1 @c Copyright (C) 1988,1989,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001
2 @c Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
7 @chapter Target Description Macros and Functions
8 @cindex machine description macros
9 @cindex target description macros
10 @cindex macros, target description
11 @cindex @file{tm.h} macros
13 In addition to the file @file{@var{machine}.md}, a machine description
14 includes a C header file conventionally given the name
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
21 source file defines a variable @code{targetm}, which is a structure
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.
28 * Target Structure:: The @code{targetm} variable.
29 * Driver:: Controlling how the driver runs the compilation passes.
30 * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}.
31 * Per-Function Data:: Defining data structures for per-function information.
32 * Storage Layout:: Defining sizes and alignments of data.
33 * Type Layout:: Defining sizes and properties of basic user data types.
34 * Escape Sequences:: Defining the value of target character escape sequences
35 * Registers:: Naming and describing the hardware registers.
36 * Register Classes:: Defining the classes of hardware registers.
37 * Stack and Calling:: Defining which way the stack grows and by how much.
38 * Varargs:: Defining the varargs macros.
39 * Trampolines:: Code set up at run time to enter a nested function.
40 * Library Calls:: Controlling how library routines are implicitly called.
41 * Addressing Modes:: Defining addressing modes valid for memory operands.
42 * Condition Code:: Defining how insns update the condition code.
43 * Costs:: Defining relative costs of different operations.
44 * Scheduling:: Adjusting the behavior of the instruction scheduler.
45 * Sections:: Dividing storage into text, data, and other sections.
46 * PIC:: Macros for position independent code.
47 * Assembler Format:: Defining how to write insns and pseudo-ops to output.
48 * Debugging Info:: Defining the format of debugging output.
49 * Cross-compilation:: Handling floating point for cross-compilers.
50 * Mode Switching:: Insertion of mode-switching instructions.
51 * Target Attributes:: Defining target-specific uses of @code{__attribute__}.
52 * Misc:: Everything else.
55 @node Target Structure
56 @section The Global @code{targetm} Variable
58 @cindex target functions
60 @deftypevar {struct gcc_target} targetm
61 The target @file{.c} file must define the global @code{targetm} variable
62 which contains pointers to functions and data relating to the target
63 machine. The variable is declared in @file{target.h};
64 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
65 used to initialize the variable, and macros for the default initializers
66 for elements of the structure. The @file{.c} file should override those
67 macros for which the default definition is inappropriate. For example:
70 #include "target-def.h"
72 /* @r{Initialize the GCC target structure.} */
74 #undef TARGET_COMP_TYPE_ATTRIBUTES
75 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
77 struct gcc_target targetm = TARGET_INITIALIZER;
81 Where a macro should be defined in the @file{.c} file in this manner to
82 form part of the @code{targetm} structure, it is documented below as a
83 ``Target Hook'' with a prototype. Many macros will change in future
84 from being defined in the @file{.h} file to being part of the
85 @code{targetm} structure.
88 @section Controlling the Compilation Driver, @file{gcc}
90 @cindex controlling the compilation driver
92 @c prevent bad page break with this line
93 You can control the compilation driver.
96 @findex SWITCH_TAKES_ARG
97 @item SWITCH_TAKES_ARG (@var{char})
98 A C expression which determines whether the option @option{-@var{char}}
99 takes arguments. The value should be the number of arguments that
100 option takes--zero, for many options.
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
109 @findex WORD_SWITCH_TAKES_ARG
110 @item WORD_SWITCH_TAKES_ARG (@var{name})
111 A C expression which determines whether the option @option{-@var{name}}
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.
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
123 @findex SWITCH_CURTAILS_COMPILATION
124 @item SWITCH_CURTAILS_COMPILATION (@var{char})
125 A C expression which determines whether the option @option{-@var{char}}
126 stops compilation before the generation of an executable. The value is
127 boolean, nonzero if the option does stop an executable from being
128 generated, zero otherwise.
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.
138 @findex SWITCHES_NEED_SPACES
139 @item SWITCHES_NEED_SPACES
140 A string-valued C expression which enumerates the options for which
141 the linker needs a space between the option and its argument.
143 If this macro is not defined, the default value is @code{""}.
145 @findex TARGET_OPTION_TRANSLATE_TABLE
146 @item TARGET_OPTION_TRANSLATE_TABLE
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
156 multilibs. Example nonsensical definition, where @code{-malt-abi},
157 @code{-EB}, and @code{-mspoo} cause different multilibs to be chosen:
160 #define TARGET_OPTION_TRANSLATE_TABLE \
161 @{ "-fast", "-march=fast-foo -malt-abi -I/usr/fast-foo" @}, \
162 @{ "-compat", "-EB -malign=4 -mspoo" @}
167 A C string constant that tells the GCC driver program options to
168 pass to CPP@. It can also specify how to translate options you
169 give to GCC into options for GCC to pass to the CPP@.
171 Do not define this macro if it does not need to do anything.
173 @findex CPLUSPLUS_CPP_SPEC
174 @item CPLUSPLUS_CPP_SPEC
175 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
176 than C@. If you do not define this macro, then the value of
177 @code{CPP_SPEC} (if any) will be used instead.
179 @findex NO_BUILTIN_SIZE_TYPE
180 @item NO_BUILTIN_SIZE_TYPE
181 If this macro is defined, the preprocessor will not define the built-in macro
182 @code{__SIZE_TYPE__}. The macro @code{__SIZE_TYPE__} must then be defined
183 by @code{CPP_SPEC} instead.
185 This should be defined if @code{SIZE_TYPE} depends on target dependent flags
186 which are not accessible to the preprocessor. Otherwise, it should not
189 @findex NO_BUILTIN_PTRDIFF_TYPE
190 @item NO_BUILTIN_PTRDIFF_TYPE
191 If this macro is defined, the preprocessor will not define the built-in macro
192 @code{__PTRDIFF_TYPE__}. The macro @code{__PTRDIFF_TYPE__} must then be
193 defined by @code{CPP_SPEC} instead.
195 This should be defined if @code{PTRDIFF_TYPE} depends on target dependent flags
196 which are not accessible to the preprocessor. Otherwise, it should not
199 @findex NO_BUILTIN_WCHAR_TYPE
200 @item NO_BUILTIN_WCHAR_TYPE
201 If this macro is defined, the preprocessor will not define the built-in macro
202 @code{__WCHAR_TYPE__}. The macro @code{__WCHAR_TYPE__} must then be
203 defined by @code{CPP_SPEC} instead.
205 This should be defined if @code{WCHAR_TYPE} depends on target dependent flags
206 which are not accessible to the preprocessor. Otherwise, it should not
209 @findex NO_BUILTIN_WINT_TYPE
210 @item NO_BUILTIN_WINT_TYPE
211 If this macro is defined, the preprocessor will not define the built-in macro
212 @code{__WINT_TYPE__}. The macro @code{__WINT_TYPE__} must then be
213 defined by @code{CPP_SPEC} instead.
215 This should be defined if @code{WINT_TYPE} depends on target dependent flags
216 which are not accessible to the preprocessor. Otherwise, it should not
219 @findex SIGNED_CHAR_SPEC
220 @item SIGNED_CHAR_SPEC
221 A C string constant that tells the GCC driver program options to
222 pass to CPP@. By default, this macro is defined to pass the option
223 @option{-D__CHAR_UNSIGNED__} to CPP if @code{char} will be treated as
224 @code{unsigned char} by @code{cc1}.
226 Do not define this macro unless you need to override the default
231 A C string constant that tells the GCC driver program options to
232 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
234 It can also specify how to translate options you give to GCC into options
235 for GCC to pass to front ends.
237 Do not define this macro if it does not need to do anything.
241 A C string constant that tells the GCC driver program options to
242 pass to @code{cc1plus}. It can also specify how to translate options you
243 give to GCC into options for GCC to pass to the @code{cc1plus}.
245 Do not define this macro if it does not need to do anything.
246 Note that everything defined in CC1_SPEC is already passed to
247 @code{cc1plus} so there is no need to duplicate the contents of
248 CC1_SPEC in CC1PLUS_SPEC@.
252 A C string constant that tells the GCC driver program options to
253 pass to the assembler. It can also specify how to translate options
254 you give to GCC into options for GCC to pass to the assembler.
255 See the file @file{sun3.h} for an example of this.
257 Do not define this macro if it does not need to do anything.
259 @findex ASM_FINAL_SPEC
261 A C string constant that tells the GCC driver program how to
262 run any programs which cleanup after the normal assembler.
263 Normally, this is not needed. See the file @file{mips.h} for
266 Do not define this macro if it does not need to do anything.
270 A C string constant that tells the GCC driver program options to
271 pass to the linker. It can also specify how to translate options you
272 give to GCC into options for GCC to pass to the linker.
274 Do not define this macro if it does not need to do anything.
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.
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}.
287 Another C string constant that tells the GCC driver program
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}.
292 If this macro is not defined, the GCC driver provides a default that
293 passes the string @option{-lgcc} to the linker.
295 @findex STARTFILE_SPEC
297 Another C string constant used much like @code{LINK_SPEC}. The
298 difference between the two is that @code{STARTFILE_SPEC} is used at
299 the very beginning of the command given to the linker.
301 If this macro is not defined, a default is provided that loads the
302 standard C startup file from the usual place. See @file{gcc.c}.
306 Another C string constant used much like @code{LINK_SPEC}. The
307 difference between the two is that @code{ENDFILE_SPEC} is used at
308 the very end of the command given to the linker.
310 Do not define this macro if it does not need to do anything.
312 @findex THREAD_MODEL_SPEC
313 @item THREAD_MODEL_SPEC
314 GCC @code{-v} will print the thread model GCC was configured to use.
315 However, this doesn't work on platforms that are multilibbed on thread
316 models, such as AIX 4.3. On such platforms, define
317 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
318 blanks that names one of the recognized thread models. @code{%*}, the
319 default value of this macro, will expand to the value of
320 @code{thread_file} set in @file{config.gcc}.
324 Define this macro to provide additional specifications to put in the
325 @file{specs} file that can be used in various specifications like
328 The definition should be an initializer for an array of structures,
329 containing a string constant, that defines the specification name, and a
330 string constant that provides the specification.
332 Do not define this macro if it does not need to do anything.
334 @code{EXTRA_SPECS} is useful when an architecture contains several
335 related targets, which have various @code{@dots{}_SPECS} which are similar
336 to each other, and the maintainer would like one central place to keep
339 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
340 define either @code{_CALL_SYSV} when the System V calling sequence is
341 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
344 The @file{config/rs6000/rs6000.h} target file defines:
347 #define EXTRA_SPECS \
348 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
350 #define CPP_SYS_DEFAULT ""
353 The @file{config/rs6000/sysv.h} target file defines:
357 "%@{posix: -D_POSIX_SOURCE @} \
358 %@{mcall-sysv: -D_CALL_SYSV @} %@{mcall-aix: -D_CALL_AIX @} \
359 %@{!mcall-sysv: %@{!mcall-aix: %(cpp_sysv_default) @}@} \
360 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
362 #undef CPP_SYSV_DEFAULT
363 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
366 while the @file{config/rs6000/eabiaix.h} target file defines
367 @code{CPP_SYSV_DEFAULT} as:
370 #undef CPP_SYSV_DEFAULT
371 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
374 @findex LINK_LIBGCC_SPECIAL
375 @item LINK_LIBGCC_SPECIAL
376 Define this macro if the driver program should find the library
377 @file{libgcc.a} itself and should not pass @option{-L} options to the
378 linker. If you do not define this macro, the driver program will pass
379 the argument @option{-lgcc} to tell the linker to do the search and will
380 pass @option{-L} options to it.
382 @findex LINK_LIBGCC_SPECIAL_1
383 @item LINK_LIBGCC_SPECIAL_1
384 Define this macro if the driver program should find the library
385 @file{libgcc.a}. If you do not define this macro, the driver program will pass
386 the argument @option{-lgcc} to tell the linker to do the search.
387 This macro is similar to @code{LINK_LIBGCC_SPECIAL}, except that it does
388 not affect @option{-L} options.
