 5008 | @smallexample |
bbf5a54d AJ |
5009 | int foo () |
5010 | @{ | |
5011 | int x = 42; | |
5012 | int *y = &x; | |
5013 | int result; | |
5014 | asm ("magic stuff accessing an 'int' pointed to by '%1'" | |
5015 | "=&d" (r) : "a" (y), "m" (*y)); | |
f0eb93a8 | 5016 | return result; |
bbf5a54d | 5017 | @} |
cd1a8088 | 5018 | @end smallexample |
c1f7febf | 5019 | |
c85f7c16 | 5020 | You can put multiple assembler instructions together in a single |
8720914b HPN |
5021 | @code{asm} template, separated by the characters normally used in assembly |
5022 | code for the system. A combination that works in most places is a newline | |
5023 | to break the line, plus a tab character to move to the instruction field | |
5024 | (written as @samp{\n\t}). Sometimes semicolons can be used, if the | |
5025 | assembler allows semicolons as a line-breaking character. Note that some | |
5026 | assembler dialects use semicolons to start a comment. | |
5027 | The input operands are guaranteed not to use any of the clobbered | |
c85f7c16 JL |
5028 | registers, and neither will the output operands' addresses, so you can |
5029 | read and write the clobbered registers as many times as you like. Here | |
5030 | is an example of multiple instructions in a template; it assumes the | |
5031 | subroutine @code{_foo} accepts arguments in registers 9 and 10: | |
c1f7febf | 5032 | |
3ab51846 | 5033 | @smallexample |
8720914b | 5034 | asm ("movl %0,r9\n\tmovl %1,r10\n\tcall _foo" |
c1f7febf RK |
5035 | : /* no outputs */ |
5036 | : "g" (from), "g" (to) | |
5037 | : "r9", "r10"); | |
3ab51846 | 5038 | @end smallexample |
c1f7febf | 5039 | |
f0523f02 | 5040 | Unless an output operand has the @samp{&} constraint modifier, GCC |
c85f7c16 JL |
5041 | may allocate it in the same register as an unrelated input operand, on |
5042 | the assumption the inputs are consumed before the outputs are produced. | |
c1f7febf RK |
5043 | This assumption may be false if the assembler code actually consists of |
5044 | more than one instruction. In such a case, use @samp{&} for each output | |
c85f7c16 | 5045 | operand that may not overlap an input. @xref{Modifiers}. |
c1f7febf | 5046 | |
c85f7c16 JL |
5047 | If you want to test the condition code produced by an assembler |
5048 | instruction, you must include a branch and a label in the @code{asm} | |
5049 | construct, as follows: | |
c1f7febf | 5050 | |
3ab51846 | 5051 | @smallexample |
8720914b | 5052 | asm ("clr %0\n\tfrob %1\n\tbeq 0f\n\tmov #1,%0\n0:" |
c1f7febf RK |
5053 | : "g" (result) |
5054 | : "g" (input)); | |
3ab51846 | 5055 | @end smallexample |
c1f7febf RK |
5056 | |
5057 | @noindent | |
5058 | This assumes your assembler supports local labels, as the GNU assembler | |
5059 | and most Unix assemblers do. | |
5060 | ||
5061 | Speaking of labels, jumps from one @code{asm} to another are not | |
c85f7c16 JL |
5062 | supported. The compiler's optimizers do not know about these jumps, and |
5063 | therefore they cannot take account of them when deciding how to | |
c1f7febf RK |
5064 | optimize. |
5065 | ||
5066 | @cindex macros containing @code{asm} | |
5067 | Usually the most convenient way to use these @code{asm} instructions is to | |
5068 | encapsulate them in macros that look like functions. For example, | |
5069 | ||
3ab51846 | 5070 | @smallexample |
c1f7febf RK |
5071 | #define sin(x) \ |
5072 | (@{ double __value, __arg = (x); \ | |
5073 | asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \ | |
5074 | __value; @}) | |
3ab51846 | 5075 | @end smallexample |
c1f7febf RK |
5076 | |
5077 | @noindent | |
5078 | Here the variable @code{__arg} is used to make sure that the instruction | |
5079 | operates on a proper @code{double} value, and to accept only those | |
5080 | arguments @code{x} which can convert automatically to a @code{double}. | |
5081 | ||
c85f7c16 JL |
5082 | Another way to make sure the instruction operates on the correct data |
5083 | type is to use a cast in the @code{asm}. This is different from using a | |
c1f7febf RK |
5084 | variable @code{__arg} in that it converts more different types. For |
5085 | example, if the desired type were @code{int}, casting the argument to | |
5086 | @code{int} would accept a pointer with no complaint, while assigning the | |
5087 | argument to an @code{int} variable named @code{__arg} would warn about | |
5088 | using a pointer unless the caller explicitly casts it. | |
5089 | ||
f0523f02 | 5090 | If an @code{asm} has output operands, GCC assumes for optimization |
c85f7c16 JL |
5091 | purposes the instruction has no side effects except to change the output |
5092 | operands. This does not mean instructions with a side effect cannot be | |
5093 | used, but you must be careful, because the compiler may eliminate them | |
5094 | if the output operands aren't used, or move them out of loops, or | |
5095 | replace two with one if they constitute a common subexpression. Also, | |
5096 | if your instruction does have a side effect on a variable that otherwise | |
5097 | appears not to change, the old value of the variable may be reused later | |
5098 | if it happens to be found in a register. | |
c1f7febf | 5099 | |
2f59e40e DJ |
5100 | You can prevent an @code{asm} instruction from being deleted |
5101 | by writing the keyword @code{volatile} after | |
c1f7febf RK |
5102 | the @code{asm}. For example: |
5103 | ||
3ab51846 | 5104 | @smallexample |
310668e8 JM |
5105 | #define get_and_set_priority(new) \ |
5106 | (@{ int __old; \ | |
5107 | asm volatile ("get_and_set_priority %0, %1" \ | |
5108 | : "=g" (__old) : "g" (new)); \ | |
c85f7c16 | 5109 | __old; @}) |
3ab51846 | 5110 | @end smallexample |
c1f7febf RK |
5111 | |
5112 | @noindent | |
e71b34aa MM |
5113 | The @code{volatile} keyword indicates that the instruction has |
5114 | important side-effects. GCC will not delete a volatile @code{asm} if | |
5115 | it is reachable. (The instruction can still be deleted if GCC can | |
5116 | prove that control-flow will never reach the location of the | |
f0eb93a8 | 5117 | instruction.) Note that even a volatile @code{asm} instruction |
2f59e40e | 5118 | can be moved relative to other code, including across jump |
f0eb93a8 JM |
5119 | instructions. For example, on many targets there is a system |
5120 | register which can be set to control the rounding mode of | |
2f59e40e DJ |
5121 | floating point operations. You might try |
5122 | setting it with a volatile @code{asm}, like this PowerPC example: | |
e71b34aa | 5123 | |
3ab51846 | 5124 | @smallexample |
2f59e40e DJ |
5125 | asm volatile("mtfsf 255,%0" : : "f" (fpenv)); |
5126 | sum = x + y; | |
3ab51846 | 5127 | @end smallexample |
e71b34aa | 5128 | |
ebb48a4d | 5129 | @noindent |
2f59e40e DJ |
5130 | This will not work reliably, as the compiler may move the addition back |
5131 | before the volatile @code{asm}. To make it work you need to add an | |
5132 | artificial dependency to the @code{asm} referencing a variable in the code | |
5133 | you don't want moved, for example: | |
5134 | ||
5135 | @smallexample | |
5136 | asm volatile ("mtfsf 255,%1" : "=X"(sum): "f"(fpenv)); | |
5137 | sum = x + y; | |
5138 | @end smallexample | |
5139 | ||
5140 | Similarly, you can't expect a | |
5141 | sequence of volatile @code{asm} instructions to remain perfectly | |
5142 | consecutive. If you want consecutive output, use a single @code{asm}. | |
5143 | Also, GCC will perform some optimizations across a volatile @code{asm} | |
5144 | instruction; GCC does not ``forget everything'' when it encounters | |
5145 | a volatile @code{asm} instruction the way some other compilers do. | |
5146 | ||
5147 | An @code{asm} instruction without any output operands will be treated | |
5148 | identically to a volatile @code{asm} instruction. | |
c1f7febf RK |
5149 | |
5150 | It is a natural idea to look for a way to give access to the condition | |
5151 | code left by the assembler instruction. However, when we attempted to | |
5152 | implement this, we found no way to make it work reliably. The problem | |
5153 | is that output operands might need reloading, which would result in | |
5154 | additional following ``store'' instructions. On most machines, these | |
5155 | instructions would alter the condition code before there was time to | |
5156 | test it. This problem doesn't arise for ordinary ``test'' and | |
5157 | ``compare'' instructions because they don't have any output operands. | |
5158 | ||
eda3fbbe GB |
5159 | For reasons similar to those described above, it is not possible to give |
5160 | an assembler instruction access to the condition code left by previous | |
5161 | instructions. | |
5162 | ||
5490d604 | 5163 | If you are writing a header file that should be includable in ISO C |
c1f7febf RK |
5164 | programs, write @code{__asm__} instead of @code{asm}. @xref{Alternate |
5165 | Keywords}. | |
5166 | ||
ece7fc1c RE |
5167 | @subsection Size of an @code{asm} |
5168 | ||
5169 | Some targets require that GCC track the size of each instruction used in | |
5170 | order to generate correct code. Because the final length of an | |
5171 | @code{asm} is only known by the assembler, GCC must make an estimate as | |
5172 | to how big it will be. The estimate is formed by counting the number of | |
5173 | statements in the pattern of the @code{asm} and multiplying that by the | |
5174 | length of the longest instruction on that processor. Statements in the | |
5175 | @code{asm} are identified by newline characters and whatever statement | |
5176 | separator characters are supported by the assembler; on most processors | |
5177 | this is the `@code{;}' character. | |
5178 | ||
5179 | Normally, GCC's estimate is perfectly adequate to ensure that correct | |
5180 | code is generated, but it is possible to confuse the compiler if you use | |
5181 | pseudo instructions or assembler macros that expand into multiple real | |
5182 | instructions or if you use assembler directives that expand to more | |
5183 | space in the object file than would be needed for a single instruction. | |
5184 | If this happens then the assembler will produce a diagnostic saying that | |
5185 | a label is unreachable. | |
5186 | ||
fe0ce426 JH |
5187 | @subsection i386 floating point asm operands |
5188 | ||
5189 | There are several rules on the usage of stack-like regs in | |
5190 | asm_operands insns. These rules apply only to the operands that are | |
5191 | stack-like regs: | |
5192 | ||
5193 | @enumerate | |
5194 | @item | |
5195 | Given a set of input regs that die in an asm_operands, it is | |
5196 | necessary to know which are implicitly popped by the asm, and | |
5197 | which must be explicitly popped by gcc. | |
5198 | ||
5199 | An input reg that is implicitly popped by the asm must be | |
5200 | explicitly clobbered, unless it is constrained to match an | |
5201 | output operand. | |
5202 | ||
5203 | @item | |
5204 | For any input reg that is implicitly popped by an asm, it is | |
5205 | necessary to know how to adjust the stack to compensate for the pop. | |
5206 | If any non-popped input is closer to the top of the reg-stack than | |
5207 | the implicitly popped reg, it would not be possible to know what the | |
84330467 | 5208 | stack looked like---it's not clear how the rest of the stack ``slides |
fe0ce426 JH |
5209 | up''. |
5210 | ||
5211 | All implicitly popped input regs must be closer to the top of | |
5212 | the reg-stack than any input that is not implicitly popped. | |
5213 | ||
5214 | It is possible that if an input dies in an insn, reload might | |
5215 | use the input reg for an output reload. Consider this example: | |
5216 | ||
3ab51846 | 5217 | @smallexample |
fe0ce426 | 5218 | asm ("foo" : "=t" (a) : "f" (b)); |
3ab51846 | 5219 | @end smallexample |
fe0ce426 JH |
5220 | |
5221 | This asm says that input B is not popped by the asm, and that | |
c771326b | 5222 | the asm pushes a result onto the reg-stack, i.e., the stack is one |
fe0ce426 JH |
5223 | deeper after the asm than it was before. But, it is possible that |
5224 | reload will think that it can use the same reg for both the input and | |
5225 | the output, if input B dies in this insn. | |
5226 | ||
5227 | If any input operand uses the @code{f} constraint, all output reg | |
5228 | constraints must use the @code{&} earlyclobber. | |
5229 | ||
5230 | The asm above would be written as | |
5231 | ||
3ab51846 | 5232 | @smallexample |
fe0ce426 | 5233 | asm ("foo" : "=&t" (a) : "f" (b)); |
3ab51846 | 5234 | @end smallexample |
fe0ce426 JH |
5235 | |
5236 | @item | |
5237 | Some operands need to be in particular places on the stack. All | |
84330467 | 5238 | output operands fall in this category---there is no other way to |
fe0ce426 JH |
5239 | know which regs the outputs appear in unless the user indicates |
5240 | this in the constraints. | |
5241 | ||
5242 | Output operands must specifically indicate which reg an output | |
5243 | appears in after an asm. @code{=f} is not allowed: the operand | |
5244 | constraints must select a class with a single reg. | |
5245 | ||
5246 | @item | |
5247 | Output operands may not be ``inserted'' between existing stack regs. | |
5248 | Since no 387 opcode uses a read/write operand, all output operands | |
5249 | are dead before the asm_operands, and are pushed by the asm_operands. | |
5250 | It makes no sense to push anywhere but the top of the reg-stack. | |
5251 | ||
5252 | Output operands must start at the top of the reg-stack: output | |
5253 | operands may not ``skip'' a reg. | |
5254 | ||
5255 | @item | |
5256 | Some asm statements may need extra stack space for internal | |
5257 | calculations. This can be guaranteed by clobbering stack registers | |
5258 | unrelated to the inputs and outputs. | |
5259 | ||
5260 | @end enumerate | |
5261 | ||
5262 | Here are a couple of reasonable asms to want to write. This asm | |
5263 | takes one input, which is internally popped, and produces two outputs. | |
5264 | ||
3ab51846 | 5265 | @smallexample |
fe0ce426 | 5266 | asm ("fsincos" : "=t" (cos), "=u" (sin) : "0" (inp)); |
3ab51846 | 5267 | @end smallexample |
fe0ce426 JH |
5268 | |
5269 | This asm takes two inputs, which are popped by the @code{fyl2xp1} opcode, | |
5270 | and replaces them with one output. The user must code the @code{st(1)} | |
5271 | clobber for reg-stack.c to know that @code{fyl2xp1} pops both inputs. | |
5272 | ||
3ab51846 | 5273 | @smallexample |
fe0ce426 | 5274 | asm ("fyl2xp1" : "=t" (result) : "0" (x), "u" (y) : "st(1)"); |
3ab51846 | 5275 | @end smallexample |
fe0ce426 | 5276 | |
c1f7febf | 5277 | @include md.texi |
c1f7febf RK |
5278 | |
5279 | @node Asm Labels | |
5280 | @section Controlling Names Used in Assembler Code | |
5281 | @cindex assembler names for identifiers | |
5282 | @cindex names used in assembler code | |
5283 | @cindex identifiers, names in assembler code | |
5284 | ||
5285 | You can specify the name to be used in the assembler code for a C | |
5286 | function or variable by writing the @code{asm} (or @code{__asm__}) | |
5287 | keyword after the declarator as follows: | |
5288 | ||
3ab51846 | 5289 | @smallexample |
c1f7febf | 5290 | int foo asm ("myfoo") = 2; |
3ab51846 | 5291 | @end smallexample |
c1f7febf RK |
5292 | |
5293 | @noindent | |
5294 | This specifies that the name to be used for the variable @code{foo} in | |
5295 | the assembler code should be @samp{myfoo} rather than the usual | |
5296 | @samp{_foo}. | |
5297 | ||
5298 | On systems where an underscore is normally prepended to the name of a C | |
5299 | function or variable, this feature allows you to define names for the | |
5300 | linker that do not start with an underscore. | |
5301 | ||
0adc3c19 MM |
5302 | It does not make sense to use this feature with a non-static local |
5303 | variable since such variables do not have assembler names. If you are | |
5304 | trying to put the variable in a particular register, see @ref{Explicit | |
5305 | Reg Vars}. GCC presently accepts such code with a warning, but will | |
5306 | probably be changed to issue an error, rather than a warning, in the | |
5307 | future. | |
5308 | ||
c1f7febf RK |
5309 | You cannot use @code{asm} in this way in a function @emph{definition}; but |
5310 | you can get the same effect by writing a declaration for the function | |
5311 | before its definition and putting @code{asm} there, like this: | |
5312 | ||
3ab51846 | 5313 | @smallexample |
c1f7febf RK |
5314 | extern func () asm ("FUNC"); |
5315 | ||
5316 | func (x, y) | |
5317 | int x, y; | |
0d893a63 | 5318 | /* @r{@dots{}} */ |
3ab51846 | 5319 | @end smallexample |
c1f7febf RK |
5320 | |
5321 | It is up to you to make sure that the assembler names you choose do not | |
5322 | conflict with any other assembler symbols. Also, you must not use a | |
f0523f02 JM |
5323 | register name; that would produce completely invalid assembler code. GCC |
5324 | does not as yet have the ability to store static variables in registers. | |
c1f7febf RK |
5325 | Perhaps that will be added. |
5326 | ||
5327 | @node Explicit Reg Vars | |
5328 | @section Variables in Specified Registers | |
5329 | @cindex explicit register variables | |
5330 | @cindex variables in specified registers | |
5331 | @cindex specified registers | |
5332 | @cindex registers, global allocation | |
5333 | ||
5334 | GNU C allows you to put a few global variables into specified hardware | |
5335 | registers. You can also specify the register in which an ordinary | |
5336 | register variable should be allocated. | |
5337 | ||
5338 | @itemize @bullet | |
5339 | @item | |
5340 | Global register variables reserve registers throughout the program. | |
5341 | This may be useful in programs such as programming language | |
5342 | interpreters which have a couple of global variables that are accessed | |
5343 | very often. | |
5344 | ||
5345 | @item | |
5346 | Local register variables in specific registers do not reserve the | |
805c33df HPN |
5347 | registers, except at the point where they are used as input or output |
5348 | operands in an @code{asm} statement and the @code{asm} statement itself is | |
5349 | not deleted. The compiler's data flow analysis is capable of determining | |
c1f7febf | 5350 | where the specified registers contain live values, and where they are |
8d344fbc | 5351 | available for other uses. Stores into local register variables may be deleted |
0deaf590 JL |
5352 | when they appear to be dead according to dataflow analysis. References |
5353 | to local register variables may be deleted or moved or simplified. | |
c1f7febf RK |
5354 | |
5355 | These local variables are sometimes convenient for use with the extended | |
5356 | @code{asm} feature (@pxref{Extended Asm}), if you want to write one | |
5357 | output of the assembler instruction directly into a particular register. | |
5358 | (This will work provided the register you specify fits the constraints | |
5359 | specified for that operand in the @code{asm}.) | |
5360 | @end itemize | |
5361 | ||
5362 | @menu | |
5363 | * Global Reg Vars:: | |
5364 | * Local Reg Vars:: | |
5365 | @end menu | |
5366 | ||
5367 | @node Global Reg Vars | |
5368 | @subsection Defining Global Register Variables | |
5369 | @cindex global register variables | |
5370 | @cindex registers, global variables in | |
5371 | ||
5372 | You can define a global register variable in GNU C like this: | |
5373 | ||
3ab51846 | 5374 | @smallexample |
c1f7febf | 5375 | register int *foo asm ("a5"); |
3ab51846 | 5376 | @end smallexample |
c1f7febf RK |
5377 | |
5378 | @noindent | |
5379 | Here @code{a5} is the name of the register which should be used. Choose a | |
5380 | register which is normally saved and restored by function calls on your | |
5381 | machine, so that library routines will not clobber it. | |
5382 | ||
5383 | Naturally the register name is cpu-dependent, so you would need to | |
5384 | conditionalize your program according to cpu type. The register | |
5385 | @code{a5} would be a good choice on a 68000 for a variable of pointer | |
5386 | type. On machines with register windows, be sure to choose a ``global'' | |
5387 | register that is not affected magically by the function call mechanism. | |
5388 | ||
5389 | In addition, operating systems on one type of cpu may differ in how they | |
5390 | name the registers; then you would need additional conditionals. For | |
5391 | example, some 68000 operating systems call this register @code{%a5}. | |
5392 | ||
5393 | Eventually there may be a way of asking the compiler to choose a register | |
5394 | automatically, but first we need to figure out how it should choose and | |
5395 | how to enable you to guide the choice. No solution is evident. | |
5396 | ||
5397 | Defining a global register variable in a certain register reserves that | |
5398 | register entirely for this use, at least within the current compilation. | |
5399 | The register will not be allocated for any other purpose in the functions | |
5400 | in the current compilation. The register will not be saved and restored by | |
5401 | these functions. Stores into this register are never deleted even if they | |
5402 | would appear to be dead, but references may be deleted or moved or | |
5403 | simplified. | |
5404 | ||
5405 | It is not safe to access the global register variables from signal | |
5406 | handlers, or from more than one thread of control, because the system | |
5407 | library routines may temporarily use the register for other things (unless | |
5408 | you recompile them specially for the task at hand). | |
5409 | ||
5410 | @cindex @code{qsort}, and global register variables | |
5411 | It is not safe for one function that uses a global register variable to | |
5412 | call another such function @code{foo} by way of a third function | |
e979f9e8 | 5413 | @code{lose} that was compiled without knowledge of this variable (i.e.@: in a |
c1f7febf RK |
5414 | different source file in which the variable wasn't declared). This is |
5415 | because @code{lose} might save the register and put some other value there. | |
5416 | For example, you can't expect a global register variable to be available in | |
5417 | the comparison-function that you pass to @code{qsort}, since @code{qsort} | |
5418 | might have put something else in that register. (If you are prepared to | |
5419 | recompile @code{qsort} with the same global register variable, you can | |
5420 | solve this problem.) | |
5421 | ||
5422 | If you want to recompile @code{qsort} or other source files which do not | |
5423 | actually use your global register variable, so that they will not use that | |
5424 | register for any other purpose, then it suffices to specify the compiler | |
84330467 | 5425 | option @option{-ffixed-@var{reg}}. You need not actually add a global |
c1f7febf RK |
5426 | register declaration to their source code. |
5427 | ||
5428 | A function which can alter the value of a global register variable cannot | |
5429 | safely be called from a function compiled without this variable, because it | |
5430 | could clobber the value the caller expects to find there on return. | |
5431 | Therefore, the function which is the entry point into the part of the | |
5432 | program that uses the global register variable must explicitly save and | |
5433 | restore the value which belongs to its caller. | |
5434 | ||
5435 | @cindex register variable after @code{longjmp} | |
5436 | @cindex global register after @code{longjmp} | |
5437 | @cindex value after @code{longjmp} | |
5438 | @findex longjmp | |
5439 | @findex setjmp | |
5440 | On most machines, @code{longjmp} will restore to each global register | |
5441 | variable the value it had at the time of the @code{setjmp}. On some | |
5442 | machines, however, @code{longjmp} will not change the value of global | |
5443 | register variables. To be portable, the function that called @code{setjmp} | |
5444 | should make other arrangements to save the values of the global register | |
5445 | variables, and to restore them in a @code{longjmp}. This way, the same | |
5446 | thing will happen regardless of what @code{longjmp} does. | |
5447 | ||
5448 | All global register variable declarations must precede all function | |
5449 | definitions. If such a declaration could appear after function | |
5450 | definitions, the declaration would be too late to prevent the register from | |
5451 | being used for other purposes in the preceding functions. | |
5452 | ||
5453 | Global register variables may not have initial values, because an | |
5454 | executable file has no means to supply initial contents for a register. | |
5455 | ||
981f6289 | 5456 | On the SPARC, there are reports that g3 @dots{} g7 are suitable |
c1f7febf RK |
5457 | registers, but certain library functions, such as @code{getwd}, as well |
5458 | as the subroutines for division and remainder, modify g3 and g4. g1 and | |
5459 | g2 are local temporaries. | |
5460 | ||
5461 | On the 68000, a2 @dots{} a5 should be suitable, as should d2 @dots{} d7. | |
5462 | Of course, it will not do to use more than a few of those. | |
5463 | ||
5464 | @node Local Reg Vars | |
5465 | @subsection Specifying Registers for Local Variables | |
5466 | @cindex local variables, specifying registers | |
5467 | @cindex specifying registers for local variables | |
5468 | @cindex registers for local variables | |
5469 | ||
5470 | You can define a local register variable with a specified register | |
5471 | like this: | |
5472 | ||
3ab51846 | 5473 | @smallexample |
c1f7febf | 5474 | register int *foo asm ("a5"); |
3ab51846 | 5475 | @end smallexample |
c1f7febf RK |
5476 | |
5477 | @noindent | |
5478 | Here @code{a5} is the name of the register which should be used. Note | |
5479 | that this is the same syntax used for defining global register | |
5480 | variables, but for a local variable it would appear within a function. | |
5481 | ||
5482 | Naturally the register name is cpu-dependent, but this is not a | |
5483 | problem, since specific registers are most often useful with explicit | |
5484 | assembler instructions (@pxref{Extended Asm}). Both of these things | |
5485 | generally require that you conditionalize your program according to | |
5486 | cpu type. | |
5487 | ||
5488 | In addition, operating systems on one type of cpu may differ in how they | |
5489 | name the registers; then you would need additional conditionals. For | |
5490 | example, some 68000 operating systems call this register @code{%a5}. | |
5491 | ||
c1f7febf RK |
5492 | Defining such a register variable does not reserve the register; it |
5493 | remains available for other uses in places where flow control determines | |
d754127f | 5494 | the variable's value is not live. |
e5e809f4 | 5495 | |
f0523f02 | 5496 | This option does not guarantee that GCC will generate code that has |
e5e809f4 | 5497 | this variable in the register you specify at all times. You may not |
805c33df HPN |
5498 | code an explicit reference to this register in the @emph{assembler |
5499 | instruction template} part of an @code{asm} statement and assume it will | |
5500 | always refer to this variable. However, using the variable as an | |
5501 | @code{asm} @emph{operand} guarantees that the specified register is used | |
5502 | for the operand. | |
c1f7febf | 5503 | |
8d344fbc | 5504 | Stores into local register variables may be deleted when they appear to be dead |
0deaf590 JL |
5505 | according to dataflow analysis. References to local register variables may |
5506 | be deleted or moved or simplified. | |
5507 | ||
b55d5746 HPN |
5508 | As for global register variables, it's recommended that you choose a |
5509 | register which is normally saved and restored by function calls on | |
5510 | your machine, so that library routines will not clobber it. A common | |
5511 | pitfall is to initialize multiple call-clobbered registers with | |
5512 | arbitrary expressions, where a function call or library call for an | |
5513 | arithmetic operator will overwrite a register value from a previous | |
5514 | assignment, for example @code{r0} below: | |
5515 | @smallexample | |
5516 | register int *p1 asm ("r0") = @dots{}; | |
5517 | register int *p2 asm ("r1") = @dots{}; | |
5518 | @end smallexample | |
5519 | In those cases, a solution is to use a temporary variable for | |
5520 | each arbitrary expression. @xref{Example of asm with clobbered asm reg}. | |
5521 | ||
c1f7febf RK |
5522 | @node Alternate Keywords |
5523 | @section Alternate Keywords | |
5524 | @cindex alternate keywords | |
5525 | @cindex keywords, alternate | |
5526 | ||
5490d604 | 5527 | @option{-ansi} and the various @option{-std} options disable certain |
f458d1d5 ZW |
5528 | keywords. This causes trouble when you want to use GNU C extensions, or |
5529 | a general-purpose header file that should be usable by all programs, | |
5530 | including ISO C programs. The keywords @code{asm}, @code{typeof} and | |
5531 | @code{inline} are not available in programs compiled with | |
5532 | @option{-ansi} or @option{-std} (although @code{inline} can be used in a | |
5533 | program compiled with @option{-std=c99}). The ISO C99 keyword | |
5490d604 JM |
5534 | @code{restrict} is only available when @option{-std=gnu99} (which will |
5535 | eventually be the default) or @option{-std=c99} (or the equivalent | |
bd819a4a | 5536 | @option{-std=iso9899:1999}) is used. |
c1f7febf RK |
5537 | |
5538 | The way to solve these problems is to put @samp{__} at the beginning and | |
5539 | end of each problematical keyword. For example, use @code{__asm__} | |
f458d1d5 | 5540 | instead of @code{asm}, and @code{__inline__} instead of @code{inline}. |
c1f7febf RK |
5541 | |
5542 | Other C compilers won't accept these alternative keywords; if you want to | |
5543 | compile with another compiler, you can define the alternate keywords as | |
5544 | macros to replace them with the customary keywords. It looks like this: | |
5545 | ||
3ab51846 | 5546 | @smallexample |
c1f7febf RK |
5547 | #ifndef __GNUC__ |
5548 | #define __asm__ asm | |
5549 | #endif | |
3ab51846 | 5550 | @end smallexample |
c1f7febf | 5551 | |
6e6b0525 | 5552 | @findex __extension__ |
84330467 JM |
5553 | @opindex pedantic |
5554 | @option{-pedantic} and other options cause warnings for many GNU C extensions. | |
dbe519e0 | 5555 | You can |
c1f7febf RK |
5556 | prevent such warnings within one expression by writing |
5557 | @code{__extension__} before the expression. @code{__extension__} has no | |
5558 | effect aside from this. | |
5559 | ||
5560 | @node Incomplete Enums | |
5561 | @section Incomplete @code{enum} Types | |
5562 | ||
5563 | You can define an @code{enum} tag without specifying its possible values. | |
5564 | This results in an incomplete type, much like what you get if you write | |
5565 | @code{struct foo} without describing the elements. A later declaration | |
5566 | which does specify the possible values completes the type. | |
5567 | ||
5568 | You can't allocate variables or storage using the type while it is | |
5569 | incomplete. However, you can work with pointers to that type. | |
5570 | ||
5571 | This extension may not be very useful, but it makes the handling of | |
5572 | @code{enum} more consistent with the way @code{struct} and @code{union} | |
5573 | are handled. | |
5574 | ||
5575 | This extension is not supported by GNU C++. | |
5576 | ||
5577 | @node Function Names | |
5578 | @section Function Names as Strings | |
e6cc3a24 | 5579 | @cindex @code{__func__} identifier |
4b404517 JM |
5580 | @cindex @code{__FUNCTION__} identifier |
5581 | @cindex @code{__PRETTY_FUNCTION__} identifier | |
c1f7febf | 5582 | |
e6cc3a24 ZW |
5583 | GCC provides three magic variables which hold the name of the current |
5584 | function, as a string. The first of these is @code{__func__}, which | |
5585 | is part of the C99 standard: | |
5586 | ||
e6cc3a24 ZW |
5587 | The identifier @code{__func__} is implicitly declared by the translator |
5588 | as if, immediately following the opening brace of each function | |
5589 | definition, the declaration | |
5590 | ||
5591 | @smallexample | |
5592 | static const char __func__[] = "function-name"; | |
5593 | @end smallexample | |
c1f7febf | 5594 | |
38bb2b65 | 5595 | @noindent |
e6cc3a24 ZW |
5596 | appeared, where function-name is the name of the lexically-enclosing |
5597 | function. This name is the unadorned name of the function. | |
e6cc3a24 ZW |
5598 | |
5599 | @code{__FUNCTION__} is another name for @code{__func__}. Older | |
5600 | versions of GCC recognize only this name. However, it is not | |
5601 | standardized. For maximum portability, we recommend you use | |
5602 | @code{__func__}, but provide a fallback definition with the | |
5603 | preprocessor: | |
5604 | ||
5605 | @smallexample | |
5606 | #if __STDC_VERSION__ < 199901L | |
5607 | # if __GNUC__ >= 2 | |
5608 | # define __func__ __FUNCTION__ | |
5609 | # else | |
5610 | # define __func__ "<unknown>" | |
5611 | # endif | |
5612 | #endif | |
5613 | @end smallexample | |
5614 | ||
5615 | In C, @code{__PRETTY_FUNCTION__} is yet another name for | |
5616 | @code{__func__}. However, in C++, @code{__PRETTY_FUNCTION__} contains | |
5617 | the type signature of the function as well as its bare name. For | |
5618 | example, this program: | |
c1f7febf RK |
5619 | |
5620 | @smallexample | |
5621 | extern "C" @{ | |
5622 | extern int printf (char *, ...); | |
5623 | @} | |
5624 | ||
5625 | class a @{ | |
5626 | public: | |
a721a601 | 5627 | void sub (int i) |
c1f7febf RK |
5628 | @{ |
5629 | printf ("__FUNCTION__ = %s\n", __FUNCTION__); | |
5630 | printf ("__PRETTY_FUNCTION__ = %s\n", __PRETTY_FUNCTION__); | |
5631 | @} | |
5632 | @}; | |
5633 | ||
5634 | int | |
5635 | main (void) | |
5636 | @{ | |
5637 | a ax; | |
5638 | ax.sub (0); | |
5639 | return 0; | |
5640 | @} | |
5641 | @end smallexample | |
5642 | ||
5643 | @noindent | |
5644 | gives this output: | |
5645 | ||
5646 | @smallexample | |
5647 | __FUNCTION__ = sub | |
e6cc3a24 | 5648 | __PRETTY_FUNCTION__ = void a::sub(int) |
22acfb79 NM |
5649 | @end smallexample |
5650 | ||
e6cc3a24 ZW |
5651 | These identifiers are not preprocessor macros. In GCC 3.3 and |
5652 | earlier, in C only, @code{__FUNCTION__} and @code{__PRETTY_FUNCTION__} | |
5653 | were treated as string literals; they could be used to initialize | |
5654 | @code{char} arrays, and they could be concatenated with other string | |
5655 | literals. GCC 3.4 and later treat them as variables, like | |
5656 | @code{__func__}. In C++, @code{__FUNCTION__} and | |
5657 | @code{__PRETTY_FUNCTION__} have always been variables. | |
22acfb79 | 5658 | |
c1f7febf RK |
5659 | @node Return Address |
5660 | @section Getting the Return or Frame Address of a Function | |
5661 | ||
5662 | These functions may be used to get information about the callers of a | |
5663 | function. | |
5664 | ||
84330467 | 5665 | @deftypefn {Built-in Function} {void *} __builtin_return_address (unsigned int @var{level}) |
c1f7febf RK |
5666 | This function returns the return address of the current function, or of |
5667 | one of its callers. The @var{level} argument is number of frames to | |
5668 | scan up the call stack. A value of @code{0} yields the return address | |
5669 | of the current function, a value of @code{1} yields the return address | |
8a36672b | 5670 | of the caller of the current function, and so forth. When inlining |
95b1627e EC |
5671 | the expected behavior is that the function will return the address of |
5672 | the function that will be returned to. To work around this behavior use | |
5673 | the @code{noinline} function attribute. | |
c1f7febf RK |
5674 | |
5675 | The @var{level} argument must be a constant integer. | |
5676 | ||
5677 | On some machines it may be impossible to determine the return address of | |
5678 | any function other than the current one; in such cases, or when the top | |
dd96fbc5 | 5679 | of the stack has been reached, this function will return @code{0} or a |
8a36672b | 5680 | random value. In addition, @code{__builtin_frame_address} may be used |
dd96fbc5 | 5681 | to determine if the top of the stack has been reached. |
c1f7febf | 5682 | |
df2a54e9 | 5683 | This function should only be used with a nonzero argument for debugging |
c1f7febf | 5684 | purposes. |
84330467 | 5685 | @end deftypefn |
c1f7febf | 5686 | |
84330467 | 5687 | @deftypefn {Built-in Function} {void *} __builtin_frame_address (unsigned int @var{level}) |
c1f7febf RK |
5688 | This function is similar to @code{__builtin_return_address}, but it |
5689 | returns the address of the function frame rather than the return address | |
5690 | of the function. Calling @code{__builtin_frame_address} with a value of | |
5691 | @code{0} yields the frame address of the current function, a value of | |
5692 | @code{1} yields the frame address of the caller of the current function, | |
5693 | and so forth. | |
5694 | ||
5695 | The frame is the area on the stack which holds local variables and saved | |
5696 | registers. The frame address is normally the address of the first word | |
5697 | pushed on to the stack by the function. However, the exact definition | |
5698 | depends upon the processor and the calling convention. If the processor | |
5699 | has a dedicated frame pointer register, and the function has a frame, | |
5700 | then @code{__builtin_frame_address} will return the value of the frame | |
5701 | pointer register. | |
5702 | ||
dd96fbc5 L |
5703 | On some machines it may be impossible to determine the frame address of |
5704 | any function other than the current one; in such cases, or when the top | |
5705 | of the stack has been reached, this function will return @code{0} if | |
5706 | the first frame pointer is properly initialized by the startup code. | |
5707 | ||
df2a54e9 | 5708 | This function should only be used with a nonzero argument for debugging |
dd96fbc5 | 5709 | purposes. |
84330467 | 5710 | @end deftypefn |
c1f7febf | 5711 | |
1255c85c BS |
5712 | @node Vector Extensions |
5713 | @section Using vector instructions through built-in functions | |
5714 | ||
5715 | On some targets, the instruction set contains SIMD vector instructions that | |
5716 | operate on multiple values contained in one large register at the same time. | |
5717 | For example, on the i386 the MMX, 3Dnow! and SSE extensions can be used | |
5718 | this way. | |
5719 | ||
5720 | The first step in using these extensions is to provide the necessary data | |
5721 | types. This should be done using an appropriate @code{typedef}: | |
5722 | ||
3ab51846 | 5723 | @smallexample |
4a5eab38 | 5724 | typedef int v4si __attribute__ ((vector_size (16))); |
3ab51846 | 5725 | @end smallexample |
1255c85c | 5726 | |
4a5eab38 PB |
5727 | The @code{int} type specifies the base type, while the attribute specifies |
5728 | the vector size for the variable, measured in bytes. For example, the | |
5729 | declaration above causes the compiler to set the mode for the @code{v4si} | |
5730 | type to be 16 bytes wide and divided into @code{int} sized units. For | |
5731 | a 32-bit @code{int} this means a vector of 4 units of 4 bytes, and the | |
5732 | corresponding mode of @code{foo} will be @acronym{V4SI}. | |
1255c85c | 5733 | |
4a5eab38 PB |
5734 | The @code{vector_size} attribute is only applicable to integral and |
5735 | float scalars, although arrays, pointers, and function return values | |
5736 | are allowed in conjunction with this construct. | |
5737 | ||
5738 | All the basic integer types can be used as base types, both as signed | |
5739 | and as unsigned: @code{char}, @code{short}, @code{int}, @code{long}, | |
5740 | @code{long long}. In addition, @code{float} and @code{double} can be | |
5741 | used to build floating-point vector types. | |
1255c85c | 5742 | |
cb2a532e | 5743 | Specifying a combination that is not valid for the current architecture |
2dd76960 | 5744 | will cause GCC to synthesize the instructions using a narrower mode. |
cb2a532e | 5745 | For example, if you specify a variable of type @code{V4SI} and your |
2dd76960 | 5746 | architecture does not allow for this specific SIMD type, GCC will |
cb2a532e AH |
5747 | produce code that uses 4 @code{SIs}. |
5748 | ||
5749 | The types defined in this manner can be used with a subset of normal C | |
2dd76960 | 5750 | operations. Currently, GCC will allow using the following operators |
5cfd5d9b | 5751 | on these types: @code{+, -, *, /, unary minus, ^, |, &, ~, %}@. |
cb2a532e AH |
5752 | |
5753 | The operations behave like C++ @code{valarrays}. Addition is defined as | |
5754 | the addition of the corresponding elements of the operands. For | |
5755 | example, in the code below, each of the 4 elements in @var{a} will be | |
5756 | added to the corresponding 4 elements in @var{b} and the resulting | |
5757 | vector will be stored in @var{c}. | |
5758 | ||
3ab51846 | 5759 | @smallexample |
4a5eab38 | 5760 | typedef int v4si __attribute__ ((vector_size (16))); |
cb2a532e AH |
5761 | |
5762 | v4si a, b, c; | |
5763 | ||
5764 | c = a + b; | |
3ab51846 | 5765 | @end smallexample |
cb2a532e | 5766 | |
3a3e1600 GK |
5767 | Subtraction, multiplication, division, and the logical operations |
5768 | operate in a similar manner. Likewise, the result of using the unary | |
5769 | minus or complement operators on a vector type is a vector whose | |
5770 | elements are the negative or complemented values of the corresponding | |
cb2a532e AH |
5771 | elements in the operand. |
5772 | ||
5773 | You can declare variables and use them in function calls and returns, as | |
5774 | well as in assignments and some casts. You can specify a vector type as | |
5775 | a return type for a function. Vector types can also be used as function | |
5776 | arguments. It is possible to cast from one vector type to another, | |
5777 | provided they are of the same size (in fact, you can also cast vectors | |
5778 | to and from other datatypes of the same size). | |
5779 | ||
5780 | You cannot operate between vectors of different lengths or different | |
90a21764 | 5781 | signedness without a cast. |
cb2a532e AH |
5782 | |
5783 | A port that supports hardware vector operations, usually provides a set | |
5784 | of built-in functions that can be used to operate on vectors. For | |
5785 | example, a function to add two vectors and multiply the result by a | |
5786 | third could look like this: | |
1255c85c | 5787 | |
3ab51846 | 5788 | @smallexample |
1255c85c BS |
5789 | v4si f (v4si a, v4si b, v4si c) |
5790 | @{ | |
5791 | v4si tmp = __builtin_addv4si (a, b); | |
5792 | return __builtin_mulv4si (tmp, c); | |
5793 | @} | |
5794 | ||
3ab51846 | 5795 | @end smallexample |
1255c85c | 5796 | |
7a3ea201 RH |
5797 | @node Offsetof |
5798 | @section Offsetof | |
5799 | @findex __builtin_offsetof | |
5800 | ||
5801 | GCC implements for both C and C++ a syntactic extension to implement | |
5802 | the @code{offsetof} macro. | |
5803 | ||
5804 | @smallexample | |
5805 | primary: | |
6ccde948 | 5806 | "__builtin_offsetof" "(" @code{typename} "," offsetof_member_designator ")" |
7a3ea201 RH |
5807 | |
5808 | offsetof_member_designator: | |
6ccde948 RW |
5809 | @code{identifier} |
5810 | | offsetof_member_designator "." @code{identifier} | |
5811 | | offsetof_member_designator "[" @code{expr} "]" | |
7a3ea201 RH |
5812 | @end smallexample |
5813 | ||
5814 | This extension is sufficient such that | |
5815 | ||
5816 | @smallexample | |
5817 | #define offsetof(@var{type}, @var{member}) __builtin_offsetof (@var{type}, @var{member}) | |
5818 | @end smallexample | |
5819 | ||
5820 | is a suitable definition of the @code{offsetof} macro. In C++, @var{type} | |
5821 | may be dependent. In either case, @var{member} may consist of a single | |
5822 | identifier, or a sequence of member accesses and array references. | |
5823 | ||
48ae6c13 RH |
5824 | @node Atomic Builtins |
5825 | @section Built-in functions for atomic memory access | |
5826 | ||
5827 | The following builtins are intended to be compatible with those described | |
5828 | in the @cite{Intel Itanium Processor-specific Application Binary Interface}, | |
5829 | section 7.4. As such, they depart from the normal GCC practice of using | |
5830 | the ``__builtin_'' prefix, and further that they are overloaded such that | |
5831 | they work on multiple types. | |
5832 | ||
5833 | The definition given in the Intel documentation allows only for the use of | |
5834 | the types @code{int}, @code{long}, @code{long long} as well as their unsigned | |
5835 | counterparts. GCC will allow any integral scalar or pointer type that is | |
5836 | 1, 2, 4 or 8 bytes in length. | |
5837 | ||
5838 | Not all operations are supported by all target processors. If a particular | |
5839 | operation cannot be implemented on the target processor, a warning will be | |
5840 | generated and a call an external function will be generated. The external | |
5841 | function will carry the same name as the builtin, with an additional suffix | |
5842 | @samp{_@var{n}} where @var{n} is the size of the data type. | |
5843 | ||
5844 | @c ??? Should we have a mechanism to suppress this warning? This is almost | |
5845 | @c useful for implementing the operation under the control of an external | |
5846 | @c mutex. | |
5847 | ||
5848 | In most cases, these builtins are considered a @dfn{full barrier}. That is, | |
5849 | no memory operand will be moved across the operation, either forward or | |
5850 | backward. Further, instructions will be issued as necessary to prevent the | |
5851 | processor from speculating loads across the operation and from queuing stores | |
5852 | after the operation. | |
5853 | ||
d1facce0 | 5854 | All of the routines are described in the Intel documentation to take |
48ae6c13 RH |
5855 | ``an optional list of variables protected by the memory barrier''. It's |
5856 | not clear what is meant by that; it could mean that @emph{only} the | |
5857 | following variables are protected, or it could mean that these variables | |
5858 | should in addition be protected. At present GCC ignores this list and | |
5859 | protects all variables which are globally accessible. If in the future | |
5860 | we make some use of this list, an empty list will continue to mean all | |
5861 | globally accessible variables. | |
5862 | ||
5863 | @table @code | |
5864 | @item @var{type} __sync_fetch_and_add (@var{type} *ptr, @var{type} value, ...) | |
5865 | @itemx @var{type} __sync_fetch_and_sub (@var{type} *ptr, @var{type} value, ...) | |
5866 | @itemx @var{type} __sync_fetch_and_or (@var{type} *ptr, @var{type} value, ...) | |
5867 | @itemx @var{type} __sync_fetch_and_and (@var{type} *ptr, @var{type} value, ...) | |
5868 | @itemx @var{type} __sync_fetch_and_xor (@var{type} *ptr, @var{type} value, ...) | |
5869 | @itemx @var{type} __sync_fetch_and_nand (@var{type} *ptr, @var{type} value, ...) | |
5870 | @findex __sync_fetch_and_add | |
5871 | @findex __sync_fetch_and_sub | |
5872 | @findex __sync_fetch_and_or | |
5873 | @findex __sync_fetch_and_and | |
5874 | @findex __sync_fetch_and_xor | |
5875 | @findex __sync_fetch_and_nand | |
5876 | These builtins perform the operation suggested by the name, and | |
5877 | returns the value that had previously been in memory. That is, | |
5878 | ||
5879 | @smallexample | |
5880 | @{ tmp = *ptr; *ptr @var{op}= value; return tmp; @} | |
23462d4d | 5881 | @{ tmp = *ptr; *ptr = ~(tmp & value); return tmp; @} // nand |
48ae6c13 RH |
5882 | @end smallexample |
5883 | ||
23462d4d UB |
5884 | @emph{Note:} GCC 4.4 and later implement @code{__sync_fetch_and_nand} |
5885 | builtin as @code{*ptr = ~(tmp & value)} instead of @code{*ptr = ~tmp & value}. | |
5886 | ||
48ae6c13 RH |
5887 | @item @var{type} __sync_add_and_fetch (@var{type} *ptr, @var{type} value, ...) |
5888 | @itemx @var{type} __sync_sub_and_fetch (@var{type} *ptr, @var{type} value, ...) | |
5889 | @itemx @var{type} __sync_or_and_fetch (@var{type} *ptr, @var{type} value, ...) | |
5890 | @itemx @var{type} __sync_and_and_fetch (@var{type} *ptr, @var{type} value, ...) | |
5891 | @itemx @var{type} __sync_xor_and_fetch (@var{type} *ptr, @var{type} value, ...) | |
5892 | @itemx @var{type} __sync_nand_and_fetch (@var{type} *ptr, @var{type} value, ...) | |
5893 | @findex __sync_add_and_fetch | |
5894 | @findex __sync_sub_and_fetch | |
5895 | @findex __sync_or_and_fetch | |
5896 | @findex __sync_and_and_fetch | |
5897 | @findex __sync_xor_and_fetch | |
5898 | @findex __sync_nand_and_fetch | |
5899 | These builtins perform the operation suggested by the name, and | |
5900 | return the new value. That is, | |
5901 | ||
5902 | @smallexample | |
5903 | @{ *ptr @var{op}= value; return *ptr; @} | |
23462d4d | 5904 | @{ *ptr = ~(*ptr & value); return *ptr; @} // nand |
48ae6c13 RH |
5905 | @end smallexample |
5906 | ||
23462d4d UB |
5907 | @emph{Note:} GCC 4.4 and later implement @code{__sync_nand_and_fetch} |
5908 | builtin as @code{*ptr = ~(*ptr & value)} instead of | |
5909 | @code{*ptr = ~*ptr & value}. | |
5910 | ||
48ae6c13 RH |
5911 | @item bool __sync_bool_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...) |
5912 | @itemx @var{type} __sync_val_compare_and_swap (@var{type} *ptr, @var{type} oldval @var{type} newval, ...) | |
5913 | @findex __sync_bool_compare_and_swap | |
5914 | @findex __sync_val_compare_and_swap | |
5915 | These builtins perform an atomic compare and swap. That is, if the current | |
5916 | value of @code{*@var{ptr}} is @var{oldval}, then write @var{newval} into | |
5917 | @code{*@var{ptr}}. | |
5918 | ||
0ac11108 | 5919 | The ``bool'' version returns true if the comparison is successful and |
48ae6c13 | 5920 | @var{newval} was written. The ``val'' version returns the contents |
f12b785d | 5921 | of @code{*@var{ptr}} before the operation. |
48ae6c13 RH |
5922 | |
5923 | @item __sync_synchronize (...) | |
5924 | @findex __sync_synchronize | |
5925 | This builtin issues a full memory barrier. | |
5926 | ||
5927 | @item @var{type} __sync_lock_test_and_set (@var{type} *ptr, @var{type} value, ...) | |
5928 | @findex __sync_lock_test_and_set | |
5929 | This builtin, as described by Intel, is not a traditional test-and-set | |
5930 | operation, but rather an atomic exchange operation. It writes @var{value} | |
5931 | into @code{*@var{ptr}}, and returns the previous contents of | |
5932 | @code{*@var{ptr}}. | |
5933 | ||
5934 | Many targets have only minimal support for such locks, and do not support | |
5935 | a full exchange operation. In this case, a target may support reduced | |
5936 | functionality here by which the @emph{only} valid value to store is the | |
5937 | immediate constant 1. The exact value actually stored in @code{*@var{ptr}} | |
5938 | is implementation defined. | |
5939 | ||
5940 | This builtin is not a full barrier, but rather an @dfn{acquire barrier}. | |
5941 | This means that references after the builtin cannot move to (or be | |
5942 | speculated to) before the builtin, but previous memory stores may not | |
0ac11108 | 5943 | be globally visible yet, and previous memory loads may not yet be |
48ae6c13 RH |
5944 | satisfied. |
5945 | ||
5946 | @item void __sync_lock_release (@var{type} *ptr, ...) | |
5947 | @findex __sync_lock_release | |
5948 | This builtin releases the lock acquired by @code{__sync_lock_test_and_set}. | |
5949 | Normally this means writing the constant 0 to @code{*@var{ptr}}. | |
5950 | ||
5951 | This builtin is not a full barrier, but rather a @dfn{release barrier}. | |
5952 | This means that all previous memory stores are globally visible, and all | |
5953 | previous memory loads have been satisfied, but following memory reads | |
5954 | are not prevented from being speculated to before the barrier. | |
5955 | @end table | |
5956 | ||
10a0d495 JJ |
5957 | @node Object Size Checking |
5958 | @section Object Size Checking Builtins | |
5959 | @findex __builtin_object_size | |
5960 | @findex __builtin___memcpy_chk | |
5961 | @findex __builtin___mempcpy_chk | |
5962 | @findex __builtin___memmove_chk | |
5963 | @findex __builtin___memset_chk | |
5964 | @findex __builtin___strcpy_chk | |
5965 | @findex __builtin___stpcpy_chk | |
5966 | @findex __builtin___strncpy_chk | |
5967 | @findex __builtin___strcat_chk | |
5968 | @findex __builtin___strncat_chk | |
5969 | @findex __builtin___sprintf_chk | |
5970 | @findex __builtin___snprintf_chk | |
5971 | @findex __builtin___vsprintf_chk | |
5972 | @findex __builtin___vsnprintf_chk | |
5973 | @findex __builtin___printf_chk | |
5974 | @findex __builtin___vprintf_chk | |
5975 | @findex __builtin___fprintf_chk | |
5976 | @findex __builtin___vfprintf_chk | |
5977 | ||
5978 | GCC implements a limited buffer overflow protection mechanism | |
5979 | that can prevent some buffer overflow attacks. | |
5980 | ||
5981 | @deftypefn {Built-in Function} {size_t} __builtin_object_size (void * @var{ptr}, int @var{type}) | |
5982 | is a built-in construct that returns a constant number of bytes from | |
5983 | @var{ptr} to the end of the object @var{ptr} pointer points to | |
5984 | (if known at compile time). @code{__builtin_object_size} never evaluates | |
5985 | its arguments for side-effects. If there are any side-effects in them, it | |
5986 | returns @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0} | |
5987 | for @var{type} 2 or 3. If there are multiple objects @var{ptr} can | |
5988 | point to and all of them are known at compile time, the returned number | |
5989 | is the maximum of remaining byte counts in those objects if @var{type} & 2 is | |
a4d05547 | 5990 | 0 and minimum if nonzero. If it is not possible to determine which objects |
10a0d495 JJ |
5991 | @var{ptr} points to at compile time, @code{__builtin_object_size} should |
5992 | return @code{(size_t) -1} for @var{type} 0 or 1 and @code{(size_t) 0} | |
5993 | for @var{type} 2 or 3. | |
5994 | ||
5995 | @var{type} is an integer constant from 0 to 3. If the least significant | |
5996 | bit is clear, objects are whole variables, if it is set, a closest | |
5997 | surrounding subobject is considered the object a pointer points to. | |
5998 | The second bit determines if maximum or minimum of remaining bytes | |
5999 | is computed. | |
6000 | ||
6001 | @smallexample | |
6002 | struct V @{ char buf1[10]; int b; char buf2[10]; @} var; | |
6003 | char *p = &var.buf1[1], *q = &var.b; | |
6004 | ||
6005 | /* Here the object p points to is var. */ | |
6006 | assert (__builtin_object_size (p, 0) == sizeof (var) - 1); | |
