]>
Commit | Line | Data |
---|---|---|
c58b209a | 1 | @c Copyright (C) 1988,1989,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001, |
c5c367ac | 2 | @c 2002, 2003, 2004, 2005 Free Software Foundation, Inc. |
feca2ed3 JW |
3 | @c This is part of the GCC manual. |
4 | @c For copying conditions, see the file gcc.texi. | |
5 | ||
6 | @node Target Macros | |
672a6f42 | 7 | @chapter Target Description Macros and Functions |
feca2ed3 JW |
8 | @cindex machine description macros |
9 | @cindex target description macros | |
10 | @cindex macros, target description | |
11 | @cindex @file{tm.h} macros | |
12 | ||
13 | In addition to the file @file{@var{machine}.md}, a machine description | |
14 | includes a C header file conventionally given the name | |
672a6f42 NB |
15 | @file{@var{machine}.h} and a C source file named @file{@var{machine}.c}. |
16 | The header file defines numerous macros that convey the information | |
17 | about the target machine that does not fit into the scheme of the | |
18 | @file{.md} file. The file @file{tm.h} should be a link to | |
19 | @file{@var{machine}.h}. The header file @file{config.h} includes | |
20 | @file{tm.h} and most compiler source files include @file{config.h}. The | |
648c546a | 21 | source file defines a variable @code{targetm}, which is a structure |
672a6f42 NB |
22 | containing pointers to functions and data relating to the target |
23 | machine. @file{@var{machine}.c} should also contain their definitions, | |
24 | if they are not defined elsewhere in GCC, and other functions called | |
25 | through the macros defined in the @file{.h} file. | |
feca2ed3 JW |
26 | |
27 | @menu | |
648c546a | 28 | * Target Structure:: The @code{targetm} variable. |
feca2ed3 | 29 | * Driver:: Controlling how the driver runs the compilation passes. |
630d3d5a | 30 | * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}. |
414c4dc4 | 31 | * Per-Function Data:: Defining data structures for per-function information. |
feca2ed3 JW |
32 | * Storage Layout:: Defining sizes and alignments of data. |
33 | * Type Layout:: Defining sizes and properties of basic user data types. | |
34 | * Registers:: Naming and describing the hardware registers. | |
35 | * Register Classes:: Defining the classes of hardware registers. | |
36 | * Stack and Calling:: Defining which way the stack grows and by how much. | |
37 | * Varargs:: Defining the varargs macros. | |
38 | * Trampolines:: Code set up at run time to enter a nested function. | |
39 | * Library Calls:: Controlling how library routines are implicitly called. | |
40 | * Addressing Modes:: Defining addressing modes valid for memory operands. | |
41 | * Condition Code:: Defining how insns update the condition code. | |
42 | * Costs:: Defining relative costs of different operations. | |
c237e94a | 43 | * Scheduling:: Adjusting the behavior of the instruction scheduler. |
feca2ed3 JW |
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. | |
b216cd4a | 48 | * Floating Point:: Handling floating point for cross-compilers. |
9f09b1f2 | 49 | * Mode Switching:: Insertion of mode-switching instructions. |
91d231cb | 50 | * Target Attributes:: Defining target-specific uses of @code{__attribute__}. |
d604bca3 | 51 | * MIPS Coprocessors:: MIPS coprocessor support and how to customize it. |
7bb1ad93 | 52 | * PCH Target:: Validity checking for precompiled headers. |
4185ae53 | 53 | * C++ ABI:: Controlling C++ ABI changes. |
feca2ed3 JW |
54 | * Misc:: Everything else. |
55 | @end menu | |
56 | ||
672a6f42 | 57 | @node Target Structure |
648c546a | 58 | @section The Global @code{targetm} Variable |
672a6f42 NB |
59 | @cindex target hooks |
60 | @cindex target functions | |
61 | ||
f6897b10 SS |
62 | @deftypevar {struct gcc_target} targetm |
63 | The target @file{.c} file must define the global @code{targetm} variable | |
672a6f42 NB |
64 | which contains pointers to functions and data relating to the target |
65 | machine. The variable is declared in @file{target.h}; | |
66 | @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is | |
67 | used to initialize the variable, and macros for the default initializers | |
68 | for elements of the structure. The @file{.c} file should override those | |
69 | macros for which the default definition is inappropriate. For example: | |
70 | @smallexample | |
71 | #include "target.h" | |
72 | #include "target-def.h" | |
73 | ||
74 | /* @r{Initialize the GCC target structure.} */ | |
75 | ||
91d231cb JM |
76 | #undef TARGET_COMP_TYPE_ATTRIBUTES |
77 | #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes | |
672a6f42 | 78 | |
f6897b10 | 79 | struct gcc_target targetm = TARGET_INITIALIZER; |
672a6f42 NB |
80 | @end smallexample |
81 | @end deftypevar | |
82 | ||
83 | Where a macro should be defined in the @file{.c} file in this manner to | |
648c546a | 84 | form part of the @code{targetm} structure, it is documented below as a |
672a6f42 NB |
85 | ``Target Hook'' with a prototype. Many macros will change in future |
86 | from being defined in the @file{.h} file to being part of the | |
648c546a | 87 | @code{targetm} structure. |
672a6f42 | 88 | |
feca2ed3 JW |
89 | @node Driver |
90 | @section Controlling the Compilation Driver, @file{gcc} | |
91 | @cindex driver | |
92 | @cindex controlling the compilation driver | |
93 | ||
94 | @c prevent bad page break with this line | |
95 | You can control the compilation driver. | |
96 | ||
a2c4f8e0 | 97 | @defmac SWITCH_TAKES_ARG (@var{char}) |
630d3d5a | 98 | A C expression which determines whether the option @option{-@var{char}} |
feca2ed3 JW |
99 | takes arguments. The value should be the number of arguments that |
100 | option takes--zero, for many options. | |
101 | ||
102 | By default, this macro is defined as | |
103 | @code{DEFAULT_SWITCH_TAKES_ARG}, which handles the standard options | |
104 | properly. You need not define @code{SWITCH_TAKES_ARG} unless you | |
105 | wish to add additional options which take arguments. Any redefinition | |
106 | should call @code{DEFAULT_SWITCH_TAKES_ARG} and then check for | |
107 | additional options. | |
a2c4f8e0 | 108 | @end defmac |
feca2ed3 | 109 | |
a2c4f8e0 | 110 | @defmac WORD_SWITCH_TAKES_ARG (@var{name}) |
630d3d5a | 111 | A C expression which determines whether the option @option{-@var{name}} |
feca2ed3 JW |
112 | takes arguments. The value should be the number of arguments that |
113 | option takes--zero, for many options. This macro rather than | |
114 | @code{SWITCH_TAKES_ARG} is used for multi-character option names. | |
115 | ||
116 | By default, this macro is defined as | |
117 | @code{DEFAULT_WORD_SWITCH_TAKES_ARG}, which handles the standard options | |
118 | properly. You need not define @code{WORD_SWITCH_TAKES_ARG} unless you | |
119 | wish to add additional options which take arguments. Any redefinition | |
120 | should call @code{DEFAULT_WORD_SWITCH_TAKES_ARG} and then check for | |
121 | additional options. | |
a2c4f8e0 | 122 | @end defmac |
feca2ed3 | 123 | |
a2c4f8e0 | 124 | @defmac SWITCH_CURTAILS_COMPILATION (@var{char}) |
630d3d5a | 125 | A C expression which determines whether the option @option{-@var{char}} |
88117d44 | 126 | stops compilation before the generation of an executable. The value is |
df2a54e9 | 127 | boolean, nonzero if the option does stop an executable from being |
88117d44 NC |
128 | generated, zero otherwise. |
129 | ||
130 | By default, this macro is defined as | |
131 | @code{DEFAULT_SWITCH_CURTAILS_COMPILATION}, which handles the standard | |
132 | options properly. You need not define | |
133 | @code{SWITCH_CURTAILS_COMPILATION} unless you wish to add additional | |
134 | options which affect the generation of an executable. Any redefinition | |
135 | should call @code{DEFAULT_SWITCH_CURTAILS_COMPILATION} and then check | |
136 | for additional options. | |
a2c4f8e0 | 137 | @end defmac |
88117d44 | 138 | |
a2c4f8e0 | 139 | @defmac SWITCHES_NEED_SPACES |
feca2ed3 JW |
140 | A string-valued C expression which enumerates the options for which |
141 | the linker needs a space between the option and its argument. | |
142 | ||
143 | If this macro is not defined, the default value is @code{""}. | |
a2c4f8e0 | 144 | @end defmac |
feca2ed3 | 145 | |
a2c4f8e0 | 146 | @defmac TARGET_OPTION_TRANSLATE_TABLE |
0259b07a DD |
147 | If defined, a list of pairs of strings, the first of which is a |
148 | potential command line target to the @file{gcc} driver program, and the | |
149 | second of which is a space-separated (tabs and other whitespace are not | |
150 | supported) list of options with which to replace the first option. The | |
151 | target defining this list is responsible for assuring that the results | |
152 | are valid. Replacement options may not be the @code{--opt} style, they | |
153 | must be the @code{-opt} style. It is the intention of this macro to | |
154 | provide a mechanism for substitution that affects the multilibs chosen, | |
155 | such as one option that enables many options, some of which select | |
4ec7afd7 KH |
156 | multilibs. Example nonsensical definition, where @option{-malt-abi}, |
157 | @option{-EB}, and @option{-mspoo} cause different multilibs to be chosen: | |
0259b07a | 158 | |
478c9e72 | 159 | @smallexample |
0259b07a DD |
160 | #define TARGET_OPTION_TRANSLATE_TABLE \ |
161 | @{ "-fast", "-march=fast-foo -malt-abi -I/usr/fast-foo" @}, \ | |
162 | @{ "-compat", "-EB -malign=4 -mspoo" @} | |
478c9e72 | 163 | @end smallexample |
a2c4f8e0 | 164 | @end defmac |
0259b07a | 165 | |
a2c4f8e0 | 166 | @defmac DRIVER_SELF_SPECS |
db36994b RS |
167 | A list of specs for the driver itself. It should be a suitable |
168 | initializer for an array of strings, with no surrounding braces. | |
169 | ||
3bd6d4c4 AO |
170 | The driver applies these specs to its own command line between loading |
171 | default @file{specs} files (but not command-line specified ones) and | |
172 | choosing the multilib directory or running any subcommands. It | |
173 | applies them in the order given, so each spec can depend on the | |
174 | options added by earlier ones. It is also possible to remove options | |
175 | using @samp{%<@var{option}} in the usual way. | |
db36994b RS |
176 | |
177 | This macro can be useful when a port has several interdependent target | |
178 | options. It provides a way of standardizing the command line so | |
179 | that the other specs are easier to write. | |
180 | ||
181 | Do not define this macro if it does not need to do anything. | |
a2c4f8e0 | 182 | @end defmac |
db36994b | 183 | |
a2c4f8e0 | 184 | @defmac OPTION_DEFAULT_SPECS |
7816bea0 DJ |
185 | A list of specs used to support configure-time default options (i.e.@: |
186 | @option{--with} options) in the driver. It should be a suitable initializer | |
187 | for an array of structures, each containing two strings, without the | |
188 | outermost pair of surrounding braces. | |
189 | ||
190 | The first item in the pair is the name of the default. This must match | |
191 | the code in @file{config.gcc} for the target. The second item is a spec | |
192 | to apply if a default with this name was specified. The string | |
193 | @samp{%(VALUE)} in the spec will be replaced by the value of the default | |
194 | everywhere it occurs. | |
195 | ||
196 | The driver will apply these specs to its own command line between loading | |
197 | default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using | |
198 | the same mechanism as @code{DRIVER_SELF_SPECS}. | |
199 | ||
200 | Do not define this macro if it does not need to do anything. | |
a2c4f8e0 | 201 | @end defmac |
7816bea0 | 202 | |
a2c4f8e0 | 203 | @defmac CPP_SPEC |
a3a15b4d | 204 | A C string constant that tells the GCC driver program options to |
161d7b59 JM |
205 | pass to CPP@. It can also specify how to translate options you |
206 | give to GCC into options for GCC to pass to the CPP@. | |
feca2ed3 JW |
207 | |
208 | Do not define this macro if it does not need to do anything. | |
a2c4f8e0 | 209 | @end defmac |
feca2ed3 | 210 | |
a2c4f8e0 | 211 | @defmac CPLUSPLUS_CPP_SPEC |
a9374841 | 212 | This macro is just like @code{CPP_SPEC}, but is used for C++, rather |
161d7b59 | 213 | than C@. If you do not define this macro, then the value of |
a9374841 | 214 | @code{CPP_SPEC} (if any) will be used instead. |
a2c4f8e0 | 215 | @end defmac |
a9374841 | 216 | |
a2c4f8e0 | 217 | @defmac CC1_SPEC |
a3a15b4d | 218 | A C string constant that tells the GCC driver program options to |
66519c70 JL |
219 | pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language |
220 | front ends. | |
a3a15b4d | 221 | It can also specify how to translate options you give to GCC into options |
630d3d5a | 222 | for GCC to pass to front ends. |
feca2ed3 JW |
223 | |
224 | Do not define this macro if it does not need to do anything. | |
a2c4f8e0 | 225 | @end defmac |
feca2ed3 | 226 | |
a2c4f8e0 | 227 | @defmac CC1PLUS_SPEC |
a3a15b4d | 228 | A C string constant that tells the GCC driver program options to |
feca2ed3 | 229 | pass to @code{cc1plus}. It can also specify how to translate options you |
a3a15b4d | 230 | give to GCC into options for GCC to pass to the @code{cc1plus}. |
feca2ed3 JW |
231 | |
232 | Do not define this macro if it does not need to do anything. | |
1d96e5b4 FF |
233 | Note that everything defined in CC1_SPEC is already passed to |
234 | @code{cc1plus} so there is no need to duplicate the contents of | |
161d7b59 | 235 | CC1_SPEC in CC1PLUS_SPEC@. |
a2c4f8e0 | 236 | @end defmac |
feca2ed3 | 237 | |
a2c4f8e0 | 238 | @defmac ASM_SPEC |
a3a15b4d | 239 | A C string constant that tells the GCC driver program options to |
feca2ed3 | 240 | pass to the assembler. It can also specify how to translate options |
a3a15b4d | 241 | you give to GCC into options for GCC to pass to the assembler. |
feca2ed3 JW |
242 | See the file @file{sun3.h} for an example of this. |
243 | ||
244 | Do not define this macro if it does not need to do anything. | |
a2c4f8e0 | 245 | @end defmac |
feca2ed3 | 246 | |
a2c4f8e0 | 247 | @defmac ASM_FINAL_SPEC |
a3a15b4d | 248 | A C string constant that tells the GCC driver program how to |
feca2ed3 JW |
249 | run any programs which cleanup after the normal assembler. |
250 | Normally, this is not needed. See the file @file{mips.h} for | |
251 | an example of this. | |
252 | ||
253 | Do not define this macro if it does not need to do anything. | |
a2c4f8e0 | 254 | @end defmac |
feca2ed3 | 255 | |
a2c4f8e0 | 256 | @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT |
4977bab6 ZW |
257 | Define this macro, with no value, if the driver should give the assembler |
258 | an argument consisting of a single dash, @option{-}, to instruct it to | |
259 | read from its standard input (which will be a pipe connected to the | |
260 | output of the compiler proper). This argument is given after any | |
261 | @option{-o} option specifying the name of the output file. | |
262 | ||
263 | If you do not define this macro, the assembler is assumed to read its | |
264 | standard input if given no non-option arguments. If your assembler | |
265 | cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct; | |
266 | see @file{mips.h} for instance. | |
a2c4f8e0 | 267 | @end defmac |
4977bab6 | 268 | |
a2c4f8e0 | 269 | @defmac LINK_SPEC |
a3a15b4d | 270 | A C string constant that tells the GCC driver program options to |
feca2ed3 | 271 | pass to the linker. It can also specify how to translate options you |
a3a15b4d | 272 | give to GCC into options for GCC to pass to the linker. |
feca2ed3 JW |
273 | |
274 | Do not define this macro if it does not need to do anything. | |
a2c4f8e0 | 275 | @end defmac |
feca2ed3 | 276 | |
a2c4f8e0 | 277 | @defmac LIB_SPEC |
feca2ed3 JW |
278 | Another C string constant used much like @code{LINK_SPEC}. The difference |
279 | between the two is that @code{LIB_SPEC} is used at the end of the | |
280 | command given to the linker. | |
281 | ||
282 | If this macro is not defined, a default is provided that | |
283 | loads the standard C library from the usual place. See @file{gcc.c}. | |
a2c4f8e0 | 284 | @end defmac |
feca2ed3 | 285 | |
a2c4f8e0 | 286 | @defmac LIBGCC_SPEC |
a3a15b4d | 287 | Another C string constant that tells the GCC driver program |
feca2ed3 JW |
288 | how and when to place a reference to @file{libgcc.a} into the |
289 | linker command line. This constant is placed both before and after | |
290 | the value of @code{LIB_SPEC}. | |
291 | ||
a3a15b4d | 292 | If this macro is not defined, the GCC driver provides a default that |
630d3d5a | 293 | passes the string @option{-lgcc} to the linker. |
a2c4f8e0 | 294 | @end defmac |
feca2ed3 | 295 | |
328163dc MA |
296 | @defmac REAL_LIBGCC_SPEC |
297 | By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the | |
298 | @code{LIBGCC_SPEC} is not directly used by the driver program but is | |
299 | instead modified to refer to different versions of @file{libgcc.a} | |
4ec7afd7 KH |
300 | depending on the values of the command line flags @option{-static}, |
301 | @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On | |
328163dc MA |
302 | targets where these modifications are inappropriate, define |
303 | @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the | |
304 | driver how to place a reference to @file{libgcc} on the link command | |
305 | line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified. | |
306 | @end defmac | |
307 | ||
743eeb5a AM |
308 | @defmac USE_LD_AS_NEEDED |
309 | A macro that controls the modifications to @code{LIBGCC_SPEC} | |
310 | mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be | |
311 | generated that uses --as-needed and the shared libgcc in place of the | |
312 | static exception handler library, when linking without any of | |
313 | @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}. | |
314 | @end defmac | |
315 | ||
316 | @defmac LINK_EH_SPEC | |
317 | If defined, this C string constant is added to @code{LINK_SPEC}. | |
318 | When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects | |
319 | the modifications to @code{LIBGCC_SPEC} mentioned in | |
320 | @code{REAL_LIBGCC_SPEC}. | |
321 | @end defmac | |
322 | ||
a2c4f8e0 | 323 | @defmac STARTFILE_SPEC |
feca2ed3 JW |
324 | Another C string constant used much like @code{LINK_SPEC}. The |
325 | difference between the two is that @code{STARTFILE_SPEC} is used at | |
326 | the very beginning of the command given to the linker. | |
327 | ||
328 | If this macro is not defined, a default is provided that loads the | |
329 | standard C startup file from the usual place. See @file{gcc.c}. | |
a2c4f8e0 | 330 | @end defmac |
feca2ed3 | 331 | |
a2c4f8e0 | 332 | @defmac ENDFILE_SPEC |
feca2ed3 JW |
333 | Another C string constant used much like @code{LINK_SPEC}. The |
334 | difference between the two is that @code{ENDFILE_SPEC} is used at | |
335 | the very end of the command given to the linker. | |
336 | ||
337 | Do not define this macro if it does not need to do anything. | |
a2c4f8e0 | 338 | @end defmac |
feca2ed3 | 339 | |
a2c4f8e0 | 340 | @defmac THREAD_MODEL_SPEC |
008355a6 AO |
341 | GCC @code{-v} will print the thread model GCC was configured to use. |
342 | However, this doesn't work on platforms that are multilibbed on thread | |
343 | models, such as AIX 4.3. On such platforms, define | |
344 | @code{THREAD_MODEL_SPEC} such that it evaluates to a string without | |
345 | blanks that names one of the recognized thread models. @code{%*}, the | |
346 | default value of this macro, will expand to the value of | |
347 | @code{thread_file} set in @file{config.gcc}. | |
a2c4f8e0 | 348 | @end defmac |
008355a6 | 349 | |
a2c4f8e0 | 350 | @defmac SYSROOT_SUFFIX_SPEC |
73774972 EC |
351 | Define this macro to add a suffix to the target sysroot when GCC is |
352 | configured with a sysroot. This will cause GCC to search for usr/lib, | |
353 | et al, within sysroot+suffix. | |
a2c4f8e0 | 354 | @end defmac |
e7f13528 | 355 | |
a2c4f8e0 | 356 | @defmac SYSROOT_HEADERS_SUFFIX_SPEC |
73774972 EC |
357 | Define this macro to add a headers_suffix to the target sysroot when |
358 | GCC is configured with a sysroot. This will cause GCC to pass the | |
f4314bb6 | 359 | updated sysroot+headers_suffix to CPP, causing it to search for |
e7f13528 | 360 | usr/include, et al, within sysroot+headers_suffix. |
a2c4f8e0 | 361 | @end defmac |
e7f13528 | 362 | |
a2c4f8e0 | 363 | @defmac EXTRA_SPECS |
feca2ed3 JW |
364 | Define this macro to provide additional specifications to put in the |
365 | @file{specs} file that can be used in various specifications like | |
366 | @code{CC1_SPEC}. | |
367 | ||
368 | The definition should be an initializer for an array of structures, | |
369 | containing a string constant, that defines the specification name, and a | |
370 | string constant that provides the specification. | |
371 | ||
372 | Do not define this macro if it does not need to do anything. | |
373 | ||
374 | @code{EXTRA_SPECS} is useful when an architecture contains several | |
630d3d5a | 375 | related targets, which have various @code{@dots{}_SPECS} which are similar |
feca2ed3 JW |
376 | to each other, and the maintainer would like one central place to keep |
377 | these definitions. | |
378 | ||
379 | For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to | |
380 | define either @code{_CALL_SYSV} when the System V calling sequence is | |
381 | used or @code{_CALL_AIX} when the older AIX-based calling sequence is | |
382 | used. | |
383 | ||
384 | The @file{config/rs6000/rs6000.h} target file defines: | |
385 | ||
3ab51846 | 386 | @smallexample |
feca2ed3 JW |
387 | #define EXTRA_SPECS \ |
388 | @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @}, | |
389 | ||
390 | #define CPP_SYS_DEFAULT "" | |
3ab51846 | 391 | @end smallexample |
feca2ed3 JW |
392 | |
393 | The @file{config/rs6000/sysv.h} target file defines: | |
394 | @smallexample | |
395 | #undef CPP_SPEC | |
396 | #define CPP_SPEC \ | |
397 | "%@{posix: -D_POSIX_SOURCE @} \ | |
50d440bc NC |
398 | %@{mcall-sysv: -D_CALL_SYSV @} \ |
399 | %@{!mcall-sysv: %(cpp_sysv_default) @} \ | |
feca2ed3 JW |
400 | %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}" |
401 | ||
402 | #undef CPP_SYSV_DEFAULT | |
403 | #define CPP_SYSV_DEFAULT "-D_CALL_SYSV" | |
404 | @end smallexample | |
405 | ||
406 | while the @file{config/rs6000/eabiaix.h} target file defines | |
407 | @code{CPP_SYSV_DEFAULT} as: | |
408 | ||
409 | @smallexample | |
410 | #undef CPP_SYSV_DEFAULT | |
411 | #define CPP_SYSV_DEFAULT "-D_CALL_AIX" | |
412 | @end smallexample | |
a2c4f8e0 | 413 | @end defmac |
feca2ed3 | 414 | |
a2c4f8e0 | 415 | @defmac LINK_LIBGCC_SPECIAL_1 |
feca2ed3 JW |
416 | Define this macro if the driver program should find the library |
417 | @file{libgcc.a}. If you do not define this macro, the driver program will pass | |
630d3d5a | 418 | the argument @option{-lgcc} to tell the linker to do the search. |
a2c4f8e0 | 419 | @end defmac |
feca2ed3 | 420 | |
a2c4f8e0 | 421 | @defmac LINK_GCC_C_SEQUENCE_SPEC |
bbd7687d DM |
422 | The sequence in which libgcc and libc are specified to the linker. |
423 | By default this is @code{%G %L %G}. | |
a2c4f8e0 | 424 | @end defmac |
bbd7687d | 425 | |
a2c4f8e0 | 426 | @defmac LINK_COMMAND_SPEC |
9ec36da5 JL |
427 | A C string constant giving the complete command line need to execute the |
428 | linker. When you do this, you will need to update your port each time a | |
429 | change is made to the link command line within @file{gcc.c}. Therefore, | |
430 | define this macro only if you need to completely redefine the command | |
431 | line for invoking the linker and there is no other way to accomplish | |
bbd7687d DM |
432 | the effect you need. Overriding this macro may be avoidable by overriding |
433 | @code{LINK_GCC_C_SEQUENCE_SPEC} instead. | |
a2c4f8e0 | 434 | @end defmac |
9ec36da5 | 435 | |
a2c4f8e0 | 436 | @defmac LINK_ELIMINATE_DUPLICATE_LDIRECTORIES |
2cc07db4 | 437 | A nonzero value causes @command{collect2} to remove duplicate @option{-L@var{directory}} search |
5897739e JO |
438 | directories from linking commands. Do not give it a nonzero value if |
439 | removing duplicate search directories changes the linker's semantics. | |
a2c4f8e0 | 440 | @end defmac |
5897739e | 441 | |
a2c4f8e0 | 442 | @defmac MULTILIB_DEFAULTS |
feca2ed3 JW |
443 | Define this macro as a C expression for the initializer of an array of |
444 | string to tell the driver program which options are defaults for this | |
445 | target and thus do not need to be handled specially when using | |
446 | @code{MULTILIB_OPTIONS}. | |
447 | ||
448 | Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in | |
449 | the target makefile fragment or if none of the options listed in | |
450 | @code{MULTILIB_OPTIONS} are set by default. | |
451 | @xref{Target Fragment}. | |
a2c4f8e0 | 452 | @end defmac |
feca2ed3 | 453 | |
a2c4f8e0 | 454 | @defmac RELATIVE_PREFIX_NOT_LINKDIR |
05739753 | 455 | Define this macro to tell @command{gcc} that it should only translate |
630d3d5a | 456 | a @option{-B} prefix into a @option{-L} linker option if the prefix |
feca2ed3 | 457 | indicates an absolute file name. |
a2c4f8e0 | 458 | @end defmac |
feca2ed3 | 459 | |
a2c4f8e0 | 460 | @defmac MD_EXEC_PREFIX |
feca2ed3 JW |
461 | If defined, this macro is an additional prefix to try after |
462 | @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched | |
630d3d5a | 463 | when the @option{-b} option is used, or the compiler is built as a cross |
5505263f JL |
464 | compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it |
465 | to the list of directories used to find the assembler in @file{configure.in}. | |
a2c4f8e0 | 466 | @end defmac |
feca2ed3 | 467 | |
a2c4f8e0 | 468 | @defmac STANDARD_STARTFILE_PREFIX |
feca2ed3 | 469 | Define this macro as a C string constant if you wish to override the |
0d037580 | 470 | standard choice of @code{libdir} as the default prefix to |
feca2ed3 | 471 | try when searching for startup files such as @file{crt0.o}. |
0d037580 DJ |
472 | @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler |
473 | is built as a cross compiler. | |
a2c4f8e0 | 474 | @end defmac |
feca2ed3 | 475 | |
656c7a3a AL |
476 | @defmac STANDARD_STARTFILE_PREFIX_1 |
477 | Define this macro as a C string constant if you wish to override the | |
478 | standard choice of @code{/lib} as a prefix to try after the default prefix | |
479 | when searching for startup files such as @file{crt0.o}. | |
480 | @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler | |
481 | is built as a cross compiler. | |
482 | @end defmac | |
483 | ||
484 | @defmac STANDARD_STARTFILE_PREFIX_2 | |
485 | Define this macro as a C string constant if you wish to override the | |
486 | standard choice of @code{/lib} as yet another prefix to try after the | |
487 | default prefix when searching for startup files such as @file{crt0.o}. | |
488 | @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler | |
489 | is built as a cross compiler. | |
490 | @end defmac | |
491 | ||
a2c4f8e0 | 492 | @defmac MD_STARTFILE_PREFIX |
feca2ed3 JW |
493 | If defined, this macro supplies an additional prefix to try after the |
494 | standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the | |
630d3d5a | 495 | @option{-b} option is used, or when the compiler is built as a cross |
feca2ed3 | 496 | compiler. |
a2c4f8e0 | 497 | @end defmac |
feca2ed3 | 498 | |
a2c4f8e0 | 499 | @defmac MD_STARTFILE_PREFIX_1 |
feca2ed3 | 500 | If defined, this macro supplies yet another prefix to try after the |
630d3d5a | 501 | standard prefixes. It is not searched when the @option{-b} option is |
feca2ed3 | 502 | used, or when the compiler is built as a cross compiler. |
a2c4f8e0 | 503 | @end defmac |
feca2ed3 | 504 | |
a2c4f8e0 | 505 | @defmac INIT_ENVIRONMENT |
e9a25f70 | 506 | Define this macro as a C string constant if you wish to set environment |
feca2ed3 JW |
507 | variables for programs called by the driver, such as the assembler and |
508 | loader. The driver passes the value of this macro to @code{putenv} to | |
509 | initialize the necessary environment variables. | |
a2c4f8e0 | 510 | @end defmac |
feca2ed3 | 511 | |
a2c4f8e0 | 512 | @defmac LOCAL_INCLUDE_DIR |
feca2ed3 JW |
513 | Define this macro as a C string constant if you wish to override the |
514 | standard choice of @file{/usr/local/include} as the default prefix to | |
515 | try when searching for local header files. @code{LOCAL_INCLUDE_DIR} | |
516 | comes before @code{SYSTEM_INCLUDE_DIR} in the search order. | |
517 | ||
c237e94a ZW |
518 | Cross compilers do not search either @file{/usr/local/include} or its |
519 | replacement. | |
a2c4f8e0 | 520 | @end defmac |
feca2ed3 | 521 | |
a2c4f8e0 ZW |
522 | @defmac MODIFY_TARGET_NAME |
523 | Define this macro if you wish to define command-line switches that | |
524 | modify the default target name. | |
dc36ec2c RK |
525 | |
526 | For each switch, you can include a string to be appended to the first | |
527 | part of the configuration name or a string to be deleted from the | |
528 | configuration name, if present. The definition should be an initializer | |
529 | for an array of structures. Each array element should have three | |
530 | elements: the switch name (a string constant, including the initial | |
531 | dash), one of the enumeration codes @code{ADD} or @code{DELETE} to | |
532 | indicate whether the string should be inserted or deleted, and the string | |
533 | to be inserted or deleted (a string constant). | |
534 | ||
535 | For example, on a machine where @samp{64} at the end of the | |
630d3d5a JM |
536 | configuration name denotes a 64-bit target and you want the @option{-32} |
537 | and @option{-64} switches to select between 32- and 64-bit targets, you would | |
dc36ec2c RK |
538 | code |
539 | ||
540 | @smallexample | |
541 | #define MODIFY_TARGET_NAME \ | |
542 | @{ @{ "-32", DELETE, "64"@}, \ | |
543 | @{"-64", ADD, "64"@}@} | |
544 | @end smallexample | |
a2c4f8e0 | 545 | @end defmac |
dc36ec2c | 546 | |
a2c4f8e0 | 547 | @defmac SYSTEM_INCLUDE_DIR |
feca2ed3 JW |
548 | Define this macro as a C string constant if you wish to specify a |
549 | system-specific directory to search for header files before the standard | |
550 | directory. @code{SYSTEM_INCLUDE_DIR} comes before | |
551 | @code{STANDARD_INCLUDE_DIR} in the search order. | |
552 | ||
553 | Cross compilers do not use this macro and do not search the directory | |
554 | specified. | |
a2c4f8e0 | 555 | @end defmac |
feca2ed3 | 556 | |
a2c4f8e0 | 557 | @defmac STANDARD_INCLUDE_DIR |
feca2ed3 JW |
558 | Define this macro as a C string constant if you wish to override the |
559 | standard choice of @file{/usr/include} as the default prefix to | |
560 | try when searching for header files. | |
561 | ||
a2c4f8e0 | 562 | Cross compilers ignore this macro and do not search either |
feca2ed3 | 563 | @file{/usr/include} or its replacement. |
a2c4f8e0 | 564 | @end defmac |
feca2ed3 | 565 | |
a2c4f8e0 | 566 | @defmac STANDARD_INCLUDE_COMPONENT |
e9a25f70 JL |
567 | The ``component'' corresponding to @code{STANDARD_INCLUDE_DIR}. |
568 | See @code{INCLUDE_DEFAULTS}, below, for the description of components. | |
569 | If you do not define this macro, no component is used. | |
a2c4f8e0 | 570 | @end defmac |
e9a25f70 | 571 | |
a2c4f8e0 | 572 | @defmac INCLUDE_DEFAULTS |
feca2ed3 | 573 | Define this macro if you wish to override the entire default search path |
e9a25f70 JL |
574 | for include files. For a native compiler, the default search path |
575 | usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR}, | |
feca2ed3 JW |
576 | @code{SYSTEM_INCLUDE_DIR}, @code{GPLUSPLUS_INCLUDE_DIR}, and |
577 | @code{STANDARD_INCLUDE_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR} | |
578 | and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile}, | |
161d7b59 | 579 | and specify private search areas for GCC@. The directory |
feca2ed3 JW |
580 | @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs. |
581 | ||
582 | The definition should be an initializer for an array of structures. | |
e9a25f70 | 583 | Each array element should have four elements: the directory name (a |
9f6dc500 HPN |
584 | string constant), the component name (also a string constant), a flag |
585 | for C++-only directories, | |
e9a25f70 JL |
586 | and a flag showing that the includes in the directory don't need to be |
587 | wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of | |
588 | the array with a null element. | |
589 | ||
590 | The component name denotes what GNU package the include file is part of, | |
4bd0bee9 | 591 | if any, in all uppercase letters. For example, it might be @samp{GCC} |
9f6dc500 | 592 | or @samp{BINUTILS}. If the package is part of a vendor-supplied |
e9a25f70 JL |
593 | operating system, code the component name as @samp{0}. |
594 | ||
e9a25f70 | 595 | For example, here is the definition used for VAX/VMS: |
feca2ed3 | 596 | |
3ab51846 | 597 | @smallexample |
feca2ed3 JW |
598 | #define INCLUDE_DEFAULTS \ |
599 | @{ \ | |
e9a25f70 JL |
600 | @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \ |
601 | @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \ | |
602 | @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \ | |
603 | @{ ".", 0, 0, 0@}, \ | |
604 | @{ 0, 0, 0, 0@} \ | |
feca2ed3 | 605 | @} |
3ab51846 | 606 | @end smallexample |
a2c4f8e0 | 607 | @end defmac |
feca2ed3 JW |
608 | |
609 | Here is the order of prefixes tried for exec files: | |
610 | ||
611 | @enumerate | |
612 | @item | |
630d3d5a | 613 | Any prefixes specified by the user with @option{-B}. |
feca2ed3 JW |
614 | |
615 | @item | |
616 | The environment variable @code{GCC_EXEC_PREFIX}, if any. | |
617 | ||
618 | @item | |
619 | The directories specified by the environment variable @code{COMPILER_PATH}. | |
620 | ||
621 | @item | |
622 | The macro @code{STANDARD_EXEC_PREFIX}. | |
623 | ||
624 | @item | |
625 | @file{/usr/lib/gcc/}. | |
626 | ||
627 | @item | |
628 | The macro @code{MD_EXEC_PREFIX}, if any. | |
629 | @end enumerate | |
630 | ||
631 | Here is the order of prefixes tried for startfiles: | |
632 | ||
633 | @enumerate | |
634 | @item | |
630d3d5a | 635 | Any prefixes specified by the user with @option{-B}. |
feca2ed3 JW |
636 | |
637 | @item | |
638 | The environment variable @code{GCC_EXEC_PREFIX}, if any. | |
639 | ||
640 | @item | |
641 | The directories specified by the environment variable @code{LIBRARY_PATH} | |
512b62fb | 642 | (or port-specific name; native only, cross compilers do not use this). |
feca2ed3 JW |
643 | |
644 | @item | |
645 | The macro @code{STANDARD_EXEC_PREFIX}. | |
646 | ||
647 | @item | |
648 | @file{/usr/lib/gcc/}. | |
649 | ||
650 | @item | |
651 | The macro @code{MD_EXEC_PREFIX}, if any. | |
652 | ||
653 | @item | |
654 | The macro @code{MD_STARTFILE_PREFIX}, if any. | |
655 | ||
656 | @item | |
657 | The macro @code{STANDARD_STARTFILE_PREFIX}. | |
658 | ||
659 | @item | |
660 | @file{/lib/}. | |
661 | ||
662 | @item | |
663 | @file{/usr/lib/}. | |
664 | @end enumerate | |
665 | ||
666 | @node Run-time Target | |
667 | @section Run-time Target Specification | |
668 | @cindex run-time target specification | |
669 | @cindex predefined macros | |
670 | @cindex target specifications | |
671 | ||
672 | @c prevent bad page break with this line | |
673 | Here are run-time target specifications. | |
674 | ||
a2c4f8e0 | 675 | @defmac TARGET_CPU_CPP_BUILTINS () |
12a41c22 NB |
676 | This function-like macro expands to a block of code that defines |
677 | built-in preprocessor macros and assertions for the target cpu, using | |
1f95326c | 678 | the functions @code{builtin_define}, @code{builtin_define_std} and |
cb60f38d | 679 | @code{builtin_assert}. When the front end |
12a41c22 NB |
680 | calls this macro it provides a trailing semicolon, and since it has |
681 | finished command line option processing your code can use those | |
682 | results freely. | |
3df89291 NB |
683 | |
684 | @code{builtin_assert} takes a string in the form you pass to the | |
685 | command-line option @option{-A}, such as @code{cpu=mips}, and creates | |
d90a95fb | 686 | the assertion. @code{builtin_define} takes a string in the form |
3df89291 NB |
687 | accepted by option @option{-D} and unconditionally defines the macro. |
688 | ||
d90a95fb | 689 | @code{builtin_define_std} takes a string representing the name of an |
3df89291 | 690 | object-like macro. If it doesn't lie in the user's namespace, |
d90a95fb | 691 | @code{builtin_define_std} defines it unconditionally. Otherwise, it |
3df89291 NB |
692 | defines a version with two leading underscores, and another version |
693 | with two leading and trailing underscores, and defines the original | |
694 | only if an ISO standard was not requested on the command line. For | |
695 | example, passing @code{unix} defines @code{__unix}, @code{__unix__} | |
696 | and possibly @code{unix}; passing @code{_mips} defines @code{__mips}, | |
697 | @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64} | |
698 | defines only @code{_ABI64}. | |
699 | ||
e0322d5c NB |
700 | You can also test for the C dialect being compiled. The variable |
701 | @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus} | |
702 | or @code{clk_objective_c}. Note that if we are preprocessing | |
703 | assembler, this variable will be @code{clk_c} but the function-like | |
704 | macro @code{preprocessing_asm_p()} will return true, so you might want | |
ce3649d2 | 705 | to check for that first. If you need to check for strict ANSI, the |
c219e1da JDA |
706 | variable @code{flag_iso} can be used. The function-like macro |
707 | @code{preprocessing_trad_p()} can be used to check for traditional | |
708 | preprocessing. | |
a2c4f8e0 | 709 | @end defmac |
e0322d5c | 710 | |
a2c4f8e0 | 711 | @defmac TARGET_OS_CPP_BUILTINS () |
12a41c22 NB |
712 | Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional |
713 | and is used for the target operating system instead. | |
a2c4f8e0 | 714 | @end defmac |
12a41c22 | 715 | |
a2c4f8e0 | 716 | @defmac TARGET_OBJFMT_CPP_BUILTINS () |
4e2e315f NB |
717 | Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional |
718 | and is used for the target object format. @file{elfos.h} uses this | |
719 | macro to define @code{__ELF__}, so you probably do not need to define | |
720 | it yourself. | |
a2c4f8e0 | 721 | @end defmac |
4e2e315f | 722 | |
a2c4f8e0 | 723 | @deftypevar {extern int} target_flags |
75685792 RS |
724 | This variable is declared in @file{options.h}, which is included before |
725 | any target-specific headers. | |
a2c4f8e0 | 726 | @end deftypevar |
feca2ed3 | 727 | |
75685792 RS |
728 | @deftypevar {Target Hook} int TARGET_DEFAULT_TARGET_FLAGS |
729 | This variable specifies the initial value of @code{target_flags}. | |
730 | Its default setting is 0. | |
75685792 RS |
731 | @end deftypevar |
732 | ||
bacf5b96 RS |
733 | @cindex optional hardware or system features |
734 | @cindex features, optional, in system conventions | |
735 | ||
75685792 RS |
736 | @deftypefn {Target Hook} bool TARGET_HANDLE_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value}) |
737 | This hook is called whenever the user specifies one of the | |
738 | target-specific options described by the @file{.opt} definition files | |
739 | (@pxref{Options}). It has the opportunity to do some option-specific | |
740 | processing and should return true if the option is valid. The default | |
741 | definition does nothing but return true. | |
742 | ||
743 | @var{code} specifies the @code{OPT_@var{name}} enumeration value | |
744 | associated with the selected option; @var{name} is just a rendering of | |
745 | the option name in which non-alphanumeric characters are replaced by | |
746 | underscores. @var{arg} specifies the string argument and is null if | |
747 | no argument was given. If the option is flagged as a @code{UInteger} | |
748 | (@pxref{Option properties}), @var{value} is the numeric value of the | |
749 | argument. Otherwise @var{value} is 1 if the positive form of the | |
750 | option was used and 0 if the ``no-'' form was. | |
751 | @end deftypefn | |
752 | ||
a2c4f8e0 | 753 | @defmac TARGET_VERSION |
feca2ed3 JW |
754 | This macro is a C statement to print on @code{stderr} a string |
755 | describing the particular machine description choice. Every machine | |
756 | description should define @code{TARGET_VERSION}. For example: | |
757 | ||
758 | @smallexample | |
759 | #ifdef MOTOROLA | |
760 | #define TARGET_VERSION \ | |
761 | fprintf (stderr, " (68k, Motorola syntax)"); | |
762 | #else | |
763 | #define TARGET_VERSION \ | |
764 | fprintf (stderr, " (68k, MIT syntax)"); | |
765 | #endif | |
766 | @end smallexample | |
a2c4f8e0 | 767 | @end defmac |
feca2ed3 | 768 | |
a2c4f8e0 | 769 | @defmac OVERRIDE_OPTIONS |
feca2ed3 JW |
770 | Sometimes certain combinations of command options do not make sense on |
771 | a particular target machine. You can define a macro | |
772 | @code{OVERRIDE_OPTIONS} to take account of this. This macro, if | |
773 | defined, is executed once just after all the command options have been | |
774 | parsed. | |
775 | ||
776 | Don't use this macro to turn on various extra optimizations for | |
630d3d5a | 777 | @option{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for. |
a2c4f8e0 | 778 | @end defmac |
feca2ed3 | 779 | |
a2c4f8e0 | 780 | @defmac OPTIMIZATION_OPTIONS (@var{level}, @var{size}) |
feca2ed3 JW |
781 | Some machines may desire to change what optimizations are performed for |
782 | various optimization levels. This macro, if defined, is executed once | |
783 | just after the optimization level is determined and before the remainder | |
784 | of the command options have been parsed. Values set in this macro are | |
785 | used as the default values for the other command line options. | |
786 | ||
630d3d5a JM |
787 | @var{level} is the optimization level specified; 2 if @option{-O2} is |
788 | specified, 1 if @option{-O} is specified, and 0 if neither is specified. | |
feca2ed3 | 789 | |
df2a54e9 | 790 | @var{size} is nonzero if @option{-Os} is specified and zero otherwise. |
c6aded7c | 791 | |
feca2ed3 JW |
792 | You should not use this macro to change options that are not |
793 | machine-specific. These should uniformly selected by the same | |
794 | optimization level on all supported machines. Use this macro to enable | |
795 | machine-specific optimizations. | |
796 | ||
797 | @strong{Do not examine @code{write_symbols} in | |
798 | this macro!} The debugging options are not supposed to alter the | |
799 | generated code. | |
a2c4f8e0 | 800 | @end defmac |
feca2ed3 | 801 | |
a2c4f8e0 | 802 | @defmac CAN_DEBUG_WITHOUT_FP |
feca2ed3 | 803 | Define this macro if debugging can be performed even without a frame |
a3a15b4d | 804 | pointer. If this macro is defined, GCC will turn on the |
630d3d5a | 805 | @option{-fomit-frame-pointer} option whenever @option{-O} is specified. |
a2c4f8e0 | 806 | @end defmac |
feca2ed3 | 807 | |
414c4dc4 NC |
808 | @node Per-Function Data |
809 | @section Defining data structures for per-function information. | |
810 | @cindex per-function data | |
811 | @cindex data structures | |
812 | ||
813 | If the target needs to store information on a per-function basis, GCC | |
814 | provides a macro and a couple of variables to allow this. Note, just | |
815 | using statics to store the information is a bad idea, since GCC supports | |
816 | nested functions, so you can be halfway through encoding one function | |
817 | when another one comes along. | |
818 | ||
819 | GCC defines a data structure called @code{struct function} which | |
820 | contains all of the data specific to an individual function. This | |
821 | structure contains a field called @code{machine} whose type is | |
822 | @code{struct machine_function *}, which can be used by targets to point | |
823 | to their own specific data. | |
824 | ||
825 | If a target needs per-function specific data it should define the type | |
e2500fed GK |
826 | @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}. |
827 | This macro should be used to initialize the function pointer | |
828 | @code{init_machine_status}. This pointer is explained below. | |
414c4dc4 NC |
829 | |
830 | One typical use of per-function, target specific data is to create an | |
831 | RTX to hold the register containing the function's return address. This | |
832 | RTX can then be used to implement the @code{__builtin_return_address} | |
833 | function, for level 0. | |
834 | ||
aee96fe9 | 835 | Note---earlier implementations of GCC used a single data area to hold |
414c4dc4 NC |
836 | all of the per-function information. Thus when processing of a nested |
837 | function began the old per-function data had to be pushed onto a | |
838 | stack, and when the processing was finished, it had to be popped off the | |
839 | stack. GCC used to provide function pointers called | |
02f52e19 | 840 | @code{save_machine_status} and @code{restore_machine_status} to handle |
414c4dc4 NC |
841 | the saving and restoring of the target specific information. Since the |
842 | single data area approach is no longer used, these pointers are no | |
843 | longer supported. | |
844 | ||
a2c4f8e0 | 845 | @defmac INIT_EXPANDERS |
c21cd8b1 | 846 | Macro called to initialize any target specific information. This macro |
414c4dc4 | 847 | is called once per function, before generation of any RTL has begun. |
c21cd8b1 | 848 | The intention of this macro is to allow the initialization of the |
a2c4f8e0 ZW |
849 | function pointer @code{init_machine_status}. |
850 | @end defmac | |
414c4dc4 | 851 | |
a2c4f8e0 ZW |
852 | @deftypevar {void (*)(struct function *)} init_machine_status |
853 | If this function pointer is non-@code{NULL} it will be called once per | |
854 | function, before function compilation starts, in order to allow the | |
855 | target to perform any target specific initialization of the | |
856 | @code{struct function} structure. It is intended that this would be | |
857 | used to initialize the @code{machine} of that structure. | |
414c4dc4 | 858 | |
8a36672b | 859 | @code{struct machine_function} structures are expected to be freed by GC@. |
e2500fed GK |
860 | Generally, any memory that they reference must be allocated by using |
861 | @code{ggc_alloc}, including the structure itself. | |
a2c4f8e0 | 862 | @end deftypevar |
414c4dc4 | 863 | |
feca2ed3 JW |
864 | @node Storage Layout |
865 | @section Storage Layout | |
866 | @cindex storage layout | |
867 | ||
868 | Note that the definitions of the macros in this table which are sizes or | |
869 | alignments measured in bits do not need to be constant. They can be C | |
870 | expressions that refer to static variables, such as the @code{target_flags}. | |
871 | @xref{Run-time Target}. | |
872 | ||
a2c4f8e0 | 873 | @defmac BITS_BIG_ENDIAN |
feca2ed3 JW |
874 | Define this macro to have the value 1 if the most significant bit in a |
875 | byte has the lowest number; otherwise define it to have the value zero. | |
876 | This means that bit-field instructions count from the most significant | |
877 | bit. If the machine has no bit-field instructions, then this must still | |
878 | be defined, but it doesn't matter which value it is defined to. This | |
879 | macro need not be a constant. | |
880 | ||
881 | This macro does not affect the way structure fields are packed into | |
882 | bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}. | |
a2c4f8e0 | 883 | @end defmac |
feca2ed3 | 884 | |
a2c4f8e0 | 885 | @defmac BYTES_BIG_ENDIAN |
feca2ed3 JW |
886 | Define this macro to have the value 1 if the most significant byte in a |
887 | word has the lowest number. This macro need not be a constant. | |
a2c4f8e0 | 888 | @end defmac |
feca2ed3 | 889 | |
a2c4f8e0 | 890 | @defmac WORDS_BIG_ENDIAN |
feca2ed3 JW |
891 | Define this macro to have the value 1 if, in a multiword object, the |
892 | most significant word has the lowest number. This applies to both | |
a3a15b4d | 893 | memory locations and registers; GCC fundamentally assumes that the |
feca2ed3 JW |
894 | order of words in memory is the same as the order in registers. This |
895 | macro need not be a constant. | |
a2c4f8e0 | 896 | @end defmac |
feca2ed3 | 897 | |
a2c4f8e0 | 898 | @defmac LIBGCC2_WORDS_BIG_ENDIAN |
aee96fe9 JM |
899 | Define this macro if @code{WORDS_BIG_ENDIAN} is not constant. This must be a |
900 | constant value with the same meaning as @code{WORDS_BIG_ENDIAN}, which will be | |
901 | used only when compiling @file{libgcc2.c}. Typically the value will be set | |
feca2ed3 | 902 | based on preprocessor defines. |
a2c4f8e0 | 903 | @end defmac |
feca2ed3 | 904 | |
a2c4f8e0 | 905 | @defmac FLOAT_WORDS_BIG_ENDIAN |
feca2ed3 JW |
906 | Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or |
907 | @code{TFmode} floating point numbers are stored in memory with the word | |
908 | containing the sign bit at the lowest address; otherwise define it to | |
909 | have the value 0. This macro need not be a constant. | |
910 | ||
911 | You need not define this macro if the ordering is the same as for | |
912 | multi-word integers. | |
a2c4f8e0 | 913 | @end defmac |
feca2ed3 | 914 | |
a2c4f8e0 | 915 | @defmac BITS_PER_UNIT |
feca2ed3 | 916 | Define this macro to be the number of bits in an addressable storage |
5c60f03d | 917 | unit (byte). If you do not define this macro the default is 8. |
a2c4f8e0 | 918 | @end defmac |
feca2ed3 | 919 | |
a2c4f8e0 | 920 | @defmac BITS_PER_WORD |
e81dd381 KG |
921 | Number of bits in a word. If you do not define this macro, the default |
922 | is @code{BITS_PER_UNIT * UNITS_PER_WORD}. | |
a2c4f8e0 | 923 | @end defmac |
feca2ed3 | 924 | |
a2c4f8e0 | 925 | @defmac MAX_BITS_PER_WORD |
feca2ed3 JW |
926 | Maximum number of bits in a word. If this is undefined, the default is |
927 | @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the | |
928 | largest value that @code{BITS_PER_WORD} can have at run-time. | |
a2c4f8e0 | 929 | @end defmac |
feca2ed3 | 930 | |
a2c4f8e0 | 931 | @defmac UNITS_PER_WORD |
c4336539 PB |
932 | Number of storage units in a word; normally the size of a general-purpose |
933 | register, a power of two from 1 or 8. | |
a2c4f8e0 | 934 | @end defmac |
feca2ed3 | 935 | |
a2c4f8e0 | 936 | @defmac MIN_UNITS_PER_WORD |
feca2ed3 JW |
937 | Minimum number of units in a word. If this is undefined, the default is |
938 | @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the | |
939 | smallest value that @code{UNITS_PER_WORD} can have at run-time. | |
a2c4f8e0 | 940 | @end defmac |
feca2ed3 | 941 | |
c4336539 PB |
942 | @defmac UNITS_PER_SIMD_WORD |
943 | Number of units in the vectors that the vectorizer can produce. | |
944 | The default is equal to @code{UNITS_PER_WORD}, because the vectorizer | |
945 | can do some transformations even in absence of specialized @acronym{SIMD} | |
946 | hardware. | |
947 | @end defmac | |
948 | ||
a2c4f8e0 | 949 | @defmac POINTER_SIZE |
feca2ed3 JW |
950 | Width of a pointer, in bits. You must specify a value no wider than the |
951 | width of @code{Pmode}. If it is not equal to the width of @code{Pmode}, | |
2465bf76 KG |
952 | you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify |
953 | a value the default is @code{BITS_PER_WORD}. | |
a2c4f8e0 | 954 | @end defmac |
feca2ed3 | 955 | |
a2c4f8e0 | 956 | @defmac POINTERS_EXTEND_UNSIGNED |
6dd12198 | 957 | A C expression whose value is greater than zero if pointers that need to be |
f5963e61 | 958 | extended from being @code{POINTER_SIZE} bits wide to @code{Pmode} are to |
6dd12198 SE |
959 | be zero-extended and zero if they are to be sign-extended. If the value |
960 | is less then zero then there must be an "ptr_extend" instruction that | |
961 | extends a pointer from @code{POINTER_SIZE} to @code{Pmode}. | |
feca2ed3 JW |
962 | |
963 | You need not define this macro if the @code{POINTER_SIZE} is equal | |
964 | to the width of @code{Pmode}. | |
a2c4f8e0 | 965 | @end defmac |
feca2ed3 | 966 | |
a2c4f8e0 | 967 | @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type}) |
feca2ed3 JW |
968 | A macro to update @var{m} and @var{unsignedp} when an object whose type |
969 | is @var{type} and which has the specified mode and signedness is to be | |
970 | stored in a register. This macro is only called when @var{type} is a | |
971 | scalar type. | |
972 | ||
973 | On most RISC machines, which only have operations that operate on a full | |
974 | register, define this macro to set @var{m} to @code{word_mode} if | |
975 | @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most | |
976 | cases, only integer modes should be widened because wider-precision | |
977 | floating-point operations are usually more expensive than their narrower | |
978 | counterparts. | |
979 | ||
980 | For most machines, the macro definition does not change @var{unsignedp}. | |
981 | However, some machines, have instructions that preferentially handle | |
982 | either signed or unsigned quantities of certain modes. For example, on | |
983 | the DEC Alpha, 32-bit loads from memory and 32-bit add instructions | |
984 | sign-extend the result to 64 bits. On such machines, set | |
985 | @var{unsignedp} according to which kind of extension is more efficient. | |
986 | ||
987 | Do not define this macro if it would never modify @var{m}. | |
a2c4f8e0 | 988 | @end defmac |
feca2ed3 | 989 | |
d4453b7a PB |
990 | @defmac PROMOTE_FUNCTION_MODE |
991 | Like @code{PROMOTE_MODE}, but is applied to outgoing function arguments or | |
992 | function return values, as specified by @code{TARGET_PROMOTE_FUNCTION_ARGS} | |
993 | and @code{TARGET_PROMOTE_FUNCTION_RETURN}, respectively. | |
994 | ||
995 | The default is @code{PROMOTE_MODE}. | |
996 | @end defmac | |
997 | ||
61f71b34 DD |
998 | @deftypefn {Target Hook} bool TARGET_PROMOTE_FUNCTION_ARGS (tree @var{fntype}) |
999 | This target hook should return @code{true} if the promotion described by | |
d4453b7a PB |
1000 | @code{PROMOTE_FUNCTION_MODE} should be done for outgoing function |
1001 | arguments. | |
61f71b34 | 1002 | @end deftypefn |
feca2ed3 | 1003 | |
61f71b34 DD |
1004 | @deftypefn {Target Hook} bool TARGET_PROMOTE_FUNCTION_RETURN (tree @var{fntype}) |
1005 | This target hook should return @code{true} if the promotion described by | |
d4453b7a | 1006 | @code{PROMOTE_FUNCTION_MODE} should be done for the return value of |
61f71b34 | 1007 | functions. |
feca2ed3 | 1008 | |
61f71b34 | 1009 | If this target hook returns @code{true}, @code{FUNCTION_VALUE} must |
5b4ef0b1 | 1010 | perform the same promotions done by @code{PROMOTE_FUNCTION_MODE}. |
61f71b34 | 1011 | @end deftypefn |
feca2ed3 | 1012 | |
a2c4f8e0 | 1013 | @defmac PARM_BOUNDARY |
feca2ed3 JW |
1014 | Normal alignment required for function parameters on the stack, in |
1015 | bits. All stack parameters receive at least this much alignment | |
1016 | regardless of data type. On most machines, this is the same as the | |
1017 | size of an integer. | |
a2c4f8e0 | 1018 | @end defmac |
feca2ed3 | 1019 | |
a2c4f8e0 | 1020 | @defmac STACK_BOUNDARY |
31cdd499 ZW |
1021 | Define this macro to the minimum alignment enforced by hardware for the |
1022 | stack pointer on this machine. The definition is a C expression for the | |
1023 | desired alignment (measured in bits). This value is used as a default | |
1024 | if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines, | |
1025 | this should be the same as @code{PARM_BOUNDARY}. | |
a2c4f8e0 | 1026 | @end defmac |
c795bca9 | 1027 | |
a2c4f8e0 | 1028 | @defmac PREFERRED_STACK_BOUNDARY |
31cdd499 ZW |
1029 | Define this macro if you wish to preserve a certain alignment for the |
1030 | stack pointer, greater than what the hardware enforces. The definition | |
1031 | is a C expression for the desired alignment (measured in bits). This | |
1032 | macro must evaluate to a value equal to or larger than | |
1033 | @code{STACK_BOUNDARY}. | |
a2c4f8e0 | 1034 | @end defmac |
feca2ed3 | 1035 | |
a2c4f8e0 | 1036 | @defmac FORCE_PREFERRED_STACK_BOUNDARY_IN_MAIN |
1d482056 RH |
1037 | A C expression that evaluates true if @code{PREFERRED_STACK_BOUNDARY} is |
1038 | not guaranteed by the runtime and we should emit code to align the stack | |
1039 | at the beginning of @code{main}. | |
1040 | ||
c795bca9 | 1041 | @cindex @code{PUSH_ROUNDING}, interaction with @code{PREFERRED_STACK_BOUNDARY} |
feca2ed3 | 1042 | If @code{PUSH_ROUNDING} is not defined, the stack will always be aligned |
c795bca9 BS |
1043 | to the specified boundary. If @code{PUSH_ROUNDING} is defined and specifies |
1044 | a less strict alignment than @code{PREFERRED_STACK_BOUNDARY}, the stack may | |
1045 | be momentarily unaligned while pushing arguments. | |
a2c4f8e0 | 1046 | @end defmac |
feca2ed3 | 1047 | |
a2c4f8e0 | 1048 | @defmac FUNCTION_BOUNDARY |
feca2ed3 | 1049 | Alignment required for a function entry point, in bits. |
a2c4f8e0 | 1050 | @end defmac |
feca2ed3 | 1051 | |
a2c4f8e0 | 1052 | @defmac BIGGEST_ALIGNMENT |
feca2ed3 | 1053 | Biggest alignment that any data type can require on this machine, in bits. |
a2c4f8e0 | 1054 | @end defmac |
feca2ed3 | 1055 | |
a2c4f8e0 | 1056 | @defmac MINIMUM_ATOMIC_ALIGNMENT |
861bb6c1 JL |
1057 | If defined, the smallest alignment, in bits, that can be given to an |
1058 | object that can be referenced in one operation, without disturbing any | |
1059 | nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger | |
1060 | on machines that don't have byte or half-word store operations. | |
a2c4f8e0 | 1061 | @end defmac |
861bb6c1 | 1062 | |
a2c4f8e0 | 1063 | @defmac BIGGEST_FIELD_ALIGNMENT |
11cf4d18 JJ |
1064 | Biggest alignment that any structure or union field can require on this |
1065 | machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for | |
1066 | structure and union fields only, unless the field alignment has been set | |
1067 | by the @code{__attribute__ ((aligned (@var{n})))} construct. | |
a2c4f8e0 | 1068 | @end defmac |
feca2ed3 | 1069 | |
a2c4f8e0 | 1070 | @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed}) |
feca2ed3 | 1071 | An expression for the alignment of a structure field @var{field} if the |
ad9335eb JJ |
1072 | alignment computed in the usual way (including applying of |
1073 | @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the | |
1074 | alignment) is @var{computed}. It overrides alignment only if the | |
1075 | field alignment has not been set by the | |
1076 | @code{__attribute__ ((aligned (@var{n})))} construct. | |
a2c4f8e0 | 1077 | @end defmac |
feca2ed3 | 1078 | |
a2c4f8e0 | 1079 | @defmac MAX_OFILE_ALIGNMENT |
feca2ed3 JW |
1080 | Biggest alignment supported by the object file format of this machine. |
1081 | Use this macro to limit the alignment which can be specified using the | |
1082 | @code{__attribute__ ((aligned (@var{n})))} construct. If not defined, | |
1083 | the default value is @code{BIGGEST_ALIGNMENT}. | |
a2c4f8e0 | 1084 | @end defmac |
feca2ed3 | 1085 | |
a2c4f8e0 | 1086 | @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align}) |
a8d1550a | 1087 | If defined, a C expression to compute the alignment for a variable in |
8a198bd2 JW |
1088 | the static store. @var{type} is the data type, and @var{basic-align} is |
1089 | the alignment that the object would ordinarily have. The value of this | |
feca2ed3 JW |
1090 | macro is used instead of that alignment to align the object. |
1091 | ||
1092 | If this macro is not defined, then @var{basic-align} is used. | |
1093 | ||
1094 | @findex strcpy | |
1095 | One use of this macro is to increase alignment of medium-size data to | |
1096 | make it all fit in fewer cache lines. Another is to cause character | |
1097 | arrays to be word-aligned so that @code{strcpy} calls that copy | |
1098 | constants to character arrays can be done inline. | |
a2c4f8e0 | 1099 | @end defmac |
feca2ed3 | 1100 | |
a2c4f8e0 | 1101 | @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align}) |
feca2ed3 JW |
1102 | If defined, a C expression to compute the alignment given to a constant |
1103 | that is being placed in memory. @var{constant} is the constant and | |
1104 | @var{basic-align} is the alignment that the object would ordinarily | |
1105 | have. The value of this macro is used instead of that alignment to | |
1106 | align the object. | |
1107 | ||
1108 | If this macro is not defined, then @var{basic-align} is used. | |
1109 | ||
1110 | The typical use of this macro is to increase alignment for string | |
1111 | constants to be word aligned so that @code{strcpy} calls that copy | |
1112 | constants can be done inline. | |
a2c4f8e0 | 1113 | @end defmac |
feca2ed3 | 1114 | |
a2c4f8e0 | 1115 | @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align}) |
a8d1550a | 1116 | If defined, a C expression to compute the alignment for a variable in |
d16790f2 JW |
1117 | the local store. @var{type} is the data type, and @var{basic-align} is |
1118 | the alignment that the object would ordinarily have. The value of this | |
1119 | macro is used instead of that alignment to align the object. | |
1120 | ||
1121 | If this macro is not defined, then @var{basic-align} is used. | |
1122 | ||
1123 | One use of this macro is to increase alignment of medium-size data to | |
1124 | make it all fit in fewer cache lines. | |
a2c4f8e0 | 1125 | @end defmac |
d16790f2 | 1126 | |
a2c4f8e0 | 1127 | @defmac EMPTY_FIELD_BOUNDARY |
c771326b | 1128 | Alignment in bits to be given to a structure bit-field that follows an |
feca2ed3 JW |
1129 | empty field such as @code{int : 0;}. |
1130 | ||
78d55cc8 | 1131 | If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro. |
a2c4f8e0 | 1132 | @end defmac |
feca2ed3 | 1133 | |
a2c4f8e0 | 1134 | @defmac STRUCTURE_SIZE_BOUNDARY |
feca2ed3 JW |
1135 | Number of bits which any structure or union's size must be a multiple of. |
1136 | Each structure or union's size is rounded up to a multiple of this. | |
1137 | ||
1138 | If you do not define this macro, the default is the same as | |
1139 | @code{BITS_PER_UNIT}. | |
a2c4f8e0 | 1140 | @end defmac |
feca2ed3 | 1141 | |
a2c4f8e0 | 1142 | @defmac STRICT_ALIGNMENT |
feca2ed3 JW |
1143 | Define this macro to be the value 1 if instructions will fail to work |
1144 | if given data not on the nominal alignment. If instructions will merely | |
1145 | go slower in that case, define this macro as 0. | |
a2c4f8e0 | 1146 | @end defmac |
feca2ed3 | 1147 | |
a2c4f8e0 | 1148 | @defmac PCC_BITFIELD_TYPE_MATTERS |
feca2ed3 | 1149 | Define this if you wish to imitate the way many other C compilers handle |
c771326b | 1150 | alignment of bit-fields and the structures that contain them. |
feca2ed3 | 1151 | |
8dc65b6e MM |
1152 | The behavior is that the type written for a named bit-field (@code{int}, |
1153 | @code{short}, or other integer type) imposes an alignment for the entire | |
1154 | structure, as if the structure really did contain an ordinary field of | |
1155 | that type. In addition, the bit-field is placed within the structure so | |
1156 | that it would fit within such a field, not crossing a boundary for it. | |
1157 | ||
1158 | Thus, on most machines, a named bit-field whose type is written as | |
1159 | @code{int} would not cross a four-byte boundary, and would force | |
1160 | four-byte alignment for the whole structure. (The alignment used may | |
1161 | not be four bytes; it is controlled by the other alignment parameters.) | |
1162 | ||
1163 | An unnamed bit-field will not affect the alignment of the containing | |
1164 | structure. | |
feca2ed3 JW |
1165 | |
1166 | If the macro is defined, its definition should be a C expression; | |
1167 | a nonzero value for the expression enables this behavior. | |
1168 | ||
1169 | Note that if this macro is not defined, or its value is zero, some | |
c771326b | 1170 | bit-fields may cross more than one alignment boundary. The compiler can |
feca2ed3 JW |
1171 | support such references if there are @samp{insv}, @samp{extv}, and |
1172 | @samp{extzv} insns that can directly reference memory. | |
1173 | ||
c771326b | 1174 | The other known way of making bit-fields work is to define |
feca2ed3 JW |
1175 | @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}. |
1176 | Then every structure can be accessed with fullwords. | |
1177 | ||
c771326b | 1178 | Unless the machine has bit-field instructions or you define |
feca2ed3 JW |
1179 | @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define |
1180 | @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value. | |
1181 | ||
a3a15b4d | 1182 | If your aim is to make GCC use the same conventions for laying out |
c771326b | 1183 | bit-fields as are used by another compiler, here is how to investigate |
feca2ed3 JW |
1184 | what the other compiler does. Compile and run this program: |
1185 | ||
3ab51846 | 1186 | @smallexample |
feca2ed3 JW |
1187 | struct foo1 |
1188 | @{ | |
1189 | char x; | |
1190 | char :0; | |
1191 | char y; | |
1192 | @}; | |
1193 | ||
1194 | struct foo2 | |
1195 | @{ | |
1196 | char x; | |
1197 | int :0; | |
1198 | char y; | |
1199 | @}; | |
1200 | ||
1201 | main () | |
1202 | @{ | |
1203 | printf ("Size of foo1 is %d\n", | |
1204 | sizeof (struct foo1)); | |
1205 | printf ("Size of foo2 is %d\n", | |
1206 | sizeof (struct foo2)); | |
1207 | exit (0); | |
1208 | @} | |
3ab51846 | 1209 | @end smallexample |
feca2ed3 JW |
1210 | |
1211 | If this prints 2 and 5, then the compiler's behavior is what you would | |
1212 | get from @code{PCC_BITFIELD_TYPE_MATTERS}. | |
a2c4f8e0 | 1213 | @end defmac |
feca2ed3 | 1214 | |
a2c4f8e0 | 1215 | @defmac BITFIELD_NBYTES_LIMITED |
f913c102 AO |
1216 | Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited |
1217 | to aligning a bit-field within the structure. | |
a2c4f8e0 | 1218 | @end defmac |
feca2ed3 | 1219 | |
13c1cd82 PB |
1220 | @deftypefn {Target Hook} bool TARGET_ALIGN_ANON_BITFIELDS (void) |
1221 | When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine | |
1222 | whether unnamed bitfields affect the alignment of the containing | |
1223 | structure. The hook should return true if the structure should inherit | |
1224 | the alignment requirements of an unnamed bitfield's type. | |
1225 | @end deftypefn | |
1226 | ||
a2c4f8e0 | 1227 | @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode}) |
31a02448 | 1228 | Return 1 if a structure or array containing @var{field} should be accessed using |
9f6dc500 HPN |
1229 | @code{BLKMODE}. |
1230 | ||
182e515e AH |
1231 | If @var{field} is the only field in the structure, @var{mode} is its |
1232 | mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the | |
1233 | case where structures of one field would require the structure's mode to | |
1234 | retain the field's mode. | |
1235 | ||
9f6dc500 HPN |
1236 | Normally, this is not needed. See the file @file{c4x.h} for an example |
1237 | of how to use this macro to prevent a structure having a floating point | |
1238 | field from being accessed in an integer mode. | |
a2c4f8e0 | 1239 | @end defmac |
9f6dc500 | 1240 | |
a2c4f8e0 | 1241 | @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified}) |
0003feb2 VM |
1242 | Define this macro as an expression for the alignment of a type (given |
1243 | by @var{type} as a tree node) if the alignment computed in the usual | |
1244 | way is @var{computed} and the alignment explicitly specified was | |
feca2ed3 JW |
1245 | @var{specified}. |
1246 | ||
1247 | The default is to use @var{specified} if it is larger; otherwise, use | |
1248 | the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT} | |
a2c4f8e0 | 1249 | @end defmac |
feca2ed3 | 1250 | |
a2c4f8e0 | 1251 | @defmac MAX_FIXED_MODE_SIZE |
feca2ed3 JW |
1252 | An integer expression for the size in bits of the largest integer |
1253 | machine mode that should actually be used. All integer machine modes of | |
1254 | this size or smaller can be used for structures and unions with the | |
1255 | appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE | |
1256 | (DImode)} is assumed. | |
a2c4f8e0 | 1257 | @end defmac |
feca2ed3 | 1258 | |
a2c4f8e0 | 1259 | @defmac STACK_SAVEAREA_MODE (@var{save_level}) |
73c8090f | 1260 | If defined, an expression of type @code{enum machine_mode} that |
39403d82 DE |
1261 | specifies the mode of the save area operand of a |
1262 | @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}). | |
1263 | @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or | |
1264 | @code{SAVE_NONLOCAL} and selects which of the three named patterns is | |
1265 | having its mode specified. | |
73c8090f DE |
1266 | |
1267 | You need not define this macro if it always returns @code{Pmode}. You | |
1268 | would most commonly define this macro if the | |
1269 | @code{save_stack_@var{level}} patterns need to support both a 32- and a | |
1270 | 64-bit mode. | |
a2c4f8e0 | 1271 | @end defmac |
73c8090f | 1272 | |
a2c4f8e0 | 1273 | @defmac STACK_SIZE_MODE |
39403d82 DE |
1274 | If defined, an expression of type @code{enum machine_mode} that |
1275 | specifies the mode of the size increment operand of an | |
1276 | @code{allocate_stack} named pattern (@pxref{Standard Names}). | |
1277 | ||
1278 | You need not define this macro if it always returns @code{word_mode}. | |
1279 | You would most commonly define this macro if the @code{allocate_stack} | |
1280 | pattern needs to support both a 32- and a 64-bit mode. | |
a2c4f8e0 | 1281 | @end defmac |
39403d82 | 1282 | |
a2c4f8e0 | 1283 | @defmac TARGET_FLOAT_FORMAT |
feca2ed3 | 1284 | A code distinguishing the floating point format of the target machine. |
