]>
Commit | Line | Data |
---|---|---|
f1717362 | 1 | @c Copyright (C) 1988-2016 Free Software Foundation, Inc. |
202d6e5f | 2 | @c This is part of the GCC manual. |
3 | @c For copying conditions, see the file gcc.texi. | |
4 | ||
5 | @node Target Macros | |
6 | @chapter Target Description Macros and Functions | |
7 | @cindex machine description macros | |
8 | @cindex target description macros | |
9 | @cindex macros, target description | |
10 | @cindex @file{tm.h} macros | |
11 | ||
12 | In addition to the file @file{@var{machine}.md}, a machine description | |
13 | includes a C header file conventionally given the name | |
14 | @file{@var{machine}.h} and a C source file named @file{@var{machine}.c}. | |
15 | The header file defines numerous macros that convey the information | |
16 | about the target machine that does not fit into the scheme of the | |
17 | @file{.md} file. The file @file{tm.h} should be a link to | |
18 | @file{@var{machine}.h}. The header file @file{config.h} includes | |
19 | @file{tm.h} and most compiler source files include @file{config.h}. The | |
20 | source file defines a variable @code{targetm}, which is a structure | |
21 | containing pointers to functions and data relating to the target | |
22 | machine. @file{@var{machine}.c} should also contain their definitions, | |
23 | if they are not defined elsewhere in GCC, and other functions called | |
24 | through the macros defined in the @file{.h} file. | |
25 | ||
26 | @menu | |
27 | * Target Structure:: The @code{targetm} variable. | |
28 | * Driver:: Controlling how the driver runs the compilation passes. | |
29 | * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}. | |
30 | * Per-Function Data:: Defining data structures for per-function information. | |
31 | * Storage Layout:: Defining sizes and alignments of data. | |
32 | * Type Layout:: Defining sizes and properties of basic user data types. | |
33 | * Registers:: Naming and describing the hardware registers. | |
34 | * Register Classes:: Defining the classes of hardware registers. | |
202d6e5f | 35 | * Stack and Calling:: Defining which way the stack grows and by how much. |
36 | * Varargs:: Defining the varargs macros. | |
37 | * Trampolines:: Code set up at run time to enter a nested function. | |
38 | * Library Calls:: Controlling how library routines are implicitly called. | |
39 | * Addressing Modes:: Defining addressing modes valid for memory operands. | |
40 | * Anchored Addresses:: Defining how @option{-fsection-anchors} should work. | |
41 | * Condition Code:: Defining how insns update the condition code. | |
42 | * Costs:: Defining relative costs of different operations. | |
43 | * Scheduling:: Adjusting the behavior of the instruction scheduler. | |
44 | * Sections:: Dividing storage into text, data, and other sections. | |
45 | * PIC:: Macros for position independent code. | |
46 | * Assembler Format:: Defining how to write insns and pseudo-ops to output. | |
47 | * Debugging Info:: Defining the format of debugging output. | |
48 | * Floating Point:: Handling floating point for cross-compilers. | |
49 | * Mode Switching:: Insertion of mode-switching instructions. | |
50 | * Target Attributes:: Defining target-specific uses of @code{__attribute__}. | |
51 | * Emulated TLS:: Emulated TLS support. | |
52 | * MIPS Coprocessors:: MIPS coprocessor support and how to customize it. | |
53 | * PCH Target:: Validity checking for precompiled headers. | |
54 | * C++ ABI:: Controlling C++ ABI changes. | |
55 | * Named Address Spaces:: Adding support for named address spaces | |
56 | * Misc:: Everything else. | |
57 | @end menu | |
58 | ||
59 | @node Target Structure | |
60 | @section The Global @code{targetm} Variable | |
61 | @cindex target hooks | |
62 | @cindex target functions | |
63 | ||
64 | @deftypevar {struct gcc_target} targetm | |
65 | The target @file{.c} file must define the global @code{targetm} variable | |
66 | which contains pointers to functions and data relating to the target | |
67 | machine. The variable is declared in @file{target.h}; | |
68 | @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is | |
69 | used to initialize the variable, and macros for the default initializers | |
70 | for elements of the structure. The @file{.c} file should override those | |
71 | macros for which the default definition is inappropriate. For example: | |
72 | @smallexample | |
73 | #include "target.h" | |
74 | #include "target-def.h" | |
75 | ||
76 | /* @r{Initialize the GCC target structure.} */ | |
77 | ||
78 | #undef TARGET_COMP_TYPE_ATTRIBUTES | |
79 | #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes | |
80 | ||
81 | struct gcc_target targetm = TARGET_INITIALIZER; | |
82 | @end smallexample | |
83 | @end deftypevar | |
84 | ||
85 | Where a macro should be defined in the @file{.c} file in this manner to | |
86 | form part of the @code{targetm} structure, it is documented below as a | |
87 | ``Target Hook'' with a prototype. Many macros will change in future | |
88 | from being defined in the @file{.h} file to being part of the | |
89 | @code{targetm} structure. | |
90 | ||
c94b1d0e | 91 | Similarly, there is a @code{targetcm} variable for hooks that are |
92 | specific to front ends for C-family languages, documented as ``C | |
93 | Target Hook''. This is declared in @file{c-family/c-target.h}, the | |
851d9296 | 94 | initializer @code{TARGETCM_INITIALIZER} in |
c94b1d0e | 95 | @file{c-family/c-target-def.h}. If targets initialize @code{targetcm} |
96 | themselves, they should set @code{target_has_targetcm=yes} in | |
97 | @file{config.gcc}; otherwise a default definition is used. | |
98 | ||
3e87b980 | 99 | Similarly, there is a @code{targetm_common} variable for hooks that |
100 | are shared between the compiler driver and the compilers proper, | |
101 | documented as ``Common Target Hook''. This is declared in | |
102 | @file{common/common-target.h}, the initializer | |
103 | @code{TARGETM_COMMON_INITIALIZER} in | |
104 | @file{common/common-target-def.h}. If targets initialize | |
105 | @code{targetm_common} themselves, they should set | |
106 | @code{target_has_targetm_common=yes} in @file{config.gcc}; otherwise a | |
107 | default definition is used. | |
108 | ||
202d6e5f | 109 | @node Driver |
110 | @section Controlling the Compilation Driver, @file{gcc} | |
111 | @cindex driver | |
112 | @cindex controlling the compilation driver | |
113 | ||
114 | @c prevent bad page break with this line | |
115 | You can control the compilation driver. | |
116 | ||
202d6e5f | 117 | @defmac DRIVER_SELF_SPECS |
118 | A list of specs for the driver itself. It should be a suitable | |
119 | initializer for an array of strings, with no surrounding braces. | |
120 | ||
121 | The driver applies these specs to its own command line between loading | |
122 | default @file{specs} files (but not command-line specified ones) and | |
123 | choosing the multilib directory or running any subcommands. It | |
124 | applies them in the order given, so each spec can depend on the | |
125 | options added by earlier ones. It is also possible to remove options | |
126 | using @samp{%<@var{option}} in the usual way. | |
127 | ||
128 | This macro can be useful when a port has several interdependent target | |
129 | options. It provides a way of standardizing the command line so | |
130 | that the other specs are easier to write. | |
131 | ||
132 | Do not define this macro if it does not need to do anything. | |
133 | @end defmac | |
134 | ||
135 | @defmac OPTION_DEFAULT_SPECS | |
136 | A list of specs used to support configure-time default options (i.e.@: | |
137 | @option{--with} options) in the driver. It should be a suitable initializer | |
138 | for an array of structures, each containing two strings, without the | |
139 | outermost pair of surrounding braces. | |
140 | ||
141 | The first item in the pair is the name of the default. This must match | |
142 | the code in @file{config.gcc} for the target. The second item is a spec | |
143 | to apply if a default with this name was specified. The string | |
144 | @samp{%(VALUE)} in the spec will be replaced by the value of the default | |
145 | everywhere it occurs. | |
146 | ||
147 | The driver will apply these specs to its own command line between loading | |
148 | default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using | |
149 | the same mechanism as @code{DRIVER_SELF_SPECS}. | |
150 | ||
151 | Do not define this macro if it does not need to do anything. | |
152 | @end defmac | |
153 | ||
154 | @defmac CPP_SPEC | |
155 | A C string constant that tells the GCC driver program options to | |
156 | pass to CPP@. It can also specify how to translate options you | |
157 | give to GCC into options for GCC to pass to the CPP@. | |
158 | ||
159 | Do not define this macro if it does not need to do anything. | |
160 | @end defmac | |
161 | ||
162 | @defmac CPLUSPLUS_CPP_SPEC | |
163 | This macro is just like @code{CPP_SPEC}, but is used for C++, rather | |
164 | than C@. If you do not define this macro, then the value of | |
165 | @code{CPP_SPEC} (if any) will be used instead. | |
166 | @end defmac | |
167 | ||
168 | @defmac CC1_SPEC | |
169 | A C string constant that tells the GCC driver program options to | |
170 | pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language | |
171 | front ends. | |
172 | It can also specify how to translate options you give to GCC into options | |
173 | for GCC to pass to front ends. | |
174 | ||
175 | Do not define this macro if it does not need to do anything. | |
176 | @end defmac | |
177 | ||
178 | @defmac CC1PLUS_SPEC | |
179 | A C string constant that tells the GCC driver program options to | |
180 | pass to @code{cc1plus}. It can also specify how to translate options you | |
181 | give to GCC into options for GCC to pass to the @code{cc1plus}. | |
182 | ||
183 | Do not define this macro if it does not need to do anything. | |
184 | Note that everything defined in CC1_SPEC is already passed to | |
185 | @code{cc1plus} so there is no need to duplicate the contents of | |
186 | CC1_SPEC in CC1PLUS_SPEC@. | |
187 | @end defmac | |
188 | ||
189 | @defmac ASM_SPEC | |
190 | A C string constant that tells the GCC driver program options to | |
191 | pass to the assembler. It can also specify how to translate options | |
192 | you give to GCC into options for GCC to pass to the assembler. | |
193 | See the file @file{sun3.h} for an example of this. | |
194 | ||
195 | Do not define this macro if it does not need to do anything. | |
196 | @end defmac | |
197 | ||
198 | @defmac ASM_FINAL_SPEC | |
199 | A C string constant that tells the GCC driver program how to | |
200 | run any programs which cleanup after the normal assembler. | |
201 | Normally, this is not needed. See the file @file{mips.h} for | |
202 | an example of this. | |
203 | ||
204 | Do not define this macro if it does not need to do anything. | |
205 | @end defmac | |
206 | ||
207 | @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT | |
208 | Define this macro, with no value, if the driver should give the assembler | |
209 | an argument consisting of a single dash, @option{-}, to instruct it to | |
210 | read from its standard input (which will be a pipe connected to the | |
211 | output of the compiler proper). This argument is given after any | |
212 | @option{-o} option specifying the name of the output file. | |
213 | ||
214 | If you do not define this macro, the assembler is assumed to read its | |
215 | standard input if given no non-option arguments. If your assembler | |
216 | cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct; | |
217 | see @file{mips.h} for instance. | |
218 | @end defmac | |
219 | ||
220 | @defmac LINK_SPEC | |
221 | A C string constant that tells the GCC driver program options to | |
222 | pass to the linker. It can also specify how to translate options you | |
223 | give to GCC into options for GCC to pass to the linker. | |
224 | ||
225 | Do not define this macro if it does not need to do anything. | |
226 | @end defmac | |
227 | ||
228 | @defmac LIB_SPEC | |
229 | Another C string constant used much like @code{LINK_SPEC}. The difference | |
230 | between the two is that @code{LIB_SPEC} is used at the end of the | |
231 | command given to the linker. | |
232 | ||
233 | If this macro is not defined, a default is provided that | |
234 | loads the standard C library from the usual place. See @file{gcc.c}. | |
235 | @end defmac | |
236 | ||
237 | @defmac LIBGCC_SPEC | |
238 | Another C string constant that tells the GCC driver program | |
239 | how and when to place a reference to @file{libgcc.a} into the | |
240 | linker command line. This constant is placed both before and after | |
241 | the value of @code{LIB_SPEC}. | |
242 | ||
243 | If this macro is not defined, the GCC driver provides a default that | |
244 | passes the string @option{-lgcc} to the linker. | |
245 | @end defmac | |
246 | ||
247 | @defmac REAL_LIBGCC_SPEC | |
248 | By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the | |
249 | @code{LIBGCC_SPEC} is not directly used by the driver program but is | |
250 | instead modified to refer to different versions of @file{libgcc.a} | |
251 | depending on the values of the command line flags @option{-static}, | |
252 | @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On | |
253 | targets where these modifications are inappropriate, define | |
254 | @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the | |
255 | driver how to place a reference to @file{libgcc} on the link command | |
256 | line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified. | |
257 | @end defmac | |
258 | ||
259 | @defmac USE_LD_AS_NEEDED | |
260 | A macro that controls the modifications to @code{LIBGCC_SPEC} | |
261 | mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be | |
8aaed91d | 262 | generated that uses @option{--as-needed} or equivalent options and the |
263 | shared @file{libgcc} in place of the | |
202d6e5f | 264 | static exception handler library, when linking without any of |
265 | @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}. | |
266 | @end defmac | |
267 | ||
268 | @defmac LINK_EH_SPEC | |
269 | If defined, this C string constant is added to @code{LINK_SPEC}. | |
270 | When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects | |
271 | the modifications to @code{LIBGCC_SPEC} mentioned in | |
272 | @code{REAL_LIBGCC_SPEC}. | |
273 | @end defmac | |
274 | ||
275 | @defmac STARTFILE_SPEC | |
276 | Another C string constant used much like @code{LINK_SPEC}. The | |
277 | difference between the two is that @code{STARTFILE_SPEC} is used at | |
278 | the very beginning of the command given to the linker. | |
279 | ||
280 | If this macro is not defined, a default is provided that loads the | |
281 | standard C startup file from the usual place. See @file{gcc.c}. | |
282 | @end defmac | |
283 | ||
284 | @defmac ENDFILE_SPEC | |
285 | Another C string constant used much like @code{LINK_SPEC}. The | |
286 | difference between the two is that @code{ENDFILE_SPEC} is used at | |
287 | the very end of the command given to the linker. | |
288 | ||
289 | Do not define this macro if it does not need to do anything. | |
290 | @end defmac | |
291 | ||
292 | @defmac THREAD_MODEL_SPEC | |
293 | GCC @code{-v} will print the thread model GCC was configured to use. | |
294 | However, this doesn't work on platforms that are multilibbed on thread | |
295 | models, such as AIX 4.3. On such platforms, define | |
296 | @code{THREAD_MODEL_SPEC} such that it evaluates to a string without | |
297 | blanks that names one of the recognized thread models. @code{%*}, the | |
298 | default value of this macro, will expand to the value of | |
299 | @code{thread_file} set in @file{config.gcc}. | |
300 | @end defmac | |
301 | ||
302 | @defmac SYSROOT_SUFFIX_SPEC | |
303 | Define this macro to add a suffix to the target sysroot when GCC is | |
304 | configured with a sysroot. This will cause GCC to search for usr/lib, | |
305 | et al, within sysroot+suffix. | |
306 | @end defmac | |
307 | ||
308 | @defmac SYSROOT_HEADERS_SUFFIX_SPEC | |
309 | Define this macro to add a headers_suffix to the target sysroot when | |
310 | GCC is configured with a sysroot. This will cause GCC to pass the | |
311 | updated sysroot+headers_suffix to CPP, causing it to search for | |
312 | usr/include, et al, within sysroot+headers_suffix. | |
313 | @end defmac | |
314 | ||
315 | @defmac EXTRA_SPECS | |
316 | Define this macro to provide additional specifications to put in the | |
317 | @file{specs} file that can be used in various specifications like | |
318 | @code{CC1_SPEC}. | |
319 | ||
320 | The definition should be an initializer for an array of structures, | |
321 | containing a string constant, that defines the specification name, and a | |
322 | string constant that provides the specification. | |
323 | ||
324 | Do not define this macro if it does not need to do anything. | |
325 | ||
326 | @code{EXTRA_SPECS} is useful when an architecture contains several | |
327 | related targets, which have various @code{@dots{}_SPECS} which are similar | |
328 | to each other, and the maintainer would like one central place to keep | |
329 | these definitions. | |
330 | ||
331 | For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to | |
332 | define either @code{_CALL_SYSV} when the System V calling sequence is | |
333 | used or @code{_CALL_AIX} when the older AIX-based calling sequence is | |
334 | used. | |
335 | ||
336 | The @file{config/rs6000/rs6000.h} target file defines: | |
337 | ||
338 | @smallexample | |
339 | #define EXTRA_SPECS \ | |
340 | @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @}, | |
341 | ||
342 | #define CPP_SYS_DEFAULT "" | |
343 | @end smallexample | |
344 | ||
345 | The @file{config/rs6000/sysv.h} target file defines: | |
346 | @smallexample | |
347 | #undef CPP_SPEC | |
348 | #define CPP_SPEC \ | |
349 | "%@{posix: -D_POSIX_SOURCE @} \ | |
350 | %@{mcall-sysv: -D_CALL_SYSV @} \ | |
351 | %@{!mcall-sysv: %(cpp_sysv_default) @} \ | |
352 | %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}" | |
353 | ||
354 | #undef CPP_SYSV_DEFAULT | |
355 | #define CPP_SYSV_DEFAULT "-D_CALL_SYSV" | |
356 | @end smallexample | |
357 | ||
358 | while the @file{config/rs6000/eabiaix.h} target file defines | |
359 | @code{CPP_SYSV_DEFAULT} as: | |
360 | ||
361 | @smallexample | |
362 | #undef CPP_SYSV_DEFAULT | |
363 | #define CPP_SYSV_DEFAULT "-D_CALL_AIX" | |
364 | @end smallexample | |
365 | @end defmac | |
366 | ||
367 | @defmac LINK_LIBGCC_SPECIAL_1 | |
368 | Define this macro if the driver program should find the library | |
369 | @file{libgcc.a}. If you do not define this macro, the driver program will pass | |
370 | the argument @option{-lgcc} to tell the linker to do the search. | |
371 | @end defmac | |
372 | ||
373 | @defmac LINK_GCC_C_SEQUENCE_SPEC | |
374 | The sequence in which libgcc and libc are specified to the linker. | |
375 | By default this is @code{%G %L %G}. | |
376 | @end defmac | |
377 | ||
0ed9e37f | 378 | @defmac POST_LINK_SPEC |
379 | Define this macro to add additional steps to be executed after linker. | |
380 | The default value of this macro is empty string. | |
381 | @end defmac | |
382 | ||
202d6e5f | 383 | @defmac LINK_COMMAND_SPEC |
384 | A C string constant giving the complete command line need to execute the | |
385 | linker. When you do this, you will need to update your port each time a | |
386 | change is made to the link command line within @file{gcc.c}. Therefore, | |
387 | define this macro only if you need to completely redefine the command | |
388 | line for invoking the linker and there is no other way to accomplish | |
389 | the effect you need. Overriding this macro may be avoidable by overriding | |
390 | @code{LINK_GCC_C_SEQUENCE_SPEC} instead. | |
391 | @end defmac | |
392 | ||
3e87b980 | 393 | @hook TARGET_ALWAYS_STRIP_DOTDOT |
394 | ||
202d6e5f | 395 | @defmac MULTILIB_DEFAULTS |
396 | Define this macro as a C expression for the initializer of an array of | |
397 | string to tell the driver program which options are defaults for this | |
398 | target and thus do not need to be handled specially when using | |
399 | @code{MULTILIB_OPTIONS}. | |
400 | ||
401 | Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in | |
402 | the target makefile fragment or if none of the options listed in | |
403 | @code{MULTILIB_OPTIONS} are set by default. | |
404 | @xref{Target Fragment}. | |
405 | @end defmac | |
406 | ||
407 | @defmac RELATIVE_PREFIX_NOT_LINKDIR | |
408 | Define this macro to tell @command{gcc} that it should only translate | |
409 | a @option{-B} prefix into a @option{-L} linker option if the prefix | |
410 | indicates an absolute file name. | |
411 | @end defmac | |
412 | ||
413 | @defmac MD_EXEC_PREFIX | |
414 | If defined, this macro is an additional prefix to try after | |
415 | @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched | |
416 | when the compiler is built as a cross | |
417 | compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it | |
e757a6a8 | 418 | to the list of directories used to find the assembler in @file{configure.ac}. |
202d6e5f | 419 | @end defmac |
420 | ||
421 | @defmac STANDARD_STARTFILE_PREFIX | |
422 | Define this macro as a C string constant if you wish to override the | |
423 | standard choice of @code{libdir} as the default prefix to | |
424 | try when searching for startup files such as @file{crt0.o}. | |
425 | @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler | |
426 | is built as a cross compiler. | |
427 | @end defmac | |
428 | ||
429 | @defmac STANDARD_STARTFILE_PREFIX_1 | |
430 | Define this macro as a C string constant if you wish to override the | |
431 | standard choice of @code{/lib} as a prefix to try after the default prefix | |
432 | when searching for startup files such as @file{crt0.o}. | |
433 | @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler | |
434 | is built as a cross compiler. | |
435 | @end defmac | |
436 | ||
437 | @defmac STANDARD_STARTFILE_PREFIX_2 | |
438 | Define this macro as a C string constant if you wish to override the | |
439 | standard choice of @code{/lib} as yet another prefix to try after the | |
440 | default prefix when searching for startup files such as @file{crt0.o}. | |
441 | @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler | |
442 | is built as a cross compiler. | |
443 | @end defmac | |
444 | ||
445 | @defmac MD_STARTFILE_PREFIX | |
446 | If defined, this macro supplies an additional prefix to try after the | |
447 | standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the | |
448 | compiler is built as a cross compiler. | |
449 | @end defmac | |
450 | ||
451 | @defmac MD_STARTFILE_PREFIX_1 | |
452 | If defined, this macro supplies yet another prefix to try after the | |
453 | standard prefixes. It is not searched when the compiler is built as a | |
454 | cross compiler. | |
455 | @end defmac | |
456 | ||
457 | @defmac INIT_ENVIRONMENT | |
458 | Define this macro as a C string constant if you wish to set environment | |
459 | variables for programs called by the driver, such as the assembler and | |
460 | loader. The driver passes the value of this macro to @code{putenv} to | |
461 | initialize the necessary environment variables. | |
462 | @end defmac | |
463 | ||
464 | @defmac LOCAL_INCLUDE_DIR | |
465 | Define this macro as a C string constant if you wish to override the | |
466 | standard choice of @file{/usr/local/include} as the default prefix to | |
467 | try when searching for local header files. @code{LOCAL_INCLUDE_DIR} | |
638454a1 | 468 | comes before @code{NATIVE_SYSTEM_HEADER_DIR} (set in |
469 | @file{config.gcc}, normally @file{/usr/include}) in the search order. | |
202d6e5f | 470 | |
471 | Cross compilers do not search either @file{/usr/local/include} or its | |
472 | replacement. | |
473 | @end defmac | |
474 | ||
638454a1 | 475 | @defmac NATIVE_SYSTEM_HEADER_COMPONENT |
476 | The ``component'' corresponding to @code{NATIVE_SYSTEM_HEADER_DIR}. | |
202d6e5f | 477 | See @code{INCLUDE_DEFAULTS}, below, for the description of components. |
478 | If you do not define this macro, no component is used. | |
479 | @end defmac | |
480 | ||
481 | @defmac INCLUDE_DEFAULTS | |
482 | Define this macro if you wish to override the entire default search path | |
483 | for include files. For a native compiler, the default search path | |
484 | usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR}, | |
638454a1 | 485 | @code{GPLUSPLUS_INCLUDE_DIR}, and |
486 | @code{NATIVE_SYSTEM_HEADER_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR} | |
202d6e5f | 487 | and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile}, |
488 | and specify private search areas for GCC@. The directory | |
489 | @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs. | |
490 | ||
491 | The definition should be an initializer for an array of structures. | |
492 | Each array element should have four elements: the directory name (a | |
493 | string constant), the component name (also a string constant), a flag | |
494 | for C++-only directories, | |
495 | and a flag showing that the includes in the directory don't need to be | |
496 | wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of | |
497 | the array with a null element. | |
498 | ||
499 | The component name denotes what GNU package the include file is part of, | |
500 | if any, in all uppercase letters. For example, it might be @samp{GCC} | |
501 | or @samp{BINUTILS}. If the package is part of a vendor-supplied | |
502 | operating system, code the component name as @samp{0}. | |
503 | ||
504 | For example, here is the definition used for VAX/VMS: | |
505 | ||
506 | @smallexample | |
507 | #define INCLUDE_DEFAULTS \ | |
508 | @{ \ | |
509 | @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \ | |
510 | @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \ | |
511 | @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \ | |
512 | @{ ".", 0, 0, 0@}, \ | |
513 | @{ 0, 0, 0, 0@} \ | |
514 | @} | |
515 | @end smallexample | |
516 | @end defmac | |
517 | ||
518 | Here is the order of prefixes tried for exec files: | |
519 | ||
520 | @enumerate | |
521 | @item | |
522 | Any prefixes specified by the user with @option{-B}. | |
523 | ||
524 | @item | |
525 | The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX} | |
15b474a2 | 526 | is not set and the compiler has not been installed in the configure-time |
202d6e5f | 527 | @var{prefix}, the location in which the compiler has actually been installed. |
528 | ||
529 | @item | |
530 | The directories specified by the environment variable @code{COMPILER_PATH}. | |
531 | ||
532 | @item | |
533 | The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed | |
15b474a2 | 534 | in the configured-time @var{prefix}. |
202d6e5f | 535 | |
536 | @item | |
15b474a2 | 537 | The location @file{/usr/libexec/gcc/}, but only if this is a native compiler. |
202d6e5f | 538 | |
539 | @item | |
15b474a2 | 540 | The location @file{/usr/lib/gcc/}, but only if this is a native compiler. |
202d6e5f | 541 | |
542 | @item | |
15b474a2 | 543 | The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native |
202d6e5f | 544 | compiler. |
545 | @end enumerate | |
546 | ||
547 | Here is the order of prefixes tried for startfiles: | |
548 | ||
549 | @enumerate | |
550 | @item | |
551 | Any prefixes specified by the user with @option{-B}. | |
552 | ||
553 | @item | |
554 | The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined | |
555 | value based on the installed toolchain location. | |
556 | ||
557 | @item | |
558 | The directories specified by the environment variable @code{LIBRARY_PATH} | |
559 | (or port-specific name; native only, cross compilers do not use this). | |
560 | ||
561 | @item | |
562 | The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed | |
15b474a2 | 563 | in the configured @var{prefix} or this is a native compiler. |
202d6e5f | 564 | |
565 | @item | |
566 | The location @file{/usr/lib/gcc/}, but only if this is a native compiler. | |
567 | ||
568 | @item | |
15b474a2 | 569 | The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native |
202d6e5f | 570 | compiler. |
571 | ||
572 | @item | |
15b474a2 | 573 | The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a |
202d6e5f | 574 | native compiler, or we have a target system root. |
575 | ||
576 | @item | |
15b474a2 | 577 | The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a |
202d6e5f | 578 | native compiler, or we have a target system root. |
579 | ||
580 | @item | |
581 | The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications. | |
582 | If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and | |
583 | the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix. | |
584 | ||
585 | @item | |
586 | The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native | |
587 | compiler, or we have a target system root. The default for this macro is | |
588 | @file{/lib/}. | |
589 | ||
590 | @item | |
591 | The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native | |
592 | compiler, or we have a target system root. The default for this macro is | |
593 | @file{/usr/lib/}. | |
594 | @end enumerate | |
595 | ||
596 | @node Run-time Target | |
597 | @section Run-time Target Specification | |
598 | @cindex run-time target specification | |
599 | @cindex predefined macros | |
600 | @cindex target specifications | |
601 | ||
602 | @c prevent bad page break with this line | |
603 | Here are run-time target specifications. | |
604 | ||
605 | @defmac TARGET_CPU_CPP_BUILTINS () | |
606 | This function-like macro expands to a block of code that defines | |
607 | built-in preprocessor macros and assertions for the target CPU, using | |
608 | the functions @code{builtin_define}, @code{builtin_define_std} and | |
609 | @code{builtin_assert}. When the front end | |
610 | calls this macro it provides a trailing semicolon, and since it has | |
611 | finished command line option processing your code can use those | |
612 | results freely. | |
613 | ||
614 | @code{builtin_assert} takes a string in the form you pass to the | |
615 | command-line option @option{-A}, such as @code{cpu=mips}, and creates | |
616 | the assertion. @code{builtin_define} takes a string in the form | |
617 | accepted by option @option{-D} and unconditionally defines the macro. | |
618 | ||
619 | @code{builtin_define_std} takes a string representing the name of an | |
620 | object-like macro. If it doesn't lie in the user's namespace, | |
621 | @code{builtin_define_std} defines it unconditionally. Otherwise, it | |
622 | defines a version with two leading underscores, and another version | |
623 | with two leading and trailing underscores, and defines the original | |
624 | only if an ISO standard was not requested on the command line. For | |
625 | example, passing @code{unix} defines @code{__unix}, @code{__unix__} | |
626 | and possibly @code{unix}; passing @code{_mips} defines @code{__mips}, | |
627 | @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64} | |
628 | defines only @code{_ABI64}. | |
629 | ||
630 | You can also test for the C dialect being compiled. The variable | |
631 | @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus} | |
632 | or @code{clk_objective_c}. Note that if we are preprocessing | |
633 | assembler, this variable will be @code{clk_c} but the function-like | |
634 | macro @code{preprocessing_asm_p()} will return true, so you might want | |
635 | to check for that first. If you need to check for strict ANSI, the | |
636 | variable @code{flag_iso} can be used. The function-like macro | |
637 | @code{preprocessing_trad_p()} can be used to check for traditional | |
638 | preprocessing. | |
639 | @end defmac | |
640 | ||
641 | @defmac TARGET_OS_CPP_BUILTINS () | |
642 | Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional | |
643 | and is used for the target operating system instead. | |
644 | @end defmac | |
645 | ||
646 | @defmac TARGET_OBJFMT_CPP_BUILTINS () | |
647 | Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional | |
648 | and is used for the target object format. @file{elfos.h} uses this | |
649 | macro to define @code{__ELF__}, so you probably do not need to define | |
650 | it yourself. | |
651 | @end defmac | |
652 | ||
653 | @deftypevar {extern int} target_flags | |
654 | This variable is declared in @file{options.h}, which is included before | |
655 | any target-specific headers. | |
656 | @end deftypevar | |
657 | ||
658 | @hook TARGET_DEFAULT_TARGET_FLAGS | |
659 | This variable specifies the initial value of @code{target_flags}. | |
660 | Its default setting is 0. | |
661 | @end deftypevr | |
662 | ||
663 | @cindex optional hardware or system features | |
664 | @cindex features, optional, in system conventions | |
665 | ||
666 | @hook TARGET_HANDLE_OPTION | |
667 | This hook is called whenever the user specifies one of the | |
668 | target-specific options described by the @file{.opt} definition files | |
669 | (@pxref{Options}). It has the opportunity to do some option-specific | |
670 | processing and should return true if the option is valid. The default | |
671 | definition does nothing but return true. | |
672 | ||
fba5dd52 | 673 | @var{decoded} specifies the option and its arguments. @var{opts} and |
674 | @var{opts_set} are the @code{gcc_options} structures to be used for | |
675 | storing option state, and @var{loc} is the location at which the | |
676 | option was passed (@code{UNKNOWN_LOCATION} except for options passed | |
677 | via attributes). | |
202d6e5f | 678 | @end deftypefn |
679 | ||
680 | @hook TARGET_HANDLE_C_OPTION | |
681 | This target hook is called whenever the user specifies one of the | |
682 | target-specific C language family options described by the @file{.opt} | |
683 | definition files(@pxref{Options}). It has the opportunity to do some | |
684 | option-specific processing and should return true if the option is | |
685 | valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The | |
686 | default definition does nothing but return false. | |
687 | ||
688 | In general, you should use @code{TARGET_HANDLE_OPTION} to handle | |
689 | options. However, if processing an option requires routines that are | |
690 | only available in the C (and related language) front ends, then you | |
691 | should use @code{TARGET_HANDLE_C_OPTION} instead. | |
692 | @end deftypefn | |
693 | ||
1f6616ee | 694 | @hook TARGET_OBJC_CONSTRUCT_STRING_OBJECT |
695 | ||
ff6624bc | 696 | @hook TARGET_OBJC_DECLARE_UNRESOLVED_CLASS_REFERENCE |
697 | ||
698 | @hook TARGET_OBJC_DECLARE_CLASS_DEFINITION | |
699 | ||
1f6616ee | 700 | @hook TARGET_STRING_OBJECT_REF_TYPE_P |
701 | ||
702 | @hook TARGET_CHECK_STRING_OBJECT_FORMAT_ARG | |
d4238e8b | 703 | |
202d6e5f | 704 | @hook TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE |
202d6e5f | 705 | |
706 | @defmac C_COMMON_OVERRIDE_OPTIONS | |
4c834714 | 707 | This is similar to the @code{TARGET_OPTION_OVERRIDE} hook |
708 | but is only used in the C | |
202d6e5f | 709 | language frontends (C, Objective-C, C++, Objective-C++) and so can be |
710 | used to alter option flag variables which only exist in those | |
711 | frontends. | |
712 | @end defmac | |
713 | ||
c17f64cc | 714 | @hook TARGET_OPTION_OPTIMIZATION_TABLE |
202d6e5f | 715 | Some machines may desire to change what optimizations are performed for |
c17f64cc | 716 | various optimization levels. This variable, if defined, describes |
717 | options to enable at particular sets of optimization levels. These | |
718 | options are processed once | |
202d6e5f | 719 | just after the optimization level is determined and before the remainder |
c17f64cc | 720 | of the command options have been parsed, so may be overridden by other |
b59688ee | 721 | options passed explicitly. |
202d6e5f | 722 | |
c17f64cc | 723 | This processing is run once at program startup and when the optimization |
202d6e5f | 724 | options are changed via @code{#pragma GCC optimize} or by using the |
725 | @code{optimize} attribute. | |
c17f64cc | 726 | @end deftypevr |
202d6e5f | 727 | |
cc07c468 | 728 | @hook TARGET_OPTION_INIT_STRUCT |
729 | ||
686e2769 | 730 | @hook TARGET_OPTION_DEFAULT_PARAMS |
731 | ||
821d4118 | 732 | @defmac SWITCHABLE_TARGET |
733 | Some targets need to switch between substantially different subtargets | |
734 | during compilation. For example, the MIPS target has one subtarget for | |
735 | the traditional MIPS architecture and another for MIPS16. Source code | |
736 | can switch between these two subarchitectures using the @code{mips16} | |
737 | and @code{nomips16} attributes. | |
738 | ||
739 | Such subtargets can differ in things like the set of available | |
740 | registers, the set of available instructions, the costs of various | |
741 | operations, and so on. GCC caches a lot of this type of information | |
742 | in global variables, and recomputing them for each subtarget takes a | |
743 | significant amount of time. The compiler therefore provides a facility | |
744 | for maintaining several versions of the global variables and quickly | |
745 | switching between them; see @file{target-globals.h} for details. | |
746 | ||
747 | Define this macro to 1 if your target needs this facility. The default | |
748 | is 0. | |
749 | @end defmac | |
750 | ||
4c866b9b | 751 | @hook TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P |
752 | ||
202d6e5f | 753 | @node Per-Function Data |
754 | @section Defining data structures for per-function information. | |
755 | @cindex per-function data | |
756 | @cindex data structures | |
757 | ||
758 | If the target needs to store information on a per-function basis, GCC | |
759 | provides a macro and a couple of variables to allow this. Note, just | |
760 | using statics to store the information is a bad idea, since GCC supports | |
761 | nested functions, so you can be halfway through encoding one function | |
762 | when another one comes along. | |
763 | ||
764 | GCC defines a data structure called @code{struct function} which | |
765 | contains all of the data specific to an individual function. This | |
766 | structure contains a field called @code{machine} whose type is | |
767 | @code{struct machine_function *}, which can be used by targets to point | |
768 | to their own specific data. | |
769 | ||
770 | If a target needs per-function specific data it should define the type | |
771 | @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}. | |
772 | This macro should be used to initialize the function pointer | |
773 | @code{init_machine_status}. This pointer is explained below. | |
774 | ||
775 | One typical use of per-function, target specific data is to create an | |
776 | RTX to hold the register containing the function's return address. This | |
777 | RTX can then be used to implement the @code{__builtin_return_address} | |
778 | function, for level 0. | |
779 | ||
780 | Note---earlier implementations of GCC used a single data area to hold | |
781 | all of the per-function information. Thus when processing of a nested | |
782 | function began the old per-function data had to be pushed onto a | |
783 | stack, and when the processing was finished, it had to be popped off the | |
784 | stack. GCC used to provide function pointers called | |
785 | @code{save_machine_status} and @code{restore_machine_status} to handle | |
786 | the saving and restoring of the target specific information. Since the | |
787 | single data area approach is no longer used, these pointers are no | |
788 | longer supported. | |
789 | ||
790 | @defmac INIT_EXPANDERS | |
791 | Macro called to initialize any target specific information. This macro | |
792 | is called once per function, before generation of any RTL has begun. | |
793 | The intention of this macro is to allow the initialization of the | |
794 | function pointer @code{init_machine_status}. | |
795 | @end defmac | |
796 | ||
797 | @deftypevar {void (*)(struct function *)} init_machine_status | |
798 | If this function pointer is non-@code{NULL} it will be called once per | |
799 | function, before function compilation starts, in order to allow the | |
800 | target to perform any target specific initialization of the | |
801 | @code{struct function} structure. It is intended that this would be | |
802 | used to initialize the @code{machine} of that structure. | |
803 | ||
804 | @code{struct machine_function} structures are expected to be freed by GC@. | |
805 | Generally, any memory that they reference must be allocated by using | |
806 | GC allocation, including the structure itself. | |
807 | @end deftypevar | |
808 | ||
809 | @node Storage Layout | |
810 | @section Storage Layout | |
811 | @cindex storage layout | |
812 | ||
813 | Note that the definitions of the macros in this table which are sizes or | |
814 | alignments measured in bits do not need to be constant. They can be C | |
815 | expressions that refer to static variables, such as the @code{target_flags}. | |
816 | @xref{Run-time Target}. | |
817 | ||
818 | @defmac BITS_BIG_ENDIAN | |
819 | Define this macro to have the value 1 if the most significant bit in a | |
820 | byte has the lowest number; otherwise define it to have the value zero. | |
821 | This means that bit-field instructions count from the most significant | |
822 | bit. If the machine has no bit-field instructions, then this must still | |
823 | be defined, but it doesn't matter which value it is defined to. This | |
824 | macro need not be a constant. | |
825 | ||
826 | This macro does not affect the way structure fields are packed into | |
827 | bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}. | |
828 | @end defmac | |
829 | ||
830 | @defmac BYTES_BIG_ENDIAN | |
831 | Define this macro to have the value 1 if the most significant byte in a | |
832 | word has the lowest number. This macro need not be a constant. | |
833 | @end defmac | |
834 | ||
835 | @defmac WORDS_BIG_ENDIAN | |
836 | Define this macro to have the value 1 if, in a multiword object, the | |
837 | most significant word has the lowest number. This applies to both | |
76c64076 | 838 | memory locations and registers; see @code{REG_WORDS_BIG_ENDIAN} if the |
839 | order of words in memory is not the same as the order in registers. This | |
202d6e5f | 840 | macro need not be a constant. |
841 | @end defmac | |
842 | ||
76c64076 | 843 | @defmac REG_WORDS_BIG_ENDIAN |
844 | On some machines, the order of words in a multiword object differs between | |
845 | registers in memory. In such a situation, define this macro to describe | |
846 | the order of words in a register. The macro @code{WORDS_BIG_ENDIAN} controls | |
847 | the order of words in memory. | |
848 | @end defmac | |
849 | ||
202d6e5f | 850 | @defmac FLOAT_WORDS_BIG_ENDIAN |
851 | Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or | |
852 | @code{TFmode} floating point numbers are stored in memory with the word | |
853 | containing the sign bit at the lowest address; otherwise define it to | |
854 | have the value 0. This macro need not be a constant. | |
855 | ||
856 | You need not define this macro if the ordering is the same as for | |
857 | multi-word integers. | |
858 | @end defmac | |
859 | ||
202d6e5f | 860 | @defmac BITS_PER_WORD |
861 | Number of bits in a word. If you do not define this macro, the default | |
862 | is @code{BITS_PER_UNIT * UNITS_PER_WORD}. | |
863 | @end defmac | |
864 | ||
865 | @defmac MAX_BITS_PER_WORD | |
866 | Maximum number of bits in a word. If this is undefined, the default is | |
867 | @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the | |
868 | largest value that @code{BITS_PER_WORD} can have at run-time. | |
869 | @end defmac | |
870 | ||
871 | @defmac UNITS_PER_WORD | |
872 | Number of storage units in a word; normally the size of a general-purpose | |
873 | register, a power of two from 1 or 8. | |
874 | @end defmac | |
875 | ||
876 | @defmac MIN_UNITS_PER_WORD | |
877 | Minimum number of units in a word. If this is undefined, the default is | |
878 | @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the | |
879 | smallest value that @code{UNITS_PER_WORD} can have at run-time. | |
880 | @end defmac | |
881 | ||
202d6e5f | 882 | @defmac POINTER_SIZE |
883 | Width of a pointer, in bits. You must specify a value no wider than the | |
884 | width of @code{Pmode}. If it is not equal to the width of @code{Pmode}, | |
885 | you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify | |
886 | a value the default is @code{BITS_PER_WORD}. | |
887 | @end defmac | |
888 | ||
889 | @defmac POINTERS_EXTEND_UNSIGNED | |
890 | A C expression that determines how pointers should be extended from | |
891 | @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is | |
892 | greater than zero if pointers should be zero-extended, zero if they | |
893 | should be sign-extended, and negative if some other sort of conversion | |
894 | is needed. In the last case, the extension is done by the target's | |
895 | @code{ptr_extend} instruction. | |
896 | ||
897 | You need not define this macro if the @code{ptr_mode}, @code{Pmode} | |
898 | and @code{word_mode} are all the same width. | |
899 | @end defmac | |
900 | ||
901 | @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type}) | |
902 | A macro to update @var{m} and @var{unsignedp} when an object whose type | |
903 | is @var{type} and which has the specified mode and signedness is to be | |
904 | stored in a register. This macro is only called when @var{type} is a | |
905 | scalar type. | |
906 | ||
907 | On most RISC machines, which only have operations that operate on a full | |
908 | register, define this macro to set @var{m} to @code{word_mode} if | |
909 | @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most | |
910 | cases, only integer modes should be widened because wider-precision | |
911 | floating-point operations are usually more expensive than their narrower | |
912 | counterparts. | |
913 | ||
914 | For most machines, the macro definition does not change @var{unsignedp}. | |
915 | However, some machines, have instructions that preferentially handle | |
916 | either signed or unsigned quantities of certain modes. For example, on | |
917 | the DEC Alpha, 32-bit loads from memory and 32-bit add instructions | |
918 | sign-extend the result to 64 bits. On such machines, set | |
919 | @var{unsignedp} according to which kind of extension is more efficient. | |
920 | ||
921 | Do not define this macro if it would never modify @var{m}. | |
922 | @end defmac | |
923 | ||
924 | @hook TARGET_PROMOTE_FUNCTION_MODE | |
202d6e5f | 925 | |
926 | @defmac PARM_BOUNDARY | |
927 | Normal alignment required for function parameters on the stack, in | |
928 | bits. All stack parameters receive at least this much alignment | |
929 | regardless of data type. On most machines, this is the same as the | |
930 | size of an integer. | |
931 | @end defmac | |
932 | ||
933 | @defmac STACK_BOUNDARY | |
934 | Define this macro to the minimum alignment enforced by hardware for the | |
935 | stack pointer on this machine. The definition is a C expression for the | |
936 | desired alignment (measured in bits). This value is used as a default | |
937 | if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines, | |
938 | this should be the same as @code{PARM_BOUNDARY}. | |
939 | @end defmac | |
940 | ||
941 | @defmac PREFERRED_STACK_BOUNDARY | |
942 | Define this macro if you wish to preserve a certain alignment for the | |
943 | stack pointer, greater than what the hardware enforces. The definition | |
944 | is a C expression for the desired alignment (measured in bits). This | |
945 | macro must evaluate to a value equal to or larger than | |
946 | @code{STACK_BOUNDARY}. | |
947 | @end defmac | |
948 | ||
949 | @defmac INCOMING_STACK_BOUNDARY | |
950 | Define this macro if the incoming stack boundary may be different | |
951 | from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate | |
952 | to a value equal to or larger than @code{STACK_BOUNDARY}. | |
953 | @end defmac | |
954 | ||
955 | @defmac FUNCTION_BOUNDARY | |
956 | Alignment required for a function entry point, in bits. | |
957 | @end defmac | |
958 | ||
959 | @defmac BIGGEST_ALIGNMENT | |
960 | Biggest alignment that any data type can require on this machine, in | |
961 | bits. Note that this is not the biggest alignment that is supported, | |
962 | just the biggest alignment that, when violated, may cause a fault. | |
963 | @end defmac | |
964 | ||
41971a20 | 965 | @hook TARGET_ABSOLUTE_BIGGEST_ALIGNMENT |
966 | ||
202d6e5f | 967 | @defmac MALLOC_ABI_ALIGNMENT |
968 | Alignment, in bits, a C conformant malloc implementation has to | |
969 | provide. If not defined, the default value is @code{BITS_PER_WORD}. | |
970 | @end defmac | |
971 | ||
972 | @defmac ATTRIBUTE_ALIGNED_VALUE | |
973 | Alignment used by the @code{__attribute__ ((aligned))} construct. If | |
974 | not defined, the default value is @code{BIGGEST_ALIGNMENT}. | |
975 | @end defmac | |
976 | ||
977 | @defmac MINIMUM_ATOMIC_ALIGNMENT | |
978 | If defined, the smallest alignment, in bits, that can be given to an | |
979 | object that can be referenced in one operation, without disturbing any | |
980 | nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger | |
981 | on machines that don't have byte or half-word store operations. | |
982 | @end defmac | |
983 | ||
984 | @defmac BIGGEST_FIELD_ALIGNMENT | |
985 | Biggest alignment that any structure or union field can require on this | |
986 | machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for | |
987 | structure and union fields only, unless the field alignment has been set | |
988 | by the @code{__attribute__ ((aligned (@var{n})))} construct. | |
989 | @end defmac | |
990 | ||
991 | @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed}) | |
992 | An expression for the alignment of a structure field @var{field} if the | |
993 | alignment computed in the usual way (including applying of | |
994 | @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the | |
995 | alignment) is @var{computed}. It overrides alignment only if the | |
996 | field alignment has not been set by the | |
997 | @code{__attribute__ ((aligned (@var{n})))} construct. | |
998 | @end defmac | |
999 | ||
1000 | @defmac MAX_STACK_ALIGNMENT | |
1001 | Biggest stack alignment guaranteed by the backend. Use this macro | |
1002 | to specify the maximum alignment of a variable on stack. | |
1003 | ||
1004 | If not defined, the default value is @code{STACK_BOUNDARY}. | |
1005 | ||
1006 | @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}. | |
1007 | @c But the fix for PR 32893 indicates that we can only guarantee | |
1008 | @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not | |
1009 | @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported. | |
1010 | @end defmac | |
1011 | ||
1012 | @defmac MAX_OFILE_ALIGNMENT | |
1013 | Biggest alignment supported by the object file format of this machine. | |
1014 | Use this macro to limit the alignment which can be specified using the | |
1015 | @code{__attribute__ ((aligned (@var{n})))} construct. If not defined, | |
1016 | the default value is @code{BIGGEST_ALIGNMENT}. | |
1017 | ||
1018 | On systems that use ELF, the default (in @file{config/elfos.h}) is | |
1019 | the largest supported 32-bit ELF section alignment representable on | |
3a4303e7 | 1020 | a 32-bit host e.g. @samp{(((uint64_t) 1 << 28) * 8)}. |
202d6e5f | 1021 | On 32-bit ELF the largest supported section alignment in bits is |
1022 | @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts. | |
1023 | @end defmac | |
1024 | ||
1025 | @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align}) | |
1026 | If defined, a C expression to compute the alignment for a variable in | |
1027 | the static store. @var{type} is the data type, and @var{basic-align} is | |
1028 | the alignment that the object would ordinarily have. The value of this | |
1029 | macro is used instead of that alignment to align the object. | |
1030 | ||
1031 | If this macro is not defined, then @var{basic-align} is used. | |
1032 | ||
1033 | @findex strcpy | |
1034 | One use of this macro is to increase alignment of medium-size data to | |
1035 | make it all fit in fewer cache lines. Another is to cause character | |
1036 | arrays to be word-aligned so that @code{strcpy} calls that copy | |
1037 | constants to character arrays can be done inline. | |
1038 | @end defmac | |
1039 | ||
a16734cd | 1040 | @defmac DATA_ABI_ALIGNMENT (@var{type}, @var{basic-align}) |
1041 | Similar to @code{DATA_ALIGNMENT}, but for the cases where the ABI mandates | |
1042 | some alignment increase, instead of optimization only purposes. E.g.@ | |
1043 | AMD x86-64 psABI says that variables with array type larger than 15 bytes | |
1044 | must be aligned to 16 byte boundaries. | |
1045 | ||
1046 | If this macro is not defined, then @var{basic-align} is used. | |
1047 | @end defmac | |
1048 | ||