6007 | /* The subobject p points to is var.buf1. */ | |
6008 | assert (__builtin_object_size (p, 1) == sizeof (var.buf1) - 1); | |
6009 | /* The object q points to is var. */ | |
6010 | assert (__builtin_object_size (q, 0) | |
6ccde948 | 6011 | == (char *) (&var + 1) - (char *) &var.b); |
10a0d495 JJ |
6012 | /* The subobject q points to is var.b. */ |
6013 | assert (__builtin_object_size (q, 1) == sizeof (var.b)); | |
6014 | @end smallexample | |
6015 | @end deftypefn | |
6016 | ||
6017 | There are built-in functions added for many common string operation | |
021efafc | 6018 | functions, e.g., for @code{memcpy} @code{__builtin___memcpy_chk} |
10a0d495 JJ |
6019 | built-in is provided. This built-in has an additional last argument, |
6020 | which is the number of bytes remaining in object the @var{dest} | |
6021 | argument points to or @code{(size_t) -1} if the size is not known. | |
6022 | ||
6023 | The built-in functions are optimized into the normal string functions | |
6024 | like @code{memcpy} if the last argument is @code{(size_t) -1} or if | |
6025 | it is known at compile time that the destination object will not | |
6026 | be overflown. If the compiler can determine at compile time the | |
6027 | object will be always overflown, it issues a warning. | |
6028 | ||
6029 | The intended use can be e.g. | |
6030 | ||
6031 | @smallexample | |
6032 | #undef memcpy | |
6033 | #define bos0(dest) __builtin_object_size (dest, 0) | |
6034 | #define memcpy(dest, src, n) \ | |
6035 | __builtin___memcpy_chk (dest, src, n, bos0 (dest)) | |
6036 | ||
6037 | char *volatile p; | |
6038 | char buf[10]; | |
6039 | /* It is unknown what object p points to, so this is optimized | |
6040 | into plain memcpy - no checking is possible. */ | |
6041 | memcpy (p, "abcde", n); | |
6042 | /* Destination is known and length too. It is known at compile | |
6043 | time there will be no overflow. */ | |
6044 | memcpy (&buf[5], "abcde", 5); | |
6045 | /* Destination is known, but the length is not known at compile time. | |
6046 | This will result in __memcpy_chk call that can check for overflow | |
6047 | at runtime. */ | |
6048 | memcpy (&buf[5], "abcde", n); | |
6049 | /* Destination is known and it is known at compile time there will | |
6050 | be overflow. There will be a warning and __memcpy_chk call that | |
6051 | will abort the program at runtime. */ | |
6052 | memcpy (&buf[6], "abcde", 5); | |
6053 | @end smallexample | |
6054 | ||
6055 | Such built-in functions are provided for @code{memcpy}, @code{mempcpy}, | |
6056 | @code{memmove}, @code{memset}, @code{strcpy}, @code{stpcpy}, @code{strncpy}, | |
6057 | @code{strcat} and @code{strncat}. | |
6058 | ||
6059 | There are also checking built-in functions for formatted output functions. | |
6060 | @smallexample | |
6061 | int __builtin___sprintf_chk (char *s, int flag, size_t os, const char *fmt, ...); | |
6062 | int __builtin___snprintf_chk (char *s, size_t maxlen, int flag, size_t os, | |
6ccde948 | 6063 | const char *fmt, ...); |
10a0d495 | 6064 | int __builtin___vsprintf_chk (char *s, int flag, size_t os, const char *fmt, |
6ccde948 | 6065 | va_list ap); |
10a0d495 | 6066 | int __builtin___vsnprintf_chk (char *s, size_t maxlen, int flag, size_t os, |
6ccde948 | 6067 | const char *fmt, va_list ap); |
10a0d495 JJ |
6068 | @end smallexample |
6069 | ||
6070 | The added @var{flag} argument is passed unchanged to @code{__sprintf_chk} | |
021efafc | 6071 | etc.@: functions and can contain implementation specific flags on what |
10a0d495 JJ |
6072 | additional security measures the checking function might take, such as |
6073 | handling @code{%n} differently. | |
6074 | ||
6075 | The @var{os} argument is the object size @var{s} points to, like in the | |
a4d05547 | 6076 | other built-in functions. There is a small difference in the behavior |
10a0d495 JJ |
6077 | though, if @var{os} is @code{(size_t) -1}, the built-in functions are |
6078 | optimized into the non-checking functions only if @var{flag} is 0, otherwise | |
6079 | the checking function is called with @var{os} argument set to | |
6080 | @code{(size_t) -1}. | |
6081 | ||
6082 | In addition to this, there are checking built-in functions | |
6083 | @code{__builtin___printf_chk}, @code{__builtin___vprintf_chk}, | |
6084 | @code{__builtin___fprintf_chk} and @code{__builtin___vfprintf_chk}. | |
6085 | These have just one additional argument, @var{flag}, right before | |
6086 | format string @var{fmt}. If the compiler is able to optimize them to | |
021efafc | 6087 | @code{fputc} etc.@: functions, it will, otherwise the checking function |
10a0d495 JJ |
6088 | should be called and the @var{flag} argument passed to it. |
6089 | ||
185ebd6c | 6090 | @node Other Builtins |
f0523f02 | 6091 | @section Other built-in functions provided by GCC |
c771326b | 6092 | @cindex built-in functions |
3bf5906b | 6093 | @findex __builtin_fpclassify |
0c8d3c2b | 6094 | @findex __builtin_isfinite |
8a91c45b | 6095 | @findex __builtin_isnormal |
01702459 JM |
6096 | @findex __builtin_isgreater |
6097 | @findex __builtin_isgreaterequal | |
05f41289 | 6098 | @findex __builtin_isinf_sign |
01702459 JM |
6099 | @findex __builtin_isless |
6100 | @findex __builtin_islessequal | |
6101 | @findex __builtin_islessgreater | |
6102 | @findex __builtin_isunordered | |
17684d46 RG |
6103 | @findex __builtin_powi |
6104 | @findex __builtin_powif | |
6105 | @findex __builtin_powil | |
98ff7c4d KG |
6106 | @findex _Exit |
6107 | @findex _exit | |
01702459 JM |
6108 | @findex abort |
6109 | @findex abs | |
98ff7c4d KG |
6110 | @findex acos |
6111 | @findex acosf | |
6112 | @findex acosh | |
6113 | @findex acoshf | |
6114 | @findex acoshl | |
6115 | @findex acosl | |
01702459 | 6116 | @findex alloca |
98ff7c4d KG |
6117 | @findex asin |
6118 | @findex asinf | |
6119 | @findex asinh | |
6120 | @findex asinhf | |
6121 | @findex asinhl | |
6122 | @findex asinl | |
29f523be | 6123 | @findex atan |
46847aa6 RS |
6124 | @findex atan2 |
6125 | @findex atan2f | |
6126 | @findex atan2l | |
29f523be | 6127 | @findex atanf |
98ff7c4d KG |
6128 | @findex atanh |
6129 | @findex atanhf | |
6130 | @findex atanhl | |
29f523be | 6131 | @findex atanl |
01702459 JM |
6132 | @findex bcmp |
6133 | @findex bzero | |
075ec276 RS |
6134 | @findex cabs |
6135 | @findex cabsf | |
6136 | @findex cabsl | |
11bf0eb0 KG |
6137 | @findex cacos |
6138 | @findex cacosf | |
6139 | @findex cacosh | |
6140 | @findex cacoshf | |
6141 | @findex cacoshl | |
6142 | @findex cacosl | |
1331d16f | 6143 | @findex calloc |
11bf0eb0 KG |
6144 | @findex carg |
6145 | @findex cargf | |
6146 | @findex cargl | |
6147 | @findex casin | |
6148 | @findex casinf | |
6149 | @findex casinh | |
6150 | @findex casinhf | |
6151 | @findex casinhl | |
6152 | @findex casinl | |
6153 | @findex catan | |
6154 | @findex catanf | |
6155 | @findex catanh | |
6156 | @findex catanhf | |
6157 | @findex catanhl | |
6158 | @findex catanl | |
98ff7c4d KG |
6159 | @findex cbrt |
6160 | @findex cbrtf | |
6161 | @findex cbrtl | |
11bf0eb0 KG |
6162 | @findex ccos |
6163 | @findex ccosf | |
6164 | @findex ccosh | |
6165 | @findex ccoshf | |
6166 | @findex ccoshl | |
6167 | @findex ccosl | |
b052d8ee RS |
6168 | @findex ceil |
6169 | @findex ceilf | |
6170 | @findex ceill | |
11bf0eb0 KG |
6171 | @findex cexp |
6172 | @findex cexpf | |
6173 | @findex cexpl | |
341e3d11 JM |
6174 | @findex cimag |
6175 | @findex cimagf | |
6176 | @findex cimagl | |
c3887ef2 PC |
6177 | @findex clog |
6178 | @findex clogf | |
6179 | @findex clogl | |
341e3d11 JM |
6180 | @findex conj |
6181 | @findex conjf | |
6182 | @findex conjl | |
98ff7c4d KG |
6183 | @findex copysign |
6184 | @findex copysignf | |
6185 | @findex copysignl | |
01702459 JM |
6186 | @findex cos |
6187 | @findex cosf | |
98ff7c4d KG |
6188 | @findex cosh |
6189 | @findex coshf | |
6190 | @findex coshl | |
01702459 | 6191 | @findex cosl |
11bf0eb0 KG |
6192 | @findex cpow |
6193 | @findex cpowf | |
6194 | @findex cpowl | |
6195 | @findex cproj | |
6196 | @findex cprojf | |
6197 | @findex cprojl | |
341e3d11 JM |
6198 | @findex creal |
6199 | @findex crealf | |
6200 | @findex creall | |
11bf0eb0 KG |
6201 | @findex csin |
6202 | @findex csinf | |
6203 | @findex csinh | |
6204 | @findex csinhf | |
6205 | @findex csinhl | |
6206 | @findex csinl | |
6207 | @findex csqrt | |
6208 | @findex csqrtf | |
6209 | @findex csqrtl | |
6210 | @findex ctan | |
6211 | @findex ctanf | |
6212 | @findex ctanh | |
6213 | @findex ctanhf | |
6214 | @findex ctanhl | |
6215 | @findex ctanl | |
178b2b9f RS |
6216 | @findex dcgettext |
6217 | @findex dgettext | |
98ff7c4d KG |
6218 | @findex drem |
6219 | @findex dremf | |
6220 | @findex dreml | |
488f17e1 KG |
6221 | @findex erf |
6222 | @findex erfc | |
6223 | @findex erfcf | |
6224 | @findex erfcl | |
6225 | @findex erff | |
6226 | @findex erfl | |
01702459 | 6227 | @findex exit |
e7b489c8 | 6228 | @findex exp |
98ff7c4d KG |
6229 | @findex exp10 |
6230 | @findex exp10f | |
6231 | @findex exp10l | |
6232 | @findex exp2 | |
6233 | @findex exp2f | |
6234 | @findex exp2l | |
e7b489c8 RS |
6235 | @findex expf |
6236 | @findex expl | |
98ff7c4d KG |
6237 | @findex expm1 |
6238 | @findex expm1f | |
6239 | @findex expm1l | |
01702459 JM |
6240 | @findex fabs |
6241 | @findex fabsf | |
6242 | @findex fabsl | |
98ff7c4d KG |
6243 | @findex fdim |
6244 | @findex fdimf | |
6245 | @findex fdiml | |
01702459 | 6246 | @findex ffs |
b052d8ee RS |
6247 | @findex floor |
6248 | @findex floorf | |
6249 | @findex floorl | |
98ff7c4d KG |
6250 | @findex fma |
6251 | @findex fmaf | |
6252 | @findex fmal | |
6253 | @findex fmax | |
6254 | @findex fmaxf | |
6255 | @findex fmaxl | |
6256 | @findex fmin | |
6257 | @findex fminf | |
6258 | @findex fminl | |
b052d8ee RS |
6259 | @findex fmod |
6260 | @findex fmodf | |
6261 | @findex fmodl | |
18f988a0 | 6262 | @findex fprintf |
b4c984fb | 6263 | @findex fprintf_unlocked |
01702459 | 6264 | @findex fputs |
b4c984fb | 6265 | @findex fputs_unlocked |
a2a919aa KG |
6266 | @findex frexp |
6267 | @findex frexpf | |
6268 | @findex frexpl | |
178b2b9f | 6269 | @findex fscanf |
488f17e1 KG |
6270 | @findex gamma |
6271 | @findex gammaf | |
6272 | @findex gammal | |
bf460eec KG |
6273 | @findex gamma_r |
6274 | @findex gammaf_r | |
6275 | @findex gammal_r | |
178b2b9f | 6276 | @findex gettext |
98ff7c4d KG |
6277 | @findex hypot |
6278 | @findex hypotf | |
6279 | @findex hypotl | |
6280 | @findex ilogb | |
6281 | @findex ilogbf | |
6282 | @findex ilogbl | |
e78f4a97 | 6283 | @findex imaxabs |
c7b6c6cd | 6284 | @findex index |
740e5b6f KG |
6285 | @findex isalnum |
6286 | @findex isalpha | |
6287 | @findex isascii | |
6288 | @findex isblank | |
6289 | @findex iscntrl | |
6290 | @findex isdigit | |
6291 | @findex isgraph | |
6292 | @findex islower | |
6293 | @findex isprint | |
6294 | @findex ispunct | |
6295 | @findex isspace | |
6296 | @findex isupper | |
ca4944e1 KG |
6297 | @findex iswalnum |
6298 | @findex iswalpha | |
6299 | @findex iswblank | |
6300 | @findex iswcntrl | |
6301 | @findex iswdigit | |
6302 | @findex iswgraph | |
6303 | @findex iswlower | |
6304 | @findex iswprint | |
6305 | @findex iswpunct | |
6306 | @findex iswspace | |
6307 | @findex iswupper | |
6308 | @findex iswxdigit | |
740e5b6f | 6309 | @findex isxdigit |
488f17e1 KG |
6310 | @findex j0 |
6311 | @findex j0f | |
6312 | @findex j0l | |
6313 | @findex j1 | |
6314 | @findex j1f | |
6315 | @findex j1l | |
6316 | @findex jn | |
6317 | @findex jnf | |
6318 | @findex jnl | |
01702459 | 6319 | @findex labs |
98ff7c4d KG |
6320 | @findex ldexp |
6321 | @findex ldexpf | |
6322 | @findex ldexpl | |
488f17e1 KG |
6323 | @findex lgamma |
6324 | @findex lgammaf | |
6325 | @findex lgammal | |
bf460eec KG |
6326 | @findex lgamma_r |
6327 | @findex lgammaf_r | |
6328 | @findex lgammal_r | |
01702459 | 6329 | @findex llabs |
98ff7c4d KG |
6330 | @findex llrint |
6331 | @findex llrintf | |
6332 | @findex llrintl | |
6333 | @findex llround | |
6334 | @findex llroundf | |
6335 | @findex llroundl | |
e7b489c8 | 6336 | @findex log |
98ff7c4d KG |
6337 | @findex log10 |
6338 | @findex log10f | |
6339 | @findex log10l | |
6340 | @findex log1p | |
6341 | @findex log1pf | |
6342 | @findex log1pl | |
6343 | @findex log2 | |
6344 | @findex log2f | |
6345 | @findex log2l | |
6346 | @findex logb | |
6347 | @findex logbf | |
6348 | @findex logbl | |
e7b489c8 RS |
6349 | @findex logf |
6350 | @findex logl | |
98ff7c4d KG |
6351 | @findex lrint |
6352 | @findex lrintf | |
6353 | @findex lrintl | |
6354 | @findex lround | |
6355 | @findex lroundf | |
6356 | @findex lroundl | |
1331d16f | 6357 | @findex malloc |
2a5fce6d | 6358 | @findex memchr |
01702459 JM |
6359 | @findex memcmp |
6360 | @findex memcpy | |
9cb65f92 | 6361 | @findex mempcpy |
01702459 | 6362 | @findex memset |
a2a919aa KG |
6363 | @findex modf |
6364 | @findex modff | |
6365 | @findex modfl | |
b052d8ee RS |
6366 | @findex nearbyint |
6367 | @findex nearbyintf | |
6368 | @findex nearbyintl | |
98ff7c4d KG |
6369 | @findex nextafter |
6370 | @findex nextafterf | |
6371 | @findex nextafterl | |
6372 | @findex nexttoward | |
6373 | @findex nexttowardf | |
6374 | @findex nexttowardl | |
46847aa6 | 6375 | @findex pow |
98ff7c4d KG |
6376 | @findex pow10 |
6377 | @findex pow10f | |
6378 | @findex pow10l | |
46847aa6 RS |
6379 | @findex powf |
6380 | @findex powl | |
01702459 | 6381 | @findex printf |
b4c984fb | 6382 | @findex printf_unlocked |
08291658 RS |
6383 | @findex putchar |
6384 | @findex puts | |
98ff7c4d KG |
6385 | @findex remainder |
6386 | @findex remainderf | |
6387 | @findex remainderl | |
a2a919aa KG |
6388 | @findex remquo |
6389 | @findex remquof | |
6390 | @findex remquol | |
c7b6c6cd | 6391 | @findex rindex |
98ff7c4d KG |
6392 | @findex rint |
6393 | @findex rintf | |
6394 | @findex rintl | |
b052d8ee RS |
6395 | @findex round |
6396 | @findex roundf | |
6397 | @findex roundl | |
98ff7c4d KG |
6398 | @findex scalb |
6399 | @findex scalbf | |
6400 | @findex scalbl | |
6401 | @findex scalbln | |
6402 | @findex scalblnf | |
6403 | @findex scalblnf | |
6404 | @findex scalbn | |
6405 | @findex scalbnf | |
6406 | @findex scanfnl | |
ef79730c RS |
6407 | @findex signbit |
6408 | @findex signbitf | |
6409 | @findex signbitl | |
44aea9ac JJ |
6410 | @findex signbitd32 |
6411 | @findex signbitd64 | |
6412 | @findex signbitd128 | |
488f17e1 KG |
6413 | @findex significand |
6414 | @findex significandf | |
6415 | @findex significandl | |
01702459 | 6416 | @findex sin |
a2a919aa KG |
6417 | @findex sincos |
6418 | @findex sincosf | |
6419 | @findex sincosl | |
01702459 | 6420 | @findex sinf |
98ff7c4d KG |
6421 | @findex sinh |
6422 | @findex sinhf | |
6423 | @findex sinhl | |
01702459 | 6424 | @findex sinl |
08291658 RS |
6425 | @findex snprintf |
6426 | @findex sprintf | |
01702459 JM |
6427 | @findex sqrt |
6428 | @findex sqrtf | |
6429 | @findex sqrtl | |
08291658 | 6430 | @findex sscanf |
9cb65f92 | 6431 | @findex stpcpy |
e905ac64 KG |
6432 | @findex stpncpy |
6433 | @findex strcasecmp | |
d118937d | 6434 | @findex strcat |
01702459 JM |
6435 | @findex strchr |
6436 | @findex strcmp | |
6437 | @findex strcpy | |
d118937d | 6438 | @findex strcspn |
1331d16f | 6439 | @findex strdup |
178b2b9f RS |
6440 | @findex strfmon |
6441 | @findex strftime | |
01702459 | 6442 | @findex strlen |
e905ac64 | 6443 | @findex strncasecmp |
d118937d | 6444 | @findex strncat |
da9e9f08 KG |
6445 | @findex strncmp |
6446 | @findex strncpy | |
e905ac64 | 6447 | @findex strndup |
01702459 JM |
6448 | @findex strpbrk |
6449 | @findex strrchr | |
d118937d | 6450 | @findex strspn |
01702459 | 6451 | @findex strstr |
29f523be RS |
6452 | @findex tan |
6453 | @findex tanf | |
98ff7c4d KG |
6454 | @findex tanh |
6455 | @findex tanhf | |
6456 | @findex tanhl | |
29f523be | 6457 | @findex tanl |
488f17e1 KG |
6458 | @findex tgamma |
6459 | @findex tgammaf | |
6460 | @findex tgammal | |
740e5b6f KG |
6461 | @findex toascii |
6462 | @findex tolower | |
6463 | @findex toupper | |
ca4944e1 KG |
6464 | @findex towlower |
6465 | @findex towupper | |
4977bab6 ZW |
6466 | @findex trunc |
6467 | @findex truncf | |
6468 | @findex truncl | |
178b2b9f RS |
6469 | @findex vfprintf |
6470 | @findex vfscanf | |
08291658 RS |
6471 | @findex vprintf |
6472 | @findex vscanf | |
6473 | @findex vsnprintf | |
6474 | @findex vsprintf | |
6475 | @findex vsscanf | |
488f17e1 KG |
6476 | @findex y0 |
6477 | @findex y0f | |
6478 | @findex y0l | |
6479 | @findex y1 | |
6480 | @findex y1f | |
6481 | @findex y1l | |
6482 | @findex yn | |
6483 | @findex ynf | |
6484 | @findex ynl | |
185ebd6c | 6485 | |
f0523f02 | 6486 | GCC provides a large number of built-in functions other than the ones |
185ebd6c RH |
6487 | mentioned above. Some of these are for internal use in the processing |
6488 | of exceptions or variable-length argument lists and will not be | |
6489 | documented here because they may change from time to time; we do not | |
6490 | recommend general use of these functions. | |
6491 | ||
6492 | The remaining functions are provided for optimization purposes. | |
6493 | ||
84330467 | 6494 | @opindex fno-builtin |
9c34dbbf ZW |
6495 | GCC includes built-in versions of many of the functions in the standard |
6496 | C library. The versions prefixed with @code{__builtin_} will always be | |
6497 | treated as having the same meaning as the C library function even if you | |
8a36672b | 6498 | specify the @option{-fno-builtin} option. (@pxref{C Dialect Options}) |
9c34dbbf | 6499 | Many of these functions are only optimized in certain cases; if they are |
01702459 JM |
6500 | not optimized in a particular case, a call to the library function will |
6501 | be emitted. | |
6502 | ||
84330467 JM |
6503 | @opindex ansi |
6504 | @opindex std | |
b052d8ee | 6505 | Outside strict ISO C mode (@option{-ansi}, @option{-std=c89} or |
98ff7c4d KG |
6506 | @option{-std=c99}), the functions |
6507 | @code{_exit}, @code{alloca}, @code{bcmp}, @code{bzero}, | |
6508 | @code{dcgettext}, @code{dgettext}, @code{dremf}, @code{dreml}, | |
6509 | @code{drem}, @code{exp10f}, @code{exp10l}, @code{exp10}, @code{ffsll}, | |
bf460eec KG |
6510 | @code{ffsl}, @code{ffs}, @code{fprintf_unlocked}, |
6511 | @code{fputs_unlocked}, @code{gammaf}, @code{gammal}, @code{gamma}, | |
6512 | @code{gammaf_r}, @code{gammal_r}, @code{gamma_r}, @code{gettext}, | |
740e5b6f KG |
6513 | @code{index}, @code{isascii}, @code{j0f}, @code{j0l}, @code{j0}, |
6514 | @code{j1f}, @code{j1l}, @code{j1}, @code{jnf}, @code{jnl}, @code{jn}, | |
bf460eec KG |
6515 | @code{lgammaf_r}, @code{lgammal_r}, @code{lgamma_r}, @code{mempcpy}, |
6516 | @code{pow10f}, @code{pow10l}, @code{pow10}, @code{printf_unlocked}, | |
6517 | @code{rindex}, @code{scalbf}, @code{scalbl}, @code{scalb}, | |
6518 | @code{signbit}, @code{signbitf}, @code{signbitl}, @code{signbitd32}, | |
6519 | @code{signbitd64}, @code{signbitd128}, @code{significandf}, | |
6520 | @code{significandl}, @code{significand}, @code{sincosf}, | |
6521 | @code{sincosl}, @code{sincos}, @code{stpcpy}, @code{stpncpy}, | |
6522 | @code{strcasecmp}, @code{strdup}, @code{strfmon}, @code{strncasecmp}, | |
6523 | @code{strndup}, @code{toascii}, @code{y0f}, @code{y0l}, @code{y0}, | |
6524 | @code{y1f}, @code{y1l}, @code{y1}, @code{ynf}, @code{ynl} and | |
6525 | @code{yn} | |
1331d16f | 6526 | may be handled as built-in functions. |
b052d8ee | 6527 | All these functions have corresponding versions |
9c34dbbf ZW |
6528 | prefixed with @code{__builtin_}, which may be used even in strict C89 |
6529 | mode. | |
01702459 | 6530 | |
075ec276 | 6531 | The ISO C99 functions |
98ff7c4d KG |
6532 | @code{_Exit}, @code{acoshf}, @code{acoshl}, @code{acosh}, @code{asinhf}, |
6533 | @code{asinhl}, @code{asinh}, @code{atanhf}, @code{atanhl}, @code{atanh}, | |
11bf0eb0 KG |
6534 | @code{cabsf}, @code{cabsl}, @code{cabs}, @code{cacosf}, @code{cacoshf}, |
6535 | @code{cacoshl}, @code{cacosh}, @code{cacosl}, @code{cacos}, | |
6536 | @code{cargf}, @code{cargl}, @code{carg}, @code{casinf}, @code{casinhf}, | |
6537 | @code{casinhl}, @code{casinh}, @code{casinl}, @code{casin}, | |
6538 | @code{catanf}, @code{catanhf}, @code{catanhl}, @code{catanh}, | |
6539 | @code{catanl}, @code{catan}, @code{cbrtf}, @code{cbrtl}, @code{cbrt}, | |
6540 | @code{ccosf}, @code{ccoshf}, @code{ccoshl}, @code{ccosh}, @code{ccosl}, | |
6541 | @code{ccos}, @code{cexpf}, @code{cexpl}, @code{cexp}, @code{cimagf}, | |
c3887ef2 PC |
6542 | @code{cimagl}, @code{cimag}, @code{clogf}, @code{clogl}, @code{clog}, |
6543 | @code{conjf}, @code{conjl}, @code{conj}, @code{copysignf}, @code{copysignl}, | |
6544 | @code{copysign}, @code{cpowf}, @code{cpowl}, @code{cpow}, @code{cprojf}, | |
6545 | @code{cprojl}, @code{cproj}, @code{crealf}, @code{creall}, @code{creal}, | |
6546 | @code{csinf}, @code{csinhf}, @code{csinhl}, @code{csinh}, @code{csinl}, | |
6547 | @code{csin}, @code{csqrtf}, @code{csqrtl}, @code{csqrt}, @code{ctanf}, | |
6548 | @code{ctanhf}, @code{ctanhl}, @code{ctanh}, @code{ctanl}, @code{ctan}, | |
6549 | @code{erfcf}, @code{erfcl}, @code{erfc}, @code{erff}, @code{erfl}, | |
6550 | @code{erf}, @code{exp2f}, @code{exp2l}, @code{exp2}, @code{expm1f}, | |
6551 | @code{expm1l}, @code{expm1}, @code{fdimf}, @code{fdiml}, @code{fdim}, | |
6552 | @code{fmaf}, @code{fmal}, @code{fmaxf}, @code{fmaxl}, @code{fmax}, | |
6553 | @code{fma}, @code{fminf}, @code{fminl}, @code{fmin}, @code{hypotf}, | |
6554 | @code{hypotl}, @code{hypot}, @code{ilogbf}, @code{ilogbl}, @code{ilogb}, | |
6555 | @code{imaxabs}, @code{isblank}, @code{iswblank}, @code{lgammaf}, | |
6556 | @code{lgammal}, @code{lgamma}, @code{llabs}, @code{llrintf}, @code{llrintl}, | |
ca4944e1 KG |
6557 | @code{llrint}, @code{llroundf}, @code{llroundl}, @code{llround}, |
6558 | @code{log1pf}, @code{log1pl}, @code{log1p}, @code{log2f}, @code{log2l}, | |
6559 | @code{log2}, @code{logbf}, @code{logbl}, @code{logb}, @code{lrintf}, | |
6560 | @code{lrintl}, @code{lrint}, @code{lroundf}, @code{lroundl}, | |
6561 | @code{lround}, @code{nearbyintf}, @code{nearbyintl}, @code{nearbyint}, | |
740e5b6f KG |
6562 | @code{nextafterf}, @code{nextafterl}, @code{nextafter}, |
6563 | @code{nexttowardf}, @code{nexttowardl}, @code{nexttoward}, | |
6564 | @code{remainderf}, @code{remainderl}, @code{remainder}, @code{remquof}, | |
6565 | @code{remquol}, @code{remquo}, @code{rintf}, @code{rintl}, @code{rint}, | |
6566 | @code{roundf}, @code{roundl}, @code{round}, @code{scalblnf}, | |
6567 | @code{scalblnl}, @code{scalbln}, @code{scalbnf}, @code{scalbnl}, | |
6568 | @code{scalbn}, @code{snprintf}, @code{tgammaf}, @code{tgammal}, | |
6569 | @code{tgamma}, @code{truncf}, @code{truncl}, @code{trunc}, | |
6570 | @code{vfscanf}, @code{vscanf}, @code{vsnprintf} and @code{vsscanf} | |
08291658 | 6571 | are handled as built-in functions |
b052d8ee | 6572 | except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}). |
46847aa6 | 6573 | |
98ff7c4d KG |
6574 | There are also built-in versions of the ISO C99 functions |
6575 | @code{acosf}, @code{acosl}, @code{asinf}, @code{asinl}, @code{atan2f}, | |
29f523be | 6576 | @code{atan2l}, @code{atanf}, @code{atanl}, @code{ceilf}, @code{ceill}, |
98ff7c4d KG |
6577 | @code{cosf}, @code{coshf}, @code{coshl}, @code{cosl}, @code{expf}, |
6578 | @code{expl}, @code{fabsf}, @code{fabsl}, @code{floorf}, @code{floorl}, | |
a2a919aa KG |
6579 | @code{fmodf}, @code{fmodl}, @code{frexpf}, @code{frexpl}, @code{ldexpf}, |
6580 | @code{ldexpl}, @code{log10f}, @code{log10l}, @code{logf}, @code{logl}, | |
6581 | @code{modfl}, @code{modf}, @code{powf}, @code{powl}, @code{sinf}, | |
6582 | @code{sinhf}, @code{sinhl}, @code{sinl}, @code{sqrtf}, @code{sqrtl}, | |
6583 | @code{tanf}, @code{tanhf}, @code{tanhl} and @code{tanl} | |
46847aa6 RS |
6584 | that are recognized in any mode since ISO C90 reserves these names for |
6585 | the purpose to which ISO C99 puts them. All these functions have | |
6586 | corresponding versions prefixed with @code{__builtin_}. | |
6587 | ||
ca4944e1 KG |
6588 | The ISO C94 functions |
6589 | @code{iswalnum}, @code{iswalpha}, @code{iswcntrl}, @code{iswdigit}, | |
6590 | @code{iswgraph}, @code{iswlower}, @code{iswprint}, @code{iswpunct}, | |
6591 | @code{iswspace}, @code{iswupper}, @code{iswxdigit}, @code{towlower} and | |
6592 | @code{towupper} | |
6593 | are handled as built-in functions | |
6594 | except in strict ISO C90 mode (@option{-ansi} or @option{-std=c89}). | |
6595 | ||
98ff7c4d KG |
6596 | The ISO C90 functions |
6597 | @code{abort}, @code{abs}, @code{acos}, @code{asin}, @code{atan2}, | |
6598 | @code{atan}, @code{calloc}, @code{ceil}, @code{cosh}, @code{cos}, | |
6599 | @code{exit}, @code{exp}, @code{fabs}, @code{floor}, @code{fmod}, | |
740e5b6f KG |
6600 | @code{fprintf}, @code{fputs}, @code{frexp}, @code{fscanf}, |
6601 | @code{isalnum}, @code{isalpha}, @code{iscntrl}, @code{isdigit}, | |
6602 | @code{isgraph}, @code{islower}, @code{isprint}, @code{ispunct}, | |
6603 | @code{isspace}, @code{isupper}, @code{isxdigit}, @code{tolower}, | |
6604 | @code{toupper}, @code{labs}, @code{ldexp}, @code{log10}, @code{log}, | |
2a5fce6d PC |
6605 | @code{malloc}, @code{memchr}, @code{memcmp}, @code{memcpy}, |
6606 | @code{memset}, @code{modf}, @code{pow}, @code{printf}, @code{putchar}, | |
6607 | @code{puts}, @code{scanf}, @code{sinh}, @code{sin}, @code{snprintf}, | |
6608 | @code{sprintf}, @code{sqrt}, @code{sscanf}, @code{strcat}, | |
6609 | @code{strchr}, @code{strcmp}, @code{strcpy}, @code{strcspn}, | |
6610 | @code{strlen}, @code{strncat}, @code{strncmp}, @code{strncpy}, | |
6611 | @code{strpbrk}, @code{strrchr}, @code{strspn}, @code{strstr}, | |
6612 | @code{tanh}, @code{tan}, @code{vfprintf}, @code{vprintf} and @code{vsprintf} | |
08291658 | 6613 | are all recognized as built-in functions unless |
46847aa6 RS |
6614 | @option{-fno-builtin} is specified (or @option{-fno-builtin-@var{function}} |
6615 | is specified for an individual function). All of these functions have | |
4977bab6 | 6616 | corresponding versions prefixed with @code{__builtin_}. |
9c34dbbf ZW |
6617 | |
6618 | GCC provides built-in versions of the ISO C99 floating point comparison | |
6619 | macros that avoid raising exceptions for unordered operands. They have | |
6620 | the same names as the standard macros ( @code{isgreater}, | |
6621 | @code{isgreaterequal}, @code{isless}, @code{islessequal}, | |
6622 | @code{islessgreater}, and @code{isunordered}) , with @code{__builtin_} | |
6623 | prefixed. We intend for a library implementor to be able to simply | |
6624 | @code{#define} each standard macro to its built-in equivalent. | |
3bf5906b KG |
6625 | In the same fashion, GCC provides @code{fpclassify}, @code{isfinite}, |
6626 | @code{isinf_sign} and @code{isnormal} built-ins used with | |
6627 | @code{__builtin_} prefixed. The @code{isinf} and @code{isnan} | |
6628 | builtins appear both with and without the @code{__builtin_} prefix. | |
185ebd6c | 6629 | |
ecbcf7b3 AH |
6630 | @deftypefn {Built-in Function} int __builtin_types_compatible_p (@var{type1}, @var{type2}) |
6631 | ||
6632 | You can use the built-in function @code{__builtin_types_compatible_p} to | |
6633 | determine whether two types are the same. | |
6634 | ||
6635 | This built-in function returns 1 if the unqualified versions of the | |
6636 | types @var{type1} and @var{type2} (which are types, not expressions) are | |
6637 | compatible, 0 otherwise. The result of this built-in function can be | |
6638 | used in integer constant expressions. | |
6639 | ||
6640 | This built-in function ignores top level qualifiers (e.g., @code{const}, | |
6641 | @code{volatile}). For example, @code{int} is equivalent to @code{const | |
6642 | int}. | |
6643 | ||
6644 | The type @code{int[]} and @code{int[5]} are compatible. On the other | |
6645 | hand, @code{int} and @code{char *} are not compatible, even if the size | |
6646 | of their types, on the particular architecture are the same. Also, the | |
6647 | amount of pointer indirection is taken into account when determining | |
6648 | similarity. Consequently, @code{short *} is not similar to | |
6649 | @code{short **}. Furthermore, two types that are typedefed are | |
6650 | considered compatible if their underlying types are compatible. | |
6651 | ||
bca63328 JM |
6652 | An @code{enum} type is not considered to be compatible with another |
6653 | @code{enum} type even if both are compatible with the same integer | |
6654 | type; this is what the C standard specifies. | |
6655 | For example, @code{enum @{foo, bar@}} is not similar to | |
ecbcf7b3 AH |
6656 | @code{enum @{hot, dog@}}. |
6657 | ||
6658 | You would typically use this function in code whose execution varies | |
6659 | depending on the arguments' types. For example: | |
6660 | ||
6661 | @smallexample | |
6e5bb5ad JM |
6662 | #define foo(x) \ |
6663 | (@{ \ | |
b7886f14 | 6664 | typeof (x) tmp = (x); \ |
6e5bb5ad JM |
6665 | if (__builtin_types_compatible_p (typeof (x), long double)) \ |
6666 | tmp = foo_long_double (tmp); \ | |
6667 | else if (__builtin_types_compatible_p (typeof (x), double)) \ | |
6668 | tmp = foo_double (tmp); \ | |
6669 | else if (__builtin_types_compatible_p (typeof (x), float)) \ | |
6670 | tmp = foo_float (tmp); \ | |
6671 | else \ | |
6672 | abort (); \ | |
6673 | tmp; \ | |
ecbcf7b3 AH |
6674 | @}) |
6675 | @end smallexample | |
6676 | ||
8a36672b | 6677 | @emph{Note:} This construct is only available for C@. |
ecbcf7b3 AH |
6678 | |
6679 | @end deftypefn | |
6680 | ||
6681 | @deftypefn {Built-in Function} @var{type} __builtin_choose_expr (@var{const_exp}, @var{exp1}, @var{exp2}) | |
6682 | ||
6683 | You can use the built-in function @code{__builtin_choose_expr} to | |
6684 | evaluate code depending on the value of a constant expression. This | |
928c19bb JM |
6685 | built-in function returns @var{exp1} if @var{const_exp}, which is an |
6686 | integer constant expression, is nonzero. Otherwise it returns 0. | |
ecbcf7b3 AH |
6687 | |
6688 | This built-in function is analogous to the @samp{? :} operator in C, | |
6689 | except that the expression returned has its type unaltered by promotion | |
6690 | rules. Also, the built-in function does not evaluate the expression | |
6691 | that was not chosen. For example, if @var{const_exp} evaluates to true, | |
6692 | @var{exp2} is not evaluated even if it has side-effects. | |
6693 | ||
6694 | This built-in function can return an lvalue if the chosen argument is an | |
6695 | lvalue. | |
6696 | ||
6697 | If @var{exp1} is returned, the return type is the same as @var{exp1}'s | |
6698 | type. Similarly, if @var{exp2} is returned, its return type is the same | |
6699 | as @var{exp2}. | |
6700 | ||
6701 | Example: | |
6702 | ||
6703 | @smallexample | |
478c9e72 JJ |
6704 | #define foo(x) \ |
6705 | __builtin_choose_expr ( \ | |
6706 | __builtin_types_compatible_p (typeof (x), double), \ | |
6707 | foo_double (x), \ | |
6708 | __builtin_choose_expr ( \ | |
6709 | __builtin_types_compatible_p (typeof (x), float), \ | |
6710 | foo_float (x), \ | |
6711 | /* @r{The void expression results in a compile-time error} \ | |
6712 | @r{when assigning the result to something.} */ \ | |
ecbcf7b3 AH |
6713 | (void)0)) |
6714 | @end smallexample | |
6715 | ||
8a36672b | 6716 | @emph{Note:} This construct is only available for C@. Furthermore, the |
ecbcf7b3 AH |
6717 | unused expression (@var{exp1} or @var{exp2} depending on the value of |
6718 | @var{const_exp}) may still generate syntax errors. This may change in | |
6719 | future revisions. | |
6720 | ||
6721 | @end deftypefn | |
6722 | ||
84330467 JM |
6723 | @deftypefn {Built-in Function} int __builtin_constant_p (@var{exp}) |
6724 | You can use the built-in function @code{__builtin_constant_p} to | |
185ebd6c | 6725 | determine if a value is known to be constant at compile-time and hence |
f0523f02 | 6726 | that GCC can perform constant-folding on expressions involving that |
185ebd6c RH |
6727 | value. The argument of the function is the value to test. The function |
6728 | returns the integer 1 if the argument is known to be a compile-time | |
6729 | constant and 0 if it is not known to be a compile-time constant. A | |
6730 | return of 0 does not indicate that the value is @emph{not} a constant, | |
f0523f02 | 6731 | but merely that GCC cannot prove it is a constant with the specified |
84330467 | 6732 | value of the @option{-O} option. |
185ebd6c RH |
6733 | |
6734 | You would typically use this function in an embedded application where | |
6735 | memory was a critical resource. If you have some complex calculation, | |
6736 | you may want it to be folded if it involves constants, but need to call | |
6737 | a function if it does not. For example: | |
6738 | ||
4d390518 | 6739 | @smallexample |
310668e8 JM |
6740 | #define Scale_Value(X) \ |
6741 | (__builtin_constant_p (X) \ | |
6742 | ? ((X) * SCALE + OFFSET) : Scale (X)) | |
185ebd6c RH |
6743 | @end smallexample |
6744 | ||
84330467 | 6745 | You may use this built-in function in either a macro or an inline |
185ebd6c | 6746 | function. However, if you use it in an inlined function and pass an |
f0523f02 | 6747 | argument of the function as the argument to the built-in, GCC will |
185ebd6c | 6748 | never return 1 when you call the inline function with a string constant |
4b404517 | 6749 | or compound literal (@pxref{Compound Literals}) and will not return 1 |
185ebd6c | 6750 | when you pass a constant numeric value to the inline function unless you |
84330467 | 6751 | specify the @option{-O} option. |
13104975 ZW |
6752 | |
6753 | You may also use @code{__builtin_constant_p} in initializers for static | |
6754 | data. For instance, you can write | |
6755 | ||
6756 | @smallexample | |
79323c50 | 6757 | static const int table[] = @{ |
13104975 | 6758 | __builtin_constant_p (EXPRESSION) ? (EXPRESSION) : -1, |
0d893a63 | 6759 | /* @r{@dots{}} */ |
79323c50 | 6760 | @}; |
13104975 ZW |
6761 | @end smallexample |
6762 | ||
6763 | @noindent | |
6764 | This is an acceptable initializer even if @var{EXPRESSION} is not a | |
928c19bb JM |
6765 | constant expression, including the case where |
6766 | @code{__builtin_constant_p} returns 1 because @var{EXPRESSION} can be | |
6767 | folded to a constant but @var{EXPRESSION} contains operands that would | |
c782f1dd | 6768 | not otherwise be permitted in a static initializer (for example, |
928c19bb | 6769 | @code{0 && foo ()}). GCC must be more conservative about evaluating the |
13104975 ZW |
6770 | built-in in this case, because it has no opportunity to perform |
6771 | optimization. | |
6772 | ||
6773 | Previous versions of GCC did not accept this built-in in data | |
6774 | initializers. The earliest version where it is completely safe is | |
6775 | 3.0.1. | |
84330467 | 6776 | @end deftypefn |
185ebd6c | 6777 | |
84330467 JM |
6778 | @deftypefn {Built-in Function} long __builtin_expect (long @var{exp}, long @var{c}) |
6779 | @opindex fprofile-arcs | |
02f52e19 | 6780 | You may use @code{__builtin_expect} to provide the compiler with |
994a57cd | 6781 | branch prediction information. In general, you should prefer to |
84330467 | 6782 | use actual profile feedback for this (@option{-fprofile-arcs}), as |
994a57cd | 6783 | programmers are notoriously bad at predicting how their programs |
60b6e1f5 | 6784 | actually perform. However, there are applications in which this |
994a57cd RH |
6785 | data is hard to collect. |
6786 | ||
ef950eba JH |
6787 | The return value is the value of @var{exp}, which should be an integral |
6788 | expression. The semantics of the built-in are that it is expected that | |
6789 | @var{exp} == @var{c}. For example: | |
994a57cd RH |
6790 | |
6791 | @smallexample | |
6792 | if (__builtin_expect (x, 0)) | |
6793 | foo (); | |
6794 | @end smallexample | |
6795 | ||
6796 | @noindent | |
6797 | would indicate that we do not expect to call @code{foo}, since | |
6798 | we expect @code{x} to be zero. Since you are limited to integral | |
6799 | expressions for @var{exp}, you should use constructions such as | |
6800 | ||
6801 | @smallexample | |
6802 | if (__builtin_expect (ptr != NULL, 1)) | |
6803 | error (); | |
6804 | @end smallexample | |
6805 | ||
6806 | @noindent | |
6807 | when testing pointer or floating-point values. | |
84330467 | 6808 | @end deftypefn |
994a57cd | 6809 | |
a18c20ec AP |
6810 | @deftypefn {Built-in Function} void __builtin_trap (void) |
6811 | This function causes the program to exit abnormally. GCC implements | |
6812 | this function by using a target-dependent mechanism (such as | |
6813 | intentionally executing an illegal instruction) or by calling | |
6814 | @code{abort}. The mechanism used may vary from release to release so | |
6815 | you should not rely on any particular implementation. | |
6816 | @end deftypefn | |
6817 | ||
677feb77 DD |
6818 | @deftypefn {Built-in Function} void __builtin___clear_cache (char *@var{begin}, char *@var{end}) |
6819 | This function is used to flush the processor's instruction cache for | |
6820 | the region of memory between @var{begin} inclusive and @var{end} | |
6821 | exclusive. Some targets require that the instruction cache be | |
6822 | flushed, after modifying memory containing code, in order to obtain | |
6823 | deterministic behavior. | |
6824 | ||
6825 | If the target does not require instruction cache flushes, | |
6826 | @code{__builtin___clear_cache} has no effect. Otherwise either | |
6827 | instructions are emitted in-line to clear the instruction cache or a | |
6828 | call to the @code{__clear_cache} function in libgcc is made. | |
6829 | @end deftypefn | |
6830 | ||
3bca17dd | 6831 | @deftypefn {Built-in Function} void __builtin_prefetch (const void *@var{addr}, ...) |
a9ccbb60 JJ |
6832 | This function is used to minimize cache-miss latency by moving data into |
6833 | a cache before it is accessed. | |
6834 | You can insert calls to @code{__builtin_prefetch} into code for which | |
6835 | you know addresses of data in memory that is likely to be accessed soon. | |
6836 | If the target supports them, data prefetch instructions will be generated. | |
6837 | If the prefetch is done early enough before the access then the data will | |
6838 | be in the cache by the time it is accessed. | |
6839 | ||
6840 | The value of @var{addr} is the address of the memory to prefetch. | |
e83d297b | 6841 | There are two optional arguments, @var{rw} and @var{locality}. |
a9ccbb60 | 6842 | The value of @var{rw} is a compile-time constant one or zero; one |
e83d297b JJ |
6843 | means that the prefetch is preparing for a write to the memory address |
6844 | and zero, the default, means that the prefetch is preparing for a read. | |
a9ccbb60 JJ |
6845 | The value @var{locality} must be a compile-time constant integer between |
6846 | zero and three. A value of zero means that the data has no temporal | |
6847 | locality, so it need not be left in the cache after the access. A value | |
6848 | of three means that the data has a high degree of temporal locality and | |
6849 | should be left in all levels of cache possible. Values of one and two | |
e83d297b JJ |
6850 | mean, respectively, a low or moderate degree of temporal locality. The |
6851 | default is three. | |
a9ccbb60 JJ |
6852 | |
6853 | @smallexample | |
6854 | for (i = 0; i < n; i++) | |
6855 | @{ | |
6856 | a[i] = a[i] + b[i]; | |
6857 | __builtin_prefetch (&a[i+j], 1, 1); | |
6858 | __builtin_prefetch (&b[i+j], 0, 1); | |
0d893a63 | 6859 | /* @r{@dots{}} */ |
a9ccbb60 JJ |
6860 | @} |
6861 | @end smallexample | |
6862 | ||
f282ffb3 | 6863 | Data prefetch does not generate faults if @var{addr} is invalid, but |
a9ccbb60 JJ |
6864 | the address expression itself must be valid. For example, a prefetch |
6865 | of @code{p->next} will not fault if @code{p->next} is not a valid | |
6866 | address, but evaluation will fault if @code{p} is not a valid address. | |
6867 | ||
6868 | If the target does not support data prefetch, the address expression | |
6869 | is evaluated if it includes side effects but no other code is generated | |
6870 | and GCC does not issue a warning. | |
6871 | @end deftypefn | |
6872 | ||
ab5e2615 RH |
6873 | @deftypefn {Built-in Function} double __builtin_huge_val (void) |
6874 | Returns a positive infinity, if supported by the floating-point format, | |
6875 | else @code{DBL_MAX}. This function is suitable for implementing the | |
6876 | ISO C macro @code{HUGE_VAL}. | |
6877 | @end deftypefn | |
6878 | ||
6879 | @deftypefn {Built-in Function} float __builtin_huge_valf (void) | |
6880 | Similar to @code{__builtin_huge_val}, except the return type is @code{float}. | |
6881 | @end deftypefn | |
6882 | ||
dad78426 | 6883 | @deftypefn {Built-in Function} {long double} __builtin_huge_vall (void) |
ab5e2615 RH |
6884 | Similar to @code{__builtin_huge_val}, except the return |
6885 | type is @code{long double}. | |
6886 | @end deftypefn | |
6887 | ||
3bf5906b KG |
6888 | @deftypefn {Built-in Function} int __builtin_fpclassify (int, int, int, int, int, ...) |
6889 | This built-in implements the C99 fpclassify functionality. The first | |
6890 | five int arguments should be the target library's notion of the | |
6891 | possible FP classes and are used for return values. They must be | |
6892 | constant values and they must appear in this order: @code{FP_NAN}, | |
32101f99 | 6893 | @code{FP_INFINITE}, @code{FP_NORMAL}, @code{FP_SUBNORMAL} and |
3bf5906b KG |
6894 | @code{FP_ZERO}. The ellipsis is for exactly one floating point value |
6895 | to classify. GCC treats the last argument as type-generic, which | |
6896 | means it does not do default promotion from float to double. | |
6897 | @end deftypefn | |
6898 | ||
ab5e2615 RH |
6899 | @deftypefn {Built-in Function} double __builtin_inf (void) |
6900 | Similar to @code{__builtin_huge_val}, except a warning is generated | |
6901 | if the target floating-point format does not support infinities. | |
ab5e2615 RH |
6902 | @end deftypefn |
6903 | ||
9a8ce21f JG |
6904 | @deftypefn {Built-in Function} _Decimal32 __builtin_infd32 (void) |
6905 | Similar to @code{__builtin_inf}, except the return type is @code{_Decimal32}. | |
6906 | @end deftypefn | |
6907 | ||
6908 | @deftypefn {Built-in Function} _Decimal64 __builtin_infd64 (void) | |
6909 | Similar to @code{__builtin_inf}, except the return type is @code{_Decimal64}. | |
6910 | @end deftypefn | |
6911 | ||
6912 | @deftypefn {Built-in Function} _Decimal128 __builtin_infd128 (void) | |
6913 | Similar to @code{__builtin_inf}, except the return type is @code{_Decimal128}. | |
6914 | @end deftypefn | |
6915 | ||
ab5e2615 RH |
6916 | @deftypefn {Built-in Function} float __builtin_inff (void) |
6917 | Similar to @code{__builtin_inf}, except the return type is @code{float}. | |
9c86fc0b | 6918 | This function is suitable for implementing the ISO C99 macro @code{INFINITY}. |
ab5e2615 RH |
6919 | @end deftypefn |
6920 | ||
dad78426 | 6921 | @deftypefn {Built-in Function} {long double} __builtin_infl (void) |
ab5e2615 RH |
6922 | Similar to @code{__builtin_inf}, except the return |
6923 | type is @code{long double}. | |
6924 | @end deftypefn | |
6925 | ||
05f41289 KG |
6926 | @deftypefn {Built-in Function} int __builtin_isinf_sign (...) |
6927 | Similar to @code{isinf}, except the return value will be negative for | |
6928 | an argument of @code{-Inf}. Note while the parameter list is an | |
6929 | ellipsis, this function only accepts exactly one floating point | |
6930 | argument. GCC treats this parameter as type-generic, which means it | |
6931 | does not do default promotion from float to double. | |
6932 | @end deftypefn | |
6933 | ||
1472e41c RH |
6934 | @deftypefn {Built-in Function} double __builtin_nan (const char *str) |
6935 | This is an implementation of the ISO C99 function @code{nan}. | |
6936 | ||
6937 | Since ISO C99 defines this function in terms of @code{strtod}, which we | |
c0478a66 | 6938 | do not implement, a description of the parsing is in order. The string |
1472e41c RH |
6939 | is parsed as by @code{strtol}; that is, the base is recognized by |
6940 | leading @samp{0} or @samp{0x} prefixes. The number parsed is placed | |
6941 | in the significand such that the least significant bit of the number | |
daf2f129 | 6942 | is at the least significant bit of the significand. The number is |
1472e41c | 6943 | truncated to fit the significand field provided. The significand is |
8a36672b | 6944 | forced to be a quiet NaN@. |
1472e41c | 6945 | |
a7d37464 GK |
6946 | This function, if given a string literal all of which would have been |
6947 | consumed by strtol, is evaluated early enough that it is considered a | |
6948 | compile-time constant. | |
1472e41c RH |
6949 | @end deftypefn |
6950 | ||
9a8ce21f JG |
6951 | @deftypefn {Built-in Function} _Decimal32 __builtin_nand32 (const char *str) |
6952 | Similar to @code{__builtin_nan}, except the return type is @code{_Decimal32}. | |
6953 | @end deftypefn | |
6954 | ||
6955 | @deftypefn {Built-in Function} _Decimal64 __builtin_nand64 (const char *str) | |
6956 | Similar to @code{__builtin_nan}, except the return type is @code{_Decimal64}. | |
6957 | @end deftypefn | |
6958 | ||
6959 | @deftypefn {Built-in Function} _Decimal128 __builtin_nand128 (const char *str) | |
6960 | Similar to @code{__builtin_nan}, except the return type is @code{_Decimal128}. | |
6961 | @end deftypefn | |
6962 | ||
1472e41c RH |
6963 | @deftypefn {Built-in Function} float __builtin_nanf (const char *str) |
6964 | Similar to @code{__builtin_nan}, except the return type is @code{float}. | |
6965 | @end deftypefn | |
6966 | ||
dad78426 | 6967 | @deftypefn {Built-in Function} {long double} __builtin_nanl (const char *str) |
1472e41c RH |
6968 | Similar to @code{__builtin_nan}, except the return type is @code{long double}. |
6969 | @end deftypefn | |
6970 | ||
6971 | @deftypefn {Built-in Function} double __builtin_nans (const char *str) | |
daf2f129 | 6972 | Similar to @code{__builtin_nan}, except the significand is forced |
8a36672b | 6973 | to be a signaling NaN@. The @code{nans} function is proposed by |
aaa67502 | 6974 | @uref{http://www.open-std.org/jtc1/sc22/wg14/www/docs/n965.htm,,WG14 N965}. |
1472e41c RH |
6975 | @end deftypefn |
6976 | ||
6977 | @deftypefn {Built-in Function} float __builtin_nansf (const char *str) | |
6978 | Similar to @code{__builtin_nans}, except the return type is @code{float}. | |
6979 | @end deftypefn | |
6980 | ||
dad78426 | 6981 | @deftypefn {Built-in Function} {long double} __builtin_nansl (const char *str) |
1472e41c RH |
6982 | Similar to @code{__builtin_nans}, except the return type is @code{long double}. |
6983 | @end deftypefn | |
6984 | ||
2928cd7a RH |
6985 | @deftypefn {Built-in Function} int __builtin_ffs (unsigned int x) |
6986 | Returns one plus the index of the least significant 1-bit of @var{x}, or | |
6987 | if @var{x} is zero, returns zero. | |
6988 | @end deftypefn | |
6989 | ||
6990 | @deftypefn {Built-in Function} int __builtin_clz (unsigned int x) | |
6991 | Returns the number of leading 0-bits in @var{x}, starting at the most | |
6992 | significant bit position. If @var{x} is 0, the result is undefined. | |
6993 | @end deftypefn | |
6994 | ||
6995 | @deftypefn {Built-in Function} int __builtin_ctz (unsigned int x) | |
6996 | Returns the number of trailing 0-bits in @var{x}, starting at the least | |
6997 | significant bit position. If @var{x} is 0, the result is undefined. | |
6998 | @end deftypefn | |
6999 | ||
7000 | @deftypefn {Built-in Function} int __builtin_popcount (unsigned int x) | |
7001 | Returns the number of 1-bits in @var{x}. | |
7002 | @end deftypefn | |
7003 | ||
7004 | @deftypefn {Built-in Function} int __builtin_parity (unsigned int x) | |
8a36672b | 7005 | Returns the parity of @var{x}, i.e.@: the number of 1-bits in @var{x} |
2928cd7a RH |
7006 | modulo 2. |
7007 | @end deftypefn | |
7008 | ||
7009 | @deftypefn {Built-in Function} int __builtin_ffsl (unsigned long) | |
7010 | Similar to @code{__builtin_ffs}, except the argument type is | |
7011 | @code{unsigned long}. | |
7012 | @end deftypefn | |
7013 | ||
7014 | @deftypefn {Built-in Function} int __builtin_clzl (unsigned long) | |
7015 | Similar to @code{__builtin_clz}, except the argument type is | |
7016 | @code{unsigned long}. | |
7017 | @end deftypefn | |
7018 | ||
7019 | @deftypefn {Built-in Function} int __builtin_ctzl (unsigned long) | |
7020 | Similar to @code{__builtin_ctz}, except the argument type is | |
7021 | @code{unsigned long}. | |
7022 | @end deftypefn | |
7023 | ||
7024 | @deftypefn {Built-in Function} int __builtin_popcountl (unsigned long) | |
7025 | Similar to @code{__builtin_popcount}, except the argument type is | |
7026 | @code{unsigned long}. | |
7027 | @end deftypefn | |
7028 | ||
7029 | @deftypefn {Built-in Function} int __builtin_parityl (unsigned long) | |
7030 | Similar to @code{__builtin_parity}, except the argument type is | |
7031 | @code{unsigned long}. | |
7032 | @end deftypefn | |
7033 | ||
7034 | @deftypefn {Built-in Function} int __builtin_ffsll (unsigned long long) | |
7035 | Similar to @code{__builtin_ffs}, except the argument type is | |
7036 | @code{unsigned long long}. | |
7037 | @end deftypefn | |
7038 | ||
7039 | @deftypefn {Built-in Function} int __builtin_clzll (unsigned long long) | |
7040 | Similar to @code{__builtin_clz}, except the argument type is | |
7041 | @code{unsigned long long}. | |
7042 | @end deftypefn | |
7043 | ||
7044 | @deftypefn {Built-in Function} int __builtin_ctzll (unsigned long long) | |
7045 | Similar to @code{__builtin_ctz}, except the argument type is | |
7046 | @code{unsigned long long}. | |
7047 | @end deftypefn | |
7048 | ||
7049 | @deftypefn {Built-in Function} int __builtin_popcountll (unsigned long long) | |
7050 | Similar to @code{__builtin_popcount}, except the argument type is | |
7051 | @code{unsigned long long}. | |
7052 | @end deftypefn | |
7053 | ||
7054 | @deftypefn {Built-in Function} int __builtin_parityll (unsigned long long) | |
7055 | Similar to @code{__builtin_parity}, except the argument type is | |
7056 | @code{unsigned long long}. | |
7057 | @end deftypefn | |
7058 | ||
17684d46 RG |
7059 | @deftypefn {Built-in Function} double __builtin_powi (double, int) |
7060 | Returns the first argument raised to the power of the second. Unlike the | |
7061 | @code{pow} function no guarantees about precision and rounding are made. | |
7062 | @end deftypefn | |
7063 | ||
7064 | @deftypefn {Built-in Function} float __builtin_powif (float, int) | |
7065 | Similar to @code{__builtin_powi}, except the argument and return types | |
7066 | are @code{float}. | |
7067 | @end deftypefn | |
7068 | ||
7069 | @deftypefn {Built-in Function} {long double} __builtin_powil (long double, int) | |
7070 | Similar to @code{__builtin_powi}, except the argument and return types | |
7071 | are @code{long double}. | |
7072 | @end deftypefn | |
7073 | ||
167fa32c EC |
7074 | @deftypefn {Built-in Function} int32_t __builtin_bswap32 (int32_t x) |
7075 | Returns @var{x} with the order of the bytes reversed; for example, | |
7076 | @code{0xaabbccdd} becomes @code{0xddccbbaa}. Byte here always means | |
7077 | exactly 8 bits. | |
7078 | @end deftypefn | |
7079 | ||
7080 | @deftypefn {Built-in Function} int64_t __builtin_bswap64 (int64_t x) | |
7081 | Similar to @code{__builtin_bswap32}, except the argument and return types | |
7082 | are 64-bit. | |
7083 | @end deftypefn | |
2928cd7a | 7084 | |
0975678f JM |
7085 | @node Target Builtins |
7086 | @section Built-in Functions Specific to Particular Target Machines | |
7087 | ||
7088 | On some target machines, GCC supports many built-in functions specific | |
7089 | to those machines. Generally these generate calls to specific machine | |
7090 | instructions, but allow the compiler to schedule those calls. | |
7091 | ||
7092 | @menu | |
6d8fd7bb | 7093 | * Alpha Built-in Functions:: |
88f77cba JB |
7094 | * ARM iWMMXt Built-in Functions:: |
7095 | * ARM NEON Intrinsics:: | |
161c21b6 | 7096 | * Blackfin Built-in Functions:: |
c3ee0579 | 7097 | * FR-V Built-in Functions:: |
0975678f | 7098 | * X86 Built-in Functions:: |
118ea793 | 7099 | * MIPS DSP Built-in Functions:: |
d840bfd3 | 7100 | * MIPS Paired-Single Support:: |
93581857 | 7101 | * MIPS Loongson Built-in Functions:: |
4d210b07 | 7102 | * Other MIPS Built-in Functions:: |
358da97e | 7103 | * picoChip Built-in Functions:: |
333c8841 | 7104 | * PowerPC AltiVec Built-in Functions:: |
c5145ceb | 7105 | * SPARC VIS Built-in Functions:: |
85d9c13c | 7106 | * SPU Built-in Functions:: |
0975678f JM |
7107 | @end menu |
7108 | ||
6d8fd7bb RH |
7109 | @node Alpha Built-in Functions |
7110 | @subsection Alpha Built-in Functions | |
7111 | ||
7112 | These built-in functions are available for the Alpha family of | |
7113 | processors, depending on the command-line switches used. | |
7114 | ||
95b1627e | 7115 | The following built-in functions are always available. They |
6d8fd7bb RH |
7116 | all generate the machine instruction that is part of the name. |
7117 | ||
3ab51846 | 7118 | @smallexample |
6d8fd7bb RH |
7119 | long __builtin_alpha_implver (void) |
7120 | long __builtin_alpha_rpcc (void) | |
7121 | long __builtin_alpha_amask (long) | |
7122 | long __builtin_alpha_cmpbge (long, long) | |
c4b50f1a RH |
7123 | long __builtin_alpha_extbl (long, long) |
7124 | long __builtin_alpha_extwl (long, long) | |
7125 | long __builtin_alpha_extll (long, long) | |
6d8fd7bb | 7126 | long __builtin_alpha_extql (long, long) |
c4b50f1a RH |
7127 | long __builtin_alpha_extwh (long, long) |
7128 | long __builtin_alpha_extlh (long, long) | |
6d8fd7bb | 7129 | long __builtin_alpha_extqh (long, long) |
c4b50f1a RH |
7130 | long __builtin_alpha_insbl (long, long) |
7131 | long __builtin_alpha_inswl (long, long) | |
7132 | long __builtin_alpha_insll (long, long) | |
7133 | long __builtin_alpha_insql (long, long) | |
7134 | long __builtin_alpha_inswh (long, long) | |
7135 | long __builtin_alpha_inslh (long, long) | |
7136 | long __builtin_alpha_insqh (long, long) | |
7137 | long __builtin_alpha_mskbl (long, long) | |
7138 | long __builtin_alpha_mskwl (long, long) | |
7139 | long __builtin_alpha_mskll (long, long) | |
7140 | long __builtin_alpha_mskql (long, long) | |
7141 | long __builtin_alpha_mskwh (long, long) | |
7142 | long __builtin_alpha_msklh (long, long) | |
7143 | long __builtin_alpha_mskqh (long, long) | |