390 @findex LINK_COMMAND_SPEC
391 @item LINK_COMMAND_SPEC
392 A C string constant giving the complete command line need to execute the
393 linker. When you do this, you will need to update your port each time a
394 change is made to the link command line within @file{gcc.c}. Therefore,
395 define this macro only if you need to completely redefine the command
396 line for invoking the linker and there is no other way to accomplish
399 @findex LINK_ELIMINATE_DUPLICATE_LDIRECTORIES
400 @item LINK_ELIMINATE_DUPLICATE_LDIRECTORIES
401 A nonzero value causes @command{collect2} to remove duplicate @option{-L@var{directory}} search
402 directories from linking commands. Do not give it a nonzero value if
403 removing duplicate search directories changes the linker's semantics.
405 @findex MULTILIB_DEFAULTS
406 @item MULTILIB_DEFAULTS
407 Define this macro as a C expression for the initializer of an array of
408 string to tell the driver program which options are defaults for this
409 target and thus do not need to be handled specially when using
410 @code{MULTILIB_OPTIONS}.
412 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
413 the target makefile fragment or if none of the options listed in
414 @code{MULTILIB_OPTIONS} are set by default.
415 @xref{Target Fragment}.
417 @findex RELATIVE_PREFIX_NOT_LINKDIR
418 @item RELATIVE_PREFIX_NOT_LINKDIR
419 Define this macro to tell @code{gcc} that it should only translate
420 a @option{-B} prefix into a @option{-L} linker option if the prefix
421 indicates an absolute file name.
423 @findex STANDARD_EXEC_PREFIX
424 @item STANDARD_EXEC_PREFIX
425 Define this macro as a C string constant if you wish to override the
426 standard choice of @file{/usr/local/lib/gcc-lib/} as the default prefix to
427 try when searching for the executable files of the compiler.
429 @findex MD_EXEC_PREFIX
431 If defined, this macro is an additional prefix to try after
432 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
433 when the @option{-b} option is used, or the compiler is built as a cross
434 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
435 to the list of directories used to find the assembler in @file{configure.in}.
437 @findex STANDARD_STARTFILE_PREFIX
438 @item STANDARD_STARTFILE_PREFIX
439 Define this macro as a C string constant if you wish to override the
440 standard choice of @file{/usr/local/lib/} as the default prefix to
441 try when searching for startup files such as @file{crt0.o}.
443 @findex MD_STARTFILE_PREFIX
444 @item MD_STARTFILE_PREFIX
445 If defined, this macro supplies an additional prefix to try after the
446 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
447 @option{-b} option is used, or when the compiler is built as a cross
450 @findex MD_STARTFILE_PREFIX_1
451 @item MD_STARTFILE_PREFIX_1
452 If defined, this macro supplies yet another prefix to try after the
453 standard prefixes. It is not searched when the @option{-b} option is
454 used, or when the compiler is built as a cross compiler.
456 @findex INIT_ENVIRONMENT
457 @item INIT_ENVIRONMENT
458 Define this macro as a C string constant if you wish to set environment
459 variables for programs called by the driver, such as the assembler and
460 loader. The driver passes the value of this macro to @code{putenv} to
461 initialize the necessary environment variables.
463 @findex LOCAL_INCLUDE_DIR
464 @item LOCAL_INCLUDE_DIR
465 Define this macro as a C string constant if you wish to override the
466 standard choice of @file{/usr/local/include} as the default prefix to
467 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
468 comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
470 Cross compilers do not search either @file{/usr/local/include} or its
473 @findex MODIFY_TARGET_NAME
474 @item MODIFY_TARGET_NAME
475 Define this macro if you with to define command-line switches that modify the
478 For each switch, you can include a string to be appended to the first
479 part of the configuration name or a string to be deleted from the
480 configuration name, if present. The definition should be an initializer
481 for an array of structures. Each array element should have three
482 elements: the switch name (a string constant, including the initial
483 dash), one of the enumeration codes @code{ADD} or @code{DELETE} to
484 indicate whether the string should be inserted or deleted, and the string
485 to be inserted or deleted (a string constant).
487 For example, on a machine where @samp{64} at the end of the
488 configuration name denotes a 64-bit target and you want the @option{-32}
489 and @option{-64} switches to select between 32- and 64-bit targets, you would
493 #define MODIFY_TARGET_NAME \
494 @{ @{ "-32", DELETE, "64"@}, \
495 @{"-64", ADD, "64"@}@}
499 @findex SYSTEM_INCLUDE_DIR
500 @item SYSTEM_INCLUDE_DIR
501 Define this macro as a C string constant if you wish to specify a
502 system-specific directory to search for header files before the standard
503 directory. @code{SYSTEM_INCLUDE_DIR} comes before
504 @code{STANDARD_INCLUDE_DIR} in the search order.
506 Cross compilers do not use this macro and do not search the directory
509 @findex STANDARD_INCLUDE_DIR
510 @item STANDARD_INCLUDE_DIR
511 Define this macro as a C string constant if you wish to override the
512 standard choice of @file{/usr/include} as the default prefix to
513 try when searching for header files.
515 Cross compilers do not use this macro and do not search either
516 @file{/usr/include} or its replacement.
518 @findex STANDARD_INCLUDE_COMPONENT
519 @item STANDARD_INCLUDE_COMPONENT
520 The ``component'' corresponding to @code{STANDARD_INCLUDE_DIR}.
521 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
522 If you do not define this macro, no component is used.
524 @findex INCLUDE_DEFAULTS
525 @item INCLUDE_DEFAULTS
526 Define this macro if you wish to override the entire default search path
527 for include files. For a native compiler, the default search path
528 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
529 @code{SYSTEM_INCLUDE_DIR}, @code{GPLUSPLUS_INCLUDE_DIR}, and
530 @code{STANDARD_INCLUDE_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
531 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
532 and specify private search areas for GCC@. The directory
533 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
535 The definition should be an initializer for an array of structures.
536 Each array element should have four elements: the directory name (a
537 string constant), the component name (also a string constant), a flag
538 for C++-only directories,
539 and a flag showing that the includes in the directory don't need to be
540 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
541 the array with a null element.
543 The component name denotes what GNU package the include file is part of,
544 if any, in all upper-case letters. For example, it might be @samp{GCC}
545 or @samp{BINUTILS}. If the package is part of a vendor-supplied
546 operating system, code the component name as @samp{0}.
548 For example, here is the definition used for VAX/VMS:
551 #define INCLUDE_DEFAULTS \
553 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
554 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
555 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
562 Here is the order of prefixes tried for exec files:
566 Any prefixes specified by the user with @option{-B}.
569 The environment variable @code{GCC_EXEC_PREFIX}, if any.
572 The directories specified by the environment variable @code{COMPILER_PATH}.
575 The macro @code{STANDARD_EXEC_PREFIX}.
578 @file{/usr/lib/gcc/}.
581 The macro @code{MD_EXEC_PREFIX}, if any.
584 Here is the order of prefixes tried for startfiles:
588 Any prefixes specified by the user with @option{-B}.
591 The environment variable @code{GCC_EXEC_PREFIX}, if any.
594 The directories specified by the environment variable @code{LIBRARY_PATH}
595 (or port-specific name; native only, cross compilers do not use this).
598 The macro @code{STANDARD_EXEC_PREFIX}.
601 @file{/usr/lib/gcc/}.
604 The macro @code{MD_EXEC_PREFIX}, if any.
607 The macro @code{MD_STARTFILE_PREFIX}, if any.
610 The macro @code{STANDARD_STARTFILE_PREFIX}.
619 @node Run-time Target
620 @section Run-time Target Specification
621 @cindex run-time target specification
622 @cindex predefined macros
623 @cindex target specifications
625 @c prevent bad page break with this line
626 Here are run-time target specifications.
629 @findex CPP_PREDEFINES
631 Define this to be a string constant containing @option{-D} options to
632 define the predefined macros that identify this machine and system.
633 These macros will be predefined unless the @option{-ansi} option (or a
634 @option{-std} option for strict ISO C conformance) is specified.
636 In addition, a parallel set of macros are predefined, whose names are
637 made by appending @samp{__} at the beginning and at the end. These
638 @samp{__} macros are permitted by the ISO standard, so they are
639 predefined regardless of whether @option{-ansi} or a @option{-std} option
642 For example, on the Sun, one can use the following value:
645 "-Dmc68000 -Dsun -Dunix"
648 The result is to define the macros @code{__mc68000__}, @code{__sun__}
649 and @code{__unix__} unconditionally, and the macros @code{mc68000},
650 @code{sun} and @code{unix} provided @option{-ansi} is not specified.
652 @findex extern int target_flags
653 @item extern int target_flags;
654 This declaration should be present.
656 @cindex optional hardware or system features
657 @cindex features, optional, in system conventions
659 This series of macros is to allow compiler command arguments to
660 enable or disable the use of optional features of the target machine.
661 For example, one machine description serves both the 68000 and
662 the 68020; a command argument tells the compiler whether it should
663 use 68020-only instructions or not. This command argument works
664 by means of a macro @code{TARGET_68020} that tests a bit in
667 Define a macro @code{TARGET_@var{featurename}} for each such option.
668 Its definition should test a bit in @code{target_flags}. It is
669 recommended that a helper macro @code{TARGET_MASK_@var{featurename}}
670 is defined for each bit-value to test, and used in
671 @code{TARGET_@var{featurename}} and @code{TARGET_SWITCHES}. For
675 #define TARGET_MASK_68020 1
676 #define TARGET_68020 (target_flags & TARGET_MASK_68020)
679 One place where these macros are used is in the condition-expressions
680 of instruction patterns. Note how @code{TARGET_68020} appears
681 frequently in the 68000 machine description file, @file{m68k.md}.
682 Another place they are used is in the definitions of the other
683 macros in the @file{@var{machine}.h} file.
685 @findex TARGET_SWITCHES
686 @item TARGET_SWITCHES
687 This macro defines names of command options to set and clear
688 bits in @code{target_flags}. Its definition is an initializer
689 with a subgrouping for each command option.
691 Each subgrouping contains a string constant, that defines the option
692 name, a number, which contains the bits to set in
693 @code{target_flags}, and a second string which is the description
694 displayed by @option{--help}. If the number is negative then the bits specified
695 by the number are cleared instead of being set. If the description
696 string is present but empty, then no help information will be displayed
697 for that option, but it will not count as an undocumented option. The
698 actual option name is made by appending @samp{-m} to the specified name.
699 Non-empty description strings should be marked with @code{N_(@dots{})} for
700 @command{xgettext}. In addition to the description for @option{--help},
701 more detailed documentation for each option should be added to
704 One of the subgroupings should have a null string. The number in
705 this grouping is the default value for @code{target_flags}. Any
706 target options act starting with that value.
708 Here is an example which defines @option{-m68000} and @option{-m68020}
709 with opposite meanings, and picks the latter as the default:
712 #define TARGET_SWITCHES \
713 @{ @{ "68020", TARGET_MASK_68020, "" @}, \
714 @{ "68000", -TARGET_MASK_68020, \
715 N_("Compile for the 68000") @}, \
716 @{ "", TARGET_MASK_68020, "" @}@}
719 @findex TARGET_OPTIONS
721 This macro is similar to @code{TARGET_SWITCHES} but defines names of command
722 options that have values. Its definition is an initializer with a
723 subgrouping for each command option.
725 Each subgrouping contains a string constant, that defines the fixed part
726 of the option name, the address of a variable, and a description string
727 (which should again be marked with @code{N_(@dots{})}).
728 The variable, type @code{char *}, is set to the variable part of the
729 given option if the fixed part matches. The actual option name is made
730 by appending @samp{-m} to the specified name. Again, each option should
731 also be documented in @file{invoke.texi}.
733 Here is an example which defines @option{-mshort-data-@var{number}}. If the
734 given option is @option{-mshort-data-512}, the variable @code{m88k_short_data}
735 will be set to the string @code{"512"}.