a2c4f8e0 | 1285 | There are four defined values: |
feca2ed3 | 1286 | |
a2c4f8e0 | 1287 | @ftable @code |
feca2ed3 JW |
1288 | @item IEEE_FLOAT_FORMAT |
1289 | This code indicates IEEE floating point. It is the default; there is no | |
a2c4f8e0 | 1290 | need to define @code{TARGET_FLOAT_FORMAT} when the format is IEEE@. |
feca2ed3 | 1291 | |
feca2ed3 | 1292 | @item VAX_FLOAT_FORMAT |
4226378a PK |
1293 | This code indicates the ``F float'' (for @code{float}) and ``D float'' |
1294 | or ``G float'' formats (for @code{double}) used on the VAX and PDP-11@. | |
feca2ed3 | 1295 | |
68eb4fb9 LB |
1296 | @item IBM_FLOAT_FORMAT |
1297 | This code indicates the format used on the IBM System/370. | |
1298 | ||
68eb4fb9 LB |
1299 | @item C4X_FLOAT_FORMAT |
1300 | This code indicates the format used on the TMS320C3x/C4x. | |
a2c4f8e0 | 1301 | @end ftable |
68eb4fb9 | 1302 | |
a2c4f8e0 ZW |
1303 | If your target uses a floating point format other than these, you must |
1304 | define a new @var{name}_FLOAT_FORMAT code for it, and add support for | |
1305 | it to @file{real.c}. | |
feca2ed3 JW |
1306 | |
1307 | The ordering of the component words of floating point values stored in | |
807633e5 | 1308 | memory is controlled by @code{FLOAT_WORDS_BIG_ENDIAN}. |
a2c4f8e0 | 1309 | @end defmac |
e9a25f70 | 1310 | |
a2c4f8e0 | 1311 | @defmac MODE_HAS_NANS (@var{mode}) |
71925bc0 RS |
1312 | When defined, this macro should be true if @var{mode} has a NaN |
1313 | representation. The compiler assumes that NaNs are not equal to | |
1314 | anything (including themselves) and that addition, subtraction, | |
1315 | multiplication and division all return NaNs when one operand is | |
1316 | NaN@. | |
1317 | ||
1318 | By default, this macro is true if @var{mode} is a floating-point | |
1319 | mode and the target floating-point format is IEEE@. | |
a2c4f8e0 | 1320 | @end defmac |
71925bc0 | 1321 | |
a2c4f8e0 | 1322 | @defmac MODE_HAS_INFINITIES (@var{mode}) |
71925bc0 RS |
1323 | This macro should be true if @var{mode} can represent infinity. At |
1324 | present, the compiler uses this macro to decide whether @samp{x - x} | |
1325 | is always defined. By default, the macro is true when @var{mode} | |
1326 | is a floating-point mode and the target format is IEEE@. | |
a2c4f8e0 | 1327 | @end defmac |
71925bc0 | 1328 | |
a2c4f8e0 | 1329 | @defmac MODE_HAS_SIGNED_ZEROS (@var{mode}) |
71925bc0 RS |
1330 | True if @var{mode} distinguishes between positive and negative zero. |
1331 | The rules are expected to follow the IEEE standard: | |
1332 | ||
1333 | @itemize @bullet | |
1334 | @item | |
1335 | @samp{x + x} has the same sign as @samp{x}. | |
1336 | ||
1337 | @item | |
1338 | If the sum of two values with opposite sign is zero, the result is | |
1339 | positive for all rounding modes expect towards @minus{}infinity, for | |
1340 | which it is negative. | |
1341 | ||
1342 | @item | |
1343 | The sign of a product or quotient is negative when exactly one | |
1344 | of the operands is negative. | |
1345 | @end itemize | |
1346 | ||
1347 | The default definition is true if @var{mode} is a floating-point | |
1348 | mode and the target format is IEEE@. | |
a2c4f8e0 | 1349 | @end defmac |
71925bc0 | 1350 | |
a2c4f8e0 | 1351 | @defmac MODE_HAS_SIGN_DEPENDENT_ROUNDING (@var{mode}) |
71925bc0 RS |
1352 | If defined, this macro should be true for @var{mode} if it has at |
1353 | least one rounding mode in which @samp{x} and @samp{-x} can be | |
1354 | rounded to numbers of different magnitude. Two such modes are | |
1355 | towards @minus{}infinity and towards +infinity. | |
1356 | ||
1357 | The default definition of this macro is true if @var{mode} is | |
1358 | a floating-point mode and the target format is IEEE@. | |
a2c4f8e0 | 1359 | @end defmac |
3fcaac1d | 1360 | |
a2c4f8e0 | 1361 | @defmac ROUND_TOWARDS_ZERO |
3fcaac1d RS |
1362 | If defined, this macro should be true if the prevailing rounding |
1363 | mode is towards zero. A true value has the following effects: | |
1364 | ||
1365 | @itemize @bullet | |
1366 | @item | |
1367 | @code{MODE_HAS_SIGN_DEPENDENT_ROUNDING} will be false for all modes. | |
1368 | ||
1369 | @item | |
1370 | @file{libgcc.a}'s floating-point emulator will round towards zero | |
1371 | rather than towards nearest. | |
1372 | ||
1373 | @item | |
1374 | The compiler's floating-point emulator will round towards zero after | |
1375 | doing arithmetic, and when converting from the internal float format to | |
1376 | the target format. | |
1377 | @end itemize | |
1378 | ||
1379 | The macro does not affect the parsing of string literals. When the | |
1380 | primary rounding mode is towards zero, library functions like | |
1381 | @code{strtod} might still round towards nearest, and the compiler's | |
1382 | parser should behave like the target's @code{strtod} where possible. | |
1383 | ||
1384 | Not defining this macro is equivalent to returning zero. | |
a2c4f8e0 | 1385 | @end defmac |
3fcaac1d | 1386 | |
a2c4f8e0 | 1387 | @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size}) |
4226378a | 1388 | This macro should return true if floats with @var{size} |
3fcaac1d RS |
1389 | bits do not have a NaN or infinity representation, but use the largest |
1390 | exponent for normal numbers instead. | |
1391 | ||
1392 | Defining this macro to true for @var{size} causes @code{MODE_HAS_NANS} | |
1393 | and @code{MODE_HAS_INFINITIES} to be false for @var{size}-bit modes. | |
1394 | It also affects the way @file{libgcc.a} and @file{real.c} emulate | |
1395 | floating-point arithmetic. | |
1396 | ||
1397 | The default definition of this macro returns false for all sizes. | |
a2c4f8e0 | 1398 | @end defmac |
feca2ed3 | 1399 | |
c8e4f0e9 AH |
1400 | @deftypefn {Target Hook} bool TARGET_VECTOR_OPAQUE_P (tree @var{type}) |
1401 | This target hook should return @code{true} a vector is opaque. That | |
1402 | is, if no cast is needed when copying a vector value of type | |
1403 | @var{type} into another vector lvalue of the same size. Vector opaque | |
1404 | types cannot be initialized. The default is that there are no such | |
1405 | types. | |
62e1dfcf NC |
1406 | @end deftypefn |
1407 | ||
f913c102 AO |
1408 | @deftypefn {Target Hook} bool TARGET_MS_BITFIELD_LAYOUT_P (tree @var{record_type}) |
1409 | This target hook returns @code{true} if bit-fields in the given | |
1410 | @var{record_type} are to be laid out following the rules of Microsoft | |
1411 | Visual C/C++, namely: (i) a bit-field won't share the same storage | |
1412 | unit with the previous bit-field if their underlying types have | |
1413 | different sizes, and the bit-field will be aligned to the highest | |
1414 | alignment of the underlying types of itself and of the previous | |
1415 | bit-field; (ii) a zero-sized bit-field will affect the alignment of | |
1416 | the whole enclosing structure, even if it is unnamed; except that | |
1417 | (iii) a zero-sized bit-field will be disregarded unless it follows | |
6335b0aa | 1418 | another bit-field of nonzero size. If this hook returns @code{true}, |
f913c102 | 1419 | other macros that control bit-field layout are ignored. |
e4850f36 DR |
1420 | |
1421 | When a bit-field is inserted into a packed record, the whole size | |
1422 | of the underlying type is used by one or more same-size adjacent | |
1423 | bit-fields (that is, if its long:3, 32 bits is used in the record, | |
1424 | and any additional adjacent long bit-fields are packed into the same | |
8a36672b JM |
1425 | chunk of 32 bits. However, if the size changes, a new field of that |
1426 | size is allocated). In an unpacked record, this is the same as using | |
e4850f36 DR |
1427 | alignment, but not equivalent when packing. |
1428 | ||
1429 | If both MS bit-fields and @samp{__attribute__((packed))} are used, | |
8a36672b | 1430 | the latter will take precedence. If @samp{__attribute__((packed))} is |
e4850f36 DR |
1431 | used on a single field when MS bit-fields are in use, it will take |
1432 | precedence for that field, but the alignment of the rest of the structure | |
1433 | may affect its placement. | |
f913c102 AO |
1434 | @end deftypefn |
1435 | ||
f18eca82 ZL |
1436 | @deftypefn {Target Hook} {const char *} TARGET_MANGLE_FUNDAMENTAL_TYPE (tree @var{type}) |
1437 | If your target defines any fundamental types, define this hook to | |
1438 | return the appropriate encoding for these types as part of a C++ | |
1439 | mangled name. The @var{type} argument is the tree structure | |
1440 | representing the type to be mangled. The hook may be applied to trees | |
1441 | which are not target-specific fundamental types; it should return | |
1442 | @code{NULL} for all such types, as well as arguments it does not | |
1443 | recognize. If the return value is not @code{NULL}, it must point to | |
1444 | a statically-allocated string constant. | |
1445 | ||
1446 | Target-specific fundamental types might be new fundamental types or | |
1447 | qualified versions of ordinary fundamental types. Encode new | |
1448 | fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name} | |
1449 | is the name used for the type in source code, and @var{n} is the | |
1450 | length of @var{name} in decimal. Encode qualified versions of | |
1451 | ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where | |
1452 | @var{name} is the name used for the type qualifier in source code, | |
1453 | @var{n} is the length of @var{name} as above, and @var{code} is the | |
1454 | code used to represent the unqualified version of this type. (See | |
1455 | @code{write_builtin_type} in @file{cp/mangle.c} for the list of | |
1456 | codes.) In both cases the spaces are for clarity; do not include any | |
1457 | spaces in your string. | |
1458 | ||
1459 | The default version of this hook always returns @code{NULL}, which is | |
1460 | appropriate for a target that does not define any new fundamental | |
1461 | types. | |
1462 | @end deftypefn | |
1463 | ||
feca2ed3 JW |
1464 | @node Type Layout |
1465 | @section Layout of Source Language Data Types | |
1466 | ||
1467 | These macros define the sizes and other characteristics of the standard | |
1468 | basic data types used in programs being compiled. Unlike the macros in | |
1469 | the previous section, these apply to specific features of C and related | |
1470 | languages, rather than to fundamental aspects of storage layout. | |
1471 | ||
a2c4f8e0 | 1472 | @defmac INT_TYPE_SIZE |
feca2ed3 JW |
1473 | A C expression for the size in bits of the type @code{int} on the |
1474 | target machine. If you don't define this, the default is one word. | |
a2c4f8e0 | 1475 | @end defmac |
feca2ed3 | 1476 | |
a2c4f8e0 | 1477 | @defmac SHORT_TYPE_SIZE |
feca2ed3 JW |
1478 | A C expression for the size in bits of the type @code{short} on the |
1479 | target machine. If you don't define this, the default is half a word. | |
1480 | (If this would be less than one storage unit, it is rounded up to one | |
1481 | unit.) | |
a2c4f8e0 | 1482 | @end defmac |
feca2ed3 | 1483 | |
a2c4f8e0 | 1484 | @defmac LONG_TYPE_SIZE |
feca2ed3 JW |
1485 | A C expression for the size in bits of the type @code{long} on the |
1486 | target machine. If you don't define this, the default is one word. | |
a2c4f8e0 | 1487 | @end defmac |
feca2ed3 | 1488 | |
a2c4f8e0 | 1489 | @defmac ADA_LONG_TYPE_SIZE |
1615c261 | 1490 | On some machines, the size used for the Ada equivalent of the type |
8a36672b | 1491 | @code{long} by a native Ada compiler differs from that used by C@. In |
1615c261 RK |
1492 | that situation, define this macro to be a C expression to be used for |
1493 | the size of that type. If you don't define this, the default is the | |
1494 | value of @code{LONG_TYPE_SIZE}. | |
a2c4f8e0 | 1495 | @end defmac |
1615c261 | 1496 | |
a2c4f8e0 | 1497 | @defmac LONG_LONG_TYPE_SIZE |
feca2ed3 JW |
1498 | A C expression for the size in bits of the type @code{long long} on the |
1499 | target machine. If you don't define this, the default is two | |
047c1c92 | 1500 | words. If you want to support GNU Ada on your machine, the value of this |
feca2ed3 | 1501 | macro must be at least 64. |
a2c4f8e0 | 1502 | @end defmac |
feca2ed3 | 1503 | |
a2c4f8e0 | 1504 | @defmac CHAR_TYPE_SIZE |
feca2ed3 | 1505 | A C expression for the size in bits of the type @code{char} on the |
c294bd99 HPN |
1506 | target machine. If you don't define this, the default is |
1507 | @code{BITS_PER_UNIT}. | |
a2c4f8e0 | 1508 | @end defmac |
feca2ed3 | 1509 | |
a2c4f8e0 | 1510 | @defmac BOOL_TYPE_SIZE |
3d1ad9e5 JM |
1511 | A C expression for the size in bits of the C++ type @code{bool} and |
1512 | C99 type @code{_Bool} on the target machine. If you don't define | |
1513 | this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}. | |
a2c4f8e0 | 1514 | @end defmac |
68eb4fb9 | 1515 | |
a2c4f8e0 | 1516 | @defmac FLOAT_TYPE_SIZE |
feca2ed3 JW |
1517 | A C expression for the size in bits of the type @code{float} on the |
1518 | target machine. If you don't define this, the default is one word. | |
a2c4f8e0 | 1519 | @end defmac |
feca2ed3 | 1520 | |
a2c4f8e0 | 1521 | @defmac DOUBLE_TYPE_SIZE |
feca2ed3 JW |
1522 | A C expression for the size in bits of the type @code{double} on the |
1523 | target machine. If you don't define this, the default is two | |
1524 | words. | |
a2c4f8e0 | 1525 | @end defmac |
feca2ed3 | 1526 | |
a2c4f8e0 | 1527 | @defmac LONG_DOUBLE_TYPE_SIZE |
feca2ed3 JW |
1528 | A C expression for the size in bits of the type @code{long double} on |
1529 | the target machine. If you don't define this, the default is two | |
1530 | words. | |
a2c4f8e0 | 1531 | @end defmac |
feca2ed3 | 1532 | |
4e9db8b2 SE |
1533 | @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE |
1534 | Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or | |
1535 | if you want routines in @file{libgcc2.a} for a size other than | |
1536 | @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the | |
1537 | default is @code{LONG_DOUBLE_TYPE_SIZE}. | |
1538 | @end defmac | |
1539 | ||
1540 | @defmac LIBGCC2_HAS_DF_MODE | |
1541 | Define this macro if neither @code{LIBGCC2_DOUBLE_TYPE_SIZE} nor | |
1542 | @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is | |
1543 | @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a} | |
1544 | anyway. If you don't define this and either @code{LIBGCC2_DOUBLE_TYPE_SIZE} | |
1545 | or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1, | |
1546 | otherwise it is 0. | |
1547 | @end defmac | |
1548 | ||
1549 | @defmac LIBGCC2_HAS_XF_MODE | |
1550 | Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not | |
1551 | @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a} | |
1552 | anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} | |
1553 | is 80 then the default is 1, otherwise it is 0. | |
1554 | @end defmac | |
1555 | ||
1556 | @defmac LIBGCC2_HAS_TF_MODE | |
1557 | Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not | |
1558 | @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a} | |
1559 | anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} | |
1560 | is 128 then the default is 1, otherwise it is 0. | |
1561 | @end defmac | |
1562 | ||
a2c4f8e0 | 1563 | @defmac TARGET_FLT_EVAL_METHOD |
d57a4b98 RH |
1564 | A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h}, |
1565 | assuming, if applicable, that the floating-point control word is in its | |
1566 | default state. If you do not define this macro the value of | |
1567 | @code{FLT_EVAL_METHOD} will be zero. | |
a2c4f8e0 | 1568 | @end defmac |
aaa2e8ef | 1569 | |
a2c4f8e0 | 1570 | @defmac WIDEST_HARDWARE_FP_SIZE |
e9a25f70 JL |
1571 | A C expression for the size in bits of the widest floating-point format |
1572 | supported by the hardware. If you define this macro, you must specify a | |
1573 | value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}. | |
1574 | If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE} | |
1575 | is the default. | |
a2c4f8e0 | 1576 | @end defmac |
e9a25f70 | 1577 | |
a2c4f8e0 | 1578 | @defmac DEFAULT_SIGNED_CHAR |
feca2ed3 JW |
1579 | An expression whose value is 1 or 0, according to whether the type |
1580 | @code{char} should be signed or unsigned by default. The user can | |
630d3d5a JM |
1581 | always override this default with the options @option{-fsigned-char} |
1582 | and @option{-funsigned-char}. | |
a2c4f8e0 | 1583 | @end defmac |
feca2ed3 | 1584 | |
221ee7c9 KH |
1585 | @deftypefn {Target Hook} bool TARGET_DEFAULT_SHORT_ENUMS (void) |
1586 | This target hook should return true if the compiler should give an | |
1587 | @code{enum} type only as many bytes as it takes to represent the range | |
1588 | of possible values of that type. It should return false if all | |
35afa569 KH |
1589 | @code{enum} types should be allocated like @code{int}. |
1590 | ||
221ee7c9 KH |
1591 | The default is to return false. |
1592 | @end deftypefn | |
35afa569 | 1593 | |
a2c4f8e0 | 1594 | @defmac SIZE_TYPE |
feca2ed3 JW |
1595 | A C expression for a string describing the name of the data type to use |
1596 | for size values. The typedef name @code{size_t} is defined using the | |
1597 | contents of the string. | |
1598 | ||
1599 | The string can contain more than one keyword. If so, separate them with | |
1600 | spaces, and write first any length keyword, then @code{unsigned} if | |
1601 | appropriate, and finally @code{int}. The string must exactly match one | |
1602 | of the data type names defined in the function | |
1603 | @code{init_decl_processing} in the file @file{c-decl.c}. You may not | |
1604 | omit @code{int} or change the order---that would cause the compiler to | |
1605 | crash on startup. | |
1606 | ||
1607 | If you don't define this macro, the default is @code{"long unsigned | |
1608 | int"}. | |
a2c4f8e0 | 1609 | @end defmac |
feca2ed3 | 1610 | |
a2c4f8e0 | 1611 | @defmac PTRDIFF_TYPE |
feca2ed3 JW |
1612 | A C expression for a string describing the name of the data type to use |
1613 | for the result of subtracting two pointers. The typedef name | |
1614 | @code{ptrdiff_t} is defined using the contents of the string. See | |
1615 | @code{SIZE_TYPE} above for more information. | |
1616 | ||
1617 | If you don't define this macro, the default is @code{"long int"}. | |
a2c4f8e0 | 1618 | @end defmac |
feca2ed3 | 1619 | |
a2c4f8e0 | 1620 | @defmac WCHAR_TYPE |
feca2ed3 JW |
1621 | A C expression for a string describing the name of the data type to use |
1622 | for wide characters. The typedef name @code{wchar_t} is defined using | |
1623 | the contents of the string. See @code{SIZE_TYPE} above for more | |
1624 | information. | |
1625 | ||
1626 | If you don't define this macro, the default is @code{"int"}. | |
a2c4f8e0 | 1627 | @end defmac |
feca2ed3 | 1628 | |
a2c4f8e0 | 1629 | @defmac WCHAR_TYPE_SIZE |
feca2ed3 JW |
1630 | A C expression for the size in bits of the data type for wide |
1631 | characters. This is used in @code{cpp}, which cannot make use of | |
1632 | @code{WCHAR_TYPE}. | |
a2c4f8e0 | 1633 | @end defmac |
feca2ed3 | 1634 | |
a2c4f8e0 | 1635 | @defmac WINT_TYPE |
1a67c7d3 JL |
1636 | A C expression for a string describing the name of the data type to |
1637 | use for wide characters passed to @code{printf} and returned from | |
1638 | @code{getwc}. The typedef name @code{wint_t} is defined using the | |
1639 | contents of the string. See @code{SIZE_TYPE} above for more | |
1640 | information. | |
1641 | ||
1642 | If you don't define this macro, the default is @code{"unsigned int"}. | |
a2c4f8e0 | 1643 | @end defmac |
1a67c7d3 | 1644 | |
a2c4f8e0 | 1645 | @defmac INTMAX_TYPE |
b15ad712 JM |
1646 | A C expression for a string describing the name of the data type that |
1647 | can represent any value of any standard or extended signed integer type. | |
1648 | The typedef name @code{intmax_t} is defined using the contents of the | |
1649 | string. See @code{SIZE_TYPE} above for more information. | |
1650 | ||
1651 | If you don't define this macro, the default is the first of | |
1652 | @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as | |
1653 | much precision as @code{long long int}. | |
a2c4f8e0 | 1654 | @end defmac |
b15ad712 | 1655 | |
a2c4f8e0 | 1656 | @defmac UINTMAX_TYPE |
b15ad712 JM |
1657 | A C expression for a string describing the name of the data type that |
1658 | can represent any value of any standard or extended unsigned integer | |
1659 | type. The typedef name @code{uintmax_t} is defined using the contents | |
1660 | of the string. See @code{SIZE_TYPE} above for more information. | |
1661 | ||
1662 | If you don't define this macro, the default is the first of | |
1663 | @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long | |
1664 | unsigned int"} that has as much precision as @code{long long unsigned | |
1665 | int}. | |
a2c4f8e0 | 1666 | @end defmac |
b15ad712 | 1667 | |
a2c4f8e0 | 1668 | @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION |
f3c55c97 AO |
1669 | The C++ compiler represents a pointer-to-member-function with a struct |
1670 | that looks like: | |
1671 | ||
3ab51846 | 1672 | @smallexample |
f3c55c97 AO |
1673 | struct @{ |
1674 | union @{ | |
1675 | void (*fn)(); | |
1676 | ptrdiff_t vtable_index; | |
1677 | @}; | |
1678 | ptrdiff_t delta; | |
1679 | @}; | |
3ab51846 | 1680 | @end smallexample |
f3c55c97 AO |
1681 | |
1682 | @noindent | |
1683 | The C++ compiler must use one bit to indicate whether the function that | |
1684 | will be called through a pointer-to-member-function is virtual. | |
1685 | Normally, we assume that the low-order bit of a function pointer must | |
1686 | always be zero. Then, by ensuring that the vtable_index is odd, we can | |
1687 | distinguish which variant of the union is in use. But, on some | |
1688 | platforms function pointers can be odd, and so this doesn't work. In | |
1689 | that case, we use the low-order bit of the @code{delta} field, and shift | |
1690 | the remainder of the @code{delta} field to the left. | |
1691 | ||
1692 | GCC will automatically make the right selection about where to store | |
1693 | this bit using the @code{FUNCTION_BOUNDARY} setting for your platform. | |
1694 | However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY} | |
1695 | set such that functions always start at even addresses, but the lowest | |
1696 | bit of pointers to functions indicate whether the function at that | |
1697 | address is in ARM or Thumb mode. If this is the case of your | |
1698 | architecture, you should define this macro to | |
1699 | @code{ptrmemfunc_vbit_in_delta}. | |
1700 | ||
1701 | In general, you should not have to define this macro. On architectures | |
1702 | in which function addresses are always even, according to | |
1703 | @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to | |
1704 | @code{ptrmemfunc_vbit_in_pfn}. | |
a2c4f8e0 | 1705 | @end defmac |
67231816 | 1706 | |
a2c4f8e0 | 1707 | @defmac TARGET_VTABLE_USES_DESCRIPTORS |
67231816 | 1708 | Normally, the C++ compiler uses function pointers in vtables. This |
f282ffb3 | 1709 | macro allows the target to change to use ``function descriptors'' |
67231816 RH |
1710 | instead. Function descriptors are found on targets for whom a |
1711 | function pointer is actually a small data structure. Normally the | |
f282ffb3 | 1712 | data structure consists of the actual code address plus a data |
67231816 RH |
1713 | pointer to which the function's data is relative. |
1714 | ||
1715 | If vtables are used, the value of this macro should be the number | |
1716 | of words that the function descriptor occupies. | |
a2c4f8e0 | 1717 | @end defmac |
a6f5e048 | 1718 | |
a2c4f8e0 | 1719 | @defmac TARGET_VTABLE_ENTRY_ALIGN |
a6f5e048 RH |
1720 | By default, the vtable entries are void pointers, the so the alignment |
1721 | is the same as pointer alignment. The value of this macro specifies | |
1722 | the alignment of the vtable entry in bits. It should be defined only | |
1723 | when special alignment is necessary. */ | |
a2c4f8e0 | 1724 | @end defmac |
a6f5e048 | 1725 | |
a2c4f8e0 | 1726 | @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE |
a6f5e048 RH |
1727 | There are a few non-descriptor entries in the vtable at offsets below |
1728 | zero. If these entries must be padded (say, to preserve the alignment | |
1729 | specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number | |
1730 | of words in each data entry. | |
a2c4f8e0 | 1731 | @end defmac |
b2b263e1 | 1732 | |
feca2ed3 JW |
1733 | @node Registers |
1734 | @section Register Usage | |
1735 | @cindex register usage | |
1736 | ||
1737 | This section explains how to describe what registers the target machine | |
1738 | has, and how (in general) they can be used. | |
1739 | ||
1740 | The description of which registers a specific instruction can use is | |
1741 | done with register classes; see @ref{Register Classes}. For information | |
1742 | on using registers to access a stack frame, see @ref{Frame Registers}. | |
1743 | For passing values in registers, see @ref{Register Arguments}. | |
1744 | For returning values in registers, see @ref{Scalar Return}. | |
1745 | ||
1746 | @menu | |
1747 | * Register Basics:: Number and kinds of registers. | |
1748 | * Allocation Order:: Order in which registers are allocated. | |
1749 | * Values in Registers:: What kinds of values each reg can hold. | |
1750 | * Leaf Functions:: Renumbering registers for leaf functions. | |
1751 | * Stack Registers:: Handling a register stack such as 80387. | |
feca2ed3 JW |
1752 | @end menu |
1753 | ||
1754 | @node Register Basics | |
1755 | @subsection Basic Characteristics of Registers | |
1756 | ||
1757 | @c prevent bad page break with this line | |
1758 | Registers have various characteristics. | |
1759 | ||
a2c4f8e0 | 1760 | @defmac FIRST_PSEUDO_REGISTER |
feca2ed3 JW |
1761 | Number of hardware registers known to the compiler. They receive |
1762 | numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first | |
1763 | pseudo register's number really is assigned the number | |
1764 | @code{FIRST_PSEUDO_REGISTER}. | |
a2c4f8e0 | 1765 | @end defmac |
feca2ed3 | 1766 | |
a2c4f8e0 | 1767 | @defmac FIXED_REGISTERS |
feca2ed3 JW |
1768 | @cindex fixed register |
1769 | An initializer that says which registers are used for fixed purposes | |
1770 | all throughout the compiled code and are therefore not available for | |
1771 | general allocation. These would include the stack pointer, the frame | |
1772 | pointer (except on machines where that can be used as a general | |
1773 | register when no frame pointer is needed), the program counter on | |
1774 | machines where that is considered one of the addressable registers, | |
1775 | and any other numbered register with a standard use. | |
1776 | ||
1777 | This information is expressed as a sequence of numbers, separated by | |
1778 | commas and surrounded by braces. The @var{n}th number is 1 if | |
1779 | register @var{n} is fixed, 0 otherwise. | |
1780 | ||
1781 | The table initialized from this macro, and the table initialized by | |
1782 | the following one, may be overridden at run time either automatically, | |
1783 | by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by | |
630d3d5a JM |
1784 | the user with the command options @option{-ffixed-@var{reg}}, |
1785 | @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}. | |
a2c4f8e0 | 1786 | @end defmac |
feca2ed3 | 1787 | |
a2c4f8e0 | 1788 | @defmac CALL_USED_REGISTERS |
feca2ed3 JW |
1789 | @cindex call-used register |
1790 | @cindex call-clobbered register | |
1791 | @cindex call-saved register | |
1792 | Like @code{FIXED_REGISTERS} but has 1 for each register that is | |
1793 | clobbered (in general) by function calls as well as for fixed | |
1794 | registers. This macro therefore identifies the registers that are not | |
1795 | available for general allocation of values that must live across | |
1796 | function calls. | |
1797 | ||
1798 | If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler | |
1799 | automatically saves it on function entry and restores it on function | |
1800 | exit, if the register is used within the function. | |
a2c4f8e0 | 1801 | @end defmac |
feca2ed3 | 1802 | |
a2c4f8e0 | 1803 | @defmac CALL_REALLY_USED_REGISTERS |
fc1296b7 AM |
1804 | @cindex call-used register |
1805 | @cindex call-clobbered register | |
1806 | @cindex call-saved register | |
f282ffb3 JM |
1807 | Like @code{CALL_USED_REGISTERS} except this macro doesn't require |
1808 | that the entire set of @code{FIXED_REGISTERS} be included. | |
fc1296b7 | 1809 | (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}). |
f282ffb3 | 1810 | This macro is optional. If not specified, it defaults to the value |
fc1296b7 | 1811 | of @code{CALL_USED_REGISTERS}. |
a2c4f8e0 | 1812 | @end defmac |
fc1296b7 | 1813 | |
a2c4f8e0 | 1814 | @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode}) |
1e326708 MH |
1815 | @cindex call-used register |
1816 | @cindex call-clobbered register | |
1817 | @cindex call-saved register | |
df2a54e9 | 1818 | A C expression that is nonzero if it is not permissible to store a |
1e326708 MH |
1819 | value of mode @var{mode} in hard register number @var{regno} across a |
1820 | call without some part of it being clobbered. For most machines this | |
1821 | macro need not be defined. It is only required for machines that do not | |
1822 | preserve the entire contents of a register across a call. | |
a2c4f8e0 | 1823 | @end defmac |
1e326708 | 1824 | |
feca2ed3 JW |
1825 | @findex fixed_regs |
1826 | @findex call_used_regs | |
a2c4f8e0 ZW |
1827 | @findex global_regs |
1828 | @findex reg_names | |
1829 | @findex reg_class_contents | |
1830 | @defmac CONDITIONAL_REGISTER_USAGE | |
055177dc NC |
1831 | Zero or more C statements that may conditionally modify five variables |
1832 | @code{fixed_regs}, @code{call_used_regs}, @code{global_regs}, | |
c237e94a ZW |
1833 | @code{reg_names}, and @code{reg_class_contents}, to take into account |
1834 | any dependence of these register sets on target flags. The first three | |
1835 | of these are of type @code{char []} (interpreted as Boolean vectors). | |
1836 | @code{global_regs} is a @code{const char *[]}, and | |
1837 | @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is | |
1838 | called, @code{fixed_regs}, @code{call_used_regs}, | |
1839 | @code{reg_class_contents}, and @code{reg_names} have been initialized | |
055177dc | 1840 | from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS}, |
c237e94a | 1841 | @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively. |
630d3d5a | 1842 | @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}}, |
c237e94a ZW |
1843 | @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}} |
1844 | command options have been applied. | |
feca2ed3 JW |
1845 | |
1846 | You need not define this macro if it has no work to do. | |
1847 | ||
1848 | @cindex disabling certain registers | |
1849 | @cindex controlling register usage | |
1850 | If the usage of an entire class of registers depends on the target | |
1851 | flags, you may indicate this to GCC by using this macro to modify | |
1852 | @code{fixed_regs} and @code{call_used_regs} to 1 for each of the | |
161d7b59 | 1853 | registers in the classes which should not be used by GCC@. Also define |
97488870 R |
1854 | the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT} |
1855 | to return @code{NO_REGS} if it | |
feca2ed3 JW |
1856 | is called with a letter for a class that shouldn't be used. |
1857 | ||
1858 | (However, if this class is not included in @code{GENERAL_REGS} and all | |
1859 | of the insn patterns whose constraints permit this class are | |
1860 | controlled by target switches, then GCC will automatically avoid using | |
1861 | these registers when the target switches are opposed to them.) | |
a2c4f8e0 | 1862 | @end defmac |
feca2ed3 | 1863 | |
a2c4f8e0 | 1864 | @defmac INCOMING_REGNO (@var{out}) |
feca2ed3 JW |
1865 | Define this macro if the target machine has register windows. This C |
1866 | expression returns the register number as seen by the called function | |
1867 | corresponding to the register number @var{out} as seen by the calling | |
1868 | function. Return @var{out} if register number @var{out} is not an | |
1869 | outbound register. | |
a2c4f8e0 | 1870 | @end defmac |
feca2ed3 | 1871 | |
a2c4f8e0 | 1872 | @defmac OUTGOING_REGNO (@var{in}) |
feca2ed3 JW |
1873 | Define this macro if the target machine has register windows. This C |
1874 | expression returns the register number as seen by the calling function | |
1875 | corresponding to the register number @var{in} as seen by the called | |
1876 | function. Return @var{in} if register number @var{in} is not an inbound | |
1877 | register. | |
a2c4f8e0 | 1878 | @end defmac |
feca2ed3 | 1879 | |
a2c4f8e0 | 1880 | @defmac LOCAL_REGNO (@var{regno}) |
fa80e43d JL |
1881 | Define this macro if the target machine has register windows. This C |
1882 | expression returns true if the register is call-saved but is in the | |
1883 | register window. Unlike most call-saved registers, such registers | |
1884 | need not be explicitly restored on function exit or during non-local | |
1885 | gotos. | |
a2c4f8e0 | 1886 | @end defmac |
fa80e43d | 1887 | |
a2c4f8e0 | 1888 | @defmac PC_REGNUM |
feca2ed3 JW |
1889 | If the program counter has a register number, define this as that |
1890 | register number. Otherwise, do not define it. | |
a2c4f8e0 | 1891 | @end defmac |
feca2ed3 JW |
1892 | |
1893 | @node Allocation Order | |
1894 | @subsection Order of Allocation of Registers | |
1895 | @cindex order of register allocation | |
1896 | @cindex register allocation order | |
1897 | ||
1898 | @c prevent bad page break with this line | |
1899 | Registers are allocated in order. | |
1900 | ||
a2c4f8e0 | 1901 | @defmac REG_ALLOC_ORDER |
feca2ed3 | 1902 | If defined, an initializer for a vector of integers, containing the |
a3a15b4d | 1903 | numbers of hard registers in the order in which GCC should prefer |
feca2ed3 JW |
1904 | to use them (from most preferred to least). |
1905 | ||
1906 | If this macro is not defined, registers are used lowest numbered first | |
1907 | (all else being equal). | |
1908 | ||
1909 | One use of this macro is on machines where the highest numbered | |
1910 | registers must always be saved and the save-multiple-registers | |
1911 | instruction supports only sequences of consecutive registers. On such | |
1912 | machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists | |
956d6950 | 1913 | the highest numbered allocable register first. |
a2c4f8e0 | 1914 | @end defmac |
feca2ed3 | 1915 | |
a2c4f8e0 | 1916 | @defmac ORDER_REGS_FOR_LOCAL_ALLOC |
feca2ed3 JW |
1917 | A C statement (sans semicolon) to choose the order in which to allocate |
1918 | hard registers for pseudo-registers local to a basic block. | |
1919 | ||
1920 | Store the desired register order in the array @code{reg_alloc_order}. | |
1921 | Element 0 should be the register to allocate first; element 1, the next | |
1922 | register; and so on. | |
1923 | ||
1924 | The macro body should not assume anything about the contents of | |
1925 | @code{reg_alloc_order} before execution of the macro. | |
1926 | ||
1927 | On most machines, it is not necessary to define this macro. | |
a2c4f8e0 | 1928 | @end defmac |
feca2ed3 JW |
1929 | |
1930 | @node Values in Registers | |
1931 | @subsection How Values Fit in Registers | |
1932 | ||
1933 | This section discusses the macros that describe which kinds of values | |
1934 | (specifically, which machine modes) each register can hold, and how many | |
1935 | consecutive registers are needed for a given mode. | |
1936 | ||
a2c4f8e0 | 1937 | @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode}) |
feca2ed3 JW |
1938 | A C expression for the number of consecutive hard registers, starting |
1939 | at register number @var{regno}, required to hold a value of mode | |
1940 | @var{mode}. | |
1941 | ||
1942 | On a machine where all registers are exactly one word, a suitable | |
1943 | definition of this macro is | |
1944 | ||
1945 | @smallexample | |
1946 | #define HARD_REGNO_NREGS(REGNO, MODE) \ | |
1947 | ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \ | |
32bd3974 | 1948 | / UNITS_PER_WORD) |
feca2ed3 | 1949 | @end smallexample |
a2c4f8e0 | 1950 | @end defmac |
feca2ed3 | 1951 | |
ca0b6e3b EB |
1952 | @defmac REGMODE_NATURAL_SIZE (@var{mode}) |
1953 | Define this macro if the natural size of registers that hold values | |
1954 | of mode @var{mode} is not the word size. It is a C expression that | |
1955 | should give the natural size in bytes for the specified mode. It is | |
1956 | used by the register allocator to try to optimize its results. This | |
1957 | happens for example on SPARC 64-bit where the natural size of | |
1958 | floating-point registers is still 32-bit. | |
1959 | @end defmac | |
1960 | ||
a2c4f8e0 | 1961 | @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode}) |
feca2ed3 JW |
1962 | A C expression that is nonzero if it is permissible to store a value |
1963 | of mode @var{mode} in hard register number @var{regno} (or in several | |
1964 | registers starting with that one). For a machine where all registers | |
1965 | are equivalent, a suitable definition is | |
1966 | ||
1967 | @smallexample | |
1968 | #define HARD_REGNO_MODE_OK(REGNO, MODE) 1 | |
1969 | @end smallexample | |
1970 | ||
e9a25f70 JL |
1971 | You need not include code to check for the numbers of fixed registers, |
1972 | because the allocation mechanism considers them to be always occupied. | |
feca2ed3 JW |
1973 | |
1974 | @cindex register pairs | |
1975 | On some machines, double-precision values must be kept in even/odd | |
e9a25f70 JL |
1976 | register pairs. You can implement that by defining this macro to reject |
1977 | odd register numbers for such modes. | |
feca2ed3 JW |
1978 | |
1979 | The minimum requirement for a mode to be OK in a register is that the | |
1980 | @samp{mov@var{mode}} instruction pattern support moves between the | |
e9a25f70 JL |
1981 | register and other hard register in the same class and that moving a |
1982 | value into the register and back out not alter it. | |
feca2ed3 | 1983 | |
e9a25f70 JL |
1984 | Since the same instruction used to move @code{word_mode} will work for |
1985 | all narrower integer modes, it is not necessary on any machine for | |
feca2ed3 JW |
1986 | @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided |
1987 | you define patterns @samp{movhi}, etc., to take advantage of this. This | |
1988 | is useful because of the interaction between @code{HARD_REGNO_MODE_OK} | |
1989 | and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes | |
1990 | to be tieable. | |
1991 | ||
1992 | Many machines have special registers for floating point arithmetic. | |
1993 | Often people assume that floating point machine modes are allowed only | |
1994 | in floating point registers. This is not true. Any registers that | |
1995 | can hold integers can safely @emph{hold} a floating point machine | |
1996 | mode, whether or not floating arithmetic can be done on it in those | |
1997 | registers. Integer move instructions can be used to move the values. | |
1998 | ||
1999 | On some machines, though, the converse is true: fixed-point machine | |
2000 | modes may not go in floating registers. This is true if the floating | |
2001 | registers normalize any value stored in them, because storing a | |
2002 | non-floating value there would garble it. In this case, | |
2003 | @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in | |
2004 | floating registers. But if the floating registers do not automatically | |
2005 | normalize, if you can store any bit pattern in one and retrieve it | |
2006 | unchanged without a trap, then any machine mode may go in a floating | |
2007 | register, so you can define this macro to say so. | |
2008 | ||
2009 | The primary significance of special floating registers is rather that | |
2010 | they are the registers acceptable in floating point arithmetic | |
2011 | instructions. However, this is of no concern to | |
2012 | @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper | |
2013 | constraints for those instructions. | |
2014 | ||
2015 | On some machines, the floating registers are especially slow to access, | |
2016 | so that it is better to store a value in a stack frame than in such a | |
2017 | register if floating point arithmetic is not being done. As long as the | |
2018 | floating registers are not in class @code{GENERAL_REGS}, they will not | |
2019 | be used unless some pattern's constraint asks for one. | |
a2c4f8e0 | 2020 | @end defmac |
feca2ed3 | 2021 | |
150c9fe8 KH |
2022 | @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to}) |
2023 | A C expression that is nonzero if it is OK to rename a hard register | |
2024 | @var{from} to another hard register @var{to}. | |
2025 | ||
2026 | One common use of this macro is to prevent renaming of a register to | |
2027 | another register that is not saved by a prologue in an interrupt | |
2028 | handler. | |
2029 | ||
2030 | The default is always nonzero. | |
2031 | @end defmac | |
2032 | ||
a2c4f8e0 | 2033 | @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2}) |
e9a25f70 | 2034 | A C expression that is nonzero if a value of mode |
956d6950 | 2035 | @var{mode1} is accessible in mode @var{mode2} without copying. |
feca2ed3 JW |
2036 | |
2037 | If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and | |
e9a25f70 JL |
2038 | @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for |
2039 | any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})} | |
2040 | should be nonzero. If they differ for any @var{r}, you should define | |
2041 | this macro to return zero unless some other mechanism ensures the | |
956d6950 | 2042 | accessibility of the value in a narrower mode. |
e9a25f70 JL |
2043 | |
2044 | You should define this macro to return nonzero in as many cases as | |
a3a15b4d | 2045 | possible since doing so will allow GCC to perform better register |
e9a25f70 | 2046 | allocation. |
a2c4f8e0 | 2047 | @end defmac |
7506f491 | 2048 | |
a2c4f8e0 | 2049 | @defmac AVOID_CCMODE_COPIES |
7506f491 | 2050 | Define this macro if the compiler should avoid copies to/from @code{CCmode} |
a89608cb | 2051 | registers. You should only define this macro if support for copying to/from |
7506f491 | 2052 | @code{CCmode} is incomplete. |
a2c4f8e0 | 2053 | @end defmac |
feca2ed3 JW |
2054 | |
2055 | @node Leaf Functions | |
2056 | @subsection Handling Leaf Functions | |
2057 | ||
2058 | @cindex leaf functions | |
2059 | @cindex functions, leaf | |
2060 | On some machines, a leaf function (i.e., one which makes no calls) can run | |
2061 | more efficiently if it does not make its own register window. Often this | |
2062 | means it is required to receive its arguments in the registers where they | |
2063 | are passed by the caller, instead of the registers where they would | |
2064 | normally arrive. | |
2065 | ||
2066 | The special treatment for leaf functions generally applies only when | |
2067 | other conditions are met; for example, often they may use only those | |
2068 | registers for its own variables and temporaries. We use the term ``leaf | |
2069 | function'' to mean a function that is suitable for this special | |
2070 | handling, so that functions with no calls are not necessarily ``leaf | |
2071 | functions''. | |
2072 | ||
a3a15b4d | 2073 | GCC assigns register numbers before it knows whether the function is |
feca2ed3 JW |
2074 | suitable for leaf function treatment. So it needs to renumber the |
2075 | registers in order to output a leaf function. The following macros | |
2076 | accomplish this. | |
2077 | ||
a2c4f8e0 | 2078 | @defmac LEAF_REGISTERS |
7d167afd | 2079 | Name of a char vector, indexed by hard register number, which |
feca2ed3 JW |
2080 | contains 1 for a register that is allowable in a candidate for leaf |
2081 | function treatment. | |
2082 | ||
2083 | If leaf function treatment involves renumbering the registers, then the | |
2084 | registers marked here should be the ones before renumbering---those that | |
a3a15b4d | 2085 | GCC would ordinarily allocate. The registers which will actually be |
feca2ed3 JW |
2086 | used in the assembler code, after renumbering, should not be marked with 1 |
2087 | in this vector. | |
2088 | ||
2089 | Define this macro only if the target machine offers a way to optimize | |
2090 | the treatment of leaf functions. | |
a2c4f8e0 | 2091 | @end defmac |
feca2ed3 | 2092 | |
a2c4f8e0 | 2093 | @defmac LEAF_REG_REMAP (@var{regno}) |
feca2ed3 JW |
2094 | A C expression whose value is the register number to which @var{regno} |
2095 | should be renumbered, when a function is treated as a leaf function. | |
2096 | ||
2097 | If @var{regno} is a register number which should not appear in a leaf | |
630d3d5a | 2098 | function before renumbering, then the expression should yield @minus{}1, which |
feca2ed3 JW |
2099 | will cause the compiler to abort. |
2100 | ||
2101 | Define this macro only if the target machine offers a way to optimize the | |
2102 | treatment of leaf functions, and registers need to be renumbered to do | |
2103 | this. | |
a2c4f8e0 | 2104 | @end defmac |
feca2ed3 | 2105 | |
54ff41b7 JW |
2106 | @findex current_function_is_leaf |
2107 | @findex current_function_uses_only_leaf_regs | |
c237e94a ZW |
2108 | @code{TARGET_ASM_FUNCTION_PROLOGUE} and |
2109 | @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions | |
2110 | specially. They can test the C variable @code{current_function_is_leaf} | |
2111 | which is nonzero for leaf functions. @code{current_function_is_leaf} is | |
2112 | set prior to local register allocation and is valid for the remaining | |
08c148a8 NB |
2113 | compiler passes. They can also test the C variable |
2114 | @code{current_function_uses_only_leaf_regs} which is nonzero for leaf | |
2115 | functions which only use leaf registers. | |
9ac617d4 EB |
2116 | @code{current_function_uses_only_leaf_regs} is valid after all passes |
2117 | that modify the instructions have been run and is only useful if | |
2118 | @code{LEAF_REGISTERS} is defined. | |
feca2ed3 JW |
2119 | @c changed this to fix overfull. ALSO: why the "it" at the beginning |
2120 | @c of the next paragraph?! --mew 2feb93 | |
2121 | ||
2122 | @node Stack Registers | |
2123 | @subsection Registers That Form a Stack | |
2124 | ||
2125 | There are special features to handle computers where some of the | |
a2c4f8e0 ZW |
2126 | ``registers'' form a stack. Stack registers are normally written by |
2127 | pushing onto the stack, and are numbered relative to the top of the | |
2128 | stack. | |
feca2ed3 | 2129 | |
a3a15b4d | 2130 | Currently, GCC can only handle one group of stack-like registers, and |
a2c4f8e0 ZW |
2131 | they must be consecutively numbered. Furthermore, the existing |
2132 | support for stack-like registers is specific to the 80387 floating | |
2133 | point coprocessor. If you have a new architecture that uses | |
2134 | stack-like registers, you will need to do substantial work on | |
2135 | @file{reg-stack.c} and write your machine description to cooperate | |
2136 | with it, as well as defining these macros. | |
2137 | ||
2138 | @defmac STACK_REGS | |
feca2ed3 | 2139 | Define this if the machine has any stack-like registers. |
a2c4f8e0 | 2140 | @end defmac |
feca2ed3 | 2141 | |
a2c4f8e0 | 2142 | @defmac FIRST_STACK_REG |
feca2ed3 JW |
2143 | The number of the first stack-like register. This one is the top |
2144 | of the stack. | |
a2c4f8e0 | 2145 | @end defmac |
feca2ed3 | 2146 | |
a2c4f8e0 | 2147 | @defmac LAST_STACK_REG |
feca2ed3 JW |
2148 | The number of the last stack-like register. This one is the bottom of |
2149 | the stack. | |
a2c4f8e0 | 2150 | @end defmac |
feca2ed3 | 2151 | |
feca2ed3 JW |
2152 | @node Register Classes |
2153 | @section Register Classes | |
2154 | @cindex register class definitions | |
2155 | @cindex class definitions, register | |
2156 | ||
2157 | On many machines, the numbered registers are not all equivalent. | |
2158 | For example, certain registers may not be allowed for indexed addressing; | |
2159 | certain registers may not be allowed in some instructions. These machine | |
2160 | restrictions are described to the compiler using @dfn{register classes}. | |
2161 | ||
2162 | You define a number of register classes, giving each one a name and saying | |
2163 | which of the registers belong to it. Then you can specify register classes | |
2164 | that are allowed as operands to particular instruction patterns. | |
2165 | ||
2166 | @findex ALL_REGS | |
2167 | @findex NO_REGS | |
2168 | In general, each register will belong to several classes. In fact, one | |
2169 | class must be named @code{ALL_REGS} and contain all the registers. Another | |
2170 | class must be named @code{NO_REGS} and contain no registers. Often the | |
2171 | union of two classes will be another class; however, this is not required. | |
2172 | ||
2173 | @findex GENERAL_REGS | |
2174 | One of the classes must be named @code{GENERAL_REGS}. There is nothing | |
2175 | terribly special about the name, but the operand constraint letters | |
2176 | @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is | |
2177 | the same as @code{ALL_REGS}, just define it as a macro which expands | |
2178 | to @code{ALL_REGS}. | |
2179 | ||
2180 | Order the classes so that if class @var{x} is contained in class @var{y} | |
2181 | then @var{x} has a lower class number than @var{y}. | |
2182 | ||
2183 | The way classes other than @code{GENERAL_REGS} are specified in operand | |
2184 | constraints is through machine-dependent operand constraint letters. | |
2185 | You can define such letters to correspond to various classes, then use | |
2186 | them in operand constraints. | |
2187 | ||
2188 | You should define a class for the union of two classes whenever some | |
2189 | instruction allows both classes. For example, if an instruction allows | |
2190 | either a floating point (coprocessor) register or a general register for a | |
2191 | certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS} | |
2192 | which includes both of them. Otherwise you will get suboptimal code. | |
2193 | ||
2194 | You must also specify certain redundant information about the register | |
2195 | classes: for each class, which classes contain it and which ones are | |
2196 | contained in it; for each pair of classes, the largest class contained | |
2197 | in their union. | |
2198 | ||
2199 | When a value occupying several consecutive registers is expected in a | |
2200 | certain class, all the registers used must belong to that class. | |
2201 | Therefore, register classes cannot be used to enforce a requirement for | |
2202 | a register pair to start with an even-numbered register. The way to | |
2203 | specify this requirement is with @code{HARD_REGNO_MODE_OK}. | |
2204 | ||
2205 | Register classes used for input-operands of bitwise-and or shift | |
2206 | instructions have a special requirement: each such class must have, for | |
2207 | each fixed-point machine mode, a subclass whose registers can transfer that | |
2208 | mode to or from memory. For example, on some machines, the operations for | |
2209 | single-byte values (@code{QImode}) are limited to certain registers. When | |
2210 | this is so, each register class that is used in a bitwise-and or shift | |
2211 | instruction must have a subclass consisting of registers from which | |
2212 | single-byte values can be loaded or stored. This is so that | |
2213 | @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return. | |
2214 | ||
a2c4f8e0 | 2215 | @deftp {Data type} {enum reg_class} |
2eac577f JM |
2216 | An enumerated type that must be defined with all the register class names |
2217 | as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS} | |
2218 | must be the last register class, followed by one more enumerated value, | |
feca2ed3 JW |
2219 | @code{LIM_REG_CLASSES}, which is not a register class but rather |
2220 | tells how many classes there are. | |
2221 | ||
2222 | Each register class has a number, which is the value of casting | |
2223 | the class name to type @code{int}. The number serves as an index | |
2224 | in many of the tables described below. | |
a2c4f8e0 | 2225 | @end deftp |
feca2ed3 | 2226 | |
a2c4f8e0 | 2227 | @defmac N_REG_CLASSES |
feca2ed3 JW |
2228 | The number of distinct register classes, defined as follows: |
2229 | ||
3ab51846 | 2230 | @smallexample |
feca2ed3 | 2231 | #define N_REG_CLASSES (int) LIM_REG_CLASSES |
3ab51846 | 2232 | @end smallexample |
a2c4f8e0 | 2233 | @end defmac |
feca2ed3 | 2234 | |
a2c4f8e0 | 2235 | @defmac REG_CLASS_NAMES |
feca2ed3 JW |
2236 | An initializer containing the names of the register classes as C string |
2237 | constants. These names are used in writing some of the debugging dumps. | |
a2c4f8e0 | 2238 | @end defmac |
feca2ed3 | 2239 | |
a2c4f8e0 | 2240 | @defmac REG_CLASS_CONTENTS |
feca2ed3 JW |
2241 | An initializer containing the contents of the register classes, as integers |
2242 | which are bit masks. The @var{n}th integer specifies the contents of class | |
2243 | @var{n}. The way the integer @var{mask} is interpreted is that | |
2244 | register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1. | |
2245 | ||
2246 | When the machine has more than 32 registers, an integer does not suffice. | |
2247 | Then the integers are replaced by sub-initializers, braced groupings containing | |
2248 | several integers. Each sub-initializer must be suitable as an initializer | |
2249 | for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}. | |
7c272079 MP |
2250 | In this situation, the first integer in each sub-initializer corresponds to |
2251 | registers 0 through 31, the second integer to registers 32 through 63, and | |
2252 | so on. | |
a2c4f8e0 | 2253 | @end defmac |
feca2ed3 | 2254 | |
a2c4f8e0 | 2255 | @defmac REGNO_REG_CLASS (@var{regno}) |
feca2ed3 JW |
2256 | A C expression whose value is a register class containing hard register |
2257 | @var{regno}. In general there is more than one such class; choose a class | |
2258 | which is @dfn{minimal}, meaning that no smaller class also contains the | |
2259 | register. | |
a2c4f8e0 | 2260 | @end defmac |
feca2ed3 | 2261 | |
a2c4f8e0 | 2262 | @defmac BASE_REG_CLASS |
feca2ed3 JW |
2263 | A macro whose definition is the name of the class to which a valid |
2264 | base register must belong. A base register is one used in an address | |
2265 | which is the register value plus a displacement. | |
a2c4f8e0 | 2266 | @end defmac |
feca2ed3 | 2267 | |
a2c4f8e0 | 2268 | @defmac MODE_BASE_REG_CLASS (@var{mode}) |
3dcc68a4 | 2269 | This is a variation of the @code{BASE_REG_CLASS} macro which allows |
c0478a66 | 2270 | the selection of a base register in a mode dependent manner. If |
3dcc68a4 NC |
2271 | @var{mode} is VOIDmode then it should return the same value as |
2272 | @code{BASE_REG_CLASS}. | |
a2c4f8e0 | 2273 | @end defmac |
3dcc68a4 | 2274 | |
888d2cd6 DJ |
2275 | @defmac MODE_BASE_REG_REG_CLASS (@var{mode}) |
2276 | A C expression whose value is the register class to which a valid | |
2277 | base register must belong in order to be used in a base plus index | |
2278 | register address. You should define this macro if base plus index | |
2279 | addresses have different requirements than other base register uses. | |
2280 | @end defmac | |
2281 | ||
a2c4f8e0 | 2282 | @defmac INDEX_REG_CLASS |
feca2ed3 JW |
2283 | A macro whose definition is the name of the class to which a valid |
2284 | index register must belong. An index register is one used in an | |
2285 | address where its value is either multiplied by a scale factor or | |
2286 | added to another register (as well as added to a displacement). | |
a2c4f8e0 | 2287 | @end defmac |
feca2ed3 | 2288 | |
a2c4f8e0 | 2289 | @defmac CONSTRAINT_LEN (@var{char}, @var{str}) |
97488870 R |
2290 | For the constraint at the start of @var{str}, which starts with the letter |
2291 | @var{c}, return the length. This allows you to have register class / | |
2292 | constant / extra constraints that are longer than a single letter; | |
2293 | you don't need to define this macro if you can do with single-letter | |
2294 | constraints only. The definition of this macro should use | |
2295 | DEFAULT_CONSTRAINT_LEN for all the characters that you don't want | |
2296 | to handle specially. | |
2297 | There are some sanity checks in genoutput.c that check the constraint lengths | |
2298 | for the md file, so you can also use this macro to help you while you are | |
2299 | transitioning from a byzantine single-letter-constraint scheme: when you | |
2300 | return a negative length for a constraint you want to re-use, genoutput | |
2301 | will complain about every instance where it is used in the md file. | |
a2c4f8e0 | 2302 | @end defmac |
97488870 | 2303 | |
a2c4f8e0 | 2304 | @defmac REG_CLASS_FROM_LETTER (@var{char}) |
feca2ed3 JW |
2305 | A C expression which defines the machine-dependent operand constraint |
2306 | letters for register classes. If @var{char} is such a letter, the | |
2307 | value should be the register class corresponding to it. Otherwise, | |
2308 | the value should be @code{NO_REGS}. The register letter @samp{r}, | |
2309 | corresponding to class @code{GENERAL_REGS}, will not be passed | |
2310 | to this macro; you do not need to handle it. | |
a2c4f8e0 | 2311 | @end defmac |
feca2ed3 | 2312 | |
a2c4f8e0 | 2313 | @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str}) |
97488870 R |
2314 | Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string |
2315 | passed in @var{str}, so that you can use suffixes to distinguish between | |
2316 | different variants. | |
a2c4f8e0 | 2317 | @end defmac |
97488870 | 2318 | |
a2c4f8e0 | 2319 | @defmac REGNO_OK_FOR_BASE_P (@var{num}) |
feca2ed3 JW |
2320 | A C expression which is nonzero if register number @var{num} is |
2321 | suitable for use as a base register in operand addresses. It may be | |
2322 | either a suitable hard register or a pseudo register that has been | |
2323 | allocated such a hard register. | |
a2c4f8e0 | 2324 | @end defmac |
feca2ed3 | 2325 | |
a2c4f8e0 | 2326 | @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode}) |
861bb6c1 JL |
2327 | A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that |
2328 | that expression may examine the mode of the memory reference in | |
2329 | @var{mode}. You should define this macro if the mode of the memory | |
2330 | reference affects whether a register may be used as a base register. If | |
2331 | you define this macro, the compiler will use it instead of | |
2332 | @code{REGNO_OK_FOR_BASE_P}. | |
a2c4f8e0 | 2333 | @end defmac |
861bb6c1 | 2334 | |
888d2cd6 DJ |
2335 | @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode}) |
2336 | A C expression which is nonzero if register number @var{num} is suitable for | |
2337 | use as a base register in base plus index operand addresses, accessing | |
2338 | memory in mode @var{mode}. It may be either a suitable hard register or a | |
2339 | pseudo register that has been allocated such a hard register. You should | |
2340 | define this macro if base plus index addresses have different requirements | |
2341 | than other base register uses. | |
2342 | @end defmac | |
2343 | ||
a2c4f8e0 | 2344 | @defmac REGNO_OK_FOR_INDEX_P (@var{num}) |
feca2ed3 JW |
2345 | A C expression which is nonzero if register number @var{num} is |
2346 | suitable for use as an index register in operand addresses. It may be | |
2347 | either a suitable hard register or a pseudo register that has been | |
2348 | allocated such a hard register. | |
2349 | ||
2350 | The difference between an index register and a base register is that | |
2351 | the index register may be scaled. If an address involves the sum of | |
2352 | two registers, neither one of them scaled, then either one may be | |
2353 | labeled the ``base'' and the other the ``index''; but whichever | |
2354 | labeling is used must fit the machine's constraints of which registers | |
2355 | may serve in each capacity. The compiler will try both labelings, | |
2356 | looking for one that is valid, and will reload one or both registers | |
2357 | only if neither labeling works. | |
a2c4f8e0 | 2358 | @end defmac |
feca2ed3 | 2359 | |
a2c4f8e0 | 2360 | @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class}) |
feca2ed3 JW |
2361 | A C expression that places additional restrictions on the register class |
2362 | to use when it is necessary to copy value @var{x} into a register in class | |
2363 | @var{class}. The value is a register class; perhaps @var{class}, or perhaps | |
2364 | another, smaller class. On many machines, the following definition is | |
2365 | safe: | |
2366 | ||
3ab51846 | 2367 | @smallexample |
feca2ed3 | 2368 | #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS |
3ab51846 | 2369 | @end smallexample |
feca2ed3 JW |
2370 | |
2371 | Sometimes returning a more restrictive class makes better code. For | |
2372 | example, on the 68000, when @var{x} is an integer constant that is in range | |
2373 | for a @samp{moveq} instruction, the value of this macro is always | |
2374 | @code{DATA_REGS} as long as @var{class} includes the data registers. | |
2375 | Requiring a data register guarantees that a @samp{moveq} will be used. | |
2376 | ||
222a2f1a GK |
2377 | One case where @code{PREFERRED_RELOAD_CLASS} must not return |
2378 | @var{class} is if @var{x} is a legitimate constant which cannot be | |
2379 | loaded into some register class. By returning @code{NO_REGS} you can | |
2380 | force @var{x} into a memory location. For example, rs6000 can load | |
2381 | immediate values into general-purpose registers, but does not have an | |
2382 | instruction for loading an immediate value into a floating-point | |
2383 | register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when | |
2384 | @var{x} is a floating-point constant. If the constant can't be loaded | |
2385 | into any kind of register, code generation will be better if | |
2386 | @code{LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead | |
2387 | of using @code{PREFERRED_RELOAD_CLASS}. | |
a2c4f8e0 | 2388 | @end defmac |
feca2ed3 | 2389 | |
a2c4f8e0 | 2390 | @defmac PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class}) |
feca2ed3 JW |
2391 | Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of |
2392 | input reloads. If you don't define this macro, the default is to use | |
2393 | @var{class}, unchanged. | |
a2c4f8e0 | 2394 | @end defmac |
feca2ed3 | 2395 | |
a2c4f8e0 | 2396 | @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class}) |
feca2ed3 JW |
2397 | A C expression that places additional restrictions on the register class |
2398 | to use when it is necessary to be able to hold a value of mode | |
2399 | @var{mode} in a reload register for which class @var{class} would | |
2400 | ordinarily be used. | |
2401 | ||
2402 | Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when | |
2403 | there are certain modes that simply can't go in certain reload classes. | |
2404 | ||
2405 | The value is a register class; perhaps @var{class}, or perhaps another, | |
2406 | smaller class. | |
2407 | ||
2408 | Don't define this macro unless the target machine has limitations which | |
2409 | require the macro to do something nontrivial. | |
a2c4f8e0 | 2410 | @end defmac |
feca2ed3 | 2411 | |
a2c4f8e0 ZW |
2412 | @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x}) |
2413 | @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x}) | |
2414 | @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x}) | |
feca2ed3 JW |
2415 | Many machines have some registers that cannot be copied directly to or |
2416 | from memory or even from other types of registers. An example is the | |
2417 | @samp{MQ} register, which on most machines, can only be copied to or | |
2418 | from general registers, but not memory. Some machines allow copying all | |
2419 | registers to and from memory, but require a scratch register for stores | |
2420 | to some memory locations (e.g., those with symbolic address on the RT, | |
981f6289 | 2421 | and those with certain symbolic address on the SPARC when compiling |
161d7b59 | 2422 | PIC)@. In some cases, both an intermediate and a scratch register are |
feca2ed3 JW |
2423 | required. |
2424 | ||
2425 | You should define these macros to indicate to the reload phase that it may | |
2426 | need to allocate at least one register for a reload in addition to the | |
2427 | register to contain the data. Specifically, if copying @var{x} to a | |
2428 | register @var{class} in @var{mode} requires an intermediate register, | |
2429 | you should define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the | |
2430 | largest register class all of whose registers can be used as | |
2431 | intermediate registers or scratch registers. | |
2432 | ||
2433 | If copying a register @var{class} in @var{mode} to @var{x} requires an | |
2434 | intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS} | |
2435 | should be defined to return the largest register class required. If the | |
2436 | requirements for input and output reloads are the same, the macro | |
2437 | @code{SECONDARY_RELOAD_CLASS} should be used instead of defining both | |
2438 | macros identically. | |
2439 | ||
2440 | The values returned by these macros are often @code{GENERAL_REGS}. | |
2441 | Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x} | |
2442 | can be directly copied to or from a register of @var{class} in | |
2443 | @var{mode} without requiring a scratch register. Do not define this | |
2444 | macro if it would always return @code{NO_REGS}. | |
2445 | ||
2446 | If a scratch register is required (either with or without an | |
2447 | intermediate register), you should define patterns for | |
2448 | @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required | |
2449 | (@pxref{Standard Names}. These patterns, which will normally be | |
2450 | implemented with a @code{define_expand}, should be similar to the | |
2451 | @samp{mov@var{m}} patterns, except that operand 2 is the scratch | |
2452 | register. | |
2453 | ||
2454 | Define constraints for the reload register and scratch register that | |
2455 | contain a single register class. If the original reload register (whose | |
2456 | class is @var{class}) can meet the constraint given in the pattern, the | |
2457 | value returned by these macros is used for the class of the scratch | |
2458 | register. Otherwise, two additional reload registers are required. | |
2459 | Their classes are obtained from the constraints in the insn pattern. | |
2460 | ||
2461 | @var{x} might be a pseudo-register or a @code{subreg} of a | |
2462 | pseudo-register, which could either be in a hard register or in memory. | |
630d3d5a | 2463 | Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is |
feca2ed3 JW |
2464 | in memory and the hard register number if it is in a register. |
2465 | ||
2466 | These macros should not be used in the case where a particular class of | |
2467 | registers can only be copied to memory and not to another class of | |
2468 | registers. In that case, secondary reload registers are not needed and | |
2469 | would not be helpful. Instead, a stack location must be used to perform | |
a8154559 | 2470 | the copy and the @code{mov@var{m}} pattern should use memory as an |
feca2ed3 JW |
2471 | intermediate storage. This case often occurs between floating-point and |
2472 | general registers. | |
a2c4f8e0 | 2473 | @end defmac |
feca2ed3 | 2474 | |
a2c4f8e0 | 2475 | @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m}) |
feca2ed3 JW |
2476 | Certain machines have the property that some registers cannot be copied |
2477 | to some other registers without using memory. Define this macro on | |
df2a54e9 | 2478 | those machines to be a C expression that is nonzero if objects of mode |
feca2ed3 JW |
2479 | @var{m} in registers of @var{class1} can only be copied to registers of |
2480 | class @var{class2} by storing a register of @var{class1} into memory | |
2481 | and loading that memory location into a register of @var{class2}. | |
2482 | ||
2483 | Do not define this macro if its value would always be zero. | |
a2c4f8e0 | 2484 | @end defmac |
feca2ed3 | 2485 | |
a2c4f8e0 | 2486 | @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode}) |
feca2ed3 JW |
2487 | Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler |
2488 | allocates a stack slot for a memory location needed for register copies. | |
2489 | If this macro is defined, the compiler instead uses the memory location | |
2490 | defined by this macro. | |
2491 | ||
2492 | Do not define this macro if you do not define | |
2493 | @code{SECONDARY_MEMORY_NEEDED}. | |
a2c4f8e0 | 2494 | @end defmac |
feca2ed3 | 2495 | |
a2c4f8e0 | 2496 | @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode}) |
feca2ed3 JW |
2497 | When the compiler needs a secondary memory location to copy between two |
2498 | registers of mode @var{mode}, it normally allocates sufficient memory to | |
2499 | hold a quantity of @code{BITS_PER_WORD} bits and performs the store and | |
2500 | load operations in a mode that many bits wide and whose class is the | |
2501 | same as that of @var{mode}. | |
2502 | ||
2503 | This is right thing to do on most machines because it ensures that all | |
2504 | bits of the register are copied and prevents accesses to the registers | |