7144 | long __builtin_alpha_umulh (long, long) | |
6d8fd7bb RH |
7145 | long __builtin_alpha_zap (long, long) |
7146 | long __builtin_alpha_zapnot (long, long) | |
3ab51846 | 7147 | @end smallexample |
6d8fd7bb RH |
7148 | |
7149 | The following built-in functions are always with @option{-mmax} | |
7150 | or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{pca56} or | |
7151 | later. They all generate the machine instruction that is part | |
7152 | of the name. | |
7153 | ||
3ab51846 | 7154 | @smallexample |
6d8fd7bb RH |
7155 | long __builtin_alpha_pklb (long) |
7156 | long __builtin_alpha_pkwb (long) | |
7157 | long __builtin_alpha_unpkbl (long) | |
7158 | long __builtin_alpha_unpkbw (long) | |
7159 | long __builtin_alpha_minub8 (long, long) | |
7160 | long __builtin_alpha_minsb8 (long, long) | |
7161 | long __builtin_alpha_minuw4 (long, long) | |
7162 | long __builtin_alpha_minsw4 (long, long) | |
7163 | long __builtin_alpha_maxub8 (long, long) | |
7164 | long __builtin_alpha_maxsb8 (long, long) | |
7165 | long __builtin_alpha_maxuw4 (long, long) | |
7166 | long __builtin_alpha_maxsw4 (long, long) | |
7167 | long __builtin_alpha_perr (long, long) | |
3ab51846 | 7168 | @end smallexample |
6d8fd7bb | 7169 | |
c4b50f1a RH |
7170 | The following built-in functions are always with @option{-mcix} |
7171 | or @option{-mcpu=@var{cpu}} where @var{cpu} is @code{ev67} or | |
7172 | later. They all generate the machine instruction that is part | |
7173 | of the name. | |
7174 | ||
3ab51846 | 7175 | @smallexample |
c4b50f1a RH |
7176 | long __builtin_alpha_cttz (long) |
7177 | long __builtin_alpha_ctlz (long) | |
7178 | long __builtin_alpha_ctpop (long) | |
3ab51846 | 7179 | @end smallexample |
c4b50f1a | 7180 | |
116b7a5e RH |
7181 | The following builtins are available on systems that use the OSF/1 |
7182 | PALcode. Normally they invoke the @code{rduniq} and @code{wruniq} | |
7183 | PAL calls, but when invoked with @option{-mtls-kernel}, they invoke | |
7184 | @code{rdval} and @code{wrval}. | |
7185 | ||
3ab51846 | 7186 | @smallexample |
116b7a5e RH |
7187 | void *__builtin_thread_pointer (void) |
7188 | void __builtin_set_thread_pointer (void *) | |
3ab51846 | 7189 | @end smallexample |
116b7a5e | 7190 | |
88f77cba JB |
7191 | @node ARM iWMMXt Built-in Functions |
7192 | @subsection ARM iWMMXt Built-in Functions | |
4bc73018 NC |
7193 | |
7194 | These built-in functions are available for the ARM family of | |
88f77cba | 7195 | processors when the @option{-mcpu=iwmmxt} switch is used: |
4bc73018 | 7196 | |
3ab51846 | 7197 | @smallexample |
d63851eb ILT |
7198 | typedef int v2si __attribute__ ((vector_size (8))); |
7199 | typedef short v4hi __attribute__ ((vector_size (8))); | |
7200 | typedef char v8qi __attribute__ ((vector_size (8))); | |
7201 | ||
7202 | int __builtin_arm_getwcx (int) | |
7203 | void __builtin_arm_setwcx (int, int) | |
7204 | int __builtin_arm_textrmsb (v8qi, int) | |
7205 | int __builtin_arm_textrmsh (v4hi, int) | |
7206 | int __builtin_arm_textrmsw (v2si, int) | |
7207 | int __builtin_arm_textrmub (v8qi, int) | |
7208 | int __builtin_arm_textrmuh (v4hi, int) | |
7209 | int __builtin_arm_textrmuw (v2si, int) | |
7210 | v8qi __builtin_arm_tinsrb (v8qi, int) | |
7211 | v4hi __builtin_arm_tinsrh (v4hi, int) | |
7212 | v2si __builtin_arm_tinsrw (v2si, int) | |
7213 | long long __builtin_arm_tmia (long long, int, int) | |
7214 | long long __builtin_arm_tmiabb (long long, int, int) | |
7215 | long long __builtin_arm_tmiabt (long long, int, int) | |
7216 | long long __builtin_arm_tmiaph (long long, int, int) | |
7217 | long long __builtin_arm_tmiatb (long long, int, int) | |
7218 | long long __builtin_arm_tmiatt (long long, int, int) | |
7219 | int __builtin_arm_tmovmskb (v8qi) | |
7220 | int __builtin_arm_tmovmskh (v4hi) | |
7221 | int __builtin_arm_tmovmskw (v2si) | |
7222 | long long __builtin_arm_waccb (v8qi) | |
7223 | long long __builtin_arm_wacch (v4hi) | |
7224 | long long __builtin_arm_waccw (v2si) | |
7225 | v8qi __builtin_arm_waddb (v8qi, v8qi) | |
7226 | v8qi __builtin_arm_waddbss (v8qi, v8qi) | |
7227 | v8qi __builtin_arm_waddbus (v8qi, v8qi) | |
7228 | v4hi __builtin_arm_waddh (v4hi, v4hi) | |
7229 | v4hi __builtin_arm_waddhss (v4hi, v4hi) | |
7230 | v4hi __builtin_arm_waddhus (v4hi, v4hi) | |
4bc73018 | 7231 | v2si __builtin_arm_waddw (v2si, v2si) |
4bc73018 | 7232 | v2si __builtin_arm_waddwss (v2si, v2si) |
4bc73018 | 7233 | v2si __builtin_arm_waddwus (v2si, v2si) |
d63851eb ILT |
7234 | v8qi __builtin_arm_walign (v8qi, v8qi, int) |
7235 | long long __builtin_arm_wand(long long, long long) | |
7236 | long long __builtin_arm_wandn (long long, long long) | |
7237 | v8qi __builtin_arm_wavg2b (v8qi, v8qi) | |
7238 | v8qi __builtin_arm_wavg2br (v8qi, v8qi) | |
7239 | v4hi __builtin_arm_wavg2h (v4hi, v4hi) | |
7240 | v4hi __builtin_arm_wavg2hr (v4hi, v4hi) | |
7241 | v8qi __builtin_arm_wcmpeqb (v8qi, v8qi) | |
7242 | v4hi __builtin_arm_wcmpeqh (v4hi, v4hi) | |
4bc73018 | 7243 | v2si __builtin_arm_wcmpeqw (v2si, v2si) |
d63851eb ILT |
7244 | v8qi __builtin_arm_wcmpgtsb (v8qi, v8qi) |
7245 | v4hi __builtin_arm_wcmpgtsh (v4hi, v4hi) | |
4bc73018 | 7246 | v2si __builtin_arm_wcmpgtsw (v2si, v2si) |
d63851eb ILT |
7247 | v8qi __builtin_arm_wcmpgtub (v8qi, v8qi) |
7248 | v4hi __builtin_arm_wcmpgtuh (v4hi, v4hi) | |
7249 | v2si __builtin_arm_wcmpgtuw (v2si, v2si) | |
7250 | long long __builtin_arm_wmacs (long long, v4hi, v4hi) | |
7251 | long long __builtin_arm_wmacsz (v4hi, v4hi) | |
7252 | long long __builtin_arm_wmacu (long long, v4hi, v4hi) | |
7253 | long long __builtin_arm_wmacuz (v4hi, v4hi) | |
7254 | v4hi __builtin_arm_wmadds (v4hi, v4hi) | |
7255 | v4hi __builtin_arm_wmaddu (v4hi, v4hi) | |
7256 | v8qi __builtin_arm_wmaxsb (v8qi, v8qi) | |
7257 | v4hi __builtin_arm_wmaxsh (v4hi, v4hi) | |
4bc73018 | 7258 | v2si __builtin_arm_wmaxsw (v2si, v2si) |
d63851eb ILT |
7259 | v8qi __builtin_arm_wmaxub (v8qi, v8qi) |
7260 | v4hi __builtin_arm_wmaxuh (v4hi, v4hi) | |
4bc73018 | 7261 | v2si __builtin_arm_wmaxuw (v2si, v2si) |
d63851eb ILT |
7262 | v8qi __builtin_arm_wminsb (v8qi, v8qi) |
7263 | v4hi __builtin_arm_wminsh (v4hi, v4hi) | |
4bc73018 | 7264 | v2si __builtin_arm_wminsw (v2si, v2si) |
d63851eb ILT |
7265 | v8qi __builtin_arm_wminub (v8qi, v8qi) |
7266 | v4hi __builtin_arm_wminuh (v4hi, v4hi) | |
4bc73018 | 7267 | v2si __builtin_arm_wminuw (v2si, v2si) |
d63851eb ILT |
7268 | v4hi __builtin_arm_wmulsm (v4hi, v4hi) |
7269 | v4hi __builtin_arm_wmulul (v4hi, v4hi) | |
7270 | v4hi __builtin_arm_wmulum (v4hi, v4hi) | |
7271 | long long __builtin_arm_wor (long long, long long) | |
7272 | v2si __builtin_arm_wpackdss (long long, long long) | |
7273 | v2si __builtin_arm_wpackdus (long long, long long) | |
7274 | v8qi __builtin_arm_wpackhss (v4hi, v4hi) | |
7275 | v8qi __builtin_arm_wpackhus (v4hi, v4hi) | |
7276 | v4hi __builtin_arm_wpackwss (v2si, v2si) | |
7277 | v4hi __builtin_arm_wpackwus (v2si, v2si) | |
7278 | long long __builtin_arm_wrord (long long, long long) | |
7279 | long long __builtin_arm_wrordi (long long, int) | |
7280 | v4hi __builtin_arm_wrorh (v4hi, long long) | |
7281 | v4hi __builtin_arm_wrorhi (v4hi, int) | |
7282 | v2si __builtin_arm_wrorw (v2si, long long) | |
7283 | v2si __builtin_arm_wrorwi (v2si, int) | |
7284 | v2si __builtin_arm_wsadb (v8qi, v8qi) | |
7285 | v2si __builtin_arm_wsadbz (v8qi, v8qi) | |
7286 | v2si __builtin_arm_wsadh (v4hi, v4hi) | |
7287 | v2si __builtin_arm_wsadhz (v4hi, v4hi) | |
7288 | v4hi __builtin_arm_wshufh (v4hi, int) | |
7289 | long long __builtin_arm_wslld (long long, long long) | |
7290 | long long __builtin_arm_wslldi (long long, int) | |
7291 | v4hi __builtin_arm_wsllh (v4hi, long long) | |
7292 | v4hi __builtin_arm_wsllhi (v4hi, int) | |
7293 | v2si __builtin_arm_wsllw (v2si, long long) | |
4bc73018 | 7294 | v2si __builtin_arm_wsllwi (v2si, int) |
d63851eb ILT |
7295 | long long __builtin_arm_wsrad (long long, long long) |
7296 | long long __builtin_arm_wsradi (long long, int) | |
7297 | v4hi __builtin_arm_wsrah (v4hi, long long) | |
7298 | v4hi __builtin_arm_wsrahi (v4hi, int) | |
7299 | v2si __builtin_arm_wsraw (v2si, long long) | |
4bc73018 | 7300 | v2si __builtin_arm_wsrawi (v2si, int) |
d63851eb ILT |
7301 | long long __builtin_arm_wsrld (long long, long long) |
7302 | long long __builtin_arm_wsrldi (long long, int) | |
7303 | v4hi __builtin_arm_wsrlh (v4hi, long long) | |
7304 | v4hi __builtin_arm_wsrlhi (v4hi, int) | |
7305 | v2si __builtin_arm_wsrlw (v2si, long long) | |
4bc73018 | 7306 | v2si __builtin_arm_wsrlwi (v2si, int) |
d63851eb ILT |
7307 | v8qi __builtin_arm_wsubb (v8qi, v8qi) |
7308 | v8qi __builtin_arm_wsubbss (v8qi, v8qi) | |
7309 | v8qi __builtin_arm_wsubbus (v8qi, v8qi) | |
7310 | v4hi __builtin_arm_wsubh (v4hi, v4hi) | |
7311 | v4hi __builtin_arm_wsubhss (v4hi, v4hi) | |
7312 | v4hi __builtin_arm_wsubhus (v4hi, v4hi) | |
7313 | v2si __builtin_arm_wsubw (v2si, v2si) | |
7314 | v2si __builtin_arm_wsubwss (v2si, v2si) | |
7315 | v2si __builtin_arm_wsubwus (v2si, v2si) | |
7316 | v4hi __builtin_arm_wunpckehsb (v8qi) | |
7317 | v2si __builtin_arm_wunpckehsh (v4hi) | |
7318 | long long __builtin_arm_wunpckehsw (v2si) | |
7319 | v4hi __builtin_arm_wunpckehub (v8qi) | |
7320 | v2si __builtin_arm_wunpckehuh (v4hi) | |
7321 | long long __builtin_arm_wunpckehuw (v2si) | |
7322 | v4hi __builtin_arm_wunpckelsb (v8qi) | |
7323 | v2si __builtin_arm_wunpckelsh (v4hi) | |
7324 | long long __builtin_arm_wunpckelsw (v2si) | |
7325 | v4hi __builtin_arm_wunpckelub (v8qi) | |
7326 | v2si __builtin_arm_wunpckeluh (v4hi) | |
7327 | long long __builtin_arm_wunpckeluw (v2si) | |
7328 | v8qi __builtin_arm_wunpckihb (v8qi, v8qi) | |
7329 | v4hi __builtin_arm_wunpckihh (v4hi, v4hi) | |
4bc73018 | 7330 | v2si __builtin_arm_wunpckihw (v2si, v2si) |
d63851eb ILT |
7331 | v8qi __builtin_arm_wunpckilb (v8qi, v8qi) |
7332 | v4hi __builtin_arm_wunpckilh (v4hi, v4hi) | |
4bc73018 | 7333 | v2si __builtin_arm_wunpckilw (v2si, v2si) |
d63851eb ILT |
7334 | long long __builtin_arm_wxor (long long, long long) |
7335 | long long __builtin_arm_wzero () | |
3ab51846 | 7336 | @end smallexample |
4bc73018 | 7337 | |
88f77cba JB |
7338 | @node ARM NEON Intrinsics |
7339 | @subsection ARM NEON Intrinsics | |
7340 | ||
7341 | These built-in intrinsics for the ARM Advanced SIMD extension are available | |
7342 | when the @option{-mfpu=neon} switch is used: | |
7343 | ||
7344 | @include arm-neon-intrinsics.texi | |
7345 | ||
161c21b6 BS |
7346 | @node Blackfin Built-in Functions |
7347 | @subsection Blackfin Built-in Functions | |
7348 | ||
7349 | Currently, there are two Blackfin-specific built-in functions. These are | |
7350 | used for generating @code{CSYNC} and @code{SSYNC} machine insns without | |
7351 | using inline assembly; by using these built-in functions the compiler can | |
7352 | automatically add workarounds for hardware errata involving these | |
7353 | instructions. These functions are named as follows: | |
7354 | ||
7355 | @smallexample | |
7356 | void __builtin_bfin_csync (void) | |
7357 | void __builtin_bfin_ssync (void) | |
7358 | @end smallexample | |
7359 | ||
c3ee0579 RS |
7360 | @node FR-V Built-in Functions |
7361 | @subsection FR-V Built-in Functions | |
7362 | ||
7363 | GCC provides many FR-V-specific built-in functions. In general, | |
7364 | these functions are intended to be compatible with those described | |
7365 | by @cite{FR-V Family, Softune C/C++ Compiler Manual (V6), Fujitsu | |
7366 | Semiconductor}. The two exceptions are @code{__MDUNPACKH} and | |
7367 | @code{__MBTOHE}, the gcc forms of which pass 128-bit values by | |
7368 | pointer rather than by value. | |
7369 | ||
7370 | Most of the functions are named after specific FR-V instructions. | |
27ef2cdd | 7371 | Such functions are said to be ``directly mapped'' and are summarized |
c3ee0579 RS |
7372 | here in tabular form. |
7373 | ||
7374 | @menu | |
7375 | * Argument Types:: | |
7376 | * Directly-mapped Integer Functions:: | |
7377 | * Directly-mapped Media Functions:: | |
c14ff86e | 7378 | * Raw read/write Functions:: |
c3ee0579 RS |
7379 | * Other Built-in Functions:: |
7380 | @end menu | |
7381 | ||
7382 | @node Argument Types | |
7383 | @subsubsection Argument Types | |
7384 | ||
7385 | The arguments to the built-in functions can be divided into three groups: | |
7386 | register numbers, compile-time constants and run-time values. In order | |
7387 | to make this classification clear at a glance, the arguments and return | |
7388 | values are given the following pseudo types: | |
7389 | ||
7390 | @multitable @columnfractions .20 .30 .15 .35 | |
7391 | @item Pseudo type @tab Real C type @tab Constant? @tab Description | |
7392 | @item @code{uh} @tab @code{unsigned short} @tab No @tab an unsigned halfword | |
7393 | @item @code{uw1} @tab @code{unsigned int} @tab No @tab an unsigned word | |
7394 | @item @code{sw1} @tab @code{int} @tab No @tab a signed word | |
7395 | @item @code{uw2} @tab @code{unsigned long long} @tab No | |
7396 | @tab an unsigned doubleword | |
7397 | @item @code{sw2} @tab @code{long long} @tab No @tab a signed doubleword | |
7398 | @item @code{const} @tab @code{int} @tab Yes @tab an integer constant | |
7399 | @item @code{acc} @tab @code{int} @tab Yes @tab an ACC register number | |
7400 | @item @code{iacc} @tab @code{int} @tab Yes @tab an IACC register number | |
7401 | @end multitable | |
7402 | ||
7403 | These pseudo types are not defined by GCC, they are simply a notational | |
7404 | convenience used in this manual. | |
7405 | ||
7406 | Arguments of type @code{uh}, @code{uw1}, @code{sw1}, @code{uw2} | |
7407 | and @code{sw2} are evaluated at run time. They correspond to | |
7408 | register operands in the underlying FR-V instructions. | |
7409 | ||
7410 | @code{const} arguments represent immediate operands in the underlying | |
7411 | FR-V instructions. They must be compile-time constants. | |
7412 | ||
7413 | @code{acc} arguments are evaluated at compile time and specify the number | |
7414 | of an accumulator register. For example, an @code{acc} argument of 2 | |
7415 | will select the ACC2 register. | |
7416 | ||
7417 | @code{iacc} arguments are similar to @code{acc} arguments but specify the | |
7418 | number of an IACC register. See @pxref{Other Built-in Functions} | |
7419 | for more details. | |
7420 | ||
7421 | @node Directly-mapped Integer Functions | |
7422 | @subsubsection Directly-mapped Integer Functions | |
7423 | ||
7424 | The functions listed below map directly to FR-V I-type instructions. | |
7425 | ||
7426 | @multitable @columnfractions .45 .32 .23 | |
7427 | @item Function prototype @tab Example usage @tab Assembly output | |
7428 | @item @code{sw1 __ADDSS (sw1, sw1)} | |
7429 | @tab @code{@var{c} = __ADDSS (@var{a}, @var{b})} | |
7430 | @tab @code{ADDSS @var{a},@var{b},@var{c}} | |
7431 | @item @code{sw1 __SCAN (sw1, sw1)} | |
7432 | @tab @code{@var{c} = __SCAN (@var{a}, @var{b})} | |
7433 | @tab @code{SCAN @var{a},@var{b},@var{c}} | |
7434 | @item @code{sw1 __SCUTSS (sw1)} | |
7435 | @tab @code{@var{b} = __SCUTSS (@var{a})} | |
7436 | @tab @code{SCUTSS @var{a},@var{b}} | |
7437 | @item @code{sw1 __SLASS (sw1, sw1)} | |
7438 | @tab @code{@var{c} = __SLASS (@var{a}, @var{b})} | |
7439 | @tab @code{SLASS @var{a},@var{b},@var{c}} | |
7440 | @item @code{void __SMASS (sw1, sw1)} | |
7441 | @tab @code{__SMASS (@var{a}, @var{b})} | |
7442 | @tab @code{SMASS @var{a},@var{b}} | |
7443 | @item @code{void __SMSSS (sw1, sw1)} | |
7444 | @tab @code{__SMSSS (@var{a}, @var{b})} | |
7445 | @tab @code{SMSSS @var{a},@var{b}} | |
7446 | @item @code{void __SMU (sw1, sw1)} | |
7447 | @tab @code{__SMU (@var{a}, @var{b})} | |
7448 | @tab @code{SMU @var{a},@var{b}} | |
7449 | @item @code{sw2 __SMUL (sw1, sw1)} | |
7450 | @tab @code{@var{c} = __SMUL (@var{a}, @var{b})} | |
7451 | @tab @code{SMUL @var{a},@var{b},@var{c}} | |
7452 | @item @code{sw1 __SUBSS (sw1, sw1)} | |
7453 | @tab @code{@var{c} = __SUBSS (@var{a}, @var{b})} | |
7454 | @tab @code{SUBSS @var{a},@var{b},@var{c}} | |
7455 | @item @code{uw2 __UMUL (uw1, uw1)} | |
7456 | @tab @code{@var{c} = __UMUL (@var{a}, @var{b})} | |
7457 | @tab @code{UMUL @var{a},@var{b},@var{c}} | |
7458 | @end multitable | |
7459 | ||
7460 | @node Directly-mapped Media Functions | |
7461 | @subsubsection Directly-mapped Media Functions | |
7462 | ||
7463 | The functions listed below map directly to FR-V M-type instructions. | |
7464 | ||
7465 | @multitable @columnfractions .45 .32 .23 | |
7466 | @item Function prototype @tab Example usage @tab Assembly output | |
7467 | @item @code{uw1 __MABSHS (sw1)} | |
7468 | @tab @code{@var{b} = __MABSHS (@var{a})} | |
7469 | @tab @code{MABSHS @var{a},@var{b}} | |
7470 | @item @code{void __MADDACCS (acc, acc)} | |
7471 | @tab @code{__MADDACCS (@var{b}, @var{a})} | |
7472 | @tab @code{MADDACCS @var{a},@var{b}} | |
7473 | @item @code{sw1 __MADDHSS (sw1, sw1)} | |
7474 | @tab @code{@var{c} = __MADDHSS (@var{a}, @var{b})} | |
7475 | @tab @code{MADDHSS @var{a},@var{b},@var{c}} | |
7476 | @item @code{uw1 __MADDHUS (uw1, uw1)} | |
7477 | @tab @code{@var{c} = __MADDHUS (@var{a}, @var{b})} | |
7478 | @tab @code{MADDHUS @var{a},@var{b},@var{c}} | |
7479 | @item @code{uw1 __MAND (uw1, uw1)} | |
7480 | @tab @code{@var{c} = __MAND (@var{a}, @var{b})} | |
7481 | @tab @code{MAND @var{a},@var{b},@var{c}} | |
7482 | @item @code{void __MASACCS (acc, acc)} | |
7483 | @tab @code{__MASACCS (@var{b}, @var{a})} | |
7484 | @tab @code{MASACCS @var{a},@var{b}} | |
7485 | @item @code{uw1 __MAVEH (uw1, uw1)} | |
7486 | @tab @code{@var{c} = __MAVEH (@var{a}, @var{b})} | |
7487 | @tab @code{MAVEH @var{a},@var{b},@var{c}} | |
7488 | @item @code{uw2 __MBTOH (uw1)} | |
7489 | @tab @code{@var{b} = __MBTOH (@var{a})} | |
7490 | @tab @code{MBTOH @var{a},@var{b}} | |
7491 | @item @code{void __MBTOHE (uw1 *, uw1)} | |
7492 | @tab @code{__MBTOHE (&@var{b}, @var{a})} | |
7493 | @tab @code{MBTOHE @var{a},@var{b}} | |
7494 | @item @code{void __MCLRACC (acc)} | |
7495 | @tab @code{__MCLRACC (@var{a})} | |
7496 | @tab @code{MCLRACC @var{a}} | |
7497 | @item @code{void __MCLRACCA (void)} | |
7498 | @tab @code{__MCLRACCA ()} | |
7499 | @tab @code{MCLRACCA} | |
7500 | @item @code{uw1 __Mcop1 (uw1, uw1)} | |
7501 | @tab @code{@var{c} = __Mcop1 (@var{a}, @var{b})} | |
7502 | @tab @code{Mcop1 @var{a},@var{b},@var{c}} | |
7503 | @item @code{uw1 __Mcop2 (uw1, uw1)} | |
7504 | @tab @code{@var{c} = __Mcop2 (@var{a}, @var{b})} | |
7505 | @tab @code{Mcop2 @var{a},@var{b},@var{c}} | |
7506 | @item @code{uw1 __MCPLHI (uw2, const)} | |
7507 | @tab @code{@var{c} = __MCPLHI (@var{a}, @var{b})} | |
7508 | @tab @code{MCPLHI @var{a},#@var{b},@var{c}} | |
7509 | @item @code{uw1 __MCPLI (uw2, const)} | |
7510 | @tab @code{@var{c} = __MCPLI (@var{a}, @var{b})} | |
7511 | @tab @code{MCPLI @var{a},#@var{b},@var{c}} | |
7512 | @item @code{void __MCPXIS (acc, sw1, sw1)} | |
7513 | @tab @code{__MCPXIS (@var{c}, @var{a}, @var{b})} | |
7514 | @tab @code{MCPXIS @var{a},@var{b},@var{c}} | |
7515 | @item @code{void __MCPXIU (acc, uw1, uw1)} | |
7516 | @tab @code{__MCPXIU (@var{c}, @var{a}, @var{b})} | |
7517 | @tab @code{MCPXIU @var{a},@var{b},@var{c}} | |
7518 | @item @code{void __MCPXRS (acc, sw1, sw1)} | |
7519 | @tab @code{__MCPXRS (@var{c}, @var{a}, @var{b})} | |
7520 | @tab @code{MCPXRS @var{a},@var{b},@var{c}} | |
7521 | @item @code{void __MCPXRU (acc, uw1, uw1)} | |
7522 | @tab @code{__MCPXRU (@var{c}, @var{a}, @var{b})} | |
7523 | @tab @code{MCPXRU @var{a},@var{b},@var{c}} | |
7524 | @item @code{uw1 __MCUT (acc, uw1)} | |
7525 | @tab @code{@var{c} = __MCUT (@var{a}, @var{b})} | |
7526 | @tab @code{MCUT @var{a},@var{b},@var{c}} | |
7527 | @item @code{uw1 __MCUTSS (acc, sw1)} | |
7528 | @tab @code{@var{c} = __MCUTSS (@var{a}, @var{b})} | |
7529 | @tab @code{MCUTSS @var{a},@var{b},@var{c}} | |
7530 | @item @code{void __MDADDACCS (acc, acc)} | |
7531 | @tab @code{__MDADDACCS (@var{b}, @var{a})} | |
7532 | @tab @code{MDADDACCS @var{a},@var{b}} | |
7533 | @item @code{void __MDASACCS (acc, acc)} | |
7534 | @tab @code{__MDASACCS (@var{b}, @var{a})} | |
7535 | @tab @code{MDASACCS @var{a},@var{b}} | |
7536 | @item @code{uw2 __MDCUTSSI (acc, const)} | |
7537 | @tab @code{@var{c} = __MDCUTSSI (@var{a}, @var{b})} | |
7538 | @tab @code{MDCUTSSI @var{a},#@var{b},@var{c}} | |
7539 | @item @code{uw2 __MDPACKH (uw2, uw2)} | |
7540 | @tab @code{@var{c} = __MDPACKH (@var{a}, @var{b})} | |
7541 | @tab @code{MDPACKH @var{a},@var{b},@var{c}} | |
7542 | @item @code{uw2 __MDROTLI (uw2, const)} | |
7543 | @tab @code{@var{c} = __MDROTLI (@var{a}, @var{b})} | |
7544 | @tab @code{MDROTLI @var{a},#@var{b},@var{c}} | |
7545 | @item @code{void __MDSUBACCS (acc, acc)} | |
7546 | @tab @code{__MDSUBACCS (@var{b}, @var{a})} | |
7547 | @tab @code{MDSUBACCS @var{a},@var{b}} | |
7548 | @item @code{void __MDUNPACKH (uw1 *, uw2)} | |
7549 | @tab @code{__MDUNPACKH (&@var{b}, @var{a})} | |
7550 | @tab @code{MDUNPACKH @var{a},@var{b}} | |
7551 | @item @code{uw2 __MEXPDHD (uw1, const)} | |
7552 | @tab @code{@var{c} = __MEXPDHD (@var{a}, @var{b})} | |
7553 | @tab @code{MEXPDHD @var{a},#@var{b},@var{c}} | |
7554 | @item @code{uw1 __MEXPDHW (uw1, const)} | |
7555 | @tab @code{@var{c} = __MEXPDHW (@var{a}, @var{b})} | |
7556 | @tab @code{MEXPDHW @var{a},#@var{b},@var{c}} | |
7557 | @item @code{uw1 __MHDSETH (uw1, const)} | |
7558 | @tab @code{@var{c} = __MHDSETH (@var{a}, @var{b})} | |
7559 | @tab @code{MHDSETH @var{a},#@var{b},@var{c}} | |
7560 | @item @code{sw1 __MHDSETS (const)} | |
7561 | @tab @code{@var{b} = __MHDSETS (@var{a})} | |
7562 | @tab @code{MHDSETS #@var{a},@var{b}} | |
7563 | @item @code{uw1 __MHSETHIH (uw1, const)} | |
7564 | @tab @code{@var{b} = __MHSETHIH (@var{b}, @var{a})} | |
7565 | @tab @code{MHSETHIH #@var{a},@var{b}} | |
7566 | @item @code{sw1 __MHSETHIS (sw1, const)} | |
7567 | @tab @code{@var{b} = __MHSETHIS (@var{b}, @var{a})} | |
7568 | @tab @code{MHSETHIS #@var{a},@var{b}} | |
7569 | @item @code{uw1 __MHSETLOH (uw1, const)} | |
7570 | @tab @code{@var{b} = __MHSETLOH (@var{b}, @var{a})} | |
7571 | @tab @code{MHSETLOH #@var{a},@var{b}} | |
7572 | @item @code{sw1 __MHSETLOS (sw1, const)} | |
7573 | @tab @code{@var{b} = __MHSETLOS (@var{b}, @var{a})} | |
7574 | @tab @code{MHSETLOS #@var{a},@var{b}} | |
7575 | @item @code{uw1 __MHTOB (uw2)} | |
7576 | @tab @code{@var{b} = __MHTOB (@var{a})} | |
7577 | @tab @code{MHTOB @var{a},@var{b}} | |
7578 | @item @code{void __MMACHS (acc, sw1, sw1)} | |
7579 | @tab @code{__MMACHS (@var{c}, @var{a}, @var{b})} | |
7580 | @tab @code{MMACHS @var{a},@var{b},@var{c}} | |
7581 | @item @code{void __MMACHU (acc, uw1, uw1)} | |
7582 | @tab @code{__MMACHU (@var{c}, @var{a}, @var{b})} | |
7583 | @tab @code{MMACHU @var{a},@var{b},@var{c}} | |
7584 | @item @code{void __MMRDHS (acc, sw1, sw1)} | |
7585 | @tab @code{__MMRDHS (@var{c}, @var{a}, @var{b})} | |
7586 | @tab @code{MMRDHS @var{a},@var{b},@var{c}} | |
7587 | @item @code{void __MMRDHU (acc, uw1, uw1)} | |
7588 | @tab @code{__MMRDHU (@var{c}, @var{a}, @var{b})} | |
7589 | @tab @code{MMRDHU @var{a},@var{b},@var{c}} | |
7590 | @item @code{void __MMULHS (acc, sw1, sw1)} | |
7591 | @tab @code{__MMULHS (@var{c}, @var{a}, @var{b})} | |
7592 | @tab @code{MMULHS @var{a},@var{b},@var{c}} | |
7593 | @item @code{void __MMULHU (acc, uw1, uw1)} | |
7594 | @tab @code{__MMULHU (@var{c}, @var{a}, @var{b})} | |
7595 | @tab @code{MMULHU @var{a},@var{b},@var{c}} | |
7596 | @item @code{void __MMULXHS (acc, sw1, sw1)} | |
7597 | @tab @code{__MMULXHS (@var{c}, @var{a}, @var{b})} | |
7598 | @tab @code{MMULXHS @var{a},@var{b},@var{c}} | |
7599 | @item @code{void __MMULXHU (acc, uw1, uw1)} | |
7600 | @tab @code{__MMULXHU (@var{c}, @var{a}, @var{b})} | |
7601 | @tab @code{MMULXHU @var{a},@var{b},@var{c}} | |
7602 | @item @code{uw1 __MNOT (uw1)} | |
7603 | @tab @code{@var{b} = __MNOT (@var{a})} | |
7604 | @tab @code{MNOT @var{a},@var{b}} | |
7605 | @item @code{uw1 __MOR (uw1, uw1)} | |
7606 | @tab @code{@var{c} = __MOR (@var{a}, @var{b})} | |
7607 | @tab @code{MOR @var{a},@var{b},@var{c}} | |
7608 | @item @code{uw1 __MPACKH (uh, uh)} | |
7609 | @tab @code{@var{c} = __MPACKH (@var{a}, @var{b})} | |
7610 | @tab @code{MPACKH @var{a},@var{b},@var{c}} | |
7611 | @item @code{sw2 __MQADDHSS (sw2, sw2)} | |
7612 | @tab @code{@var{c} = __MQADDHSS (@var{a}, @var{b})} | |
7613 | @tab @code{MQADDHSS @var{a},@var{b},@var{c}} | |
7614 | @item @code{uw2 __MQADDHUS (uw2, uw2)} | |
7615 | @tab @code{@var{c} = __MQADDHUS (@var{a}, @var{b})} | |
7616 | @tab @code{MQADDHUS @var{a},@var{b},@var{c}} | |
7617 | @item @code{void __MQCPXIS (acc, sw2, sw2)} | |
7618 | @tab @code{__MQCPXIS (@var{c}, @var{a}, @var{b})} | |
7619 | @tab @code{MQCPXIS @var{a},@var{b},@var{c}} | |
7620 | @item @code{void __MQCPXIU (acc, uw2, uw2)} | |
7621 | @tab @code{__MQCPXIU (@var{c}, @var{a}, @var{b})} | |
7622 | @tab @code{MQCPXIU @var{a},@var{b},@var{c}} | |
7623 | @item @code{void __MQCPXRS (acc, sw2, sw2)} | |
7624 | @tab @code{__MQCPXRS (@var{c}, @var{a}, @var{b})} | |
7625 | @tab @code{MQCPXRS @var{a},@var{b},@var{c}} | |
7626 | @item @code{void __MQCPXRU (acc, uw2, uw2)} | |
7627 | @tab @code{__MQCPXRU (@var{c}, @var{a}, @var{b})} | |
7628 | @tab @code{MQCPXRU @var{a},@var{b},@var{c}} | |
7629 | @item @code{sw2 __MQLCLRHS (sw2, sw2)} | |
7630 | @tab @code{@var{c} = __MQLCLRHS (@var{a}, @var{b})} | |
7631 | @tab @code{MQLCLRHS @var{a},@var{b},@var{c}} | |
7632 | @item @code{sw2 __MQLMTHS (sw2, sw2)} | |
7633 | @tab @code{@var{c} = __MQLMTHS (@var{a}, @var{b})} | |
7634 | @tab @code{MQLMTHS @var{a},@var{b},@var{c}} | |
7635 | @item @code{void __MQMACHS (acc, sw2, sw2)} | |
7636 | @tab @code{__MQMACHS (@var{c}, @var{a}, @var{b})} | |
7637 | @tab @code{MQMACHS @var{a},@var{b},@var{c}} | |
7638 | @item @code{void __MQMACHU (acc, uw2, uw2)} | |
7639 | @tab @code{__MQMACHU (@var{c}, @var{a}, @var{b})} | |
7640 | @tab @code{MQMACHU @var{a},@var{b},@var{c}} | |
7641 | @item @code{void __MQMACXHS (acc, sw2, sw2)} | |
7642 | @tab @code{__MQMACXHS (@var{c}, @var{a}, @var{b})} | |
7643 | @tab @code{MQMACXHS @var{a},@var{b},@var{c}} | |
7644 | @item @code{void __MQMULHS (acc, sw2, sw2)} | |
7645 | @tab @code{__MQMULHS (@var{c}, @var{a}, @var{b})} | |
7646 | @tab @code{MQMULHS @var{a},@var{b},@var{c}} | |
7647 | @item @code{void __MQMULHU (acc, uw2, uw2)} | |
7648 | @tab @code{__MQMULHU (@var{c}, @var{a}, @var{b})} | |
7649 | @tab @code{MQMULHU @var{a},@var{b},@var{c}} | |
7650 | @item @code{void __MQMULXHS (acc, sw2, sw2)} | |
7651 | @tab @code{__MQMULXHS (@var{c}, @var{a}, @var{b})} | |
7652 | @tab @code{MQMULXHS @var{a},@var{b},@var{c}} | |
7653 | @item @code{void __MQMULXHU (acc, uw2, uw2)} | |
7654 | @tab @code{__MQMULXHU (@var{c}, @var{a}, @var{b})} | |
7655 | @tab @code{MQMULXHU @var{a},@var{b},@var{c}} | |
7656 | @item @code{sw2 __MQSATHS (sw2, sw2)} | |
7657 | @tab @code{@var{c} = __MQSATHS (@var{a}, @var{b})} | |
7658 | @tab @code{MQSATHS @var{a},@var{b},@var{c}} | |
7659 | @item @code{uw2 __MQSLLHI (uw2, int)} | |
7660 | @tab @code{@var{c} = __MQSLLHI (@var{a}, @var{b})} | |
7661 | @tab @code{MQSLLHI @var{a},@var{b},@var{c}} | |
7662 | @item @code{sw2 __MQSRAHI (sw2, int)} | |
7663 | @tab @code{@var{c} = __MQSRAHI (@var{a}, @var{b})} | |
7664 | @tab @code{MQSRAHI @var{a},@var{b},@var{c}} | |
7665 | @item @code{sw2 __MQSUBHSS (sw2, sw2)} | |
7666 | @tab @code{@var{c} = __MQSUBHSS (@var{a}, @var{b})} | |
7667 | @tab @code{MQSUBHSS @var{a},@var{b},@var{c}} | |
7668 | @item @code{uw2 __MQSUBHUS (uw2, uw2)} | |
7669 | @tab @code{@var{c} = __MQSUBHUS (@var{a}, @var{b})} | |
7670 | @tab @code{MQSUBHUS @var{a},@var{b},@var{c}} | |
7671 | @item @code{void __MQXMACHS (acc, sw2, sw2)} | |
7672 | @tab @code{__MQXMACHS (@var{c}, @var{a}, @var{b})} | |
7673 | @tab @code{MQXMACHS @var{a},@var{b},@var{c}} | |
7674 | @item @code{void __MQXMACXHS (acc, sw2, sw2)} | |
7675 | @tab @code{__MQXMACXHS (@var{c}, @var{a}, @var{b})} | |
7676 | @tab @code{MQXMACXHS @var{a},@var{b},@var{c}} | |
7677 | @item @code{uw1 __MRDACC (acc)} | |
7678 | @tab @code{@var{b} = __MRDACC (@var{a})} | |
7679 | @tab @code{MRDACC @var{a},@var{b}} | |
7680 | @item @code{uw1 __MRDACCG (acc)} | |
7681 | @tab @code{@var{b} = __MRDACCG (@var{a})} | |
7682 | @tab @code{MRDACCG @var{a},@var{b}} | |
7683 | @item @code{uw1 __MROTLI (uw1, const)} | |
7684 | @tab @code{@var{c} = __MROTLI (@var{a}, @var{b})} | |
7685 | @tab @code{MROTLI @var{a},#@var{b},@var{c}} | |
7686 | @item @code{uw1 __MROTRI (uw1, const)} | |
7687 | @tab @code{@var{c} = __MROTRI (@var{a}, @var{b})} | |
7688 | @tab @code{MROTRI @var{a},#@var{b},@var{c}} | |
7689 | @item @code{sw1 __MSATHS (sw1, sw1)} | |
7690 | @tab @code{@var{c} = __MSATHS (@var{a}, @var{b})} | |
7691 | @tab @code{MSATHS @var{a},@var{b},@var{c}} | |
7692 | @item @code{uw1 __MSATHU (uw1, uw1)} | |
7693 | @tab @code{@var{c} = __MSATHU (@var{a}, @var{b})} | |
7694 | @tab @code{MSATHU @var{a},@var{b},@var{c}} | |
7695 | @item @code{uw1 __MSLLHI (uw1, const)} | |
7696 | @tab @code{@var{c} = __MSLLHI (@var{a}, @var{b})} | |
7697 | @tab @code{MSLLHI @var{a},#@var{b},@var{c}} | |
7698 | @item @code{sw1 __MSRAHI (sw1, const)} | |
7699 | @tab @code{@var{c} = __MSRAHI (@var{a}, @var{b})} | |
7700 | @tab @code{MSRAHI @var{a},#@var{b},@var{c}} | |
7701 | @item @code{uw1 __MSRLHI (uw1, const)} | |
7702 | @tab @code{@var{c} = __MSRLHI (@var{a}, @var{b})} | |
7703 | @tab @code{MSRLHI @var{a},#@var{b},@var{c}} | |
7704 | @item @code{void __MSUBACCS (acc, acc)} | |
7705 | @tab @code{__MSUBACCS (@var{b}, @var{a})} | |
7706 | @tab @code{MSUBACCS @var{a},@var{b}} | |
7707 | @item @code{sw1 __MSUBHSS (sw1, sw1)} | |
7708 | @tab @code{@var{c} = __MSUBHSS (@var{a}, @var{b})} | |
7709 | @tab @code{MSUBHSS @var{a},@var{b},@var{c}} | |
7710 | @item @code{uw1 __MSUBHUS (uw1, uw1)} | |
7711 | @tab @code{@var{c} = __MSUBHUS (@var{a}, @var{b})} | |
7712 | @tab @code{MSUBHUS @var{a},@var{b},@var{c}} | |
7713 | @item @code{void __MTRAP (void)} | |
7714 | @tab @code{__MTRAP ()} | |
7715 | @tab @code{MTRAP} | |
7716 | @item @code{uw2 __MUNPACKH (uw1)} | |
7717 | @tab @code{@var{b} = __MUNPACKH (@var{a})} | |
7718 | @tab @code{MUNPACKH @var{a},@var{b}} | |
7719 | @item @code{uw1 __MWCUT (uw2, uw1)} | |
7720 | @tab @code{@var{c} = __MWCUT (@var{a}, @var{b})} | |
7721 | @tab @code{MWCUT @var{a},@var{b},@var{c}} | |
7722 | @item @code{void __MWTACC (acc, uw1)} | |
7723 | @tab @code{__MWTACC (@var{b}, @var{a})} | |
7724 | @tab @code{MWTACC @var{a},@var{b}} | |
7725 | @item @code{void __MWTACCG (acc, uw1)} | |
7726 | @tab @code{__MWTACCG (@var{b}, @var{a})} | |
7727 | @tab @code{MWTACCG @var{a},@var{b}} | |
7728 | @item @code{uw1 __MXOR (uw1, uw1)} | |
7729 | @tab @code{@var{c} = __MXOR (@var{a}, @var{b})} | |
7730 | @tab @code{MXOR @var{a},@var{b},@var{c}} | |
7731 | @end multitable | |
7732 | ||
c14ff86e AH |
7733 | @node Raw read/write Functions |
7734 | @subsubsection Raw read/write Functions | |
7735 | ||
7736 | This sections describes built-in functions related to read and write | |
7737 | instructions to access memory. These functions generate | |
7738 | @code{membar} instructions to flush the I/O load and stores where | |
7739 | appropriate, as described in Fujitsu's manual described above. | |
7740 | ||
7741 | @table @code | |
7742 | ||
7743 | @item unsigned char __builtin_read8 (void *@var{data}) | |
7744 | @item unsigned short __builtin_read16 (void *@var{data}) | |
7745 | @item unsigned long __builtin_read32 (void *@var{data}) | |
7746 | @item unsigned long long __builtin_read64 (void *@var{data}) | |
7747 | ||
7748 | @item void __builtin_write8 (void *@var{data}, unsigned char @var{datum}) | |
7749 | @item void __builtin_write16 (void *@var{data}, unsigned short @var{datum}) | |
7750 | @item void __builtin_write32 (void *@var{data}, unsigned long @var{datum}) | |
7751 | @item void __builtin_write64 (void *@var{data}, unsigned long long @var{datum}) | |
7752 | @end table | |
7753 | ||
c3ee0579 RS |
7754 | @node Other Built-in Functions |
7755 | @subsubsection Other Built-in Functions | |
7756 | ||
7757 | This section describes built-in functions that are not named after | |
7758 | a specific FR-V instruction. | |
7759 | ||
7760 | @table @code | |
7761 | @item sw2 __IACCreadll (iacc @var{reg}) | |
7762 | Return the full 64-bit value of IACC0@. The @var{reg} argument is reserved | |
7763 | for future expansion and must be 0. | |
7764 | ||
7765 | @item sw1 __IACCreadl (iacc @var{reg}) | |
7766 | Return the value of IACC0H if @var{reg} is 0 and IACC0L if @var{reg} is 1. | |
7767 | Other values of @var{reg} are rejected as invalid. | |
7768 | ||
7769 | @item void __IACCsetll (iacc @var{reg}, sw2 @var{x}) | |
7770 | Set the full 64-bit value of IACC0 to @var{x}. The @var{reg} argument | |
7771 | is reserved for future expansion and must be 0. | |
7772 | ||
7773 | @item void __IACCsetl (iacc @var{reg}, sw1 @var{x}) | |
7774 | Set IACC0H to @var{x} if @var{reg} is 0 and IACC0L to @var{x} if @var{reg} | |
7775 | is 1. Other values of @var{reg} are rejected as invalid. | |
7776 | ||
7777 | @item void __data_prefetch0 (const void *@var{x}) | |
7778 | Use the @code{dcpl} instruction to load the contents of address @var{x} | |
7779 | into the data cache. | |
7780 | ||
7781 | @item void __data_prefetch (const void *@var{x}) | |
7782 | Use the @code{nldub} instruction to load the contents of address @var{x} | |
7783 | into the data cache. The instruction will be issued in slot I1@. | |
7784 | @end table | |
7785 | ||
0975678f JM |
7786 | @node X86 Built-in Functions |
7787 | @subsection X86 Built-in Functions | |
7788 | ||
7789 | These built-in functions are available for the i386 and x86-64 family | |
7790 | of computers, depending on the command-line switches used. | |
7791 | ||
75576871 BB |
7792 | Note that, if you specify command-line switches such as @option{-msse}, |
7793 | the compiler could use the extended instruction sets even if the built-ins | |
7794 | are not used explicitly in the program. For this reason, applications | |
7795 | which perform runtime CPU detection must compile separate files for each | |
7796 | supported architecture, using the appropriate flags. In particular, | |
7797 | the file containing the CPU detection code should be compiled without | |
7798 | these options. | |
7799 | ||
0975678f | 7800 | The following machine modes are available for use with MMX built-in functions |
333c8841 AH |
7801 | (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32-bit integers, |
7802 | @code{V4HI} for a vector of four 16-bit integers, and @code{V8QI} for a | |
7803 | vector of eight 8-bit integers. Some of the built-in functions operate on | |
75d8b30e | 7804 | MMX registers as a whole 64-bit entity, these use @code{V1DI} as their mode. |
0975678f JM |
7805 | |
7806 | If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector | |
333c8841 | 7807 | of two 32-bit floating point values. |
0975678f | 7808 | |
333c8841 AH |
7809 | If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32-bit |
7810 | floating point values. Some instructions use a vector of four 32-bit | |
0975678f | 7811 | integers, these use @code{V4SI}. Finally, some instructions operate on an |
333c8841 | 7812 | entire vector register, interpreting it as a 128-bit integer, these use mode |
0975678f JM |
7813 | @code{TI}. |
7814 | ||
27f56cb1 | 7815 | In 64-bit mode, the x86-64 family of processors uses additional built-in |
5513e239 UB |
7816 | functions for efficient use of @code{TF} (@code{__float128}) 128-bit |
7817 | floating point and @code{TC} 128-bit complex floating point values. | |
7818 | ||
27f56cb1 GP |
7819 | The following floating point built-in functions are available in 64-bit |
7820 | mode. All of them implement the function that is part of the name. | |
5513e239 UB |
7821 | |
7822 | @smallexample | |
7823 | __float128 __builtin_fabsq (__float128) | |
7824 | __float128 __builtin_copysignq (__float128, __float128) | |
7825 | @end smallexample | |
7826 | ||
7827 | The following floating point built-in functions are made available in the | |
7828 | 64-bit mode. | |
7829 | ||
7830 | @table @code | |
7831 | @item __float128 __builtin_infq (void) | |
7832 | Similar to @code{__builtin_inf}, except the return type is @code{__float128}. | |
593812b6 BE |
7833 | @findex __builtin_infq |
7834 | ||
7835 | @item __float128 __builtin_huge_valq (void) | |
7836 | Similar to @code{__builtin_huge_val}, except the return type is @code{__float128}. | |
7837 | @findex __builtin_huge_valq | |
5513e239 UB |
7838 | @end table |
7839 | ||
0975678f JM |
7840 | The following built-in functions are made available by @option{-mmmx}. |
7841 | All of them generate the machine instruction that is part of the name. | |
7842 | ||
3ab51846 | 7843 | @smallexample |
0975678f JM |
7844 | v8qi __builtin_ia32_paddb (v8qi, v8qi) |
7845 | v4hi __builtin_ia32_paddw (v4hi, v4hi) | |
7846 | v2si __builtin_ia32_paddd (v2si, v2si) | |
7847 | v8qi __builtin_ia32_psubb (v8qi, v8qi) | |
7848 | v4hi __builtin_ia32_psubw (v4hi, v4hi) | |
7849 | v2si __builtin_ia32_psubd (v2si, v2si) | |
7850 | v8qi __builtin_ia32_paddsb (v8qi, v8qi) | |
7851 | v4hi __builtin_ia32_paddsw (v4hi, v4hi) | |
7852 | v8qi __builtin_ia32_psubsb (v8qi, v8qi) | |
7853 | v4hi __builtin_ia32_psubsw (v4hi, v4hi) | |
7854 | v8qi __builtin_ia32_paddusb (v8qi, v8qi) | |
7855 | v4hi __builtin_ia32_paddusw (v4hi, v4hi) | |
7856 | v8qi __builtin_ia32_psubusb (v8qi, v8qi) | |
7857 | v4hi __builtin_ia32_psubusw (v4hi, v4hi) | |
7858 | v4hi __builtin_ia32_pmullw (v4hi, v4hi) | |
7859 | v4hi __builtin_ia32_pmulhw (v4hi, v4hi) | |
7860 | di __builtin_ia32_pand (di, di) | |
7861 | di __builtin_ia32_pandn (di,di) | |
7862 | di __builtin_ia32_por (di, di) | |
7863 | di __builtin_ia32_pxor (di, di) | |
7864 | v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi) | |
7865 | v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi) | |
7866 | v2si __builtin_ia32_pcmpeqd (v2si, v2si) | |
7867 | v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi) | |
7868 | v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi) | |
7869 | v2si __builtin_ia32_pcmpgtd (v2si, v2si) | |
7870 | v8qi __builtin_ia32_punpckhbw (v8qi, v8qi) | |
7871 | v4hi __builtin_ia32_punpckhwd (v4hi, v4hi) | |
7872 | v2si __builtin_ia32_punpckhdq (v2si, v2si) | |
7873 | v8qi __builtin_ia32_punpcklbw (v8qi, v8qi) | |
7874 | v4hi __builtin_ia32_punpcklwd (v4hi, v4hi) | |
7875 | v2si __builtin_ia32_punpckldq (v2si, v2si) | |
7876 | v8qi __builtin_ia32_packsswb (v4hi, v4hi) | |
7877 | v4hi __builtin_ia32_packssdw (v2si, v2si) | |
7878 | v8qi __builtin_ia32_packuswb (v4hi, v4hi) | |
10a97ae6 | 7879 | |
52eaae97 UB |
7880 | v4hi __builtin_ia32_psllw (v4hi, v4hi) |
7881 | v2si __builtin_ia32_pslld (v2si, v2si) | |
7882 | v1di __builtin_ia32_psllq (v1di, v1di) | |
7883 | v4hi __builtin_ia32_psrlw (v4hi, v4hi) | |
7884 | v2si __builtin_ia32_psrld (v2si, v2si) | |
7885 | v1di __builtin_ia32_psrlq (v1di, v1di) | |
7886 | v4hi __builtin_ia32_psraw (v4hi, v4hi) | |
7887 | v2si __builtin_ia32_psrad (v2si, v2si) | |
7888 | v4hi __builtin_ia32_psllwi (v4hi, int) | |
7889 | v2si __builtin_ia32_pslldi (v2si, int) | |
7890 | v1di __builtin_ia32_psllqi (v1di, int) | |
7891 | v4hi __builtin_ia32_psrlwi (v4hi, int) | |
7892 | v2si __builtin_ia32_psrldi (v2si, int) | |
7893 | v1di __builtin_ia32_psrlqi (v1di, int) | |
7894 | v4hi __builtin_ia32_psrawi (v4hi, int) | |
7895 | v2si __builtin_ia32_psradi (v2si, int) | |
10a97ae6 | 7896 | |
3ab51846 | 7897 | @end smallexample |
0975678f JM |
7898 | |
7899 | The following built-in functions are made available either with | |
7900 | @option{-msse}, or with a combination of @option{-m3dnow} and | |
7901 | @option{-march=athlon}. All of them generate the machine | |
7902 | instruction that is part of the name. | |
7903 | ||
3ab51846 | 7904 | @smallexample |
0975678f JM |
7905 | v4hi __builtin_ia32_pmulhuw (v4hi, v4hi) |
7906 | v8qi __builtin_ia32_pavgb (v8qi, v8qi) | |
7907 | v4hi __builtin_ia32_pavgw (v4hi, v4hi) | |
ab555a5b | 7908 | v1di __builtin_ia32_psadbw (v8qi, v8qi) |
0975678f JM |
7909 | v8qi __builtin_ia32_pmaxub (v8qi, v8qi) |
7910 | v4hi __builtin_ia32_pmaxsw (v4hi, v4hi) | |
7911 | v8qi __builtin_ia32_pminub (v8qi, v8qi) | |
7912 | v4hi __builtin_ia32_pminsw (v4hi, v4hi) | |
7913 | int __builtin_ia32_pextrw (v4hi, int) | |
7914 | v4hi __builtin_ia32_pinsrw (v4hi, int, int) | |
7915 | int __builtin_ia32_pmovmskb (v8qi) | |
7916 | void __builtin_ia32_maskmovq (v8qi, v8qi, char *) | |
7917 | void __builtin_ia32_movntq (di *, di) | |
7918 | void __builtin_ia32_sfence (void) | |
3ab51846 | 7919 | @end smallexample |
0975678f JM |
7920 | |
7921 | The following built-in functions are available when @option{-msse} is used. | |
7922 | All of them generate the machine instruction that is part of the name. | |
7923 | ||
3ab51846 | 7924 | @smallexample |
0975678f JM |
7925 | int __builtin_ia32_comieq (v4sf, v4sf) |
7926 | int __builtin_ia32_comineq (v4sf, v4sf) | |
7927 | int __builtin_ia32_comilt (v4sf, v4sf) | |
7928 | int __builtin_ia32_comile (v4sf, v4sf) | |
7929 | int __builtin_ia32_comigt (v4sf, v4sf) | |
7930 | int __builtin_ia32_comige (v4sf, v4sf) | |
7931 | int __builtin_ia32_ucomieq (v4sf, v4sf) | |
7932 | int __builtin_ia32_ucomineq (v4sf, v4sf) | |
7933 | int __builtin_ia32_ucomilt (v4sf, v4sf) | |
7934 | int __builtin_ia32_ucomile (v4sf, v4sf) | |
7935 | int __builtin_ia32_ucomigt (v4sf, v4sf) | |
7936 | int __builtin_ia32_ucomige (v4sf, v4sf) | |
7937 | v4sf __builtin_ia32_addps (v4sf, v4sf) | |
7938 | v4sf __builtin_ia32_subps (v4sf, v4sf) | |
7939 | v4sf __builtin_ia32_mulps (v4sf, v4sf) | |
7940 | v4sf __builtin_ia32_divps (v4sf, v4sf) | |
7941 | v4sf __builtin_ia32_addss (v4sf, v4sf) | |
7942 | v4sf __builtin_ia32_subss (v4sf, v4sf) | |
7943 | v4sf __builtin_ia32_mulss (v4sf, v4sf) | |
7944 | v4sf __builtin_ia32_divss (v4sf, v4sf) | |
7945 | v4si __builtin_ia32_cmpeqps (v4sf, v4sf) | |
7946 | v4si __builtin_ia32_cmpltps (v4sf, v4sf) | |
7947 | v4si __builtin_ia32_cmpleps (v4sf, v4sf) | |
7948 | v4si __builtin_ia32_cmpgtps (v4sf, v4sf) | |
7949 | v4si __builtin_ia32_cmpgeps (v4sf, v4sf) | |
7950 | v4si __builtin_ia32_cmpunordps (v4sf, v4sf) | |
7951 | v4si __builtin_ia32_cmpneqps (v4sf, v4sf) | |
7952 | v4si __builtin_ia32_cmpnltps (v4sf, v4sf) | |
7953 | v4si __builtin_ia32_cmpnleps (v4sf, v4sf) | |
7954 | v4si __builtin_ia32_cmpngtps (v4sf, v4sf) | |
7955 | v4si __builtin_ia32_cmpngeps (v4sf, v4sf) | |
7956 | v4si __builtin_ia32_cmpordps (v4sf, v4sf) | |
7957 | v4si __builtin_ia32_cmpeqss (v4sf, v4sf) | |
7958 | v4si __builtin_ia32_cmpltss (v4sf, v4sf) | |
7959 | v4si __builtin_ia32_cmpless (v4sf, v4sf) | |
0975678f JM |
7960 | v4si __builtin_ia32_cmpunordss (v4sf, v4sf) |
7961 | v4si __builtin_ia32_cmpneqss (v4sf, v4sf) | |
7962 | v4si __builtin_ia32_cmpnlts (v4sf, v4sf) | |
7963 | v4si __builtin_ia32_cmpnless (v4sf, v4sf) | |
0975678f JM |
7964 | v4si __builtin_ia32_cmpordss (v4sf, v4sf) |
7965 | v4sf __builtin_ia32_maxps (v4sf, v4sf) | |
7966 | v4sf __builtin_ia32_maxss (v4sf, v4sf) | |
7967 | v4sf __builtin_ia32_minps (v4sf, v4sf) | |
7968 | v4sf __builtin_ia32_minss (v4sf, v4sf) | |
7969 | v4sf __builtin_ia32_andps (v4sf, v4sf) | |
7970 | v4sf __builtin_ia32_andnps (v4sf, v4sf) | |
7971 | v4sf __builtin_ia32_orps (v4sf, v4sf) | |
7972 | v4sf __builtin_ia32_xorps (v4sf, v4sf) | |
7973 | v4sf __builtin_ia32_movss (v4sf, v4sf) | |
7974 | v4sf __builtin_ia32_movhlps (v4sf, v4sf) | |
7975 | v4sf __builtin_ia32_movlhps (v4sf, v4sf) | |
7976 | v4sf __builtin_ia32_unpckhps (v4sf, v4sf) | |
7977 | v4sf __builtin_ia32_unpcklps (v4sf, v4sf) | |
7978 | v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si) | |
7979 | v4sf __builtin_ia32_cvtsi2ss (v4sf, int) | |
7980 | v2si __builtin_ia32_cvtps2pi (v4sf) | |
7981 | int __builtin_ia32_cvtss2si (v4sf) | |
7982 | v2si __builtin_ia32_cvttps2pi (v4sf) | |
7983 | int __builtin_ia32_cvttss2si (v4sf) | |
7984 | v4sf __builtin_ia32_rcpps (v4sf) | |
7985 | v4sf __builtin_ia32_rsqrtps (v4sf) | |
7986 | v4sf __builtin_ia32_sqrtps (v4sf) | |
7987 | v4sf __builtin_ia32_rcpss (v4sf) | |
7988 | v4sf __builtin_ia32_rsqrtss (v4sf) | |
7989 | v4sf __builtin_ia32_sqrtss (v4sf) | |
7990 | v4sf __builtin_ia32_shufps (v4sf, v4sf, int) | |
7991 | void __builtin_ia32_movntps (float *, v4sf) | |
7992 | int __builtin_ia32_movmskps (v4sf) | |
3ab51846 | 7993 | @end smallexample |
0975678f JM |
7994 | |
7995 | The following built-in functions are available when @option{-msse} is used. | |
7996 | ||
7997 | @table @code | |
7998 | @item v4sf __builtin_ia32_loadaps (float *) | |
7999 | Generates the @code{movaps} machine instruction as a load from memory. | |
8000 | @item void __builtin_ia32_storeaps (float *, v4sf) | |
8001 | Generates the @code{movaps} machine instruction as a store to memory. | |
8002 | @item v4sf __builtin_ia32_loadups (float *) | |
8003 | Generates the @code{movups} machine instruction as a load from memory. | |
8004 | @item void __builtin_ia32_storeups (float *, v4sf) | |
8005 | Generates the @code{movups} machine instruction as a store to memory. | |
8006 | @item v4sf __builtin_ia32_loadsss (float *) | |
8007 | Generates the @code{movss} machine instruction as a load from memory. | |
8008 | @item void __builtin_ia32_storess (float *, v4sf) | |
8009 | Generates the @code{movss} machine instruction as a store to memory. | |
bb1418c1 | 8010 | @item v4sf __builtin_ia32_loadhps (v4sf, const v2sf *) |
0975678f | 8011 | Generates the @code{movhps} machine instruction as a load from memory. |
bb1418c1 | 8012 | @item v4sf __builtin_ia32_loadlps (v4sf, const v2sf *) |
0975678f | 8013 | Generates the @code{movlps} machine instruction as a load from memory |
bb1418c1 | 8014 | @item void __builtin_ia32_storehps (v2sf *, v4sf) |
0975678f | 8015 | Generates the @code{movhps} machine instruction as a store to memory. |
bb1418c1 | 8016 | @item void __builtin_ia32_storelps (v2sf *, v4sf) |
0975678f JM |
8017 | Generates the @code{movlps} machine instruction as a store to memory. |
8018 | @end table | |
8019 | ||
d7aa4788 RG |
8020 | The following built-in functions are available when @option{-msse2} is used. |
8021 | All of them generate the machine instruction that is part of the name. | |
8022 | ||
8023 | @smallexample | |
8024 | int __builtin_ia32_comisdeq (v2df, v2df) | |
8025 | int __builtin_ia32_comisdlt (v2df, v2df) | |
8026 | int __builtin_ia32_comisdle (v2df, v2df) | |
8027 | int __builtin_ia32_comisdgt (v2df, v2df) | |
8028 | int __builtin_ia32_comisdge (v2df, v2df) | |
8029 | int __builtin_ia32_comisdneq (v2df, v2df) | |
8030 | int __builtin_ia32_ucomisdeq (v2df, v2df) | |
8031 | int __builtin_ia32_ucomisdlt (v2df, v2df) | |
8032 | int __builtin_ia32_ucomisdle (v2df, v2df) | |
8033 | int __builtin_ia32_ucomisdgt (v2df, v2df) | |
8034 | int __builtin_ia32_ucomisdge (v2df, v2df) | |
8035 | int __builtin_ia32_ucomisdneq (v2df, v2df) | |
8036 | v2df __builtin_ia32_cmpeqpd (v2df, v2df) | |
8037 | v2df __builtin_ia32_cmpltpd (v2df, v2df) | |
8038 | v2df __builtin_ia32_cmplepd (v2df, v2df) | |
8039 | v2df __builtin_ia32_cmpgtpd (v2df, v2df) | |
8040 | v2df __builtin_ia32_cmpgepd (v2df, v2df) | |
8041 | v2df __builtin_ia32_cmpunordpd (v2df, v2df) | |
8042 | v2df __builtin_ia32_cmpneqpd (v2df, v2df) | |
8043 | v2df __builtin_ia32_cmpnltpd (v2df, v2df) | |
8044 | v2df __builtin_ia32_cmpnlepd (v2df, v2df) | |
8045 | v2df __builtin_ia32_cmpngtpd (v2df, v2df) | |
8046 | v2df __builtin_ia32_cmpngepd (v2df, v2df) | |
8047 | v2df __builtin_ia32_cmpordpd (v2df, v2df) | |
8048 | v2df __builtin_ia32_cmpeqsd (v2df, v2df) | |
8049 | v2df __builtin_ia32_cmpltsd (v2df, v2df) | |
8050 | v2df __builtin_ia32_cmplesd (v2df, v2df) | |
8051 | v2df __builtin_ia32_cmpunordsd (v2df, v2df) | |
8052 | v2df __builtin_ia32_cmpneqsd (v2df, v2df) | |
8053 | v2df __builtin_ia32_cmpnltsd (v2df, v2df) | |
8054 | v2df __builtin_ia32_cmpnlesd (v2df, v2df) | |
8055 | v2df __builtin_ia32_cmpordsd (v2df, v2df) | |
8056 | v2di __builtin_ia32_paddq (v2di, v2di) | |
8057 | v2di __builtin_ia32_psubq (v2di, v2di) | |
8058 | v2df __builtin_ia32_addpd (v2df, v2df) | |
8059 | v2df __builtin_ia32_subpd (v2df, v2df) | |
8060 | v2df __builtin_ia32_mulpd (v2df, v2df) | |
8061 | v2df __builtin_ia32_divpd (v2df, v2df) | |
8062 | v2df __builtin_ia32_addsd (v2df, v2df) | |
8063 | v2df __builtin_ia32_subsd (v2df, v2df) | |
8064 | v2df __builtin_ia32_mulsd (v2df, v2df) | |
8065 | v2df __builtin_ia32_divsd (v2df, v2df) | |
8066 | v2df __builtin_ia32_minpd (v2df, v2df) | |
8067 | v2df __builtin_ia32_maxpd (v2df, v2df) | |
8068 | v2df __builtin_ia32_minsd (v2df, v2df) | |
8069 | v2df __builtin_ia32_maxsd (v2df, v2df) | |
8070 | v2df __builtin_ia32_andpd (v2df, v2df) | |
8071 | v2df __builtin_ia32_andnpd (v2df, v2df) | |
8072 | v2df __builtin_ia32_orpd (v2df, v2df) | |
8073 | v2df __builtin_ia32_xorpd (v2df, v2df) | |
8074 | v2df __builtin_ia32_movsd (v2df, v2df) | |
8075 | v2df __builtin_ia32_unpckhpd (v2df, v2df) | |
8076 | v2df __builtin_ia32_unpcklpd (v2df, v2df) | |
8077 | v16qi __builtin_ia32_paddb128 (v16qi, v16qi) | |
8078 | v8hi __builtin_ia32_paddw128 (v8hi, v8hi) | |
8079 | v4si __builtin_ia32_paddd128 (v4si, v4si) | |
8080 | v2di __builtin_ia32_paddq128 (v2di, v2di) | |
8081 | v16qi __builtin_ia32_psubb128 (v16qi, v16qi) | |
8082 | v8hi __builtin_ia32_psubw128 (v8hi, v8hi) | |
8083 | v4si __builtin_ia32_psubd128 (v4si, v4si) | |
8084 | v2di __builtin_ia32_psubq128 (v2di, v2di) | |
8085 | v8hi __builtin_ia32_pmullw128 (v8hi, v8hi) | |
8086 | v8hi __builtin_ia32_pmulhw128 (v8hi, v8hi) | |
8087 | v2di __builtin_ia32_pand128 (v2di, v2di) | |
8088 | v2di __builtin_ia32_pandn128 (v2di, v2di) | |
8089 | v2di __builtin_ia32_por128 (v2di, v2di) | |
8090 | v2di __builtin_ia32_pxor128 (v2di, v2di) | |
8091 | v16qi __builtin_ia32_pavgb128 (v16qi, v16qi) | |
8092 | v8hi __builtin_ia32_pavgw128 (v8hi, v8hi) | |
8093 | v16qi __builtin_ia32_pcmpeqb128 (v16qi, v16qi) | |
8094 | v8hi __builtin_ia32_pcmpeqw128 (v8hi, v8hi) | |
8095 | v4si __builtin_ia32_pcmpeqd128 (v4si, v4si) | |
8096 | v16qi __builtin_ia32_pcmpgtb128 (v16qi, v16qi) | |
8097 | v8hi __builtin_ia32_pcmpgtw128 (v8hi, v8hi) | |
8098 | v4si __builtin_ia32_pcmpgtd128 (v4si, v4si) | |
8099 | v16qi __builtin_ia32_pmaxub128 (v16qi, v16qi) | |
8100 | v8hi __builtin_ia32_pmaxsw128 (v8hi, v8hi) | |
8101 | v16qi __builtin_ia32_pminub128 (v16qi, v16qi) | |
8102 | v8hi __builtin_ia32_pminsw128 (v8hi, v8hi) | |
8103 | v16qi __builtin_ia32_punpckhbw128 (v16qi, v16qi) | |
8104 | v8hi __builtin_ia32_punpckhwd128 (v8hi, v8hi) | |
8105 | v4si __builtin_ia32_punpckhdq128 (v4si, v4si) | |
8106 | v2di __builtin_ia32_punpckhqdq128 (v2di, v2di) | |
8107 | v16qi __builtin_ia32_punpcklbw128 (v16qi, v16qi) | |
8108 | v8hi __builtin_ia32_punpcklwd128 (v8hi, v8hi) | |
8109 | v4si __builtin_ia32_punpckldq128 (v4si, v4si) | |
8110 | v2di __builtin_ia32_punpcklqdq128 (v2di, v2di) | |
1b667c82 L |
8111 | v16qi __builtin_ia32_packsswb128 (v8hi, v8hi) |
8112 | v8hi __builtin_ia32_packssdw128 (v4si, v4si) | |
8113 | v16qi __builtin_ia32_packuswb128 (v8hi, v8hi) | |
d7aa4788 RG |
8114 | v8hi __builtin_ia32_pmulhuw128 (v8hi, v8hi) |
8115 | void __builtin_ia32_maskmovdqu (v16qi, v16qi) | |
8116 | v2df __builtin_ia32_loadupd (double *) | |
8117 | void __builtin_ia32_storeupd (double *, v2df) | |
bb1418c1 L |
8118 | v2df __builtin_ia32_loadhpd (v2df, double const *) |
8119 | v2df __builtin_ia32_loadlpd (v2df, double const *) | |
d7aa4788 RG |
8120 | int __builtin_ia32_movmskpd (v2df) |
8121 | int __builtin_ia32_pmovmskb128 (v16qi) | |
8122 | void __builtin_ia32_movnti (int *, int) | |
8123 | void __builtin_ia32_movntpd (double *, v2df) | |
8124 | void __builtin_ia32_movntdq (v2df *, v2df) | |
8125 | v4si __builtin_ia32_pshufd (v4si, int) | |
8126 | v8hi __builtin_ia32_pshuflw (v8hi, int) | |
8127 | v8hi __builtin_ia32_pshufhw (v8hi, int) | |
8128 | v2di __builtin_ia32_psadbw128 (v16qi, v16qi) | |
8129 | v2df __builtin_ia32_sqrtpd (v2df) | |
8130 | v2df __builtin_ia32_sqrtsd (v2df) | |
8131 | v2df __builtin_ia32_shufpd (v2df, v2df, int) | |
8132 | v2df __builtin_ia32_cvtdq2pd (v4si) | |
8133 | v4sf __builtin_ia32_cvtdq2ps (v4si) | |
8134 | v4si __builtin_ia32_cvtpd2dq (v2df) | |
8135 | v2si __builtin_ia32_cvtpd2pi (v2df) | |
8136 | v4sf __builtin_ia32_cvtpd2ps (v2df) | |
8137 | v4si __builtin_ia32_cvttpd2dq (v2df) | |
8138 | v2si __builtin_ia32_cvttpd2pi (v2df) | |
8139 | v2df __builtin_ia32_cvtpi2pd (v2si) | |
8140 | int __builtin_ia32_cvtsd2si (v2df) | |
8141 | int __builtin_ia32_cvttsd2si (v2df) | |
8142 | long long __builtin_ia32_cvtsd2si64 (v2df) | |
8143 | long long __builtin_ia32_cvttsd2si64 (v2df) | |
8144 | v4si __builtin_ia32_cvtps2dq (v4sf) | |
8145 | v2df __builtin_ia32_cvtps2pd (v4sf) | |
8146 | v4si __builtin_ia32_cvttps2dq (v4sf) | |
8147 | v2df __builtin_ia32_cvtsi2sd (v2df, int) | |
8148 | v2df __builtin_ia32_cvtsi642sd (v2df, long long) | |
8149 | v4sf __builtin_ia32_cvtsd2ss (v4sf, v2df) | |
8150 | v2df __builtin_ia32_cvtss2sd (v2df, v4sf) | |
8151 | void __builtin_ia32_clflush (const void *) | |
8152 | void __builtin_ia32_lfence (void) | |
8153 | void __builtin_ia32_mfence (void) | |
8154 | v16qi __builtin_ia32_loaddqu (const char *) | |
8155 | void __builtin_ia32_storedqu (char *, v16qi) | |
ab555a5b | 8156 | v1di __builtin_ia32_pmuludq (v2si, v2si) |
d7aa4788 | 8157 | v2di __builtin_ia32_pmuludq128 (v4si, v4si) |
52eaae97 UB |
8158 | v8hi __builtin_ia32_psllw128 (v8hi, v8hi) |
8159 | v4si __builtin_ia32_pslld128 (v4si, v4si) | |
8160 | v2di __builtin_ia32_psllq128 (v2di, v2di) | |
8161 | v8hi __builtin_ia32_psrlw128 (v8hi, v8hi) | |
8162 | v4si __builtin_ia32_psrld128 (v4si, v4si) | |
d7aa4788 | 8163 | v2di __builtin_ia32_psrlq128 (v2di, v2di) |
52eaae97 UB |
8164 | v8hi __builtin_ia32_psraw128 (v8hi, v8hi) |
8165 | v4si __builtin_ia32_psrad128 (v4si, v4si) | |
d7aa4788 RG |
8166 | v2di __builtin_ia32_pslldqi128 (v2di, int) |
8167 | v8hi __builtin_ia32_psllwi128 (v8hi, int) | |
8168 | v4si __builtin_ia32_pslldi128 (v4si, int) | |
8169 | v2di __builtin_ia32_psllqi128 (v2di, int) | |
8170 | v2di __builtin_ia32_psrldqi128 (v2di, int) | |
8171 | v8hi __builtin_ia32_psrlwi128 (v8hi, int) | |
8172 | v4si __builtin_ia32_psrldi128 (v4si, int) | |
8173 | v2di __builtin_ia32_psrlqi128 (v2di, int) | |
8174 | v8hi __builtin_ia32_psrawi128 (v8hi, int) | |
8175 | v4si __builtin_ia32_psradi128 (v4si, int) | |
8176 | v4si __builtin_ia32_pmaddwd128 (v8hi, v8hi) | |
b53b23dc | 8177 | v2di __builtin_ia32_movq128 (v2di) |
d7aa4788 RG |