738 extern char *m88k_short_data;
739 #define TARGET_OPTIONS \
740 @{ @{ "short-data-", &m88k_short_data, \
741 N_("Specify the size of the short data section") @} @}
744 @findex TARGET_VERSION
746 This macro is a C statement to print on @code{stderr} a string
747 describing the particular machine description choice. Every machine
748 description should define @code{TARGET_VERSION}. For example:
752 #define TARGET_VERSION \
753 fprintf (stderr, " (68k, Motorola syntax)");
755 #define TARGET_VERSION \
756 fprintf (stderr, " (68k, MIT syntax)");
760 @findex OVERRIDE_OPTIONS
761 @item OVERRIDE_OPTIONS
762 Sometimes certain combinations of command options do not make sense on
763 a particular target machine. You can define a macro
764 @code{OVERRIDE_OPTIONS} to take account of this. This macro, if
765 defined, is executed once just after all the command options have been
768 Don't use this macro to turn on various extra optimizations for
769 @option{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for.
771 @findex OPTIMIZATION_OPTIONS
772 @item OPTIMIZATION_OPTIONS (@var{level}, @var{size})
773 Some machines may desire to change what optimizations are performed for
774 various optimization levels. This macro, if defined, is executed once
775 just after the optimization level is determined and before the remainder
776 of the command options have been parsed. Values set in this macro are
777 used as the default values for the other command line options.
779 @var{level} is the optimization level specified; 2 if @option{-O2} is
780 specified, 1 if @option{-O} is specified, and 0 if neither is specified.
782 @var{size} is nonzero if @option{-Os} is specified and zero otherwise.
784 You should not use this macro to change options that are not
785 machine-specific. These should uniformly selected by the same
786 optimization level on all supported machines. Use this macro to enable
787 machine-specific optimizations.
789 @strong{Do not examine @code{write_symbols} in
790 this macro!} The debugging options are not supposed to alter the
793 @findex CAN_DEBUG_WITHOUT_FP
794 @item CAN_DEBUG_WITHOUT_FP
795 Define this macro if debugging can be performed even without a frame
796 pointer. If this macro is defined, GCC will turn on the
797 @option{-fomit-frame-pointer} option whenever @option{-O} is specified.
800 @node Per-Function Data
801 @section Defining data structures for per-function information.
802 @cindex per-function data
803 @cindex data structures
805 If the target needs to store information on a per-function basis, GCC
806 provides a macro and a couple of variables to allow this. Note, just
807 using statics to store the information is a bad idea, since GCC supports
808 nested functions, so you can be halfway through encoding one function
809 when another one comes along.
811 GCC defines a data structure called @code{struct function} which
812 contains all of the data specific to an individual function. This
813 structure contains a field called @code{machine} whose type is
814 @code{struct machine_function *}, which can be used by targets to point
815 to their own specific data.
817 If a target needs per-function specific data it should define the type
818 @code{struct machine_function} and also the macro
819 @code{INIT_EXPANDERS}. This macro should be used to initialize some or
820 all of the function pointers @code{init_machine_status},
821 @code{free_machine_status} and @code{mark_machine_status}. These
822 pointers are explained below.
824 One typical use of per-function, target specific data is to create an
825 RTX to hold the register containing the function's return address. This
826 RTX can then be used to implement the @code{__builtin_return_address}
827 function, for level 0.
829 Note---earlier implementations of GCC used a single data area to hold
830 all of the per-function information. Thus when processing of a nested
831 function began the old per-function data had to be pushed onto a
832 stack, and when the processing was finished, it had to be popped off the
833 stack. GCC used to provide function pointers called
834 @code{save_machine_status} and @code{restore_machine_status} to handle
835 the saving and restoring of the target specific information. Since the
836 single data area approach is no longer used, these pointers are no
839 The macro and function pointers are described below.
842 @findex INIT_EXPANDERS
844 Macro called to initialize any target specific information. This macro
845 is called once per function, before generation of any RTL has begun.
846 The intention of this macro is to allow the initialization of the
847 function pointers below.
849 @findex init_machine_status
850 @item init_machine_status
851 This is a @code{void (*)(struct function *)} function pointer. If this
852 pointer is non-@code{NULL} it will be called once per function, before function
853 compilation starts, in order to allow the target to perform any target
854 specific initialization of the @code{struct function} structure. It is
855 intended that this would be used to initialize the @code{machine} of
858 @findex free_machine_status
859 @item free_machine_status
860 This is a @code{void (*)(struct function *)} function pointer. If this
861 pointer is non-@code{NULL} it will be called once per function, after the
862 function has been compiled, in order to allow any memory allocated
863 during the @code{init_machine_status} function call to be freed.
865 @findex mark_machine_status
866 @item mark_machine_status
867 This is a @code{void (*)(struct function *)} function pointer. If this
868 pointer is non-@code{NULL} it will be called once per function in order to mark
869 any data items in the @code{struct machine_function} structure which
870 need garbage collection.
875 @section Storage Layout
876 @cindex storage layout
878 Note that the definitions of the macros in this table which are sizes or
879 alignments measured in bits do not need to be constant. They can be C
880 expressions that refer to static variables, such as the @code{target_flags}.
881 @xref{Run-time Target}.
884 @findex BITS_BIG_ENDIAN
885 @item BITS_BIG_ENDIAN
886 Define this macro to have the value 1 if the most significant bit in a
887 byte has the lowest number; otherwise define it to have the value zero.
888 This means that bit-field instructions count from the most significant
889 bit. If the machine has no bit-field instructions, then this must still
890 be defined, but it doesn't matter which value it is defined to. This
891 macro need not be a constant.
893 This macro does not affect the way structure fields are packed into
894 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
896 @findex BYTES_BIG_ENDIAN
897 @item BYTES_BIG_ENDIAN
898 Define this macro to have the value 1 if the most significant byte in a
899 word has the lowest number. This macro need not be a constant.
901 @findex WORDS_BIG_ENDIAN
902 @item WORDS_BIG_ENDIAN
903 Define this macro to have the value 1 if, in a multiword object, the
904 most significant word has the lowest number. This applies to both
905 memory locations and registers; GCC fundamentally assumes that the
906 order of words in memory is the same as the order in registers. This
907 macro need not be a constant.
909 @findex LIBGCC2_WORDS_BIG_ENDIAN
910 @item LIBGCC2_WORDS_BIG_ENDIAN
911 Define this macro if @code{WORDS_BIG_ENDIAN} is not constant. This must be a
912 constant value with the same meaning as @code{WORDS_BIG_ENDIAN}, which will be
913 used only when compiling @file{libgcc2.c}. Typically the value will be set
914 based on preprocessor defines.
916 @findex FLOAT_WORDS_BIG_ENDIAN
917 @item FLOAT_WORDS_BIG_ENDIAN
918 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
919 @code{TFmode} floating point numbers are stored in memory with the word
920 containing the sign bit at the lowest address; otherwise define it to
921 have the value 0. This macro need not be a constant.
923 You need not define this macro if the ordering is the same as for
926 @findex BITS_PER_UNIT
928 Define this macro to be the number of bits in an addressable storage
929 unit (byte); normally 8.
931 @findex BITS_PER_WORD
933 Number of bits in a word; normally 32.
935 @findex MAX_BITS_PER_WORD
936 @item MAX_BITS_PER_WORD
937 Maximum number of bits in a word. If this is undefined, the default is
938 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
939 largest value that @code{BITS_PER_WORD} can have at run-time.
941 @findex UNITS_PER_WORD
943 Number of storage units in a word; normally 4.
945 @findex MIN_UNITS_PER_WORD
946 @item MIN_UNITS_PER_WORD
947 Minimum number of units in a word. If this is undefined, the default is
948 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
949 smallest value that @code{UNITS_PER_WORD} can have at run-time.
953 Width of a pointer, in bits. You must specify a value no wider than the
954 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
955 you must define @code{POINTERS_EXTEND_UNSIGNED}.
957 @findex POINTERS_EXTEND_UNSIGNED
958 @item POINTERS_EXTEND_UNSIGNED
959 A C expression whose value is greater than zero if pointers that need to be
960 extended from being @code{POINTER_SIZE} bits wide to @code{Pmode} are to
961 be zero-extended and zero if they are to be sign-extended. If the value
962 is less then zero then there must be an "ptr_extend" instruction that
963 extends a pointer from @code{POINTER_SIZE} to @code{Pmode}.
965 You need not define this macro if the @code{POINTER_SIZE} is equal
966 to the width of @code{Pmode}.
969 @item PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
970 A macro to update @var{m} and @var{unsignedp} when an object whose type
971 is @var{type} and which has the specified mode and signedness is to be
972 stored in a register. This macro is only called when @var{type} is a
975 On most RISC machines, which only have operations that operate on a full
976 register, define this macro to set @var{m} to @code{word_mode} if
977 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
978 cases, only integer modes should be widened because wider-precision
979 floating-point operations are usually more expensive than their narrower
982 For most machines, the macro definition does not change @var{unsignedp}.
983 However, some machines, have instructions that preferentially handle
984 either signed or unsigned quantities of certain modes. For example, on
985 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
986 sign-extend the result to 64 bits. On such machines, set
987 @var{unsignedp} according to which kind of extension is more efficient.
989 Do not define this macro if it would never modify @var{m}.
991 @findex PROMOTE_FUNCTION_ARGS
992 @item PROMOTE_FUNCTION_ARGS
993 Define this macro if the promotion described by @code{PROMOTE_MODE}
994 should also be done for outgoing function arguments.
996 @findex PROMOTE_FUNCTION_RETURN
997 @item PROMOTE_FUNCTION_RETURN
998 Define this macro if the promotion described by @code{PROMOTE_MODE}
999 should also be done for the return value of functions.
1001 If this macro is defined, @code{FUNCTION_VALUE} must perform the same
1002 promotions done by @code{PROMOTE_MODE}.
1004 @findex PROMOTE_FOR_CALL_ONLY
1005 @item PROMOTE_FOR_CALL_ONLY
1006 Define this macro if the promotion described by @code{PROMOTE_MODE}
1007 should @emph{only} be performed for outgoing function arguments or
1008 function return values, as specified by @code{PROMOTE_FUNCTION_ARGS}
1009 and @code{PROMOTE_FUNCTION_RETURN}, respectively.
1011 @findex PARM_BOUNDARY
1013 Normal alignment required for function parameters on the stack, in
1014 bits. All stack parameters receive at least this much alignment
1015 regardless of data type. On most machines, this is the same as the
1018 @findex STACK_BOUNDARY
1019 @item STACK_BOUNDARY
1020 Define this macro to the minimum alignment enforced by hardware for the
1021 stack pointer on this machine. The definition is a C expression for the
1022 desired alignment (measured in bits). This value is used as a default
1023 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
1024 this should be the same as @code{PARM_BOUNDARY}.
1026 @findex PREFERRED_STACK_BOUNDARY
1027 @item PREFERRED_STACK_BOUNDARY
1028 Define this macro if you wish to preserve a certain alignment for the
1029 stack pointer, greater than what the hardware enforces. The definition
1030 is a C expression for the desired alignment (measured in bits). This
1031 macro must evaluate to a value equal to or larger than
1032 @code{STACK_BOUNDARY}.
1034 @findex FORCE_PREFERRED_STACK_BOUNDARY_IN_MAIN
1035 @item FORCE_PREFERRED_STACK_BOUNDARY_IN_MAIN
1036 A C expression that evaluates true if @code{PREFERRED_STACK_BOUNDARY} is
1037 not guaranteed by the runtime and we should emit code to align the stack
1038 at the beginning of @code{main}.
1040 @cindex @code{PUSH_ROUNDING}, interaction with @code{PREFERRED_STACK_BOUNDARY}
1041 If @code{PUSH_ROUNDING} is not defined, the stack will always be aligned
1042 to the specified boundary. If @code{PUSH_ROUNDING} is defined and specifies
1043 a less strict alignment than @code{PREFERRED_STACK_BOUNDARY}, the stack may
1044 be momentarily unaligned while pushing arguments.