2505 | in a narrower mode, which some machines prohibit for floating-point | |
2506 | registers. | |
2507 | ||
2508 | However, this default behavior is not correct on some machines, such as | |
2509 | the DEC Alpha, that store short integers in floating-point registers | |
2510 | differently than in integer registers. On those machines, the default | |
2511 | widening will not work correctly and you must define this macro to | |
2512 | suppress that widening in some cases. See the file @file{alpha.h} for | |
2513 | details. | |
2514 | ||
2515 | Do not define this macro if you do not define | |
2516 | @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that | |
2517 | is @code{BITS_PER_WORD} bits wide is correct for your machine. | |
a2c4f8e0 | 2518 | @end defmac |
feca2ed3 | 2519 | |
a2c4f8e0 | 2520 | @defmac SMALL_REGISTER_CLASSES |
faa9eb19 BS |
2521 | On some machines, it is risky to let hard registers live across arbitrary |
2522 | insns. Typically, these machines have instructions that require values | |
2523 | to be in specific registers (like an accumulator), and reload will fail | |
2524 | if the required hard register is used for another purpose across such an | |
2525 | insn. | |
feca2ed3 | 2526 | |
df2a54e9 JM |
2527 | Define @code{SMALL_REGISTER_CLASSES} to be an expression with a nonzero |
2528 | value on these machines. When this macro has a nonzero value, the | |
faa9eb19 | 2529 | compiler will try to minimize the lifetime of hard registers. |
feca2ed3 | 2530 | |
df2a54e9 | 2531 | It is always safe to define this macro with a nonzero value, but if you |
861bb6c1 JL |
2532 | unnecessarily define it, you will reduce the amount of optimizations |
2533 | that can be performed in some cases. If you do not define this macro | |
df2a54e9 | 2534 | with a nonzero value when it is required, the compiler will run out of |
861bb6c1 JL |
2535 | spill registers and print a fatal error message. For most machines, you |
2536 | should not define this macro at all. | |
a2c4f8e0 | 2537 | @end defmac |
feca2ed3 | 2538 | |
a2c4f8e0 | 2539 | @defmac CLASS_LIKELY_SPILLED_P (@var{class}) |
feca2ed3 JW |
2540 | A C expression whose value is nonzero if pseudos that have been assigned |
2541 | to registers of class @var{class} would likely be spilled because | |
2542 | registers of @var{class} are needed for spill registers. | |
2543 | ||
2544 | The default value of this macro returns 1 if @var{class} has exactly one | |
2545 | register and zero otherwise. On most machines, this default should be | |
40687a9e | 2546 | used. Only define this macro to some other expression if pseudos |
feca2ed3 JW |
2547 | allocated by @file{local-alloc.c} end up in memory because their hard |
2548 | registers were needed for spill registers. If this macro returns nonzero | |
2549 | for those classes, those pseudos will only be allocated by | |
2550 | @file{global.c}, which knows how to reallocate the pseudo to another | |
2551 | register. If there would not be another register available for | |
2552 | reallocation, you should not change the definition of this macro since | |
2553 | the only effect of such a definition would be to slow down register | |
2554 | allocation. | |
a2c4f8e0 | 2555 | @end defmac |
feca2ed3 | 2556 | |
a2c4f8e0 | 2557 | @defmac CLASS_MAX_NREGS (@var{class}, @var{mode}) |
feca2ed3 JW |
2558 | A C expression for the maximum number of consecutive registers |
2559 | of class @var{class} needed to hold a value of mode @var{mode}. | |
2560 | ||
2561 | This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact, | |
2562 | the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})} | |
2563 | should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno}, | |
2564 | @var{mode})} for all @var{regno} values in the class @var{class}. | |
2565 | ||
2566 | This macro helps control the handling of multiple-word values | |
2567 | in the reload pass. | |
a2c4f8e0 | 2568 | @end defmac |
feca2ed3 | 2569 | |
a2c4f8e0 | 2570 | @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class}) |
b0c42aed JH |
2571 | If defined, a C expression that returns nonzero for a @var{class} for which |
2572 | a change from mode @var{from} to mode @var{to} is invalid. | |
feca2ed3 JW |
2573 | |
2574 | For the example, loading 32-bit integer or floating-point objects into | |
57694e40 | 2575 | floating-point registers on the Alpha extends them to 64 bits. |
feca2ed3 | 2576 | Therefore loading a 64-bit object and then storing it as a 32-bit object |
57694e40 | 2577 | does not store the low-order 32 bits, as would be the case for a normal |
cff9f8d5 AH |
2578 | register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS} |
2579 | as below: | |
02188693 | 2580 | |
3ab51846 | 2581 | @smallexample |
b0c42aed JH |
2582 | #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \ |
2583 | (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \ | |
2584 | ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0) | |
3ab51846 | 2585 | @end smallexample |
a2c4f8e0 | 2586 | @end defmac |
feca2ed3 JW |
2587 | |
2588 | Three other special macros describe which operands fit which constraint | |
2589 | letters. | |
2590 | ||
a2c4f8e0 | 2591 | @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c}) |
e119b68c MM |
2592 | A C expression that defines the machine-dependent operand constraint |
2593 | letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify | |
2594 | particular ranges of integer values. If @var{c} is one of those | |
2595 | letters, the expression should check that @var{value}, an integer, is in | |
2596 | the appropriate range and return 1 if so, 0 otherwise. If @var{c} is | |
2597 | not one of those letters, the value should be 0 regardless of | |
2598 | @var{value}. | |
a2c4f8e0 | 2599 | @end defmac |
feca2ed3 | 2600 | |
a2c4f8e0 | 2601 | @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str}) |
97488870 R |
2602 | Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint |
2603 | string passed in @var{str}, so that you can use suffixes to distinguish | |
2604 | between different variants. | |
a2c4f8e0 | 2605 | @end defmac |
97488870 | 2606 | |
a2c4f8e0 | 2607 | @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c}) |
feca2ed3 | 2608 | A C expression that defines the machine-dependent operand constraint |
e119b68c MM |
2609 | letters that specify particular ranges of @code{const_double} values |
2610 | (@samp{G} or @samp{H}). | |
feca2ed3 JW |
2611 | |
2612 | If @var{c} is one of those letters, the expression should check that | |
2613 | @var{value}, an RTX of code @code{const_double}, is in the appropriate | |
2614 | range and return 1 if so, 0 otherwise. If @var{c} is not one of those | |
2615 | letters, the value should be 0 regardless of @var{value}. | |
2616 | ||
2617 | @code{const_double} is used for all floating-point constants and for | |
2618 | @code{DImode} fixed-point constants. A given letter can accept either | |
2619 | or both kinds of values. It can use @code{GET_MODE} to distinguish | |
2620 | between these kinds. | |
a2c4f8e0 | 2621 | @end defmac |
feca2ed3 | 2622 | |
a2c4f8e0 | 2623 | @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str}) |
97488870 R |
2624 | Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint |
2625 | string passed in @var{str}, so that you can use suffixes to distinguish | |
2626 | between different variants. | |
a2c4f8e0 | 2627 | @end defmac |
97488870 | 2628 | |
a2c4f8e0 | 2629 | @defmac EXTRA_CONSTRAINT (@var{value}, @var{c}) |
feca2ed3 | 2630 | A C expression that defines the optional machine-dependent constraint |
c2cba7a9 RH |
2631 | letters that can be used to segregate specific types of operands, usually |
2632 | memory references, for the target machine. Any letter that is not | |
97488870 R |
2633 | elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} / |
2634 | @code{REG_CLASS_FROM_CONSTRAINT} | |
c2cba7a9 RH |
2635 | may be used. Normally this macro will not be defined. |
2636 | ||
2637 | If it is required for a particular target machine, it should return 1 | |
2638 | if @var{value} corresponds to the operand type represented by the | |
2639 | constraint letter @var{c}. If @var{c} is not defined as an extra | |
e119b68c | 2640 | constraint, the value returned should be 0 regardless of @var{value}. |
feca2ed3 | 2641 | |
c2cba7a9 RH |
2642 | For example, on the ROMP, load instructions cannot have their output |
2643 | in r0 if the memory reference contains a symbolic address. Constraint | |
2644 | letter @samp{Q} is defined as representing a memory address that does | |
feca2ed3 JW |
2645 | @emph{not} contain a symbolic address. An alternative is specified with |
2646 | a @samp{Q} constraint on the input and @samp{r} on the output. The next | |
2647 | alternative specifies @samp{m} on the input and a register class that | |
2648 | does not include r0 on the output. | |
a2c4f8e0 | 2649 | @end defmac |
ccfc6cc8 | 2650 | |
a2c4f8e0 | 2651 | @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str}) |
97488870 R |
2652 | Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed |
2653 | in @var{str}, so that you can use suffixes to distinguish between different | |
2654 | variants. | |
a2c4f8e0 | 2655 | @end defmac |
97488870 | 2656 | |
a2c4f8e0 | 2657 | @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str}) |
ccfc6cc8 UW |
2658 | A C expression that defines the optional machine-dependent constraint |
2659 | letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should | |
2660 | be treated like memory constraints by the reload pass. | |
2661 | ||
73774972 | 2662 | It should return 1 if the operand type represented by the constraint |
97488870 R |
2663 | at the start of @var{str}, the first letter of which is the letter @var{c}, |
2664 | comprises a subset of all memory references including | |
73774972 EC |
2665 | all those whose address is simply a base register. This allows the reload |
2666 | pass to reload an operand, if it does not directly correspond to the operand | |
ccfc6cc8 UW |
2667 | type of @var{c}, by copying its address into a base register. |
2668 | ||
2669 | For example, on the S/390, some instructions do not accept arbitrary | |
2670 | memory references, but only those that do not make use of an index | |
2671 | register. The constraint letter @samp{Q} is defined via | |
2672 | @code{EXTRA_CONSTRAINT} as representing a memory address of this type. | |
2673 | If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT}, | |
2674 | a @samp{Q} constraint can handle any memory operand, because the | |
2675 | reload pass knows it can be reloaded by copying the memory address | |
2676 | into a base register if required. This is analogous to the way | |
2677 | a @samp{o} constraint can handle any memory operand. | |
a2c4f8e0 | 2678 | @end defmac |
ccfc6cc8 | 2679 | |
a2c4f8e0 | 2680 | @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str}) |
ccfc6cc8 | 2681 | A C expression that defines the optional machine-dependent constraint |
97488870 R |
2682 | letters, amongst those accepted by @code{EXTRA_CONSTRAINT} / |
2683 | @code{EXTRA_CONSTRAINT_STR}, that should | |
ccfc6cc8 UW |
2684 | be treated like address constraints by the reload pass. |
2685 | ||
73774972 | 2686 | It should return 1 if the operand type represented by the constraint |
3a6e2189 | 2687 | at the start of @var{str}, which starts with the letter @var{c}, comprises |
97488870 | 2688 | a subset of all memory addresses including |
73774972 EC |
2689 | all those that consist of just a base register. This allows the reload |
2690 | pass to reload an operand, if it does not directly correspond to the operand | |
97488870 | 2691 | type of @var{str}, by copying it into a base register. |
ccfc6cc8 UW |
2692 | |
2693 | Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only | |
73774972 | 2694 | be used with the @code{address_operand} predicate. It is treated |
ccfc6cc8 | 2695 | analogously to the @samp{p} constraint. |
a2c4f8e0 | 2696 | @end defmac |
feca2ed3 JW |
2697 | |
2698 | @node Stack and Calling | |
2699 | @section Stack Layout and Calling Conventions | |
2700 | @cindex calling conventions | |
2701 | ||
2702 | @c prevent bad page break with this line | |
2703 | This describes the stack layout and calling conventions. | |
2704 | ||
2705 | @menu | |
2706 | * Frame Layout:: | |
7c16328b | 2707 | * Exception Handling:: |
861bb6c1 | 2708 | * Stack Checking:: |
feca2ed3 JW |
2709 | * Frame Registers:: |
2710 | * Elimination:: | |
2711 | * Stack Arguments:: | |
2712 | * Register Arguments:: | |
2713 | * Scalar Return:: | |
2714 | * Aggregate Return:: | |
2715 | * Caller Saves:: | |
2716 | * Function Entry:: | |
2717 | * Profiling:: | |
91d231cb | 2718 | * Tail Calls:: |
7d69de61 | 2719 | * Stack Smashing Protection:: |
feca2ed3 JW |
2720 | @end menu |
2721 | ||
2722 | @node Frame Layout | |
2723 | @subsection Basic Stack Layout | |
2724 | @cindex stack frame layout | |
2725 | @cindex frame layout | |
2726 | ||
2727 | @c prevent bad page break with this line | |
2728 | Here is the basic stack layout. | |
2729 | ||
a2c4f8e0 | 2730 | @defmac STACK_GROWS_DOWNWARD |
feca2ed3 JW |
2731 | Define this macro if pushing a word onto the stack moves the stack |
2732 | pointer to a smaller address. | |
2733 | ||
d78aa55c | 2734 | When we say, ``define this macro if @dots{}'', it means that the |
feca2ed3 JW |
2735 | compiler checks this macro only with @code{#ifdef} so the precise |
2736 | definition used does not matter. | |
a2c4f8e0 | 2737 | @end defmac |
feca2ed3 | 2738 | |
a2c4f8e0 | 2739 | @defmac STACK_PUSH_CODE |
918a6124 GK |
2740 | This macro defines the operation used when something is pushed |
2741 | on the stack. In RTL, a push operation will be | |
04a5176a | 2742 | @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})} |
918a6124 GK |
2743 | |
2744 | The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC}, | |
2745 | and @code{POST_INC}. Which of these is correct depends on | |
2746 | the stack direction and on whether the stack pointer points | |
2747 | to the last item on the stack or whether it points to the | |
2748 | space for the next item on the stack. | |
2749 | ||
2750 | The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is | |
2751 | defined, which is almost always right, and @code{PRE_INC} otherwise, | |
2752 | which is often wrong. | |
a2c4f8e0 | 2753 | @end defmac |
918a6124 | 2754 | |
a2c4f8e0 | 2755 | @defmac FRAME_GROWS_DOWNWARD |
a4d05547 | 2756 | Define this macro to nonzero value if the addresses of local variable slots |
f62c8a5c | 2757 | are at negative offsets from the frame pointer. |
a2c4f8e0 | 2758 | @end defmac |
feca2ed3 | 2759 | |
a2c4f8e0 | 2760 | @defmac ARGS_GROW_DOWNWARD |
feca2ed3 JW |
2761 | Define this macro if successive arguments to a function occupy decreasing |
2762 | addresses on the stack. | |
a2c4f8e0 | 2763 | @end defmac |
feca2ed3 | 2764 | |
a2c4f8e0 | 2765 | @defmac STARTING_FRAME_OFFSET |
feca2ed3 JW |
2766 | Offset from the frame pointer to the first local variable slot to be allocated. |
2767 | ||
2768 | If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by | |
2769 | subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}. | |
2770 | Otherwise, it is found by adding the length of the first slot to the | |
2771 | value @code{STARTING_FRAME_OFFSET}. | |
2772 | @c i'm not sure if the above is still correct.. had to change it to get | |
2773 | @c rid of an overfull. --mew 2feb93 | |
a2c4f8e0 | 2774 | @end defmac |
feca2ed3 | 2775 | |
a2c4f8e0 | 2776 | @defmac STACK_ALIGNMENT_NEEDED |
95f3f59e | 2777 | Define to zero to disable final alignment of the stack during reload. |
0b4be7de | 2778 | The nonzero default for this macro is suitable for most ports. |
95f3f59e | 2779 | |
0b4be7de | 2780 | On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there |
95f3f59e JDA |
2781 | is a register save block following the local block that doesn't require |
2782 | alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable | |
2783 | stack alignment and do it in the backend. | |
a2c4f8e0 | 2784 | @end defmac |
95f3f59e | 2785 | |
a2c4f8e0 | 2786 | @defmac STACK_POINTER_OFFSET |
feca2ed3 JW |
2787 | Offset from the stack pointer register to the first location at which |
2788 | outgoing arguments are placed. If not specified, the default value of | |
2789 | zero is used. This is the proper value for most machines. | |
2790 | ||
2791 | If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above | |
2792 | the first location at which outgoing arguments are placed. | |
a2c4f8e0 | 2793 | @end defmac |
feca2ed3 | 2794 | |
a2c4f8e0 | 2795 | @defmac FIRST_PARM_OFFSET (@var{fundecl}) |
feca2ed3 JW |
2796 | Offset from the argument pointer register to the first argument's |
2797 | address. On some machines it may depend on the data type of the | |
2798 | function. | |
2799 | ||
2800 | If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above | |
2801 | the first argument's address. | |
a2c4f8e0 | 2802 | @end defmac |
feca2ed3 | 2803 | |
a2c4f8e0 | 2804 | @defmac STACK_DYNAMIC_OFFSET (@var{fundecl}) |
feca2ed3 JW |
2805 | Offset from the stack pointer register to an item dynamically allocated |
2806 | on the stack, e.g., by @code{alloca}. | |
2807 | ||
2808 | The default value for this macro is @code{STACK_POINTER_OFFSET} plus the | |
2809 | length of the outgoing arguments. The default is correct for most | |
2810 | machines. See @file{function.c} for details. | |
a2c4f8e0 | 2811 | @end defmac |
feca2ed3 | 2812 | |
c6d01079 AK |
2813 | @defmac INITIAL_FRAME_ADDRESS_RTX |
2814 | A C expression whose value is RTL representing the address of the initial | |
c8f27794 JW |
2815 | stack frame. This address is passed to @code{RETURN_ADDR_RTX} and |
2816 | @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable | |
2817 | default value will be used. Define this macro in order to make frame pointer | |
2818 | elimination work in the presence of @code{__builtin_frame_address (count)} and | |
2819 | @code{__builtin_return_address (count)} for @code{count} not equal to zero. | |
c6d01079 AK |
2820 | @end defmac |
2821 | ||
a2c4f8e0 | 2822 | @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr}) |
feca2ed3 JW |
2823 | A C expression whose value is RTL representing the address in a stack |
2824 | frame where the pointer to the caller's frame is stored. Assume that | |
2825 | @var{frameaddr} is an RTL expression for the address of the stack frame | |
2826 | itself. | |
2827 | ||
2828 | If you don't define this macro, the default is to return the value | |
2829 | of @var{frameaddr}---that is, the stack frame address is also the | |
2830 | address of the stack word that points to the previous frame. | |
a2c4f8e0 | 2831 | @end defmac |
feca2ed3 | 2832 | |
a2c4f8e0 | 2833 | @defmac SETUP_FRAME_ADDRESSES |
feca2ed3 JW |
2834 | If defined, a C expression that produces the machine-specific code to |
2835 | setup the stack so that arbitrary frames can be accessed. For example, | |
981f6289 | 2836 | on the SPARC, we must flush all of the register windows to the stack |
0bc02db4 MS |
2837 | before we can access arbitrary stack frames. You will seldom need to |
2838 | define this macro. | |
a2c4f8e0 | 2839 | @end defmac |
0bc02db4 | 2840 | |
d6da68b9 KH |
2841 | @deftypefn {Target Hook} bool TARGET_BUILTIN_SETJMP_FRAME_VALUE () |
2842 | This target hook should return an rtx that is used to store | |
0bc02db4 MS |
2843 | the address of the current frame into the built in @code{setjmp} buffer. |
2844 | The default value, @code{virtual_stack_vars_rtx}, is correct for most | |
d6da68b9 | 2845 | machines. One reason you may need to define this target hook is if |
0bc02db4 | 2846 | @code{hard_frame_pointer_rtx} is the appropriate value on your machine. |
d6da68b9 | 2847 | @end deftypefn |
feca2ed3 | 2848 | |
a2c4f8e0 | 2849 | @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr}) |
feca2ed3 | 2850 | A C expression whose value is RTL representing the value of the return |
861bb6c1 JL |
2851 | address for the frame @var{count} steps up from the current frame, after |
2852 | the prologue. @var{frameaddr} is the frame pointer of the @var{count} | |
2853 | frame, or the frame pointer of the @var{count} @minus{} 1 frame if | |
feca2ed3 JW |
2854 | @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined. |
2855 | ||
e9a25f70 JL |
2856 | The value of the expression must always be the correct address when |
2857 | @var{count} is zero, but may be @code{NULL_RTX} if there is not way to | |
2858 | determine the return address of other frames. | |
a2c4f8e0 | 2859 | @end defmac |
e9a25f70 | 2860 | |
a2c4f8e0 | 2861 | @defmac RETURN_ADDR_IN_PREVIOUS_FRAME |
feca2ed3 JW |
2862 | Define this if the return address of a particular stack frame is accessed |
2863 | from the frame pointer of the previous stack frame. | |
a2c4f8e0 | 2864 | @end defmac |
861bb6c1 | 2865 | |
a2c4f8e0 | 2866 | @defmac INCOMING_RETURN_ADDR_RTX |
861bb6c1 JL |
2867 | A C expression whose value is RTL representing the location of the |
2868 | incoming return address at the beginning of any function, before the | |
2869 | prologue. This RTL is either a @code{REG}, indicating that the return | |
2870 | value is saved in @samp{REG}, or a @code{MEM} representing a location in | |
2871 | the stack. | |
2872 | ||
2873 | You only need to define this macro if you want to support call frame | |
2874 | debugging information like that provided by DWARF 2. | |
2875 | ||
2c849145 | 2876 | If this RTL is a @code{REG}, you should also define |
aee96fe9 | 2877 | @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}. |
a2c4f8e0 | 2878 | @end defmac |
2c849145 | 2879 | |
ed80cd68 | 2880 | @defmac DWARF_ALT_FRAME_RETURN_COLUMN |
73774972 | 2881 | A C expression whose value is an integer giving a DWARF 2 column |
ed80cd68 | 2882 | number that may be used as an alternate return column. This should |
73774972 | 2883 | be defined only if @code{DWARF_FRAME_RETURN_COLUMN} is set to a |
ed80cd68 RH |
2884 | general register, but an alternate column needs to be used for |
2885 | signal frames. | |
2886 | @end defmac | |
2887 | ||
282efe1c RH |
2888 | @defmac DWARF_ZERO_REG |
2889 | A C expression whose value is an integer giving a DWARF 2 register | |
2890 | number that is considered to always have the value zero. This should | |
2891 | only be defined if the target has an architected zero register, and | |
2892 | someone decided it was a good idea to use that register number to | |
2893 | terminate the stack backtrace. New ports should avoid this. | |
2894 | @end defmac | |
2895 | ||
e54c7471 EB |
2896 | @deftypefn {Target Hook} void TARGET_DWARF_HANDLE_FRAME_UNSPEC (const char *@var{label}, rtx @var{pattern}, int @var{index}) |
2897 | This target hook allows the backend to emit frame-related insns that | |
2898 | contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging | |
2899 | info engine will invoke it on insns of the form | |
2900 | @smallexample | |
2901 | (set (reg) (unspec [...] UNSPEC_INDEX)) | |
2902 | @end smallexample | |
2903 | and | |
2904 | @smallexample | |
2905 | (set (reg) (unspec_volatile [...] UNSPECV_INDEX)). | |
2906 | @end smallexample | |
2907 | to let the backend emit the call frame instructions. @var{label} is | |
2908 | the CFI label attached to the insn, @var{pattern} is the pattern of | |
2909 | the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}. | |
2910 | @end deftypefn | |
2911 | ||
a2c4f8e0 | 2912 | @defmac INCOMING_FRAME_SP_OFFSET |
861bb6c1 JL |
2913 | A C expression whose value is an integer giving the offset, in bytes, |
2914 | from the value of the stack pointer register to the top of the stack | |
2915 | frame at the beginning of any function, before the prologue. The top of | |
2916 | the frame is defined to be the value of the stack pointer in the | |
2917 | previous frame, just before the call instruction. | |
2918 | ||
71038426 RH |
2919 | You only need to define this macro if you want to support call frame |
2920 | debugging information like that provided by DWARF 2. | |
a2c4f8e0 | 2921 | @end defmac |
71038426 | 2922 | |
a2c4f8e0 | 2923 | @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl}) |
71038426 RH |
2924 | A C expression whose value is an integer giving the offset, in bytes, |
2925 | from the argument pointer to the canonical frame address (cfa). The | |
02f52e19 | 2926 | final value should coincide with that calculated by |
71038426 RH |
2927 | @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable |
2928 | during virtual register instantiation. | |
2929 | ||
2c849145 JM |
2930 | The default value for this macro is @code{FIRST_PARM_OFFSET (fundecl)}, |
2931 | which is correct for most machines; in general, the arguments are found | |
208e52d9 JM |
2932 | immediately before the stack frame. Note that this is not the case on |
2933 | some targets that save registers into the caller's frame, such as SPARC | |
2934 | and rs6000, and so such targets need to define this macro. | |
2c849145 | 2935 | |
208e52d9 | 2936 | You only need to define this macro if the default is incorrect, and you |
2c849145 JM |
2937 | want to support call frame debugging information like that provided by |
2938 | DWARF 2. | |
a2c4f8e0 | 2939 | @end defmac |
512b62fb | 2940 | |
7c16328b RH |
2941 | @node Exception Handling |
2942 | @subsection Exception Handling Support | |
2943 | @cindex exception handling | |
2944 | ||
a2c4f8e0 | 2945 | @defmac EH_RETURN_DATA_REGNO (@var{N}) |
52a11cbf RH |
2946 | A C expression whose value is the @var{N}th register number used for |
2947 | data by exception handlers, or @code{INVALID_REGNUM} if fewer than | |
2948 | @var{N} registers are usable. | |
2949 | ||
2950 | The exception handling library routines communicate with the exception | |
2951 | handlers via a set of agreed upon registers. Ideally these registers | |
2952 | should be call-clobbered; it is possible to use call-saved registers, | |
2953 | but may negatively impact code size. The target must support at least | |
2954 | 2 data registers, but should define 4 if there are enough free registers. | |
2955 | ||
2956 | You must define this macro if you want to support call frame exception | |
2957 | handling like that provided by DWARF 2. | |
a2c4f8e0 | 2958 | @end defmac |
52a11cbf | 2959 | |
a2c4f8e0 | 2960 | @defmac EH_RETURN_STACKADJ_RTX |
52a11cbf RH |
2961 | A C expression whose value is RTL representing a location in which |
2962 | to store a stack adjustment to be applied before function return. | |
2963 | This is used to unwind the stack to an exception handler's call frame. | |
2964 | It will be assigned zero on code paths that return normally. | |
2965 | ||
02f52e19 | 2966 | Typically this is a call-clobbered hard register that is otherwise |
52a11cbf RH |
2967 | untouched by the epilogue, but could also be a stack slot. |
2968 | ||
34dc173c | 2969 | Do not define this macro if the stack pointer is saved and restored |
73774972 EC |
2970 | by the regular prolog and epilog code in the call frame itself; in |
2971 | this case, the exception handling library routines will update the | |
2972 | stack location to be restored in place. Otherwise, you must define | |
2973 | this macro if you want to support call frame exception handling like | |
34dc173c | 2974 | that provided by DWARF 2. |
a2c4f8e0 | 2975 | @end defmac |
52a11cbf | 2976 | |
a2c4f8e0 | 2977 | @defmac EH_RETURN_HANDLER_RTX |
52a11cbf | 2978 | A C expression whose value is RTL representing a location in which |
02f52e19 | 2979 | to store the address of an exception handler to which we should |
52a11cbf RH |
2980 | return. It will not be assigned on code paths that return normally. |
2981 | ||
2982 | Typically this is the location in the call frame at which the normal | |
02f52e19 AJ |
2983 | return address is stored. For targets that return by popping an |
2984 | address off the stack, this might be a memory address just below | |
52a11cbf | 2985 | the @emph{target} call frame rather than inside the current call |
73774972 EC |
2986 | frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already |
2987 | been assigned, so it may be used to calculate the location of the | |
34dc173c | 2988 | target call frame. |
52a11cbf RH |
2989 | |
2990 | Some targets have more complex requirements than storing to an | |
2991 | address calculable during initial code generation. In that case | |
2992 | the @code{eh_return} instruction pattern should be used instead. | |
2993 | ||
2994 | If you want to support call frame exception handling, you must | |
2995 | define either this macro or the @code{eh_return} instruction pattern. | |
a2c4f8e0 | 2996 | @end defmac |
52a11cbf | 2997 | |
1e60c057 R |
2998 | @defmac RETURN_ADDR_OFFSET |
2999 | If defined, an integer-valued C expression for which rtl will be generated | |
3000 | to add it to the exception handler address before it is searched in the | |
3001 | exception handling tables, and to subtract it again from the address before | |
3002 | using it to return to the exception handler. | |
3003 | @end defmac | |
3004 | ||
a2c4f8e0 | 3005 | @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global}) |
2a1ee410 RH |
3006 | This macro chooses the encoding of pointers embedded in the exception |
3007 | handling sections. If at all possible, this should be defined such | |
3008 | that the exception handling section will not require dynamic relocations, | |
3009 | and so may be read-only. | |
3010 | ||
aee96fe9 JM |
3011 | @var{code} is 0 for data, 1 for code labels, 2 for function pointers. |
3012 | @var{global} is true if the symbol may be affected by dynamic relocations. | |
2a1ee410 RH |
3013 | The macro should return a combination of the @code{DW_EH_PE_*} defines |
3014 | as found in @file{dwarf2.h}. | |
3015 | ||
ebb48a4d | 3016 | If this macro is not defined, pointers will not be encoded but |
2a1ee410 | 3017 | represented directly. |
a2c4f8e0 | 3018 | @end defmac |
2a1ee410 | 3019 | |
a2c4f8e0 | 3020 | @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done}) |
2a1ee410 RH |
3021 | This macro allows the target to emit whatever special magic is required |
3022 | to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}. | |
3023 | Generic code takes care of pc-relative and indirect encodings; this must | |
3024 | be defined if the target uses text-relative or data-relative encodings. | |
3025 | ||
aee96fe9 JM |
3026 | This is a C statement that branches to @var{done} if the format was |
3027 | handled. @var{encoding} is the format chosen, @var{size} is the number | |
3028 | of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF} | |
2a1ee410 | 3029 | to be emitted. |
a2c4f8e0 | 3030 | @end defmac |
2a1ee410 | 3031 | |
8662eb14 AM |
3032 | @defmac MD_UNWIND_SUPPORT |
3033 | A string specifying a file to be #include'd in unwind-dw2.c. The file | |
3034 | so included typically defines @code{MD_FALLBACK_FRAME_STATE_FOR}. | |
3035 | @end defmac | |
3036 | ||
3037 | @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs}) | |
7c16328b RH |
3038 | This macro allows the target to add cpu and operating system specific |
3039 | code to the call-frame unwinder for use when there is no unwind data | |
3040 | available. The most common reason to implement this macro is to unwind | |
3041 | through signal frames. | |
3042 | ||
3043 | This macro is called from @code{uw_frame_state_for} in @file{unwind-dw2.c} | |
3044 | and @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context}; | |
3045 | @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra} | |
3046 | for the address of the code being executed and @code{context->cfa} for | |
3047 | the stack pointer value. If the frame can be decoded, the register save | |
8662eb14 AM |
3048 | addresses should be updated in @var{fs} and the macro should evaluate to |
3049 | @code{_URC_NO_REASON}. If the frame cannot be decoded, the macro should | |
3050 | evaluate to @code{_URC_END_OF_STACK}. | |
8207b189 FS |
3051 | |
3052 | For proper signal handling in Java this macro is accompanied by | |
3053 | @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers. | |
a2c4f8e0 | 3054 | @end defmac |
861bb6c1 | 3055 | |
3950dcdf JJ |
3056 | @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs}) |
3057 | This macro allows the target to add operating system specific code to the | |
3058 | call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive, | |
3059 | usually used for signal or interrupt frames. | |
3060 | ||
3061 | This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}. | |
3062 | @var{context} is an @code{_Unwind_Context}; | |
3063 | @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi} | |
3064 | for the abi and context in the @code{.unwabi} directive. If the | |
3065 | @code{.unwabi} directive can be handled, the register save addresses should | |
3066 | be updated in @var{fs}. | |
3067 | @end defmac | |
3068 | ||
4746cf84 MA |
3069 | @defmac TARGET_USES_WEAK_UNWIND_INFO |
3070 | A C expression that evaluates to true if the target requires unwind | |
3071 | info to be given comdat linkage. Define it to be @code{1} if comdat | |
3072 | linkage is necessary. The default is @code{0}. | |
3073 | @end defmac | |
3074 | ||
861bb6c1 JL |
3075 | @node Stack Checking |
3076 | @subsection Specifying How Stack Checking is Done | |
3077 | ||
a3a15b4d | 3078 | GCC will check that stack references are within the boundaries of |
630d3d5a | 3079 | the stack, if the @option{-fstack-check} is specified, in one of three ways: |
861bb6c1 JL |
3080 | |
3081 | @enumerate | |
3082 | @item | |
a3a15b4d | 3083 | If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC |
861bb6c1 JL |
3084 | will assume that you have arranged for stack checking to be done at |
3085 | appropriate places in the configuration files, e.g., in | |
08c148a8 NB |
3086 | @code{TARGET_ASM_FUNCTION_PROLOGUE}. GCC will do not other special |
3087 | processing. | |
861bb6c1 JL |
3088 | |
3089 | @item | |
3090 | If @code{STACK_CHECK_BUILTIN} is zero and you defined a named pattern | |
a3a15b4d | 3091 | called @code{check_stack} in your @file{md} file, GCC will call that |
861bb6c1 JL |
3092 | pattern with one argument which is the address to compare the stack |
3093 | value against. You must arrange for this pattern to report an error if | |
3094 | the stack pointer is out of range. | |
3095 | ||
3096 | @item | |
a3a15b4d | 3097 | If neither of the above are true, GCC will generate code to periodically |
861bb6c1 JL |
3098 | ``probe'' the stack pointer using the values of the macros defined below. |
3099 | @end enumerate | |
3100 | ||
a3a15b4d | 3101 | Normally, you will use the default values of these macros, so GCC |
861bb6c1 JL |
3102 | will use the third approach. |
3103 | ||
a2c4f8e0 | 3104 | @defmac STACK_CHECK_BUILTIN |
861bb6c1 | 3105 | A nonzero value if stack checking is done by the configuration files in a |
02f52e19 AJ |
3106 | machine-dependent manner. You should define this macro if stack checking |
3107 | is require by the ABI of your machine or if you would like to have to stack | |
a3a15b4d | 3108 | checking in some more efficient way than GCC's portable approach. |
861bb6c1 | 3109 | The default value of this macro is zero. |
a2c4f8e0 | 3110 | @end defmac |
861bb6c1 | 3111 | |
a2c4f8e0 | 3112 | @defmac STACK_CHECK_PROBE_INTERVAL |
a3a15b4d | 3113 | An integer representing the interval at which GCC must generate stack |
861bb6c1 JL |
3114 | probe instructions. You will normally define this macro to be no larger |
3115 | than the size of the ``guard pages'' at the end of a stack area. The | |
3116 | default value of 4096 is suitable for most systems. | |
a2c4f8e0 | 3117 | @end defmac |
861bb6c1 | 3118 | |
a2c4f8e0 | 3119 | @defmac STACK_CHECK_PROBE_LOAD |
02f52e19 | 3120 | A integer which is nonzero if GCC should perform the stack probe |
a3a15b4d | 3121 | as a load instruction and zero if GCC should use a store instruction. |
861bb6c1 | 3122 | The default is zero, which is the most efficient choice on most systems. |
a2c4f8e0 | 3123 | @end defmac |
861bb6c1 | 3124 | |
a2c4f8e0 | 3125 | @defmac STACK_CHECK_PROTECT |
861bb6c1 JL |
3126 | The number of bytes of stack needed to recover from a stack overflow, |
3127 | for languages where such a recovery is supported. The default value of | |
3128 | 75 words should be adequate for most machines. | |
a2c4f8e0 | 3129 | @end defmac |
861bb6c1 | 3130 | |
a2c4f8e0 | 3131 | @defmac STACK_CHECK_MAX_FRAME_SIZE |
a3a15b4d | 3132 | The maximum size of a stack frame, in bytes. GCC will generate probe |
861bb6c1 JL |
3133 | instructions in non-leaf functions to ensure at least this many bytes of |
3134 | stack are available. If a stack frame is larger than this size, stack | |
a3a15b4d JL |
3135 | checking will not be reliable and GCC will issue a warning. The |
3136 | default is chosen so that GCC only generates one instruction on most | |
861bb6c1 | 3137 | systems. You should normally not change the default value of this macro. |
a2c4f8e0 | 3138 | @end defmac |
861bb6c1 | 3139 | |
a2c4f8e0 | 3140 | @defmac STACK_CHECK_FIXED_FRAME_SIZE |
a3a15b4d | 3141 | GCC uses this value to generate the above warning message. It |
861bb6c1 JL |
3142 | represents the amount of fixed frame used by a function, not including |
3143 | space for any callee-saved registers, temporaries and user variables. | |
3144 | You need only specify an upper bound for this amount and will normally | |
3145 | use the default of four words. | |
a2c4f8e0 | 3146 | @end defmac |
861bb6c1 | 3147 | |
a2c4f8e0 | 3148 | @defmac STACK_CHECK_MAX_VAR_SIZE |
a3a15b4d | 3149 | The maximum size, in bytes, of an object that GCC will place in the |
861bb6c1 | 3150 | fixed area of the stack frame when the user specifies |
630d3d5a | 3151 | @option{-fstack-check}. |
a3a15b4d | 3152 | GCC computed the default from the values of the above macros and you will |
861bb6c1 | 3153 | normally not need to override that default. |
a2c4f8e0 | 3154 | @end defmac |
feca2ed3 JW |
3155 | |
3156 | @need 2000 | |
3157 | @node Frame Registers | |
3158 | @subsection Registers That Address the Stack Frame | |
3159 | ||
3160 | @c prevent bad page break with this line | |
3161 | This discusses registers that address the stack frame. | |
3162 | ||
a2c4f8e0 | 3163 | @defmac STACK_POINTER_REGNUM |
feca2ed3 JW |
3164 | The register number of the stack pointer register, which must also be a |
3165 | fixed register according to @code{FIXED_REGISTERS}. On most machines, | |
3166 | the hardware determines which register this is. | |
a2c4f8e0 | 3167 | @end defmac |
feca2ed3 | 3168 | |
a2c4f8e0 | 3169 | @defmac FRAME_POINTER_REGNUM |
feca2ed3 JW |
3170 | The register number of the frame pointer register, which is used to |
3171 | access automatic variables in the stack frame. On some machines, the | |
3172 | hardware determines which register this is. On other machines, you can | |
3173 | choose any register you wish for this purpose. | |
a2c4f8e0 | 3174 | @end defmac |
feca2ed3 | 3175 | |
a2c4f8e0 | 3176 | @defmac HARD_FRAME_POINTER_REGNUM |
feca2ed3 JW |
3177 | On some machines the offset between the frame pointer and starting |
3178 | offset of the automatic variables is not known until after register | |
3179 | allocation has been done (for example, because the saved registers are | |
3180 | between these two locations). On those machines, define | |
3181 | @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to | |
3182 | be used internally until the offset is known, and define | |
556e0f21 | 3183 | @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number |
feca2ed3 JW |
3184 | used for the frame pointer. |
3185 | ||
3186 | You should define this macro only in the very rare circumstances when it | |
3187 | is not possible to calculate the offset between the frame pointer and | |
3188 | the automatic variables until after register allocation has been | |
3189 | completed. When this macro is defined, you must also indicate in your | |
3190 | definition of @code{ELIMINABLE_REGS} how to eliminate | |
3191 | @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM} | |
3192 | or @code{STACK_POINTER_REGNUM}. | |
3193 | ||
3194 | Do not define this macro if it would be the same as | |
3195 | @code{FRAME_POINTER_REGNUM}. | |
a2c4f8e0 | 3196 | @end defmac |
feca2ed3 | 3197 | |
a2c4f8e0 | 3198 | @defmac ARG_POINTER_REGNUM |
feca2ed3 JW |
3199 | The register number of the arg pointer register, which is used to access |
3200 | the function's argument list. On some machines, this is the same as the | |
3201 | frame pointer register. On some machines, the hardware determines which | |
3202 | register this is. On other machines, you can choose any register you | |
3203 | wish for this purpose. If this is not the same register as the frame | |
3204 | pointer register, then you must mark it as a fixed register according to | |
3205 | @code{FIXED_REGISTERS}, or arrange to be able to eliminate it | |
3206 | (@pxref{Elimination}). | |
a2c4f8e0 | 3207 | @end defmac |
feca2ed3 | 3208 | |
a2c4f8e0 | 3209 | @defmac RETURN_ADDRESS_POINTER_REGNUM |
feca2ed3 JW |
3210 | The register number of the return address pointer register, which is used to |
3211 | access the current function's return address from the stack. On some | |
3212 | machines, the return address is not at a fixed offset from the frame | |
3213 | pointer or stack pointer or argument pointer. This register can be defined | |
3214 | to point to the return address on the stack, and then be converted by | |
3215 | @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer. | |
3216 | ||
3217 | Do not define this macro unless there is no other way to get the return | |
3218 | address from the stack. | |
a2c4f8e0 | 3219 | @end defmac |
feca2ed3 | 3220 | |
a2c4f8e0 ZW |
3221 | @defmac STATIC_CHAIN_REGNUM |
3222 | @defmacx STATIC_CHAIN_INCOMING_REGNUM | |
feca2ed3 JW |
3223 | Register numbers used for passing a function's static chain pointer. If |
3224 | register windows are used, the register number as seen by the called | |
3225 | function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register | |
3226 | number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If | |
3227 | these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need | |
bd819a4a | 3228 | not be defined. |
feca2ed3 JW |
3229 | |
3230 | The static chain register need not be a fixed register. | |
3231 | ||
3232 | If the static chain is passed in memory, these macros should not be | |
3233 | defined; instead, the next two macros should be defined. | |
a2c4f8e0 | 3234 | @end defmac |
feca2ed3 | 3235 | |
a2c4f8e0 ZW |
3236 | @defmac STATIC_CHAIN |
3237 | @defmacx STATIC_CHAIN_INCOMING | |
feca2ed3 JW |
3238 | If the static chain is passed in memory, these macros provide rtx giving |
3239 | @code{mem} expressions that denote where they are stored. | |
3240 | @code{STATIC_CHAIN} and @code{STATIC_CHAIN_INCOMING} give the locations | |
3241 | as seen by the calling and called functions, respectively. Often the former | |
3242 | will be at an offset from the stack pointer and the latter at an offset from | |
bd819a4a | 3243 | the frame pointer. |
feca2ed3 JW |
3244 | |
3245 | @findex stack_pointer_rtx | |
3246 | @findex frame_pointer_rtx | |
3247 | @findex arg_pointer_rtx | |
3248 | The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and | |
3249 | @code{arg_pointer_rtx} will have been initialized prior to the use of these | |
3250 | macros and should be used to refer to those items. | |
3251 | ||
3252 | If the static chain is passed in a register, the two previous macros should | |
3253 | be defined instead. | |
a2c4f8e0 | 3254 | @end defmac |
919543ab | 3255 | |
a2c4f8e0 | 3256 | @defmac DWARF_FRAME_REGISTERS |
919543ab AH |
3257 | This macro specifies the maximum number of hard registers that can be |
3258 | saved in a call frame. This is used to size data structures used in | |
3259 | DWARF2 exception handling. | |
3260 | ||
3261 | Prior to GCC 3.0, this macro was needed in order to establish a stable | |
3262 | exception handling ABI in the face of adding new hard registers for ISA | |
3263 | extensions. In GCC 3.0 and later, the EH ABI is insulated from changes | |
3264 | in the number of hard registers. Nevertheless, this macro can still be | |
3265 | used to reduce the runtime memory requirements of the exception handling | |
3266 | routines, which can be substantial if the ISA contains a lot of | |
3267 | registers that are not call-saved. | |
3268 | ||
3269 | If this macro is not defined, it defaults to | |
3270 | @code{FIRST_PSEUDO_REGISTER}. | |
a2c4f8e0 | 3271 | @end defmac |
919543ab | 3272 | |
a2c4f8e0 | 3273 | @defmac PRE_GCC3_DWARF_FRAME_REGISTERS |
919543ab AH |
3274 | |
3275 | This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided | |
3276 | for backward compatibility in pre GCC 3.0 compiled code. | |
3277 | ||
3278 | If this macro is not defined, it defaults to | |
3279 | @code{DWARF_FRAME_REGISTERS}. | |
a2c4f8e0 | 3280 | @end defmac |
919543ab | 3281 | |
a2c4f8e0 | 3282 | @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno}) |
41f3a930 AH |
3283 | |
3284 | Define this macro if the target's representation for dwarf registers | |
3285 | is different than the internal representation for unwind column. | |
61aeb06f | 3286 | Given a dwarf register, this macro should return the internal unwind |
41f3a930 AH |
3287 | column number to use instead. |
3288 | ||
73774972 | 3289 | See the PowerPC's SPE target for an example. |
a2c4f8e0 | 3290 | @end defmac |
feca2ed3 | 3291 | |
34c80057 AM |
3292 | @defmac DWARF_FRAME_REGNUM (@var{regno}) |
3293 | ||
3294 | Define this macro if the target's representation for dwarf registers | |
3295 | used in .eh_frame or .debug_frame is different from that used in other | |
2dd76960 | 3296 | debug info sections. Given a GCC hard register number, this macro |
34c80057 AM |
3297 | should return the .eh_frame register number. The default is |
3298 | @code{DBX_REGISTER_NUMBER (@var{regno})}. | |
3299 | ||
3300 | @end defmac | |
3301 | ||
3302 | @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh}) | |
3303 | ||
3304 | Define this macro to map register numbers held in the call frame info | |
2dd76960 | 3305 | that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that |
34c80057 | 3306 | should be output in .debug_frame (@code{@var{for_eh}} is zero) and |
f676971a | 3307 | .eh_frame (@code{@var{for_eh}} is nonzero). The default is to |
34c80057 AM |
3308 | return @code{@var{regno}}. |
3309 | ||
3310 | @end defmac | |
3311 | ||
feca2ed3 JW |
3312 | @node Elimination |
3313 | @subsection Eliminating Frame Pointer and Arg Pointer | |
3314 | ||
3315 | @c prevent bad page break with this line | |
3316 | This is about eliminating the frame pointer and arg pointer. | |
3317 | ||
a2c4f8e0 | 3318 | @defmac FRAME_POINTER_REQUIRED |
feca2ed3 JW |
3319 | A C expression which is nonzero if a function must have and use a frame |
3320 | pointer. This expression is evaluated in the reload pass. If its value is | |
3321 | nonzero the function will have a frame pointer. | |
3322 | ||
3323 | The expression can in principle examine the current function and decide | |
3324 | according to the facts, but on most machines the constant 0 or the | |
3325 | constant 1 suffices. Use 0 when the machine allows code to be generated | |
3326 | with no frame pointer, and doing so saves some time or space. Use 1 | |
3327 | when there is no possible advantage to avoiding a frame pointer. | |
3328 | ||
3329 | In certain cases, the compiler does not know how to produce valid code | |
3330 | without a frame pointer. The compiler recognizes those cases and | |
3331 | automatically gives the function a frame pointer regardless of what | |
3332 | @code{FRAME_POINTER_REQUIRED} says. You don't need to worry about | |
bd819a4a | 3333 | them. |
feca2ed3 JW |
3334 | |
3335 | In a function that does not require a frame pointer, the frame pointer | |
3336 | register can be allocated for ordinary usage, unless you mark it as a | |
3337 | fixed register. See @code{FIXED_REGISTERS} for more information. | |
a2c4f8e0 | 3338 | @end defmac |
feca2ed3 | 3339 | |
feca2ed3 | 3340 | @findex get_frame_size |
a2c4f8e0 | 3341 | @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var}) |
feca2ed3 JW |
3342 | A C statement to store in the variable @var{depth-var} the difference |
3343 | between the frame pointer and the stack pointer values immediately after | |
3344 | the function prologue. The value would be computed from information | |
3345 | such as the result of @code{get_frame_size ()} and the tables of | |
3346 | registers @code{regs_ever_live} and @code{call_used_regs}. | |
3347 | ||
3348 | If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and | |
3349 | need not be defined. Otherwise, it must be defined even if | |
3350 | @code{FRAME_POINTER_REQUIRED} is defined to always be true; in that | |
3351 | case, you may set @var{depth-var} to anything. | |
a2c4f8e0 | 3352 | @end defmac |
feca2ed3 | 3353 | |
a2c4f8e0 | 3354 | @defmac ELIMINABLE_REGS |
feca2ed3 JW |
3355 | If defined, this macro specifies a table of register pairs used to |
3356 | eliminate unneeded registers that point into the stack frame. If it is not | |
3357 | defined, the only elimination attempted by the compiler is to replace | |
3358 | references to the frame pointer with references to the stack pointer. | |
3359 | ||
3360 | The definition of this macro is a list of structure initializations, each | |
3361 | of which specifies an original and replacement register. | |
3362 | ||
3363 | On some machines, the position of the argument pointer is not known until | |
3364 | the compilation is completed. In such a case, a separate hard register | |
3365 | must be used for the argument pointer. This register can be eliminated by | |
3366 | replacing it with either the frame pointer or the argument pointer, | |
3367 | depending on whether or not the frame pointer has been eliminated. | |
3368 | ||
3369 | In this case, you might specify: | |
3ab51846 | 3370 | @smallexample |
feca2ed3 JW |
3371 | #define ELIMINABLE_REGS \ |
3372 | @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \ | |
3373 | @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \ | |
3374 | @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@} | |
3ab51846 | 3375 | @end smallexample |
feca2ed3 JW |
3376 | |
3377 | Note that the elimination of the argument pointer with the stack pointer is | |
3378 | specified first since that is the preferred elimination. | |
a2c4f8e0 | 3379 | @end defmac |
feca2ed3 | 3380 | |
a2c4f8e0 | 3381 | @defmac CAN_ELIMINATE (@var{from-reg}, @var{to-reg}) |
df2a54e9 | 3382 | A C expression that returns nonzero if the compiler is allowed to try |
feca2ed3 JW |
3383 | to replace register number @var{from-reg} with register number |
3384 | @var{to-reg}. This macro need only be defined if @code{ELIMINABLE_REGS} | |
3385 | is defined, and will usually be the constant 1, since most of the cases | |
3386 | preventing register elimination are things that the compiler already | |
3387 | knows about. | |
a2c4f8e0 | 3388 | @end defmac |
feca2ed3 | 3389 | |
a2c4f8e0 | 3390 | @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var}) |
feca2ed3 JW |
3391 | This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It |
3392 | specifies the initial difference between the specified pair of | |
3393 | registers. This macro must be defined if @code{ELIMINABLE_REGS} is | |
3394 | defined. | |
a2c4f8e0 | 3395 | @end defmac |
feca2ed3 JW |
3396 | |
3397 | @node Stack Arguments | |
3398 | @subsection Passing Function Arguments on the Stack | |
3399 | @cindex arguments on stack | |
3400 | @cindex stack arguments | |
3401 | ||
3402 | The macros in this section control how arguments are passed | |
3403 | on the stack. See the following section for other macros that | |
3404 | control passing certain arguments in registers. | |
3405 | ||
61f71b34 DD |
3406 | @deftypefn {Target Hook} bool TARGET_PROMOTE_PROTOTYPES (tree @var{fntype}) |
3407 | This target hook returns @code{true} if an argument declared in a | |
3408 | prototype as an integral type smaller than @code{int} should actually be | |
3409 | passed as an @code{int}. In addition to avoiding errors in certain | |
3410 | cases of mismatch, it also makes for better code on certain machines. | |
3411 | The default is to not promote prototypes. | |
3412 | @end deftypefn | |
feca2ed3 | 3413 | |
a2c4f8e0 | 3414 | @defmac PUSH_ARGS |
767094dd | 3415 | A C expression. If nonzero, push insns will be used to pass |
f73ad30e JH |
3416 | outgoing arguments. |
3417 | If the target machine does not have a push instruction, set it to zero. | |
3418 | That directs GCC to use an alternate strategy: to | |
3419 | allocate the entire argument block and then store the arguments into | |
aee96fe9 | 3420 | it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too. |
a2c4f8e0 | 3421 | @end defmac |
f73ad30e | 3422 | |
9d6bef95 JM |
3423 | @defmac PUSH_ARGS_REVERSED |
3424 | A C expression. If nonzero, function arguments will be evaluated from | |
3425 | last to first, rather than from first to last. If this macro is not | |
3426 | defined, it defaults to @code{PUSH_ARGS} on targets where the stack | |
3427 | and args grow in opposite directions, and 0 otherwise. | |
3428 | @end defmac | |
3429 | ||
a2c4f8e0 | 3430 | @defmac PUSH_ROUNDING (@var{npushed}) |
feca2ed3 JW |
3431 | A C expression that is the number of bytes actually pushed onto the |
3432 | stack when an instruction attempts to push @var{npushed} bytes. | |
feca2ed3 JW |
3433 | |
3434 | On some machines, the definition | |
3435 | ||
3ab51846 | 3436 | @smallexample |
feca2ed3 | 3437 | #define PUSH_ROUNDING(BYTES) (BYTES) |
3ab51846 | 3438 | @end smallexample |
feca2ed3 JW |
3439 | |
3440 | @noindent | |
3441 | will suffice. But on other machines, instructions that appear | |
3442 | to push one byte actually push two bytes in an attempt to maintain | |
3443 | alignment. Then the definition should be | |
3444 | ||
3ab51846 | 3445 | @smallexample |
feca2ed3 | 3446 | #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1) |
3ab51846 | 3447 | @end smallexample |
a2c4f8e0 | 3448 | @end defmac |
feca2ed3 | 3449 | |
feca2ed3 | 3450 | @findex current_function_outgoing_args_size |
a2c4f8e0 | 3451 | @defmac ACCUMULATE_OUTGOING_ARGS |
767094dd | 3452 | A C expression. If nonzero, the maximum amount of space required for outgoing arguments |
feca2ed3 JW |
3453 | will be computed and placed into the variable |
3454 | @code{current_function_outgoing_args_size}. No space will be pushed | |
3455 | onto the stack for each call; instead, the function prologue should | |
3456 | increase the stack frame size by this amount. | |
3457 | ||
f73ad30e | 3458 | Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS} |
feca2ed3 | 3459 | is not proper. |
a2c4f8e0 | 3460 | @end defmac |
feca2ed3 | 3461 | |
a2c4f8e0 | 3462 | @defmac REG_PARM_STACK_SPACE (@var{fndecl}) |
feca2ed3 JW |
3463 | Define this macro if functions should assume that stack space has been |
3464 | allocated for arguments even when their values are passed in | |
3465 | registers. | |
3466 | ||
3467 | The value of this macro is the size, in bytes, of the area reserved for | |
ab87f8c8 | 3468 | arguments passed in registers for the function represented by @var{fndecl}, |
a3a15b4d | 3469 | which can be zero if GCC is calling a library function. |
feca2ed3 JW |
3470 | |
3471 | This space can be allocated by the caller, or be a part of the | |
3472 | machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says | |
3473 | which. | |
a2c4f8e0 | 3474 | @end defmac |
feca2ed3 JW |
3475 | @c above is overfull. not sure what to do. --mew 5feb93 did |
3476 | @c something, not sure if it looks good. --mew 10feb93 | |
3477 | ||
a2c4f8e0 | 3478 | @defmac OUTGOING_REG_PARM_STACK_SPACE |
feca2ed3 JW |
3479 | Define this if it is the responsibility of the caller to allocate the area |
3480 | reserved for arguments passed in registers. | |
3481 | ||
3482 | If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls | |
3483 | whether the space for these arguments counts in the value of | |
3484 | @code{current_function_outgoing_args_size}. | |
a2c4f8e0 | 3485 | @end defmac |
feca2ed3 | 3486 | |
a2c4f8e0 | 3487 | @defmac STACK_PARMS_IN_REG_PARM_AREA |
feca2ed3 JW |
3488 | Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the |
3489 | stack parameters don't skip the area specified by it. | |
3490 | @c i changed this, makes more sens and it should have taken care of the | |
3491 | @c overfull.. not as specific, tho. --mew 5feb93 | |
3492 | ||
3493 | Normally, when a parameter is not passed in registers, it is placed on the | |
3494 | stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro | |
3495 | suppresses this behavior and causes the parameter to be passed on the | |
3496 | stack in its natural location. | |
a2c4f8e0 | 3497 | @end defmac |
feca2ed3 | 3498 | |
a2c4f8e0 | 3499 | @defmac RETURN_POPS_ARGS (@var{fundecl}, @var{funtype}, @var{stack-size}) |
feca2ed3 JW |
3500 | A C expression that should indicate the number of bytes of its own |
3501 | arguments that a function pops on returning, or 0 if the | |
3502 | function pops no arguments and the caller must therefore pop them all | |
3503 | after the function returns. | |
3504 | ||
3505 | @var{fundecl} is a C variable whose value is a tree node that describes | |
3506 | the function in question. Normally it is a node of type | |
3507 | @code{FUNCTION_DECL} that describes the declaration of the function. | |
91d231cb | 3508 | From this you can obtain the @code{DECL_ATTRIBUTES} of the function. |
feca2ed3 JW |
3509 | |
3510 | @var{funtype} is a C variable whose value is a tree node that | |
3511 | describes the function in question. Normally it is a node of type | |
3512 | @code{FUNCTION_TYPE} that describes the data type of the function. | |
3513 | From this it is possible to obtain the data types of the value and | |
3514 | arguments (if known). | |
3515 | ||
861bb6c1 | 3516 | When a call to a library function is being considered, @var{fundecl} |
feca2ed3 JW |
3517 | will contain an identifier node for the library function. Thus, if |
3518 | you need to distinguish among various library functions, you can do so | |
3519 | by their names. Note that ``library function'' in this context means | |
3520 | a function used to perform arithmetic, whose name is known specially | |
3521 | in the compiler and was not mentioned in the C code being compiled. | |
3522 | ||
3523 | @var{stack-size} is the number of bytes of arguments passed on the | |
3524 | stack. If a variable number of bytes is passed, it is zero, and | |
3525 | argument popping will always be the responsibility of the calling function. | |
3526 | ||
8aeea6e6 | 3527 | On the VAX, all functions always pop their arguments, so the definition |
feca2ed3 JW |
3528 | of this macro is @var{stack-size}. On the 68000, using the standard |
3529 | calling convention, no functions pop their arguments, so the value of | |
3530 | the macro is always 0 in this case. But an alternative calling | |
3531 | convention is available in which functions that take a fixed number of | |
3532 | arguments pop them but other functions (such as @code{printf}) pop | |
3533 | nothing (the caller pops all). When this convention is in use, | |
3534 | @var{funtype} is examined to determine whether a function takes a fixed | |
3535 | number of arguments. | |
a2c4f8e0 | 3536 | @end defmac |
fa5322fa | 3537 | |
a2c4f8e0 | 3538 | @defmac CALL_POPS_ARGS (@var{cum}) |
fa5322fa AO |
3539 | A C expression that should indicate the number of bytes a call sequence |
3540 | pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS} | |
3541 | when compiling a function call. | |
3542 | ||
3543 | @var{cum} is the variable in which all arguments to the called function | |
3544 | have been accumulated. | |
3545 | ||
3546 | On certain architectures, such as the SH5, a call trampoline is used | |
3547 | that pops certain registers off the stack, depending on the arguments | |
3548 | that have been passed to the function. Since this is a property of the | |
3549 | call site, not of the called function, @code{RETURN_POPS_ARGS} is not | |
3550 | appropriate. | |
a2c4f8e0 | 3551 | @end defmac |
feca2ed3 JW |
3552 | |
3553 | @node Register Arguments | |
3554 | @subsection Passing Arguments in Registers | |
3555 | @cindex arguments in registers | |
3556 | @cindex registers arguments | |
3557 | ||
3558 | This section describes the macros which let you control how various | |
3559 | types of arguments are passed in registers or how they are arranged in | |
3560 | the stack. | |
3561 | ||
a2c4f8e0 | 3562 | @defmac FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named}) |
feca2ed3 JW |
3563 | A C expression that controls whether a function argument is passed |
3564 | in a register, and which register. | |
3565 | ||
3566 | The arguments are @var{cum}, which summarizes all the previous | |
3567 | arguments; @var{mode}, the machine mode of the argument; @var{type}, | |
3568 | the data type of the argument as a tree node or 0 if that is not known | |
3569 | (which happens for C support library functions); and @var{named}, | |
3570 | which is 1 for an ordinary argument and 0 for nameless arguments that | |
3571 | correspond to @samp{@dots{}} in the called function's prototype. | |
3719d27b JO |
3572 | @var{type} can be an incomplete type if a syntax error has previously |
3573 | occurred. | |
feca2ed3 JW |
3574 | |
3575 | The value of the expression is usually either a @code{reg} RTX for the | |
3576 | hard register in which to pass the argument, or zero to pass the | |
3577 | argument on the stack. | |
3578 | ||
8aeea6e6 | 3579 | For machines like the VAX and 68000, where normally all arguments are |
feca2ed3 JW |
3580 | pushed, zero suffices as a definition. |
3581 | ||
161d7b59 | 3582 | The value of the expression can also be a @code{parallel} RTX@. This is |
feca2ed3 | 3583 | used when an argument is passed in multiple locations. The mode of the |
ce376beb | 3584 | @code{parallel} should be the mode of the entire argument. The |
feca2ed3 | 3585 | @code{parallel} holds any number of @code{expr_list} pairs; each one |
f797c10b NC |
3586 | describes where part of the argument is passed. In each |
3587 | @code{expr_list} the first operand must be a @code{reg} RTX for the hard | |
3588 | register in which to pass this part of the argument, and the mode of the | |
3589 | register RTX indicates how large this part of the argument is. The | |
3590 | second operand of the @code{expr_list} is a @code{const_int} which gives | |
3591 | the offset in bytes into the entire argument of where this part starts. | |
02f52e19 | 3592 | As a special exception the first @code{expr_list} in the @code{parallel} |
c980b85b NC |
3593 | RTX may have a first operand of zero. This indicates that the entire |
3594 | argument is also stored on the stack. | |
feca2ed3 | 3595 | |
1cc5e432 GK |
3596 | The last time this macro is called, it is called with @code{MODE == |
3597 | VOIDmode}, and its result is passed to the @code{call} or @code{call_value} | |
3598 | pattern as operands 2 and 3 respectively. | |
3599 | ||
feca2ed3 | 3600 | @cindex @file{stdarg.h} and register arguments |
5490d604 | 3601 | The usual way to make the ISO library @file{stdarg.h} work on a machine |
feca2ed3 JW |
3602 | where some arguments are usually passed in registers, is to cause |
3603 | nameless arguments to be passed on the stack instead. This is done | |
3604 | by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0. | |
3605 | ||
fe984136 | 3606 | @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG} |
feca2ed3 | 3607 | @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG} |
fe984136 | 3608 | You may use the hook @code{targetm.calls.must_pass_in_stack} |
feca2ed3 JW |
3609 | in the definition of this macro to determine if this argument is of a |
3610 | type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE} | |
df2a54e9 | 3611 | is not defined and @code{FUNCTION_ARG} returns nonzero for such an |
feca2ed3 JW |
3612 | argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is |
3613 | defined, the argument will be computed in the stack and then loaded into | |
3614 | a register. | |
a2c4f8e0 | 3615 | @end defmac |
feca2ed3 | 3616 | |
fe984136 RH |
3617 | @deftypefn {Target Hook} bool TARGET_MUST_PASS_IN_STACK (enum machine_mode @var{mode}, tree @var{type}) |
3618 | This target hook should return @code{true} if we should not pass @var{type} | |
3619 | solely in registers. The file @file{expr.h} defines a | |
d9a4ee00 JL |
3620 | definition that is usually appropriate, refer to @file{expr.h} for additional |
3621 | documentation. | |
fe984136 | 3622 | @end deftypefn |
d9a4ee00 | 3623 | |
a2c4f8e0 | 3624 | @defmac FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named}) |
feca2ed3 JW |
3625 | Define this macro if the target machine has ``register windows'', so |
3626 | that the register in which a function sees an arguments is not | |
3627 | necessarily the same as the one in which the caller passed the | |
3628 | argument. | |
3629 | ||
3630 | For such machines, @code{FUNCTION_ARG} computes the register in which | |
3631 | the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should | |
3632 | be defined in a similar fashion to tell the function being called | |
3633 | where the arguments will arrive. | |
3634 | ||
3635 | If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG} | |
bd819a4a | 3636 | serves both purposes. |
a2c4f8e0 | 3637 | @end defmac |
feca2ed3 | 3638 | |
78a52f11 RH |
3639 | @deftypefn {Target Hook} int TARGET_ARG_PARTIAL_BYTES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named}) |
3640 | This target hook returns the number of bytes at the beginning of an | |
3641 | argument that must be put in registers. The value must be zero for | |
feca2ed3 JW |
3642 | arguments that are passed entirely in registers or that are entirely |
3643 | pushed on the stack. | |
3644 | ||
3645 | On some machines, certain arguments must be passed partially in | |
3646 | registers and partially in memory. On these machines, typically the | |
78a52f11 | 3647 | first few words of arguments are passed in registers, and the rest |
feca2ed3 JW |
3648 | on the stack. If a multi-word argument (a @code{double} or a |
3649 | structure) crosses that boundary, its first few words must be passed | |
3650 | in registers and the rest must be pushed. This macro tells the | |
78a52f11 | 3651 | compiler when this occurs, and how many bytes should go in registers. |
feca2ed3 JW |
3652 | |
3653 | @code{FUNCTION_ARG} for these arguments should return the first | |
3654 | register to be used by the caller for this argument; likewise | |
3655 | @code{FUNCTION_INCOMING_ARG}, for the called function. | |
78a52f11 | 3656 | @end deftypefn |
feca2ed3 | 3657 | |
8cd5a4e0 | 3658 | @deftypefn {Target Hook} bool TARGET_PASS_BY_REFERENCE (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named}) |
f676971a | 3659 | This target hook should return @code{true} if an argument at the |
8cd5a4e0 | 3660 | position indicated by @var{cum} should be passed by reference. This |
f676971a | 3661 | predicate is queried after target independent reasons for being |
8cd5a4e0 RH |
3662 | passed by reference, such as @code{TREE_ADDRESSABLE (type)}. |
3663 | ||
3664 | If the hook returns true, a copy of that argument is made in memory and a | |
feca2ed3 JW |
3665 | pointer to the argument is passed instead of the argument itself. |
3666 | The pointer is passed in whatever way is appropriate for passing a pointer | |
3667 | to that type. | |
8cd5a4e0 | 3668 | @end deftypefn |
feca2ed3 | 3669 | |
6cdd5672 RH |
3670 | @deftypefn {Target Hook} bool TARGET_CALLEE_COPIES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named}) |
3671 | The function argument described by the parameters to this hook is | |
3672 | known to be passed by reference. The hook should return true if the | |
3673 | function argument should be copied by the callee instead of copied | |
3674 | by the caller. | |
3675 | ||
a1c496cb | 3676 | For any argument for which the hook returns true, if it can be |
6cdd5672 RH |
3677 | determined that the argument is not modified, then a copy need |
3678 | not be generated. | |
3679 | ||
3680 | The default version of this hook always returns false. | |
3681 | @end deftypefn | |
feca2ed3 | 3682 | |
a2c4f8e0 | 3683 | @defmac CUMULATIVE_ARGS |
feca2ed3 JW |
3684 | A C type for declaring a variable that is used as the first argument of |
3685 | @code{FUNCTION_ARG} and other related values. For some target machines, | |
3686 | the type @code{int} suffices and can hold the number of bytes of | |
3687 | argument so far. | |
3688 | ||
3689 | There is no need to record in @code{CUMULATIVE_ARGS} anything about the | |
3690 | arguments that have been passed on the stack. The compiler has other | |
3691 | variables to keep track of that. For target machines on which all | |
3692 | arguments are passed on the stack, there is no need to store anything in | |
3693 | @code{CUMULATIVE_ARGS}; however, the data structure must exist and | |
3694 | should not be empty, so use @code{int}. | |
a2c4f8e0 | 3695 | @end defmac |
feca2ed3 | 3696 | |
0f6937fe | 3697 | @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args}) |
a2c4f8e0 ZW |
3698 | A C statement (sans semicolon) for initializing the variable |
3699 | @var{cum} for the state at the beginning of the argument list. The | |