8178 | @end smallexample |
8179 | ||
9e200aaf | 8180 | The following built-in functions are available when @option{-msse3} is used. |
22c7c85e L |
8181 | All of them generate the machine instruction that is part of the name. |
8182 | ||
3ab51846 | 8183 | @smallexample |
22c7c85e | 8184 | v2df __builtin_ia32_addsubpd (v2df, v2df) |
d7aa4788 | 8185 | v4sf __builtin_ia32_addsubps (v4sf, v4sf) |
22c7c85e | 8186 | v2df __builtin_ia32_haddpd (v2df, v2df) |
d7aa4788 | 8187 | v4sf __builtin_ia32_haddps (v4sf, v4sf) |
22c7c85e | 8188 | v2df __builtin_ia32_hsubpd (v2df, v2df) |
d7aa4788 | 8189 | v4sf __builtin_ia32_hsubps (v4sf, v4sf) |
22c7c85e L |
8190 | v16qi __builtin_ia32_lddqu (char const *) |
8191 | void __builtin_ia32_monitor (void *, unsigned int, unsigned int) | |
8192 | v2df __builtin_ia32_movddup (v2df) | |
8193 | v4sf __builtin_ia32_movshdup (v4sf) | |
8194 | v4sf __builtin_ia32_movsldup (v4sf) | |
8195 | void __builtin_ia32_mwait (unsigned int, unsigned int) | |
3ab51846 | 8196 | @end smallexample |
22c7c85e | 8197 | |
9e200aaf | 8198 | The following built-in functions are available when @option{-msse3} is used. |
22c7c85e L |
8199 | |
8200 | @table @code | |
8201 | @item v2df __builtin_ia32_loadddup (double const *) | |
8202 | Generates the @code{movddup} machine instruction as a load from memory. | |
8203 | @end table | |
8204 | ||
b1875f52 L |
8205 | The following built-in functions are available when @option{-mssse3} is used. |
8206 | All of them generate the machine instruction that is part of the name | |
8207 | with MMX registers. | |
8208 | ||
8209 | @smallexample | |
8210 | v2si __builtin_ia32_phaddd (v2si, v2si) | |
8211 | v4hi __builtin_ia32_phaddw (v4hi, v4hi) | |
8212 | v4hi __builtin_ia32_phaddsw (v4hi, v4hi) | |
8213 | v2si __builtin_ia32_phsubd (v2si, v2si) | |
8214 | v4hi __builtin_ia32_phsubw (v4hi, v4hi) | |
8215 | v4hi __builtin_ia32_phsubsw (v4hi, v4hi) | |
1b667c82 | 8216 | v4hi __builtin_ia32_pmaddubsw (v8qi, v8qi) |
b1875f52 L |
8217 | v4hi __builtin_ia32_pmulhrsw (v4hi, v4hi) |
8218 | v8qi __builtin_ia32_pshufb (v8qi, v8qi) | |
8219 | v8qi __builtin_ia32_psignb (v8qi, v8qi) | |
8220 | v2si __builtin_ia32_psignd (v2si, v2si) | |
8221 | v4hi __builtin_ia32_psignw (v4hi, v4hi) | |
99c25ac1 | 8222 | v1di __builtin_ia32_palignr (v1di, v1di, int) |
b1875f52 L |
8223 | v8qi __builtin_ia32_pabsb (v8qi) |
8224 | v2si __builtin_ia32_pabsd (v2si) | |
8225 | v4hi __builtin_ia32_pabsw (v4hi) | |
8226 | @end smallexample | |
8227 | ||
8228 | The following built-in functions are available when @option{-mssse3} is used. | |
8229 | All of them generate the machine instruction that is part of the name | |
8230 | with SSE registers. | |
8231 | ||
8232 | @smallexample | |
8233 | v4si __builtin_ia32_phaddd128 (v4si, v4si) | |
8234 | v8hi __builtin_ia32_phaddw128 (v8hi, v8hi) | |
8235 | v8hi __builtin_ia32_phaddsw128 (v8hi, v8hi) | |
8236 | v4si __builtin_ia32_phsubd128 (v4si, v4si) | |
8237 | v8hi __builtin_ia32_phsubw128 (v8hi, v8hi) | |
8238 | v8hi __builtin_ia32_phsubsw128 (v8hi, v8hi) | |
1b667c82 | 8239 | v8hi __builtin_ia32_pmaddubsw128 (v16qi, v16qi) |
b1875f52 L |
8240 | v8hi __builtin_ia32_pmulhrsw128 (v8hi, v8hi) |
8241 | v16qi __builtin_ia32_pshufb128 (v16qi, v16qi) | |
8242 | v16qi __builtin_ia32_psignb128 (v16qi, v16qi) | |
8243 | v4si __builtin_ia32_psignd128 (v4si, v4si) | |
8244 | v8hi __builtin_ia32_psignw128 (v8hi, v8hi) | |
858e5e79 | 8245 | v2di __builtin_ia32_palignr128 (v2di, v2di, int) |
b1875f52 L |
8246 | v16qi __builtin_ia32_pabsb128 (v16qi) |
8247 | v4si __builtin_ia32_pabsd128 (v4si) | |
8248 | v8hi __builtin_ia32_pabsw128 (v8hi) | |
8249 | @end smallexample | |
8250 | ||
9a5cee02 L |
8251 | The following built-in functions are available when @option{-msse4.1} is |
8252 | used. All of them generate the machine instruction that is part of the | |
8253 | name. | |
8254 | ||
8255 | @smallexample | |
8256 | v2df __builtin_ia32_blendpd (v2df, v2df, const int) | |
8257 | v4sf __builtin_ia32_blendps (v4sf, v4sf, const int) | |
8258 | v2df __builtin_ia32_blendvpd (v2df, v2df, v2df) | |
8259 | v4sf __builtin_ia32_blendvps (v4sf, v4sf, v4sf) | |
291d9a2d | 8260 | v2df __builtin_ia32_dppd (v2df, v2df, const int) |
9a5cee02 L |
8261 | v4sf __builtin_ia32_dpps (v4sf, v4sf, const int) |
8262 | v4sf __builtin_ia32_insertps128 (v4sf, v4sf, const int) | |
8263 | v2di __builtin_ia32_movntdqa (v2di *); | |
8264 | v16qi __builtin_ia32_mpsadbw128 (v16qi, v16qi, const int) | |
8265 | v8hi __builtin_ia32_packusdw128 (v4si, v4si) | |
8266 | v16qi __builtin_ia32_pblendvb128 (v16qi, v16qi, v16qi) | |
8267 | v8hi __builtin_ia32_pblendw128 (v8hi, v8hi, const int) | |
8268 | v2di __builtin_ia32_pcmpeqq (v2di, v2di) | |
8269 | v8hi __builtin_ia32_phminposuw128 (v8hi) | |
8270 | v16qi __builtin_ia32_pmaxsb128 (v16qi, v16qi) | |
8271 | v4si __builtin_ia32_pmaxsd128 (v4si, v4si) | |
8272 | v4si __builtin_ia32_pmaxud128 (v4si, v4si) | |
8273 | v8hi __builtin_ia32_pmaxuw128 (v8hi, v8hi) | |
8274 | v16qi __builtin_ia32_pminsb128 (v16qi, v16qi) | |
8275 | v4si __builtin_ia32_pminsd128 (v4si, v4si) | |
8276 | v4si __builtin_ia32_pminud128 (v4si, v4si) | |
8277 | v8hi __builtin_ia32_pminuw128 (v8hi, v8hi) | |
8278 | v4si __builtin_ia32_pmovsxbd128 (v16qi) | |
8279 | v2di __builtin_ia32_pmovsxbq128 (v16qi) | |
8280 | v8hi __builtin_ia32_pmovsxbw128 (v16qi) | |
8281 | v2di __builtin_ia32_pmovsxdq128 (v4si) | |
8282 | v4si __builtin_ia32_pmovsxwd128 (v8hi) | |
8283 | v2di __builtin_ia32_pmovsxwq128 (v8hi) | |
8284 | v4si __builtin_ia32_pmovzxbd128 (v16qi) | |
8285 | v2di __builtin_ia32_pmovzxbq128 (v16qi) | |
8286 | v8hi __builtin_ia32_pmovzxbw128 (v16qi) | |
8287 | v2di __builtin_ia32_pmovzxdq128 (v4si) | |
8288 | v4si __builtin_ia32_pmovzxwd128 (v8hi) | |
8289 | v2di __builtin_ia32_pmovzxwq128 (v8hi) | |
8290 | v2di __builtin_ia32_pmuldq128 (v4si, v4si) | |
8291 | v4si __builtin_ia32_pmulld128 (v4si, v4si) | |
8292 | int __builtin_ia32_ptestc128 (v2di, v2di) | |
8293 | int __builtin_ia32_ptestnzc128 (v2di, v2di) | |
8294 | int __builtin_ia32_ptestz128 (v2di, v2di) | |
8295 | v2df __builtin_ia32_roundpd (v2df, const int) | |
8296 | v4sf __builtin_ia32_roundps (v4sf, const int) | |
8297 | v2df __builtin_ia32_roundsd (v2df, v2df, const int) | |
8298 | v4sf __builtin_ia32_roundss (v4sf, v4sf, const int) | |
8299 | @end smallexample | |
8300 | ||
8301 | The following built-in functions are available when @option{-msse4.1} is | |
8302 | used. | |
8303 | ||
8304 | @table @code | |
8305 | @item v4sf __builtin_ia32_vec_set_v4sf (v4sf, float, const int) | |
8306 | Generates the @code{insertps} machine instruction. | |
8307 | @item int __builtin_ia32_vec_ext_v16qi (v16qi, const int) | |
8308 | Generates the @code{pextrb} machine instruction. | |
8309 | @item v16qi __builtin_ia32_vec_set_v16qi (v16qi, int, const int) | |
8310 | Generates the @code{pinsrb} machine instruction. | |
8311 | @item v4si __builtin_ia32_vec_set_v4si (v4si, int, const int) | |
8312 | Generates the @code{pinsrd} machine instruction. | |
8313 | @item v2di __builtin_ia32_vec_set_v2di (v2di, long long, const int) | |
8314 | Generates the @code{pinsrq} machine instruction in 64bit mode. | |
8315 | @end table | |
8316 | ||
8317 | The following built-in functions are changed to generate new SSE4.1 | |
8318 | instructions when @option{-msse4.1} is used. | |
8319 | ||
8320 | @table @code | |
8321 | @item float __builtin_ia32_vec_ext_v4sf (v4sf, const int) | |
8322 | Generates the @code{extractps} machine instruction. | |
8323 | @item int __builtin_ia32_vec_ext_v4si (v4si, const int) | |
8324 | Generates the @code{pextrd} machine instruction. | |
8325 | @item long long __builtin_ia32_vec_ext_v2di (v2di, const int) | |
8326 | Generates the @code{pextrq} machine instruction in 64bit mode. | |
8327 | @end table | |
8328 | ||
3b8dd071 L |
8329 | The following built-in functions are available when @option{-msse4.2} is |
8330 | used. All of them generate the machine instruction that is part of the | |
8331 | name. | |
8332 | ||
8333 | @smallexample | |
8334 | v16qi __builtin_ia32_pcmpestrm128 (v16qi, int, v16qi, int, const int) | |
8335 | int __builtin_ia32_pcmpestri128 (v16qi, int, v16qi, int, const int) | |
8336 | int __builtin_ia32_pcmpestria128 (v16qi, int, v16qi, int, const int) | |
8337 | int __builtin_ia32_pcmpestric128 (v16qi, int, v16qi, int, const int) | |
8338 | int __builtin_ia32_pcmpestrio128 (v16qi, int, v16qi, int, const int) | |
8339 | int __builtin_ia32_pcmpestris128 (v16qi, int, v16qi, int, const int) | |
8340 | int __builtin_ia32_pcmpestriz128 (v16qi, int, v16qi, int, const int) | |
8341 | v16qi __builtin_ia32_pcmpistrm128 (v16qi, v16qi, const int) | |
8342 | int __builtin_ia32_pcmpistri128 (v16qi, v16qi, const int) | |
8343 | int __builtin_ia32_pcmpistria128 (v16qi, v16qi, const int) | |
8344 | int __builtin_ia32_pcmpistric128 (v16qi, v16qi, const int) | |
8345 | int __builtin_ia32_pcmpistrio128 (v16qi, v16qi, const int) | |
8346 | int __builtin_ia32_pcmpistris128 (v16qi, v16qi, const int) | |
8347 | int __builtin_ia32_pcmpistriz128 (v16qi, v16qi, const int) | |
291d9a2d | 8348 | v2di __builtin_ia32_pcmpgtq (v2di, v2di) |
3b8dd071 L |
8349 | @end smallexample |
8350 | ||
8351 | The following built-in functions are available when @option{-msse4.2} is | |
8352 | used. | |
8353 | ||
8354 | @table @code | |
291d9a2d | 8355 | @item unsigned int __builtin_ia32_crc32qi (unsigned int, unsigned char) |
3b8dd071 | 8356 | Generates the @code{crc32b} machine instruction. |
291d9a2d | 8357 | @item unsigned int __builtin_ia32_crc32hi (unsigned int, unsigned short) |
3b8dd071 | 8358 | Generates the @code{crc32w} machine instruction. |
291d9a2d | 8359 | @item unsigned int __builtin_ia32_crc32si (unsigned int, unsigned int) |
3b8dd071 | 8360 | Generates the @code{crc32l} machine instruction. |
a44acfb9 | 8361 | @item unsigned long long __builtin_ia32_crc32di (unsigned long long, unsigned long long) |
3b8dd071 L |
8362 | @end table |
8363 | ||
8364 | The following built-in functions are changed to generate new SSE4.2 | |
8365 | instructions when @option{-msse4.2} is used. | |
8366 | ||
8367 | @table @code | |
291d9a2d | 8368 | @item int __builtin_popcount (unsigned int) |
3b8dd071 | 8369 | Generates the @code{popcntl} machine instruction. |
291d9a2d | 8370 | @item int __builtin_popcountl (unsigned long) |
3b8dd071 L |
8371 | Generates the @code{popcntl} or @code{popcntq} machine instruction, |
8372 | depending on the size of @code{unsigned long}. | |
291d9a2d | 8373 | @item int __builtin_popcountll (unsigned long long) |
3b8dd071 L |
8374 | Generates the @code{popcntq} machine instruction. |
8375 | @end table | |
8376 | ||
31cb596a JY |
8377 | The following built-in functions are available when @option{-mavx} is |
8378 | used. All of them generate the machine instruction that is part of the | |
8379 | name. | |
8380 | ||
8381 | @smallexample | |
8382 | v4df __builtin_ia32_addpd256 (v4df,v4df) | |
8383 | v8sf __builtin_ia32_addps256 (v8sf,v8sf) | |
8384 | v4df __builtin_ia32_addsubpd256 (v4df,v4df) | |
8385 | v8sf __builtin_ia32_addsubps256 (v8sf,v8sf) | |
8386 | v4df __builtin_ia32_andnpd256 (v4df,v4df) | |
8387 | v8sf __builtin_ia32_andnps256 (v8sf,v8sf) | |
8388 | v4df __builtin_ia32_andpd256 (v4df,v4df) | |
8389 | v8sf __builtin_ia32_andps256 (v8sf,v8sf) | |
8390 | v4df __builtin_ia32_blendpd256 (v4df,v4df,int) | |
8391 | v8sf __builtin_ia32_blendps256 (v8sf,v8sf,int) | |
8392 | v4df __builtin_ia32_blendvpd256 (v4df,v4df,v4df) | |
8393 | v8sf __builtin_ia32_blendvps256 (v8sf,v8sf,v8sf) | |
8394 | v2df __builtin_ia32_cmppd (v2df,v2df,int) | |
8395 | v4df __builtin_ia32_cmppd256 (v4df,v4df,int) | |
8396 | v4sf __builtin_ia32_cmpps (v4sf,v4sf,int) | |
8397 | v8sf __builtin_ia32_cmpps256 (v8sf,v8sf,int) | |
8398 | v2df __builtin_ia32_cmpsd (v2df,v2df,int) | |
8399 | v4sf __builtin_ia32_cmpss (v4sf,v4sf,int) | |
8400 | v4df __builtin_ia32_cvtdq2pd256 (v4si) | |
8401 | v8sf __builtin_ia32_cvtdq2ps256 (v8si) | |
8402 | v4si __builtin_ia32_cvtpd2dq256 (v4df) | |
8403 | v4sf __builtin_ia32_cvtpd2ps256 (v4df) | |
8404 | v8si __builtin_ia32_cvtps2dq256 (v8sf) | |
8405 | v4df __builtin_ia32_cvtps2pd256 (v4sf) | |
8406 | v4si __builtin_ia32_cvttpd2dq256 (v4df) | |
8407 | v8si __builtin_ia32_cvttps2dq256 (v8sf) | |
8408 | v4df __builtin_ia32_divpd256 (v4df,v4df) | |
8409 | v8sf __builtin_ia32_divps256 (v8sf,v8sf) | |
8410 | v8sf __builtin_ia32_dpps256 (v8sf,v8sf,int) | |
8411 | v4df __builtin_ia32_haddpd256 (v4df,v4df) | |
8412 | v8sf __builtin_ia32_haddps256 (v8sf,v8sf) | |
8413 | v4df __builtin_ia32_hsubpd256 (v4df,v4df) | |
8414 | v8sf __builtin_ia32_hsubps256 (v8sf,v8sf) | |
8415 | v32qi __builtin_ia32_lddqu256 (pcchar) | |
8416 | v32qi __builtin_ia32_loaddqu256 (pcchar) | |
8417 | v4df __builtin_ia32_loadupd256 (pcdouble) | |
8418 | v8sf __builtin_ia32_loadups256 (pcfloat) | |
8419 | v2df __builtin_ia32_maskloadpd (pcv2df,v2df) | |
8420 | v4df __builtin_ia32_maskloadpd256 (pcv4df,v4df) | |
8421 | v4sf __builtin_ia32_maskloadps (pcv4sf,v4sf) | |
8422 | v8sf __builtin_ia32_maskloadps256 (pcv8sf,v8sf) | |
8423 | void __builtin_ia32_maskstorepd (pv2df,v2df,v2df) | |
8424 | void __builtin_ia32_maskstorepd256 (pv4df,v4df,v4df) | |
8425 | void __builtin_ia32_maskstoreps (pv4sf,v4sf,v4sf) | |
8426 | void __builtin_ia32_maskstoreps256 (pv8sf,v8sf,v8sf) | |
8427 | v4df __builtin_ia32_maxpd256 (v4df,v4df) | |
8428 | v8sf __builtin_ia32_maxps256 (v8sf,v8sf) | |
8429 | v4df __builtin_ia32_minpd256 (v4df,v4df) | |
8430 | v8sf __builtin_ia32_minps256 (v8sf,v8sf) | |
8431 | v4df __builtin_ia32_movddup256 (v4df) | |
8432 | int __builtin_ia32_movmskpd256 (v4df) | |
8433 | int __builtin_ia32_movmskps256 (v8sf) | |
8434 | v8sf __builtin_ia32_movshdup256 (v8sf) | |
8435 | v8sf __builtin_ia32_movsldup256 (v8sf) | |
8436 | v4df __builtin_ia32_mulpd256 (v4df,v4df) | |
8437 | v8sf __builtin_ia32_mulps256 (v8sf,v8sf) | |
8438 | v4df __builtin_ia32_orpd256 (v4df,v4df) | |
8439 | v8sf __builtin_ia32_orps256 (v8sf,v8sf) | |
8440 | v2df __builtin_ia32_pd_pd256 (v4df) | |
8441 | v4df __builtin_ia32_pd256_pd (v2df) | |
8442 | v4sf __builtin_ia32_ps_ps256 (v8sf) | |
8443 | v8sf __builtin_ia32_ps256_ps (v4sf) | |
8444 | int __builtin_ia32_ptestc256 (v4di,v4di,ptest) | |
8445 | int __builtin_ia32_ptestnzc256 (v4di,v4di,ptest) | |
8446 | int __builtin_ia32_ptestz256 (v4di,v4di,ptest) | |
8447 | v8sf __builtin_ia32_rcpps256 (v8sf) | |
8448 | v4df __builtin_ia32_roundpd256 (v4df,int) | |
8449 | v8sf __builtin_ia32_roundps256 (v8sf,int) | |
8450 | v8sf __builtin_ia32_rsqrtps_nr256 (v8sf) | |
8451 | v8sf __builtin_ia32_rsqrtps256 (v8sf) | |
8452 | v4df __builtin_ia32_shufpd256 (v4df,v4df,int) | |
8453 | v8sf __builtin_ia32_shufps256 (v8sf,v8sf,int) | |
8454 | v4si __builtin_ia32_si_si256 (v8si) | |
8455 | v8si __builtin_ia32_si256_si (v4si) | |
8456 | v4df __builtin_ia32_sqrtpd256 (v4df) | |
8457 | v8sf __builtin_ia32_sqrtps_nr256 (v8sf) | |
8458 | v8sf __builtin_ia32_sqrtps256 (v8sf) | |
8459 | void __builtin_ia32_storedqu256 (pchar,v32qi) | |
8460 | void __builtin_ia32_storeupd256 (pdouble,v4df) | |
8461 | void __builtin_ia32_storeups256 (pfloat,v8sf) | |
8462 | v4df __builtin_ia32_subpd256 (v4df,v4df) | |
8463 | v8sf __builtin_ia32_subps256 (v8sf,v8sf) | |
8464 | v4df __builtin_ia32_unpckhpd256 (v4df,v4df) | |
8465 | v8sf __builtin_ia32_unpckhps256 (v8sf,v8sf) | |
8466 | v4df __builtin_ia32_unpcklpd256 (v4df,v4df) | |
8467 | v8sf __builtin_ia32_unpcklps256 (v8sf,v8sf) | |
8468 | v4df __builtin_ia32_vbroadcastf128_pd256 (pcv2df) | |
8469 | v8sf __builtin_ia32_vbroadcastf128_ps256 (pcv4sf) | |
8470 | v4df __builtin_ia32_vbroadcastsd256 (pcdouble) | |
8471 | v4sf __builtin_ia32_vbroadcastss (pcfloat) | |
8472 | v8sf __builtin_ia32_vbroadcastss256 (pcfloat) | |
8473 | v2df __builtin_ia32_vextractf128_pd256 (v4df,int) | |
8474 | v4sf __builtin_ia32_vextractf128_ps256 (v8sf,int) | |
8475 | v4si __builtin_ia32_vextractf128_si256 (v8si,int) | |
8476 | v4df __builtin_ia32_vinsertf128_pd256 (v4df,v2df,int) | |
8477 | v8sf __builtin_ia32_vinsertf128_ps256 (v8sf,v4sf,int) | |
8478 | v8si __builtin_ia32_vinsertf128_si256 (v8si,v4si,int) | |
8479 | v4df __builtin_ia32_vperm2f128_pd256 (v4df,v4df,int) | |
8480 | v8sf __builtin_ia32_vperm2f128_ps256 (v8sf,v8sf,int) | |
8481 | v8si __builtin_ia32_vperm2f128_si256 (v8si,v8si,int) | |
8482 | v2df __builtin_ia32_vpermil2pd (v2df,v2df,v2di,int) | |
8483 | v4df __builtin_ia32_vpermil2pd256 (v4df,v4df,v4di,int) | |
8484 | v4sf __builtin_ia32_vpermil2ps (v4sf,v4sf,v4si,int) | |
8485 | v8sf __builtin_ia32_vpermil2ps256 (v8sf,v8sf,v8si,int) | |
8486 | v2df __builtin_ia32_vpermilpd (v2df,int) | |
8487 | v4df __builtin_ia32_vpermilpd256 (v4df,int) | |
8488 | v4sf __builtin_ia32_vpermilps (v4sf,int) | |
8489 | v8sf __builtin_ia32_vpermilps256 (v8sf,int) | |
8490 | v2df __builtin_ia32_vpermilvarpd (v2df,v2di) | |
8491 | v4df __builtin_ia32_vpermilvarpd256 (v4df,v4di) | |
8492 | v4sf __builtin_ia32_vpermilvarps (v4sf,v4si) | |
8493 | v8sf __builtin_ia32_vpermilvarps256 (v8sf,v8si) | |
8494 | int __builtin_ia32_vtestcpd (v2df,v2df,ptest) | |
8495 | int __builtin_ia32_vtestcpd256 (v4df,v4df,ptest) | |
8496 | int __builtin_ia32_vtestcps (v4sf,v4sf,ptest) | |
8497 | int __builtin_ia32_vtestcps256 (v8sf,v8sf,ptest) | |
8498 | int __builtin_ia32_vtestnzcpd (v2df,v2df,ptest) | |
8499 | int __builtin_ia32_vtestnzcpd256 (v4df,v4df,ptest) | |
8500 | int __builtin_ia32_vtestnzcps (v4sf,v4sf,ptest) | |
8501 | int __builtin_ia32_vtestnzcps256 (v8sf,v8sf,ptest) | |
8502 | int __builtin_ia32_vtestzpd (v2df,v2df,ptest) | |
8503 | int __builtin_ia32_vtestzpd256 (v4df,v4df,ptest) | |
8504 | int __builtin_ia32_vtestzps (v4sf,v4sf,ptest) | |
8505 | int __builtin_ia32_vtestzps256 (v8sf,v8sf,ptest) | |
8506 | void __builtin_ia32_vzeroall (void) | |
8507 | void __builtin_ia32_vzeroupper (void) | |
8508 | v4df __builtin_ia32_xorpd256 (v4df,v4df) | |
8509 | v8sf __builtin_ia32_xorps256 (v8sf,v8sf) | |
8510 | @end smallexample | |
8511 | ||
8b96a312 L |
8512 | The following built-in functions are available when @option{-maes} is |
8513 | used. All of them generate the machine instruction that is part of the | |
8514 | name. | |
8515 | ||
8516 | @smallexample | |
8517 | v2di __builtin_ia32_aesenc128 (v2di, v2di) | |
8518 | v2di __builtin_ia32_aesenclast128 (v2di, v2di) | |
8519 | v2di __builtin_ia32_aesdec128 (v2di, v2di) | |
8520 | v2di __builtin_ia32_aesdeclast128 (v2di, v2di) | |
8521 | v2di __builtin_ia32_aeskeygenassist128 (v2di, const int) | |
8522 | v2di __builtin_ia32_aesimc128 (v2di) | |
8523 | @end smallexample | |
8524 | ||
8525 | The following built-in function is available when @option{-mpclmul} is | |
8526 | used. | |
8527 | ||
8528 | @table @code | |
8529 | @item v2di __builtin_ia32_pclmulqdq128 (v2di, v2di, const int) | |
8530 | Generates the @code{pclmulqdq} machine instruction. | |
8531 | @end table | |
8532 | ||
21efb4d4 | 8533 | The following built-in functions are available when @option{-msse4a} is used. |
291d9a2d | 8534 | All of them generate the machine instruction that is part of the name. |
21efb4d4 HJ |
8535 | |
8536 | @smallexample | |
291d9a2d UB |
8537 | void __builtin_ia32_movntsd (double *, v2df) |
8538 | void __builtin_ia32_movntss (float *, v4sf) | |
8539 | v2di __builtin_ia32_extrq (v2di, v16qi) | |
8540 | v2di __builtin_ia32_extrqi (v2di, const unsigned int, const unsigned int) | |
8541 | v2di __builtin_ia32_insertq (v2di, v2di) | |
8542 | v2di __builtin_ia32_insertqi (v2di, v2di, const unsigned int, const unsigned int) | |
21efb4d4 HJ |
8543 | @end smallexample |
8544 | ||
04e1d06b MM |
8545 | The following built-in functions are available when @option{-msse5} is used. |
8546 | All of them generate the machine instruction that is part of the name | |
8547 | with MMX registers. | |
8548 | ||
8549 | @smallexample | |
8550 | v2df __builtin_ia32_comeqpd (v2df, v2df) | |
8551 | v2df __builtin_ia32_comeqps (v2df, v2df) | |
8552 | v4sf __builtin_ia32_comeqsd (v4sf, v4sf) | |
8553 | v4sf __builtin_ia32_comeqss (v4sf, v4sf) | |
8554 | v2df __builtin_ia32_comfalsepd (v2df, v2df) | |
8555 | v2df __builtin_ia32_comfalseps (v2df, v2df) | |
8556 | v4sf __builtin_ia32_comfalsesd (v4sf, v4sf) | |
8557 | v4sf __builtin_ia32_comfalsess (v4sf, v4sf) | |
8558 | v2df __builtin_ia32_comgepd (v2df, v2df) | |
8559 | v2df __builtin_ia32_comgeps (v2df, v2df) | |
8560 | v4sf __builtin_ia32_comgesd (v4sf, v4sf) | |
8561 | v4sf __builtin_ia32_comgess (v4sf, v4sf) | |
8562 | v2df __builtin_ia32_comgtpd (v2df, v2df) | |
8563 | v2df __builtin_ia32_comgtps (v2df, v2df) | |
8564 | v4sf __builtin_ia32_comgtsd (v4sf, v4sf) | |
8565 | v4sf __builtin_ia32_comgtss (v4sf, v4sf) | |
8566 | v2df __builtin_ia32_comlepd (v2df, v2df) | |
8567 | v2df __builtin_ia32_comleps (v2df, v2df) | |
8568 | v4sf __builtin_ia32_comlesd (v4sf, v4sf) | |
8569 | v4sf __builtin_ia32_comless (v4sf, v4sf) | |
8570 | v2df __builtin_ia32_comltpd (v2df, v2df) | |
8571 | v2df __builtin_ia32_comltps (v2df, v2df) | |
8572 | v4sf __builtin_ia32_comltsd (v4sf, v4sf) | |
8573 | v4sf __builtin_ia32_comltss (v4sf, v4sf) | |
8574 | v2df __builtin_ia32_comnepd (v2df, v2df) | |
8575 | v2df __builtin_ia32_comneps (v2df, v2df) | |
8576 | v4sf __builtin_ia32_comnesd (v4sf, v4sf) | |
8577 | v4sf __builtin_ia32_comness (v4sf, v4sf) | |
8578 | v2df __builtin_ia32_comordpd (v2df, v2df) | |
8579 | v2df __builtin_ia32_comordps (v2df, v2df) | |
8580 | v4sf __builtin_ia32_comordsd (v4sf, v4sf) | |
8581 | v4sf __builtin_ia32_comordss (v4sf, v4sf) | |
8582 | v2df __builtin_ia32_comtruepd (v2df, v2df) | |
8583 | v2df __builtin_ia32_comtrueps (v2df, v2df) | |
8584 | v4sf __builtin_ia32_comtruesd (v4sf, v4sf) | |
8585 | v4sf __builtin_ia32_comtruess (v4sf, v4sf) | |
8586 | v2df __builtin_ia32_comueqpd (v2df, v2df) | |
8587 | v2df __builtin_ia32_comueqps (v2df, v2df) | |
8588 | v4sf __builtin_ia32_comueqsd (v4sf, v4sf) | |
8589 | v4sf __builtin_ia32_comueqss (v4sf, v4sf) | |
8590 | v2df __builtin_ia32_comugepd (v2df, v2df) | |
8591 | v2df __builtin_ia32_comugeps (v2df, v2df) | |
8592 | v4sf __builtin_ia32_comugesd (v4sf, v4sf) | |
8593 | v4sf __builtin_ia32_comugess (v4sf, v4sf) | |
8594 | v2df __builtin_ia32_comugtpd (v2df, v2df) | |
8595 | v2df __builtin_ia32_comugtps (v2df, v2df) | |
8596 | v4sf __builtin_ia32_comugtsd (v4sf, v4sf) | |
8597 | v4sf __builtin_ia32_comugtss (v4sf, v4sf) | |
8598 | v2df __builtin_ia32_comulepd (v2df, v2df) | |
8599 | v2df __builtin_ia32_comuleps (v2df, v2df) | |
8600 | v4sf __builtin_ia32_comulesd (v4sf, v4sf) | |
8601 | v4sf __builtin_ia32_comuless (v4sf, v4sf) | |
8602 | v2df __builtin_ia32_comultpd (v2df, v2df) | |
8603 | v2df __builtin_ia32_comultps (v2df, v2df) | |
8604 | v4sf __builtin_ia32_comultsd (v4sf, v4sf) | |
8605 | v4sf __builtin_ia32_comultss (v4sf, v4sf) | |
8606 | v2df __builtin_ia32_comunepd (v2df, v2df) | |
8607 | v2df __builtin_ia32_comuneps (v2df, v2df) | |
8608 | v4sf __builtin_ia32_comunesd (v4sf, v4sf) | |
8609 | v4sf __builtin_ia32_comuness (v4sf, v4sf) | |
8610 | v2df __builtin_ia32_comunordpd (v2df, v2df) | |
8611 | v2df __builtin_ia32_comunordps (v2df, v2df) | |
8612 | v4sf __builtin_ia32_comunordsd (v4sf, v4sf) | |
8613 | v4sf __builtin_ia32_comunordss (v4sf, v4sf) | |
8614 | v2df __builtin_ia32_fmaddpd (v2df, v2df, v2df) | |
8615 | v4sf __builtin_ia32_fmaddps (v4sf, v4sf, v4sf) | |
8616 | v2df __builtin_ia32_fmaddsd (v2df, v2df, v2df) | |
8617 | v4sf __builtin_ia32_fmaddss (v4sf, v4sf, v4sf) | |
8618 | v2df __builtin_ia32_fmsubpd (v2df, v2df, v2df) | |
8619 | v4sf __builtin_ia32_fmsubps (v4sf, v4sf, v4sf) | |
8620 | v2df __builtin_ia32_fmsubsd (v2df, v2df, v2df) | |
8621 | v4sf __builtin_ia32_fmsubss (v4sf, v4sf, v4sf) | |
8622 | v2df __builtin_ia32_fnmaddpd (v2df, v2df, v2df) | |
8623 | v4sf __builtin_ia32_fnmaddps (v4sf, v4sf, v4sf) | |
8624 | v2df __builtin_ia32_fnmaddsd (v2df, v2df, v2df) | |
8625 | v4sf __builtin_ia32_fnmaddss (v4sf, v4sf, v4sf) | |
8626 | v2df __builtin_ia32_fnmsubpd (v2df, v2df, v2df) | |
8627 | v4sf __builtin_ia32_fnmsubps (v4sf, v4sf, v4sf) | |
8628 | v2df __builtin_ia32_fnmsubsd (v2df, v2df, v2df) | |
8629 | v4sf __builtin_ia32_fnmsubss (v4sf, v4sf, v4sf) | |
8630 | v2df __builtin_ia32_frczpd (v2df) | |
8631 | v4sf __builtin_ia32_frczps (v4sf) | |
8632 | v2df __builtin_ia32_frczsd (v2df, v2df) | |
8633 | v4sf __builtin_ia32_frczss (v4sf, v4sf) | |
8634 | v2di __builtin_ia32_pcmov (v2di, v2di, v2di) | |
8635 | v2di __builtin_ia32_pcmov_v2di (v2di, v2di, v2di) | |
8636 | v4si __builtin_ia32_pcmov_v4si (v4si, v4si, v4si) | |
8637 | v8hi __builtin_ia32_pcmov_v8hi (v8hi, v8hi, v8hi) | |
8638 | v16qi __builtin_ia32_pcmov_v16qi (v16qi, v16qi, v16qi) | |
8639 | v2df __builtin_ia32_pcmov_v2df (v2df, v2df, v2df) | |
8640 | v4sf __builtin_ia32_pcmov_v4sf (v4sf, v4sf, v4sf) | |
8641 | v16qi __builtin_ia32_pcomeqb (v16qi, v16qi) | |
8642 | v8hi __builtin_ia32_pcomeqw (v8hi, v8hi) | |
8643 | v4si __builtin_ia32_pcomeqd (v4si, v4si) | |
8644 | v2di __builtin_ia32_pcomeqq (v2di, v2di) | |
8645 | v16qi __builtin_ia32_pcomequb (v16qi, v16qi) | |
8646 | v4si __builtin_ia32_pcomequd (v4si, v4si) | |
8647 | v2di __builtin_ia32_pcomequq (v2di, v2di) | |
8648 | v8hi __builtin_ia32_pcomequw (v8hi, v8hi) | |
8649 | v8hi __builtin_ia32_pcomeqw (v8hi, v8hi) | |
8650 | v16qi __builtin_ia32_pcomfalseb (v16qi, v16qi) | |
8651 | v4si __builtin_ia32_pcomfalsed (v4si, v4si) | |
8652 | v2di __builtin_ia32_pcomfalseq (v2di, v2di) | |
8653 | v16qi __builtin_ia32_pcomfalseub (v16qi, v16qi) | |
8654 | v4si __builtin_ia32_pcomfalseud (v4si, v4si) | |
8655 | v2di __builtin_ia32_pcomfalseuq (v2di, v2di) | |
8656 | v8hi __builtin_ia32_pcomfalseuw (v8hi, v8hi) | |
8657 | v8hi __builtin_ia32_pcomfalsew (v8hi, v8hi) | |
8658 | v16qi __builtin_ia32_pcomgeb (v16qi, v16qi) | |
8659 | v4si __builtin_ia32_pcomged (v4si, v4si) | |
8660 | v2di __builtin_ia32_pcomgeq (v2di, v2di) | |
8661 | v16qi __builtin_ia32_pcomgeub (v16qi, v16qi) | |
8662 | v4si __builtin_ia32_pcomgeud (v4si, v4si) | |
8663 | v2di __builtin_ia32_pcomgeuq (v2di, v2di) | |
8664 | v8hi __builtin_ia32_pcomgeuw (v8hi, v8hi) | |
8665 | v8hi __builtin_ia32_pcomgew (v8hi, v8hi) | |
8666 | v16qi __builtin_ia32_pcomgtb (v16qi, v16qi) | |
8667 | v4si __builtin_ia32_pcomgtd (v4si, v4si) | |
8668 | v2di __builtin_ia32_pcomgtq (v2di, v2di) | |
8669 | v16qi __builtin_ia32_pcomgtub (v16qi, v16qi) | |
8670 | v4si __builtin_ia32_pcomgtud (v4si, v4si) | |
8671 | v2di __builtin_ia32_pcomgtuq (v2di, v2di) | |
8672 | v8hi __builtin_ia32_pcomgtuw (v8hi, v8hi) | |
8673 | v8hi __builtin_ia32_pcomgtw (v8hi, v8hi) | |
8674 | v16qi __builtin_ia32_pcomleb (v16qi, v16qi) | |
8675 | v4si __builtin_ia32_pcomled (v4si, v4si) | |
8676 | v2di __builtin_ia32_pcomleq (v2di, v2di) | |
8677 | v16qi __builtin_ia32_pcomleub (v16qi, v16qi) | |
8678 | v4si __builtin_ia32_pcomleud (v4si, v4si) | |
8679 | v2di __builtin_ia32_pcomleuq (v2di, v2di) | |
8680 | v8hi __builtin_ia32_pcomleuw (v8hi, v8hi) | |
8681 | v8hi __builtin_ia32_pcomlew (v8hi, v8hi) | |
8682 | v16qi __builtin_ia32_pcomltb (v16qi, v16qi) | |
8683 | v4si __builtin_ia32_pcomltd (v4si, v4si) | |
8684 | v2di __builtin_ia32_pcomltq (v2di, v2di) | |
8685 | v16qi __builtin_ia32_pcomltub (v16qi, v16qi) | |
8686 | v4si __builtin_ia32_pcomltud (v4si, v4si) | |
8687 | v2di __builtin_ia32_pcomltuq (v2di, v2di) | |
8688 | v8hi __builtin_ia32_pcomltuw (v8hi, v8hi) | |
8689 | v8hi __builtin_ia32_pcomltw (v8hi, v8hi) | |
8690 | v16qi __builtin_ia32_pcomneb (v16qi, v16qi) | |
8691 | v4si __builtin_ia32_pcomned (v4si, v4si) | |
8692 | v2di __builtin_ia32_pcomneq (v2di, v2di) | |
8693 | v16qi __builtin_ia32_pcomneub (v16qi, v16qi) | |
8694 | v4si __builtin_ia32_pcomneud (v4si, v4si) | |
8695 | v2di __builtin_ia32_pcomneuq (v2di, v2di) | |
8696 | v8hi __builtin_ia32_pcomneuw (v8hi, v8hi) | |
8697 | v8hi __builtin_ia32_pcomnew (v8hi, v8hi) | |
8698 | v16qi __builtin_ia32_pcomtrueb (v16qi, v16qi) | |
8699 | v4si __builtin_ia32_pcomtrued (v4si, v4si) | |
8700 | v2di __builtin_ia32_pcomtrueq (v2di, v2di) | |
8701 | v16qi __builtin_ia32_pcomtrueub (v16qi, v16qi) | |
8702 | v4si __builtin_ia32_pcomtrueud (v4si, v4si) | |
8703 | v2di __builtin_ia32_pcomtrueuq (v2di, v2di) | |
8704 | v8hi __builtin_ia32_pcomtrueuw (v8hi, v8hi) | |
8705 | v8hi __builtin_ia32_pcomtruew (v8hi, v8hi) | |
8706 | v4df __builtin_ia32_permpd (v2df, v2df, v16qi) | |
8707 | v4sf __builtin_ia32_permps (v4sf, v4sf, v16qi) | |
8708 | v4si __builtin_ia32_phaddbd (v16qi) | |
8709 | v2di __builtin_ia32_phaddbq (v16qi) | |
8710 | v8hi __builtin_ia32_phaddbw (v16qi) | |
8711 | v2di __builtin_ia32_phadddq (v4si) | |
8712 | v4si __builtin_ia32_phaddubd (v16qi) | |
8713 | v2di __builtin_ia32_phaddubq (v16qi) | |
8714 | v8hi __builtin_ia32_phaddubw (v16qi) | |
8715 | v2di __builtin_ia32_phaddudq (v4si) | |
8716 | v4si __builtin_ia32_phadduwd (v8hi) | |
8717 | v2di __builtin_ia32_phadduwq (v8hi) | |
8718 | v4si __builtin_ia32_phaddwd (v8hi) | |
8719 | v2di __builtin_ia32_phaddwq (v8hi) | |
8720 | v8hi __builtin_ia32_phsubbw (v16qi) | |
8721 | v2di __builtin_ia32_phsubdq (v4si) | |
8722 | v4si __builtin_ia32_phsubwd (v8hi) | |
8723 | v4si __builtin_ia32_pmacsdd (v4si, v4si, v4si) | |
8724 | v2di __builtin_ia32_pmacsdqh (v4si, v4si, v2di) | |
8725 | v2di __builtin_ia32_pmacsdql (v4si, v4si, v2di) | |
8726 | v4si __builtin_ia32_pmacssdd (v4si, v4si, v4si) | |
8727 | v2di __builtin_ia32_pmacssdqh (v4si, v4si, v2di) | |
8728 | v2di __builtin_ia32_pmacssdql (v4si, v4si, v2di) | |
8729 | v4si __builtin_ia32_pmacsswd (v8hi, v8hi, v4si) | |
8730 | v8hi __builtin_ia32_pmacssww (v8hi, v8hi, v8hi) | |
8731 | v4si __builtin_ia32_pmacswd (v8hi, v8hi, v4si) | |
8732 | v8hi __builtin_ia32_pmacsww (v8hi, v8hi, v8hi) | |
8733 | v4si __builtin_ia32_pmadcsswd (v8hi, v8hi, v4si) | |
8734 | v4si __builtin_ia32_pmadcswd (v8hi, v8hi, v4si) | |
8735 | v16qi __builtin_ia32_pperm (v16qi, v16qi, v16qi) | |
8736 | v16qi __builtin_ia32_protb (v16qi, v16qi) | |
8737 | v4si __builtin_ia32_protd (v4si, v4si) | |
8738 | v2di __builtin_ia32_protq (v2di, v2di) | |
8739 | v8hi __builtin_ia32_protw (v8hi, v8hi) | |
8740 | v16qi __builtin_ia32_pshab (v16qi, v16qi) | |
8741 | v4si __builtin_ia32_pshad (v4si, v4si) | |
8742 | v2di __builtin_ia32_pshaq (v2di, v2di) | |
8743 | v8hi __builtin_ia32_pshaw (v8hi, v8hi) | |
8744 | v16qi __builtin_ia32_pshlb (v16qi, v16qi) | |
8745 | v4si __builtin_ia32_pshld (v4si, v4si) | |
8746 | v2di __builtin_ia32_pshlq (v2di, v2di) | |
8747 | v8hi __builtin_ia32_pshlw (v8hi, v8hi) | |
8748 | @end smallexample | |
8749 | ||
84fbffb2 | 8750 | The following builtin-in functions are available when @option{-msse5} |
04e1d06b MM |
8751 | is used. The second argument must be an integer constant and generate |
8752 | the machine instruction that is part of the name with the @samp{_imm} | |
8753 | suffix removed. | |
8754 | ||
8755 | @smallexample | |
8756 | v16qi __builtin_ia32_protb_imm (v16qi, int) | |
8757 | v4si __builtin_ia32_protd_imm (v4si, int) | |
8758 | v2di __builtin_ia32_protq_imm (v2di, int) | |
8759 | v8hi __builtin_ia32_protw_imm (v8hi, int) | |
8760 | @end smallexample | |
8761 | ||
0975678f JM |
8762 | The following built-in functions are available when @option{-m3dnow} is used. |
8763 | All of them generate the machine instruction that is part of the name. | |
8764 | ||
3ab51846 | 8765 | @smallexample |
0975678f JM |
8766 | void __builtin_ia32_femms (void) |
8767 | v8qi __builtin_ia32_pavgusb (v8qi, v8qi) | |
8768 | v2si __builtin_ia32_pf2id (v2sf) | |
8769 | v2sf __builtin_ia32_pfacc (v2sf, v2sf) | |
8770 | v2sf __builtin_ia32_pfadd (v2sf, v2sf) | |
8771 | v2si __builtin_ia32_pfcmpeq (v2sf, v2sf) | |
8772 | v2si __builtin_ia32_pfcmpge (v2sf, v2sf) | |
8773 | v2si __builtin_ia32_pfcmpgt (v2sf, v2sf) | |
8774 | v2sf __builtin_ia32_pfmax (v2sf, v2sf) | |
8775 | v2sf __builtin_ia32_pfmin (v2sf, v2sf) | |
8776 | v2sf __builtin_ia32_pfmul (v2sf, v2sf) | |
8777 | v2sf __builtin_ia32_pfrcp (v2sf) | |
8778 | v2sf __builtin_ia32_pfrcpit1 (v2sf, v2sf) | |
8779 | v2sf __builtin_ia32_pfrcpit2 (v2sf, v2sf) | |
8780 | v2sf __builtin_ia32_pfrsqrt (v2sf) | |
8781 | v2sf __builtin_ia32_pfrsqrtit1 (v2sf, v2sf) | |
8782 | v2sf __builtin_ia32_pfsub (v2sf, v2sf) | |
8783 | v2sf __builtin_ia32_pfsubr (v2sf, v2sf) | |
8784 | v2sf __builtin_ia32_pi2fd (v2si) | |
8785 | v4hi __builtin_ia32_pmulhrw (v4hi, v4hi) | |
3ab51846 | 8786 | @end smallexample |
0975678f JM |
8787 | |
8788 | The following built-in functions are available when both @option{-m3dnow} | |
8789 | and @option{-march=athlon} are used. All of them generate the machine | |
8790 | instruction that is part of the name. | |
8791 | ||
3ab51846 | 8792 | @smallexample |
0975678f JM |
8793 | v2si __builtin_ia32_pf2iw (v2sf) |
8794 | v2sf __builtin_ia32_pfnacc (v2sf, v2sf) | |
8795 | v2sf __builtin_ia32_pfpnacc (v2sf, v2sf) | |
8796 | v2sf __builtin_ia32_pi2fw (v2si) | |
8797 | v2sf __builtin_ia32_pswapdsf (v2sf) | |
8798 | v2si __builtin_ia32_pswapdsi (v2si) | |
3ab51846 | 8799 | @end smallexample |
0975678f | 8800 | |
118ea793 CF |
8801 | @node MIPS DSP Built-in Functions |
8802 | @subsection MIPS DSP Built-in Functions | |
8803 | ||
8804 | The MIPS DSP Application-Specific Extension (ASE) includes new | |
8805 | instructions that are designed to improve the performance of DSP and | |
8806 | media applications. It provides instructions that operate on packed | |
32041385 | 8807 | 8-bit/16-bit integer data, Q7, Q15 and Q31 fractional data. |
118ea793 CF |
8808 | |
8809 | GCC supports MIPS DSP operations using both the generic | |
8810 | vector extensions (@pxref{Vector Extensions}) and a collection of | |
8811 | MIPS-specific built-in functions. Both kinds of support are | |
8812 | enabled by the @option{-mdsp} command-line option. | |
8813 | ||
32041385 CF |
8814 | Revision 2 of the ASE was introduced in the second half of 2006. |
8815 | This revision adds extra instructions to the original ASE, but is | |
8816 | otherwise backwards-compatible with it. You can select revision 2 | |
8817 | using the command-line option @option{-mdspr2}; this option implies | |
8818 | @option{-mdsp}. | |
8819 | ||
1e27273f CM |
8820 | The SCOUNT and POS bits of the DSP control register are global. The |
8821 | WRDSP, EXTPDP, EXTPDPV and MTHLIP instructions modify the SCOUNT and | |
8822 | POS bits. During optimization, the compiler will not delete these | |
8823 | instructions and it will not delete calls to functions containing | |
8824 | these instructions. | |
8825 | ||
118ea793 CF |
8826 | At present, GCC only provides support for operations on 32-bit |
8827 | vectors. The vector type associated with 8-bit integer data is | |
32041385 CF |
8828 | usually called @code{v4i8}, the vector type associated with Q7 |
8829 | is usually called @code{v4q7}, the vector type associated with 16-bit | |
8830 | integer data is usually called @code{v2i16}, and the vector type | |
8831 | associated with Q15 is usually called @code{v2q15}. They can be | |
8832 | defined in C as follows: | |
118ea793 CF |
8833 | |
8834 | @smallexample | |
32041385 CF |
8835 | typedef signed char v4i8 __attribute__ ((vector_size(4))); |
8836 | typedef signed char v4q7 __attribute__ ((vector_size(4))); | |
8837 | typedef short v2i16 __attribute__ ((vector_size(4))); | |
118ea793 CF |
8838 | typedef short v2q15 __attribute__ ((vector_size(4))); |
8839 | @end smallexample | |
8840 | ||
32041385 CF |
8841 | @code{v4i8}, @code{v4q7}, @code{v2i16} and @code{v2q15} values are |
8842 | initialized in the same way as aggregates. For example: | |
118ea793 CF |
8843 | |
8844 | @smallexample | |
8845 | v4i8 a = @{1, 2, 3, 4@}; | |
8846 | v4i8 b; | |
8847 | b = (v4i8) @{5, 6, 7, 8@}; | |
8848 | ||
8849 | v2q15 c = @{0x0fcb, 0x3a75@}; | |
8850 | v2q15 d; | |
8851 | d = (v2q15) @{0.1234 * 0x1.0p15, 0.4567 * 0x1.0p15@}; | |
8852 | @end smallexample | |
8853 | ||
8854 | @emph{Note:} The CPU's endianness determines the order in which values | |
8855 | are packed. On little-endian targets, the first value is the least | |
8856 | significant and the last value is the most significant. The opposite | |
8857 | order applies to big-endian targets. For example, the code above will | |
8858 | set the lowest byte of @code{a} to @code{1} on little-endian targets | |
8859 | and @code{4} on big-endian targets. | |
8860 | ||
32041385 | 8861 | @emph{Note:} Q7, Q15 and Q31 values must be initialized with their integer |
118ea793 | 8862 | representation. As shown in this example, the integer representation |
32041385 CF |
8863 | of a Q7 value can be obtained by multiplying the fractional value by |
8864 | @code{0x1.0p7}. The equivalent for Q15 values is to multiply by | |
118ea793 CF |
8865 | @code{0x1.0p15}. The equivalent for Q31 values is to multiply by |
8866 | @code{0x1.0p31}. | |
8867 | ||
8868 | The table below lists the @code{v4i8} and @code{v2q15} operations for which | |
8869 | hardware support exists. @code{a} and @code{b} are @code{v4i8} values, | |
8870 | and @code{c} and @code{d} are @code{v2q15} values. | |
8871 | ||
8872 | @multitable @columnfractions .50 .50 | |
8873 | @item C code @tab MIPS instruction | |
8874 | @item @code{a + b} @tab @code{addu.qb} | |
8875 | @item @code{c + d} @tab @code{addq.ph} | |
8876 | @item @code{a - b} @tab @code{subu.qb} | |
8877 | @item @code{c - d} @tab @code{subq.ph} | |
8878 | @end multitable | |
8879 | ||
32041385 CF |
8880 | The table below lists the @code{v2i16} operation for which |
8881 | hardware support exists for the DSP ASE REV 2. @code{e} and @code{f} are | |
8882 | @code{v2i16} values. | |
8883 | ||
8884 | @multitable @columnfractions .50 .50 | |
8885 | @item C code @tab MIPS instruction | |
8886 | @item @code{e * f} @tab @code{mul.ph} | |
8887 | @end multitable | |
8888 | ||
118ea793 CF |
8889 | It is easier to describe the DSP built-in functions if we first define |
8890 | the following types: | |
8891 | ||
8892 | @smallexample | |
8893 | typedef int q31; | |
8894 | typedef int i32; | |
32041385 | 8895 | typedef unsigned int ui32; |
118ea793 CF |
8896 | typedef long long a64; |
8897 | @end smallexample | |
8898 | ||
8899 | @code{q31} and @code{i32} are actually the same as @code{int}, but we | |
8900 | use @code{q31} to indicate a Q31 fractional value and @code{i32} to | |
8901 | indicate a 32-bit integer value. Similarly, @code{a64} is the same as | |
8902 | @code{long long}, but we use @code{a64} to indicate values that will | |
8903 | be placed in one of the four DSP accumulators (@code{$ac0}, | |
8904 | @code{$ac1}, @code{$ac2} or @code{$ac3}). | |
8905 | ||
8906 | Also, some built-in functions prefer or require immediate numbers as | |
8907 | parameters, because the corresponding DSP instructions accept both immediate | |
8908 | numbers and register operands, or accept immediate numbers only. The | |
8909 | immediate parameters are listed as follows. | |
8910 | ||
8911 | @smallexample | |
32041385 | 8912 | imm0_3: 0 to 3. |
118ea793 CF |
8913 | imm0_7: 0 to 7. |
8914 | imm0_15: 0 to 15. | |
8915 | imm0_31: 0 to 31. | |
8916 | imm0_63: 0 to 63. | |
8917 | imm0_255: 0 to 255. | |
8918 | imm_n32_31: -32 to 31. | |
8919 | imm_n512_511: -512 to 511. | |
8920 | @end smallexample | |
8921 | ||
8922 | The following built-in functions map directly to a particular MIPS DSP | |
8923 | instruction. Please refer to the architecture specification | |
8924 | for details on what each instruction does. | |
8925 | ||
8926 | @smallexample | |
8927 | v2q15 __builtin_mips_addq_ph (v2q15, v2q15) | |
8928 | v2q15 __builtin_mips_addq_s_ph (v2q15, v2q15) | |
8929 | q31 __builtin_mips_addq_s_w (q31, q31) | |
8930 | v4i8 __builtin_mips_addu_qb (v4i8, v4i8) | |
8931 | v4i8 __builtin_mips_addu_s_qb (v4i8, v4i8) | |
8932 | v2q15 __builtin_mips_subq_ph (v2q15, v2q15) | |
8933 | v2q15 __builtin_mips_subq_s_ph (v2q15, v2q15) | |
8934 | q31 __builtin_mips_subq_s_w (q31, q31) | |
8935 | v4i8 __builtin_mips_subu_qb (v4i8, v4i8) | |
8936 | v4i8 __builtin_mips_subu_s_qb (v4i8, v4i8) | |
8937 | i32 __builtin_mips_addsc (i32, i32) | |
8938 | i32 __builtin_mips_addwc (i32, i32) | |
8939 | i32 __builtin_mips_modsub (i32, i32) | |
8940 | i32 __builtin_mips_raddu_w_qb (v4i8) | |
8941 | v2q15 __builtin_mips_absq_s_ph (v2q15) | |
8942 | q31 __builtin_mips_absq_s_w (q31) | |
8943 | v4i8 __builtin_mips_precrq_qb_ph (v2q15, v2q15) | |
8944 | v2q15 __builtin_mips_precrq_ph_w (q31, q31) | |
8945 | v2q15 __builtin_mips_precrq_rs_ph_w (q31, q31) | |
8946 | v4i8 __builtin_mips_precrqu_s_qb_ph (v2q15, v2q15) | |
8947 | q31 __builtin_mips_preceq_w_phl (v2q15) | |
8948 | q31 __builtin_mips_preceq_w_phr (v2q15) | |
8949 | v2q15 __builtin_mips_precequ_ph_qbl (v4i8) | |
8950 | v2q15 __builtin_mips_precequ_ph_qbr (v4i8) | |
8951 | v2q15 __builtin_mips_precequ_ph_qbla (v4i8) | |
8952 | v2q15 __builtin_mips_precequ_ph_qbra (v4i8) | |
8953 | v2q15 __builtin_mips_preceu_ph_qbl (v4i8) | |
8954 | v2q15 __builtin_mips_preceu_ph_qbr (v4i8) | |
8955 | v2q15 __builtin_mips_preceu_ph_qbla (v4i8) | |
8956 | v2q15 __builtin_mips_preceu_ph_qbra (v4i8) | |
8957 | v4i8 __builtin_mips_shll_qb (v4i8, imm0_7) | |
8958 | v4i8 __builtin_mips_shll_qb (v4i8, i32) | |
8959 | v2q15 __builtin_mips_shll_ph (v2q15, imm0_15) | |
8960 | v2q15 __builtin_mips_shll_ph (v2q15, i32) | |
8961 | v2q15 __builtin_mips_shll_s_ph (v2q15, imm0_15) | |
8962 | v2q15 __builtin_mips_shll_s_ph (v2q15, i32) | |
8963 | q31 __builtin_mips_shll_s_w (q31, imm0_31) | |
8964 | q31 __builtin_mips_shll_s_w (q31, i32) | |
8965 | v4i8 __builtin_mips_shrl_qb (v4i8, imm0_7) | |
8966 | v4i8 __builtin_mips_shrl_qb (v4i8, i32) | |
8967 | v2q15 __builtin_mips_shra_ph (v2q15, imm0_15) | |
8968 | v2q15 __builtin_mips_shra_ph (v2q15, i32) | |
8969 | v2q15 __builtin_mips_shra_r_ph (v2q15, imm0_15) | |
8970 | v2q15 __builtin_mips_shra_r_ph (v2q15, i32) | |
8971 | q31 __builtin_mips_shra_r_w (q31, imm0_31) | |
8972 | q31 __builtin_mips_shra_r_w (q31, i32) | |
8973 | v2q15 __builtin_mips_muleu_s_ph_qbl (v4i8, v2q15) | |
8974 | v2q15 __builtin_mips_muleu_s_ph_qbr (v4i8, v2q15) | |
8975 | v2q15 __builtin_mips_mulq_rs_ph (v2q15, v2q15) | |
8976 | q31 __builtin_mips_muleq_s_w_phl (v2q15, v2q15) | |
8977 | q31 __builtin_mips_muleq_s_w_phr (v2q15, v2q15) | |
8978 | a64 __builtin_mips_dpau_h_qbl (a64, v4i8, v4i8) | |
8979 | a64 __builtin_mips_dpau_h_qbr (a64, v4i8, v4i8) | |
8980 | a64 __builtin_mips_dpsu_h_qbl (a64, v4i8, v4i8) | |
8981 | a64 __builtin_mips_dpsu_h_qbr (a64, v4i8, v4i8) | |
8982 | a64 __builtin_mips_dpaq_s_w_ph (a64, v2q15, v2q15) | |
8983 | a64 __builtin_mips_dpaq_sa_l_w (a64, q31, q31) | |
8984 | a64 __builtin_mips_dpsq_s_w_ph (a64, v2q15, v2q15) | |
8985 | a64 __builtin_mips_dpsq_sa_l_w (a64, q31, q31) | |
8986 | a64 __builtin_mips_mulsaq_s_w_ph (a64, v2q15, v2q15) | |
8987 | a64 __builtin_mips_maq_s_w_phl (a64, v2q15, v2q15) | |
8988 | a64 __builtin_mips_maq_s_w_phr (a64, v2q15, v2q15) | |
8989 | a64 __builtin_mips_maq_sa_w_phl (a64, v2q15, v2q15) | |
8990 | a64 __builtin_mips_maq_sa_w_phr (a64, v2q15, v2q15) | |
8991 | i32 __builtin_mips_bitrev (i32) | |
8992 | i32 __builtin_mips_insv (i32, i32) | |
8993 | v4i8 __builtin_mips_repl_qb (imm0_255) | |
8994 | v4i8 __builtin_mips_repl_qb (i32) | |
8995 | v2q15 __builtin_mips_repl_ph (imm_n512_511) | |
8996 | v2q15 __builtin_mips_repl_ph (i32) | |
8997 | void __builtin_mips_cmpu_eq_qb (v4i8, v4i8) | |
8998 | void __builtin_mips_cmpu_lt_qb (v4i8, v4i8) | |
8999 | void __builtin_mips_cmpu_le_qb (v4i8, v4i8) | |
9000 | i32 __builtin_mips_cmpgu_eq_qb (v4i8, v4i8) | |
9001 | i32 __builtin_mips_cmpgu_lt_qb (v4i8, v4i8) | |
9002 | i32 __builtin_mips_cmpgu_le_qb (v4i8, v4i8) | |
9003 | void __builtin_mips_cmp_eq_ph (v2q15, v2q15) | |
9004 | void __builtin_mips_cmp_lt_ph (v2q15, v2q15) | |
9005 | void __builtin_mips_cmp_le_ph (v2q15, v2q15) | |
9006 | v4i8 __builtin_mips_pick_qb (v4i8, v4i8) | |
9007 | v2q15 __builtin_mips_pick_ph (v2q15, v2q15) | |
9008 | v2q15 __builtin_mips_packrl_ph (v2q15, v2q15) | |
9009 | i32 __builtin_mips_extr_w (a64, imm0_31) | |
9010 | i32 __builtin_mips_extr_w (a64, i32) | |
9011 | i32 __builtin_mips_extr_r_w (a64, imm0_31) | |
9012 | i32 __builtin_mips_extr_s_h (a64, i32) | |
9013 | i32 __builtin_mips_extr_rs_w (a64, imm0_31) | |
9014 | i32 __builtin_mips_extr_rs_w (a64, i32) | |
9015 | i32 __builtin_mips_extr_s_h (a64, imm0_31) | |
9016 | i32 __builtin_mips_extr_r_w (a64, i32) | |
9017 | i32 __builtin_mips_extp (a64, imm0_31) | |
9018 | i32 __builtin_mips_extp (a64, i32) | |
9019 | i32 __builtin_mips_extpdp (a64, imm0_31) | |
9020 | i32 __builtin_mips_extpdp (a64, i32) | |
9021 | a64 __builtin_mips_shilo (a64, imm_n32_31) | |
9022 | a64 __builtin_mips_shilo (a64, i32) | |
9023 | a64 __builtin_mips_mthlip (a64, i32) | |
9024 | void __builtin_mips_wrdsp (i32, imm0_63) | |
9025 | i32 __builtin_mips_rddsp (imm0_63) | |
9026 | i32 __builtin_mips_lbux (void *, i32) | |
9027 | i32 __builtin_mips_lhx (void *, i32) | |
9028 | i32 __builtin_mips_lwx (void *, i32) | |
9029 | i32 __builtin_mips_bposge32 (void) | |
9030 | @end smallexample | |
9031 | ||
32041385 CF |
9032 | The following built-in functions map directly to a particular MIPS DSP REV 2 |
9033 | instruction. Please refer to the architecture specification | |
9034 | for details on what each instruction does. | |
9035 | ||
9036 | @smallexample | |
9037 | v4q7 __builtin_mips_absq_s_qb (v4q7); | |
9038 | v2i16 __builtin_mips_addu_ph (v2i16, v2i16); | |
9039 | v2i16 __builtin_mips_addu_s_ph (v2i16, v2i16); | |
9040 | v4i8 __builtin_mips_adduh_qb (v4i8, v4i8); | |
9041 | v4i8 __builtin_mips_adduh_r_qb (v4i8, v4i8); | |
9042 | i32 __builtin_mips_append (i32, i32, imm0_31); | |
9043 | i32 __builtin_mips_balign (i32, i32, imm0_3); | |
9044 | i32 __builtin_mips_cmpgdu_eq_qb (v4i8, v4i8); | |
9045 | i32 __builtin_mips_cmpgdu_lt_qb (v4i8, v4i8); | |
9046 | i32 __builtin_mips_cmpgdu_le_qb (v4i8, v4i8); | |
9047 | a64 __builtin_mips_dpa_w_ph (a64, v2i16, v2i16); | |
9048 | a64 __builtin_mips_dps_w_ph (a64, v2i16, v2i16); | |
9049 | a64 __builtin_mips_madd (a64, i32, i32); | |
9050 | a64 __builtin_mips_maddu (a64, ui32, ui32); | |
9051 | a64 __builtin_mips_msub (a64, i32, i32); | |
9052 | a64 __builtin_mips_msubu (a64, ui32, ui32); | |
9053 | v2i16 __builtin_mips_mul_ph (v2i16, v2i16); | |
9054 | v2i16 __builtin_mips_mul_s_ph (v2i16, v2i16); | |
9055 | q31 __builtin_mips_mulq_rs_w (q31, q31); | |
9056 | v2q15 __builtin_mips_mulq_s_ph (v2q15, v2q15); | |
9057 | q31 __builtin_mips_mulq_s_w (q31, q31); | |
9058 | a64 __builtin_mips_mulsa_w_ph (a64, v2i16, v2i16); | |
9059 | a64 __builtin_mips_mult (i32, i32); | |
9060 | a64 __builtin_mips_multu (ui32, ui32); | |
9061 | v4i8 __builtin_mips_precr_qb_ph (v2i16, v2i16); | |
9062 | v2i16 __builtin_mips_precr_sra_ph_w (i32, i32, imm0_31); | |
9063 | v2i16 __builtin_mips_precr_sra_r_ph_w (i32, i32, imm0_31); | |
9064 | i32 __builtin_mips_prepend (i32, i32, imm0_31); | |
9065 | v4i8 __builtin_mips_shra_qb (v4i8, imm0_7); | |
9066 | v4i8 __builtin_mips_shra_r_qb (v4i8, imm0_7); | |
9067 | v4i8 __builtin_mips_shra_qb (v4i8, i32); | |
9068 | v4i8 __builtin_mips_shra_r_qb (v4i8, i32); | |
9069 | v2i16 __builtin_mips_shrl_ph (v2i16, imm0_15); | |
9070 | v2i16 __builtin_mips_shrl_ph (v2i16, i32); | |
9071 | v2i16 __builtin_mips_subu_ph (v2i16, v2i16); | |
9072 | v2i16 __builtin_mips_subu_s_ph (v2i16, v2i16); | |
9073 | v4i8 __builtin_mips_subuh_qb (v4i8, v4i8); | |
9074 | v4i8 __builtin_mips_subuh_r_qb (v4i8, v4i8); | |
9075 | v2q15 __builtin_mips_addqh_ph (v2q15, v2q15); | |
9076 | v2q15 __builtin_mips_addqh_r_ph (v2q15, v2q15); | |
9077 | q31 __builtin_mips_addqh_w (q31, q31); | |
9078 | q31 __builtin_mips_addqh_r_w (q31, q31); | |
9079 | v2q15 __builtin_mips_subqh_ph (v2q15, v2q15); | |
9080 | v2q15 __builtin_mips_subqh_r_ph (v2q15, v2q15); | |
9081 | q31 __builtin_mips_subqh_w (q31, q31); | |
9082 | q31 __builtin_mips_subqh_r_w (q31, q31); | |
9083 | a64 __builtin_mips_dpax_w_ph (a64, v2i16, v2i16); | |
9084 | a64 __builtin_mips_dpsx_w_ph (a64, v2i16, v2i16); | |
9085 | a64 __builtin_mips_dpaqx_s_w_ph (a64, v2q15, v2q15); | |
9086 | a64 __builtin_mips_dpaqx_sa_w_ph (a64, v2q15, v2q15); | |
9087 | a64 __builtin_mips_dpsqx_s_w_ph (a64, v2q15, v2q15); | |
9088 | a64 __builtin_mips_dpsqx_sa_w_ph (a64, v2q15, v2q15); | |
9089 | @end smallexample | |
9090 | ||
9091 | ||
d840bfd3 CF |
9092 | @node MIPS Paired-Single Support |
9093 | @subsection MIPS Paired-Single Support | |
9094 | ||
9095 | The MIPS64 architecture includes a number of instructions that | |
9096 | operate on pairs of single-precision floating-point values. | |
9097 | Each pair is packed into a 64-bit floating-point register, | |
9098 | with one element being designated the ``upper half'' and | |
9099 | the other being designated the ``lower half''. | |
9100 | ||
9101 | GCC supports paired-single operations using both the generic | |
9102 | vector extensions (@pxref{Vector Extensions}) and a collection of | |
9103 | MIPS-specific built-in functions. Both kinds of support are | |
9104 | enabled by the @option{-mpaired-single} command-line option. | |
9105 | ||
9106 | The vector type associated with paired-single values is usually | |
9107 | called @code{v2sf}. It can be defined in C as follows: | |
9108 | ||
9109 | @smallexample | |
9110 | typedef float v2sf __attribute__ ((vector_size (8))); | |
9111 | @end smallexample | |
9112 | ||
9113 | @code{v2sf} values are initialized in the same way as aggregates. | |
9114 | For example: | |
9115 | ||
9116 | @smallexample | |
9117 | v2sf a = @{1.5, 9.1@}; | |
9118 | v2sf b; | |
9119 | float e, f; | |
9120 | b = (v2sf) @{e, f@}; | |
9121 | @end smallexample | |
9122 | ||
9123 | @emph{Note:} The CPU's endianness determines which value is stored in | |
9124 | the upper half of a register and which value is stored in the lower half. | |
9125 | On little-endian targets, the first value is the lower one and the second | |
9126 | value is the upper one. The opposite order applies to big-endian targets. | |
9127 | For example, the code above will set the lower half of @code{a} to | |
9128 | @code{1.5} on little-endian targets and @code{9.1} on big-endian targets. | |
9129 | ||
93581857 MS |
9130 | @node MIPS Loongson Built-in Functions |
9131 | @subsection MIPS Loongson Built-in Functions | |
9132 | ||
9133 | GCC provides intrinsics to access the SIMD instructions provided by the | |
9134 | ST Microelectronics Loongson-2E and -2F processors. These intrinsics, | |
9135 | available after inclusion of the @code{loongson.h} header file, | |
9136 | operate on the following 64-bit vector types: | |
9137 | ||
9138 | @itemize | |
9139 | @item @code{uint8x8_t}, a vector of eight unsigned 8-bit integers; | |
9140 | @item @code{uint16x4_t}, a vector of four unsigned 16-bit integers; | |
9141 | @item @code{uint32x2_t}, a vector of two unsigned 32-bit integers; | |
9142 | @item @code{int8x8_t}, a vector of eight signed 8-bit integers; | |
9143 | @item @code{int16x4_t}, a vector of four signed 16-bit integers; | |
9144 | @item @code{int32x2_t}, a vector of two signed 32-bit integers. | |
9145 | @end itemize | |
9146 | ||
9147 | The intrinsics provided are listed below; each is named after the | |
9148 | machine instruction to which it corresponds, with suffixes added as | |
9149 | appropriate to distinguish intrinsics that expand to the same machine | |
9150 | instruction yet have different argument types. Refer to the architecture | |
9151 | documentation for a description of the functionality of each | |
9152 | instruction. | |
9153 | ||
9154 | @smallexample | |
9155 | int16x4_t packsswh (int32x2_t s, int32x2_t t); | |
9156 | int8x8_t packsshb (int16x4_t s, int16x4_t t); | |
9157 | uint8x8_t packushb (uint16x4_t s, uint16x4_t t); | |
9158 | uint32x2_t paddw_u (uint32x2_t s, uint32x2_t t); | |
9159 | uint16x4_t paddh_u (uint16x4_t s, uint16x4_t t); | |
9160 | uint8x8_t paddb_u (uint8x8_t s, uint8x8_t t); | |
9161 | int32x2_t paddw_s (int32x2_t s, int32x2_t t); | |
9162 | int16x4_t paddh_s (int16x4_t s, int16x4_t t); | |
9163 | int8x8_t paddb_s (int8x8_t s, int8x8_t t); | |
9164 | uint64_t paddd_u (uint64_t s, uint64_t t); | |
9165 | int64_t paddd_s (int64_t s, int64_t t); | |
9166 | int16x4_t paddsh (int16x4_t s, int16x4_t t); | |
9167 | int8x8_t paddsb (int8x8_t s, int8x8_t t); | |
9168 | uint16x4_t paddush (uint16x4_t s, uint16x4_t t); | |
9169 | uint8x8_t paddusb (uint8x8_t s, uint8x8_t t); | |
9170 | uint64_t pandn_ud (uint64_t s, uint64_t t); | |
9171 | uint32x2_t pandn_uw (uint32x2_t s, uint32x2_t t); | |
9172 | uint16x4_t pandn_uh (uint16x4_t s, uint16x4_t t); | |
9173 | uint8x8_t pandn_ub (uint8x8_t s, uint8x8_t t); | |
9174 | int64_t pandn_sd (int64_t s, int64_t t); | |
9175 | int32x2_t pandn_sw (int32x2_t s, int32x2_t t); | |
9176 | int16x4_t pandn_sh (int16x4_t s, int16x4_t t); | |
9177 | int8x8_t pandn_sb (int8x8_t s, int8x8_t t); | |
9178 | uint16x4_t pavgh (uint16x4_t s, uint16x4_t t); | |
9179 | uint8x8_t pavgb (uint8x8_t s, uint8x8_t t); | |
9180 | uint32x2_t pcmpeqw_u (uint32x2_t s, uint32x2_t t); | |
9181 | uint16x4_t pcmpeqh_u (uint16x4_t s, uint16x4_t t); | |
9182 | uint8x8_t pcmpeqb_u (uint8x8_t s, uint8x8_t t); | |
9183 | int32x2_t pcmpeqw_s (int32x2_t s, int32x2_t t); | |
9184 | int16x4_t pcmpeqh_s (int16x4_t s, int16x4_t t); | |
9185 | int8x8_t pcmpeqb_s (int8x8_t s, int8x8_t t); | |
9186 | uint32x2_t pcmpgtw_u (uint32x2_t s, uint32x2_t t); | |
9187 | uint16x4_t pcmpgth_u (uint16x4_t s, uint16x4_t t); | |
9188 | uint8x8_t pcmpgtb_u (uint8x8_t s, uint8x8_t t); | |
9189 | int32x2_t pcmpgtw_s (int32x2_t s, int32x2_t t); | |
9190 | int16x4_t pcmpgth_s (int16x4_t s, int16x4_t t); | |
9191 | int8x8_t pcmpgtb_s (int8x8_t s, int8x8_t t); | |
9192 | uint16x4_t pextrh_u (uint16x4_t s, int field); | |
9193 | int16x4_t pextrh_s (int16x4_t s, int field); | |
9194 | uint16x4_t pinsrh_0_u (uint16x4_t s, uint16x4_t t); | |
9195 | uint16x4_t pinsrh_1_u (uint16x4_t s, uint16x4_t t); | |
9196 | uint16x4_t pinsrh_2_u (uint16x4_t s, uint16x4_t t); | |
9197 | uint16x4_t pinsrh_3_u (uint16x4_t s, uint16x4_t t); | |
9198 | int16x4_t pinsrh_0_s (int16x4_t s, int16x4_t t); | |
9199 | int16x4_t pinsrh_1_s (int16x4_t s, int16x4_t t); | |
9200 | int16x4_t pinsrh_2_s (int16x4_t s, int16x4_t t); | |
9201 | int16x4_t pinsrh_3_s (int16x4_t s, int16x4_t t); | |
9202 | int32x2_t pmaddhw (int16x4_t s, int16x4_t t); | |
9203 | int16x4_t pmaxsh (int16x4_t s, int16x4_t t); | |
9204 | uint8x8_t pmaxub (uint8x8_t s, uint8x8_t t); | |
9205 | int16x4_t pminsh (int16x4_t s, int16x4_t t); | |