1046 @findex FUNCTION_BOUNDARY
1047 @item FUNCTION_BOUNDARY
1048 Alignment required for a function entry point, in bits.
1050 @findex BIGGEST_ALIGNMENT
1051 @item BIGGEST_ALIGNMENT
1052 Biggest alignment that any data type can require on this machine, in bits.
1054 @findex MINIMUM_ATOMIC_ALIGNMENT
1055 @item MINIMUM_ATOMIC_ALIGNMENT
1056 If defined, the smallest alignment, in bits, that can be given to an
1057 object that can be referenced in one operation, without disturbing any
1058 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1059 on machines that don't have byte or half-word store operations.
1061 @findex BIGGEST_FIELD_ALIGNMENT
1062 @item BIGGEST_FIELD_ALIGNMENT
1063 Biggest alignment that any structure or union field can require on this
1064 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1065 structure and union fields only, unless the field alignment has been set
1066 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1068 @findex ADJUST_FIELD_ALIGN
1069 @item ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1070 An expression for the alignment of a structure field @var{field} if the
1071 alignment computed in the usual way is @var{computed}. GCC uses
1072 this value instead of the value in @code{BIGGEST_ALIGNMENT} or
1073 @code{BIGGEST_FIELD_ALIGNMENT}, if defined.
1075 @findex MAX_OFILE_ALIGNMENT
1076 @item MAX_OFILE_ALIGNMENT
1077 Biggest alignment supported by the object file format of this machine.
1078 Use this macro to limit the alignment which can be specified using the
1079 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1080 the default value is @code{BIGGEST_ALIGNMENT}.
1082 @findex DATA_ALIGNMENT
1083 @item DATA_ALIGNMENT (@var{type}, @var{basic-align})
1084 If defined, a C expression to compute the alignment for a variable in
1085 the static store. @var{type} is the data type, and @var{basic-align} is
1086 the alignment that the object would ordinarily have. The value of this
1087 macro is used instead of that alignment to align the object.
1089 If this macro is not defined, then @var{basic-align} is used.
1092 One use of this macro is to increase alignment of medium-size data to
1093 make it all fit in fewer cache lines. Another is to cause character
1094 arrays to be word-aligned so that @code{strcpy} calls that copy
1095 constants to character arrays can be done inline.
1097 @findex CONSTANT_ALIGNMENT
1098 @item CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1099 If defined, a C expression to compute the alignment given to a constant
1100 that is being placed in memory. @var{constant} is the constant and
1101 @var{basic-align} is the alignment that the object would ordinarily
1102 have. The value of this macro is used instead of that alignment to
1105 If this macro is not defined, then @var{basic-align} is used.
1107 The typical use of this macro is to increase alignment for string
1108 constants to be word aligned so that @code{strcpy} calls that copy
1109 constants can be done inline.
1111 @findex LOCAL_ALIGNMENT
1112 @item LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1113 If defined, a C expression to compute the alignment for a variable in
1114 the local store. @var{type} is the data type, and @var{basic-align} is
1115 the alignment that the object would ordinarily have. The value of this
1116 macro is used instead of that alignment to align the object.
1118 If this macro is not defined, then @var{basic-align} is used.
1120 One use of this macro is to increase alignment of medium-size data to
1121 make it all fit in fewer cache lines.
1123 @findex EMPTY_FIELD_BOUNDARY
1124 @item EMPTY_FIELD_BOUNDARY
1125 Alignment in bits to be given to a structure bit-field that follows an
1126 empty field such as @code{int : 0;}.
1128 Note that @code{PCC_BITFIELD_TYPE_MATTERS} also affects the alignment
1129 that results from an empty field.
1131 @findex STRUCTURE_SIZE_BOUNDARY
1132 @item STRUCTURE_SIZE_BOUNDARY
1133 Number of bits which any structure or union's size must be a multiple of.
1134 Each structure or union's size is rounded up to a multiple of this.
1136 If you do not define this macro, the default is the same as
1137 @code{BITS_PER_UNIT}.
1139 @findex STRICT_ALIGNMENT
1140 @item STRICT_ALIGNMENT
1141 Define this macro to be the value 1 if instructions will fail to work
1142 if given data not on the nominal alignment. If instructions will merely
1143 go slower in that case, define this macro as 0.
1145 @findex PCC_BITFIELD_TYPE_MATTERS
1146 @item PCC_BITFIELD_TYPE_MATTERS
1147 Define this if you wish to imitate the way many other C compilers handle
1148 alignment of bit-fields and the structures that contain them.
1150 The behavior is that the type written for a bit-field (@code{int},
1151 @code{short}, or other integer type) imposes an alignment for the
1152 entire structure, as if the structure really did contain an ordinary
1153 field of that type. In addition, the bit-field is placed within the
1154 structure so that it would fit within such a field, not crossing a
1157 Thus, on most machines, a bit-field whose type is written as @code{int}
1158 would not cross a four-byte boundary, and would force four-byte
1159 alignment for the whole structure. (The alignment used may not be four
1160 bytes; it is controlled by the other alignment parameters.)
1162 If the macro is defined, its definition should be a C expression;
1163 a nonzero value for the expression enables this behavior.
1165 Note that if this macro is not defined, or its value is zero, some
1166 bit-fields may cross more than one alignment boundary. The compiler can
1167 support such references if there are @samp{insv}, @samp{extv}, and
1168 @samp{extzv} insns that can directly reference memory.
1170 The other known way of making bit-fields work is to define
1171 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1172 Then every structure can be accessed with fullwords.
1174 Unless the machine has bit-field instructions or you define
1175 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1176 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1178 If your aim is to make GCC use the same conventions for laying out
1179 bit-fields as are used by another compiler, here is how to investigate
1180 what the other compiler does. Compile and run this program:
1199 printf ("Size of foo1 is %d\n",
1200 sizeof (struct foo1));
1201 printf ("Size of foo2 is %d\n",
1202 sizeof (struct foo2));
1207 If this prints 2 and 5, then the compiler's behavior is what you would
1208 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1210 @findex BITFIELD_NBYTES_LIMITED
1211 @item BITFIELD_NBYTES_LIMITED
1212 Like PCC_BITFIELD_TYPE_MATTERS except that its effect is limited to
1213 aligning a bit-field within the structure.
1215 @findex MEMBER_TYPE_FORCES_BLK
1216 @item MEMBER_TYPE_FORCES_BLK (@var{field})
1217 Return 1 if a structure or array containing @var{field} should be accessed using
1220 Normally, this is not needed. See the file @file{c4x.h} for an example
1221 of how to use this macro to prevent a structure having a floating point
1222 field from being accessed in an integer mode.
1224 @findex ROUND_TYPE_SIZE
1225 @item ROUND_TYPE_SIZE (@var{type}, @var{computed}, @var{specified})
1226 Define this macro as an expression for the overall size of a type
1227 (given by @var{type} as a tree node) when the size computed in the
1228 usual way is @var{computed} and the alignment is @var{specified}.
1230 The default is to round @var{computed} up to a multiple of @var{specified}.
1232 @findex ROUND_TYPE_SIZE_UNIT
1233 @item ROUND_TYPE_SIZE_UNIT (@var{type}, @var{computed}, @var{specified})
1234 Similar to @code{ROUND_TYPE_SIZE}, but sizes and alignments are
1235 specified in units (bytes). If you define @code{ROUND_TYPE_SIZE},
1236 you must also define this macro and they must be defined consistently
1239 @findex ROUND_TYPE_ALIGN
1240 @item ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1241 Define this macro as an expression for the alignment of a type (given
1242 by @var{type} as a tree node) if the alignment computed in the usual
1243 way is @var{computed} and the alignment explicitly specified was
1246 The default is to use @var{specified} if it is larger; otherwise, use
1247 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1249 @findex MAX_FIXED_MODE_SIZE
1250 @item MAX_FIXED_MODE_SIZE
1251 An integer expression for the size in bits of the largest integer
1252 machine mode that should actually be used. All integer machine modes of
1253 this size or smaller can be used for structures and unions with the
1254 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1255 (DImode)} is assumed.
1257 @findex VECTOR_MODE_SUPPORTED_P
1258 @item VECTOR_MODE_SUPPORTED_P(@var{mode})
1259 Define this macro to be nonzero if the port is prepared to handle insns
1260 involving vector mode @var{mode}. At the very least, it must have move
1261 patterns for this mode.
1263 @findex STACK_SAVEAREA_MODE
1264 @item STACK_SAVEAREA_MODE (@var{save_level})
1265 If defined, an expression of type @code{enum machine_mode} that
1266 specifies the mode of the save area operand of a
1267 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1268 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1269 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1270 having its mode specified.
1272 You need not define this macro if it always returns @code{Pmode}. You
1273 would most commonly define this macro if the
1274 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1277 @findex STACK_SIZE_MODE
1278 @item STACK_SIZE_MODE
1279 If defined, an expression of type @code{enum machine_mode} that
1280 specifies the mode of the size increment operand of an
1281 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1283 You need not define this macro if it always returns @code{word_mode}.
1284 You would most commonly define this macro if the @code{allocate_stack}
1285 pattern needs to support both a 32- and a 64-bit mode.
1287 @findex CHECK_FLOAT_VALUE
1288 @item CHECK_FLOAT_VALUE (@var{mode}, @var{value}, @var{overflow})
1289 A C statement to validate the value @var{value} (of type
1290 @code{double}) for mode @var{mode}. This means that you check whether
1291 @var{value} fits within the possible range of values for mode
1292 @var{mode} on this target machine. The mode @var{mode} is always
1293 a mode of class @code{MODE_FLOAT}. @var{overflow} is nonzero if
1294 the value is already known to be out of range.
1296 If @var{value} is not valid or if @var{overflow} is nonzero, you should
1297 set @var{overflow} to 1 and then assign some valid value to @var{value}.
1298 Allowing an invalid value to go through the compiler can produce
1299 incorrect assembler code which may even cause Unix assemblers to crash.
1301 This macro need not be defined if there is no work for it to do.
1303 @findex TARGET_FLOAT_FORMAT
1304 @item TARGET_FLOAT_FORMAT
1305 A code distinguishing the floating point format of the target machine.
1306 There are five defined values:
1309 @findex IEEE_FLOAT_FORMAT
1310 @item IEEE_FLOAT_FORMAT
1311 This code indicates IEEE floating point. It is the default; there is no
1312 need to define this macro when the format is IEEE@.
1314 @findex VAX_FLOAT_FORMAT
1315 @item VAX_FLOAT_FORMAT
1316 This code indicates the peculiar format used on the VAX.
1318 @findex IBM_FLOAT_FORMAT
1319 @item IBM_FLOAT_FORMAT
1320 This code indicates the format used on the IBM System/370.
1322 @findex C4X_FLOAT_FORMAT
1323 @item C4X_FLOAT_FORMAT
1324 This code indicates the format used on the TMS320C3x/C4x.
1326 @findex UNKNOWN_FLOAT_FORMAT
1327 @item UNKNOWN_FLOAT_FORMAT
1328 This code indicates any other format.
1331 The value of this macro is compared with @code{HOST_FLOAT_FORMAT}
1332 (@pxref{Config}) to determine whether the target machine has the same
1333 format as the host machine. If any other formats are actually in use on
1334 supported machines, new codes should be defined for them.
1336 The ordering of the component words of floating point values stored in
1337 memory is controlled by @code{FLOAT_WORDS_BIG_ENDIAN} for the target
1338 machine and @code{HOST_FLOAT_WORDS_BIG_ENDIAN} for the host.
1343 @section Layout of Source Language Data Types
1345 These macros define the sizes and other characteristics of the standard
1346 basic data types used in programs being compiled. Unlike the macros in
1347 the previous section, these apply to specific features of C and related
1348 languages, rather than to fundamental aspects of storage layout.
1351 @findex INT_TYPE_SIZE
1353 A C expression for the size in bits of the type @code{int} on the
1354 target machine. If you don't define this, the default is one word.
1356 @findex MAX_INT_TYPE_SIZE
1357 @item MAX_INT_TYPE_SIZE
1358 Maximum number for the size in bits of the type @code{int} on the target
1359 machine. If this is undefined, the default is @code{INT_TYPE_SIZE}.