202d6e5f | 1049 | @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align}) |
1050 | If defined, a C expression to compute the alignment given to a constant | |
1051 | that is being placed in memory. @var{constant} is the constant and | |
1052 | @var{basic-align} is the alignment that the object would ordinarily | |
1053 | have. The value of this macro is used instead of that alignment to | |
1054 | align the object. | |
1055 | ||
e532afed | 1056 | The default definition just returns @var{basic-align}. |
202d6e5f | 1057 | |
1058 | The typical use of this macro is to increase alignment for string | |
1059 | constants to be word aligned so that @code{strcpy} calls that copy | |
1060 | constants can be done inline. | |
1061 | @end defmac | |
1062 | ||
1063 | @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align}) | |
1064 | If defined, a C expression to compute the alignment for a variable in | |
1065 | the local store. @var{type} is the data type, and @var{basic-align} is | |
1066 | the alignment that the object would ordinarily have. The value of this | |
1067 | macro is used instead of that alignment to align the object. | |
1068 | ||
1069 | If this macro is not defined, then @var{basic-align} is used. | |
1070 | ||
1071 | One use of this macro is to increase alignment of medium-size data to | |
1072 | make it all fit in fewer cache lines. | |
c6e790c5 | 1073 | |
2f05705b | 1074 | If the value of this macro has a type, it should be an unsigned type. |
202d6e5f | 1075 | @end defmac |
1076 | ||
482a44fa | 1077 | @hook TARGET_VECTOR_ALIGNMENT |
1078 | ||
202d6e5f | 1079 | @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align}) |
1080 | If defined, a C expression to compute the alignment for stack slot. | |
1081 | @var{type} is the data type, @var{mode} is the widest mode available, | |
1082 | and @var{basic-align} is the alignment that the slot would ordinarily | |
1083 | have. The value of this macro is used instead of that alignment to | |
1084 | align the slot. | |
1085 | ||
1086 | If this macro is not defined, then @var{basic-align} is used when | |
1087 | @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will | |
1088 | be used. | |
1089 | ||
1090 | This macro is to set alignment of stack slot to the maximum alignment | |
1091 | of all possible modes which the slot may have. | |
c6e790c5 | 1092 | |
2f05705b | 1093 | If the value of this macro has a type, it should be an unsigned type. |
202d6e5f | 1094 | @end defmac |
1095 | ||
1096 | @defmac LOCAL_DECL_ALIGNMENT (@var{decl}) | |
1097 | If defined, a C expression to compute the alignment for a local | |
1098 | variable @var{decl}. | |
1099 | ||
1100 | If this macro is not defined, then | |
1101 | @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))} | |
1102 | is used. | |
1103 | ||
1104 | One use of this macro is to increase alignment of medium-size data to | |
1105 | make it all fit in fewer cache lines. | |
c6e790c5 | 1106 | |
2f05705b | 1107 | If the value of this macro has a type, it should be an unsigned type. |
202d6e5f | 1108 | @end defmac |
1109 | ||
1110 | @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align}) | |
1111 | If defined, a C expression to compute the minimum required alignment | |
1112 | for dynamic stack realignment purposes for @var{exp} (a type or decl), | |
1113 | @var{mode}, assuming normal alignment @var{align}. | |
1114 | ||
1115 | If this macro is not defined, then @var{align} will be used. | |
1116 | @end defmac | |
1117 | ||
1118 | @defmac EMPTY_FIELD_BOUNDARY | |
1119 | Alignment in bits to be given to a structure bit-field that follows an | |
1120 | empty field such as @code{int : 0;}. | |
1121 | ||
1122 | If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro. | |
1123 | @end defmac | |
1124 | ||
1125 | @defmac STRUCTURE_SIZE_BOUNDARY | |
1126 | Number of bits which any structure or union's size must be a multiple of. | |
1127 | Each structure or union's size is rounded up to a multiple of this. | |
1128 | ||
1129 | If you do not define this macro, the default is the same as | |
1130 | @code{BITS_PER_UNIT}. | |
1131 | @end defmac | |
1132 | ||
1133 | @defmac STRICT_ALIGNMENT | |
1134 | Define this macro to be the value 1 if instructions will fail to work | |
1135 | if given data not on the nominal alignment. If instructions will merely | |
1136 | go slower in that case, define this macro as 0. | |
1137 | @end defmac | |
1138 | ||
1139 | @defmac PCC_BITFIELD_TYPE_MATTERS | |
1140 | Define this if you wish to imitate the way many other C compilers handle | |
1141 | alignment of bit-fields and the structures that contain them. | |
1142 | ||
1143 | The behavior is that the type written for a named bit-field (@code{int}, | |
1144 | @code{short}, or other integer type) imposes an alignment for the entire | |
1145 | structure, as if the structure really did contain an ordinary field of | |
1146 | that type. In addition, the bit-field is placed within the structure so | |
1147 | that it would fit within such a field, not crossing a boundary for it. | |
1148 | ||
1149 | Thus, on most machines, a named bit-field whose type is written as | |
1150 | @code{int} would not cross a four-byte boundary, and would force | |
1151 | four-byte alignment for the whole structure. (The alignment used may | |
1152 | not be four bytes; it is controlled by the other alignment parameters.) | |
1153 | ||
1154 | An unnamed bit-field will not affect the alignment of the containing | |
1155 | structure. | |
1156 | ||
1157 | If the macro is defined, its definition should be a C expression; | |
1158 | a nonzero value for the expression enables this behavior. | |
1159 | ||
1160 | Note that if this macro is not defined, or its value is zero, some | |
1161 | bit-fields may cross more than one alignment boundary. The compiler can | |
1162 | support such references if there are @samp{insv}, @samp{extv}, and | |
1163 | @samp{extzv} insns that can directly reference memory. | |
1164 | ||
1165 | The other known way of making bit-fields work is to define | |
1166 | @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}. | |
1167 | Then every structure can be accessed with fullwords. | |
1168 | ||
1169 | Unless the machine has bit-field instructions or you define | |
1170 | @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define | |
1171 | @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value. | |
1172 | ||
1173 | If your aim is to make GCC use the same conventions for laying out | |
1174 | bit-fields as are used by another compiler, here is how to investigate | |
1175 | what the other compiler does. Compile and run this program: | |
1176 | ||
1177 | @smallexample | |
1178 | struct foo1 | |
1179 | @{ | |
1180 | char x; | |
1181 | char :0; | |
1182 | char y; | |
1183 | @}; | |
1184 | ||
1185 | struct foo2 | |
1186 | @{ | |
1187 | char x; | |
1188 | int :0; | |
1189 | char y; | |
1190 | @}; | |
1191 | ||
1192 | main () | |
1193 | @{ | |
1194 | printf ("Size of foo1 is %d\n", | |
1195 | sizeof (struct foo1)); | |
1196 | printf ("Size of foo2 is %d\n", | |
1197 | sizeof (struct foo2)); | |
1198 | exit (0); | |
1199 | @} | |
1200 | @end smallexample | |
1201 | ||
1202 | If this prints 2 and 5, then the compiler's behavior is what you would | |
1203 | get from @code{PCC_BITFIELD_TYPE_MATTERS}. | |
1204 | @end defmac | |
1205 | ||
1206 | @defmac BITFIELD_NBYTES_LIMITED | |
1207 | Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited | |
1208 | to aligning a bit-field within the structure. | |
1209 | @end defmac | |
1210 | ||
1211 | @hook TARGET_ALIGN_ANON_BITFIELD | |
202d6e5f | 1212 | |
1213 | @hook TARGET_NARROW_VOLATILE_BITFIELD | |
202d6e5f | 1214 | |
f91ed644 | 1215 | @hook TARGET_MEMBER_TYPE_FORCES_BLK |
202d6e5f | 1216 | |
1217 | @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified}) | |
1218 | Define this macro as an expression for the alignment of a type (given | |
1219 | by @var{type} as a tree node) if the alignment computed in the usual | |
1220 | way is @var{computed} and the alignment explicitly specified was | |
1221 | @var{specified}. | |
1222 | ||
1223 | The default is to use @var{specified} if it is larger; otherwise, use | |
1224 | the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT} | |
1225 | @end defmac | |
1226 | ||
1227 | @defmac MAX_FIXED_MODE_SIZE | |
1228 | An integer expression for the size in bits of the largest integer | |
1229 | machine mode that should actually be used. All integer machine modes of | |
1230 | this size or smaller can be used for structures and unions with the | |
1231 | appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE | |
1232 | (DImode)} is assumed. | |
1233 | @end defmac | |
1234 | ||
1235 | @defmac STACK_SAVEAREA_MODE (@var{save_level}) | |
3754d046 | 1236 | If defined, an expression of type @code{machine_mode} that |
202d6e5f | 1237 | specifies the mode of the save area operand of a |
1238 | @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}). | |
1239 | @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or | |
1240 | @code{SAVE_NONLOCAL} and selects which of the three named patterns is | |
1241 | having its mode specified. | |
1242 | ||
1243 | You need not define this macro if it always returns @code{Pmode}. You | |
1244 | would most commonly define this macro if the | |
1245 | @code{save_stack_@var{level}} patterns need to support both a 32- and a | |
1246 | 64-bit mode. | |
1247 | @end defmac | |
1248 | ||
1249 | @defmac STACK_SIZE_MODE | |
3754d046 | 1250 | If defined, an expression of type @code{machine_mode} that |
202d6e5f | 1251 | specifies the mode of the size increment operand of an |
1252 | @code{allocate_stack} named pattern (@pxref{Standard Names}). | |
1253 | ||
1254 | You need not define this macro if it always returns @code{word_mode}. | |
1255 | You would most commonly define this macro if the @code{allocate_stack} | |
1256 | pattern needs to support both a 32- and a 64-bit mode. | |
1257 | @end defmac | |
1258 | ||
1259 | @hook TARGET_LIBGCC_CMP_RETURN_MODE | |
202d6e5f | 1260 | |
1261 | @hook TARGET_LIBGCC_SHIFT_COUNT_MODE | |
202d6e5f | 1262 | |
1263 | @hook TARGET_UNWIND_WORD_MODE | |
202d6e5f | 1264 | |
202d6e5f | 1265 | @hook TARGET_MS_BITFIELD_LAYOUT_P |
202d6e5f | 1266 | |
1267 | @hook TARGET_DECIMAL_FLOAT_SUPPORTED_P | |
202d6e5f | 1268 | |
1269 | @hook TARGET_FIXED_POINT_SUPPORTED_P | |
202d6e5f | 1270 | |
1271 | @hook TARGET_EXPAND_TO_RTL_HOOK | |
202d6e5f | 1272 | |
1273 | @hook TARGET_INSTANTIATE_DECLS | |
202d6e5f | 1274 | |
1275 | @hook TARGET_MANGLE_TYPE | |
202d6e5f | 1276 | |
1277 | @node Type Layout | |
1278 | @section Layout of Source Language Data Types | |
1279 | ||
1280 | These macros define the sizes and other characteristics of the standard | |
1281 | basic data types used in programs being compiled. Unlike the macros in | |
1282 | the previous section, these apply to specific features of C and related | |
1283 | languages, rather than to fundamental aspects of storage layout. | |
1284 | ||
1285 | @defmac INT_TYPE_SIZE | |
1286 | A C expression for the size in bits of the type @code{int} on the | |
1287 | target machine. If you don't define this, the default is one word. | |
1288 | @end defmac | |
1289 | ||
1290 | @defmac SHORT_TYPE_SIZE | |
1291 | A C expression for the size in bits of the type @code{short} on the | |
1292 | target machine. If you don't define this, the default is half a word. | |
1293 | (If this would be less than one storage unit, it is rounded up to one | |
1294 | unit.) | |
1295 | @end defmac | |
1296 | ||
1297 | @defmac LONG_TYPE_SIZE | |
1298 | A C expression for the size in bits of the type @code{long} on the | |
1299 | target machine. If you don't define this, the default is one word. | |
1300 | @end defmac | |
1301 | ||
1302 | @defmac ADA_LONG_TYPE_SIZE | |
1303 | On some machines, the size used for the Ada equivalent of the type | |
1304 | @code{long} by a native Ada compiler differs from that used by C@. In | |
1305 | that situation, define this macro to be a C expression to be used for | |
1306 | the size of that type. If you don't define this, the default is the | |
1307 | value of @code{LONG_TYPE_SIZE}. | |
1308 | @end defmac | |
1309 | ||
1310 | @defmac LONG_LONG_TYPE_SIZE | |
1311 | A C expression for the size in bits of the type @code{long long} on the | |
1312 | target machine. If you don't define this, the default is two | |
1313 | words. If you want to support GNU Ada on your machine, the value of this | |
1314 | macro must be at least 64. | |
1315 | @end defmac | |
1316 | ||
1317 | @defmac CHAR_TYPE_SIZE | |
1318 | A C expression for the size in bits of the type @code{char} on the | |
1319 | target machine. If you don't define this, the default is | |
1320 | @code{BITS_PER_UNIT}. | |
1321 | @end defmac | |
1322 | ||
1323 | @defmac BOOL_TYPE_SIZE | |
1324 | A C expression for the size in bits of the C++ type @code{bool} and | |
1325 | C99 type @code{_Bool} on the target machine. If you don't define | |
1326 | this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}. | |
1327 | @end defmac | |
1328 | ||
1329 | @defmac FLOAT_TYPE_SIZE | |
1330 | A C expression for the size in bits of the type @code{float} on the | |
1331 | target machine. If you don't define this, the default is one word. | |
1332 | @end defmac | |
1333 | ||
1334 | @defmac DOUBLE_TYPE_SIZE | |
1335 | A C expression for the size in bits of the type @code{double} on the | |
1336 | target machine. If you don't define this, the default is two | |
1337 | words. | |
1338 | @end defmac | |
1339 | ||
1340 | @defmac LONG_DOUBLE_TYPE_SIZE | |
1341 | A C expression for the size in bits of the type @code{long double} on | |
1342 | the target machine. If you don't define this, the default is two | |
1343 | words. | |
1344 | @end defmac | |
1345 | ||
1346 | @defmac SHORT_FRACT_TYPE_SIZE | |
1347 | A C expression for the size in bits of the type @code{short _Fract} on | |
1348 | the target machine. If you don't define this, the default is | |
1349 | @code{BITS_PER_UNIT}. | |
1350 | @end defmac | |
1351 | ||
1352 | @defmac FRACT_TYPE_SIZE | |
1353 | A C expression for the size in bits of the type @code{_Fract} on | |
1354 | the target machine. If you don't define this, the default is | |
1355 | @code{BITS_PER_UNIT * 2}. | |
1356 | @end defmac | |
1357 | ||
1358 | @defmac LONG_FRACT_TYPE_SIZE | |
1359 | A C expression for the size in bits of the type @code{long _Fract} on | |
1360 | the target machine. If you don't define this, the default is | |
1361 | @code{BITS_PER_UNIT * 4}. | |
1362 | @end defmac | |
1363 | ||
1364 | @defmac LONG_LONG_FRACT_TYPE_SIZE | |
1365 | A C expression for the size in bits of the type @code{long long _Fract} on | |
1366 | the target machine. If you don't define this, the default is | |
1367 | @code{BITS_PER_UNIT * 8}. | |
1368 | @end defmac | |
1369 | ||
1370 | @defmac SHORT_ACCUM_TYPE_SIZE | |
1371 | A C expression for the size in bits of the type @code{short _Accum} on | |
1372 | the target machine. If you don't define this, the default is | |
1373 | @code{BITS_PER_UNIT * 2}. | |
1374 | @end defmac | |
1375 | ||
1376 | @defmac ACCUM_TYPE_SIZE | |
1377 | A C expression for the size in bits of the type @code{_Accum} on | |
1378 | the target machine. If you don't define this, the default is | |
1379 | @code{BITS_PER_UNIT * 4}. | |
1380 | @end defmac | |
1381 | ||
1382 | @defmac LONG_ACCUM_TYPE_SIZE | |
1383 | A C expression for the size in bits of the type @code{long _Accum} on | |
1384 | the target machine. If you don't define this, the default is | |
1385 | @code{BITS_PER_UNIT * 8}. | |
1386 | @end defmac | |
1387 | ||
1388 | @defmac LONG_LONG_ACCUM_TYPE_SIZE | |
1389 | A C expression for the size in bits of the type @code{long long _Accum} on | |
1390 | the target machine. If you don't define this, the default is | |
1391 | @code{BITS_PER_UNIT * 16}. | |
1392 | @end defmac | |
1393 | ||
f308a9b2 | 1394 | @defmac LIBGCC2_GNU_PREFIX |
1395 | This macro corresponds to the @code{TARGET_LIBFUNC_GNU_PREFIX} target | |
1396 | hook and should be defined if that hook is overriden to be true. It | |
1397 | causes function names in libgcc to be changed to use a @code{__gnu_} | |
1398 | prefix for their name rather than the default @code{__}. A port which | |
1399 | uses this macro should also arrange to use @file{t-gnu-prefix} in | |
1400 | the libgcc @file{config.host}. | |
1401 | @end defmac | |
1402 | ||
202d6e5f | 1403 | @defmac TARGET_FLT_EVAL_METHOD |
1404 | A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h}, | |
1405 | assuming, if applicable, that the floating-point control word is in its | |
1406 | default state. If you do not define this macro the value of | |
1407 | @code{FLT_EVAL_METHOD} will be zero. | |
1408 | @end defmac | |
1409 | ||
1410 | @defmac WIDEST_HARDWARE_FP_SIZE | |
1411 | A C expression for the size in bits of the widest floating-point format | |
1412 | supported by the hardware. If you define this macro, you must specify a | |
1413 | value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}. | |
1414 | If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE} | |
1415 | is the default. | |
1416 | @end defmac | |
1417 | ||
1418 | @defmac DEFAULT_SIGNED_CHAR | |
1419 | An expression whose value is 1 or 0, according to whether the type | |
1420 | @code{char} should be signed or unsigned by default. The user can | |
1421 | always override this default with the options @option{-fsigned-char} | |
1422 | and @option{-funsigned-char}. | |
1423 | @end defmac | |
1424 | ||
1425 | @hook TARGET_DEFAULT_SHORT_ENUMS | |
202d6e5f | 1426 | |
1427 | @defmac SIZE_TYPE | |
1428 | A C expression for a string describing the name of the data type to use | |
1429 | for size values. The typedef name @code{size_t} is defined using the | |
1430 | contents of the string. | |
1431 | ||
1432 | The string can contain more than one keyword. If so, separate them with | |
1433 | spaces, and write first any length keyword, then @code{unsigned} if | |
1434 | appropriate, and finally @code{int}. The string must exactly match one | |
1435 | of the data type names defined in the function | |
816c3ff2 | 1436 | @code{c_common_nodes_and_builtins} in the file @file{c-family/c-common.c}. |
1437 | You may not omit @code{int} or change the order---that would cause the | |
1438 | compiler to crash on startup. | |
202d6e5f | 1439 | |
1440 | If you don't define this macro, the default is @code{"long unsigned | |
1441 | int"}. | |
1442 | @end defmac | |
1443 | ||
748e5d45 | 1444 | @defmac SIZETYPE |
1445 | GCC defines internal types (@code{sizetype}, @code{ssizetype}, | |
1446 | @code{bitsizetype} and @code{sbitsizetype}) for expressions | |
1447 | dealing with size. This macro is a C expression for a string describing | |
1448 | the name of the data type from which the precision of @code{sizetype} | |
1449 | is extracted. | |
1450 | ||
1451 | The string has the same restrictions as @code{SIZE_TYPE} string. | |
1452 | ||
1453 | If you don't define this macro, the default is @code{SIZE_TYPE}. | |
1454 | @end defmac | |
1455 | ||
202d6e5f | 1456 | @defmac PTRDIFF_TYPE |
1457 | A C expression for a string describing the name of the data type to use | |
1458 | for the result of subtracting two pointers. The typedef name | |
1459 | @code{ptrdiff_t} is defined using the contents of the string. See | |
1460 | @code{SIZE_TYPE} above for more information. | |
1461 | ||
1462 | If you don't define this macro, the default is @code{"long int"}. | |
1463 | @end defmac | |
1464 | ||
1465 | @defmac WCHAR_TYPE | |
1466 | A C expression for a string describing the name of the data type to use | |
1467 | for wide characters. The typedef name @code{wchar_t} is defined using | |
1468 | the contents of the string. See @code{SIZE_TYPE} above for more | |
1469 | information. | |
1470 | ||
1471 | If you don't define this macro, the default is @code{"int"}. | |
1472 | @end defmac | |
1473 | ||
1474 | @defmac WCHAR_TYPE_SIZE | |
1475 | A C expression for the size in bits of the data type for wide | |
1476 | characters. This is used in @code{cpp}, which cannot make use of | |
1477 | @code{WCHAR_TYPE}. | |
1478 | @end defmac | |
1479 | ||
1480 | @defmac WINT_TYPE | |
1481 | A C expression for a string describing the name of the data type to | |
1482 | use for wide characters passed to @code{printf} and returned from | |
1483 | @code{getwc}. The typedef name @code{wint_t} is defined using the | |
1484 | contents of the string. See @code{SIZE_TYPE} above for more | |
1485 | information. | |
1486 | ||
1487 | If you don't define this macro, the default is @code{"unsigned int"}. | |
1488 | @end defmac | |
1489 | ||
1490 | @defmac INTMAX_TYPE | |
1491 | A C expression for a string describing the name of the data type that | |
1492 | can represent any value of any standard or extended signed integer type. | |
1493 | The typedef name @code{intmax_t} is defined using the contents of the | |
1494 | string. See @code{SIZE_TYPE} above for more information. | |
1495 | ||
1496 | If you don't define this macro, the default is the first of | |
1497 | @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as | |
1498 | much precision as @code{long long int}. | |
1499 | @end defmac | |
1500 | ||
1501 | @defmac UINTMAX_TYPE | |
1502 | A C expression for a string describing the name of the data type that | |
1503 | can represent any value of any standard or extended unsigned integer | |
1504 | type. The typedef name @code{uintmax_t} is defined using the contents | |
1505 | of the string. See @code{SIZE_TYPE} above for more information. | |
1506 | ||
1507 | If you don't define this macro, the default is the first of | |
1508 | @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long | |
1509 | unsigned int"} that has as much precision as @code{long long unsigned | |
1510 | int}. | |
1511 | @end defmac | |
1512 | ||
1513 | @defmac SIG_ATOMIC_TYPE | |
1514 | @defmacx INT8_TYPE | |
1515 | @defmacx INT16_TYPE | |
1516 | @defmacx INT32_TYPE | |
1517 | @defmacx INT64_TYPE | |
1518 | @defmacx UINT8_TYPE | |
1519 | @defmacx UINT16_TYPE | |
1520 | @defmacx UINT32_TYPE | |
1521 | @defmacx UINT64_TYPE | |
1522 | @defmacx INT_LEAST8_TYPE | |
1523 | @defmacx INT_LEAST16_TYPE | |
1524 | @defmacx INT_LEAST32_TYPE | |
1525 | @defmacx INT_LEAST64_TYPE | |
1526 | @defmacx UINT_LEAST8_TYPE | |
1527 | @defmacx UINT_LEAST16_TYPE | |
1528 | @defmacx UINT_LEAST32_TYPE | |
1529 | @defmacx UINT_LEAST64_TYPE | |
1530 | @defmacx INT_FAST8_TYPE | |
1531 | @defmacx INT_FAST16_TYPE | |
1532 | @defmacx INT_FAST32_TYPE | |
1533 | @defmacx INT_FAST64_TYPE | |
1534 | @defmacx UINT_FAST8_TYPE | |
1535 | @defmacx UINT_FAST16_TYPE | |
1536 | @defmacx UINT_FAST32_TYPE | |
1537 | @defmacx UINT_FAST64_TYPE | |
1538 | @defmacx INTPTR_TYPE | |
1539 | @defmacx UINTPTR_TYPE | |
1540 | C expressions for the standard types @code{sig_atomic_t}, | |
1541 | @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t}, | |
1542 | @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t}, | |
1543 | @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t}, | |
1544 | @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t}, | |
1545 | @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t}, | |
1546 | @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t}, | |
1547 | @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t}, | |
1548 | @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See | |
1549 | @code{SIZE_TYPE} above for more information. | |
1550 | ||
1551 | If any of these macros evaluates to a null pointer, the corresponding | |
1552 | type is not supported; if GCC is configured to provide | |
1553 | @code{<stdint.h>} in such a case, the header provided may not conform | |
1554 | to C99, depending on the type in question. The defaults for all of | |
1555 | these macros are null pointers. | |
1556 | @end defmac | |
1557 | ||
1558 | @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION | |
1559 | The C++ compiler represents a pointer-to-member-function with a struct | |
1560 | that looks like: | |
1561 | ||
1562 | @smallexample | |
1563 | struct @{ | |
1564 | union @{ | |
1565 | void (*fn)(); | |
1566 | ptrdiff_t vtable_index; | |
1567 | @}; | |
1568 | ptrdiff_t delta; | |
1569 | @}; | |
1570 | @end smallexample | |
1571 | ||
1572 | @noindent | |
1573 | The C++ compiler must use one bit to indicate whether the function that | |
1574 | will be called through a pointer-to-member-function is virtual. | |
1575 | Normally, we assume that the low-order bit of a function pointer must | |
1576 | always be zero. Then, by ensuring that the vtable_index is odd, we can | |
1577 | distinguish which variant of the union is in use. But, on some | |
1578 | platforms function pointers can be odd, and so this doesn't work. In | |
1579 | that case, we use the low-order bit of the @code{delta} field, and shift | |
1580 | the remainder of the @code{delta} field to the left. | |
1581 | ||
1582 | GCC will automatically make the right selection about where to store | |
1583 | this bit using the @code{FUNCTION_BOUNDARY} setting for your platform. | |
1584 | However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY} | |
1585 | set such that functions always start at even addresses, but the lowest | |
1586 | bit of pointers to functions indicate whether the function at that | |
1587 | address is in ARM or Thumb mode. If this is the case of your | |
1588 | architecture, you should define this macro to | |
1589 | @code{ptrmemfunc_vbit_in_delta}. | |
1590 | ||
1591 | In general, you should not have to define this macro. On architectures | |
1592 | in which function addresses are always even, according to | |
1593 | @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to | |
1594 | @code{ptrmemfunc_vbit_in_pfn}. | |
1595 | @end defmac | |
1596 | ||
1597 | @defmac TARGET_VTABLE_USES_DESCRIPTORS | |
1598 | Normally, the C++ compiler uses function pointers in vtables. This | |
1599 | macro allows the target to change to use ``function descriptors'' | |
1600 | instead. Function descriptors are found on targets for whom a | |
1601 | function pointer is actually a small data structure. Normally the | |
1602 | data structure consists of the actual code address plus a data | |
1603 | pointer to which the function's data is relative. | |
1604 | ||
1605 | If vtables are used, the value of this macro should be the number | |
1606 | of words that the function descriptor occupies. | |
1607 | @end defmac | |
1608 | ||
1609 | @defmac TARGET_VTABLE_ENTRY_ALIGN | |
1610 | By default, the vtable entries are void pointers, the so the alignment | |
1611 | is the same as pointer alignment. The value of this macro specifies | |
1612 | the alignment of the vtable entry in bits. It should be defined only | |
1613 | when special alignment is necessary. */ | |
1614 | @end defmac | |
1615 | ||
1616 | @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE | |
1617 | There are a few non-descriptor entries in the vtable at offsets below | |
1618 | zero. If these entries must be padded (say, to preserve the alignment | |
1619 | specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number | |
1620 | of words in each data entry. | |
1621 | @end defmac | |
1622 | ||
1623 | @node Registers | |
1624 | @section Register Usage | |
1625 | @cindex register usage | |
1626 | ||
1627 | This section explains how to describe what registers the target machine | |
1628 | has, and how (in general) they can be used. | |
1629 | ||
1630 | The description of which registers a specific instruction can use is | |
1631 | done with register classes; see @ref{Register Classes}. For information | |
1632 | on using registers to access a stack frame, see @ref{Frame Registers}. | |
1633 | For passing values in registers, see @ref{Register Arguments}. | |
1634 | For returning values in registers, see @ref{Scalar Return}. | |
1635 | ||
1636 | @menu | |
1637 | * Register Basics:: Number and kinds of registers. | |
1638 | * Allocation Order:: Order in which registers are allocated. | |
1639 | * Values in Registers:: What kinds of values each reg can hold. | |
1640 | * Leaf Functions:: Renumbering registers for leaf functions. | |
1641 | * Stack Registers:: Handling a register stack such as 80387. | |
1642 | @end menu | |
1643 | ||
1644 | @node Register Basics | |
1645 | @subsection Basic Characteristics of Registers | |
1646 | ||
1647 | @c prevent bad page break with this line | |
1648 | Registers have various characteristics. | |
1649 | ||
1650 | @defmac FIRST_PSEUDO_REGISTER | |
1651 | Number of hardware registers known to the compiler. They receive | |
1652 | numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first | |
1653 | pseudo register's number really is assigned the number | |
1654 | @code{FIRST_PSEUDO_REGISTER}. | |
1655 | @end defmac | |
1656 | ||
1657 | @defmac FIXED_REGISTERS | |
1658 | @cindex fixed register | |
1659 | An initializer that says which registers are used for fixed purposes | |
1660 | all throughout the compiled code and are therefore not available for | |
1661 | general allocation. These would include the stack pointer, the frame | |
1662 | pointer (except on machines where that can be used as a general | |
1663 | register when no frame pointer is needed), the program counter on | |
1664 | machines where that is considered one of the addressable registers, | |
1665 | and any other numbered register with a standard use. | |
1666 | ||
1667 | This information is expressed as a sequence of numbers, separated by | |
1668 | commas and surrounded by braces. The @var{n}th number is 1 if | |
1669 | register @var{n} is fixed, 0 otherwise. | |
1670 | ||
1671 | The table initialized from this macro, and the table initialized by | |
1672 | the following one, may be overridden at run time either automatically, | |
1673 | by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by | |
1674 | the user with the command options @option{-ffixed-@var{reg}}, | |
1675 | @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}. | |
1676 | @end defmac | |
1677 | ||
1678 | @defmac CALL_USED_REGISTERS | |
1679 | @cindex call-used register | |
1680 | @cindex call-clobbered register | |
1681 | @cindex call-saved register | |
1682 | Like @code{FIXED_REGISTERS} but has 1 for each register that is | |
1683 | clobbered (in general) by function calls as well as for fixed | |
1684 | registers. This macro therefore identifies the registers that are not | |
1685 | available for general allocation of values that must live across | |
1686 | function calls. | |
1687 | ||
1688 | If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler | |
1689 | automatically saves it on function entry and restores it on function | |
1690 | exit, if the register is used within the function. | |
1691 | @end defmac | |
1692 | ||
1693 | @defmac CALL_REALLY_USED_REGISTERS | |
1694 | @cindex call-used register | |
1695 | @cindex call-clobbered register | |
1696 | @cindex call-saved register | |
1697 | Like @code{CALL_USED_REGISTERS} except this macro doesn't require | |
1698 | that the entire set of @code{FIXED_REGISTERS} be included. | |
1699 | (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}). | |
1700 | This macro is optional. If not specified, it defaults to the value | |
1701 | of @code{CALL_USED_REGISTERS}. | |
1702 | @end defmac | |
1703 | ||
1704 | @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode}) | |
1705 | @cindex call-used register | |
1706 | @cindex call-clobbered register | |
1707 | @cindex call-saved register | |
1708 | A C expression that is nonzero if it is not permissible to store a | |
1709 | value of mode @var{mode} in hard register number @var{regno} across a | |
1710 | call without some part of it being clobbered. For most machines this | |
1711 | macro need not be defined. It is only required for machines that do not | |
1712 | preserve the entire contents of a register across a call. | |
1713 | @end defmac | |
1714 | ||
1715 | @findex fixed_regs | |
1716 | @findex call_used_regs | |
1717 | @findex global_regs | |
1718 | @findex reg_names | |
1719 | @findex reg_class_contents | |
b2d7ede1 | 1720 | @hook TARGET_CONDITIONAL_REGISTER_USAGE |
202d6e5f | 1721 | |
1722 | @defmac INCOMING_REGNO (@var{out}) | |
1723 | Define this macro if the target machine has register windows. This C | |
1724 | expression returns the register number as seen by the called function | |
1725 | corresponding to the register number @var{out} as seen by the calling | |
1726 | function. Return @var{out} if register number @var{out} is not an | |
1727 | outbound register. | |
1728 | @end defmac | |
1729 | ||
1730 | @defmac OUTGOING_REGNO (@var{in}) | |
1731 | Define this macro if the target machine has register windows. This C | |
1732 | expression returns the register number as seen by the calling function | |
1733 | corresponding to the register number @var{in} as seen by the called | |
1734 | function. Return @var{in} if register number @var{in} is not an inbound | |
1735 | register. | |
1736 | @end defmac | |
1737 | ||
1738 | @defmac LOCAL_REGNO (@var{regno}) | |
1739 | Define this macro if the target machine has register windows. This C | |
1740 | expression returns true if the register is call-saved but is in the | |
1741 | register window. Unlike most call-saved registers, such registers | |
1742 | need not be explicitly restored on function exit or during non-local | |
1743 | gotos. | |
1744 | @end defmac | |
1745 | ||
1746 | @defmac PC_REGNUM | |
1747 | If the program counter has a register number, define this as that | |
1748 | register number. Otherwise, do not define it. | |
1749 | @end defmac | |
1750 | ||
1751 | @node Allocation Order | |
1752 | @subsection Order of Allocation of Registers | |
1753 | @cindex order of register allocation | |
1754 | @cindex register allocation order | |
1755 | ||
1756 | @c prevent bad page break with this line | |
1757 | Registers are allocated in order. | |
1758 | ||
1759 | @defmac REG_ALLOC_ORDER | |
1760 | If defined, an initializer for a vector of integers, containing the | |
1761 | numbers of hard registers in the order in which GCC should prefer | |
1762 | to use them (from most preferred to least). | |
1763 | ||
1764 | If this macro is not defined, registers are used lowest numbered first | |
1765 | (all else being equal). | |
1766 | ||
1767 | One use of this macro is on machines where the highest numbered | |
1768 | registers must always be saved and the save-multiple-registers | |
1769 | instruction supports only sequences of consecutive registers. On such | |
1770 | machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists | |
1771 | the highest numbered allocable register first. | |
1772 | @end defmac | |
1773 | ||
1774 | @defmac ADJUST_REG_ALLOC_ORDER | |
1775 | A C statement (sans semicolon) to choose the order in which to allocate | |
1776 | hard registers for pseudo-registers local to a basic block. | |
1777 | ||
1778 | Store the desired register order in the array @code{reg_alloc_order}. | |
1779 | Element 0 should be the register to allocate first; element 1, the next | |
1780 | register; and so on. | |
1781 | ||
1782 | The macro body should not assume anything about the contents of | |
1783 | @code{reg_alloc_order} before execution of the macro. | |
1784 | ||
1785 | On most machines, it is not necessary to define this macro. | |
1786 | @end defmac | |
1787 | ||
1788 | @defmac HONOR_REG_ALLOC_ORDER | |
1789 | Normally, IRA tries to estimate the costs for saving a register in the | |
1790 | prologue and restoring it in the epilogue. This discourages it from | |
1791 | using call-saved registers. If a machine wants to ensure that IRA | |
1792 | allocates registers in the order given by REG_ALLOC_ORDER even if some | |
d44f2f7c | 1793 | call-saved registers appear earlier than call-used ones, then define this |
1794 | macro as a C expression to nonzero. Default is 0. | |
202d6e5f | 1795 | @end defmac |
1796 | ||
1797 | @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno}) | |
1798 | In some case register allocation order is not enough for the | |
1799 | Integrated Register Allocator (@acronym{IRA}) to generate a good code. | |
1800 | If this macro is defined, it should return a floating point value | |
1801 | based on @var{regno}. The cost of using @var{regno} for a pseudo will | |
1802 | be increased by approximately the pseudo's usage frequency times the | |
1803 | value returned by this macro. Not defining this macro is equivalent | |
1804 | to having it always return @code{0.0}. | |
1805 | ||
1806 | On most machines, it is not necessary to define this macro. | |
1807 | @end defmac | |
1808 | ||
1809 | @node Values in Registers | |
1810 | @subsection How Values Fit in Registers | |
1811 | ||
1812 | This section discusses the macros that describe which kinds of values | |
1813 | (specifically, which machine modes) each register can hold, and how many | |
1814 | consecutive registers are needed for a given mode. | |
1815 | ||
1816 | @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode}) | |
1817 | A C expression for the number of consecutive hard registers, starting | |
1818 | at register number @var{regno}, required to hold a value of mode | |
1819 | @var{mode}. This macro must never return zero, even if a register | |
1820 | cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK | |
1821 | and/or CANNOT_CHANGE_MODE_CLASS instead. | |
1822 | ||
1823 | On a machine where all registers are exactly one word, a suitable | |
1824 | definition of this macro is | |
1825 | ||
1826 | @smallexample | |
1827 | #define HARD_REGNO_NREGS(REGNO, MODE) \ | |
1828 | ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \ | |
1829 | / UNITS_PER_WORD) | |
1830 | @end smallexample | |
1831 | @end defmac | |
1832 | ||
1833 | @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode}) | |
1834 | A C expression that is nonzero if a value of mode @var{mode}, stored | |
1835 | in memory, ends with padding that causes it to take up more space than | |
1836 | in registers starting at register number @var{regno} (as determined by | |
1837 | multiplying GCC's notion of the size of the register when containing | |
1838 | this mode by the number of registers returned by | |
1839 | @code{HARD_REGNO_NREGS}). By default this is zero. | |
1840 | ||
1841 | For example, if a floating-point value is stored in three 32-bit | |
1842 | registers but takes up 128 bits in memory, then this would be | |
1843 | nonzero. | |
1844 | ||
1845 | This macros only needs to be defined if there are cases where | |
1846 | @code{subreg_get_info} | |
1847 | would otherwise wrongly determine that a @code{subreg} can be | |
1848 | represented by an offset to the register number, when in fact such a | |
1849 | @code{subreg} would contain some of the padding not stored in | |
1850 | registers and so not be representable. | |
1851 | @end defmac | |
1852 | ||
1853 | @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode}) | |
1854 | For values of @var{regno} and @var{mode} for which | |
1855 | @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression | |
1856 | returning the greater number of registers required to hold the value | |
1857 | including any padding. In the example above, the value would be four. | |
1858 | @end defmac | |
1859 | ||
1860 | @defmac REGMODE_NATURAL_SIZE (@var{mode}) | |
1861 | Define this macro if the natural size of registers that hold values | |
1862 | of mode @var{mode} is not the word size. It is a C expression that | |
1863 | should give the natural size in bytes for the specified mode. It is | |
1864 | used by the register allocator to try to optimize its results. This | |
1865 | happens for example on SPARC 64-bit where the natural size of | |
1866 | floating-point registers is still 32-bit. | |
1867 | @end defmac | |
1868 | ||
1869 | @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode}) | |
1870 | A C expression that is nonzero if it is permissible to store a value | |
1871 | of mode @var{mode} in hard register number @var{regno} (or in several | |
1872 | registers starting with that one). For a machine where all registers | |
1873 | are equivalent, a suitable definition is | |
1874 | ||
1875 | @smallexample | |
1876 | #define HARD_REGNO_MODE_OK(REGNO, MODE) 1 | |
1877 | @end smallexample | |
1878 | ||
1879 | You need not include code to check for the numbers of fixed registers, | |
1880 | because the allocation mechanism considers them to be always occupied. | |
1881 | ||
1882 | @cindex register pairs | |
1883 | On some machines, double-precision values must be kept in even/odd | |
1884 | register pairs. You can implement that by defining this macro to reject | |
1885 | odd register numbers for such modes. | |
1886 | ||
1887 | The minimum requirement for a mode to be OK in a register is that the | |
1888 | @samp{mov@var{mode}} instruction pattern support moves between the | |
1889 | register and other hard register in the same class and that moving a | |
1890 | value into the register and back out not alter it. | |
1891 | ||
1892 | Since the same instruction used to move @code{word_mode} will work for | |
1893 | all narrower integer modes, it is not necessary on any machine for | |
1894 | @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided | |
1895 | you define patterns @samp{movhi}, etc., to take advantage of this. This | |
1896 | is useful because of the interaction between @code{HARD_REGNO_MODE_OK} | |
1897 | and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes | |
1898 | to be tieable. | |
1899 | ||
1900 | Many machines have special registers for floating point arithmetic. | |
1901 | Often people assume that floating point machine modes are allowed only | |
1902 | in floating point registers. This is not true. Any registers that | |
1903 | can hold integers can safely @emph{hold} a floating point machine | |
1904 | mode, whether or not floating arithmetic can be done on it in those | |
1905 | registers. Integer move instructions can be used to move the values. | |
1906 | ||
1907 | On some machines, though, the converse is true: fixed-point machine | |
1908 | modes may not go in floating registers. This is true if the floating | |
1909 | registers normalize any value stored in them, because storing a | |
1910 | non-floating value there would garble it. In this case, | |
1911 | @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in | |
1912 | floating registers. But if the floating registers do not automatically | |
1913 | normalize, if you can store any bit pattern in one and retrieve it | |
1914 | unchanged without a trap, then any machine mode may go in a floating | |
1915 | register, so you can define this macro to say so. | |
1916 | ||
1917 | The primary significance of special floating registers is rather that | |
1918 | they are the registers acceptable in floating point arithmetic | |
1919 | instructions. However, this is of no concern to | |
1920 | @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper | |
1921 | constraints for those instructions. | |
1922 | ||
1923 | On some machines, the floating registers are especially slow to access, | |
1924 | so that it is better to store a value in a stack frame than in such a | |
1925 | register if floating point arithmetic is not being done. As long as the | |
1926 | floating registers are not in class @code{GENERAL_REGS}, they will not | |
1927 | be used unless some pattern's constraint asks for one. | |
1928 | @end defmac | |
1929 | ||
1930 | @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to}) | |
1931 | A C expression that is nonzero if it is OK to rename a hard register | |
1932 | @var{from} to another hard register @var{to}. | |
1933 | ||
1934 | One common use of this macro is to prevent renaming of a register to | |
1935 | another register that is not saved by a prologue in an interrupt | |
1936 | handler. | |
1937 | ||
1938 | The default is always nonzero. | |
1939 | @end defmac | |
1940 | ||
1941 | @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2}) | |
1942 | A C expression that is nonzero if a value of mode | |
1943 | @var{mode1} is accessible in mode @var{mode2} without copying. | |
1944 | ||
1945 | If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and | |
1946 | @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for | |
1947 | any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})} | |
1948 | should be nonzero. If they differ for any @var{r}, you should define | |
1949 | this macro to return zero unless some other mechanism ensures the | |
1950 | accessibility of the value in a narrower mode. | |
1951 | ||
1952 | You should define this macro to return nonzero in as many cases as | |
1953 | possible since doing so will allow GCC to perform better register | |
1954 | allocation. | |
1955 | @end defmac | |
1956 | ||
1957 | @hook TARGET_HARD_REGNO_SCRATCH_OK | |
202d6e5f | 1958 | |
1959 | @defmac AVOID_CCMODE_COPIES | |
1960 | Define this macro if the compiler should avoid copies to/from @code{CCmode} | |
1961 | registers. You should only define this macro if support for copying to/from | |
1962 | @code{CCmode} is incomplete. | |
1963 | @end defmac | |
1964 | ||
1965 | @node Leaf Functions | |
1966 | @subsection Handling Leaf Functions | |
1967 | ||
1968 | @cindex leaf functions | |
1969 | @cindex functions, leaf | |
1970 | On some machines, a leaf function (i.e., one which makes no calls) can run | |
1971 | more efficiently if it does not make its own register window. Often this | |
1972 | means it is required to receive its arguments in the registers where they | |
1973 | are passed by the caller, instead of the registers where they would | |
1974 | normally arrive. | |
1975 | ||
1976 | The special treatment for leaf functions generally applies only when | |
1977 | other conditions are met; for example, often they may use only those | |
1978 | registers for its own variables and temporaries. We use the term ``leaf | |
1979 | function'' to mean a function that is suitable for this special | |
1980 | handling, so that functions with no calls are not necessarily ``leaf | |
1981 | functions''. | |
1982 | ||
1983 | GCC assigns register numbers before it knows whether the function is | |
1984 | suitable for leaf function treatment. So it needs to renumber the | |
1985 | registers in order to output a leaf function. The following macros | |
1986 | accomplish this. | |
1987 | ||
1988 | @defmac LEAF_REGISTERS | |
1989 | Name of a char vector, indexed by hard register number, which | |
1990 | contains 1 for a register that is allowable in a candidate for leaf | |
1991 | function treatment. | |
1992 | ||
1993 | If leaf function treatment involves renumbering the registers, then the | |
1994 | registers marked here should be the ones before renumbering---those that | |
1995 | GCC would ordinarily allocate. The registers which will actually be | |
1996 | used in the assembler code, after renumbering, should not be marked with 1 | |
1997 | in this vector. | |
1998 | ||
1999 | Define this macro only if the target machine offers a way to optimize | |
2000 | the treatment of leaf functions. | |
2001 | @end defmac | |
2002 | ||
2003 | @defmac LEAF_REG_REMAP (@var{regno}) | |
2004 | A C expression whose value is the register number to which @var{regno} | |
2005 | should be renumbered, when a function is treated as a leaf function. | |
2006 | ||
2007 | If @var{regno} is a register number which should not appear in a leaf | |
2008 | function before renumbering, then the expression should yield @minus{}1, which | |
2009 | will cause the compiler to abort. | |
2010 | ||
2011 | Define this macro only if the target machine offers a way to optimize the | |
2012 | treatment of leaf functions, and registers need to be renumbered to do | |
2013 | this. | |
2014 | @end defmac | |
2015 | ||
2016 | @findex current_function_is_leaf | |
2017 | @findex current_function_uses_only_leaf_regs | |
2018 | @code{TARGET_ASM_FUNCTION_PROLOGUE} and | |
2019 | @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions | |
2020 | specially. They can test the C variable @code{current_function_is_leaf} | |
2021 | which is nonzero for leaf functions. @code{current_function_is_leaf} is | |
2022 | set prior to local register allocation and is valid for the remaining | |
2023 | compiler passes. They can also test the C variable | |
2024 | @code{current_function_uses_only_leaf_regs} which is nonzero for leaf | |
2025 | functions which only use leaf registers. | |
2026 | @code{current_function_uses_only_leaf_regs} is valid after all passes | |
2027 | that modify the instructions have been run and is only useful if | |
2028 | @code{LEAF_REGISTERS} is defined. | |
2029 | @c changed this to fix overfull. ALSO: why the "it" at the beginning | |
2030 | @c of the next paragraph?! --mew 2feb93 | |
2031 | ||
2032 | @node Stack Registers | |
2033 | @subsection Registers That Form a Stack | |
2034 | ||
2035 | There are special features to handle computers where some of the | |
2036 | ``registers'' form a stack. Stack registers are normally written by | |
2037 | pushing onto the stack, and are numbered relative to the top of the | |
2038 | stack. | |
2039 | ||
2040 | Currently, GCC can only handle one group of stack-like registers, and | |
2041 | they must be consecutively numbered. Furthermore, the existing | |
2042 | support for stack-like registers is specific to the 80387 floating | |
2043 | point coprocessor. If you have a new architecture that uses | |
2044 | stack-like registers, you will need to do substantial work on | |
2045 | @file{reg-stack.c} and write your machine description to cooperate | |
2046 | with it, as well as defining these macros. | |
2047 | ||
2048 | @defmac STACK_REGS | |
2049 | Define this if the machine has any stack-like registers. | |
2050 | @end defmac | |
2051 | ||
2052 | @defmac STACK_REG_COVER_CLASS | |
2053 | This is a cover class containing the stack registers. Define this if | |
2054 | the machine has any stack-like registers. | |
2055 | @end defmac | |
2056 | ||
2057 | @defmac FIRST_STACK_REG | |
2058 | The number of the first stack-like register. This one is the top | |
2059 | of the stack. | |
2060 | @end defmac | |
2061 | ||
2062 | @defmac LAST_STACK_REG | |
2063 | The number of the last stack-like register. This one is the bottom of | |
2064 | the stack. | |
2065 | @end defmac | |
2066 | ||
2067 | @node Register Classes | |
2068 | @section Register Classes | |
2069 | @cindex register class definitions | |
2070 | @cindex class definitions, register | |
2071 | ||
2072 | On many machines, the numbered registers are not all equivalent. | |
2073 | For example, certain registers may not be allowed for indexed addressing; | |
2074 | certain registers may not be allowed in some instructions. These machine | |
2075 | restrictions are described to the compiler using @dfn{register classes}. | |
2076 | ||
2077 | You define a number of register classes, giving each one a name and saying | |
2078 | which of the registers belong to it. Then you can specify register classes | |
2079 | that are allowed as operands to particular instruction patterns. | |
2080 | ||
2081 | @findex ALL_REGS | |
2082 | @findex NO_REGS | |
2083 | In general, each register will belong to several classes. In fact, one | |
2084 | class must be named @code{ALL_REGS} and contain all the registers. Another | |
2085 | class must be named @code{NO_REGS} and contain no registers. Often the | |
2086 | union of two classes will be another class; however, this is not required. | |
2087 | ||
2088 | @findex GENERAL_REGS | |
2089 | One of the classes must be named @code{GENERAL_REGS}. There is nothing | |
2090 | terribly special about the name, but the operand constraint letters | |
2091 | @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is | |
2092 | the same as @code{ALL_REGS}, just define it as a macro which expands | |
2093 | to @code{ALL_REGS}. | |
2094 | ||
2095 | Order the classes so that if class @var{x} is contained in class @var{y} | |
2096 | then @var{x} has a lower class number than @var{y}. | |
2097 | ||
2098 | The way classes other than @code{GENERAL_REGS} are specified in operand | |
2099 | constraints is through machine-dependent operand constraint letters. | |
2100 | You can define such letters to correspond to various classes, then use | |
2101 | them in operand constraints. | |
2102 | ||
a4621139 | 2103 | You must define the narrowest register classes for allocatable |
2104 | registers, so that each class either has no subclasses, or that for | |
2105 | some mode, the move cost between registers within the class is | |
2106 | cheaper than moving a register in the class to or from memory | |
2107 | (@pxref{Costs}). | |
2108 | ||
202d6e5f | 2109 | You should define a class for the union of two classes whenever some |
2110 | instruction allows both classes. For example, if an instruction allows | |
2111 | either a floating point (coprocessor) register or a general register for a | |
2112 | certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS} | |