9206 | uint8x8_t pminub (uint8x8_t s, uint8x8_t t); | |
9207 | uint8x8_t pmovmskb_u (uint8x8_t s); | |
9208 | int8x8_t pmovmskb_s (int8x8_t s); | |
9209 | uint16x4_t pmulhuh (uint16x4_t s, uint16x4_t t); | |
9210 | int16x4_t pmulhh (int16x4_t s, int16x4_t t); | |
9211 | int16x4_t pmullh (int16x4_t s, int16x4_t t); | |
9212 | int64_t pmuluw (uint32x2_t s, uint32x2_t t); | |
9213 | uint8x8_t pasubub (uint8x8_t s, uint8x8_t t); | |
9214 | uint16x4_t biadd (uint8x8_t s); | |
9215 | uint16x4_t psadbh (uint8x8_t s, uint8x8_t t); | |
9216 | uint16x4_t pshufh_u (uint16x4_t dest, uint16x4_t s, uint8_t order); | |
9217 | int16x4_t pshufh_s (int16x4_t dest, int16x4_t s, uint8_t order); | |
9218 | uint16x4_t psllh_u (uint16x4_t s, uint8_t amount); | |
9219 | int16x4_t psllh_s (int16x4_t s, uint8_t amount); | |
9220 | uint32x2_t psllw_u (uint32x2_t s, uint8_t amount); | |
9221 | int32x2_t psllw_s (int32x2_t s, uint8_t amount); | |
9222 | uint16x4_t psrlh_u (uint16x4_t s, uint8_t amount); | |
9223 | int16x4_t psrlh_s (int16x4_t s, uint8_t amount); | |
9224 | uint32x2_t psrlw_u (uint32x2_t s, uint8_t amount); | |
9225 | int32x2_t psrlw_s (int32x2_t s, uint8_t amount); | |
9226 | uint16x4_t psrah_u (uint16x4_t s, uint8_t amount); | |
9227 | int16x4_t psrah_s (int16x4_t s, uint8_t amount); | |
9228 | uint32x2_t psraw_u (uint32x2_t s, uint8_t amount); | |
9229 | int32x2_t psraw_s (int32x2_t s, uint8_t amount); | |
9230 | uint32x2_t psubw_u (uint32x2_t s, uint32x2_t t); | |
9231 | uint16x4_t psubh_u (uint16x4_t s, uint16x4_t t); | |
9232 | uint8x8_t psubb_u (uint8x8_t s, uint8x8_t t); | |
9233 | int32x2_t psubw_s (int32x2_t s, int32x2_t t); | |
9234 | int16x4_t psubh_s (int16x4_t s, int16x4_t t); | |
9235 | int8x8_t psubb_s (int8x8_t s, int8x8_t t); | |
9236 | uint64_t psubd_u (uint64_t s, uint64_t t); | |
9237 | int64_t psubd_s (int64_t s, int64_t t); | |
9238 | int16x4_t psubsh (int16x4_t s, int16x4_t t); | |
9239 | int8x8_t psubsb (int8x8_t s, int8x8_t t); | |
9240 | uint16x4_t psubush (uint16x4_t s, uint16x4_t t); | |
9241 | uint8x8_t psubusb (uint8x8_t s, uint8x8_t t); | |
9242 | uint32x2_t punpckhwd_u (uint32x2_t s, uint32x2_t t); | |
9243 | uint16x4_t punpckhhw_u (uint16x4_t s, uint16x4_t t); | |
9244 | uint8x8_t punpckhbh_u (uint8x8_t s, uint8x8_t t); | |
9245 | int32x2_t punpckhwd_s (int32x2_t s, int32x2_t t); | |
9246 | int16x4_t punpckhhw_s (int16x4_t s, int16x4_t t); | |
9247 | int8x8_t punpckhbh_s (int8x8_t s, int8x8_t t); | |
9248 | uint32x2_t punpcklwd_u (uint32x2_t s, uint32x2_t t); | |
9249 | uint16x4_t punpcklhw_u (uint16x4_t s, uint16x4_t t); | |
9250 | uint8x8_t punpcklbh_u (uint8x8_t s, uint8x8_t t); | |
9251 | int32x2_t punpcklwd_s (int32x2_t s, int32x2_t t); | |
9252 | int16x4_t punpcklhw_s (int16x4_t s, int16x4_t t); | |
9253 | int8x8_t punpcklbh_s (int8x8_t s, int8x8_t t); | |
9254 | @end smallexample | |
9255 | ||
d840bfd3 CF |
9256 | @menu |
9257 | * Paired-Single Arithmetic:: | |
9258 | * Paired-Single Built-in Functions:: | |
9259 | * MIPS-3D Built-in Functions:: | |
9260 | @end menu | |
9261 | ||
9262 | @node Paired-Single Arithmetic | |
9263 | @subsubsection Paired-Single Arithmetic | |
9264 | ||
9265 | The table below lists the @code{v2sf} operations for which hardware | |
9266 | support exists. @code{a}, @code{b} and @code{c} are @code{v2sf} | |
9267 | values and @code{x} is an integral value. | |
9268 | ||
9269 | @multitable @columnfractions .50 .50 | |
9270 | @item C code @tab MIPS instruction | |
9271 | @item @code{a + b} @tab @code{add.ps} | |
9272 | @item @code{a - b} @tab @code{sub.ps} | |
9273 | @item @code{-a} @tab @code{neg.ps} | |
9274 | @item @code{a * b} @tab @code{mul.ps} | |
9275 | @item @code{a * b + c} @tab @code{madd.ps} | |
9276 | @item @code{a * b - c} @tab @code{msub.ps} | |
9277 | @item @code{-(a * b + c)} @tab @code{nmadd.ps} | |
9278 | @item @code{-(a * b - c)} @tab @code{nmsub.ps} | |
9279 | @item @code{x ? a : b} @tab @code{movn.ps}/@code{movz.ps} | |
9280 | @end multitable | |
9281 | ||
9282 | Note that the multiply-accumulate instructions can be disabled | |
9283 | using the command-line option @code{-mno-fused-madd}. | |
9284 | ||
9285 | @node Paired-Single Built-in Functions | |
9286 | @subsubsection Paired-Single Built-in Functions | |
9287 | ||
9288 | The following paired-single functions map directly to a particular | |
9289 | MIPS instruction. Please refer to the architecture specification | |
9290 | for details on what each instruction does. | |
9291 | ||
9292 | @table @code | |
9293 | @item v2sf __builtin_mips_pll_ps (v2sf, v2sf) | |
9294 | Pair lower lower (@code{pll.ps}). | |
9295 | ||
9296 | @item v2sf __builtin_mips_pul_ps (v2sf, v2sf) | |
9297 | Pair upper lower (@code{pul.ps}). | |
9298 | ||
9299 | @item v2sf __builtin_mips_plu_ps (v2sf, v2sf) | |
9300 | Pair lower upper (@code{plu.ps}). | |
9301 | ||
9302 | @item v2sf __builtin_mips_puu_ps (v2sf, v2sf) | |
9303 | Pair upper upper (@code{puu.ps}). | |
9304 | ||
9305 | @item v2sf __builtin_mips_cvt_ps_s (float, float) | |
9306 | Convert pair to paired single (@code{cvt.ps.s}). | |
9307 | ||
9308 | @item float __builtin_mips_cvt_s_pl (v2sf) | |
9309 | Convert pair lower to single (@code{cvt.s.pl}). | |
9310 | ||
9311 | @item float __builtin_mips_cvt_s_pu (v2sf) | |
9312 | Convert pair upper to single (@code{cvt.s.pu}). | |
9313 | ||
9314 | @item v2sf __builtin_mips_abs_ps (v2sf) | |
9315 | Absolute value (@code{abs.ps}). | |
9316 | ||
9317 | @item v2sf __builtin_mips_alnv_ps (v2sf, v2sf, int) | |
9318 | Align variable (@code{alnv.ps}). | |
9319 | ||
9320 | @emph{Note:} The value of the third parameter must be 0 or 4 | |
9321 | modulo 8, otherwise the result will be unpredictable. Please read the | |
9322 | instruction description for details. | |
9323 | @end table | |
9324 | ||
9325 | The following multi-instruction functions are also available. | |
9326 | In each case, @var{cond} can be any of the 16 floating-point conditions: | |
9327 | @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult}, | |
9328 | @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, @code{ngl}, | |
9329 | @code{lt}, @code{nge}, @code{le} or @code{ngt}. | |
9330 | ||
9331 | @table @code | |
9332 | @item v2sf __builtin_mips_movt_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
9333 | @itemx v2sf __builtin_mips_movf_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
9334 | Conditional move based on floating point comparison (@code{c.@var{cond}.ps}, | |
9335 | @code{movt.ps}/@code{movf.ps}). | |
9336 | ||
9337 | The @code{movt} functions return the value @var{x} computed by: | |
9338 | ||
9339 | @smallexample | |
9340 | c.@var{cond}.ps @var{cc},@var{a},@var{b} | |
9341 | mov.ps @var{x},@var{c} | |
9342 | movt.ps @var{x},@var{d},@var{cc} | |
9343 | @end smallexample | |
9344 | ||
9345 | The @code{movf} functions are similar but use @code{movf.ps} instead | |
9346 | of @code{movt.ps}. | |
9347 | ||
9348 | @item int __builtin_mips_upper_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
9349 | @itemx int __builtin_mips_lower_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
9350 | Comparison of two paired-single values (@code{c.@var{cond}.ps}, | |
9351 | @code{bc1t}/@code{bc1f}). | |
9352 | ||
9353 | These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps} | |
9354 | and return either the upper or lower half of the result. For example: | |
9355 | ||
9356 | @smallexample | |
9357 | v2sf a, b; | |
9358 | if (__builtin_mips_upper_c_eq_ps (a, b)) | |
9359 | upper_halves_are_equal (); | |
9360 | else | |
9361 | upper_halves_are_unequal (); | |
9362 | ||
9363 | if (__builtin_mips_lower_c_eq_ps (a, b)) | |
9364 | lower_halves_are_equal (); | |
9365 | else | |
9366 | lower_halves_are_unequal (); | |
9367 | @end smallexample | |
9368 | @end table | |
9369 | ||
9370 | @node MIPS-3D Built-in Functions | |
9371 | @subsubsection MIPS-3D Built-in Functions | |
9372 | ||
9373 | The MIPS-3D Application-Specific Extension (ASE) includes additional | |
9374 | paired-single instructions that are designed to improve the performance | |
9375 | of 3D graphics operations. Support for these instructions is controlled | |
9376 | by the @option{-mips3d} command-line option. | |
9377 | ||
9378 | The functions listed below map directly to a particular MIPS-3D | |
9379 | instruction. Please refer to the architecture specification for | |
9380 | more details on what each instruction does. | |
9381 | ||
9382 | @table @code | |
9383 | @item v2sf __builtin_mips_addr_ps (v2sf, v2sf) | |
9384 | Reduction add (@code{addr.ps}). | |
9385 | ||
9386 | @item v2sf __builtin_mips_mulr_ps (v2sf, v2sf) | |
9387 | Reduction multiply (@code{mulr.ps}). | |
9388 | ||
9389 | @item v2sf __builtin_mips_cvt_pw_ps (v2sf) | |
9390 | Convert paired single to paired word (@code{cvt.pw.ps}). | |
9391 | ||
9392 | @item v2sf __builtin_mips_cvt_ps_pw (v2sf) | |
9393 | Convert paired word to paired single (@code{cvt.ps.pw}). | |
9394 | ||
9395 | @item float __builtin_mips_recip1_s (float) | |
9396 | @itemx double __builtin_mips_recip1_d (double) | |
9397 | @itemx v2sf __builtin_mips_recip1_ps (v2sf) | |
9398 | Reduced precision reciprocal (sequence step 1) (@code{recip1.@var{fmt}}). | |
9399 | ||
9400 | @item float __builtin_mips_recip2_s (float, float) | |
9401 | @itemx double __builtin_mips_recip2_d (double, double) | |
9402 | @itemx v2sf __builtin_mips_recip2_ps (v2sf, v2sf) | |
9403 | Reduced precision reciprocal (sequence step 2) (@code{recip2.@var{fmt}}). | |
9404 | ||
9405 | @item float __builtin_mips_rsqrt1_s (float) | |
9406 | @itemx double __builtin_mips_rsqrt1_d (double) | |
9407 | @itemx v2sf __builtin_mips_rsqrt1_ps (v2sf) | |
9408 | Reduced precision reciprocal square root (sequence step 1) | |
9409 | (@code{rsqrt1.@var{fmt}}). | |
9410 | ||
9411 | @item float __builtin_mips_rsqrt2_s (float, float) | |
9412 | @itemx double __builtin_mips_rsqrt2_d (double, double) | |
9413 | @itemx v2sf __builtin_mips_rsqrt2_ps (v2sf, v2sf) | |
9414 | Reduced precision reciprocal square root (sequence step 2) | |
9415 | (@code{rsqrt2.@var{fmt}}). | |
9416 | @end table | |
9417 | ||
9418 | The following multi-instruction functions are also available. | |
9419 | In each case, @var{cond} can be any of the 16 floating-point conditions: | |
9420 | @code{f}, @code{un}, @code{eq}, @code{ueq}, @code{olt}, @code{ult}, | |
9421 | @code{ole}, @code{ule}, @code{sf}, @code{ngle}, @code{seq}, | |
9422 | @code{ngl}, @code{lt}, @code{nge}, @code{le} or @code{ngt}. | |
9423 | ||
9424 | @table @code | |
9425 | @item int __builtin_mips_cabs_@var{cond}_s (float @var{a}, float @var{b}) | |
9426 | @itemx int __builtin_mips_cabs_@var{cond}_d (double @var{a}, double @var{b}) | |
9427 | Absolute comparison of two scalar values (@code{cabs.@var{cond}.@var{fmt}}, | |
9428 | @code{bc1t}/@code{bc1f}). | |
9429 | ||
9430 | These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.s} | |
9431 | or @code{cabs.@var{cond}.d} and return the result as a boolean value. | |
9432 | For example: | |
9433 | ||
9434 | @smallexample | |
9435 | float a, b; | |
9436 | if (__builtin_mips_cabs_eq_s (a, b)) | |
9437 | true (); | |
9438 | else | |
9439 | false (); | |
9440 | @end smallexample | |
9441 | ||
9442 | @item int __builtin_mips_upper_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
9443 | @itemx int __builtin_mips_lower_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
9444 | Absolute comparison of two paired-single values (@code{cabs.@var{cond}.ps}, | |
9445 | @code{bc1t}/@code{bc1f}). | |
9446 | ||
9447 | These functions compare @var{a} and @var{b} using @code{cabs.@var{cond}.ps} | |
9448 | and return either the upper or lower half of the result. For example: | |
9449 | ||
9450 | @smallexample | |
9451 | v2sf a, b; | |
9452 | if (__builtin_mips_upper_cabs_eq_ps (a, b)) | |
9453 | upper_halves_are_equal (); | |
9454 | else | |
9455 | upper_halves_are_unequal (); | |
9456 | ||
9457 | if (__builtin_mips_lower_cabs_eq_ps (a, b)) | |
9458 | lower_halves_are_equal (); | |
9459 | else | |
9460 | lower_halves_are_unequal (); | |
9461 | @end smallexample | |
9462 | ||
9463 | @item v2sf __builtin_mips_movt_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
9464 | @itemx v2sf __builtin_mips_movf_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
9465 | Conditional move based on absolute comparison (@code{cabs.@var{cond}.ps}, | |
9466 | @code{movt.ps}/@code{movf.ps}). | |
9467 | ||
9468 | The @code{movt} functions return the value @var{x} computed by: | |
9469 | ||
9470 | @smallexample | |
9471 | cabs.@var{cond}.ps @var{cc},@var{a},@var{b} | |
9472 | mov.ps @var{x},@var{c} | |
9473 | movt.ps @var{x},@var{d},@var{cc} | |
9474 | @end smallexample | |
9475 | ||
9476 | The @code{movf} functions are similar but use @code{movf.ps} instead | |
9477 | of @code{movt.ps}. | |
9478 | ||
9479 | @item int __builtin_mips_any_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
9480 | @itemx int __builtin_mips_all_c_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
9481 | @itemx int __builtin_mips_any_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
9482 | @itemx int __builtin_mips_all_cabs_@var{cond}_ps (v2sf @var{a}, v2sf @var{b}) | |
9483 | Comparison of two paired-single values | |
9484 | (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps}, | |
9485 | @code{bc1any2t}/@code{bc1any2f}). | |
9486 | ||
9487 | These functions compare @var{a} and @var{b} using @code{c.@var{cond}.ps} | |
9488 | or @code{cabs.@var{cond}.ps}. The @code{any} forms return true if either | |
9489 | result is true and the @code{all} forms return true if both results are true. | |
9490 | For example: | |
9491 | ||
9492 | @smallexample | |
9493 | v2sf a, b; | |
9494 | if (__builtin_mips_any_c_eq_ps (a, b)) | |
9495 | one_is_true (); | |
9496 | else | |
9497 | both_are_false (); | |
9498 | ||
9499 | if (__builtin_mips_all_c_eq_ps (a, b)) | |
9500 | both_are_true (); | |
9501 | else | |
9502 | one_is_false (); | |
9503 | @end smallexample | |
9504 | ||
9505 | @item int __builtin_mips_any_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
9506 | @itemx int __builtin_mips_all_c_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
9507 | @itemx int __builtin_mips_any_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
9508 | @itemx int __builtin_mips_all_cabs_@var{cond}_4s (v2sf @var{a}, v2sf @var{b}, v2sf @var{c}, v2sf @var{d}) | |
9509 | Comparison of four paired-single values | |
9510 | (@code{c.@var{cond}.ps}/@code{cabs.@var{cond}.ps}, | |
9511 | @code{bc1any4t}/@code{bc1any4f}). | |
9512 | ||
9513 | These functions use @code{c.@var{cond}.ps} or @code{cabs.@var{cond}.ps} | |
9514 | to compare @var{a} with @var{b} and to compare @var{c} with @var{d}. | |
9515 | The @code{any} forms return true if any of the four results are true | |
9516 | and the @code{all} forms return true if all four results are true. | |
9517 | For example: | |
9518 | ||
9519 | @smallexample | |
9520 | v2sf a, b, c, d; | |
9521 | if (__builtin_mips_any_c_eq_4s (a, b, c, d)) | |
9522 | some_are_true (); | |
9523 | else | |
9524 | all_are_false (); | |
9525 | ||
9526 | if (__builtin_mips_all_c_eq_4s (a, b, c, d)) | |
9527 | all_are_true (); | |
9528 | else | |
9529 | some_are_false (); | |
9530 | @end smallexample | |
9531 | @end table | |
9532 | ||
358da97e HS |
9533 | @node picoChip Built-in Functions |
9534 | @subsection picoChip Built-in Functions | |
9535 | ||
9536 | GCC provides an interface to selected machine instructions from the | |
9537 | picoChip instruction set. | |
9538 | ||
9539 | @table @code | |
9540 | @item int __builtin_sbc (int @var{value}) | |
9541 | Sign bit count. Return the number of consecutive bits in @var{value} | |
9542 | which have the same value as the sign-bit. The result is the number of | |
9543 | leading sign bits minus one, giving the number of redundant sign bits in | |
9544 | @var{value}. | |
9545 | ||
9546 | @item int __builtin_byteswap (int @var{value}) | |
9547 | Byte swap. Return the result of swapping the upper and lower bytes of | |
9548 | @var{value}. | |
9549 | ||
9550 | @item int __builtin_brev (int @var{value}) | |
9551 | Bit reversal. Return the result of reversing the bits in | |
9552 | @var{value}. Bit 15 is swapped with bit 0, bit 14 is swapped with bit 1, | |
9553 | and so on. | |
9554 | ||
9555 | @item int __builtin_adds (int @var{x}, int @var{y}) | |
9556 | Saturating addition. Return the result of adding @var{x} and @var{y}, | |
9557 | storing the value 32767 if the result overflows. | |
9558 | ||
9559 | @item int __builtin_subs (int @var{x}, int @var{y}) | |
9560 | Saturating subtraction. Return the result of subtracting @var{y} from | |
9561 | @var{x}, storing the value -32768 if the result overflows. | |
9562 | ||
9563 | @item void __builtin_halt (void) | |
9564 | Halt. The processor will stop execution. This built-in is useful for | |
9565 | implementing assertions. | |
9566 | ||
9567 | @end table | |
9568 | ||
4d210b07 RS |
9569 | @node Other MIPS Built-in Functions |
9570 | @subsection Other MIPS Built-in Functions | |
9571 | ||
9572 | GCC provides other MIPS-specific built-in functions: | |
9573 | ||
9574 | @table @code | |
9575 | @item void __builtin_mips_cache (int @var{op}, const volatile void *@var{addr}) | |
9576 | Insert a @samp{cache} instruction with operands @var{op} and @var{addr}. | |
9577 | GCC defines the preprocessor macro @code{___GCC_HAVE_BUILTIN_MIPS_CACHE} | |
9578 | when this function is available. | |
9579 | @end table | |
9580 | ||
333c8841 AH |
9581 | @node PowerPC AltiVec Built-in Functions |
9582 | @subsection PowerPC AltiVec Built-in Functions | |
9583 | ||
b0b343db JJ |
9584 | GCC provides an interface for the PowerPC family of processors to access |
9585 | the AltiVec operations described in Motorola's AltiVec Programming | |
9586 | Interface Manual. The interface is made available by including | |
9587 | @code{<altivec.h>} and using @option{-maltivec} and | |
9588 | @option{-mabi=altivec}. The interface supports the following vector | |
9589 | types. | |
333c8841 | 9590 | |
b0b343db JJ |
9591 | @smallexample |
9592 | vector unsigned char | |
9593 | vector signed char | |
9594 | vector bool char | |
333c8841 | 9595 | |
b0b343db JJ |
9596 | vector unsigned short |
9597 | vector signed short | |
9598 | vector bool short | |
9599 | vector pixel | |
9600 | ||
9601 | vector unsigned int | |
9602 | vector signed int | |
9603 | vector bool int | |
9604 | vector float | |
9605 | @end smallexample | |
9606 | ||
9607 | GCC's implementation of the high-level language interface available from | |
9608 | C and C++ code differs from Motorola's documentation in several ways. | |
9609 | ||
9610 | @itemize @bullet | |
9611 | ||
9612 | @item | |
9613 | A vector constant is a list of constant expressions within curly braces. | |
9614 | ||
9615 | @item | |
9616 | A vector initializer requires no cast if the vector constant is of the | |
9617 | same type as the variable it is initializing. | |
333c8841 | 9618 | |
b0b343db | 9619 | @item |
5edea4c6 JJ |
9620 | If @code{signed} or @code{unsigned} is omitted, the signedness of the |
9621 | vector type is the default signedness of the base type. The default | |
9622 | varies depending on the operating system, so a portable program should | |
9623 | always specify the signedness. | |
4e6e4e4c JJ |
9624 | |
9625 | @item | |
9626 | Compiling with @option{-maltivec} adds keywords @code{__vector}, | |
5950c3c9 BE |
9627 | @code{vector}, @code{__pixel}, @code{pixel}, @code{__bool} and |
9628 | @code{bool}. When compiling ISO C, the context-sensitive substitution | |
9629 | of the keywords @code{vector}, @code{pixel} and @code{bool} is | |
9630 | disabled. To use them, you must include @code{<altivec.h>} instead. | |
4e6e4e4c JJ |
9631 | |
9632 | @item | |
9633 | GCC allows using a @code{typedef} name as the type specifier for a | |
9634 | vector type. | |
b0b343db JJ |
9635 | |
9636 | @item | |
9637 | For C, overloaded functions are implemented with macros so the following | |
9638 | does not work: | |
90989b26 AH |
9639 | |
9640 | @smallexample | |
8254cb45 | 9641 | vec_add ((vector signed int)@{1, 2, 3, 4@}, foo); |
90989b26 AH |
9642 | @end smallexample |
9643 | ||
b0b343db JJ |
9644 | Since @code{vec_add} is a macro, the vector constant in the example |
9645 | is treated as four separate arguments. Wrap the entire argument in | |
9646 | parentheses for this to work. | |
9647 | @end itemize | |
90989b26 | 9648 | |
ae4b4a02 AH |
9649 | @emph{Note:} Only the @code{<altivec.h>} interface is supported. |
9650 | Internally, GCC uses built-in functions to achieve the functionality in | |
9651 | the aforementioned header file, but they are not supported and are | |
9652 | subject to change without notice. | |
9653 | ||
b0b343db JJ |
9654 | The following interfaces are supported for the generic and specific |
9655 | AltiVec operations and the AltiVec predicates. In cases where there | |
9656 | is a direct mapping between generic and specific operations, only the | |
9657 | generic names are shown here, although the specific operations can also | |
9658 | be used. | |
333c8841 | 9659 | |
b0b343db JJ |
9660 | Arguments that are documented as @code{const int} require literal |
9661 | integral values within the range required for that operation. | |
333c8841 | 9662 | |
b0b343db JJ |
9663 | @smallexample |
9664 | vector signed char vec_abs (vector signed char); | |
9665 | vector signed short vec_abs (vector signed short); | |
9666 | vector signed int vec_abs (vector signed int); | |
9667 | vector float vec_abs (vector float); | |
333c8841 | 9668 | |
b0b343db JJ |
9669 | vector signed char vec_abss (vector signed char); |
9670 | vector signed short vec_abss (vector signed short); | |
9671 | vector signed int vec_abss (vector signed int); | |
333c8841 | 9672 | |
b0b343db JJ |
9673 | vector signed char vec_add (vector bool char, vector signed char); |
9674 | vector signed char vec_add (vector signed char, vector bool char); | |
9675 | vector signed char vec_add (vector signed char, vector signed char); | |
9676 | vector unsigned char vec_add (vector bool char, vector unsigned char); | |
9677 | vector unsigned char vec_add (vector unsigned char, vector bool char); | |
924fcc4e JM |
9678 | vector unsigned char vec_add (vector unsigned char, |
9679 | vector unsigned char); | |
b0b343db JJ |
9680 | vector signed short vec_add (vector bool short, vector signed short); |
9681 | vector signed short vec_add (vector signed short, vector bool short); | |
333c8841 | 9682 | vector signed short vec_add (vector signed short, vector signed short); |
b0b343db | 9683 | vector unsigned short vec_add (vector bool short, |
924fcc4e JM |
9684 | vector unsigned short); |
9685 | vector unsigned short vec_add (vector unsigned short, | |
b0b343db | 9686 | vector bool short); |
6e5bb5ad JM |
9687 | vector unsigned short vec_add (vector unsigned short, |
9688 | vector unsigned short); | |
b0b343db JJ |
9689 | vector signed int vec_add (vector bool int, vector signed int); |
9690 | vector signed int vec_add (vector signed int, vector bool int); | |
333c8841 | 9691 | vector signed int vec_add (vector signed int, vector signed int); |
b0b343db JJ |
9692 | vector unsigned int vec_add (vector bool int, vector unsigned int); |
9693 | vector unsigned int vec_add (vector unsigned int, vector bool int); | |
333c8841 AH |
9694 | vector unsigned int vec_add (vector unsigned int, vector unsigned int); |
9695 | vector float vec_add (vector float, vector float); | |
9696 | ||
b0b343db JJ |
9697 | vector float vec_vaddfp (vector float, vector float); |
9698 | ||
9699 | vector signed int vec_vadduwm (vector bool int, vector signed int); | |
9700 | vector signed int vec_vadduwm (vector signed int, vector bool int); | |
9701 | vector signed int vec_vadduwm (vector signed int, vector signed int); | |
9702 | vector unsigned int vec_vadduwm (vector bool int, vector unsigned int); | |
9703 | vector unsigned int vec_vadduwm (vector unsigned int, vector bool int); | |
9704 | vector unsigned int vec_vadduwm (vector unsigned int, | |
9705 | vector unsigned int); | |
9706 | ||
9707 | vector signed short vec_vadduhm (vector bool short, | |
9708 | vector signed short); | |
9709 | vector signed short vec_vadduhm (vector signed short, | |
9710 | vector bool short); | |
9711 | vector signed short vec_vadduhm (vector signed short, | |
9712 | vector signed short); | |
9713 | vector unsigned short vec_vadduhm (vector bool short, | |
9714 | vector unsigned short); | |
9715 | vector unsigned short vec_vadduhm (vector unsigned short, | |
9716 | vector bool short); | |
9717 | vector unsigned short vec_vadduhm (vector unsigned short, | |
9718 | vector unsigned short); | |
9719 | ||
9720 | vector signed char vec_vaddubm (vector bool char, vector signed char); | |
9721 | vector signed char vec_vaddubm (vector signed char, vector bool char); | |
9722 | vector signed char vec_vaddubm (vector signed char, vector signed char); | |
9723 | vector unsigned char vec_vaddubm (vector bool char, | |
9724 | vector unsigned char); | |
9725 | vector unsigned char vec_vaddubm (vector unsigned char, | |
9726 | vector bool char); | |
9727 | vector unsigned char vec_vaddubm (vector unsigned char, | |
9728 | vector unsigned char); | |
9729 | ||
333c8841 AH |
9730 | vector unsigned int vec_addc (vector unsigned int, vector unsigned int); |
9731 | ||
b0b343db JJ |
9732 | vector unsigned char vec_adds (vector bool char, vector unsigned char); |
9733 | vector unsigned char vec_adds (vector unsigned char, vector bool char); | |
924fcc4e JM |
9734 | vector unsigned char vec_adds (vector unsigned char, |
9735 | vector unsigned char); | |
b0b343db JJ |
9736 | vector signed char vec_adds (vector bool char, vector signed char); |
9737 | vector signed char vec_adds (vector signed char, vector bool char); | |
333c8841 | 9738 | vector signed char vec_adds (vector signed char, vector signed char); |
b0b343db | 9739 | vector unsigned short vec_adds (vector bool short, |
924fcc4e JM |
9740 | vector unsigned short); |
9741 | vector unsigned short vec_adds (vector unsigned short, | |
b0b343db | 9742 | vector bool short); |
6e5bb5ad JM |
9743 | vector unsigned short vec_adds (vector unsigned short, |
9744 | vector unsigned short); | |
b0b343db JJ |
9745 | vector signed short vec_adds (vector bool short, vector signed short); |
9746 | vector signed short vec_adds (vector signed short, vector bool short); | |
333c8841 | 9747 | vector signed short vec_adds (vector signed short, vector signed short); |
b0b343db JJ |
9748 | vector unsigned int vec_adds (vector bool int, vector unsigned int); |
9749 | vector unsigned int vec_adds (vector unsigned int, vector bool int); | |
333c8841 | 9750 | vector unsigned int vec_adds (vector unsigned int, vector unsigned int); |
b0b343db JJ |
9751 | vector signed int vec_adds (vector bool int, vector signed int); |
9752 | vector signed int vec_adds (vector signed int, vector bool int); | |
333c8841 AH |
9753 | vector signed int vec_adds (vector signed int, vector signed int); |
9754 | ||
b0b343db JJ |
9755 | vector signed int vec_vaddsws (vector bool int, vector signed int); |
9756 | vector signed int vec_vaddsws (vector signed int, vector bool int); | |
9757 | vector signed int vec_vaddsws (vector signed int, vector signed int); | |
9758 | ||
9759 | vector unsigned int vec_vadduws (vector bool int, vector unsigned int); | |
9760 | vector unsigned int vec_vadduws (vector unsigned int, vector bool int); | |
9761 | vector unsigned int vec_vadduws (vector unsigned int, | |
9762 | vector unsigned int); | |
9763 | ||
9764 | vector signed short vec_vaddshs (vector bool short, | |
9765 | vector signed short); | |
9766 | vector signed short vec_vaddshs (vector signed short, | |
9767 | vector bool short); | |
9768 | vector signed short vec_vaddshs (vector signed short, | |
9769 | vector signed short); | |
9770 | ||
9771 | vector unsigned short vec_vadduhs (vector bool short, | |
9772 | vector unsigned short); | |
9773 | vector unsigned short vec_vadduhs (vector unsigned short, | |
9774 | vector bool short); | |
9775 | vector unsigned short vec_vadduhs (vector unsigned short, | |
9776 | vector unsigned short); | |
9777 | ||
9778 | vector signed char vec_vaddsbs (vector bool char, vector signed char); | |
9779 | vector signed char vec_vaddsbs (vector signed char, vector bool char); | |
9780 | vector signed char vec_vaddsbs (vector signed char, vector signed char); | |
9781 | ||
9782 | vector unsigned char vec_vaddubs (vector bool char, | |
9783 | vector unsigned char); | |
9784 | vector unsigned char vec_vaddubs (vector unsigned char, | |
9785 | vector bool char); | |
9786 | vector unsigned char vec_vaddubs (vector unsigned char, | |
9787 | vector unsigned char); | |
9788 | ||
333c8841 | 9789 | vector float vec_and (vector float, vector float); |
b0b343db JJ |
9790 | vector float vec_and (vector float, vector bool int); |
9791 | vector float vec_and (vector bool int, vector float); | |
9792 | vector bool int vec_and (vector bool int, vector bool int); | |
9793 | vector signed int vec_and (vector bool int, vector signed int); | |
9794 | vector signed int vec_and (vector signed int, vector bool int); | |
333c8841 | 9795 | vector signed int vec_and (vector signed int, vector signed int); |
b0b343db JJ |
9796 | vector unsigned int vec_and (vector bool int, vector unsigned int); |
9797 | vector unsigned int vec_and (vector unsigned int, vector bool int); | |
333c8841 | 9798 | vector unsigned int vec_and (vector unsigned int, vector unsigned int); |
b0b343db JJ |
9799 | vector bool short vec_and (vector bool short, vector bool short); |
9800 | vector signed short vec_and (vector bool short, vector signed short); | |
9801 | vector signed short vec_and (vector signed short, vector bool short); | |
333c8841 | 9802 | vector signed short vec_and (vector signed short, vector signed short); |
b0b343db | 9803 | vector unsigned short vec_and (vector bool short, |
924fcc4e JM |
9804 | vector unsigned short); |
9805 | vector unsigned short vec_and (vector unsigned short, | |
b0b343db | 9806 | vector bool short); |
6e5bb5ad JM |
9807 | vector unsigned short vec_and (vector unsigned short, |
9808 | vector unsigned short); | |
b0b343db JJ |
9809 | vector signed char vec_and (vector bool char, vector signed char); |
9810 | vector bool char vec_and (vector bool char, vector bool char); | |
9811 | vector signed char vec_and (vector signed char, vector bool char); | |
333c8841 | 9812 | vector signed char vec_and (vector signed char, vector signed char); |
b0b343db JJ |
9813 | vector unsigned char vec_and (vector bool char, vector unsigned char); |
9814 | vector unsigned char vec_and (vector unsigned char, vector bool char); | |
924fcc4e JM |
9815 | vector unsigned char vec_and (vector unsigned char, |
9816 | vector unsigned char); | |
333c8841 AH |
9817 | |
9818 | vector float vec_andc (vector float, vector float); | |
b0b343db JJ |
9819 | vector float vec_andc (vector float, vector bool int); |
9820 | vector float vec_andc (vector bool int, vector float); | |
9821 | vector bool int vec_andc (vector bool int, vector bool int); | |
9822 | vector signed int vec_andc (vector bool int, vector signed int); | |
9823 | vector signed int vec_andc (vector signed int, vector bool int); | |
333c8841 | 9824 | vector signed int vec_andc (vector signed int, vector signed int); |
b0b343db JJ |
9825 | vector unsigned int vec_andc (vector bool int, vector unsigned int); |
9826 | vector unsigned int vec_andc (vector unsigned int, vector bool int); | |
333c8841 | 9827 | vector unsigned int vec_andc (vector unsigned int, vector unsigned int); |
b0b343db JJ |
9828 | vector bool short vec_andc (vector bool short, vector bool short); |
9829 | vector signed short vec_andc (vector bool short, vector signed short); | |
9830 | vector signed short vec_andc (vector signed short, vector bool short); | |
333c8841 | 9831 | vector signed short vec_andc (vector signed short, vector signed short); |
b0b343db | 9832 | vector unsigned short vec_andc (vector bool short, |
924fcc4e JM |
9833 | vector unsigned short); |
9834 | vector unsigned short vec_andc (vector unsigned short, | |
b0b343db | 9835 | vector bool short); |
6e5bb5ad JM |
9836 | vector unsigned short vec_andc (vector unsigned short, |
9837 | vector unsigned short); | |
b0b343db JJ |
9838 | vector signed char vec_andc (vector bool char, vector signed char); |
9839 | vector bool char vec_andc (vector bool char, vector bool char); | |
9840 | vector signed char vec_andc (vector signed char, vector bool char); | |
333c8841 | 9841 | vector signed char vec_andc (vector signed char, vector signed char); |
b0b343db JJ |
9842 | vector unsigned char vec_andc (vector bool char, vector unsigned char); |
9843 | vector unsigned char vec_andc (vector unsigned char, vector bool char); | |
924fcc4e JM |
9844 | vector unsigned char vec_andc (vector unsigned char, |
9845 | vector unsigned char); | |
333c8841 | 9846 | |
924fcc4e JM |
9847 | vector unsigned char vec_avg (vector unsigned char, |
9848 | vector unsigned char); | |
333c8841 | 9849 | vector signed char vec_avg (vector signed char, vector signed char); |
6e5bb5ad JM |
9850 | vector unsigned short vec_avg (vector unsigned short, |
9851 | vector unsigned short); | |
333c8841 AH |
9852 | vector signed short vec_avg (vector signed short, vector signed short); |
9853 | vector unsigned int vec_avg (vector unsigned int, vector unsigned int); | |
9854 | vector signed int vec_avg (vector signed int, vector signed int); | |
9855 | ||
b0b343db JJ |
9856 | vector signed int vec_vavgsw (vector signed int, vector signed int); |
9857 | ||
9858 | vector unsigned int vec_vavguw (vector unsigned int, | |
9859 | vector unsigned int); | |
9860 | ||
9861 | vector signed short vec_vavgsh (vector signed short, | |
9862 | vector signed short); | |
9863 | ||
9864 | vector unsigned short vec_vavguh (vector unsigned short, | |
9865 | vector unsigned short); | |
9866 | ||
9867 | vector signed char vec_vavgsb (vector signed char, vector signed char); | |
9868 | ||
9869 | vector unsigned char vec_vavgub (vector unsigned char, | |
9870 | vector unsigned char); | |
9871 | ||
333c8841 AH |
9872 | vector float vec_ceil (vector float); |
9873 | ||
9874 | vector signed int vec_cmpb (vector float, vector float); | |
9875 | ||
b0b343db JJ |
9876 | vector bool char vec_cmpeq (vector signed char, vector signed char); |
9877 | vector bool char vec_cmpeq (vector unsigned char, vector unsigned char); | |
9878 | vector bool short vec_cmpeq (vector signed short, vector signed short); | |
9879 | vector bool short vec_cmpeq (vector unsigned short, | |
9880 | vector unsigned short); | |
9881 | vector bool int vec_cmpeq (vector signed int, vector signed int); | |
9882 | vector bool int vec_cmpeq (vector unsigned int, vector unsigned int); | |
9883 | vector bool int vec_cmpeq (vector float, vector float); | |
333c8841 | 9884 | |
b0b343db | 9885 | vector bool int vec_vcmpeqfp (vector float, vector float); |
333c8841 | 9886 | |
b0b343db JJ |
9887 | vector bool int vec_vcmpequw (vector signed int, vector signed int); |
9888 | vector bool int vec_vcmpequw (vector unsigned int, vector unsigned int); | |
9889 | ||
9890 | vector bool short vec_vcmpequh (vector signed short, | |
9891 | vector signed short); | |
9892 | vector bool short vec_vcmpequh (vector unsigned short, | |
9893 | vector unsigned short); | |
333c8841 | 9894 | |
b0b343db JJ |
9895 | vector bool char vec_vcmpequb (vector signed char, vector signed char); |
9896 | vector bool char vec_vcmpequb (vector unsigned char, | |
9897 | vector unsigned char); | |
333c8841 | 9898 | |
b0b343db | 9899 | vector bool int vec_cmpge (vector float, vector float); |
333c8841 | 9900 | |
b0b343db JJ |
9901 | vector bool char vec_cmpgt (vector unsigned char, vector unsigned char); |
9902 | vector bool char vec_cmpgt (vector signed char, vector signed char); | |
9903 | vector bool short vec_cmpgt (vector unsigned short, | |
9904 | vector unsigned short); | |
9905 | vector bool short vec_cmpgt (vector signed short, vector signed short); | |
9906 | vector bool int vec_cmpgt (vector unsigned int, vector unsigned int); | |
9907 | vector bool int vec_cmpgt (vector signed int, vector signed int); | |
9908 | vector bool int vec_cmpgt (vector float, vector float); | |
9909 | ||
9910 | vector bool int vec_vcmpgtfp (vector float, vector float); | |
9911 | ||
9912 | vector bool int vec_vcmpgtsw (vector signed int, vector signed int); | |
9913 | ||
9914 | vector bool int vec_vcmpgtuw (vector unsigned int, vector unsigned int); | |
333c8841 | 9915 | |
b0b343db JJ |
9916 | vector bool short vec_vcmpgtsh (vector signed short, |
9917 | vector signed short); | |
9918 | ||
9919 | vector bool short vec_vcmpgtuh (vector unsigned short, | |
9920 | vector unsigned short); | |
9921 | ||
9922 | vector bool char vec_vcmpgtsb (vector signed char, vector signed char); | |
9923 | ||
9924 | vector bool char vec_vcmpgtub (vector unsigned char, | |
9925 | vector unsigned char); | |
9926 | ||
9927 | vector bool int vec_cmple (vector float, vector float); | |
9928 | ||
9929 | vector bool char vec_cmplt (vector unsigned char, vector unsigned char); | |
9930 | vector bool char vec_cmplt (vector signed char, vector signed char); | |
9931 | vector bool short vec_cmplt (vector unsigned short, | |
9932 | vector unsigned short); | |
9933 | vector bool short vec_cmplt (vector signed short, vector signed short); | |
9934 | vector bool int vec_cmplt (vector unsigned int, vector unsigned int); | |
9935 | vector bool int vec_cmplt (vector signed int, vector signed int); | |
9936 | vector bool int vec_cmplt (vector float, vector float); | |
333c8841 | 9937 | |
b0b343db JJ |
9938 | vector float vec_ctf (vector unsigned int, const int); |
9939 | vector float vec_ctf (vector signed int, const int); | |
333c8841 | 9940 | |
b0b343db JJ |
9941 | vector float vec_vcfsx (vector signed int, const int); |
9942 | ||
9943 | vector float vec_vcfux (vector unsigned int, const int); | |
9944 | ||
9945 | vector signed int vec_cts (vector float, const int); | |
9946 | ||
9947 | vector unsigned int vec_ctu (vector float, const int); | |
9948 | ||
9949 | void vec_dss (const int); | |
333c8841 AH |
9950 | |
9951 | void vec_dssall (void); | |
9952 | ||
b0b343db JJ |
9953 | void vec_dst (const vector unsigned char *, int, const int); |
9954 | void vec_dst (const vector signed char *, int, const int); | |
9955 | void vec_dst (const vector bool char *, int, const int); | |
9956 | void vec_dst (const vector unsigned short *, int, const int); | |
9957 | void vec_dst (const vector signed short *, int, const int); | |
9958 | void vec_dst (const vector bool short *, int, const int); | |
9959 | void vec_dst (const vector pixel *, int, const int); | |
9960 | void vec_dst (const vector unsigned int *, int, const int); | |
9961 | void vec_dst (const vector signed int *, int, const int); | |
9962 | void vec_dst (const vector bool int *, int, const int); | |
9963 | void vec_dst (const vector float *, int, const int); | |
9964 | void vec_dst (const unsigned char *, int, const int); | |
9965 | void vec_dst (const signed char *, int, const int); | |
9966 | void vec_dst (const unsigned short *, int, const int); | |
9967 | void vec_dst (const short *, int, const int); | |
9968 | void vec_dst (const unsigned int *, int, const int); | |
9969 | void vec_dst (const int *, int, const int); | |
9970 | void vec_dst (const unsigned long *, int, const int); | |
9971 | void vec_dst (const long *, int, const int); | |
9972 | void vec_dst (const float *, int, const int); | |
9973 | ||
9974 | void vec_dstst (const vector unsigned char *, int, const int); | |
9975 | void vec_dstst (const vector signed char *, int, const int); | |
9976 | void vec_dstst (const vector bool char *, int, const int); | |
9977 | void vec_dstst (const vector unsigned short *, int, const int); | |
9978 | void vec_dstst (const vector signed short *, int, const int); | |
9979 | void vec_dstst (const vector bool short *, int, const int); | |
9980 | void vec_dstst (const vector pixel *, int, const int); | |
9981 | void vec_dstst (const vector unsigned int *, int, const int); | |
9982 | void vec_dstst (const vector signed int *, int, const int); | |
9983 | void vec_dstst (const vector bool int *, int, const int); | |
9984 | void vec_dstst (const vector float *, int, const int); | |
9985 | void vec_dstst (const unsigned char *, int, const int); | |
9986 | void vec_dstst (const signed char *, int, const int); | |
9987 | void vec_dstst (const unsigned short *, int, const int); | |
9988 | void vec_dstst (const short *, int, const int); | |
9989 | void vec_dstst (const unsigned int *, int, const int); | |
9990 | void vec_dstst (const int *, int, const int); | |
9991 | void vec_dstst (const unsigned long *, int, const int); | |
9992 | void vec_dstst (const long *, int, const int); | |
9993 | void vec_dstst (const float *, int, const int); | |
9994 | ||
9995 | void vec_dststt (const vector unsigned char *, int, const int); | |
9996 | void vec_dststt (const vector signed char *, int, const int); | |
9997 | void vec_dststt (const vector bool char *, int, const int); | |
9998 | void vec_dststt (const vector unsigned short *, int, const int); | |
9999 | void vec_dststt (const vector signed short *, int, const int); | |
10000 | void vec_dststt (const vector bool short *, int, const int); | |
10001 | void vec_dststt (const vector pixel *, int, const int); | |
10002 | void vec_dststt (const vector unsigned int *, int, const int); | |
10003 | void vec_dststt (const vector signed int *, int, const int); | |
10004 | void vec_dststt (const vector bool int *, int, const int); | |
10005 | void vec_dststt (const vector float *, int, const int); | |
10006 | void vec_dststt (const unsigned char *, int, const int); | |
10007 | void vec_dststt (const signed char *, int, const int); | |
10008 | void vec_dststt (const unsigned short *, int, const int); | |
10009 | void vec_dststt (const short *, int, const int); | |
10010 | void vec_dststt (const unsigned int *, int, const int); | |
10011 | void vec_dststt (const int *, int, const int); | |
10012 | void vec_dststt (const unsigned long *, int, const int); | |
10013 | void vec_dststt (const long *, int, const int); | |
10014 | void vec_dststt (const float *, int, const int); | |
10015 | ||
10016 | void vec_dstt (const vector unsigned char *, int, const int); | |
10017 | void vec_dstt (const vector signed char *, int, const int); | |
10018 | void vec_dstt (const vector bool char *, int, const int); | |
10019 | void vec_dstt (const vector unsigned short *, int, const int); | |
10020 | void vec_dstt (const vector signed short *, int, const int); | |
10021 | void vec_dstt (const vector bool short *, int, const int); | |
10022 | void vec_dstt (const vector pixel *, int, const int); | |
10023 | void vec_dstt (const vector unsigned int *, int, const int); | |
10024 | void vec_dstt (const vector signed int *, int, const int); | |
10025 | void vec_dstt (const vector bool int *, int, const int); | |
10026 | void vec_dstt (const vector float *, int, const int); | |
10027 | void vec_dstt (const unsigned char *, int, const int); | |
10028 | void vec_dstt (const signed char *, int, const int); | |
10029 | void vec_dstt (const unsigned short *, int, const int); | |
10030 | void vec_dstt (const short *, int, const int); | |
10031 | void vec_dstt (const unsigned int *, int, const int); | |
10032 | void vec_dstt (const int *, int, const int); | |
10033 | void vec_dstt (const unsigned long *, int, const int); | |
10034 | void vec_dstt (const long *, int, const int); | |
10035 | void vec_dstt (const float *, int, const int); | |
10036 | ||
10037 | vector float vec_expte (vector float); | |
10038 | ||
10039 | vector float vec_floor (vector float); | |
10040 | ||
10041 | vector float vec_ld (int, const vector float *); | |
10042 | vector float vec_ld (int, const float *); | |
10043 | vector bool int vec_ld (int, const vector bool int *); | |
10044 | vector signed int vec_ld (int, const vector signed int *); | |
10045 | vector signed int vec_ld (int, const int *); | |
10046 | vector signed int vec_ld (int, const long *); | |
10047 | vector unsigned int vec_ld (int, const vector unsigned int *); | |
10048 | vector unsigned int vec_ld (int, const unsigned int *); | |
10049 | vector unsigned int vec_ld (int, const unsigned long *); | |
10050 | vector bool short vec_ld (int, const vector bool short *); | |
10051 | vector pixel vec_ld (int, const vector pixel *); | |
10052 | vector signed short vec_ld (int, const vector signed short *); | |
10053 | vector signed short vec_ld (int, const short *); | |
10054 | vector unsigned short vec_ld (int, const vector unsigned short *); | |
10055 | vector unsigned short vec_ld (int, const unsigned short *); | |
10056 | vector bool char vec_ld (int, const vector bool char *); | |
10057 | vector signed char vec_ld (int, const vector signed char *); | |
10058 | vector signed char vec_ld (int, const signed char *); | |
10059 | vector unsigned char vec_ld (int, const vector unsigned char *); | |
10060 | vector unsigned char vec_ld (int, const unsigned char *); | |
10061 | ||
10062 | vector signed char vec_lde (int, const signed char *); | |
10063 | vector unsigned char vec_lde (int, const unsigned char *); | |
10064 | vector signed short vec_lde (int, const short *); | |
10065 | vector unsigned short vec_lde (int, const unsigned short *); | |
10066 | vector float vec_lde (int, const float *); | |
10067 | vector signed int vec_lde (int, const int *); | |
10068 | vector unsigned int vec_lde (int, const unsigned int *); | |
10069 | vector signed int vec_lde (int, const long *); | |
10070 | vector unsigned int vec_lde (int, const unsigned long *); | |
10071 | ||
10072 | vector float vec_lvewx (int, float *); | |
10073 | vector signed int vec_lvewx (int, int *); | |
10074 | vector unsigned int vec_lvewx (int, unsigned int *); | |
10075 | vector signed int vec_lvewx (int, long *); | |
10076 | vector unsigned int vec_lvewx (int, unsigned long *); | |
10077 | ||
10078 | vector signed short vec_lvehx (int, short *); | |
10079 | vector unsigned short vec_lvehx (int, unsigned short *); | |
10080 | ||
10081 | vector signed char vec_lvebx (int, char *); | |
10082 | vector unsigned char vec_lvebx (int, unsigned char *); | |
10083 | ||
10084 | vector float vec_ldl (int, const vector float *); | |
10085 | vector float vec_ldl (int, const float *); | |
10086 | vector bool int vec_ldl (int, const vector bool int *); | |
10087 | vector signed int vec_ldl (int, const vector signed int *); | |
10088 | vector signed int vec_ldl (int, const int *); | |
10089 | vector signed int vec_ldl (int, const long *); | |
10090 | vector unsigned int vec_ldl (int, const vector unsigned int *); | |
10091 | vector unsigned int vec_ldl (int, const unsigned int *); | |
10092 | vector unsigned int vec_ldl (int, const unsigned long *); | |
10093 | vector bool short vec_ldl (int, const vector bool short *); | |
10094 | vector pixel vec_ldl (int, const vector pixel *); | |
10095 | vector signed short vec_ldl (int, const vector signed short *); | |
10096 | vector signed short vec_ldl (int, const short *); | |
10097 | vector unsigned short vec_ldl (int, const vector unsigned short *); | |
10098 | vector unsigned short vec_ldl (int, const unsigned short *); | |
10099 | vector bool char vec_ldl (int, const vector bool char *); | |
10100 | vector signed char vec_ldl (int, const vector signed char *); | |
10101 | vector signed char vec_ldl (int, const signed char *); | |
10102 | vector unsigned char vec_ldl (int, const vector unsigned char *); | |
10103 | vector unsigned char vec_ldl (int, const unsigned char *); | |
333c8841 AH |
10104 | |
10105 | vector float vec_loge (vector float); | |
10106 | ||
b0b343db JJ |
10107 | vector unsigned char vec_lvsl (int, const volatile unsigned char *); |
10108 | vector unsigned char vec_lvsl (int, const volatile signed char *); | |
10109 | vector unsigned char vec_lvsl (int, const volatile unsigned short *); | |
10110 | vector unsigned char vec_lvsl (int, const volatile short *); | |
10111 | vector unsigned char vec_lvsl (int, const volatile unsigned int *); | |
10112 | vector unsigned char vec_lvsl (int, const volatile int *); | |
10113 | vector unsigned char vec_lvsl (int, const volatile unsigned long *); | |
10114 | vector unsigned char vec_lvsl (int, const volatile long *); | |
10115 | vector unsigned char vec_lvsl (int, const volatile float *); | |
10116 | ||
10117 | vector unsigned char vec_lvsr (int, const volatile unsigned char *); | |
10118 | vector unsigned char vec_lvsr (int, const volatile signed char *); | |
10119 | vector unsigned char vec_lvsr (int, const volatile unsigned short *); | |
10120 | vector unsigned char vec_lvsr (int, const volatile short *); | |
10121 | vector unsigned char vec_lvsr (int, const volatile unsigned int *); | |
10122 | vector unsigned char vec_lvsr (int, const volatile int *); | |
10123 | vector unsigned char vec_lvsr (int, const volatile unsigned long *); | |
10124 | vector unsigned char vec_lvsr (int, const volatile long *); | |
10125 | vector unsigned char vec_lvsr (int, const volatile float *); | |
333c8841 AH |
10126 | |
10127 | vector float vec_madd (vector float, vector float, vector float); | |
10128 | ||
b0b343db JJ |
10129 | vector signed short vec_madds (vector signed short, |
10130 | vector signed short, | |
6e5bb5ad | 10131 | vector signed short); |
333c8841 | 10132 | |
b0b343db JJ |
10133 | vector unsigned char vec_max (vector bool char, vector unsigned char); |
10134 | vector unsigned char vec_max (vector unsigned char, vector bool char); | |
924fcc4e JM |
10135 | vector unsigned char vec_max (vector unsigned char, |
10136 | vector unsigned char); | |
b0b343db JJ |
10137 | vector signed char vec_max (vector bool char, vector signed char); |
10138 | vector signed char vec_max (vector signed char, vector bool char); | |
333c8841 | 10139 | vector signed char vec_max (vector signed char, vector signed char); |
b0b343db | 10140 | vector unsigned short vec_max (vector bool short, |
924fcc4e JM |
10141 | vector unsigned short); |
10142 | vector unsigned short vec_max (vector unsigned short, | |
b0b343db | 10143 | vector bool short); |
6e5bb5ad JM |
10144 | vector unsigned short vec_max (vector unsigned short, |
10145 | vector unsigned short); | |
b0b343db JJ |
10146 | vector signed short vec_max (vector bool short, vector signed short); |
10147 | vector signed short vec_max (vector signed short, vector bool short); | |
333c8841 | 10148 | vector signed short vec_max (vector signed short, vector signed short); |
b0b343db JJ |
10149 | vector unsigned int vec_max (vector bool int, vector unsigned int); |
10150 | vector unsigned int vec_max (vector unsigned int, vector bool int); | |
333c8841 | 10151 | vector unsigned int vec_max (vector unsigned int, vector unsigned int); |
b0b343db JJ |
10152 | vector signed int vec_max (vector bool int, vector signed int); |
10153 | vector signed int vec_max (vector signed int, vector bool int); | |
333c8841 AH |
10154 | vector signed int vec_max (vector signed int, vector signed int); |
10155 | vector float vec_max (vector float, vector float); | |
10156 | ||
b0b343db JJ |
10157 | vector float vec_vmaxfp (vector float, vector float); |
10158 | ||
10159 | vector signed int vec_vmaxsw (vector bool int, vector signed int); | |
10160 | vector signed int vec_vmaxsw (vector signed int, vector bool int); | |
10161 | vector signed int vec_vmaxsw (vector signed int, vector signed int); | |
10162 | ||
10163 | vector unsigned int vec_vmaxuw (vector bool int, vector unsigned int); | |
10164 | vector unsigned int vec_vmaxuw (vector unsigned int, vector bool int); | |
10165 | vector unsigned int vec_vmaxuw (vector unsigned int, | |
10166 | vector unsigned int); | |
10167 | ||
10168 | vector signed short vec_vmaxsh (vector bool short, vector signed short); | |
10169 | vector signed short vec_vmaxsh (vector signed short, vector bool short); | |
10170 | vector signed short vec_vmaxsh (vector signed short, | |
10171 | vector signed short); | |
10172 | ||
10173 | vector unsigned short vec_vmaxuh (vector bool short, | |
10174 | vector unsigned short); | |
10175 | vector unsigned short vec_vmaxuh (vector unsigned short, | |
10176 | vector bool short); | |
10177 | vector unsigned short vec_vmaxuh (vector unsigned short, | |
10178 | vector unsigned short); | |
10179 | ||
10180 | vector signed char vec_vmaxsb (vector bool char, vector signed char); | |
10181 | vector signed char vec_vmaxsb (vector signed char, vector bool char); | |
10182 | vector signed char vec_vmaxsb (vector signed char, vector signed char); | |
10183 | ||
10184 | vector unsigned char vec_vmaxub (vector bool char, | |
10185 | vector unsigned char); | |
10186 | vector unsigned char vec_vmaxub (vector unsigned char, | |
10187 | vector bool char); | |
10188 | vector unsigned char vec_vmaxub (vector unsigned char, | |
10189 | vector unsigned char); | |
10190 | ||
10191 | vector bool char vec_mergeh (vector bool char, vector bool char); | |
333c8841 | 10192 | vector signed char vec_mergeh (vector signed char, vector signed char); |
6e5bb5ad JM |
10193 | vector unsigned char vec_mergeh (vector unsigned char, |
10194 | vector unsigned char); | |
b0b343db JJ |
10195 | vector bool short vec_mergeh (vector bool short, vector bool short); |
10196 | vector pixel vec_mergeh (vector pixel, vector pixel); | |
924fcc4e JM |
10197 | vector signed short vec_mergeh (vector signed short, |
10198 | vector signed short); | |
6e5bb5ad JM |
10199 | vector unsigned short vec_mergeh (vector unsigned short, |
10200 | vector unsigned short); | |
333c8841 | 10201 | vector float vec_mergeh (vector float, vector float); |
b0b343db | 10202 | vector bool int vec_mergeh (vector bool int, vector bool int); |
333c8841 | 10203 | vector signed int vec_mergeh (vector signed int, vector signed int); |
924fcc4e JM |
10204 | vector unsigned int vec_mergeh (vector unsigned int, |
10205 | vector unsigned int); | |
333c8841 | 10206 | |
b0b343db JJ |
10207 | vector float vec_vmrghw (vector float, vector float); |
10208 | vector bool int vec_vmrghw (vector bool int, vector bool int); | |
10209 | vector signed int vec_vmrghw (vector signed int, vector signed int); | |
10210 | vector unsigned int vec_vmrghw (vector unsigned int, | |
10211 | vector unsigned int); | |
10212 | ||