1360 Otherwise, it is the constant value that is the largest value that
1361 @code{INT_TYPE_SIZE} can have at run-time. This is used in @code{cpp}.
1363 @findex SHORT_TYPE_SIZE
1364 @item SHORT_TYPE_SIZE
1365 A C expression for the size in bits of the type @code{short} on the
1366 target machine. If you don't define this, the default is half a word.
1367 (If this would be less than one storage unit, it is rounded up to one
1370 @findex LONG_TYPE_SIZE
1371 @item LONG_TYPE_SIZE
1372 A C expression for the size in bits of the type @code{long} on the
1373 target machine. If you don't define this, the default is one word.
1375 @findex ADA_LONG_TYPE_SIZE
1376 @item ADA_LONG_TYPE_SIZE
1377 On some machines, the size used for the Ada equivalent of the type
1378 @code{long} by a native Ada compiler differs from that used by C. In
1379 that situation, define this macro to be a C expression to be used for
1380 the size of that type. If you don't define this, the default is the
1381 value of @code{LONG_TYPE_SIZE}.
1383 @findex MAX_LONG_TYPE_SIZE
1384 @item MAX_LONG_TYPE_SIZE
1385 Maximum number for the size in bits of the type @code{long} on the
1386 target machine. If this is undefined, the default is
1387 @code{LONG_TYPE_SIZE}. Otherwise, it is the constant value that is the
1388 largest value that @code{LONG_TYPE_SIZE} can have at run-time. This is
1391 @findex LONG_LONG_TYPE_SIZE
1392 @item LONG_LONG_TYPE_SIZE
1393 A C expression for the size in bits of the type @code{long long} on the
1394 target machine. If you don't define this, the default is two
1395 words. If you want to support GNU Ada on your machine, the value of this
1396 macro must be at least 64.
1398 @findex CHAR_TYPE_SIZE
1399 @item CHAR_TYPE_SIZE
1400 A C expression for the size in bits of the type @code{char} on the
1401 target machine. If you don't define this, the default is
1402 @code{BITS_PER_UNIT}.
1404 @findex MAX_CHAR_TYPE_SIZE
1405 @item MAX_CHAR_TYPE_SIZE
1406 Maximum number for the size in bits of the type @code{char} on the
1407 target machine. If this is undefined, the default is
1408 @code{CHAR_TYPE_SIZE}. Otherwise, it is the constant value that is the
1409 largest value that @code{CHAR_TYPE_SIZE} can have at run-time. This is
1412 @findex BOOL_TYPE_SIZE
1413 @item BOOL_TYPE_SIZE
1414 A C expression for the size in bits of the C++ type @code{bool} on the
1415 target machine. If you don't define this, the default is
1416 @code{CHAR_TYPE_SIZE}.
1418 @findex FLOAT_TYPE_SIZE
1419 @item FLOAT_TYPE_SIZE
1420 A C expression for the size in bits of the type @code{float} on the
1421 target machine. If you don't define this, the default is one word.
1423 @findex DOUBLE_TYPE_SIZE
1424 @item DOUBLE_TYPE_SIZE
1425 A C expression for the size in bits of the type @code{double} on the
1426 target machine. If you don't define this, the default is two
1429 @findex LONG_DOUBLE_TYPE_SIZE
1430 @item LONG_DOUBLE_TYPE_SIZE
1431 A C expression for the size in bits of the type @code{long double} on
1432 the target machine. If you don't define this, the default is two
1435 @findex MAX_LONG_DOUBLE_TYPE_SIZE
1436 Maximum number for the size in bits of the type @code{long double} on the
1437 target machine. If this is undefined, the default is
1438 @code{LONG_DOUBLE_TYPE_SIZE}. Otherwise, it is the constant value that is
1439 the largest value that @code{LONG_DOUBLE_TYPE_SIZE} can have at run-time.
1440 This is used in @code{cpp}.
1442 @findex INTEL_EXTENDED_IEEE_FORMAT
1443 Define this macro to be 1 if the target machine uses 80-bit floating-point
1444 values with 128-bit size and alignment. This is used in @file{real.c}.
1446 @findex WIDEST_HARDWARE_FP_SIZE
1447 @item WIDEST_HARDWARE_FP_SIZE
1448 A C expression for the size in bits of the widest floating-point format
1449 supported by the hardware. If you define this macro, you must specify a
1450 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1451 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1454 @findex DEFAULT_SIGNED_CHAR
1455 @item DEFAULT_SIGNED_CHAR
1456 An expression whose value is 1 or 0, according to whether the type
1457 @code{char} should be signed or unsigned by default. The user can
1458 always override this default with the options @option{-fsigned-char}
1459 and @option{-funsigned-char}.
1461 @findex DEFAULT_SHORT_ENUMS
1462 @item DEFAULT_SHORT_ENUMS
1463 A C expression to determine whether to give an @code{enum} type
1464 only as many bytes as it takes to represent the range of possible values
1465 of that type. A nonzero value means to do that; a zero value means all
1466 @code{enum} types should be allocated like @code{int}.
1468 If you don't define the macro, the default is 0.
1472 A C expression for a string describing the name of the data type to use
1473 for size values. The typedef name @code{size_t} is defined using the
1474 contents of the string.
1476 The string can contain more than one keyword. If so, separate them with
1477 spaces, and write first any length keyword, then @code{unsigned} if
1478 appropriate, and finally @code{int}. The string must exactly match one
1479 of the data type names defined in the function
1480 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1481 omit @code{int} or change the order---that would cause the compiler to
1484 If you don't define this macro, the default is @code{"long unsigned
1487 @findex PTRDIFF_TYPE
1489 A C expression for a string describing the name of the data type to use
1490 for the result of subtracting two pointers. The typedef name
1491 @code{ptrdiff_t} is defined using the contents of the string. See
1492 @code{SIZE_TYPE} above for more information.
1494 If you don't define this macro, the default is @code{"long int"}.
1498 A C expression for a string describing the name of the data type to use
1499 for wide characters. The typedef name @code{wchar_t} is defined using
1500 the contents of the string. See @code{SIZE_TYPE} above for more
1503 If you don't define this macro, the default is @code{"int"}.
1505 @findex WCHAR_TYPE_SIZE
1506 @item WCHAR_TYPE_SIZE
1507 A C expression for the size in bits of the data type for wide
1508 characters. This is used in @code{cpp}, which cannot make use of
1511 @findex MAX_WCHAR_TYPE_SIZE
1512 @item MAX_WCHAR_TYPE_SIZE
1513 Maximum number for the size in bits of the data type for wide
1514 characters. If this is undefined, the default is
1515 @code{WCHAR_TYPE_SIZE}. Otherwise, it is the constant value that is the
1516 largest value that @code{WCHAR_TYPE_SIZE} can have at run-time. This is
1519 @findex GCOV_TYPE_SIZE
1520 @item GCOV_TYPE_SIZE
1521 A C expression for the size in bits of the type used for gcov counters on the
1522 target machine. If you don't define this, the default is one
1523 @code{LONG_TYPE_SIZE} in case it is greater or equal to 64-bit and
1524 @code{LONG_LONG_TYPE_SIZE} otherwise. You may want to re-define the type to
1525 ensure atomicity for counters in multithreaded programs.
1529 A C expression for a string describing the name of the data type to
1530 use for wide characters passed to @code{printf} and returned from
1531 @code{getwc}. The typedef name @code{wint_t} is defined using the
1532 contents of the string. See @code{SIZE_TYPE} above for more
1535 If you don't define this macro, the default is @code{"unsigned int"}.
1539 A C expression for a string describing the name of the data type that
1540 can represent any value of any standard or extended signed integer type.
1541 The typedef name @code{intmax_t} is defined using the contents of the
1542 string. See @code{SIZE_TYPE} above for more information.
1544 If you don't define this macro, the default is the first of
1545 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1546 much precision as @code{long long int}.
1548 @findex UINTMAX_TYPE
1550 A C expression for a string describing the name of the data type that
1551 can represent any value of any standard or extended unsigned integer
1552 type. The typedef name @code{uintmax_t} is defined using the contents
1553 of the string. See @code{SIZE_TYPE} above for more information.
1555 If you don't define this macro, the default is the first of
1556 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1557 unsigned int"} that has as much precision as @code{long long unsigned
1560 @findex OBJC_SELECTORS_WITHOUT_LABELS
1561 @item OBJC_SELECTORS_WITHOUT_LABELS
1562 Define this macro if the compiler can group all the selectors together
1563 into a vector and use just one label at the beginning of the vector.
1564 Otherwise, the compiler must give each selector its own assembler
1567 On certain machines, it is important to have a separate label for each
1568 selector because this enables the linker to eliminate duplicate selectors.
1570 @findex TARGET_PTRMEMFUNC_VBIT_LOCATION
1571 @item TARGET_PTRMEMFUNC_VBIT_LOCATION
1572 The C++ compiler represents a pointer-to-member-function with a struct
1579 ptrdiff_t vtable_index;
1586 The C++ compiler must use one bit to indicate whether the function that
1587 will be called through a pointer-to-member-function is virtual.
1588 Normally, we assume that the low-order bit of a function pointer must
1589 always be zero. Then, by ensuring that the vtable_index is odd, we can
1590 distinguish which variant of the union is in use. But, on some
1591 platforms function pointers can be odd, and so this doesn't work. In
1592 that case, we use the low-order bit of the @code{delta} field, and shift
1593 the remainder of the @code{delta} field to the left.
1595 GCC will automatically make the right selection about where to store
1596 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1597 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1598 set such that functions always start at even addresses, but the lowest
1599 bit of pointers to functions indicate whether the function at that
1600 address is in ARM or Thumb mode. If this is the case of your
1601 architecture, you should define this macro to
1602 @code{ptrmemfunc_vbit_in_delta}.
1604 In general, you should not have to define this macro. On architectures
1605 in which function addresses are always even, according to
1606 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1607 @code{ptrmemfunc_vbit_in_pfn}.
1609 @findex TARGET_VTABLE_USES_DESCRIPTORS
1610 @item TARGET_VTABLE_USES_DESCRIPTORS
1611 Normally, the C++ compiler uses function pointers in vtables. This
1612 macro allows the target to change to use ``function descriptors''
1613 instead. Function descriptors are found on targets for whom a
1614 function pointer is actually a small data structure. Normally the
1615 data structure consists of the actual code address plus a data
1616 pointer to which the function's data is relative.
1618 If vtables are used, the value of this macro should be the number
1619 of words that the function descriptor occupies.
1622 @node Escape Sequences
1623 @section Target Character Escape Sequences
1624 @cindex escape sequences
1626 By default, GCC assumes that the C character escape sequences take on
1627 their ASCII values for the target. If this is not correct, you must
1628 explicitly define all of the macros below.
1633 A C constant expression for the integer value for escape sequence
1638 A C constant expression for the integer value of the target escape
1639 character. As an extension, GCC evaluates the escape sequences
1640 @samp{\e} and @samp{\E} to this.
1644 @findex TARGET_NEWLINE
1647 @itemx TARGET_NEWLINE
1648 C constant expressions for the integer values for escape sequences
1649 @samp{\b}, @samp{\t} and @samp{\n}.
1657 C constant expressions for the integer values for escape sequences
1658 @samp{\v}, @samp{\f} and @samp{\r}.
1662 @section Register Usage
1663 @cindex register usage
1665 This section explains how to describe what registers the target machine
1666 has, and how (in general) they can be used.
1668 The description of which registers a specific instruction can use is
1669 done with register classes; see @ref{Register Classes}. For information
1670 on using registers to access a stack frame, see @ref{Frame Registers}.
1671 For passing values in registers, see @ref{Register Arguments}.
1672 For returning values in registers, see @ref{Scalar Return}.
1675 * Register Basics:: Number and kinds of registers.
1676 * Allocation Order:: Order in which registers are allocated.
1677 * Values in Registers:: What kinds of values each reg can hold.
1678 * Leaf Functions:: Renumbering registers for leaf functions.
1679 * Stack Registers:: Handling a register stack such as 80387.