3700 | variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype} | |
3701 | is the tree node for the data type of the function which will receive | |
3702 | the args, or 0 if the args are to a compiler support library function. | |
3703 | For direct calls that are not libcalls, @var{fndecl} contain the | |
3704 | declaration node of the function. @var{fndecl} is also set when | |
3705 | @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function | |
0f6937fe AM |
3706 | being compiled. @var{n_named_args} is set to the number of named |
3707 | arguments, including a structure return address if it is passed as a | |
3708 | parameter, when making a call. When processing incoming arguments, | |
78466c0e | 3709 | @var{n_named_args} is set to @minus{}1. |
feca2ed3 JW |
3710 | |
3711 | When processing a call to a compiler support library function, | |
3712 | @var{libname} identifies which one. It is a @code{symbol_ref} rtx which | |
3713 | contains the name of the function, as a string. @var{libname} is 0 when | |
3714 | an ordinary C function call is being processed. Thus, each time this | |
3715 | macro is called, either @var{libname} or @var{fntype} is nonzero, but | |
3716 | never both of them at once. | |
a2c4f8e0 | 3717 | @end defmac |
feca2ed3 | 3718 | |
a2c4f8e0 | 3719 | @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname}) |
97fc4caf AO |
3720 | Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls, |
3721 | it gets a @code{MODE} argument instead of @var{fntype}, that would be | |
3722 | @code{NULL}. @var{indirect} would always be zero, too. If this macro | |
3723 | is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname, | |
3724 | 0)} is used instead. | |
a2c4f8e0 | 3725 | @end defmac |
97fc4caf | 3726 | |
a2c4f8e0 | 3727 | @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname}) |
feca2ed3 JW |
3728 | Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of |
3729 | finding the arguments for the function being compiled. If this macro is | |
3730 | undefined, @code{INIT_CUMULATIVE_ARGS} is used instead. | |
3731 | ||
3732 | The value passed for @var{libname} is always 0, since library routines | |
161d7b59 | 3733 | with special calling conventions are never compiled with GCC@. The |
feca2ed3 JW |
3734 | argument @var{libname} exists for symmetry with |
3735 | @code{INIT_CUMULATIVE_ARGS}. | |
3736 | @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe. | |
3737 | @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93 | |
a2c4f8e0 | 3738 | @end defmac |
feca2ed3 | 3739 | |
a2c4f8e0 | 3740 | @defmac FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named}) |
feca2ed3 JW |
3741 | A C statement (sans semicolon) to update the summarizer variable |
3742 | @var{cum} to advance past an argument in the argument list. The | |
3743 | values @var{mode}, @var{type} and @var{named} describe that argument. | |
3744 | Once this is done, the variable @var{cum} is suitable for analyzing | |
bd819a4a | 3745 | the @emph{following} argument with @code{FUNCTION_ARG}, etc. |
feca2ed3 JW |
3746 | |
3747 | This macro need not do anything if the argument in question was passed | |
3748 | on the stack. The compiler knows how to track the amount of stack space | |
3749 | used for arguments without any special help. | |
a2c4f8e0 | 3750 | @end defmac |
feca2ed3 | 3751 | |
a2c4f8e0 | 3752 | @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type}) |
feca2ed3 JW |
3753 | If defined, a C expression which determines whether, and in which direction, |
3754 | to pad out an argument with extra space. The value should be of type | |
3755 | @code{enum direction}: either @code{upward} to pad above the argument, | |
3756 | @code{downward} to pad below, or @code{none} to inhibit padding. | |
3757 | ||
3758 | The @emph{amount} of padding is always just enough to reach the next | |
3759 | multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control | |
3760 | it. | |
3761 | ||
3762 | This macro has a default definition which is right for most systems. | |
3763 | For little-endian machines, the default is to pad upward. For | |
3764 | big-endian machines, the default is to pad downward for an argument of | |
3765 | constant size shorter than an @code{int}, and upward otherwise. | |
a2c4f8e0 | 3766 | @end defmac |
feca2ed3 | 3767 | |
a2c4f8e0 | 3768 | @defmac PAD_VARARGS_DOWN |
02f52e19 AJ |
3769 | If defined, a C expression which determines whether the default |
3770 | implementation of va_arg will attempt to pad down before reading the | |
5e4f6244 CP |
3771 | next argument, if that argument is smaller than its aligned space as |
3772 | controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such | |
3773 | arguments are padded down if @code{BYTES_BIG_ENDIAN} is true. | |
a2c4f8e0 | 3774 | @end defmac |
5e4f6244 | 3775 | |
6e985040 AM |
3776 | @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first}) |
3777 | Specify padding for the last element of a block move between registers and | |
3778 | memory. @var{first} is nonzero if this is the only element. Defining this | |
3779 | macro allows better control of register function parameters on big-endian | |
3780 | machines, without using @code{PARALLEL} rtl. In particular, | |
3781 | @code{MUST_PASS_IN_STACK} need not test padding and mode of types in | |
3782 | registers, as there is no longer a "wrong" part of a register; For example, | |
3783 | a three byte aggregate may be passed in the high part of a register if so | |
3784 | required. | |
3785 | @end defmac | |
3786 | ||
a2c4f8e0 | 3787 | @defmac FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type}) |
feca2ed3 JW |
3788 | If defined, a C expression that gives the alignment boundary, in bits, |
3789 | of an argument with the specified mode and type. If it is not defined, | |
3790 | @code{PARM_BOUNDARY} is used for all arguments. | |
a2c4f8e0 | 3791 | @end defmac |
feca2ed3 | 3792 | |
a2c4f8e0 | 3793 | @defmac FUNCTION_ARG_REGNO_P (@var{regno}) |
feca2ed3 JW |
3794 | A C expression that is nonzero if @var{regno} is the number of a hard |
3795 | register in which function arguments are sometimes passed. This does | |
3796 | @emph{not} include implicit arguments such as the static chain and | |
3797 | the structure-value address. On many machines, no registers can be | |
3798 | used for this purpose since all function arguments are pushed on the | |
3799 | stack. | |
a2c4f8e0 | 3800 | @end defmac |
bb1b857a | 3801 | |
42ba5130 RH |
3802 | @deftypefn {Target Hook} bool TARGET_SPLIT_COMPLEX_ARG (tree @var{type}) |
3803 | This hook should return true if parameter of type @var{type} are passed | |
3804 | as two scalar parameters. By default, GCC will attempt to pack complex | |
3805 | arguments into the target's word size. Some ABIs require complex arguments | |
3806 | to be split and treated as their individual components. For example, on | |
3807 | AIX64, complex floats should be passed in a pair of floating point | |
3808 | registers, even though a complex float would fit in one 64-bit floating | |
3809 | point register. | |
3810 | ||
3811 | The default value of this hook is @code{NULL}, which is treated as always | |
3812 | false. | |
3813 | @end deftypefn | |
ded9bf77 | 3814 | |
d3da4d14 RH |
3815 | @deftypefn {Target Hook} tree TARGET_BUILD_BUILTIN_VA_LIST (void) |
3816 | This hook returns a type node for @code{va_list} for the target. | |
3817 | The default version of the hook returns @code{void*}. | |
3818 | @end deftypefn | |
3819 | ||
23a60a04 JM |
3820 | @deftypefn {Target Hook} tree TARGET_GIMPLIFY_VA_ARG_EXPR (tree @var{valist}, tree @var{type}, tree *@var{pre_p}, tree *@var{post_p}) |
3821 | This hook performs target-specific gimplification of | |
3822 | @code{VA_ARG_EXPR}. The first two parameters correspond to the | |
3823 | arguments to @code{va_arg}; the latter two are as in | |
3824 | @code{gimplify.c:gimplify_expr}. | |
23a60a04 JM |
3825 | @end deftypefn |
3826 | ||
e09ec166 EC |
3827 | @deftypefn {Target Hook} bool TARGET_VALID_POINTER_MODE (enum machine_mode @var{mode}) |
3828 | Define this to return nonzero if the port can handle pointers | |
3829 | with machine mode @var{mode}. The default version of this | |
3830 | hook returns true for both @code{ptr_mode} and @code{Pmode}. | |
3831 | @end deftypefn | |
3832 | ||
6dd53648 RH |
3833 | @deftypefn {Target Hook} bool TARGET_SCALAR_MODE_SUPPORTED_P (enum machine_mode @var{mode}) |
3834 | Define this to return nonzero if the port is prepared to handle | |
3835 | insns involving scalar mode @var{mode}. For a scalar mode to be | |
3836 | considered supported, all the basic arithmetic and comparisons | |
3837 | must work. | |
3838 | ||
3839 | The default version of this hook returns true for any mode | |
3840 | required to handle the basic C types (as defined by the port). | |
3841 | Included here are the double-word arithmetic supported by the | |
3842 | code in @file{optabs.c}. | |
3843 | @end deftypefn | |
3844 | ||
f676971a EC |
3845 | @deftypefn {Target Hook} bool TARGET_VECTOR_MODE_SUPPORTED_P (enum machine_mode @var{mode}) |
3846 | Define this to return nonzero if the port is prepared to handle | |
3847 | insns involving vector mode @var{mode}. At the very least, it | |
3848 | must have move patterns for this mode. | |
3849 | @end deftypefn | |
3850 | ||
feca2ed3 JW |
3851 | @node Scalar Return |
3852 | @subsection How Scalar Function Values Are Returned | |
3853 | @cindex return values in registers | |
3854 | @cindex values, returned by functions | |
3855 | @cindex scalars, returned as values | |
3856 | ||
3857 | This section discusses the macros that control returning scalars as | |
3858 | values---values that can fit in registers. | |
3859 | ||
a2c4f8e0 | 3860 | @defmac FUNCTION_VALUE (@var{valtype}, @var{func}) |
feca2ed3 JW |
3861 | A C expression to create an RTX representing the place where a |
3862 | function returns a value of data type @var{valtype}. @var{valtype} is | |
3863 | a tree node representing a data type. Write @code{TYPE_MODE | |
3864 | (@var{valtype})} to get the machine mode used to represent that type. | |
3865 | On many machines, only the mode is relevant. (Actually, on most | |
3866 | machines, scalar values are returned in the same place regardless of | |
bd819a4a | 3867 | mode). |
feca2ed3 JW |
3868 | |
3869 | The value of the expression is usually a @code{reg} RTX for the hard | |
3870 | register where the return value is stored. The value can also be a | |
3871 | @code{parallel} RTX, if the return value is in multiple places. See | |
3872 | @code{FUNCTION_ARG} for an explanation of the @code{parallel} form. | |
3873 | ||
04ab46a4 | 3874 | If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply the same |
feca2ed3 JW |
3875 | promotion rules specified in @code{PROMOTE_MODE} if @var{valtype} is a |
3876 | scalar type. | |
3877 | ||
3878 | If the precise function being called is known, @var{func} is a tree | |
3879 | node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null | |
3880 | pointer. This makes it possible to use a different value-returning | |
3881 | convention for specific functions when all their calls are | |
bd819a4a | 3882 | known. |
feca2ed3 JW |
3883 | |
3884 | @code{FUNCTION_VALUE} is not used for return vales with aggregate data | |
3885 | types, because these are returned in another way. See | |
cea28603 | 3886 | @code{TARGET_STRUCT_VALUE_RTX} and related macros, below. |
a2c4f8e0 | 3887 | @end defmac |
feca2ed3 | 3888 | |
a2c4f8e0 | 3889 | @defmac FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func}) |
feca2ed3 JW |
3890 | Define this macro if the target machine has ``register windows'' |
3891 | so that the register in which a function returns its value is not | |
3892 | the same as the one in which the caller sees the value. | |
3893 | ||
3894 | For such machines, @code{FUNCTION_VALUE} computes the register in which | |
3895 | the caller will see the value. @code{FUNCTION_OUTGOING_VALUE} should be | |
3896 | defined in a similar fashion to tell the function where to put the | |
bd819a4a | 3897 | value. |
feca2ed3 JW |
3898 | |
3899 | If @code{FUNCTION_OUTGOING_VALUE} is not defined, | |
bd819a4a | 3900 | @code{FUNCTION_VALUE} serves both purposes. |
feca2ed3 JW |
3901 | |
3902 | @code{FUNCTION_OUTGOING_VALUE} is not used for return vales with | |
3903 | aggregate data types, because these are returned in another way. See | |
cea28603 | 3904 | @code{TARGET_STRUCT_VALUE_RTX} and related macros, below. |
a2c4f8e0 | 3905 | @end defmac |
feca2ed3 | 3906 | |
a2c4f8e0 | 3907 | @defmac LIBCALL_VALUE (@var{mode}) |
feca2ed3 JW |
3908 | A C expression to create an RTX representing the place where a library |
3909 | function returns a value of mode @var{mode}. If the precise function | |
3910 | being called is known, @var{func} is a tree node | |
3911 | (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null | |
3912 | pointer. This makes it possible to use a different value-returning | |
3913 | convention for specific functions when all their calls are | |
bd819a4a | 3914 | known. |
feca2ed3 JW |
3915 | |
3916 | Note that ``library function'' in this context means a compiler | |
3917 | support routine, used to perform arithmetic, whose name is known | |
3918 | specially by the compiler and was not mentioned in the C code being | |
3919 | compiled. | |
3920 | ||
3921 | The definition of @code{LIBRARY_VALUE} need not be concerned aggregate | |
3922 | data types, because none of the library functions returns such types. | |
a2c4f8e0 | 3923 | @end defmac |
feca2ed3 | 3924 | |
a2c4f8e0 | 3925 | @defmac FUNCTION_VALUE_REGNO_P (@var{regno}) |
feca2ed3 JW |
3926 | A C expression that is nonzero if @var{regno} is the number of a hard |
3927 | register in which the values of called function may come back. | |
3928 | ||
3929 | A register whose use for returning values is limited to serving as the | |
3930 | second of a pair (for a value of type @code{double}, say) need not be | |
3931 | recognized by this macro. So for most machines, this definition | |
3932 | suffices: | |
3933 | ||
3ab51846 | 3934 | @smallexample |
feca2ed3 | 3935 | #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0) |
3ab51846 | 3936 | @end smallexample |
feca2ed3 JW |
3937 | |
3938 | If the machine has register windows, so that the caller and the called | |
3939 | function use different registers for the return value, this macro | |
3940 | should recognize only the caller's register numbers. | |
a2c4f8e0 | 3941 | @end defmac |
feca2ed3 | 3942 | |
a2c4f8e0 | 3943 | @defmac APPLY_RESULT_SIZE |
feca2ed3 JW |
3944 | Define this macro if @samp{untyped_call} and @samp{untyped_return} |
3945 | need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for | |
3946 | saving and restoring an arbitrary return value. | |
a2c4f8e0 | 3947 | @end defmac |
feca2ed3 | 3948 | |
c988af2b RS |
3949 | @deftypefn {Target Hook} bool TARGET_RETURN_IN_MSB (tree @var{type}) |
3950 | This hook should return true if values of type @var{type} are returned | |
3951 | at the most significant end of a register (in other words, if they are | |
3952 | padded at the least significant end). You can assume that @var{type} | |
3953 | is returned in a register; the caller is required to check this. | |
3954 | ||
3955 | Note that the register provided by @code{FUNCTION_VALUE} must be able | |
3956 | to hold the complete return value. For example, if a 1-, 2- or 3-byte | |
3957 | structure is returned at the most significant end of a 4-byte register, | |
3958 | @code{FUNCTION_VALUE} should provide an @code{SImode} rtx. | |
3959 | @end deftypefn | |
3960 | ||
feca2ed3 JW |
3961 | @node Aggregate Return |
3962 | @subsection How Large Values Are Returned | |
3963 | @cindex aggregates as return values | |
3964 | @cindex large return values | |
3965 | @cindex returning aggregate values | |
3966 | @cindex structure value address | |
3967 | ||
3968 | When a function value's mode is @code{BLKmode} (and in some other | |
3969 | cases), the value is not returned according to @code{FUNCTION_VALUE} | |
3970 | (@pxref{Scalar Return}). Instead, the caller passes the address of a | |
3971 | block of memory in which the value should be stored. This address | |
3972 | is called the @dfn{structure value address}. | |
3973 | ||
3974 | This section describes how to control returning structure values in | |
3975 | memory. | |
3976 | ||
d624465f | 3977 | @deftypefn {Target Hook} bool TARGET_RETURN_IN_MEMORY (tree @var{type}, tree @var{fntype}) |
61f71b34 DD |
3978 | This target hook should return a nonzero value to say to return the |
3979 | function value in memory, just as large structures are always returned. | |
3980 | Here @var{type} will be the data type of the value, and @var{fntype} | |
3981 | will be the type of the function doing the returning, or @code{NULL} for | |
3982 | libcalls. | |
feca2ed3 JW |
3983 | |
3984 | Note that values of mode @code{BLKmode} must be explicitly handled | |
61f71b34 | 3985 | by this function. Also, the option @option{-fpcc-struct-return} |
feca2ed3 | 3986 | takes effect regardless of this macro. On most systems, it is |
61f71b34 | 3987 | possible to leave the hook undefined; this causes a default |
feca2ed3 JW |
3988 | definition to be used, whose value is the constant 1 for @code{BLKmode} |
3989 | values, and 0 otherwise. | |
3990 | ||
61f71b34 | 3991 | Do not use this hook to indicate that structures and unions should always |
feca2ed3 JW |
3992 | be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN} |
3993 | to indicate this. | |
61f71b34 | 3994 | @end deftypefn |
feca2ed3 | 3995 | |
a2c4f8e0 | 3996 | @defmac DEFAULT_PCC_STRUCT_RETURN |
feca2ed3 JW |
3997 | Define this macro to be 1 if all structure and union return values must be |
3998 | in memory. Since this results in slower code, this should be defined | |
161d7b59 | 3999 | only if needed for compatibility with other compilers or with an ABI@. |
feca2ed3 | 4000 | If you define this macro to be 0, then the conventions used for structure |
d624465f KH |
4001 | and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY} |
4002 | target hook. | |
feca2ed3 JW |
4003 | |
4004 | If not defined, this defaults to the value 1. | |
a2c4f8e0 | 4005 | @end defmac |
feca2ed3 | 4006 | |
61f71b34 DD |
4007 | @deftypefn {Target Hook} rtx TARGET_STRUCT_VALUE_RTX (tree @var{fndecl}, int @var{incoming}) |
4008 | This target hook should return the location of the structure value | |
4009 | address (normally a @code{mem} or @code{reg}), or 0 if the address is | |
4010 | passed as an ``invisible'' first argument. Note that @var{fndecl} may | |
1f6acb82 KH |
4011 | be @code{NULL}, for libcalls. You do not need to define this target |
4012 | hook if the address is always passed as an ``invisible'' first | |
4013 | argument. | |
feca2ed3 | 4014 | |
feca2ed3 JW |
4015 | On some architectures the place where the structure value address |
4016 | is found by the called function is not the same place that the | |
4017 | caller put it. This can be due to register windows, or it could | |
4018 | be because the function prologue moves it to a different place. | |
61f71b34 DD |
4019 | @var{incoming} is @code{true} when the location is needed in |
4020 | the context of the called function, and @code{false} in the context of | |
4021 | the caller. | |
feca2ed3 | 4022 | |
61f71b34 DD |
4023 | If @var{incoming} is @code{true} and the address is to be found on the |
4024 | stack, return a @code{mem} which refers to the frame pointer. | |
4025 | @end deftypefn | |
feca2ed3 | 4026 | |
a2c4f8e0 | 4027 | @defmac PCC_STATIC_STRUCT_RETURN |
feca2ed3 JW |
4028 | Define this macro if the usual system convention on the target machine |
4029 | for returning structures and unions is for the called function to return | |
4030 | the address of a static variable containing the value. | |
4031 | ||
4032 | Do not define this if the usual system convention is for the caller to | |
4033 | pass an address to the subroutine. | |
4034 | ||
630d3d5a JM |
4035 | This macro has effect in @option{-fpcc-struct-return} mode, but it does |
4036 | nothing when you use @option{-freg-struct-return} mode. | |
a2c4f8e0 | 4037 | @end defmac |
feca2ed3 JW |
4038 | |
4039 | @node Caller Saves | |
4040 | @subsection Caller-Saves Register Allocation | |
4041 | ||
a3a15b4d | 4042 | If you enable it, GCC can save registers around function calls. This |
feca2ed3 JW |
4043 | makes it possible to use call-clobbered registers to hold variables that |
4044 | must live across calls. | |
4045 | ||
a2c4f8e0 | 4046 | @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls}) |
feca2ed3 JW |
4047 | A C expression to determine whether it is worthwhile to consider placing |
4048 | a pseudo-register in a call-clobbered hard register and saving and | |
4049 | restoring it around each function call. The expression should be 1 when | |
4050 | this is worth doing, and 0 otherwise. | |
4051 | ||
4052 | If you don't define this macro, a default is used which is good on most | |
4053 | machines: @code{4 * @var{calls} < @var{refs}}. | |
a2c4f8e0 | 4054 | @end defmac |
8d5c8167 | 4055 | |
a2c4f8e0 | 4056 | @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs}) |
8d5c8167 JL |
4057 | A C expression specifying which mode is required for saving @var{nregs} |
4058 | of a pseudo-register in call-clobbered hard register @var{regno}. If | |
4059 | @var{regno} is unsuitable for caller save, @code{VOIDmode} should be | |
4060 | returned. For most machines this macro need not be defined since GCC | |
4061 | will select the smallest suitable mode. | |
a2c4f8e0 | 4062 | @end defmac |
feca2ed3 JW |
4063 | |
4064 | @node Function Entry | |
4065 | @subsection Function Entry and Exit | |
4066 | @cindex function entry and exit | |
4067 | @cindex prologue | |
4068 | @cindex epilogue | |
4069 | ||
4070 | This section describes the macros that output function entry | |
4071 | (@dfn{prologue}) and exit (@dfn{epilogue}) code. | |
4072 | ||
08c148a8 NB |
4073 | @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size}) |
4074 | If defined, a function that outputs the assembler code for entry to a | |
feca2ed3 JW |
4075 | function. The prologue is responsible for setting up the stack frame, |
4076 | initializing the frame pointer register, saving registers that must be | |
4077 | saved, and allocating @var{size} additional bytes of storage for the | |
4078 | local variables. @var{size} is an integer. @var{file} is a stdio | |
4079 | stream to which the assembler code should be output. | |
4080 | ||
4081 | The label for the beginning of the function need not be output by this | |
4082 | macro. That has already been done when the macro is run. | |
4083 | ||
4084 | @findex regs_ever_live | |
4085 | To determine which registers to save, the macro can refer to the array | |
4086 | @code{regs_ever_live}: element @var{r} is nonzero if hard register | |
4087 | @var{r} is used anywhere within the function. This implies the function | |
4088 | prologue should save register @var{r}, provided it is not one of the | |
08c148a8 | 4089 | call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use |
feca2ed3 JW |
4090 | @code{regs_ever_live}.) |
4091 | ||
4092 | On machines that have ``register windows'', the function entry code does | |
4093 | not save on the stack the registers that are in the windows, even if | |
4094 | they are supposed to be preserved by function calls; instead it takes | |
4095 | appropriate steps to ``push'' the register stack, if any non-call-used | |
4096 | registers are used in the function. | |
4097 | ||
4098 | @findex frame_pointer_needed | |
4099 | On machines where functions may or may not have frame-pointers, the | |
4100 | function entry code must vary accordingly; it must set up the frame | |
4101 | pointer if one is wanted, and not otherwise. To determine whether a | |
4102 | frame pointer is in wanted, the macro can refer to the variable | |
4103 | @code{frame_pointer_needed}. The variable's value will be 1 at run | |
4104 | time in a function that needs a frame pointer. @xref{Elimination}. | |
4105 | ||
4106 | The function entry code is responsible for allocating any stack space | |
4107 | required for the function. This stack space consists of the regions | |
4108 | listed below. In most cases, these regions are allocated in the | |
4109 | order listed, with the last listed region closest to the top of the | |
4110 | stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and | |
4111 | the highest address if it is not defined). You can use a different order | |
4112 | for a machine if doing so is more convenient or required for | |
4113 | compatibility reasons. Except in cases where required by standard | |
4114 | or by a debugger, there is no reason why the stack layout used by GCC | |
4115 | need agree with that used by other compilers for a machine. | |
08c148a8 NB |
4116 | @end deftypefn |
4117 | ||
17b53c33 NB |
4118 | @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file}) |
4119 | If defined, a function that outputs assembler code at the end of a | |
4120 | prologue. This should be used when the function prologue is being | |
4121 | emitted as RTL, and you have some extra assembler that needs to be | |
4122 | emitted. @xref{prologue instruction pattern}. | |
4123 | @end deftypefn | |
4124 | ||
4125 | @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file}) | |
4126 | If defined, a function that outputs assembler code at the start of an | |
4127 | epilogue. This should be used when the function epilogue is being | |
4128 | emitted as RTL, and you have some extra assembler that needs to be | |
4129 | emitted. @xref{epilogue instruction pattern}. | |
4130 | @end deftypefn | |
4131 | ||
08c148a8 NB |
4132 | @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size}) |
4133 | If defined, a function that outputs the assembler code for exit from a | |
4134 | function. The epilogue is responsible for restoring the saved | |
4135 | registers and stack pointer to their values when the function was | |
4136 | called, and returning control to the caller. This macro takes the | |
4137 | same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the | |
4138 | registers to restore are determined from @code{regs_ever_live} and | |
4139 | @code{CALL_USED_REGISTERS} in the same way. | |
4140 | ||
4141 | On some machines, there is a single instruction that does all the work | |
4142 | of returning from the function. On these machines, give that | |
4143 | instruction the name @samp{return} and do not define the macro | |
4144 | @code{TARGET_ASM_FUNCTION_EPILOGUE} at all. | |
4145 | ||
4146 | Do not define a pattern named @samp{return} if you want the | |
4147 | @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target | |
4148 | switches to control whether return instructions or epilogues are used, | |
4149 | define a @samp{return} pattern with a validity condition that tests the | |
4150 | target switches appropriately. If the @samp{return} pattern's validity | |
4151 | condition is false, epilogues will be used. | |
4152 | ||
4153 | On machines where functions may or may not have frame-pointers, the | |
4154 | function exit code must vary accordingly. Sometimes the code for these | |
4155 | two cases is completely different. To determine whether a frame pointer | |
4156 | is wanted, the macro can refer to the variable | |
4157 | @code{frame_pointer_needed}. The variable's value will be 1 when compiling | |
4158 | a function that needs a frame pointer. | |
4159 | ||
4160 | Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and | |
4161 | @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially. | |
4162 | The C variable @code{current_function_is_leaf} is nonzero for such a | |
4163 | function. @xref{Leaf Functions}. | |
4164 | ||
4165 | On some machines, some functions pop their arguments on exit while | |
4166 | others leave that for the caller to do. For example, the 68020 when | |
4167 | given @option{-mrtd} pops arguments in functions that take a fixed | |
4168 | number of arguments. | |
4169 | ||
4170 | @findex current_function_pops_args | |
4171 | Your definition of the macro @code{RETURN_POPS_ARGS} decides which | |
4172 | functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE} | |
4173 | needs to know what was decided. The variable that is called | |
4174 | @code{current_function_pops_args} is the number of bytes of its | |
4175 | arguments that a function should pop. @xref{Scalar Return}. | |
4176 | @c what is the "its arguments" in the above sentence referring to, pray | |
4177 | @c tell? --mew 5feb93 | |
4178 | @end deftypefn | |
4179 | ||
feca2ed3 JW |
4180 | @itemize @bullet |
4181 | @item | |
4182 | @findex current_function_pretend_args_size | |
4183 | A region of @code{current_function_pretend_args_size} bytes of | |
4184 | uninitialized space just underneath the first argument arriving on the | |
4185 | stack. (This may not be at the very start of the allocated stack region | |
4186 | if the calling sequence has pushed anything else since pushing the stack | |
4187 | arguments. But usually, on such machines, nothing else has been pushed | |
4188 | yet, because the function prologue itself does all the pushing.) This | |
4189 | region is used on machines where an argument may be passed partly in | |
4190 | registers and partly in memory, and, in some cases to support the | |
6c535c69 | 4191 | features in @code{<stdarg.h>}. |
feca2ed3 JW |
4192 | |
4193 | @item | |
4194 | An area of memory used to save certain registers used by the function. | |
4195 | The size of this area, which may also include space for such things as | |
4196 | the return address and pointers to previous stack frames, is | |
4197 | machine-specific and usually depends on which registers have been used | |
4198 | in the function. Machines with register windows often do not require | |
4199 | a save area. | |
4200 | ||
4201 | @item | |
4202 | A region of at least @var{size} bytes, possibly rounded up to an allocation | |
4203 | boundary, to contain the local variables of the function. On some machines, | |
4204 | this region and the save area may occur in the opposite order, with the | |
4205 | save area closer to the top of the stack. | |
4206 | ||
4207 | @item | |
4208 | @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames | |
4209 | Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of | |
4210 | @code{current_function_outgoing_args_size} bytes to be used for outgoing | |
4211 | argument lists of the function. @xref{Stack Arguments}. | |
4212 | @end itemize | |
4213 | ||
a2c4f8e0 | 4214 | @defmac EXIT_IGNORE_STACK |
feca2ed3 JW |
4215 | Define this macro as a C expression that is nonzero if the return |
4216 | instruction or the function epilogue ignores the value of the stack | |
4217 | pointer; in other words, if it is safe to delete an instruction to | |
9d05bbce KH |
4218 | adjust the stack pointer before a return from the function. The |
4219 | default is 0. | |
feca2ed3 JW |
4220 | |
4221 | Note that this macro's value is relevant only for functions for which | |
4222 | frame pointers are maintained. It is never safe to delete a final | |
4223 | stack adjustment in a function that has no frame pointer, and the | |
4224 | compiler knows this regardless of @code{EXIT_IGNORE_STACK}. | |
a2c4f8e0 | 4225 | @end defmac |
feca2ed3 | 4226 | |
a2c4f8e0 | 4227 | @defmac EPILOGUE_USES (@var{regno}) |
8760eaae | 4228 | Define this macro as a C expression that is nonzero for registers that are |
feca2ed3 JW |
4229 | used by the epilogue or the @samp{return} pattern. The stack and frame |
4230 | pointer registers are already be assumed to be used as needed. | |
a2c4f8e0 | 4231 | @end defmac |
feca2ed3 | 4232 | |
a2c4f8e0 | 4233 | @defmac EH_USES (@var{regno}) |
15b5aef3 RH |
4234 | Define this macro as a C expression that is nonzero for registers that are |
4235 | used by the exception handling mechanism, and so should be considered live | |
4236 | on entry to an exception edge. | |
a2c4f8e0 | 4237 | @end defmac |
15b5aef3 | 4238 | |
a2c4f8e0 | 4239 | @defmac DELAY_SLOTS_FOR_EPILOGUE |
feca2ed3 JW |
4240 | Define this macro if the function epilogue contains delay slots to which |
4241 | instructions from the rest of the function can be ``moved''. The | |
4242 | definition should be a C expression whose value is an integer | |
4243 | representing the number of delay slots there. | |
a2c4f8e0 | 4244 | @end defmac |
feca2ed3 | 4245 | |
a2c4f8e0 | 4246 | @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n}) |
feca2ed3 JW |
4247 | A C expression that returns 1 if @var{insn} can be placed in delay |
4248 | slot number @var{n} of the epilogue. | |
4249 | ||
4250 | The argument @var{n} is an integer which identifies the delay slot now | |
4251 | being considered (since different slots may have different rules of | |
4252 | eligibility). It is never negative and is always less than the number | |
4253 | of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns). | |
4254 | If you reject a particular insn for a given delay slot, in principle, it | |
4255 | may be reconsidered for a subsequent delay slot. Also, other insns may | |
4256 | (at least in principle) be considered for the so far unfilled delay | |
4257 | slot. | |
4258 | ||
4259 | @findex current_function_epilogue_delay_list | |
4260 | @findex final_scan_insn | |
4261 | The insns accepted to fill the epilogue delay slots are put in an RTL | |
4262 | list made with @code{insn_list} objects, stored in the variable | |
4263 | @code{current_function_epilogue_delay_list}. The insn for the first | |
4264 | delay slot comes first in the list. Your definition of the macro | |
08c148a8 NB |
4265 | @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by |
4266 | outputting the insns in this list, usually by calling | |
4267 | @code{final_scan_insn}. | |
feca2ed3 JW |
4268 | |
4269 | You need not define this macro if you did not define | |
4270 | @code{DELAY_SLOTS_FOR_EPILOGUE}. | |
a2c4f8e0 | 4271 | @end defmac |
feca2ed3 | 4272 | |
65e71cd6 | 4273 | @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_MI_THUNK (FILE *@var{file}, tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, tree @var{function}) |
483ab821 | 4274 | A function that outputs the assembler code for a thunk |
feca2ed3 JW |
4275 | function, used to implement C++ virtual function calls with multiple |
4276 | inheritance. The thunk acts as a wrapper around a virtual function, | |
4277 | adjusting the implicit object parameter before handing control off to | |
4278 | the real function. | |
4279 | ||
4280 | First, emit code to add the integer @var{delta} to the location that | |
4281 | contains the incoming first argument. Assume that this argument | |
4282 | contains a pointer, and is the one used to pass the @code{this} pointer | |
4283 | in C++. This is the incoming argument @emph{before} the function prologue, | |
e979f9e8 | 4284 | e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of |
feca2ed3 JW |
4285 | all other incoming arguments. |
4286 | ||
65e71cd6 EB |
4287 | Then, if @var{vcall_offset} is nonzero, an additional adjustment should be |
4288 | made after adding @code{delta}. In particular, if @var{p} is the | |
4289 | adjusted pointer, the following adjustment should be made: | |
4290 | ||
4291 | @smallexample | |
4292 | p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)] | |
4293 | @end smallexample | |
4294 | ||
4295 | After the additions, emit code to jump to @var{function}, which is a | |
feca2ed3 JW |
4296 | @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does |
4297 | not touch the return address. Hence returning from @var{FUNCTION} will | |
4298 | return to whoever called the current @samp{thunk}. | |
4299 | ||
4300 | The effect must be as if @var{function} had been called directly with | |
4301 | the adjusted first argument. This macro is responsible for emitting all | |
08c148a8 NB |
4302 | of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE} |
4303 | and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked. | |
feca2ed3 JW |
4304 | |
4305 | The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function} | |
4306 | have already been extracted from it.) It might possibly be useful on | |
4307 | some targets, but probably not. | |
4308 | ||
861bb6c1 | 4309 | If you do not define this macro, the target-independent code in the C++ |
c771326b | 4310 | front end will generate a less efficient heavyweight thunk that calls |
861bb6c1 JL |
4311 | @var{function} instead of jumping to it. The generic approach does |
4312 | not support varargs. | |
483ab821 MM |
4313 | @end deftypefn |
4314 | ||
65e71cd6 EB |
4315 | @deftypefn {Target Hook} bool TARGET_ASM_CAN_OUTPUT_MI_THUNK (tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, tree @var{function}) |
4316 | A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able | |
4317 | to output the assembler code for the thunk function specified by the | |
4318 | arguments it is passed, and false otherwise. In the latter case, the | |
4319 | generic approach will be used by the C++ front end, with the limitations | |
4320 | previously exposed. | |
483ab821 | 4321 | @end deftypefn |
feca2ed3 JW |
4322 | |
4323 | @node Profiling | |
4324 | @subsection Generating Code for Profiling | |
4325 | @cindex profiling, code generation | |
4326 | ||
4327 | These macros will help you generate code for profiling. | |
4328 | ||
a2c4f8e0 | 4329 | @defmac FUNCTION_PROFILER (@var{file}, @var{labelno}) |
feca2ed3 JW |
4330 | A C statement or compound statement to output to @var{file} some |
4331 | assembler code to call the profiling subroutine @code{mcount}. | |
feca2ed3 JW |
4332 | |
4333 | @findex mcount | |
980e2067 | 4334 | The details of how @code{mcount} expects to be called are determined by |
161d7b59 | 4335 | your operating system environment, not by GCC@. To figure them out, |
980e2067 JL |
4336 | compile a small program for profiling using the system's installed C |
4337 | compiler and look at the assembler code that results. | |
4338 | ||
4339 | Older implementations of @code{mcount} expect the address of a counter | |
4340 | variable to be loaded into some register. The name of this variable is | |
4341 | @samp{LP} followed by the number @var{labelno}, so you would generate | |
4342 | the name using @samp{LP%d} in a @code{fprintf}. | |
a2c4f8e0 | 4343 | @end defmac |
980e2067 | 4344 | |
a2c4f8e0 | 4345 | @defmac PROFILE_HOOK |
411707f4 CC |
4346 | A C statement or compound statement to output to @var{file} some assembly |
4347 | code to call the profiling subroutine @code{mcount} even the target does | |
4348 | not support profiling. | |
a2c4f8e0 | 4349 | @end defmac |
411707f4 | 4350 | |
a2c4f8e0 | 4351 | @defmac NO_PROFILE_COUNTERS |
980e2067 JL |
4352 | Define this macro if the @code{mcount} subroutine on your system does |
4353 | not need a counter variable allocated for each function. This is true | |
4354 | for almost all modern implementations. If you define this macro, you | |
4355 | must not use the @var{labelno} argument to @code{FUNCTION_PROFILER}. | |
a2c4f8e0 | 4356 | @end defmac |
feca2ed3 | 4357 | |
a2c4f8e0 | 4358 | @defmac PROFILE_BEFORE_PROLOGUE |
feca2ed3 JW |
4359 | Define this macro if the code for function profiling should come before |
4360 | the function prologue. Normally, the profiling code comes after. | |
a2c4f8e0 | 4361 | @end defmac |
feca2ed3 | 4362 | |
91d231cb JM |
4363 | @node Tail Calls |
4364 | @subsection Permitting tail calls | |
4365 | @cindex tail calls | |
b36f4ed3 | 4366 | |
4977bab6 ZW |
4367 | @deftypefn {Target Hook} bool TARGET_FUNCTION_OK_FOR_SIBCALL (tree @var{decl}, tree @var{exp}) |
4368 | True if it is ok to do sibling call optimization for the specified | |
4369 | call expression @var{exp}. @var{decl} will be the called function, | |
4370 | or @code{NULL} if this is an indirect call. | |
4cb1433c RH |
4371 | |
4372 | It is not uncommon for limitations of calling conventions to prevent | |
4373 | tail calls to functions outside the current unit of translation, or | |
4977bab6 | 4374 | during PIC compilation. The hook is used to enforce these restrictions, |
02f52e19 | 4375 | as the @code{sibcall} md pattern can not fail, or fall over to a |
4977bab6 ZW |
4376 | ``normal'' call. The criteria for successful sibling call optimization |
4377 | may vary greatly between different architectures. | |
4378 | @end deftypefn | |
4cb1433c | 4379 | |
7d69de61 RH |
4380 | @node Stack Smashing Protection |
4381 | @subsection Stack smashing protection | |
4382 | @cindex stack smashing protection | |
4383 | ||
4384 | @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_GUARD (void) | |
4385 | This hook returns a @code{DECL} node for the external variable to use | |
4386 | for the stack protection guard. This variable is initialized by the | |
4387 | runtime to some random value and is used to initialize the guard value | |
4388 | that is placed at the top of the local stack frame. The type of this | |
4389 | variable must be @code{ptr_type_node}. | |
4390 | ||
4391 | The default version of this hook creates a variable called | |
4392 | @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}. | |
4393 | @end deftypefn | |
4394 | ||
4395 | @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_FAIL (void) | |
4396 | This hook returns a tree expression that alerts the runtime that the | |
4397 | stack protect guard variable has been modified. This expression should | |
4398 | involve a call to a @code{noreturn} function. | |
4399 | ||
4400 | The default version of this hook invokes a function called | |
4401 | @samp{__stack_chk_fail}, taking no arguments. This function is | |
4402 | normally defined in @file{libgcc2.c}. | |
4403 | @end deftypefn | |
4404 | ||
feca2ed3 JW |
4405 | @node Varargs |
4406 | @section Implementing the Varargs Macros | |
4407 | @cindex varargs implementation | |
4408 | ||
aee96fe9 JM |
4409 | GCC comes with an implementation of @code{<varargs.h>} and |
4410 | @code{<stdarg.h>} that work without change on machines that pass arguments | |
feca2ed3 JW |
4411 | on the stack. Other machines require their own implementations of |
4412 | varargs, and the two machine independent header files must have | |
4413 | conditionals to include it. | |
4414 | ||
aee96fe9 | 4415 | ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in |
feca2ed3 JW |
4416 | the calling convention for @code{va_start}. The traditional |
4417 | implementation takes just one argument, which is the variable in which | |
5490d604 | 4418 | to store the argument pointer. The ISO implementation of |
feca2ed3 JW |
4419 | @code{va_start} takes an additional second argument. The user is |
4420 | supposed to write the last named argument of the function here. | |
4421 | ||
4422 | However, @code{va_start} should not use this argument. The way to find | |
4423 | the end of the named arguments is with the built-in functions described | |
4424 | below. | |
4425 | ||
a2c4f8e0 | 4426 | @defmac __builtin_saveregs () |
feca2ed3 | 4427 | Use this built-in function to save the argument registers in memory so |
5490d604 | 4428 | that the varargs mechanism can access them. Both ISO and traditional |
feca2ed3 | 4429 | versions of @code{va_start} must use @code{__builtin_saveregs}, unless |
c2379679 | 4430 | you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead. |
feca2ed3 JW |
4431 | |
4432 | On some machines, @code{__builtin_saveregs} is open-coded under the | |
f61c92c3 KH |
4433 | control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On |
4434 | other machines, it calls a routine written in assembler language, | |
4435 | found in @file{libgcc2.c}. | |
feca2ed3 JW |
4436 | |
4437 | Code generated for the call to @code{__builtin_saveregs} appears at the | |
4438 | beginning of the function, as opposed to where the call to | |
4439 | @code{__builtin_saveregs} is written, regardless of what the code is. | |
4440 | This is because the registers must be saved before the function starts | |
4441 | to use them for its own purposes. | |
4442 | @c i rewrote the first sentence above to fix an overfull hbox. --mew | |
4443 | @c 10feb93 | |
a2c4f8e0 | 4444 | @end defmac |
feca2ed3 | 4445 | |
a2c4f8e0 | 4446 | @defmac __builtin_args_info (@var{category}) |
feca2ed3 JW |
4447 | Use this built-in function to find the first anonymous arguments in |
4448 | registers. | |
4449 | ||
4450 | In general, a machine may have several categories of registers used for | |
4451 | arguments, each for a particular category of data types. (For example, | |
4452 | on some machines, floating-point registers are used for floating-point | |
4453 | arguments while other arguments are passed in the general registers.) | |
4454 | To make non-varargs functions use the proper calling convention, you | |
4455 | have defined the @code{CUMULATIVE_ARGS} data type to record how many | |
4456 | registers in each category have been used so far | |
4457 | ||
4458 | @code{__builtin_args_info} accesses the same data structure of type | |
4459 | @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished | |
4460 | with it, with @var{category} specifying which word to access. Thus, the | |
4461 | value indicates the first unused register in a given category. | |
4462 | ||
4463 | Normally, you would use @code{__builtin_args_info} in the implementation | |
4464 | of @code{va_start}, accessing each category just once and storing the | |
4465 | value in the @code{va_list} object. This is because @code{va_list} will | |
4466 | have to update the values, and there is no way to alter the | |
4467 | values accessed by @code{__builtin_args_info}. | |
a2c4f8e0 | 4468 | @end defmac |
feca2ed3 | 4469 | |
a2c4f8e0 | 4470 | @defmac __builtin_next_arg (@var{lastarg}) |
feca2ed3 JW |
4471 | This is the equivalent of @code{__builtin_args_info}, for stack |
4472 | arguments. It returns the address of the first anonymous stack | |
767094dd | 4473 | argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it |
feca2ed3 JW |
4474 | returns the address of the location above the first anonymous stack |
4475 | argument. Use it in @code{va_start} to initialize the pointer for | |
4476 | fetching arguments from the stack. Also use it in @code{va_start} to | |
4477 | verify that the second parameter @var{lastarg} is the last named argument | |
4478 | of the current function. | |
a2c4f8e0 | 4479 | @end defmac |
feca2ed3 | 4480 | |
a2c4f8e0 | 4481 | @defmac __builtin_classify_type (@var{object}) |
feca2ed3 JW |
4482 | Since each machine has its own conventions for which data types are |
4483 | passed in which kind of register, your implementation of @code{va_arg} | |
4484 | has to embody these conventions. The easiest way to categorize the | |
4485 | specified data type is to use @code{__builtin_classify_type} together | |
4486 | with @code{sizeof} and @code{__alignof__}. | |
4487 | ||
4488 | @code{__builtin_classify_type} ignores the value of @var{object}, | |
4489 | considering only its data type. It returns an integer describing what | |
4490 | kind of type that is---integer, floating, pointer, structure, and so on. | |
4491 | ||
4492 | The file @file{typeclass.h} defines an enumeration that you can use to | |
4493 | interpret the values of @code{__builtin_classify_type}. | |
a2c4f8e0 | 4494 | @end defmac |
feca2ed3 JW |
4495 | |
4496 | These machine description macros help implement varargs: | |
4497 | ||
61f71b34 DD |
4498 | @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN_SAVEREGS (void) |
4499 | If defined, this hook produces the machine-specific code for a call to | |
4500 | @code{__builtin_saveregs}. This code will be moved to the very | |
4501 | beginning of the function, before any parameter access are made. The | |
4502 | return value of this function should be an RTX that contains the value | |
4503 | to use as the return of @code{__builtin_saveregs}. | |
4504 | @end deftypefn | |
feca2ed3 | 4505 | |
61f71b34 DD |
4506 | @deftypefn {Target Hook} void TARGET_SETUP_INCOMING_VARARGS (CUMULATIVE_ARGS *@var{args_so_far}, enum machine_mode @var{mode}, tree @var{type}, int *@var{pretend_args_size}, int @var{second_time}) |
4507 | This target hook offers an alternative to using | |
4508 | @code{__builtin_saveregs} and defining the hook | |
4509 | @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous | |
4510 | register arguments into the stack so that all the arguments appear to | |
4511 | have been passed consecutively on the stack. Once this is done, you can | |
4512 | use the standard implementation of varargs that works for machines that | |
4513 | pass all their arguments on the stack. | |
feca2ed3 | 4514 | |
61f71b34 | 4515 | The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data |
8760eaae | 4516 | structure, containing the values that are obtained after processing the |
feca2ed3 JW |
4517 | named arguments. The arguments @var{mode} and @var{type} describe the |
4518 | last named argument---its machine mode and its data type as a tree node. | |
4519 | ||
61f71b34 DD |
4520 | The target hook should do two things: first, push onto the stack all the |
4521 | argument registers @emph{not} used for the named arguments, and second, | |
4522 | store the size of the data thus pushed into the @code{int}-valued | |
4523 | variable pointed to by @var{pretend_args_size}. The value that you | |
4524 | store here will serve as additional offset for setting up the stack | |
4525 | frame. | |
feca2ed3 JW |
4526 | |
4527 | Because you must generate code to push the anonymous arguments at | |
4528 | compile time without knowing their data types, | |
61f71b34 DD |
4529 | @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that |
4530 | have just a single category of argument register and use it uniformly | |
4531 | for all data types. | |
feca2ed3 JW |
4532 | |
4533 | If the argument @var{second_time} is nonzero, it means that the | |
4534 | arguments of the function are being analyzed for the second time. This | |
4535 | happens for an inline function, which is not actually compiled until the | |
61f71b34 | 4536 | end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should |
feca2ed3 | 4537 | not generate any instructions in this case. |
61f71b34 | 4538 | @end deftypefn |
feca2ed3 | 4539 | |
61f71b34 DD |
4540 | @deftypefn {Target Hook} bool TARGET_STRICT_ARGUMENT_NAMING (CUMULATIVE_ARGS *@var{ca}) |
4541 | Define this hook to return @code{true} if the location where a function | |
e5e809f4 | 4542 | argument is passed depends on whether or not it is a named argument. |
feca2ed3 | 4543 | |
61f71b34 DD |
4544 | This hook controls how the @var{named} argument to @code{FUNCTION_ARG} |
4545 | is set for varargs and stdarg functions. If this hook returns | |
4546 | @code{true}, the @var{named} argument is always true for named | |
4547 | arguments, and false for unnamed arguments. If it returns @code{false}, | |
5b4ef0b1 | 4548 | but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true}, |
61f71b34 DD |
4549 | then all arguments are treated as named. Otherwise, all named arguments |
4550 | except the last are treated as named. | |
e5e809f4 | 4551 | |
61f71b34 DD |
4552 | You need not define this hook if it always returns zero. |
4553 | @end deftypefn | |
9ab70a9b | 4554 | |
61f71b34 | 4555 | @deftypefn {Target Hook} bool TARGET_PRETEND_OUTGOING_VARARGS_NAMED |
9ab70a9b | 4556 | If you need to conditionally change ABIs so that one works with |
61f71b34 DD |
4557 | @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither |
4558 | @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was | |
4559 | defined, then define this hook to return @code{true} if | |
c2379679 | 4560 | @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise. |
61f71b34 DD |
4561 | Otherwise, you should not define this hook. |
4562 | @end deftypefn | |
feca2ed3 JW |
4563 | |
4564 | @node Trampolines | |
4565 | @section Trampolines for Nested Functions | |
4566 | @cindex trampolines for nested functions | |
4567 | @cindex nested functions, trampolines for | |
4568 | ||
4569 | A @dfn{trampoline} is a small piece of code that is created at run time | |
4570 | when the address of a nested function is taken. It normally resides on | |
4571 | the stack, in the stack frame of the containing function. These macros | |
a3a15b4d | 4572 | tell GCC how to generate code to allocate and initialize a |
feca2ed3 JW |
4573 | trampoline. |
4574 | ||
4575 | The instructions in the trampoline must do two things: load a constant | |
4576 | address into the static chain register, and jump to the real address of | |
4577 | the nested function. On CISC machines such as the m68k, this requires | |
4578 | two instructions, a move immediate and a jump. Then the two addresses | |
4579 | exist in the trampoline as word-long immediate operands. On RISC | |
4580 | machines, it is often necessary to load each address into a register in | |
4581 | two parts. Then pieces of each address form separate immediate | |
4582 | operands. | |
4583 | ||
4584 | The code generated to initialize the trampoline must store the variable | |
4585 | parts---the static chain value and the function address---into the | |
4586 | immediate operands of the instructions. On a CISC machine, this is | |
4587 | simply a matter of copying each address to a memory reference at the | |
4588 | proper offset from the start of the trampoline. On a RISC machine, it | |
4589 | may be necessary to take out pieces of the address and store them | |
4590 | separately. | |
4591 | ||
a2c4f8e0 | 4592 | @defmac TRAMPOLINE_TEMPLATE (@var{file}) |
feca2ed3 JW |
4593 | A C statement to output, on the stream @var{file}, assembler code for a |
4594 | block of data that contains the constant parts of a trampoline. This | |
4595 | code should not include a label---the label is taken care of | |
4596 | automatically. | |
4597 | ||
4598 | If you do not define this macro, it means no template is needed | |
4599 | for the target. Do not define this macro on systems where the block move | |
4600 | code to copy the trampoline into place would be larger than the code | |
4601 | to generate it on the spot. | |
a2c4f8e0 | 4602 | @end defmac |
feca2ed3 | 4603 | |
a2c4f8e0 | 4604 | @defmac TRAMPOLINE_SECTION |
feca2ed3 JW |
4605 | The name of a subroutine to switch to the section in which the |
4606 | trampoline template is to be placed (@pxref{Sections}). The default is | |
4607 | a value of @samp{readonly_data_section}, which places the trampoline in | |
4608 | the section containing read-only data. | |
a2c4f8e0 | 4609 | @end defmac |
feca2ed3 | 4610 | |
a2c4f8e0 | 4611 | @defmac TRAMPOLINE_SIZE |
feca2ed3 | 4612 | A C expression for the size in bytes of the trampoline, as an integer. |
a2c4f8e0 | 4613 | @end defmac |
feca2ed3 | 4614 | |
a2c4f8e0 | 4615 | @defmac TRAMPOLINE_ALIGNMENT |
feca2ed3 JW |
4616 | Alignment required for trampolines, in bits. |
4617 | ||
4618 | If you don't define this macro, the value of @code{BIGGEST_ALIGNMENT} | |
4619 | is used for aligning trampolines. | |
a2c4f8e0 | 4620 | @end defmac |
feca2ed3 | 4621 | |
a2c4f8e0 | 4622 | @defmac INITIALIZE_TRAMPOLINE (@var{addr}, @var{fnaddr}, @var{static_chain}) |
feca2ed3 JW |
4623 | A C statement to initialize the variable parts of a trampoline. |
4624 | @var{addr} is an RTX for the address of the trampoline; @var{fnaddr} is | |
4625 | an RTX for the address of the nested function; @var{static_chain} is an | |
4626 | RTX for the static chain value that should be passed to the function | |
4627 | when it is called. | |
a2c4f8e0 | 4628 | @end defmac |
feca2ed3 | 4629 | |
a2c4f8e0 | 4630 | @defmac TRAMPOLINE_ADJUST_ADDRESS (@var{addr}) |
b33493e3 AO |
4631 | A C statement that should perform any machine-specific adjustment in |
4632 | the address of the trampoline. Its argument contains the address that | |
4633 | was passed to @code{INITIALIZE_TRAMPOLINE}. In case the address to be | |
4634 | used for a function call should be different from the address in which | |
4635 | the template was stored, the different address should be assigned to | |
4636 | @var{addr}. If this macro is not defined, @var{addr} will be used for | |
4637 | function calls. | |
4638 | ||
08c148a8 NB |
4639 | @cindex @code{TARGET_ASM_FUNCTION_EPILOGUE} and trampolines |
4640 | @cindex @code{TARGET_ASM_FUNCTION_PROLOGUE} and trampolines | |
feca2ed3 JW |
4641 | If this macro is not defined, by default the trampoline is allocated as |
4642 | a stack slot. This default is right for most machines. The exceptions | |
4643 | are machines where it is impossible to execute instructions in the stack | |
4644 | area. On such machines, you may have to implement a separate stack, | |
08c148a8 NB |
4645 | using this macro in conjunction with @code{TARGET_ASM_FUNCTION_PROLOGUE} |
4646 | and @code{TARGET_ASM_FUNCTION_EPILOGUE}. | |
feca2ed3 JW |
4647 | |
4648 | @var{fp} points to a data structure, a @code{struct function}, which | |
4649 | describes the compilation status of the immediate containing function of | |
0d569849 | 4650 | the function which the trampoline is for. The stack slot for the |
feca2ed3 JW |
4651 | trampoline is in the stack frame of this containing function. Other |
4652 | allocation strategies probably must do something analogous with this | |
4653 | information. | |
a2c4f8e0 | 4654 | @end defmac |
feca2ed3 JW |
4655 | |
4656 | Implementing trampolines is difficult on many machines because they have | |
4657 | separate instruction and data caches. Writing into a stack location | |
4658 | fails to clear the memory in the instruction cache, so when the program | |
4659 | jumps to that location, it executes the old contents. | |
4660 | ||
4661 | Here are two possible solutions. One is to clear the relevant parts of | |
4662 | the instruction cache whenever a trampoline is set up. The other is to | |
4663 | make all trampolines identical, by having them jump to a standard | |
4664 | subroutine. The former technique makes trampoline execution faster; the | |
4665 | latter makes initialization faster. | |
4666 | ||
4667 | To clear the instruction cache when a trampoline is initialized, define | |
f691dc3b | 4668 | the following macro. |
feca2ed3 | 4669 | |
a2c4f8e0 | 4670 | @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end}) |
feca2ed3 | 4671 | If defined, expands to a C expression clearing the @emph{instruction |
f691dc3b AJ |
4672 | cache} in the specified interval. The definition of this macro would |
4673 | typically be a series of @code{asm} statements. Both @var{beg} and | |
4674 | @var{end} are both pointer expressions. | |
a2c4f8e0 | 4675 | @end defmac |
feca2ed3 | 4676 | |
e7a742ec EB |
4677 | The operating system may also require the stack to be made executable |
4678 | before calling the trampoline. To implement this requirement, define | |
4679 | the following macro. | |
4680 | ||
4681 | @defmac ENABLE_EXECUTE_STACK | |
4682 | Define this macro if certain operations must be performed before executing | |
4683 | code located on the stack. The macro should expand to a series of C | |
431ae0bf | 4684 | file-scope constructs (e.g.@: functions) and provide a unique entry point |
e7a742ec EB |
4685 | named @code{__enable_execute_stack}. The target is responsible for |
4686 | emitting calls to the entry point in the code, for example from the | |
4687 | @code{INITIALIZE_TRAMPOLINE} macro. | |
4688 | @end defmac | |
4689 | ||
feca2ed3 JW |
4690 | To use a standard subroutine, define the following macro. In addition, |
4691 | you must make sure that the instructions in a trampoline fill an entire | |
4692 | cache line with identical instructions, or else ensure that the | |
4693 | beginning of the trampoline code is always aligned at the same point in | |
4694 | its cache line. Look in @file{m68k.h} as a guide. | |
4695 | ||
a2c4f8e0 | 4696 | @defmac TRANSFER_FROM_TRAMPOLINE |
feca2ed3 JW |
4697 | Define this macro if trampolines need a special subroutine to do their |
4698 | work. The macro should expand to a series of @code{asm} statements | |
161d7b59 | 4699 | which will be compiled with GCC@. They go in a library function named |
feca2ed3 JW |
4700 | @code{__transfer_from_trampoline}. |
4701 | ||
4702 | If you need to avoid executing the ordinary prologue code of a compiled | |
4703 | C function when you jump to the subroutine, you can do so by placing a | |
4704 | special label of your own in the assembler code. Use one @code{asm} | |
4705 | statement to generate an assembler label, and another to make the label | |
4706 | global. Then trampolines can use that label to jump directly to your | |
4707 | special assembler code. | |
a2c4f8e0 | 4708 | @end defmac |
feca2ed3 JW |
4709 | |
4710 | @node Library Calls | |
4711 | @section Implicit Calls to Library Routines | |
4712 | @cindex library subroutine names | |
4713 | @cindex @file{libgcc.a} | |
4714 | ||
4715 | @c prevent bad page break with this line | |
4716 | Here is an explanation of implicit calls to library routines. | |
4717 | ||
a2c4f8e0 | 4718 | @defmac DECLARE_LIBRARY_RENAMES |
d8088c6f BS |
4719 | This macro, if defined, should expand to a piece of C code that will get |
4720 | expanded when compiling functions for libgcc.a. It can be used to | |
2dd76960 | 4721 | provide alternate names for GCC's internal library functions if there |
d8088c6f | 4722 | are ABI-mandated names that the compiler should provide. |
a2c4f8e0 | 4723 | @end defmac |
d8088c6f | 4724 | |
c15c90bb ZW |
4725 | @findex init_one_libfunc |
4726 | @findex set_optab_libfunc | |
4727 | @deftypefn {Target Hook} void TARGET_INIT_LIBFUNCS (void) | |
4728 | This hook should declare additional library routines or rename | |
4729 | existing ones, using the functions @code{set_optab_libfunc} and | |
4730 | @code{init_one_libfunc} defined in @file{optabs.c}. | |
4731 | @code{init_optabs} calls this macro after initializing all the normal | |
4732 | library routines. | |
feca2ed3 | 4733 | |
c15c90bb ZW |
4734 | The default is to do nothing. Most ports don't need to define this hook. |
4735 | @end deftypefn | |
c5c60e15 | 4736 | |
9c917669 | 4737 | @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison}) |
c15c90bb ZW |
4738 | This macro should return @code{true} if the library routine that |
4739 | implements the floating point comparison operator @var{comparison} in | |
4740 | mode @var{mode} will return a boolean, and @var{false} if it will | |
4741 | return a tristate. | |
4742 | ||
4743 | GCC's own floating point libraries return tristates from the | |
4744 | comparison operators, so the default returns false always. Most ports | |
4745 | don't need to define this macro. | |
4746 | @end defmac | |
4747 | ||
b3f8d95d MM |
4748 | @defmac TARGET_LIB_INT_CMP_BIASED |
4749 | This macro should evaluate to @code{true} if the integer comparison | |
4750 | functions (like @code{__cmpdi2}) return 0 to indicate that the first | |
4751 | operand is smaller than the second, 1 to indicate that they are equal, | |
4752 | and 2 to indicate that the first operand is greater than the second. | |
38b974a6 | 4753 | If this macro evaluates to @code{false} the comparison functions return |
78466c0e | 4754 | @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines |
b3f8d95d MM |
4755 | in @file{libgcc.a}, you do not need to define this macro. |
4756 | @end defmac | |
4757 | ||
c15c90bb ZW |
4758 | @cindex US Software GOFAST, floating point emulation library |
4759 | @cindex floating point emulation library, US Software GOFAST | |
4760 | @cindex GOFAST, floating point emulation library | |
4761 | @findex gofast_maybe_init_libfuncs | |
4762 | @defmac US_SOFTWARE_GOFAST | |
4763 | Define this macro if your system C library uses the US Software GOFAST | |
73774972 | 4764 | library to provide floating point emulation. |
c15c90bb ZW |
4765 | |
4766 | In addition to defining this macro, your architecture must set | |
4767 | @code{TARGET_INIT_LIBFUNCS} to @code{gofast_maybe_init_libfuncs}, or | |
4768 | else call that function from its version of that hook. It is defined | |
4769 | in @file{config/gofast.h}, which must be included by your | |
4770 | architecture's @file{@var{cpu}.c} file. See @file{sparc/sparc.c} for | |
4771 | an example. | |
4772 | ||
4773 | If this macro is defined, the | |
4774 | @code{TARGET_FLOAT_LIB_COMPARE_RETURNS_BOOL} target hook must return | |
4775 | false for @code{SFmode} and @code{DFmode} comparisons. | |
a2c4f8e0 | 4776 | @end defmac |
c5c60e15 | 4777 | |
feca2ed3 | 4778 | @cindex @code{EDOM}, implicit usage |
a2c4f8e0 ZW |
4779 | @findex matherr |
4780 | @defmac TARGET_EDOM | |
feca2ed3 | 4781 | The value of @code{EDOM} on the target machine, as a C integer constant |
a3a15b4d | 4782 | expression. If you don't define this macro, GCC does not attempt to |
feca2ed3 JW |
4783 | deposit the value of @code{EDOM} into @code{errno} directly. Look in |
4784 | @file{/usr/include/errno.h} to find the value of @code{EDOM} on your | |
4785 | system. | |
4786 | ||
4787 | If you do not define @code{TARGET_EDOM}, then compiled code reports | |
4788 | domain errors by calling the library function and letting it report the | |
4789 | error. If mathematical functions on your system use @code{matherr} when | |
4790 | there is an error, then you should leave @code{TARGET_EDOM} undefined so | |
4791 | that @code{matherr} is used normally. | |
a2c4f8e0 | 4792 | @end defmac |
feca2ed3 | 4793 | |
feca2ed3 | 4794 | @cindex @code{errno}, implicit usage |
a2c4f8e0 | 4795 | @defmac GEN_ERRNO_RTX |
feca2ed3 JW |
4796 | Define this macro as a C expression to create an rtl expression that |
4797 | refers to the global ``variable'' @code{errno}. (On certain systems, | |
4798 | @code{errno} may not actually be a variable.) If you don't define this | |
4799 | macro, a reasonable default is used. | |
a2c4f8e0 | 4800 | @end defmac |
feca2ed3 | 4801 | |
272f51a3 | 4802 | @cindex C99 math functions, implicit usage |
a2c4f8e0 | 4803 | @defmac TARGET_C99_FUNCTIONS |
272f51a3 | 4804 | When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into |
3bcf1b13 | 4805 | @code{sinf} and similarly for other functions defined by C99 standard. The |
272f51a3 JH |
4806 | default is nonzero that should be proper value for most modern systems, however |
4807 | number of existing systems lacks support for these functions in the runtime so | |
4808 | they needs this macro to be redefined to 0. | |
a2c4f8e0 | 4809 | @end defmac |
272f51a3 | 4810 | |
a2c4f8e0 | 4811 | @defmac NEXT_OBJC_RUNTIME |
2147b154 | 4812 | Define this macro to generate code for Objective-C message sending using |
feca2ed3 JW |
4813 | the calling convention of the NeXT system. This calling convention |
4814 | involves passing the object, the selector and the method arguments all | |
4815 | at once to the method-lookup library function. | |
4816 | ||
4817 | The default calling convention passes just the object and the selector | |
4818 | to the lookup function, which returns a pointer to the method. | |
a2c4f8e0 | 4819 | @end defmac |
feca2ed3 JW |
4820 | |
4821 | @node Addressing Modes | |
4822 | @section Addressing Modes | |
4823 | @cindex addressing modes | |
4824 | ||
4825 | @c prevent bad page break with this line | |
4826 | This is about addressing modes. | |
4827 | ||
a2c4f8e0 ZW |
4828 | @defmac HAVE_PRE_INCREMENT |
4829 | @defmacx HAVE_PRE_DECREMENT | |
4830 | @defmacx HAVE_POST_INCREMENT | |
4831 | @defmacx HAVE_POST_DECREMENT | |
df2a54e9 | 4832 | A C expression that is nonzero if the machine supports pre-increment, |
7a6bd5ae | 4833 | pre-decrement, post-increment, or post-decrement addressing respectively. |
a2c4f8e0 | 4834 | @end defmac |
feca2ed3 | 4835 | |
a2c4f8e0 ZW |
4836 | @defmac HAVE_PRE_MODIFY_DISP |
4837 | @defmacx HAVE_POST_MODIFY_DISP | |
df2a54e9 | 4838 | A C expression that is nonzero if the machine supports pre- or |
7a6bd5ae JL |
4839 | post-address side-effect generation involving constants other than |
4840 | the size of the memory operand. | |
a2c4f8e0 | 4841 | @end defmac |
864bcaa7 | 4842 | |
a2c4f8e0 ZW |
4843 | @defmac HAVE_PRE_MODIFY_REG |
4844 | @defmacx HAVE_POST_MODIFY_REG | |
df2a54e9 | 4845 | A C expression that is nonzero if the machine supports pre- or |
7a6bd5ae | 4846 | post-address side-effect generation involving a register displacement. |
a2c4f8e0 | 4847 | @end defmac |
864bcaa7 | 4848 | |
a2c4f8e0 | 4849 | @defmac CONSTANT_ADDRESS_P (@var{x}) |
feca2ed3 JW |
4850 | A C expression that is 1 if the RTX @var{x} is a constant which |
4851 | is a valid address. On most machines, this can be defined as | |
4852 | @code{CONSTANT_P (@var{x})}, but a few machines are more restrictive | |
4853 | in which constant addresses are supported. | |
a2c4f8e0 | 4854 | @end defmac |
feca2ed3 | 4855 | |
a2c4f8e0 ZW |
4856 | @defmac CONSTANT_P (@var{x}) |
4857 | @code{CONSTANT_P}, which is defined by target-independent code, | |
4858 | accepts integer-values expressions whose values are not explicitly | |
4859 | known, such as @code{symbol_ref}, @code{label_ref}, and @code{high} | |
4860 | expressions and @code{const} arithmetic expressions, in addition to | |
4861 | @code{const_int} and @code{const_double} expressions. | |
4862 | @end defmac | |
feca2ed3 | 4863 | |
a2c4f8e0 | 4864 | @defmac MAX_REGS_PER_ADDRESS |
feca2ed3 JW |
4865 | A number, the maximum number of registers that can appear in a valid |
4866 | memory address. Note that it is up to you to specify a value equal to | |