10213 | vector bool short vec_vmrghh (vector bool short, vector bool short); | |
10214 | vector signed short vec_vmrghh (vector signed short, | |
10215 | vector signed short); | |
10216 | vector unsigned short vec_vmrghh (vector unsigned short, | |
10217 | vector unsigned short); | |
10218 | vector pixel vec_vmrghh (vector pixel, vector pixel); | |
10219 | ||
10220 | vector bool char vec_vmrghb (vector bool char, vector bool char); | |
10221 | vector signed char vec_vmrghb (vector signed char, vector signed char); | |
10222 | vector unsigned char vec_vmrghb (vector unsigned char, | |
10223 | vector unsigned char); | |
10224 | ||
10225 | vector bool char vec_mergel (vector bool char, vector bool char); | |
333c8841 | 10226 | vector signed char vec_mergel (vector signed char, vector signed char); |
6e5bb5ad JM |
10227 | vector unsigned char vec_mergel (vector unsigned char, |
10228 | vector unsigned char); | |
b0b343db JJ |
10229 | vector bool short vec_mergel (vector bool short, vector bool short); |
10230 | vector pixel vec_mergel (vector pixel, vector pixel); | |
924fcc4e JM |
10231 | vector signed short vec_mergel (vector signed short, |
10232 | vector signed short); | |
6e5bb5ad JM |
10233 | vector unsigned short vec_mergel (vector unsigned short, |
10234 | vector unsigned short); | |
333c8841 | 10235 | vector float vec_mergel (vector float, vector float); |
b0b343db | 10236 | vector bool int vec_mergel (vector bool int, vector bool int); |
333c8841 | 10237 | vector signed int vec_mergel (vector signed int, vector signed int); |
924fcc4e JM |
10238 | vector unsigned int vec_mergel (vector unsigned int, |
10239 | vector unsigned int); | |
333c8841 | 10240 | |
b0b343db JJ |
10241 | vector float vec_vmrglw (vector float, vector float); |
10242 | vector signed int vec_vmrglw (vector signed int, vector signed int); | |
10243 | vector unsigned int vec_vmrglw (vector unsigned int, | |
10244 | vector unsigned int); | |
10245 | vector bool int vec_vmrglw (vector bool int, vector bool int); | |
333c8841 | 10246 | |
b0b343db JJ |
10247 | vector bool short vec_vmrglh (vector bool short, vector bool short); |
10248 | vector signed short vec_vmrglh (vector signed short, | |
10249 | vector signed short); | |
10250 | vector unsigned short vec_vmrglh (vector unsigned short, | |
10251 | vector unsigned short); | |
10252 | vector pixel vec_vmrglh (vector pixel, vector pixel); | |
10253 | ||
10254 | vector bool char vec_vmrglb (vector bool char, vector bool char); | |
10255 | vector signed char vec_vmrglb (vector signed char, vector signed char); | |
10256 | vector unsigned char vec_vmrglb (vector unsigned char, | |
10257 | vector unsigned char); | |
333c8841 | 10258 | |
b0b343db | 10259 | vector unsigned short vec_mfvscr (void); |
333c8841 | 10260 | |
b0b343db JJ |
10261 | vector unsigned char vec_min (vector bool char, vector unsigned char); |
10262 | vector unsigned char vec_min (vector unsigned char, vector bool char); | |
924fcc4e JM |
10263 | vector unsigned char vec_min (vector unsigned char, |
10264 | vector unsigned char); | |
b0b343db JJ |
10265 | vector signed char vec_min (vector bool char, vector signed char); |
10266 | vector signed char vec_min (vector signed char, vector bool char); | |
333c8841 | 10267 | vector signed char vec_min (vector signed char, vector signed char); |
b0b343db | 10268 | vector unsigned short vec_min (vector bool short, |
924fcc4e JM |
10269 | vector unsigned short); |
10270 | vector unsigned short vec_min (vector unsigned short, | |
b0b343db | 10271 | vector bool short); |
6e5bb5ad JM |
10272 | vector unsigned short vec_min (vector unsigned short, |
10273 | vector unsigned short); | |
b0b343db JJ |
10274 | vector signed short vec_min (vector bool short, vector signed short); |
10275 | vector signed short vec_min (vector signed short, vector bool short); | |
333c8841 | 10276 | vector signed short vec_min (vector signed short, vector signed short); |
b0b343db JJ |
10277 | vector unsigned int vec_min (vector bool int, vector unsigned int); |
10278 | vector unsigned int vec_min (vector unsigned int, vector bool int); | |
333c8841 | 10279 | vector unsigned int vec_min (vector unsigned int, vector unsigned int); |
b0b343db JJ |
10280 | vector signed int vec_min (vector bool int, vector signed int); |
10281 | vector signed int vec_min (vector signed int, vector bool int); | |
333c8841 AH |
10282 | vector signed int vec_min (vector signed int, vector signed int); |
10283 | vector float vec_min (vector float, vector float); | |
10284 | ||
b0b343db JJ |
10285 | vector float vec_vminfp (vector float, vector float); |
10286 | ||
10287 | vector signed int vec_vminsw (vector bool int, vector signed int); | |
10288 | vector signed int vec_vminsw (vector signed int, vector bool int); | |
10289 | vector signed int vec_vminsw (vector signed int, vector signed int); | |
10290 | ||
10291 | vector unsigned int vec_vminuw (vector bool int, vector unsigned int); | |
10292 | vector unsigned int vec_vminuw (vector unsigned int, vector bool int); | |
10293 | vector unsigned int vec_vminuw (vector unsigned int, | |
10294 | vector unsigned int); | |
10295 | ||
10296 | vector signed short vec_vminsh (vector bool short, vector signed short); | |
10297 | vector signed short vec_vminsh (vector signed short, vector bool short); | |
10298 | vector signed short vec_vminsh (vector signed short, | |
10299 | vector signed short); | |
10300 | ||
10301 | vector unsigned short vec_vminuh (vector bool short, | |
10302 | vector unsigned short); | |
10303 | vector unsigned short vec_vminuh (vector unsigned short, | |
10304 | vector bool short); | |
10305 | vector unsigned short vec_vminuh (vector unsigned short, | |
10306 | vector unsigned short); | |
10307 | ||
10308 | vector signed char vec_vminsb (vector bool char, vector signed char); | |
10309 | vector signed char vec_vminsb (vector signed char, vector bool char); | |
10310 | vector signed char vec_vminsb (vector signed char, vector signed char); | |
10311 | ||
10312 | vector unsigned char vec_vminub (vector bool char, | |
10313 | vector unsigned char); | |
10314 | vector unsigned char vec_vminub (vector unsigned char, | |
10315 | vector bool char); | |
10316 | vector unsigned char vec_vminub (vector unsigned char, | |
10317 | vector unsigned char); | |
10318 | ||
10319 | vector signed short vec_mladd (vector signed short, | |
10320 | vector signed short, | |
6e5bb5ad | 10321 | vector signed short); |
924fcc4e JM |
10322 | vector signed short vec_mladd (vector signed short, |
10323 | vector unsigned short, | |
6e5bb5ad | 10324 | vector unsigned short); |
924fcc4e JM |
10325 | vector signed short vec_mladd (vector unsigned short, |
10326 | vector signed short, | |
6e5bb5ad JM |
10327 | vector signed short); |
10328 | vector unsigned short vec_mladd (vector unsigned short, | |
10329 | vector unsigned short, | |
10330 | vector unsigned short); | |
10331 | ||
924fcc4e JM |
10332 | vector signed short vec_mradds (vector signed short, |
10333 | vector signed short, | |
6e5bb5ad JM |
10334 | vector signed short); |
10335 | ||
924fcc4e JM |
10336 | vector unsigned int vec_msum (vector unsigned char, |
10337 | vector unsigned char, | |
6e5bb5ad | 10338 | vector unsigned int); |
b0b343db JJ |
10339 | vector signed int vec_msum (vector signed char, |
10340 | vector unsigned char, | |
6e5bb5ad | 10341 | vector signed int); |
924fcc4e JM |
10342 | vector unsigned int vec_msum (vector unsigned short, |
10343 | vector unsigned short, | |
6e5bb5ad | 10344 | vector unsigned int); |
b0b343db JJ |
10345 | vector signed int vec_msum (vector signed short, |
10346 | vector signed short, | |
6e5bb5ad JM |
10347 | vector signed int); |
10348 | ||
b0b343db JJ |
10349 | vector signed int vec_vmsumshm (vector signed short, |
10350 | vector signed short, | |
10351 | vector signed int); | |
10352 | ||
10353 | vector unsigned int vec_vmsumuhm (vector unsigned short, | |
10354 | vector unsigned short, | |
10355 | vector unsigned int); | |
10356 | ||
10357 | vector signed int vec_vmsummbm (vector signed char, | |
10358 | vector unsigned char, | |
10359 | vector signed int); | |
10360 | ||
10361 | vector unsigned int vec_vmsumubm (vector unsigned char, | |
10362 | vector unsigned char, | |
10363 | vector unsigned int); | |
10364 | ||
6e5bb5ad | 10365 | vector unsigned int vec_msums (vector unsigned short, |
924fcc4e JM |
10366 | vector unsigned short, |
10367 | vector unsigned int); | |
b0b343db JJ |
10368 | vector signed int vec_msums (vector signed short, |
10369 | vector signed short, | |
6e5bb5ad | 10370 | vector signed int); |
333c8841 | 10371 | |
b0b343db JJ |
10372 | vector signed int vec_vmsumshs (vector signed short, |
10373 | vector signed short, | |
10374 | vector signed int); | |
10375 | ||
10376 | vector unsigned int vec_vmsumuhs (vector unsigned short, | |
10377 | vector unsigned short, | |
10378 | vector unsigned int); | |
10379 | ||
333c8841 AH |
10380 | void vec_mtvscr (vector signed int); |
10381 | void vec_mtvscr (vector unsigned int); | |
b0b343db | 10382 | void vec_mtvscr (vector bool int); |
333c8841 AH |
10383 | void vec_mtvscr (vector signed short); |
10384 | void vec_mtvscr (vector unsigned short); | |
b0b343db JJ |
10385 | void vec_mtvscr (vector bool short); |
10386 | void vec_mtvscr (vector pixel); | |
333c8841 AH |
10387 | void vec_mtvscr (vector signed char); |
10388 | void vec_mtvscr (vector unsigned char); | |
b0b343db | 10389 | void vec_mtvscr (vector bool char); |
333c8841 | 10390 | |
924fcc4e JM |
10391 | vector unsigned short vec_mule (vector unsigned char, |
10392 | vector unsigned char); | |
b0b343db JJ |
10393 | vector signed short vec_mule (vector signed char, |
10394 | vector signed char); | |
924fcc4e JM |
10395 | vector unsigned int vec_mule (vector unsigned short, |
10396 | vector unsigned short); | |
333c8841 AH |
10397 | vector signed int vec_mule (vector signed short, vector signed short); |
10398 | ||
b0b343db JJ |
10399 | vector signed int vec_vmulesh (vector signed short, |
10400 | vector signed short); | |
10401 | ||
10402 | vector unsigned int vec_vmuleuh (vector unsigned short, | |
10403 | vector unsigned short); | |
10404 | ||
10405 | vector signed short vec_vmulesb (vector signed char, | |
10406 | vector signed char); | |
10407 | ||
10408 | vector unsigned short vec_vmuleub (vector unsigned char, | |
10409 | vector unsigned char); | |
10410 | ||
924fcc4e JM |
10411 | vector unsigned short vec_mulo (vector unsigned char, |
10412 | vector unsigned char); | |
333c8841 | 10413 | vector signed short vec_mulo (vector signed char, vector signed char); |
924fcc4e JM |
10414 | vector unsigned int vec_mulo (vector unsigned short, |
10415 | vector unsigned short); | |
333c8841 AH |
10416 | vector signed int vec_mulo (vector signed short, vector signed short); |
10417 | ||
b0b343db JJ |
10418 | vector signed int vec_vmulosh (vector signed short, |
10419 | vector signed short); | |
10420 | ||
10421 | vector unsigned int vec_vmulouh (vector unsigned short, | |
10422 | vector unsigned short); | |
10423 | ||
10424 | vector signed short vec_vmulosb (vector signed char, | |
10425 | vector signed char); | |
10426 | ||
10427 | vector unsigned short vec_vmuloub (vector unsigned char, | |
10428 | vector unsigned char); | |
10429 | ||
333c8841 AH |
10430 | vector float vec_nmsub (vector float, vector float, vector float); |
10431 | ||
10432 | vector float vec_nor (vector float, vector float); | |
10433 | vector signed int vec_nor (vector signed int, vector signed int); | |
10434 | vector unsigned int vec_nor (vector unsigned int, vector unsigned int); | |
b0b343db | 10435 | vector bool int vec_nor (vector bool int, vector bool int); |
333c8841 | 10436 | vector signed short vec_nor (vector signed short, vector signed short); |
6e5bb5ad JM |
10437 | vector unsigned short vec_nor (vector unsigned short, |
10438 | vector unsigned short); | |
b0b343db | 10439 | vector bool short vec_nor (vector bool short, vector bool short); |
333c8841 | 10440 | vector signed char vec_nor (vector signed char, vector signed char); |
924fcc4e JM |
10441 | vector unsigned char vec_nor (vector unsigned char, |
10442 | vector unsigned char); | |
b0b343db | 10443 | vector bool char vec_nor (vector bool char, vector bool char); |
333c8841 AH |
10444 | |
10445 | vector float vec_or (vector float, vector float); | |
b0b343db JJ |
10446 | vector float vec_or (vector float, vector bool int); |
10447 | vector float vec_or (vector bool int, vector float); | |
10448 | vector bool int vec_or (vector bool int, vector bool int); | |
10449 | vector signed int vec_or (vector bool int, vector signed int); | |
10450 | vector signed int vec_or (vector signed int, vector bool int); | |
333c8841 | 10451 | vector signed int vec_or (vector signed int, vector signed int); |
b0b343db JJ |
10452 | vector unsigned int vec_or (vector bool int, vector unsigned int); |
10453 | vector unsigned int vec_or (vector unsigned int, vector bool int); | |
333c8841 | 10454 | vector unsigned int vec_or (vector unsigned int, vector unsigned int); |
b0b343db JJ |
10455 | vector bool short vec_or (vector bool short, vector bool short); |
10456 | vector signed short vec_or (vector bool short, vector signed short); | |
10457 | vector signed short vec_or (vector signed short, vector bool short); | |
333c8841 | 10458 | vector signed short vec_or (vector signed short, vector signed short); |
b0b343db JJ |
10459 | vector unsigned short vec_or (vector bool short, vector unsigned short); |
10460 | vector unsigned short vec_or (vector unsigned short, vector bool short); | |
924fcc4e JM |
10461 | vector unsigned short vec_or (vector unsigned short, |
10462 | vector unsigned short); | |
b0b343db JJ |
10463 | vector signed char vec_or (vector bool char, vector signed char); |
10464 | vector bool char vec_or (vector bool char, vector bool char); | |
10465 | vector signed char vec_or (vector signed char, vector bool char); | |
333c8841 | 10466 | vector signed char vec_or (vector signed char, vector signed char); |
b0b343db JJ |
10467 | vector unsigned char vec_or (vector bool char, vector unsigned char); |
10468 | vector unsigned char vec_or (vector unsigned char, vector bool char); | |
924fcc4e JM |
10469 | vector unsigned char vec_or (vector unsigned char, |
10470 | vector unsigned char); | |
333c8841 AH |
10471 | |
10472 | vector signed char vec_pack (vector signed short, vector signed short); | |
6e5bb5ad JM |
10473 | vector unsigned char vec_pack (vector unsigned short, |
10474 | vector unsigned short); | |
b0b343db | 10475 | vector bool char vec_pack (vector bool short, vector bool short); |
333c8841 | 10476 | vector signed short vec_pack (vector signed int, vector signed int); |
924fcc4e JM |
10477 | vector unsigned short vec_pack (vector unsigned int, |
10478 | vector unsigned int); | |
b0b343db | 10479 | vector bool short vec_pack (vector bool int, vector bool int); |
333c8841 | 10480 | |
b0b343db JJ |
10481 | vector bool short vec_vpkuwum (vector bool int, vector bool int); |
10482 | vector signed short vec_vpkuwum (vector signed int, vector signed int); | |
10483 | vector unsigned short vec_vpkuwum (vector unsigned int, | |
10484 | vector unsigned int); | |
10485 | ||
10486 | vector bool char vec_vpkuhum (vector bool short, vector bool short); | |
10487 | vector signed char vec_vpkuhum (vector signed short, | |
10488 | vector signed short); | |
10489 | vector unsigned char vec_vpkuhum (vector unsigned short, | |
10490 | vector unsigned short); | |
10491 | ||
10492 | vector pixel vec_packpx (vector unsigned int, vector unsigned int); | |
333c8841 | 10493 | |
6e5bb5ad JM |
10494 | vector unsigned char vec_packs (vector unsigned short, |
10495 | vector unsigned short); | |
333c8841 | 10496 | vector signed char vec_packs (vector signed short, vector signed short); |
924fcc4e JM |
10497 | vector unsigned short vec_packs (vector unsigned int, |
10498 | vector unsigned int); | |
333c8841 AH |
10499 | vector signed short vec_packs (vector signed int, vector signed int); |
10500 | ||
b0b343db JJ |
10501 | vector signed short vec_vpkswss (vector signed int, vector signed int); |
10502 | ||
10503 | vector unsigned short vec_vpkuwus (vector unsigned int, | |
10504 | vector unsigned int); | |
10505 | ||
10506 | vector signed char vec_vpkshss (vector signed short, | |
10507 | vector signed short); | |
10508 | ||
10509 | vector unsigned char vec_vpkuhus (vector unsigned short, | |
10510 | vector unsigned short); | |
10511 | ||
6e5bb5ad JM |
10512 | vector unsigned char vec_packsu (vector unsigned short, |
10513 | vector unsigned short); | |
924fcc4e JM |
10514 | vector unsigned char vec_packsu (vector signed short, |
10515 | vector signed short); | |
10516 | vector unsigned short vec_packsu (vector unsigned int, | |
10517 | vector unsigned int); | |
333c8841 AH |
10518 | vector unsigned short vec_packsu (vector signed int, vector signed int); |
10519 | ||
b0b343db JJ |
10520 | vector unsigned short vec_vpkswus (vector signed int, |
10521 | vector signed int); | |
10522 | ||
10523 | vector unsigned char vec_vpkshus (vector signed short, | |
10524 | vector signed short); | |
10525 | ||
10526 | vector float vec_perm (vector float, | |
10527 | vector float, | |
924fcc4e | 10528 | vector unsigned char); |
b0b343db JJ |
10529 | vector signed int vec_perm (vector signed int, |
10530 | vector signed int, | |
6e5bb5ad | 10531 | vector unsigned char); |
b0b343db JJ |
10532 | vector unsigned int vec_perm (vector unsigned int, |
10533 | vector unsigned int, | |
6e5bb5ad | 10534 | vector unsigned char); |
b0b343db JJ |
10535 | vector bool int vec_perm (vector bool int, |
10536 | vector bool int, | |
10537 | vector unsigned char); | |
10538 | vector signed short vec_perm (vector signed short, | |
10539 | vector signed short, | |
6e5bb5ad JM |
10540 | vector unsigned char); |
10541 | vector unsigned short vec_perm (vector unsigned short, | |
10542 | vector unsigned short, | |
10543 | vector unsigned char); | |
b0b343db JJ |
10544 | vector bool short vec_perm (vector bool short, |
10545 | vector bool short, | |
10546 | vector unsigned char); | |
10547 | vector pixel vec_perm (vector pixel, | |
10548 | vector pixel, | |
10549 | vector unsigned char); | |
10550 | vector signed char vec_perm (vector signed char, | |
10551 | vector signed char, | |
6e5bb5ad | 10552 | vector unsigned char); |
924fcc4e JM |
10553 | vector unsigned char vec_perm (vector unsigned char, |
10554 | vector unsigned char, | |
6e5bb5ad | 10555 | vector unsigned char); |
b0b343db JJ |
10556 | vector bool char vec_perm (vector bool char, |
10557 | vector bool char, | |
10558 | vector unsigned char); | |
333c8841 AH |
10559 | |
10560 | vector float vec_re (vector float); | |
10561 | ||
b0b343db JJ |
10562 | vector signed char vec_rl (vector signed char, |
10563 | vector unsigned char); | |
924fcc4e JM |
10564 | vector unsigned char vec_rl (vector unsigned char, |
10565 | vector unsigned char); | |
333c8841 | 10566 | vector signed short vec_rl (vector signed short, vector unsigned short); |
924fcc4e JM |
10567 | vector unsigned short vec_rl (vector unsigned short, |
10568 | vector unsigned short); | |
333c8841 AH |
10569 | vector signed int vec_rl (vector signed int, vector unsigned int); |
10570 | vector unsigned int vec_rl (vector unsigned int, vector unsigned int); | |
10571 | ||
b0b343db JJ |
10572 | vector signed int vec_vrlw (vector signed int, vector unsigned int); |
10573 | vector unsigned int vec_vrlw (vector unsigned int, vector unsigned int); | |
10574 | ||
10575 | vector signed short vec_vrlh (vector signed short, | |
10576 | vector unsigned short); | |
10577 | vector unsigned short vec_vrlh (vector unsigned short, | |
10578 | vector unsigned short); | |
10579 | ||
10580 | vector signed char vec_vrlb (vector signed char, vector unsigned char); | |
10581 | vector unsigned char vec_vrlb (vector unsigned char, | |
10582 | vector unsigned char); | |
10583 | ||
333c8841 AH |
10584 | vector float vec_round (vector float); |
10585 | ||
10586 | vector float vec_rsqrte (vector float); | |
10587 | ||
b0b343db | 10588 | vector float vec_sel (vector float, vector float, vector bool int); |
333c8841 | 10589 | vector float vec_sel (vector float, vector float, vector unsigned int); |
b0b343db JJ |
10590 | vector signed int vec_sel (vector signed int, |
10591 | vector signed int, | |
10592 | vector bool int); | |
10593 | vector signed int vec_sel (vector signed int, | |
10594 | vector signed int, | |
6e5bb5ad | 10595 | vector unsigned int); |
b0b343db JJ |
10596 | vector unsigned int vec_sel (vector unsigned int, |
10597 | vector unsigned int, | |
10598 | vector bool int); | |
10599 | vector unsigned int vec_sel (vector unsigned int, | |
10600 | vector unsigned int, | |
6e5bb5ad | 10601 | vector unsigned int); |
b0b343db JJ |
10602 | vector bool int vec_sel (vector bool int, |
10603 | vector bool int, | |
10604 | vector bool int); | |
10605 | vector bool int vec_sel (vector bool int, | |
10606 | vector bool int, | |
10607 | vector unsigned int); | |
10608 | vector signed short vec_sel (vector signed short, | |
10609 | vector signed short, | |
10610 | vector bool short); | |
10611 | vector signed short vec_sel (vector signed short, | |
10612 | vector signed short, | |
6e5bb5ad JM |
10613 | vector unsigned short); |
10614 | vector unsigned short vec_sel (vector unsigned short, | |
924fcc4e | 10615 | vector unsigned short, |
b0b343db | 10616 | vector bool short); |
6e5bb5ad JM |
10617 | vector unsigned short vec_sel (vector unsigned short, |
10618 | vector unsigned short, | |
10619 | vector unsigned short); | |
b0b343db JJ |
10620 | vector bool short vec_sel (vector bool short, |
10621 | vector bool short, | |
10622 | vector bool short); | |
10623 | vector bool short vec_sel (vector bool short, | |
10624 | vector bool short, | |
10625 | vector unsigned short); | |
10626 | vector signed char vec_sel (vector signed char, | |
10627 | vector signed char, | |
10628 | vector bool char); | |
10629 | vector signed char vec_sel (vector signed char, | |
10630 | vector signed char, | |
6e5bb5ad | 10631 | vector unsigned char); |
924fcc4e JM |
10632 | vector unsigned char vec_sel (vector unsigned char, |
10633 | vector unsigned char, | |
b0b343db | 10634 | vector bool char); |
924fcc4e JM |
10635 | vector unsigned char vec_sel (vector unsigned char, |
10636 | vector unsigned char, | |
6e5bb5ad | 10637 | vector unsigned char); |
b0b343db JJ |
10638 | vector bool char vec_sel (vector bool char, |
10639 | vector bool char, | |
10640 | vector bool char); | |
10641 | vector bool char vec_sel (vector bool char, | |
10642 | vector bool char, | |
10643 | vector unsigned char); | |
10644 | ||
10645 | vector signed char vec_sl (vector signed char, | |
10646 | vector unsigned char); | |
924fcc4e JM |
10647 | vector unsigned char vec_sl (vector unsigned char, |
10648 | vector unsigned char); | |
333c8841 | 10649 | vector signed short vec_sl (vector signed short, vector unsigned short); |
924fcc4e JM |
10650 | vector unsigned short vec_sl (vector unsigned short, |
10651 | vector unsigned short); | |
333c8841 AH |
10652 | vector signed int vec_sl (vector signed int, vector unsigned int); |
10653 | vector unsigned int vec_sl (vector unsigned int, vector unsigned int); | |
10654 | ||
b0b343db JJ |
10655 | vector signed int vec_vslw (vector signed int, vector unsigned int); |
10656 | vector unsigned int vec_vslw (vector unsigned int, vector unsigned int); | |
10657 | ||
10658 | vector signed short vec_vslh (vector signed short, | |
10659 | vector unsigned short); | |
10660 | vector unsigned short vec_vslh (vector unsigned short, | |
10661 | vector unsigned short); | |
10662 | ||
10663 | vector signed char vec_vslb (vector signed char, vector unsigned char); | |
10664 | vector unsigned char vec_vslb (vector unsigned char, | |
10665 | vector unsigned char); | |
10666 | ||
10667 | vector float vec_sld (vector float, vector float, const int); | |
10668 | vector signed int vec_sld (vector signed int, | |
10669 | vector signed int, | |
10670 | const int); | |
10671 | vector unsigned int vec_sld (vector unsigned int, | |
10672 | vector unsigned int, | |
10673 | const int); | |
10674 | vector bool int vec_sld (vector bool int, | |
10675 | vector bool int, | |
10676 | const int); | |
10677 | vector signed short vec_sld (vector signed short, | |
10678 | vector signed short, | |
10679 | const int); | |
6e5bb5ad | 10680 | vector unsigned short vec_sld (vector unsigned short, |
b0b343db JJ |
10681 | vector unsigned short, |
10682 | const int); | |
10683 | vector bool short vec_sld (vector bool short, | |
10684 | vector bool short, | |
10685 | const int); | |
10686 | vector pixel vec_sld (vector pixel, | |
10687 | vector pixel, | |
10688 | const int); | |
10689 | vector signed char vec_sld (vector signed char, | |
10690 | vector signed char, | |
10691 | const int); | |
924fcc4e JM |
10692 | vector unsigned char vec_sld (vector unsigned char, |
10693 | vector unsigned char, | |
b0b343db JJ |
10694 | const int); |
10695 | vector bool char vec_sld (vector bool char, | |
10696 | vector bool char, | |
10697 | const int); | |
333c8841 | 10698 | |
b0b343db JJ |
10699 | vector signed int vec_sll (vector signed int, |
10700 | vector unsigned int); | |
10701 | vector signed int vec_sll (vector signed int, | |
10702 | vector unsigned short); | |
10703 | vector signed int vec_sll (vector signed int, | |
10704 | vector unsigned char); | |
10705 | vector unsigned int vec_sll (vector unsigned int, | |
10706 | vector unsigned int); | |
924fcc4e JM |
10707 | vector unsigned int vec_sll (vector unsigned int, |
10708 | vector unsigned short); | |
b0b343db JJ |
10709 | vector unsigned int vec_sll (vector unsigned int, |
10710 | vector unsigned char); | |
10711 | vector bool int vec_sll (vector bool int, | |
10712 | vector unsigned int); | |
10713 | vector bool int vec_sll (vector bool int, | |
10714 | vector unsigned short); | |
10715 | vector bool int vec_sll (vector bool int, | |
10716 | vector unsigned char); | |
10717 | vector signed short vec_sll (vector signed short, | |
10718 | vector unsigned int); | |
924fcc4e JM |
10719 | vector signed short vec_sll (vector signed short, |
10720 | vector unsigned short); | |
b0b343db JJ |
10721 | vector signed short vec_sll (vector signed short, |
10722 | vector unsigned char); | |
924fcc4e JM |
10723 | vector unsigned short vec_sll (vector unsigned short, |
10724 | vector unsigned int); | |
6e5bb5ad JM |
10725 | vector unsigned short vec_sll (vector unsigned short, |
10726 | vector unsigned short); | |
924fcc4e JM |
10727 | vector unsigned short vec_sll (vector unsigned short, |
10728 | vector unsigned char); | |
b0b343db JJ |
10729 | vector bool short vec_sll (vector bool short, vector unsigned int); |
10730 | vector bool short vec_sll (vector bool short, vector unsigned short); | |
10731 | vector bool short vec_sll (vector bool short, vector unsigned char); | |
10732 | vector pixel vec_sll (vector pixel, vector unsigned int); | |
10733 | vector pixel vec_sll (vector pixel, vector unsigned short); | |
10734 | vector pixel vec_sll (vector pixel, vector unsigned char); | |
333c8841 AH |
10735 | vector signed char vec_sll (vector signed char, vector unsigned int); |
10736 | vector signed char vec_sll (vector signed char, vector unsigned short); | |
10737 | vector signed char vec_sll (vector signed char, vector unsigned char); | |
924fcc4e JM |
10738 | vector unsigned char vec_sll (vector unsigned char, |
10739 | vector unsigned int); | |
10740 | vector unsigned char vec_sll (vector unsigned char, | |
10741 | vector unsigned short); | |
10742 | vector unsigned char vec_sll (vector unsigned char, | |
10743 | vector unsigned char); | |
b0b343db JJ |
10744 | vector bool char vec_sll (vector bool char, vector unsigned int); |
10745 | vector bool char vec_sll (vector bool char, vector unsigned short); | |
10746 | vector bool char vec_sll (vector bool char, vector unsigned char); | |
333c8841 AH |
10747 | |
10748 | vector float vec_slo (vector float, vector signed char); | |
10749 | vector float vec_slo (vector float, vector unsigned char); | |
10750 | vector signed int vec_slo (vector signed int, vector signed char); | |
10751 | vector signed int vec_slo (vector signed int, vector unsigned char); | |
10752 | vector unsigned int vec_slo (vector unsigned int, vector signed char); | |
10753 | vector unsigned int vec_slo (vector unsigned int, vector unsigned char); | |
333c8841 AH |
10754 | vector signed short vec_slo (vector signed short, vector signed char); |
10755 | vector signed short vec_slo (vector signed short, vector unsigned char); | |
924fcc4e JM |
10756 | vector unsigned short vec_slo (vector unsigned short, |
10757 | vector signed char); | |
10758 | vector unsigned short vec_slo (vector unsigned short, | |
10759 | vector unsigned char); | |
b0b343db JJ |
10760 | vector pixel vec_slo (vector pixel, vector signed char); |
10761 | vector pixel vec_slo (vector pixel, vector unsigned char); | |
333c8841 AH |
10762 | vector signed char vec_slo (vector signed char, vector signed char); |
10763 | vector signed char vec_slo (vector signed char, vector unsigned char); | |
10764 | vector unsigned char vec_slo (vector unsigned char, vector signed char); | |
924fcc4e JM |
10765 | vector unsigned char vec_slo (vector unsigned char, |
10766 | vector unsigned char); | |
333c8841 | 10767 | |
b0b343db JJ |
10768 | vector signed char vec_splat (vector signed char, const int); |
10769 | vector unsigned char vec_splat (vector unsigned char, const int); | |
10770 | vector bool char vec_splat (vector bool char, const int); | |
10771 | vector signed short vec_splat (vector signed short, const int); | |
10772 | vector unsigned short vec_splat (vector unsigned short, const int); | |
10773 | vector bool short vec_splat (vector bool short, const int); | |
10774 | vector pixel vec_splat (vector pixel, const int); | |
10775 | vector float vec_splat (vector float, const int); | |
10776 | vector signed int vec_splat (vector signed int, const int); | |
10777 | vector unsigned int vec_splat (vector unsigned int, const int); | |
10778 | vector bool int vec_splat (vector bool int, const int); | |
10779 | ||
10780 | vector float vec_vspltw (vector float, const int); | |
10781 | vector signed int vec_vspltw (vector signed int, const int); | |
10782 | vector unsigned int vec_vspltw (vector unsigned int, const int); | |
10783 | vector bool int vec_vspltw (vector bool int, const int); | |
10784 | ||
10785 | vector bool short vec_vsplth (vector bool short, const int); | |
10786 | vector signed short vec_vsplth (vector signed short, const int); | |
10787 | vector unsigned short vec_vsplth (vector unsigned short, const int); | |
10788 | vector pixel vec_vsplth (vector pixel, const int); | |
10789 | ||
10790 | vector signed char vec_vspltb (vector signed char, const int); | |
10791 | vector unsigned char vec_vspltb (vector unsigned char, const int); | |
10792 | vector bool char vec_vspltb (vector bool char, const int); | |
333c8841 | 10793 | |
b0b343db | 10794 | vector signed char vec_splat_s8 (const int); |
333c8841 | 10795 | |
b0b343db | 10796 | vector signed short vec_splat_s16 (const int); |
333c8841 | 10797 | |
b0b343db | 10798 | vector signed int vec_splat_s32 (const int); |
333c8841 | 10799 | |
b0b343db | 10800 | vector unsigned char vec_splat_u8 (const int); |
333c8841 | 10801 | |
b0b343db | 10802 | vector unsigned short vec_splat_u16 (const int); |
333c8841 | 10803 | |
b0b343db | 10804 | vector unsigned int vec_splat_u32 (const int); |
333c8841 AH |
10805 | |
10806 | vector signed char vec_sr (vector signed char, vector unsigned char); | |
924fcc4e JM |
10807 | vector unsigned char vec_sr (vector unsigned char, |
10808 | vector unsigned char); | |
b0b343db JJ |
10809 | vector signed short vec_sr (vector signed short, |
10810 | vector unsigned short); | |
924fcc4e JM |
10811 | vector unsigned short vec_sr (vector unsigned short, |
10812 | vector unsigned short); | |
333c8841 AH |
10813 | vector signed int vec_sr (vector signed int, vector unsigned int); |
10814 | vector unsigned int vec_sr (vector unsigned int, vector unsigned int); | |
10815 | ||
b0b343db JJ |
10816 | vector signed int vec_vsrw (vector signed int, vector unsigned int); |
10817 | vector unsigned int vec_vsrw (vector unsigned int, vector unsigned int); | |
10818 | ||
10819 | vector signed short vec_vsrh (vector signed short, | |
10820 | vector unsigned short); | |
10821 | vector unsigned short vec_vsrh (vector unsigned short, | |
10822 | vector unsigned short); | |
10823 | ||
10824 | vector signed char vec_vsrb (vector signed char, vector unsigned char); | |
10825 | vector unsigned char vec_vsrb (vector unsigned char, | |
10826 | vector unsigned char); | |
10827 | ||
333c8841 | 10828 | vector signed char vec_sra (vector signed char, vector unsigned char); |
924fcc4e JM |
10829 | vector unsigned char vec_sra (vector unsigned char, |
10830 | vector unsigned char); | |
10831 | vector signed short vec_sra (vector signed short, | |
10832 | vector unsigned short); | |
6e5bb5ad JM |
10833 | vector unsigned short vec_sra (vector unsigned short, |
10834 | vector unsigned short); | |
333c8841 AH |
10835 | vector signed int vec_sra (vector signed int, vector unsigned int); |
10836 | vector unsigned int vec_sra (vector unsigned int, vector unsigned int); | |
10837 | ||
b0b343db JJ |
10838 | vector signed int vec_vsraw (vector signed int, vector unsigned int); |
10839 | vector unsigned int vec_vsraw (vector unsigned int, | |
10840 | vector unsigned int); | |
10841 | ||
10842 | vector signed short vec_vsrah (vector signed short, | |
10843 | vector unsigned short); | |
10844 | vector unsigned short vec_vsrah (vector unsigned short, | |
10845 | vector unsigned short); | |
10846 | ||
10847 | vector signed char vec_vsrab (vector signed char, vector unsigned char); | |
10848 | vector unsigned char vec_vsrab (vector unsigned char, | |
10849 | vector unsigned char); | |
10850 | ||
333c8841 AH |
10851 | vector signed int vec_srl (vector signed int, vector unsigned int); |
10852 | vector signed int vec_srl (vector signed int, vector unsigned short); | |
10853 | vector signed int vec_srl (vector signed int, vector unsigned char); | |
10854 | vector unsigned int vec_srl (vector unsigned int, vector unsigned int); | |
924fcc4e JM |
10855 | vector unsigned int vec_srl (vector unsigned int, |
10856 | vector unsigned short); | |
333c8841 | 10857 | vector unsigned int vec_srl (vector unsigned int, vector unsigned char); |
b0b343db JJ |
10858 | vector bool int vec_srl (vector bool int, vector unsigned int); |
10859 | vector bool int vec_srl (vector bool int, vector unsigned short); | |
10860 | vector bool int vec_srl (vector bool int, vector unsigned char); | |
333c8841 | 10861 | vector signed short vec_srl (vector signed short, vector unsigned int); |
924fcc4e JM |
10862 | vector signed short vec_srl (vector signed short, |
10863 | vector unsigned short); | |
333c8841 | 10864 | vector signed short vec_srl (vector signed short, vector unsigned char); |
924fcc4e JM |
10865 | vector unsigned short vec_srl (vector unsigned short, |
10866 | vector unsigned int); | |
6e5bb5ad JM |
10867 | vector unsigned short vec_srl (vector unsigned short, |
10868 | vector unsigned short); | |
924fcc4e JM |
10869 | vector unsigned short vec_srl (vector unsigned short, |
10870 | vector unsigned char); | |
b0b343db JJ |
10871 | vector bool short vec_srl (vector bool short, vector unsigned int); |
10872 | vector bool short vec_srl (vector bool short, vector unsigned short); | |
10873 | vector bool short vec_srl (vector bool short, vector unsigned char); | |
10874 | vector pixel vec_srl (vector pixel, vector unsigned int); | |
10875 | vector pixel vec_srl (vector pixel, vector unsigned short); | |
10876 | vector pixel vec_srl (vector pixel, vector unsigned char); | |
333c8841 AH |
10877 | vector signed char vec_srl (vector signed char, vector unsigned int); |
10878 | vector signed char vec_srl (vector signed char, vector unsigned short); | |
10879 | vector signed char vec_srl (vector signed char, vector unsigned char); | |
924fcc4e JM |
10880 | vector unsigned char vec_srl (vector unsigned char, |
10881 | vector unsigned int); | |
10882 | vector unsigned char vec_srl (vector unsigned char, | |
10883 | vector unsigned short); | |
10884 | vector unsigned char vec_srl (vector unsigned char, | |
10885 | vector unsigned char); | |
b0b343db JJ |
10886 | vector bool char vec_srl (vector bool char, vector unsigned int); |
10887 | vector bool char vec_srl (vector bool char, vector unsigned short); | |
10888 | vector bool char vec_srl (vector bool char, vector unsigned char); | |
333c8841 AH |
10889 | |
10890 | vector float vec_sro (vector float, vector signed char); | |
10891 | vector float vec_sro (vector float, vector unsigned char); | |
10892 | vector signed int vec_sro (vector signed int, vector signed char); | |
10893 | vector signed int vec_sro (vector signed int, vector unsigned char); | |
10894 | vector unsigned int vec_sro (vector unsigned int, vector signed char); | |
10895 | vector unsigned int vec_sro (vector unsigned int, vector unsigned char); | |
333c8841 AH |
10896 | vector signed short vec_sro (vector signed short, vector signed char); |
10897 | vector signed short vec_sro (vector signed short, vector unsigned char); | |
924fcc4e JM |
10898 | vector unsigned short vec_sro (vector unsigned short, |
10899 | vector signed char); | |
10900 | vector unsigned short vec_sro (vector unsigned short, | |
10901 | vector unsigned char); | |
b0b343db JJ |
10902 | vector pixel vec_sro (vector pixel, vector signed char); |
10903 | vector pixel vec_sro (vector pixel, vector unsigned char); | |
333c8841 AH |
10904 | vector signed char vec_sro (vector signed char, vector signed char); |
10905 | vector signed char vec_sro (vector signed char, vector unsigned char); | |
10906 | vector unsigned char vec_sro (vector unsigned char, vector signed char); | |
924fcc4e JM |
10907 | vector unsigned char vec_sro (vector unsigned char, |
10908 | vector unsigned char); | |
333c8841 | 10909 | |
333c8841 | 10910 | void vec_st (vector float, int, vector float *); |
b0b343db JJ |
10911 | void vec_st (vector float, int, float *); |
10912 | void vec_st (vector signed int, int, vector signed int *); | |
333c8841 | 10913 | void vec_st (vector signed int, int, int *); |
333c8841 | 10914 | void vec_st (vector unsigned int, int, vector unsigned int *); |
b0b343db JJ |
10915 | void vec_st (vector unsigned int, int, unsigned int *); |
10916 | void vec_st (vector bool int, int, vector bool int *); | |
10917 | void vec_st (vector bool int, int, unsigned int *); | |
10918 | void vec_st (vector bool int, int, int *); | |
333c8841 | 10919 | void vec_st (vector signed short, int, vector signed short *); |
b0b343db | 10920 | void vec_st (vector signed short, int, short *); |
333c8841 | 10921 | void vec_st (vector unsigned short, int, vector unsigned short *); |
b0b343db JJ |
10922 | void vec_st (vector unsigned short, int, unsigned short *); |
10923 | void vec_st (vector bool short, int, vector bool short *); | |
10924 | void vec_st (vector bool short, int, unsigned short *); | |
10925 | void vec_st (vector pixel, int, vector pixel *); | |
10926 | void vec_st (vector pixel, int, unsigned short *); | |
10927 | void vec_st (vector pixel, int, short *); | |
10928 | void vec_st (vector bool short, int, short *); | |
333c8841 | 10929 | void vec_st (vector signed char, int, vector signed char *); |
b0b343db | 10930 | void vec_st (vector signed char, int, signed char *); |
333c8841 | 10931 | void vec_st (vector unsigned char, int, vector unsigned char *); |
b0b343db JJ |
10932 | void vec_st (vector unsigned char, int, unsigned char *); |
10933 | void vec_st (vector bool char, int, vector bool char *); | |
10934 | void vec_st (vector bool char, int, unsigned char *); | |
10935 | void vec_st (vector bool char, int, signed char *); | |
333c8841 | 10936 | |
333c8841 AH |
10937 | void vec_ste (vector signed char, int, signed char *); |
10938 | void vec_ste (vector unsigned char, int, unsigned char *); | |
b0b343db JJ |
10939 | void vec_ste (vector bool char, int, signed char *); |
10940 | void vec_ste (vector bool char, int, unsigned char *); | |
333c8841 | 10941 | void vec_ste (vector signed short, int, short *); |
b0b343db JJ |
10942 | void vec_ste (vector unsigned short, int, unsigned short *); |
10943 | void vec_ste (vector bool short, int, short *); | |
10944 | void vec_ste (vector bool short, int, unsigned short *); | |
10945 | void vec_ste (vector pixel, int, short *); | |
10946 | void vec_ste (vector pixel, int, unsigned short *); | |
10947 | void vec_ste (vector float, int, float *); | |
333c8841 AH |
10948 | void vec_ste (vector signed int, int, int *); |
10949 | void vec_ste (vector unsigned int, int, unsigned int *); | |
b0b343db JJ |
10950 | void vec_ste (vector bool int, int, int *); |
10951 | void vec_ste (vector bool int, int, unsigned int *); | |
10952 | ||
10953 | void vec_stvewx (vector float, int, float *); | |
10954 | void vec_stvewx (vector signed int, int, int *); | |
10955 | void vec_stvewx (vector unsigned int, int, unsigned int *); | |
10956 | void vec_stvewx (vector bool int, int, int *); | |
10957 | void vec_stvewx (vector bool int, int, unsigned int *); | |
10958 | ||
10959 | void vec_stvehx (vector signed short, int, short *); | |
10960 | void vec_stvehx (vector unsigned short, int, unsigned short *); | |
10961 | void vec_stvehx (vector bool short, int, short *); | |
10962 | void vec_stvehx (vector bool short, int, unsigned short *); | |
10963 | void vec_stvehx (vector pixel, int, short *); | |
10964 | void vec_stvehx (vector pixel, int, unsigned short *); | |
10965 | ||
10966 | void vec_stvebx (vector signed char, int, signed char *); | |
10967 | void vec_stvebx (vector unsigned char, int, unsigned char *); | |
10968 | void vec_stvebx (vector bool char, int, signed char *); | |
10969 | void vec_stvebx (vector bool char, int, unsigned char *); | |
333c8841 AH |
10970 | |
10971 | void vec_stl (vector float, int, vector float *); | |
10972 | void vec_stl (vector float, int, float *); | |
10973 | void vec_stl (vector signed int, int, vector signed int *); | |
10974 | void vec_stl (vector signed int, int, int *); | |
333c8841 AH |
10975 | void vec_stl (vector unsigned int, int, vector unsigned int *); |
10976 | void vec_stl (vector unsigned int, int, unsigned int *); | |
b0b343db JJ |
10977 | void vec_stl (vector bool int, int, vector bool int *); |
10978 | void vec_stl (vector bool int, int, unsigned int *); | |
10979 | void vec_stl (vector bool int, int, int *); | |
333c8841 | 10980 | void vec_stl (vector signed short, int, vector signed short *); |
b0b343db JJ |
10981 | void vec_stl (vector signed short, int, short *); |
10982 | void vec_stl (vector unsigned short, int, vector unsigned short *); | |
333c8841 | 10983 | void vec_stl (vector unsigned short, int, unsigned short *); |
b0b343db JJ |
10984 | void vec_stl (vector bool short, int, vector bool short *); |
10985 | void vec_stl (vector bool short, int, unsigned short *); | |
10986 | void vec_stl (vector bool short, int, short *); | |
10987 | void vec_stl (vector pixel, int, vector pixel *); | |
10988 | void vec_stl (vector pixel, int, unsigned short *); | |
10989 | void vec_stl (vector pixel, int, short *); | |
333c8841 | 10990 | void vec_stl (vector signed char, int, vector signed char *); |
b0b343db | 10991 | void vec_stl (vector signed char, int, signed char *); |
333c8841 | 10992 | void vec_stl (vector unsigned char, int, vector unsigned char *); |
b0b343db JJ |
10993 | void vec_stl (vector unsigned char, int, unsigned char *); |
10994 | void vec_stl (vector bool char, int, vector bool char *); | |
10995 | void vec_stl (vector bool char, int, unsigned char *); | |
10996 | void vec_stl (vector bool char, int, signed char *); | |
333c8841 | 10997 | |
b0b343db JJ |
10998 | vector signed char vec_sub (vector bool char, vector signed char); |
10999 | vector signed char vec_sub (vector signed char, vector bool char); | |
333c8841 | 11000 | vector signed char vec_sub (vector signed char, vector signed char); |
b0b343db JJ |
11001 | vector unsigned char vec_sub (vector bool char, vector unsigned char); |
11002 | vector unsigned char vec_sub (vector unsigned char, vector bool char); | |
924fcc4e JM |
11003 | vector unsigned char vec_sub (vector unsigned char, |
11004 | vector unsigned char); | |
b0b343db JJ |
11005 | vector signed short vec_sub (vector bool short, vector signed short); |
11006 | vector signed short vec_sub (vector signed short, vector bool short); | |
333c8841 | 11007 | vector signed short vec_sub (vector signed short, vector signed short); |
b0b343db | 11008 | vector unsigned short vec_sub (vector bool short, |
924fcc4e JM |
11009 | vector unsigned short); |
11010 | vector unsigned short vec_sub (vector unsigned short, | |
b0b343db | 11011 | vector bool short); |
6e5bb5ad JM |
11012 | vector unsigned short vec_sub (vector unsigned short, |
11013 | vector unsigned short); | |
b0b343db JJ |
11014 | vector signed int vec_sub (vector bool int, vector signed int); |
11015 | vector signed int vec_sub (vector signed int, vector bool int); | |
333c8841 | 11016 | vector signed int vec_sub (vector signed int, vector signed int); |
b0b343db JJ |
11017 | vector unsigned int vec_sub (vector bool int, vector unsigned int); |
11018 | vector unsigned int vec_sub (vector unsigned int, vector bool int); | |
333c8841 AH |
11019 | vector unsigned int vec_sub (vector unsigned int, vector unsigned int); |
11020 | vector float vec_sub (vector float, vector float); | |
11021 | ||
b0b343db JJ |
11022 | vector float vec_vsubfp (vector float, vector float); |
11023 | ||
11024 | vector signed int vec_vsubuwm (vector bool int, vector signed int); | |
11025 | vector signed int vec_vsubuwm (vector signed int, vector bool int); | |
11026 | vector signed int vec_vsubuwm (vector signed int, vector signed int); | |
11027 | vector unsigned int vec_vsubuwm (vector bool int, vector unsigned int); | |
11028 | vector unsigned int vec_vsubuwm (vector unsigned int, vector bool int); | |
11029 | vector unsigned int vec_vsubuwm (vector unsigned int, | |
11030 | vector unsigned int); | |
11031 | ||
11032 | vector signed short vec_vsubuhm (vector bool short, | |
11033 | vector signed short); | |
11034 | vector signed short vec_vsubuhm (vector signed short, | |
11035 | vector bool short); | |
11036 | vector signed short vec_vsubuhm (vector signed short, | |
11037 | vector signed short); | |
11038 | vector unsigned short vec_vsubuhm (vector bool short, | |
11039 | vector unsigned short); | |
11040 | vector unsigned short vec_vsubuhm (vector unsigned short, | |
11041 | vector bool short); | |
11042 | vector unsigned short vec_vsubuhm (vector unsigned short, | |
11043 | vector unsigned short); | |
11044 | ||
11045 | vector signed char vec_vsububm (vector bool char, vector signed char); | |
11046 | vector signed char vec_vsububm (vector signed char, vector bool char); | |
11047 | vector signed char vec_vsububm (vector signed char, vector signed char); | |
11048 | vector unsigned char vec_vsububm (vector bool char, | |
11049 | vector unsigned char); | |
11050 | vector unsigned char vec_vsububm (vector unsigned char, | |
11051 | vector bool char); | |
11052 | vector unsigned char vec_vsububm (vector unsigned char, | |
11053 | vector unsigned char); | |
11054 | ||
333c8841 AH |
11055 | vector unsigned int vec_subc (vector unsigned int, vector unsigned int); |
11056 | ||
b0b343db JJ |
11057 | vector unsigned char vec_subs (vector bool char, vector unsigned char); |
11058 | vector unsigned char vec_subs (vector unsigned char, vector bool char); | |
924fcc4e JM |
11059 | vector unsigned char vec_subs (vector unsigned char, |
11060 | vector unsigned char); | |
b0b343db JJ |
11061 | vector signed char vec_subs (vector bool char, vector signed char); |
11062 | vector signed char vec_subs (vector signed char, vector bool char); | |
333c8841 | 11063 | vector signed char vec_subs (vector signed char, vector signed char); |
b0b343db | 11064 | vector unsigned short vec_subs (vector bool short, |
924fcc4e JM |
11065 | vector unsigned short); |
11066 | vector unsigned short vec_subs (vector unsigned short, | |
b0b343db | 11067 | vector bool short); |
6e5bb5ad JM |
11068 | vector unsigned short vec_subs (vector unsigned short, |
11069 | vector unsigned short); | |
b0b343db JJ |
11070 | vector signed short vec_subs (vector bool short, vector signed short); |
11071 | vector signed short vec_subs (vector signed short, vector bool short); | |
333c8841 | 11072 | vector signed short vec_subs (vector signed short, vector signed short); |
b0b343db JJ |
11073 | vector unsigned int vec_subs (vector bool int, vector unsigned int); |
11074 | vector unsigned int vec_subs (vector unsigned int, vector bool int); | |
333c8841 | 11075 | vector unsigned int vec_subs (vector unsigned int, vector unsigned int); |
b0b343db JJ |
11076 | vector signed int vec_subs (vector bool int, vector signed int); |
11077 | vector signed int vec_subs (vector signed int, vector bool int); | |
333c8841 AH |
11078 | vector signed int vec_subs (vector signed int, vector signed int); |
11079 | ||
b0b343db JJ |
11080 | vector signed int vec_vsubsws (vector bool int, vector signed int); |
11081 | vector signed int vec_vsubsws (vector signed int, vector bool int); | |
11082 | vector signed int vec_vsubsws (vector signed int, vector signed int); | |
11083 | ||
11084 | vector unsigned int vec_vsubuws (vector bool int, vector unsigned int); | |
11085 | vector unsigned int vec_vsubuws (vector unsigned int, vector bool int); | |
11086 | vector unsigned int vec_vsubuws (vector unsigned int, | |
11087 | vector unsigned int); | |
11088 | ||
11089 | vector signed short vec_vsubshs (vector bool short, | |
11090 | vector signed short); | |
11091 | vector signed short vec_vsubshs (vector signed short, | |
11092 | vector bool short); | |
11093 | vector signed short vec_vsubshs (vector signed short, | |
11094 | vector signed short); | |
11095 | ||
11096 | vector unsigned short vec_vsubuhs (vector bool short, | |
11097 | vector unsigned short); | |
11098 | vector unsigned short vec_vsubuhs (vector unsigned short, | |
11099 | vector bool short); | |
11100 | vector unsigned short vec_vsubuhs (vector unsigned short, | |
11101 | vector unsigned short); | |
11102 | ||
11103 | vector signed char vec_vsubsbs (vector bool char, vector signed char); | |
11104 | vector signed char vec_vsubsbs (vector signed char, vector bool char); | |
11105 | vector signed char vec_vsubsbs (vector signed char, vector signed char); | |
11106 | ||
11107 | vector unsigned char vec_vsububs (vector bool char, | |
11108 | vector unsigned char); | |
11109 | vector unsigned char vec_vsububs (vector unsigned char, | |
11110 | vector bool char); | |
11111 | vector unsigned char vec_vsububs (vector unsigned char, | |
11112 | vector unsigned char); | |
11113 | ||
924fcc4e JM |
11114 | vector unsigned int vec_sum4s (vector unsigned char, |
11115 | vector unsigned int); | |
333c8841 AH |
11116 | vector signed int vec_sum4s (vector signed char, vector signed int); |
11117 | vector signed int vec_sum4s (vector signed short, vector signed int); | |
11118 | ||
b0b343db JJ |
11119 | vector signed int vec_vsum4shs (vector signed short, vector signed int); |
11120 | ||
11121 | vector signed int vec_vsum4sbs (vector signed char, vector signed int); | |
11122 | ||
11123 | vector unsigned int vec_vsum4ubs (vector unsigned char, | |
11124 | vector unsigned int); | |
11125 | ||
333c8841 AH |
11126 | vector signed int vec_sum2s (vector signed int, vector signed int); |
11127 | ||
11128 | vector signed int vec_sums (vector signed int, vector signed int); | |
11129 | ||
11130 | vector float vec_trunc (vector float); | |
11131 | ||
11132 | vector signed short vec_unpackh (vector signed char); | |
b0b343db | 11133 | vector bool short vec_unpackh (vector bool char); |
333c8841 | 11134 | vector signed int vec_unpackh (vector signed short); |
b0b343db JJ |
11135 | vector bool int vec_unpackh (vector bool short); |
11136 | vector unsigned int vec_unpackh (vector pixel); | |
11137 | ||
11138 | vector bool int vec_vupkhsh (vector bool short); | |
11139 | vector signed int vec_vupkhsh (vector signed short); | |
11140 | ||
11141 | vector unsigned int vec_vupkhpx (vector pixel); | |
11142 | ||
11143 | vector bool short vec_vupkhsb (vector bool char); | |
11144 | vector signed short vec_vupkhsb (vector signed char); | |
333c8841 AH |
11145 | |
11146 | vector signed short vec_unpackl (vector signed char); | |
b0b343db JJ |
11147 | vector bool short vec_unpackl (vector bool char); |
11148 | vector unsigned int vec_unpackl (vector pixel); | |
333c8841 | 11149 | vector signed int vec_unpackl (vector signed short); |
b0b343db JJ |
11150 | vector bool int vec_unpackl (vector bool short); |
11151 | ||
11152 | vector unsigned int vec_vupklpx (vector pixel); | |
11153 | ||
11154 | vector bool int vec_vupklsh (vector bool short); | |
11155 | vector signed int vec_vupklsh (vector signed short); | |
11156 | ||
11157 | vector bool short vec_vupklsb (vector bool char); | |
11158 | vector signed short vec_vupklsb (vector signed char); | |
333c8841 AH |
11159 | |
11160 | vector float vec_xor (vector float, vector float); | |
b0b343db JJ |
11161 | vector float vec_xor (vector float, vector bool int); |
11162 | vector float vec_xor (vector bool int, vector float); | |
11163 | vector bool int vec_xor (vector bool int, vector bool int); | |
11164 | vector signed int vec_xor (vector bool int, vector signed int); | |
11165 | vector signed int vec_xor (vector signed int, vector bool int); | |
333c8841 | 11166 | vector signed int vec_xor (vector signed int, vector signed int); |
b0b343db JJ |
11167 | vector unsigned int vec_xor (vector bool int, vector unsigned int); |
11168 | vector unsigned int vec_xor (vector unsigned int, vector bool int); | |
333c8841 | 11169 | vector unsigned int vec_xor (vector unsigned int, vector unsigned int); |
b0b343db JJ |
11170 | vector bool short vec_xor (vector bool short, vector bool short); |
11171 | vector signed short vec_xor (vector bool short, vector signed short); | |
11172 | vector signed short vec_xor (vector signed short, vector bool short); | |
333c8841 | 11173 | vector signed short vec_xor (vector signed short, vector signed short); |