1682 @node Register Basics
1683 @subsection Basic Characteristics of Registers
1685 @c prevent bad page break with this line
1686 Registers have various characteristics.
1689 @findex FIRST_PSEUDO_REGISTER
1690 @item FIRST_PSEUDO_REGISTER
1691 Number of hardware registers known to the compiler. They receive
1692 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1693 pseudo register's number really is assigned the number
1694 @code{FIRST_PSEUDO_REGISTER}.
1696 @item FIXED_REGISTERS
1697 @findex FIXED_REGISTERS
1698 @cindex fixed register
1699 An initializer that says which registers are used for fixed purposes
1700 all throughout the compiled code and are therefore not available for
1701 general allocation. These would include the stack pointer, the frame
1702 pointer (except on machines where that can be used as a general
1703 register when no frame pointer is needed), the program counter on
1704 machines where that is considered one of the addressable registers,
1705 and any other numbered register with a standard use.
1707 This information is expressed as a sequence of numbers, separated by
1708 commas and surrounded by braces. The @var{n}th number is 1 if
1709 register @var{n} is fixed, 0 otherwise.
1711 The table initialized from this macro, and the table initialized by
1712 the following one, may be overridden at run time either automatically,
1713 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1714 the user with the command options @option{-ffixed-@var{reg}},
1715 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1717 @findex CALL_USED_REGISTERS
1718 @item CALL_USED_REGISTERS
1719 @cindex call-used register
1720 @cindex call-clobbered register
1721 @cindex call-saved register
1722 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1723 clobbered (in general) by function calls as well as for fixed
1724 registers. This macro therefore identifies the registers that are not
1725 available for general allocation of values that must live across
1728 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1729 automatically saves it on function entry and restores it on function
1730 exit, if the register is used within the function.
1732 @findex CALL_REALLY_USED_REGISTERS
1733 @item CALL_REALLY_USED_REGISTERS
1734 @cindex call-used register
1735 @cindex call-clobbered register
1736 @cindex call-saved register
1737 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1738 that the entire set of @code{FIXED_REGISTERS} be included.
1739 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1740 This macro is optional. If not specified, it defaults to the value
1741 of @code{CALL_USED_REGISTERS}.
1743 @findex HARD_REGNO_CALL_PART_CLOBBERED
1744 @item HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1745 @cindex call-used register
1746 @cindex call-clobbered register
1747 @cindex call-saved register
1748 A C expression that is nonzero if it is not permissible to store a
1749 value of mode @var{mode} in hard register number @var{regno} across a
1750 call without some part of it being clobbered. For most machines this
1751 macro need not be defined. It is only required for machines that do not
1752 preserve the entire contents of a register across a call.
1754 @findex CONDITIONAL_REGISTER_USAGE
1756 @findex call_used_regs
1757 @item CONDITIONAL_REGISTER_USAGE
1758 Zero or more C statements that may conditionally modify five variables
1759 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1760 @code{reg_names}, and @code{reg_class_contents}, to take into account
1761 any dependence of these register sets on target flags. The first three
1762 of these are of type @code{char []} (interpreted as Boolean vectors).
1763 @code{global_regs} is a @code{const char *[]}, and
1764 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1765 called, @code{fixed_regs}, @code{call_used_regs},
1766 @code{reg_class_contents}, and @code{reg_names} have been initialized
1767 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1768 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1769 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1770 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1771 command options have been applied.
1773 You need not define this macro if it has no work to do.
1775 @cindex disabling certain registers
1776 @cindex controlling register usage
1777 If the usage of an entire class of registers depends on the target
1778 flags, you may indicate this to GCC by using this macro to modify
1779 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
1780 registers in the classes which should not be used by GCC@. Also define
1781 the macro @code{REG_CLASS_FROM_LETTER} to return @code{NO_REGS} if it
1782 is called with a letter for a class that shouldn't be used.
1784 (However, if this class is not included in @code{GENERAL_REGS} and all
1785 of the insn patterns whose constraints permit this class are
1786 controlled by target switches, then GCC will automatically avoid using
1787 these registers when the target switches are opposed to them.)
1789 @findex NON_SAVING_SETJMP
1790 @item NON_SAVING_SETJMP
1791 If this macro is defined and has a nonzero value, it means that
1792 @code{setjmp} and related functions fail to save the registers, or that
1793 @code{longjmp} fails to restore them. To compensate, the compiler
1794 avoids putting variables in registers in functions that use
1797 @findex INCOMING_REGNO
1798 @item INCOMING_REGNO (@var{out})
1799 Define this macro if the target machine has register windows. This C
1800 expression returns the register number as seen by the called function
1801 corresponding to the register number @var{out} as seen by the calling
1802 function. Return @var{out} if register number @var{out} is not an
1805 @findex OUTGOING_REGNO
1806 @item OUTGOING_REGNO (@var{in})
1807 Define this macro if the target machine has register windows. This C
1808 expression returns the register number as seen by the calling function
1809 corresponding to the register number @var{in} as seen by the called
1810 function. Return @var{in} if register number @var{in} is not an inbound
1814 @item LOCAL_REGNO (@var{regno})
1815 Define this macro if the target machine has register windows. This C
1816 expression returns true if the register is call-saved but is in the
1817 register window. Unlike most call-saved registers, such registers
1818 need not be explicitly restored on function exit or during non-local
1824 If the program counter has a register number, define this as that
1825 register number. Otherwise, do not define it.
1829 @node Allocation Order
1830 @subsection Order of Allocation of Registers
1831 @cindex order of register allocation
1832 @cindex register allocation order
1834 @c prevent bad page break with this line
1835 Registers are allocated in order.
1838 @findex REG_ALLOC_ORDER
1839 @item REG_ALLOC_ORDER
1840 If defined, an initializer for a vector of integers, containing the
1841 numbers of hard registers in the order in which GCC should prefer
1842 to use them (from most preferred to least).
1844 If this macro is not defined, registers are used lowest numbered first
1845 (all else being equal).
1847 One use of this macro is on machines where the highest numbered
1848 registers must always be saved and the save-multiple-registers
1849 instruction supports only sequences of consecutive registers. On such
1850 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
1851 the highest numbered allocable register first.
1853 @findex ORDER_REGS_FOR_LOCAL_ALLOC
1854 @item ORDER_REGS_FOR_LOCAL_ALLOC
1855 A C statement (sans semicolon) to choose the order in which to allocate
1856 hard registers for pseudo-registers local to a basic block.
1858 Store the desired register order in the array @code{reg_alloc_order}.
1859 Element 0 should be the register to allocate first; element 1, the next
1860 register; and so on.
1862 The macro body should not assume anything about the contents of
1863 @code{reg_alloc_order} before execution of the macro.
1865 On most machines, it is not necessary to define this macro.
1868 @node Values in Registers
1869 @subsection How Values Fit in Registers
1871 This section discusses the macros that describe which kinds of values
1872 (specifically, which machine modes) each register can hold, and how many
1873 consecutive registers are needed for a given mode.
1876 @findex HARD_REGNO_NREGS
1877 @item HARD_REGNO_NREGS (@var{regno}, @var{mode})
1878 A C expression for the number of consecutive hard registers, starting
1879 at register number @var{regno}, required to hold a value of mode
1882 On a machine where all registers are exactly one word, a suitable
1883 definition of this macro is
1886 #define HARD_REGNO_NREGS(REGNO, MODE) \
1887 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
1891 @findex HARD_REGNO_MODE_OK
1892 @item HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
1893 A C expression that is nonzero if it is permissible to store a value
1894 of mode @var{mode} in hard register number @var{regno} (or in several
1895 registers starting with that one). For a machine where all registers
1896 are equivalent, a suitable definition is
1899 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
1902 You need not include code to check for the numbers of fixed registers,
1903 because the allocation mechanism considers them to be always occupied.
1905 @cindex register pairs
1906 On some machines, double-precision values must be kept in even/odd
1907 register pairs. You can implement that by defining this macro to reject
1908 odd register numbers for such modes.
1910 The minimum requirement for a mode to be OK in a register is that the
1911 @samp{mov@var{mode}} instruction pattern support moves between the
1912 register and other hard register in the same class and that moving a
1913 value into the register and back out not alter it.
1915 Since the same instruction used to move @code{word_mode} will work for
1916 all narrower integer modes, it is not necessary on any machine for
1917 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
1918 you define patterns @samp{movhi}, etc., to take advantage of this. This
1919 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
1920 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
1923 Many machines have special registers for floating point arithmetic.
1924 Often people assume that floating point machine modes are allowed only
1925 in floating point registers. This is not true. Any registers that
1926 can hold integers can safely @emph{hold} a floating point machine
1927 mode, whether or not floating arithmetic can be done on it in those
1928 registers. Integer move instructions can be used to move the values.
1930 On some machines, though, the converse is true: fixed-point machine
1931 modes may not go in floating registers. This is true if the floating
1932 registers normalize any value stored in them, because storing a
1933 non-floating value there would garble it. In this case,
1934 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
1935 floating registers. But if the floating registers do not automatically
1936 normalize, if you can store any bit pattern in one and retrieve it
1937 unchanged without a trap, then any machine mode may go in a floating
1938 register, so you can define this macro to say so.
1940 The primary significance of special floating registers is rather that
1941 they are the registers acceptable in floating point arithmetic
1942 instructions. However, this is of no concern to
1943 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
1944 constraints for those instructions.
1946 On some machines, the floating registers are especially slow to access,
1947 so that it is better to store a value in a stack frame than in such a
1948 register if floating point arithmetic is not being done. As long as the
1949 floating registers are not in class @code{GENERAL_REGS}, they will not
1950 be used unless some pattern's constraint asks for one.
1952 @findex MODES_TIEABLE_P
1953 @item MODES_TIEABLE_P (@var{mode1}, @var{mode2})
1954 A C expression that is nonzero if a value of mode
1955 @var{mode1} is accessible in mode @var{mode2} without copying.
1957 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
1958 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
1959 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
1960 should be nonzero. If they differ for any @var{r}, you should define
1961 this macro to return zero unless some other mechanism ensures the
1962 accessibility of the value in a narrower mode.
1964 You should define this macro to return nonzero in as many cases as
1965 possible since doing so will allow GCC to perform better register
1968 @findex AVOID_CCMODE_COPIES
1969 @item AVOID_CCMODE_COPIES
1970 Define this macro if the compiler should avoid copies to/from @code{CCmode}
1971 registers. You should only define this macro if support for copying to/from
1972 @code{CCmode} is incomplete.
1974 @findex SUBREG_REGNO_OFFSET
1975 @item SUBREG_REGNO_OFFSET
1976 Define this macro if the compiler needs to handle subregs in a non-standard
1977 way. The macro returns the correct regno offset for mode @code{YMODE} given
1978 a subreg of type @code{XMODE}.
1979 This macro takes 4 parameters:
1982 A regno of an inner hard subreg_reg (or what will become one).
1988 The mode of a top level SUBREG (or what may become one).
1990 The default function can be found in @file{rtlanal.c}, function
1991 @code{subreg_regno_offset}. Normally this does not need to be defined.
1994 @node Leaf Functions
1995 @subsection Handling Leaf Functions
1997 @cindex leaf functions
1998 @cindex functions, leaf
1999 On some machines, a leaf function (i.e., one which makes no calls) can run
2000 more efficiently if it does not make its own register window. Often this
2001 means it is required to receive its arguments in the registers where they
2002 are passed by the caller, instead of the registers where they would
2005 The special treatment for leaf functions generally applies only when
2006 other conditions are met; for example, often they may use only those
2007 registers for its own variables and temporaries. We use the term ``leaf
2008 function'' to mean a function that is suitable for this special
2009 handling, so that functions with no calls are not necessarily ``leaf
2012 GCC assigns register numbers before it knows whether the function is
2013 suitable for leaf function treatment. So it needs to renumber the
2014 registers in order to output a leaf function. The following macros
2018 @findex LEAF_REGISTERS
2019 @item LEAF_REGISTERS
2020 Name of a char vector, indexed by hard register number, which
2021 contains 1 for a register that is allowable in a candidate for leaf
2024 If leaf function treatment involves renumbering the registers, then the
2025 registers marked here should be the ones before renumbering---those that
2026 GCC would ordinarily allocate. The registers which will actually be
2027 used in the assembler code, after renumbering, should not be marked with 1
2030 Define this macro only if the target machine offers a way to optimize
2031 the treatment of leaf functions.