4867 | the maximum number that @code{GO_IF_LEGITIMATE_ADDRESS} would ever | |
4868 | accept. | |
a2c4f8e0 | 4869 | @end defmac |
feca2ed3 | 4870 | |
a2c4f8e0 | 4871 | @defmac GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label}) |
feca2ed3 JW |
4872 | A C compound statement with a conditional @code{goto @var{label};} |
4873 | executed if @var{x} (an RTX) is a legitimate memory address on the | |
4874 | target machine for a memory operand of mode @var{mode}. | |
4875 | ||
4876 | It usually pays to define several simpler macros to serve as | |
4877 | subroutines for this one. Otherwise it may be too complicated to | |
4878 | understand. | |
4879 | ||
4880 | This macro must exist in two variants: a strict variant and a | |
4881 | non-strict one. The strict variant is used in the reload pass. It | |
4882 | must be defined so that any pseudo-register that has not been | |
4883 | allocated a hard register is considered a memory reference. In | |
4884 | contexts where some kind of register is required, a pseudo-register | |
4885 | with no hard register must be rejected. | |
4886 | ||
4887 | The non-strict variant is used in other passes. It must be defined to | |
4888 | accept all pseudo-registers in every context where some kind of | |
4889 | register is required. | |
4890 | ||
4891 | @findex REG_OK_STRICT | |
4892 | Compiler source files that want to use the strict variant of this | |
4893 | macro define the macro @code{REG_OK_STRICT}. You should use an | |
4894 | @code{#ifdef REG_OK_STRICT} conditional to define the strict variant | |
4895 | in that case and the non-strict variant otherwise. | |
4896 | ||
4897 | Subroutines to check for acceptable registers for various purposes (one | |
4898 | for base registers, one for index registers, and so on) are typically | |
4899 | among the subroutines used to define @code{GO_IF_LEGITIMATE_ADDRESS}. | |
4900 | Then only these subroutine macros need have two variants; the higher | |
bd819a4a | 4901 | levels of macros may be the same whether strict or not. |
feca2ed3 JW |
4902 | |
4903 | Normally, constant addresses which are the sum of a @code{symbol_ref} | |
4904 | and an integer are stored inside a @code{const} RTX to mark them as | |
4905 | constant. Therefore, there is no need to recognize such sums | |
4906 | specifically as legitimate addresses. Normally you would simply | |
4907 | recognize any @code{const} as legitimate. | |
4908 | ||
4909 | Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant | |
4910 | sums that are not marked with @code{const}. It assumes that a naked | |
4911 | @code{plus} indicates indexing. If so, then you @emph{must} reject such | |
4912 | naked constant sums as illegitimate addresses, so that none of them will | |
4913 | be given to @code{PRINT_OPERAND_ADDRESS}. | |
4914 | ||
fb49053f | 4915 | @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation |
feca2ed3 JW |
4916 | On some machines, whether a symbolic address is legitimate depends on |
4917 | the section that the address refers to. On these machines, define the | |
fb49053f RH |
4918 | target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information |
4919 | into the @code{symbol_ref}, and then check for it here. When you see a | |
feca2ed3 JW |
4920 | @code{const}, you will have to look inside it to find the |
4921 | @code{symbol_ref} in order to determine the section. @xref{Assembler | |
4922 | Format}. | |
a2c4f8e0 | 4923 | @end defmac |
feca2ed3 | 4924 | |
a2c4f8e0 | 4925 | @defmac REG_OK_FOR_BASE_P (@var{x}) |
feca2ed3 JW |
4926 | A C expression that is nonzero if @var{x} (assumed to be a @code{reg} |
4927 | RTX) is valid for use as a base register. For hard registers, it | |
4928 | should always accept those which the hardware permits and reject the | |
4929 | others. Whether the macro accepts or rejects pseudo registers must be | |
4930 | controlled by @code{REG_OK_STRICT} as described above. This usually | |
4931 | requires two variant definitions, of which @code{REG_OK_STRICT} | |
4932 | controls the one actually used. | |
a2c4f8e0 | 4933 | @end defmac |
feca2ed3 | 4934 | |
a2c4f8e0 | 4935 | @defmac REG_MODE_OK_FOR_BASE_P (@var{x}, @var{mode}) |
861bb6c1 JL |
4936 | A C expression that is just like @code{REG_OK_FOR_BASE_P}, except that |
4937 | that expression may examine the mode of the memory reference in | |
4938 | @var{mode}. You should define this macro if the mode of the memory | |
4939 | reference affects whether a register may be used as a base register. If | |
4940 | you define this macro, the compiler will use it instead of | |
4941 | @code{REG_OK_FOR_BASE_P}. | |
a2c4f8e0 | 4942 | @end defmac |
861bb6c1 | 4943 | |
888d2cd6 DJ |
4944 | @defmac REG_MODE_OK_FOR_REG_BASE_P (@var{x}, @var{mode}) |
4945 | A C expression which is nonzero if @var{x} (assumed to be a @code{reg} RTX) | |
4946 | is suitable for use as a base register in base plus index operand addresses, | |
4947 | accessing memory in mode @var{mode}. It may be either a suitable hard | |
f0eb93a8 | 4948 | register or a pseudo register that has been allocated such a hard register. |
888d2cd6 DJ |
4949 | You should define this macro if base plus index addresses have different |
4950 | requirements than other base register uses. | |
4951 | @end defmac | |
4952 | ||
a2c4f8e0 | 4953 | @defmac REG_OK_FOR_INDEX_P (@var{x}) |
feca2ed3 JW |
4954 | A C expression that is nonzero if @var{x} (assumed to be a @code{reg} |
4955 | RTX) is valid for use as an index register. | |
4956 | ||
4957 | The difference between an index register and a base register is that | |
4958 | the index register may be scaled. If an address involves the sum of | |
4959 | two registers, neither one of them scaled, then either one may be | |
4960 | labeled the ``base'' and the other the ``index''; but whichever | |
4961 | labeling is used must fit the machine's constraints of which registers | |
4962 | may serve in each capacity. The compiler will try both labelings, | |
4963 | looking for one that is valid, and will reload one or both registers | |
4964 | only if neither labeling works. | |
a2c4f8e0 | 4965 | @end defmac |
feca2ed3 | 4966 | |
a2c4f8e0 | 4967 | @defmac FIND_BASE_TERM (@var{x}) |
b949ea8b JW |
4968 | A C expression to determine the base term of address @var{x}. |
4969 | This macro is used in only one place: `find_base_term' in alias.c. | |
4970 | ||
4971 | It is always safe for this macro to not be defined. It exists so | |
4972 | that alias analysis can understand machine-dependent addresses. | |
4973 | ||
4974 | The typical use of this macro is to handle addresses containing | |
161d7b59 | 4975 | a label_ref or symbol_ref within an UNSPEC@. |
a2c4f8e0 | 4976 | @end defmac |
b949ea8b | 4977 | |
a2c4f8e0 | 4978 | @defmac LEGITIMIZE_ADDRESS (@var{x}, @var{oldx}, @var{mode}, @var{win}) |
feca2ed3 JW |
4979 | A C compound statement that attempts to replace @var{x} with a valid |
4980 | memory address for an operand of mode @var{mode}. @var{win} will be a | |
4981 | C statement label elsewhere in the code; the macro definition may use | |
4982 | ||
3ab51846 | 4983 | @smallexample |
feca2ed3 | 4984 | GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{win}); |
3ab51846 | 4985 | @end smallexample |
feca2ed3 JW |
4986 | |
4987 | @noindent | |
4988 | to avoid further processing if the address has become legitimate. | |
4989 | ||
4990 | @findex break_out_memory_refs | |
4991 | @var{x} will always be the result of a call to @code{break_out_memory_refs}, | |
4992 | and @var{oldx} will be the operand that was given to that function to produce | |
4993 | @var{x}. | |
4994 | ||
4995 | The code generated by this macro should not alter the substructure of | |
4996 | @var{x}. If it transforms @var{x} into a more legitimate form, it | |
4997 | should assign @var{x} (which will always be a C variable) a new value. | |
4998 | ||
4999 | It is not necessary for this macro to come up with a legitimate | |
5000 | address. The compiler has standard ways of doing so in all cases. In | |
3e759eda | 5001 | fact, it is safe to omit this macro. But often a |
feca2ed3 | 5002 | machine-dependent strategy can generate better code. |
a2c4f8e0 | 5003 | @end defmac |
feca2ed3 | 5004 | |
a2c4f8e0 | 5005 | @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win}) |
a9a2595b JR |
5006 | A C compound statement that attempts to replace @var{x}, which is an address |
5007 | that needs reloading, with a valid memory address for an operand of mode | |
5008 | @var{mode}. @var{win} will be a C statement label elsewhere in the code. | |
5009 | It is not necessary to define this macro, but it might be useful for | |
02f52e19 | 5010 | performance reasons. |
a9a2595b JR |
5011 | |
5012 | For example, on the i386, it is sometimes possible to use a single | |
5013 | reload register instead of two by reloading a sum of two pseudo | |
5014 | registers into a register. On the other hand, for number of RISC | |
5015 | processors offsets are limited so that often an intermediate address | |
5016 | needs to be generated in order to address a stack slot. By defining | |
aee96fe9 | 5017 | @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses |
a9a2595b JR |
5018 | generated for adjacent some stack slots can be made identical, and thus |
5019 | be shared. | |
5020 | ||
39bdfaa0 RH |
5021 | @emph{Note}: This macro should be used with caution. It is necessary |
5022 | to know something of how reload works in order to effectively use this, | |
5023 | and it is quite easy to produce macros that build in too much knowledge | |
5024 | of reload internals. | |
a9a2595b | 5025 | |
5f0c590d JL |
5026 | @emph{Note}: This macro must be able to reload an address created by a |
5027 | previous invocation of this macro. If it fails to handle such addresses | |
5028 | then the compiler may generate incorrect code or abort. | |
5029 | ||
a9a2595b | 5030 | @findex push_reload |
39bdfaa0 RH |
5031 | The macro definition should use @code{push_reload} to indicate parts that |
5032 | need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually | |
5033 | suitable to be passed unaltered to @code{push_reload}. | |
a9a2595b | 5034 | |
39bdfaa0 | 5035 | The code generated by this macro must not alter the substructure of |
a9a2595b JR |
5036 | @var{x}. If it transforms @var{x} into a more legitimate form, it |
5037 | should assign @var{x} (which will always be a C variable) a new value. | |
5038 | This also applies to parts that you change indirectly by calling | |
5039 | @code{push_reload}. | |
5040 | ||
39bdfaa0 RH |
5041 | @findex strict_memory_address_p |
5042 | The macro definition may use @code{strict_memory_address_p} to test if | |
5043 | the address has become legitimate. | |
5044 | ||
a9a2595b JR |
5045 | @findex copy_rtx |
5046 | If you want to change only a part of @var{x}, one standard way of doing | |
5047 | this is to use @code{copy_rtx}. Note, however, that is unshares only a | |
5048 | single level of rtl. Thus, if the part to be changed is not at the | |
c771326b | 5049 | top level, you'll need to replace first the top level. |
a9a2595b JR |
5050 | It is not necessary for this macro to come up with a legitimate |
5051 | address; but often a machine-dependent strategy can generate better code. | |
a2c4f8e0 | 5052 | @end defmac |
a9a2595b | 5053 | |
a2c4f8e0 | 5054 | @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label}) |
feca2ed3 JW |
5055 | A C statement or compound statement with a conditional @code{goto |
5056 | @var{label};} executed if memory address @var{x} (an RTX) can have | |
5057 | different meanings depending on the machine mode of the memory | |
5058 | reference it is used for or if the address is valid for some modes | |
5059 | but not others. | |
5060 | ||
5061 | Autoincrement and autodecrement addresses typically have mode-dependent | |
5062 | effects because the amount of the increment or decrement is the size | |
5063 | of the operand being addressed. Some machines have other mode-dependent | |
5064 | addresses. Many RISC machines have no mode-dependent addresses. | |
5065 | ||
5066 | You may assume that @var{addr} is a valid address for the machine. | |
a2c4f8e0 | 5067 | @end defmac |
feca2ed3 | 5068 | |
a2c4f8e0 | 5069 | @defmac LEGITIMATE_CONSTANT_P (@var{x}) |
feca2ed3 JW |
5070 | A C expression that is nonzero if @var{x} is a legitimate constant for |
5071 | an immediate operand on the target machine. You can assume that | |
5072 | @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact, | |
5073 | @samp{1} is a suitable definition for this macro on machines where | |
bd819a4a | 5074 | anything @code{CONSTANT_P} is valid. |
a2c4f8e0 | 5075 | @end defmac |
feca2ed3 | 5076 | |
73f8783a RS |
5077 | @deftypefn {Target Hook} rtx TARGET_DELEGITIMIZE_ADDRESS (rtx @var{x}) |
5078 | This hook is used to undo the possibly obfuscating effects of the | |
5079 | @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target | |
5080 | macros. Some backend implementations of these macros wrap symbol | |
5081 | references inside an @code{UNSPEC} rtx to represent PIC or similar | |
5082 | addressing modes. This target hook allows GCC's optimizers to understand | |
5083 | the semantics of these opaque @code{UNSPEC}s by converting them back | |
5084 | into their original form. | |
5085 | @end deftypefn | |
5086 | ||
d3da4d14 RH |
5087 | @deftypefn {Target Hook} bool TARGET_CANNOT_FORCE_CONST_MEM (rtx @var{x}) |
5088 | This hook should return true if @var{x} is of a form that cannot (or | |
5089 | should not) be spilled to the constant pool. The default version of | |
5090 | this hook returns false. | |
5091 | ||
5092 | The primary reason to define this hook is to prevent reload from | |
5093 | deciding that a non-legitimate constant would be better reloaded | |
5094 | from the constant pool instead of spilling and reloading a register | |
5095 | holding the constant. This restriction is often true of addresses | |
5096 | of TLS symbols for various targets. | |
5097 | @end deftypefn | |
5098 | ||
d16b59fa DN |
5099 | @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD (void) |
5100 | This hook should return the DECL of a function @var{f} that given an | |
5101 | address @var{addr} as an argument returns a mask @var{m} that can be | |
5102 | used to extract from two vectors the relevant data that resides in | |
5103 | @var{addr} in case @var{addr} is not properly aligned. | |
5104 | ||
5105 | The autovectrizer, when vectorizing a load operation from an address | |
5106 | @var{addr} that may be unaligned, will generate two vector loads from | |
5107 | the two aligned addresses around @var{addr}. It then generates a | |
5108 | @code{REALIGN_LOAD} operation to extract the relevant data from the | |
5109 | two loaded vectors. The first two arguments to @code{REALIGN_LOAD}, | |
5110 | @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and | |
5111 | the third argument, @var{OFF}, defines how the data will be extracted | |
5112 | from these two vectors: if @var{OFF} is 0, then the returned vector is | |
5113 | @var{v2}; otherwise, the returned vector is composed from the last | |
5114 | @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first | |
5115 | @var{OFF} elements of @var{v2}. | |
5116 | ||
5117 | If this hook is defined, the autovectorizer will generate a call | |
5118 | to @var{f} (using the DECL tree that this hook returns) and will | |
5119 | use the return value of @var{f} as the argument @var{OFF} to | |
5120 | @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f} | |
5121 | should comply with the semantics expected by @code{REALIGN_LOAD} | |
5122 | described above. | |
5123 | If this hook is not defined, then @var{addr} will be used as | |
5124 | the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low | |
5125 | log2(@var{VS})-1 bits of @var{addr} will be considered. | |
5126 | @end deftypefn | |
5127 | ||
feca2ed3 JW |
5128 | @node Condition Code |
5129 | @section Condition Code Status | |
5130 | @cindex condition code status | |
5131 | ||
5132 | @c prevent bad page break with this line | |
5133 | This describes the condition code status. | |
5134 | ||
5135 | @findex cc_status | |
5136 | The file @file{conditions.h} defines a variable @code{cc_status} to | |
5137 | describe how the condition code was computed (in case the interpretation of | |
5138 | the condition code depends on the instruction that it was set by). This | |
5139 | variable contains the RTL expressions on which the condition code is | |
5140 | currently based, and several standard flags. | |
5141 | ||
5142 | Sometimes additional machine-specific flags must be defined in the machine | |
5143 | description header file. It can also add additional machine-specific | |
5144 | information by defining @code{CC_STATUS_MDEP}. | |
5145 | ||
a2c4f8e0 | 5146 | @defmac CC_STATUS_MDEP |
feca2ed3 JW |
5147 | C code for a data type which is used for declaring the @code{mdep} |
5148 | component of @code{cc_status}. It defaults to @code{int}. | |
5149 | ||
5150 | This macro is not used on machines that do not use @code{cc0}. | |
a2c4f8e0 | 5151 | @end defmac |
feca2ed3 | 5152 | |
a2c4f8e0 | 5153 | @defmac CC_STATUS_MDEP_INIT |
feca2ed3 JW |
5154 | A C expression to initialize the @code{mdep} field to ``empty''. |
5155 | The default definition does nothing, since most machines don't use | |
5156 | the field anyway. If you want to use the field, you should probably | |
5157 | define this macro to initialize it. | |
5158 | ||
5159 | This macro is not used on machines that do not use @code{cc0}. | |
a2c4f8e0 | 5160 | @end defmac |
feca2ed3 | 5161 | |
a2c4f8e0 | 5162 | @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn}) |
feca2ed3 JW |
5163 | A C compound statement to set the components of @code{cc_status} |
5164 | appropriately for an insn @var{insn} whose body is @var{exp}. It is | |
5165 | this macro's responsibility to recognize insns that set the condition | |
5166 | code as a byproduct of other activity as well as those that explicitly | |
5167 | set @code{(cc0)}. | |
5168 | ||
5169 | This macro is not used on machines that do not use @code{cc0}. | |
5170 | ||
5171 | If there are insns that do not set the condition code but do alter | |
5172 | other machine registers, this macro must check to see whether they | |
5173 | invalidate the expressions that the condition code is recorded as | |
5174 | reflecting. For example, on the 68000, insns that store in address | |
5175 | registers do not set the condition code, which means that usually | |
5176 | @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such | |
5177 | insns. But suppose that the previous insn set the condition code | |
5178 | based on location @samp{a4@@(102)} and the current insn stores a new | |
5179 | value in @samp{a4}. Although the condition code is not changed by | |
5180 | this, it will no longer be true that it reflects the contents of | |
5181 | @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter | |
5182 | @code{cc_status} in this case to say that nothing is known about the | |
5183 | condition code value. | |
5184 | ||
5185 | The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal | |
5186 | with the results of peephole optimization: insns whose patterns are | |
5187 | @code{parallel} RTXs containing various @code{reg}, @code{mem} or | |
5188 | constants which are just the operands. The RTL structure of these | |
5189 | insns is not sufficient to indicate what the insns actually do. What | |
5190 | @code{NOTICE_UPDATE_CC} should do when it sees one is just to run | |
5191 | @code{CC_STATUS_INIT}. | |
5192 | ||
5193 | A possible definition of @code{NOTICE_UPDATE_CC} is to call a function | |
5194 | that looks at an attribute (@pxref{Insn Attributes}) named, for example, | |
5195 | @samp{cc}. This avoids having detailed information about patterns in | |
5196 | two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}. | |
a2c4f8e0 | 5197 | @end defmac |
feca2ed3 | 5198 | |
a2c4f8e0 | 5199 | @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y}) |
feca2ed3 JW |
5200 | Returns a mode from class @code{MODE_CC} to be used when comparison |
5201 | operation code @var{op} is applied to rtx @var{x} and @var{y}. For | |
981f6289 | 5202 | example, on the SPARC, @code{SELECT_CC_MODE} is defined as (see |
feca2ed3 JW |
5203 | @pxref{Jump Patterns} for a description of the reason for this |
5204 | definition) | |
5205 | ||
5206 | @smallexample | |
5207 | #define SELECT_CC_MODE(OP,X,Y) \ | |
5208 | (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \ | |
5209 | ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \ | |
5210 | : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \ | |
5211 | || GET_CODE (X) == NEG) \ | |
5212 | ? CC_NOOVmode : CCmode)) | |
5213 | @end smallexample | |
5214 | ||
94134f42 ZW |
5215 | You should define this macro if and only if you define extra CC modes |
5216 | in @file{@var{machine}-modes.def}. | |
a2c4f8e0 | 5217 | @end defmac |
feca2ed3 | 5218 | |
a2c4f8e0 | 5219 | @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1}) |
8760eaae | 5220 | On some machines not all possible comparisons are defined, but you can |
feca2ed3 JW |
5221 | convert an invalid comparison into a valid one. For example, the Alpha |
5222 | does not have a @code{GT} comparison, but you can use an @code{LT} | |
5223 | comparison instead and swap the order of the operands. | |
5224 | ||
5225 | On such machines, define this macro to be a C statement to do any | |
5226 | required conversions. @var{code} is the initial comparison code | |
5227 | and @var{op0} and @var{op1} are the left and right operands of the | |
5228 | comparison, respectively. You should modify @var{code}, @var{op0}, and | |
5229 | @var{op1} as required. | |
5230 | ||
a3a15b4d | 5231 | GCC will not assume that the comparison resulting from this macro is |
feca2ed3 JW |
5232 | valid but will see if the resulting insn matches a pattern in the |
5233 | @file{md} file. | |
5234 | ||
5235 | You need not define this macro if it would never change the comparison | |
5236 | code or operands. | |
a2c4f8e0 | 5237 | @end defmac |
feca2ed3 | 5238 | |
a2c4f8e0 | 5239 | @defmac REVERSIBLE_CC_MODE (@var{mode}) |
feca2ed3 JW |
5240 | A C expression whose value is one if it is always safe to reverse a |
5241 | comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE} | |
5242 | can ever return @var{mode} for a floating-point inequality comparison, | |
5243 | then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero. | |
5244 | ||
5245 | You need not define this macro if it would always returns zero or if the | |
5246 | floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}. | |
981f6289 | 5247 | For example, here is the definition used on the SPARC, where floating-point |
feca2ed3 JW |
5248 | inequality comparisons are always given @code{CCFPEmode}: |
5249 | ||
5250 | @smallexample | |
5251 | #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode) | |
5252 | @end smallexample | |
a2c4f8e0 | 5253 | @end defmac |
feca2ed3 | 5254 | |
a2c4f8e0 | 5255 | @defmac REVERSE_CONDITION (@var{code}, @var{mode}) |
9e7adcb3 JH |
5256 | A C expression whose value is reversed condition code of the @var{code} for |
5257 | comparison done in CC_MODE @var{mode}. The macro is used only in case | |
5258 | @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case | |
5259 | machine has some non-standard way how to reverse certain conditionals. For | |
5260 | instance in case all floating point conditions are non-trapping, compiler may | |
5261 | freely convert unordered compares to ordered one. Then definition may look | |
5262 | like: | |
5263 | ||
5264 | @smallexample | |
5265 | #define REVERSE_CONDITION(CODE, MODE) \ | |
c771326b | 5266 | ((MODE) != CCFPmode ? reverse_condition (CODE) \ |
9e7adcb3 JH |
5267 | : reverse_condition_maybe_unordered (CODE)) |
5268 | @end smallexample | |
a2c4f8e0 | 5269 | @end defmac |
9e7adcb3 | 5270 | |
15dce812 | 5271 | @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{op1}, @var{op2}) |
7e6d8ba1 | 5272 | A C expression that returns true if the conditional execution predicate |
15dce812 RE |
5273 | @var{op1}, a comparison operation, is the inverse of @var{op2} and vice |
5274 | versa. Define this to return 0 if the target has conditional execution | |
5275 | predicates that cannot be reversed safely. There is no need to validate | |
5276 | that the arguments of op1 and op2 are the same, this is done separately. | |
5277 | If no expansion is specified, this macro is defined as follows: | |
7e6d8ba1 AH |
5278 | |
5279 | @smallexample | |
aee96fe9 | 5280 | #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \ |
15dce812 | 5281 | (GET_CODE ((x)) == reversed_comparison_code ((y), NULL)) |
7e6d8ba1 | 5282 | @end smallexample |
a2c4f8e0 | 5283 | @end defmac |
feca2ed3 | 5284 | |
e129d93a ILT |
5285 | @deftypefn {Target Hook} bool TARGET_FIXED_CONDITION_CODE_REGS (unsigned int *, unsigned int *) |
5286 | On targets which do not use @code{(cc0)}, and which use a hard | |
5287 | register rather than a pseudo-register to hold condition codes, the | |
5288 | regular CSE passes are often not able to identify cases in which the | |
5289 | hard register is set to a common value. Use this hook to enable a | |
5290 | small pass which optimizes such cases. This hook should return true | |
5291 | to enable this pass, and it should set the integers to which its | |
5292 | arguments point to the hard register numbers used for condition codes. | |
5293 | When there is only one such register, as is true on most systems, the | |
5294 | integer pointed to by the second argument should be set to | |
5295 | @code{INVALID_REGNUM}. | |
5296 | ||
5297 | The default version of this hook returns false. | |
5298 | @end deftypefn | |
5299 | ||
5300 | @deftypefn {Target Hook} enum machine_mode TARGET_CC_MODES_COMPATIBLE (enum machine_mode, enum machine_mode) | |
5301 | On targets which use multiple condition code modes in class | |
5302 | @code{MODE_CC}, it is sometimes the case that a comparison can be | |
5303 | validly done in more than one mode. On such a system, define this | |
5304 | target hook to take two mode arguments and to return a mode in which | |
5305 | both comparisons may be validly done. If there is no such mode, | |
5306 | return @code{VOIDmode}. | |
5307 | ||
5308 | The default version of this hook checks whether the modes are the | |
5309 | same. If they are, it returns that mode. If they are different, it | |
5310 | returns @code{VOIDmode}. | |
5311 | @end deftypefn | |
5312 | ||
feca2ed3 JW |
5313 | @node Costs |
5314 | @section Describing Relative Costs of Operations | |
5315 | @cindex costs of instructions | |
5316 | @cindex relative costs | |
5317 | @cindex speed of instructions | |
5318 | ||
5319 | These macros let you describe the relative speed of various operations | |
5320 | on the target machine. | |
5321 | ||
a2c4f8e0 | 5322 | @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to}) |
e56b4594 AO |
5323 | A C expression for the cost of moving data of mode @var{mode} from a |
5324 | register in class @var{from} to one in class @var{to}. The classes are | |
5325 | expressed using the enumeration values such as @code{GENERAL_REGS}. A | |
5326 | value of 2 is the default; other values are interpreted relative to | |
5327 | that. | |
feca2ed3 JW |
5328 | |
5329 | It is not required that the cost always equal 2 when @var{from} is the | |
5330 | same as @var{to}; on some machines it is expensive to move between | |
5331 | registers if they are not general registers. | |
5332 | ||
5333 | If reload sees an insn consisting of a single @code{set} between two | |
5334 | hard registers, and if @code{REGISTER_MOVE_COST} applied to their | |
5335 | classes returns a value of 2, reload does not check to ensure that the | |
5336 | constraints of the insn are met. Setting a cost of other than 2 will | |
5337 | allow reload to verify that the constraints are met. You should do this | |
5338 | if the @samp{mov@var{m}} pattern's constraints do not allow such copying. | |
a2c4f8e0 | 5339 | @end defmac |
feca2ed3 | 5340 | |
a2c4f8e0 | 5341 | @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in}) |
cbd5b9a2 KR |
5342 | A C expression for the cost of moving data of mode @var{mode} between a |
5343 | register of class @var{class} and memory; @var{in} is zero if the value | |
df2a54e9 | 5344 | is to be written to memory, nonzero if it is to be read in. This cost |
473fe49b KR |
5345 | is relative to those in @code{REGISTER_MOVE_COST}. If moving between |
5346 | registers and memory is more expensive than between two registers, you | |
5347 | should define this macro to express the relative cost. | |
5348 | ||
a3a15b4d | 5349 | If you do not define this macro, GCC uses a default cost of 4 plus |
38e01259 | 5350 | the cost of copying via a secondary reload register, if one is |
473fe49b KR |
5351 | needed. If your machine requires a secondary reload register to copy |
5352 | between memory and a register of @var{class} but the reload mechanism is | |
5353 | more complex than copying via an intermediate, define this macro to | |
5354 | reflect the actual cost of the move. | |
5355 | ||
a3a15b4d | 5356 | GCC defines the function @code{memory_move_secondary_cost} if |
473fe49b KR |
5357 | secondary reloads are needed. It computes the costs due to copying via |
5358 | a secondary register. If your machine copies from memory using a | |
5359 | secondary register in the conventional way but the default base value of | |
5360 | 4 is not correct for your machine, define this macro to add some other | |
5361 | value to the result of that function. The arguments to that function | |
5362 | are the same as to this macro. | |
a2c4f8e0 | 5363 | @end defmac |
cbd5b9a2 | 5364 | |
a2c4f8e0 | 5365 | @defmac BRANCH_COST |
feca2ed3 JW |
5366 | A C expression for the cost of a branch instruction. A value of 1 is |
5367 | the default; other values are interpreted relative to that. | |
a2c4f8e0 | 5368 | @end defmac |
feca2ed3 JW |
5369 | |
5370 | Here are additional macros which do not specify precise relative costs, | |
a3a15b4d | 5371 | but only that certain actions are more expensive than GCC would |
feca2ed3 JW |
5372 | ordinarily expect. |
5373 | ||
a2c4f8e0 | 5374 | @defmac SLOW_BYTE_ACCESS |
feca2ed3 | 5375 | Define this macro as a C expression which is nonzero if accessing less |
e979f9e8 | 5376 | than a word of memory (i.e.@: a @code{char} or a @code{short}) is no |
feca2ed3 JW |
5377 | faster than accessing a word of memory, i.e., if such access |
5378 | require more than one instruction or if there is no difference in cost | |
5379 | between byte and (aligned) word loads. | |
5380 | ||
5381 | When this macro is not defined, the compiler will access a field by | |
5382 | finding the smallest containing object; when it is defined, a fullword | |
5383 | load will be used if alignment permits. Unless bytes accesses are | |
5384 | faster than word accesses, using word accesses is preferable since it | |
5385 | may eliminate subsequent memory access if subsequent accesses occur to | |
5386 | other fields in the same word of the structure, but to different bytes. | |
a2c4f8e0 | 5387 | @end defmac |
feca2ed3 | 5388 | |
a2c4f8e0 | 5389 | @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment}) |
5fad8ebf DE |
5390 | Define this macro to be the value 1 if memory accesses described by the |
5391 | @var{mode} and @var{alignment} parameters have a cost many times greater | |
5392 | than aligned accesses, for example if they are emulated in a trap | |
5393 | handler. | |
feca2ed3 | 5394 | |
df2a54e9 JM |
5395 | When this macro is nonzero, the compiler will act as if |
5396 | @code{STRICT_ALIGNMENT} were nonzero when generating code for block | |
feca2ed3 | 5397 | moves. This can cause significantly more instructions to be produced. |
df2a54e9 | 5398 | Therefore, do not set this macro nonzero if unaligned accesses only add a |
feca2ed3 JW |
5399 | cycle or two to the time for a memory access. |
5400 | ||
6be57663 | 5401 | If the value of this macro is always zero, it need not be defined. If |
df2a54e9 JM |
5402 | this macro is defined, it should produce a nonzero value when |
5403 | @code{STRICT_ALIGNMENT} is nonzero. | |
a2c4f8e0 | 5404 | @end defmac |
feca2ed3 | 5405 | |
a2c4f8e0 | 5406 | @defmac MOVE_RATIO |
9862dea9 | 5407 | The threshold of number of scalar memory-to-memory move insns, @emph{below} |
c5c76735 | 5408 | which a sequence of insns should be generated instead of a |
feca2ed3 JW |
5409 | string move insn or a library call. Increasing the value will always |
5410 | make code faster, but eventually incurs high cost in increased code size. | |
5411 | ||
c5c76735 JL |
5412 | Note that on machines where the corresponding move insn is a |
5413 | @code{define_expand} that emits a sequence of insns, this macro counts | |
5414 | the number of such sequences. | |
9862dea9 | 5415 | |
feca2ed3 | 5416 | If you don't define this, a reasonable default is used. |
a2c4f8e0 | 5417 | @end defmac |
feca2ed3 | 5418 | |
a2c4f8e0 | 5419 | @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment}) |
fbe1758d AM |
5420 | A C expression used to determine whether @code{move_by_pieces} will be used to |
5421 | copy a chunk of memory, or whether some other block move mechanism | |
6e01bd94 | 5422 | will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less |
fbe1758d | 5423 | than @code{MOVE_RATIO}. |
a2c4f8e0 | 5424 | @end defmac |
fbe1758d | 5425 | |
a2c4f8e0 | 5426 | @defmac MOVE_MAX_PIECES |
fbe1758d | 5427 | A C expression used by @code{move_by_pieces} to determine the largest unit |
6e01bd94 | 5428 | a load or store used to copy memory is. Defaults to @code{MOVE_MAX}. |
a2c4f8e0 | 5429 | @end defmac |
fbe1758d | 5430 | |
a2c4f8e0 | 5431 | @defmac CLEAR_RATIO |
78762e3b RS |
5432 | The threshold of number of scalar move insns, @emph{below} which a sequence |
5433 | of insns should be generated to clear memory instead of a string clear insn | |
5434 | or a library call. Increasing the value will always make code faster, but | |
5435 | eventually incurs high cost in increased code size. | |
5436 | ||
5437 | If you don't define this, a reasonable default is used. | |
a2c4f8e0 | 5438 | @end defmac |
78762e3b | 5439 | |
a2c4f8e0 | 5440 | @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment}) |
78762e3b RS |
5441 | A C expression used to determine whether @code{clear_by_pieces} will be used |
5442 | to clear a chunk of memory, or whether some other block clear mechanism | |
5443 | will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less | |
5444 | than @code{CLEAR_RATIO}. | |
a2c4f8e0 | 5445 | @end defmac |
78762e3b | 5446 | |
a2c4f8e0 | 5447 | @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment}) |
4977bab6 ZW |
5448 | A C expression used to determine whether @code{store_by_pieces} will be |
5449 | used to set a chunk of memory to a constant value, or whether some other | |
5450 | mechanism will be used. Used by @code{__builtin_memset} when storing | |
5451 | values other than constant zero and by @code{__builtin_strcpy} when | |
5452 | when called with a constant source string. | |
0bdcd332 | 5453 | Defaults to 1 if @code{move_by_pieces_ninsns} returns less |
45d78e7f | 5454 | than @code{MOVE_RATIO}. |
a2c4f8e0 | 5455 | @end defmac |
4977bab6 | 5456 | |
a2c4f8e0 | 5457 | @defmac USE_LOAD_POST_INCREMENT (@var{mode}) |
6e01bd94 MH |
5458 | A C expression used to determine whether a load postincrement is a good |
5459 | thing to use for a given mode. Defaults to the value of | |
5460 | @code{HAVE_POST_INCREMENT}. | |
a2c4f8e0 | 5461 | @end defmac |
6e01bd94 | 5462 | |
a2c4f8e0 | 5463 | @defmac USE_LOAD_POST_DECREMENT (@var{mode}) |
6e01bd94 MH |
5464 | A C expression used to determine whether a load postdecrement is a good |
5465 | thing to use for a given mode. Defaults to the value of | |
5466 | @code{HAVE_POST_DECREMENT}. | |
a2c4f8e0 | 5467 | @end defmac |
fbe1758d | 5468 | |
a2c4f8e0 | 5469 | @defmac USE_LOAD_PRE_INCREMENT (@var{mode}) |
6e01bd94 MH |
5470 | A C expression used to determine whether a load preincrement is a good |
5471 | thing to use for a given mode. Defaults to the value of | |
5472 | @code{HAVE_PRE_INCREMENT}. | |
a2c4f8e0 | 5473 | @end defmac |
6e01bd94 | 5474 | |
a2c4f8e0 | 5475 | @defmac USE_LOAD_PRE_DECREMENT (@var{mode}) |
6e01bd94 MH |
5476 | A C expression used to determine whether a load predecrement is a good |
5477 | thing to use for a given mode. Defaults to the value of | |
5478 | @code{HAVE_PRE_DECREMENT}. | |
a2c4f8e0 | 5479 | @end defmac |
fbe1758d | 5480 | |
a2c4f8e0 | 5481 | @defmac USE_STORE_POST_INCREMENT (@var{mode}) |
6e01bd94 MH |
5482 | A C expression used to determine whether a store postincrement is a good |
5483 | thing to use for a given mode. Defaults to the value of | |
5484 | @code{HAVE_POST_INCREMENT}. | |
a2c4f8e0 | 5485 | @end defmac |
6e01bd94 | 5486 | |
a2c4f8e0 | 5487 | @defmac USE_STORE_POST_DECREMENT (@var{mode}) |
c771326b | 5488 | A C expression used to determine whether a store postdecrement is a good |
6e01bd94 MH |
5489 | thing to use for a given mode. Defaults to the value of |
5490 | @code{HAVE_POST_DECREMENT}. | |
a2c4f8e0 | 5491 | @end defmac |
fbe1758d | 5492 | |
a2c4f8e0 | 5493 | @defmac USE_STORE_PRE_INCREMENT (@var{mode}) |
6e01bd94 MH |
5494 | This macro is used to determine whether a store preincrement is a good |
5495 | thing to use for a given mode. Defaults to the value of | |
5496 | @code{HAVE_PRE_INCREMENT}. | |
a2c4f8e0 | 5497 | @end defmac |
6e01bd94 | 5498 | |
a2c4f8e0 | 5499 | @defmac USE_STORE_PRE_DECREMENT (@var{mode}) |
6e01bd94 MH |
5500 | This macro is used to determine whether a store predecrement is a good |
5501 | thing to use for a given mode. Defaults to the value of | |
5502 | @code{HAVE_PRE_DECREMENT}. | |
a2c4f8e0 | 5503 | @end defmac |
fbe1758d | 5504 | |
a2c4f8e0 | 5505 | @defmac NO_FUNCTION_CSE |
feca2ed3 JW |
5506 | Define this macro if it is as good or better to call a constant |
5507 | function address than to call an address kept in a register. | |
a2c4f8e0 | 5508 | @end defmac |
feca2ed3 | 5509 | |
a2c4f8e0 | 5510 | @defmac RANGE_TEST_NON_SHORT_CIRCUIT |
85e50b6b DE |
5511 | Define this macro if a non-short-circuit operation produced by |
5512 | @samp{fold_range_test ()} is optimal. This macro defaults to true if | |
5513 | @code{BRANCH_COST} is greater than or equal to the value 2. | |
a2c4f8e0 | 5514 | @end defmac |
feca2ed3 | 5515 | |
3c50106f RH |
5516 | @deftypefn {Target Hook} bool TARGET_RTX_COSTS (rtx @var{x}, int @var{code}, int @var{outer_code}, int *@var{total}) |
5517 | This target hook describes the relative costs of RTL expressions. | |
5518 | ||
5519 | The cost may depend on the precise form of the expression, which is | |
5520 | available for examination in @var{x}, and the rtx code of the expression | |
5521 | in which it is contained, found in @var{outer_code}. @var{code} is the | |
5522 | expression code---redundant, since it can be obtained with | |
5523 | @code{GET_CODE (@var{x})}. | |
5524 | ||
5525 | In implementing this hook, you can use the construct | |
5526 | @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast | |
5527 | instructions. | |
5528 | ||
5529 | On entry to the hook, @code{*@var{total}} contains a default estimate | |
5530 | for the cost of the expression. The hook should modify this value as | |
31a52b86 RS |
5531 | necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)} |
5532 | for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus | |
5533 | operations, and @code{COSTS_N_INSNS (1)} for all other operations. | |
5534 | ||
8a36672b | 5535 | When optimizing for code size, i.e.@: when @code{optimize_size} is |
1a141fe1 | 5536 | nonzero, this target hook should be used to estimate the relative |
31a52b86 | 5537 | size cost of an expression, again relative to @code{COSTS_N_INSNS}. |
3c50106f RH |
5538 | |
5539 | The hook returns true when all subexpressions of @var{x} have been | |
5540 | processed, and false when @code{rtx_cost} should recurse. | |
5541 | @end deftypefn | |
5542 | ||
10154ff8 RH |
5543 | @deftypefn {Target Hook} int TARGET_ADDRESS_COST (rtx @var{address}) |
5544 | This hook computes the cost of an addressing mode that contains | |
5545 | @var{address}. If not defined, the cost is computed from | |
5546 | the @var{address} expression and the @code{TARGET_RTX_COST} hook. | |
5547 | ||
5548 | For most CISC machines, the default cost is a good approximation of the | |
5549 | true cost of the addressing mode. However, on RISC machines, all | |
5550 | instructions normally have the same length and execution time. Hence | |
5551 | all addresses will have equal costs. | |
5552 | ||
5553 | In cases where more than one form of an address is known, the form with | |
5554 | the lowest cost will be used. If multiple forms have the same, lowest, | |
5555 | cost, the one that is the most complex will be used. | |
5556 | ||
5557 | For example, suppose an address that is equal to the sum of a register | |
5558 | and a constant is used twice in the same basic block. When this macro | |
5559 | is not defined, the address will be computed in a register and memory | |
5560 | references will be indirect through that register. On machines where | |
5561 | the cost of the addressing mode containing the sum is no higher than | |
5562 | that of a simple indirect reference, this will produce an additional | |
5563 | instruction and possibly require an additional register. Proper | |
5564 | specification of this macro eliminates this overhead for such machines. | |
5565 | ||
5566 | This hook is never called with an invalid address. | |
5567 | ||
5568 | On machines where an address involving more than one register is as | |
5569 | cheap as an address computation involving only one register, defining | |
5570 | @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to | |
5571 | be live over a region of code where only one would have been if | |
5572 | @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect | |
5573 | should be considered in the definition of this macro. Equivalent costs | |
5574 | should probably only be given to addresses with different numbers of | |
5575 | registers on machines with lots of registers. | |
5576 | @end deftypefn | |
5577 | ||
c237e94a ZW |
5578 | @node Scheduling |
5579 | @section Adjusting the Instruction Scheduler | |
5580 | ||
5581 | The instruction scheduler may need a fair amount of machine-specific | |
5582 | adjustment in order to produce good code. GCC provides several target | |
5583 | hooks for this purpose. It is usually enough to define just a few of | |
5584 | them: try the first ones in this list first. | |
5585 | ||
5586 | @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void) | |
fae15c93 VM |
5587 | This hook returns the maximum number of instructions that can ever |
5588 | issue at the same time on the target machine. The default is one. | |
5589 | Although the insn scheduler can define itself the possibility of issue | |
5590 | an insn on the same cycle, the value can serve as an additional | |
5591 | constraint to issue insns on the same simulated processor cycle (see | |
5592 | hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}). | |
5593 | This value must be constant over the entire compilation. If you need | |
5594 | it to vary depending on what the instructions are, you must use | |
c237e94a | 5595 | @samp{TARGET_SCHED_VARIABLE_ISSUE}. |
fae15c93 | 5596 | |
fa0aee89 PB |
5597 | You could define this hook to return the value of the macro |
5598 | @code{MAX_DFA_ISSUE_RATE}. | |
c237e94a ZW |
5599 | @end deftypefn |
5600 | ||
5601 | @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more}) | |
5602 | This hook is executed by the scheduler after it has scheduled an insn | |
5603 | from the ready list. It should return the number of insns which can | |
3ee04299 DE |
5604 | still be issued in the current cycle. The default is |
5605 | @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and | |
5606 | @code{USE}, which normally are not counted against the issue rate. | |
5607 | You should define this hook if some insns take more machine resources | |
5608 | than others, so that fewer insns can follow them in the same cycle. | |
5609 | @var{file} is either a null pointer, or a stdio stream to write any | |
5610 | debug output to. @var{verbose} is the verbose level provided by | |
5611 | @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that | |
5612 | was scheduled. | |
c237e94a ZW |
5613 | @end deftypefn |
5614 | ||
5615 | @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost}) | |
fae15c93 VM |
5616 | This function corrects the value of @var{cost} based on the |
5617 | relationship between @var{insn} and @var{dep_insn} through the | |
5618 | dependence @var{link}. It should return the new value. The default | |
5619 | is to make no adjustment to @var{cost}. This can be used for example | |
5620 | to specify to the scheduler using the traditional pipeline description | |
c237e94a | 5621 | that an output- or anti-dependence does not incur the same cost as a |
fae15c93 VM |
5622 | data-dependence. If the scheduler using the automaton based pipeline |
5623 | description, the cost of anti-dependence is zero and the cost of | |
5624 | output-dependence is maximum of one and the difference of latency | |
5625 | times of the first and the second insns. If these values are not | |
5626 | acceptable, you could use the hook to modify them too. See also | |
fa0aee89 | 5627 | @pxref{Processor pipeline description}. |
c237e94a ZW |
5628 | @end deftypefn |
5629 | ||
5630 | @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority}) | |
5631 | This hook adjusts the integer scheduling priority @var{priority} of | |
5632 | @var{insn}. It should return the new priority. Reduce the priority to | |
5633 | execute @var{insn} earlier, increase the priority to execute @var{insn} | |
5634 | later. Do not define this hook if you do not need to adjust the | |
5635 | scheduling priorities of insns. | |
5636 | @end deftypefn | |
5637 | ||
5638 | @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock}) | |
5639 | This hook is executed by the scheduler after it has scheduled the ready | |
5640 | list, to allow the machine description to reorder it (for example to | |
5641 | combine two small instructions together on @samp{VLIW} machines). | |
5642 | @var{file} is either a null pointer, or a stdio stream to write any | |
5643 | debug output to. @var{verbose} is the verbose level provided by | |
5644 | @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready | |
5645 | list of instructions that are ready to be scheduled. @var{n_readyp} is | |
5646 | a pointer to the number of elements in the ready list. The scheduler | |
5647 | reads the ready list in reverse order, starting with | |
5648 | @var{ready}[@var{*n_readyp}-1] and going to @var{ready}[0]. @var{clock} | |
5649 | is the timer tick of the scheduler. You may modify the ready list and | |
5650 | the number of ready insns. The return value is the number of insns that | |
5651 | can issue this cycle; normally this is just @code{issue_rate}. See also | |
5652 | @samp{TARGET_SCHED_REORDER2}. | |
5653 | @end deftypefn | |
5654 | ||
5655 | @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_ready}, @var{clock}) | |
5656 | Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That | |
5657 | function is called whenever the scheduler starts a new cycle. This one | |
5658 | is called once per iteration over a cycle, immediately after | |
5659 | @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and | |
5660 | return the number of insns to be scheduled in the same cycle. Defining | |
5661 | this hook can be useful if there are frequent situations where | |
5662 | scheduling one insn causes other insns to become ready in the same | |
5663 | cycle. These other insns can then be taken into account properly. | |
5664 | @end deftypefn | |
5665 | ||
30028c85 VM |
5666 | @deftypefn {Target Hook} void TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK (rtx @var{head}, rtx @var{tail}) |
5667 | This hook is called after evaluation forward dependencies of insns in | |
5668 | chain given by two parameter values (@var{head} and @var{tail} | |
5669 | correspondingly) but before insns scheduling of the insn chain. For | |
5670 | example, it can be used for better insn classification if it requires | |
5671 | analysis of dependencies. This hook can use backward and forward | |
5672 | dependencies of the insn scheduler because they are already | |
5673 | calculated. | |
5674 | @end deftypefn | |
5675 | ||
c237e94a ZW |
5676 | @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready}) |
5677 | This hook is executed by the scheduler at the beginning of each block of | |
5678 | instructions that are to be scheduled. @var{file} is either a null | |
5679 | pointer, or a stdio stream to write any debug output to. @var{verbose} | |
5680 | is the verbose level provided by @option{-fsched-verbose-@var{n}}. | |
5681 | @var{max_ready} is the maximum number of insns in the current scheduling | |
5682 | region that can be live at the same time. This can be used to allocate | |
431ae0bf | 5683 | scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}. |
c237e94a ZW |
5684 | @end deftypefn |
5685 | ||
5686 | @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose}) | |
5687 | This hook is executed by the scheduler at the end of each block of | |
5688 | instructions that are to be scheduled. It can be used to perform | |
5689 | cleanup of any actions done by the other scheduling hooks. @var{file} | |
5690 | is either a null pointer, or a stdio stream to write any debug output | |
5691 | to. @var{verbose} is the verbose level provided by | |
5692 | @option{-fsched-verbose-@var{n}}. | |
5693 | @end deftypefn | |
5694 | ||
58565a33 SKG |
5695 | @deftypefn {Target Hook} void TARGET_SCHED_INIT_GLOBAL (FILE *@var{file}, int @var{verbose}, int @var{old_max_uid}) |
5696 | This hook is executed by the scheduler after function level initializations. | |
5697 | @var{file} is either a null pointer, or a stdio stream to write any debug output to. | |
5698 | @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}. | |
5699 | @var{old_max_uid} is the maximum insn uid when scheduling begins. | |
5700 | @end deftypefn | |
5701 | ||
5702 | @deftypefn {Target Hook} void TARGET_SCHED_FINISH_GLOBAL (FILE *@var{file}, int @var{verbose}) | |
8a36672b | 5703 | This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}. |
58565a33 SKG |
5704 | @var{file} is either a null pointer, or a stdio stream to write any debug output to. |
5705 | @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}. | |
5706 | @end deftypefn | |
5707 | ||
fae15c93 VM |
5708 | @deftypefn {Target Hook} int TARGET_SCHED_DFA_PRE_CYCLE_INSN (void) |
5709 | The hook returns an RTL insn. The automaton state used in the | |
5710 | pipeline hazard recognizer is changed as if the insn were scheduled | |
5711 | when the new simulated processor cycle starts. Usage of the hook may | |
5712 | simplify the automaton pipeline description for some @acronym{VLIW} | |
5713 | processors. If the hook is defined, it is used only for the automaton | |
5714 | based pipeline description. The default is not to change the state | |
5715 | when the new simulated processor cycle starts. | |
5716 | @end deftypefn | |
5717 | ||
5718 | @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN (void) | |
5719 | The hook can be used to initialize data used by the previous hook. | |
5720 | @end deftypefn | |
5721 | ||
5722 | @deftypefn {Target Hook} int TARGET_SCHED_DFA_POST_CYCLE_INSN (void) | |
5723 | The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used | |
5724 | to changed the state as if the insn were scheduled when the new | |
5725 | simulated processor cycle finishes. | |
5726 | @end deftypefn | |
5727 | ||
5728 | @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN (void) | |
5729 | The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but | |
5730 | used to initialize data used by the previous hook. | |
5731 | @end deftypefn | |
5732 | ||
5733 | @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD (void) | |
5734 | This hook controls better choosing an insn from the ready insn queue | |
5735 | for the @acronym{DFA}-based insn scheduler. Usually the scheduler | |
5736 | chooses the first insn from the queue. If the hook returns a positive | |
5737 | value, an additional scheduler code tries all permutations of | |
5738 | @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()} | |
5739 | subsequent ready insns to choose an insn whose issue will result in | |
5740 | maximal number of issued insns on the same cycle. For the | |
5741 | @acronym{VLIW} processor, the code could actually solve the problem of | |
5742 | packing simple insns into the @acronym{VLIW} insn. Of course, if the | |
5743 | rules of @acronym{VLIW} packing are described in the automaton. | |
5744 | ||
5745 | This code also could be used for superscalar @acronym{RISC} | |
5746 | processors. Let us consider a superscalar @acronym{RISC} processor | |
5747 | with 3 pipelines. Some insns can be executed in pipelines @var{A} or | |
5748 | @var{B}, some insns can be executed only in pipelines @var{B} or | |
5749 | @var{C}, and one insn can be executed in pipeline @var{B}. The | |
5750 | processor may issue the 1st insn into @var{A} and the 2nd one into | |
5751 | @var{B}. In this case, the 3rd insn will wait for freeing @var{B} | |
5752 | until the next cycle. If the scheduler issues the 3rd insn the first, | |
5753 | the processor could issue all 3 insns per cycle. | |
5754 | ||
5755 | Actually this code demonstrates advantages of the automaton based | |
5756 | pipeline hazard recognizer. We try quickly and easy many insn | |
5757 | schedules to choose the best one. | |
5758 | ||
5759 | The default is no multipass scheduling. | |
5760 | @end deftypefn | |
5761 | ||
30028c85 VM |
5762 | @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD (rtx) |
5763 | ||
5764 | This hook controls what insns from the ready insn queue will be | |
5765 | considered for the multipass insn scheduling. If the hook returns | |
5766 | zero for insn passed as the parameter, the insn will be not chosen to | |
5767 | be issued. | |
5768 | ||
62b9c42c | 5769 | The default is that any ready insns can be chosen to be issued. |
30028c85 VM |
5770 | @end deftypefn |
5771 | ||
5772 | @deftypefn {Target Hook} int TARGET_SCHED_DFA_NEW_CYCLE (FILE *, int, rtx, int, int, int *) | |
5773 | ||
5774 | This hook is called by the insn scheduler before issuing insn passed | |
5775 | as the third parameter on given cycle. If the hook returns nonzero, | |
5776 | the insn is not issued on given processors cycle. Instead of that, | |
5777 | the processor cycle is advanced. If the value passed through the last | |
5778 | parameter is zero, the insn ready queue is not sorted on the new cycle | |
5779 | start as usually. The first parameter passes file for debugging | |
5780 | output. The second one passes the scheduler verbose level of the | |
5781 | debugging output. The forth and the fifth parameter values are | |
5782 | correspondingly processor cycle on which the previous insn has been | |
5783 | issued and the current processor cycle. | |
5784 | @end deftypefn | |
5785 | ||
1b2c3767 | 5786 | @deftypefn {Target Hook} bool TARGET_SCHED_IS_COSTLY_DEPENDENCE (rtx @var{insn1}, rtx @var{insn2}, rtx @var{dep_link}, int @var{dep_cost}, int @var{distance}) |
569fa502 | 5787 | This hook is used to define which dependences are considered costly by |
daf2f129 | 5788 | the target, so costly that it is not advisable to schedule the insns that |
569fa502 | 5789 | are involved in the dependence too close to one another. The parameters |
daf2f129 JM |
5790 | to this hook are as follows: The second parameter @var{insn2} is dependent |
5791 | upon the first parameter @var{insn1}. The dependence between @var{insn1} | |
5792 | and @var{insn2} is represented by the third parameter @var{dep_link}. The | |
5793 | fourth parameter @var{cost} is the cost of the dependence, and the fifth | |
5794 | parameter @var{distance} is the distance in cycles between the two insns. | |
569fa502 DN |
5795 | The hook returns @code{true} if considering the distance between the two |
5796 | insns the dependence between them is considered costly by the target, | |
5797 | and @code{false} otherwise. | |
5798 | ||
5799 | Defining this hook can be useful in multiple-issue out-of-order machines, | |
daf2f129 | 5800 | where (a) it's practically hopeless to predict the actual data/resource |
569fa502 | 5801 | delays, however: (b) there's a better chance to predict the actual grouping |
daf2f129 | 5802 | that will be formed, and (c) correctly emulating the grouping can be very |
569fa502 | 5803 | important. In such targets one may want to allow issuing dependent insns |
78466c0e | 5804 | closer to one another---i.e., closer than the dependence distance; however, |
569fa502 DN |
5805 | not in cases of "costly dependences", which this hooks allows to define. |
5806 | @end deftypefn | |
5807 | ||
fae15c93 VM |
5808 | Macros in the following table are generated by the program |
5809 | @file{genattr} and can be useful for writing the hooks. | |
5810 | ||
a2c4f8e0 | 5811 | @defmac MAX_DFA_ISSUE_RATE |
fae15c93 VM |
5812 | The macro definition is generated in the automaton based pipeline |
5813 | description interface. Its value is calculated from the automaton | |
5814 | based pipeline description and is equal to maximal number of all insns | |
5815 | described in constructions @samp{define_insn_reservation} which can be | |
5816 | issued on the same processor cycle. | |
a2c4f8e0 | 5817 | @end defmac |
fae15c93 | 5818 | |
feca2ed3 JW |
5819 | @node Sections |
5820 | @section Dividing the Output into Sections (Texts, Data, @dots{}) | |
5821 | @c the above section title is WAY too long. maybe cut the part between | |
5822 | @c the (...)? --mew 10feb93 | |
5823 | ||
5824 | An object file is divided into sections containing different types of | |
5825 | data. In the most common case, there are three sections: the @dfn{text | |
5826 | section}, which holds instructions and read-only data; the @dfn{data | |
5827 | section}, which holds initialized writable data; and the @dfn{bss | |
5828 | section}, which holds uninitialized data. Some systems have other kinds | |
5829 | of sections. | |
5830 | ||
5831 | The compiler must tell the assembler when to switch sections. These | |
5832 | macros control what commands to output to tell the assembler this. You | |
5833 | can also define additional sections. | |
5834 | ||
a2c4f8e0 | 5835 | @defmac TEXT_SECTION_ASM_OP |
047c1c92 HPN |
5836 | A C expression whose value is a string, including spacing, containing the |
5837 | assembler operation that should precede instructions and read-only data. | |
5838 | Normally @code{"\t.text"} is right. | |
a2c4f8e0 | 5839 | @end defmac |
33c09f2f | 5840 | |
a2c4f8e0 | 5841 | @defmac HOT_TEXT_SECTION_NAME |
194734e9 JH |
5842 | If defined, a C string constant for the name of the section containing most |
5843 | frequently executed functions of the program. If not defined, GCC will provide | |
5844 | a default definition if the target supports named sections. | |
a2c4f8e0 | 5845 | @end defmac |
194734e9 | 5846 | |
a2c4f8e0 | 5847 | @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME |
194734e9 JH |
5848 | If defined, a C string constant for the name of the section containing unlikely |
5849 | executed functions in the program. | |
a2c4f8e0 | 5850 | @end defmac |
194734e9 | 5851 | |
a2c4f8e0 | 5852 | @defmac DATA_SECTION_ASM_OP |
047c1c92 HPN |
5853 | A C expression whose value is a string, including spacing, containing the |
5854 | assembler operation to identify the following data as writable initialized | |
5855 | data. Normally @code{"\t.data"} is right. | |
a2c4f8e0 | 5856 | @end defmac |
feca2ed3 | 5857 | |
a2c4f8e0 | 5858 | @defmac READONLY_DATA_SECTION_ASM_OP |
d48bc59a RH |
5859 | A C expression whose value is a string, including spacing, containing the |
5860 | assembler operation to identify the following data as read-only initialized | |
5861 | data. | |
a2c4f8e0 | 5862 | @end defmac |
d48bc59a | 5863 | |
a2c4f8e0 | 5864 | @defmac READONLY_DATA_SECTION |
d48bc59a RH |
5865 | A macro naming a function to call to switch to the proper section for |
5866 | read-only data. The default is to use @code{READONLY_DATA_SECTION_ASM_OP} | |
5867 | if defined, else fall back to @code{text_section}. | |
5868 | ||
5869 | The most common definition will be @code{data_section}, if the target | |
5870 | does not have a special read-only data section, and does not put data | |
5871 | in the text section. | |
a2c4f8e0 | 5872 | @end defmac |
d48bc59a | 5873 | |
a2c4f8e0 | 5874 | @defmac BSS_SECTION_ASM_OP |
047c1c92 HPN |
5875 | If defined, a C expression whose value is a string, including spacing, |
5876 | containing the assembler operation to identify the following data as | |
5877 | uninitialized global data. If not defined, and neither | |
5878 | @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined, | |
5879 | uninitialized global data will be output in the data section if | |
630d3d5a | 5880 | @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be |
047c1c92 | 5881 | used. |
a2c4f8e0 | 5882 | @end defmac |
feca2ed3 | 5883 | |
a2c4f8e0 | 5884 | @defmac INIT_SECTION_ASM_OP |
047c1c92 HPN |
5885 | If defined, a C expression whose value is a string, including spacing, |
5886 | containing the assembler operation to identify the following data as | |
5887 | initialization code. If not defined, GCC will assume such a section does | |
5888 | not exist. | |
a2c4f8e0 | 5889 | @end defmac |
feca2ed3 | 5890 | |
a2c4f8e0 | 5891 | @defmac FINI_SECTION_ASM_OP |
047c1c92 HPN |
5892 | If defined, a C expression whose value is a string, including spacing, |
5893 | containing the assembler operation to identify the following data as | |
5894 | finalization code. If not defined, GCC will assume such a section does | |
5895 | not exist. | |
a2c4f8e0 | 5896 | @end defmac |
750054a2 | 5897 | |
7abc66b1 JB |
5898 | @defmac INIT_ARRAY_SECTION_ASM_OP |
5899 | If defined, a C expression whose value is a string, including spacing, | |
5900 | containing the assembler operation to identify the following data as | |
5901 | part of the @code{.init_array} (or equivalent) section. If not | |
5902 | defined, GCC will assume such a section does not exist. Do not define | |
5903 | both this macro and @code{INIT_SECTION_ASM_OP}. | |
5904 | @end defmac | |
5905 | ||
5906 | @defmac FINI_ARRAY_SECTION_ASM_OP | |
5907 | If defined, a C expression whose value is a string, including spacing, | |
5908 | containing the assembler operation to identify the following data as | |
5909 | part of the @code{.fini_array} (or equivalent) section. If not | |
5910 | defined, GCC will assume such a section does not exist. Do not define | |
5911 | both this macro and @code{FINI_SECTION_ASM_OP}. | |
5912 | @end defmac | |
5913 | ||
a2c4f8e0 | 5914 | @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function}) |
cea3bd3e RH |
5915 | If defined, an ASM statement that switches to a different section |
5916 | via @var{section_op}, calls @var{function}, and switches back to | |
5917 | the text section. This is used in @file{crtstuff.c} if | |
5918 | @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls | |
5919 | to initialization and finalization functions from the init and fini | |
5920 | sections. By default, this macro uses a simple function call. Some | |
1b2dd04a AO |
5921 | ports need hand-crafted assembly code to avoid dependencies on |
5922 | registers initialized in the function prologue or to ensure that | |
5923 | constant pools don't end up too far way in the text section. | |
a2c4f8e0 | 5924 | @end defmac |
1b2dd04a | 5925 | |
a2c4f8e0 | 5926 | @defmac FORCE_CODE_SECTION_ALIGN |
cea3bd3e RH |
5927 | If defined, an ASM statement that aligns a code section to some |
5928 | arbitrary boundary. This is used to force all fragments of the | |
5929 | @code{.init} and @code{.fini} sections to have to same alignment | |
5930 | and thus prevent the linker from having to add any padding. | |
a2c4f8e0 | 5931 | @end defmac |
cea3bd3e | 5932 | |
feca2ed3 JW |
5933 | @findex in_text |
5934 | @findex in_data | |
a2c4f8e0 | 5935 | @defmac EXTRA_SECTIONS |
feca2ed3 JW |
5936 | A list of names for sections other than the standard two, which are |
5937 | @code{in_text} and @code{in_data}. You need not define this macro | |
5938 | on a system with no other sections (that GCC needs to use). | |
a2c4f8e0 | 5939 | @end defmac |
feca2ed3 | 5940 | |
feca2ed3 JW |
5941 | @findex text_section |
5942 | @findex data_section | |
a2c4f8e0 | 5943 | @defmac EXTRA_SECTION_FUNCTIONS |
feca2ed3 JW |
5944 | One or more functions to be defined in @file{varasm.c}. These |
5945 | functions should do jobs analogous to those of @code{text_section} and | |
5946 | @code{data_section}, for your additional sections. Do not define this | |
5947 | macro if you do not define @code{EXTRA_SECTIONS}. | |
a2c4f8e0 | 5948 | @end defmac |
feca2ed3 | 5949 | |
a2c4f8e0 | 5950 | @defmac JUMP_TABLES_IN_TEXT_SECTION |
df2a54e9 | 5951 | Define this macro to be an expression with a nonzero value if jump |
75197b37 BS |
5952 | tables (for @code{tablejump} insns) should be output in the text |
5953 | section, along with the assembler instructions. Otherwise, the | |
5954 | readonly data section is used. | |
feca2ed3 JW |
5955 | |
5956 | This macro is irrelevant if there is no separate readonly data section. | |
a2c4f8e0 | 5957 | @end defmac |
feca2ed3 | 5958 | |
ae46c4e0 RH |
5959 | @deftypefn {Target Hook} void TARGET_ASM_SELECT_SECTION (tree @var{exp}, int @var{reloc}, unsigned HOST_WIDE_INT @var{align}) |
5960 | Switches to the appropriate section for output of @var{exp}. You can | |
5961 | assume that @var{exp} is either a @code{VAR_DECL} node or a constant of | |
5962 | some sort. @var{reloc} indicates whether the initial value of @var{exp} | |
5963 | requires link-time relocations. Bit 0 is set when variable contains | |
5964 | local relocations only, while bit 1 is set for global relocations. | |
5965 | Select the section by calling @code{data_section} or one of the | |
5966 | alternatives for other sections. @var{align} is the constant alignment | |
5967 | in bits. | |
5968 | ||
5969 | The default version of this function takes care of putting read-only | |
5970 | variables in @code{readonly_data_section}. | |
09afda70 GK |
5971 | |
5972 | See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}. | |
ae46c4e0 RH |
5973 | @end deftypefn |
5974 | ||
09afda70 GK |
5975 | @defmac USE_SELECT_SECTION_FOR_FUNCTIONS |
5976 | Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called | |
5977 | for @code{FUNCTION_DECL}s as well as for variables and constants. | |
5978 | ||
5979 | In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the | |
5980 | function has been determined to be likely to be called, and nonzero if | |
5981 | it is unlikely to be called. | |
5982 | @end defmac | |
5983 | ||
ae46c4e0 RH |
5984 | @deftypefn {Target Hook} void TARGET_ASM_UNIQUE_SECTION (tree @var{decl}, int @var{reloc}) |
5985 | Build up a unique section name, expressed as a @code{STRING_CST} node, | |
5986 | and assign it to @samp{DECL_SECTION_NAME (@var{decl})}. | |
5987 | As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether | |
5988 | the initial value of @var{exp} requires link-time relocations. | |
5989 | ||
5990 | The default version of this function appends the symbol name to the | |
5991 | ELF section name that would normally be used for the symbol. For | |
5992 | example, the function @code{foo} would be placed in @code{.text.foo}. | |
5993 | Whatever the actual target object format, this is often good enough. | |
5994 | @end deftypefn | |
5995 | ||
ab5c8549 JJ |
5996 | @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_RODATA_SECTION (tree @var{decl}) |
5997 | Switches to a readonly data section associated with | |
5998 | @samp{DECL_SECTION_NAME (@var{decl})}. | |
5999 | The default version of this function switches to @code{.gnu.linkonce.r.name} | |
6000 | section if function's section is @code{.gnu.linkonce.t.name}, to | |
6001 | @code{.rodata.name} if function is in @code{.text.name} section | |
6002 | and otherwise switches to the normal readonly data section. | |
6003 | @end deftypefn | |
6004 | ||
b64a1b53 RH |
6005 | @deftypefn {Target Hook} void TARGET_ASM_SELECT_RTX_SECTION (enum machine_mode @var{mode}, rtx @var{x}, unsigned HOST_WIDE_INT @var{align}) |
6006 | Switches to the appropriate section for output of constant pool entry | |
6007 | @var{x} in @var{mode}. You can assume that @var{x} is some kind of | |
6008 | constant in RTL@. The argument @var{mode} is redundant except in the | |
6009 | case of a @code{const_int} rtx. Select the section by calling | |
6010 | @code{readonly_data_section} or one of the alternatives for other | |
6011 | sections. @var{align} is the constant alignment in bits. | |
6012 | ||
6013 | The default version of this function takes care of putting symbolic | |
6014 | constants in @code{flag_pic} mode in @code{data_section} and everything | |
6015 | else in @code{readonly_data_section}. | |
6016 | @end deftypefn | |
6017 | ||