15edf55e | 2113 | which includes both of them. Otherwise you will get suboptimal code, |
2114 | or even internal compiler errors when reload cannot find a register in the | |
851d9296 | 2115 | class computed via @code{reg_class_subunion}. |
202d6e5f | 2116 | |
2117 | You must also specify certain redundant information about the register | |
2118 | classes: for each class, which classes contain it and which ones are | |
2119 | contained in it; for each pair of classes, the largest class contained | |
2120 | in their union. | |
2121 | ||
2122 | When a value occupying several consecutive registers is expected in a | |
2123 | certain class, all the registers used must belong to that class. | |
2124 | Therefore, register classes cannot be used to enforce a requirement for | |
2125 | a register pair to start with an even-numbered register. The way to | |
2126 | specify this requirement is with @code{HARD_REGNO_MODE_OK}. | |
2127 | ||
2128 | Register classes used for input-operands of bitwise-and or shift | |
2129 | instructions have a special requirement: each such class must have, for | |
2130 | each fixed-point machine mode, a subclass whose registers can transfer that | |
2131 | mode to or from memory. For example, on some machines, the operations for | |
2132 | single-byte values (@code{QImode}) are limited to certain registers. When | |
2133 | this is so, each register class that is used in a bitwise-and or shift | |
2134 | instruction must have a subclass consisting of registers from which | |
2135 | single-byte values can be loaded or stored. This is so that | |
2136 | @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return. | |
2137 | ||
2138 | @deftp {Data type} {enum reg_class} | |
2139 | An enumerated type that must be defined with all the register class names | |
2140 | as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS} | |
2141 | must be the last register class, followed by one more enumerated value, | |
2142 | @code{LIM_REG_CLASSES}, which is not a register class but rather | |
2143 | tells how many classes there are. | |
2144 | ||
2145 | Each register class has a number, which is the value of casting | |
2146 | the class name to type @code{int}. The number serves as an index | |
2147 | in many of the tables described below. | |
2148 | @end deftp | |
2149 | ||
2150 | @defmac N_REG_CLASSES | |
2151 | The number of distinct register classes, defined as follows: | |
2152 | ||
2153 | @smallexample | |
2154 | #define N_REG_CLASSES (int) LIM_REG_CLASSES | |
2155 | @end smallexample | |
2156 | @end defmac | |
2157 | ||
2158 | @defmac REG_CLASS_NAMES | |
2159 | An initializer containing the names of the register classes as C string | |
2160 | constants. These names are used in writing some of the debugging dumps. | |
2161 | @end defmac | |
2162 | ||
2163 | @defmac REG_CLASS_CONTENTS | |
2164 | An initializer containing the contents of the register classes, as integers | |
2165 | which are bit masks. The @var{n}th integer specifies the contents of class | |
2166 | @var{n}. The way the integer @var{mask} is interpreted is that | |
2167 | register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1. | |
2168 | ||
2169 | When the machine has more than 32 registers, an integer does not suffice. | |
2170 | Then the integers are replaced by sub-initializers, braced groupings containing | |
2171 | several integers. Each sub-initializer must be suitable as an initializer | |
2172 | for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}. | |
2173 | In this situation, the first integer in each sub-initializer corresponds to | |
2174 | registers 0 through 31, the second integer to registers 32 through 63, and | |
2175 | so on. | |
2176 | @end defmac | |
2177 | ||
2178 | @defmac REGNO_REG_CLASS (@var{regno}) | |
2179 | A C expression whose value is a register class containing hard register | |
2180 | @var{regno}. In general there is more than one such class; choose a class | |
2181 | which is @dfn{minimal}, meaning that no smaller class also contains the | |
2182 | register. | |
2183 | @end defmac | |
2184 | ||
2185 | @defmac BASE_REG_CLASS | |
2186 | A macro whose definition is the name of the class to which a valid | |
2187 | base register must belong. A base register is one used in an address | |
2188 | which is the register value plus a displacement. | |
2189 | @end defmac | |
2190 | ||
2191 | @defmac MODE_BASE_REG_CLASS (@var{mode}) | |
2192 | This is a variation of the @code{BASE_REG_CLASS} macro which allows | |
2193 | the selection of a base register in a mode dependent manner. If | |
2194 | @var{mode} is VOIDmode then it should return the same value as | |
2195 | @code{BASE_REG_CLASS}. | |
2196 | @end defmac | |
2197 | ||
2198 | @defmac MODE_BASE_REG_REG_CLASS (@var{mode}) | |
2199 | A C expression whose value is the register class to which a valid | |
2200 | base register must belong in order to be used in a base plus index | |
2201 | register address. You should define this macro if base plus index | |
2202 | addresses have different requirements than other base register uses. | |
2203 | @end defmac | |
2204 | ||
f8a8fc7b | 2205 | @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{address_space}, @var{outer_code}, @var{index_code}) |
202d6e5f | 2206 | A C expression whose value is the register class to which a valid |
f8a8fc7b | 2207 | base register for a memory reference in mode @var{mode} to address |
2208 | space @var{address_space} must belong. @var{outer_code} and @var{index_code} | |
2209 | define the context in which the base register occurs. @var{outer_code} is | |
2210 | the code of the immediately enclosing expression (@code{MEM} for the top level | |
2211 | of an address, @code{ADDRESS} for something that occurs in an | |
202d6e5f | 2212 | @code{address_operand}). @var{index_code} is the code of the corresponding |
2213 | index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise. | |
2214 | @end defmac | |
2215 | ||
2216 | @defmac INDEX_REG_CLASS | |
2217 | A macro whose definition is the name of the class to which a valid | |
2218 | index register must belong. An index register is one used in an | |
2219 | address where its value is either multiplied by a scale factor or | |
2220 | added to another register (as well as added to a displacement). | |
2221 | @end defmac | |
2222 | ||
2223 | @defmac REGNO_OK_FOR_BASE_P (@var{num}) | |
2224 | A C expression which is nonzero if register number @var{num} is | |
2225 | suitable for use as a base register in operand addresses. | |
202d6e5f | 2226 | @end defmac |
2227 | ||
2228 | @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode}) | |
2229 | A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that | |
2230 | that expression may examine the mode of the memory reference in | |
2231 | @var{mode}. You should define this macro if the mode of the memory | |
2232 | reference affects whether a register may be used as a base register. If | |
2233 | you define this macro, the compiler will use it instead of | |
2234 | @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for | |
2235 | addresses that appear outside a @code{MEM}, i.e., as an | |
2236 | @code{address_operand}. | |
202d6e5f | 2237 | @end defmac |
2238 | ||
2239 | @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode}) | |
2240 | A C expression which is nonzero if register number @var{num} is suitable for | |
2241 | use as a base register in base plus index operand addresses, accessing | |
2242 | memory in mode @var{mode}. It may be either a suitable hard register or a | |
2243 | pseudo register that has been allocated such a hard register. You should | |
2244 | define this macro if base plus index addresses have different requirements | |
2245 | than other base register uses. | |
2246 | ||
2247 | Use of this macro is deprecated; please use the more general | |
2248 | @code{REGNO_MODE_CODE_OK_FOR_BASE_P}. | |
202d6e5f | 2249 | @end defmac |
2250 | ||
f8a8fc7b | 2251 | @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{address_space}, @var{outer_code}, @var{index_code}) |
2252 | A C expression which is nonzero if register number @var{num} is | |
2253 | suitable for use as a base register in operand addresses, accessing | |
2254 | memory in mode @var{mode} in address space @var{address_space}. | |
2255 | This is similar to @code{REGNO_MODE_OK_FOR_BASE_P}, except | |
202d6e5f | 2256 | that that expression may examine the context in which the register |
2257 | appears in the memory reference. @var{outer_code} is the code of the | |
2258 | immediately enclosing expression (@code{MEM} if at the top level of the | |
2259 | address, @code{ADDRESS} for something that occurs in an | |
2260 | @code{address_operand}). @var{index_code} is the code of the | |
2261 | corresponding index expression if @var{outer_code} is @code{PLUS}; | |
2262 | @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses | |
2263 | that appear outside a @code{MEM}, i.e., as an @code{address_operand}. | |
202d6e5f | 2264 | @end defmac |
2265 | ||
2266 | @defmac REGNO_OK_FOR_INDEX_P (@var{num}) | |
2267 | A C expression which is nonzero if register number @var{num} is | |
2268 | suitable for use as an index register in operand addresses. It may be | |
2269 | either a suitable hard register or a pseudo register that has been | |
2270 | allocated such a hard register. | |
2271 | ||
2272 | The difference between an index register and a base register is that | |
2273 | the index register may be scaled. If an address involves the sum of | |
2274 | two registers, neither one of them scaled, then either one may be | |
2275 | labeled the ``base'' and the other the ``index''; but whichever | |
2276 | labeling is used must fit the machine's constraints of which registers | |
2277 | may serve in each capacity. The compiler will try both labelings, | |
2278 | looking for one that is valid, and will reload one or both registers | |
2279 | only if neither labeling works. | |
202d6e5f | 2280 | @end defmac |
2281 | ||
d78118a3 | 2282 | @hook TARGET_PREFERRED_RENAME_CLASS |
2283 | ||
09a17585 | 2284 | @hook TARGET_PREFERRED_RELOAD_CLASS |
09a17585 | 2285 | |
202d6e5f | 2286 | @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class}) |
2287 | A C expression that places additional restrictions on the register class | |
2288 | to use when it is necessary to copy value @var{x} into a register in class | |
2289 | @var{class}. The value is a register class; perhaps @var{class}, or perhaps | |
2290 | another, smaller class. On many machines, the following definition is | |
2291 | safe: | |
2292 | ||
2293 | @smallexample | |
2294 | #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS | |
2295 | @end smallexample | |
2296 | ||
2297 | Sometimes returning a more restrictive class makes better code. For | |
2298 | example, on the 68000, when @var{x} is an integer constant that is in range | |
2299 | for a @samp{moveq} instruction, the value of this macro is always | |
2300 | @code{DATA_REGS} as long as @var{class} includes the data registers. | |
2301 | Requiring a data register guarantees that a @samp{moveq} will be used. | |
2302 | ||
2303 | One case where @code{PREFERRED_RELOAD_CLASS} must not return | |
2304 | @var{class} is if @var{x} is a legitimate constant which cannot be | |
2305 | loaded into some register class. By returning @code{NO_REGS} you can | |
2306 | force @var{x} into a memory location. For example, rs6000 can load | |
2307 | immediate values into general-purpose registers, but does not have an | |
2308 | instruction for loading an immediate value into a floating-point | |
2309 | register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when | |
2310 | @var{x} is a floating-point constant. If the constant can't be loaded | |
2311 | into any kind of register, code generation will be better if | |
ca316360 | 2312 | @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead |
2313 | of using @code{TARGET_PREFERRED_RELOAD_CLASS}. | |
202d6e5f | 2314 | |
2315 | If an insn has pseudos in it after register allocation, reload will go | |
2316 | through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS} | |
2317 | to find the best one. Returning @code{NO_REGS}, in this case, makes | |
2318 | reload add a @code{!} in front of the constraint: the x86 back-end uses | |
2319 | this feature to discourage usage of 387 registers when math is done in | |
2320 | the SSE registers (and vice versa). | |
2321 | @end defmac | |
2322 | ||
71db0d8b | 2323 | @hook TARGET_PREFERRED_OUTPUT_RELOAD_CLASS |
71db0d8b | 2324 | |
202d6e5f | 2325 | @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class}) |
2326 | A C expression that places additional restrictions on the register class | |
2327 | to use when it is necessary to be able to hold a value of mode | |
2328 | @var{mode} in a reload register for which class @var{class} would | |
2329 | ordinarily be used. | |
2330 | ||
2331 | Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when | |
2332 | there are certain modes that simply can't go in certain reload classes. | |
2333 | ||
2334 | The value is a register class; perhaps @var{class}, or perhaps another, | |
2335 | smaller class. | |
2336 | ||
2337 | Don't define this macro unless the target machine has limitations which | |
2338 | require the macro to do something nontrivial. | |
2339 | @end defmac | |
2340 | ||
2341 | @hook TARGET_SECONDARY_RELOAD | |
202d6e5f | 2342 | |
2343 | @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x}) | |
2344 | @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x}) | |
2345 | @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x}) | |
2346 | These macros are obsolete, new ports should use the target hook | |
2347 | @code{TARGET_SECONDARY_RELOAD} instead. | |
2348 | ||
2349 | These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD} | |
2350 | target hook. Older ports still define these macros to indicate to the | |
2351 | reload phase that it may | |
2352 | need to allocate at least one register for a reload in addition to the | |
2353 | register to contain the data. Specifically, if copying @var{x} to a | |
2354 | register @var{class} in @var{mode} requires an intermediate register, | |
2355 | you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the | |
2356 | largest register class all of whose registers can be used as | |
2357 | intermediate registers or scratch registers. | |
2358 | ||
2359 | If copying a register @var{class} in @var{mode} to @var{x} requires an | |
2360 | intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS} | |
2361 | was supposed to be defined be defined to return the largest register | |
2362 | class required. If the | |
2363 | requirements for input and output reloads were the same, the macro | |
2364 | @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both | |
2365 | macros identically. | |
2366 | ||
2367 | The values returned by these macros are often @code{GENERAL_REGS}. | |
2368 | Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x} | |
2369 | can be directly copied to or from a register of @var{class} in | |
2370 | @var{mode} without requiring a scratch register. Do not define this | |
2371 | macro if it would always return @code{NO_REGS}. | |
2372 | ||
2373 | If a scratch register is required (either with or without an | |
2374 | intermediate register), you were supposed to define patterns for | |
2375 | @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required | |
2376 | (@pxref{Standard Names}. These patterns, which were normally | |
2377 | implemented with a @code{define_expand}, should be similar to the | |
2378 | @samp{mov@var{m}} patterns, except that operand 2 is the scratch | |
2379 | register. | |
2380 | ||
2381 | These patterns need constraints for the reload register and scratch | |
2382 | register that | |
2383 | contain a single register class. If the original reload register (whose | |
2384 | class is @var{class}) can meet the constraint given in the pattern, the | |
2385 | value returned by these macros is used for the class of the scratch | |
2386 | register. Otherwise, two additional reload registers are required. | |
2387 | Their classes are obtained from the constraints in the insn pattern. | |
2388 | ||
2389 | @var{x} might be a pseudo-register or a @code{subreg} of a | |
2390 | pseudo-register, which could either be in a hard register or in memory. | |
2391 | Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is | |
2392 | in memory and the hard register number if it is in a register. | |
2393 | ||
2394 | These macros should not be used in the case where a particular class of | |
2395 | registers can only be copied to memory and not to another class of | |
2396 | registers. In that case, secondary reload registers are not needed and | |
2397 | would not be helpful. Instead, a stack location must be used to perform | |
2398 | the copy and the @code{mov@var{m}} pattern should use memory as an | |
2399 | intermediate storage. This case often occurs between floating-point and | |
2400 | general registers. | |
2401 | @end defmac | |
2402 | ||
2403 | @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m}) | |
2404 | Certain machines have the property that some registers cannot be copied | |
2405 | to some other registers without using memory. Define this macro on | |
2406 | those machines to be a C expression that is nonzero if objects of mode | |
2407 | @var{m} in registers of @var{class1} can only be copied to registers of | |
2408 | class @var{class2} by storing a register of @var{class1} into memory | |
2409 | and loading that memory location into a register of @var{class2}. | |
2410 | ||
2411 | Do not define this macro if its value would always be zero. | |
2412 | @end defmac | |
2413 | ||
2414 | @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode}) | |
2415 | Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler | |
2416 | allocates a stack slot for a memory location needed for register copies. | |
2417 | If this macro is defined, the compiler instead uses the memory location | |
2418 | defined by this macro. | |
2419 | ||
2420 | Do not define this macro if you do not define | |
2421 | @code{SECONDARY_MEMORY_NEEDED}. | |
2422 | @end defmac | |
2423 | ||
2424 | @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode}) | |
2425 | When the compiler needs a secondary memory location to copy between two | |
2426 | registers of mode @var{mode}, it normally allocates sufficient memory to | |
2427 | hold a quantity of @code{BITS_PER_WORD} bits and performs the store and | |
2428 | load operations in a mode that many bits wide and whose class is the | |
2429 | same as that of @var{mode}. | |
2430 | ||
2431 | This is right thing to do on most machines because it ensures that all | |
2432 | bits of the register are copied and prevents accesses to the registers | |
2433 | in a narrower mode, which some machines prohibit for floating-point | |
2434 | registers. | |
2435 | ||
2436 | However, this default behavior is not correct on some machines, such as | |
2437 | the DEC Alpha, that store short integers in floating-point registers | |
2438 | differently than in integer registers. On those machines, the default | |
2439 | widening will not work correctly and you must define this macro to | |
2440 | suppress that widening in some cases. See the file @file{alpha.h} for | |
2441 | details. | |
2442 | ||
2443 | Do not define this macro if you do not define | |
2444 | @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that | |
2445 | is @code{BITS_PER_WORD} bits wide is correct for your machine. | |
2446 | @end defmac | |
2447 | ||
24dd0668 | 2448 | @hook TARGET_CLASS_LIKELY_SPILLED_P |
d3ba22dc | 2449 | |
2450 | @hook TARGET_CLASS_MAX_NREGS | |
24dd0668 | 2451 | |
202d6e5f | 2452 | @defmac CLASS_MAX_NREGS (@var{class}, @var{mode}) |
2453 | A C expression for the maximum number of consecutive registers | |
2454 | of class @var{class} needed to hold a value of mode @var{mode}. | |
2455 | ||
2456 | This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact, | |
2457 | the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})} | |
2458 | should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno}, | |
2459 | @var{mode})} for all @var{regno} values in the class @var{class}. | |
2460 | ||
2461 | This macro helps control the handling of multiple-word values | |
2462 | in the reload pass. | |
2463 | @end defmac | |
2464 | ||
2465 | @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class}) | |
2466 | If defined, a C expression that returns nonzero for a @var{class} for which | |
2467 | a change from mode @var{from} to mode @var{to} is invalid. | |
2468 | ||
e94743fa | 2469 | For example, loading 32-bit integer or floating-point objects into |
2470 | floating-point registers on Alpha extends them to 64 bits. | |
202d6e5f | 2471 | Therefore loading a 64-bit object and then storing it as a 32-bit object |
2472 | does not store the low-order 32 bits, as would be the case for a normal | |
2473 | register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS} | |
2474 | as below: | |
2475 | ||
2476 | @smallexample | |
2477 | #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \ | |
2478 | (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \ | |
2479 | ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0) | |
2480 | @end smallexample | |
e94743fa | 2481 | |
2482 | Even if storing from a register in mode @var{to} would be valid, | |
2483 | if both @var{from} and @code{raw_reg_mode} for @var{class} are wider | |
2484 | than @code{word_mode}, then we must prevent @var{to} narrowing the | |
2485 | mode. This happens when the middle-end assumes that it can load | |
2486 | or store pieces of an @var{N}-word pseudo, and that the pseudo will | |
2487 | eventually be allocated to @var{N} @code{word_mode} hard registers. | |
2488 | Failure to prevent this kind of mode change will result in the | |
2489 | entire @code{raw_reg_mode} being modified instead of the partial | |
2490 | value that the middle-end intended. | |
2491 | ||
202d6e5f | 2492 | @end defmac |
2493 | ||
20c3c7fc | 2494 | @hook TARGET_IRA_CHANGE_PSEUDO_ALLOCNO_CLASS |
2495 | ||
c6a6cdaa | 2496 | @hook TARGET_LRA_P |
2497 | ||
2498 | @hook TARGET_REGISTER_PRIORITY | |
2499 | ||
4a2ca8f3 | 2500 | @hook TARGET_REGISTER_USAGE_LEVELING_P |
2501 | ||
c6a6cdaa | 2502 | @hook TARGET_DIFFERENT_ADDR_DISPLACEMENT_P |
2503 | ||
f4447329 | 2504 | @hook TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P |
2505 | ||
968ba45e | 2506 | @hook TARGET_LEGITIMIZE_ADDRESS_DISPLACEMENT |
2507 | ||
c6a6cdaa | 2508 | @hook TARGET_SPILL_CLASS |
2509 | ||
17f446a0 | 2510 | @hook TARGET_CSTORE_MODE |
2511 | ||
202d6e5f | 2512 | @node Stack and Calling |
2513 | @section Stack Layout and Calling Conventions | |
2514 | @cindex calling conventions | |
2515 | ||
2516 | @c prevent bad page break with this line | |
2517 | This describes the stack layout and calling conventions. | |
2518 | ||
2519 | @menu | |
2520 | * Frame Layout:: | |
2521 | * Exception Handling:: | |
2522 | * Stack Checking:: | |
2523 | * Frame Registers:: | |
2524 | * Elimination:: | |
2525 | * Stack Arguments:: | |
2526 | * Register Arguments:: | |
2527 | * Scalar Return:: | |
2528 | * Aggregate Return:: | |
2529 | * Caller Saves:: | |
2530 | * Function Entry:: | |
2531 | * Profiling:: | |
2532 | * Tail Calls:: | |
2533 | * Stack Smashing Protection:: | |
e9eaaa6a | 2534 | * Miscellaneous Register Hooks:: |
202d6e5f | 2535 | @end menu |
2536 | ||
2537 | @node Frame Layout | |
2538 | @subsection Basic Stack Layout | |
2539 | @cindex stack frame layout | |
2540 | @cindex frame layout | |
2541 | ||
2542 | @c prevent bad page break with this line | |
2543 | Here is the basic stack layout. | |
2544 | ||
2545 | @defmac STACK_GROWS_DOWNWARD | |
2b785411 | 2546 | Define this macro to be true if pushing a word onto the stack moves the stack |
2547 | pointer to a smaller address, and false otherwise. | |
202d6e5f | 2548 | @end defmac |
2549 | ||
2550 | @defmac STACK_PUSH_CODE | |
2551 | This macro defines the operation used when something is pushed | |
2552 | on the stack. In RTL, a push operation will be | |
2553 | @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})} | |
2554 | ||
2555 | The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC}, | |
2556 | and @code{POST_INC}. Which of these is correct depends on | |
2557 | the stack direction and on whether the stack pointer points | |
2558 | to the last item on the stack or whether it points to the | |
2559 | space for the next item on the stack. | |
2560 | ||
2561 | The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is | |
2b785411 | 2562 | true, which is almost always right, and @code{PRE_INC} otherwise, |
202d6e5f | 2563 | which is often wrong. |
2564 | @end defmac | |
2565 | ||
2566 | @defmac FRAME_GROWS_DOWNWARD | |
2567 | Define this macro to nonzero value if the addresses of local variable slots | |
2568 | are at negative offsets from the frame pointer. | |
2569 | @end defmac | |
2570 | ||
2571 | @defmac ARGS_GROW_DOWNWARD | |
2572 | Define this macro if successive arguments to a function occupy decreasing | |
2573 | addresses on the stack. | |
2574 | @end defmac | |
2575 | ||
2576 | @defmac STARTING_FRAME_OFFSET | |
2577 | Offset from the frame pointer to the first local variable slot to be allocated. | |
2578 | ||
2579 | If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by | |
2580 | subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}. | |
2581 | Otherwise, it is found by adding the length of the first slot to the | |
2582 | value @code{STARTING_FRAME_OFFSET}. | |
2583 | @c i'm not sure if the above is still correct.. had to change it to get | |
2584 | @c rid of an overfull. --mew 2feb93 | |
2585 | @end defmac | |
2586 | ||
2587 | @defmac STACK_ALIGNMENT_NEEDED | |
2588 | Define to zero to disable final alignment of the stack during reload. | |
2589 | The nonzero default for this macro is suitable for most ports. | |
2590 | ||
2591 | On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there | |
2592 | is a register save block following the local block that doesn't require | |
2593 | alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable | |
2594 | stack alignment and do it in the backend. | |
2595 | @end defmac | |
2596 | ||
2597 | @defmac STACK_POINTER_OFFSET | |
2598 | Offset from the stack pointer register to the first location at which | |
2599 | outgoing arguments are placed. If not specified, the default value of | |
2600 | zero is used. This is the proper value for most machines. | |
2601 | ||
2602 | If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above | |
2603 | the first location at which outgoing arguments are placed. | |
2604 | @end defmac | |
2605 | ||
2606 | @defmac FIRST_PARM_OFFSET (@var{fundecl}) | |
2607 | Offset from the argument pointer register to the first argument's | |
2608 | address. On some machines it may depend on the data type of the | |
2609 | function. | |
2610 | ||
2611 | If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above | |
2612 | the first argument's address. | |
2613 | @end defmac | |
2614 | ||
2615 | @defmac STACK_DYNAMIC_OFFSET (@var{fundecl}) | |
2616 | Offset from the stack pointer register to an item dynamically allocated | |
2617 | on the stack, e.g., by @code{alloca}. | |
2618 | ||
2619 | The default value for this macro is @code{STACK_POINTER_OFFSET} plus the | |
2620 | length of the outgoing arguments. The default is correct for most | |
2621 | machines. See @file{function.c} for details. | |
2622 | @end defmac | |
2623 | ||
2624 | @defmac INITIAL_FRAME_ADDRESS_RTX | |
2625 | A C expression whose value is RTL representing the address of the initial | |
2626 | stack frame. This address is passed to @code{RETURN_ADDR_RTX} and | |
2627 | @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable | |
2628 | default value will be used. Define this macro in order to make frame pointer | |
2629 | elimination work in the presence of @code{__builtin_frame_address (count)} and | |
2630 | @code{__builtin_return_address (count)} for @code{count} not equal to zero. | |
2631 | @end defmac | |
2632 | ||
2633 | @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr}) | |
2634 | A C expression whose value is RTL representing the address in a stack | |
2635 | frame where the pointer to the caller's frame is stored. Assume that | |
2636 | @var{frameaddr} is an RTL expression for the address of the stack frame | |
2637 | itself. | |
2638 | ||
2639 | If you don't define this macro, the default is to return the value | |
2640 | of @var{frameaddr}---that is, the stack frame address is also the | |
2641 | address of the stack word that points to the previous frame. | |
2642 | @end defmac | |
2643 | ||
2644 | @defmac SETUP_FRAME_ADDRESSES | |
3dd01ce6 | 2645 | A C expression that produces the machine-specific code to |
202d6e5f | 2646 | setup the stack so that arbitrary frames can be accessed. For example, |
2647 | on the SPARC, we must flush all of the register windows to the stack | |
2648 | before we can access arbitrary stack frames. You will seldom need to | |
3dd01ce6 | 2649 | define this macro. The default is to do nothing. |
202d6e5f | 2650 | @end defmac |
2651 | ||
2652 | @hook TARGET_BUILTIN_SETJMP_FRAME_VALUE | |
202d6e5f | 2653 | |
2654 | @defmac FRAME_ADDR_RTX (@var{frameaddr}) | |
2655 | A C expression whose value is RTL representing the value of the frame | |
2656 | address for the current frame. @var{frameaddr} is the frame pointer | |
2657 | of the current frame. This is used for __builtin_frame_address. | |
2658 | You need only define this macro if the frame address is not the same | |
2659 | as the frame pointer. Most machines do not need to define it. | |
2660 | @end defmac | |
2661 | ||
2662 | @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr}) | |
2663 | A C expression whose value is RTL representing the value of the return | |
2664 | address for the frame @var{count} steps up from the current frame, after | |
2665 | the prologue. @var{frameaddr} is the frame pointer of the @var{count} | |
2666 | frame, or the frame pointer of the @var{count} @minus{} 1 frame if | |
a26d6c60 | 2667 | @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is nonzero. |
202d6e5f | 2668 | |
2669 | The value of the expression must always be the correct address when | |
2670 | @var{count} is zero, but may be @code{NULL_RTX} if there is no way to | |
2671 | determine the return address of other frames. | |
2672 | @end defmac | |
2673 | ||
2674 | @defmac RETURN_ADDR_IN_PREVIOUS_FRAME | |
a26d6c60 | 2675 | Define this macro to nonzero value if the return address of a particular |
2676 | stack frame is accessed from the frame pointer of the previous stack | |
2677 | frame. The zero default for this macro is suitable for most ports. | |
202d6e5f | 2678 | @end defmac |
2679 | ||
2680 | @defmac INCOMING_RETURN_ADDR_RTX | |
2681 | A C expression whose value is RTL representing the location of the | |
2682 | incoming return address at the beginning of any function, before the | |
2683 | prologue. This RTL is either a @code{REG}, indicating that the return | |
2684 | value is saved in @samp{REG}, or a @code{MEM} representing a location in | |
2685 | the stack. | |
2686 | ||
2687 | You only need to define this macro if you want to support call frame | |
2688 | debugging information like that provided by DWARF 2. | |
2689 | ||
2690 | If this RTL is a @code{REG}, you should also define | |
2691 | @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}. | |
2692 | @end defmac | |
2693 | ||
2694 | @defmac DWARF_ALT_FRAME_RETURN_COLUMN | |
2695 | A C expression whose value is an integer giving a DWARF 2 column | |
2696 | number that may be used as an alternative return column. The column | |
2697 | must not correspond to any gcc hard register (that is, it must not | |
2698 | be in the range of @code{DWARF_FRAME_REGNUM}). | |
2699 | ||
2700 | This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a | |
2701 | general register, but an alternative column needs to be used for signal | |
2702 | frames. Some targets have also used different frame return columns | |
2703 | over time. | |
2704 | @end defmac | |
2705 | ||
2706 | @defmac DWARF_ZERO_REG | |
2707 | A C expression whose value is an integer giving a DWARF 2 register | |
2708 | number that is considered to always have the value zero. This should | |
2709 | only be defined if the target has an architected zero register, and | |
2710 | someone decided it was a good idea to use that register number to | |
2711 | terminate the stack backtrace. New ports should avoid this. | |
2712 | @end defmac | |
2713 | ||
2714 | @hook TARGET_DWARF_HANDLE_FRAME_UNSPEC | |
202d6e5f | 2715 | |
2716 | @defmac INCOMING_FRAME_SP_OFFSET | |
2717 | A C expression whose value is an integer giving the offset, in bytes, | |
2718 | from the value of the stack pointer register to the top of the stack | |
2719 | frame at the beginning of any function, before the prologue. The top of | |
2720 | the frame is defined to be the value of the stack pointer in the | |
2721 | previous frame, just before the call instruction. | |
2722 | ||
2723 | You only need to define this macro if you want to support call frame | |
2724 | debugging information like that provided by DWARF 2. | |
2725 | @end defmac | |
2726 | ||
2727 | @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl}) | |
2728 | A C expression whose value is an integer giving the offset, in bytes, | |
2729 | from the argument pointer to the canonical frame address (cfa). The | |
2730 | final value should coincide with that calculated by | |
2731 | @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable | |
2732 | during virtual register instantiation. | |
2733 | ||
2734 | The default value for this macro is | |
2735 | @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size}, | |
2736 | which is correct for most machines; in general, the arguments are found | |
2737 | immediately before the stack frame. Note that this is not the case on | |
2738 | some targets that save registers into the caller's frame, such as SPARC | |
2739 | and rs6000, and so such targets need to define this macro. | |
2740 | ||
2741 | You only need to define this macro if the default is incorrect, and you | |
2742 | want to support call frame debugging information like that provided by | |
2743 | DWARF 2. | |
2744 | @end defmac | |
2745 | ||
2746 | @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl}) | |
2747 | If defined, a C expression whose value is an integer giving the offset | |
2748 | in bytes from the frame pointer to the canonical frame address (cfa). | |
2749 | The final value should coincide with that calculated by | |
2750 | @code{INCOMING_FRAME_SP_OFFSET}. | |
2751 | ||
2752 | Normally the CFA is calculated as an offset from the argument pointer, | |
2753 | via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is | |
2754 | variable due to the ABI, this may not be possible. If this macro is | |
2755 | defined, it implies that the virtual register instantiation should be | |
2756 | based on the frame pointer instead of the argument pointer. Only one | |
2757 | of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET} | |
2758 | should be defined. | |
2759 | @end defmac | |
2760 | ||
2761 | @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl}) | |
2762 | If defined, a C expression whose value is an integer giving the offset | |
2763 | in bytes from the canonical frame address (cfa) to the frame base used | |
2764 | in DWARF 2 debug information. The default is zero. A different value | |
2765 | may reduce the size of debug information on some ports. | |
2766 | @end defmac | |
2767 | ||
2768 | @node Exception Handling | |
2769 | @subsection Exception Handling Support | |
2770 | @cindex exception handling | |
2771 | ||
2772 | @defmac EH_RETURN_DATA_REGNO (@var{N}) | |
2773 | A C expression whose value is the @var{N}th register number used for | |
2774 | data by exception handlers, or @code{INVALID_REGNUM} if fewer than | |
2775 | @var{N} registers are usable. | |
2776 | ||
2777 | The exception handling library routines communicate with the exception | |
2778 | handlers via a set of agreed upon registers. Ideally these registers | |
2779 | should be call-clobbered; it is possible to use call-saved registers, | |
2780 | but may negatively impact code size. The target must support at least | |
2781 | 2 data registers, but should define 4 if there are enough free registers. | |
2782 | ||
2783 | You must define this macro if you want to support call frame exception | |
2784 | handling like that provided by DWARF 2. | |
2785 | @end defmac | |
2786 | ||
2787 | @defmac EH_RETURN_STACKADJ_RTX | |
2788 | A C expression whose value is RTL representing a location in which | |
2789 | to store a stack adjustment to be applied before function return. | |
2790 | This is used to unwind the stack to an exception handler's call frame. | |
2791 | It will be assigned zero on code paths that return normally. | |
2792 | ||
2793 | Typically this is a call-clobbered hard register that is otherwise | |
2794 | untouched by the epilogue, but could also be a stack slot. | |
2795 | ||
2796 | Do not define this macro if the stack pointer is saved and restored | |
2797 | by the regular prolog and epilog code in the call frame itself; in | |
2798 | this case, the exception handling library routines will update the | |
2799 | stack location to be restored in place. Otherwise, you must define | |
2800 | this macro if you want to support call frame exception handling like | |
2801 | that provided by DWARF 2. | |
2802 | @end defmac | |
2803 | ||
2804 | @defmac EH_RETURN_HANDLER_RTX | |
2805 | A C expression whose value is RTL representing a location in which | |
2806 | to store the address of an exception handler to which we should | |
2807 | return. It will not be assigned on code paths that return normally. | |
2808 | ||
2809 | Typically this is the location in the call frame at which the normal | |
2810 | return address is stored. For targets that return by popping an | |
2811 | address off the stack, this might be a memory address just below | |
2812 | the @emph{target} call frame rather than inside the current call | |
2813 | frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already | |
2814 | been assigned, so it may be used to calculate the location of the | |
2815 | target call frame. | |
2816 | ||
2817 | Some targets have more complex requirements than storing to an | |
2818 | address calculable during initial code generation. In that case | |
2819 | the @code{eh_return} instruction pattern should be used instead. | |
2820 | ||
2821 | If you want to support call frame exception handling, you must | |
2822 | define either this macro or the @code{eh_return} instruction pattern. | |
2823 | @end defmac | |
2824 | ||
2825 | @defmac RETURN_ADDR_OFFSET | |
2826 | If defined, an integer-valued C expression for which rtl will be generated | |
2827 | to add it to the exception handler address before it is searched in the | |
2828 | exception handling tables, and to subtract it again from the address before | |
2829 | using it to return to the exception handler. | |
2830 | @end defmac | |
2831 | ||
2832 | @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global}) | |
2833 | This macro chooses the encoding of pointers embedded in the exception | |
2834 | handling sections. If at all possible, this should be defined such | |
2835 | that the exception handling section will not require dynamic relocations, | |
2836 | and so may be read-only. | |
2837 | ||
2838 | @var{code} is 0 for data, 1 for code labels, 2 for function pointers. | |
2839 | @var{global} is true if the symbol may be affected by dynamic relocations. | |
2840 | The macro should return a combination of the @code{DW_EH_PE_*} defines | |
2841 | as found in @file{dwarf2.h}. | |
2842 | ||
2843 | If this macro is not defined, pointers will not be encoded but | |
2844 | represented directly. | |
2845 | @end defmac | |
2846 | ||
2847 | @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done}) | |
2848 | This macro allows the target to emit whatever special magic is required | |
2849 | to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}. | |
2850 | Generic code takes care of pc-relative and indirect encodings; this must | |
2851 | be defined if the target uses text-relative or data-relative encodings. | |
2852 | ||
2853 | This is a C statement that branches to @var{done} if the format was | |
2854 | handled. @var{encoding} is the format chosen, @var{size} is the number | |
2855 | of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF} | |
2856 | to be emitted. | |
2857 | @end defmac | |
2858 | ||
202d6e5f | 2859 | @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs}) |
2860 | This macro allows the target to add CPU and operating system specific | |
2861 | code to the call-frame unwinder for use when there is no unwind data | |
2862 | available. The most common reason to implement this macro is to unwind | |
2863 | through signal frames. | |
2864 | ||
2865 | This macro is called from @code{uw_frame_state_for} in | |
2866 | @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and | |
2867 | @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context}; | |
2868 | @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra} | |
2869 | for the address of the code being executed and @code{context->cfa} for | |
2870 | the stack pointer value. If the frame can be decoded, the register | |
2871 | save addresses should be updated in @var{fs} and the macro should | |
2872 | evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded, | |
2873 | the macro should evaluate to @code{_URC_END_OF_STACK}. | |
2874 | ||
2875 | For proper signal handling in Java this macro is accompanied by | |
2876 | @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers. | |
2877 | @end defmac | |
2878 | ||
2879 | @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs}) | |
2880 | This macro allows the target to add operating system specific code to the | |
2881 | call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive, | |
2882 | usually used for signal or interrupt frames. | |
2883 | ||
c07f27fd | 2884 | This macro is called from @code{uw_update_context} in libgcc's |
2885 | @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context}; | |
202d6e5f | 2886 | @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi} |
2887 | for the abi and context in the @code{.unwabi} directive. If the | |
2888 | @code{.unwabi} directive can be handled, the register save addresses should | |
2889 | be updated in @var{fs}. | |
2890 | @end defmac | |
2891 | ||
2892 | @defmac TARGET_USES_WEAK_UNWIND_INFO | |
2893 | A C expression that evaluates to true if the target requires unwind | |
2894 | info to be given comdat linkage. Define it to be @code{1} if comdat | |
2895 | linkage is necessary. The default is @code{0}. | |
2896 | @end defmac | |
2897 | ||
2898 | @node Stack Checking | |
2899 | @subsection Specifying How Stack Checking is Done | |
2900 | ||
2901 | GCC will check that stack references are within the boundaries of the | |
2902 | stack, if the option @option{-fstack-check} is specified, in one of | |
2903 | three ways: | |
2904 | ||
2905 | @enumerate | |
2906 | @item | |
2907 | If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC | |
2908 | will assume that you have arranged for full stack checking to be done | |
2909 | at appropriate places in the configuration files. GCC will not do | |
2910 | other special processing. | |
2911 | ||
2912 | @item | |
2913 | If @code{STACK_CHECK_BUILTIN} is zero and the value of the | |
2914 | @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume | |
2915 | that you have arranged for static stack checking (checking of the | |
2916 | static stack frame of functions) to be done at appropriate places | |
2917 | in the configuration files. GCC will only emit code to do dynamic | |
2918 | stack checking (checking on dynamic stack allocations) using the third | |
2919 | approach below. | |
2920 | ||
2921 | @item | |
2922 | If neither of the above are true, GCC will generate code to periodically | |
2923 | ``probe'' the stack pointer using the values of the macros defined below. | |
2924 | @end enumerate | |
2925 | ||
2926 | If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined, | |
2927 | GCC will change its allocation strategy for large objects if the option | |
2928 | @option{-fstack-check} is specified: they will always be allocated | |
2929 | dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes. | |
2930 | ||
2931 | @defmac STACK_CHECK_BUILTIN | |
2932 | A nonzero value if stack checking is done by the configuration files in a | |
2933 | machine-dependent manner. You should define this macro if stack checking | |
2934 | is required by the ABI of your machine or if you would like to do stack | |
2935 | checking in some more efficient way than the generic approach. The default | |
2936 | value of this macro is zero. | |
2937 | @end defmac | |
2938 | ||
2939 | @defmac STACK_CHECK_STATIC_BUILTIN | |
2940 | A nonzero value if static stack checking is done by the configuration files | |
2941 | in a machine-dependent manner. You should define this macro if you would | |
2942 | like to do static stack checking in some more efficient way than the generic | |
2943 | approach. The default value of this macro is zero. | |
2944 | @end defmac | |
2945 | ||
2946 | @defmac STACK_CHECK_PROBE_INTERVAL_EXP | |
2947 | An integer specifying the interval at which GCC must generate stack probe | |
2948 | instructions, defined as 2 raised to this integer. You will normally | |
2949 | define this macro so that the interval be no larger than the size of | |
2950 | the ``guard pages'' at the end of a stack area. The default value | |
2951 | of 12 (4096-byte interval) is suitable for most systems. | |
2952 | @end defmac | |
2953 | ||
2954 | @defmac STACK_CHECK_MOVING_SP | |
2955 | An integer which is nonzero if GCC should move the stack pointer page by page | |
2956 | when doing probes. This can be necessary on systems where the stack pointer | |
2957 | contains the bottom address of the memory area accessible to the executing | |
2958 | thread at any point in time. In this situation an alternate signal stack | |
2959 | is required in order to be able to recover from a stack overflow. The | |
2960 | default value of this macro is zero. | |
2961 | @end defmac | |
2962 | ||
2963 | @defmac STACK_CHECK_PROTECT | |
2964 | The number of bytes of stack needed to recover from a stack overflow, for | |
5ee1325b | 2965 | languages where such a recovery is supported. The default value of 4KB/8KB |
202d6e5f | 2966 | with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and |
5ee1325b | 2967 | 8KB/12KB with other exception handling mechanisms should be adequate for most |
2968 | architectures and operating systems. | |
202d6e5f | 2969 | @end defmac |
2970 | ||
2971 | The following macros are relevant only if neither STACK_CHECK_BUILTIN | |
2972 | nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether | |
2973 | in the opposite case. | |
2974 | ||
2975 | @defmac STACK_CHECK_MAX_FRAME_SIZE | |
2976 | The maximum size of a stack frame, in bytes. GCC will generate probe | |
2977 | instructions in non-leaf functions to ensure at least this many bytes of | |
2978 | stack are available. If a stack frame is larger than this size, stack | |
2979 | checking will not be reliable and GCC will issue a warning. The | |
2980 | default is chosen so that GCC only generates one instruction on most | |
2981 | systems. You should normally not change the default value of this macro. | |
2982 | @end defmac | |
2983 | ||
2984 | @defmac STACK_CHECK_FIXED_FRAME_SIZE | |
2985 | GCC uses this value to generate the above warning message. It | |
2986 | represents the amount of fixed frame used by a function, not including | |
2987 | space for any callee-saved registers, temporaries and user variables. | |
2988 | You need only specify an upper bound for this amount and will normally | |
2989 | use the default of four words. | |
2990 | @end defmac | |
2991 | ||
2992 | @defmac STACK_CHECK_MAX_VAR_SIZE | |
2993 | The maximum size, in bytes, of an object that GCC will place in the | |
2994 | fixed area of the stack frame when the user specifies | |
2995 | @option{-fstack-check}. | |
2996 | GCC computed the default from the values of the above macros and you will | |
2997 | normally not need to override that default. | |
2998 | @end defmac | |
2999 | ||
3000 | @need 2000 | |
3001 | @node Frame Registers | |
3002 | @subsection Registers That Address the Stack Frame | |
3003 | ||
3004 | @c prevent bad page break with this line | |
3005 | This discusses registers that address the stack frame. | |
3006 | ||
3007 | @defmac STACK_POINTER_REGNUM | |
3008 | The register number of the stack pointer register, which must also be a | |
3009 | fixed register according to @code{FIXED_REGISTERS}. On most machines, | |
3010 | the hardware determines which register this is. | |
3011 | @end defmac | |
3012 | ||
3013 | @defmac FRAME_POINTER_REGNUM | |
3014 | The register number of the frame pointer register, which is used to | |
3015 | access automatic variables in the stack frame. On some machines, the | |
3016 | hardware determines which register this is. On other machines, you can | |
3017 | choose any register you wish for this purpose. | |
3018 | @end defmac | |
3019 | ||
3020 | @defmac HARD_FRAME_POINTER_REGNUM | |
3021 | On some machines the offset between the frame pointer and starting | |
3022 | offset of the automatic variables is not known until after register | |
3023 | allocation has been done (for example, because the saved registers are | |
3024 | between these two locations). On those machines, define | |
3025 | @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to | |
3026 | be used internally until the offset is known, and define | |
3027 | @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number | |
3028 | used for the frame pointer. | |
3029 | ||
3030 | You should define this macro only in the very rare circumstances when it | |
3031 | is not possible to calculate the offset between the frame pointer and | |
3032 | the automatic variables until after register allocation has been | |
3033 | completed. When this macro is defined, you must also indicate in your | |
3034 | definition of @code{ELIMINABLE_REGS} how to eliminate | |
3035 | @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM} | |
3036 | or @code{STACK_POINTER_REGNUM}. | |
3037 | ||
3038 | Do not define this macro if it would be the same as | |
3039 | @code{FRAME_POINTER_REGNUM}. | |
3040 | @end defmac | |
3041 | ||
3042 | @defmac ARG_POINTER_REGNUM | |
3043 | The register number of the arg pointer register, which is used to access | |
3044 | the function's argument list. On some machines, this is the same as the | |
3045 | frame pointer register. On some machines, the hardware determines which | |
3046 | register this is. On other machines, you can choose any register you | |
3047 | wish for this purpose. If this is not the same register as the frame | |
3048 | pointer register, then you must mark it as a fixed register according to | |
3049 | @code{FIXED_REGISTERS}, or arrange to be able to eliminate it | |
3050 | (@pxref{Elimination}). | |
3051 | @end defmac | |
3052 | ||
5ae82d58 | 3053 | @defmac HARD_FRAME_POINTER_IS_FRAME_POINTER |
3054 | Define this to a preprocessor constant that is nonzero if | |
3055 | @code{hard_frame_pointer_rtx} and @code{frame_pointer_rtx} should be | |
3056 | the same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM | |
3057 | == FRAME_POINTER_REGNUM)}; you only need to define this macro if that | |
3058 | definition is not suitable for use in preprocessor conditionals. | |
3059 | @end defmac | |
3060 | ||
3061 | @defmac HARD_FRAME_POINTER_IS_ARG_POINTER | |
3062 | Define this to a preprocessor constant that is nonzero if | |
3063 | @code{hard_frame_pointer_rtx} and @code{arg_pointer_rtx} should be the | |
3064 | same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM == | |
3065 | ARG_POINTER_REGNUM)}; you only need to define this macro if that | |
3066 | definition is not suitable for use in preprocessor conditionals. | |
3067 | @end defmac | |
3068 | ||
202d6e5f | 3069 | @defmac RETURN_ADDRESS_POINTER_REGNUM |
3070 | The register number of the return address pointer register, which is used to | |
3071 | access the current function's return address from the stack. On some | |
3072 | machines, the return address is not at a fixed offset from the frame | |
3073 | pointer or stack pointer or argument pointer. This register can be defined | |
3074 | to point to the return address on the stack, and then be converted by | |
3075 | @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer. | |
3076 | ||
3077 | Do not define this macro unless there is no other way to get the return | |