2033 @findex LEAF_REG_REMAP
2034 @item LEAF_REG_REMAP (@var{regno})
2035 A C expression whose value is the register number to which @var{regno}
2036 should be renumbered, when a function is treated as a leaf function.
2038 If @var{regno} is a register number which should not appear in a leaf
2039 function before renumbering, then the expression should yield @minus{}1, which
2040 will cause the compiler to abort.
2042 Define this macro only if the target machine offers a way to optimize the
2043 treatment of leaf functions, and registers need to be renumbered to do
2047 @findex current_function_is_leaf
2048 @findex current_function_uses_only_leaf_regs
2049 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2050 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2051 specially. They can test the C variable @code{current_function_is_leaf}
2052 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2053 set prior to local register allocation and is valid for the remaining
2054 compiler passes. They can also test the C variable
2055 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2056 functions which only use leaf registers.
2057 @code{current_function_uses_only_leaf_regs} is valid after reload and is
2058 only useful if @code{LEAF_REGISTERS} is defined.
2059 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2060 @c of the next paragraph?! --mew 2feb93
2062 @node Stack Registers
2063 @subsection Registers That Form a Stack
2065 There are special features to handle computers where some of the
2066 ``registers'' form a stack, as in the 80387 coprocessor for the 80386.
2067 Stack registers are normally written by pushing onto the stack, and are
2068 numbered relative to the top of the stack.
2070 Currently, GCC can only handle one group of stack-like registers, and
2071 they must be consecutively numbered.
2076 Define this if the machine has any stack-like registers.
2078 @findex FIRST_STACK_REG
2079 @item FIRST_STACK_REG
2080 The number of the first stack-like register. This one is the top
2083 @findex LAST_STACK_REG
2084 @item LAST_STACK_REG
2085 The number of the last stack-like register. This one is the bottom of
2089 @node Register Classes
2090 @section Register Classes
2091 @cindex register class definitions
2092 @cindex class definitions, register
2094 On many machines, the numbered registers are not all equivalent.
2095 For example, certain registers may not be allowed for indexed addressing;
2096 certain registers may not be allowed in some instructions. These machine
2097 restrictions are described to the compiler using @dfn{register classes}.
2099 You define a number of register classes, giving each one a name and saying
2100 which of the registers belong to it. Then you can specify register classes
2101 that are allowed as operands to particular instruction patterns.
2105 In general, each register will belong to several classes. In fact, one
2106 class must be named @code{ALL_REGS} and contain all the registers. Another
2107 class must be named @code{NO_REGS} and contain no registers. Often the
2108 union of two classes will be another class; however, this is not required.
2110 @findex GENERAL_REGS
2111 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2112 terribly special about the name, but the operand constraint letters
2113 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2114 the same as @code{ALL_REGS}, just define it as a macro which expands
2117 Order the classes so that if class @var{x} is contained in class @var{y}
2118 then @var{x} has a lower class number than @var{y}.
2120 The way classes other than @code{GENERAL_REGS} are specified in operand
2121 constraints is through machine-dependent operand constraint letters.
2122 You can define such letters to correspond to various classes, then use
2123 them in operand constraints.
2125 You should define a class for the union of two classes whenever some
2126 instruction allows both classes. For example, if an instruction allows
2127 either a floating point (coprocessor) register or a general register for a
2128 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2129 which includes both of them. Otherwise you will get suboptimal code.
2131 You must also specify certain redundant information about the register
2132 classes: for each class, which classes contain it and which ones are
2133 contained in it; for each pair of classes, the largest class contained
2136 When a value occupying several consecutive registers is expected in a
2137 certain class, all the registers used must belong to that class.
2138 Therefore, register classes cannot be used to enforce a requirement for
2139 a register pair to start with an even-numbered register. The way to
2140 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2142 Register classes used for input-operands of bitwise-and or shift
2143 instructions have a special requirement: each such class must have, for
2144 each fixed-point machine mode, a subclass whose registers can transfer that
2145 mode to or from memory. For example, on some machines, the operations for
2146 single-byte values (@code{QImode}) are limited to certain registers. When
2147 this is so, each register class that is used in a bitwise-and or shift
2148 instruction must have a subclass consisting of registers from which
2149 single-byte values can be loaded or stored. This is so that
2150 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2153 @findex enum reg_class
2154 @item enum reg_class
2155 An enumeral type that must be defined with all the register class names
2156 as enumeral values. @code{NO_REGS} must be first. @code{ALL_REGS}
2157 must be the last register class, followed by one more enumeral value,
2158 @code{LIM_REG_CLASSES}, which is not a register class but rather
2159 tells how many classes there are.
2161 Each register class has a number, which is the value of casting
2162 the class name to type @code{int}. The number serves as an index
2163 in many of the tables described below.
2165 @findex N_REG_CLASSES
2167 The number of distinct register classes, defined as follows:
2170 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2173 @findex REG_CLASS_NAMES
2174 @item REG_CLASS_NAMES
2175 An initializer containing the names of the register classes as C string
2176 constants. These names are used in writing some of the debugging dumps.
2178 @findex REG_CLASS_CONTENTS
2179 @item REG_CLASS_CONTENTS
2180 An initializer containing the contents of the register classes, as integers
2181 which are bit masks. The @var{n}th integer specifies the contents of class
2182 @var{n}. The way the integer @var{mask} is interpreted is that
2183 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2185 When the machine has more than 32 registers, an integer does not suffice.
2186 Then the integers are replaced by sub-initializers, braced groupings containing
2187 several integers. Each sub-initializer must be suitable as an initializer
2188 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2189 In this situation, the first integer in each sub-initializer corresponds to
2190 registers 0 through 31, the second integer to registers 32 through 63, and
2193 @findex REGNO_REG_CLASS
2194 @item REGNO_REG_CLASS (@var{regno})
2195 A C expression whose value is a register class containing hard register
2196 @var{regno}. In general there is more than one such class; choose a class
2197 which is @dfn{minimal}, meaning that no smaller class also contains the
2200 @findex BASE_REG_CLASS
2201 @item BASE_REG_CLASS
2202 A macro whose definition is the name of the class to which a valid
2203 base register must belong. A base register is one used in an address
2204 which is the register value plus a displacement.
2206 @findex INDEX_REG_CLASS
2207 @item INDEX_REG_CLASS
2208 A macro whose definition is the name of the class to which a valid
2209 index register must belong. An index register is one used in an
2210 address where its value is either multiplied by a scale factor or
2211 added to another register (as well as added to a displacement).
2213 @findex REG_CLASS_FROM_LETTER
2214 @item REG_CLASS_FROM_LETTER (@var{char})
2215 A C expression which defines the machine-dependent operand constraint
2216 letters for register classes. If @var{char} is such a letter, the
2217 value should be the register class corresponding to it. Otherwise,
2218 the value should be @code{NO_REGS}. The register letter @samp{r},
2219 corresponding to class @code{GENERAL_REGS}, will not be passed
2220 to this macro; you do not need to handle it.
2222 @findex REGNO_OK_FOR_BASE_P
2223 @item REGNO_OK_FOR_BASE_P (@var{num})
2224 A C expression which is nonzero if register number @var{num} is
2225 suitable for use as a base register in operand addresses. It may be
2226 either a suitable hard register or a pseudo register that has been
2227 allocated such a hard register.
2229 @findex REGNO_MODE_OK_FOR_BASE_P
2230 @item REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2231 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2232 that expression may examine the mode of the memory reference in
2233 @var{mode}. You should define this macro if the mode of the memory
2234 reference affects whether a register may be used as a base register. If
2235 you define this macro, the compiler will use it instead of
2236 @code{REGNO_OK_FOR_BASE_P}.
2238 @findex REGNO_OK_FOR_INDEX_P
2239 @item REGNO_OK_FOR_INDEX_P (@var{num})
2240 A C expression which is nonzero if register number @var{num} is
2241 suitable for use as an index register in operand addresses. It may be
2242 either a suitable hard register or a pseudo register that has been
2243 allocated such a hard register.
2245 The difference between an index register and a base register is that
2246 the index register may be scaled. If an address involves the sum of
2247 two registers, neither one of them scaled, then either one may be
2248 labeled the ``base'' and the other the ``index''; but whichever
2249 labeling is used must fit the machine's constraints of which registers
2250 may serve in each capacity. The compiler will try both labelings,
2251 looking for one that is valid, and will reload one or both registers
2252 only if neither labeling works.
2254 @findex PREFERRED_RELOAD_CLASS
2255 @item PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2256 A C expression that places additional restrictions on the register class
2257 to use when it is necessary to copy value @var{x} into a register in class
2258 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2259 another, smaller class. On many machines, the following definition is
2263 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2266 Sometimes returning a more restrictive class makes better code. For
2267 example, on the 68000, when @var{x} is an integer constant that is in range
2268 for a @samp{moveq} instruction, the value of this macro is always
2269 @code{DATA_REGS} as long as @var{class} includes the data registers.
2270 Requiring a data register guarantees that a @samp{moveq} will be used.
2272 If @var{x} is a @code{const_double}, by returning @code{NO_REGS}
2273 you can force @var{x} into a memory constant. This is useful on
2274 certain machines where immediate floating values cannot be loaded into
2275 certain kinds of registers.
2277 @findex PREFERRED_OUTPUT_RELOAD_CLASS
2278 @item PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2279 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2280 input reloads. If you don't define this macro, the default is to use
2281 @var{class}, unchanged.
2283 @findex LIMIT_RELOAD_CLASS
2284 @item LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2285 A C expression that places additional restrictions on the register class
2286 to use when it is necessary to be able to hold a value of mode
2287 @var{mode} in a reload register for which class @var{class} would
2290 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2291 there are certain modes that simply can't go in certain reload classes.
2293 The value is a register class; perhaps @var{class}, or perhaps another,
2296 Don't define this macro unless the target machine has limitations which
2297 require the macro to do something nontrivial.
2299 @findex SECONDARY_RELOAD_CLASS
2300 @findex SECONDARY_INPUT_RELOAD_CLASS
2301 @findex SECONDARY_OUTPUT_RELOAD_CLASS
2302 @item SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2303 @itemx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2304 @itemx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2305 Many machines have some registers that cannot be copied directly to or
2306 from memory or even from other types of registers. An example is the
2307 @samp{MQ} register, which on most machines, can only be copied to or
2308 from general registers, but not memory. Some machines allow copying all
2309 registers to and from memory, but require a scratch register for stores
2310 to some memory locations (e.g., those with symbolic address on the RT,
2311 and those with certain symbolic address on the Sparc when compiling
2312 PIC)@. In some cases, both an intermediate and a scratch register are
2315 You should define these macros to indicate to the reload phase that it may
2316 need to allocate at least one register for a reload in addition to the
2317 register to contain the data. Specifically, if copying @var{x} to a
2318 register @var{class} in @var{mode} requires an intermediate register,
2319 you should define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2320 largest register class all of whose registers can be used as
2321 intermediate registers or scratch registers.
2323 If copying a register @var{class} in @var{mode} to @var{x} requires an
2324 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2325 should be defined to return the largest register class required. If the
2326 requirements for input and output reloads are the same, the macro
2327 @code{SECONDARY_RELOAD_CLASS} should be used instead of defining both
2330 The values returned by these macros are often @code{GENERAL_REGS}.
2331 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2332 can be directly copied to or from a register of @var{class} in
2333 @var{mode} without requiring a scratch register. Do not define this
2334 macro if it would always return @code{NO_REGS}.
2336 If a scratch register is required (either with or without an
2337 intermediate register), you should define patterns for
2338 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2339 (@pxref{Standard Names}. These patterns, which will normally be
2340 implemented with a @code{define_expand}, should be similar to the
2341 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2344 Define constraints for the reload register and scratch register that
2345 contain a single register class. If the original reload register (whose
2346 class is @var{class}) can meet the constraint given in the pattern, the
2347 value returned by these macros is used for the class of the scratch
2348 register. Otherwise, two additional reload registers are required.