b0b343db | 11174 | vector unsigned short vec_xor (vector bool short, |
924fcc4e JM |
11175 | vector unsigned short); |
11176 | vector unsigned short vec_xor (vector unsigned short, | |
b0b343db | 11177 | vector bool short); |
6e5bb5ad JM |
11178 | vector unsigned short vec_xor (vector unsigned short, |
11179 | vector unsigned short); | |
b0b343db JJ |
11180 | vector signed char vec_xor (vector bool char, vector signed char); |
11181 | vector bool char vec_xor (vector bool char, vector bool char); | |
11182 | vector signed char vec_xor (vector signed char, vector bool char); | |
333c8841 | 11183 | vector signed char vec_xor (vector signed char, vector signed char); |
b0b343db JJ |
11184 | vector unsigned char vec_xor (vector bool char, vector unsigned char); |
11185 | vector unsigned char vec_xor (vector unsigned char, vector bool char); | |
924fcc4e JM |
11186 | vector unsigned char vec_xor (vector unsigned char, |
11187 | vector unsigned char); | |
333c8841 | 11188 | |
b0b343db JJ |
11189 | int vec_all_eq (vector signed char, vector bool char); |
11190 | int vec_all_eq (vector signed char, vector signed char); | |
11191 | int vec_all_eq (vector unsigned char, vector bool char); | |
11192 | int vec_all_eq (vector unsigned char, vector unsigned char); | |
11193 | int vec_all_eq (vector bool char, vector bool char); | |
11194 | int vec_all_eq (vector bool char, vector unsigned char); | |
11195 | int vec_all_eq (vector bool char, vector signed char); | |
11196 | int vec_all_eq (vector signed short, vector bool short); | |
11197 | int vec_all_eq (vector signed short, vector signed short); | |
11198 | int vec_all_eq (vector unsigned short, vector bool short); | |
11199 | int vec_all_eq (vector unsigned short, vector unsigned short); | |
11200 | int vec_all_eq (vector bool short, vector bool short); | |
11201 | int vec_all_eq (vector bool short, vector unsigned short); | |
11202 | int vec_all_eq (vector bool short, vector signed short); | |
11203 | int vec_all_eq (vector pixel, vector pixel); | |
11204 | int vec_all_eq (vector signed int, vector bool int); | |
11205 | int vec_all_eq (vector signed int, vector signed int); | |
11206 | int vec_all_eq (vector unsigned int, vector bool int); | |
11207 | int vec_all_eq (vector unsigned int, vector unsigned int); | |
11208 | int vec_all_eq (vector bool int, vector bool int); | |
11209 | int vec_all_eq (vector bool int, vector unsigned int); | |
11210 | int vec_all_eq (vector bool int, vector signed int); | |
11211 | int vec_all_eq (vector float, vector float); | |
11212 | ||
11213 | int vec_all_ge (vector bool char, vector unsigned char); | |
11214 | int vec_all_ge (vector unsigned char, vector bool char); | |
11215 | int vec_all_ge (vector unsigned char, vector unsigned char); | |
11216 | int vec_all_ge (vector bool char, vector signed char); | |
11217 | int vec_all_ge (vector signed char, vector bool char); | |
11218 | int vec_all_ge (vector signed char, vector signed char); | |
11219 | int vec_all_ge (vector bool short, vector unsigned short); | |
11220 | int vec_all_ge (vector unsigned short, vector bool short); | |
11221 | int vec_all_ge (vector unsigned short, vector unsigned short); | |
11222 | int vec_all_ge (vector signed short, vector signed short); | |
11223 | int vec_all_ge (vector bool short, vector signed short); | |
11224 | int vec_all_ge (vector signed short, vector bool short); | |
11225 | int vec_all_ge (vector bool int, vector unsigned int); | |
11226 | int vec_all_ge (vector unsigned int, vector bool int); | |
11227 | int vec_all_ge (vector unsigned int, vector unsigned int); | |
11228 | int vec_all_ge (vector bool int, vector signed int); | |
11229 | int vec_all_ge (vector signed int, vector bool int); | |
11230 | int vec_all_ge (vector signed int, vector signed int); | |
11231 | int vec_all_ge (vector float, vector float); | |
11232 | ||
11233 | int vec_all_gt (vector bool char, vector unsigned char); | |
11234 | int vec_all_gt (vector unsigned char, vector bool char); | |
11235 | int vec_all_gt (vector unsigned char, vector unsigned char); | |
11236 | int vec_all_gt (vector bool char, vector signed char); | |
11237 | int vec_all_gt (vector signed char, vector bool char); | |
11238 | int vec_all_gt (vector signed char, vector signed char); | |
11239 | int vec_all_gt (vector bool short, vector unsigned short); | |
11240 | int vec_all_gt (vector unsigned short, vector bool short); | |
11241 | int vec_all_gt (vector unsigned short, vector unsigned short); | |
11242 | int vec_all_gt (vector bool short, vector signed short); | |
11243 | int vec_all_gt (vector signed short, vector bool short); | |
11244 | int vec_all_gt (vector signed short, vector signed short); | |
11245 | int vec_all_gt (vector bool int, vector unsigned int); | |
11246 | int vec_all_gt (vector unsigned int, vector bool int); | |
11247 | int vec_all_gt (vector unsigned int, vector unsigned int); | |
11248 | int vec_all_gt (vector bool int, vector signed int); | |
11249 | int vec_all_gt (vector signed int, vector bool int); | |
11250 | int vec_all_gt (vector signed int, vector signed int); | |
11251 | int vec_all_gt (vector float, vector float); | |
11252 | ||
11253 | int vec_all_in (vector float, vector float); | |
11254 | ||
11255 | int vec_all_le (vector bool char, vector unsigned char); | |
11256 | int vec_all_le (vector unsigned char, vector bool char); | |
11257 | int vec_all_le (vector unsigned char, vector unsigned char); | |
11258 | int vec_all_le (vector bool char, vector signed char); | |
11259 | int vec_all_le (vector signed char, vector bool char); | |
11260 | int vec_all_le (vector signed char, vector signed char); | |
11261 | int vec_all_le (vector bool short, vector unsigned short); | |
11262 | int vec_all_le (vector unsigned short, vector bool short); | |
11263 | int vec_all_le (vector unsigned short, vector unsigned short); | |
11264 | int vec_all_le (vector bool short, vector signed short); | |
11265 | int vec_all_le (vector signed short, vector bool short); | |
11266 | int vec_all_le (vector signed short, vector signed short); | |
11267 | int vec_all_le (vector bool int, vector unsigned int); | |
11268 | int vec_all_le (vector unsigned int, vector bool int); | |
11269 | int vec_all_le (vector unsigned int, vector unsigned int); | |
11270 | int vec_all_le (vector bool int, vector signed int); | |
11271 | int vec_all_le (vector signed int, vector bool int); | |
11272 | int vec_all_le (vector signed int, vector signed int); | |
11273 | int vec_all_le (vector float, vector float); | |
11274 | ||
11275 | int vec_all_lt (vector bool char, vector unsigned char); | |
11276 | int vec_all_lt (vector unsigned char, vector bool char); | |
11277 | int vec_all_lt (vector unsigned char, vector unsigned char); | |
11278 | int vec_all_lt (vector bool char, vector signed char); | |
11279 | int vec_all_lt (vector signed char, vector bool char); | |
11280 | int vec_all_lt (vector signed char, vector signed char); | |
11281 | int vec_all_lt (vector bool short, vector unsigned short); | |
11282 | int vec_all_lt (vector unsigned short, vector bool short); | |
11283 | int vec_all_lt (vector unsigned short, vector unsigned short); | |
11284 | int vec_all_lt (vector bool short, vector signed short); | |
11285 | int vec_all_lt (vector signed short, vector bool short); | |
11286 | int vec_all_lt (vector signed short, vector signed short); | |
11287 | int vec_all_lt (vector bool int, vector unsigned int); | |
11288 | int vec_all_lt (vector unsigned int, vector bool int); | |
11289 | int vec_all_lt (vector unsigned int, vector unsigned int); | |
11290 | int vec_all_lt (vector bool int, vector signed int); | |
11291 | int vec_all_lt (vector signed int, vector bool int); | |
11292 | int vec_all_lt (vector signed int, vector signed int); | |
11293 | int vec_all_lt (vector float, vector float); | |
11294 | ||
11295 | int vec_all_nan (vector float); | |
11296 | ||
11297 | int vec_all_ne (vector signed char, vector bool char); | |
11298 | int vec_all_ne (vector signed char, vector signed char); | |
11299 | int vec_all_ne (vector unsigned char, vector bool char); | |
11300 | int vec_all_ne (vector unsigned char, vector unsigned char); | |
11301 | int vec_all_ne (vector bool char, vector bool char); | |
11302 | int vec_all_ne (vector bool char, vector unsigned char); | |
11303 | int vec_all_ne (vector bool char, vector signed char); | |
11304 | int vec_all_ne (vector signed short, vector bool short); | |
11305 | int vec_all_ne (vector signed short, vector signed short); | |
11306 | int vec_all_ne (vector unsigned short, vector bool short); | |
11307 | int vec_all_ne (vector unsigned short, vector unsigned short); | |
11308 | int vec_all_ne (vector bool short, vector bool short); | |
11309 | int vec_all_ne (vector bool short, vector unsigned short); | |
11310 | int vec_all_ne (vector bool short, vector signed short); | |
11311 | int vec_all_ne (vector pixel, vector pixel); | |
11312 | int vec_all_ne (vector signed int, vector bool int); | |
11313 | int vec_all_ne (vector signed int, vector signed int); | |
11314 | int vec_all_ne (vector unsigned int, vector bool int); | |
11315 | int vec_all_ne (vector unsigned int, vector unsigned int); | |
11316 | int vec_all_ne (vector bool int, vector bool int); | |
11317 | int vec_all_ne (vector bool int, vector unsigned int); | |
11318 | int vec_all_ne (vector bool int, vector signed int); | |
11319 | int vec_all_ne (vector float, vector float); | |
11320 | ||
11321 | int vec_all_nge (vector float, vector float); | |
11322 | ||
11323 | int vec_all_ngt (vector float, vector float); | |
11324 | ||
11325 | int vec_all_nle (vector float, vector float); | |
11326 | ||
11327 | int vec_all_nlt (vector float, vector float); | |
11328 | ||
11329 | int vec_all_numeric (vector float); | |
11330 | ||
11331 | int vec_any_eq (vector signed char, vector bool char); | |
11332 | int vec_any_eq (vector signed char, vector signed char); | |
11333 | int vec_any_eq (vector unsigned char, vector bool char); | |
11334 | int vec_any_eq (vector unsigned char, vector unsigned char); | |
11335 | int vec_any_eq (vector bool char, vector bool char); | |
11336 | int vec_any_eq (vector bool char, vector unsigned char); | |
11337 | int vec_any_eq (vector bool char, vector signed char); | |
11338 | int vec_any_eq (vector signed short, vector bool short); | |
11339 | int vec_any_eq (vector signed short, vector signed short); | |
11340 | int vec_any_eq (vector unsigned short, vector bool short); | |
11341 | int vec_any_eq (vector unsigned short, vector unsigned short); | |
11342 | int vec_any_eq (vector bool short, vector bool short); | |
11343 | int vec_any_eq (vector bool short, vector unsigned short); | |
11344 | int vec_any_eq (vector bool short, vector signed short); | |
11345 | int vec_any_eq (vector pixel, vector pixel); | |
11346 | int vec_any_eq (vector signed int, vector bool int); | |
11347 | int vec_any_eq (vector signed int, vector signed int); | |
11348 | int vec_any_eq (vector unsigned int, vector bool int); | |
11349 | int vec_any_eq (vector unsigned int, vector unsigned int); | |
11350 | int vec_any_eq (vector bool int, vector bool int); | |
11351 | int vec_any_eq (vector bool int, vector unsigned int); | |
11352 | int vec_any_eq (vector bool int, vector signed int); | |
11353 | int vec_any_eq (vector float, vector float); | |
11354 | ||
11355 | int vec_any_ge (vector signed char, vector bool char); | |
11356 | int vec_any_ge (vector unsigned char, vector bool char); | |
11357 | int vec_any_ge (vector unsigned char, vector unsigned char); | |
11358 | int vec_any_ge (vector signed char, vector signed char); | |
11359 | int vec_any_ge (vector bool char, vector unsigned char); | |
11360 | int vec_any_ge (vector bool char, vector signed char); | |
11361 | int vec_any_ge (vector unsigned short, vector bool short); | |
11362 | int vec_any_ge (vector unsigned short, vector unsigned short); | |
11363 | int vec_any_ge (vector signed short, vector signed short); | |
11364 | int vec_any_ge (vector signed short, vector bool short); | |
11365 | int vec_any_ge (vector bool short, vector unsigned short); | |
11366 | int vec_any_ge (vector bool short, vector signed short); | |
11367 | int vec_any_ge (vector signed int, vector bool int); | |
11368 | int vec_any_ge (vector unsigned int, vector bool int); | |
11369 | int vec_any_ge (vector unsigned int, vector unsigned int); | |
11370 | int vec_any_ge (vector signed int, vector signed int); | |
11371 | int vec_any_ge (vector bool int, vector unsigned int); | |
11372 | int vec_any_ge (vector bool int, vector signed int); | |
11373 | int vec_any_ge (vector float, vector float); | |
11374 | ||
11375 | int vec_any_gt (vector bool char, vector unsigned char); | |
11376 | int vec_any_gt (vector unsigned char, vector bool char); | |
11377 | int vec_any_gt (vector unsigned char, vector unsigned char); | |
11378 | int vec_any_gt (vector bool char, vector signed char); | |
11379 | int vec_any_gt (vector signed char, vector bool char); | |
11380 | int vec_any_gt (vector signed char, vector signed char); | |
11381 | int vec_any_gt (vector bool short, vector unsigned short); | |
11382 | int vec_any_gt (vector unsigned short, vector bool short); | |
11383 | int vec_any_gt (vector unsigned short, vector unsigned short); | |
11384 | int vec_any_gt (vector bool short, vector signed short); | |
11385 | int vec_any_gt (vector signed short, vector bool short); | |
11386 | int vec_any_gt (vector signed short, vector signed short); | |
11387 | int vec_any_gt (vector bool int, vector unsigned int); | |
11388 | int vec_any_gt (vector unsigned int, vector bool int); | |
11389 | int vec_any_gt (vector unsigned int, vector unsigned int); | |
11390 | int vec_any_gt (vector bool int, vector signed int); | |
11391 | int vec_any_gt (vector signed int, vector bool int); | |
11392 | int vec_any_gt (vector signed int, vector signed int); | |
11393 | int vec_any_gt (vector float, vector float); | |
11394 | ||
11395 | int vec_any_le (vector bool char, vector unsigned char); | |
11396 | int vec_any_le (vector unsigned char, vector bool char); | |
11397 | int vec_any_le (vector unsigned char, vector unsigned char); | |
11398 | int vec_any_le (vector bool char, vector signed char); | |
11399 | int vec_any_le (vector signed char, vector bool char); | |
11400 | int vec_any_le (vector signed char, vector signed char); | |
11401 | int vec_any_le (vector bool short, vector unsigned short); | |
11402 | int vec_any_le (vector unsigned short, vector bool short); | |
11403 | int vec_any_le (vector unsigned short, vector unsigned short); | |
11404 | int vec_any_le (vector bool short, vector signed short); | |
11405 | int vec_any_le (vector signed short, vector bool short); | |
11406 | int vec_any_le (vector signed short, vector signed short); | |
11407 | int vec_any_le (vector bool int, vector unsigned int); | |
11408 | int vec_any_le (vector unsigned int, vector bool int); | |
11409 | int vec_any_le (vector unsigned int, vector unsigned int); | |
11410 | int vec_any_le (vector bool int, vector signed int); | |
11411 | int vec_any_le (vector signed int, vector bool int); | |
11412 | int vec_any_le (vector signed int, vector signed int); | |
11413 | int vec_any_le (vector float, vector float); | |
11414 | ||
11415 | int vec_any_lt (vector bool char, vector unsigned char); | |
11416 | int vec_any_lt (vector unsigned char, vector bool char); | |
11417 | int vec_any_lt (vector unsigned char, vector unsigned char); | |
11418 | int vec_any_lt (vector bool char, vector signed char); | |
11419 | int vec_any_lt (vector signed char, vector bool char); | |
11420 | int vec_any_lt (vector signed char, vector signed char); | |
11421 | int vec_any_lt (vector bool short, vector unsigned short); | |
11422 | int vec_any_lt (vector unsigned short, vector bool short); | |
11423 | int vec_any_lt (vector unsigned short, vector unsigned short); | |
11424 | int vec_any_lt (vector bool short, vector signed short); | |
11425 | int vec_any_lt (vector signed short, vector bool short); | |
11426 | int vec_any_lt (vector signed short, vector signed short); | |
11427 | int vec_any_lt (vector bool int, vector unsigned int); | |
11428 | int vec_any_lt (vector unsigned int, vector bool int); | |
11429 | int vec_any_lt (vector unsigned int, vector unsigned int); | |
11430 | int vec_any_lt (vector bool int, vector signed int); | |
11431 | int vec_any_lt (vector signed int, vector bool int); | |
11432 | int vec_any_lt (vector signed int, vector signed int); | |
11433 | int vec_any_lt (vector float, vector float); | |
11434 | ||
11435 | int vec_any_nan (vector float); | |
11436 | ||
11437 | int vec_any_ne (vector signed char, vector bool char); | |
11438 | int vec_any_ne (vector signed char, vector signed char); | |
11439 | int vec_any_ne (vector unsigned char, vector bool char); | |
11440 | int vec_any_ne (vector unsigned char, vector unsigned char); | |
11441 | int vec_any_ne (vector bool char, vector bool char); | |
11442 | int vec_any_ne (vector bool char, vector unsigned char); | |
11443 | int vec_any_ne (vector bool char, vector signed char); | |
11444 | int vec_any_ne (vector signed short, vector bool short); | |
11445 | int vec_any_ne (vector signed short, vector signed short); | |
11446 | int vec_any_ne (vector unsigned short, vector bool short); | |
11447 | int vec_any_ne (vector unsigned short, vector unsigned short); | |
11448 | int vec_any_ne (vector bool short, vector bool short); | |
11449 | int vec_any_ne (vector bool short, vector unsigned short); | |
11450 | int vec_any_ne (vector bool short, vector signed short); | |
11451 | int vec_any_ne (vector pixel, vector pixel); | |
11452 | int vec_any_ne (vector signed int, vector bool int); | |
11453 | int vec_any_ne (vector signed int, vector signed int); | |
11454 | int vec_any_ne (vector unsigned int, vector bool int); | |
11455 | int vec_any_ne (vector unsigned int, vector unsigned int); | |
11456 | int vec_any_ne (vector bool int, vector bool int); | |
11457 | int vec_any_ne (vector bool int, vector unsigned int); | |
11458 | int vec_any_ne (vector bool int, vector signed int); | |
11459 | int vec_any_ne (vector float, vector float); | |
11460 | ||
11461 | int vec_any_nge (vector float, vector float); | |
11462 | ||
11463 | int vec_any_ngt (vector float, vector float); | |
11464 | ||
11465 | int vec_any_nle (vector float, vector float); | |
11466 | ||
11467 | int vec_any_nlt (vector float, vector float); | |
11468 | ||
11469 | int vec_any_numeric (vector float); | |
11470 | ||
11471 | int vec_any_out (vector float, vector float); | |
333c8841 AH |
11472 | @end smallexample |
11473 | ||
c5145ceb JM |
11474 | @node SPARC VIS Built-in Functions |
11475 | @subsection SPARC VIS Built-in Functions | |
11476 | ||
11477 | GCC supports SIMD operations on the SPARC using both the generic vector | |
2fd13506 | 11478 | extensions (@pxref{Vector Extensions}) as well as built-in functions for |
c5145ceb JM |
11479 | the SPARC Visual Instruction Set (VIS). When you use the @option{-mvis} |
11480 | switch, the VIS extension is exposed as the following built-in functions: | |
11481 | ||
11482 | @smallexample | |
11483 | typedef int v2si __attribute__ ((vector_size (8))); | |
11484 | typedef short v4hi __attribute__ ((vector_size (8))); | |
11485 | typedef short v2hi __attribute__ ((vector_size (4))); | |
11486 | typedef char v8qi __attribute__ ((vector_size (8))); | |
11487 | typedef char v4qi __attribute__ ((vector_size (4))); | |
11488 | ||
11489 | void * __builtin_vis_alignaddr (void *, long); | |
11490 | int64_t __builtin_vis_faligndatadi (int64_t, int64_t); | |
11491 | v2si __builtin_vis_faligndatav2si (v2si, v2si); | |
11492 | v4hi __builtin_vis_faligndatav4hi (v4si, v4si); | |
11493 | v8qi __builtin_vis_faligndatav8qi (v8qi, v8qi); | |
11494 | ||
11495 | v4hi __builtin_vis_fexpand (v4qi); | |
11496 | ||
11497 | v4hi __builtin_vis_fmul8x16 (v4qi, v4hi); | |
11498 | v4hi __builtin_vis_fmul8x16au (v4qi, v4hi); | |
11499 | v4hi __builtin_vis_fmul8x16al (v4qi, v4hi); | |
11500 | v4hi __builtin_vis_fmul8sux16 (v8qi, v4hi); | |
11501 | v4hi __builtin_vis_fmul8ulx16 (v8qi, v4hi); | |
11502 | v2si __builtin_vis_fmuld8sux16 (v4qi, v2hi); | |
11503 | v2si __builtin_vis_fmuld8ulx16 (v4qi, v2hi); | |
11504 | ||
11505 | v4qi __builtin_vis_fpack16 (v4hi); | |
11506 | v8qi __builtin_vis_fpack32 (v2si, v2si); | |
11507 | v2hi __builtin_vis_fpackfix (v2si); | |
11508 | v8qi __builtin_vis_fpmerge (v4qi, v4qi); | |
11509 | ||
11510 | int64_t __builtin_vis_pdist (v8qi, v8qi, int64_t); | |
11511 | @end smallexample | |
11512 | ||
85d9c13c TS |
11513 | @node SPU Built-in Functions |
11514 | @subsection SPU Built-in Functions | |
11515 | ||
11516 | GCC provides extensions for the SPU processor as described in the | |
11517 | Sony/Toshiba/IBM SPU Language Extensions Specification, which can be | |
11518 | found at @uref{http://cell.scei.co.jp/} or | |
11519 | @uref{http://www.ibm.com/developerworks/power/cell/}. GCC's | |
11520 | implementation differs in several ways. | |
11521 | ||
11522 | @itemize @bullet | |
11523 | ||
11524 | @item | |
11525 | The optional extension of specifying vector constants in parentheses is | |
11526 | not supported. | |
11527 | ||
11528 | @item | |
11529 | A vector initializer requires no cast if the vector constant is of the | |
11530 | same type as the variable it is initializing. | |
11531 | ||
11532 | @item | |
11533 | If @code{signed} or @code{unsigned} is omitted, the signedness of the | |
11534 | vector type is the default signedness of the base type. The default | |
11535 | varies depending on the operating system, so a portable program should | |
11536 | always specify the signedness. | |
11537 | ||
11538 | @item | |
11539 | By default, the keyword @code{__vector} is added. The macro | |
11540 | @code{vector} is defined in @code{<spu_intrinsics.h>} and can be | |
11541 | undefined. | |
11542 | ||
11543 | @item | |
11544 | GCC allows using a @code{typedef} name as the type specifier for a | |
11545 | vector type. | |
11546 | ||
11547 | @item | |
11548 | For C, overloaded functions are implemented with macros so the following | |
11549 | does not work: | |
11550 | ||
11551 | @smallexample | |
11552 | spu_add ((vector signed int)@{1, 2, 3, 4@}, foo); | |
11553 | @end smallexample | |
11554 | ||
11555 | Since @code{spu_add} is a macro, the vector constant in the example | |
11556 | is treated as four separate arguments. Wrap the entire argument in | |
11557 | parentheses for this to work. | |
11558 | ||
11559 | @item | |
11560 | The extended version of @code{__builtin_expect} is not supported. | |
11561 | ||
11562 | @end itemize | |
11563 | ||
5681c208 | 11564 | @emph{Note:} Only the interface described in the aforementioned |
85d9c13c TS |
11565 | specification is supported. Internally, GCC uses built-in functions to |
11566 | implement the required functionality, but these are not supported and | |
11567 | are subject to change without notice. | |
11568 | ||
a2bec818 DJ |
11569 | @node Target Format Checks |
11570 | @section Format Checks Specific to Particular Target Machines | |
11571 | ||
11572 | For some target machines, GCC supports additional options to the | |
11573 | format attribute | |
11574 | (@pxref{Function Attributes,,Declaring Attributes of Functions}). | |
11575 | ||
11576 | @menu | |
11577 | * Solaris Format Checks:: | |
11578 | @end menu | |
11579 | ||
11580 | @node Solaris Format Checks | |
11581 | @subsection Solaris Format Checks | |
11582 | ||
11583 | Solaris targets support the @code{cmn_err} (or @code{__cmn_err__}) format | |
11584 | check. @code{cmn_err} accepts a subset of the standard @code{printf} | |
11585 | conversions, and the two-argument @code{%b} conversion for displaying | |
11586 | bit-fields. See the Solaris man page for @code{cmn_err} for more information. | |
11587 | ||
0168a849 SS |
11588 | @node Pragmas |
11589 | @section Pragmas Accepted by GCC | |
11590 | @cindex pragmas | |
11591 | @cindex #pragma | |
11592 | ||
11593 | GCC supports several types of pragmas, primarily in order to compile | |
11594 | code originally written for other compilers. Note that in general | |
11595 | we do not recommend the use of pragmas; @xref{Function Attributes}, | |
11596 | for further explanation. | |
11597 | ||
11598 | @menu | |
11599 | * ARM Pragmas:: | |
38b2d076 | 11600 | * M32C Pragmas:: |
a5c76ee6 | 11601 | * RS/6000 and PowerPC Pragmas:: |
0168a849 | 11602 | * Darwin Pragmas:: |
07a43492 | 11603 | * Solaris Pragmas:: |
84b8b0e0 | 11604 | * Symbol-Renaming Pragmas:: |
467cecf3 | 11605 | * Structure-Packing Pragmas:: |
52eb57df | 11606 | * Weak Pragmas:: |
79cf5994 | 11607 | * Diagnostic Pragmas:: |
b9e75696 | 11608 | * Visibility Pragmas:: |
20cef83a | 11609 | * Push/Pop Macro Pragmas:: |
ab442df7 | 11610 | * Function Specific Option Pragmas:: |
0168a849 SS |
11611 | @end menu |
11612 | ||
11613 | @node ARM Pragmas | |
11614 | @subsection ARM Pragmas | |
11615 | ||
11616 | The ARM target defines pragmas for controlling the default addition of | |
11617 | @code{long_call} and @code{short_call} attributes to functions. | |
11618 | @xref{Function Attributes}, for information about the effects of these | |
11619 | attributes. | |
11620 | ||
11621 | @table @code | |
11622 | @item long_calls | |
11623 | @cindex pragma, long_calls | |
11624 | Set all subsequent functions to have the @code{long_call} attribute. | |
11625 | ||
11626 | @item no_long_calls | |
11627 | @cindex pragma, no_long_calls | |
11628 | Set all subsequent functions to have the @code{short_call} attribute. | |
11629 | ||
11630 | @item long_calls_off | |
11631 | @cindex pragma, long_calls_off | |
11632 | Do not affect the @code{long_call} or @code{short_call} attributes of | |
11633 | subsequent functions. | |
11634 | @end table | |
11635 | ||
38b2d076 DD |
11636 | @node M32C Pragmas |
11637 | @subsection M32C Pragmas | |
11638 | ||
11639 | @table @code | |
11640 | @item memregs @var{number} | |
11641 | @cindex pragma, memregs | |
11642 | Overrides the command line option @code{-memregs=} for the current | |
11643 | file. Use with care! This pragma must be before any function in the | |
11644 | file, and mixing different memregs values in different objects may | |
11645 | make them incompatible. This pragma is useful when a | |
11646 | performance-critical function uses a memreg for temporary values, | |
11647 | as it may allow you to reduce the number of memregs used. | |
11648 | ||
11649 | @end table | |
11650 | ||
a5c76ee6 ZW |
11651 | @node RS/6000 and PowerPC Pragmas |
11652 | @subsection RS/6000 and PowerPC Pragmas | |
11653 | ||
11654 | The RS/6000 and PowerPC targets define one pragma for controlling | |
11655 | whether or not the @code{longcall} attribute is added to function | |
11656 | declarations by default. This pragma overrides the @option{-mlongcall} | |
95b1627e | 11657 | option, but not the @code{longcall} and @code{shortcall} attributes. |
a5c76ee6 ZW |
11658 | @xref{RS/6000 and PowerPC Options}, for more information about when long |
11659 | calls are and are not necessary. | |
11660 | ||
11661 | @table @code | |
11662 | @item longcall (1) | |
11663 | @cindex pragma, longcall | |
11664 | Apply the @code{longcall} attribute to all subsequent function | |
11665 | declarations. | |
11666 | ||
11667 | @item longcall (0) | |
11668 | Do not apply the @code{longcall} attribute to subsequent function | |
11669 | declarations. | |
11670 | @end table | |
11671 | ||
0168a849 | 11672 | @c Describe h8300 pragmas here. |
0168a849 SS |
11673 | @c Describe sh pragmas here. |
11674 | @c Describe v850 pragmas here. | |
11675 | ||
11676 | @node Darwin Pragmas | |
11677 | @subsection Darwin Pragmas | |
11678 | ||
11679 | The following pragmas are available for all architectures running the | |
11680 | Darwin operating system. These are useful for compatibility with other | |
85ebf0c6 | 11681 | Mac OS compilers. |
0168a849 SS |
11682 | |
11683 | @table @code | |
11684 | @item mark @var{tokens}@dots{} | |
11685 | @cindex pragma, mark | |
11686 | This pragma is accepted, but has no effect. | |
11687 | ||
11688 | @item options align=@var{alignment} | |
11689 | @cindex pragma, options align | |
11690 | This pragma sets the alignment of fields in structures. The values of | |
11691 | @var{alignment} may be @code{mac68k}, to emulate m68k alignment, or | |
11692 | @code{power}, to emulate PowerPC alignment. Uses of this pragma nest | |
11693 | properly; to restore the previous setting, use @code{reset} for the | |
11694 | @var{alignment}. | |
11695 | ||
11696 | @item segment @var{tokens}@dots{} | |
11697 | @cindex pragma, segment | |
11698 | This pragma is accepted, but has no effect. | |
11699 | ||
11700 | @item unused (@var{var} [, @var{var}]@dots{}) | |
11701 | @cindex pragma, unused | |
11702 | This pragma declares variables to be possibly unused. GCC will not | |
11703 | produce warnings for the listed variables. The effect is similar to | |
11704 | that of the @code{unused} attribute, except that this pragma may appear | |
11705 | anywhere within the variables' scopes. | |
11706 | @end table | |
11707 | ||
07a43492 DJ |
11708 | @node Solaris Pragmas |
11709 | @subsection Solaris Pragmas | |
11710 | ||
11711 | The Solaris target supports @code{#pragma redefine_extname} | |
11712 | (@pxref{Symbol-Renaming Pragmas}). It also supports additional | |
11713 | @code{#pragma} directives for compatibility with the system compiler. | |
11714 | ||
11715 | @table @code | |
11716 | @item align @var{alignment} (@var{variable} [, @var{variable}]...) | |
11717 | @cindex pragma, align | |
11718 | ||
11719 | Increase the minimum alignment of each @var{variable} to @var{alignment}. | |
11720 | This is the same as GCC's @code{aligned} attribute @pxref{Variable | |
b5b3e36a | 11721 | Attributes}). Macro expansion occurs on the arguments to this pragma |
0ee2ea09 | 11722 | when compiling C and Objective-C@. It does not currently occur when |
b5b3e36a DJ |
11723 | compiling C++, but this is a bug which may be fixed in a future |
11724 | release. | |
07a43492 DJ |
11725 | |
11726 | @item fini (@var{function} [, @var{function}]...) | |
11727 | @cindex pragma, fini | |
11728 | ||
11729 | This pragma causes each listed @var{function} to be called after | |
11730 | main, or during shared module unloading, by adding a call to the | |
11731 | @code{.fini} section. | |
11732 | ||
11733 | @item init (@var{function} [, @var{function}]...) | |
11734 | @cindex pragma, init | |
11735 | ||
11736 | This pragma causes each listed @var{function} to be called during | |
11737 | initialization (before @code{main}) or during shared module loading, by | |
11738 | adding a call to the @code{.init} section. | |
11739 | ||
11740 | @end table | |
11741 | ||
84b8b0e0 ZW |
11742 | @node Symbol-Renaming Pragmas |
11743 | @subsection Symbol-Renaming Pragmas | |
41c64394 | 11744 | |
84b8b0e0 ZW |
11745 | For compatibility with the Solaris and Tru64 UNIX system headers, GCC |
11746 | supports two @code{#pragma} directives which change the name used in | |
11747 | assembly for a given declaration. These pragmas are only available on | |
11748 | platforms whose system headers need them. To get this effect on all | |
11749 | platforms supported by GCC, use the asm labels extension (@pxref{Asm | |
11750 | Labels}). | |
41c64394 RH |
11751 | |
11752 | @table @code | |
11753 | @item redefine_extname @var{oldname} @var{newname} | |
11754 | @cindex pragma, redefine_extname | |
11755 | ||
84b8b0e0 ZW |
11756 | This pragma gives the C function @var{oldname} the assembly symbol |
11757 | @var{newname}. The preprocessor macro @code{__PRAGMA_REDEFINE_EXTNAME} | |
11758 | will be defined if this pragma is available (currently only on | |
11759 | Solaris). | |
41c64394 | 11760 | |
41c64394 RH |
11761 | @item extern_prefix @var{string} |
11762 | @cindex pragma, extern_prefix | |
11763 | ||
84b8b0e0 ZW |
11764 | This pragma causes all subsequent external function and variable |
11765 | declarations to have @var{string} prepended to their assembly symbols. | |
11766 | This effect may be terminated with another @code{extern_prefix} pragma | |
11767 | whose argument is an empty string. The preprocessor macro | |
11768 | @code{__PRAGMA_EXTERN_PREFIX} will be defined if this pragma is | |
8a36672b | 11769 | available (currently only on Tru64 UNIX)@. |
41c64394 RH |
11770 | @end table |
11771 | ||
84b8b0e0 ZW |
11772 | These pragmas and the asm labels extension interact in a complicated |
11773 | manner. Here are some corner cases you may want to be aware of. | |
11774 | ||
11775 | @enumerate | |
11776 | @item Both pragmas silently apply only to declarations with external | |
11777 | linkage. Asm labels do not have this restriction. | |
11778 | ||
11779 | @item In C++, both pragmas silently apply only to declarations with | |
11780 | ``C'' linkage. Again, asm labels do not have this restriction. | |
11781 | ||
11782 | @item If any of the three ways of changing the assembly name of a | |
11783 | declaration is applied to a declaration whose assembly name has | |
11784 | already been determined (either by a previous use of one of these | |
11785 | features, or because the compiler needed the assembly name in order to | |
11786 | generate code), and the new name is different, a warning issues and | |
11787 | the name does not change. | |
11788 | ||
11789 | @item The @var{oldname} used by @code{#pragma redefine_extname} is | |
11790 | always the C-language name. | |
11791 | ||
11792 | @item If @code{#pragma extern_prefix} is in effect, and a declaration | |
11793 | occurs with an asm label attached, the prefix is silently ignored for | |
11794 | that declaration. | |
11795 | ||
11796 | @item If @code{#pragma extern_prefix} and @code{#pragma redefine_extname} | |
11797 | apply to the same declaration, whichever triggered first wins, and a | |
11798 | warning issues if they contradict each other. (We would like to have | |
11799 | @code{#pragma redefine_extname} always win, for consistency with asm | |
11800 | labels, but if @code{#pragma extern_prefix} triggers first we have no | |
11801 | way of knowing that that happened.) | |
11802 | @end enumerate | |
11803 | ||
467cecf3 JB |
11804 | @node Structure-Packing Pragmas |
11805 | @subsection Structure-Packing Pragmas | |
11806 | ||
20cef83a DS |
11807 | For compatibility with Microsoft Windows compilers, GCC supports a |
11808 | set of @code{#pragma} directives which change the maximum alignment of | |
11809 | members of structures (other than zero-width bitfields), unions, and | |
11810 | classes subsequently defined. The @var{n} value below always is required | |
11811 | to be a small power of two and specifies the new alignment in bytes. | |
467cecf3 JB |
11812 | |
11813 | @enumerate | |
11814 | @item @code{#pragma pack(@var{n})} simply sets the new alignment. | |
11815 | @item @code{#pragma pack()} sets the alignment to the one that was in | |
11816 | effect when compilation started (see also command line option | |
11817 | @option{-fpack-struct[=<n>]} @pxref{Code Gen Options}). | |
11818 | @item @code{#pragma pack(push[,@var{n}])} pushes the current alignment | |
11819 | setting on an internal stack and then optionally sets the new alignment. | |
11820 | @item @code{#pragma pack(pop)} restores the alignment setting to the one | |
11821 | saved at the top of the internal stack (and removes that stack entry). | |
11822 | Note that @code{#pragma pack([@var{n}])} does not influence this internal | |
11823 | stack; thus it is possible to have @code{#pragma pack(push)} followed by | |
11824 | multiple @code{#pragma pack(@var{n})} instances and finalized by a single | |
11825 | @code{#pragma pack(pop)}. | |
11826 | @end enumerate | |
11827 | ||
021efafc | 11828 | Some targets, e.g.@: i386 and powerpc, support the @code{ms_struct} |
6bb7beac EC |
11829 | @code{#pragma} which lays out a structure as the documented |
11830 | @code{__attribute__ ((ms_struct))}. | |
11831 | @enumerate | |
11832 | @item @code{#pragma ms_struct on} turns on the layout for structures | |
11833 | declared. | |
11834 | @item @code{#pragma ms_struct off} turns off the layout for structures | |
11835 | declared. | |
11836 | @item @code{#pragma ms_struct reset} goes back to the default layout. | |
11837 | @end enumerate | |
11838 | ||
52eb57df RH |
11839 | @node Weak Pragmas |
11840 | @subsection Weak Pragmas | |
11841 | ||
11842 | For compatibility with SVR4, GCC supports a set of @code{#pragma} | |
11843 | directives for declaring symbols to be weak, and defining weak | |
11844 | aliases. | |
11845 | ||
11846 | @table @code | |
11847 | @item #pragma weak @var{symbol} | |
11848 | @cindex pragma, weak | |
11849 | This pragma declares @var{symbol} to be weak, as if the declaration | |
11850 | had the attribute of the same name. The pragma may appear before | |
0ac11108 | 11851 | or after the declaration of @var{symbol}, but must appear before |
52eb57df RH |
11852 | either its first use or its definition. It is not an error for |
11853 | @var{symbol} to never be defined at all. | |
11854 | ||
11855 | @item #pragma weak @var{symbol1} = @var{symbol2} | |
11856 | This pragma declares @var{symbol1} to be a weak alias of @var{symbol2}. | |
11857 | It is an error if @var{symbol2} is not defined in the current | |
11858 | translation unit. | |
11859 | @end table | |
11860 | ||
79cf5994 DD |
11861 | @node Diagnostic Pragmas |
11862 | @subsection Diagnostic Pragmas | |
11863 | ||
11864 | GCC allows the user to selectively enable or disable certain types of | |
11865 | diagnostics, and change the kind of the diagnostic. For example, a | |
11866 | project's policy might require that all sources compile with | |
11867 | @option{-Werror} but certain files might have exceptions allowing | |
11868 | specific types of warnings. Or, a project might selectively enable | |
11869 | diagnostics and treat them as errors depending on which preprocessor | |
11870 | macros are defined. | |
11871 | ||
11872 | @table @code | |
11873 | @item #pragma GCC diagnostic @var{kind} @var{option} | |
11874 | @cindex pragma, diagnostic | |
11875 | ||
11876 | Modifies the disposition of a diagnostic. Note that not all | |
1eaf20ec | 11877 | diagnostics are modifiable; at the moment only warnings (normally |
923158be | 11878 | controlled by @samp{-W@dots{}}) can be controlled, and not all of them. |
79cf5994 DD |
11879 | Use @option{-fdiagnostics-show-option} to determine which diagnostics |
11880 | are controllable and which option controls them. | |
11881 | ||
11882 | @var{kind} is @samp{error} to treat this diagnostic as an error, | |
11883 | @samp{warning} to treat it like a warning (even if @option{-Werror} is | |
11884 | in effect), or @samp{ignored} if the diagnostic is to be ignored. | |
11885 | @var{option} is a double quoted string which matches the command line | |
11886 | option. | |
11887 | ||
11888 | @example | |
11889 | #pragma GCC diagnostic warning "-Wformat" | |
c116cd05 MLI |
11890 | #pragma GCC diagnostic error "-Wformat" |
11891 | #pragma GCC diagnostic ignored "-Wformat" | |
79cf5994 DD |
11892 | @end example |
11893 | ||
11894 | Note that these pragmas override any command line options. Also, | |
11895 | while it is syntactically valid to put these pragmas anywhere in your | |
11896 | sources, the only supported location for them is before any data or | |
11897 | functions are defined. Doing otherwise may result in unpredictable | |
11898 | results depending on how the optimizer manages your sources. If the | |
11899 | same option is listed multiple times, the last one specified is the | |
11900 | one that is in effect. This pragma is not intended to be a general | |
11901 | purpose replacement for command line options, but for implementing | |
11902 | strict control over project policies. | |
11903 | ||
11904 | @end table | |
11905 | ||
0d48657d SB |
11906 | GCC also offers a simple mechanism for printing messages during |
11907 | compilation. | |
11908 | ||
11909 | @table @code | |
11910 | @item #pragma message @var{string} | |
11911 | @cindex pragma, diagnostic | |
11912 | ||
11913 | Prints @var{string} as a compiler message on compilation. The message | |
11914 | is informational only, and is neither a compilation warning nor an error. | |
11915 | ||
11916 | @smallexample | |
11917 | #pragma message "Compiling " __FILE__ "..." | |
11918 | @end smallexample | |
11919 | ||
11920 | @var{string} may be parenthesized, and is printed with location | |
11921 | information. For example, | |
11922 | ||
11923 | @smallexample | |
11924 | #define DO_PRAGMA(x) _Pragma (#x) | |
11925 | #define TODO(x) DO_PRAGMA(message ("TODO - " #x)) | |
11926 | ||
11927 | TODO(Remember to fix this) | |
11928 | @end smallexample | |
11929 | ||
11930 | prints @samp{/tmp/file.c:4: note: #pragma message: | |
11931 | TODO - Remember to fix this}. | |
11932 | ||
11933 | @end table | |
11934 | ||
b9e75696 JM |
11935 | @node Visibility Pragmas |
11936 | @subsection Visibility Pragmas | |
11937 | ||
11938 | @table @code | |
11939 | @item #pragma GCC visibility push(@var{visibility}) | |
11940 | @itemx #pragma GCC visibility pop | |
11941 | @cindex pragma, visibility | |
11942 | ||
11943 | This pragma allows the user to set the visibility for multiple | |
11944 | declarations without having to give each a visibility attribute | |
11945 | @xref{Function Attributes}, for more information about visibility and | |
11946 | the attribute syntax. | |
11947 | ||
11948 | In C++, @samp{#pragma GCC visibility} affects only namespace-scope | |
11949 | declarations. Class members and template specializations are not | |
11950 | affected; if you want to override the visibility for a particular | |
11951 | member or instantiation, you must use an attribute. | |
11952 | ||
11953 | @end table | |
11954 | ||
20cef83a DS |
11955 | |
11956 | @node Push/Pop Macro Pragmas | |
11957 | @subsection Push/Pop Macro Pragmas | |
11958 | ||
11959 | For compatibility with Microsoft Windows compilers, GCC supports | |
11960 | @samp{#pragma push_macro(@var{"macro_name"})} | |
11961 | and @samp{#pragma pop_macro(@var{"macro_name"})}. | |
11962 | ||
11963 | @table @code | |
11964 | @item #pragma push_macro(@var{"macro_name"}) | |
11965 | @cindex pragma, push_macro | |
11966 | This pragma saves the value of the macro named as @var{macro_name} to | |
11967 | the top of the stack for this macro. | |
11968 | ||
11969 | @item #pragma pop_macro(@var{"macro_name"}) | |
11970 | @cindex pragma, pop_macro | |
11971 | This pragma sets the value of the macro named as @var{macro_name} to | |
11972 | the value on top of the stack for this macro. If the stack for | |
11973 | @var{macro_name} is empty, the value of the macro remains unchanged. | |
11974 | @end table | |
11975 | ||
11976 | For example: | |
11977 | ||
11978 | @smallexample | |
11979 | #define X 1 | |
11980 | #pragma push_macro("X") | |
11981 | #undef X | |
11982 | #define X -1 | |
11983 | #pragma pop_macro("X") | |
11984 | int x [X]; | |
11985 | @end smallexample | |
11986 | ||
11987 | In this example, the definition of X as 1 is saved by @code{#pragma | |
11988 | push_macro} and restored by @code{#pragma pop_macro}. | |
11989 | ||
ab442df7 MM |
11990 | @node Function Specific Option Pragmas |
11991 | @subsection Function Specific Option Pragmas | |
11992 | ||
11993 | @table @code | |
5779e713 MM |
11994 | @item #pragma GCC target (@var{"string"}...) |
11995 | @cindex pragma GCC target | |
ab442df7 MM |
11996 | |
11997 | This pragma allows you to set target specific options for functions | |
11998 | defined later in the source file. One or more strings can be | |
11999 | specified. Each function that is defined after this point will be as | |
5779e713 | 12000 | if @code{attribute((target("STRING")))} was specified for that |
ab442df7 MM |
12001 | function. The parenthesis around the options is optional. |
12002 | @xref{Function Attributes}, for more information about the | |
5779e713 | 12003 | @code{target} attribute and the attribute syntax. |
ab442df7 | 12004 | |
5779e713 | 12005 | The @samp{#pragma GCC target} pragma is not implemented in GCC |
ab442df7 | 12006 | versions earlier than 4.4, and is currently only implemented for the |
5779e713 | 12007 | 386 and x86_64 backends. |
ab442df7 MM |
12008 | @end table |
12009 | ||
ab442df7 MM |
12010 | @table @code |
12011 | @item #pragma GCC optimize (@var{"string"}...) | |
12012 | @cindex pragma GCC optimize | |
12013 | ||
12014 | This pragma allows you to set global optimization options for functions | |
12015 | defined later in the source file. One or more strings can be | |
12016 | specified. Each function that is defined after this point will be as | |
12017 | if @code{attribute((optimize("STRING")))} was specified for that | |
12018 | function. The parenthesis around the options is optional. | |
12019 | @xref{Function Attributes}, for more information about the | |
12020 | @code{optimize} attribute and the attribute syntax. | |
12021 | ||
12022 | The @samp{#pragma GCC optimize} pragma is not implemented in GCC | |
12023 | versions earlier than 4.4. | |
12024 | @end table | |
12025 | ||
12026 | @table @code | |
5779e713 MM |
12027 | @item #pragma GCC push_options |
12028 | @itemx #pragma GCC pop_options | |
12029 | @cindex pragma GCC push_options | |
12030 | @cindex pragma GCC pop_options | |
12031 | ||
12032 | These pragmas maintain a stack of the current target and optimization | |
12033 | options. It is intended for include files where you temporarily want | |
12034 | to switch to using a different @samp{#pragma GCC target} or | |
12035 | @samp{#pragma GCC optimize} and then to pop back to the previous | |
12036 | options. | |
12037 | ||
12038 | The @samp{#pragma GCC push_options} and @samp{#pragma GCC pop_options} | |
12039 | pragmas are not implemented in GCC versions earlier than 4.4. | |
ab442df7 MM |
12040 | @end table |
12041 | ||
12042 | @table @code | |
5779e713 MM |
12043 | @item #pragma GCC reset_options |
12044 | @cindex pragma GCC reset_options | |
ab442df7 | 12045 | |
5779e713 MM |
12046 | This pragma clears the current @code{#pragma GCC target} and |
12047 | @code{#pragma GCC optimize} to use the default switches as specified | |
12048 | on the command line. | |
12049 | ||
12050 | The @samp{#pragma GCC reset_options} pragma is not implemented in GCC | |
12051 | versions earlier than 4.4. | |
ab442df7 MM |
12052 | @end table |
12053 | ||
3e96a2fd | 12054 | @node Unnamed Fields |
2fbebc71 | 12055 | @section Unnamed struct/union fields within structs/unions |
3e96a2fd DD |
12056 | @cindex struct |
12057 | @cindex union | |
12058 | ||
12059 | For compatibility with other compilers, GCC allows you to define | |
12060 | a structure or union that contains, as fields, structures and unions | |
12061 | without names. For example: | |
12062 | ||
3ab51846 | 12063 | @smallexample |
3e96a2fd DD |
12064 | struct @{ |
12065 | int a; | |
12066 | union @{ | |
12067 | int b; | |
12068 | float c; | |
12069 | @}; | |
12070 | int d; | |
12071 | @} foo; | |
3ab51846 | 12072 | @end smallexample |
3e96a2fd DD |
12073 | |
12074 | In this example, the user would be able to access members of the unnamed | |
12075 | union with code like @samp{foo.b}. Note that only unnamed structs and | |
12076 | unions are allowed, you may not have, for example, an unnamed | |
12077 | @code{int}. | |
12078 | ||
12079 | You must never create such structures that cause ambiguous field definitions. | |
12080 | For example, this structure: | |
12081 | ||
3ab51846 | 12082 | @smallexample |
3e96a2fd DD |
12083 | struct @{ |
12084 | int a; | |
12085 | struct @{ | |
12086 | int a; | |
12087 | @}; | |
12088 | @} foo; | |
3ab51846 | 12089 | @end smallexample |
3e96a2fd DD |
12090 | |
12091 | It is ambiguous which @code{a} is being referred to with @samp{foo.a}. | |
12092 | Such constructs are not supported and must be avoided. In the future, | |
12093 | such constructs may be detected and treated as compilation errors. | |
12094 | ||
2fbebc71 JM |
12095 | @opindex fms-extensions |
12096 | Unless @option{-fms-extensions} is used, the unnamed field must be a | |
12097 | structure or union definition without a tag (for example, @samp{struct | |
12098 | @{ int a; @};}). If @option{-fms-extensions} is used, the field may | |
12099 | also be a definition with a tag such as @samp{struct foo @{ int a; | |
12100 | @};}, a reference to a previously defined structure or union such as | |
12101 | @samp{struct foo;}, or a reference to a @code{typedef} name for a | |
12102 | previously defined structure or union type. | |
12103 | ||
3d78f2e9 RH |
12104 | @node Thread-Local |
12105 | @section Thread-Local Storage | |
12106 | @cindex Thread-Local Storage | |
9217ef40 | 12107 | @cindex @acronym{TLS} |
3d78f2e9 RH |
12108 | @cindex __thread |
12109 | ||
9217ef40 RH |
12110 | Thread-local storage (@acronym{TLS}) is a mechanism by which variables |
12111 | are allocated such that there is one instance of the variable per extant | |
3d78f2e9 RH |
12112 | thread. The run-time model GCC uses to implement this originates |
12113 | in the IA-64 processor-specific ABI, but has since been migrated | |
12114 | to other processors as well. It requires significant support from | |
12115 | the linker (@command{ld}), dynamic linker (@command{ld.so}), and | |
12116 | system libraries (@file{libc.so} and @file{libpthread.so}), so it | |
9217ef40 | 12117 | is not available everywhere. |
3d78f2e9 RH |
12118 | |
12119 | At the user level, the extension is visible with a new storage | |
12120 | class keyword: @code{__thread}. For example: | |
12121 | ||
3ab51846 | 12122 | @smallexample |
3d78f2e9 RH |
12123 | __thread int i; |
12124 | extern __thread struct state s; | |
12125 | static __thread char *p; | |
3ab51846 | 12126 | @end smallexample |
3d78f2e9 RH |
12127 | |
12128 | The @code{__thread} specifier may be used alone, with the @code{extern} | |
12129 | or @code{static} specifiers, but with no other storage class specifier. | |
12130 | When used with @code{extern} or @code{static}, @code{__thread} must appear | |
12131 | immediately after the other storage class specifier. | |
12132 | ||
12133 | The @code{__thread} specifier may be applied to any global, file-scoped | |
244c2241 RH |
12134 | static, function-scoped static, or static data member of a class. It may |
12135 | not be applied to block-scoped automatic or non-static data member. | |
3d78f2e9 RH |
12136 | |
12137 | When the address-of operator is applied to a thread-local variable, it is | |
12138 | evaluated at run-time and returns the address of the current thread's | |
12139 | instance of that variable. An address so obtained may be used by any | |
12140 | thread. When a thread terminates, any pointers to thread-local variables | |
12141 | in that thread become invalid. | |
12142 | ||
12143 | No static initialization may refer to the address of a thread-local variable. | |
12144 | ||
244c2241 RH |
12145 | In C++, if an initializer is present for a thread-local variable, it must |
12146 | be a @var{constant-expression}, as defined in 5.19.2 of the ANSI/ISO C++ | |
12147 | standard. | |
3d78f2e9 RH |
12148 | |
12149 | See @uref{http://people.redhat.com/drepper/tls.pdf, | |
12150 | ELF Handling For Thread-Local Storage} for a detailed explanation of | |
12151 | the four thread-local storage addressing models, and how the run-time | |
12152 | is expected to function. | |
12153 | ||
9217ef40 RH |
12154 | @menu |
12155 | * C99 Thread-Local Edits:: | |
12156 | * C++98 Thread-Local Edits:: | |
12157 | @end menu | |
12158 | ||
12159 | @node C99 Thread-Local Edits | |
12160 | @subsection ISO/IEC 9899:1999 Edits for Thread-Local Storage | |
12161 | ||
12162 | The following are a set of changes to ISO/IEC 9899:1999 (aka C99) | |
12163 | that document the exact semantics of the language extension. | |
12164 | ||
12165 | @itemize @bullet | |
12166 | @item | |
12167 | @cite{5.1.2 Execution environments} | |
12168 | ||
12169 | Add new text after paragraph 1 | |
12170 | ||
12171 | @quotation | |
12172 | Within either execution environment, a @dfn{thread} is a flow of | |
12173 | control within a program. It is implementation defined whether | |
12174 | or not there may be more than one thread associated with a program. | |
12175 | It is implementation defined how threads beyond the first are | |
12176 | created, the name and type of the function called at thread | |
12177 | startup, and how threads may be terminated. However, objects | |
12178 | with thread storage duration shall be initialized before thread | |
12179 | startup. | |
12180 | @end quotation | |
12181 | ||
12182 | @item | |
12183 | @cite{6.2.4 Storage durations of objects} | |
12184 | ||
12185 | Add new text before paragraph 3 | |
12186 | ||
12187 | @quotation | |
12188 | An object whose identifier is declared with the storage-class | |
12189 | specifier @w{@code{__thread}} has @dfn{thread storage duration}. | |
12190 | Its lifetime is the entire execution of the thread, and its | |
12191 | stored value is initialized only once, prior to thread startup. | |
12192 | @end quotation | |
12193 | ||
12194 | @item | |
12195 | @cite{6.4.1 Keywords} | |
12196 | ||
12197 | Add @code{__thread}. | |
12198 | ||
12199 | @item | |
12200 | @cite{6.7.1 Storage-class specifiers} | |
12201 | ||
12202 | Add @code{__thread} to the list of storage class specifiers in | |
12203 | paragraph 1. | |
12204 | ||
12205 | Change paragraph 2 to | |
12206 | ||
12207 | @quotation | |
12208 | With the exception of @code{__thread}, at most one storage-class | |
12209 | specifier may be given [@dots{}]. The @code{__thread} specifier may | |
12210 | be used alone, or immediately following @code{extern} or | |
12211 | @code{static}. | |
12212 | @end quotation | |
12213 | ||
12214 | Add new text after paragraph 6 | |
12215 | ||
12216 | @quotation | |
12217 | The declaration of an identifier for a variable that has | |
12218 | block scope that specifies @code{__thread} shall also | |
12219 | specify either @code{extern} or @code{static}. | |
12220 | ||
12221 | The @code{__thread} specifier shall be used only with | |
12222 | variables. | |
12223 | @end quotation | |
12224 | @end itemize | |
12225 | ||
12226 | @node C++98 Thread-Local Edits | |
12227 | @subsection ISO/IEC 14882:1998 Edits for Thread-Local Storage | |