3078 | address from the stack. | |
3079 | @end defmac | |
3080 | ||
3081 | @defmac STATIC_CHAIN_REGNUM | |
3082 | @defmacx STATIC_CHAIN_INCOMING_REGNUM | |
3083 | Register numbers used for passing a function's static chain pointer. If | |
3084 | register windows are used, the register number as seen by the called | |
3085 | function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register | |
3086 | number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If | |
3087 | these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need | |
3088 | not be defined. | |
3089 | ||
3090 | The static chain register need not be a fixed register. | |
3091 | ||
3092 | If the static chain is passed in memory, these macros should not be | |
3093 | defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used. | |
3094 | @end defmac | |
3095 | ||
3096 | @hook TARGET_STATIC_CHAIN | |
202d6e5f | 3097 | |
3098 | @defmac DWARF_FRAME_REGISTERS | |
3099 | This macro specifies the maximum number of hard registers that can be | |
3100 | saved in a call frame. This is used to size data structures used in | |
3101 | DWARF2 exception handling. | |
3102 | ||
3103 | Prior to GCC 3.0, this macro was needed in order to establish a stable | |
3104 | exception handling ABI in the face of adding new hard registers for ISA | |
3105 | extensions. In GCC 3.0 and later, the EH ABI is insulated from changes | |
3106 | in the number of hard registers. Nevertheless, this macro can still be | |
3107 | used to reduce the runtime memory requirements of the exception handling | |
3108 | routines, which can be substantial if the ISA contains a lot of | |
3109 | registers that are not call-saved. | |
3110 | ||
3111 | If this macro is not defined, it defaults to | |
3112 | @code{FIRST_PSEUDO_REGISTER}. | |
3113 | @end defmac | |
3114 | ||
3115 | @defmac PRE_GCC3_DWARF_FRAME_REGISTERS | |
3116 | ||
3117 | This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided | |
3118 | for backward compatibility in pre GCC 3.0 compiled code. | |
3119 | ||
3120 | If this macro is not defined, it defaults to | |
3121 | @code{DWARF_FRAME_REGISTERS}. | |
3122 | @end defmac | |
3123 | ||
3124 | @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno}) | |
3125 | ||
3126 | Define this macro if the target's representation for dwarf registers | |
3127 | is different than the internal representation for unwind column. | |
3128 | Given a dwarf register, this macro should return the internal unwind | |
3129 | column number to use instead. | |
3130 | ||
3131 | See the PowerPC's SPE target for an example. | |
3132 | @end defmac | |
3133 | ||
3134 | @defmac DWARF_FRAME_REGNUM (@var{regno}) | |
3135 | ||
3136 | Define this macro if the target's representation for dwarf registers | |
3137 | used in .eh_frame or .debug_frame is different from that used in other | |
3138 | debug info sections. Given a GCC hard register number, this macro | |
3139 | should return the .eh_frame register number. The default is | |
3140 | @code{DBX_REGISTER_NUMBER (@var{regno})}. | |
3141 | ||
3142 | @end defmac | |
3143 | ||
3144 | @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh}) | |
3145 | ||
3146 | Define this macro to map register numbers held in the call frame info | |
3147 | that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that | |
3148 | should be output in .debug_frame (@code{@var{for_eh}} is zero) and | |
3149 | .eh_frame (@code{@var{for_eh}} is nonzero). The default is to | |
3150 | return @code{@var{regno}}. | |
3151 | ||
3152 | @end defmac | |
3153 | ||
e451a093 | 3154 | @defmac REG_VALUE_IN_UNWIND_CONTEXT |
3155 | ||
3156 | Define this macro if the target stores register values as | |
3157 | @code{_Unwind_Word} type in unwind context. It should be defined if | |
3158 | target register size is larger than the size of @code{void *}. The | |
3159 | default is to store register values as @code{void *} type. | |
3160 | ||
3161 | @end defmac | |
3162 | ||
3163 | @defmac ASSUME_EXTENDED_UNWIND_CONTEXT | |
3164 | ||
3165 | Define this macro to be 1 if the target always uses extended unwind | |
3166 | context with version, args_size and by_value fields. If it is undefined, | |
3167 | it will be defined to 1 when @code{REG_VALUE_IN_UNWIND_CONTEXT} is | |
3168 | defined and 0 otherwise. | |
3169 | ||
3170 | @end defmac | |
3171 | ||
202d6e5f | 3172 | @node Elimination |
3173 | @subsection Eliminating Frame Pointer and Arg Pointer | |
3174 | ||
3175 | @c prevent bad page break with this line | |
3176 | This is about eliminating the frame pointer and arg pointer. | |
3177 | ||
3178 | @hook TARGET_FRAME_POINTER_REQUIRED | |
202d6e5f | 3179 | |
3180 | @findex get_frame_size | |
3181 | @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var}) | |
3182 | A C statement to store in the variable @var{depth-var} the difference | |
3183 | between the frame pointer and the stack pointer values immediately after | |
3184 | the function prologue. The value would be computed from information | |
3185 | such as the result of @code{get_frame_size ()} and the tables of | |
3186 | registers @code{regs_ever_live} and @code{call_used_regs}. | |
3187 | ||
3188 | If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and | |
3189 | need not be defined. Otherwise, it must be defined even if | |
3190 | @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that | |
3191 | case, you may set @var{depth-var} to anything. | |
3192 | @end defmac | |
3193 | ||
3194 | @defmac ELIMINABLE_REGS | |
3195 | If defined, this macro specifies a table of register pairs used to | |
3196 | eliminate unneeded registers that point into the stack frame. If it is not | |
3197 | defined, the only elimination attempted by the compiler is to replace | |
3198 | references to the frame pointer with references to the stack pointer. | |
3199 | ||
3200 | The definition of this macro is a list of structure initializations, each | |
3201 | of which specifies an original and replacement register. | |
3202 | ||
3203 | On some machines, the position of the argument pointer is not known until | |
3204 | the compilation is completed. In such a case, a separate hard register | |
3205 | must be used for the argument pointer. This register can be eliminated by | |
3206 | replacing it with either the frame pointer or the argument pointer, | |
3207 | depending on whether or not the frame pointer has been eliminated. | |
3208 | ||
3209 | In this case, you might specify: | |
3210 | @smallexample | |
3211 | #define ELIMINABLE_REGS \ | |
3212 | @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \ | |
3213 | @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \ | |
3214 | @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@} | |
3215 | @end smallexample | |
3216 | ||
3217 | Note that the elimination of the argument pointer with the stack pointer is | |
3218 | specified first since that is the preferred elimination. | |
3219 | @end defmac | |
3220 | ||
3221 | @hook TARGET_CAN_ELIMINATE | |
202d6e5f | 3222 | |
3223 | @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var}) | |
3224 | This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It | |
3225 | specifies the initial difference between the specified pair of | |
3226 | registers. This macro must be defined if @code{ELIMINABLE_REGS} is | |
3227 | defined. | |
3228 | @end defmac | |
3229 | ||
3230 | @node Stack Arguments | |
3231 | @subsection Passing Function Arguments on the Stack | |
3232 | @cindex arguments on stack | |
3233 | @cindex stack arguments | |
3234 | ||
3235 | The macros in this section control how arguments are passed | |
3236 | on the stack. See the following section for other macros that | |
3237 | control passing certain arguments in registers. | |
3238 | ||
3239 | @hook TARGET_PROMOTE_PROTOTYPES | |
202d6e5f | 3240 | |
3241 | @defmac PUSH_ARGS | |
3242 | A C expression. If nonzero, push insns will be used to pass | |
3243 | outgoing arguments. | |
3244 | If the target machine does not have a push instruction, set it to zero. | |
3245 | That directs GCC to use an alternate strategy: to | |
3246 | allocate the entire argument block and then store the arguments into | |
3247 | it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too. | |
3248 | @end defmac | |
3249 | ||
3250 | @defmac PUSH_ARGS_REVERSED | |
3251 | A C expression. If nonzero, function arguments will be evaluated from | |
3252 | last to first, rather than from first to last. If this macro is not | |
3253 | defined, it defaults to @code{PUSH_ARGS} on targets where the stack | |
3254 | and args grow in opposite directions, and 0 otherwise. | |
3255 | @end defmac | |
3256 | ||
3257 | @defmac PUSH_ROUNDING (@var{npushed}) | |
3258 | A C expression that is the number of bytes actually pushed onto the | |
3259 | stack when an instruction attempts to push @var{npushed} bytes. | |
3260 | ||
3261 | On some machines, the definition | |
3262 | ||
3263 | @smallexample | |
3264 | #define PUSH_ROUNDING(BYTES) (BYTES) | |
3265 | @end smallexample | |
3266 | ||
3267 | @noindent | |
3268 | will suffice. But on other machines, instructions that appear | |
3269 | to push one byte actually push two bytes in an attempt to maintain | |
3270 | alignment. Then the definition should be | |
3271 | ||
3272 | @smallexample | |
3273 | #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1) | |
3274 | @end smallexample | |
c6e790c5 | 3275 | |
2f05705b | 3276 | If the value of this macro has a type, it should be an unsigned type. |
202d6e5f | 3277 | @end defmac |
3278 | ||
81e483dd | 3279 | @findex outgoing_args_size |
3280 | @findex crtl->outgoing_args_size | |
202d6e5f | 3281 | @defmac ACCUMULATE_OUTGOING_ARGS |
3282 | A C expression. If nonzero, the maximum amount of space required for outgoing arguments | |
81e483dd | 3283 | will be computed and placed into |
3284 | @code{crtl->outgoing_args_size}. No space will be pushed | |
202d6e5f | 3285 | onto the stack for each call; instead, the function prologue should |
3286 | increase the stack frame size by this amount. | |
3287 | ||
3288 | Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS} | |
3289 | is not proper. | |
3290 | @end defmac | |
3291 | ||
3292 | @defmac REG_PARM_STACK_SPACE (@var{fndecl}) | |
3293 | Define this macro if functions should assume that stack space has been | |
3294 | allocated for arguments even when their values are passed in | |
3295 | registers. | |
3296 | ||
3297 | The value of this macro is the size, in bytes, of the area reserved for | |
3298 | arguments passed in registers for the function represented by @var{fndecl}, | |
3299 | which can be zero if GCC is calling a library function. | |
3300 | The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself | |
3301 | of the function. | |
3302 | ||
3303 | This space can be allocated by the caller, or be a part of the | |
3304 | machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says | |
3305 | which. | |
3306 | @end defmac | |
3307 | @c above is overfull. not sure what to do. --mew 5feb93 did | |
3308 | @c something, not sure if it looks good. --mew 10feb93 | |
3309 | ||
02114c95 | 3310 | @defmac INCOMING_REG_PARM_STACK_SPACE (@var{fndecl}) |
3311 | Like @code{REG_PARM_STACK_SPACE}, but for incoming register arguments. | |
3312 | Define this macro if space guaranteed when compiling a function body | |
3313 | is different to space required when making a call, a situation that | |
3314 | can arise with K&R style function definitions. | |
3315 | @end defmac | |
3316 | ||
202d6e5f | 3317 | @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype}) |
3318 | Define this to a nonzero value if it is the responsibility of the | |
3319 | caller to allocate the area reserved for arguments passed in registers | |
3320 | when calling a function of @var{fntype}. @var{fntype} may be NULL | |
3321 | if the function called is a library function. | |
3322 | ||
3323 | If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls | |
3324 | whether the space for these arguments counts in the value of | |
81e483dd | 3325 | @code{crtl->outgoing_args_size}. |
202d6e5f | 3326 | @end defmac |
3327 | ||
3328 | @defmac STACK_PARMS_IN_REG_PARM_AREA | |
3329 | Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the | |
3330 | stack parameters don't skip the area specified by it. | |
3331 | @c i changed this, makes more sens and it should have taken care of the | |
3332 | @c overfull.. not as specific, tho. --mew 5feb93 | |
3333 | ||
3334 | Normally, when a parameter is not passed in registers, it is placed on the | |
3335 | stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro | |
3336 | suppresses this behavior and causes the parameter to be passed on the | |
3337 | stack in its natural location. | |
3338 | @end defmac | |
3339 | ||
fe0b4016 | 3340 | @hook TARGET_RETURN_POPS_ARGS |
202d6e5f | 3341 | |
3342 | @defmac CALL_POPS_ARGS (@var{cum}) | |
3343 | A C expression that should indicate the number of bytes a call sequence | |
3344 | pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS} | |
3345 | when compiling a function call. | |
3346 | ||
3347 | @var{cum} is the variable in which all arguments to the called function | |
3348 | have been accumulated. | |
3349 | ||
3350 | On certain architectures, such as the SH5, a call trampoline is used | |
3351 | that pops certain registers off the stack, depending on the arguments | |
3352 | that have been passed to the function. Since this is a property of the | |
3353 | call site, not of the called function, @code{RETURN_POPS_ARGS} is not | |
3354 | appropriate. | |
3355 | @end defmac | |
3356 | ||
3357 | @node Register Arguments | |
3358 | @subsection Passing Arguments in Registers | |
3359 | @cindex arguments in registers | |
3360 | @cindex registers arguments | |
3361 | ||
3362 | This section describes the macros which let you control how various | |
3363 | types of arguments are passed in registers or how they are arranged in | |
3364 | the stack. | |
3365 | ||
81bcd36c | 3366 | @hook TARGET_FUNCTION_ARG |
202d6e5f | 3367 | |
3368 | @hook TARGET_MUST_PASS_IN_STACK | |
202d6e5f | 3369 | |
81bcd36c | 3370 | @hook TARGET_FUNCTION_INCOMING_ARG |
202d6e5f | 3371 | |
a9d8ab38 | 3372 | @hook TARGET_USE_PSEUDO_PIC_REG |
3373 | ||
3374 | @hook TARGET_INIT_PIC_REG | |
3375 | ||
202d6e5f | 3376 | @hook TARGET_ARG_PARTIAL_BYTES |
202d6e5f | 3377 | |
b4c24d67 | 3378 | @hook TARGET_PASS_BY_REFERENCE |
202d6e5f | 3379 | |
3380 | @hook TARGET_CALLEE_COPIES | |
202d6e5f | 3381 | |
3382 | @defmac CUMULATIVE_ARGS | |
81bcd36c | 3383 | A C type for declaring a variable that is used as the first argument |
3384 | of @code{TARGET_FUNCTION_ARG} and other related values. For some | |
3385 | target machines, the type @code{int} suffices and can hold the number | |
3386 | of bytes of argument so far. | |
202d6e5f | 3387 | |
3388 | There is no need to record in @code{CUMULATIVE_ARGS} anything about the | |
3389 | arguments that have been passed on the stack. The compiler has other | |
3390 | variables to keep track of that. For target machines on which all | |
3391 | arguments are passed on the stack, there is no need to store anything in | |
3392 | @code{CUMULATIVE_ARGS}; however, the data structure must exist and | |
3393 | should not be empty, so use @code{int}. | |
3394 | @end defmac | |
3395 | ||
3396 | @defmac OVERRIDE_ABI_FORMAT (@var{fndecl}) | |
3397 | If defined, this macro is called before generating any code for a | |
3398 | function, but after the @var{cfun} descriptor for the function has been | |
3399 | created. The back end may use this macro to update @var{cfun} to | |
3400 | reflect an ABI other than that which would normally be used by default. | |
3401 | If the compiler is generating code for a compiler-generated function, | |
3402 | @var{fndecl} may be @code{NULL}. | |
3403 | @end defmac | |
3404 | ||
3405 | @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args}) | |
3406 | A C statement (sans semicolon) for initializing the variable | |
3407 | @var{cum} for the state at the beginning of the argument list. The | |
3408 | variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype} | |
3409 | is the tree node for the data type of the function which will receive | |
3410 | the args, or 0 if the args are to a compiler support library function. | |
3411 | For direct calls that are not libcalls, @var{fndecl} contain the | |
3412 | declaration node of the function. @var{fndecl} is also set when | |
3413 | @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function | |
3414 | being compiled. @var{n_named_args} is set to the number of named | |
3415 | arguments, including a structure return address if it is passed as a | |
3416 | parameter, when making a call. When processing incoming arguments, | |
3417 | @var{n_named_args} is set to @minus{}1. | |
3418 | ||
3419 | When processing a call to a compiler support library function, | |
3420 | @var{libname} identifies which one. It is a @code{symbol_ref} rtx which | |
3421 | contains the name of the function, as a string. @var{libname} is 0 when | |
3422 | an ordinary C function call is being processed. Thus, each time this | |
3423 | macro is called, either @var{libname} or @var{fntype} is nonzero, but | |
3424 | never both of them at once. | |
3425 | @end defmac | |
3426 | ||
3427 | @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname}) | |
3428 | Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls, | |
3429 | it gets a @code{MODE} argument instead of @var{fntype}, that would be | |
3430 | @code{NULL}. @var{indirect} would always be zero, too. If this macro | |
3431 | is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname, | |
3432 | 0)} is used instead. | |
3433 | @end defmac | |
3434 | ||
3435 | @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname}) | |
3436 | Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of | |
3437 | finding the arguments for the function being compiled. If this macro is | |
3438 | undefined, @code{INIT_CUMULATIVE_ARGS} is used instead. | |
3439 | ||
3440 | The value passed for @var{libname} is always 0, since library routines | |
3441 | with special calling conventions are never compiled with GCC@. The | |
3442 | argument @var{libname} exists for symmetry with | |
3443 | @code{INIT_CUMULATIVE_ARGS}. | |
3444 | @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe. | |
3445 | @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93 | |
3446 | @end defmac | |
3447 | ||
81bcd36c | 3448 | @hook TARGET_FUNCTION_ARG_ADVANCE |
202d6e5f | 3449 | |
3450 | @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type}) | |
3451 | If defined, a C expression that is the number of bytes to add to the | |
3452 | offset of the argument passed in memory. This is needed for the SPU, | |
3453 | which passes @code{char} and @code{short} arguments in the preferred | |
3454 | slot that is in the middle of the quad word instead of starting at the | |
3455 | top. | |
3456 | @end defmac | |
3457 | ||
3458 | @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type}) | |
3459 | If defined, a C expression which determines whether, and in which direction, | |
3460 | to pad out an argument with extra space. The value should be of type | |
3461 | @code{enum direction}: either @code{upward} to pad above the argument, | |
3462 | @code{downward} to pad below, or @code{none} to inhibit padding. | |
3463 | ||
17bfc2bc | 3464 | The @emph{amount} of padding is not controlled by this macro, but by the |
3465 | target hook @code{TARGET_FUNCTION_ARG_ROUND_BOUNDARY}. It is | |
3466 | always just enough to reach the next multiple of that boundary. | |
202d6e5f | 3467 | |
3468 | This macro has a default definition which is right for most systems. | |
3469 | For little-endian machines, the default is to pad upward. For | |
3470 | big-endian machines, the default is to pad downward for an argument of | |
3471 | constant size shorter than an @code{int}, and upward otherwise. | |
3472 | @end defmac | |
3473 | ||
3474 | @defmac PAD_VARARGS_DOWN | |
3475 | If defined, a C expression which determines whether the default | |
3476 | implementation of va_arg will attempt to pad down before reading the | |
3477 | next argument, if that argument is smaller than its aligned space as | |
3478 | controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such | |
3479 | arguments are padded down if @code{BYTES_BIG_ENDIAN} is true. | |
3480 | @end defmac | |
3481 | ||
3482 | @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first}) | |
3483 | Specify padding for the last element of a block move between registers and | |
3484 | memory. @var{first} is nonzero if this is the only element. Defining this | |
3485 | macro allows better control of register function parameters on big-endian | |
3486 | machines, without using @code{PARALLEL} rtl. In particular, | |
3487 | @code{MUST_PASS_IN_STACK} need not test padding and mode of types in | |
3488 | registers, as there is no longer a "wrong" part of a register; For example, | |
3489 | a three byte aggregate may be passed in the high part of a register if so | |
3490 | required. | |
3491 | @end defmac | |
3492 | ||
bd99ba64 | 3493 | @hook TARGET_FUNCTION_ARG_BOUNDARY |
202d6e5f | 3494 | |
17bfc2bc | 3495 | @hook TARGET_FUNCTION_ARG_ROUND_BOUNDARY |
3496 | ||
202d6e5f | 3497 | @defmac FUNCTION_ARG_REGNO_P (@var{regno}) |
3498 | A C expression that is nonzero if @var{regno} is the number of a hard | |
3499 | register in which function arguments are sometimes passed. This does | |
3500 | @emph{not} include implicit arguments such as the static chain and | |
3501 | the structure-value address. On many machines, no registers can be | |
3502 | used for this purpose since all function arguments are pushed on the | |
3503 | stack. | |
3504 | @end defmac | |
3505 | ||
3506 | @hook TARGET_SPLIT_COMPLEX_ARG | |
202d6e5f | 3507 | |
3508 | @hook TARGET_BUILD_BUILTIN_VA_LIST | |
202d6e5f | 3509 | |
303ece6c | 3510 | @hook TARGET_ENUM_VA_LIST_P |
202d6e5f | 3511 | |
3512 | @hook TARGET_FN_ABI_VA_LIST | |
202d6e5f | 3513 | |
3514 | @hook TARGET_CANONICAL_VA_LIST_TYPE | |
202d6e5f | 3515 | |
3516 | @hook TARGET_GIMPLIFY_VA_ARG_EXPR | |
202d6e5f | 3517 | |
3518 | @hook TARGET_VALID_POINTER_MODE | |
202d6e5f | 3519 | |
be97d4b6 | 3520 | @hook TARGET_REF_MAY_ALIAS_ERRNO |
3521 | ||
202d6e5f | 3522 | @hook TARGET_SCALAR_MODE_SUPPORTED_P |
202d6e5f | 3523 | |
3524 | @hook TARGET_VECTOR_MODE_SUPPORTED_P | |
202d6e5f | 3525 | |
13d3ceb9 | 3526 | @hook TARGET_ARRAY_MODE_SUPPORTED_P |
3527 | ||
d5957f0d | 3528 | @hook TARGET_LIBGCC_FLOATING_MODE_SUPPORTED_P |
3529 | ||
202d6e5f | 3530 | @hook TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P |
202d6e5f | 3531 | |
3532 | @node Scalar Return | |
3533 | @subsection How Scalar Function Values Are Returned | |
3534 | @cindex return values in registers | |
3535 | @cindex values, returned by functions | |
3536 | @cindex scalars, returned as values | |
3537 | ||
3538 | This section discusses the macros that control returning scalars as | |
3539 | values---values that can fit in registers. | |
3540 | ||
3541 | @hook TARGET_FUNCTION_VALUE | |
3542 | ||
202d6e5f | 3543 | @defmac FUNCTION_VALUE (@var{valtype}, @var{func}) |
3544 | This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for | |
3545 | a new target instead. | |
3546 | @end defmac | |
3547 | ||
3548 | @defmac LIBCALL_VALUE (@var{mode}) | |
3549 | A C expression to create an RTX representing the place where a library | |
3550 | function returns a value of mode @var{mode}. | |
3551 | ||
3552 | Note that ``library function'' in this context means a compiler | |
3553 | support routine, used to perform arithmetic, whose name is known | |
3554 | specially by the compiler and was not mentioned in the C code being | |
3555 | compiled. | |
3556 | @end defmac | |
3557 | ||
3558 | @hook TARGET_LIBCALL_VALUE | |
202d6e5f | 3559 | |
3560 | @defmac FUNCTION_VALUE_REGNO_P (@var{regno}) | |
3561 | A C expression that is nonzero if @var{regno} is the number of a hard | |
3562 | register in which the values of called function may come back. | |
3563 | ||
3564 | A register whose use for returning values is limited to serving as the | |
3565 | second of a pair (for a value of type @code{double}, say) need not be | |
3566 | recognized by this macro. So for most machines, this definition | |
3567 | suffices: | |
3568 | ||
3569 | @smallexample | |
3570 | #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0) | |
3571 | @end smallexample | |
3572 | ||
3573 | If the machine has register windows, so that the caller and the called | |
3574 | function use different registers for the return value, this macro | |
3575 | should recognize only the caller's register numbers. | |
3576 | ||
3577 | This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P} | |
3578 | for a new target instead. | |
3579 | @end defmac | |
3580 | ||
3581 | @hook TARGET_FUNCTION_VALUE_REGNO_P | |
202d6e5f | 3582 | |
3583 | @defmac APPLY_RESULT_SIZE | |
3584 | Define this macro if @samp{untyped_call} and @samp{untyped_return} | |
3585 | need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for | |
3586 | saving and restoring an arbitrary return value. | |
3587 | @end defmac | |
3588 | ||
809140f3 | 3589 | @hook TARGET_OMIT_STRUCT_RETURN_REG |
3590 | ||
202d6e5f | 3591 | @hook TARGET_RETURN_IN_MSB |
202d6e5f | 3592 | |
3593 | @node Aggregate Return | |
3594 | @subsection How Large Values Are Returned | |
3595 | @cindex aggregates as return values | |
3596 | @cindex large return values | |
3597 | @cindex returning aggregate values | |
3598 | @cindex structure value address | |
3599 | ||
3600 | When a function value's mode is @code{BLKmode} (and in some other | |
3601 | cases), the value is not returned according to | |
3602 | @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the | |
3603 | caller passes the address of a block of memory in which the value | |
3604 | should be stored. This address is called the @dfn{structure value | |
3605 | address}. | |
3606 | ||
3607 | This section describes how to control returning structure values in | |
3608 | memory. | |
3609 | ||
3610 | @hook TARGET_RETURN_IN_MEMORY | |
202d6e5f | 3611 | |
3612 | @defmac DEFAULT_PCC_STRUCT_RETURN | |
3613 | Define this macro to be 1 if all structure and union return values must be | |
3614 | in memory. Since this results in slower code, this should be defined | |
3615 | only if needed for compatibility with other compilers or with an ABI@. | |
3616 | If you define this macro to be 0, then the conventions used for structure | |
3617 | and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY} | |
3618 | target hook. | |
3619 | ||
3620 | If not defined, this defaults to the value 1. | |
3621 | @end defmac | |
3622 | ||
3623 | @hook TARGET_STRUCT_VALUE_RTX | |
202d6e5f | 3624 | |
3625 | @defmac PCC_STATIC_STRUCT_RETURN | |
3626 | Define this macro if the usual system convention on the target machine | |
3627 | for returning structures and unions is for the called function to return | |
3628 | the address of a static variable containing the value. | |
3629 | ||
3630 | Do not define this if the usual system convention is for the caller to | |
3631 | pass an address to the subroutine. | |
3632 | ||
3633 | This macro has effect in @option{-fpcc-struct-return} mode, but it does | |
3634 | nothing when you use @option{-freg-struct-return} mode. | |
3635 | @end defmac | |
3636 | ||
4bac51c9 | 3637 | @hook TARGET_GET_RAW_RESULT_MODE |
3638 | ||
3639 | @hook TARGET_GET_RAW_ARG_MODE | |
3640 | ||
202d6e5f | 3641 | @node Caller Saves |
3642 | @subsection Caller-Saves Register Allocation | |
3643 | ||
3644 | If you enable it, GCC can save registers around function calls. This | |
3645 | makes it possible to use call-clobbered registers to hold variables that | |
3646 | must live across calls. | |
3647 | ||
202d6e5f | 3648 | @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs}) |
3649 | A C expression specifying which mode is required for saving @var{nregs} | |
3650 | of a pseudo-register in call-clobbered hard register @var{regno}. If | |
3651 | @var{regno} is unsuitable for caller save, @code{VOIDmode} should be | |
3652 | returned. For most machines this macro need not be defined since GCC | |
3653 | will select the smallest suitable mode. | |
3654 | @end defmac | |
3655 | ||
3656 | @node Function Entry | |
3657 | @subsection Function Entry and Exit | |
3658 | @cindex function entry and exit | |
3659 | @cindex prologue | |
3660 | @cindex epilogue | |
3661 | ||
3662 | This section describes the macros that output function entry | |
3663 | (@dfn{prologue}) and exit (@dfn{epilogue}) code. | |
3664 | ||
3665 | @hook TARGET_ASM_FUNCTION_PROLOGUE | |
202d6e5f | 3666 | |
3667 | @hook TARGET_ASM_FUNCTION_END_PROLOGUE | |
202d6e5f | 3668 | |
3669 | @hook TARGET_ASM_FUNCTION_BEGIN_EPILOGUE | |
202d6e5f | 3670 | |
3671 | @hook TARGET_ASM_FUNCTION_EPILOGUE | |
202d6e5f | 3672 | |
3673 | @itemize @bullet | |
3674 | @item | |
81e483dd | 3675 | @findex pretend_args_size |
3676 | @findex crtl->args.pretend_args_size | |
3677 | A region of @code{crtl->args.pretend_args_size} bytes of | |
202d6e5f | 3678 | uninitialized space just underneath the first argument arriving on the |
3679 | stack. (This may not be at the very start of the allocated stack region | |
3680 | if the calling sequence has pushed anything else since pushing the stack | |
3681 | arguments. But usually, on such machines, nothing else has been pushed | |
3682 | yet, because the function prologue itself does all the pushing.) This | |
3683 | region is used on machines where an argument may be passed partly in | |
3684 | registers and partly in memory, and, in some cases to support the | |
3685 | features in @code{<stdarg.h>}. | |
3686 | ||
3687 | @item | |
3688 | An area of memory used to save certain registers used by the function. | |
3689 | The size of this area, which may also include space for such things as | |
3690 | the return address and pointers to previous stack frames, is | |
3691 | machine-specific and usually depends on which registers have been used | |
3692 | in the function. Machines with register windows often do not require | |
3693 | a save area. | |
3694 | ||
3695 | @item | |
3696 | A region of at least @var{size} bytes, possibly rounded up to an allocation | |
3697 | boundary, to contain the local variables of the function. On some machines, | |
3698 | this region and the save area may occur in the opposite order, with the | |
3699 | save area closer to the top of the stack. | |
3700 | ||
3701 | @item | |
3702 | @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames | |
3703 | Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of | |
81e483dd | 3704 | @code{crtl->outgoing_args_size} bytes to be used for outgoing |
202d6e5f | 3705 | argument lists of the function. @xref{Stack Arguments}. |
3706 | @end itemize | |
3707 | ||
3708 | @defmac EXIT_IGNORE_STACK | |
3709 | Define this macro as a C expression that is nonzero if the return | |
3710 | instruction or the function epilogue ignores the value of the stack | |
3711 | pointer; in other words, if it is safe to delete an instruction to | |
3712 | adjust the stack pointer before a return from the function. The | |
3713 | default is 0. | |
3714 | ||
3715 | Note that this macro's value is relevant only for functions for which | |
3716 | frame pointers are maintained. It is never safe to delete a final | |
3717 | stack adjustment in a function that has no frame pointer, and the | |
3718 | compiler knows this regardless of @code{EXIT_IGNORE_STACK}. | |
3719 | @end defmac | |
3720 | ||
3721 | @defmac EPILOGUE_USES (@var{regno}) | |
3722 | Define this macro as a C expression that is nonzero for registers that are | |
3723 | used by the epilogue or the @samp{return} pattern. The stack and frame | |
3724 | pointer registers are already assumed to be used as needed. | |
3725 | @end defmac | |
3726 | ||
3727 | @defmac EH_USES (@var{regno}) | |
3728 | Define this macro as a C expression that is nonzero for registers that are | |
3729 | used by the exception handling mechanism, and so should be considered live | |
3730 | on entry to an exception edge. | |
3731 | @end defmac | |
3732 | ||
202d6e5f | 3733 | @hook TARGET_ASM_OUTPUT_MI_THUNK |
202d6e5f | 3734 | |
3735 | @hook TARGET_ASM_CAN_OUTPUT_MI_THUNK | |
202d6e5f | 3736 | |
3737 | @node Profiling | |
3738 | @subsection Generating Code for Profiling | |
3739 | @cindex profiling, code generation | |
3740 | ||
3741 | These macros will help you generate code for profiling. | |
3742 | ||
3743 | @defmac FUNCTION_PROFILER (@var{file}, @var{labelno}) | |
3744 | A C statement or compound statement to output to @var{file} some | |
3745 | assembler code to call the profiling subroutine @code{mcount}. | |
3746 | ||
3747 | @findex mcount | |
3748 | The details of how @code{mcount} expects to be called are determined by | |
3749 | your operating system environment, not by GCC@. To figure them out, | |
3750 | compile a small program for profiling using the system's installed C | |
3751 | compiler and look at the assembler code that results. | |
3752 | ||
3753 | Older implementations of @code{mcount} expect the address of a counter | |
3754 | variable to be loaded into some register. The name of this variable is | |
3755 | @samp{LP} followed by the number @var{labelno}, so you would generate | |
3756 | the name using @samp{LP%d} in a @code{fprintf}. | |
3757 | @end defmac | |
3758 | ||
3759 | @defmac PROFILE_HOOK | |
3760 | A C statement or compound statement to output to @var{file} some assembly | |
3761 | code to call the profiling subroutine @code{mcount} even the target does | |
3762 | not support profiling. | |
3763 | @end defmac | |
3764 | ||
3765 | @defmac NO_PROFILE_COUNTERS | |
3766 | Define this macro to be an expression with a nonzero value if the | |
3767 | @code{mcount} subroutine on your system does not need a counter variable | |
3768 | allocated for each function. This is true for almost all modern | |
3769 | implementations. If you define this macro, you must not use the | |
3770 | @var{labelno} argument to @code{FUNCTION_PROFILER}. | |
3771 | @end defmac | |
3772 | ||
3773 | @defmac PROFILE_BEFORE_PROLOGUE | |
3774 | Define this macro if the code for function profiling should come before | |
3775 | the function prologue. Normally, the profiling code comes after. | |
3776 | @end defmac | |
3777 | ||
d44f2f7c | 3778 | @hook TARGET_KEEP_LEAF_WHEN_PROFILED |
3779 | ||
202d6e5f | 3780 | @node Tail Calls |
3781 | @subsection Permitting tail calls | |
3782 | @cindex tail calls | |
3783 | ||
3784 | @hook TARGET_FUNCTION_OK_FOR_SIBCALL | |
202d6e5f | 3785 | |
3786 | @hook TARGET_EXTRA_LIVE_ON_ENTRY | |
202d6e5f | 3787 | |
2e851bb8 | 3788 | @hook TARGET_SET_UP_BY_PROLOGUE |
3789 | ||
08c6cbd2 | 3790 | @hook TARGET_WARN_FUNC_RETURN |
3791 | ||
202d6e5f | 3792 | @node Stack Smashing Protection |
3793 | @subsection Stack smashing protection | |
3794 | @cindex stack smashing protection | |
3795 | ||
3796 | @hook TARGET_STACK_PROTECT_GUARD | |
202d6e5f | 3797 | |
3798 | @hook TARGET_STACK_PROTECT_FAIL | |
202d6e5f | 3799 | |
48b14f50 | 3800 | @hook TARGET_SUPPORTS_SPLIT_STACK |
3801 | ||
e9eaaa6a | 3802 | @node Miscellaneous Register Hooks |
3803 | @subsection Miscellaneous register hooks | |
3804 | @cindex miscellaneous register hooks | |
3805 | ||
3806 | @hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS | |
3807 | ||
202d6e5f | 3808 | @node Varargs |
3809 | @section Implementing the Varargs Macros | |
3810 | @cindex varargs implementation | |
3811 | ||
3812 | GCC comes with an implementation of @code{<varargs.h>} and | |
3813 | @code{<stdarg.h>} that work without change on machines that pass arguments | |
3814 | on the stack. Other machines require their own implementations of | |
3815 | varargs, and the two machine independent header files must have | |
3816 | conditionals to include it. | |
3817 | ||
3818 | ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in | |
3819 | the calling convention for @code{va_start}. The traditional | |
3820 | implementation takes just one argument, which is the variable in which | |
3821 | to store the argument pointer. The ISO implementation of | |
3822 | @code{va_start} takes an additional second argument. The user is | |
3823 | supposed to write the last named argument of the function here. | |
3824 | ||
3825 | However, @code{va_start} should not use this argument. The way to find | |
3826 | the end of the named arguments is with the built-in functions described | |
3827 | below. | |
3828 | ||
3829 | @defmac __builtin_saveregs () | |
3830 | Use this built-in function to save the argument registers in memory so | |
3831 | that the varargs mechanism can access them. Both ISO and traditional | |
3832 | versions of @code{va_start} must use @code{__builtin_saveregs}, unless | |
3833 | you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead. | |
3834 | ||
3835 | On some machines, @code{__builtin_saveregs} is open-coded under the | |
3836 | control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On | |
3837 | other machines, it calls a routine written in assembler language, | |
3838 | found in @file{libgcc2.c}. | |
3839 | ||
3840 | Code generated for the call to @code{__builtin_saveregs} appears at the | |
3841 | beginning of the function, as opposed to where the call to | |
3842 | @code{__builtin_saveregs} is written, regardless of what the code is. | |
3843 | This is because the registers must be saved before the function starts | |
3844 | to use them for its own purposes. | |
3845 | @c i rewrote the first sentence above to fix an overfull hbox. --mew | |
3846 | @c 10feb93 | |
3847 | @end defmac | |
3848 | ||
202d6e5f | 3849 | @defmac __builtin_next_arg (@var{lastarg}) |
f739ab33 | 3850 | This builtin returns the address of the first anonymous stack |
202d6e5f | 3851 | argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it |
3852 | returns the address of the location above the first anonymous stack | |
3853 | argument. Use it in @code{va_start} to initialize the pointer for | |
3854 | fetching arguments from the stack. Also use it in @code{va_start} to | |
3855 | verify that the second parameter @var{lastarg} is the last named argument | |
3856 | of the current function. | |
3857 | @end defmac | |
3858 | ||
3859 | @defmac __builtin_classify_type (@var{object}) | |
3860 | Since each machine has its own conventions for which data types are | |
3861 | passed in which kind of register, your implementation of @code{va_arg} | |
3862 | has to embody these conventions. The easiest way to categorize the | |
3863 | specified data type is to use @code{__builtin_classify_type} together | |
3864 | with @code{sizeof} and @code{__alignof__}. | |
3865 | ||
3866 | @code{__builtin_classify_type} ignores the value of @var{object}, | |
3867 | considering only its data type. It returns an integer describing what | |
3868 | kind of type that is---integer, floating, pointer, structure, and so on. | |
3869 | ||
3870 | The file @file{typeclass.h} defines an enumeration that you can use to | |
3871 | interpret the values of @code{__builtin_classify_type}. | |
3872 | @end defmac | |
3873 | ||
3874 | These machine description macros help implement varargs: | |
3875 | ||
3876 | @hook TARGET_EXPAND_BUILTIN_SAVEREGS | |
202d6e5f | 3877 | |
3878 | @hook TARGET_SETUP_INCOMING_VARARGS | |
202d6e5f | 3879 | |
3880 | @hook TARGET_STRICT_ARGUMENT_NAMING | |
202d6e5f | 3881 | |
53597a55 | 3882 | @hook TARGET_CALL_ARGS |
3883 | ||
3884 | @hook TARGET_END_CALL_ARGS | |
3885 | ||
202d6e5f | 3886 | @hook TARGET_PRETEND_OUTGOING_VARARGS_NAMED |
202d6e5f | 3887 | |
058a1b7a | 3888 | @hook TARGET_LOAD_BOUNDS_FOR_ARG |
3889 | ||
3890 | @hook TARGET_STORE_BOUNDS_FOR_ARG | |
3891 | ||
3892 | @hook TARGET_LOAD_RETURNED_BOUNDS | |
3893 | ||
3894 | @hook TARGET_STORE_RETURNED_BOUNDS | |
3895 | ||
3896 | @hook TARGET_CHKP_FUNCTION_VALUE_BOUNDS | |
3897 | ||
3898 | @hook TARGET_SETUP_INCOMING_VARARG_BOUNDS | |
3899 | ||
202d6e5f | 3900 | @node Trampolines |
3901 | @section Trampolines for Nested Functions | |
3902 | @cindex trampolines for nested functions | |
3903 | @cindex nested functions, trampolines for | |
3904 | ||
3905 | A @dfn{trampoline} is a small piece of code that is created at run time | |
3906 | when the address of a nested function is taken. It normally resides on | |
3907 | the stack, in the stack frame of the containing function. These macros | |
3908 | tell GCC how to generate code to allocate and initialize a | |
3909 | trampoline. | |
3910 | ||
3911 | The instructions in the trampoline must do two things: load a constant | |
3912 | address into the static chain register, and jump to the real address of | |
3913 | the nested function. On CISC machines such as the m68k, this requires | |
3914 | two instructions, a move immediate and a jump. Then the two addresses | |
3915 | exist in the trampoline as word-long immediate operands. On RISC | |
3916 | machines, it is often necessary to load each address into a register in | |
3917 | two parts. Then pieces of each address form separate immediate | |
3918 | operands. | |
3919 | ||
3920 | The code generated to initialize the trampoline must store the variable | |
3921 | parts---the static chain value and the function address---into the | |
3922 | immediate operands of the instructions. On a CISC machine, this is | |
3923 | simply a matter of copying each address to a memory reference at the | |
3924 | proper offset from the start of the trampoline. On a RISC machine, it | |
3925 | may be necessary to take out pieces of the address and store them | |
3926 | separately. | |
3927 | ||
3928 | @hook TARGET_ASM_TRAMPOLINE_TEMPLATE | |
202d6e5f | 3929 | |
3930 | @defmac TRAMPOLINE_SECTION | |
3931 | Return the section into which the trampoline template is to be placed | |
3932 | (@pxref{Sections}). The default value is @code{readonly_data_section}. | |
3933 | @end defmac | |
3934 | ||
3935 | @defmac TRAMPOLINE_SIZE | |
3936 | A C expression for the size in bytes of the trampoline, as an integer. | |
3937 | @end defmac | |
3938 | ||
3939 | @defmac TRAMPOLINE_ALIGNMENT | |
3940 | Alignment required for trampolines, in bits. | |
3941 | ||
3942 | If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT} | |
3943 | is used for aligning trampolines. | |
3944 | @end defmac | |
3945 | ||
3946 | @hook TARGET_TRAMPOLINE_INIT | |
202d6e5f | 3947 | |
3948 | @hook TARGET_TRAMPOLINE_ADJUST_ADDRESS | |
202d6e5f | 3949 | |
3950 | Implementing trampolines is difficult on many machines because they have | |
3951 | separate instruction and data caches. Writing into a stack location | |
3952 | fails to clear the memory in the instruction cache, so when the program | |
3953 | jumps to that location, it executes the old contents. | |
3954 | ||
3955 | Here are two possible solutions. One is to clear the relevant parts of | |
3956 | the instruction cache whenever a trampoline is set up. The other is to | |
3957 | make all trampolines identical, by having them jump to a standard | |
3958 | subroutine. The former technique makes trampoline execution faster; the | |
3959 | latter makes initialization faster. | |
3960 | ||
3961 | To clear the instruction cache when a trampoline is initialized, define | |
3962 | the following macro. | |
3963 | ||
3964 | @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end}) | |
3965 | If defined, expands to a C expression clearing the @emph{instruction | |
3966 | cache} in the specified interval. The definition of this macro would | |
3967 | typically be a series of @code{asm} statements. Both @var{beg} and | |
3968 | @var{end} are both pointer expressions. | |
3969 | @end defmac | |
3970 | ||
202d6e5f | 3971 | To use a standard subroutine, define the following macro. In addition, |
3972 | you must make sure that the instructions in a trampoline fill an entire | |
3973 | cache line with identical instructions, or else ensure that the | |
3974 | beginning of the trampoline code is always aligned at the same point in | |
3975 | its cache line. Look in @file{m68k.h} as a guide. | |
3976 | ||
3977 | @defmac TRANSFER_FROM_TRAMPOLINE | |
3978 | Define this macro if trampolines need a special subroutine to do their | |
3979 | work. The macro should expand to a series of @code{asm} statements | |
3980 | which will be compiled with GCC@. They go in a library function named | |
3981 | @code{__transfer_from_trampoline}. | |
3982 | ||
3983 | If you need to avoid executing the ordinary prologue code of a compiled | |
3984 | C function when you jump to the subroutine, you can do so by placing a | |
3985 | special label of your own in the assembler code. Use one @code{asm} | |
3986 | statement to generate an assembler label, and another to make the label | |
3987 | global. Then trampolines can use that label to jump directly to your | |
3988 | special assembler code. | |
3989 | @end defmac | |
3990 | ||
3991 | @node Library Calls | |
3992 | @section Implicit Calls to Library Routines | |
3993 | @cindex library subroutine names | |
3994 | @cindex @file{libgcc.a} | |
3995 | ||
3996 | @c prevent bad page break with this line | |
3997 | Here is an explanation of implicit calls to library routines. | |
3998 | ||
3999 | @defmac DECLARE_LIBRARY_RENAMES | |
4000 | This macro, if defined, should expand to a piece of C code that will get | |
4001 | expanded when compiling functions for libgcc.a. It can be used to | |
4002 | provide alternate names for GCC's internal library functions if there | |
4003 | are ABI-mandated names that the compiler should provide. | |
4004 | @end defmac | |
4005 | ||
4006 | @findex set_optab_libfunc | |
4007 | @findex init_one_libfunc | |
4008 | @hook TARGET_INIT_LIBFUNCS | |
202d6e5f | 4009 | |
f308a9b2 | 4010 | @hook TARGET_LIBFUNC_GNU_PREFIX |
4011 | ||
202d6e5f | 4012 | @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison}) |
4013 | This macro should return @code{true} if the library routine that | |
4014 | implements the floating point comparison operator @var{comparison} in | |
4015 | mode @var{mode} will return a boolean, and @var{false} if it will | |
4016 | return a tristate. | |
4017 | ||
4018 | GCC's own floating point libraries return tristates from the | |
4019 | comparison operators, so the default returns false always. Most ports | |
4020 | don't need to define this macro. | |
4021 | @end defmac | |
4022 | ||
4023 | @defmac TARGET_LIB_INT_CMP_BIASED | |
4024 | This macro should evaluate to @code{true} if the integer comparison | |
4025 | functions (like @code{__cmpdi2}) return 0 to indicate that the first | |
4026 | operand is smaller than the second, 1 to indicate that they are equal, | |
4027 | and 2 to indicate that the first operand is greater than the second. | |
4028 | If this macro evaluates to @code{false} the comparison functions return | |
4029 | @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines | |
4030 | in @file{libgcc.a}, you do not need to define this macro. | |
4031 | @end defmac | |
4032 | ||
b214253d | 4033 | @defmac TARGET_HAS_NO_HW_DIVIDE |
4034 | This macro should be defined if the target has no hardware divide | |
4035 | instructions. If this macro is defined, GCC will use an algorithm which | |
4036 | make use of simple logical and arithmetic operations for 64-bit | |
4037 | division. If the macro is not defined, GCC will use an algorithm which | |
4038 | make use of a 64-bit by 32-bit divide primitive. | |
4039 | @end defmac | |
4040 | ||
202d6e5f | 4041 | @cindex @code{EDOM}, implicit usage |
4042 | @findex matherr | |
4043 | @defmac TARGET_EDOM | |
4044 | The value of @code{EDOM} on the target machine, as a C integer constant | |
4045 | expression. If you don't define this macro, GCC does not attempt to | |
4046 | deposit the value of @code{EDOM} into @code{errno} directly. Look in | |
4047 | @file{/usr/include/errno.h} to find the value of @code{EDOM} on your | |
4048 | system. | |
4049 | ||
4050 | If you do not define @code{TARGET_EDOM}, then compiled code reports | |
4051 | domain errors by calling the library function and letting it report the | |
4052 | error. If mathematical functions on your system use @code{matherr} when | |
4053 | there is an error, then you should leave @code{TARGET_EDOM} undefined so | |
4054 | that @code{matherr} is used normally. | |
4055 | @end defmac | |
4056 | ||
4057 | @cindex @code{errno}, implicit usage | |
4058 | @defmac GEN_ERRNO_RTX | |
4059 | Define this macro as a C expression to create an rtl expression that | |
4060 | refers to the global ``variable'' @code{errno}. (On certain systems, | |
4061 | @code{errno} may not actually be a variable.) If you don't define this | |
4062 | macro, a reasonable default is used. | |
4063 | @end defmac | |
4064 | ||
30f690e0 | 4065 | @hook TARGET_LIBC_HAS_FUNCTION |
202d6e5f | 4066 | |
4067 | @defmac NEXT_OBJC_RUNTIME | |
3f3d5ad4 | 4068 | Set this macro to 1 to use the "NeXT" Objective-C message sending conventions |
4069 | by default. This calling convention involves passing the object, the selector | |
4070 | and the method arguments all at once to the method-lookup library function. | |
4071 | This is the usual setting when targeting Darwin/Mac OS X systems, which have | |
4072 | the NeXT runtime installed. | |
4073 | ||
4074 | If the macro is set to 0, the "GNU" Objective-C message sending convention | |
4075 | will be used by default. This convention passes just the object and the | |
4076 | selector to the method-lookup function, which returns a pointer to the method. | |
4077 | ||
4078 | In either case, it remains possible to select code-generation for the alternate | |
4079 | scheme, by means of compiler command line switches. | |
202d6e5f | 4080 | @end defmac |
4081 | ||
4082 | @node Addressing Modes | |
4083 | @section Addressing Modes | |
4084 | @cindex addressing modes | |
4085 | ||
4086 | @c prevent bad page break with this line | |
4087 | This is about addressing modes. | |
4088 | ||
4089 | @defmac HAVE_PRE_INCREMENT | |
4090 | @defmacx HAVE_PRE_DECREMENT | |
4091 | @defmacx HAVE_POST_INCREMENT | |
4092 | @defmacx HAVE_POST_DECREMENT | |
4093 | A C expression that is nonzero if the machine supports pre-increment, | |
4094 | pre-decrement, post-increment, or post-decrement addressing respectively. | |
4095 | @end defmac | |
4096 | ||
4097 | @defmac HAVE_PRE_MODIFY_DISP | |
4098 | @defmacx HAVE_POST_MODIFY_DISP | |
4099 | A C expression that is nonzero if the machine supports pre- or | |
4100 | post-address side-effect generation involving constants other than | |
4101 | the size of the memory operand. | |
4102 | @end defmac | |
4103 | ||
4104 | @defmac HAVE_PRE_MODIFY_REG | |
4105 | @defmacx HAVE_POST_MODIFY_REG | |
4106 | A C expression that is nonzero if the machine supports pre- or | |
4107 | post-address side-effect generation involving a register displacement. | |
4108 | @end defmac | |
4109 | ||
4110 | @defmac CONSTANT_ADDRESS_P (@var{x}) | |
4111 | A C expression that is 1 if the RTX @var{x} is a constant which | |
4112 | is a valid address. On most machines the default definition of | |
4113 | @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)} | |
4114 | is acceptable, but a few machines are more restrictive as to which | |
15b474a2 | 4115 | constant addresses are supported. |
202d6e5f | 4116 | @end defmac |
4117 | ||
4118 | @defmac CONSTANT_P (@var{x}) | |
4119 | @code{CONSTANT_P}, which is defined by target-independent code, | |
4120 | accepts integer-values expressions whose values are not explicitly | |
4121 | known, such as @code{symbol_ref}, @code{label_ref}, and @code{high} | |
4122 | expressions and @code{const} arithmetic expressions, in addition to | |
4123 | @code{const_int} and @code{const_double} expressions. | |
4124 | @end defmac | |
4125 | ||
4126 | @defmac MAX_REGS_PER_ADDRESS | |
4127 | A number, the maximum number of registers that can appear in a valid | |
4128 | memory address. Note that it is up to you to specify a value equal to | |
4129 | the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever | |
4130 | accept. | |
4131 | @end defmac | |
4132 | ||
4133 | @hook TARGET_LEGITIMATE_ADDRESS_P | |
202d6e5f | 4134 | |
4135 | @defmac TARGET_MEM_CONSTRAINT | |
4136 | A single character to be used instead of the default @code{'m'} | |
4137 | character for general memory addresses. This defines the constraint | |
4138 | letter which matches the memory addresses accepted by | |
4139 | @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to | |
4140 | support new address formats in your back end without changing the | |
4141 | semantics of the @code{'m'} constraint. This is necessary in order to | |
4142 | preserve functionality of inline assembly constructs using the | |
4143 | @code{'m'} constraint. | |
4144 | @end defmac | |
4145 | ||
4146 | @defmac FIND_BASE_TERM (@var{x}) | |
4147 | A C expression to determine the base term of address @var{x}, | |
4148 | or to provide a simplified version of @var{x} from which @file{alias.c} | |