c6a2438a | 6018 | @deftypefn {Target Hook} void TARGET_ENCODE_SECTION_INFO (tree @var{decl}, rtx @var{rtl}, int @var{new_decl_p}) |
fb49053f RH |
6019 | Define this hook if references to a symbol or a constant must be |
6020 | treated differently depending on something about the variable or | |
6021 | function named by the symbol (such as what section it is in). | |
6022 | ||
c6a2438a ZW |
6023 | The hook is executed immediately after rtl has been created for |
6024 | @var{decl}, which may be a variable or function declaration or | |
6025 | an entry in the constant pool. In either case, @var{rtl} is the | |
6026 | rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})} | |
0864034e | 6027 | in this hook; that field may not have been initialized yet. |
c6a2438a ZW |
6028 | |
6029 | In the case of a constant, it is safe to assume that the rtl is | |
6030 | a @code{mem} whose address is a @code{symbol_ref}. Most decls | |
6031 | will also have this form, but that is not guaranteed. Global | |
6032 | register variables, for instance, will have a @code{reg} for their | |
6033 | rtl. (Normally the right thing to do with such unusual rtl is | |
6034 | leave it alone.) | |
fb49053f RH |
6035 | |
6036 | The @var{new_decl_p} argument will be true if this is the first time | |
c6a2438a | 6037 | that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will |
fb49053f RH |
6038 | be false for subsequent invocations, which will happen for duplicate |
6039 | declarations. Whether or not anything must be done for the duplicate | |
6040 | declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}. | |
c6a2438a | 6041 | @var{new_decl_p} is always true when the hook is called for a constant. |
fb49053f RH |
6042 | |
6043 | @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO} | |
c6a2438a ZW |
6044 | The usual thing for this hook to do is to record flags in the |
6045 | @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}. | |
6046 | Historically, the name string was modified if it was necessary to | |
6047 | encode more than one bit of information, but this practice is now | |
6048 | discouraged; use @code{SYMBOL_REF_FLAGS}. | |
6049 | ||
6050 | The default definition of this hook, @code{default_encode_section_info} | |
6051 | in @file{varasm.c}, sets a number of commonly-useful bits in | |
6052 | @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need | |
6053 | before overriding it. | |
fb49053f RH |
6054 | @end deftypefn |
6055 | ||
772c5265 RH |
6056 | @deftypefn {Target Hook} const char *TARGET_STRIP_NAME_ENCODING (const char *name) |
6057 | Decode @var{name} and return the real name part, sans | |
6058 | the characters that @code{TARGET_ENCODE_SECTION_INFO} | |
6059 | may have added. | |
6060 | @end deftypefn | |
6061 | ||
47754fd5 RH |
6062 | @deftypefn {Target Hook} bool TARGET_IN_SMALL_DATA_P (tree @var{exp}) |
6063 | Returns true if @var{exp} should be placed into a ``small data'' section. | |
6064 | The default version of this hook always returns false. | |
6065 | @end deftypefn | |
6066 | ||
e2a6476e DE |
6067 | @deftypevar {Target Hook} bool TARGET_HAVE_SRODATA_SECTION |
6068 | Contains the value true if the target places read-only | |
6069 | ``small data'' into a separate section. The default value is false. | |
6070 | @end deftypevar | |
6071 | ||
47754fd5 RH |
6072 | @deftypefn {Target Hook} bool TARGET_BINDS_LOCAL_P (tree @var{exp}) |
6073 | Returns true if @var{exp} names an object for which name resolution | |
6074 | rules must resolve to the current ``module'' (dynamic shared library | |
6075 | or executable image). | |
6076 | ||
6077 | The default version of this hook implements the name resolution rules | |
6078 | for ELF, which has a looser model of global name binding than other | |
6079 | currently supported object file formats. | |
6080 | @end deftypefn | |
6081 | ||
e2a6476e DE |
6082 | @deftypevar {Target Hook} bool TARGET_HAVE_TLS |
6083 | Contains the value true if the target supports thread-local storage. | |
6084 | The default value is false. | |
6085 | @end deftypevar | |
6086 | ||
6087 | ||
feca2ed3 JW |
6088 | @node PIC |
6089 | @section Position Independent Code | |
6090 | @cindex position independent code | |
6091 | @cindex PIC | |
6092 | ||
6093 | This section describes macros that help implement generation of position | |
6094 | independent code. Simply defining these macros is not enough to | |
6095 | generate valid PIC; you must also add support to the macros | |
6096 | @code{GO_IF_LEGITIMATE_ADDRESS} and @code{PRINT_OPERAND_ADDRESS}, as | |
6097 | well as @code{LEGITIMIZE_ADDRESS}. You must modify the definition of | |
6098 | @samp{movsi} to do something appropriate when the source operand | |
6099 | contains a symbolic address. You may also need to alter the handling of | |
6100 | switch statements so that they use relative addresses. | |
6101 | @c i rearranged the order of the macros above to try to force one of | |
6102 | @c them to the next line, to eliminate an overfull hbox. --mew 10feb93 | |
6103 | ||
a2c4f8e0 | 6104 | @defmac PIC_OFFSET_TABLE_REGNUM |
feca2ed3 JW |
6105 | The register number of the register used to address a table of static |
6106 | data addresses in memory. In some cases this register is defined by a | |
161d7b59 | 6107 | processor's ``application binary interface'' (ABI)@. When this macro |
feca2ed3 JW |
6108 | is defined, RTL is generated for this register once, as with the stack |
6109 | pointer and frame pointer registers. If this macro is not defined, it | |
6110 | is up to the machine-dependent files to allocate such a register (if | |
003b9f78 | 6111 | necessary). Note that this register must be fixed when in use (e.g.@: |
12beba6f | 6112 | when @code{flag_pic} is true). |
a2c4f8e0 | 6113 | @end defmac |
feca2ed3 | 6114 | |
a2c4f8e0 | 6115 | @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED |
feca2ed3 JW |
6116 | Define this macro if the register defined by |
6117 | @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define | |
ed4db1ee | 6118 | this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined. |
a2c4f8e0 | 6119 | @end defmac |
feca2ed3 | 6120 | |
a2c4f8e0 | 6121 | @defmac LEGITIMATE_PIC_OPERAND_P (@var{x}) |
feca2ed3 JW |
6122 | A C expression that is nonzero if @var{x} is a legitimate immediate |
6123 | operand on the target machine when generating position independent code. | |
6124 | You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not | |
6125 | check this. You can also assume @var{flag_pic} is true, so you need not | |
6126 | check it either. You need not define this macro if all constants | |
6127 | (including @code{SYMBOL_REF}) can be immediate operands when generating | |
6128 | position independent code. | |
a2c4f8e0 | 6129 | @end defmac |
feca2ed3 JW |
6130 | |
6131 | @node Assembler Format | |
6132 | @section Defining the Output Assembler Language | |
6133 | ||
6134 | This section describes macros whose principal purpose is to describe how | |
648c546a | 6135 | to write instructions in assembler language---rather than what the |
feca2ed3 JW |
6136 | instructions do. |
6137 | ||
6138 | @menu | |
6139 | * File Framework:: Structural information for the assembler file. | |
6140 | * Data Output:: Output of constants (numbers, strings, addresses). | |
6141 | * Uninitialized Data:: Output of uninitialized variables. | |
6142 | * Label Output:: Output and generation of labels. | |
6143 | * Initialization:: General principles of initialization | |
6144 | and termination routines. | |
6145 | * Macros for Initialization:: | |
6146 | Specific macros that control the handling of | |
6147 | initialization and termination routines. | |
6148 | * Instruction Output:: Output of actual instructions. | |
6149 | * Dispatch Tables:: Output of jump tables. | |
6150 | * Exception Region Output:: Output of exception region code. | |
6151 | * Alignment Output:: Pseudo ops for alignment and skipping data. | |
6152 | @end menu | |
6153 | ||
6154 | @node File Framework | |
6155 | @subsection The Overall Framework of an Assembler File | |
6156 | @cindex assembler format | |
6157 | @cindex output of assembler code | |
6158 | ||
6159 | @c prevent bad page break with this line | |
1bc7c5b6 ZW |
6160 | This describes the overall framework of an assembly file. |
6161 | ||
6162 | @deftypefn {Target Hook} void TARGET_ASM_FILE_START () | |
6163 | @findex default_file_start | |
6164 | Output to @code{asm_out_file} any text which the assembler expects to | |
6165 | find at the beginning of a file. The default behavior is controlled | |
6166 | by two flags, documented below. Unless your target's assembler is | |
6167 | quite unusual, if you override the default, you should call | |
6168 | @code{default_file_start} at some point in your target hook. This | |
6169 | lets other target files rely on these variables. | |
6170 | @end deftypefn | |
feca2ed3 | 6171 | |
1bc7c5b6 ZW |
6172 | @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_APP_OFF |
6173 | If this flag is true, the text of the macro @code{ASM_APP_OFF} will be | |
6174 | printed as the very first line in the assembly file, unless | |
6175 | @option{-fverbose-asm} is in effect. (If that macro has been defined | |
6176 | to the empty string, this variable has no effect.) With the normal | |
6177 | definition of @code{ASM_APP_OFF}, the effect is to notify the GNU | |
6178 | assembler that it need not bother stripping comments or extra | |
6179 | whitespace from its input. This allows it to work a bit faster. | |
6180 | ||
6181 | The default is false. You should not set it to true unless you have | |
6182 | verified that your port does not generate any extra whitespace or | |
6183 | comments that will cause GAS to issue errors in NO_APP mode. | |
6184 | @end deftypevr | |
6185 | ||
6186 | @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_FILE_DIRECTIVE | |
6187 | If this flag is true, @code{output_file_directive} will be called | |
6188 | for the primary source file, immediately after printing | |
6189 | @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect | |
6190 | this to be done. The default is false. | |
6191 | @end deftypevr | |
feca2ed3 | 6192 | |
a5fe455b ZW |
6193 | @deftypefn {Target Hook} void TARGET_ASM_FILE_END () |
6194 | Output to @code{asm_out_file} any text which the assembler expects | |
6195 | to find at the end of a file. The default is to output nothing. | |
6196 | @end deftypefn | |
feca2ed3 | 6197 | |
a5fe455b ZW |
6198 | @deftypefun void file_end_indicate_exec_stack () |
6199 | Some systems use a common convention, the @samp{.note.GNU-stack} | |
6200 | special section, to indicate whether or not an object file relies on | |
6201 | the stack being executable. If your system uses this convention, you | |
6202 | should define @code{TARGET_ASM_FILE_END} to this function. If you | |
6203 | need to do other things in that hook, have your hook function call | |
6204 | this function. | |
6205 | @end deftypefun | |
feca2ed3 | 6206 | |
a2c4f8e0 | 6207 | @defmac ASM_COMMENT_START |
feca2ed3 JW |
6208 | A C string constant describing how to begin a comment in the target |
6209 | assembler language. The compiler assumes that the comment will end at | |
6210 | the end of the line. | |
a2c4f8e0 | 6211 | @end defmac |
feca2ed3 | 6212 | |
a2c4f8e0 | 6213 | @defmac ASM_APP_ON |
feca2ed3 JW |
6214 | A C string constant for text to be output before each @code{asm} |
6215 | statement or group of consecutive ones. Normally this is | |
6216 | @code{"#APP"}, which is a comment that has no effect on most | |
6217 | assemblers but tells the GNU assembler that it must check the lines | |
6218 | that follow for all valid assembler constructs. | |
a2c4f8e0 | 6219 | @end defmac |
feca2ed3 | 6220 | |
a2c4f8e0 | 6221 | @defmac ASM_APP_OFF |
feca2ed3 JW |
6222 | A C string constant for text to be output after each @code{asm} |
6223 | statement or group of consecutive ones. Normally this is | |
6224 | @code{"#NO_APP"}, which tells the GNU assembler to resume making the | |
6225 | time-saving assumptions that are valid for ordinary compiler output. | |
a2c4f8e0 | 6226 | @end defmac |
feca2ed3 | 6227 | |
a2c4f8e0 | 6228 | @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name}) |
feca2ed3 JW |
6229 | A C statement to output COFF information or DWARF debugging information |
6230 | which indicates that filename @var{name} is the current source file to | |
6231 | the stdio stream @var{stream}. | |
6232 | ||
6233 | This macro need not be defined if the standard form of output | |
6234 | for the file format in use is appropriate. | |
a2c4f8e0 | 6235 | @end defmac |
feca2ed3 | 6236 | |
a2c4f8e0 | 6237 | @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string}) |
e9a25f70 JL |
6238 | A C statement to output the string @var{string} to the stdio stream |
6239 | @var{stream}. If you do not call the function @code{output_quoted_string} | |
a3a15b4d | 6240 | in your config files, GCC will only call it to output filenames to |
e9a25f70 JL |
6241 | the assembler source. So you can use it to canonicalize the format |
6242 | of the filename using this macro. | |
a2c4f8e0 | 6243 | @end defmac |
e9a25f70 | 6244 | |
a2c4f8e0 | 6245 | @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string}) |
feca2ed3 JW |
6246 | A C statement to output something to the assembler file to handle a |
6247 | @samp{#ident} directive containing the text @var{string}. If this | |
6248 | macro is not defined, nothing is output for a @samp{#ident} directive. | |
a2c4f8e0 | 6249 | @end defmac |
feca2ed3 | 6250 | |
7c262518 RH |
6251 | @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, unsigned int @var{align}) |
6252 | Output assembly directives to switch to section @var{name}. The section | |
6253 | should have attributes as specified by @var{flags}, which is a bit mask | |
6254 | of the @code{SECTION_*} flags defined in @file{output.h}. If @var{align} | |
df2a54e9 | 6255 | is nonzero, it contains an alignment in bytes to be used for the section, |
f282ffb3 | 6256 | otherwise some target default should be used. Only targets that must |
7c262518 RH |
6257 | specify an alignment within the section directive need pay attention to |
6258 | @var{align} -- we will still use @code{ASM_OUTPUT_ALIGN}. | |
6259 | @end deftypefn | |
6260 | ||
6261 | @deftypefn {Target Hook} bool TARGET_HAVE_NAMED_SECTIONS | |
6262 | This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}. | |
6263 | @end deftypefn | |
6264 | ||
6265 | @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc}) | |
6266 | Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION} | |
6267 | based on a variable or function decl, a section name, and whether or not the | |
6268 | declaration's initializer may contain runtime relocations. @var{decl} may be | |
6269 | null, in which case read-write data should be assumed. | |
6270 | ||
6271 | The default version if this function handles choosing code vs data, | |
6272 | read-only vs read-write data, and @code{flag_pic}. You should only | |
6273 | need to override this if your target has special flags that might be | |
6274 | set via @code{__attribute__}. | |
6275 | @end deftypefn | |
6276 | ||
feca2ed3 JW |
6277 | @need 2000 |
6278 | @node Data Output | |
6279 | @subsection Output of Data | |
6280 | ||
301d03af RS |
6281 | |
6282 | @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP | |
6283 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP | |
6284 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP | |
6285 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP | |
6286 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP | |
6287 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP | |
6288 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP | |
6289 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP | |
6290 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP | |
6291 | These hooks specify assembly directives for creating certain kinds | |
6292 | of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a | |
6293 | byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an | |
6294 | aligned two-byte object, and so on. Any of the hooks may be | |
6295 | @code{NULL}, indicating that no suitable directive is available. | |
6296 | ||
6297 | The compiler will print these strings at the start of a new line, | |
6298 | followed immediately by the object's initial value. In most cases, | |
6299 | the string should contain a tab, a pseudo-op, and then another tab. | |
6300 | @end deftypevr | |
6301 | ||
6302 | @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p}) | |
6303 | The @code{assemble_integer} function uses this hook to output an | |
6304 | integer object. @var{x} is the object's value, @var{size} is its size | |
6305 | in bytes and @var{aligned_p} indicates whether it is aligned. The | |
6306 | function should return @code{true} if it was able to output the | |
6307 | object. If it returns false, @code{assemble_integer} will try to | |
6308 | split the object into smaller parts. | |
6309 | ||
6310 | The default implementation of this hook will use the | |
6311 | @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false} | |
6312 | when the relevant string is @code{NULL}. | |
6313 | @end deftypefn | |
feca2ed3 | 6314 | |
a2c4f8e0 | 6315 | @defmac OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail}) |
422be3c3 AO |
6316 | A C statement to recognize @var{rtx} patterns that |
6317 | @code{output_addr_const} can't deal with, and output assembly code to | |
6318 | @var{stream} corresponding to the pattern @var{x}. This may be used to | |
6319 | allow machine-dependent @code{UNSPEC}s to appear within constants. | |
6320 | ||
6321 | If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must | |
6322 | @code{goto fail}, so that a standard error message is printed. If it | |
6323 | prints an error message itself, by calling, for example, | |
6324 | @code{output_operand_lossage}, it may just complete normally. | |
a2c4f8e0 | 6325 | @end defmac |
422be3c3 | 6326 | |
a2c4f8e0 | 6327 | @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len}) |
feca2ed3 JW |
6328 | A C statement to output to the stdio stream @var{stream} an assembler |
6329 | instruction to assemble a string constant containing the @var{len} | |
6330 | bytes at @var{ptr}. @var{ptr} will be a C expression of type | |
6331 | @code{char *} and @var{len} a C expression of type @code{int}. | |
6332 | ||
6333 | If the assembler has a @code{.ascii} pseudo-op as found in the | |
6334 | Berkeley Unix assembler, do not define the macro | |
6335 | @code{ASM_OUTPUT_ASCII}. | |
a2c4f8e0 | 6336 | @end defmac |
feca2ed3 | 6337 | |
a2c4f8e0 | 6338 | @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n}) |
67231816 RH |
6339 | A C statement to output word @var{n} of a function descriptor for |
6340 | @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS} | |
6341 | is defined, and is otherwise unused. | |
a2c4f8e0 | 6342 | @end defmac |
67231816 | 6343 | |
a2c4f8e0 | 6344 | @defmac CONSTANT_POOL_BEFORE_FUNCTION |
861bb6c1 | 6345 | You may define this macro as a C expression. You should define the |
df2a54e9 | 6346 | expression to have a nonzero value if GCC should output the constant |
861bb6c1 | 6347 | pool for a function before the code for the function, or a zero value if |
a3a15b4d JL |
6348 | GCC should output the constant pool after the function. If you do |
6349 | not define this macro, the usual case, GCC will output the constant | |
861bb6c1 | 6350 | pool before the function. |
a2c4f8e0 | 6351 | @end defmac |
861bb6c1 | 6352 | |
a2c4f8e0 | 6353 | @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size}) |
feca2ed3 JW |
6354 | A C statement to output assembler commands to define the start of the |
6355 | constant pool for a function. @var{funname} is a string giving | |
6356 | the name of the function. Should the return type of the function | |
6357 | be required, it can be obtained via @var{fundecl}. @var{size} | |
6358 | is the size, in bytes, of the constant pool that will be written | |
6359 | immediately after this call. | |
6360 | ||
6361 | If no constant-pool prefix is required, the usual case, this macro need | |
6362 | not be defined. | |
a2c4f8e0 | 6363 | @end defmac |
feca2ed3 | 6364 | |
a2c4f8e0 | 6365 | @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto}) |
feca2ed3 JW |
6366 | A C statement (with or without semicolon) to output a constant in the |
6367 | constant pool, if it needs special treatment. (This macro need not do | |
6368 | anything for RTL expressions that can be output normally.) | |
6369 | ||
6370 | The argument @var{file} is the standard I/O stream to output the | |
6371 | assembler code on. @var{x} is the RTL expression for the constant to | |
6372 | output, and @var{mode} is the machine mode (in case @var{x} is a | |
6373 | @samp{const_int}). @var{align} is the required alignment for the value | |
6374 | @var{x}; you should output an assembler directive to force this much | |
6375 | alignment. | |
6376 | ||
6377 | The argument @var{labelno} is a number to use in an internal label for | |
6378 | the address of this pool entry. The definition of this macro is | |
6379 | responsible for outputting the label definition at the proper place. | |
6380 | Here is how to do this: | |
6381 | ||
3ab51846 | 6382 | @smallexample |
4977bab6 | 6383 | @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno}); |
3ab51846 | 6384 | @end smallexample |
feca2ed3 JW |
6385 | |
6386 | When you output a pool entry specially, you should end with a | |
6387 | @code{goto} to the label @var{jumpto}. This will prevent the same pool | |
6388 | entry from being output a second time in the usual manner. | |
6389 | ||
6390 | You need not define this macro if it would do nothing. | |
a2c4f8e0 | 6391 | @end defmac |
feca2ed3 | 6392 | |
a2c4f8e0 | 6393 | @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size}) |
861bb6c1 JL |
6394 | A C statement to output assembler commands to at the end of the constant |
6395 | pool for a function. @var{funname} is a string giving the name of the | |
6396 | function. Should the return type of the function be required, you can | |
6397 | obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the | |
a3a15b4d | 6398 | constant pool that GCC wrote immediately before this call. |
861bb6c1 JL |
6399 | |
6400 | If no constant-pool epilogue is required, the usual case, you need not | |
6401 | define this macro. | |
a2c4f8e0 | 6402 | @end defmac |
861bb6c1 | 6403 | |
a2c4f8e0 | 6404 | @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}) |
feca2ed3 JW |
6405 | Define this macro as a C expression which is nonzero if @var{C} is |
6406 | used as a logical line separator by the assembler. | |
6407 | ||
6408 | If you do not define this macro, the default is that only | |
6409 | the character @samp{;} is treated as a logical line separator. | |
a2c4f8e0 | 6410 | @end defmac |
feca2ed3 | 6411 | |
8ca83838 | 6412 | @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN |
baed53ac | 6413 | @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN |
17b53c33 NB |
6414 | These target hooks are C string constants, describing the syntax in the |
6415 | assembler for grouping arithmetic expressions. If not overridden, they | |
6416 | default to normal parentheses, which is correct for most assemblers. | |
8ca83838 | 6417 | @end deftypevr |
17b53c33 | 6418 | |
feca2ed3 JW |
6419 | These macros are provided by @file{real.h} for writing the definitions |
6420 | of @code{ASM_OUTPUT_DOUBLE} and the like: | |
6421 | ||
a2c4f8e0 ZW |
6422 | @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l}) |
6423 | @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l}) | |
6424 | @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l}) | |
feca2ed3 | 6425 | These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the target's |
b216cd4a ZW |
6426 | floating point representation, and store its bit pattern in the variable |
6427 | @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE}, this variable should | |
6428 | be a simple @code{long int}. For the others, it should be an array of | |
6429 | @code{long int}. The number of elements in this array is determined by | |
6430 | the size of the desired target floating point data type: 32 bits of it | |
6431 | go in each @code{long int} array element. Each array element holds 32 | |
6432 | bits of the result, even if @code{long int} is wider than 32 bits on the | |
6433 | host machine. | |
feca2ed3 JW |
6434 | |
6435 | The array element values are designed so that you can print them out | |
6436 | using @code{fprintf} in the order they should appear in the target | |
6437 | machine's memory. | |
a2c4f8e0 | 6438 | @end defmac |
feca2ed3 JW |
6439 | |
6440 | @node Uninitialized Data | |
6441 | @subsection Output of Uninitialized Variables | |
6442 | ||
6443 | Each of the macros in this section is used to do the whole job of | |
6444 | outputting a single uninitialized variable. | |
6445 | ||
a2c4f8e0 | 6446 | @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded}) |
feca2ed3 JW |
6447 | A C statement (sans semicolon) to output to the stdio stream |
6448 | @var{stream} the assembler definition of a common-label named | |
6449 | @var{name} whose size is @var{size} bytes. The variable @var{rounded} | |
6450 | is the size rounded up to whatever alignment the caller wants. | |
6451 | ||
6452 | Use the expression @code{assemble_name (@var{stream}, @var{name})} to | |
6453 | output the name itself; before and after that, output the additional | |
6454 | assembler syntax for defining the name, and a newline. | |
6455 | ||
6456 | This macro controls how the assembler definitions of uninitialized | |
6457 | common global variables are output. | |
a2c4f8e0 | 6458 | @end defmac |
feca2ed3 | 6459 | |
a2c4f8e0 | 6460 | @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment}) |
feca2ed3 JW |
6461 | Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a |
6462 | separate, explicit argument. If you define this macro, it is used in | |
6463 | place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in | |
6464 | handling the required alignment of the variable. The alignment is specified | |
6465 | as the number of bits. | |
a2c4f8e0 | 6466 | @end defmac |
feca2ed3 | 6467 | |
a2c4f8e0 | 6468 | @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment}) |
e9a25f70 JL |
6469 | Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the |
6470 | variable to be output, if there is one, or @code{NULL_TREE} if there | |
8760eaae | 6471 | is no corresponding variable. If you define this macro, GCC will use it |
e9a25f70 JL |
6472 | in place of both @code{ASM_OUTPUT_COMMON} and |
6473 | @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see | |
6474 | the variable's decl in order to chose what to output. | |
a2c4f8e0 | 6475 | @end defmac |
e9a25f70 | 6476 | |
a2c4f8e0 | 6477 | @defmac ASM_OUTPUT_SHARED_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded}) |
feca2ed3 JW |
6478 | If defined, it is similar to @code{ASM_OUTPUT_COMMON}, except that it |
6479 | is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_COMMON} | |
6480 | will be used. | |
a2c4f8e0 | 6481 | @end defmac |
feca2ed3 | 6482 | |
a2c4f8e0 | 6483 | @defmac ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded}) |
feca2ed3 JW |
6484 | A C statement (sans semicolon) to output to the stdio stream |
6485 | @var{stream} the assembler definition of uninitialized global @var{decl} named | |
6486 | @var{name} whose size is @var{size} bytes. The variable @var{rounded} | |
6487 | is the size rounded up to whatever alignment the caller wants. | |
6488 | ||
6489 | Try to use function @code{asm_output_bss} defined in @file{varasm.c} when | |
6490 | defining this macro. If unable, use the expression | |
6491 | @code{assemble_name (@var{stream}, @var{name})} to output the name itself; | |
6492 | before and after that, output the additional assembler syntax for defining | |
6493 | the name, and a newline. | |
6494 | ||
6495 | This macro controls how the assembler definitions of uninitialized global | |
6496 | variables are output. This macro exists to properly support languages like | |
aee96fe9 | 6497 | C++ which do not have @code{common} data. However, this macro currently |
feca2ed3 JW |
6498 | is not defined for all targets. If this macro and |
6499 | @code{ASM_OUTPUT_ALIGNED_BSS} are not defined then @code{ASM_OUTPUT_COMMON} | |
e9a25f70 JL |
6500 | or @code{ASM_OUTPUT_ALIGNED_COMMON} or |
6501 | @code{ASM_OUTPUT_ALIGNED_DECL_COMMON} is used. | |
a2c4f8e0 | 6502 | @end defmac |
feca2ed3 | 6503 | |
a2c4f8e0 | 6504 | @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment}) |
feca2ed3 JW |
6505 | Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a |
6506 | separate, explicit argument. If you define this macro, it is used in | |
6507 | place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in | |
6508 | handling the required alignment of the variable. The alignment is specified | |
6509 | as the number of bits. | |
6510 | ||
6511 | Try to use function @code{asm_output_aligned_bss} defined in file | |
6512 | @file{varasm.c} when defining this macro. | |
a2c4f8e0 | 6513 | @end defmac |
feca2ed3 | 6514 | |
a2c4f8e0 | 6515 | @defmac ASM_OUTPUT_SHARED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded}) |
feca2ed3 JW |
6516 | If defined, it is similar to @code{ASM_OUTPUT_BSS}, except that it |
6517 | is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_BSS} | |
6518 | will be used. | |
a2c4f8e0 | 6519 | @end defmac |
feca2ed3 | 6520 | |
a2c4f8e0 | 6521 | @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded}) |
feca2ed3 JW |
6522 | A C statement (sans semicolon) to output to the stdio stream |
6523 | @var{stream} the assembler definition of a local-common-label named | |
6524 | @var{name} whose size is @var{size} bytes. The variable @var{rounded} | |
6525 | is the size rounded up to whatever alignment the caller wants. | |
6526 | ||
6527 | Use the expression @code{assemble_name (@var{stream}, @var{name})} to | |
6528 | output the name itself; before and after that, output the additional | |
6529 | assembler syntax for defining the name, and a newline. | |
6530 | ||
6531 | This macro controls how the assembler definitions of uninitialized | |
6532 | static variables are output. | |
a2c4f8e0 | 6533 | @end defmac |
feca2ed3 | 6534 | |
a2c4f8e0 | 6535 | @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment}) |
feca2ed3 JW |
6536 | Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a |
6537 | separate, explicit argument. If you define this macro, it is used in | |
6538 | place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in | |
6539 | handling the required alignment of the variable. The alignment is specified | |
6540 | as the number of bits. | |
a2c4f8e0 | 6541 | @end defmac |
feca2ed3 | 6542 | |
a2c4f8e0 | 6543 | @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment}) |
e9a25f70 JL |
6544 | Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the |
6545 | variable to be output, if there is one, or @code{NULL_TREE} if there | |
8760eaae | 6546 | is no corresponding variable. If you define this macro, GCC will use it |
e9a25f70 JL |
6547 | in place of both @code{ASM_OUTPUT_DECL} and |
6548 | @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see | |
6549 | the variable's decl in order to chose what to output. | |
a2c4f8e0 | 6550 | @end defmac |
e9a25f70 | 6551 | |
a2c4f8e0 | 6552 | @defmac ASM_OUTPUT_SHARED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded}) |
feca2ed3 JW |
6553 | If defined, it is similar to @code{ASM_OUTPUT_LOCAL}, except that it |
6554 | is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_LOCAL} | |
6555 | will be used. | |
a2c4f8e0 | 6556 | @end defmac |
feca2ed3 JW |
6557 | |
6558 | @node Label Output | |
6559 | @subsection Output and Generation of Labels | |
6560 | ||
6561 | @c prevent bad page break with this line | |
6562 | This is about outputting labels. | |
6563 | ||
feca2ed3 | 6564 | @findex assemble_name |
a2c4f8e0 | 6565 | @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name}) |
feca2ed3 JW |
6566 | A C statement (sans semicolon) to output to the stdio stream |
6567 | @var{stream} the assembler definition of a label named @var{name}. | |
6568 | Use the expression @code{assemble_name (@var{stream}, @var{name})} to | |
6569 | output the name itself; before and after that, output the additional | |
4ad5e05d KG |
6570 | assembler syntax for defining the name, and a newline. A default |
6571 | definition of this macro is provided which is correct for most systems. | |
a2c4f8e0 | 6572 | @end defmac |
feca2ed3 | 6573 | |
57829bc4 MM |
6574 | @findex assemble_name_raw |
6575 | @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name}) | |
e374d5c9 | 6576 | Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known |
57829bc4 MM |
6577 | to refer to a compiler-generated label. The default definition uses |
6578 | @code{assemble_name_raw}, which is like @code{assemble_name} except | |
6579 | that it is more efficient. | |
6580 | @end defmac | |
6581 | ||
a2c4f8e0 | 6582 | @defmac SIZE_ASM_OP |
2be2ac70 ZW |
6583 | A C string containing the appropriate assembler directive to specify the |
6584 | size of a symbol, without any arguments. On systems that use ELF, the | |
6585 | default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other | |
6586 | systems, the default is not to define this macro. | |
6587 | ||
6588 | Define this macro only if it is correct to use the default definitions | |
6589 | of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE} | |
6590 | for your system. If you need your own custom definitions of those | |
6591 | macros, or if you do not need explicit symbol sizes at all, do not | |
6592 | define this macro. | |
a2c4f8e0 | 6593 | @end defmac |
2be2ac70 | 6594 | |
a2c4f8e0 | 6595 | @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size}) |
2be2ac70 ZW |
6596 | A C statement (sans semicolon) to output to the stdio stream |
6597 | @var{stream} a directive telling the assembler that the size of the | |
6598 | symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}. | |
6599 | If you define @code{SIZE_ASM_OP}, a default definition of this macro is | |
6600 | provided. | |
a2c4f8e0 | 6601 | @end defmac |
2be2ac70 | 6602 | |
a2c4f8e0 | 6603 | @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name}) |
2be2ac70 ZW |
6604 | A C statement (sans semicolon) to output to the stdio stream |
6605 | @var{stream} a directive telling the assembler to calculate the size of | |
99086d59 | 6606 | the symbol @var{name} by subtracting its address from the current |
73774972 | 6607 | address. |
99086d59 ZW |
6608 | |
6609 | If you define @code{SIZE_ASM_OP}, a default definition of this macro is | |
6610 | provided. The default assumes that the assembler recognizes a special | |
6611 | @samp{.} symbol as referring to the current address, and can calculate | |
6612 | the difference between this and another symbol. If your assembler does | |
6613 | not recognize @samp{.} or cannot do calculations with it, you will need | |
6614 | to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique. | |
a2c4f8e0 | 6615 | @end defmac |
2be2ac70 | 6616 | |
a2c4f8e0 | 6617 | @defmac TYPE_ASM_OP |
2be2ac70 ZW |
6618 | A C string containing the appropriate assembler directive to specify the |
6619 | type of a symbol, without any arguments. On systems that use ELF, the | |
6620 | default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other | |
6621 | systems, the default is not to define this macro. | |
6622 | ||
6623 | Define this macro only if it is correct to use the default definition of | |
6624 | @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own | |
6625 | custom definition of this macro, or if you do not need explicit symbol | |
6626 | types at all, do not define this macro. | |
a2c4f8e0 | 6627 | @end defmac |
2be2ac70 | 6628 | |
a2c4f8e0 | 6629 | @defmac TYPE_OPERAND_FMT |
2be2ac70 ZW |
6630 | A C string which specifies (using @code{printf} syntax) the format of |
6631 | the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the | |
6632 | default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems, | |
6633 | the default is not to define this macro. | |
6634 | ||
6635 | Define this macro only if it is correct to use the default definition of | |
6636 | @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own | |
6637 | custom definition of this macro, or if you do not need explicit symbol | |
6638 | types at all, do not define this macro. | |
a2c4f8e0 | 6639 | @end defmac |
2be2ac70 | 6640 | |
a2c4f8e0 | 6641 | @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type}) |
2be2ac70 ZW |
6642 | A C statement (sans semicolon) to output to the stdio stream |
6643 | @var{stream} a directive telling the assembler that the type of the | |
6644 | symbol @var{name} is @var{type}. @var{type} is a C string; currently, | |
6645 | that string is always either @samp{"function"} or @samp{"object"}, but | |
6646 | you should not count on this. | |
6647 | ||
6648 | If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default | |
6649 | definition of this macro is provided. | |
a2c4f8e0 | 6650 | @end defmac |
2be2ac70 | 6651 | |
a2c4f8e0 | 6652 | @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl}) |
feca2ed3 JW |
6653 | A C statement (sans semicolon) to output to the stdio stream |
6654 | @var{stream} any text necessary for declaring the name @var{name} of a | |
6655 | function which is being defined. This macro is responsible for | |
6656 | outputting the label definition (perhaps using | |
6657 | @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the | |
6658 | @code{FUNCTION_DECL} tree node representing the function. | |
6659 | ||
6660 | If this macro is not defined, then the function name is defined in the | |
6661 | usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}). | |
6662 | ||
2be2ac70 ZW |
6663 | You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition |
6664 | of this macro. | |
a2c4f8e0 | 6665 | @end defmac |
2be2ac70 | 6666 | |
a2c4f8e0 | 6667 | @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl}) |
feca2ed3 JW |
6668 | A C statement (sans semicolon) to output to the stdio stream |
6669 | @var{stream} any text necessary for declaring the size of a function | |
6670 | which is being defined. The argument @var{name} is the name of the | |
6671 | function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node | |
6672 | representing the function. | |
6673 | ||
6674 | If this macro is not defined, then the function size is not defined. | |
6675 | ||
2be2ac70 ZW |
6676 | You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition |
6677 | of this macro. | |
a2c4f8e0 | 6678 | @end defmac |
2be2ac70 | 6679 | |
a2c4f8e0 | 6680 | @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl}) |
feca2ed3 JW |
6681 | A C statement (sans semicolon) to output to the stdio stream |
6682 | @var{stream} any text necessary for declaring the name @var{name} of an | |
6683 | initialized variable which is being defined. This macro must output the | |
6684 | label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument | |
6685 | @var{decl} is the @code{VAR_DECL} tree node representing the variable. | |
6686 | ||
6687 | If this macro is not defined, then the variable name is defined in the | |
6688 | usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}). | |
6689 | ||
2be2ac70 ZW |
6690 | You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or |
6691 | @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro. | |
a2c4f8e0 | 6692 | @end defmac |
2be2ac70 | 6693 | |
18f3e349 GK |
6694 | @defmac ASM_DECLARE_CONSTANT_NAME (@var{stream}, @var{name}, @var{exp}, @var{size}) |
6695 | A C statement (sans semicolon) to output to the stdio stream | |
6696 | @var{stream} any text necessary for declaring the name @var{name} of a | |
6697 | constant which is being defined. This macro is responsible for | |
6698 | outputting the label definition (perhaps using | |
6699 | @code{ASM_OUTPUT_LABEL}). The argument @var{exp} is the | |
6700 | value of the constant, and @var{size} is the size of the constant | |
6701 | in bytes. @var{name} will be an internal label. | |
6702 | ||
6703 | If this macro is not defined, then the @var{name} is defined in the | |
6704 | usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}). | |
6705 | ||
6706 | You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition | |
6707 | of this macro. | |
6708 | @end defmac | |
6709 | ||
a2c4f8e0 | 6710 | @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name}) |
1cb36a98 RH |
6711 | A C statement (sans semicolon) to output to the stdio stream |
6712 | @var{stream} any text necessary for claiming a register @var{regno} | |
6713 | for a global variable @var{decl} with name @var{name}. | |
6714 | ||
6715 | If you don't define this macro, that is equivalent to defining it to do | |
6716 | nothing. | |
a2c4f8e0 | 6717 | @end defmac |
1cb36a98 | 6718 | |
a2c4f8e0 | 6719 | @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend}) |
feca2ed3 JW |
6720 | A C statement (sans semicolon) to finish up declaring a variable name |
6721 | once the compiler has processed its initializer fully and thus has had a | |
6722 | chance to determine the size of an array when controlled by an | |
6723 | initializer. This is used on systems where it's necessary to declare | |
6724 | something about the size of the object. | |
6725 | ||
6726 | If you don't define this macro, that is equivalent to defining it to do | |
6727 | nothing. | |
6728 | ||
2be2ac70 ZW |
6729 | You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or |
6730 | @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro. | |
a2c4f8e0 | 6731 | @end defmac |
2be2ac70 | 6732 | |
5eb99654 KG |
6733 | @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_LABEL (FILE *@var{stream}, const char *@var{name}) |
6734 | This target hook is a function to output to the stdio stream | |
feca2ed3 | 6735 | @var{stream} some commands that will make the label @var{name} global; |
5eb99654 | 6736 | that is, available for reference from other files. |
feca2ed3 | 6737 | |
5eb99654 KG |
6738 | The default implementation relies on a proper definition of |
6739 | @code{GLOBAL_ASM_OP}. | |
b65d23aa | 6740 | @end deftypefn |
072cdaed | 6741 | |
a2c4f8e0 | 6742 | @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name}) |
feca2ed3 JW |
6743 | A C statement (sans semicolon) to output to the stdio stream |
6744 | @var{stream} some commands that will make the label @var{name} weak; | |
6745 | that is, available for reference from other files but only used if | |
6746 | no other definition is available. Use the expression | |
6747 | @code{assemble_name (@var{stream}, @var{name})} to output the name | |
6748 | itself; before and after that, output the additional assembler syntax | |
6749 | for making that name weak, and a newline. | |
6750 | ||
79c4e63f AM |
6751 | If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not |
6752 | support weak symbols and you should not define the @code{SUPPORTS_WEAK} | |
6753 | macro. | |
a2c4f8e0 | 6754 | @end defmac |
79c4e63f | 6755 | |
a2c4f8e0 | 6756 | @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value}) |
79c4e63f AM |
6757 | Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and |
6758 | @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function | |
6759 | or variable decl. If @var{value} is not @code{NULL}, this C statement | |
6760 | should output to the stdio stream @var{stream} assembler code which | |
6761 | defines (equates) the weak symbol @var{name} to have the value | |
6762 | @var{value}. If @var{value} is @code{NULL}, it should output commands | |
6763 | to make @var{name} weak. | |
a2c4f8e0 | 6764 | @end defmac |
feca2ed3 | 6765 | |
a2c4f8e0 | 6766 | @defmac SUPPORTS_WEAK |
feca2ed3 JW |
6767 | A C expression which evaluates to true if the target supports weak symbols. |
6768 | ||
6769 | If you don't define this macro, @file{defaults.h} provides a default | |
79c4e63f AM |
6770 | definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL} |
6771 | is defined, the default definition is @samp{1}; otherwise, it is | |
6772 | @samp{0}. Define this macro if you want to control weak symbol support | |
6773 | with a compiler flag such as @option{-melf}. | |
a2c4f8e0 | 6774 | @end defmac |
feca2ed3 | 6775 | |
a2c4f8e0 | 6776 | @defmac MAKE_DECL_ONE_ONLY (@var{decl}) |
feca2ed3 JW |
6777 | A C statement (sans semicolon) to mark @var{decl} to be emitted as a |
6778 | public symbol such that extra copies in multiple translation units will | |
6779 | be discarded by the linker. Define this macro if your object file | |
6780 | format provides support for this concept, such as the @samp{COMDAT} | |
6781 | section flags in the Microsoft Windows PE/COFF format, and this support | |
6782 | requires changes to @var{decl}, such as putting it in a separate section. | |
a2c4f8e0 | 6783 | @end defmac |
feca2ed3 | 6784 | |
a2c4f8e0 | 6785 | @defmac SUPPORTS_ONE_ONLY |
feca2ed3 JW |
6786 | A C expression which evaluates to true if the target supports one-only |
6787 | semantics. | |
6788 | ||
6789 | If you don't define this macro, @file{varasm.c} provides a default | |
6790 | definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default | |
6791 | definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if | |
e9a25f70 | 6792 | you want to control one-only symbol support with a compiler flag, or if |
feca2ed3 JW |
6793 | setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to |
6794 | be emitted as one-only. | |
a2c4f8e0 | 6795 | @end defmac |
feca2ed3 | 6796 | |
93638d7a AM |
6797 | @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_VISIBILITY (tree @var{decl}, const char *@var{visibility}) |
6798 | This target hook is a function to output to @var{asm_out_file} some | |
6799 | commands that will make the symbol(s) associated with @var{decl} have | |
6800 | hidden, protected or internal visibility as specified by @var{visibility}. | |
6801 | @end deftypefn | |
6802 | ||
0524c91d | 6803 | @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC |
4746cf84 | 6804 | A C expression that evaluates to true if the target's linker expects |
0524c91d | 6805 | that weak symbols do not appear in a static archive's table of contents. |
f676971a | 6806 | The default is @code{0}. |
0524c91d MA |
6807 | |
6808 | Leaving weak symbols out of an archive's table of contents means that, | |
6809 | if a symbol will only have a definition in one translation unit and | |
6810 | will have undefined references from other translation units, that | |
6811 | symbol should not be weak. Defining this macro to be nonzero will | |
6812 | thus have the effect that certain symbols that would normally be weak | |
6813 | (explicit template instantiations, and vtables for polymorphic classes | |
6814 | with noninline key methods) will instead be nonweak. | |
6815 | ||
6816 | The C++ ABI requires this macro to be zero. Define this macro for | |
6817 | targets where full C++ ABI compliance is impossible and where linker | |
6818 | restrictions require weak symbols to be left out of a static archive's | |
6819 | table of contents. | |
4746cf84 MA |
6820 | @end defmac |
6821 | ||
a2c4f8e0 | 6822 | @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name}) |
feca2ed3 JW |
6823 | A C statement (sans semicolon) to output to the stdio stream |
6824 | @var{stream} any text necessary for declaring the name of an external | |
6825 | symbol named @var{name} which is referenced in this compilation but | |
6826 | not defined. The value of @var{decl} is the tree node for the | |
6827 | declaration. | |
6828 | ||
6829 | This macro need not be defined if it does not need to output anything. | |
6830 | The GNU assembler and most Unix assemblers don't require anything. | |
a2c4f8e0 | 6831 | @end defmac |
feca2ed3 | 6832 | |
6773a41c RO |
6833 | @deftypefn {Target Hook} void TARGET_ASM_EXTERNAL_LIBCALL (rtx @var{symref}) |
6834 | This target hook is a function to output to @var{asm_out_file} an assembler | |
feca2ed3 | 6835 | pseudo-op to declare a library function name external. The name of the |
6773a41c RO |
6836 | library function is given by @var{symref}, which is a @code{symbol_ref}. |
6837 | @end deftypefn | |
feca2ed3 | 6838 | |
8e3e233b DP |
6839 | @deftypefn {Target Hook} void TARGET_ASM_MARK_DECL_PRESERVED (tree @var{decl}) |
6840 | This target hook is a function to output to @var{asm_out_file} an assembler | |
6841 | directive to annotate used symbol. Darwin target use .no_dead_code_strip | |
6842 | directive. | |
6843 | @end deftypefn | |
6844 | ||
a2c4f8e0 | 6845 | @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name}) |
feca2ed3 JW |
6846 | A C statement (sans semicolon) to output to the stdio stream |
6847 | @var{stream} a reference in assembler syntax to a label named | |
6848 | @var{name}. This should add @samp{_} to the front of the name, if that | |
6849 | is customary on your operating system, as it is in most Berkeley Unix | |
6850 | systems. This macro is used in @code{assemble_name}. | |
a2c4f8e0 | 6851 | @end defmac |
feca2ed3 | 6852 | |
a2c4f8e0 | 6853 | @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym}) |
99c8c61c | 6854 | A C statement (sans semicolon) to output a reference to |
2f0b7af6 | 6855 | @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name} |
99c8c61c AO |
6856 | will be used to output the name of the symbol. This macro may be used |
6857 | to modify the way a symbol is referenced depending on information | |
fb49053f | 6858 | encoded by @code{TARGET_ENCODE_SECTION_INFO}. |
a2c4f8e0 | 6859 | @end defmac |
99c8c61c | 6860 | |
a2c4f8e0 | 6861 | @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf}) |
2f0b7af6 | 6862 | A C statement (sans semicolon) to output a reference to @var{buf}, the |
4226378a | 6863 | result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined, |
2f0b7af6 GK |
6864 | @code{assemble_name} will be used to output the name of the symbol. |
6865 | This macro is not used by @code{output_asm_label}, or the @code{%l} | |
6866 | specifier that calls it; the intention is that this macro should be set | |
4226378a PK |
6867 | when it is necessary to output a label differently when its address is |
6868 | being taken. | |
a2c4f8e0 | 6869 | @end defmac |
2f0b7af6 | 6870 | |
4977bab6 ZW |
6871 | @deftypefn {Target Hook} void TARGET_ASM_INTERNAL_LABEL (FILE *@var{stream}, const char *@var{prefix}, unsigned long @var{labelno}) |
6872 | A function to output to the stdio stream @var{stream} a label whose | |
6873 | name is made from the string @var{prefix} and the number @var{labelno}. | |
feca2ed3 JW |
6874 | |
6875 | It is absolutely essential that these labels be distinct from the labels | |
6876 | used for user-level functions and variables. Otherwise, certain programs | |
6877 | will have name conflicts with internal labels. | |
6878 | ||
6879 | It is desirable to exclude internal labels from the symbol table of the | |
6880 | object file. Most assemblers have a naming convention for labels that | |
6881 | should be excluded; on many systems, the letter @samp{L} at the | |
6882 | beginning of a label has this effect. You should find out what | |
6883 | convention your system uses, and follow it. | |
6884 | ||
8a36672b | 6885 | The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}. |
4977bab6 | 6886 | @end deftypefn |
feca2ed3 | 6887 | |
a2c4f8e0 | 6888 | @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num}) |
8215347e JW |
6889 | A C statement to output to the stdio stream @var{stream} a debug info |
6890 | label whose name is made from the string @var{prefix} and the number | |
6891 | @var{num}. This is useful for VLIW targets, where debug info labels | |
6892 | may need to be treated differently than branch target labels. On some | |
6893 | systems, branch target labels must be at the beginning of instruction | |
6894 | bundles, but debug info labels can occur in the middle of instruction | |
6895 | bundles. | |
6896 | ||
4977bab6 | 6897 | If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be |
8215347e | 6898 | used. |
a2c4f8e0 | 6899 | @end defmac |
8215347e | 6900 | |
a2c4f8e0 | 6901 | @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num}) |
feca2ed3 JW |
6902 | A C statement to store into the string @var{string} a label whose name |
6903 | is made from the string @var{prefix} and the number @var{num}. | |
6904 | ||
6905 | This string, when output subsequently by @code{assemble_name}, should | |
4977bab6 | 6906 | produce the output that @code{(*targetm.asm_out.internal_label)} would produce |
feca2ed3 JW |
6907 | with the same @var{prefix} and @var{num}. |
6908 | ||
6909 | If the string begins with @samp{*}, then @code{assemble_name} will | |
6910 | output the rest of the string unchanged. It is often convenient for | |
6911 | @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the | |
6912 | string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets | |
6913 | to output the string, and may change it. (Of course, | |
6914 | @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so | |
6915 | you should know what it does on your machine.) | |
a2c4f8e0 | 6916 | @end defmac |
feca2ed3 | 6917 | |
a2c4f8e0 | 6918 | @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number}) |
feca2ed3 JW |
6919 | A C expression to assign to @var{outvar} (which is a variable of type |
6920 | @code{char *}) a newly allocated string made from the string | |
6921 | @var{name} and the number @var{number}, with some suitable punctuation | |
6922 | added. Use @code{alloca} to get space for the string. | |
6923 | ||
6924 | The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to | |
6925 | produce an assembler label for an internal static variable whose name is | |
6926 | @var{name}. Therefore, the string must be such as to result in valid | |
6927 | assembler code. The argument @var{number} is different each time this | |
6928 | macro is executed; it prevents conflicts between similarly-named | |
6929 | internal static variables in different scopes. | |
6930 | ||
6931 | Ideally this string should not be a valid C identifier, to prevent any | |
6932 | conflict with the user's own symbols. Most assemblers allow periods | |
6933 | or percent signs in assembler symbols; putting at least one of these | |
6934 | between the name and the number will suffice. | |
6935 | ||
4977bab6 ZW |
6936 | If this macro is not defined, a default definition will be provided |
6937 | which is correct for most systems. | |
a2c4f8e0 | 6938 | @end defmac |
4977bab6 | 6939 | |
a2c4f8e0 | 6940 | @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value}) |
feca2ed3 JW |
6941 | A C statement to output to the stdio stream @var{stream} assembler code |
6942 | which defines (equates) the symbol @var{name} to have the value @var{value}. | |
6943 | ||
203cb4ef | 6944 | @findex SET_ASM_OP |
aee96fe9 | 6945 | If @code{SET_ASM_OP} is defined, a default definition is provided which is |
feca2ed3 | 6946 | correct for most systems. |
a2c4f8e0 | 6947 | @end defmac |
810e3c45 | 6948 | |
a2c4f8e0 | 6949 | @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value}) |
e4faf1eb | 6950 | A C statement to output to the stdio stream @var{stream} assembler code |
3b7a2e58 | 6951 | which defines (equates) the symbol whose tree node is @var{decl_of_name} |
e4faf1eb NC |
6952 | to have the value of the tree node @var{decl_of_value}. This macro will |
6953 | be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if | |
6954 | the tree nodes are available. | |
6955 | ||
203cb4ef | 6956 | @findex SET_ASM_OP |
aee96fe9 | 6957 | If @code{SET_ASM_OP} is defined, a default definition is provided which is |
956d6950 | 6958 | correct for most systems. |
a2c4f8e0 | 6959 | @end defmac |
956d6950 | 6960 | |
083b6717 JDA |
6961 | @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value}) |
6962 | A C statement that evaluates to true if the assembler code which defines | |
6963 | (equates) the symbol whose tree node is @var{decl_of_name} to have the value | |
6964 | of the tree node @var{decl_of_value} should be emitted near the end of the | |
6965 | current compilation unit. The default is to not defer output of defines. | |
6966 | This macro affects defines output by @samp{ASM_OUTPUT_DEF} and | |
6967 | @samp{ASM_OUTPUT_DEF_FROM_DECLS}. | |
6968 | @end defmac | |
6969 | ||
a2c4f8e0 | 6970 | @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value}) |
810e3c45 JM |
6971 | A C statement to output to the stdio stream @var{stream} assembler code |
6972 | which defines (equates) the weak symbol @var{name} to have the value | |
3aa8ab7b L |
6973 | @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as |
6974 | an undefined weak symbol. | |
810e3c45 JM |
6975 | |
6976 | Define this macro if the target only supports weak aliases; define | |
aee96fe9 | 6977 | @code{ASM_OUTPUT_DEF} instead if possible. |
a2c4f8e0 | 6978 | @end defmac |
810e3c45 | 6979 | |
a2c4f8e0 | 6980 | @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name}) |
feca2ed3 | 6981 | Define this macro to override the default assembler names used for |
2147b154 | 6982 | Objective-C methods. |
feca2ed3 JW |
6983 | |
6984 | The default name is a unique method number followed by the name of the | |
6985 | class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of | |
6986 | the category is also included in the assembler name (e.g.@: | |
6987 | @samp{_1_Foo_Bar}). | |
6988 | ||
6989 | These names are safe on most systems, but make debugging difficult since | |
6990 | the method's selector is not present in the name. Therefore, particular | |
6991 | systems define other ways of computing names. | |
6992 | ||
6993 | @var{buf} is an expression of type @code{char *} which gives you a | |
6994 | buffer in which to store the name; its length is as long as | |
6995 | @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus | |
6996 | 50 characters extra. | |
6997 | ||
6998 | The argument @var{is_inst} specifies whether the method is an instance | |
6999 | method or a class method; @var{class_name} is the name of the class; | |
59d42021 | 7000 | @var{cat_name} is the name of the category (or @code{NULL} if the method is not |
feca2ed3 JW |
7001 | in a category); and @var{sel_name} is the name of the selector. |
7002 | ||
7003 | On systems where the assembler can handle quoted names, you can use this | |
7004 | macro to provide more human-readable names. | |
a2c4f8e0 | 7005 | @end defmac |
28df0b5a | 7006 | |
a2c4f8e0 | 7007 | @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name}) |
f60b945b SS |
7008 | A C statement (sans semicolon) to output to the stdio stream |
7009 | @var{stream} commands to declare that the label @var{name} is an | |
7010 | Objective-C class reference. This is only needed for targets whose | |
7011 | linkers have special support for NeXT-style runtimes. | |
a2c4f8e0 | 7012 | @end defmac |
f60b945b | 7013 | |
a2c4f8e0 | 7014 | @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name}) |
28df0b5a SS |
7015 | A C statement (sans semicolon) to output to the stdio stream |
7016 | @var{stream} commands to declare that the label @var{name} is an | |
7017 | unresolved Objective-C class reference. This is only needed for targets | |
7018 | whose linkers have special support for NeXT-style runtimes. | |
a2c4f8e0 | 7019 | @end defmac |
feca2ed3 JW |
7020 | |
7021 | @node Initialization | |
7022 | @subsection How Initialization Functions Are Handled | |
7023 | @cindex initialization routines | |
7024 | @cindex termination routines | |
7025 | @cindex constructors, output of | |
7026 | @cindex destructors, output of | |
7027 | ||
7028 | The compiled code for certain languages includes @dfn{constructors} | |
7029 | (also called @dfn{initialization routines})---functions to initialize | |
7030 | data in the program when the program is started. These functions need | |
7031 | to be called before the program is ``started''---that is to say, before | |
7032 | @code{main} is called. | |
7033 | ||
7034 | Compiling some languages generates @dfn{destructors} (also called | |
7035 | @dfn{termination routines}) that should be called when the program | |
7036 | terminates. | |
7037 | ||
7038 | To make the initialization and termination functions work, the compiler | |
7039 | must output something in the assembler code to cause those functions to | |
7040 | be called at the appropriate time. When you port the compiler to a new | |
7041 | system, you need to specify how to do this. | |
7042 | ||
7043 | There are two major ways that GCC currently supports the execution of | |
7044 | initialization and termination functions. Each way has two variants. | |
7045 | Much of the structure is common to all four variations. | |
7046 | ||
7047 | @findex __CTOR_LIST__ | |
7048 | @findex __DTOR_LIST__ | |
7049 | The linker must build two lists of these functions---a list of | |
7050 | initialization functions, called @code{__CTOR_LIST__}, and a list of | |
7051 | termination functions, called @code{__DTOR_LIST__}. | |
7052 | ||
7053 | Each list always begins with an ignored function pointer (which may hold | |
7054 | 0, @minus{}1, or a count of the function pointers after it, depending on | |
7055 | the environment). This is followed by a series of zero or more function | |
7056 | pointers to constructors (or destructors), followed by a function | |
7057 | pointer containing zero. | |
7058 | ||
7059 | Depending on the operating system and its executable file format, either | |
7060 | @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup | |
7061 | time and exit time. Constructors are called in reverse order of the | |
7062 | list; destructors in forward order. | |
7063 | ||
7064 | The best way to handle static constructors works only for object file | |
7065 | formats which provide arbitrarily-named sections. A section is set | |
7066 | aside for a list of constructors, and another for a list of destructors. | |
7067 | Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each | |
7068 | object file that defines an initialization function also puts a word in | |
7069 | the constructor section to point to that function. The linker | |
7070 | accumulates all these words into one contiguous @samp{.ctors} section. | |
7071 | Termination functions are handled similarly. | |
7072 | ||
2cc07db4 RH |
7073 | This method will be chosen as the default by @file{target-def.h} if |
7074 | @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not | |
f282ffb3 | 7075 | support arbitrary sections, but does support special designated |
2cc07db4 RH |
7076 | constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP} |
7077 | and @code{DTORS_SECTION_ASM_OP} to achieve the same effect. | |
feca2ed3 JW |
7078 | |
7079 | When arbitrary sections are available, there are two variants, depending | |
7080 | upon how the code in @file{crtstuff.c} is called. On systems that | |
2cc07db4 | 7081 | support a @dfn{.init} section which is executed at program startup, |
feca2ed3 | 7082 | parts of @file{crtstuff.c} are compiled into that section. The |
05739753 | 7083 | program is linked by the @command{gcc} driver like this: |
feca2ed3 | 7084 | |
3ab51846 | 7085 | @smallexample |
2cc07db4 | 7086 | ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o |
3ab51846 | 7087 | @end smallexample |
feca2ed3 | 7088 | |
2cc07db4 RH |
7089 | The prologue of a function (@code{__init}) appears in the @code{.init} |
7090 | section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise | |
7091 | for the function @code{__fini} in the @dfn{.fini} section. Normally these | |
7092 | files are provided by the operating system or by the GNU C library, but | |
7093 | are provided by GCC for a few targets. | |
7094 | ||
7095 | The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets) | |
7096 | compiled from @file{crtstuff.c}. They contain, among other things, code | |
7097 | fragments within the @code{.init} and @code{.fini} sections that branch | |
7098 | to routines in the @code{.text} section. The linker will pull all parts | |
7099 | of a section together, which results in a complete @code{__init} function | |
7100 | that invokes the routines we need at startup. | |
feca2ed3 JW |
7101 | |
7102 | To use this variant, you must define the @code{INIT_SECTION_ASM_OP} | |
7103 | macro properly. | |
7104 | ||
2cc07db4 RH |
7105 | If no init section is available, when GCC compiles any function called |
7106 | @code{main} (or more accurately, any function designated as a program | |
7107 | entry point by the language front end calling @code{expand_main_function}), | |
7108 | it inserts a procedure call to @code{__main} as the first executable code | |
7109 | after the function prologue. The @code{__main} function is defined | |
7110 | in @file{libgcc2.c} and runs the global constructors. | |
feca2ed3 JW |
7111 | |
7112 | In file formats that don't support arbitrary sections, there are again | |
7113 | two variants. In the simplest variant, the GNU linker (GNU @code{ld}) | |
7114 | and an `a.out' format must be used. In this case, | |
2cc07db4 | 7115 | @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs} |
feca2ed3 JW |
7116 | entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__}, |
7117 | and with the address of the void function containing the initialization | |
7118 | code as its value. The GNU linker recognizes this as a request to add | |
2cc07db4 | 7119 | the value to a @dfn{set}; the values are accumulated, and are eventually |
feca2ed3 JW |
7120 | placed in the executable as a vector in the format described above, with |
7121 | a leading (ignored) count and a trailing zero element. | |
2cc07db4 | 7122 | @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init |
feca2ed3 JW |
7123 | section is available, the absence of @code{INIT_SECTION_ASM_OP} causes |
7124 | the compilation of @code{main} to call @code{__main} as above, starting | |
7125 | the initialization process. | |
7126 | ||
7127 | The last variant uses neither arbitrary sections nor the GNU linker. | |
7128 | This is preferable when you want to do dynamic linking and when using | |
161d7b59 | 7129 | file formats which the GNU linker does not support, such as `ECOFF'@. In |
2cc07db4 RH |
7130 | this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and |
7131 | termination functions are recognized simply by their names. This requires | |
7132 | an extra program in the linkage step, called @command{collect2}. This program | |
7133 | pretends to be the linker, for use with GCC; it does its job by running | |
7134 | the ordinary linker, but also arranges to include the vectors of | |
7135 | initialization and termination functions. These functions are called | |
7136 | via @code{__main} as described above. In order to use this method, | |
7137 | @code{use_collect2} must be defined in the target in @file{config.gcc}. | |
feca2ed3 JW |
7138 | |
7139 | @ifinfo | |
7140 | The following section describes the specific macros that control and | |
7141 | customize the handling of initialization and termination functions. | |
7142 | @end ifinfo | |
7143 | ||
7144 | @node Macros for Initialization | |
7145 | @subsection Macros Controlling Initialization Routines | |
7146 | ||
7147 | Here are the macros that control how the compiler handles initialization | |
7148 | and termination functions: | |
7149 | ||
a2c4f8e0 | 7150 | @defmac INIT_SECTION_ASM_OP |
047c1c92 HPN |
7151 | If defined, a C string constant, including spacing, for the assembler |
7152 | operation to identify the following data as initialization code. If not | |
7153 | defined, GCC will assume such a section does not exist. When you are | |
7154 | using special sections for initialization and termination functions, this | |
7155 | macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to | |
7156 | run the initialization functions. | |
a2c4f8e0 | 7157 | @end defmac |
feca2ed3 | 7158 | |
a2c4f8e0 | 7159 | @defmac HAS_INIT_SECTION |
feca2ed3 | 7160 | If defined, @code{main} will not call @code{__main} as described above. |
2cc07db4 RH |
7161 | This macro should be defined for systems that control start-up code |
7162 | on a symbol-by-symbol basis, such as OSF/1, and should not | |
7163 | be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}. | |
a2c4f8e0 | 7164 | @end defmac |
feca2ed3 | 7165 | |
a2c4f8e0 | 7166 | @defmac LD_INIT_SWITCH |
feca2ed3 JW |
7167 | If defined, a C string constant for a switch that tells the linker that |
7168 | the following symbol is an initialization routine. | |
a2c4f8e0 | 7169 | @end defmac |
feca2ed3 | 7170 | |
a2c4f8e0 | 7171 | @defmac LD_FINI_SWITCH |
feca2ed3 JW |
7172 | If defined, a C string constant for a switch that tells the linker that |
7173 | the following symbol is a finalization routine. | |
a2c4f8e0 | 7174 | @end defmac |
feca2ed3 | 7175 | |
a2c4f8e0 | 7176 | @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func}) |
414e05cf RE |
7177 | If defined, a C statement that will write a function that can be |
7178 | automatically called when a shared library is loaded. The function | |
7179 | should call @var{func}, which takes no arguments. If not defined, and | |
7180 | the object format requires an explicit initialization function, then a | |
172270b3 | 7181 | function called @code{_GLOBAL__DI} will be generated. |
414e05cf RE |
7182 | |
7183 | This function and the following one are used by collect2 when linking a | |