12228 | ||
12229 | The following are a set of changes to ISO/IEC 14882:1998 (aka C++98) | |
12230 | that document the exact semantics of the language extension. | |
12231 | ||
12232 | @itemize @bullet | |
8d23a2c8 | 12233 | @item |
9217ef40 RH |
12234 | @b{[intro.execution]} |
12235 | ||
12236 | New text after paragraph 4 | |
12237 | ||
12238 | @quotation | |
12239 | A @dfn{thread} is a flow of control within the abstract machine. | |
12240 | It is implementation defined whether or not there may be more than | |
12241 | one thread. | |
12242 | @end quotation | |
12243 | ||
12244 | New text after paragraph 7 | |
12245 | ||
12246 | @quotation | |
95b1627e | 12247 | It is unspecified whether additional action must be taken to |
9217ef40 RH |
12248 | ensure when and whether side effects are visible to other threads. |
12249 | @end quotation | |
12250 | ||
12251 | @item | |
12252 | @b{[lex.key]} | |
12253 | ||
12254 | Add @code{__thread}. | |
12255 | ||
12256 | @item | |
12257 | @b{[basic.start.main]} | |
12258 | ||
12259 | Add after paragraph 5 | |
12260 | ||
12261 | @quotation | |
12262 | The thread that begins execution at the @code{main} function is called | |
95b1627e | 12263 | the @dfn{main thread}. It is implementation defined how functions |
9217ef40 RH |
12264 | beginning threads other than the main thread are designated or typed. |
12265 | A function so designated, as well as the @code{main} function, is called | |
12266 | a @dfn{thread startup function}. It is implementation defined what | |
12267 | happens if a thread startup function returns. It is implementation | |
12268 | defined what happens to other threads when any thread calls @code{exit}. | |
12269 | @end quotation | |
12270 | ||
12271 | @item | |
12272 | @b{[basic.start.init]} | |
12273 | ||
12274 | Add after paragraph 4 | |
12275 | ||
12276 | @quotation | |
12277 | The storage for an object of thread storage duration shall be | |
c0478a66 | 12278 | statically initialized before the first statement of the thread startup |
9217ef40 RH |
12279 | function. An object of thread storage duration shall not require |
12280 | dynamic initialization. | |
12281 | @end quotation | |
12282 | ||
12283 | @item | |
12284 | @b{[basic.start.term]} | |
12285 | ||
12286 | Add after paragraph 3 | |
12287 | ||
12288 | @quotation | |
244c2241 RH |
12289 | The type of an object with thread storage duration shall not have a |
12290 | non-trivial destructor, nor shall it be an array type whose elements | |
12291 | (directly or indirectly) have non-trivial destructors. | |
9217ef40 RH |
12292 | @end quotation |
12293 | ||
12294 | @item | |
12295 | @b{[basic.stc]} | |
12296 | ||
12297 | Add ``thread storage duration'' to the list in paragraph 1. | |
12298 | ||
12299 | Change paragraph 2 | |
12300 | ||
12301 | @quotation | |
12302 | Thread, static, and automatic storage durations are associated with | |
12303 | objects introduced by declarations [@dots{}]. | |
12304 | @end quotation | |
12305 | ||
12306 | Add @code{__thread} to the list of specifiers in paragraph 3. | |
12307 | ||
12308 | @item | |
12309 | @b{[basic.stc.thread]} | |
12310 | ||
12311 | New section before @b{[basic.stc.static]} | |
12312 | ||
12313 | @quotation | |
63519d23 | 12314 | The keyword @code{__thread} applied to a non-local object gives the |
9217ef40 RH |
12315 | object thread storage duration. |
12316 | ||
12317 | A local variable or class data member declared both @code{static} | |
12318 | and @code{__thread} gives the variable or member thread storage | |
12319 | duration. | |
12320 | @end quotation | |
12321 | ||
12322 | @item | |
12323 | @b{[basic.stc.static]} | |
12324 | ||
12325 | Change paragraph 1 | |
12326 | ||
12327 | @quotation | |
12328 | All objects which have neither thread storage duration, dynamic | |
12329 | storage duration nor are local [@dots{}]. | |
12330 | @end quotation | |
12331 | ||
12332 | @item | |
12333 | @b{[dcl.stc]} | |
12334 | ||
12335 | Add @code{__thread} to the list in paragraph 1. | |
12336 | ||
12337 | Change paragraph 1 | |
12338 | ||
12339 | @quotation | |
12340 | With the exception of @code{__thread}, at most one | |
12341 | @var{storage-class-specifier} shall appear in a given | |
12342 | @var{decl-specifier-seq}. The @code{__thread} specifier may | |
12343 | be used alone, or immediately following the @code{extern} or | |
12344 | @code{static} specifiers. [@dots{}] | |
12345 | @end quotation | |
12346 | ||
12347 | Add after paragraph 5 | |
12348 | ||
12349 | @quotation | |
12350 | The @code{__thread} specifier can be applied only to the names of objects | |
12351 | and to anonymous unions. | |
12352 | @end quotation | |
12353 | ||
12354 | @item | |
12355 | @b{[class.mem]} | |
12356 | ||
12357 | Add after paragraph 6 | |
12358 | ||
12359 | @quotation | |
12360 | Non-@code{static} members shall not be @code{__thread}. | |
12361 | @end quotation | |
12362 | @end itemize | |
12363 | ||
f7fd775f JW |
12364 | @node Binary constants |
12365 | @section Binary constants using the @samp{0b} prefix | |
12366 | @cindex Binary constants using the @samp{0b} prefix | |
12367 | ||
12368 | Integer constants can be written as binary constants, consisting of a | |
12369 | sequence of @samp{0} and @samp{1} digits, prefixed by @samp{0b} or | |
12370 | @samp{0B}. This is particularly useful in environments that operate a | |
12371 | lot on the bit-level (like microcontrollers). | |
12372 | ||
12373 | The following statements are identical: | |
12374 | ||
12375 | @smallexample | |
12376 | i = 42; | |
12377 | i = 0x2a; | |
12378 | i = 052; | |
12379 | i = 0b101010; | |
12380 | @end smallexample | |
12381 | ||
12382 | The type of these constants follows the same rules as for octal or | |
12383 | hexadecimal integer constants, so suffixes like @samp{L} or @samp{UL} | |
12384 | can be applied. | |
12385 | ||
c1f7febf RK |
12386 | @node C++ Extensions |
12387 | @chapter Extensions to the C++ Language | |
12388 | @cindex extensions, C++ language | |
12389 | @cindex C++ language extensions | |
12390 | ||
12391 | The GNU compiler provides these extensions to the C++ language (and you | |
12392 | can also use most of the C language extensions in your C++ programs). If you | |
12393 | want to write code that checks whether these features are available, you can | |
12394 | test for the GNU compiler the same way as for C programs: check for a | |
12395 | predefined macro @code{__GNUC__}. You can also use @code{__GNUG__} to | |
48795525 GP |
12396 | test specifically for GNU C++ (@pxref{Common Predefined Macros,, |
12397 | Predefined Macros,cpp,The GNU C Preprocessor}). | |
c1f7febf RK |
12398 | |
12399 | @menu | |
6ccde948 | 12400 | * Volatiles:: What constitutes an access to a volatile object. |
49419c8f | 12401 | * Restricted Pointers:: C99 restricted pointers and references. |
7a81cf7f | 12402 | * Vague Linkage:: Where G++ puts inlines, vtables and such. |
c1f7febf | 12403 | * C++ Interface:: You can use a single C++ header file for both |
e6f3b89d | 12404 | declarations and definitions. |
c1f7febf | 12405 | * Template Instantiation:: Methods for ensuring that exactly one copy of |
e6f3b89d | 12406 | each needed template instantiation is emitted. |
0ded1f18 JM |
12407 | * Bound member functions:: You can extract a function pointer to the |
12408 | method denoted by a @samp{->*} or @samp{.*} expression. | |
e6f3b89d | 12409 | * C++ Attributes:: Variable, function, and type attributes for C++ only. |
664a90c0 | 12410 | * Namespace Association:: Strong using-directives for namespace association. |
cb68ec50 | 12411 | * Type Traits:: Compiler support for type traits |
1f730ff7 | 12412 | * Java Exceptions:: Tweaking exception handling to work with Java. |
90ea7324 | 12413 | * Deprecated Features:: Things will disappear from g++. |
e6f3b89d | 12414 | * Backwards Compatibility:: Compatibilities with earlier definitions of C++. |
c1f7febf RK |
12415 | @end menu |
12416 | ||
02cac427 NS |
12417 | @node Volatiles |
12418 | @section When is a Volatile Object Accessed? | |
12419 | @cindex accessing volatiles | |
12420 | @cindex volatile read | |
12421 | @cindex volatile write | |
12422 | @cindex volatile access | |
12423 | ||
767094dd JM |
12424 | Both the C and C++ standard have the concept of volatile objects. These |
12425 | are normally accessed by pointers and used for accessing hardware. The | |
a9e64c63 EB |
12426 | standards encourage compilers to refrain from optimizations concerning |
12427 | accesses to volatile objects. The C standard leaves it implementation | |
12428 | defined as to what constitutes a volatile access. The C++ standard omits | |
12429 | to specify this, except to say that C++ should behave in a similar manner | |
767094dd | 12430 | to C with respect to volatiles, where possible. The minimum either |
8117da65 | 12431 | standard specifies is that at a sequence point all previous accesses to |
02cac427 | 12432 | volatile objects have stabilized and no subsequent accesses have |
767094dd | 12433 | occurred. Thus an implementation is free to reorder and combine |
02cac427 | 12434 | volatile accesses which occur between sequence points, but cannot do so |
767094dd | 12435 | for accesses across a sequence point. The use of volatiles does not |
02cac427 NS |
12436 | allow you to violate the restriction on updating objects multiple times |
12437 | within a sequence point. | |
12438 | ||
a9e64c63 | 12439 | @xref{Qualifiers implementation, , Volatile qualifier and the C compiler}. |
02cac427 | 12440 | |
a9e64c63 | 12441 | The behavior differs slightly between C and C++ in the non-obvious cases: |
02cac427 | 12442 | |
3ab51846 | 12443 | @smallexample |
c771326b | 12444 | volatile int *src = @var{somevalue}; |
02cac427 | 12445 | *src; |
3ab51846 | 12446 | @end smallexample |
02cac427 | 12447 | |
a9e64c63 EB |
12448 | With C, such expressions are rvalues, and GCC interprets this either as a |
12449 | read of the volatile object being pointed to or only as request to evaluate | |
12450 | the side-effects. The C++ standard specifies that such expressions do not | |
12451 | undergo lvalue to rvalue conversion, and that the type of the dereferenced | |
767094dd | 12452 | object may be incomplete. The C++ standard does not specify explicitly |
a9e64c63 | 12453 | that it is this lvalue to rvalue conversion which may be responsible for |
767094dd JM |
12454 | causing an access. However, there is reason to believe that it is, |
12455 | because otherwise certain simple expressions become undefined. However, | |
f0523f02 | 12456 | because it would surprise most programmers, G++ treats dereferencing a |
a9e64c63 | 12457 | pointer to volatile object of complete type when the value is unused as |
0ee2ea09 | 12458 | GCC would do for an equivalent type in C@. When the object has incomplete |
a9e64c63 EB |
12459 | type, G++ issues a warning; if you wish to force an error, you must |
12460 | force a conversion to rvalue with, for instance, a static cast. | |
02cac427 | 12461 | |
f0523f02 | 12462 | When using a reference to volatile, G++ does not treat equivalent |
02cac427 | 12463 | expressions as accesses to volatiles, but instead issues a warning that |
767094dd | 12464 | no volatile is accessed. The rationale for this is that otherwise it |
02cac427 NS |
12465 | becomes difficult to determine where volatile access occur, and not |
12466 | possible to ignore the return value from functions returning volatile | |
767094dd | 12467 | references. Again, if you wish to force a read, cast the reference to |
02cac427 NS |
12468 | an rvalue. |
12469 | ||
535233a8 NS |
12470 | @node Restricted Pointers |
12471 | @section Restricting Pointer Aliasing | |
12472 | @cindex restricted pointers | |
12473 | @cindex restricted references | |
12474 | @cindex restricted this pointer | |
12475 | ||
2dd76960 | 12476 | As with the C front end, G++ understands the C99 feature of restricted pointers, |
535233a8 | 12477 | specified with the @code{__restrict__}, or @code{__restrict} type |
767094dd | 12478 | qualifier. Because you cannot compile C++ by specifying the @option{-std=c99} |
535233a8 NS |
12479 | language flag, @code{restrict} is not a keyword in C++. |
12480 | ||
12481 | In addition to allowing restricted pointers, you can specify restricted | |
12482 | references, which indicate that the reference is not aliased in the local | |
12483 | context. | |
12484 | ||
3ab51846 | 12485 | @smallexample |
535233a8 NS |
12486 | void fn (int *__restrict__ rptr, int &__restrict__ rref) |
12487 | @{ | |
0d893a63 | 12488 | /* @r{@dots{}} */ |
535233a8 | 12489 | @} |
3ab51846 | 12490 | @end smallexample |
535233a8 NS |
12491 | |
12492 | @noindent | |
12493 | In the body of @code{fn}, @var{rptr} points to an unaliased integer and | |
12494 | @var{rref} refers to a (different) unaliased integer. | |
12495 | ||
12496 | You may also specify whether a member function's @var{this} pointer is | |
12497 | unaliased by using @code{__restrict__} as a member function qualifier. | |
12498 | ||
3ab51846 | 12499 | @smallexample |
535233a8 NS |
12500 | void T::fn () __restrict__ |
12501 | @{ | |
0d893a63 | 12502 | /* @r{@dots{}} */ |
535233a8 | 12503 | @} |
3ab51846 | 12504 | @end smallexample |
535233a8 NS |
12505 | |
12506 | @noindent | |
12507 | Within the body of @code{T::fn}, @var{this} will have the effective | |
767094dd | 12508 | definition @code{T *__restrict__ const this}. Notice that the |
535233a8 NS |
12509 | interpretation of a @code{__restrict__} member function qualifier is |
12510 | different to that of @code{const} or @code{volatile} qualifier, in that it | |
767094dd | 12511 | is applied to the pointer rather than the object. This is consistent with |
535233a8 NS |
12512 | other compilers which implement restricted pointers. |
12513 | ||
12514 | As with all outermost parameter qualifiers, @code{__restrict__} is | |
767094dd | 12515 | ignored in function definition matching. This means you only need to |
535233a8 NS |
12516 | specify @code{__restrict__} in a function definition, rather than |
12517 | in a function prototype as well. | |
12518 | ||
7a81cf7f JM |
12519 | @node Vague Linkage |
12520 | @section Vague Linkage | |
12521 | @cindex vague linkage | |
12522 | ||
12523 | There are several constructs in C++ which require space in the object | |
12524 | file but are not clearly tied to a single translation unit. We say that | |
12525 | these constructs have ``vague linkage''. Typically such constructs are | |
12526 | emitted wherever they are needed, though sometimes we can be more | |
12527 | clever. | |
12528 | ||
12529 | @table @asis | |
12530 | @item Inline Functions | |
12531 | Inline functions are typically defined in a header file which can be | |
12532 | included in many different compilations. Hopefully they can usually be | |
12533 | inlined, but sometimes an out-of-line copy is necessary, if the address | |
12534 | of the function is taken or if inlining fails. In general, we emit an | |
12535 | out-of-line copy in all translation units where one is needed. As an | |
12536 | exception, we only emit inline virtual functions with the vtable, since | |
12537 | it will always require a copy. | |
12538 | ||
12539 | Local static variables and string constants used in an inline function | |
12540 | are also considered to have vague linkage, since they must be shared | |
12541 | between all inlined and out-of-line instances of the function. | |
12542 | ||
12543 | @item VTables | |
12544 | @cindex vtable | |
12545 | C++ virtual functions are implemented in most compilers using a lookup | |
12546 | table, known as a vtable. The vtable contains pointers to the virtual | |
12547 | functions provided by a class, and each object of the class contains a | |
12548 | pointer to its vtable (or vtables, in some multiple-inheritance | |
12549 | situations). If the class declares any non-inline, non-pure virtual | |
12550 | functions, the first one is chosen as the ``key method'' for the class, | |
12551 | and the vtable is only emitted in the translation unit where the key | |
12552 | method is defined. | |
12553 | ||
12554 | @emph{Note:} If the chosen key method is later defined as inline, the | |
12555 | vtable will still be emitted in every translation unit which defines it. | |
12556 | Make sure that any inline virtuals are declared inline in the class | |
12557 | body, even if they are not defined there. | |
12558 | ||
12559 | @item type_info objects | |
12560 | @cindex type_info | |
12561 | @cindex RTTI | |
12562 | C++ requires information about types to be written out in order to | |
12563 | implement @samp{dynamic_cast}, @samp{typeid} and exception handling. | |
12564 | For polymorphic classes (classes with virtual functions), the type_info | |
12565 | object is written out along with the vtable so that @samp{dynamic_cast} | |
12566 | can determine the dynamic type of a class object at runtime. For all | |
12567 | other types, we write out the type_info object when it is used: when | |
12568 | applying @samp{typeid} to an expression, throwing an object, or | |
12569 | referring to a type in a catch clause or exception specification. | |
12570 | ||
12571 | @item Template Instantiations | |
12572 | Most everything in this section also applies to template instantiations, | |
12573 | but there are other options as well. | |
12574 | @xref{Template Instantiation,,Where's the Template?}. | |
12575 | ||
12576 | @end table | |
12577 | ||
12578 | When used with GNU ld version 2.8 or later on an ELF system such as | |
95fef11f | 12579 | GNU/Linux or Solaris 2, or on Microsoft Windows, duplicate copies of |
7a81cf7f JM |
12580 | these constructs will be discarded at link time. This is known as |
12581 | COMDAT support. | |
12582 | ||
12583 | On targets that don't support COMDAT, but do support weak symbols, GCC | |
12584 | will use them. This way one copy will override all the others, but | |
12585 | the unused copies will still take up space in the executable. | |
12586 | ||
12587 | For targets which do not support either COMDAT or weak symbols, | |
12588 | most entities with vague linkage will be emitted as local symbols to | |
12589 | avoid duplicate definition errors from the linker. This will not happen | |
12590 | for local statics in inlines, however, as having multiple copies will | |
12591 | almost certainly break things. | |
12592 | ||
12593 | @xref{C++ Interface,,Declarations and Definitions in One Header}, for | |
12594 | another way to control placement of these constructs. | |
12595 | ||
c1f7febf | 12596 | @node C++ Interface |
fc72b380 | 12597 | @section #pragma interface and implementation |
c1f7febf RK |
12598 | |
12599 | @cindex interface and implementation headers, C++ | |
12600 | @cindex C++ interface and implementation headers | |
c1f7febf | 12601 | @cindex pragmas, interface and implementation |
c1f7febf | 12602 | |
fc72b380 JM |
12603 | @code{#pragma interface} and @code{#pragma implementation} provide the |
12604 | user with a way of explicitly directing the compiler to emit entities | |
12605 | with vague linkage (and debugging information) in a particular | |
12606 | translation unit. | |
c1f7febf | 12607 | |
fc72b380 JM |
12608 | @emph{Note:} As of GCC 2.7.2, these @code{#pragma}s are not useful in |
12609 | most cases, because of COMDAT support and the ``key method'' heuristic | |
12610 | mentioned in @ref{Vague Linkage}. Using them can actually cause your | |
27ef2cdd | 12611 | program to grow due to unnecessary out-of-line copies of inline |
fc72b380 JM |
12612 | functions. Currently (3.4) the only benefit of these |
12613 | @code{#pragma}s is reduced duplication of debugging information, and | |
12614 | that should be addressed soon on DWARF 2 targets with the use of | |
12615 | COMDAT groups. | |
c1f7febf RK |
12616 | |
12617 | @table @code | |
12618 | @item #pragma interface | |
12619 | @itemx #pragma interface "@var{subdir}/@var{objects}.h" | |
12620 | @kindex #pragma interface | |
12621 | Use this directive in @emph{header files} that define object classes, to save | |
12622 | space in most of the object files that use those classes. Normally, | |
12623 | local copies of certain information (backup copies of inline member | |
12624 | functions, debugging information, and the internal tables that implement | |
12625 | virtual functions) must be kept in each object file that includes class | |
12626 | definitions. You can use this pragma to avoid such duplication. When a | |
12627 | header file containing @samp{#pragma interface} is included in a | |
12628 | compilation, this auxiliary information will not be generated (unless | |
12629 | the main input source file itself uses @samp{#pragma implementation}). | |
12630 | Instead, the object files will contain references to be resolved at link | |
12631 | time. | |
12632 | ||
12633 | The second form of this directive is useful for the case where you have | |
12634 | multiple headers with the same name in different directories. If you | |
12635 | use this form, you must specify the same string to @samp{#pragma | |
12636 | implementation}. | |
12637 | ||
12638 | @item #pragma implementation | |
12639 | @itemx #pragma implementation "@var{objects}.h" | |
12640 | @kindex #pragma implementation | |
12641 | Use this pragma in a @emph{main input file}, when you want full output from | |
12642 | included header files to be generated (and made globally visible). The | |
12643 | included header file, in turn, should use @samp{#pragma interface}. | |
12644 | Backup copies of inline member functions, debugging information, and the | |
12645 | internal tables used to implement virtual functions are all generated in | |
12646 | implementation files. | |
12647 | ||
12648 | @cindex implied @code{#pragma implementation} | |
12649 | @cindex @code{#pragma implementation}, implied | |
12650 | @cindex naming convention, implementation headers | |
12651 | If you use @samp{#pragma implementation} with no argument, it applies to | |
12652 | an include file with the same basename@footnote{A file's @dfn{basename} | |
12653 | was the name stripped of all leading path information and of trailing | |
12654 | suffixes, such as @samp{.h} or @samp{.C} or @samp{.cc}.} as your source | |
12655 | file. For example, in @file{allclass.cc}, giving just | |
12656 | @samp{#pragma implementation} | |
12657 | by itself is equivalent to @samp{#pragma implementation "allclass.h"}. | |
12658 | ||
12659 | In versions of GNU C++ prior to 2.6.0 @file{allclass.h} was treated as | |
12660 | an implementation file whenever you would include it from | |
12661 | @file{allclass.cc} even if you never specified @samp{#pragma | |
12662 | implementation}. This was deemed to be more trouble than it was worth, | |
12663 | however, and disabled. | |
12664 | ||
c1f7febf RK |
12665 | Use the string argument if you want a single implementation file to |
12666 | include code from multiple header files. (You must also use | |
12667 | @samp{#include} to include the header file; @samp{#pragma | |
12668 | implementation} only specifies how to use the file---it doesn't actually | |
12669 | include it.) | |
12670 | ||
12671 | There is no way to split up the contents of a single header file into | |
12672 | multiple implementation files. | |
12673 | @end table | |
12674 | ||
12675 | @cindex inlining and C++ pragmas | |
12676 | @cindex C++ pragmas, effect on inlining | |
12677 | @cindex pragmas in C++, effect on inlining | |
12678 | @samp{#pragma implementation} and @samp{#pragma interface} also have an | |
12679 | effect on function inlining. | |
12680 | ||
12681 | If you define a class in a header file marked with @samp{#pragma | |
fc72b380 JM |
12682 | interface}, the effect on an inline function defined in that class is |
12683 | similar to an explicit @code{extern} declaration---the compiler emits | |
12684 | no code at all to define an independent version of the function. Its | |
12685 | definition is used only for inlining with its callers. | |
c1f7febf | 12686 | |
84330467 | 12687 | @opindex fno-implement-inlines |
c1f7febf RK |
12688 | Conversely, when you include the same header file in a main source file |
12689 | that declares it as @samp{#pragma implementation}, the compiler emits | |
12690 | code for the function itself; this defines a version of the function | |
12691 | that can be found via pointers (or by callers compiled without | |
12692 | inlining). If all calls to the function can be inlined, you can avoid | |
84330467 | 12693 | emitting the function by compiling with @option{-fno-implement-inlines}. |
c1f7febf RK |
12694 | If any calls were not inlined, you will get linker errors. |
12695 | ||
12696 | @node Template Instantiation | |
12697 | @section Where's the Template? | |
c1f7febf RK |
12698 | @cindex template instantiation |
12699 | ||
12700 | C++ templates are the first language feature to require more | |
12701 | intelligence from the environment than one usually finds on a UNIX | |
12702 | system. Somehow the compiler and linker have to make sure that each | |
12703 | template instance occurs exactly once in the executable if it is needed, | |
12704 | and not at all otherwise. There are two basic approaches to this | |
962e6e00 | 12705 | problem, which are referred to as the Borland model and the Cfront model. |
c1f7febf RK |
12706 | |
12707 | @table @asis | |
12708 | @item Borland model | |
12709 | Borland C++ solved the template instantiation problem by adding the code | |
469b759e JM |
12710 | equivalent of common blocks to their linker; the compiler emits template |
12711 | instances in each translation unit that uses them, and the linker | |
12712 | collapses them together. The advantage of this model is that the linker | |
12713 | only has to consider the object files themselves; there is no external | |
12714 | complexity to worry about. This disadvantage is that compilation time | |
12715 | is increased because the template code is being compiled repeatedly. | |
12716 | Code written for this model tends to include definitions of all | |
12717 | templates in the header file, since they must be seen to be | |
12718 | instantiated. | |
c1f7febf RK |
12719 | |
12720 | @item Cfront model | |
12721 | The AT&T C++ translator, Cfront, solved the template instantiation | |
12722 | problem by creating the notion of a template repository, an | |
469b759e JM |
12723 | automatically maintained place where template instances are stored. A |
12724 | more modern version of the repository works as follows: As individual | |
12725 | object files are built, the compiler places any template definitions and | |
12726 | instantiations encountered in the repository. At link time, the link | |
12727 | wrapper adds in the objects in the repository and compiles any needed | |
12728 | instances that were not previously emitted. The advantages of this | |
12729 | model are more optimal compilation speed and the ability to use the | |
12730 | system linker; to implement the Borland model a compiler vendor also | |
c1f7febf | 12731 | needs to replace the linker. The disadvantages are vastly increased |
469b759e JM |
12732 | complexity, and thus potential for error; for some code this can be |
12733 | just as transparent, but in practice it can been very difficult to build | |
c1f7febf | 12734 | multiple programs in one directory and one program in multiple |
469b759e JM |
12735 | directories. Code written for this model tends to separate definitions |
12736 | of non-inline member templates into a separate file, which should be | |
12737 | compiled separately. | |
c1f7febf RK |
12738 | @end table |
12739 | ||
469b759e | 12740 | When used with GNU ld version 2.8 or later on an ELF system such as |
2dd76960 JM |
12741 | GNU/Linux or Solaris 2, or on Microsoft Windows, G++ supports the |
12742 | Borland model. On other systems, G++ implements neither automatic | |
a4b3b54a | 12743 | model. |
469b759e | 12744 | |
2dd76960 | 12745 | A future version of G++ will support a hybrid model whereby the compiler |
469b759e JM |
12746 | will emit any instantiations for which the template definition is |
12747 | included in the compile, and store template definitions and | |
12748 | instantiation context information into the object file for the rest. | |
12749 | The link wrapper will extract that information as necessary and invoke | |
12750 | the compiler to produce the remaining instantiations. The linker will | |
12751 | then combine duplicate instantiations. | |
12752 | ||
12753 | In the mean time, you have the following options for dealing with | |
12754 | template instantiations: | |
c1f7febf RK |
12755 | |
12756 | @enumerate | |
d863830b | 12757 | @item |
84330467 JM |
12758 | @opindex frepo |
12759 | Compile your template-using code with @option{-frepo}. The compiler will | |
d863830b JL |
12760 | generate files with the extension @samp{.rpo} listing all of the |
12761 | template instantiations used in the corresponding object files which | |
12762 | could be instantiated there; the link wrapper, @samp{collect2}, will | |
12763 | then update the @samp{.rpo} files to tell the compiler where to place | |
12764 | those instantiations and rebuild any affected object files. The | |
12765 | link-time overhead is negligible after the first pass, as the compiler | |
12766 | will continue to place the instantiations in the same files. | |
12767 | ||
12768 | This is your best option for application code written for the Borland | |
12769 | model, as it will just work. Code written for the Cfront model will | |
12770 | need to be modified so that the template definitions are available at | |
12771 | one or more points of instantiation; usually this is as simple as adding | |
12772 | @code{#include <tmethods.cc>} to the end of each template header. | |
12773 | ||
12774 | For library code, if you want the library to provide all of the template | |
12775 | instantiations it needs, just try to link all of its object files | |
12776 | together; the link will fail, but cause the instantiations to be | |
12777 | generated as a side effect. Be warned, however, that this may cause | |
12778 | conflicts if multiple libraries try to provide the same instantiations. | |
12779 | For greater control, use explicit instantiation as described in the next | |
12780 | option. | |
12781 | ||
c1f7febf | 12782 | @item |
84330467 JM |
12783 | @opindex fno-implicit-templates |
12784 | Compile your code with @option{-fno-implicit-templates} to disable the | |
c1f7febf RK |
12785 | implicit generation of template instances, and explicitly instantiate |
12786 | all the ones you use. This approach requires more knowledge of exactly | |
12787 | which instances you need than do the others, but it's less | |
12788 | mysterious and allows greater control. You can scatter the explicit | |
12789 | instantiations throughout your program, perhaps putting them in the | |
12790 | translation units where the instances are used or the translation units | |
12791 | that define the templates themselves; you can put all of the explicit | |
12792 | instantiations you need into one big file; or you can create small files | |
12793 | like | |
12794 | ||
3ab51846 | 12795 | @smallexample |
c1f7febf RK |
12796 | #include "Foo.h" |
12797 | #include "Foo.cc" | |
12798 | ||
12799 | template class Foo<int>; | |
12800 | template ostream& operator << | |
12801 | (ostream&, const Foo<int>&); | |
3ab51846 | 12802 | @end smallexample |
c1f7febf RK |
12803 | |
12804 | for each of the instances you need, and create a template instantiation | |
12805 | library from those. | |
12806 | ||
12807 | If you are using Cfront-model code, you can probably get away with not | |
84330467 | 12808 | using @option{-fno-implicit-templates} when compiling files that don't |
c1f7febf RK |
12809 | @samp{#include} the member template definitions. |
12810 | ||
12811 | If you use one big file to do the instantiations, you may want to | |
84330467 | 12812 | compile it without @option{-fno-implicit-templates} so you get all of the |
c1f7febf RK |
12813 | instances required by your explicit instantiations (but not by any |
12814 | other files) without having to specify them as well. | |
12815 | ||
2dd76960 | 12816 | G++ has extended the template instantiation syntax given in the ISO |
6d9c4c83 | 12817 | standard to allow forward declaration of explicit instantiations |
4003d7f9 | 12818 | (with @code{extern}), instantiation of the compiler support data for a |
e979f9e8 | 12819 | template class (i.e.@: the vtable) without instantiating any of its |
4003d7f9 JM |
12820 | members (with @code{inline}), and instantiation of only the static data |
12821 | members of a template class, without the support data or member | |
12822 | functions (with (@code{static}): | |
c1f7febf | 12823 | |
3ab51846 | 12824 | @smallexample |
c1f7febf | 12825 | extern template int max (int, int); |
c1f7febf | 12826 | inline template class Foo<int>; |
4003d7f9 | 12827 | static template class Foo<int>; |
3ab51846 | 12828 | @end smallexample |
c1f7febf RK |
12829 | |
12830 | @item | |
2dd76960 | 12831 | Do nothing. Pretend G++ does implement automatic instantiation |
c1f7febf RK |
12832 | management. Code written for the Borland model will work fine, but |
12833 | each translation unit will contain instances of each of the templates it | |
12834 | uses. In a large program, this can lead to an unacceptable amount of code | |
12835 | duplication. | |
c1f7febf RK |
12836 | @end enumerate |
12837 | ||
0ded1f18 JM |
12838 | @node Bound member functions |
12839 | @section Extracting the function pointer from a bound pointer to member function | |
0ded1f18 JM |
12840 | @cindex pmf |
12841 | @cindex pointer to member function | |
12842 | @cindex bound pointer to member function | |
12843 | ||
12844 | In C++, pointer to member functions (PMFs) are implemented using a wide | |
12845 | pointer of sorts to handle all the possible call mechanisms; the PMF | |
12846 | needs to store information about how to adjust the @samp{this} pointer, | |
12847 | and if the function pointed to is virtual, where to find the vtable, and | |
12848 | where in the vtable to look for the member function. If you are using | |
12849 | PMFs in an inner loop, you should really reconsider that decision. If | |
12850 | that is not an option, you can extract the pointer to the function that | |
12851 | would be called for a given object/PMF pair and call it directly inside | |
12852 | the inner loop, to save a bit of time. | |
12853 | ||
12854 | Note that you will still be paying the penalty for the call through a | |
12855 | function pointer; on most modern architectures, such a call defeats the | |
161d7b59 | 12856 | branch prediction features of the CPU@. This is also true of normal |
0ded1f18 JM |
12857 | virtual function calls. |
12858 | ||
12859 | The syntax for this extension is | |
12860 | ||
3ab51846 | 12861 | @smallexample |
0ded1f18 JM |
12862 | extern A a; |
12863 | extern int (A::*fp)(); | |
12864 | typedef int (*fptr)(A *); | |
12865 | ||
12866 | fptr p = (fptr)(a.*fp); | |
3ab51846 | 12867 | @end smallexample |
0ded1f18 | 12868 | |
e979f9e8 | 12869 | For PMF constants (i.e.@: expressions of the form @samp{&Klasse::Member}), |
767094dd | 12870 | no object is needed to obtain the address of the function. They can be |
0fb6bbf5 ML |
12871 | converted to function pointers directly: |
12872 | ||
3ab51846 | 12873 | @smallexample |
0fb6bbf5 | 12874 | fptr p1 = (fptr)(&A::foo); |
3ab51846 | 12875 | @end smallexample |
0fb6bbf5 | 12876 | |
84330467 JM |
12877 | @opindex Wno-pmf-conversions |
12878 | You must specify @option{-Wno-pmf-conversions} to use this extension. | |
0ded1f18 | 12879 | |
5c25e11d PE |
12880 | @node C++ Attributes |
12881 | @section C++-Specific Variable, Function, and Type Attributes | |
12882 | ||
12883 | Some attributes only make sense for C++ programs. | |
12884 | ||
12885 | @table @code | |
12886 | @item init_priority (@var{priority}) | |
12887 | @cindex init_priority attribute | |
12888 | ||
12889 | ||
12890 | In Standard C++, objects defined at namespace scope are guaranteed to be | |
12891 | initialized in an order in strict accordance with that of their definitions | |
12892 | @emph{in a given translation unit}. No guarantee is made for initializations | |
12893 | across translation units. However, GNU C++ allows users to control the | |
3844cd2e | 12894 | order of initialization of objects defined at namespace scope with the |
5c25e11d PE |
12895 | @code{init_priority} attribute by specifying a relative @var{priority}, |
12896 | a constant integral expression currently bounded between 101 and 65535 | |
12897 | inclusive. Lower numbers indicate a higher priority. | |
12898 | ||
12899 | In the following example, @code{A} would normally be created before | |
12900 | @code{B}, but the @code{init_priority} attribute has reversed that order: | |
12901 | ||
478c9e72 | 12902 | @smallexample |
5c25e11d PE |
12903 | Some_Class A __attribute__ ((init_priority (2000))); |
12904 | Some_Class B __attribute__ ((init_priority (543))); | |
478c9e72 | 12905 | @end smallexample |
5c25e11d PE |
12906 | |
12907 | @noindent | |
12908 | Note that the particular values of @var{priority} do not matter; only their | |
12909 | relative ordering. | |
12910 | ||
60c87482 BM |
12911 | @item java_interface |
12912 | @cindex java_interface attribute | |
12913 | ||
02f52e19 | 12914 | This type attribute informs C++ that the class is a Java interface. It may |
60c87482 | 12915 | only be applied to classes declared within an @code{extern "Java"} block. |
02f52e19 AJ |
12916 | Calls to methods declared in this interface will be dispatched using GCJ's |
12917 | interface table mechanism, instead of regular virtual table dispatch. | |
60c87482 | 12918 | |
5c25e11d PE |
12919 | @end table |
12920 | ||
38bb2b65 | 12921 | See also @ref{Namespace Association}. |
86098eb8 | 12922 | |
664a90c0 JM |
12923 | @node Namespace Association |
12924 | @section Namespace Association | |
86098eb8 | 12925 | |
fea77ed9 MM |
12926 | @strong{Caution:} The semantics of this extension are not fully |
12927 | defined. Users should refrain from using this extension as its | |
12928 | semantics may change subtly over time. It is possible that this | |
664a90c0 | 12929 | extension will be removed in future versions of G++. |
fea77ed9 | 12930 | |
86098eb8 JM |
12931 | A using-directive with @code{__attribute ((strong))} is stronger |
12932 | than a normal using-directive in two ways: | |
12933 | ||
12934 | @itemize @bullet | |
12935 | @item | |
664a90c0 JM |
12936 | Templates from the used namespace can be specialized and explicitly |
12937 | instantiated as though they were members of the using namespace. | |
86098eb8 JM |
12938 | |
12939 | @item | |
12940 | The using namespace is considered an associated namespace of all | |
12941 | templates in the used namespace for purposes of argument-dependent | |
12942 | name lookup. | |
12943 | @end itemize | |
12944 | ||
664a90c0 JM |
12945 | The used namespace must be nested within the using namespace so that |
12946 | normal unqualified lookup works properly. | |
12947 | ||
86098eb8 JM |
12948 | This is useful for composing a namespace transparently from |
12949 | implementation namespaces. For example: | |
12950 | ||
12951 | @smallexample | |
12952 | namespace std @{ | |
12953 | namespace debug @{ | |
12954 | template <class T> struct A @{ @}; | |
12955 | @} | |
12956 | using namespace debug __attribute ((__strong__)); | |
cd1a8088 | 12957 | template <> struct A<int> @{ @}; // @r{ok to specialize} |
86098eb8 JM |
12958 | |
12959 | template <class T> void f (A<T>); | |
12960 | @} | |
12961 | ||
12962 | int main() | |
12963 | @{ | |
cd1a8088 | 12964 | f (std::A<float>()); // @r{lookup finds} std::f |
86098eb8 JM |
12965 | f (std::A<int>()); |
12966 | @} | |
12967 | @end smallexample | |
12968 | ||
cb68ec50 PC |
12969 | @node Type Traits |
12970 | @section Type Traits | |
12971 | ||
12972 | The C++ front-end implements syntactic extensions that allow to | |
12973 | determine at compile time various characteristics of a type (or of a | |
12974 | pair of types). | |
12975 | ||
12976 | @table @code | |
12977 | @item __has_nothrow_assign (type) | |
b29441ec PC |
12978 | If @code{type} is const qualified or is a reference type then the trait is |
12979 | false. Otherwise if @code{__has_trivial_assign (type)} is true then the trait | |
12980 | is true, else if @code{type} is a cv class or union type with copy assignment | |
12981 | operators that are known not to throw an exception then the trait is true, | |
12982 | else it is false. Requires: @code{type} shall be a complete type, an array | |
12983 | type of unknown bound, or is a @code{void} type. | |
cb68ec50 PC |
12984 | |
12985 | @item __has_nothrow_copy (type) | |
12986 | If @code{__has_trivial_copy (type)} is true then the trait is true, else if | |
12987 | @code{type} is a cv class or union type with copy constructors that | |
12988 | are known not to throw an exception then the trait is true, else it is false. | |
12989 | Requires: @code{type} shall be a complete type, an array type of | |
12990 | unknown bound, or is a @code{void} type. | |
12991 | ||
12992 | @item __has_nothrow_constructor (type) | |
12993 | If @code{__has_trivial_constructor (type)} is true then the trait is | |
12994 | true, else if @code{type} is a cv class or union type (or array | |
12995 | thereof) with a default constructor that is known not to throw an | |
12996 | exception then the trait is true, else it is false. Requires: | |
12997 | @code{type} shall be a complete type, an array type of unknown bound, | |
12998 | or is a @code{void} type. | |
12999 | ||
13000 | @item __has_trivial_assign (type) | |
13001 | If @code{type} is const qualified or is a reference type then the trait is | |
13002 | false. Otherwise if @code{__is_pod (type)} is true then the trait is | |
13003 | true, else if @code{type} is a cv class or union type with a trivial | |
13004 | copy assignment ([class.copy]) then the trait is true, else it is | |
13005 | false. Requires: @code{type} shall be a complete type, an array type | |
13006 | of unknown bound, or is a @code{void} type. | |
13007 | ||
13008 | @item __has_trivial_copy (type) | |
13009 | If @code{__is_pod (type)} is true or @code{type} is a reference type | |
13010 | then the trait is true, else if @code{type} is a cv class or union type | |
13011 | with a trivial copy constructor ([class.copy]) then the trait | |
13012 | is true, else it is false. Requires: @code{type} shall be a complete | |
13013 | type, an array type of unknown bound, or is a @code{void} type. | |
13014 | ||
13015 | @item __has_trivial_constructor (type) | |
13016 | If @code{__is_pod (type)} is true then the trait is true, else if | |
13017 | @code{type} is a cv class or union type (or array thereof) with a | |
13018 | trivial default constructor ([class.ctor]) then the trait is true, | |
13019 | else it is false. Requires: @code{type} shall be a complete type, an | |
13020 | array type of unknown bound, or is a @code{void} type. | |
13021 | ||
13022 | @item __has_trivial_destructor (type) | |
13023 | If @code{__is_pod (type)} is true or @code{type} is a reference type then | |
13024 | the trait is true, else if @code{type} is a cv class or union type (or | |
13025 | array thereof) with a trivial destructor ([class.dtor]) then the trait | |
13026 | is true, else it is false. Requires: @code{type} shall be a complete | |
13027 | type, an array type of unknown bound, or is a @code{void} type. | |
13028 | ||
13029 | @item __has_virtual_destructor (type) | |
13030 | If @code{type} is a class type with a virtual destructor | |
13031 | ([class.dtor]) then the trait is true, else it is false. Requires: | |
13032 | @code{type} shall be a complete type, an array type of unknown bound, | |
13033 | or is a @code{void} type. | |
13034 | ||
13035 | @item __is_abstract (type) | |
13036 | If @code{type} is an abstract class ([class.abstract]) then the trait | |
13037 | is true, else it is false. Requires: @code{type} shall be a complete | |
13038 | type, an array type of unknown bound, or is a @code{void} type. | |
13039 | ||
13040 | @item __is_base_of (base_type, derived_type) | |
13041 | If @code{base_type} is a base class of @code{derived_type} | |
13042 | ([class.derived]) then the trait is true, otherwise it is false. | |
13043 | Top-level cv qualifications of @code{base_type} and | |
13044 | @code{derived_type} are ignored. For the purposes of this trait, a | |
13045 | class type is considered is own base. Requires: if @code{__is_class | |
13046 | (base_type)} and @code{__is_class (derived_type)} are true and | |
13047 | @code{base_type} and @code{derived_type} are not the same type | |
13048 | (disregarding cv-qualifiers), @code{derived_type} shall be a complete | |
13049 | type. Diagnostic is produced if this requirement is not met. | |
13050 | ||
13051 | @item __is_class (type) | |
13052 | If @code{type} is a cv class type, and not a union type | |
d1facce0 | 13053 | ([basic.compound]) the trait is true, else it is false. |
cb68ec50 PC |
13054 | |
13055 | @item __is_empty (type) | |
13056 | If @code{__is_class (type)} is false then the trait is false. | |
13057 | Otherwise @code{type} is considered empty if and only if: @code{type} | |
13058 | has no non-static data members, or all non-static data members, if | |
d1facce0 | 13059 | any, are bit-fields of length 0, and @code{type} has no virtual |
cb68ec50 PC |
13060 | members, and @code{type} has no virtual base classes, and @code{type} |
13061 | has no base classes @code{base_type} for which | |
13062 | @code{__is_empty (base_type)} is false. Requires: @code{type} shall | |
13063 | be a complete type, an array type of unknown bound, or is a | |
13064 | @code{void} type. | |
13065 | ||
13066 | @item __is_enum (type) | |
d1facce0 | 13067 | If @code{type} is a cv enumeration type ([basic.compound]) the trait is |
cb68ec50 PC |
13068 | true, else it is false. |
13069 | ||
13070 | @item __is_pod (type) | |
13071 | If @code{type} is a cv POD type ([basic.types]) then the trait is true, | |
13072 | else it is false. Requires: @code{type} shall be a complete type, | |
13073 | an array type of unknown bound, or is a @code{void} type. | |
13074 | ||
13075 | @item __is_polymorphic (type) | |
13076 | If @code{type} is a polymorphic class ([class.virtual]) then the trait | |
13077 | is true, else it is false. Requires: @code{type} shall be a complete | |
13078 | type, an array type of unknown bound, or is a @code{void} type. | |
13079 | ||
13080 | @item __is_union (type) | |
d1facce0 | 13081 | If @code{type} is a cv union type ([basic.compound]) the trait is |
cb68ec50 PC |
13082 | true, else it is false. |
13083 | ||
13084 | @end table | |
13085 | ||
1f730ff7 ZW |
13086 | @node Java Exceptions |
13087 | @section Java Exceptions | |
13088 | ||
13089 | The Java language uses a slightly different exception handling model | |
13090 | from C++. Normally, GNU C++ will automatically detect when you are | |
13091 | writing C++ code that uses Java exceptions, and handle them | |
13092 | appropriately. However, if C++ code only needs to execute destructors | |
13093 | when Java exceptions are thrown through it, GCC will guess incorrectly. | |
9c34dbbf | 13094 | Sample problematic code is: |
1f730ff7 | 13095 | |
478c9e72 | 13096 | @smallexample |
1f730ff7 | 13097 | struct S @{ ~S(); @}; |
cd1a8088 | 13098 | extern void bar(); // @r{is written in Java, and may throw exceptions} |
1f730ff7 ZW |
13099 | void foo() |
13100 | @{ | |
13101 | S s; | |
13102 | bar(); | |
13103 | @} | |
478c9e72 | 13104 | @end smallexample |
1f730ff7 ZW |
13105 | |
13106 | @noindent | |
13107 | The usual effect of an incorrect guess is a link failure, complaining of | |
13108 | a missing routine called @samp{__gxx_personality_v0}. | |
13109 | ||
13110 | You can inform the compiler that Java exceptions are to be used in a | |
13111 | translation unit, irrespective of what it might think, by writing | |
13112 | @samp{@w{#pragma GCC java_exceptions}} at the head of the file. This | |
13113 | @samp{#pragma} must appear before any functions that throw or catch | |
13114 | exceptions, or run destructors when exceptions are thrown through them. | |
13115 | ||
13116 | You cannot mix Java and C++ exceptions in the same translation unit. It | |
13117 | is believed to be safe to throw a C++ exception from one file through | |
9c34dbbf ZW |
13118 | another file compiled for the Java exception model, or vice versa, but |
13119 | there may be bugs in this area. | |
1f730ff7 | 13120 | |
e6f3b89d PE |
13121 | @node Deprecated Features |
13122 | @section Deprecated Features | |
13123 | ||
13124 | In the past, the GNU C++ compiler was extended to experiment with new | |
767094dd | 13125 | features, at a time when the C++ language was still evolving. Now that |
e6f3b89d | 13126 | the C++ standard is complete, some of those features are superseded by |
767094dd JM |
13127 | superior alternatives. Using the old features might cause a warning in |
13128 | some cases that the feature will be dropped in the future. In other | |
e6f3b89d PE |
13129 | cases, the feature might be gone already. |
13130 | ||
13131 | While the list below is not exhaustive, it documents some of the options | |
13132 | that are now deprecated: | |
13133 | ||
13134 | @table @code | |
13135 | @item -fexternal-templates | |
13136 | @itemx -falt-external-templates | |
2dd76960 | 13137 | These are two of the many ways for G++ to implement template |
767094dd | 13138 | instantiation. @xref{Template Instantiation}. The C++ standard clearly |
e6f3b89d | 13139 | defines how template definitions have to be organized across |
2dd76960 | 13140 | implementation units. G++ has an implicit instantiation mechanism that |
e6f3b89d PE |
13141 | should work just fine for standard-conforming code. |
13142 | ||
13143 | @item -fstrict-prototype | |
13144 | @itemx -fno-strict-prototype | |
13145 | Previously it was possible to use an empty prototype parameter list to | |
13146 | indicate an unspecified number of parameters (like C), rather than no | |
767094dd | 13147 | parameters, as C++ demands. This feature has been removed, except where |
38bb2b65 | 13148 | it is required for backwards compatibility. @xref{Backwards Compatibility}. |
e6f3b89d PE |
13149 | @end table |
13150 | ||
ae209f28 NS |
13151 | G++ allows a virtual function returning @samp{void *} to be overridden |
13152 | by one returning a different pointer type. This extension to the | |
13153 | covariant return type rules is now deprecated and will be removed from a | |
13154 | future version. | |
13155 | ||
8ff24a79 MM |
13156 | The G++ minimum and maximum operators (@samp{<?} and @samp{>?}) and |
13157 | their compound forms (@samp{<?=}) and @samp{>?=}) have been deprecated | |
32e26ece GK |
13158 | and are now removed from G++. Code using these operators should be |
13159 | modified to use @code{std::min} and @code{std::max} instead. | |
8ff24a79 | 13160 | |
ad1a6d45 | 13161 | The named return value extension has been deprecated, and is now |
2dd76960 | 13162 | removed from G++. |
e6f3b89d | 13163 | |
82c18d5c | 13164 | The use of initializer lists with new expressions has been deprecated, |
2dd76960 | 13165 | and is now removed from G++. |
ad1a6d45 NS |
13166 | |
13167 | Floating and complex non-type template parameters have been deprecated, | |
2dd76960 | 13168 | and are now removed from G++. |
ad1a6d45 | 13169 | |
90ea7324 | 13170 | The implicit typename extension has been deprecated and is now |
2dd76960 | 13171 | removed from G++. |
90ea7324 | 13172 | |
1eaf20ec | 13173 | The use of default arguments in function pointers, function typedefs |
90ea7324 | 13174 | and other places where they are not permitted by the standard is |
2dd76960 | 13175 | deprecated and will be removed from a future version of G++. |
82c18d5c | 13176 | |
6871294a JW |
13177 | G++ allows floating-point literals to appear in integral constant expressions, |
13178 | e.g. @samp{ enum E @{ e = int(2.2 * 3.7) @} } | |
13179 | This extension is deprecated and will be removed from a future version. | |
13180 | ||
13181 | G++ allows static data members of const floating-point type to be declared | |
13182 | with an initializer in a class definition. The standard only allows | |
13183 | initializers for static members of const integral types and const | |
13184 | enumeration types so this extension has been deprecated and will be removed | |
13185 | from a future version. | |
13186 | ||
e6f3b89d PE |
13187 | @node Backwards Compatibility |
13188 | @section Backwards Compatibility | |
13189 | @cindex Backwards Compatibility | |
13190 | @cindex ARM [Annotated C++ Reference Manual] | |
13191 | ||
aee96fe9 | 13192 | Now that there is a definitive ISO standard C++, G++ has a specification |
767094dd | 13193 | to adhere to. The C++ language evolved over time, and features that |
e6f3b89d | 13194 | used to be acceptable in previous drafts of the standard, such as the ARM |
767094dd | 13195 | [Annotated C++ Reference Manual], are no longer accepted. In order to allow |
aee96fe9 | 13196 | compilation of C++ written to such drafts, G++ contains some backwards |
767094dd | 13197 | compatibilities. @emph{All such backwards compatibility features are |
aee96fe9 | 13198 | liable to disappear in future versions of G++.} They should be considered |
38bb2b65 | 13199 | deprecated. @xref{Deprecated Features}. |
e6f3b89d PE |
13200 | |
13201 | @table @code | |
13202 | @item For scope | |
13203 | If a variable is declared at for scope, it used to remain in scope until | |
13204 | the end of the scope which contained the for statement (rather than just | |
aee96fe9 | 13205 | within the for scope). G++ retains this, but issues a warning, if such a |
e6f3b89d PE |
13206 | variable is accessed outside the for scope. |
13207 | ||
ad1a6d45 | 13208 | @item Implicit C language |
630d3d5a | 13209 | Old C system header files did not contain an @code{extern "C" @{@dots{}@}} |
767094dd JM |
13210 | scope to set the language. On such systems, all header files are |
13211 | implicitly scoped inside a C language scope. Also, an empty prototype | |
e6f3b89d PE |
13212 | @code{()} will be treated as an unspecified number of arguments, rather |
13213 | than no arguments, as C++ demands. | |
13214 | @end table |