4149 | can easily find the base term. This macro is used in only two places: | |
4150 | @code{find_base_value} and @code{find_base_term} in @file{alias.c}. | |
4151 | ||
4152 | It is always safe for this macro to not be defined. It exists so | |
4153 | that alias analysis can understand machine-dependent addresses. | |
4154 | ||
4155 | The typical use of this macro is to handle addresses containing | |
4156 | a label_ref or symbol_ref within an UNSPEC@. | |
4157 | @end defmac | |
4158 | ||
4159 | @hook TARGET_LEGITIMIZE_ADDRESS | |
202d6e5f | 4160 | |
4161 | @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win}) | |
4162 | A C compound statement that attempts to replace @var{x}, which is an address | |
4163 | that needs reloading, with a valid memory address for an operand of mode | |
4164 | @var{mode}. @var{win} will be a C statement label elsewhere in the code. | |
4165 | It is not necessary to define this macro, but it might be useful for | |
4166 | performance reasons. | |
4167 | ||
4168 | For example, on the i386, it is sometimes possible to use a single | |
4169 | reload register instead of two by reloading a sum of two pseudo | |
4170 | registers into a register. On the other hand, for number of RISC | |
4171 | processors offsets are limited so that often an intermediate address | |
4172 | needs to be generated in order to address a stack slot. By defining | |
4173 | @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses | |
4174 | generated for adjacent some stack slots can be made identical, and thus | |
4175 | be shared. | |
4176 | ||
4177 | @emph{Note}: This macro should be used with caution. It is necessary | |
4178 | to know something of how reload works in order to effectively use this, | |
4179 | and it is quite easy to produce macros that build in too much knowledge | |
4180 | of reload internals. | |
4181 | ||
4182 | @emph{Note}: This macro must be able to reload an address created by a | |
4183 | previous invocation of this macro. If it fails to handle such addresses | |
4184 | then the compiler may generate incorrect code or abort. | |
4185 | ||
4186 | @findex push_reload | |
4187 | The macro definition should use @code{push_reload} to indicate parts that | |
4188 | need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually | |
4189 | suitable to be passed unaltered to @code{push_reload}. | |
4190 | ||
4191 | The code generated by this macro must not alter the substructure of | |
4192 | @var{x}. If it transforms @var{x} into a more legitimate form, it | |
4193 | should assign @var{x} (which will always be a C variable) a new value. | |
4194 | This also applies to parts that you change indirectly by calling | |
4195 | @code{push_reload}. | |
4196 | ||
4197 | @findex strict_memory_address_p | |
4198 | The macro definition may use @code{strict_memory_address_p} to test if | |
4199 | the address has become legitimate. | |
4200 | ||
4201 | @findex copy_rtx | |
4202 | If you want to change only a part of @var{x}, one standard way of doing | |
4203 | this is to use @code{copy_rtx}. Note, however, that it unshares only a | |
4204 | single level of rtl. Thus, if the part to be changed is not at the | |
4205 | top level, you'll need to replace first the top level. | |
4206 | It is not necessary for this macro to come up with a legitimate | |
4207 | address; but often a machine-dependent strategy can generate better code. | |
4208 | @end defmac | |
4209 | ||
4210 | @hook TARGET_MODE_DEPENDENT_ADDRESS_P | |
202d6e5f | 4211 | |
ca316360 | 4212 | @hook TARGET_LEGITIMATE_CONSTANT_P |
202d6e5f | 4213 | |
4214 | @hook TARGET_DELEGITIMIZE_ADDRESS | |
202d6e5f | 4215 | |
202e7097 | 4216 | @hook TARGET_CONST_NOT_OK_FOR_DEBUG_P |
202e7097 | 4217 | |
202d6e5f | 4218 | @hook TARGET_CANNOT_FORCE_CONST_MEM |
202d6e5f | 4219 | |
4220 | @hook TARGET_USE_BLOCKS_FOR_CONSTANT_P | |
202d6e5f | 4221 | |
80d4c05f | 4222 | @hook TARGET_USE_BLOCKS_FOR_DECL_P |
80d4c05f | 4223 | |
707e3139 | 4224 | @hook TARGET_BUILTIN_RECIPROCAL |
202d6e5f | 4225 | |
4226 | @hook TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD | |
202d6e5f | 4227 | |
202d6e5f | 4228 | @hook TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST |
202d6e5f | 4229 | |
4230 | @hook TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE | |
202d6e5f | 4231 | |
428d5ecf | 4232 | @hook TARGET_VECTORIZE_VEC_PERM_CONST_OK |
202d6e5f | 4233 | |
4234 | @hook TARGET_VECTORIZE_BUILTIN_CONVERSION | |
202d6e5f | 4235 | |
4236 | @hook TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION | |
202d6e5f | 4237 | |
b6c464fe | 4238 | @hook TARGET_VECTORIZE_BUILTIN_MD_VECTORIZED_FUNCTION |
4239 | ||
202d6e5f | 4240 | @hook TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT |
202d6e5f | 4241 | |
b24d851f | 4242 | @hook TARGET_VECTORIZE_PREFERRED_SIMD_MODE |
2101edf2 | 4243 | |
421b11b1 | 4244 | @hook TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES |
421b11b1 | 4245 | |
1f137e6d | 4246 | @hook TARGET_VECTORIZE_GET_MASK_MODE |
4247 | ||
4db2b577 | 4248 | @hook TARGET_VECTORIZE_INIT_COST |
4249 | ||
4250 | @hook TARGET_VECTORIZE_ADD_STMT_COST | |
4251 | ||
4252 | @hook TARGET_VECTORIZE_FINISH_COST | |
4253 | ||
4254 | @hook TARGET_VECTORIZE_DESTROY_COST_DATA | |
4255 | ||
4c0315d0 | 4256 | @hook TARGET_VECTORIZE_BUILTIN_TM_LOAD |
4257 | ||
4258 | @hook TARGET_VECTORIZE_BUILTIN_TM_STORE | |
4259 | ||
16dfb112 | 4260 | @hook TARGET_VECTORIZE_BUILTIN_GATHER |
16dfb112 | 4261 | |
0bd6d857 | 4262 | @hook TARGET_VECTORIZE_BUILTIN_SCATTER |
4263 | ||
d09768a4 | 4264 | @hook TARGET_SIMD_CLONE_COMPUTE_VECSIZE_AND_SIMDLEN |
4265 | ||
4266 | @hook TARGET_SIMD_CLONE_ADJUST | |
4267 | ||
4268 | @hook TARGET_SIMD_CLONE_USABLE | |
4269 | ||
0bb0f256 | 4270 | @hook TARGET_GOACC_VALIDATE_DIMS |
4271 | ||
e1037942 | 4272 | @hook TARGET_GOACC_DIM_LIMIT |
4273 | ||
c25f1934 | 4274 | @hook TARGET_GOACC_FORK_JOIN |
4275 | ||
641a0fa1 | 4276 | @hook TARGET_GOACC_REDUCTION |
4277 | ||
202d6e5f | 4278 | @node Anchored Addresses |
4279 | @section Anchored Addresses | |
4280 | @cindex anchored addresses | |
4281 | @cindex @option{-fsection-anchors} | |
4282 | ||
4283 | GCC usually addresses every static object as a separate entity. | |
4284 | For example, if we have: | |
4285 | ||
4286 | @smallexample | |
4287 | static int a, b, c; | |
4288 | int foo (void) @{ return a + b + c; @} | |
4289 | @end smallexample | |
4290 | ||
4291 | the code for @code{foo} will usually calculate three separate symbolic | |
4292 | addresses: those of @code{a}, @code{b} and @code{c}. On some targets, | |
4293 | it would be better to calculate just one symbolic address and access | |
4294 | the three variables relative to it. The equivalent pseudocode would | |
4295 | be something like: | |
4296 | ||
4297 | @smallexample | |
4298 | int foo (void) | |
4299 | @{ | |
4300 | register int *xr = &x; | |
4301 | return xr[&a - &x] + xr[&b - &x] + xr[&c - &x]; | |
4302 | @} | |
4303 | @end smallexample | |
4304 | ||
4305 | (which isn't valid C). We refer to shared addresses like @code{x} as | |
4306 | ``section anchors''. Their use is controlled by @option{-fsection-anchors}. | |
4307 | ||
4308 | The hooks below describe the target properties that GCC needs to know | |
4309 | in order to make effective use of section anchors. It won't use | |
4310 | section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET} | |
4311 | or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value. | |
4312 | ||
4313 | @hook TARGET_MIN_ANCHOR_OFFSET | |
202d6e5f | 4314 | |
4315 | @hook TARGET_MAX_ANCHOR_OFFSET | |
202d6e5f | 4316 | |
4317 | @hook TARGET_ASM_OUTPUT_ANCHOR | |
202d6e5f | 4318 | |
4319 | @hook TARGET_USE_ANCHORS_FOR_SYMBOL_P | |
202d6e5f | 4320 | |
4321 | @node Condition Code | |
4322 | @section Condition Code Status | |
4323 | @cindex condition code status | |
4324 | ||
4325 | The macros in this section can be split in two families, according to the | |
4326 | two ways of representing condition codes in GCC. | |
4327 | ||
4328 | The first representation is the so called @code{(cc0)} representation | |
4329 | (@pxref{Jump Patterns}), where all instructions can have an implicit | |
4330 | clobber of the condition codes. The second is the condition code | |
4331 | register representation, which provides better schedulability for | |
4332 | architectures that do have a condition code register, but on which | |
4333 | most instructions do not affect it. The latter category includes | |
4334 | most RISC machines. | |
4335 | ||
4336 | The implicit clobbering poses a strong restriction on the placement of | |
5b4ac966 | 4337 | the definition and use of the condition code. In the past the definition |
4338 | and use were always adjacent. However, recent changes to support trapping | |
4339 | arithmatic may result in the definition and user being in different blocks. | |
4340 | Thus, there may be a @code{NOTE_INSN_BASIC_BLOCK} between them. Additionally, | |
4341 | the definition may be the source of exception handling edges. | |
4342 | ||
4343 | These restrictions can prevent important | |
202d6e5f | 4344 | optimizations on some machines. For example, on the IBM RS/6000, there |
4345 | is a delay for taken branches unless the condition code register is set | |
4346 | three instructions earlier than the conditional branch. The instruction | |
4347 | scheduler cannot perform this optimization if it is not permitted to | |
4348 | separate the definition and use of the condition code register. | |
4349 | ||
4350 | For this reason, it is possible and suggested to use a register to | |
4351 | represent the condition code for new ports. If there is a specific | |
4352 | condition code register in the machine, use a hard register. If the | |
4353 | condition code or comparison result can be placed in any general register, | |
4354 | or if there are multiple condition registers, use a pseudo register. | |
4355 | Registers used to store the condition code value will usually have a mode | |
4356 | that is in class @code{MODE_CC}. | |
4357 | ||
4358 | Alternatively, you can use @code{BImode} if the comparison operator is | |
4359 | specified already in the compare instruction. In this case, you are not | |
4360 | interested in most macros in this section. | |
4361 | ||
4362 | @menu | |
4363 | * CC0 Condition Codes:: Old style representation of condition codes. | |
4364 | * MODE_CC Condition Codes:: Modern representation of condition codes. | |
202d6e5f | 4365 | @end menu |
4366 | ||
4367 | @node CC0 Condition Codes | |
4368 | @subsection Representation of condition codes using @code{(cc0)} | |
4369 | @findex cc0 | |
4370 | ||
4371 | @findex cc_status | |
4372 | The file @file{conditions.h} defines a variable @code{cc_status} to | |
4373 | describe how the condition code was computed (in case the interpretation of | |
4374 | the condition code depends on the instruction that it was set by). This | |
4375 | variable contains the RTL expressions on which the condition code is | |
4376 | currently based, and several standard flags. | |
4377 | ||
4378 | Sometimes additional machine-specific flags must be defined in the machine | |
4379 | description header file. It can also add additional machine-specific | |
4380 | information by defining @code{CC_STATUS_MDEP}. | |
4381 | ||
4382 | @defmac CC_STATUS_MDEP | |
4383 | C code for a data type which is used for declaring the @code{mdep} | |
4384 | component of @code{cc_status}. It defaults to @code{int}. | |
4385 | ||
4386 | This macro is not used on machines that do not use @code{cc0}. | |
4387 | @end defmac | |
4388 | ||
4389 | @defmac CC_STATUS_MDEP_INIT | |
4390 | A C expression to initialize the @code{mdep} field to ``empty''. | |
4391 | The default definition does nothing, since most machines don't use | |
4392 | the field anyway. If you want to use the field, you should probably | |
4393 | define this macro to initialize it. | |
4394 | ||
4395 | This macro is not used on machines that do not use @code{cc0}. | |
4396 | @end defmac | |
4397 | ||
4398 | @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn}) | |
4399 | A C compound statement to set the components of @code{cc_status} | |
4400 | appropriately for an insn @var{insn} whose body is @var{exp}. It is | |
4401 | this macro's responsibility to recognize insns that set the condition | |
4402 | code as a byproduct of other activity as well as those that explicitly | |
4403 | set @code{(cc0)}. | |
4404 | ||
4405 | This macro is not used on machines that do not use @code{cc0}. | |
4406 | ||
4407 | If there are insns that do not set the condition code but do alter | |
4408 | other machine registers, this macro must check to see whether they | |
4409 | invalidate the expressions that the condition code is recorded as | |
4410 | reflecting. For example, on the 68000, insns that store in address | |
4411 | registers do not set the condition code, which means that usually | |
4412 | @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such | |
4413 | insns. But suppose that the previous insn set the condition code | |
4414 | based on location @samp{a4@@(102)} and the current insn stores a new | |
4415 | value in @samp{a4}. Although the condition code is not changed by | |
4416 | this, it will no longer be true that it reflects the contents of | |
4417 | @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter | |
4418 | @code{cc_status} in this case to say that nothing is known about the | |
4419 | condition code value. | |
4420 | ||
4421 | The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal | |
4422 | with the results of peephole optimization: insns whose patterns are | |
4423 | @code{parallel} RTXs containing various @code{reg}, @code{mem} or | |
4424 | constants which are just the operands. The RTL structure of these | |
4425 | insns is not sufficient to indicate what the insns actually do. What | |
4426 | @code{NOTICE_UPDATE_CC} should do when it sees one is just to run | |
4427 | @code{CC_STATUS_INIT}. | |
4428 | ||
4429 | A possible definition of @code{NOTICE_UPDATE_CC} is to call a function | |
4430 | that looks at an attribute (@pxref{Insn Attributes}) named, for example, | |
4431 | @samp{cc}. This avoids having detailed information about patterns in | |
4432 | two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}. | |
4433 | @end defmac | |
4434 | ||
4435 | @node MODE_CC Condition Codes | |
4436 | @subsection Representation of condition codes using registers | |
4437 | @findex CCmode | |
4438 | @findex MODE_CC | |
4439 | ||
4440 | @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y}) | |
4441 | On many machines, the condition code may be produced by other instructions | |
4442 | than compares, for example the branch can use directly the condition | |
4443 | code set by a subtract instruction. However, on some machines | |
4444 | when the condition code is set this way some bits (such as the overflow | |
4445 | bit) are not set in the same way as a test instruction, so that a different | |
4446 | branch instruction must be used for some conditional branches. When | |
4447 | this happens, use the machine mode of the condition code register to | |
4448 | record different formats of the condition code register. Modes can | |
4449 | also be used to record which compare instruction (e.g. a signed or an | |
4450 | unsigned comparison) produced the condition codes. | |
4451 | ||
4452 | If other modes than @code{CCmode} are required, add them to | |
4453 | @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose | |
4454 | a mode given an operand of a compare. This is needed because the modes | |
4455 | have to be chosen not only during RTL generation but also, for example, | |
4456 | by instruction combination. The result of @code{SELECT_CC_MODE} should | |
4457 | be consistent with the mode used in the patterns; for example to support | |
4458 | the case of the add on the SPARC discussed above, we have the pattern | |
4459 | ||
4460 | @smallexample | |
4461 | (define_insn "" | |
4462 | [(set (reg:CC_NOOV 0) | |
4463 | (compare:CC_NOOV | |
4464 | (plus:SI (match_operand:SI 0 "register_operand" "%r") | |
4465 | (match_operand:SI 1 "arith_operand" "rI")) | |
4466 | (const_int 0)))] | |
4467 | "" | |
4468 | "@dots{}") | |
4469 | @end smallexample | |
4470 | ||
4471 | @noindent | |
4472 | together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode} | |
4473 | for comparisons whose argument is a @code{plus}: | |
4474 | ||
4475 | @smallexample | |
4476 | #define SELECT_CC_MODE(OP,X,Y) \ | |
123f1406 | 4477 | (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \ |
4478 | ? ((OP == LT || OP == LE || OP == GT || OP == GE) \ | |
4479 | ? CCFPEmode : CCFPmode) \ | |
4480 | : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \ | |
4481 | || GET_CODE (X) == NEG || GET_CODE (x) == ASHIFT) \ | |
202d6e5f | 4482 | ? CC_NOOVmode : CCmode)) |
4483 | @end smallexample | |
4484 | ||
4485 | Another reason to use modes is to retain information on which operands | |
4486 | were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in | |
4487 | this section. | |
4488 | ||
4489 | You should define this macro if and only if you define extra CC modes | |
4490 | in @file{@var{machine}-modes.def}. | |
4491 | @end defmac | |
4492 | ||
f8e3aa87 | 4493 | @hook TARGET_CANONICALIZE_COMPARISON |
202d6e5f | 4494 | |
4495 | @defmac REVERSIBLE_CC_MODE (@var{mode}) | |
4496 | A C expression whose value is one if it is always safe to reverse a | |
4497 | comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE} | |
4498 | can ever return @var{mode} for a floating-point inequality comparison, | |
4499 | then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero. | |
4500 | ||
4501 | You need not define this macro if it would always returns zero or if the | |
4502 | floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}. | |
4503 | For example, here is the definition used on the SPARC, where floating-point | |
123f1406 | 4504 | inequality comparisons are given either @code{CCFPEmode} or @code{CCFPmode}: |
202d6e5f | 4505 | |
4506 | @smallexample | |
123f1406 | 4507 | #define REVERSIBLE_CC_MODE(MODE) \ |
4508 | ((MODE) != CCFPEmode && (MODE) != CCFPmode) | |
202d6e5f | 4509 | @end smallexample |
4510 | @end defmac | |
4511 | ||
4512 | @defmac REVERSE_CONDITION (@var{code}, @var{mode}) | |
4513 | A C expression whose value is reversed condition code of the @var{code} for | |
4514 | comparison done in CC_MODE @var{mode}. The macro is used only in case | |
4515 | @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case | |
4516 | machine has some non-standard way how to reverse certain conditionals. For | |
4517 | instance in case all floating point conditions are non-trapping, compiler may | |
123f1406 | 4518 | freely convert unordered compares to ordered ones. Then definition may look |
202d6e5f | 4519 | like: |
4520 | ||
4521 | @smallexample | |
4522 | #define REVERSE_CONDITION(CODE, MODE) \ | |
4523 | ((MODE) != CCFPmode ? reverse_condition (CODE) \ | |
4524 | : reverse_condition_maybe_unordered (CODE)) | |
4525 | @end smallexample | |
4526 | @end defmac | |
4527 | ||
4528 | @hook TARGET_FIXED_CONDITION_CODE_REGS | |
202d6e5f | 4529 | |
4530 | @hook TARGET_CC_MODES_COMPATIBLE | |
202d6e5f | 4531 | |
3b64e9e1 | 4532 | @hook TARGET_FLAGS_REGNUM |
4533 | ||
202d6e5f | 4534 | @node Costs |
4535 | @section Describing Relative Costs of Operations | |
4536 | @cindex costs of instructions | |
4537 | @cindex relative costs | |
4538 | @cindex speed of instructions | |
4539 | ||
4540 | These macros let you describe the relative speed of various operations | |
4541 | on the target machine. | |
4542 | ||
4543 | @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to}) | |
4544 | A C expression for the cost of moving data of mode @var{mode} from a | |
4545 | register in class @var{from} to one in class @var{to}. The classes are | |
4546 | expressed using the enumeration values such as @code{GENERAL_REGS}. A | |
4547 | value of 2 is the default; other values are interpreted relative to | |
4548 | that. | |
4549 | ||
4550 | It is not required that the cost always equal 2 when @var{from} is the | |
4551 | same as @var{to}; on some machines it is expensive to move between | |
4552 | registers if they are not general registers. | |
4553 | ||
4554 | If reload sees an insn consisting of a single @code{set} between two | |
4555 | hard registers, and if @code{REGISTER_MOVE_COST} applied to their | |
4556 | classes returns a value of 2, reload does not check to ensure that the | |
4557 | constraints of the insn are met. Setting a cost of other than 2 will | |
4558 | allow reload to verify that the constraints are met. You should do this | |
4559 | if the @samp{mov@var{m}} pattern's constraints do not allow such copying. | |
4560 | ||
4561 | These macros are obsolete, new ports should use the target hook | |
4562 | @code{TARGET_REGISTER_MOVE_COST} instead. | |
4563 | @end defmac | |
4564 | ||
4565 | @hook TARGET_REGISTER_MOVE_COST | |
202d6e5f | 4566 | |
4567 | @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in}) | |
4568 | A C expression for the cost of moving data of mode @var{mode} between a | |
4569 | register of class @var{class} and memory; @var{in} is zero if the value | |
4570 | is to be written to memory, nonzero if it is to be read in. This cost | |
4571 | is relative to those in @code{REGISTER_MOVE_COST}. If moving between | |
4572 | registers and memory is more expensive than between two registers, you | |
4573 | should define this macro to express the relative cost. | |
4574 | ||
4575 | If you do not define this macro, GCC uses a default cost of 4 plus | |
4576 | the cost of copying via a secondary reload register, if one is | |
4577 | needed. If your machine requires a secondary reload register to copy | |
4578 | between memory and a register of @var{class} but the reload mechanism is | |
4579 | more complex than copying via an intermediate, define this macro to | |
4580 | reflect the actual cost of the move. | |
4581 | ||
4582 | GCC defines the function @code{memory_move_secondary_cost} if | |
4583 | secondary reloads are needed. It computes the costs due to copying via | |
4584 | a secondary register. If your machine copies from memory using a | |
4585 | secondary register in the conventional way but the default base value of | |
4586 | 4 is not correct for your machine, define this macro to add some other | |
4587 | value to the result of that function. The arguments to that function | |
4588 | are the same as to this macro. | |
4589 | ||
4590 | These macros are obsolete, new ports should use the target hook | |
4591 | @code{TARGET_MEMORY_MOVE_COST} instead. | |
4592 | @end defmac | |
4593 | ||
9de204fa | 4594 | @hook TARGET_MEMORY_MOVE_COST |
202d6e5f | 4595 | |
4596 | @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p}) | |
b56a5220 | 4597 | A C expression for the cost of a branch instruction. A value of 1 is |
4598 | the default; other values are interpreted relative to that. Parameter | |
4599 | @var{speed_p} is true when the branch in question should be optimized | |
4600 | for speed. When it is false, @code{BRANCH_COST} should return a value | |
4601 | optimal for code size rather than performance. @var{predictable_p} is | |
4602 | true for well-predicted branches. On many architectures the | |
4603 | @code{BRANCH_COST} can be reduced then. | |
202d6e5f | 4604 | @end defmac |
4605 | ||
4606 | Here are additional macros which do not specify precise relative costs, | |
4607 | but only that certain actions are more expensive than GCC would | |
4608 | ordinarily expect. | |
4609 | ||
4610 | @defmac SLOW_BYTE_ACCESS | |
4611 | Define this macro as a C expression which is nonzero if accessing less | |
4612 | than a word of memory (i.e.@: a @code{char} or a @code{short}) is no | |
4613 | faster than accessing a word of memory, i.e., if such access | |
4614 | require more than one instruction or if there is no difference in cost | |
4615 | between byte and (aligned) word loads. | |
4616 | ||
4617 | When this macro is not defined, the compiler will access a field by | |
4618 | finding the smallest containing object; when it is defined, a fullword | |
4619 | load will be used if alignment permits. Unless bytes accesses are | |
4620 | faster than word accesses, using word accesses is preferable since it | |
4621 | may eliminate subsequent memory access if subsequent accesses occur to | |
4622 | other fields in the same word of the structure, but to different bytes. | |
4623 | @end defmac | |
4624 | ||
4625 | @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment}) | |
4626 | Define this macro to be the value 1 if memory accesses described by the | |
4627 | @var{mode} and @var{alignment} parameters have a cost many times greater | |
4628 | than aligned accesses, for example if they are emulated in a trap | |
4629 | handler. | |
4630 | ||
4631 | When this macro is nonzero, the compiler will act as if | |
4632 | @code{STRICT_ALIGNMENT} were nonzero when generating code for block | |
4633 | moves. This can cause significantly more instructions to be produced. | |
4634 | Therefore, do not set this macro nonzero if unaligned accesses only add a | |
4635 | cycle or two to the time for a memory access. | |
4636 | ||
4637 | If the value of this macro is always zero, it need not be defined. If | |
4638 | this macro is defined, it should produce a nonzero value when | |
4639 | @code{STRICT_ALIGNMENT} is nonzero. | |
4640 | @end defmac | |
4641 | ||
4642 | @defmac MOVE_RATIO (@var{speed}) | |
4643 | The threshold of number of scalar memory-to-memory move insns, @emph{below} | |
4644 | which a sequence of insns should be generated instead of a | |
4645 | string move insn or a library call. Increasing the value will always | |
4646 | make code faster, but eventually incurs high cost in increased code size. | |
4647 | ||
4648 | Note that on machines where the corresponding move insn is a | |
4649 | @code{define_expand} that emits a sequence of insns, this macro counts | |
4650 | the number of such sequences. | |
4651 | ||
4652 | The parameter @var{speed} is true if the code is currently being | |
4653 | optimized for speed rather than size. | |
4654 | ||
4655 | If you don't define this, a reasonable default is used. | |
4656 | @end defmac | |
4657 | ||
0ec3791c | 4658 | @hook TARGET_USE_BY_PIECES_INFRASTRUCTURE_P |
4659 | ||
202d6e5f | 4660 | @defmac MOVE_MAX_PIECES |
4661 | A C expression used by @code{move_by_pieces} to determine the largest unit | |
4662 | a load or store used to copy memory is. Defaults to @code{MOVE_MAX}. | |
4663 | @end defmac | |
4664 | ||
4665 | @defmac CLEAR_RATIO (@var{speed}) | |
4666 | The threshold of number of scalar move insns, @emph{below} which a sequence | |
4667 | of insns should be generated to clear memory instead of a string clear insn | |
4668 | or a library call. Increasing the value will always make code faster, but | |
4669 | eventually incurs high cost in increased code size. | |
4670 | ||
4671 | The parameter @var{speed} is true if the code is currently being | |
4672 | optimized for speed rather than size. | |
4673 | ||
4674 | If you don't define this, a reasonable default is used. | |
4675 | @end defmac | |
4676 | ||
202d6e5f | 4677 | @defmac SET_RATIO (@var{speed}) |
4678 | The threshold of number of scalar move insns, @emph{below} which a sequence | |
4679 | of insns should be generated to set memory to a constant value, instead of | |
15b474a2 | 4680 | a block set insn or a library call. |
202d6e5f | 4681 | Increasing the value will always make code faster, but |
4682 | eventually incurs high cost in increased code size. | |
4683 | ||
4684 | The parameter @var{speed} is true if the code is currently being | |
4685 | optimized for speed rather than size. | |
4686 | ||
4687 | If you don't define this, it defaults to the value of @code{MOVE_RATIO}. | |
4688 | @end defmac | |
4689 | ||
202d6e5f | 4690 | @defmac USE_LOAD_POST_INCREMENT (@var{mode}) |
4691 | A C expression used to determine whether a load postincrement is a good | |
4692 | thing to use for a given mode. Defaults to the value of | |
4693 | @code{HAVE_POST_INCREMENT}. | |
4694 | @end defmac | |
4695 | ||
4696 | @defmac USE_LOAD_POST_DECREMENT (@var{mode}) | |
4697 | A C expression used to determine whether a load postdecrement is a good | |
4698 | thing to use for a given mode. Defaults to the value of | |
4699 | @code{HAVE_POST_DECREMENT}. | |
4700 | @end defmac | |
4701 | ||
4702 | @defmac USE_LOAD_PRE_INCREMENT (@var{mode}) | |
4703 | A C expression used to determine whether a load preincrement is a good | |
4704 | thing to use for a given mode. Defaults to the value of | |
4705 | @code{HAVE_PRE_INCREMENT}. | |
4706 | @end defmac | |
4707 | ||
4708 | @defmac USE_LOAD_PRE_DECREMENT (@var{mode}) | |
4709 | A C expression used to determine whether a load predecrement is a good | |
4710 | thing to use for a given mode. Defaults to the value of | |
4711 | @code{HAVE_PRE_DECREMENT}. | |
4712 | @end defmac | |
4713 | ||
4714 | @defmac USE_STORE_POST_INCREMENT (@var{mode}) | |
4715 | A C expression used to determine whether a store postincrement is a good | |
4716 | thing to use for a given mode. Defaults to the value of | |
4717 | @code{HAVE_POST_INCREMENT}. | |
4718 | @end defmac | |
4719 | ||
4720 | @defmac USE_STORE_POST_DECREMENT (@var{mode}) | |
4721 | A C expression used to determine whether a store postdecrement is a good | |
4722 | thing to use for a given mode. Defaults to the value of | |
4723 | @code{HAVE_POST_DECREMENT}. | |
4724 | @end defmac | |
4725 | ||
4726 | @defmac USE_STORE_PRE_INCREMENT (@var{mode}) | |
4727 | This macro is used to determine whether a store preincrement is a good | |
4728 | thing to use for a given mode. Defaults to the value of | |
4729 | @code{HAVE_PRE_INCREMENT}. | |
4730 | @end defmac | |
4731 | ||
4732 | @defmac USE_STORE_PRE_DECREMENT (@var{mode}) | |
4733 | This macro is used to determine whether a store predecrement is a good | |
4734 | thing to use for a given mode. Defaults to the value of | |
4735 | @code{HAVE_PRE_DECREMENT}. | |
4736 | @end defmac | |
4737 | ||
4738 | @defmac NO_FUNCTION_CSE | |
93516111 | 4739 | Define this macro to be true if it is as good or better to call a constant |
202d6e5f | 4740 | function address than to call an address kept in a register. |
4741 | @end defmac | |
4742 | ||
802c6441 | 4743 | @defmac LOGICAL_OP_NON_SHORT_CIRCUIT |
4744 | Define this macro if a non-short-circuit operation produced by | |
4745 | @samp{fold_range_test ()} is optimal. This macro defaults to true if | |
4746 | @code{BRANCH_COST} is greater than or equal to the value 2. | |
4747 | @end defmac | |
4748 | ||
acdfe9e0 | 4749 | @hook TARGET_OPTAB_SUPPORTED_P |
4750 | ||
202d6e5f | 4751 | @hook TARGET_RTX_COSTS |
202d6e5f | 4752 | |
4753 | @hook TARGET_ADDRESS_COST | |
202d6e5f | 4754 | |
0f177fa9 | 4755 | @hook TARGET_NO_SPECULATION_IN_DELAY_SLOTS_P |
4756 | ||
202d6e5f | 4757 | @node Scheduling |
4758 | @section Adjusting the Instruction Scheduler | |
4759 | ||
4760 | The instruction scheduler may need a fair amount of machine-specific | |
4761 | adjustment in order to produce good code. GCC provides several target | |
4762 | hooks for this purpose. It is usually enough to define just a few of | |
4763 | them: try the first ones in this list first. | |
4764 | ||
4765 | @hook TARGET_SCHED_ISSUE_RATE | |
202d6e5f | 4766 | |
4767 | @hook TARGET_SCHED_VARIABLE_ISSUE | |
202d6e5f | 4768 | |
4769 | @hook TARGET_SCHED_ADJUST_COST | |
202d6e5f | 4770 | |
4771 | @hook TARGET_SCHED_ADJUST_PRIORITY | |
202d6e5f | 4772 | |
4773 | @hook TARGET_SCHED_REORDER | |
202d6e5f | 4774 | |
4775 | @hook TARGET_SCHED_REORDER2 | |
202d6e5f | 4776 | |
641b578b | 4777 | @hook TARGET_SCHED_MACRO_FUSION_P |
4778 | ||
4779 | @hook TARGET_SCHED_MACRO_FUSION_PAIR_P | |
4780 | ||
202d6e5f | 4781 | @hook TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK |
202d6e5f | 4782 | |
4783 | @hook TARGET_SCHED_INIT | |
202d6e5f | 4784 | |
4785 | @hook TARGET_SCHED_FINISH | |
202d6e5f | 4786 | |
4787 | @hook TARGET_SCHED_INIT_GLOBAL | |
202d6e5f | 4788 | |
4789 | @hook TARGET_SCHED_FINISH_GLOBAL | |
202d6e5f | 4790 | |
4791 | @hook TARGET_SCHED_DFA_PRE_CYCLE_INSN | |
202d6e5f | 4792 | |
4793 | @hook TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN | |
202d6e5f | 4794 | |
4795 | @hook TARGET_SCHED_DFA_POST_CYCLE_INSN | |
202d6e5f | 4796 | |
4797 | @hook TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN | |
202d6e5f | 4798 | |
4799 | @hook TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE | |
202d6e5f | 4800 | |
4801 | @hook TARGET_SCHED_DFA_POST_ADVANCE_CYCLE | |
202d6e5f | 4802 | |
4803 | @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD | |
202d6e5f | 4804 | |
4805 | @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD | |
4806 | ||
44ad1e56 | 4807 | @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BEGIN |
44ad1e56 | 4808 | |
4809 | @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_ISSUE | |
44ad1e56 | 4810 | |
4811 | @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BACKTRACK | |
44ad1e56 | 4812 | |
4813 | @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_END | |
44ad1e56 | 4814 | |
4815 | @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_INIT | |
44ad1e56 | 4816 | |
4817 | @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_FINI | |
44ad1e56 | 4818 | |
ec1842d2 | 4819 | @hook TARGET_SCHED_DFA_NEW_CYCLE |
202d6e5f | 4820 | |
4821 | @hook TARGET_SCHED_IS_COSTLY_DEPENDENCE | |
202d6e5f | 4822 | |
4823 | @hook TARGET_SCHED_H_I_D_EXTENDED | |
202d6e5f | 4824 | |
4825 | @hook TARGET_SCHED_ALLOC_SCHED_CONTEXT | |
202d6e5f | 4826 | |
4827 | @hook TARGET_SCHED_INIT_SCHED_CONTEXT | |
202d6e5f | 4828 | |
4829 | @hook TARGET_SCHED_SET_SCHED_CONTEXT | |
202d6e5f | 4830 | |
4831 | @hook TARGET_SCHED_CLEAR_SCHED_CONTEXT | |
202d6e5f | 4832 | |
4833 | @hook TARGET_SCHED_FREE_SCHED_CONTEXT | |
202d6e5f | 4834 | |
4835 | @hook TARGET_SCHED_SPECULATE_INSN | |
202d6e5f | 4836 | |
4837 | @hook TARGET_SCHED_NEEDS_BLOCK_P | |
202d6e5f | 4838 | |
4839 | @hook TARGET_SCHED_GEN_SPEC_CHECK | |
202d6e5f | 4840 | |
202d6e5f | 4841 | @hook TARGET_SCHED_SET_SCHED_FLAGS |
202d6e5f | 4842 | |
4843 | @hook TARGET_SCHED_SMS_RES_MII | |
202d6e5f | 4844 | |
0ac9454e | 4845 | @hook TARGET_SCHED_DISPATCH |
0ac9454e | 4846 | |
4847 | @hook TARGET_SCHED_DISPATCH_DO | |
0ac9454e | 4848 | |
6c2d9e41 | 4849 | @hook TARGET_SCHED_EXPOSED_PIPELINE |
4850 | ||
5b1c765d | 4851 | @hook TARGET_SCHED_REASSOCIATION_WIDTH |
4852 | ||
012ad66c | 4853 | @hook TARGET_SCHED_FUSION_PRIORITY |
4854 | ||
202d6e5f | 4855 | @node Sections |
4856 | @section Dividing the Output into Sections (Texts, Data, @dots{}) | |
4857 | @c the above section title is WAY too long. maybe cut the part between | |
4858 | @c the (...)? --mew 10feb93 | |
4859 | ||
4860 | An object file is divided into sections containing different types of | |
4861 | data. In the most common case, there are three sections: the @dfn{text | |
4862 | section}, which holds instructions and read-only data; the @dfn{data | |
4863 | section}, which holds initialized writable data; and the @dfn{bss | |
4864 | section}, which holds uninitialized data. Some systems have other kinds | |
4865 | of sections. | |
4866 | ||
4867 | @file{varasm.c} provides several well-known sections, such as | |
4868 | @code{text_section}, @code{data_section} and @code{bss_section}. | |
4869 | The normal way of controlling a @code{@var{foo}_section} variable | |
4870 | is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro, | |
4871 | as described below. The macros are only read once, when @file{varasm.c} | |
4872 | initializes itself, so their values must be run-time constants. | |
4873 | They may however depend on command-line flags. | |
4874 | ||
4875 | @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make | |
4876 | use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them | |
4877 | to be string literals. | |
4878 | ||
4879 | Some assemblers require a different string to be written every time a | |
4880 | section is selected. If your assembler falls into this category, you | |
4881 | should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use | |
4882 | @code{get_unnamed_section} to set up the sections. | |
4883 | ||
4884 | You must always create a @code{text_section}, either by defining | |
4885 | @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section} | |
4886 | in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of | |
4887 | @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not | |
4888 | create a distinct @code{readonly_data_section}, the default is to | |
4889 | reuse @code{text_section}. | |
4890 | ||
4891 | All the other @file{varasm.c} sections are optional, and are null | |
4892 | if the target does not provide them. | |
4893 | ||
4894 | @defmac TEXT_SECTION_ASM_OP | |
4895 | A C expression whose value is a string, including spacing, containing the | |
4896 | assembler operation that should precede instructions and read-only data. | |
4897 | Normally @code{"\t.text"} is right. | |
4898 | @end defmac | |
4899 | ||
4900 | @defmac HOT_TEXT_SECTION_NAME | |
4901 | If defined, a C string constant for the name of the section containing most | |
4902 | frequently executed functions of the program. If not defined, GCC will provide | |
4903 | a default definition if the target supports named sections. | |
4904 | @end defmac | |
4905 | ||
4906 | @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME | |
4907 | If defined, a C string constant for the name of the section containing unlikely | |
4908 | executed functions in the program. | |
4909 | @end defmac | |
4910 | ||
4911 | @defmac DATA_SECTION_ASM_OP | |
4912 | A C expression whose value is a string, including spacing, containing the | |
4913 | assembler operation to identify the following data as writable initialized | |
4914 | data. Normally @code{"\t.data"} is right. | |
4915 | @end defmac | |
4916 | ||
4917 | @defmac SDATA_SECTION_ASM_OP | |
4918 | If defined, a C expression whose value is a string, including spacing, | |
4919 | containing the assembler operation to identify the following data as | |
4920 | initialized, writable small data. | |
4921 | @end defmac | |
4922 | ||
4923 | @defmac READONLY_DATA_SECTION_ASM_OP | |
4924 | A C expression whose value is a string, including spacing, containing the | |
4925 | assembler operation to identify the following data as read-only initialized | |
4926 | data. | |
4927 | @end defmac | |
4928 | ||
4929 | @defmac BSS_SECTION_ASM_OP | |
4930 | If defined, a C expression whose value is a string, including spacing, | |
4931 | containing the assembler operation to identify the following data as | |
fa8d6f0f | 4932 | uninitialized global data. If not defined, and |
4933 | @code{ASM_OUTPUT_ALIGNED_BSS} not defined, | |
202d6e5f | 4934 | uninitialized global data will be output in the data section if |
4935 | @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be | |
4936 | used. | |
4937 | @end defmac | |
4938 | ||
4939 | @defmac SBSS_SECTION_ASM_OP | |
4940 | If defined, a C expression whose value is a string, including spacing, | |
4941 | containing the assembler operation to identify the following data as | |
4942 | uninitialized, writable small data. | |
4943 | @end defmac | |
4944 | ||
4945 | @defmac TLS_COMMON_ASM_OP | |
4946 | If defined, a C expression whose value is a string containing the | |
4947 | assembler operation to identify the following data as thread-local | |
4948 | common data. The default is @code{".tls_common"}. | |
4949 | @end defmac | |
4950 | ||
4951 | @defmac TLS_SECTION_ASM_FLAG | |
4952 | If defined, a C expression whose value is a character constant | |
4953 | containing the flag used to mark a section as a TLS section. The | |
4954 | default is @code{'T'}. | |
4955 | @end defmac | |
4956 | ||
4957 | @defmac INIT_SECTION_ASM_OP | |
4958 | If defined, a C expression whose value is a string, including spacing, | |
4959 | containing the assembler operation to identify the following data as | |
4960 | initialization code. If not defined, GCC will assume such a section does | |
4961 | not exist. This section has no corresponding @code{init_section} | |
4962 | variable; it is used entirely in runtime code. | |
4963 | @end defmac | |
4964 | ||
4965 | @defmac FINI_SECTION_ASM_OP | |
4966 | If defined, a C expression whose value is a string, including spacing, | |
4967 | containing the assembler operation to identify the following data as | |
4968 | finalization code. If not defined, GCC will assume such a section does | |
4969 | not exist. This section has no corresponding @code{fini_section} | |
4970 | variable; it is used entirely in runtime code. | |
4971 | @end defmac | |
4972 | ||
4973 | @defmac INIT_ARRAY_SECTION_ASM_OP | |
4974 | If defined, a C expression whose value is a string, including spacing, | |
4975 | containing the assembler operation to identify the following data as | |
4976 | part of the @code{.init_array} (or equivalent) section. If not | |
4977 | defined, GCC will assume such a section does not exist. Do not define | |
4978 | both this macro and @code{INIT_SECTION_ASM_OP}. | |
4979 | @end defmac | |
4980 | ||
4981 | @defmac FINI_ARRAY_SECTION_ASM_OP | |
4982 | If defined, a C expression whose value is a string, including spacing, | |
4983 | containing the assembler operation to identify the following data as | |
4984 | part of the @code{.fini_array} (or equivalent) section. If not | |
4985 | defined, GCC will assume such a section does not exist. Do not define | |
4986 | both this macro and @code{FINI_SECTION_ASM_OP}. | |
4987 | @end defmac | |
4988 | ||
4989 | @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function}) | |
4990 | If defined, an ASM statement that switches to a different section | |
4991 | via @var{section_op}, calls @var{function}, and switches back to | |
4992 | the text section. This is used in @file{crtstuff.c} if | |
4993 | @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls | |
4994 | to initialization and finalization functions from the init and fini | |
4995 | sections. By default, this macro uses a simple function call. Some | |
4996 | ports need hand-crafted assembly code to avoid dependencies on | |
4997 | registers initialized in the function prologue or to ensure that | |
4998 | constant pools don't end up too far way in the text section. | |
4999 | @end defmac | |
5000 | ||
5001 | @defmac TARGET_LIBGCC_SDATA_SECTION | |
5002 | If defined, a string which names the section into which small | |
5003 | variables defined in crtstuff and libgcc should go. This is useful | |
5004 | when the target has options for optimizing access to small data, and | |
5005 | you want the crtstuff and libgcc routines to be conservative in what | |
5006 | they expect of your application yet liberal in what your application | |
5007 | expects. For example, for targets with a @code{.sdata} section (like | |
5008 | MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't | |
5009 | require small data support from your application, but use this macro | |
5010 | to put small data into @code{.sdata} so that your application can | |
5011 | access these variables whether it uses small data or not. | |
5012 | @end defmac | |
5013 | ||
5014 | @defmac FORCE_CODE_SECTION_ALIGN | |
5015 | If defined, an ASM statement that aligns a code section to some | |
5016 | arbitrary boundary. This is used to force all fragments of the | |
5017 | @code{.init} and @code{.fini} sections to have to same alignment | |
5018 | and thus prevent the linker from having to add any padding. | |
5019 | @end defmac | |
5020 | ||
5021 | @defmac JUMP_TABLES_IN_TEXT_SECTION | |
5022 | Define this macro to be an expression with a nonzero value if jump | |
5023 | tables (for @code{tablejump} insns) should be output in the text | |
5024 | section, along with the assembler instructions. Otherwise, the | |
5025 | readonly data section is used. | |
5026 | ||
5027 | This macro is irrelevant if there is no separate readonly data section. | |
5028 | @end defmac | |
5029 | ||
5030 | @hook TARGET_ASM_INIT_SECTIONS | |
202d6e5f | 5031 | |
5032 | @hook TARGET_ASM_RELOC_RW_MASK | |
202d6e5f | 5033 | |
5034 | @hook TARGET_ASM_SELECT_SECTION | |
202d6e5f | 5035 | |
5036 | @defmac USE_SELECT_SECTION_FOR_FUNCTIONS | |
5037 | Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called | |
5038 | for @code{FUNCTION_DECL}s as well as for variables and constants. | |
5039 | ||
5040 | In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the | |
5041 | function has been determined to be likely to be called, and nonzero if | |
5042 | it is unlikely to be called. | |
5043 | @end defmac | |
5044 | ||
5045 | @hook TARGET_ASM_UNIQUE_SECTION | |
202d6e5f | 5046 | |
5047 | @hook TARGET_ASM_FUNCTION_RODATA_SECTION | |
202d6e5f | 5048 | |
6e71bd21 | 5049 | @hook TARGET_ASM_MERGEABLE_RODATA_PREFIX |
5050 | ||
e3ef604c | 5051 | @hook TARGET_ASM_TM_CLONE_TABLE_SECTION |
5052 | ||
202d6e5f | 5053 | @hook TARGET_ASM_SELECT_RTX_SECTION |
202d6e5f | 5054 | |
5055 | @hook TARGET_MANGLE_DECL_ASSEMBLER_NAME | |
202d6e5f | 5056 | |
5057 | @hook TARGET_ENCODE_SECTION_INFO | |
202d6e5f | 5058 | |
5059 | @hook TARGET_STRIP_NAME_ENCODING | |
202d6e5f | 5060 | |
5061 | @hook TARGET_IN_SMALL_DATA_P | |
202d6e5f | 5062 | |
5063 | @hook TARGET_HAVE_SRODATA_SECTION | |
202d6e5f | 5064 | |
8637d6a2 | 5065 | @hook TARGET_PROFILE_BEFORE_PROLOGUE |
5066 | ||
202d6e5f | 5067 | @hook TARGET_BINDS_LOCAL_P |
202d6e5f | 5068 | |
5069 | @hook TARGET_HAVE_TLS | |
202d6e5f | 5070 | |
5071 | ||
5072 | @node PIC | |
5073 | @section Position Independent Code | |
5074 | @cindex position independent code | |
5075 | @cindex PIC | |
5076 | ||
5077 | This section describes macros that help implement generation of position | |
5078 | independent code. Simply defining these macros is not enough to | |
5079 | generate valid PIC; you must also add support to the hook | |
5080 | @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro | |
5081 | @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You | |
5082 | must modify the definition of @samp{movsi} to do something appropriate | |
5083 | when the source operand contains a symbolic address. You may also | |
5084 | need to alter the handling of switch statements so that they use | |
5085 | relative addresses. | |
15b474a2 | 5086 | @c i rearranged the order of the macros above to try to force one of |
202d6e5f | 5087 | @c them to the next line, to eliminate an overfull hbox. --mew 10feb93 |
5088 | ||
5089 | @defmac PIC_OFFSET_TABLE_REGNUM | |
5090 | The register number of the register used to address a table of static | |
5091 | data addresses in memory. In some cases this register is defined by a | |
5092 | processor's ``application binary interface'' (ABI)@. When this macro | |
5093 | is defined, RTL is generated for this register once, as with the stack | |
5094 | pointer and frame pointer registers. If this macro is not defined, it | |
5095 | is up to the machine-dependent files to allocate such a register (if | |
5096 | necessary). Note that this register must be fixed when in use (e.g.@: | |
5097 | when @code{flag_pic} is true). | |
5098 | @end defmac | |
5099 | ||
5100 | @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED | |
260e669e | 5101 | A C expression that is nonzero if the register defined by |
5102 | @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. If not defined, | |
5103 | the default is zero. Do not define | |
202d6e5f | 5104 | this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined. |
5105 | @end defmac | |
5106 | ||
5107 | @defmac LEGITIMATE_PIC_OPERAND_P (@var{x}) | |
5108 | A C expression that is nonzero if @var{x} is a legitimate immediate | |
5109 | operand on the target machine when generating position independent code. | |
5110 | You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not | |
5111 | check this. You can also assume @var{flag_pic} is true, so you need not | |
5112 | check it either. You need not define this macro if all constants | |
5113 | (including @code{SYMBOL_REF}) can be immediate operands when generating | |
5114 | position independent code. | |
5115 | @end defmac | |
5116 | ||
5117 | @node Assembler Format | |
5118 | @section Defining the Output Assembler Language | |
5119 | ||
5120 | This section describes macros whose principal purpose is to describe how | |
5121 | to write instructions in assembler language---rather than what the | |
5122 | instructions do. | |
5123 | ||
5124 | @menu | |
5125 | * File Framework:: Structural information for the assembler file. | |
5126 | * Data Output:: Output of constants (numbers, strings, addresses). | |
5127 | * Uninitialized Data:: Output of uninitialized variables. | |
5128 | * Label Output:: Output and generation of labels. | |
5129 | * Initialization:: General principles of initialization | |
5130 | and termination routines. | |
5131 | * Macros for Initialization:: | |
5132 | Specific macros that control the handling of | |
5133 | initialization and termination routines. | |
5134 | * Instruction Output:: Output of actual instructions. | |
5135 | * Dispatch Tables:: Output of jump tables. | |
5136 | * Exception Region Output:: Output of exception region code. | |
5137 | * Alignment Output:: Pseudo ops for alignment and skipping data. | |
5138 | @end menu | |
5139 | ||
5140 | @node File Framework | |
5141 | @subsection The Overall Framework of an Assembler File | |
5142 | @cindex assembler format | |
5143 | @cindex output of assembler code | |
5144 | ||
5145 | @c prevent bad page break with this line | |
5146 | This describes the overall framework of an assembly file. | |
5147 | ||
5148 | @findex default_file_start | |
5149 | @hook TARGET_ASM_FILE_START | |
202d6e5f | 5150 | |
5151 | @hook TARGET_ASM_FILE_START_APP_OFF | |
202d6e5f | 5152 | |
5153 | @hook TARGET_ASM_FILE_START_FILE_DIRECTIVE | |
202d6e5f | 5154 | |
5155 | @hook TARGET_ASM_FILE_END | |
202d6e5f | 5156 | |
5157 | @deftypefun void file_end_indicate_exec_stack () | |
5158 | Some systems use a common convention, the @samp{.note.GNU-stack} | |
5159 | special section, to indicate whether or not an object file relies on | |
5160 | the stack being executable. If your system uses this convention, you | |
5161 | should define @code{TARGET_ASM_FILE_END} to this function. If you | |
5162 | need to do other things in that hook, have your hook function call | |
5163 | this function. | |
5164 | @end deftypefun | |
5165 | ||
5166 | @hook TARGET_ASM_LTO_START | |
202d6e5f | 5167 | |
5168 | @hook TARGET_ASM_LTO_END | |
202d6e5f | 5169 | |
5170 | @hook TARGET_ASM_CODE_END | |
202d6e5f | 5171 | |
5172 | @defmac ASM_COMMENT_START | |
5173 | A C string constant describing how to begin a comment in the target | |
5174 | assembler language. The compiler assumes that the comment will end at | |
5175 | the end of the line. | |
5176 | @end defmac | |
5177 | ||
5178 | @defmac ASM_APP_ON | |
5179 | A C string constant for text to be output before each @code{asm} | |
5180 | statement or group of consecutive ones. Normally this is | |
5181 | @code{"#APP"}, which is a comment that has no effect on most | |
5182 | assemblers but tells the GNU assembler that it must check the lines | |
5183 | that follow for all valid assembler constructs. | |
5184 | @end defmac | |
5185 | ||
5186 | @defmac ASM_APP_OFF | |
5187 | A C string constant for text to be output after each @code{asm} | |
5188 | statement or group of consecutive ones. Normally this is | |
5189 | @code{"#NO_APP"}, which tells the GNU assembler to resume making the | |
5190 | time-saving assumptions that are valid for ordinary compiler output. | |
5191 | @end defmac | |
5192 | ||
5193 | @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name}) | |
5194 | A C statement to output COFF information or DWARF debugging information | |
5195 | which indicates that filename @var{name} is the current source file to | |
5196 | the stdio stream @var{stream}. | |
5197 | ||
5198 | This macro need not be defined if the standard form of output | |
5199 | for the file format in use is appropriate. | |
5200 | @end defmac | |
5201 | ||
e2823fde | 5202 | @hook TARGET_ASM_OUTPUT_SOURCE_FILENAME |
5203 | ||
367b1459 | 5204 | @hook TARGET_ASM_OUTPUT_IDENT |
5205 | ||
202d6e5f | 5206 | @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string}) |
5207 | A C statement to output the string @var{string} to the stdio stream | |
5208 | @var{stream}. If you do not call the function @code{output_quoted_string} | |
5209 | in your config files, GCC will only call it to output filenames to | |
5210 | the assembler source. So you can use it to canonicalize the format | |
5211 | of the filename using this macro. | |
5212 | @end defmac | |
5213 | ||
202d6e5f | 5214 | @hook TARGET_ASM_NAMED_SECTION |
202d6e5f | 5215 | |
3b73548b | 5216 | @hook TARGET_ASM_FUNCTION_SECTION |
3b73548b | 5217 | |
3b3edcd4 | 5218 | @hook TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS |
5219 | ||