f282ffb3 | 7184 | shared library that needs constructors or destructors, or has DWARF2 |
414e05cf | 7185 | exception tables embedded in the code. |
a2c4f8e0 | 7186 | @end defmac |
414e05cf | 7187 | |
a2c4f8e0 | 7188 | @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func}) |
414e05cf RE |
7189 | If defined, a C statement that will write a function that can be |
7190 | automatically called when a shared library is unloaded. The function | |
7191 | should call @var{func}, which takes no arguments. If not defined, and | |
7192 | the object format requires an explicit finalization function, then a | |
172270b3 | 7193 | function called @code{_GLOBAL__DD} will be generated. |
a2c4f8e0 | 7194 | @end defmac |
414e05cf | 7195 | |
a2c4f8e0 | 7196 | @defmac INVOKE__main |
feca2ed3 JW |
7197 | If defined, @code{main} will call @code{__main} despite the presence of |
7198 | @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems | |
7199 | where the init section is not actually run automatically, but is still | |
7200 | useful for collecting the lists of constructors and destructors. | |
a2c4f8e0 | 7201 | @end defmac |
feca2ed3 | 7202 | |
a2c4f8e0 | 7203 | @defmac SUPPORTS_INIT_PRIORITY |
ea4f1fce JO |
7204 | If nonzero, the C++ @code{init_priority} attribute is supported and the |
7205 | compiler should emit instructions to control the order of initialization | |
7206 | of objects. If zero, the compiler will issue an error message upon | |
7207 | encountering an @code{init_priority} attribute. | |
a2c4f8e0 | 7208 | @end defmac |
2cc07db4 RH |
7209 | |
7210 | @deftypefn {Target Hook} bool TARGET_HAVE_CTORS_DTORS | |
7211 | This value is true if the target supports some ``native'' method of | |
7212 | collecting constructors and destructors to be run at startup and exit. | |
7213 | It is false if we must use @command{collect2}. | |
7214 | @end deftypefn | |
7215 | ||
7216 | @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority}) | |
7217 | If defined, a function that outputs assembler code to arrange to call | |
7218 | the function referenced by @var{symbol} at initialization time. | |
ea4f1fce | 7219 | |
2cc07db4 RH |
7220 | Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking |
7221 | no arguments and with no return value. If the target supports initialization | |
7222 | priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY}; | |
7223 | otherwise it must be @code{DEFAULT_INIT_PRIORITY}. | |
7224 | ||
14976c58 | 7225 | If this macro is not defined by the target, a suitable default will |
2cc07db4 RH |
7226 | be chosen if (1) the target supports arbitrary section names, (2) the |
7227 | target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2} | |
7228 | is not defined. | |
7229 | @end deftypefn | |
7230 | ||
7231 | @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority}) | |
7232 | This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination | |
feca2ed3 | 7233 | functions rather than initialization functions. |
2cc07db4 | 7234 | @end deftypefn |
14686fcd | 7235 | |
2cc07db4 RH |
7236 | If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine |
7237 | generated for the generated object file will have static linkage. | |
feca2ed3 | 7238 | |
2cc07db4 RH |
7239 | If your system uses @command{collect2} as the means of processing |
7240 | constructors, then that program normally uses @command{nm} to scan | |
7241 | an object file for constructor functions to be called. | |
14686fcd | 7242 | |
4a023207 | 7243 | On certain kinds of systems, you can define this macro to make |
2cc07db4 | 7244 | @command{collect2} work faster (and, in some cases, make it work at all): |
feca2ed3 | 7245 | |
a2c4f8e0 | 7246 | @defmac OBJECT_FORMAT_COFF |
feca2ed3 | 7247 | Define this macro if the system uses COFF (Common Object File Format) |
2cc07db4 | 7248 | object files, so that @command{collect2} can assume this format and scan |
feca2ed3 | 7249 | object files directly for dynamic constructor/destructor functions. |
feca2ed3 | 7250 | |
4a023207 | 7251 | This macro is effective only in a native compiler; @command{collect2} as |
2cc07db4 | 7252 | part of a cross compiler always uses @command{nm} for the target machine. |
a2c4f8e0 | 7253 | @end defmac |
feca2ed3 | 7254 | |
a2c4f8e0 | 7255 | @defmac REAL_NM_FILE_NAME |
feca2ed3 | 7256 | Define this macro as a C string constant containing the file name to use |
2cc07db4 RH |
7257 | to execute @command{nm}. The default is to search the path normally for |
7258 | @command{nm}. | |
feca2ed3 JW |
7259 | |
7260 | If your system supports shared libraries and has a program to list the | |
7261 | dynamic dependencies of a given library or executable, you can define | |
7262 | these macros to enable support for running initialization and | |
7263 | termination functions in shared libraries: | |
a2c4f8e0 | 7264 | @end defmac |
feca2ed3 | 7265 | |
a2c4f8e0 | 7266 | @defmac LDD_SUFFIX |
2cc07db4 RH |
7267 | Define this macro to a C string constant containing the name of the program |
7268 | which lists dynamic dependencies, like @command{"ldd"} under SunOS 4. | |
a2c4f8e0 | 7269 | @end defmac |
feca2ed3 | 7270 | |
a2c4f8e0 | 7271 | @defmac PARSE_LDD_OUTPUT (@var{ptr}) |
feca2ed3 | 7272 | Define this macro to be C code that extracts filenames from the output |
aee96fe9 | 7273 | of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable |
feca2ed3 JW |
7274 | of type @code{char *} that points to the beginning of a line of output |
7275 | from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the | |
aee96fe9 JM |
7276 | code must advance @var{ptr} to the beginning of the filename on that |
7277 | line. Otherwise, it must set @var{ptr} to @code{NULL}. | |
a2c4f8e0 | 7278 | @end defmac |
feca2ed3 JW |
7279 | |
7280 | @node Instruction Output | |
7281 | @subsection Output of Assembler Instructions | |
7282 | ||
7283 | @c prevent bad page break with this line | |
7284 | This describes assembler instruction output. | |
7285 | ||
a2c4f8e0 | 7286 | @defmac REGISTER_NAMES |
feca2ed3 JW |
7287 | A C initializer containing the assembler's names for the machine |
7288 | registers, each one as a C string constant. This is what translates | |
7289 | register numbers in the compiler into assembler language. | |
a2c4f8e0 | 7290 | @end defmac |
feca2ed3 | 7291 | |
a2c4f8e0 | 7292 | @defmac ADDITIONAL_REGISTER_NAMES |
feca2ed3 JW |
7293 | If defined, a C initializer for an array of structures containing a name |
7294 | and a register number. This macro defines additional names for hard | |
7295 | registers, thus allowing the @code{asm} option in declarations to refer | |
7296 | to registers using alternate names. | |
a2c4f8e0 | 7297 | @end defmac |
feca2ed3 | 7298 | |
a2c4f8e0 | 7299 | @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr}) |
feca2ed3 JW |
7300 | Define this macro if you are using an unusual assembler that |
7301 | requires different names for the machine instructions. | |
7302 | ||
7303 | The definition is a C statement or statements which output an | |
7304 | assembler instruction opcode to the stdio stream @var{stream}. The | |
7305 | macro-operand @var{ptr} is a variable of type @code{char *} which | |
7306 | points to the opcode name in its ``internal'' form---the form that is | |
7307 | written in the machine description. The definition should output the | |
7308 | opcode name to @var{stream}, performing any translation you desire, and | |
7309 | increment the variable @var{ptr} to point at the end of the opcode | |
7310 | so that it will not be output twice. | |
7311 | ||
7312 | In fact, your macro definition may process less than the entire opcode | |
7313 | name, or more than the opcode name; but if you want to process text | |
7314 | that includes @samp{%}-sequences to substitute operands, you must take | |
7315 | care of the substitution yourself. Just be sure to increment | |
7316 | @var{ptr} over whatever text should not be output normally. | |
7317 | ||
37bef197 | 7318 | @findex recog_data.operand |
feca2ed3 | 7319 | If you need to look at the operand values, they can be found as the |
37bef197 | 7320 | elements of @code{recog_data.operand}. |
feca2ed3 JW |
7321 | |
7322 | If the macro definition does nothing, the instruction is output | |
7323 | in the usual way. | |
a2c4f8e0 | 7324 | @end defmac |
feca2ed3 | 7325 | |
a2c4f8e0 | 7326 | @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands}) |
feca2ed3 JW |
7327 | If defined, a C statement to be executed just prior to the output of |
7328 | assembler code for @var{insn}, to modify the extracted operands so | |
7329 | they will be output differently. | |
7330 | ||
7331 | Here the argument @var{opvec} is the vector containing the operands | |
7332 | extracted from @var{insn}, and @var{noperands} is the number of | |
7333 | elements of the vector which contain meaningful data for this insn. | |
7334 | The contents of this vector are what will be used to convert the insn | |
7335 | template into assembler code, so you can change the assembler output | |
7336 | by changing the contents of the vector. | |
7337 | ||
7338 | This macro is useful when various assembler syntaxes share a single | |
7339 | file of instruction patterns; by defining this macro differently, you | |
7340 | can cause a large class of instructions to be output differently (such | |
7341 | as with rearranged operands). Naturally, variations in assembler | |
7342 | syntax affecting individual insn patterns ought to be handled by | |
7343 | writing conditional output routines in those patterns. | |
7344 | ||
7345 | If this macro is not defined, it is equivalent to a null statement. | |
a2c4f8e0 | 7346 | @end defmac |
feca2ed3 | 7347 | |
a2c4f8e0 | 7348 | @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code}) |
feca2ed3 JW |
7349 | A C compound statement to output to stdio stream @var{stream} the |
7350 | assembler syntax for an instruction operand @var{x}. @var{x} is an | |
7351 | RTL expression. | |
7352 | ||
7353 | @var{code} is a value that can be used to specify one of several ways | |
7354 | of printing the operand. It is used when identical operands must be | |
7355 | printed differently depending on the context. @var{code} comes from | |
7356 | the @samp{%} specification that was used to request printing of the | |
7357 | operand. If the specification was just @samp{%@var{digit}} then | |
7358 | @var{code} is 0; if the specification was @samp{%@var{ltr} | |
7359 | @var{digit}} then @var{code} is the ASCII code for @var{ltr}. | |
7360 | ||
7361 | @findex reg_names | |
7362 | If @var{x} is a register, this macro should print the register's name. | |
7363 | The names can be found in an array @code{reg_names} whose type is | |
7364 | @code{char *[]}. @code{reg_names} is initialized from | |
7365 | @code{REGISTER_NAMES}. | |
7366 | ||
7367 | When the machine description has a specification @samp{%@var{punct}} | |
7368 | (a @samp{%} followed by a punctuation character), this macro is called | |
7369 | with a null pointer for @var{x} and the punctuation character for | |
7370 | @var{code}. | |
a2c4f8e0 | 7371 | @end defmac |
feca2ed3 | 7372 | |
a2c4f8e0 | 7373 | @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code}) |
feca2ed3 JW |
7374 | A C expression which evaluates to true if @var{code} is a valid |
7375 | punctuation character for use in the @code{PRINT_OPERAND} macro. If | |
7376 | @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no | |
7377 | punctuation characters (except for the standard one, @samp{%}) are used | |
7378 | in this way. | |
a2c4f8e0 | 7379 | @end defmac |
feca2ed3 | 7380 | |
a2c4f8e0 | 7381 | @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x}) |
feca2ed3 JW |
7382 | A C compound statement to output to stdio stream @var{stream} the |
7383 | assembler syntax for an instruction operand that is a memory reference | |
7384 | whose address is @var{x}. @var{x} is an RTL expression. | |
7385 | ||
fb49053f | 7386 | @cindex @code{TARGET_ENCODE_SECTION_INFO} usage |
feca2ed3 | 7387 | On some machines, the syntax for a symbolic address depends on the |
fb49053f RH |
7388 | section that the address refers to. On these machines, define the hook |
7389 | @code{TARGET_ENCODE_SECTION_INFO} to store the information into the | |
a2c4f8e0 ZW |
7390 | @code{symbol_ref}, and then check for it here. @xref{Assembler |
7391 | Format}. | |
7392 | @end defmac | |
feca2ed3 | 7393 | |
feca2ed3 | 7394 | @findex dbr_sequence_length |
a2c4f8e0 | 7395 | @defmac DBR_OUTPUT_SEQEND (@var{file}) |
feca2ed3 JW |
7396 | A C statement, to be executed after all slot-filler instructions have |
7397 | been output. If necessary, call @code{dbr_sequence_length} to | |
7398 | determine the number of slots filled in a sequence (zero if not | |
7399 | currently outputting a sequence), to decide how many no-ops to output, | |
7400 | or whatever. | |
7401 | ||
7402 | Don't define this macro if it has nothing to do, but it is helpful in | |
7403 | reading assembly output if the extent of the delay sequence is made | |
e979f9e8 | 7404 | explicit (e.g.@: with white space). |
a2c4f8e0 | 7405 | @end defmac |
feca2ed3 JW |
7406 | |
7407 | @findex final_sequence | |
7408 | Note that output routines for instructions with delay slots must be | |
e979f9e8 JM |
7409 | prepared to deal with not being output as part of a sequence |
7410 | (i.e.@: when the scheduling pass is not run, or when no slot fillers could be | |
feca2ed3 JW |
7411 | found.) The variable @code{final_sequence} is null when not |
7412 | processing a sequence, otherwise it contains the @code{sequence} rtx | |
7413 | being output. | |
7414 | ||
feca2ed3 | 7415 | @findex asm_fprintf |
a2c4f8e0 ZW |
7416 | @defmac REGISTER_PREFIX |
7417 | @defmacx LOCAL_LABEL_PREFIX | |
7418 | @defmacx USER_LABEL_PREFIX | |
7419 | @defmacx IMMEDIATE_PREFIX | |
feca2ed3 JW |
7420 | If defined, C string expressions to be used for the @samp{%R}, @samp{%L}, |
7421 | @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see | |
7422 | @file{final.c}). These are useful when a single @file{md} file must | |
7423 | support multiple assembler formats. In that case, the various @file{tm.h} | |
7424 | files can define these macros differently. | |
a2c4f8e0 | 7425 | @end defmac |
feca2ed3 | 7426 | |
a2c4f8e0 | 7427 | @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format}) |
3b7a2e58 | 7428 | If defined this macro should expand to a series of @code{case} |
fe0503ea NC |
7429 | statements which will be parsed inside the @code{switch} statement of |
7430 | the @code{asm_fprintf} function. This allows targets to define extra | |
7431 | printf formats which may useful when generating their assembler | |
4bd0bee9 | 7432 | statements. Note that uppercase letters are reserved for future |
fe0503ea NC |
7433 | generic extensions to asm_fprintf, and so are not available to target |
7434 | specific code. The output file is given by the parameter @var{file}. | |
7435 | The varargs input pointer is @var{argptr} and the rest of the format | |
7436 | string, starting the character after the one that is being switched | |
7437 | upon, is pointed to by @var{format}. | |
a2c4f8e0 | 7438 | @end defmac |
fe0503ea | 7439 | |
a2c4f8e0 | 7440 | @defmac ASSEMBLER_DIALECT |
feca2ed3 JW |
7441 | If your target supports multiple dialects of assembler language (such as |
7442 | different opcodes), define this macro as a C expression that gives the | |
7443 | numeric index of the assembler language dialect to use, with zero as the | |
7444 | first variant. | |
7445 | ||
7446 | If this macro is defined, you may use constructs of the form | |
c237e94a | 7447 | @smallexample |
f282ffb3 | 7448 | @samp{@{option0|option1|option2@dots{}@}} |
c237e94a ZW |
7449 | @end smallexample |
7450 | @noindent | |
7451 | in the output templates of patterns (@pxref{Output Template}) or in the | |
7452 | first argument of @code{asm_fprintf}. This construct outputs | |
7453 | @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of | |
7454 | @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters | |
7455 | within these strings retain their usual meaning. If there are fewer | |
7456 | alternatives within the braces than the value of | |
7457 | @code{ASSEMBLER_DIALECT}, the construct outputs nothing. | |
feca2ed3 JW |
7458 | |
7459 | If you do not define this macro, the characters @samp{@{}, @samp{|} and | |
7460 | @samp{@}} do not have any special meaning when used in templates or | |
7461 | operands to @code{asm_fprintf}. | |
7462 | ||
7463 | Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX}, | |
7464 | @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express | |
e5e809f4 | 7465 | the variations in assembler language syntax with that mechanism. Define |
feca2ed3 JW |
7466 | @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax |
7467 | if the syntax variant are larger and involve such things as different | |
7468 | opcodes or operand order. | |
a2c4f8e0 | 7469 | @end defmac |
feca2ed3 | 7470 | |
a2c4f8e0 | 7471 | @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno}) |
feca2ed3 JW |
7472 | A C expression to output to @var{stream} some assembler code |
7473 | which will push hard register number @var{regno} onto the stack. | |
7474 | The code need not be optimal, since this macro is used only when | |
7475 | profiling. | |
a2c4f8e0 | 7476 | @end defmac |
feca2ed3 | 7477 | |
a2c4f8e0 | 7478 | @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno}) |
feca2ed3 JW |
7479 | A C expression to output to @var{stream} some assembler code |
7480 | which will pop hard register number @var{regno} off of the stack. | |
7481 | The code need not be optimal, since this macro is used only when | |
7482 | profiling. | |
a2c4f8e0 | 7483 | @end defmac |
feca2ed3 JW |
7484 | |
7485 | @node Dispatch Tables | |
7486 | @subsection Output of Dispatch Tables | |
7487 | ||
7488 | @c prevent bad page break with this line | |
7489 | This concerns dispatch tables. | |
7490 | ||
feca2ed3 | 7491 | @cindex dispatch table |
a2c4f8e0 | 7492 | @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel}) |
feca2ed3 JW |
7493 | A C statement to output to the stdio stream @var{stream} an assembler |
7494 | pseudo-instruction to generate a difference between two labels. | |
7495 | @var{value} and @var{rel} are the numbers of two internal labels. The | |
7496 | definitions of these labels are output using | |
4977bab6 | 7497 | @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same |
feca2ed3 JW |
7498 | way here. For example, |
7499 | ||
3ab51846 | 7500 | @smallexample |
feca2ed3 JW |
7501 | fprintf (@var{stream}, "\t.word L%d-L%d\n", |
7502 | @var{value}, @var{rel}) | |
3ab51846 | 7503 | @end smallexample |
feca2ed3 JW |
7504 | |
7505 | You must provide this macro on machines where the addresses in a | |
f0523f02 | 7506 | dispatch table are relative to the table's own address. If defined, GCC |
161d7b59 | 7507 | will also use this macro on all machines when producing PIC@. |
aee96fe9 | 7508 | @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the |
33f7f353 | 7509 | mode and flags can be read. |
a2c4f8e0 | 7510 | @end defmac |
feca2ed3 | 7511 | |
a2c4f8e0 | 7512 | @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value}) |
feca2ed3 JW |
7513 | This macro should be provided on machines where the addresses |
7514 | in a dispatch table are absolute. | |
7515 | ||
7516 | The definition should be a C statement to output to the stdio stream | |
7517 | @var{stream} an assembler pseudo-instruction to generate a reference to | |
7518 | a label. @var{value} is the number of an internal label whose | |
4977bab6 | 7519 | definition is output using @code{(*targetm.asm_out.internal_label)}. |
feca2ed3 JW |
7520 | For example, |
7521 | ||
3ab51846 | 7522 | @smallexample |
feca2ed3 | 7523 | fprintf (@var{stream}, "\t.word L%d\n", @var{value}) |
3ab51846 | 7524 | @end smallexample |
a2c4f8e0 | 7525 | @end defmac |
feca2ed3 | 7526 | |
a2c4f8e0 | 7527 | @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table}) |
feca2ed3 JW |
7528 | Define this if the label before a jump-table needs to be output |
7529 | specially. The first three arguments are the same as for | |
4977bab6 | 7530 | @code{(*targetm.asm_out.internal_label)}; the fourth argument is the |
feca2ed3 JW |
7531 | jump-table which follows (a @code{jump_insn} containing an |
7532 | @code{addr_vec} or @code{addr_diff_vec}). | |
7533 | ||
7534 | This feature is used on system V to output a @code{swbeg} statement | |
7535 | for the table. | |
7536 | ||
7537 | If this macro is not defined, these labels are output with | |
4977bab6 | 7538 | @code{(*targetm.asm_out.internal_label)}. |
a2c4f8e0 | 7539 | @end defmac |
feca2ed3 | 7540 | |
a2c4f8e0 | 7541 | @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table}) |
feca2ed3 JW |
7542 | Define this if something special must be output at the end of a |
7543 | jump-table. The definition should be a C statement to be executed | |
7544 | after the assembler code for the table is written. It should write | |
7545 | the appropriate code to stdio stream @var{stream}. The argument | |
7546 | @var{table} is the jump-table insn, and @var{num} is the label-number | |
7547 | of the preceding label. | |
7548 | ||
7549 | If this macro is not defined, nothing special is output at the end of | |
7550 | the jump-table. | |
a2c4f8e0 | 7551 | @end defmac |
feca2ed3 | 7552 | |
eeab4d81 | 7553 | @deftypefn {Target Hook} void TARGET_ASM_EMIT_UNWIND_LABEL (@var{stream}, @var{decl}, @var{for_eh}, @var{empty}) |
8a36672b | 7554 | This target hook emits a label at the beginning of each FDE@. It |
4746cf84 MA |
7555 | should be defined on targets where FDEs need special labels, and it |
7556 | should write the appropriate label, for the FDE associated with the | |
7557 | function declaration @var{decl}, to the stdio stream @var{stream}. | |
eeab4d81 MS |
7558 | The third argument, @var{for_eh}, is a boolean: true if this is for an |
7559 | exception table. The fourth argument, @var{empty}, is a boolean: | |
8a36672b | 7560 | true if this is a placeholder label for an omitted FDE@. |
4746cf84 MA |
7561 | |
7562 | The default is that FDEs are not given nonlocal labels. | |
7563 | @end deftypefn | |
7564 | ||
951120ea PB |
7565 | @deftypefn {Taget Hook} void TARGET_UNWIND_EMIT (FILE * @var{stream}, rtx @var{insn}) |
7566 | This target hook emits and assembly directives required to unwind the | |
7567 | given instruction. This is only used when TARGET_UNWIND_INFO is set. | |
7568 | @end deftypefn | |
7569 | ||
02f52e19 | 7570 | @node Exception Region Output |
feca2ed3 JW |
7571 | @subsection Assembler Commands for Exception Regions |
7572 | ||
7573 | @c prevent bad page break with this line | |
7574 | ||
7575 | This describes commands marking the start and the end of an exception | |
7576 | region. | |
7577 | ||
a2c4f8e0 | 7578 | @defmac EH_FRAME_SECTION_NAME |
7c262518 RH |
7579 | If defined, a C string constant for the name of the section containing |
7580 | exception handling frame unwind information. If not defined, GCC will | |
7581 | provide a default definition if the target supports named sections. | |
7582 | @file{crtstuff.c} uses this macro to switch to the appropriate section. | |
0021b564 JM |
7583 | |
7584 | You should define this symbol if your target supports DWARF 2 frame | |
7585 | unwind information and the default definition does not work. | |
a2c4f8e0 | 7586 | @end defmac |
0021b564 | 7587 | |
a2c4f8e0 | 7588 | @defmac EH_FRAME_IN_DATA_SECTION |
02c9b1ca RH |
7589 | If defined, DWARF 2 frame unwind information will be placed in the |
7590 | data section even though the target supports named sections. This | |
7591 | might be necessary, for instance, if the system linker does garbage | |
7592 | collection and sections cannot be marked as not to be collected. | |
7593 | ||
7594 | Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is | |
7595 | also defined. | |
a2c4f8e0 | 7596 | @end defmac |
02c9b1ca | 7597 | |
1a35e62d MM |
7598 | @defmac EH_TABLES_CAN_BE_READ_ONLY |
7599 | Define this macro to 1 if your target is such that no frame unwind | |
7600 | information encoding used with non-PIC code will ever require a | |
7601 | runtime relocation, but the linker may not support merging read-only | |
7602 | and read-write sections into a single read-write section. | |
7603 | @end defmac | |
7604 | ||
a2c4f8e0 | 7605 | @defmac MASK_RETURN_ADDR |
aee96fe9 | 7606 | An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so |
feca2ed3 | 7607 | that it does not contain any extraneous set bits in it. |
a2c4f8e0 | 7608 | @end defmac |
0021b564 | 7609 | |
a2c4f8e0 | 7610 | @defmac DWARF2_UNWIND_INFO |
0021b564 JM |
7611 | Define this macro to 0 if your target supports DWARF 2 frame unwind |
7612 | information, but it does not yet work with exception handling. | |
7613 | Otherwise, if your target supports this information (if it defines | |
7614 | @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP} | |
7615 | or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of | |
7616 | 1. | |
7617 | ||
5cf58f28 PB |
7618 | If @code{TARGET_UNWIND_INFO} is defined, the target specific unwinder |
7619 | will be used in all cases. Defining this macro will enable the generation | |
7620 | of DWARF 2 frame debugging information. | |
0021b564 | 7621 | |
5cf58f28 PB |
7622 | If @code{TARGET_UNWIND_INFO} is not defined, and this macro is defined to 1, |
7623 | the DWARF 2 unwinder will be the default exception handling mechanism; | |
7624 | otherwise, @code{setjmp}/@code{longjmp} will be used by default. | |
a2c4f8e0 | 7625 | @end defmac |
0021b564 | 7626 | |
951120ea PB |
7627 | @defmac TARGET_UNWIND_INFO |
7628 | Define this macro if your target has ABI specified unwind tables. Usually | |
7629 | these will be output by @code{TARGET_UNWIND_EMIT}. | |
7630 | @end defmac | |
7631 | ||
617a1b71 PB |
7632 | @deftypevar {Target Hook} bool TARGET_UNWID_TABLES_DEFAULT |
7633 | This variable should be set to @code{true} if the target ABI requires unwinding | |
7634 | tables even when exceptions are not used. | |
7635 | @end deftypevar | |
7636 | ||
c14aea87 RO |
7637 | @defmac MUST_USE_SJLJ_EXCEPTIONS |
7638 | This macro need only be defined if @code{DWARF2_UNWIND_INFO} is | |
7639 | runtime-variable. In that case, @file{except.h} cannot correctly | |
7640 | determine the corresponding definition of | |
7641 | @code{MUST_USE_SJLJ_EXCEPTIONS}, so the target must provide it directly. | |
7642 | @end defmac | |
7643 | ||
a2c4f8e0 | 7644 | @defmac DWARF_CIE_DATA_ALIGNMENT |
27c35f4b HPN |
7645 | This macro need only be defined if the target might save registers in the |
7646 | function prologue at an offset to the stack pointer that is not aligned to | |
7647 | @code{UNITS_PER_WORD}. The definition should be the negative minimum | |
7648 | alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive | |
7649 | minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if | |
7650 | the target supports DWARF 2 frame unwind information. | |
a2c4f8e0 | 7651 | @end defmac |
feca2ed3 | 7652 | |
07c9d2eb SS |
7653 | @deftypefn {Target Hook} void TARGET_ASM_EXCEPTION_SECTION () |
7654 | If defined, a function that switches to the section in which the main | |
7655 | exception table is to be placed (@pxref{Sections}). The default is a | |
7656 | function that switches to a section named @code{.gcc_except_table} on | |
7657 | machines that support named sections via | |
7658 | @code{TARGET_ASM_NAMED_SECTION}, otherwise if @option{-fpic} or | |
7659 | @option{-fPIC} is in effect, the @code{data_section}, otherwise the | |
7660 | @code{readonly_data_section}. | |
7661 | @end deftypefn | |
7662 | ||
7663 | @deftypefn {Target Hook} void TARGET_ASM_EH_FRAME_SECTION () | |
7664 | If defined, a function that switches to the section in which the DWARF 2 | |
7665 | frame unwind information to be placed (@pxref{Sections}). The default | |
7666 | is a function that outputs a standard GAS section directive, if | |
7667 | @code{EH_FRAME_SECTION_NAME} is defined, or else a data section | |
7668 | directive followed by a synthetic label. | |
7669 | @end deftypefn | |
7670 | ||
7606e68f SS |
7671 | @deftypevar {Target Hook} bool TARGET_TERMINATE_DW2_EH_FRAME_INFO |
7672 | Contains the value true if the target should add a zero word onto the | |
7673 | end of a Dwarf-2 frame info section when used for exception handling. | |
7674 | Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and | |
7675 | true otherwise. | |
7676 | @end deftypevar | |
7677 | ||
96714395 AH |
7678 | @deftypefn {Target Hook} rtx TARGET_DWARF_REGISTER_SPAN (rtx @var{reg}) |
7679 | Given a register, this hook should return a parallel of registers to | |
7680 | represent where to find the register pieces. Define this hook if the | |
7681 | register and its mode are represented in Dwarf in non-contiguous | |
7682 | locations, or if the register should be represented in more than one | |
7683 | register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}. | |
7684 | If not defined, the default is to return @code{NULL_RTX}. | |
7685 | @end deftypefn | |
7686 | ||
617a1b71 PB |
7687 | @deftypefn {Target Hook} bool TARGET_ASM_TTYPE (rtx @var{sym}) |
7688 | This hook is used to output a reference from a frame unwinding table to | |
7689 | the type_info object identified by @var{sym}. It should return @code{true} | |
7690 | if the reference was output. Returning @code{false} will cause the | |
7691 | reference to be output using the normal Dwarf2 routines. | |
7692 | @end deftypefn | |
7693 | ||
7694 | @deftypefn {Target Hook} bool TARGET_ARM_EABI_UNWINDER | |
7695 | This hook should be set to @code{true} on targets that use an ARM EABI | |
7696 | based unwinding library, and @code{false} on other targets. This effects | |
7697 | the format of unwinding tables, and how the unwinder in entered after | |
7698 | running a cleanup. The default is @code{false}. | |
7699 | @end deftypefn | |
7700 | ||
feca2ed3 JW |
7701 | @node Alignment Output |
7702 | @subsection Assembler Commands for Alignment | |
7703 | ||
7704 | @c prevent bad page break with this line | |
7705 | This describes commands for alignment. | |
7706 | ||
a2c4f8e0 | 7707 | @defmac JUMP_ALIGN (@var{label}) |
247a370b | 7708 | The alignment (log base 2) to put in front of @var{label}, which is |
f710504c | 7709 | a common destination of jumps and has no fallthru incoming edge. |
25e22dc0 JH |
7710 | |
7711 | This macro need not be defined if you don't want any special alignment | |
7712 | to be done at such a time. Most machine descriptions do not currently | |
7713 | define the macro. | |
efa3896a | 7714 | |
3446405d JH |
7715 | Unless it's necessary to inspect the @var{label} parameter, it is better |
7716 | to set the variable @var{align_jumps} in the target's | |
c21cd8b1 | 7717 | @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's |
247a370b | 7718 | selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation. |
a2c4f8e0 | 7719 | @end defmac |
247a370b | 7720 | |
a2c4f8e0 | 7721 | @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label}) |
247a370b JH |
7722 | The alignment (log base 2) to put in front of @var{label}, which follows |
7723 | a @code{BARRIER}. | |
7724 | ||
7725 | This macro need not be defined if you don't want any special alignment | |
7726 | to be done at such a time. Most machine descriptions do not currently | |
7727 | define the macro. | |
a2c4f8e0 | 7728 | @end defmac |
3446405d | 7729 | |
a2c4f8e0 | 7730 | @defmac LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP |
02f52e19 | 7731 | The maximum number of bytes to skip when applying |
efa3896a GK |
7732 | @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if |
7733 | @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined. | |
a2c4f8e0 | 7734 | @end defmac |
efa3896a | 7735 | |
a2c4f8e0 | 7736 | @defmac LOOP_ALIGN (@var{label}) |
fc470718 | 7737 | The alignment (log base 2) to put in front of @var{label}, which follows |
aee96fe9 | 7738 | a @code{NOTE_INSN_LOOP_BEG} note. |
feca2ed3 JW |
7739 | |
7740 | This macro need not be defined if you don't want any special alignment | |
7741 | to be done at such a time. Most machine descriptions do not currently | |
7742 | define the macro. | |
7743 | ||
efa3896a | 7744 | Unless it's necessary to inspect the @var{label} parameter, it is better |
aee96fe9 | 7745 | to set the variable @code{align_loops} in the target's |
c21cd8b1 | 7746 | @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's |
aee96fe9 | 7747 | selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation. |
a2c4f8e0 | 7748 | @end defmac |
efa3896a | 7749 | |
a2c4f8e0 | 7750 | @defmac LOOP_ALIGN_MAX_SKIP |
efa3896a GK |
7751 | The maximum number of bytes to skip when applying @code{LOOP_ALIGN}. |
7752 | This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined. | |
a2c4f8e0 | 7753 | @end defmac |
efa3896a | 7754 | |
a2c4f8e0 | 7755 | @defmac LABEL_ALIGN (@var{label}) |
fc470718 | 7756 | The alignment (log base 2) to put in front of @var{label}. |
aee96fe9 | 7757 | If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment, |
fc470718 R |
7758 | the maximum of the specified values is used. |
7759 | ||
efa3896a | 7760 | Unless it's necessary to inspect the @var{label} parameter, it is better |
aee96fe9 | 7761 | to set the variable @code{align_labels} in the target's |
c21cd8b1 | 7762 | @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's |
aee96fe9 | 7763 | selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation. |
a2c4f8e0 | 7764 | @end defmac |
efa3896a | 7765 | |
a2c4f8e0 | 7766 | @defmac LABEL_ALIGN_MAX_SKIP |
efa3896a GK |
7767 | The maximum number of bytes to skip when applying @code{LABEL_ALIGN}. |
7768 | This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined. | |
a2c4f8e0 | 7769 | @end defmac |
efa3896a | 7770 | |
a2c4f8e0 | 7771 | @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes}) |
feca2ed3 JW |
7772 | A C statement to output to the stdio stream @var{stream} an assembler |
7773 | instruction to advance the location counter by @var{nbytes} bytes. | |
7774 | Those bytes should be zero when loaded. @var{nbytes} will be a C | |
7775 | expression of type @code{int}. | |
a2c4f8e0 | 7776 | @end defmac |
feca2ed3 | 7777 | |
a2c4f8e0 | 7778 | @defmac ASM_NO_SKIP_IN_TEXT |
feca2ed3 | 7779 | Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the |
556e0f21 | 7780 | text section because it fails to put zeros in the bytes that are skipped. |
feca2ed3 JW |
7781 | This is true on many Unix systems, where the pseudo--op to skip bytes |
7782 | produces no-op instructions rather than zeros when used in the text | |
7783 | section. | |
a2c4f8e0 | 7784 | @end defmac |
feca2ed3 | 7785 | |
a2c4f8e0 | 7786 | @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power}) |
feca2ed3 JW |
7787 | A C statement to output to the stdio stream @var{stream} an assembler |
7788 | command to advance the location counter to a multiple of 2 to the | |
7789 | @var{power} bytes. @var{power} will be a C expression of type @code{int}. | |
a2c4f8e0 | 7790 | @end defmac |
26f63a77 | 7791 | |
a2c4f8e0 | 7792 | @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power}) |
8e16ab99 SF |
7793 | Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used |
7794 | for padding, if necessary. | |
a2c4f8e0 | 7795 | @end defmac |
8e16ab99 | 7796 | |
a2c4f8e0 | 7797 | @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip}) |
26f63a77 JL |
7798 | A C statement to output to the stdio stream @var{stream} an assembler |
7799 | command to advance the location counter to a multiple of 2 to the | |
7800 | @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to | |
7801 | satisfy the alignment request. @var{power} and @var{max_skip} will be | |
7802 | a C expression of type @code{int}. | |
a2c4f8e0 | 7803 | @end defmac |
feca2ed3 JW |
7804 | |
7805 | @need 3000 | |
7806 | @node Debugging Info | |
7807 | @section Controlling Debugging Information Format | |
7808 | ||
7809 | @c prevent bad page break with this line | |
7810 | This describes how to specify debugging information. | |
7811 | ||
7812 | @menu | |
7813 | * All Debuggers:: Macros that affect all debugging formats uniformly. | |
7814 | * DBX Options:: Macros enabling specific options in DBX format. | |
7815 | * DBX Hooks:: Hook macros for varying DBX format. | |
7816 | * File Names and DBX:: Macros controlling output of file names in DBX format. | |
7817 | * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats. | |
5f98259a | 7818 | * VMS Debug:: Macros for VMS debug format. |
feca2ed3 JW |
7819 | @end menu |
7820 | ||
7821 | @node All Debuggers | |
7822 | @subsection Macros Affecting All Debugging Formats | |
7823 | ||
7824 | @c prevent bad page break with this line | |
7825 | These macros affect all debugging formats. | |
7826 | ||
a2c4f8e0 | 7827 | @defmac DBX_REGISTER_NUMBER (@var{regno}) |
feca2ed3 | 7828 | A C expression that returns the DBX register number for the compiler |
4617e3b5 KG |
7829 | register number @var{regno}. In the default macro provided, the value |
7830 | of this expression will be @var{regno} itself. But sometimes there are | |
7831 | some registers that the compiler knows about and DBX does not, or vice | |
7832 | versa. In such cases, some register may need to have one number in the | |
7833 | compiler and another for DBX@. | |
feca2ed3 | 7834 | |
a3a15b4d | 7835 | If two registers have consecutive numbers inside GCC, and they can be |
feca2ed3 JW |
7836 | used as a pair to hold a multiword value, then they @emph{must} have |
7837 | consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}. | |
7838 | Otherwise, debuggers will be unable to access such a pair, because they | |
7839 | expect register pairs to be consecutive in their own numbering scheme. | |
7840 | ||
7841 | If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that | |
7842 | does not preserve register pairs, then what you must do instead is | |
7843 | redefine the actual register numbering scheme. | |
a2c4f8e0 | 7844 | @end defmac |
feca2ed3 | 7845 | |
a2c4f8e0 | 7846 | @defmac DEBUGGER_AUTO_OFFSET (@var{x}) |
feca2ed3 JW |
7847 | A C expression that returns the integer offset value for an automatic |
7848 | variable having address @var{x} (an RTL expression). The default | |
7849 | computation assumes that @var{x} is based on the frame-pointer and | |
7850 | gives the offset from the frame-pointer. This is required for targets | |
7851 | that produce debugging output for DBX or COFF-style debugging output | |
7852 | for SDB and allow the frame-pointer to be eliminated when the | |
630d3d5a | 7853 | @option{-g} options is used. |
a2c4f8e0 | 7854 | @end defmac |
feca2ed3 | 7855 | |
a2c4f8e0 | 7856 | @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x}) |
feca2ed3 JW |
7857 | A C expression that returns the integer offset value for an argument |
7858 | having address @var{x} (an RTL expression). The nominal offset is | |
7859 | @var{offset}. | |
a2c4f8e0 | 7860 | @end defmac |
feca2ed3 | 7861 | |
a2c4f8e0 | 7862 | @defmac PREFERRED_DEBUGGING_TYPE |
a3a15b4d | 7863 | A C expression that returns the type of debugging output GCC should |
630d3d5a | 7864 | produce when the user specifies just @option{-g}. Define |
a3a15b4d | 7865 | this if you have arranged for GCC to support more than one format of |
e5e809f4 | 7866 | debugging output. Currently, the allowable values are @code{DBX_DEBUG}, |
5f98259a RK |
7867 | @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG}, |
7868 | @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}. | |
feca2ed3 | 7869 | |
630d3d5a | 7870 | When the user specifies @option{-ggdb}, GCC normally also uses the |
e5e809f4 | 7871 | value of this macro to select the debugging output format, but with two |
16201823 | 7872 | exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the |
e5e809f4 | 7873 | value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is |
a3a15b4d | 7874 | defined, GCC uses @code{DBX_DEBUG}. |
deabc777 | 7875 | |
feca2ed3 | 7876 | The value of this macro only affects the default debugging output; the |
630d3d5a | 7877 | user can always get a specific type of output by using @option{-gstabs}, |
def66b10 | 7878 | @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}. |
a2c4f8e0 | 7879 | @end defmac |
feca2ed3 JW |
7880 | |
7881 | @node DBX Options | |
7882 | @subsection Specific Options for DBX Output | |
7883 | ||
7884 | @c prevent bad page break with this line | |
7885 | These are specific options for DBX output. | |
7886 | ||
a2c4f8e0 | 7887 | @defmac DBX_DEBUGGING_INFO |
a3a15b4d | 7888 | Define this macro if GCC should produce debugging output for DBX |
630d3d5a | 7889 | in response to the @option{-g} option. |
a2c4f8e0 | 7890 | @end defmac |
feca2ed3 | 7891 | |
a2c4f8e0 | 7892 | @defmac XCOFF_DEBUGGING_INFO |
a3a15b4d | 7893 | Define this macro if GCC should produce XCOFF format debugging output |
630d3d5a | 7894 | in response to the @option{-g} option. This is a variant of DBX format. |
a2c4f8e0 | 7895 | @end defmac |
feca2ed3 | 7896 | |
a2c4f8e0 | 7897 | @defmac DEFAULT_GDB_EXTENSIONS |
a3a15b4d | 7898 | Define this macro to control whether GCC should by default generate |
feca2ed3 JW |
7899 | GDB's extended version of DBX debugging information (assuming DBX-format |
7900 | debugging information is enabled at all). If you don't define the | |
7901 | macro, the default is 1: always generate the extended information | |
7902 | if there is any occasion to. | |
a2c4f8e0 | 7903 | @end defmac |
feca2ed3 | 7904 | |
a2c4f8e0 | 7905 | @defmac DEBUG_SYMS_TEXT |
feca2ed3 JW |
7906 | Define this macro if all @code{.stabs} commands should be output while |
7907 | in the text section. | |
a2c4f8e0 | 7908 | @end defmac |
feca2ed3 | 7909 | |
a2c4f8e0 | 7910 | @defmac ASM_STABS_OP |
047c1c92 HPN |
7911 | A C string constant, including spacing, naming the assembler pseudo op to |
7912 | use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol. | |
7913 | If you don't define this macro, @code{"\t.stabs\t"} is used. This macro | |
7914 | applies only to DBX debugging information format. | |
a2c4f8e0 | 7915 | @end defmac |
feca2ed3 | 7916 | |
a2c4f8e0 | 7917 | @defmac ASM_STABD_OP |
047c1c92 HPN |
7918 | A C string constant, including spacing, naming the assembler pseudo op to |
7919 | use instead of @code{"\t.stabd\t"} to define a debugging symbol whose | |
7920 | value is the current location. If you don't define this macro, | |
7921 | @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging | |
7922 | information format. | |
a2c4f8e0 | 7923 | @end defmac |
feca2ed3 | 7924 | |
a2c4f8e0 | 7925 | @defmac ASM_STABN_OP |
047c1c92 HPN |
7926 | A C string constant, including spacing, naming the assembler pseudo op to |
7927 | use instead of @code{"\t.stabn\t"} to define a debugging symbol with no | |
7928 | name. If you don't define this macro, @code{"\t.stabn\t"} is used. This | |
7929 | macro applies only to DBX debugging information format. | |
a2c4f8e0 | 7930 | @end defmac |
feca2ed3 | 7931 | |
a2c4f8e0 | 7932 | @defmac DBX_NO_XREFS |
feca2ed3 JW |
7933 | Define this macro if DBX on your system does not support the construct |
7934 | @samp{xs@var{tagname}}. On some systems, this construct is used to | |
7935 | describe a forward reference to a structure named @var{tagname}. | |
7936 | On other systems, this construct is not supported at all. | |
a2c4f8e0 | 7937 | @end defmac |
feca2ed3 | 7938 | |
a2c4f8e0 | 7939 | @defmac DBX_CONTIN_LENGTH |
feca2ed3 JW |
7940 | A symbol name in DBX-format debugging information is normally |
7941 | continued (split into two separate @code{.stabs} directives) when it | |
7942 | exceeds a certain length (by default, 80 characters). On some | |
7943 | operating systems, DBX requires this splitting; on others, splitting | |
7944 | must not be done. You can inhibit splitting by defining this macro | |
7945 | with the value zero. You can override the default splitting-length by | |
7946 | defining this macro as an expression for the length you desire. | |
a2c4f8e0 | 7947 | @end defmac |
feca2ed3 | 7948 | |
a2c4f8e0 | 7949 | @defmac DBX_CONTIN_CHAR |
feca2ed3 JW |
7950 | Normally continuation is indicated by adding a @samp{\} character to |
7951 | the end of a @code{.stabs} string when a continuation follows. To use | |
7952 | a different character instead, define this macro as a character | |
7953 | constant for the character you want to use. Do not define this macro | |
7954 | if backslash is correct for your system. | |
a2c4f8e0 | 7955 | @end defmac |
feca2ed3 | 7956 | |
a2c4f8e0 | 7957 | @defmac DBX_STATIC_STAB_DATA_SECTION |
feca2ed3 JW |
7958 | Define this macro if it is necessary to go to the data section before |
7959 | outputting the @samp{.stabs} pseudo-op for a non-global static | |
7960 | variable. | |
a2c4f8e0 | 7961 | @end defmac |
feca2ed3 | 7962 | |
a2c4f8e0 | 7963 | @defmac DBX_TYPE_DECL_STABS_CODE |
feca2ed3 JW |
7964 | The value to use in the ``code'' field of the @code{.stabs} directive |
7965 | for a typedef. The default is @code{N_LSYM}. | |
a2c4f8e0 | 7966 | @end defmac |
feca2ed3 | 7967 | |
a2c4f8e0 | 7968 | @defmac DBX_STATIC_CONST_VAR_CODE |
feca2ed3 JW |
7969 | The value to use in the ``code'' field of the @code{.stabs} directive |
7970 | for a static variable located in the text section. DBX format does not | |
7971 | provide any ``right'' way to do this. The default is @code{N_FUN}. | |
a2c4f8e0 | 7972 | @end defmac |
feca2ed3 | 7973 | |
a2c4f8e0 | 7974 | @defmac DBX_REGPARM_STABS_CODE |
feca2ed3 JW |
7975 | The value to use in the ``code'' field of the @code{.stabs} directive |
7976 | for a parameter passed in registers. DBX format does not provide any | |
7977 | ``right'' way to do this. The default is @code{N_RSYM}. | |
a2c4f8e0 | 7978 | @end defmac |
feca2ed3 | 7979 | |
a2c4f8e0 | 7980 | @defmac DBX_REGPARM_STABS_LETTER |
feca2ed3 JW |
7981 | The letter to use in DBX symbol data to identify a symbol as a parameter |
7982 | passed in registers. DBX format does not customarily provide any way to | |
7983 | do this. The default is @code{'P'}. | |
a2c4f8e0 | 7984 | @end defmac |
feca2ed3 | 7985 | |
a2c4f8e0 | 7986 | @defmac DBX_FUNCTION_FIRST |
feca2ed3 JW |
7987 | Define this macro if the DBX information for a function and its |
7988 | arguments should precede the assembler code for the function. Normally, | |
7989 | in DBX format, the debugging information entirely follows the assembler | |
7990 | code. | |
a2c4f8e0 | 7991 | @end defmac |
feca2ed3 | 7992 | |
a2c4f8e0 | 7993 | @defmac DBX_BLOCKS_FUNCTION_RELATIVE |
3e487b21 ZW |
7994 | Define this macro, with value 1, if the value of a symbol describing |
7995 | the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be | |
7996 | relative to the start of the enclosing function. Normally, GCC uses | |
7997 | an absolute address. | |
7998 | @end defmac | |
7999 | ||
8000 | @defmac DBX_LINES_FUNCTION_RELATIVE | |
8001 | Define this macro, with value 1, if the value of a symbol indicating | |
8002 | the current line number (@code{N_SLINE}) should be relative to the | |
8003 | start of the enclosing function. Normally, GCC uses an absolute address. | |
a2c4f8e0 | 8004 | @end defmac |
feca2ed3 | 8005 | |
a2c4f8e0 | 8006 | @defmac DBX_USE_BINCL |
f0523f02 | 8007 | Define this macro if GCC should generate @code{N_BINCL} and |
feca2ed3 | 8008 | @code{N_EINCL} stabs for included header files, as on Sun systems. This |
f0523f02 JM |
8009 | macro also directs GCC to output a type number as a pair of a file |
8010 | number and a type number within the file. Normally, GCC does not | |
feca2ed3 JW |
8011 | generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single |
8012 | number for a type number. | |
a2c4f8e0 | 8013 | @end defmac |
feca2ed3 JW |
8014 | |
8015 | @node DBX Hooks | |
8016 | @subsection Open-Ended Hooks for DBX Format | |
8017 | ||
8018 | @c prevent bad page break with this line | |
8019 | These are hooks for DBX format. | |
8020 | ||
a2c4f8e0 | 8021 | @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name}) |
feca2ed3 JW |
8022 | Define this macro to say how to output to @var{stream} the debugging |
8023 | information for the start of a scope level for variable names. The | |
8024 | argument @var{name} is the name of an assembler symbol (for use with | |
8025 | @code{assemble_name}) whose value is the address where the scope begins. | |
a2c4f8e0 | 8026 | @end defmac |
feca2ed3 | 8027 | |
a2c4f8e0 | 8028 | @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name}) |
feca2ed3 | 8029 | Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level. |
a2c4f8e0 | 8030 | @end defmac |
feca2ed3 | 8031 | |
a2c4f8e0 | 8032 | @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl}) |
374b0b7d AM |
8033 | Define this macro if the target machine requires special handling to |
8034 | output an @code{N_FUN} entry for the function @var{decl}. | |
a2c4f8e0 | 8035 | @end defmac |
374b0b7d | 8036 | |
3e487b21 ZW |
8037 | @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter}) |
8038 | A C statement to output DBX debugging information before code for line | |
8039 | number @var{line} of the current source file to the stdio stream | |
8a36672b | 8040 | @var{stream}. @var{counter} is the number of time the macro was |
3e487b21 ZW |
8041 | invoked, including the current invocation; it is intended to generate |
8042 | unique labels in the assembly output. | |
8043 | ||
8044 | This macro should not be defined if the default output is correct, or | |
8045 | if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}. | |
8046 | @end defmac | |
8047 | ||
a2c4f8e0 | 8048 | @defmac NO_DBX_FUNCTION_END |
feca2ed3 | 8049 | Some stabs encapsulation formats (in particular ECOFF), cannot handle the |
c771326b | 8050 | @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct. |
feca2ed3 JW |
8051 | On those machines, define this macro to turn this feature off without |
8052 | disturbing the rest of the gdb extensions. | |
a2c4f8e0 | 8053 | @end defmac |
feca2ed3 | 8054 | |
5d865dac EB |
8055 | @defmac NO_DBX_BNSYM_ENSYM |
8056 | Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx | |
8057 | extension construct. On those machines, define this macro to turn this | |
8058 | feature off without disturbing the rest of the gdb extensions. | |
8059 | @end defmac | |
8060 | ||
feca2ed3 JW |
8061 | @node File Names and DBX |
8062 | @subsection File Names in DBX Format | |
8063 | ||
8064 | @c prevent bad page break with this line | |
8065 | This describes file names in DBX format. | |
8066 | ||
a2c4f8e0 | 8067 | @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name}) |
feca2ed3 | 8068 | A C statement to output DBX debugging information to the stdio stream |
93a27b7b | 8069 | @var{stream}, which indicates that file @var{name} is the main source |
feca2ed3 JW |
8070 | file---the file specified as the input file for compilation. |
8071 | This macro is called only once, at the beginning of compilation. | |
8072 | ||
8073 | This macro need not be defined if the standard form of output | |
8074 | for DBX debugging information is appropriate. | |
93a27b7b ZW |
8075 | |
8076 | It may be necessary to refer to a label equal to the beginning of the | |
8077 | text section. You can use @samp{assemble_name (stream, ltext_label_name)} | |
8078 | to do so. If you do this, you must also set the variable | |
8079 | @var{used_ltext_label_name} to @code{true}. | |
a2c4f8e0 | 8080 | @end defmac |
feca2ed3 | 8081 | |
93a27b7b ZW |
8082 | @defmac NO_DBX_MAIN_SOURCE_DIRECTORY |
8083 | Define this macro, with value 1, if GCC should not emit an indication | |
8084 | of the current directory for compilation and current source language at | |
8085 | the beginning of the file. | |
8086 | @end defmac | |
feca2ed3 | 8087 | |
93a27b7b ZW |
8088 | @defmac NO_DBX_GCC_MARKER |
8089 | Define this macro, with value 1, if GCC should not emit an indication | |
8090 | that this object file was compiled by GCC@. The default is to emit | |
8091 | an @code{N_OPT} stab at the beginning of every source file, with | |
8092 | @samp{gcc2_compiled.} for the string and value 0. | |
a2c4f8e0 | 8093 | @end defmac |
feca2ed3 | 8094 | |
a2c4f8e0 | 8095 | @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name}) |
feca2ed3 | 8096 | A C statement to output DBX debugging information at the end of |
93a27b7b ZW |
8097 | compilation of the main source file @var{name}. Output should be |
8098 | written to the stdio stream @var{stream}. | |
feca2ed3 JW |
8099 | |
8100 | If you don't define this macro, nothing special is output at the end | |
8101 | of compilation, which is correct for most machines. | |
a2c4f8e0 | 8102 | @end defmac |
feca2ed3 | 8103 | |
3e487b21 ZW |
8104 | @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END |
8105 | Define this macro @emph{instead of} defining | |
8106 | @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at | |
8107 | the end of compilation is a @code{N_SO} stab with an empty string, | |
8108 | whose value is the highest absolute text address in the file. | |
8109 | @end defmac | |
8110 | ||
feca2ed3 JW |
8111 | @need 2000 |
8112 | @node SDB and DWARF | |
8113 | @subsection Macros for SDB and DWARF Output | |
8114 | ||
8115 | @c prevent bad page break with this line | |
8116 | Here are macros for SDB and DWARF output. | |
8117 | ||
a2c4f8e0 | 8118 | @defmac SDB_DEBUGGING_INFO |
a3a15b4d | 8119 | Define this macro if GCC should produce COFF-style debugging output |
630d3d5a | 8120 | for SDB in response to the @option{-g} option. |
a2c4f8e0 | 8121 | @end defmac |
feca2ed3 | 8122 | |
a2c4f8e0 | 8123 | @defmac DWARF2_DEBUGGING_INFO |
a3a15b4d | 8124 | Define this macro if GCC should produce dwarf version 2 format |
630d3d5a | 8125 | debugging output in response to the @option{-g} option. |
f3ff3f4a | 8126 | |
a1c496cb EC |
8127 | @deftypefn {Target Hook} int TARGET_DWARF_CALLING_CONVENTION (tree @var{function}) |
8128 | Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to | |
8129 | be emitted for each function. Instead of an integer return the enum | |
8130 | value for the @code{DW_CC_} tag. | |
8131 | @end deftypefn | |
8132 | ||
861bb6c1 JL |
8133 | To support optional call frame debugging information, you must also |
8134 | define @code{INCOMING_RETURN_ADDR_RTX} and either set | |
8135 | @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the | |
8136 | prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save} | |
08c148a8 | 8137 | as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't. |
a2c4f8e0 | 8138 | @end defmac |
861bb6c1 | 8139 | |
a2c4f8e0 | 8140 | @defmac DWARF2_FRAME_INFO |
a3a15b4d | 8141 | Define this macro to a nonzero value if GCC should always output |
9ec36da5 | 8142 | Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO} |
a3a15b4d | 8143 | (@pxref{Exception Region Output} is nonzero, GCC will output this |
9ec36da5 | 8144 | information not matter how you define @code{DWARF2_FRAME_INFO}. |
a2c4f8e0 | 8145 | @end defmac |
9ec36da5 | 8146 | |
a2c4f8e0 | 8147 | @defmac DWARF2_ASM_LINE_DEBUG_INFO |
b2244e22 JW |
8148 | Define this macro to be a nonzero value if the assembler can generate Dwarf 2 |
8149 | line debug info sections. This will result in much more compact line number | |
8150 | tables, and hence is desirable if it works. | |
a2c4f8e0 | 8151 | @end defmac |
b2244e22 | 8152 | |
a2c4f8e0 | 8153 | @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2}) |
7606e68f SS |
8154 | A C statement to issue assembly directives that create a difference |
8155 | between the two given labels, using an integer of the given size. | |
a2c4f8e0 | 8156 | @end defmac |
7606e68f | 8157 | |
a2c4f8e0 | 8158 | @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}) |
7606e68f SS |
8159 | A C statement to issue assembly directives that create a |
8160 | section-relative reference to the given label, using an integer of the | |
8161 | given size. | |
a2c4f8e0 | 8162 | @end defmac |
7606e68f | 8163 | |
a2c4f8e0 | 8164 | @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label}) |
7606e68f SS |
8165 | A C statement to issue assembly directives that create a self-relative |
8166 | reference to the given label, using an integer of the given size. | |
a2c4f8e0 | 8167 | @end defmac |
7606e68f | 8168 | |
fdbe66f2 EB |
8169 | @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_DWARF_DTPREL (FILE *@var{FILE}, int @var{size}, rtx @var{x}) |
8170 | If defined, this target hook is a function which outputs a DTP-relative | |
8171 | reference to the given TLS symbol of the specified size. | |
8172 | @end deftypefn | |
8173 | ||
a2c4f8e0 | 8174 | @defmac PUT_SDB_@dots{} |
feca2ed3 JW |
8175 | Define these macros to override the assembler syntax for the special |
8176 | SDB assembler directives. See @file{sdbout.c} for a list of these | |
8177 | macros and their arguments. If the standard syntax is used, you need | |
8178 | not define them yourself. | |
a2c4f8e0 | 8179 | @end defmac |
feca2ed3 | 8180 | |
a2c4f8e0 | 8181 | @defmac SDB_DELIM |
feca2ed3 JW |
8182 | Some assemblers do not support a semicolon as a delimiter, even between |
8183 | SDB assembler directives. In that case, define this macro to be the | |
8184 | delimiter to use (usually @samp{\n}). It is not necessary to define | |
8185 | a new set of @code{PUT_SDB_@var{op}} macros if this is the only change | |
8186 | required. | |
a2c4f8e0 | 8187 | @end defmac |
feca2ed3 | 8188 | |
a2c4f8e0 | 8189 | @defmac SDB_ALLOW_UNKNOWN_REFERENCES |
feca2ed3 JW |
8190 | Define this macro to allow references to unknown structure, |
8191 | union, or enumeration tags to be emitted. Standard COFF does not | |
8192 | allow handling of unknown references, MIPS ECOFF has support for | |
8193 | it. | |
a2c4f8e0 | 8194 | @end defmac |
feca2ed3 | 8195 | |
a2c4f8e0 | 8196 | @defmac SDB_ALLOW_FORWARD_REFERENCES |
feca2ed3 JW |
8197 | Define this macro to allow references to structure, union, or |
8198 | enumeration tags that have not yet been seen to be handled. Some | |
8199 | assemblers choke if forward tags are used, while some require it. | |
a2c4f8e0 | 8200 | @end defmac |
feca2ed3 | 8201 | |
3e487b21 ZW |
8202 | @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}) |
8203 | A C statement to output SDB debugging information before code for line | |
8204 | number @var{line} of the current source file to the stdio stream | |
8205 | @var{stream}. The default is to emit an @code{.ln} directive. | |
8206 | @end defmac | |
8207 | ||
5f98259a RK |
8208 | @need 2000 |
8209 | @node VMS Debug | |
8210 | @subsection Macros for VMS Debug Format | |
8211 | ||
8212 | @c prevent bad page break with this line | |
8213 | Here are macros for VMS debug format. | |
8214 | ||
a2c4f8e0 | 8215 | @defmac VMS_DEBUGGING_INFO |
5f98259a RK |
8216 | Define this macro if GCC should produce debugging output for VMS |
8217 | in response to the @option{-g} option. The default behavior for VMS | |
8218 | is to generate minimal debug info for a traceback in the absence of | |
8219 | @option{-g} unless explicitly overridden with @option{-g0}. This | |
8220 | behavior is controlled by @code{OPTIMIZATION_OPTIONS} and | |
8221 | @code{OVERRIDE_OPTIONS}. | |
a2c4f8e0 | 8222 | @end defmac |
5f98259a | 8223 | |
b216cd4a | 8224 | @node Floating Point |
feca2ed3 JW |
8225 | @section Cross Compilation and Floating Point |
8226 | @cindex cross compilation and floating point | |
8227 | @cindex floating point and cross compilation | |
8228 | ||
b216cd4a | 8229 | While all modern machines use twos-complement representation for integers, |
feca2ed3 JW |
8230 | there are a variety of representations for floating point numbers. This |
8231 | means that in a cross-compiler the representation of floating point numbers | |
8232 | in the compiled program may be different from that used in the machine | |
8233 | doing the compilation. | |
8234 | ||
feca2ed3 | 8235 | Because different representation systems may offer different amounts of |
b216cd4a ZW |
8236 | range and precision, all floating point constants must be represented in |
8237 | the target machine's format. Therefore, the cross compiler cannot | |
8238 | safely use the host machine's floating point arithmetic; it must emulate | |
8239 | the target's arithmetic. To ensure consistency, GCC always uses | |
8240 | emulation to work with floating point values, even when the host and | |
8241 | target floating point formats are identical. | |
8242 | ||
8243 | The following macros are provided by @file{real.h} for the compiler to | |
8244 | use. All parts of the compiler which generate or optimize | |
ba31d94e ZW |
8245 | floating-point calculations must use these macros. They may evaluate |
8246 | their operands more than once, so operands must not have side effects. | |
feca2ed3 | 8247 | |
b216cd4a ZW |
8248 | @defmac REAL_VALUE_TYPE |
8249 | The C data type to be used to hold a floating point value in the target | |
8250 | machine's format. Typically this is a @code{struct} containing an | |
8251 | array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque | |
8252 | quantity. | |
8253 | @end defmac | |
8254 | ||
8255 | @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y}) | |
8256 | Compares for equality the two values, @var{x} and @var{y}. If the target | |
8257 | floating point format supports negative zeroes and/or NaNs, | |
8258 | @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and | |
8259 | @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false. | |
8260 | @end deftypefn | |
8261 | ||
8262 | @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y}) | |
8263 | Tests whether @var{x} is less than @var{y}. | |
8264 | @end deftypefn | |
8265 | ||
b216cd4a ZW |
8266 | @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x}) |
8267 | Truncates @var{x} to a signed integer, rounding toward zero. | |
8268 | @end deftypefn | |
8269 | ||
8270 | @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x}) | |
8271 | Truncates @var{x} to an unsigned integer, rounding toward zero. If | |
8272 | @var{x} is negative, returns zero. | |
8273 | @end deftypefn | |
8274 | ||
b216cd4a ZW |
8275 | @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode}) |
8276 | Converts @var{string} into a floating point number in the target machine's | |
8277 | representation for mode @var{mode}. This routine can handle both | |
8278 | decimal and hexadecimal floating point constants, using the syntax | |
8279 | defined by the C language for both. | |
8280 | @end deftypefn | |
feca2ed3 | 8281 | |
15e5ad76 | 8282 | @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x}) |
ce3649d2 | 8283 | Returns 1 if @var{x} is negative (including negative zero), 0 otherwise. |
15e5ad76 ZW |
8284 | @end deftypefn |
8285 | ||
b216cd4a ZW |
8286 | @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x}) |
8287 | Determines whether @var{x} represents infinity (positive or negative). | |
8288 | @end deftypefn | |
8289 | ||
8290 | @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x}) | |
8291 | Determines whether @var{x} represents a ``NaN'' (not-a-number). | |
8292 | @end deftypefn | |
8293 | ||
8294 | @deftypefn Macro void REAL_ARITHMETIC (REAL_VALUE_TYPE @var{output}, enum tree_code @var{code}, REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y}) | |
8295 | Calculates an arithmetic operation on the two floating point values | |
8296 | @var{x} and @var{y}, storing the result in @var{output} (which must be a | |
8297 | variable). | |
8298 | ||
8299 | The operation to be performed is specified by @var{code}. Only the | |
8300 | following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR}, | |
8301 | @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}. | |
8302 | ||
8303 | If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the | |
8304 | target's floating point format cannot represent infinity, it will call | |
8305 | @code{abort}. Callers should check for this situation first, using | |
8306 | @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}. | |
8307 | @end deftypefn | |
8308 | ||
8309 | @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x}) | |
8310 | Returns the negative of the floating point value @var{x}. | |
8311 | @end deftypefn | |
8312 | ||
15e5ad76 ZW |
8313 | @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x}) |
8314 | Returns the absolute value of @var{x}. | |
8315 | @end deftypefn | |
8316 | ||
b216cd4a ZW |
8317 | @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x}) |
8318 | Truncates the floating point value @var{x} to fit in @var{mode}. The | |
8319 | return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an | |
8320 | appropriate bit pattern to be output asa floating constant whose | |
8321 | precision accords with mode @var{mode}. | |
8322 | @end deftypefn | |
8323 | ||
8324 | @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x}) | |
8325 | Converts a floating point value @var{x} into a double-precision integer | |
8326 | which is then stored into @var{low} and @var{high}. If the value is not | |
8327 | integral, it is truncated. | |
8328 | @end deftypefn | |
8329 | ||
8330 | @deftypefn Macro void REAL_VALUE_FROM_INT (REAL_VALUE_TYPE @var{x}, HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, enum machine_mode @var{mode}) | |
b216cd4a ZW |
8331 | Converts a double-precision integer found in @var{low} and @var{high}, |
8332 | into a floating point value which is then stored into @var{x}. The | |
8333 | value is truncated to fit in mode @var{mode}. | |
8334 | @end deftypefn | |
feca2ed3 | 8335 | |
9f09b1f2 R |
8336 | @node Mode Switching |
8337 | @section Mode Switching Instructions | |
8338 | @cindex mode switching | |
8339 | The following macros control mode switching optimizations: | |
8340 | ||
a2c4f8e0 | 8341 | @defmac OPTIMIZE_MODE_SWITCHING (@var{entity}) |
9f09b1f2 R |
8342 | Define this macro if the port needs extra instructions inserted for mode |
8343 | switching in an optimizing compilation. | |
8344 | ||
8345 | For an example, the SH4 can perform both single and double precision | |
8346 | floating point operations, but to perform a single precision operation, | |
8347 | the FPSCR PR bit has to be cleared, while for a double precision | |
8348 | operation, this bit has to be set. Changing the PR bit requires a general | |
8349 | purpose register as a scratch register, hence these FPSCR sets have to | |
e979f9e8 | 8350 | be inserted before reload, i.e.@: you can't put this into instruction emitting |
18dbd950 | 8351 | or @code{TARGET_MACHINE_DEPENDENT_REORG}. |
9f09b1f2 R |
8352 | |
8353 | You can have multiple entities that are mode-switched, and select at run time | |
8354 | which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should | |
14976c58 | 8355 | return nonzero for any @var{entity} that needs mode-switching. |
9f09b1f2 R |
8356 | If you define this macro, you also have to define |
8357 | @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED}, | |
8358 | @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}. | |
73774972 EC |
8359 | @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT} |
8360 | are optional. | |
a2c4f8e0 | 8361 | @end defmac |
9f09b1f2 | 8362 | |
a2c4f8e0 | 8363 | @defmac NUM_MODES_FOR_MODE_SWITCHING |
9f09b1f2 R |
8364 | If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as |
8365 | initializer for an array of integers. Each initializer element | |
8366 | N refers to an entity that needs mode switching, and specifies the number | |
8367 | of different modes that might need to be set for this entity. | |
78466c0e JM |
8368 | The position of the initializer in the initializer---starting counting at |
8369 | zero---determines the integer that is used to refer to the mode-switched | |
9f09b1f2 R |
8370 | entity in question. |
8371 | In macros that take mode arguments / yield a mode result, modes are | |
630d3d5a | 8372 | represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode |
9f09b1f2 | 8373 | switch is needed / supplied. |
a2c4f8e0 | 8374 | @end defmac |
9f09b1f2 | 8375 | |
a2c4f8e0 | 8376 | @defmac MODE_NEEDED (@var{entity}, @var{insn}) |
9f09b1f2 R |
8377 | @var{entity} is an integer specifying a mode-switched entity. If |
8378 | @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to | |
8379 | return an integer value not larger than the corresponding element in | |
aee96fe9 JM |
8380 | @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must |
8381 | be switched into prior to the execution of @var{insn}. | |
a2c4f8e0 | 8382 | @end defmac |
9f09b1f2 | 8383 | |
73774972 EC |
8384 | @defmac MODE_AFTER (@var{mode}, @var{insn}) |
8385 | If this macro is defined, it is evaluated for every @var{insn} during | |
8a36672b | 8386 | mode switching. It determines the mode that an insn results in (if |
73774972 EC |
8387 | different from the incoming mode). |
8388 | @end defmac | |
8389 | ||
8390 | @defmac MODE_ENTRY (@var{entity}) | |
8391 | If this macro is defined, it is evaluated for every @var{entity} that needs | |
8a36672b JM |
8392 | mode switching. It should evaluate to an integer, which is a mode that |
8393 | @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY} | |
73774972 EC |
8394 | is defined then @code{MODE_EXIT} must be defined. |
8395 | @end defmac | |
8396 | ||
8397 | @defmac MODE_EXIT (@var{entity}) | |
9f09b1f2 | 8398 | If this macro is defined, it is evaluated for every @var{entity} that needs |
8a36672b JM |
8399 | mode switching. It should evaluate to an integer, which is a mode that |