202d6e5f | 5220 | @hook TARGET_HAVE_NAMED_SECTIONS |
5221 | This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}. | |
b213bf24 | 5222 | It must not be modified by command-line option processing. |
202d6e5f | 5223 | @end deftypevr |
5224 | ||
5225 | @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS} | |
5226 | @hook TARGET_HAVE_SWITCHABLE_BSS_SECTIONS | |
202d6e5f | 5227 | |
5228 | @hook TARGET_SECTION_TYPE_FLAGS | |
202d6e5f | 5229 | |
5230 | @hook TARGET_ASM_RECORD_GCC_SWITCHES | |
202d6e5f | 5231 | |
5232 | @hook TARGET_ASM_RECORD_GCC_SWITCHES_SECTION | |
202d6e5f | 5233 | |
5234 | @need 2000 | |
5235 | @node Data Output | |
5236 | @subsection Output of Data | |
5237 | ||
5238 | ||
5239 | @hook TARGET_ASM_BYTE_OP | |
202d6e5f | 5240 | |
5241 | @hook TARGET_ASM_INTEGER | |
202d6e5f | 5242 | |
d895e9da | 5243 | @hook TARGET_ASM_DECL_END |
5244 | ||
1282f299 | 5245 | @hook TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA |
1282f299 | 5246 | |
202d6e5f | 5247 | @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len}) |
5248 | A C statement to output to the stdio stream @var{stream} an assembler | |
5249 | instruction to assemble a string constant containing the @var{len} | |
5250 | bytes at @var{ptr}. @var{ptr} will be a C expression of type | |
5251 | @code{char *} and @var{len} a C expression of type @code{int}. | |
5252 | ||
5253 | If the assembler has a @code{.ascii} pseudo-op as found in the | |
5254 | Berkeley Unix assembler, do not define the macro | |
5255 | @code{ASM_OUTPUT_ASCII}. | |
5256 | @end defmac | |
5257 | ||
5258 | @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n}) | |
5259 | A C statement to output word @var{n} of a function descriptor for | |
5260 | @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS} | |
5261 | is defined, and is otherwise unused. | |
5262 | @end defmac | |
5263 | ||
5264 | @defmac CONSTANT_POOL_BEFORE_FUNCTION | |
5265 | You may define this macro as a C expression. You should define the | |
5266 | expression to have a nonzero value if GCC should output the constant | |
5267 | pool for a function before the code for the function, or a zero value if | |
5268 | GCC should output the constant pool after the function. If you do | |
5269 | not define this macro, the usual case, GCC will output the constant | |
5270 | pool before the function. | |
5271 | @end defmac | |
5272 | ||
5273 | @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size}) | |
5274 | A C statement to output assembler commands to define the start of the | |
5275 | constant pool for a function. @var{funname} is a string giving | |
5276 | the name of the function. Should the return type of the function | |
5277 | be required, it can be obtained via @var{fundecl}. @var{size} | |
5278 | is the size, in bytes, of the constant pool that will be written | |
5279 | immediately after this call. | |
5280 | ||
5281 | If no constant-pool prefix is required, the usual case, this macro need | |
5282 | not be defined. | |
5283 | @end defmac | |
5284 | ||
5285 | @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto}) | |
5286 | A C statement (with or without semicolon) to output a constant in the | |
5287 | constant pool, if it needs special treatment. (This macro need not do | |
5288 | anything for RTL expressions that can be output normally.) | |
5289 | ||
5290 | The argument @var{file} is the standard I/O stream to output the | |
5291 | assembler code on. @var{x} is the RTL expression for the constant to | |
5292 | output, and @var{mode} is the machine mode (in case @var{x} is a | |
5293 | @samp{const_int}). @var{align} is the required alignment for the value | |
5294 | @var{x}; you should output an assembler directive to force this much | |
5295 | alignment. | |
5296 | ||
5297 | The argument @var{labelno} is a number to use in an internal label for | |
5298 | the address of this pool entry. The definition of this macro is | |
5299 | responsible for outputting the label definition at the proper place. | |
5300 | Here is how to do this: | |
5301 | ||
5302 | @smallexample | |
5303 | @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno}); | |
5304 | @end smallexample | |
5305 | ||
5306 | When you output a pool entry specially, you should end with a | |
5307 | @code{goto} to the label @var{jumpto}. This will prevent the same pool | |
5308 | entry from being output a second time in the usual manner. | |
5309 | ||
5310 | You need not define this macro if it would do nothing. | |
5311 | @end defmac | |
5312 | ||
5313 | @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size}) | |
5314 | A C statement to output assembler commands to at the end of the constant | |
5315 | pool for a function. @var{funname} is a string giving the name of the | |
5316 | function. Should the return type of the function be required, you can | |
5317 | obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the | |
5318 | constant pool that GCC wrote immediately before this call. | |
5319 | ||
5320 | If no constant-pool epilogue is required, the usual case, you need not | |
5321 | define this macro. | |
5322 | @end defmac | |
5323 | ||
5324 | @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR}) | |
5325 | Define this macro as a C expression which is nonzero if @var{C} is | |
5326 | used as a logical line separator by the assembler. @var{STR} points | |
5327 | to the position in the string where @var{C} was found; this can be used if | |
5328 | a line separator uses multiple characters. | |
5329 | ||
5330 | If you do not define this macro, the default is that only | |
5331 | the character @samp{;} is treated as a logical line separator. | |
5332 | @end defmac | |
5333 | ||
5334 | @hook TARGET_ASM_OPEN_PAREN | |
202d6e5f | 5335 | |
5336 | These macros are provided by @file{real.h} for writing the definitions | |
5337 | of @code{ASM_OUTPUT_DOUBLE} and the like: | |
5338 | ||
5339 | @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l}) | |
5340 | @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l}) | |
5341 | @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l}) | |
5342 | @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l}) | |
5343 | @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l}) | |
5344 | @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l}) | |
5345 | These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the | |
5346 | target's floating point representation, and store its bit pattern in | |
5347 | the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and | |
5348 | @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a | |
5349 | simple @code{long int}. For the others, it should be an array of | |
5350 | @code{long int}. The number of elements in this array is determined | |
5351 | by the size of the desired target floating point data type: 32 bits of | |
5352 | it go in each @code{long int} array element. Each array element holds | |
5353 | 32 bits of the result, even if @code{long int} is wider than 32 bits | |
5354 | on the host machine. | |
5355 | ||
5356 | The array element values are designed so that you can print them out | |
5357 | using @code{fprintf} in the order they should appear in the target | |
5358 | machine's memory. | |
5359 | @end defmac | |
5360 | ||
5361 | @node Uninitialized Data | |
5362 | @subsection Output of Uninitialized Variables | |
5363 | ||
5364 | Each of the macros in this section is used to do the whole job of | |
5365 | outputting a single uninitialized variable. | |
5366 | ||
5367 | @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded}) | |
5368 | A C statement (sans semicolon) to output to the stdio stream | |
5369 | @var{stream} the assembler definition of a common-label named | |
5370 | @var{name} whose size is @var{size} bytes. The variable @var{rounded} | |
5371 | is the size rounded up to whatever alignment the caller wants. It is | |
5372 | possible that @var{size} may be zero, for instance if a struct with no | |
5373 | other member than a zero-length array is defined. In this case, the | |
5374 | backend must output a symbol definition that allocates at least one | |
5375 | byte, both so that the address of the resulting object does not compare | |
5376 | equal to any other, and because some object formats cannot even express | |
5377 | the concept of a zero-sized common symbol, as that is how they represent | |
5378 | an ordinary undefined external. | |
5379 | ||
5380 | Use the expression @code{assemble_name (@var{stream}, @var{name})} to | |
5381 | output the name itself; before and after that, output the additional | |
5382 | assembler syntax for defining the name, and a newline. | |
5383 | ||
5384 | This macro controls how the assembler definitions of uninitialized | |
5385 | common global variables are output. | |
5386 | @end defmac | |
5387 | ||
5388 | @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment}) | |
5389 | Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a | |
5390 | separate, explicit argument. If you define this macro, it is used in | |
5391 | place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in | |
5392 | handling the required alignment of the variable. The alignment is specified | |
5393 | as the number of bits. | |
5394 | @end defmac | |
5395 | ||
5396 | @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment}) | |
5397 | Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the | |
5398 | variable to be output, if there is one, or @code{NULL_TREE} if there | |
5399 | is no corresponding variable. If you define this macro, GCC will use it | |
5400 | in place of both @code{ASM_OUTPUT_COMMON} and | |
5401 | @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see | |
5402 | the variable's decl in order to chose what to output. | |
5403 | @end defmac | |
5404 | ||
fa8d6f0f | 5405 | @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment}) |
202d6e5f | 5406 | A C statement (sans semicolon) to output to the stdio stream |
5407 | @var{stream} the assembler definition of uninitialized global @var{decl} named | |
fa8d6f0f | 5408 | @var{name} whose size is @var{size} bytes. The variable @var{alignment} |
5409 | is the alignment specified as the number of bits. | |
202d6e5f | 5410 | |
fa8d6f0f | 5411 | Try to use function @code{asm_output_aligned_bss} defined in file |
5412 | @file{varasm.c} when defining this macro. If unable, use the expression | |
202d6e5f | 5413 | @code{assemble_name (@var{stream}, @var{name})} to output the name itself; |
5414 | before and after that, output the additional assembler syntax for defining | |
5415 | the name, and a newline. | |
5416 | ||
fa8d6f0f | 5417 | There are two ways of handling global BSS@. One is to define this macro. |
202d6e5f | 5418 | The other is to have @code{TARGET_ASM_SELECT_SECTION} return a |
5419 | switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}). | |
5420 | You do not need to do both. | |
5421 | ||
5422 | Some languages do not have @code{common} data, and require a | |
5423 | non-common form of global BSS in order to handle uninitialized globals | |
5424 | efficiently. C++ is one example of this. However, if the target does | |
5425 | not support global BSS, the front end may choose to make globals | |
5426 | common in order to save space in the object file. | |
5427 | @end defmac | |
5428 | ||
202d6e5f | 5429 | @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded}) |
5430 | A C statement (sans semicolon) to output to the stdio stream | |
5431 | @var{stream} the assembler definition of a local-common-label named | |
5432 | @var{name} whose size is @var{size} bytes. The variable @var{rounded} | |
5433 | is the size rounded up to whatever alignment the caller wants. | |
5434 | ||
5435 | Use the expression @code{assemble_name (@var{stream}, @var{name})} to | |
5436 | output the name itself; before and after that, output the additional | |
5437 | assembler syntax for defining the name, and a newline. | |
5438 | ||
5439 | This macro controls how the assembler definitions of uninitialized | |
5440 | static variables are output. | |
5441 | @end defmac | |
5442 | ||
5443 | @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment}) | |
5444 | Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a | |
5445 | separate, explicit argument. If you define this macro, it is used in | |
5446 | place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in | |
5447 | handling the required alignment of the variable. The alignment is specified | |
5448 | as the number of bits. | |
5449 | @end defmac | |
5450 | ||
5451 | @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment}) | |
5452 | Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the | |
5453 | variable to be output, if there is one, or @code{NULL_TREE} if there | |
5454 | is no corresponding variable. If you define this macro, GCC will use it | |
5455 | in place of both @code{ASM_OUTPUT_DECL} and | |
5456 | @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see | |
5457 | the variable's decl in order to chose what to output. | |
5458 | @end defmac | |
5459 | ||
5460 | @node Label Output | |
5461 | @subsection Output and Generation of Labels | |
5462 | ||
5463 | @c prevent bad page break with this line | |
5464 | This is about outputting labels. | |
5465 | ||
5466 | @findex assemble_name | |
5467 | @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name}) | |
5468 | A C statement (sans semicolon) to output to the stdio stream | |
5469 | @var{stream} the assembler definition of a label named @var{name}. | |
5470 | Use the expression @code{assemble_name (@var{stream}, @var{name})} to | |
5471 | output the name itself; before and after that, output the additional | |
5472 | assembler syntax for defining the name, and a newline. A default | |
5473 | definition of this macro is provided which is correct for most systems. | |
5474 | @end defmac | |
5475 | ||
2761b7cb | 5476 | @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl}) |
5477 | A C statement (sans semicolon) to output to the stdio stream | |
5478 | @var{stream} the assembler definition of a label named @var{name} of | |
5479 | a function. | |
5480 | Use the expression @code{assemble_name (@var{stream}, @var{name})} to | |
5481 | output the name itself; before and after that, output the additional | |
5482 | assembler syntax for defining the name, and a newline. A default | |
5483 | definition of this macro is provided which is correct for most systems. | |
5484 | ||
5485 | If this macro is not defined, then the function name is defined in the | |
5486 | usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}). | |
5487 | @end defmac | |
5488 | ||
202d6e5f | 5489 | @findex assemble_name_raw |
5490 | @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name}) | |
5491 | Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known | |
5492 | to refer to a compiler-generated label. The default definition uses | |
5493 | @code{assemble_name_raw}, which is like @code{assemble_name} except | |
5494 | that it is more efficient. | |
5495 | @end defmac | |
5496 | ||
5497 | @defmac SIZE_ASM_OP | |
5498 | A C string containing the appropriate assembler directive to specify the | |
5499 | size of a symbol, without any arguments. On systems that use ELF, the | |
5500 | default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other | |
5501 | systems, the default is not to define this macro. | |
5502 | ||
5503 | Define this macro only if it is correct to use the default definitions | |
5504 | of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE} | |
5505 | for your system. If you need your own custom definitions of those | |
5506 | macros, or if you do not need explicit symbol sizes at all, do not | |
5507 | define this macro. | |
5508 | @end defmac | |
5509 | ||
5510 | @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size}) | |
5511 | A C statement (sans semicolon) to output to the stdio stream | |
5512 | @var{stream} a directive telling the assembler that the size of the | |
5513 | symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}. | |
5514 | If you define @code{SIZE_ASM_OP}, a default definition of this macro is | |
5515 | provided. | |
5516 | @end defmac | |
5517 | ||
5518 | @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name}) | |
5519 | A C statement (sans semicolon) to output to the stdio stream | |
5520 | @var{stream} a directive telling the assembler to calculate the size of | |
5521 | the symbol @var{name} by subtracting its address from the current | |
5522 | address. | |
5523 | ||
5524 | If you define @code{SIZE_ASM_OP}, a default definition of this macro is | |
5525 | provided. The default assumes that the assembler recognizes a special | |
5526 | @samp{.} symbol as referring to the current address, and can calculate | |
5527 | the difference between this and another symbol. If your assembler does | |
5528 | not recognize @samp{.} or cannot do calculations with it, you will need | |
5529 | to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique. | |
5530 | @end defmac | |
5531 | ||
75a15fa3 | 5532 | @defmac NO_DOLLAR_IN_LABEL |
5533 | Define this macro if the assembler does not accept the character | |
5534 | @samp{$} in label names. By default constructors and destructors in | |
5535 | G++ have @samp{$} in the identifiers. If this macro is defined, | |
5536 | @samp{.} is used instead. | |
5537 | @end defmac | |
5538 | ||
5539 | @defmac NO_DOT_IN_LABEL | |
5540 | Define this macro if the assembler does not accept the character | |
5541 | @samp{.} in label names. By default constructors and destructors in G++ | |
5542 | have names that use @samp{.}. If this macro is defined, these names | |
5543 | are rewritten to avoid @samp{.}. | |
5544 | @end defmac | |
5545 | ||
202d6e5f | 5546 | @defmac TYPE_ASM_OP |
5547 | A C string containing the appropriate assembler directive to specify the | |
5548 | type of a symbol, without any arguments. On systems that use ELF, the | |
5549 | default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other | |
5550 | systems, the default is not to define this macro. | |
5551 | ||
5552 | Define this macro only if it is correct to use the default definition of | |
5553 | @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own | |
5554 | custom definition of this macro, or if you do not need explicit symbol | |
5555 | types at all, do not define this macro. | |
5556 | @end defmac | |
5557 | ||
5558 | @defmac TYPE_OPERAND_FMT | |
5559 | A C string which specifies (using @code{printf} syntax) the format of | |
5560 | the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the | |
5561 | default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems, | |
5562 | the default is not to define this macro. | |
5563 | ||
5564 | Define this macro only if it is correct to use the default definition of | |
5565 | @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own | |
5566 | custom definition of this macro, or if you do not need explicit symbol | |
5567 | types at all, do not define this macro. | |
5568 | @end defmac | |
5569 | ||
5570 | @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type}) | |
5571 | A C statement (sans semicolon) to output to the stdio stream | |
5572 | @var{stream} a directive telling the assembler that the type of the | |
5573 | symbol @var{name} is @var{type}. @var{type} is a C string; currently, | |
5574 | that string is always either @samp{"function"} or @samp{"object"}, but | |
5575 | you should not count on this. | |
5576 | ||
5577 | If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default | |
5578 | definition of this macro is provided. | |
5579 | @end defmac | |
5580 | ||
5581 | @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl}) | |
5582 | A C statement (sans semicolon) to output to the stdio stream | |
5583 | @var{stream} any text necessary for declaring the name @var{name} of a | |
5584 | function which is being defined. This macro is responsible for | |
5585 | outputting the label definition (perhaps using | |
2761b7cb | 5586 | @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the |
202d6e5f | 5587 | @code{FUNCTION_DECL} tree node representing the function. |
5588 | ||
5589 | If this macro is not defined, then the function name is defined in the | |
2761b7cb | 5590 | usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}). |
202d6e5f | 5591 | |
5592 | You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition | |
5593 | of this macro. | |
5594 | @end defmac | |
5595 | ||
5596 | @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl}) | |
5597 | A C statement (sans semicolon) to output to the stdio stream | |
5598 | @var{stream} any text necessary for declaring the size of a function | |
5599 | which is being defined. The argument @var{name} is the name of the | |
5600 | function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node | |
5601 | representing the function. | |
5602 | ||
5603 | If this macro is not defined, then the function size is not defined. | |
5604 | ||
5605 | You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition | |
5606 | of this macro. | |
5607 | @end defmac | |
5608 | ||
3e1bd9b8 | 5609 | @defmac ASM_DECLARE_COLD_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl}) |
5610 | A C statement (sans semicolon) to output to the stdio stream | |
5611 | @var{stream} any text necessary for declaring the name @var{name} of a | |
5612 | cold function partition which is being defined. This macro is responsible | |
5613 | for outputting the label definition (perhaps using | |
5614 | @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the | |
5615 | @code{FUNCTION_DECL} tree node representing the function. | |
5616 | ||
5617 | If this macro is not defined, then the cold partition name is defined in the | |
5618 | usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}). | |
5619 | ||
5620 | You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition | |
5621 | of this macro. | |
5622 | @end defmac | |
5623 | ||
5624 | @defmac ASM_DECLARE_COLD_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl}) | |
5625 | A C statement (sans semicolon) to output to the stdio stream | |
5626 | @var{stream} any text necessary for declaring the size of a cold function | |
5627 | partition which is being defined. The argument @var{name} is the name of the | |
5628 | cold partition of the function. The argument @var{decl} is the | |
5629 | @code{FUNCTION_DECL} tree node representing the function. | |
5630 | ||
5631 | If this macro is not defined, then the partition size is not defined. | |
5632 | ||
5633 | You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition | |
5634 | of this macro. | |
5635 | @end defmac | |
5636 | ||
202d6e5f | 5637 | @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl}) |
5638 | A C statement (sans semicolon) to output to the stdio stream | |
5639 | @var{stream} any text necessary for declaring the name @var{name} of an | |
5640 | initialized variable which is being defined. This macro must output the | |
5641 | label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument | |
5642 | @var{decl} is the @code{VAR_DECL} tree node representing the variable. | |
5643 | ||
5644 | If this macro is not defined, then the variable name is defined in the | |
5645 | usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}). | |
5646 | ||
5647 | You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or | |
5648 | @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro. | |
5649 | @end defmac | |
5650 | ||
8c38d114 | 5651 | @hook TARGET_ASM_DECLARE_CONSTANT_NAME |
202d6e5f | 5652 | |
5653 | @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name}) | |
5654 | A C statement (sans semicolon) to output to the stdio stream | |
5655 | @var{stream} any text necessary for claiming a register @var{regno} | |
5656 | for a global variable @var{decl} with name @var{name}. | |
5657 | ||
5658 | If you don't define this macro, that is equivalent to defining it to do | |
5659 | nothing. | |
5660 | @end defmac | |
5661 | ||
5662 | @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend}) | |
5663 | A C statement (sans semicolon) to finish up declaring a variable name | |
5664 | once the compiler has processed its initializer fully and thus has had a | |
5665 | chance to determine the size of an array when controlled by an | |
5666 | initializer. This is used on systems where it's necessary to declare | |
5667 | something about the size of the object. | |
5668 | ||
5669 | If you don't define this macro, that is equivalent to defining it to do | |
5670 | nothing. | |
5671 | ||
5672 | You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or | |
5673 | @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro. | |
5674 | @end defmac | |
5675 | ||
5676 | @hook TARGET_ASM_GLOBALIZE_LABEL | |
202d6e5f | 5677 | |
5678 | @hook TARGET_ASM_GLOBALIZE_DECL_NAME | |
202d6e5f | 5679 | |
7260826d | 5680 | @hook TARGET_ASM_ASSEMBLE_UNDEFINED_DECL |
5681 | ||
202d6e5f | 5682 | @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name}) |
5683 | A C statement (sans semicolon) to output to the stdio stream | |
5684 | @var{stream} some commands that will make the label @var{name} weak; | |
5685 | that is, available for reference from other files but only used if | |
5686 | no other definition is available. Use the expression | |
5687 | @code{assemble_name (@var{stream}, @var{name})} to output the name | |
5688 | itself; before and after that, output the additional assembler syntax | |
5689 | for making that name weak, and a newline. | |
5690 | ||
5691 | If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not | |
5692 | support weak symbols and you should not define the @code{SUPPORTS_WEAK} | |
5693 | macro. | |
5694 | @end defmac | |
5695 | ||
5696 | @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value}) | |
5697 | Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and | |
5698 | @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function | |
5699 | or variable decl. If @var{value} is not @code{NULL}, this C statement | |
5700 | should output to the stdio stream @var{stream} assembler code which | |
5701 | defines (equates) the weak symbol @var{name} to have the value | |
5702 | @var{value}. If @var{value} is @code{NULL}, it should output commands | |
5703 | to make @var{name} weak. | |
5704 | @end defmac | |
5705 | ||
5706 | @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value}) | |
5707 | Outputs a directive that enables @var{name} to be used to refer to | |
5708 | symbol @var{value} with weak-symbol semantics. @code{decl} is the | |
5709 | declaration of @code{name}. | |
5710 | @end defmac | |
5711 | ||
5712 | @defmac SUPPORTS_WEAK | |
293c8430 | 5713 | A preprocessor constant expression which evaluates to true if the target |
5714 | supports weak symbols. | |
202d6e5f | 5715 | |
5716 | If you don't define this macro, @file{defaults.h} provides a default | |
5717 | definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL} | |
293c8430 | 5718 | is defined, the default definition is @samp{1}; otherwise, it is @samp{0}. |
5719 | @end defmac | |
5720 | ||
5721 | @defmac TARGET_SUPPORTS_WEAK | |
5722 | A C expression which evaluates to true if the target supports weak symbols. | |
5723 | ||
5724 | If you don't define this macro, @file{defaults.h} provides a default | |
5725 | definition. The default definition is @samp{(SUPPORTS_WEAK)}. Define | |
5726 | this macro if you want to control weak symbol support with a compiler | |
5727 | flag such as @option{-melf}. | |
202d6e5f | 5728 | @end defmac |
5729 | ||
5730 | @defmac MAKE_DECL_ONE_ONLY (@var{decl}) | |
5731 | A C statement (sans semicolon) to mark @var{decl} to be emitted as a | |
5732 | public symbol such that extra copies in multiple translation units will | |
5733 | be discarded by the linker. Define this macro if your object file | |
5734 | format provides support for this concept, such as the @samp{COMDAT} | |
5735 | section flags in the Microsoft Windows PE/COFF format, and this support | |
5736 | requires changes to @var{decl}, such as putting it in a separate section. | |
5737 | @end defmac | |
5738 | ||
5739 | @defmac SUPPORTS_ONE_ONLY | |
5740 | A C expression which evaluates to true if the target supports one-only | |
5741 | semantics. | |
5742 | ||
5743 | If you don't define this macro, @file{varasm.c} provides a default | |
5744 | definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default | |
5745 | definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if | |
5746 | you want to control one-only symbol support with a compiler flag, or if | |
5747 | setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to | |
5748 | be emitted as one-only. | |
5749 | @end defmac | |
5750 | ||
5751 | @hook TARGET_ASM_ASSEMBLE_VISIBILITY | |
202d6e5f | 5752 | |
5753 | @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC | |
5754 | A C expression that evaluates to true if the target's linker expects | |
5755 | that weak symbols do not appear in a static archive's table of contents. | |
5756 | The default is @code{0}. | |
5757 | ||
5758 | Leaving weak symbols out of an archive's table of contents means that, | |
5759 | if a symbol will only have a definition in one translation unit and | |
5760 | will have undefined references from other translation units, that | |
5761 | symbol should not be weak. Defining this macro to be nonzero will | |
5762 | thus have the effect that certain symbols that would normally be weak | |
5763 | (explicit template instantiations, and vtables for polymorphic classes | |
5764 | with noninline key methods) will instead be nonweak. | |
5765 | ||
5766 | The C++ ABI requires this macro to be zero. Define this macro for | |
5767 | targets where full C++ ABI compliance is impossible and where linker | |
5768 | restrictions require weak symbols to be left out of a static archive's | |
5769 | table of contents. | |
5770 | @end defmac | |
5771 | ||
5772 | @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name}) | |
5773 | A C statement (sans semicolon) to output to the stdio stream | |
5774 | @var{stream} any text necessary for declaring the name of an external | |
5775 | symbol named @var{name} which is referenced in this compilation but | |
5776 | not defined. The value of @var{decl} is the tree node for the | |
5777 | declaration. | |
5778 | ||
5779 | This macro need not be defined if it does not need to output anything. | |
5780 | The GNU assembler and most Unix assemblers don't require anything. | |
5781 | @end defmac | |
5782 | ||
5783 | @hook TARGET_ASM_EXTERNAL_LIBCALL | |
202d6e5f | 5784 | |
5785 | @hook TARGET_ASM_MARK_DECL_PRESERVED | |
202d6e5f | 5786 | |
5787 | @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name}) | |
5788 | A C statement (sans semicolon) to output to the stdio stream | |
5789 | @var{stream} a reference in assembler syntax to a label named | |
5790 | @var{name}. This should add @samp{_} to the front of the name, if that | |
5791 | is customary on your operating system, as it is in most Berkeley Unix | |
5792 | systems. This macro is used in @code{assemble_name}. | |
5793 | @end defmac | |
5794 | ||
d86d364d | 5795 | @hook TARGET_MANGLE_ASSEMBLER_NAME |
5796 | ||
202d6e5f | 5797 | @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym}) |
5798 | A C statement (sans semicolon) to output a reference to | |
5799 | @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name} | |
5800 | will be used to output the name of the symbol. This macro may be used | |
5801 | to modify the way a symbol is referenced depending on information | |
5802 | encoded by @code{TARGET_ENCODE_SECTION_INFO}. | |
5803 | @end defmac | |
5804 | ||
5805 | @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf}) | |
5806 | A C statement (sans semicolon) to output a reference to @var{buf}, the | |
5807 | result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined, | |
5808 | @code{assemble_name} will be used to output the name of the symbol. | |
5809 | This macro is not used by @code{output_asm_label}, or the @code{%l} | |
5810 | specifier that calls it; the intention is that this macro should be set | |
5811 | when it is necessary to output a label differently when its address is | |
5812 | being taken. | |
5813 | @end defmac | |
5814 | ||
5815 | @hook TARGET_ASM_INTERNAL_LABEL | |
202d6e5f | 5816 | |
5817 | @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num}) | |
5818 | A C statement to output to the stdio stream @var{stream} a debug info | |
5819 | label whose name is made from the string @var{prefix} and the number | |
5820 | @var{num}. This is useful for VLIW targets, where debug info labels | |
5821 | may need to be treated differently than branch target labels. On some | |
5822 | systems, branch target labels must be at the beginning of instruction | |
5823 | bundles, but debug info labels can occur in the middle of instruction | |
5824 | bundles. | |
5825 | ||
5826 | If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be | |
5827 | used. | |
5828 | @end defmac | |
5829 | ||
5830 | @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num}) | |
5831 | A C statement to store into the string @var{string} a label whose name | |
5832 | is made from the string @var{prefix} and the number @var{num}. | |
5833 | ||
5834 | This string, when output subsequently by @code{assemble_name}, should | |
5835 | produce the output that @code{(*targetm.asm_out.internal_label)} would produce | |
5836 | with the same @var{prefix} and @var{num}. | |
5837 | ||
5838 | If the string begins with @samp{*}, then @code{assemble_name} will | |
5839 | output the rest of the string unchanged. It is often convenient for | |
5840 | @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the | |
5841 | string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets | |
5842 | to output the string, and may change it. (Of course, | |
5843 | @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so | |
5844 | you should know what it does on your machine.) | |
5845 | @end defmac | |
5846 | ||
5847 | @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number}) | |
5848 | A C expression to assign to @var{outvar} (which is a variable of type | |
5849 | @code{char *}) a newly allocated string made from the string | |
5850 | @var{name} and the number @var{number}, with some suitable punctuation | |
5851 | added. Use @code{alloca} to get space for the string. | |
5852 | ||
5853 | The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to | |
5854 | produce an assembler label for an internal static variable whose name is | |
5855 | @var{name}. Therefore, the string must be such as to result in valid | |
5856 | assembler code. The argument @var{number} is different each time this | |
5857 | macro is executed; it prevents conflicts between similarly-named | |
5858 | internal static variables in different scopes. | |
5859 | ||
5860 | Ideally this string should not be a valid C identifier, to prevent any | |
5861 | conflict with the user's own symbols. Most assemblers allow periods | |
5862 | or percent signs in assembler symbols; putting at least one of these | |
5863 | between the name and the number will suffice. | |
5864 | ||
5865 | If this macro is not defined, a default definition will be provided | |
5866 | which is correct for most systems. | |
5867 | @end defmac | |
5868 | ||
5869 | @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value}) | |
5870 | A C statement to output to the stdio stream @var{stream} assembler code | |
5871 | which defines (equates) the symbol @var{name} to have the value @var{value}. | |
5872 | ||
5873 | @findex SET_ASM_OP | |
5874 | If @code{SET_ASM_OP} is defined, a default definition is provided which is | |
5875 | correct for most systems. | |
5876 | @end defmac | |
5877 | ||
5878 | @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value}) | |
5879 | A C statement to output to the stdio stream @var{stream} assembler code | |
5880 | which defines (equates) the symbol whose tree node is @var{decl_of_name} | |
5881 | to have the value of the tree node @var{decl_of_value}. This macro will | |
5882 | be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if | |
5883 | the tree nodes are available. | |
5884 | ||
5885 | @findex SET_ASM_OP | |
5886 | If @code{SET_ASM_OP} is defined, a default definition is provided which is | |
5887 | correct for most systems. | |
5888 | @end defmac | |
5889 | ||
5890 | @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value}) | |
5891 | A C statement that evaluates to true if the assembler code which defines | |
5892 | (equates) the symbol whose tree node is @var{decl_of_name} to have the value | |
5893 | of the tree node @var{decl_of_value} should be emitted near the end of the | |
5894 | current compilation unit. The default is to not defer output of defines. | |
5895 | This macro affects defines output by @samp{ASM_OUTPUT_DEF} and | |
5896 | @samp{ASM_OUTPUT_DEF_FROM_DECLS}. | |
5897 | @end defmac | |
5898 | ||
5899 | @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value}) | |
5900 | A C statement to output to the stdio stream @var{stream} assembler code | |
5901 | which defines (equates) the weak symbol @var{name} to have the value | |
5902 | @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as | |
5903 | an undefined weak symbol. | |
5904 | ||
5905 | Define this macro if the target only supports weak aliases; define | |
5906 | @code{ASM_OUTPUT_DEF} instead if possible. | |
5907 | @end defmac | |
5908 | ||
5909 | @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name}) | |
5910 | Define this macro to override the default assembler names used for | |
5911 | Objective-C methods. | |
5912 | ||
5913 | The default name is a unique method number followed by the name of the | |
5914 | class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of | |
5915 | the category is also included in the assembler name (e.g.@: | |
5916 | @samp{_1_Foo_Bar}). | |
5917 | ||
5918 | These names are safe on most systems, but make debugging difficult since | |
5919 | the method's selector is not present in the name. Therefore, particular | |
5920 | systems define other ways of computing names. | |
5921 | ||
5922 | @var{buf} is an expression of type @code{char *} which gives you a | |
5923 | buffer in which to store the name; its length is as long as | |
5924 | @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus | |
5925 | 50 characters extra. | |
5926 | ||
5927 | The argument @var{is_inst} specifies whether the method is an instance | |
5928 | method or a class method; @var{class_name} is the name of the class; | |
5929 | @var{cat_name} is the name of the category (or @code{NULL} if the method is not | |
5930 | in a category); and @var{sel_name} is the name of the selector. | |
5931 | ||
5932 | On systems where the assembler can handle quoted names, you can use this | |
5933 | macro to provide more human-readable names. | |
5934 | @end defmac | |
5935 | ||
202d6e5f | 5936 | @node Initialization |
5937 | @subsection How Initialization Functions Are Handled | |
5938 | @cindex initialization routines | |
5939 | @cindex termination routines | |
5940 | @cindex constructors, output of | |
5941 | @cindex destructors, output of | |
5942 | ||
5943 | The compiled code for certain languages includes @dfn{constructors} | |
5944 | (also called @dfn{initialization routines})---functions to initialize | |
5945 | data in the program when the program is started. These functions need | |
5946 | to be called before the program is ``started''---that is to say, before | |
5947 | @code{main} is called. | |
5948 | ||
5949 | Compiling some languages generates @dfn{destructors} (also called | |
5950 | @dfn{termination routines}) that should be called when the program | |
5951 | terminates. | |
5952 | ||
5953 | To make the initialization and termination functions work, the compiler | |
5954 | must output something in the assembler code to cause those functions to | |
5955 | be called at the appropriate time. When you port the compiler to a new | |
5956 | system, you need to specify how to do this. | |
5957 | ||
5958 | There are two major ways that GCC currently supports the execution of | |
5959 | initialization and termination functions. Each way has two variants. | |
5960 | Much of the structure is common to all four variations. | |
5961 | ||
5962 | @findex __CTOR_LIST__ | |
5963 | @findex __DTOR_LIST__ | |
5964 | The linker must build two lists of these functions---a list of | |
5965 | initialization functions, called @code{__CTOR_LIST__}, and a list of | |
5966 | termination functions, called @code{__DTOR_LIST__}. | |
5967 | ||
5968 | Each list always begins with an ignored function pointer (which may hold | |
5969 | 0, @minus{}1, or a count of the function pointers after it, depending on | |
5970 | the environment). This is followed by a series of zero or more function | |
5971 | pointers to constructors (or destructors), followed by a function | |
5972 | pointer containing zero. | |
5973 | ||
5974 | Depending on the operating system and its executable file format, either | |
5975 | @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup | |
5976 | time and exit time. Constructors are called in reverse order of the | |
5977 | list; destructors in forward order. | |
5978 | ||
5979 | The best way to handle static constructors works only for object file | |
5980 | formats which provide arbitrarily-named sections. A section is set | |
5981 | aside for a list of constructors, and another for a list of destructors. | |
5982 | Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each | |
5983 | object file that defines an initialization function also puts a word in | |
5984 | the constructor section to point to that function. The linker | |
5985 | accumulates all these words into one contiguous @samp{.ctors} section. | |
5986 | Termination functions are handled similarly. | |
5987 | ||
5988 | This method will be chosen as the default by @file{target-def.h} if | |
5989 | @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not | |
5990 | support arbitrary sections, but does support special designated | |
5991 | constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP} | |
5992 | and @code{DTORS_SECTION_ASM_OP} to achieve the same effect. | |
5993 | ||
5994 | When arbitrary sections are available, there are two variants, depending | |
5995 | upon how the code in @file{crtstuff.c} is called. On systems that | |
5996 | support a @dfn{.init} section which is executed at program startup, | |
5997 | parts of @file{crtstuff.c} are compiled into that section. The | |
5998 | program is linked by the @command{gcc} driver like this: | |
5999 | ||
6000 | @smallexample | |
6001 | ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o | |
6002 | @end smallexample | |
6003 | ||
6004 | The prologue of a function (@code{__init}) appears in the @code{.init} | |
6005 | section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise | |
6006 | for the function @code{__fini} in the @dfn{.fini} section. Normally these | |
6007 | files are provided by the operating system or by the GNU C library, but | |
6008 | are provided by GCC for a few targets. | |
6009 | ||
6010 | The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets) | |
6011 | compiled from @file{crtstuff.c}. They contain, among other things, code | |
6012 | fragments within the @code{.init} and @code{.fini} sections that branch | |
6013 | to routines in the @code{.text} section. The linker will pull all parts | |
6014 | of a section together, which results in a complete @code{__init} function | |
6015 | that invokes the routines we need at startup. | |
6016 | ||
6017 | To use this variant, you must define the @code{INIT_SECTION_ASM_OP} | |
6018 | macro properly. | |
6019 | ||
6020 | If no init section is available, when GCC compiles any function called | |
6021 | @code{main} (or more accurately, any function designated as a program | |
6022 | entry point by the language front end calling @code{expand_main_function}), | |
6023 | it inserts a procedure call to @code{__main} as the first executable code | |
6024 | after the function prologue. The @code{__main} function is defined | |
6025 | in @file{libgcc2.c} and runs the global constructors. | |
6026 | ||
6027 | In file formats that don't support arbitrary sections, there are again | |
6028 | two variants. In the simplest variant, the GNU linker (GNU @code{ld}) | |
6029 | and an `a.out' format must be used. In this case, | |
6030 | @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs} | |
6031 | entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__}, | |
6032 | and with the address of the void function containing the initialization | |
6033 | code as its value. The GNU linker recognizes this as a request to add | |
6034 | the value to a @dfn{set}; the values are accumulated, and are eventually | |
6035 | placed in the executable as a vector in the format described above, with | |
6036 | a leading (ignored) count and a trailing zero element. | |
6037 | @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init | |
6038 | section is available, the absence of @code{INIT_SECTION_ASM_OP} causes | |
6039 | the compilation of @code{main} to call @code{__main} as above, starting | |
6040 | the initialization process. | |
6041 | ||
6042 | The last variant uses neither arbitrary sections nor the GNU linker. | |
6043 | This is preferable when you want to do dynamic linking and when using | |
6044 | file formats which the GNU linker does not support, such as `ECOFF'@. In | |
6045 | this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and | |
6046 | termination functions are recognized simply by their names. This requires | |
6047 | an extra program in the linkage step, called @command{collect2}. This program | |
6048 | pretends to be the linker, for use with GCC; it does its job by running | |
6049 | the ordinary linker, but also arranges to include the vectors of | |
6050 | initialization and termination functions. These functions are called | |
6051 | via @code{__main} as described above. In order to use this method, | |
6052 | @code{use_collect2} must be defined in the target in @file{config.gcc}. | |
6053 | ||
6054 | @ifinfo | |
6055 | The following section describes the specific macros that control and | |
6056 | customize the handling of initialization and termination functions. | |
6057 | @end ifinfo | |
6058 | ||
6059 | @node Macros for Initialization | |
6060 | @subsection Macros Controlling Initialization Routines | |
6061 | ||
6062 | Here are the macros that control how the compiler handles initialization | |
6063 | and termination functions: | |
6064 | ||
6065 | @defmac INIT_SECTION_ASM_OP | |
6066 | If defined, a C string constant, including spacing, for the assembler | |
6067 | operation to identify the following data as initialization code. If not | |
6068 | defined, GCC will assume such a section does not exist. When you are | |
6069 | using special sections for initialization and termination functions, this | |
6070 | macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to | |
6071 | run the initialization functions. | |
6072 | @end defmac | |
6073 | ||
6074 | @defmac HAS_INIT_SECTION | |
6075 | If defined, @code{main} will not call @code{__main} as described above. | |
6076 | This macro should be defined for systems that control start-up code | |
6077 | on a symbol-by-symbol basis, such as OSF/1, and should not | |
6078 | be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}. | |
6079 | @end defmac | |
6080 | ||
6081 | @defmac LD_INIT_SWITCH | |
6082 | If defined, a C string constant for a switch that tells the linker that | |
6083 | the following symbol is an initialization routine. | |
6084 | @end defmac | |
6085 | ||
6086 | @defmac LD_FINI_SWITCH | |
6087 | If defined, a C string constant for a switch that tells the linker that | |
6088 | the following symbol is a finalization routine. | |
6089 | @end defmac | |
6090 | ||
6091 | @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func}) | |
6092 | If defined, a C statement that will write a function that can be | |
6093 | automatically called when a shared library is loaded. The function | |
6094 | should call @var{func}, which takes no arguments. If not defined, and | |
6095 | the object format requires an explicit initialization function, then a | |
6096 | function called @code{_GLOBAL__DI} will be generated. | |
6097 | ||
6098 | This function and the following one are used by collect2 when linking a | |
6099 | shared library that needs constructors or destructors, or has DWARF2 | |
6100 | exception tables embedded in the code. | |
6101 | @end defmac | |
6102 | ||
6103 | @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func}) | |
6104 | If defined, a C statement that will write a function that can be | |
6105 | automatically called when a shared library is unloaded. The function | |
6106 | should call @var{func}, which takes no arguments. If not defined, and | |
6107 | the object format requires an explicit finalization function, then a | |
6108 | function called @code{_GLOBAL__DD} will be generated. | |
6109 | @end defmac | |
6110 | ||
6111 | @defmac INVOKE__main | |
6112 | If defined, @code{main} will call @code{__main} despite the presence of | |
6113 | @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems | |
6114 | where the init section is not actually run automatically, but is still | |
6115 | useful for collecting the lists of constructors and destructors. | |
6116 | @end defmac | |
6117 | ||
6118 | @defmac SUPPORTS_INIT_PRIORITY | |
6119 | If nonzero, the C++ @code{init_priority} attribute is supported and the | |
6120 | compiler should emit instructions to control the order of initialization | |
6121 | of objects. If zero, the compiler will issue an error message upon | |
6122 | encountering an @code{init_priority} attribute. | |
6123 | @end defmac | |
6124 | ||
6125 | @hook TARGET_HAVE_CTORS_DTORS | |
202d6e5f | 6126 | |
6127 | @hook TARGET_ASM_CONSTRUCTOR | |
202d6e5f | 6128 | |
6129 | @hook TARGET_ASM_DESTRUCTOR | |
202d6e5f | 6130 | |
6131 | If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine | |
6132 | generated for the generated object file will have static linkage. | |
6133 | ||
6134 | If your system uses @command{collect2} as the means of processing | |
6135 | constructors, then that program normally uses @command{nm} to scan | |
6136 | an object file for constructor functions to be called. | |
6137 | ||
6138 | On certain kinds of systems, you can define this macro to make | |
6139 | @command{collect2} work faster (and, in some cases, make it work at all): | |
6140 | ||
6141 | @defmac OBJECT_FORMAT_COFF | |
6142 | Define this macro if the system uses COFF (Common Object File Format) | |
6143 | object files, so that @command{collect2} can assume this format and scan | |
6144 | object files directly for dynamic constructor/destructor functions. | |
6145 | ||
6146 | This macro is effective only in a native compiler; @command{collect2} as | |
6147 | part of a cross compiler always uses @command{nm} for the target machine. | |
6148 | @end defmac | |
6149 | ||
6150 | @defmac REAL_NM_FILE_NAME | |
6151 | Define this macro as a C string constant containing the file name to use | |
6152 | to execute @command{nm}. The default is to search the path normally for | |
6153 | @command{nm}. | |
f369130e | 6154 | @end defmac |
6155 | ||
6156 | @defmac NM_FLAGS | |
6157 | @command{collect2} calls @command{nm} to scan object files for static | |
6158 | constructors and destructors and LTO info. By default, @option{-n} is | |
6159 | passed. Define @code{NM_FLAGS} to a C string constant if other options | |
b59688ee | 6160 | are needed to get the same output format as GNU @command{nm -n} |
f369130e | 6161 | produces. |
6162 | @end defmac | |
202d6e5f | 6163 | |
6164 | If your system supports shared libraries and has a program to list the | |
6165 | dynamic dependencies of a given library or executable, you can define | |
6166 | these macros to enable support for running initialization and | |
6167 | termination functions in shared libraries: | |
202d6e5f | 6168 | |
6169 | @defmac LDD_SUFFIX | |
6170 | Define this macro to a C string constant containing the name of the program | |
f369130e | 6171 | which lists dynamic dependencies, like @command{ldd} under SunOS 4. |
202d6e5f | 6172 | @end defmac |
6173 | ||
6174 | @defmac PARSE_LDD_OUTPUT (@var{ptr}) | |
6175 | Define this macro to be C code that extracts filenames from the output | |
6176 | of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable | |
6177 | of type @code{char *} that points to the beginning of a line of output | |
6178 | from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the | |
6179 | code must advance @var{ptr} to the beginning of the filename on that | |
6180 | line. Otherwise, it must set @var{ptr} to @code{NULL}. | |
6181 | @end defmac | |
6182 | ||
6183 | @defmac SHLIB_SUFFIX | |
6184 | Define this macro to a C string constant containing the default shared | |
6185 | library extension of the target (e.g., @samp{".so"}). @command{collect2} | |
6186 | strips version information after this suffix when generating global | |
6187 | constructor and destructor names. This define is only needed on targets | |
6188 | that use @command{collect2} to process constructors and destructors. | |
6189 | @end defmac | |
6190 | ||
6191 | @node Instruction Output | |
6192 | @subsection Output of Assembler Instructions | |
6193 | ||
6194 | @c prevent bad page break with this line | |
6195 | This describes assembler instruction output. | |
6196 | ||
6197 | @defmac REGISTER_NAMES | |
6198 | A C initializer containing the assembler's names for the machine | |
6199 | registers, each one as a C string constant. This is what translates | |
6200 | register numbers in the compiler into assembler language. | |
6201 | @end defmac | |
6202 | ||
6203 | @defmac ADDITIONAL_REGISTER_NAMES | |
6204 | If defined, a C initializer for an array of structures containing a name | |
6205 | and a register number. This macro defines additional names for hard | |
6206 | registers, thus allowing the @code{asm} option in declarations to refer | |
6207 | to registers using alternate names. | |
6208 | @end defmac | |
6209 | ||
ea26afd4 | 6210 | @defmac OVERLAPPING_REGISTER_NAMES |
6211 | If defined, a C initializer for an array of structures containing a | |
6212 | name, a register number and a count of the number of consecutive | |
6213 | machine registers the name overlaps. This macro defines additional | |
6214 | names for hard registers, thus allowing the @code{asm} option in | |
6215 | declarations to refer to registers using alternate names. Unlike | |
6216 | @code{ADDITIONAL_REGISTER_NAMES}, this macro should be used when the | |
6217 | register name implies multiple underlying registers. | |
6218 | ||
6219 | This macro should be used when it is important that a clobber in an | |
6220 | @code{asm} statement clobbers all the underlying values implied by the | |
6221 | register name. For example, on ARM, clobbering the double-precision | |
6222 | VFP register ``d0'' implies clobbering both single-precision registers | |
6223 | ``s0'' and ``s1''. | |
6224 | @end defmac | |
6225 | ||
202d6e5f | 6226 | @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr}) |
6227 | Define this macro if you are using an unusual assembler that | |
6228 | requires different names for the machine instructions. | |
6229 | ||
6230 | The definition is a C statement or statements which output an | |
6231 | assembler instruction opcode to the stdio stream @var{stream}. The | |
6232 | macro-operand @var{ptr} is a variable of type @code{char *} which | |
6233 | points to the opcode name in its ``internal'' form---the form that is | |
6234 | written in the machine description. The definition should output the | |
6235 | opcode name to @var{stream}, performing any translation you desire, and | |
6236 | increment the variable @var{ptr} to point at the end of the opcode | |
6237 | so that it will not be output twice. | |
6238 | ||
6239 | In fact, your macro definition may process less than the entire opcode | |
6240 | name, or more than the opcode name; but if you want to process text | |
6241 | that includes @samp{%}-sequences to substitute operands, you must take | |
6242 | care of the substitution yourself. Just be sure to increment | |
6243 | @var{ptr} over whatever text should not be output normally. | |
6244 | ||
6245 | @findex recog_data.operand | |
6246 | If you need to look at the operand values, they can be found as the | |
6247 | elements of @code{recog_data.operand}. | |
6248 | ||
6249 | If the macro definition does nothing, the instruction is output | |
6250 | in the usual way. | |
6251 | @end defmac | |
6252 | ||
6253 | @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands}) | |
6254 | If defined, a C statement to be executed just prior to the output of | |
6255 | assembler code for @var{insn}, to modify the extracted operands so | |
6256 | they will be output differently. | |
6257 | ||
6258 | Here the argument @var{opvec} is the vector containing the operands | |
6259 | extracted from @var{insn}, and @var{noperands} is the number of | |
6260 | elements of the vector which contain meaningful data for this insn. | |
6261 | The contents of this vector are what will be used to convert the insn | |
6262 | template into assembler code, so you can change the assembler output | |
6263 | by changing the contents of the vector. | |
6264 | ||
6265 | This macro is useful when various assembler syntaxes share a single | |
6266 | file of instruction patterns; by defining this macro differently, you | |
6267 | can cause a large class of instructions to be output differently (such | |
6268 | as with rearranged operands). Naturally, variations in assembler | |
6269 | syntax affecting individual insn patterns ought to be handled by | |
6270 | writing conditional output routines in those patterns. | |
6271 | ||
6272 | If this macro is not defined, it is equivalent to a null statement. | |
6273 | @end defmac | |
6274 | ||
6275 | @hook TARGET_ASM_FINAL_POSTSCAN_INSN | |
202d6e5f | 6276 | |
6277 | @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code}) | |
6278 | A C compound statement to output to stdio stream @var{stream} the | |
6279 | assembler syntax for an instruction operand @var{x}. @var{x} is an | |
6280 | RTL expression. | |
6281 | ||
6282 | @var{code} is a value that can be used to specify one of several ways | |
6283 | of printing the operand. It is used when identical operands must be | |
6284 | printed differently depending on the context. @var{code} comes from | |
6285 | the @samp{%} specification that was used to request printing of the | |
6286 | operand. If the specification was just @samp{%@var{digit}} then | |
6287 | @var{code} is 0; if the specification was @samp{%@var{ltr} | |
6288 | @var{digit}} then @var{code} is the ASCII code for @var{ltr}. | |
6289 | ||
6290 | @findex reg_names | |
6291 | If @var{x} is a register, this macro should print the register's name. | |
6292 | The names can be found in an array @code{reg_names} whose type is | |
6293 | @code{char *[]}. @code{reg_names} is initialized from | |
6294 | @code{REGISTER_NAMES}. | |
6295 | ||
6296 | When the machine description has a specification @samp{%@var{punct}} | |
6297 | (a @samp{%} followed by a punctuation character), this macro is called | |
6298 | with a null pointer for @var{x} and the punctuation character for | |
6299 | @var{code}. | |
6300 | @end defmac | |
6301 | ||
6302 | @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code}) | |
6303 | A C expression which evaluates to true if @var{code} is a valid | |
6304 | punctuation character for use in the @code{PRINT_OPERAND} macro. If | |
6305 | @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no | |
6306 | punctuation characters (except for the standard one, @samp{%}) are used | |
6307 | in this way. | |
6308 | @end defmac | |
6309 | ||
6310 | @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x}) | |
6311 | A C compound statement to output to stdio stream @var{stream} the | |
6312 | assembler syntax for an instruction operand that is a memory reference | |
6313 | whose address is @var{x}. @var{x} is an RTL expression. | |
6314 | ||
6315 | @cindex @code{TARGET_ENCODE_SECTION_INFO} usage | |
6316 | On some machines, the syntax for a symbolic address depends on the | |
6317 | section that the address refers to. On these machines, define the hook | |
6318 | @code{TARGET_ENCODE_SECTION_INFO} to store the information into the | |
6319 | @code{symbol_ref}, and then check for it here. @xref{Assembler | |
6320 | Format}. | |
6321 | @end defmac | |
6322 | ||
6323 | @findex dbr_sequence_length | |
6324 | @defmac DBR_OUTPUT_SEQEND (@var{file}) | |
6325 | A C statement, to be executed after all slot-filler instructions have | |
6326 | been output. If necessary, call @code{dbr_sequence_length} to | |
6327 | determine the number of slots filled in a sequence (zero if not | |
6328 | currently outputting a sequence), to decide how many no-ops to output, | |
6329 | or whatever. | |
6330 | ||
6331 | Don't define this macro if it has nothing to do, but it is helpful in | |
6332 | reading assembly output if the extent of the delay sequence is made | |
6333 | explicit (e.g.@: with white space). | |
6334 | @end defmac | |
6335 | ||
6336 | @findex final_sequence | |
6337 | Note that output routines for instructions with delay slots must be | |
6338 | prepared to deal with not being output as part of a sequence | |
6339 | (i.e.@: when the scheduling pass is not run, or when no slot fillers could be | |
6340 | found.) The variable @code{final_sequence} is null when not | |
6341 | processing a sequence, otherwise it contains the @code{sequence} rtx | |
6342 | being output. | |
6343 | ||
6344 | @findex asm_fprintf | |
6345 | @defmac REGISTER_PREFIX | |
6346 | @defmacx LOCAL_LABEL_PREFIX | |
6347 | @defmacx USER_LABEL_PREFIX | |
6348 | @defmacx IMMEDIATE_PREFIX | |
6349 | If defined, C string expressions to be used for the @samp{%R}, @samp{%L}, | |
6350 | @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see | |
6351 | @file{final.c}). These are useful when a single @file{md} file must | |
6352 | support multiple assembler formats. In that case, the various @file{tm.h} | |
6353 | files can define these macros differently. | |
6354 | @end defmac | |
6355 | ||
6356 | @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format}) | |
6357 | If defined this macro should expand to a series of @code{case} | |
6358 | statements which will be parsed inside the @code{switch} statement of | |
6359 | the @code{asm_fprintf} function. This allows targets to define extra | |
6360 | printf formats which may useful when generating their assembler | |
6361 | statements. Note that uppercase letters are reserved for future | |
6362 | generic extensions to asm_fprintf, and so are not available to target | |
6363 | specific code. The output file is given by the parameter @var{file}. | |
6364 | The varargs input pointer is @var{argptr} and the rest of the format | |
6365 | string, starting the character after the one that is being switched | |
6366 | upon, is pointed to by @var{format}. | |
6367 | @end defmac | |
6368 | ||
6369 | @defmac ASSEMBLER_DIALECT | |
6370 | If your target supports multiple dialects of assembler language (such as | |
6371 | different opcodes), define this macro as a C expression that gives the | |
6372 | numeric index of the assembler language dialect to use, with zero as the | |
6373 | first variant. | |
6374 | ||
6375 | If this macro is defined, you may use constructs of the form | |
6376 | @smallexample | |
6377 | @samp{@{option0|option1|option2@dots{}@}} | |
6378 | @end smallexample | |
6379 | @noindent | |
6380 | in the output templates of patterns (@pxref{Output Template}) or in the | |
6381 | first argument of @code{asm_fprintf}. This construct outputs | |
6382 | @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of | |
6383 | @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters | |
6384 | within these strings retain their usual meaning. If there are fewer | |
6385 | alternatives within the braces than the value of | |
82353beb | 6386 | @code{ASSEMBLER_DIALECT}, the construct outputs nothing. If it's needed |
6387 | to print curly braces or @samp{|} character in assembler output directly, | |
6388 | @samp{%@{}, @samp{%@}} and @samp{%|} can be used. | |
202d6e5f | 6389 | |
6390 | If you do not define this macro, the characters @samp{@{}, @samp{|} and | |
6391 | @samp{@}} do not have any special meaning when used in templates or | |
6392 | operands to @code{asm_fprintf}. | |
6393 | ||
6394 | Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX}, | |
6395 | @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express | |
6396 | the variations in assembler language syntax with that mechanism. Define | |
6397 | @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax | |
6398 | if the syntax variant are larger and involve such things as different | |
6399 | opcodes or operand order. | |
6400 | @end defmac | |
6401 | ||
6402 | @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno}) | |
6403 | A C expression to output to @var{stream} some assembler code | |
6404 | which will push hard register number @var{regno} onto the stack. | |
6405 | The code need not be optimal, since this macro is used only when | |
6406 | profiling. | |
6407 | @end defmac | |
6408 | ||
6409 | @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno}) | |
6410 | A C expression to output to @var{stream} some assembler code | |
6411 | which will pop hard register number @var{regno} off of the stack. | |
6412 | The code need not be optimal, since this macro is used only when | |
6413 | profiling. | |
6414 | @end defmac | |
6415 | ||
6416 | @node Dispatch Tables | |
6417 | @subsection Output of Dispatch Tables | |
6418 | ||
6419 | @c prevent bad page break with this line | |
6420 | This concerns dispatch tables. | |
6421 | ||
6422 | @cindex dispatch table | |
6423 | @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel}) | |
6424 | A C statement to output to the stdio stream @var{stream} an assembler | |
6425 | pseudo-instruction to generate a difference between two labels. | |
6426 | @var{value} and @var{rel} are the numbers of two internal labels. The | |
6427 | definitions of these labels are output using | |
6428 | @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same | |
6429 | way here. For example, | |
6430 | ||
6431 | @smallexample | |
6432 | fprintf (@var{stream}, "\t.word L%d-L%d\n", | |
6433 | @var{value}, @var{rel}) | |
6434 | @end smallexample | |
6435 | ||
6436 | You must provide this macro on machines where the addresses in a | |
6437 | dispatch table are relative to the table's own address. If defined, GCC | |
6438 | will also use this macro on all machines when producing PIC@. | |
6439 | @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the | |
6440 | mode and flags can be read. | |
6441 | @end defmac | |
6442 | ||
6443 | @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value}) | |
6444 | This macro should be provided on machines where the addresses | |
6445 | in a dispatch table are absolute. | |
6446 | ||
6447 | The definition should be a C statement to output to the stdio stream | |
6448 | @var{stream} an assembler pseudo-instruction to generate a reference to | |
6449 | a label. @var{value} is the number of an internal label whose | |
6450 | definition is output using @code{(*targetm.asm_out.internal_label)}. | |
6451 | For example, | |
6452 | ||
6453 | @smallexample | |
6454 | fprintf (@var{stream}, "\t.word L%d\n", @var{value}) | |
6455 | @end smallexample | |
6456 | @end defmac | |
6457 | ||
6458 | @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table}) | |
6459 | Define this if the label before a jump-table needs to be output | |
6460 | specially. The first three arguments are the same as for | |
6461 | @code{(*targetm.asm_out.internal_label)}; the fourth argument is the | |
9ddbb404 | 6462 | jump-table which follows (a @code{jump_table_data} containing an |
202d6e5f | 6463 | @code{addr_vec} or @code{addr_diff_vec}). |
6464 | ||
6465 | This feature is used on system V to output a @code{swbeg} statement | |
6466 | for the table. | |
6467 | ||
6468 | If this macro is not defined, these labels are output with | |
6469 | @code{(*targetm.asm_out.internal_label)}. | |
6470 | @end defmac | |
6471 | ||
6472 | @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table}) | |
6473 | Define this if something special must be output at the end of a | |
6474 | jump-table. The definition should be a C statement to be executed | |
6475 | after the assembler code for the table is written. It should write | |
6476 | the appropriate code to stdio stream @var{stream}. The argument | |
6477 | @var{table} is the jump-table insn, and @var{num} is the label-number | |
6478 | of the preceding label. | |
6479 | ||
6480 | If this macro is not defined, nothing special is output at the end of | |
6481 | the jump-table. | |
6482 | @end defmac | |
6483 | ||
6484 | @hook TARGET_ASM_EMIT_UNWIND_LABEL | |
202d6e5f | 6485 | |
6486 | @hook TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL | |
202d6e5f | 6487 | |
37966699 | 6488 | @hook TARGET_ASM_EMIT_EXCEPT_PERSONALITY |
6489 | ||
202d6e5f | 6490 | @hook TARGET_ASM_UNWIND_EMIT |
202d6e5f | 6491 | |
f3e8e368 | 6492 | @hook TARGET_ASM_UNWIND_EMIT_BEFORE_INSN |
6493 | ||
202d6e5f | 6494 | @node Exception Region Output |
6495 | @subsection Assembler Commands for Exception Regions | |
6496 | ||
6497 | @c prevent bad page break with this line | |
6498 | ||
6499 | This describes commands marking the start and the end of an exception | |
6500 | region. | |
6501 | ||
6502 | @defmac EH_FRAME_SECTION_NAME | |
6503 | If defined, a C string constant for the name of the section containing | |
6504 | exception handling frame unwind information. If not defined, GCC will | |
6505 | provide a default definition if the target supports named sections. | |
6506 | @file{crtstuff.c} uses this macro to switch to the appropriate section. | |
6507 | ||
6508 | You should define this symbol if your target supports DWARF 2 frame | |
6509 | unwind information and the default definition does not work. | |
6510 | @end defmac | |
6511 | ||
552a60d3 | 6512 | @defmac EH_FRAME_THROUGH_COLLECT2 |
6513 | If defined, DWARF 2 frame unwind information will identified by | |
6514 | specially named labels. The collect2 process will locate these | |
6515 | labels and generate code to register the frames. | |
202d6e5f | 6516 | |
552a60d3 | 6517 | This might be necessary, for instance, if the system linker will not |
6518 | place the eh_frames in-between the sentinals from @file{crtstuff.c}, | |
6519 | or if the system linker does garbage collection and sections cannot | |
6520 | be marked as not to be collected. | |
202d6e5f | 6521 | @end defmac |
6522 | ||
6523 | @defmac EH_TABLES_CAN_BE_READ_ONLY | |
6524 | Define this macro to 1 if your target is such that no frame unwind | |
6525 | information encoding used with non-PIC code will ever require a | |
6526 | runtime relocation, but the linker may not support merging read-only | |
6527 | and read-write sections into a single read-write section. | |
6528 | @end defmac | |
6529 | ||
6530 | @defmac MASK_RETURN_ADDR | |
6531 | An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so | |
6532 | that it does not contain any extraneous set bits in it. | |
6533 | @end defmac | |
6534 | ||
6535 | @defmac DWARF2_UNWIND_INFO | |
6536 | Define this macro to 0 if your target supports DWARF 2 frame unwind | |
6537 | information, but it does not yet work with exception handling. | |
6538 | Otherwise, if your target supports this information (if it defines | |
d8555e79 | 6539 | @code{INCOMING_RETURN_ADDR_RTX} and @code{OBJECT_FORMAT_ELF}), |
6540 | GCC will provide a default definition of 1. | |
cc7d6aed | 6541 | @end defmac |
202d6e5f | 6542 | |
cc7d6aed | 6543 | @hook TARGET_EXCEPT_UNWIND_INFO |
6544 | This hook defines the mechanism that will be used for exception handling | |
6545 | by the target. If the target has ABI specified unwind tables, the hook | |
6546 | should return @code{UI_TARGET}. If the target is to use the | |
6547 | @code{setjmp}/@code{longjmp}-based exception handling scheme, the hook | |
6548 | should return @code{UI_SJLJ}. If the target supports DWARF 2 frame unwind | |
6549 | information, the hook should return @code{UI_DWARF2}. | |
202d6e5f | 6550 | |
cc7d6aed | 6551 | A target may, if exceptions are disabled, choose to return @code{UI_NONE}. |
6552 | This may end up simplifying other parts of target-specific code. The | |
6553 | default implementation of this hook never returns @code{UI_NONE}. | |
202d6e5f | 6554 | |
cc7d6aed | 6555 | Note that the value returned by this hook should be constant. It should |
b213bf24 | 6556 | not depend on anything except the command-line switches described by |
6557 | @var{opts}. In particular, the | |
cc7d6aed | 6558 | setting @code{UI_SJLJ} must be fixed at compiler start-up as C pre-processor |
6559 | macros and builtin functions related to exception handling are set up | |
6560 | depending on this setting. | |
6561 | ||
6562 | The default implementation of the hook first honors the | |
6563 | @option{--enable-sjlj-exceptions} configure option, then | |
b213bf24 | 6564 | @code{DWARF2_UNWIND_INFO}, and finally defaults to @code{UI_SJLJ}. If |
6565 | @code{DWARF2_UNWIND_INFO} depends on command-line options, the target | |
6566 | must define this hook so that @var{opts} is used correctly. | |
cc7d6aed | 6567 | @end deftypefn |
202d6e5f | 6568 | |
6569 | @hook TARGET_UNWIND_TABLES_DEFAULT | |
6570 | This variable should be set to @code{true} if the target ABI requires unwinding | |
b213bf24 | 6571 | tables even when exceptions are not used. It must not be modified by |
6572 | command-line option processing. | |
202d6e5f | 6573 | @end deftypevr |
6574 | ||
202d6e5f | 6575 | @defmac DONT_USE_BUILTIN_SETJMP |
6576 | Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme | |
6577 | should use the @code{setjmp}/@code{longjmp} functions from the C library | |
6578 | instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery. | |
6579 | @end defmac | |
6580 | ||
91bb968b | 6581 | @defmac JMP_BUF_SIZE |
6582 | This macro has no effect unless @code{DONT_USE_BUILTIN_SETJMP} is also | |
6583 | defined. Define this macro if the default size of @code{jmp_buf} buffer | |
6584 | for the @code{setjmp}/@code{longjmp}-based exception handling mechanism | |
6585 | is not large enough, or if it is much too large. | |
6586 | The default size is @code{FIRST_PSEUDO_REGISTER * sizeof(void *)}. | |
6587 | @end defmac | |
6588 | ||
202d6e5f | 6589 | @defmac DWARF_CIE_DATA_ALIGNMENT |
6590 | This macro need only be defined if the target might save registers in the | |
6591 | function prologue at an offset to the stack pointer that is not aligned to | |
6592 | @code{UNITS_PER_WORD}. The definition should be the negative minimum | |
2b785411 | 6593 | alignment if @code{STACK_GROWS_DOWNWARD} is true, and the positive |
202d6e5f | 6594 | minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if |
6595 | the target supports DWARF 2 frame unwind information. | |
6596 | @end defmac | |
6597 | ||
6598 | @hook TARGET_TERMINATE_DW2_EH_FRAME_INFO | |
202d6e5f | 6599 | |
6600 | @hook TARGET_DWARF_REGISTER_SPAN | |
202d6e5f | 6601 | |
d626297e | 6602 | @hook TARGET_DWARF_FRAME_REG_MODE |
6603 | ||
202d6e5f | 6604 | @hook TARGET_INIT_DWARF_REG_SIZES_EXTRA |
202d6e5f | 6605 | |
6606 | @hook TARGET_ASM_TTYPE | |
202d6e5f | 6607 | |
6608 | @hook TARGET_ARM_EABI_UNWINDER | |
202d6e5f | 6609 | |
6610 | @node Alignment Output | |
6611 | @subsection Assembler Commands for Alignment | |
6612 | ||
6613 | @c prevent bad page break with this line | |
6614 | This describes commands for alignment. | |
6615 | ||
6616 | @defmac JUMP_ALIGN (@var{label}) | |
6617 | The alignment (log base 2) to put in front of @var{label}, which is | |
6618 | a common destination of jumps and has no fallthru incoming edge. | |
6619 | ||
6620 | This macro need not be defined if you don't want any special alignment | |
6621 | to be done at such a time. Most machine descriptions do not currently | |
6622 | define the macro. | |
6623 | ||
6624 | Unless it's necessary to inspect the @var{label} parameter, it is better | |
6625 | to set the variable @var{align_jumps} in the target's | |
6626 | @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's | |
6627 | selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation. | |
6628 | @end defmac | |
6629 | ||
ae2b9f1f | 6630 | @hook TARGET_ASM_JUMP_ALIGN_MAX_SKIP |
ae2b9f1f | 6631 | |
202d6e5f | 6632 | @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label}) |
6633 | The alignment (log base 2) to put in front of @var{label}, which follows | |
6634 | a @code{BARRIER}. | |
6635 | ||
6636 | This macro need not be defined if you don't want any special alignment | |
6637 | to be done at such a time. Most machine descriptions do not currently | |
6638 | define the macro. | |
6639 | @end defmac | |
6640 | ||
ae2b9f1f | 6641 | @hook TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP |
202d6e5f | 6642 | |
6643 | @defmac LOOP_ALIGN (@var{label}) | |
720b223d | 6644 | The alignment (log base 2) to put in front of @var{label} that heads |
6645 | a frequently executed basic block (usually the header of a loop). | |
202d6e5f | 6646 | |
6647 | This macro need not be defined if you don't want any special alignment | |
6648 | to be done at such a time. Most machine descriptions do not currently | |
6649 | define the macro. | |
6650 | ||
6651 | Unless it's necessary to inspect the @var{label} parameter, it is better | |
6652 | to set the variable @code{align_loops} in the target's | |
6653 | @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's | |
6654 | selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation. | |
6655 | @end defmac | |
6656 | ||
ae2b9f1f | 6657 | @hook TARGET_ASM_LOOP_ALIGN_MAX_SKIP |
202d6e5f | 6658 | |
6659 | @defmac LABEL_ALIGN (@var{label}) | |
6660 | The alignment (log base 2) to put in front of @var{label}. | |
6661 | If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment, | |
6662 | the maximum of the specified values is used. | |
6663 | ||
6664 | Unless it's necessary to inspect the @var{label} parameter, it is better | |
6665 | to set the variable @code{align_labels} in the target's | |
6666 | @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's | |
6667 | selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation. | |
6668 | @end defmac | |
6669 | ||
ae2b9f1f | 6670 | @hook TARGET_ASM_LABEL_ALIGN_MAX_SKIP |
202d6e5f | 6671 | |
6672 | @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes}) | |
6673 | A C statement to output to the stdio stream @var{stream} an assembler | |
6674 | instruction to advance the location counter by @var{nbytes} bytes. | |
6675 | Those bytes should be zero when loaded. @var{nbytes} will be a C | |
6676 | expression of type @code{unsigned HOST_WIDE_INT}. | |
6677 | @end defmac | |
6678 | ||
6679 | @defmac ASM_NO_SKIP_IN_TEXT | |
6680 | Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the | |
6681 | text section because it fails to put zeros in the bytes that are skipped. | |
6682 | This is true on many Unix systems, where the pseudo--op to skip bytes | |
6683 | produces no-op instructions rather than zeros when used in the text | |
6684 | section. | |
6685 | @end defmac | |
6686 | ||
6687 | @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power}) | |
6688 | A C statement to output to the stdio stream @var{stream} an assembler | |
6689 | command to advance the location counter to a multiple of 2 to the | |
6690 | @var{power} bytes. @var{power} will be a C expression of type @code{int}. | |
6691 | @end defmac | |
6692 | ||
6693 | @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power}) | |
6694 | Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used | |
6695 | for padding, if necessary. | |
6696 | @end defmac | |
6697 | ||
6698 | @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip}) | |
6699 | A C statement to output to the stdio stream @var{stream} an assembler | |
6700 | command to advance the location counter to a multiple of 2 to the | |
6701 | @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to | |
6702 | satisfy the alignment request. @var{power} and @var{max_skip} will be | |
6703 | a C expression of type @code{int}. | |
6704 | @end defmac | |
6705 | ||
6706 | @need 3000 | |
6707 | @node Debugging Info | |
6708 | @section Controlling Debugging Information Format | |
6709 | ||
6710 | @c prevent bad page break with this line | |
6711 | This describes how to specify debugging information. | |
6712 | ||
6713 | @menu | |
6714 | * All Debuggers:: Macros that affect all debugging formats uniformly. | |
6715 | * DBX Options:: Macros enabling specific options in DBX format. | |
6716 | * DBX Hooks:: Hook macros for varying DBX format. | |
6717 | * File Names and DBX:: Macros controlling output of file names in DBX format. | |
6718 | * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats. | |
6719 | * VMS Debug:: Macros for VMS debug format. | |
6720 | @end menu | |
6721 | ||
6722 | @node All Debuggers | |
6723 | @subsection Macros Affecting All Debugging Formats | |
6724 | ||
6725 | @c prevent bad page break with this line | |
6726 | These macros affect all debugging formats. | |
6727 | ||
6728 | @defmac DBX_REGISTER_NUMBER (@var{regno}) | |
6729 | A C expression that returns the DBX register number for the compiler | |
6730 | register number @var{regno}. In the default macro provided, the value | |
6731 | of this expression will be @var{regno} itself. But sometimes there are | |
6732 | some registers that the compiler knows about and DBX does not, or vice | |
6733 | versa. In such cases, some register may need to have one number in the | |
6734 | compiler and another for DBX@. | |
6735 | ||
6736 | If two registers have consecutive numbers inside GCC, and they can be | |
6737 | used as a pair to hold a multiword value, then they @emph{must} have | |
6738 | consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}. | |
6739 | Otherwise, debuggers will be unable to access such a pair, because they | |
6740 | expect register pairs to be consecutive in their own numbering scheme. | |
6741 | ||
6742 | If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that | |
6743 | does not preserve register pairs, then what you must do instead is | |
6744 | redefine the actual register numbering scheme. | |
6745 | @end defmac | |
6746 | ||
6747 | @defmac DEBUGGER_AUTO_OFFSET (@var{x}) | |
6748 | A C expression that returns the integer offset value for an automatic | |
6749 | variable having address @var{x} (an RTL expression). The default | |
6750 | computation assumes that @var{x} is based on the frame-pointer and | |
6751 | gives the offset from the frame-pointer. This is required for targets | |
6752 | that produce debugging output for DBX or COFF-style debugging output | |
6753 | for SDB and allow the frame-pointer to be eliminated when the | |
6754 | @option{-g} options is used. | |
6755 | @end defmac | |
6756 | ||
6757 | @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x}) | |
6758 | A C expression that returns the integer offset value for an argument | |
6759 | having address @var{x} (an RTL expression). The nominal offset is | |
6760 | @var{offset}. | |
6761 | @end defmac | |
6762 | ||
6763 | @defmac PREFERRED_DEBUGGING_TYPE | |
6764 | A C expression that returns the type of debugging output GCC should | |
6765 | produce when the user specifies just @option{-g}. Define | |
6766 | this if you have arranged for GCC to support more than one format of | |
6767 | debugging output. Currently, the allowable values are @code{DBX_DEBUG}, | |
6768 | @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG}, | |
6769 | @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}. | |
6770 | ||
6771 | When the user specifies @option{-ggdb}, GCC normally also uses the | |
6772 | value of this macro to select the debugging output format, but with two | |
6773 | exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the | |
6774 | value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is | |
6775 | defined, GCC uses @code{DBX_DEBUG}. | |
6776 | ||
6777 | The value of this macro only affects the default debugging output; the | |
6778 | user can always get a specific type of output by using @option{-gstabs}, | |
6779 | @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}. | |
6780 | @end defmac | |
6781 | ||
6782 | @node DBX Options | |
6783 | @subsection Specific Options for DBX Output | |
6784 | ||
6785 | @c prevent bad page break with this line | |
6786 | These are specific options for DBX output. | |
6787 | ||
6788 | @defmac DBX_DEBUGGING_INFO | |
6789 | Define this macro if GCC should produce debugging output for DBX | |
6790 | in response to the @option{-g} option. | |
6791 | @end defmac | |
6792 | ||
6793 | @defmac XCOFF_DEBUGGING_INFO | |
6794 | Define this macro if GCC should produce XCOFF format debugging output | |
6795 | in response to the @option{-g} option. This is a variant of DBX format. | |
6796 | @end defmac | |
6797 | ||
6798 | @defmac DEFAULT_GDB_EXTENSIONS | |
6799 | Define this macro to control whether GCC should by default generate | |
6800 | GDB's extended version of DBX debugging information (assuming DBX-format | |
6801 | debugging information is enabled at all). If you don't define the | |
6802 | macro, the default is 1: always generate the extended information | |
6803 | if there is any occasion to. | |
6804 | @end defmac | |
6805 | ||
6806 | @defmac DEBUG_SYMS_TEXT | |
6807 | Define this macro if all @code{.stabs} commands should be output while | |
6808 | in the text section. | |
6809 | @end defmac | |
6810 | ||
6811 | @defmac ASM_STABS_OP | |
6812 | A C string constant, including spacing, naming the assembler pseudo op to | |
6813 | use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol. | |
6814 | If you don't define this macro, @code{"\t.stabs\t"} is used. This macro | |
6815 | applies only to DBX debugging information format. | |
6816 | @end defmac | |
6817 | ||
6818 | @defmac ASM_STABD_OP | |
6819 | A C string constant, including spacing, naming the assembler pseudo op to | |
6820 | use instead of @code{"\t.stabd\t"} to define a debugging symbol whose | |
6821 | value is the current location. If you don't define this macro, | |
6822 | @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging | |
6823 | information format. | |
6824 | @end defmac | |
6825 | ||
6826 | @defmac ASM_STABN_OP | |
6827 | A C string constant, including spacing, naming the assembler pseudo op to | |
6828 | use instead of @code{"\t.stabn\t"} to define a debugging symbol with no | |
6829 | name. If you don't define this macro, @code{"\t.stabn\t"} is used. This | |
6830 | macro applies only to DBX debugging information format. | |
6831 | @end defmac | |
6832 | ||
6833 | @defmac DBX_NO_XREFS | |
6834 | Define this macro if DBX on your system does not support the construct | |
6835 | @samp{xs@var{tagname}}. On some systems, this construct is used to | |
6836 | describe a forward reference to a structure named @var{tagname}. | |
6837 | On other systems, this construct is not supported at all. | |
6838 | @end defmac | |
6839 | ||
6840 | @defmac DBX_CONTIN_LENGTH | |
6841 | A symbol name in DBX-format debugging information is normally | |
6842 | continued (split into two separate @code{.stabs} directives) when it | |
6843 | exceeds a certain length (by default, 80 characters). On some | |
6844 | operating systems, DBX requires this splitting; on others, splitting | |
6845 | must not be done. You can inhibit splitting by defining this macro | |
6846 | with the value zero. You can override the default splitting-length by | |
6847 | defining this macro as an expression for the length you desire. | |
6848 | @end defmac | |
6849 | ||
6850 | @defmac DBX_CONTIN_CHAR | |
6851 | Normally continuation is indicated by adding a @samp{\} character to | |
6852 | the end of a @code{.stabs} string when a continuation follows. To use | |
6853 | a different character instead, define this macro as a character | |
6854 | constant for the character you want to use. Do not define this macro | |
6855 | if backslash is correct for your system. | |
6856 | @end defmac | |
6857 | ||
6858 | @defmac DBX_STATIC_STAB_DATA_SECTION | |
6859 | Define this macro if it is necessary to go to the data section before | |
6860 | outputting the @samp{.stabs} pseudo-op for a non-global static | |
6861 | variable. | |
6862 | @end defmac | |
6863 | ||
6864 | @defmac DBX_TYPE_DECL_STABS_CODE | |
6865 | The value to use in the ``code'' field of the @code{.stabs} directive | |
6866 | for a typedef. The default is @code{N_LSYM}. | |
6867 | @end defmac | |
6868 | ||
6869 | @defmac DBX_STATIC_CONST_VAR_CODE | |
6870 | The value to use in the ``code'' field of the @code{.stabs} directive | |
6871 | for a static variable located in the text section. DBX format does not | |
6872 | provide any ``right'' way to do this. The default is @code{N_FUN}. | |
6873 | @end defmac | |
6874 | ||
6875 | @defmac DBX_REGPARM_STABS_CODE | |
6876 | The value to use in the ``code'' field of the @code{.stabs} directive | |
6877 | for a parameter passed in registers. DBX format does not provide any | |
6878 | ``right'' way to do this. The default is @code{N_RSYM}. | |
6879 | @end defmac | |
6880 | ||
6881 | @defmac DBX_REGPARM_STABS_LETTER | |
6882 | The letter to use in DBX symbol data to identify a symbol as a parameter | |
6883 | passed in registers. DBX format does not customarily provide any way to | |
6884 | do this. The default is @code{'P'}. | |
6885 | @end defmac | |
6886 | ||
6887 | @defmac DBX_FUNCTION_FIRST | |
6888 | Define this macro if the DBX information for a function and its | |
6889 | arguments should precede the assembler code for the function. Normally, | |
6890 | in DBX format, the debugging information entirely follows the assembler | |
6891 | code. | |
6892 | @end defmac | |
6893 | ||
6894 | @defmac DBX_BLOCKS_FUNCTION_RELATIVE | |
6895 | Define this macro, with value 1, if the value of a symbol describing | |
6896 | the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be | |
6897 | relative to the start of the enclosing function. Normally, GCC uses | |
6898 | an absolute address. | |
6899 | @end defmac | |
6900 | ||
6901 | @defmac DBX_LINES_FUNCTION_RELATIVE | |
6902 | Define this macro, with value 1, if the value of a symbol indicating | |
6903 | the current line number (@code{N_SLINE}) should be relative to the | |
6904 | start of the enclosing function. Normally, GCC uses an absolute address. | |
6905 | @end defmac | |
6906 | ||
6907 | @defmac DBX_USE_BINCL | |
6908 | Define this macro if GCC should generate @code{N_BINCL} and | |
6909 | @code{N_EINCL} stabs for included header files, as on Sun systems. This | |
6910 | macro also directs GCC to output a type number as a pair of a file | |
6911 | number and a type number within the file. Normally, GCC does not | |
6912 | generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single | |
6913 | number for a type number. | |
6914 | @end defmac | |
6915 | ||
6916 | @node DBX Hooks | |
6917 | @subsection Open-Ended Hooks for DBX Format | |
6918 | ||
6919 | @c prevent bad page break with this line | |
6920 | These are hooks for DBX format. | |
6921 | ||
202d6e5f | 6922 | @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter}) |
6923 | A C statement to output DBX debugging information before code for line | |
6924 | number @var{line} of the current source file to the stdio stream | |
6925 | @var{stream}. @var{counter} is the number of time the macro was | |
6926 | invoked, including the current invocation; it is intended to generate | |
6927 | unique labels in the assembly output. | |
6928 | ||
6929 | This macro should not be defined if the default output is correct, or | |
6930 | if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}. | |
6931 | @end defmac | |
6932 | ||
6933 | @defmac NO_DBX_FUNCTION_END | |
6934 | Some stabs encapsulation formats (in particular ECOFF), cannot handle the | |
6935 | @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct. | |
6936 | On those machines, define this macro to turn this feature off without | |
6937 | disturbing the rest of the gdb extensions. | |
6938 | @end defmac | |
6939 | ||
6940 | @defmac NO_DBX_BNSYM_ENSYM | |
6941 | Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx | |
6942 | extension construct. On those machines, define this macro to turn this | |
6943 | feature off without disturbing the rest of the gdb extensions. | |
6944 | @end defmac | |
6945 | ||
6946 | @node File Names and DBX | |
6947 | @subsection File Names in DBX Format | |
6948 | ||
6949 | @c prevent bad page break with this line | |
6950 | This describes file names in DBX format. | |
6951 | ||
6952 | @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name}) | |
6953 | A C statement to output DBX debugging information to the stdio stream | |
6954 | @var{stream}, which indicates that file @var{name} is the main source | |
6955 | file---the file specified as the input file for compilation. | |
6956 | This macro is called only once, at the beginning of compilation. | |
6957 | ||
6958 | This macro need not be defined if the standard form of output | |
6959 | for DBX debugging information is appropriate. | |
6960 | ||
6961 | It may be necessary to refer to a label equal to the beginning of the | |
6962 | text section. You can use @samp{assemble_name (stream, ltext_label_name)} | |
6963 | to do so. If you do this, you must also set the variable | |
6964 | @var{used_ltext_label_name} to @code{true}. | |
6965 | @end defmac | |
6966 | ||
6967 | @defmac NO_DBX_MAIN_SOURCE_DIRECTORY | |
6968 | Define this macro, with value 1, if GCC should not emit an indication | |
6969 | of the current directory for compilation and current source language at | |
6970 | the beginning of the file. | |
6971 | @end defmac | |
6972 | ||
6973 | @defmac NO_DBX_GCC_MARKER | |
6974 | Define this macro, with value 1, if GCC should not emit an indication | |
6975 | that this object file was compiled by GCC@. The default is to emit | |
6976 | an @code{N_OPT} stab at the beginning of every source file, with | |
6977 | @samp{gcc2_compiled.} for the string and value 0. | |
6978 | @end defmac | |
6979 | ||
6980 | @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name}) | |
6981 | A C statement to output DBX debugging information at the end of | |
6982 | compilation of the main source file @var{name}. Output should be | |
6983 | written to the stdio stream @var{stream}. | |
6984 | ||
6985 | If you don't define this macro, nothing special is output at the end | |
6986 | of compilation, which is correct for most machines. | |
6987 | @end defmac | |
6988 | ||
6989 | @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END | |
6990 | Define this macro @emph{instead of} defining | |
6991 | @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at | |
6992 | the end of compilation is an @code{N_SO} stab with an empty string, | |
6993 | whose value is the highest absolute text address in the file. | |
6994 | @end defmac | |
6995 | ||
6996 | @need 2000 | |
6997 | @node SDB and DWARF | |
6998 | @subsection Macros for SDB and DWARF Output | |
6999 | ||
7000 | @c prevent bad page break with this line | |
7001 | Here are macros for SDB and DWARF output. | |
7002 | ||
7003 | @defmac SDB_DEBUGGING_INFO | |
2e3b03ce | 7004 | Define this macro to 1 if GCC should produce COFF-style debugging output |
202d6e5f | 7005 | for SDB in response to the @option{-g} option. |
7006 | @end defmac | |
7007 | ||
7008 | @defmac DWARF2_DEBUGGING_INFO | |
7009 | Define this macro if GCC should produce dwarf version 2 format | |
7010 | debugging output in response to the @option{-g} option. | |
7011 | ||
7012 | @hook TARGET_DWARF_CALLING_CONVENTION | |
202d6e5f | 7013 | |
7014 | To support optional call frame debugging information, you must also | |
7015 | define @code{INCOMING_RETURN_ADDR_RTX} and either set | |
7016 | @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the | |
7017 | prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save} | |
7018 | as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't. | |
7019 | @end defmac | |
7020 | ||
7021 | @defmac DWARF2_FRAME_INFO | |
7022 | Define this macro to a nonzero value if GCC should always output | |
cc7d6aed | 7023 | Dwarf 2 frame information. If @code{TARGET_EXCEPT_UNWIND_INFO} |
7024 | (@pxref{Exception Region Output}) returns @code{UI_DWARF2}, and | |
7025 | exceptions are enabled, GCC will output this information not matter | |
7026 | how you define @code{DWARF2_FRAME_INFO}. | |
202d6e5f | 7027 | @end defmac |
7028 | ||
cc7d6aed | 7029 | @hook TARGET_DEBUG_UNWIND_INFO |
cc7d6aed | 7030 | |
202d6e5f | 7031 | @defmac DWARF2_ASM_LINE_DEBUG_INFO |
7032 | Define this macro to be a nonzero value if the assembler can generate Dwarf 2 | |
7033 | line debug info sections. This will result in much more compact line number | |
7034 | tables, and hence is desirable if it works. | |
7035 | @end defmac | |
7036 | ||
1651aa77 | 7037 | @hook TARGET_WANT_DEBUG_PUB_SECTIONS |
7038 | ||
effa044f | 7039 | @hook TARGET_FORCE_AT_COMP_DIR |
7040 | ||
8a42230a | 7041 | @hook TARGET_DELAY_SCHED2 |
7042 | ||
7043 | @hook TARGET_DELAY_VARTRACK | |
7044 | ||
84b574d2 | 7045 | @hook TARGET_NO_REGISTER_ALLOCATION |
7046 | ||
202d6e5f | 7047 | @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2}) |
7048 | A C statement to issue assembly directives that create a difference | |
7049 | @var{lab1} minus @var{lab2}, using an integer of the given @var{size}. | |
7050 | @end defmac | |
7051 | ||
7052 | @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2}) | |
7053 | A C statement to issue assembly directives that create a difference | |
7054 | between the two given labels in system defined units, e.g. instruction | |
7055 | slots on IA64 VMS, using an integer of the given size. | |
7056 | @end defmac | |
7057 | ||
7058 | @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section}) | |
7059 | A C statement to issue assembly directives that create a | |
7060 | section-relative reference to the given @var{label}, using an integer of the | |
7061 | given @var{size}. The label is known to be defined in the given @var{section}. | |
7062 | @end defmac | |
7063 | ||
7064 | @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label}) | |
7065 | A C statement to issue assembly directives that create a self-relative | |
7066 | reference to the given @var{label}, using an integer of the given @var{size}. | |
7067 | @end defmac | |
7068 | ||
552a60d3 | 7069 | @defmac ASM_OUTPUT_DWARF_DATAREL (@var{stream}, @var{size}, @var{label}) |
7070 | A C statement to issue assembly directives that create a reference to the | |
7071 | given @var{label} relative to the dbase, using an integer of the given @var{size}. | |
7072 | @end defmac | |
7073 | ||
202d6e5f | 7074 | @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label}) |
7075 | A C statement to issue assembly directives that create a reference to | |
7076 | the DWARF table identifier @var{label} from the current section. This | |
7077 | is used on some systems to avoid garbage collecting a DWARF table which | |
7078 | is referenced by a function. | |
7079 | @end defmac | |
7080 | ||
7081 | @hook TARGET_ASM_OUTPUT_DWARF_DTPREL | |
202d6e5f | 7082 | |
7083 | @defmac PUT_SDB_@dots{} | |
7084 | Define these macros to override the assembler syntax for the special | |
7085 | SDB assembler directives. See @file{sdbout.c} for a list of these | |
7086 | macros and their arguments. If the standard syntax is used, you need | |
7087 | not define them yourself. | |
7088 | @end defmac | |
7089 | ||
7090 | @defmac SDB_DELIM | |
7091 | Some assemblers do not support a semicolon as a delimiter, even between | |
7092 | SDB assembler directives. In that case, define this macro to be the | |
7093 | delimiter to use (usually @samp{\n}). It is not necessary to define | |
7094 | a new set of @code{PUT_SDB_@var{op}} macros if this is the only change | |
7095 | required. | |
7096 | @end defmac | |
7097 | ||
7098 | @defmac SDB_ALLOW_UNKNOWN_REFERENCES | |
7099 | Define this macro to allow references to unknown structure, | |
7100 | union, or enumeration tags to be emitted. Standard COFF does not | |
7101 | allow handling of unknown references, MIPS ECOFF has support for | |
7102 | it. | |
7103 | @end defmac | |
7104 | ||
7105 | @defmac SDB_ALLOW_FORWARD_REFERENCES | |
7106 | Define this macro to allow references to structure, union, or | |
7107 | enumeration tags that have not yet been seen to be handled. Some | |
7108 | assemblers choke if forward tags are used, while some require it. | |
7109 | @end defmac | |
7110 | ||
7111 | @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}) | |
7112 | A C statement to output SDB debugging information before code for line | |
7113 | number @var{line} of the current source file to the stdio stream | |
7114 | @var{stream}. The default is to emit an @code{.ln} directive. | |
7115 | @end defmac | |
7116 | ||
7117 | @need 2000 | |
7118 | @node VMS Debug | |
7119 | @subsection Macros for VMS Debug Format | |
7120 | ||
7121 | @c prevent bad page break with this line | |
7122 | Here are macros for VMS debug format. | |
7123 | ||
7124 | @defmac VMS_DEBUGGING_INFO | |
7125 | Define this macro if GCC should produce debugging output for VMS | |
7126 | in response to the @option{-g} option. The default behavior for VMS | |
7127 | is to generate minimal debug info for a traceback in the absence of | |
7128 | @option{-g} unless explicitly overridden with @option{-g0}. This | |
02e53c17 | 7129 | behavior is controlled by @code{TARGET_OPTION_OPTIMIZATION} and |
202d6e5f | 7130 | @code{TARGET_OPTION_OVERRIDE}. |
7131 | @end defmac | |
7132 | ||
7133 | @node Floating Point | |
7134 | @section Cross Compilation and Floating Point | |
7135 | @cindex cross compilation and floating point | |
7136 | @cindex floating point and cross compilation | |
7137 | ||
7138 | While all modern machines use twos-complement representation for integers, | |
7139 | there are a variety of representations for floating point numbers. This | |
7140 | means that in a cross-compiler the representation of floating point numbers | |
7141 | in the compiled program may be different from that used in the machine | |
7142 | doing the compilation. | |
7143 | ||
7144 | Because different representation systems may offer different amounts of | |
7145 | range and precision, all floating point constants must be represented in | |
7146 | the target machine's format. Therefore, the cross compiler cannot | |
7147 | safely use the host machine's floating point arithmetic; it must emulate | |
7148 | the target's arithmetic. To ensure consistency, GCC always uses | |
7149 | emulation to work with floating point values, even when the host and | |
7150 | target floating point formats are identical. | |
7151 | ||
7152 | The following macros are provided by @file{real.h} for the compiler to | |
7153 | use. All parts of the compiler which generate or optimize | |
7154 | floating-point calculations must use these macros. They may evaluate | |
7155 | their operands more than once, so operands must not have side effects. | |
7156 | ||
7157 | @defmac REAL_VALUE_TYPE | |
7158 | The C data type to be used to hold a floating point value in the target | |
7159 | machine's format. Typically this is a @code{struct} containing an | |
7160 | array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque | |
7161 | quantity. | |
7162 | @end defmac | |
7163 | ||
202d6e5f | 7164 | @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x}) |
7165 | Truncates @var{x} to a signed integer, rounding toward zero. | |
7166 | @end deftypefn | |
7167 | ||
7168 | @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x}) | |
7169 | Truncates @var{x} to an unsigned integer, rounding toward zero. If | |
7170 | @var{x} is negative, returns zero. | |
7171 | @end deftypefn | |
7172 | ||
3754d046 | 7173 | @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, machine_mode @var{mode}) |
202d6e5f | 7174 | Converts @var{string} into a floating point number in the target machine's |
7175 | representation for mode @var{mode}. This routine can handle both | |
7176 | decimal and hexadecimal floating point constants, using the syntax | |
7177 | defined by the C language for both. | |
7178 | @end deftypefn | |
7179 | ||
7180 | @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x}) | |
7181 | Returns 1 if @var{x} is negative (including negative zero), 0 otherwise. | |
7182 | @end deftypefn | |
7183 | ||
7184 | @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x}) | |
7185 | Determines whether @var{x} represents infinity (positive or negative). | |
7186 | @end deftypefn | |
7187 | ||
7188 | @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x}) | |
7189 | Determines whether @var{x} represents a ``NaN'' (not-a-number). | |
7190 | @end deftypefn | |
7191 | ||
202d6e5f | 7192 | @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x}) |
7193 | Returns the negative of the floating point value @var{x}. | |
7194 | @end deftypefn | |
7195 | ||
7196 | @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x}) | |
7197 | Returns the absolute value of @var{x}. | |
7198 | @end deftypefn | |
7199 | ||
202d6e5f | 7200 | @node Mode Switching |
7201 | @section Mode Switching Instructions | |
7202 | @cindex mode switching | |
7203 | The following macros control mode switching optimizations: | |
7204 | ||
7205 | @defmac OPTIMIZE_MODE_SWITCHING (@var{entity}) | |
7206 | Define this macro if the port needs extra instructions inserted for mode | |
7207 | switching in an optimizing compilation. | |