8400 | @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT} | |
73774972 | 8401 | is defined then @code{MODE_ENTRY} must be defined. |
a2c4f8e0 | 8402 | @end defmac |
9f09b1f2 | 8403 | |
a2c4f8e0 | 8404 | @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n}) |
aee96fe9 JM |
8405 | This macro specifies the order in which modes for @var{entity} are processed. |
8406 | 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the | |
9f09b1f2 | 8407 | lowest. The value of the macro should be an integer designating a mode |
aee96fe9 | 8408 | for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode} |
630d3d5a | 8409 | (@var{entity}, @var{n}) shall be a bijection in 0 @dots{} |
aee96fe9 | 8410 | @code{num_modes_for_mode_switching[@var{entity}] - 1}. |
a2c4f8e0 | 8411 | @end defmac |
9f09b1f2 | 8412 | |
a2c4f8e0 | 8413 | @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live}) |
9f09b1f2 R |
8414 | Generate one or more insns to set @var{entity} to @var{mode}. |
8415 | @var{hard_reg_live} is the set of hard registers live at the point where | |
8416 | the insn(s) are to be inserted. | |
a2c4f8e0 | 8417 | @end defmac |
9f09b1f2 | 8418 | |
91d231cb JM |
8419 | @node Target Attributes |
8420 | @section Defining target-specific uses of @code{__attribute__} | |
8421 | @cindex target attributes | |
8422 | @cindex machine attributes | |
8423 | @cindex attributes, target-specific | |
8424 | ||
8425 | Target-specific attributes may be defined for functions, data and types. | |
8426 | These are described using the following target hooks; they also need to | |
8427 | be documented in @file{extend.texi}. | |
8428 | ||
8429 | @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE | |
8430 | If defined, this target hook points to an array of @samp{struct | |
8431 | attribute_spec} (defined in @file{tree.h}) specifying the machine | |
8432 | specific attributes for this target and some of the restrictions on the | |
8433 | entities to which these attributes are applied and the arguments they | |
8434 | take. | |
8435 | @end deftypevr | |
8436 | ||
8437 | @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2}) | |
8438 | If defined, this target hook is a function which returns zero if the attributes on | |
8439 | @var{type1} and @var{type2} are incompatible, one if they are compatible, | |
8440 | and two if they are nearly compatible (which causes a warning to be | |
8441 | generated). If this is not defined, machine-specific attributes are | |
8442 | supposed always to be compatible. | |
8443 | @end deftypefn | |
8444 | ||
8445 | @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type}) | |
8446 | If defined, this target hook is a function which assigns default attributes to | |
8447 | newly defined @var{type}. | |
8448 | @end deftypefn | |
8449 | ||
8450 | @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2}) | |
8451 | Define this target hook if the merging of type attributes needs special | |
8452 | handling. If defined, the result is a list of the combined | |
8453 | @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed | |
8454 | that @code{comptypes} has already been called and returned 1. This | |
8455 | function may call @code{merge_attributes} to handle machine-independent | |
8456 | merging. | |
8457 | @end deftypefn | |
8458 | ||
8459 | @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl}) | |
8460 | Define this target hook if the merging of decl attributes needs special | |
8461 | handling. If defined, the result is a list of the combined | |
8462 | @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}. | |
8463 | @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of | |
8464 | when this is needed are when one attribute overrides another, or when an | |
8465 | attribute is nullified by a subsequent definition. This function may | |
8466 | call @code{merge_attributes} to handle machine-independent merging. | |
8467 | ||
8468 | @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES | |
b2ca3702 MM |
8469 | If the only target-specific handling you require is @samp{dllimport} |
8470 | for Microsoft Windows targets, you should define the macro | |
8471 | @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler | |
8472 | will then define a function called | |
8473 | @code{merge_dllimport_decl_attributes} which can then be defined as | |
8474 | the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also | |
8475 | add @code{handle_dll_attribute} in the attribute table for your port | |
8476 | to perform initial processing of the @samp{dllimport} and | |
8477 | @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and | |
8478 | @file{i386/i386.c}, for example. | |
91d231cb JM |
8479 | @end deftypefn |
8480 | ||
63c5b495 | 8481 | @defmac TARGET_DECLSPEC |
1a141fe1 | 8482 | Define this macro to a nonzero value if you want to treat |
63c5b495 MM |
8483 | @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By |
8484 | default, this behavior is enabled only for targets that define | |
8485 | @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation | |
8486 | of @code{__declspec} is via a built-in macro, but you should not rely | |
8487 | on this implementation detail. | |
8488 | @end defmac | |
8489 | ||
91d231cb JM |
8490 | @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr}) |
8491 | Define this target hook if you want to be able to add attributes to a decl | |
8492 | when it is being created. This is normally useful for back ends which | |
8493 | wish to implement a pragma by using the attributes which correspond to | |
8494 | the pragma's effect. The @var{node} argument is the decl which is being | |
8495 | created. The @var{attr_ptr} argument is a pointer to the attribute list | |
8496 | for this decl. The list itself should not be modified, since it may be | |
8497 | shared with other decls, but attributes may be chained on the head of | |
8498 | the list and @code{*@var{attr_ptr}} modified to point to the new | |
8499 | attributes, or a copy of the list may be made if further changes are | |
8500 | needed. | |
8501 | @end deftypefn | |
8502 | ||
8503 | @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (tree @var{fndecl}) | |
8504 | @cindex inlining | |
8505 | This target hook returns @code{true} if it is ok to inline @var{fndecl} | |
8506 | into the current function, despite its having target-specific | |
8507 | attributes, @code{false} otherwise. By default, if a function has a | |
8508 | target specific attribute attached to it, it will not be inlined. | |
8509 | @end deftypefn | |
8510 | ||
d604bca3 MH |
8511 | @node MIPS Coprocessors |
8512 | @section Defining coprocessor specifics for MIPS targets. | |
8513 | @cindex MIPS coprocessor-definition macros | |
8514 | ||
8515 | The MIPS specification allows MIPS implementations to have as many as 4 | |
2dd76960 | 8516 | coprocessors, each with as many as 32 private registers. GCC supports |
d604bca3 MH |
8517 | accessing these registers and transferring values between the registers |
8518 | and memory using asm-ized variables. For example: | |
8519 | ||
8520 | @smallexample | |
8521 | register unsigned int cp0count asm ("c0r1"); | |
8522 | unsigned int d; | |
8523 | ||
8524 | d = cp0count + 3; | |
8525 | @end smallexample | |
8526 | ||
8527 | (``c0r1'' is the default name of register 1 in coprocessor 0; alternate | |
8528 | names may be added as described below, or the default names may be | |
8529 | overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.) | |
8530 | ||
8531 | Coprocessor registers are assumed to be epilogue-used; sets to them will | |
8532 | be preserved even if it does not appear that the register is used again | |
8533 | later in the function. | |
8534 | ||
8535 | Another note: according to the MIPS spec, coprocessor 1 (if present) is | |
8a36672b | 8536 | the FPU@. One accesses COP1 registers through standard mips |
d604bca3 MH |
8537 | floating-point support; they are not included in this mechanism. |
8538 | ||
8539 | There is one macro used in defining the MIPS coprocessor interface which | |
8540 | you may want to override in subtargets; it is described below. | |
8541 | ||
a2c4f8e0 | 8542 | @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES |
d604bca3 MH |
8543 | A comma-separated list (with leading comma) of pairs describing the |
8544 | alternate names of coprocessor registers. The format of each entry should be | |
8545 | @smallexample | |
8546 | @{ @var{alternatename}, @var{register_number}@} | |
8547 | @end smallexample | |
8548 | Default: empty. | |
a2c4f8e0 | 8549 | @end defmac |
d604bca3 | 8550 | |
7bb1ad93 GK |
8551 | @node PCH Target |
8552 | @section Parameters for Precompiled Header Validity Checking | |
8553 | @cindex parameters, precompiled headers | |
8554 | ||
8d932be3 RS |
8555 | @deftypefn {Target Hook} void *TARGET_GET_PCH_VALIDITY (size_t *@var{sz}) |
8556 | This hook returns the data needed by @code{TARGET_PCH_VALID_P} and sets | |
8557 | @samp{*@var{sz}} to the size of the data in bytes. | |
7bb1ad93 GK |
8558 | @end deftypefn |
8559 | ||
8d932be3 RS |
8560 | @deftypefn {Target Hook} const char *TARGET_PCH_VALID_P (const void *@var{data}, size_t @var{sz}) |
8561 | This hook checks whether the options used to create a PCH file are | |
8562 | compatible with the current settings. It returns @code{NULL} | |
8563 | if so and a suitable error message if not. Error messages will | |
8564 | be presented to the user and must be localized using @samp{_(@var{msg})}. | |
8565 | ||
8566 | @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY} | |
8567 | when the PCH file was created and @var{sz} is the size of that data in bytes. | |
8568 | It's safe to assume that the data was created by the same version of the | |
8569 | compiler, so no format checking is needed. | |
8570 | ||
8571 | The default definition of @code{default_pch_valid_p} should be | |
8572 | suitable for most targets. | |
8573 | @end deftypefn | |
8574 | ||
8575 | @deftypefn {Target Hook} const char *TARGET_CHECK_PCH_TARGET_FLAGS (int @var{pch_flags}) | |
8576 | If this hook is nonnull, the default implementation of | |
8577 | @code{TARGET_PCH_VALID_P} will use it to check for compatible values | |
8578 | of @code{target_flags}. @var{pch_flags} specifies the value that | |
8579 | @code{target_flags} had when the PCH file was created. The return | |
8580 | value is the same as for @code{TARGET_PCH_VALID_P}. | |
7bb1ad93 GK |
8581 | @end deftypefn |
8582 | ||
4185ae53 PB |
8583 | @node C++ ABI |
8584 | @section C++ ABI parameters | |
8585 | @cindex parameters, c++ abi | |
8586 | ||
8587 | @deftypefn {Target Hook} tree TARGET_CXX_GUARD_TYPE (void) | |
8588 | Define this hook to override the integer type used for guard variables. | |
8589 | These are used to implement one-time construction of static objects. The | |
8590 | default is long_long_integer_type_node. | |
8591 | @end deftypefn | |
8592 | ||
8593 | @deftypefn {Target Hook} bool TARGET_CXX_GUARD_MASK_BIT (void) | |
f676971a | 8594 | This hook determines how guard variables are used. It should return |
4185ae53 PB |
8595 | @code{false} (the default) if first byte should be used. A return value of |
8596 | @code{true} indicates the least significant bit should be used. | |
8597 | @end deftypefn | |
8598 | ||
46e995e0 PB |
8599 | @deftypefn {Target Hook} tree TARGET_CXX_GET_COOKIE_SIZE (tree @var{type}) |
8600 | This hook returns the size of the cookie to use when allocating an array | |
8601 | whose elements have the indicated @var{type}. Assumes that it is already | |
8602 | known that a cookie is needed. The default is | |
8603 | @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the | |
8a36672b | 8604 | IA64/Generic C++ ABI@. |
46e995e0 PB |
8605 | @end deftypefn |
8606 | ||
8607 | @deftypefn {Target Hook} bool TARGET_CXX_COOKIE_HAS_SIZE (void) | |
8608 | This hook should return @code{true} if the element size should be stored in | |
8609 | array cookies. The default is to return @code{false}. | |
8610 | @end deftypefn | |
8611 | ||
d59c7b4b NC |
8612 | @deftypefn {Target Hook} int TARGET_CXX_IMPORT_EXPORT_CLASS (tree @var{type}, int @var{import_export}) |
8613 | If defined by a backend this hook allows the decision made to export | |
8614 | class @var{type} to be overruled. Upon entry @var{import_export} | |
78466c0e | 8615 | will contain 1 if the class is going to be exported, @minus{}1 if it is going |
d59c7b4b NC |
8616 | to be imported and 0 otherwise. This function should return the |
8617 | modified value and perform any other actions necessary to support the | |
8618 | backend's targeted operating system. | |
8619 | @end deftypefn | |
8620 | ||
44d10c10 PB |
8621 | @deftypefn {Target Hook} bool TARGET_CXX_CDTOR_RETURNS_THIS (void) |
8622 | This hook should return @code{true} if constructors and destructors return | |
8623 | the address of the object created/destroyed. The default is to return | |
8624 | @code{false}. | |
8625 | @end deftypefn | |
8626 | ||
af287697 MM |
8627 | @deftypefn {Target Hook} bool TARGET_CXX_KEY_METHOD_MAY_BE_INLINE (void) |
8628 | This hook returns true if the key method for a class (i.e., the method | |
8629 | which, if defined in the current translation unit, causes the virtual | |
8630 | table to be emitted) may be an inline function. Under the standard | |
8631 | Itanium C++ ABI the key method may be an inline function so long as | |
8632 | the function is not declared inline in the class definition. Under | |
8633 | some variants of the ABI, an inline function can never be the key | |
8634 | method. The default is to return @code{true}. | |
8635 | @end deftypefn | |
8636 | ||
1e731102 MM |
8637 | @deftypefn {Target Hook} void TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY (tree @var{decl}) |
8638 | @var{decl} is a virtual table, virtual table table, typeinfo object, | |
8639 | or other similar implicit class data object that will be emitted with | |
8640 | external linkage in this translation unit. No ELF visibility has been | |
8641 | explicitly specified. If the target needs to specify a visibility | |
8642 | other than that of the containing class, use this hook to set | |
8643 | @code{DECL_VISIBILITY} and @code{DECL_VISIBILITY_SPECIFIED}. | |
8644 | @end deftypefn | |
8645 | ||
8646 | @deftypefn {Target Hook} bool TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT (void) | |
8647 | This hook returns true (the default) if virtual tables and other | |
8648 | similar implicit class data objects are always COMDAT if they have | |
8649 | external linkage. If this hook returns false, then class data for | |
8650 | classes whose virtual table will be emitted in only one translation | |
8651 | unit will not be COMDAT. | |
505970fc MM |
8652 | @end deftypefn |
8653 | ||
9f62c3e3 PB |
8654 | @deftypefn {Target Hook} bool TARGET_CXX_USE_AEABI_ATEXIT (void) |
8655 | This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI) | |
8656 | should be used to register static destructors when @option{-fuse-cxa-atexit} | |
8657 | is in effect. The default is to return false to use @code{__cxa_atexit}. | |
8658 | @end deftypefn | |
8659 | ||
feca2ed3 JW |
8660 | @node Misc |
8661 | @section Miscellaneous Parameters | |
8662 | @cindex parameters, miscellaneous | |
8663 | ||
8664 | @c prevent bad page break with this line | |
8665 | Here are several miscellaneous parameters. | |
8666 | ||
e543e219 ZW |
8667 | @defmac HAS_LONG_COND_BRANCH |
8668 | Define this boolean macro to indicate whether or not your architecture | |
8669 | has conditional branches that can span all of memory. It is used in | |
8670 | conjunction with an optimization that partitions hot and cold basic | |
8671 | blocks into separate sections of the executable. If this macro is | |
8672 | set to false, gcc will convert any conditional branches that attempt | |
8673 | to cross between sections into unconditional branches or indirect jumps. | |
8674 | @end defmac | |
8675 | ||
8676 | @defmac HAS_LONG_UNCOND_BRANCH | |
8677 | Define this boolean macro to indicate whether or not your architecture | |
8678 | has unconditional branches that can span all of memory. It is used in | |
8679 | conjunction with an optimization that partitions hot and cold basic | |
8680 | blocks into separate sections of the executable. If this macro is | |
8681 | set to false, gcc will convert any unconditional branches that attempt | |
8682 | to cross between sections into indirect jumps. | |
a2c4f8e0 | 8683 | @end defmac |
8fe0ca0c | 8684 | |
a2c4f8e0 | 8685 | @defmac CASE_VECTOR_MODE |
feca2ed3 JW |
8686 | An alias for a machine mode name. This is the machine mode that |
8687 | elements of a jump-table should have. | |
a2c4f8e0 | 8688 | @end defmac |
feca2ed3 | 8689 | |
a2c4f8e0 | 8690 | @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body}) |
33f7f353 JR |
8691 | Optional: return the preferred mode for an @code{addr_diff_vec} |
8692 | when the minimum and maximum offset are known. If you define this, | |
8693 | it enables extra code in branch shortening to deal with @code{addr_diff_vec}. | |
4226378a | 8694 | To make this work, you also have to define @code{INSN_ALIGN} and |
33f7f353 | 8695 | make the alignment for @code{addr_diff_vec} explicit. |
391aaa6b | 8696 | The @var{body} argument is provided so that the offset_unsigned and scale |
33f7f353 | 8697 | flags can be updated. |
a2c4f8e0 | 8698 | @end defmac |
33f7f353 | 8699 | |
a2c4f8e0 | 8700 | @defmac CASE_VECTOR_PC_RELATIVE |
18543a22 | 8701 | Define this macro to be a C expression to indicate when jump-tables |
9c49953c KH |
8702 | should contain relative addresses. You need not define this macro if |
8703 | jump-tables never contain relative addresses, or jump-tables should | |
8704 | contain relative addresses only when @option{-fPIC} or @option{-fPIC} | |
8705 | is in effect. | |
a2c4f8e0 | 8706 | @end defmac |
feca2ed3 | 8707 | |
a2c4f8e0 | 8708 | @defmac CASE_VALUES_THRESHOLD |
feca2ed3 JW |
8709 | Define this to be the smallest number of different values for which it |
8710 | is best to use a jump-table instead of a tree of conditional branches. | |
8711 | The default is four for machines with a @code{casesi} instruction and | |
8712 | five otherwise. This is best for most machines. | |
a2c4f8e0 | 8713 | @end defmac |
feca2ed3 | 8714 | |
a2c4f8e0 | 8715 | @defmac CASE_USE_BIT_TESTS |
9bb231fd RS |
8716 | Define this macro to be a C expression to indicate whether C switch |
8717 | statements may be implemented by a sequence of bit tests. This is | |
8718 | advantageous on processors that can efficiently implement left shift | |
8719 | of 1 by the number of bits held in a register, but inappropriate on | |
8720 | targets that would require a loop. By default, this macro returns | |
8721 | @code{true} if the target defines an @code{ashlsi3} pattern, and | |
8722 | @code{false} otherwise. | |
a2c4f8e0 | 8723 | @end defmac |
9bb231fd | 8724 | |
a2c4f8e0 | 8725 | @defmac WORD_REGISTER_OPERATIONS |
feca2ed3 JW |
8726 | Define this macro if operations between registers with integral mode |
8727 | smaller than a word are always performed on the entire register. | |
8728 | Most RISC machines have this property and most CISC machines do not. | |
a2c4f8e0 | 8729 | @end defmac |
feca2ed3 | 8730 | |
7be4d808 | 8731 | @defmac LOAD_EXTEND_OP (@var{mem_mode}) |
feca2ed3 | 8732 | Define this macro to be a C expression indicating when insns that read |
7be4d808 R |
8733 | memory in @var{mem_mode}, an integral mode narrower than a word, set the |
8734 | bits outside of @var{mem_mode} to be either the sign-extension or the | |
feca2ed3 | 8735 | zero-extension of the data read. Return @code{SIGN_EXTEND} for values |
7be4d808 | 8736 | of @var{mem_mode} for which the |
feca2ed3 | 8737 | insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and |
f822d252 | 8738 | @code{UNKNOWN} for other modes. |
feca2ed3 | 8739 | |
7be4d808 | 8740 | This macro is not called with @var{mem_mode} non-integral or with a width |
feca2ed3 JW |
8741 | greater than or equal to @code{BITS_PER_WORD}, so you may return any |
8742 | value in this case. Do not define this macro if it would always return | |
f822d252 | 8743 | @code{UNKNOWN}. On machines where this macro is defined, you will normally |
feca2ed3 | 8744 | define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}. |
7be4d808 | 8745 | |
f822d252 | 8746 | You may return a non-@code{UNKNOWN} value even if for some hard registers |
7be4d808 R |
8747 | the sign extension is not performed, if for the @code{REGNO_REG_CLASS} |
8748 | of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero | |
8749 | when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any | |
8750 | integral mode larger than this but not larger than @code{word_mode}. | |
8751 | ||
f822d252 | 8752 | You must return @code{UNKNOWN} if for some hard registers that allow this |
7be4d808 R |
8753 | mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to |
8754 | @code{word_mode}, but that they can change to another integral mode that | |
8755 | is larger then @var{mem_mode} but still smaller than @code{word_mode}. | |
a2c4f8e0 | 8756 | @end defmac |
feca2ed3 | 8757 | |
a2c4f8e0 | 8758 | @defmac SHORT_IMMEDIATES_SIGN_EXTEND |
77643ab8 MM |
8759 | Define this macro if loading short immediate values into registers sign |
8760 | extends. | |
a2c4f8e0 | 8761 | @end defmac |
77643ab8 | 8762 | |
a2c4f8e0 | 8763 | @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC |
feca2ed3 JW |
8764 | Define this macro if the same instructions that convert a floating |
8765 | point number to a signed fixed point number also convert validly to an | |
8766 | unsigned one. | |
a2c4f8e0 | 8767 | @end defmac |
feca2ed3 | 8768 | |
a2c4f8e0 | 8769 | @defmac MOVE_MAX |
feca2ed3 JW |
8770 | The maximum number of bytes that a single instruction can move quickly |
8771 | between memory and registers or between two memory locations. | |
a2c4f8e0 | 8772 | @end defmac |
feca2ed3 | 8773 | |
a2c4f8e0 | 8774 | @defmac MAX_MOVE_MAX |
feca2ed3 JW |
8775 | The maximum number of bytes that a single instruction can move quickly |
8776 | between memory and registers or between two memory locations. If this | |
8777 | is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the | |
8778 | constant value that is the largest value that @code{MOVE_MAX} can have | |
8779 | at run-time. | |
a2c4f8e0 | 8780 | @end defmac |
feca2ed3 | 8781 | |
a2c4f8e0 | 8782 | @defmac SHIFT_COUNT_TRUNCATED |
feca2ed3 JW |
8783 | A C expression that is nonzero if on this machine the number of bits |
8784 | actually used for the count of a shift operation is equal to the number | |
8785 | of bits needed to represent the size of the object being shifted. When | |
df2a54e9 | 8786 | this macro is nonzero, the compiler will assume that it is safe to omit |
feca2ed3 JW |
8787 | a sign-extend, zero-extend, and certain bitwise `and' instructions that |
8788 | truncates the count of a shift operation. On machines that have | |
c771326b | 8789 | instructions that act on bit-fields at variable positions, which may |
feca2ed3 JW |
8790 | include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED} |
8791 | also enables deletion of truncations of the values that serve as | |
c771326b | 8792 | arguments to bit-field instructions. |
feca2ed3 JW |
8793 | |
8794 | If both types of instructions truncate the count (for shifts) and | |
c771326b | 8795 | position (for bit-field operations), or if no variable-position bit-field |
feca2ed3 JW |
8796 | instructions exist, you should define this macro. |
8797 | ||
8798 | However, on some machines, such as the 80386 and the 680x0, truncation | |
8799 | only applies to shift operations and not the (real or pretended) | |
c771326b | 8800 | bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on |
feca2ed3 JW |
8801 | such machines. Instead, add patterns to the @file{md} file that include |
8802 | the implied truncation of the shift instructions. | |
8803 | ||
8804 | You need not define this macro if it would always have the value of zero. | |
a2c4f8e0 | 8805 | @end defmac |
feca2ed3 | 8806 | |
273a2526 RS |
8807 | @anchor{TARGET_SHIFT_TRUNCATION_MASK} |
8808 | @deftypefn {Target Hook} int TARGET_SHIFT_TRUNCATION_MASK (enum machine_mode @var{mode}) | |
8809 | This function describes how the standard shift patterns for @var{mode} | |
8810 | deal with shifts by negative amounts or by more than the width of the mode. | |
8811 | @xref{shift patterns}. | |
8812 | ||
8813 | On many machines, the shift patterns will apply a mask @var{m} to the | |
8814 | shift count, meaning that a fixed-width shift of @var{x} by @var{y} is | |
8815 | equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If | |
8816 | this is true for mode @var{mode}, the function should return @var{m}, | |
8817 | otherwise it should return 0. A return value of 0 indicates that no | |
8818 | particular behavior is guaranteed. | |
8819 | ||
8820 | Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does | |
8821 | @emph{not} apply to general shift rtxes; it applies only to instructions | |
8822 | that are generated by the named shift patterns. | |
8823 | ||
8824 | The default implementation of this function returns | |
8825 | @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED} | |
8826 | and 0 otherwise. This definition is always safe, but if | |
8827 | @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns | |
8828 | nevertheless truncate the shift count, you may get better code | |
8829 | by overriding it. | |
8830 | @end deftypefn | |
8831 | ||
a2c4f8e0 | 8832 | @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec}) |
feca2ed3 JW |
8833 | A C expression which is nonzero if on this machine it is safe to |
8834 | ``convert'' an integer of @var{inprec} bits to one of @var{outprec} | |
8835 | bits (where @var{outprec} is smaller than @var{inprec}) by merely | |
8836 | operating on it as if it had only @var{outprec} bits. | |
8837 | ||
8838 | On many machines, this expression can be 1. | |
8839 | ||
8840 | @c rearranged this, removed the phrase "it is reported that". this was | |
8841 | @c to fix an overfull hbox. --mew 10feb93 | |
8842 | When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for | |
8843 | modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result. | |
8844 | If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in | |
8845 | such cases may improve things. | |
a2c4f8e0 | 8846 | @end defmac |
feca2ed3 | 8847 | |
a2c4f8e0 | 8848 | @defmac STORE_FLAG_VALUE |
feca2ed3 JW |
8849 | A C expression describing the value returned by a comparison operator |
8850 | with an integral mode and stored by a store-flag instruction | |
8851 | (@samp{s@var{cond}}) when the condition is true. This description must | |
8852 | apply to @emph{all} the @samp{s@var{cond}} patterns and all the | |
8853 | comparison operators whose results have a @code{MODE_INT} mode. | |
8854 | ||
630d3d5a JM |
8855 | A value of 1 or @minus{}1 means that the instruction implementing the |
8856 | comparison operator returns exactly 1 or @minus{}1 when the comparison is true | |
feca2ed3 JW |
8857 | and 0 when the comparison is false. Otherwise, the value indicates |
8858 | which bits of the result are guaranteed to be 1 when the comparison is | |
8859 | true. This value is interpreted in the mode of the comparison | |
8860 | operation, which is given by the mode of the first operand in the | |
8861 | @samp{s@var{cond}} pattern. Either the low bit or the sign bit of | |
8862 | @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by | |
8863 | the compiler. | |
8864 | ||
630d3d5a | 8865 | If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will |
feca2ed3 JW |
8866 | generate code that depends only on the specified bits. It can also |
8867 | replace comparison operators with equivalent operations if they cause | |
8868 | the required bits to be set, even if the remaining bits are undefined. | |
8869 | For example, on a machine whose comparison operators return an | |
8870 | @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as | |
8871 | @samp{0x80000000}, saying that just the sign bit is relevant, the | |
8872 | expression | |
8873 | ||
8874 | @smallexample | |
8875 | (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0)) | |
8876 | @end smallexample | |
8877 | ||
8878 | @noindent | |
8879 | can be converted to | |
8880 | ||
8881 | @smallexample | |
8882 | (ashift:SI @var{x} (const_int @var{n})) | |
8883 | @end smallexample | |
8884 | ||
8885 | @noindent | |
8886 | where @var{n} is the appropriate shift count to move the bit being | |
8887 | tested into the sign bit. | |
8888 | ||
8889 | There is no way to describe a machine that always sets the low-order bit | |
8890 | for a true value, but does not guarantee the value of any other bits, | |
8891 | but we do not know of any machine that has such an instruction. If you | |
a3a15b4d | 8892 | are trying to port GCC to such a machine, include an instruction to |
feca2ed3 | 8893 | perform a logical-and of the result with 1 in the pattern for the |
b11cc610 | 8894 | comparison operators and let us know at @email{gcc@@gcc.gnu.org}. |
feca2ed3 JW |
8895 | |
8896 | Often, a machine will have multiple instructions that obtain a value | |
8897 | from a comparison (or the condition codes). Here are rules to guide the | |
8898 | choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions | |
8899 | to be used: | |
8900 | ||
8901 | @itemize @bullet | |
8902 | @item | |
8903 | Use the shortest sequence that yields a valid definition for | |
8904 | @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to | |
8905 | ``normalize'' the value (convert it to, e.g., 1 or 0) than for the | |
8906 | comparison operators to do so because there may be opportunities to | |
8907 | combine the normalization with other operations. | |
8908 | ||
8909 | @item | |
630d3d5a | 8910 | For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being |
feca2ed3 JW |
8911 | slightly preferred on machines with expensive jumps and 1 preferred on |
8912 | other machines. | |
8913 | ||
8914 | @item | |
8915 | As a second choice, choose a value of @samp{0x80000001} if instructions | |
8916 | exist that set both the sign and low-order bits but do not define the | |
8917 | others. | |
8918 | ||
8919 | @item | |
8920 | Otherwise, use a value of @samp{0x80000000}. | |
8921 | @end itemize | |
8922 | ||
8923 | Many machines can produce both the value chosen for | |
8924 | @code{STORE_FLAG_VALUE} and its negation in the same number of | |
8925 | instructions. On those machines, you should also define a pattern for | |
8926 | those cases, e.g., one matching | |
8927 | ||
8928 | @smallexample | |
8929 | (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C}))) | |
8930 | @end smallexample | |
8931 | ||
8932 | Some machines can also perform @code{and} or @code{plus} operations on | |
8933 | condition code values with less instructions than the corresponding | |
8934 | @samp{s@var{cond}} insn followed by @code{and} or @code{plus}. On those | |
8935 | machines, define the appropriate patterns. Use the names @code{incscc} | |
8936 | and @code{decscc}, respectively, for the patterns which perform | |
8937 | @code{plus} or @code{minus} operations on condition code values. See | |
8938 | @file{rs6000.md} for some examples. The GNU Superoptizer can be used to | |
8939 | find such instruction sequences on other machines. | |
8940 | ||
06f31100 RS |
8941 | If this macro is not defined, the default value, 1, is used. You need |
8942 | not define @code{STORE_FLAG_VALUE} if the machine has no store-flag | |
8943 | instructions, or if the value generated by these instructions is 1. | |
a2c4f8e0 | 8944 | @end defmac |
feca2ed3 | 8945 | |
a2c4f8e0 | 8946 | @defmac FLOAT_STORE_FLAG_VALUE (@var{mode}) |
df2a54e9 | 8947 | A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is |
feca2ed3 | 8948 | returned when comparison operators with floating-point results are true. |
fc7ca5fd | 8949 | Define this macro on machines that have comparison operations that return |
feca2ed3 JW |
8950 | floating-point values. If there are no such operations, do not define |
8951 | this macro. | |
a2c4f8e0 | 8952 | @end defmac |
feca2ed3 | 8953 | |
fc7ca5fd | 8954 | @defmac VECTOR_STORE_FLAG_VALUE (@var{mode}) |
a4d05547 | 8955 | A C expression that gives a rtx representing the nonzero true element |
fc7ca5fd RS |
8956 | for vector comparisons. The returned rtx should be valid for the inner |
8957 | mode of @var{mode} which is guaranteed to be a vector mode. Define | |
8958 | this macro on machines that have vector comparison operations that | |
8959 | return a vector result. If there are no such operations, do not define | |
8960 | this macro. Typically, this macro is defined as @code{const1_rtx} or | |
8961 | @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent | |
8962 | the compiler optimizing such vector comparison operations for the | |
8963 | given mode. | |
8964 | @end defmac | |
8965 | ||
a2c4f8e0 ZW |
8966 | @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value}) |
8967 | @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value}) | |
7dba8395 RH |
8968 | A C expression that evaluates to true if the architecture defines a value |
8969 | for @code{clz} or @code{ctz} with a zero operand. If so, @var{value} | |
8970 | should be set to this value. If this macro is not defined, the value of | |
8971 | @code{clz} or @code{ctz} is assumed to be undefined. | |
8972 | ||
8973 | This macro must be defined if the target's expansion for @code{ffs} | |
8974 | relies on a particular value to get correct results. Otherwise it | |
8975 | is not necessary, though it may be used to optimize some corner cases. | |
8976 | ||
8977 | Note that regardless of this macro the ``definedness'' of @code{clz} | |
8978 | and @code{ctz} at zero do @emph{not} extend to the builtin functions | |
8979 | visible to the user. Thus one may be free to adjust the value at will | |
8980 | to match the target expansion of these operations without fear of | |
8a36672b | 8981 | breaking the API@. |
a2c4f8e0 | 8982 | @end defmac |
7dba8395 | 8983 | |
a2c4f8e0 | 8984 | @defmac Pmode |
feca2ed3 JW |
8985 | An alias for the machine mode for pointers. On most machines, define |
8986 | this to be the integer mode corresponding to the width of a hardware | |
8987 | pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines. | |
8988 | On some machines you must define this to be one of the partial integer | |
8989 | modes, such as @code{PSImode}. | |
8990 | ||
8991 | The width of @code{Pmode} must be at least as large as the value of | |
8992 | @code{POINTER_SIZE}. If it is not equal, you must define the macro | |
8993 | @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended | |
8994 | to @code{Pmode}. | |
a2c4f8e0 | 8995 | @end defmac |
feca2ed3 | 8996 | |
a2c4f8e0 | 8997 | @defmac FUNCTION_MODE |
feca2ed3 JW |
8998 | An alias for the machine mode used for memory references to functions |
8999 | being called, in @code{call} RTL expressions. On most machines this | |
9000 | should be @code{QImode}. | |
a2c4f8e0 | 9001 | @end defmac |
feca2ed3 | 9002 | |
a2c4f8e0 | 9003 | @defmac STDC_0_IN_SYSTEM_HEADERS |
ee773fcc NB |
9004 | In normal operation, the preprocessor expands @code{__STDC__} to the |
9005 | constant 1, to signify that GCC conforms to ISO Standard C@. On some | |
9006 | hosts, like Solaris, the system compiler uses a different convention, | |
9007 | where @code{__STDC__} is normally 0, but is 1 if the user specifies | |
9008 | strict conformance to the C Standard. | |
9009 | ||
9010 | Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host | |
9011 | convention when processing system header files, but when processing user | |
9012 | files @code{__STDC__} will always expand to 1. | |
a2c4f8e0 | 9013 | @end defmac |
ee773fcc | 9014 | |
a2c4f8e0 | 9015 | @defmac NO_IMPLICIT_EXTERN_C |
161d7b59 | 9016 | Define this macro if the system header files support C++ as well as C@. |
feca2ed3 JW |
9017 | This macro inhibits the usual method of using system header files in |
9018 | C++, which is to pretend that the file's contents are enclosed in | |
9019 | @samp{extern "C" @{@dots{}@}}. | |
a2c4f8e0 | 9020 | @end defmac |
feca2ed3 | 9021 | |
feca2ed3 JW |
9022 | @findex #pragma |
9023 | @findex pragma | |
a2c4f8e0 | 9024 | @defmac REGISTER_TARGET_PRAGMAS () |
8b97c5f8 | 9025 | Define this macro if you want to implement any target-specific pragmas. |
a5da89c6 | 9026 | If defined, it is a C expression which makes a series of calls to |
b5b3e36a DJ |
9027 | @code{c_register_pragma} or @code{c_register_pragma_with_expansion} |
9028 | for each pragma. The macro may also do any | |
a5da89c6 | 9029 | setup required for the pragmas. |
8b97c5f8 ZW |
9030 | |
9031 | The primary reason to define this macro is to provide compatibility with | |
9032 | other compilers for the same target. In general, we discourage | |
161d7b59 | 9033 | definition of target-specific pragmas for GCC@. |
feca2ed3 | 9034 | |
c237e94a | 9035 | If the pragma can be implemented by attributes then you should consider |
91d231cb | 9036 | defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well. |
f09db6e0 | 9037 | |
8b97c5f8 ZW |
9038 | Preprocessor macros that appear on pragma lines are not expanded. All |
9039 | @samp{#pragma} directives that do not match any registered pragma are | |
630d3d5a | 9040 | silently ignored, unless the user specifies @option{-Wunknown-pragmas}. |
a2c4f8e0 | 9041 | @end defmac |
8b97c5f8 | 9042 | |
c58b209a | 9043 | @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *)) |
b5b3e36a | 9044 | @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *)) |
8b97c5f8 | 9045 | |
b5b3e36a DJ |
9046 | Each call to @code{c_register_pragma} or |
9047 | @code{c_register_pragma_with_expansion} establishes one pragma. The | |
8b97c5f8 ZW |
9048 | @var{callback} routine will be called when the preprocessor encounters a |
9049 | pragma of the form | |
9050 | ||
9051 | @smallexample | |
9052 | #pragma [@var{space}] @var{name} @dots{} | |
9053 | @end smallexample | |
9054 | ||
a5da89c6 NB |
9055 | @var{space} is the case-sensitive namespace of the pragma, or |
9056 | @code{NULL} to put the pragma in the global namespace. The callback | |
9057 | routine receives @var{pfile} as its first argument, which can be passed | |
51fabca5 NB |
9058 | on to cpplib's functions if necessary. You can lex tokens after the |
9059 | @var{name} by calling @code{c_lex}. Tokens that are not read by the | |
9060 | callback will be silently ignored. The end of the line is indicated by | |
b5b3e36a DJ |
9061 | a token of type @code{CPP_EOF}. Macro expansion occurs on the |
9062 | arguments of pragmas registered with | |
9063 | @code{c_register_pragma_with_expansion} but not on the arguments of | |
9064 | pragmas registered with @code{c_register_pragma}. | |
8b97c5f8 ZW |
9065 | |
9066 | For an example use of this routine, see @file{c4x.h} and the callback | |
51fabca5 | 9067 | routines defined in @file{c4x-c.c}. |
aac69a49 NC |
9068 | |
9069 | Note that the use of @code{c_lex} is specific to the C and C++ | |
9070 | compilers. It will not work in the Java or Fortran compilers, or any | |
9071 | other language compilers for that matter. Thus if @code{c_lex} is going | |
9072 | to be called from target-specific code, it must only be done so when | |
c771326b | 9073 | building the C and C++ compilers. This can be done by defining the |
aac69a49 | 9074 | variables @code{c_target_objs} and @code{cxx_target_objs} in the |
aee96fe9 | 9075 | target entry in the @file{config.gcc} file. These variables should name |
aac69a49 NC |
9076 | the target-specific, language-specific object file which contains the |
9077 | code that uses @code{c_lex}. Note it will also be necessary to add a | |
9078 | rule to the makefile fragment pointed to by @code{tmake_file} that shows | |
9079 | how to build this object file. | |
8b97c5f8 ZW |
9080 | @end deftypefun |
9081 | ||
e2af664c NC |
9082 | @findex #pragma |
9083 | @findex pragma | |
a2c4f8e0 | 9084 | @defmac HANDLE_SYSV_PRAGMA |
e2af664c NC |
9085 | Define this macro (to a value of 1) if you want the System V style |
9086 | pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name> | |
9087 | [=<value>]} to be supported by gcc. | |
9088 | ||
9089 | The pack pragma specifies the maximum alignment (in bytes) of fields | |
9090 | within a structure, in much the same way as the @samp{__aligned__} and | |
9091 | @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets | |
c21cd8b1 | 9092 | the behavior to the default. |
e2af664c | 9093 | |
e4850f36 | 9094 | A subtlety for Microsoft Visual C/C++ style bit-field packing |
431ae0bf | 9095 | (e.g.@: -mms-bitfields) for targets that support it: |
e4850f36 DR |
9096 | When a bit-field is inserted into a packed record, the whole size |
9097 | of the underlying type is used by one or more same-size adjacent | |
9098 | bit-fields (that is, if its long:3, 32 bits is used in the record, | |
9099 | and any additional adjacent long bit-fields are packed into the same | |
8a36672b | 9100 | chunk of 32 bits. However, if the size changes, a new field of that |
e4850f36 DR |
9101 | size is allocated). |
9102 | ||
9103 | If both MS bit-fields and @samp{__attribute__((packed))} are used, | |
8a36672b | 9104 | the latter will take precedence. If @samp{__attribute__((packed))} is |
e4850f36 DR |
9105 | used on a single field when MS bit-fields are in use, it will take |
9106 | precedence for that field, but the alignment of the rest of the structure | |
9107 | may affect its placement. | |
9108 | ||
e2af664c NC |
9109 | The weak pragma only works if @code{SUPPORTS_WEAK} and |
9110 | @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation | |
9111 | of specifically named weak labels, optionally with a value. | |
a2c4f8e0 | 9112 | @end defmac |
e2af664c | 9113 | |
e2af664c NC |
9114 | @findex #pragma |
9115 | @findex pragma | |
a2c4f8e0 | 9116 | @defmac HANDLE_PRAGMA_PACK_PUSH_POP |
e2af664c | 9117 | Define this macro (to a value of 1) if you want to support the Win32 |
467cecf3 JB |
9118 | style pragmas @samp{#pragma pack(push[,@var{n}])} and @samp{#pragma |
9119 | pack(pop)}. The @samp{pack(push,[@var{n}])} pragma specifies the maximum | |
9120 | alignment (in bytes) of fields within a structure, in much the same way as | |
9121 | the @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A | |
c21cd8b1 | 9122 | pack value of zero resets the behavior to the default. Successive |
e2af664c NC |
9123 | invocations of this pragma cause the previous values to be stacked, so |
9124 | that invocations of @samp{#pragma pack(pop)} will return to the previous | |
9125 | value. | |
a2c4f8e0 | 9126 | @end defmac |
feca2ed3 | 9127 | |
b5b3e36a DJ |
9128 | @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION |
9129 | Define this macro, as well as | |
9130 | @code{HANDLE_SYSV_PRAGMA}, if macros should be expanded in the | |
9131 | arguments of @samp{#pragma pack}. | |
9132 | @end defmac | |
9133 | ||
467cecf3 JB |
9134 | @defmac TARGET_DEFAULT_PACK_STRUCT |
9135 | If your target requires a structure packing default other than 0 (meaning | |
0bdcd332 | 9136 | the machine default), define this macro to the necessary value (in bytes). |
467cecf3 JB |
9137 | This must be a value that would also valid to be used with |
9138 | @samp{#pragma pack()} (that is, a small power of two). | |
9139 | @end defmac | |
9140 | ||
a2c4f8e0 | 9141 | @defmac DOLLARS_IN_IDENTIFIERS |
b1822ccc NB |
9142 | Define this macro to control use of the character @samp{$} in |
9143 | identifier names for the C family of languages. 0 means @samp{$} is | |
9144 | not allowed by default; 1 means it is allowed. 1 is the default; | |
9145 | there is no need to define this macro in that case. | |
a2c4f8e0 | 9146 | @end defmac |
feca2ed3 | 9147 | |
a2c4f8e0 | 9148 | @defmac NO_DOLLAR_IN_LABEL |
feca2ed3 JW |
9149 | Define this macro if the assembler does not accept the character |
9150 | @samp{$} in label names. By default constructors and destructors in | |
9151 | G++ have @samp{$} in the identifiers. If this macro is defined, | |
9152 | @samp{.} is used instead. | |
a2c4f8e0 | 9153 | @end defmac |
feca2ed3 | 9154 | |
a2c4f8e0 | 9155 | @defmac NO_DOT_IN_LABEL |
feca2ed3 JW |
9156 | Define this macro if the assembler does not accept the character |
9157 | @samp{.} in label names. By default constructors and destructors in G++ | |
9158 | have names that use @samp{.}. If this macro is defined, these names | |
9159 | are rewritten to avoid @samp{.}. | |
a2c4f8e0 | 9160 | @end defmac |
feca2ed3 | 9161 | |
a2c4f8e0 | 9162 | @defmac INSN_SETS_ARE_DELAYED (@var{insn}) |
feca2ed3 JW |
9163 | Define this macro as a C expression that is nonzero if it is safe for the |
9164 | delay slot scheduler to place instructions in the delay slot of @var{insn}, | |
9165 | even if they appear to use a resource set or clobbered in @var{insn}. | |
a3a15b4d | 9166 | @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that |
feca2ed3 JW |
9167 | every @code{call_insn} has this behavior. On machines where some @code{insn} |
9168 | or @code{jump_insn} is really a function call and hence has this behavior, | |
9169 | you should define this macro. | |
9170 | ||
9171 | You need not define this macro if it would always return zero. | |
a2c4f8e0 | 9172 | @end defmac |
feca2ed3 | 9173 | |
a2c4f8e0 | 9174 | @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn}) |
feca2ed3 JW |
9175 | Define this macro as a C expression that is nonzero if it is safe for the |
9176 | delay slot scheduler to place instructions in the delay slot of @var{insn}, | |
9177 | even if they appear to set or clobber a resource referenced in @var{insn}. | |
9178 | @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where | |
9179 | some @code{insn} or @code{jump_insn} is really a function call and its operands | |
9180 | are registers whose use is actually in the subroutine it calls, you should | |
9181 | define this macro. Doing so allows the delay slot scheduler to move | |
9182 | instructions which copy arguments into the argument registers into the delay | |
9183 | slot of @var{insn}. | |
9184 | ||
9185 | You need not define this macro if it would always return zero. | |
a2c4f8e0 | 9186 | @end defmac |
feca2ed3 | 9187 | |
a2c4f8e0 | 9188 | @defmac MULTIPLE_SYMBOL_SPACES |
15072eb1 ZW |
9189 | Define this macro as a C expression that is nonzero if, in some cases, |
9190 | global symbols from one translation unit may not be bound to undefined | |
9191 | symbols in another translation unit without user intervention. For | |
9192 | instance, under Microsoft Windows symbols must be explicitly imported | |
9193 | from shared libraries (DLLs). | |
9194 | ||
9195 | You need not define this macro if it would always evaluate to zero. | |
a2c4f8e0 | 9196 | @end defmac |
861bb6c1 | 9197 | |
61158923 | 9198 | @deftypefn {Target Hook} tree TARGET_MD_ASM_CLOBBERS (tree @var{outputs}, tree @var{inputs}, tree @var{clobbers}) |
67dfe110 | 9199 | This target hook should add to @var{clobbers} @code{STRING_CST} trees for |
61158923 | 9200 | any hard regs the port wishes to automatically clobber for an asm. |
67dfe110 | 9201 | It should return the result of the last @code{tree_cons} used to add a |
61158923 HPN |
9202 | clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the |
9203 | corresponding parameters to the asm and may be inspected to avoid | |
9204 | clobbering a register that is an input or output of the asm. You can use | |
9205 | @code{decl_overlaps_hard_reg_set_p}, declared in @file{tree.h}, to test | |
9206 | for overlap with regards to asm-declared registers. | |
67dfe110 | 9207 | @end deftypefn |
57bcb97a | 9208 | |
a2c4f8e0 | 9209 | @defmac MATH_LIBRARY |
71d718e0 JM |
9210 | Define this macro as a C string constant for the linker argument to link |
9211 | in the system math library, or @samp{""} if the target does not have a | |
9212 | separate math library. | |
9213 | ||
9214 | You need only define this macro if the default of @samp{"-lm"} is wrong. | |
a2c4f8e0 | 9215 | @end defmac |
512b62fb | 9216 | |
a2c4f8e0 | 9217 | @defmac LIBRARY_PATH_ENV |
512b62fb JM |
9218 | Define this macro as a C string constant for the environment variable that |
9219 | specifies where the linker should look for libraries. | |
9220 | ||
9221 | You need only define this macro if the default of @samp{"LIBRARY_PATH"} | |
9222 | is wrong. | |
a2c4f8e0 | 9223 | @end defmac |
e09d24ff | 9224 | |
4969c0d8 L |
9225 | @defmac TARGET_POSIX_IO |
9226 | Define this macro if the target supports the following POSIX@ file | |
9227 | functions, access, mkdir and file locking with fcntl / F_SETLKW@. | |
9228 | Defining @code{TARGET_POSIX_IO} will enable the test coverage code | |
e09d24ff | 9229 | to use file locking when exiting a program, which avoids race conditions |
4969c0d8 | 9230 | if the program has forked. It will also create directories at run-time |
709a840a | 9231 | for cross-profiling. |
a2c4f8e0 | 9232 | @end defmac |
0c99ec5c | 9233 | |
a2c4f8e0 | 9234 | @defmac MAX_CONDITIONAL_EXECUTE |
0c99ec5c RH |
9235 | |
9236 | A C expression for the maximum number of instructions to execute via | |
9237 | conditional execution instructions instead of a branch. A value of | |
9238 | @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and | |
9239 | 1 if it does use cc0. | |
a2c4f8e0 | 9240 | @end defmac |
90280148 | 9241 | |
a2c4f8e0 | 9242 | @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr}) |
c05ffc49 BS |
9243 | Used if the target needs to perform machine-dependent modifications on the |
9244 | conditionals used for turning basic blocks into conditionally executed code. | |
9245 | @var{ce_info} points to a data structure, @code{struct ce_if_block}, which | |
9246 | contains information about the currently processed blocks. @var{true_expr} | |
9247 | and @var{false_expr} are the tests that are used for converting the | |
9248 | then-block and the else-block, respectively. Set either @var{true_expr} or | |
9249 | @var{false_expr} to a null pointer if the tests cannot be converted. | |
a2c4f8e0 | 9250 | @end defmac |
c05ffc49 | 9251 | |
a2c4f8e0 | 9252 | @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr}) |
c05ffc49 BS |
9253 | Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated |
9254 | if-statements into conditions combined by @code{and} and @code{or} operations. | |
9255 | @var{bb} contains the basic block that contains the test that is currently | |
9256 | being processed and about to be turned into a condition. | |
a2c4f8e0 | 9257 | @end defmac |
90280148 | 9258 | |
a2c4f8e0 | 9259 | @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn}) |
c05ffc49 BS |
9260 | A C expression to modify the @var{PATTERN} of an @var{INSN} that is to |
9261 | be converted to conditional execution format. @var{ce_info} points to | |
9262 | a data structure, @code{struct ce_if_block}, which contains information | |
9263 | about the currently processed blocks. | |
a2c4f8e0 | 9264 | @end defmac |
90280148 | 9265 | |
a2c4f8e0 | 9266 | @defmac IFCVT_MODIFY_FINAL (@var{ce_info}) |
90280148 | 9267 | A C expression to perform any final machine dependent modifications in |
c05ffc49 BS |
9268 | converting code to conditional execution. The involved basic blocks |
9269 | can be found in the @code{struct ce_if_block} structure that is pointed | |
9270 | to by @var{ce_info}. | |
a2c4f8e0 | 9271 | @end defmac |
90280148 | 9272 | |
a2c4f8e0 | 9273 | @defmac IFCVT_MODIFY_CANCEL (@var{ce_info}) |
90280148 | 9274 | A C expression to cancel any machine dependent modifications in |
c05ffc49 BS |
9275 | converting code to conditional execution. The involved basic blocks |
9276 | can be found in the @code{struct ce_if_block} structure that is pointed | |
9277 | to by @var{ce_info}. | |
a2c4f8e0 | 9278 | @end defmac |
c05ffc49 | 9279 | |
a2c4f8e0 | 9280 | @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info}) |
c05ffc49 BS |
9281 | A C expression to initialize any extra fields in a @code{struct ce_if_block} |
9282 | structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro. | |
a2c4f8e0 | 9283 | @end defmac |
c05ffc49 | 9284 | |
a2c4f8e0 | 9285 | @defmac IFCVT_EXTRA_FIELDS |
c05ffc49 | 9286 | If defined, it should expand to a set of field declarations that will be |
c0478a66 | 9287 | added to the @code{struct ce_if_block} structure. These should be initialized |
c05ffc49 | 9288 | by the @code{IFCVT_INIT_EXTRA_FIELDS} macro. |
a2c4f8e0 | 9289 | @end defmac |
c05ffc49 | 9290 | |
18dbd950 RS |
9291 | @deftypefn {Target Hook} void TARGET_MACHINE_DEPENDENT_REORG () |
9292 | If non-null, this hook performs a target-specific pass over the | |
9293 | instruction stream. The compiler will run it at all optimization levels, | |
9294 | just before the point at which it normally does delayed-branch scheduling. | |
9295 | ||
9296 | The exact purpose of the hook varies from target to target. Some use | |
9297 | it to do transformations that are necessary for correctness, such as | |
9298 | laying out in-function constant pools or avoiding hardware hazards. | |
9299 | Others use it as an opportunity to do some machine-dependent optimizations. | |
9300 | ||
9301 | You need not implement the hook if it has nothing to do. The default | |
9302 | definition is null. | |
9303 | @end deftypefn | |
9304 | ||
f6155fda SS |
9305 | @deftypefn {Target Hook} void TARGET_INIT_BUILTINS () |
9306 | Define this hook if you have any machine-specific built-in functions | |
9307 | that need to be defined. It should be a function that performs the | |
4a1d48f6 BS |
9308 | necessary setup. |
9309 | ||
c771326b | 9310 | Machine specific built-in functions can be useful to expand special machine |
4a1d48f6 BS |
9311 | instructions that would otherwise not normally be generated because |
9312 | they have no equivalent in the source language (for example, SIMD vector | |
9313 | instructions or prefetch instructions). | |
9314 | ||
6e34d3a3 JM |
9315 | To create a built-in function, call the function |
9316 | @code{lang_hooks.builtin_function} | |
c771326b | 9317 | which is defined by the language front end. You can use any type nodes set |
4a1d48f6 | 9318 | up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2}; |
c237e94a | 9319 | only language front ends that use those two functions will call |
f6155fda | 9320 | @samp{TARGET_INIT_BUILTINS}. |
acdcefcc | 9321 | @end deftypefn |
4a1d48f6 | 9322 | |
f6155fda | 9323 | @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore}) |
4a1d48f6 | 9324 | |
c771326b | 9325 | Expand a call to a machine specific built-in function that was set up by |
f6155fda SS |
9326 | @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the |
9327 | function call; the result should go to @var{target} if that is | |
9328 | convenient, and have mode @var{mode} if that is convenient. | |
9329 | @var{subtarget} may be used as the target for computing one of | |
9330 | @var{exp}'s operands. @var{ignore} is nonzero if the value is to be | |
9331 | ignored. This function should return the result of the call to the | |
9332 | built-in function. | |
acdcefcc | 9333 | @end deftypefn |
4a1d48f6 | 9334 | |
4268e4cf PB |
9335 | @deftypefn {Target Hook} tree TARGET_RESOLVE_OVERLOADED_BUILTIN (tree @var{fndecl}, tree @var{arglist}) |
9336 | ||
9337 | Select a replacement for a machine specific built-in function that | |
9338 | was set up by @samp{TARGET_INIT_BUILTINS}. This is done | |
9339 | @emph{before} regular type checking, and so allows the target to | |
9340 | implement a crude form of function overloading. @var{fndecl} is the | |
9341 | declaration of the built-in function. @var{arglist} is the list of | |
9342 | arguments passed to the built-in function. The result is a | |
9343 | complete expression that implements the operation, usually | |
9344 | another @code{CALL_EXPR}. | |
9345 | @end deftypefn | |
9346 | ||
a05a80fc KH |
9347 | @deftypefn {Target Hook} tree TARGET_FOLD_BUILTIN (tree @var{fndecl}, tree @var{arglist}, bool @var{ignore}) |
9348 | ||
9349 | Fold a call to a machine specific built-in function that was set up by | |
9350 | @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the | |
9351 | built-in function. @var{arglist} is the list of arguments passed to | |
9352 | the built-in function. The result is another tree containing a | |
9353 | simplified expression for the call's result. If @var{ignore} is true | |
9354 | the value will be ignored. | |
d6c2b67d PB |
9355 | @end deftypefn |
9356 | ||
e7e64a25 | 9357 | @deftypefn {Target Hook} const char * TARGET_INVALID_WITHIN_DOLOOP (rtx @var{insn}) |
a71a498d | 9358 | |
e7e64a25 AS |
9359 | Take an instruction in @var{insn} and return NULL if it is valid within a |
9360 | low-overhead loop, otherwise return a string why doloop could not be applied. | |
a71a498d | 9361 | |
e7e64a25 AS |
9362 | Many targets use special registers for low-overhead looping. For any |
9363 | instruction that clobbers these this function should return a string indicating | |
9364 | the reason why the doloop could not be applied. | |
a71a498d | 9365 | By default, the RTL loop optimizer does not use a present doloop pattern for |
0fa2e4df | 9366 | loops containing function calls or branch on table instructions. |
a71a498d AS |
9367 | @end deftypefn |
9368 | ||
a2c4f8e0 | 9369 | @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2}) |
6e7b03e1 | 9370 | |
4fe9b91c | 9371 | Take a branch insn in @var{branch1} and another in @var{branch2}. |
6e7b03e1 AH |
9372 | Return true if redirecting @var{branch1} to the destination of |
9373 | @var{branch2} is possible. | |
9374 | ||
9375 | On some targets, branches may have a limited range. Optimizing the | |
9376 | filling of delay slots can result in branches being redirected, and this | |
9377 | may in turn cause a branch offset to overflow. | |
a2c4f8e0 | 9378 | @end defmac |
6e7b03e1 | 9379 | |
8ddf681a R |
9380 | @deftypefn {Target Hook} bool TARGET_COMMUTATIVE_P (rtx @var{x}, @var{outer_code}) |
9381 | This target hook returns @code{true} if @var{x} is considered to be commutative. | |
9382 | Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider | |
9383 | PLUS to be commutative inside a MEM. @var{outer_code} is the rtx code | |
9384 | of the enclosing rtl, if known, otherwise it is UNKNOWN. | |
9385 | @end deftypefn | |
9386 | ||
b48f503c | 9387 | @deftypefn {Target Hook} rtx TARGET_ALLOCATE_INITIAL_VALUE (rtx @var{hard_reg}) |
385b6e2d R |
9388 | |
9389 | When the initial value of a hard register has been copied in a pseudo | |
4fe9b91c | 9390 | register, it is often not necessary to actually allocate another register |
385b6e2d | 9391 | to this pseudo register, because the original hard register or a stack slot |
b48f503c KK |
9392 | it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE} |
9393 | is called at the start of register allocation once for each hard register | |
9394 | that had its initial value copied by using | |
385b6e2d R |
9395 | @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}. |
9396 | Possible values are @code{NULL_RTX}, if you don't want | |
9397 | to do any special allocation, a @code{REG} rtx---that would typically be | |
9398 | the hard register itself, if it is known not to be clobbered---or a | |
9399 | @code{MEM}. | |
9400 | If you are returning a @code{MEM}, this is only a hint for the allocator; | |
9401 | it might decide to use another register anyways. | |
b48f503c KK |
9402 | You may use @code{current_function_leaf_function} in the hook, functions |
9403 | that use @code{REG_N_SETS}, to determine if the hard | |
385b6e2d | 9404 | register in question will not be clobbered. |
b48f503c KK |
9405 | The default value of this hook is @code{NULL}, which disables any special |
9406 | allocation. | |
9407 | @end deftypefn | |
385b6e2d | 9408 | |
a2c4f8e0 | 9409 | @defmac TARGET_OBJECT_SUFFIX |
807633e5 ZW |
9410 | Define this macro to be a C string representing the suffix for object |
9411 | files on your target machine. If you do not define this macro, GCC will | |
9412 | use @samp{.o} as the suffix for object files. | |
a2c4f8e0 | 9413 | @end defmac |
807633e5 | 9414 | |
a2c4f8e0 | 9415 | @defmac TARGET_EXECUTABLE_SUFFIX |
807633e5 ZW |
9416 | Define this macro to be a C string representing the suffix to be |
9417 | automatically added to executable files on your target machine. If you | |
9418 | do not define this macro, GCC will use the null string as the suffix for | |
9419 | executable files. | |
a2c4f8e0 | 9420 | @end defmac |
807633e5 | 9421 | |
a2c4f8e0 | 9422 | @defmac COLLECT_EXPORT_LIST |
807633e5 ZW |
9423 | If defined, @code{collect2} will scan the individual object files |
9424 | specified on its command line and create an export list for the linker. | |
9425 | Define this macro for systems like AIX, where the linker discards | |
9426 | object files that are not referenced from @code{main} and uses export | |
9427 | lists. | |
a2c4f8e0 | 9428 | @end defmac |
807633e5 | 9429 | |
a2c4f8e0 | 9430 | @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl}) |
55ae46b1 RM |
9431 | Define this macro to a C expression representing a variant of the |
9432 | method call @var{mdecl}, if Java Native Interface (JNI) methods | |
9433 | must be invoked differently from other methods on your target. | |
95fef11f | 9434 | For example, on 32-bit Microsoft Windows, JNI methods must be invoked using |
55ae46b1 RM |
9435 | the @code{stdcall} calling convention and this macro is then |
9436 | defined as this expression: | |
9437 | ||
9438 | @smallexample | |
9439 | build_type_attribute_variant (@var{mdecl}, | |
9440 | build_tree_list | |
9441 | (get_identifier ("stdcall"), | |
9442 | NULL)) | |
9443 | @end smallexample | |
a2c4f8e0 | 9444 | @end defmac |
e4ec2cac AO |
9445 | |
9446 | @deftypefn {Target Hook} bool TARGET_CANNOT_MODIFY_JUMPS_P (void) | |
9447 | This target hook returns @code{true} past the point in which new jump | |
9448 | instructions could be created. On machines that require a register for | |
9449 | every jump such as the SHmedia ISA of SH5, this point would typically be | |
9450 | reload, so this target hook should be defined to a function such as: | |
9451 | ||
9452 | @smallexample | |
9453 | static bool | |
9454 | cannot_modify_jumps_past_reload_p () | |
9455 | @{ | |
9456 | return (reload_completed || reload_in_progress); | |
9457 | @} | |
9458 | @end smallexample | |
9459 | @end deftypefn | |
fe3ad572 | 9460 | |
a3424f5c | 9461 | @deftypefn {Target Hook} int TARGET_BRANCH_TARGET_REGISTER_CLASS (void) |
fe3ad572 SC |
9462 | This target hook returns a register class for which branch target register |
9463 | optimizations should be applied. All registers in this class should be | |
c0cbdbd9 | 9464 | usable interchangeably. After reload, registers in this class will be |
fe3ad572 SC |
9465 | re-allocated and loads will be hoisted out of loops and be subjected |
9466 | to inter-block scheduling. | |
9467 | @end deftypefn | |
9468 | ||
9469 | @deftypefn {Target Hook} bool TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED (bool @var{after_prologue_epilogue_gen}) | |
9470 | Branch target register optimization will by default exclude callee-saved | |
9471 | registers | |
9472 | that are not already live during the current function; if this target hook | |
9473 | returns true, they will be included. The target code must than make sure | |
9474 | that all target registers in the class returned by | |
9475 | @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are | |
9476 | saved. @var{after_prologue_epilogue_gen} indicates if prologues and | |
9477 | epilogues have already been generated. Note, even if you only return | |
9478 | true when @var{after_prologue_epilogue_gen} is false, you still are likely | |
9479 | to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET} | |
9480 | to reserve space for caller-saved target registers. | |
9481 | @end deftypefn | |
2082e02f RS |
9482 | |
9483 | @defmac POWI_MAX_MULTS | |
73774972 | 9484 | If defined, this macro is interpreted as a signed integer C expression |
2082e02f RS |
9485 | that specifies the maximum number of floating point multiplications |
9486 | that should be emitted when expanding exponentiation by an integer | |
9487 | constant inline. When this value is defined, exponentiation requiring | |
9488 | more than this number of multiplications is implemented by calling the | |
9489 | system library's @code{pow}, @code{powf} or @code{powl} routines. | |
9490 | The default value places no upper bound on the multiplication count. | |
9491 | @end defmac | |
94d1613b | 9492 | |
3dd53121 | 9493 | @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc}) |
94d1613b MS |
9494 | This target hook should register any extra include files for the |
9495 | target. The parameter @var{stdinc} indicates if normal include files | |
3dd53121 AP |
9496 | are present. The parameter @var{sysroot} is the system root directory. |
9497 | The parameter @var{iprefix} is the prefix for the gcc directory. | |
9498 | @end deftypefn | |
9499 | ||
9500 | @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc}) | |
9501 | This target hook should register any extra include files for the | |
9502 | target before any standard headers. The parameter @var{stdinc} | |
9503 | indicates if normal include files are present. The parameter | |
9504 | @var{sysroot} is the system root directory. The parameter | |
9505 | @var{iprefix} is the prefix for the gcc directory. | |
94d1613b MS |
9506 | @end deftypefn |
9507 | ||
9508 | @deftypefn Macro void TARGET_OPTF (char *@var{path}) | |
9509 | This target hook should register special include paths for the target. | |
9510 | The parameter @var{path} is the include to register. On Darwin | |
9511 | systems, this is used for Framework includes, which have semantics | |
9512 | that are different from @option{-I}. | |
9513 | @end deftypefn | |
4a77e08c DS |
9514 | |
9515 | @deftypefn {Target Hook} bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl}) | |
9516 | This target hook returns @code{true} if it is safe to use a local alias | |
9517 | for a virtual function @var{fndecl} when constructing thunks, | |
8a36672b | 9518 | @code{false} otherwise. By default, the hook returns @code{true} for all |
431ae0bf | 9519 | functions, if a target supports aliases (i.e.@: defines |
4a77e08c DS |
9520 | @code{ASM_OUTPUT_DEF}), @code{false} otherwise, |
9521 | @end deftypefn | |
a2bec818 DJ |
9522 | |
9523 | @defmac TARGET_FORMAT_TYPES | |
9524 | If defined, this macro is the name of a global variable containing | |
9525 | target-specific format checking information for the @option{-Wformat} | |
9526 | option. The default is to have no target-specific format checks. | |
9527 | @end defmac | |
9528 | ||
9529 | @defmac TARGET_N_FORMAT_TYPES | |
9530 | If defined, this macro is the number of entries in | |
9531 | @code{TARGET_FORMAT_TYPES}. | |
9532 | @end defmac | |
e50e6b88 | 9533 | |
445cf5eb JM |
9534 | @deftypefn {Target Hook} bool TARGET_RELAXED_ORDERING |
9535 | If set to @code{true}, means that the target's memory model does not | |
9536 | guarantee that loads which do not depend on one another will access | |
9537 | main memory in the order of the instruction stream; if ordering is | |
9538 | important, an explicit memory barrier must be used. This is true of | |
9539 | many recent processors which implement a policy of ``relaxed,'' | |
9540 | ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC, | |
9541 | and ia64. The default is @code{false}. | |
9542 | @end deftypefn | |
9543 | ||
4d3e6fae FJ |
9544 | @deftypefn {Target Hook} const char *TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN (tree @var{typelist}, tree @var{funcdecl}, tree @var{val}) |
9545 | If defined, this macro returns the diagnostic message when it is | |
9546 | illegal to pass argument @var{val} to function @var{funcdecl} | |
9547 | with prototype @var{typelist}. | |
9548 | @end deftypefn | |
9549 | ||
4de67c26 JM |
9550 | @deftypefn {Target Hook} {const char *} TARGET_INVALID_CONVERSION (tree @var{fromtype}, tree @var{totype}) |
9551 | If defined, this macro returns the diagnostic message when it is | |
9552 | invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL} | |
9553 | if validity should be determined by the front end. | |
9554 | @end deftypefn | |
9555 | ||
9556 | @deftypefn {Target Hook} {const char *} TARGET_INVALID_UNARY_OP (int @var{op}, tree @var{type}) | |
9557 | If defined, this macro returns the diagnostic message when it is | |
9558 | invalid to apply operation @var{op} (where unary plus is denoted by | |
9559 | @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL} | |
9560 | if validity should be determined by the front end. | |
9561 | @end deftypefn | |
9562 | ||
9563 | @deftypefn {Target Hook} {const char *} TARGET_INVALID_BINARY_OP (int @var{op}, tree @var{type1}, tree @var{type2}) | |
9564 | If defined, this macro returns the diagnostic message when it is | |
9565 | invalid to apply operation @var{op} to operands of types @var{type1} | |
9566 | and @var{type2}, or @code{NULL} if validity should be determined by | |
9567 | the front end. | |
9568 | @end deftypefn | |
9569 | ||
e50e6b88 DS |
9570 | @defmac TARGET_USE_JCR_SECTION |
9571 | This macro determines whether to use the JCR section to register Java | |
9572 | classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both | |
9573 | SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0. | |
9574 | @end defmac |