7208 | ||
7209 | For an example, the SH4 can perform both single and double precision | |
7210 | floating point operations, but to perform a single precision operation, | |
7211 | the FPSCR PR bit has to be cleared, while for a double precision | |
7212 | operation, this bit has to be set. Changing the PR bit requires a general | |
7213 | purpose register as a scratch register, hence these FPSCR sets have to | |
7214 | be inserted before reload, i.e.@: you can't put this into instruction emitting | |
7215 | or @code{TARGET_MACHINE_DEPENDENT_REORG}. | |
7216 | ||
7217 | You can have multiple entities that are mode-switched, and select at run time | |
7218 | which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should | |
7219 | return nonzero for any @var{entity} that needs mode-switching. | |
7220 | If you define this macro, you also have to define | |
cea19dab | 7221 | @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{TARGET_MODE_NEEDED}, |
7222 | @code{TARGET_MODE_PRIORITY} and @code{TARGET_MODE_EMIT}. | |
7223 | @code{TARGET_MODE_AFTER}, @code{TARGET_MODE_ENTRY}, and @code{TARGET_MODE_EXIT} | |
202d6e5f | 7224 | are optional. |
7225 | @end defmac | |
7226 | ||
7227 | @defmac NUM_MODES_FOR_MODE_SWITCHING | |
7228 | If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as | |
7229 | initializer for an array of integers. Each initializer element | |
7230 | N refers to an entity that needs mode switching, and specifies the number | |
7231 | of different modes that might need to be set for this entity. | |
7232 | The position of the initializer in the initializer---starting counting at | |
7233 | zero---determines the integer that is used to refer to the mode-switched | |
7234 | entity in question. | |
7235 | In macros that take mode arguments / yield a mode result, modes are | |
7236 | represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode | |
7237 | switch is needed / supplied. | |
7238 | @end defmac | |
7239 | ||
cea19dab | 7240 | @hook TARGET_MODE_EMIT |
202d6e5f | 7241 | |
cea19dab | 7242 | @hook TARGET_MODE_NEEDED |
202d6e5f | 7243 | |
cea19dab | 7244 | @hook TARGET_MODE_AFTER |
202d6e5f | 7245 | |
cea19dab | 7246 | @hook TARGET_MODE_ENTRY |
202d6e5f | 7247 | |
cea19dab | 7248 | @hook TARGET_MODE_EXIT |
202d6e5f | 7249 | |
cea19dab | 7250 | @hook TARGET_MODE_PRIORITY |
202d6e5f | 7251 | |
7252 | @node Target Attributes | |
7253 | @section Defining target-specific uses of @code{__attribute__} | |
7254 | @cindex target attributes | |
7255 | @cindex machine attributes | |
7256 | @cindex attributes, target-specific | |
7257 | ||
7258 | Target-specific attributes may be defined for functions, data and types. | |
7259 | These are described using the following target hooks; they also need to | |
7260 | be documented in @file{extend.texi}. | |
7261 | ||
7262 | @hook TARGET_ATTRIBUTE_TABLE | |
202d6e5f | 7263 | |
7264 | @hook TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P | |
202d6e5f | 7265 | |
7266 | @hook TARGET_COMP_TYPE_ATTRIBUTES | |
202d6e5f | 7267 | |
7268 | @hook TARGET_SET_DEFAULT_TYPE_ATTRIBUTES | |
202d6e5f | 7269 | |
7270 | @hook TARGET_MERGE_TYPE_ATTRIBUTES | |
202d6e5f | 7271 | |
7272 | @hook TARGET_MERGE_DECL_ATTRIBUTES | |
202d6e5f | 7273 | |
7274 | @hook TARGET_VALID_DLLIMPORT_ATTRIBUTE_P | |
7275 | ||
7276 | @defmac TARGET_DECLSPEC | |
7277 | Define this macro to a nonzero value if you want to treat | |
7278 | @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By | |
7279 | default, this behavior is enabled only for targets that define | |
7280 | @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation | |
7281 | of @code{__declspec} is via a built-in macro, but you should not rely | |
7282 | on this implementation detail. | |
7283 | @end defmac | |
7284 | ||
7285 | @hook TARGET_INSERT_ATTRIBUTES | |
202d6e5f | 7286 | |
7287 | @hook TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P | |
202d6e5f | 7288 | |
7289 | @hook TARGET_OPTION_VALID_ATTRIBUTE_P | |
202d6e5f | 7290 | |
7291 | @hook TARGET_OPTION_SAVE | |
202d6e5f | 7292 | |
7293 | @hook TARGET_OPTION_RESTORE | |
202d6e5f | 7294 | |
ff7cb116 | 7295 | @hook TARGET_OPTION_POST_STREAM_IN |
7296 | ||
202d6e5f | 7297 | @hook TARGET_OPTION_PRINT |
202d6e5f | 7298 | |
04f989af | 7299 | @hook TARGET_OPTION_PRAGMA_PARSE |
202d6e5f | 7300 | |
7301 | @hook TARGET_OPTION_OVERRIDE | |
202d6e5f | 7302 | |
cc8ef84f | 7303 | @hook TARGET_OPTION_FUNCTION_VERSIONS |
cc8ef84f | 7304 | |
202d6e5f | 7305 | @hook TARGET_CAN_INLINE_P |
202d6e5f | 7306 | |
9d0e3e3a | 7307 | @hook TARGET_RELAYOUT_FUNCTION |
7308 | ||
202d6e5f | 7309 | @node Emulated TLS |
7310 | @section Emulating TLS | |
7311 | @cindex Emulated TLS | |
7312 | ||
7313 | For targets whose psABI does not provide Thread Local Storage via | |
7314 | specific relocations and instruction sequences, an emulation layer is | |
7315 | used. A set of target hooks allows this emulation layer to be | |
7316 | configured for the requirements of a particular target. For instance | |
7317 | the psABI may in fact specify TLS support in terms of an emulation | |
7318 | layer. | |
7319 | ||
7320 | The emulation layer works by creating a control object for every TLS | |
7321 | object. To access the TLS object, a lookup function is provided | |
7322 | which, when given the address of the control object, will return the | |
7323 | address of the current thread's instance of the TLS object. | |
7324 | ||
7325 | @hook TARGET_EMUTLS_GET_ADDRESS | |
202d6e5f | 7326 | |
7327 | @hook TARGET_EMUTLS_REGISTER_COMMON | |
202d6e5f | 7328 | |
7329 | @hook TARGET_EMUTLS_VAR_SECTION | |
202d6e5f | 7330 | |
7331 | @hook TARGET_EMUTLS_TMPL_SECTION | |
202d6e5f | 7332 | |
7333 | @hook TARGET_EMUTLS_VAR_PREFIX | |
202d6e5f | 7334 | |
7335 | @hook TARGET_EMUTLS_TMPL_PREFIX | |
202d6e5f | 7336 | |
7337 | @hook TARGET_EMUTLS_VAR_FIELDS | |
202d6e5f | 7338 | |
7339 | @hook TARGET_EMUTLS_VAR_INIT | |
202d6e5f | 7340 | |
7341 | @hook TARGET_EMUTLS_VAR_ALIGN_FIXED | |
202d6e5f | 7342 | |
7343 | @hook TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS | |
202d6e5f | 7344 | |
7345 | @node MIPS Coprocessors | |
7346 | @section Defining coprocessor specifics for MIPS targets. | |
7347 | @cindex MIPS coprocessor-definition macros | |
7348 | ||
7349 | The MIPS specification allows MIPS implementations to have as many as 4 | |
7350 | coprocessors, each with as many as 32 private registers. GCC supports | |
7351 | accessing these registers and transferring values between the registers | |
7352 | and memory using asm-ized variables. For example: | |
7353 | ||
7354 | @smallexample | |
7355 | register unsigned int cp0count asm ("c0r1"); | |
7356 | unsigned int d; | |
7357 | ||
7358 | d = cp0count + 3; | |
7359 | @end smallexample | |
7360 | ||
7361 | (``c0r1'' is the default name of register 1 in coprocessor 0; alternate | |
7362 | names may be added as described below, or the default names may be | |
7363 | overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.) | |
7364 | ||
7365 | Coprocessor registers are assumed to be epilogue-used; sets to them will | |
7366 | be preserved even if it does not appear that the register is used again | |
7367 | later in the function. | |
7368 | ||
7369 | Another note: according to the MIPS spec, coprocessor 1 (if present) is | |
7370 | the FPU@. One accesses COP1 registers through standard mips | |
7371 | floating-point support; they are not included in this mechanism. | |
7372 | ||
202d6e5f | 7373 | @node PCH Target |
7374 | @section Parameters for Precompiled Header Validity Checking | |
7375 | @cindex parameters, precompiled headers | |
7376 | ||
7377 | @hook TARGET_GET_PCH_VALIDITY | |
202d6e5f | 7378 | |
7379 | @hook TARGET_PCH_VALID_P | |
202d6e5f | 7380 | |
7381 | @hook TARGET_CHECK_PCH_TARGET_FLAGS | |
202d6e5f | 7382 | |
e5d92c9b | 7383 | @hook TARGET_PREPARE_PCH_SAVE |
7384 | ||
202d6e5f | 7385 | @node C++ ABI |
7386 | @section C++ ABI parameters | |
7387 | @cindex parameters, c++ abi | |
7388 | ||
7389 | @hook TARGET_CXX_GUARD_TYPE | |
202d6e5f | 7390 | |
7391 | @hook TARGET_CXX_GUARD_MASK_BIT | |
202d6e5f | 7392 | |
7393 | @hook TARGET_CXX_GET_COOKIE_SIZE | |
202d6e5f | 7394 | |
7395 | @hook TARGET_CXX_COOKIE_HAS_SIZE | |
202d6e5f | 7396 | |
7397 | @hook TARGET_CXX_IMPORT_EXPORT_CLASS | |
202d6e5f | 7398 | |
7399 | @hook TARGET_CXX_CDTOR_RETURNS_THIS | |
202d6e5f | 7400 | |
7401 | @hook TARGET_CXX_KEY_METHOD_MAY_BE_INLINE | |
202d6e5f | 7402 | |
7403 | @hook TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY | |
7404 | ||
7405 | @hook TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT | |
202d6e5f | 7406 | |
7407 | @hook TARGET_CXX_LIBRARY_RTTI_COMDAT | |
202d6e5f | 7408 | |
7409 | @hook TARGET_CXX_USE_AEABI_ATEXIT | |
202d6e5f | 7410 | |
7411 | @hook TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT | |
202d6e5f | 7412 | |
7413 | @hook TARGET_CXX_ADJUST_CLASS_AT_DEFINITION | |
7414 | ||
7bfdbc17 | 7415 | @hook TARGET_CXX_DECL_MANGLING_CONTEXT |
7416 | ||
202d6e5f | 7417 | @node Named Address Spaces |
7418 | @section Adding support for named address spaces | |
7419 | @cindex named address spaces | |
7420 | ||
7421 | The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275 | |
7422 | standards committee, @cite{Programming Languages - C - Extensions to | |
7423 | support embedded processors}, specifies a syntax for embedded | |
7424 | processors to specify alternate address spaces. You can configure a | |
7425 | GCC port to support section 5.1 of the draft report to add support for | |
7426 | address spaces other than the default address space. These address | |
7427 | spaces are new keywords that are similar to the @code{volatile} and | |
7428 | @code{const} type attributes. | |
7429 | ||
7430 | Pointers to named address spaces can have a different size than | |
7431 | pointers to the generic address space. | |
7432 | ||
7433 | For example, the SPU port uses the @code{__ea} address space to refer | |
7434 | to memory in the host processor, rather than memory local to the SPU | |
7435 | processor. Access to memory in the @code{__ea} address space involves | |
7436 | issuing DMA operations to move data between the host processor and the | |
7437 | local processor memory address space. Pointers in the @code{__ea} | |
7438 | address space are either 32 bits or 64 bits based on the | |
7439 | @option{-mea32} or @option{-mea64} switches (native SPU pointers are | |
7440 | always 32 bits). | |
7441 | ||
7442 | Internally, address spaces are represented as a small integer in the | |
7443 | range 0 to 15 with address space 0 being reserved for the generic | |
7444 | address space. | |
7445 | ||
7446 | To register a named address space qualifier keyword with the C front end, | |
7447 | the target may call the @code{c_register_addr_space} routine. For example, | |
7448 | the SPU port uses the following to declare @code{__ea} as the keyword for | |
7449 | named address space #1: | |
7450 | @smallexample | |
7451 | #define ADDR_SPACE_EA 1 | |
7452 | c_register_addr_space ("__ea", ADDR_SPACE_EA); | |
7453 | @end smallexample | |
7454 | ||
7455 | @hook TARGET_ADDR_SPACE_POINTER_MODE | |
202d6e5f | 7456 | |
7457 | @hook TARGET_ADDR_SPACE_ADDRESS_MODE | |
202d6e5f | 7458 | |
7459 | @hook TARGET_ADDR_SPACE_VALID_POINTER_MODE | |
202d6e5f | 7460 | |
7461 | @hook TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P | |
202d6e5f | 7462 | |
7463 | @hook TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS | |
202d6e5f | 7464 | |
7465 | @hook TARGET_ADDR_SPACE_SUBSET_P | |
202d6e5f | 7466 | |
9cb89654 | 7467 | @hook TARGET_ADDR_SPACE_ZERO_ADDRESS_VALID |
7468 | ||
202d6e5f | 7469 | @hook TARGET_ADDR_SPACE_CONVERT |
202d6e5f | 7470 | |
2e7a553a | 7471 | @hook TARGET_ADDR_SPACE_DEBUG |
7472 | ||
202d6e5f | 7473 | @node Misc |
7474 | @section Miscellaneous Parameters | |
7475 | @cindex parameters, miscellaneous | |
7476 | ||
7477 | @c prevent bad page break with this line | |
7478 | Here are several miscellaneous parameters. | |
7479 | ||
7480 | @defmac HAS_LONG_COND_BRANCH | |
7481 | Define this boolean macro to indicate whether or not your architecture | |
7482 | has conditional branches that can span all of memory. It is used in | |
7483 | conjunction with an optimization that partitions hot and cold basic | |
7484 | blocks into separate sections of the executable. If this macro is | |
7485 | set to false, gcc will convert any conditional branches that attempt | |
7486 | to cross between sections into unconditional branches or indirect jumps. | |
7487 | @end defmac | |
7488 | ||
7489 | @defmac HAS_LONG_UNCOND_BRANCH | |
7490 | Define this boolean macro to indicate whether or not your architecture | |
7491 | has unconditional branches that can span all of memory. It is used in | |
7492 | conjunction with an optimization that partitions hot and cold basic | |
7493 | blocks into separate sections of the executable. If this macro is | |
7494 | set to false, gcc will convert any unconditional branches that attempt | |
7495 | to cross between sections into indirect jumps. | |
7496 | @end defmac | |
7497 | ||
7498 | @defmac CASE_VECTOR_MODE | |
7499 | An alias for a machine mode name. This is the machine mode that | |
7500 | elements of a jump-table should have. | |
7501 | @end defmac | |
7502 | ||
7503 | @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body}) | |
7504 | Optional: return the preferred mode for an @code{addr_diff_vec} | |
7505 | when the minimum and maximum offset are known. If you define this, | |
7506 | it enables extra code in branch shortening to deal with @code{addr_diff_vec}. | |
7507 | To make this work, you also have to define @code{INSN_ALIGN} and | |
7508 | make the alignment for @code{addr_diff_vec} explicit. | |
7509 | The @var{body} argument is provided so that the offset_unsigned and scale | |
7510 | flags can be updated. | |
7511 | @end defmac | |
7512 | ||
7513 | @defmac CASE_VECTOR_PC_RELATIVE | |
7514 | Define this macro to be a C expression to indicate when jump-tables | |
7515 | should contain relative addresses. You need not define this macro if | |
7516 | jump-tables never contain relative addresses, or jump-tables should | |
7517 | contain relative addresses only when @option{-fPIC} or @option{-fPIC} | |
7518 | is in effect. | |
7519 | @end defmac | |
7520 | ||
7521 | @hook TARGET_CASE_VALUES_THRESHOLD | |
202d6e5f | 7522 | |
202d6e5f | 7523 | @defmac WORD_REGISTER_OPERATIONS |
94f1fba7 | 7524 | Define this macro to 1 if operations between registers with integral mode |
202d6e5f | 7525 | smaller than a word are always performed on the entire register. |
7526 | Most RISC machines have this property and most CISC machines do not. | |
7527 | @end defmac | |
7528 | ||
7529 | @defmac LOAD_EXTEND_OP (@var{mem_mode}) | |
7530 | Define this macro to be a C expression indicating when insns that read | |
7531 | memory in @var{mem_mode}, an integral mode narrower than a word, set the | |
7532 | bits outside of @var{mem_mode} to be either the sign-extension or the | |
7533 | zero-extension of the data read. Return @code{SIGN_EXTEND} for values | |
7534 | of @var{mem_mode} for which the | |
7535 | insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and | |
7536 | @code{UNKNOWN} for other modes. | |
7537 | ||
7538 | This macro is not called with @var{mem_mode} non-integral or with a width | |
7539 | greater than or equal to @code{BITS_PER_WORD}, so you may return any | |
7540 | value in this case. Do not define this macro if it would always return | |
7541 | @code{UNKNOWN}. On machines where this macro is defined, you will normally | |
7542 | define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}. | |
7543 | ||
7544 | You may return a non-@code{UNKNOWN} value even if for some hard registers | |
7545 | the sign extension is not performed, if for the @code{REGNO_REG_CLASS} | |
7546 | of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero | |
7547 | when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any | |
7548 | integral mode larger than this but not larger than @code{word_mode}. | |
7549 | ||
7550 | You must return @code{UNKNOWN} if for some hard registers that allow this | |
7551 | mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to | |
7552 | @code{word_mode}, but that they can change to another integral mode that | |
7553 | is larger then @var{mem_mode} but still smaller than @code{word_mode}. | |
7554 | @end defmac | |
7555 | ||
7556 | @defmac SHORT_IMMEDIATES_SIGN_EXTEND | |
d0b99710 | 7557 | Define this macro to 1 if loading short immediate values into registers sign |
202d6e5f | 7558 | extends. |
7559 | @end defmac | |
7560 | ||
202d6e5f | 7561 | @hook TARGET_MIN_DIVISIONS_FOR_RECIP_MUL |
202d6e5f | 7562 | |
7563 | @defmac MOVE_MAX | |
7564 | The maximum number of bytes that a single instruction can move quickly | |
7565 | between memory and registers or between two memory locations. | |
7566 | @end defmac | |
7567 | ||
7568 | @defmac MAX_MOVE_MAX | |
7569 | The maximum number of bytes that a single instruction can move quickly | |
7570 | between memory and registers or between two memory locations. If this | |
7571 | is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the | |
7572 | constant value that is the largest value that @code{MOVE_MAX} can have | |
7573 | at run-time. | |
7574 | @end defmac | |
7575 | ||
7576 | @defmac SHIFT_COUNT_TRUNCATED | |
7577 | A C expression that is nonzero if on this machine the number of bits | |
7578 | actually used for the count of a shift operation is equal to the number | |
7579 | of bits needed to represent the size of the object being shifted. When | |
7580 | this macro is nonzero, the compiler will assume that it is safe to omit | |
7581 | a sign-extend, zero-extend, and certain bitwise `and' instructions that | |
7582 | truncates the count of a shift operation. On machines that have | |
7583 | instructions that act on bit-fields at variable positions, which may | |
7584 | include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED} | |
7585 | also enables deletion of truncations of the values that serve as | |
7586 | arguments to bit-field instructions. | |
7587 | ||
7588 | If both types of instructions truncate the count (for shifts) and | |
7589 | position (for bit-field operations), or if no variable-position bit-field | |
7590 | instructions exist, you should define this macro. | |
7591 | ||
7592 | However, on some machines, such as the 80386 and the 680x0, truncation | |
7593 | only applies to shift operations and not the (real or pretended) | |
7594 | bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on | |
7595 | such machines. Instead, add patterns to the @file{md} file that include | |
7596 | the implied truncation of the shift instructions. | |
7597 | ||
7598 | You need not define this macro if it would always have the value of zero. | |
7599 | @end defmac | |
7600 | ||
7601 | @anchor{TARGET_SHIFT_TRUNCATION_MASK} | |
7602 | @hook TARGET_SHIFT_TRUNCATION_MASK | |
202d6e5f | 7603 | |
7604 | @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec}) | |
7605 | A C expression which is nonzero if on this machine it is safe to | |
7606 | ``convert'' an integer of @var{inprec} bits to one of @var{outprec} | |
7607 | bits (where @var{outprec} is smaller than @var{inprec}) by merely | |
7608 | operating on it as if it had only @var{outprec} bits. | |
7609 | ||
7610 | On many machines, this expression can be 1. | |
7611 | ||
7612 | @c rearranged this, removed the phrase "it is reported that". this was | |
7613 | @c to fix an overfull hbox. --mew 10feb93 | |
7614 | When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for | |
7615 | modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result. | |
7616 | If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in | |
7617 | such cases may improve things. | |
7618 | @end defmac | |
7619 | ||
7620 | @hook TARGET_MODE_REP_EXTENDED | |
202d6e5f | 7621 | |
7622 | @defmac STORE_FLAG_VALUE | |
7623 | A C expression describing the value returned by a comparison operator | |
7624 | with an integral mode and stored by a store-flag instruction | |
7625 | (@samp{cstore@var{mode}4}) when the condition is true. This description must | |
7626 | apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the | |
7627 | comparison operators whose results have a @code{MODE_INT} mode. | |
7628 | ||
7629 | A value of 1 or @minus{}1 means that the instruction implementing the | |
7630 | comparison operator returns exactly 1 or @minus{}1 when the comparison is true | |
7631 | and 0 when the comparison is false. Otherwise, the value indicates | |
7632 | which bits of the result are guaranteed to be 1 when the comparison is | |
7633 | true. This value is interpreted in the mode of the comparison | |
7634 | operation, which is given by the mode of the first operand in the | |
7635 | @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of | |
7636 | @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by | |
7637 | the compiler. | |
7638 | ||
7639 | If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will | |
7640 | generate code that depends only on the specified bits. It can also | |
7641 | replace comparison operators with equivalent operations if they cause | |
7642 | the required bits to be set, even if the remaining bits are undefined. | |
7643 | For example, on a machine whose comparison operators return an | |
7644 | @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as | |
7645 | @samp{0x80000000}, saying that just the sign bit is relevant, the | |
7646 | expression | |
7647 | ||
7648 | @smallexample | |
7649 | (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0)) | |
7650 | @end smallexample | |
7651 | ||
7652 | @noindent | |
7653 | can be converted to | |
7654 | ||
7655 | @smallexample | |
7656 | (ashift:SI @var{x} (const_int @var{n})) | |
7657 | @end smallexample | |
7658 | ||
7659 | @noindent | |
7660 | where @var{n} is the appropriate shift count to move the bit being | |
7661 | tested into the sign bit. | |
7662 | ||
7663 | There is no way to describe a machine that always sets the low-order bit | |
7664 | for a true value, but does not guarantee the value of any other bits, | |
7665 | but we do not know of any machine that has such an instruction. If you | |
7666 | are trying to port GCC to such a machine, include an instruction to | |
7667 | perform a logical-and of the result with 1 in the pattern for the | |
7668 | comparison operators and let us know at @email{gcc@@gcc.gnu.org}. | |
7669 | ||
7670 | Often, a machine will have multiple instructions that obtain a value | |
7671 | from a comparison (or the condition codes). Here are rules to guide the | |
7672 | choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions | |
7673 | to be used: | |
7674 | ||
7675 | @itemize @bullet | |
7676 | @item | |
7677 | Use the shortest sequence that yields a valid definition for | |
7678 | @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to | |
7679 | ``normalize'' the value (convert it to, e.g., 1 or 0) than for the | |
7680 | comparison operators to do so because there may be opportunities to | |
7681 | combine the normalization with other operations. | |
7682 | ||
7683 | @item | |
7684 | For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being | |
7685 | slightly preferred on machines with expensive jumps and 1 preferred on | |
7686 | other machines. | |
7687 | ||
7688 | @item | |
7689 | As a second choice, choose a value of @samp{0x80000001} if instructions | |
7690 | exist that set both the sign and low-order bits but do not define the | |
7691 | others. | |
7692 | ||
7693 | @item | |
7694 | Otherwise, use a value of @samp{0x80000000}. | |
7695 | @end itemize | |
7696 | ||
7697 | Many machines can produce both the value chosen for | |
7698 | @code{STORE_FLAG_VALUE} and its negation in the same number of | |
7699 | instructions. On those machines, you should also define a pattern for | |
7700 | those cases, e.g., one matching | |
7701 | ||
7702 | @smallexample | |
7703 | (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C}))) | |
7704 | @end smallexample | |
7705 | ||
7706 | Some machines can also perform @code{and} or @code{plus} operations on | |
7707 | condition code values with less instructions than the corresponding | |
7708 | @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those | |
7709 | machines, define the appropriate patterns. Use the names @code{incscc} | |
7710 | and @code{decscc}, respectively, for the patterns which perform | |
7711 | @code{plus} or @code{minus} operations on condition code values. See | |
b59688ee | 7712 | @file{rs6000.md} for some examples. The GNU Superoptimizer can be used to |
202d6e5f | 7713 | find such instruction sequences on other machines. |
7714 | ||
7715 | If this macro is not defined, the default value, 1, is used. You need | |
7716 | not define @code{STORE_FLAG_VALUE} if the machine has no store-flag | |
7717 | instructions, or if the value generated by these instructions is 1. | |
7718 | @end defmac | |
7719 | ||
7720 | @defmac FLOAT_STORE_FLAG_VALUE (@var{mode}) | |
7721 | A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is | |
7722 | returned when comparison operators with floating-point results are true. | |
7723 | Define this macro on machines that have comparison operations that return | |
7724 | floating-point values. If there are no such operations, do not define | |
7725 | this macro. | |
7726 | @end defmac | |
7727 | ||
7728 | @defmac VECTOR_STORE_FLAG_VALUE (@var{mode}) | |
7729 | A C expression that gives a rtx representing the nonzero true element | |
7730 | for vector comparisons. The returned rtx should be valid for the inner | |
7731 | mode of @var{mode} which is guaranteed to be a vector mode. Define | |
7732 | this macro on machines that have vector comparison operations that | |
7733 | return a vector result. If there are no such operations, do not define | |
7734 | this macro. Typically, this macro is defined as @code{const1_rtx} or | |
7735 | @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent | |
7736 | the compiler optimizing such vector comparison operations for the | |
7737 | given mode. | |
7738 | @end defmac | |
7739 | ||
7740 | @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value}) | |
7741 | @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value}) | |
7742 | A C expression that indicates whether the architecture defines a value | |
15b474a2 | 7743 | for @code{clz} or @code{ctz} with a zero operand. |
202d6e5f | 7744 | A result of @code{0} indicates the value is undefined. |
7745 | If the value is defined for only the RTL expression, the macro should | |
7746 | evaluate to @code{1}; if the value applies also to the corresponding optab | |
7747 | entry (which is normally the case if it expands directly into | |
15b474a2 | 7748 | the corresponding RTL), then the macro should evaluate to @code{2}. |
202d6e5f | 7749 | In the cases where the value is defined, @var{value} should be set to |
15b474a2 | 7750 | this value. |
202d6e5f | 7751 | |
7752 | If this macro is not defined, the value of @code{clz} or | |
7753 | @code{ctz} at zero is assumed to be undefined. | |
7754 | ||
7755 | This macro must be defined if the target's expansion for @code{ffs} | |
7756 | relies on a particular value to get correct results. Otherwise it | |
7757 | is not necessary, though it may be used to optimize some corner cases, and | |
7758 | to provide a default expansion for the @code{ffs} optab. | |
7759 | ||
7760 | Note that regardless of this macro the ``definedness'' of @code{clz} | |
7761 | and @code{ctz} at zero do @emph{not} extend to the builtin functions | |
7762 | visible to the user. Thus one may be free to adjust the value at will | |
7763 | to match the target expansion of these operations without fear of | |
7764 | breaking the API@. | |
7765 | @end defmac | |
7766 | ||
7767 | @defmac Pmode | |
7768 | An alias for the machine mode for pointers. On most machines, define | |
7769 | this to be the integer mode corresponding to the width of a hardware | |
7770 | pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines. | |
7771 | On some machines you must define this to be one of the partial integer | |
7772 | modes, such as @code{PSImode}. | |
7773 | ||
7774 | The width of @code{Pmode} must be at least as large as the value of | |
7775 | @code{POINTER_SIZE}. If it is not equal, you must define the macro | |
7776 | @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended | |
7777 | to @code{Pmode}. | |
7778 | @end defmac | |
7779 | ||
7780 | @defmac FUNCTION_MODE | |
7781 | An alias for the machine mode used for memory references to functions | |
7782 | being called, in @code{call} RTL expressions. On most CISC machines, | |
15b474a2 | 7783 | where an instruction can begin at any byte address, this should be |
202d6e5f | 7784 | @code{QImode}. On most RISC machines, where all instructions have fixed |
7785 | size and alignment, this should be a mode with the same size and alignment | |
7786 | as the machine instruction words - typically @code{SImode} or @code{HImode}. | |
7787 | @end defmac | |
7788 | ||
7789 | @defmac STDC_0_IN_SYSTEM_HEADERS | |
7790 | In normal operation, the preprocessor expands @code{__STDC__} to the | |
7791 | constant 1, to signify that GCC conforms to ISO Standard C@. On some | |
7792 | hosts, like Solaris, the system compiler uses a different convention, | |
7793 | where @code{__STDC__} is normally 0, but is 1 if the user specifies | |
7794 | strict conformance to the C Standard. | |
7795 | ||
7796 | Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host | |
7797 | convention when processing system header files, but when processing user | |
7798 | files @code{__STDC__} will always expand to 1. | |
7799 | @end defmac | |
7800 | ||
6adc88f8 | 7801 | @hook TARGET_C_PREINCLUDE |
7802 | ||
f9f68d35 | 7803 | @hook TARGET_CXX_IMPLICIT_EXTERN_C |
7804 | ||
202d6e5f | 7805 | @defmac NO_IMPLICIT_EXTERN_C |
7806 | Define this macro if the system header files support C++ as well as C@. | |
7807 | This macro inhibits the usual method of using system header files in | |
7808 | C++, which is to pretend that the file's contents are enclosed in | |
7809 | @samp{extern "C" @{@dots{}@}}. | |
7810 | @end defmac | |
7811 | ||
7812 | @findex #pragma | |
7813 | @findex pragma | |
7814 | @defmac REGISTER_TARGET_PRAGMAS () | |
7815 | Define this macro if you want to implement any target-specific pragmas. | |
7816 | If defined, it is a C expression which makes a series of calls to | |
7817 | @code{c_register_pragma} or @code{c_register_pragma_with_expansion} | |
7818 | for each pragma. The macro may also do any | |
7819 | setup required for the pragmas. | |
7820 | ||
7821 | The primary reason to define this macro is to provide compatibility with | |
7822 | other compilers for the same target. In general, we discourage | |
7823 | definition of target-specific pragmas for GCC@. | |
7824 | ||
7825 | If the pragma can be implemented by attributes then you should consider | |
7826 | defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well. | |
7827 | ||
7828 | Preprocessor macros that appear on pragma lines are not expanded. All | |
7829 | @samp{#pragma} directives that do not match any registered pragma are | |
7830 | silently ignored, unless the user specifies @option{-Wunknown-pragmas}. | |
7831 | @end defmac | |
7832 | ||
7833 | @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *)) | |
7834 | @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *)) | |
7835 | ||
7836 | Each call to @code{c_register_pragma} or | |
7837 | @code{c_register_pragma_with_expansion} establishes one pragma. The | |
7838 | @var{callback} routine will be called when the preprocessor encounters a | |
7839 | pragma of the form | |
7840 | ||
7841 | @smallexample | |
7842 | #pragma [@var{space}] @var{name} @dots{} | |
7843 | @end smallexample | |
7844 | ||
7845 | @var{space} is the case-sensitive namespace of the pragma, or | |
7846 | @code{NULL} to put the pragma in the global namespace. The callback | |
7847 | routine receives @var{pfile} as its first argument, which can be passed | |
7848 | on to cpplib's functions if necessary. You can lex tokens after the | |
7849 | @var{name} by calling @code{pragma_lex}. Tokens that are not read by the | |
7850 | callback will be silently ignored. The end of the line is indicated by | |
7851 | a token of type @code{CPP_EOF}. Macro expansion occurs on the | |
7852 | arguments of pragmas registered with | |
7853 | @code{c_register_pragma_with_expansion} but not on the arguments of | |
7854 | pragmas registered with @code{c_register_pragma}. | |
7855 | ||
7856 | Note that the use of @code{pragma_lex} is specific to the C and C++ | |
7857 | compilers. It will not work in the Java or Fortran compilers, or any | |
7858 | other language compilers for that matter. Thus if @code{pragma_lex} is going | |
7859 | to be called from target-specific code, it must only be done so when | |
7860 | building the C and C++ compilers. This can be done by defining the | |
7861 | variables @code{c_target_objs} and @code{cxx_target_objs} in the | |
7862 | target entry in the @file{config.gcc} file. These variables should name | |
7863 | the target-specific, language-specific object file which contains the | |
7864 | code that uses @code{pragma_lex}. Note it will also be necessary to add a | |
7865 | rule to the makefile fragment pointed to by @code{tmake_file} that shows | |
7866 | how to build this object file. | |
7867 | @end deftypefun | |
7868 | ||
202d6e5f | 7869 | @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION |
3e0e49f2 | 7870 | Define this macro if macros should be expanded in the |
202d6e5f | 7871 | arguments of @samp{#pragma pack}. |
7872 | @end defmac | |
7873 | ||
202d6e5f | 7874 | @defmac TARGET_DEFAULT_PACK_STRUCT |
7875 | If your target requires a structure packing default other than 0 (meaning | |
7876 | the machine default), define this macro to the necessary value (in bytes). | |
7877 | This must be a value that would also be valid to use with | |
7878 | @samp{#pragma pack()} (that is, a small power of two). | |
7879 | @end defmac | |
7880 | ||
7881 | @defmac DOLLARS_IN_IDENTIFIERS | |
7882 | Define this macro to control use of the character @samp{$} in | |
7883 | identifier names for the C family of languages. 0 means @samp{$} is | |
7884 | not allowed by default; 1 means it is allowed. 1 is the default; | |
7885 | there is no need to define this macro in that case. | |
7886 | @end defmac | |
7887 | ||
202d6e5f | 7888 | @defmac INSN_SETS_ARE_DELAYED (@var{insn}) |
7889 | Define this macro as a C expression that is nonzero if it is safe for the | |
7890 | delay slot scheduler to place instructions in the delay slot of @var{insn}, | |
7891 | even if they appear to use a resource set or clobbered in @var{insn}. | |
7892 | @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that | |
7893 | every @code{call_insn} has this behavior. On machines where some @code{insn} | |
7894 | or @code{jump_insn} is really a function call and hence has this behavior, | |
7895 | you should define this macro. | |
7896 | ||
7897 | You need not define this macro if it would always return zero. | |
7898 | @end defmac | |
7899 | ||
7900 | @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn}) | |
7901 | Define this macro as a C expression that is nonzero if it is safe for the | |
7902 | delay slot scheduler to place instructions in the delay slot of @var{insn}, | |
7903 | even if they appear to set or clobber a resource referenced in @var{insn}. | |
7904 | @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where | |
7905 | some @code{insn} or @code{jump_insn} is really a function call and its operands | |
7906 | are registers whose use is actually in the subroutine it calls, you should | |
7907 | define this macro. Doing so allows the delay slot scheduler to move | |
7908 | instructions which copy arguments into the argument registers into the delay | |
7909 | slot of @var{insn}. | |
7910 | ||
7911 | You need not define this macro if it would always return zero. | |
7912 | @end defmac | |
7913 | ||
7914 | @defmac MULTIPLE_SYMBOL_SPACES | |
7915 | Define this macro as a C expression that is nonzero if, in some cases, | |
7916 | global symbols from one translation unit may not be bound to undefined | |
7917 | symbols in another translation unit without user intervention. For | |
7918 | instance, under Microsoft Windows symbols must be explicitly imported | |
7919 | from shared libraries (DLLs). | |
7920 | ||
7921 | You need not define this macro if it would always evaluate to zero. | |
7922 | @end defmac | |
7923 | ||
2af3d775 | 7924 | @hook TARGET_MD_ASM_ADJUST |
202d6e5f | 7925 | |
7926 | @defmac MATH_LIBRARY | |
7927 | Define this macro as a C string constant for the linker argument to link | |
cb22f930 | 7928 | in the system math library, minus the initial @samp{"-l"}, or |
7929 | @samp{""} if the target does not have a | |
202d6e5f | 7930 | separate math library. |
7931 | ||
cb22f930 | 7932 | You need only define this macro if the default of @samp{"m"} is wrong. |
202d6e5f | 7933 | @end defmac |
7934 | ||
7935 | @defmac LIBRARY_PATH_ENV | |
7936 | Define this macro as a C string constant for the environment variable that | |
7937 | specifies where the linker should look for libraries. | |
7938 | ||
7939 | You need only define this macro if the default of @samp{"LIBRARY_PATH"} | |
7940 | is wrong. | |
7941 | @end defmac | |
7942 | ||
7943 | @defmac TARGET_POSIX_IO | |
7944 | Define this macro if the target supports the following POSIX@ file | |
7945 | functions, access, mkdir and file locking with fcntl / F_SETLKW@. | |
7946 | Defining @code{TARGET_POSIX_IO} will enable the test coverage code | |
7947 | to use file locking when exiting a program, which avoids race conditions | |
7948 | if the program has forked. It will also create directories at run-time | |
7949 | for cross-profiling. | |
7950 | @end defmac | |
7951 | ||
7952 | @defmac MAX_CONDITIONAL_EXECUTE | |
7953 | ||
7954 | A C expression for the maximum number of instructions to execute via | |
7955 | conditional execution instructions instead of a branch. A value of | |
7956 | @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and | |
7957 | 1 if it does use cc0. | |
7958 | @end defmac | |
7959 | ||
7960 | @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr}) | |
7961 | Used if the target needs to perform machine-dependent modifications on the | |
7962 | conditionals used for turning basic blocks into conditionally executed code. | |
7963 | @var{ce_info} points to a data structure, @code{struct ce_if_block}, which | |
7964 | contains information about the currently processed blocks. @var{true_expr} | |
7965 | and @var{false_expr} are the tests that are used for converting the | |
7966 | then-block and the else-block, respectively. Set either @var{true_expr} or | |
7967 | @var{false_expr} to a null pointer if the tests cannot be converted. | |
7968 | @end defmac | |
7969 | ||
7970 | @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr}) | |
7971 | Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated | |
7972 | if-statements into conditions combined by @code{and} and @code{or} operations. | |
7973 | @var{bb} contains the basic block that contains the test that is currently | |
7974 | being processed and about to be turned into a condition. | |
7975 | @end defmac | |
7976 | ||
7977 | @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn}) | |
7978 | A C expression to modify the @var{PATTERN} of an @var{INSN} that is to | |
7979 | be converted to conditional execution format. @var{ce_info} points to | |
7980 | a data structure, @code{struct ce_if_block}, which contains information | |
7981 | about the currently processed blocks. | |
7982 | @end defmac | |
7983 | ||
7984 | @defmac IFCVT_MODIFY_FINAL (@var{ce_info}) | |
7985 | A C expression to perform any final machine dependent modifications in | |
7986 | converting code to conditional execution. The involved basic blocks | |
7987 | can be found in the @code{struct ce_if_block} structure that is pointed | |
7988 | to by @var{ce_info}. | |
7989 | @end defmac | |
7990 | ||
7991 | @defmac IFCVT_MODIFY_CANCEL (@var{ce_info}) | |
7992 | A C expression to cancel any machine dependent modifications in | |
7993 | converting code to conditional execution. The involved basic blocks | |
7994 | can be found in the @code{struct ce_if_block} structure that is pointed | |
7995 | to by @var{ce_info}. | |
7996 | @end defmac | |
7997 | ||
e2ca32a4 | 7998 | @defmac IFCVT_MACHDEP_INIT (@var{ce_info}) |
7999 | A C expression to initialize any machine specific data for if-conversion | |
8000 | of the if-block in the @code{struct ce_if_block} structure that is pointed | |
8001 | to by @var{ce_info}. | |
202d6e5f | 8002 | @end defmac |
8003 | ||
8004 | @hook TARGET_MACHINE_DEPENDENT_REORG | |
202d6e5f | 8005 | |
8006 | @hook TARGET_INIT_BUILTINS | |
202d6e5f | 8007 | |
8008 | @hook TARGET_BUILTIN_DECL | |
202d6e5f | 8009 | |
8010 | @hook TARGET_EXPAND_BUILTIN | |
8011 | ||
058a1b7a | 8012 | @hook TARGET_BUILTIN_CHKP_FUNCTION |
8013 | @hook TARGET_CHKP_BOUND_TYPE | |
8014 | @hook TARGET_CHKP_BOUND_MODE | |
8015 | @hook TARGET_CHKP_MAKE_BOUNDS_CONSTANT | |
8016 | @hook TARGET_CHKP_INITIALIZE_BOUNDS | |
8017 | ||
00a1fec8 | 8018 | @hook TARGET_RESOLVE_OVERLOADED_BUILTIN |
202d6e5f | 8019 | |
bf084459 | 8020 | @hook TARGET_FOLD_BUILTIN |
202d6e5f | 8021 | |
29cad3b6 | 8022 | @hook TARGET_GIMPLE_FOLD_BUILTIN |
8023 | ||
cc8ef84f | 8024 | @hook TARGET_COMPARE_VERSION_PRIORITY |
cc8ef84f | 8025 | |
8026 | @hook TARGET_GET_FUNCTION_VERSIONS_DISPATCHER | |
cc8ef84f | 8027 | |
8028 | @hook TARGET_GENERATE_VERSION_DISPATCHER_BODY | |
cc8ef84f | 8029 | |
a04a7bec | 8030 | @hook TARGET_CAN_USE_DOLOOP_P |
8031 | ||
202d6e5f | 8032 | @hook TARGET_INVALID_WITHIN_DOLOOP |
8033 | ||
2a29bc01 | 8034 | @hook TARGET_LEGITIMATE_COMBINED_INSN |
8035 | ||
c7b4d9b2 | 8036 | @hook TARGET_CAN_FOLLOW_JUMP |
8037 | ||
202d6e5f | 8038 | @hook TARGET_COMMUTATIVE_P |
202d6e5f | 8039 | |
8040 | @hook TARGET_ALLOCATE_INITIAL_VALUE | |
8041 | ||
202d6e5f | 8042 | @hook TARGET_UNSPEC_MAY_TRAP_P |
202d6e5f | 8043 | |
8044 | @hook TARGET_SET_CURRENT_FUNCTION | |
202d6e5f | 8045 | |
8046 | @defmac TARGET_OBJECT_SUFFIX | |
8047 | Define this macro to be a C string representing the suffix for object | |
8048 | files on your target machine. If you do not define this macro, GCC will | |
8049 | use @samp{.o} as the suffix for object files. | |
8050 | @end defmac | |
8051 | ||
8052 | @defmac TARGET_EXECUTABLE_SUFFIX | |
8053 | Define this macro to be a C string representing the suffix to be | |
8054 | automatically added to executable files on your target machine. If you | |
8055 | do not define this macro, GCC will use the null string as the suffix for | |
8056 | executable files. | |
8057 | @end defmac | |
8058 | ||
8059 | @defmac COLLECT_EXPORT_LIST | |
8060 | If defined, @code{collect2} will scan the individual object files | |
8061 | specified on its command line and create an export list for the linker. | |
8062 | Define this macro for systems like AIX, where the linker discards | |
8063 | object files that are not referenced from @code{main} and uses export | |
8064 | lists. | |
8065 | @end defmac | |
8066 | ||
8067 | @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl}) | |
8068 | Define this macro to a C expression representing a variant of the | |
8069 | method call @var{mdecl}, if Java Native Interface (JNI) methods | |
8070 | must be invoked differently from other methods on your target. | |
8071 | For example, on 32-bit Microsoft Windows, JNI methods must be invoked using | |
8072 | the @code{stdcall} calling convention and this macro is then | |
8073 | defined as this expression: | |
8074 | ||
8075 | @smallexample | |
8076 | build_type_attribute_variant (@var{mdecl}, | |
8077 | build_tree_list | |
8078 | (get_identifier ("stdcall"), | |
8079 | NULL)) | |
8080 | @end smallexample | |
8081 | @end defmac | |
8082 | ||
8083 | @hook TARGET_CANNOT_MODIFY_JUMPS_P | |
202d6e5f | 8084 | |
8085 | @hook TARGET_BRANCH_TARGET_REGISTER_CLASS | |
202d6e5f | 8086 | |
8087 | @hook TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED | |
202d6e5f | 8088 | |
8089 | @hook TARGET_HAVE_CONDITIONAL_EXECUTION | |
202d6e5f | 8090 | |
f9fd4edc | 8091 | @hook TARGET_GEN_CCMP_FIRST |
8092 | ||
8093 | @hook TARGET_GEN_CCMP_NEXT | |
8094 | ||
202d6e5f | 8095 | @hook TARGET_LOOP_UNROLL_ADJUST |
202d6e5f | 8096 | |
8097 | @defmac POWI_MAX_MULTS | |
8098 | If defined, this macro is interpreted as a signed integer C expression | |
8099 | that specifies the maximum number of floating point multiplications | |
8100 | that should be emitted when expanding exponentiation by an integer | |
8101 | constant inline. When this value is defined, exponentiation requiring | |
8102 | more than this number of multiplications is implemented by calling the | |
8103 | system library's @code{pow}, @code{powf} or @code{powl} routines. | |
8104 | The default value places no upper bound on the multiplication count. | |
8105 | @end defmac | |
8106 | ||
8107 | @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc}) | |
8108 | This target hook should register any extra include files for the | |
8109 | target. The parameter @var{stdinc} indicates if normal include files | |
8110 | are present. The parameter @var{sysroot} is the system root directory. | |
8111 | The parameter @var{iprefix} is the prefix for the gcc directory. | |
8112 | @end deftypefn | |
8113 | ||
8114 | @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc}) | |
8115 | This target hook should register any extra include files for the | |
8116 | target before any standard headers. The parameter @var{stdinc} | |
8117 | indicates if normal include files are present. The parameter | |
8118 | @var{sysroot} is the system root directory. The parameter | |
8119 | @var{iprefix} is the prefix for the gcc directory. | |
8120 | @end deftypefn | |
8121 | ||
8122 | @deftypefn Macro void TARGET_OPTF (char *@var{path}) | |
8123 | This target hook should register special include paths for the target. | |
8124 | The parameter @var{path} is the include to register. On Darwin | |
8125 | systems, this is used for Framework includes, which have semantics | |
8126 | that are different from @option{-I}. | |
8127 | @end deftypefn | |
8128 | ||
8129 | @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl}) | |
8130 | This target macro returns @code{true} if it is safe to use a local alias | |
8131 | for a virtual function @var{fndecl} when constructing thunks, | |
8132 | @code{false} otherwise. By default, the macro returns @code{true} for all | |
8133 | functions, if a target supports aliases (i.e.@: defines | |
8134 | @code{ASM_OUTPUT_DEF}), @code{false} otherwise, | |
8135 | @end defmac | |
8136 | ||
8137 | @defmac TARGET_FORMAT_TYPES | |
8138 | If defined, this macro is the name of a global variable containing | |
8139 | target-specific format checking information for the @option{-Wformat} | |
8140 | option. The default is to have no target-specific format checks. | |
8141 | @end defmac | |
8142 | ||
8143 | @defmac TARGET_N_FORMAT_TYPES | |
8144 | If defined, this macro is the number of entries in | |
8145 | @code{TARGET_FORMAT_TYPES}. | |
8146 | @end defmac | |
8147 | ||
8148 | @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES | |
8149 | If defined, this macro is the name of a global variable containing | |
8150 | target-specific format overrides for the @option{-Wformat} option. The | |
8151 | default is to have no target-specific format overrides. If defined, | |
8152 | @code{TARGET_FORMAT_TYPES} must be defined, too. | |
8153 | @end defmac | |
8154 | ||
8155 | @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT | |
8156 | If defined, this macro specifies the number of entries in | |
8157 | @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}. | |
8158 | @end defmac | |
8159 | ||
8160 | @defmac TARGET_OVERRIDES_FORMAT_INIT | |
8161 | If defined, this macro specifies the optional initialization | |
8162 | routine for target specific customizations of the system printf | |
8163 | and scanf formatter settings. | |
8164 | @end defmac | |
8165 | ||
202d6e5f | 8166 | @hook TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN |
202d6e5f | 8167 | |
8168 | @hook TARGET_INVALID_CONVERSION | |
202d6e5f | 8169 | |
8170 | @hook TARGET_INVALID_UNARY_OP | |
202d6e5f | 8171 | |
8172 | @hook TARGET_INVALID_BINARY_OP | |
202d6e5f | 8173 | |
8174 | @hook TARGET_INVALID_PARAMETER_TYPE | |
202d6e5f | 8175 | |
8176 | @hook TARGET_INVALID_RETURN_TYPE | |
202d6e5f | 8177 | |
8178 | @hook TARGET_PROMOTED_TYPE | |
202d6e5f | 8179 | |
8180 | @hook TARGET_CONVERT_TO_TYPE | |
202d6e5f | 8181 | |
8182 | @defmac TARGET_USE_JCR_SECTION | |
8183 | This macro determines whether to use the JCR section to register Java | |
8184 | classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both | |
8185 | SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0. | |
8186 | @end defmac | |
8187 | ||
8188 | @defmac OBJC_JBLEN | |
8189 | This macro determines the size of the objective C jump buffer for the | |
8190 | NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value. | |
8191 | @end defmac | |
8192 | ||
8193 | @defmac LIBGCC2_UNWIND_ATTRIBUTE | |
8194 | Define this macro if any target-specific attributes need to be attached | |
15b474a2 | 8195 | to the functions in @file{libgcc} that provide low-level support for |
202d6e5f | 8196 | call stack unwinding. It is used in declarations in @file{unwind-generic.h} |
8197 | and the associated definitions of those functions. | |
8198 | @end defmac | |
8199 | ||
8200 | @hook TARGET_UPDATE_STACK_BOUNDARY | |
202d6e5f | 8201 | |
8202 | @hook TARGET_GET_DRAP_RTX | |
202d6e5f | 8203 | |
8204 | @hook TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS | |
202d6e5f | 8205 | |
8206 | @hook TARGET_CONST_ANCHOR | |
df1680c8 | 8207 | |
7ad5fd20 | 8208 | @hook TARGET_ASAN_SHADOW_OFFSET |
8209 | ||
7f738025 | 8210 | @hook TARGET_MEMMODEL_CHECK |
7f738025 | 8211 | |
df1680c8 | 8212 | @hook TARGET_ATOMIC_TEST_AND_SET_TRUEVAL |
df9f2e40 | 8213 | |
8214 | @hook TARGET_HAS_IFUNC_P | |
e913b5cd | 8215 | |
1e6e32b6 | 8216 | @hook TARGET_ATOMIC_ALIGN_FOR_MODE |
8217 | ||
b560fabd | 8218 | @hook TARGET_ATOMIC_ASSIGN_EXPAND_FENV |
55af3bae | 8219 | |
dccabdd1 | 8220 | @hook TARGET_RECORD_OFFLOAD_SYMBOL |
8221 | ||
38e21583 | 8222 | @hook TARGET_OFFLOAD_OPTIONS |
8223 | ||
e913b5cd | 8224 | @defmac TARGET_SUPPORTS_WIDE_INT |
8225 | ||
8226 | On older ports, large integers are stored in @code{CONST_DOUBLE} rtl | |
a342dbb2 | 8227 | objects. Newer ports define @code{TARGET_SUPPORTS_WIDE_INT} to be nonzero |
8228 | to indicate that large integers are stored in | |
e913b5cd | 8229 | @code{CONST_WIDE_INT} rtl objects. The @code{CONST_WIDE_INT} allows |
8230 | very large integer constants to be represented. @code{CONST_DOUBLE} | |
a342dbb2 | 8231 | is limited to twice the size of the host's @code{HOST_WIDE_INT} |
e913b5cd | 8232 | representation. |
8233 | ||
8234 | Converting a port mostly requires looking for the places where | |
a342dbb2 | 8235 | @code{CONST_DOUBLE}s are used with @code{VOIDmode} and replacing that |
e913b5cd | 8236 | code with code that accesses @code{CONST_WIDE_INT}s. @samp{"grep -i |
8237 | const_double"} at the port level gets you to 95% of the changes that | |
8238 | need to be made. There are a few places that require a deeper look. | |
8239 | ||
8240 | @itemize @bullet | |
8241 | @item | |
8242 | There is no equivalent to @code{hval} and @code{lval} for | |
8243 | @code{CONST_WIDE_INT}s. This would be difficult to express in the md | |
8244 | language since there are a variable number of elements. | |
8245 | ||
8246 | Most ports only check that @code{hval} is either 0 or -1 to see if the | |
8247 | value is small. As mentioned above, this will no longer be necessary | |
8248 | since small constants are always @code{CONST_INT}. Of course there | |
8249 | are still a few exceptions, the alpha's constraint used by the zap | |
8250 | instruction certainly requires careful examination by C code. | |
8251 | However, all the current code does is pass the hval and lval to C | |
8252 | code, so evolving the c code to look at the @code{CONST_WIDE_INT} is | |
8253 | not really a large change. | |
8254 | ||
8255 | @item | |
8256 | Because there is no standard template that ports use to materialize | |
8257 | constants, there is likely to be some futzing that is unique to each | |
8258 | port in this code. | |
8259 | ||
8260 | @item | |
8261 | The rtx costs may have to be adjusted to properly account for larger | |
8262 | constants that are represented as @code{CONST_WIDE_INT}. | |
8263 | @end itemize | |
8264 | ||
a342dbb2 | 8265 | All and all it does not take long to convert ports that the |
e913b5cd | 8266 | maintainer is familiar with. |
8267 | ||
8268 | @end defmac |