2349 Their classes are obtained from the constraints in the insn pattern.
2351 @var{x} might be a pseudo-register or a @code{subreg} of a
2352 pseudo-register, which could either be in a hard register or in memory.
2353 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2354 in memory and the hard register number if it is in a register.
2356 These macros should not be used in the case where a particular class of
2357 registers can only be copied to memory and not to another class of
2358 registers. In that case, secondary reload registers are not needed and
2359 would not be helpful. Instead, a stack location must be used to perform
2360 the copy and the @code{mov@var{m}} pattern should use memory as a
2361 intermediate storage. This case often occurs between floating-point and
2364 @findex SECONDARY_MEMORY_NEEDED
2365 @item SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2366 Certain machines have the property that some registers cannot be copied
2367 to some other registers without using memory. Define this macro on
2368 those machines to be a C expression that is nonzero if objects of mode
2369 @var{m} in registers of @var{class1} can only be copied to registers of
2370 class @var{class2} by storing a register of @var{class1} into memory
2371 and loading that memory location into a register of @var{class2}.
2373 Do not define this macro if its value would always be zero.
2375 @findex SECONDARY_MEMORY_NEEDED_RTX
2376 @item SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2377 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2378 allocates a stack slot for a memory location needed for register copies.
2379 If this macro is defined, the compiler instead uses the memory location
2380 defined by this macro.
2382 Do not define this macro if you do not define
2383 @code{SECONDARY_MEMORY_NEEDED}.
2385 @findex SECONDARY_MEMORY_NEEDED_MODE
2386 @item SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2387 When the compiler needs a secondary memory location to copy between two
2388 registers of mode @var{mode}, it normally allocates sufficient memory to
2389 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2390 load operations in a mode that many bits wide and whose class is the
2391 same as that of @var{mode}.
2393 This is right thing to do on most machines because it ensures that all
2394 bits of the register are copied and prevents accesses to the registers
2395 in a narrower mode, which some machines prohibit for floating-point
2398 However, this default behavior is not correct on some machines, such as
2399 the DEC Alpha, that store short integers in floating-point registers
2400 differently than in integer registers. On those machines, the default
2401 widening will not work correctly and you must define this macro to
2402 suppress that widening in some cases. See the file @file{alpha.h} for
2405 Do not define this macro if you do not define
2406 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2407 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2409 @findex SMALL_REGISTER_CLASSES
2410 @item SMALL_REGISTER_CLASSES
2411 On some machines, it is risky to let hard registers live across arbitrary
2412 insns. Typically, these machines have instructions that require values
2413 to be in specific registers (like an accumulator), and reload will fail
2414 if the required hard register is used for another purpose across such an
2417 Define @code{SMALL_REGISTER_CLASSES} to be an expression with a nonzero
2418 value on these machines. When this macro has a nonzero value, the
2419 compiler will try to minimize the lifetime of hard registers.
2421 It is always safe to define this macro with a nonzero value, but if you
2422 unnecessarily define it, you will reduce the amount of optimizations
2423 that can be performed in some cases. If you do not define this macro
2424 with a nonzero value when it is required, the compiler will run out of
2425 spill registers and print a fatal error message. For most machines, you
2426 should not define this macro at all.
2428 @findex CLASS_LIKELY_SPILLED_P
2429 @item CLASS_LIKELY_SPILLED_P (@var{class})
2430 A C expression whose value is nonzero if pseudos that have been assigned
2431 to registers of class @var{class} would likely be spilled because
2432 registers of @var{class} are needed for spill registers.
2434 The default value of this macro returns 1 if @var{class} has exactly one
2435 register and zero otherwise. On most machines, this default should be
2436 used. Only define this macro to some other expression if pseudos
2437 allocated by @file{local-alloc.c} end up in memory because their hard
2438 registers were needed for spill registers. If this macro returns nonzero
2439 for those classes, those pseudos will only be allocated by
2440 @file{global.c}, which knows how to reallocate the pseudo to another
2441 register. If there would not be another register available for
2442 reallocation, you should not change the definition of this macro since
2443 the only effect of such a definition would be to slow down register
2446 @findex CLASS_MAX_NREGS
2447 @item CLASS_MAX_NREGS (@var{class}, @var{mode})
2448 A C expression for the maximum number of consecutive registers
2449 of class @var{class} needed to hold a value of mode @var{mode}.
2451 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2452 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2453 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2454 @var{mode})} for all @var{regno} values in the class @var{class}.
2456 This macro helps control the handling of multiple-word values
2459 @item CLASS_CANNOT_CHANGE_MODE
2460 If defined, a C expression for a class that contains registers for
2461 which the compiler may not change modes arbitrarily.
2463 @item CLASS_CANNOT_CHANGE_MODE_P(@var{from}, @var{to})
2464 A C expression that is true if, for a register in
2465 @code{CLASS_CANNOT_CHANGE_MODE}, the requested mode punning is invalid.
2467 For the example, loading 32-bit integer or floating-point objects into
2468 floating-point registers on the Alpha extends them to 64-bits.
2469 Therefore loading a 64-bit object and then storing it as a 32-bit object
2470 does not store the low-order 32-bits, as would be the case for a normal
2471 register. Therefore, @file{alpha.h} defines @code{CLASS_CANNOT_CHANGE_MODE}
2472 as @code{FLOAT_REGS} and @code{CLASS_CANNOT_CHANGE_MODE_P} restricts
2473 mode changes to same-size modes.
2475 Compare this to IA-64, which extends floating-point values to 82-bits,
2476 and stores 64-bit integers in a different format than 64-bit doubles.
2477 Therefore @code{CLASS_CANNOT_CHANGE_MODE_P} is always true.
2480 Three other special macros describe which operands fit which constraint
2484 @findex CONST_OK_FOR_LETTER_P
2485 @item CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2486 A C expression that defines the machine-dependent operand constraint
2487 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2488 particular ranges of integer values. If @var{c} is one of those
2489 letters, the expression should check that @var{value}, an integer, is in
2490 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2491 not one of those letters, the value should be 0 regardless of
2494 @findex CONST_DOUBLE_OK_FOR_LETTER_P
2495 @item CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2496 A C expression that defines the machine-dependent operand constraint
2497 letters that specify particular ranges of @code{const_double} values
2498 (@samp{G} or @samp{H}).
2500 If @var{c} is one of those letters, the expression should check that
2501 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2502 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2503 letters, the value should be 0 regardless of @var{value}.
2505 @code{const_double} is used for all floating-point constants and for
2506 @code{DImode} fixed-point constants. A given letter can accept either
2507 or both kinds of values. It can use @code{GET_MODE} to distinguish
2508 between these kinds.
2510 @findex EXTRA_CONSTRAINT
2511 @item EXTRA_CONSTRAINT (@var{value}, @var{c})
2512 A C expression that defines the optional machine-dependent constraint
2513 letters that can be used to segregate specific types of operands, usually
2514 memory references, for the target machine. Any letter that is not
2515 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER}
2516 may be used. Normally this macro will not be defined.
2518 If it is required for a particular target machine, it should return 1
2519 if @var{value} corresponds to the operand type represented by the
2520 constraint letter @var{c}. If @var{c} is not defined as an extra
2521 constraint, the value returned should be 0 regardless of @var{value}.
2523 For example, on the ROMP, load instructions cannot have their output
2524 in r0 if the memory reference contains a symbolic address. Constraint
2525 letter @samp{Q} is defined as representing a memory address that does
2526 @emph{not} contain a symbolic address. An alternative is specified with
2527 a @samp{Q} constraint on the input and @samp{r} on the output. The next
2528 alternative specifies @samp{m} on the input and a register class that
2529 does not include r0 on the output.
2532 @node Stack and Calling
2533 @section Stack Layout and Calling Conventions
2534 @cindex calling conventions
2536 @c prevent bad page break with this line
2537 This describes the stack layout and calling conventions.
2541 * Exception Handling::
2546 * Register Arguments::
2548 * Aggregate Return::
2556 @subsection Basic Stack Layout
2557 @cindex stack frame layout
2558 @cindex frame layout
2560 @c prevent bad page break with this line
2561 Here is the basic stack layout.
2564 @findex STACK_GROWS_DOWNWARD
2565 @item STACK_GROWS_DOWNWARD
2566 Define this macro if pushing a word onto the stack moves the stack
2567 pointer to a smaller address.
2569 When we say, ``define this macro if @dots{},'' it means that the
2570 compiler checks this macro only with @code{#ifdef} so the precise
2571 definition used does not matter.
2573 @findex STACK_PUSH_CODE
2574 @item STACK_PUSH_CODE
2576 This macro defines the operation used when something is pushed
2577 on the stack. In RTL, a push operation will be
2578 @code{(set (mem (STACK_PUSH_CODE (reg sp))) ...)}
2580 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
2581 and @code{POST_INC}. Which of these is correct depends on
2582 the stack direction and on whether the stack pointer points
2583 to the last item on the stack or whether it points to the
2584 space for the next item on the stack.
2586 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
2587 defined, which is almost always right, and @code{PRE_INC} otherwise,
2588 which is often wrong.
2590 @findex FRAME_GROWS_DOWNWARD
2591 @item FRAME_GROWS_DOWNWARD
2592 Define this macro if the addresses of local variable slots are at negative
2593 offsets from the frame pointer.
2595 @findex ARGS_GROW_DOWNWARD
2596 @item ARGS_GROW_DOWNWARD
2597 Define this macro if successive arguments to a function occupy decreasing
2598 addresses on the stack.
2600 @findex STARTING_FRAME_OFFSET
2601 @item STARTING_FRAME_OFFSET
2602 Offset from the frame pointer to the first local variable slot to be allocated.
2604 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
2605 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
2606 Otherwise, it is found by adding the length of the first slot to the
2607 value @code{STARTING_FRAME_OFFSET}.
2608 @c i'm not sure if the above is still correct.. had to change it to get
2609 @c rid of an overfull. --mew 2feb93
2611 @findex STACK_POINTER_OFFSET
2612 @item STACK_POINTER_OFFSET
2613 Offset from the stack pointer register to the first location at which
2614 outgoing arguments are placed. If not specified, the default value of
2615 zero is used. This is the proper value for most machines.
2617 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2618 the first location at which outgoing arguments are placed.
2620 @findex FIRST_PARM_OFFSET
2621 @item FIRST_PARM_OFFSET (@var{fundecl})
2622 Offset from the argument pointer register to the first argument's
2623 address. On some machines it may depend on the data type of the
2626 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2627 the first argument's address.
2629 @findex STACK_DYNAMIC_OFFSET
2630 @item STACK_DYNAMIC_OFFSET (@var{fundecl})
2631 Offset from the stack pointer register to an item dynamically allocated
2632 on the stack, e.g., by @code{alloca}.
2634 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
2635 length of the outgoing arguments. The default is correct for most
2636 machines. See @file{function.c} for details.
2638 @findex DYNAMIC_CHAIN_ADDRESS
2639 @item DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
2640 A C expression whose value is RTL representing the address in a stack
2641 frame where the pointer to the caller's frame is stored. Assume that
2642 @var{frameaddr} is an RTL expression for the address of the stack frame
2645 If you don't define this macro, the default is to return the value
2646 of @var{frameaddr}---that is, the stack frame address is also the
2647 address of the stack word that points to the previous frame.
2649 @findex SETUP_FRAME_ADDRESSES
2650 @item SETUP_FRAME_ADDRESSES
2651 If defined, a C expression that produces the machine-specific code to
2652 setup the stack so that arbitrary frames can be accessed. For example,
2653 on the Sparc, we must flush all of the register windows to the stack
2654 before we can access arbitrary stack frames. You will seldom need to
2657 @findex BUILTIN_SETJMP_FRAME_VALUE
2658 @item BUILTIN_SETJMP_FRAME_VALUE
2659 If defined, a C expression that contains an rtx that is used to store
2660 the address of the current frame into the built in @code{setjmp} buffer.
2661 The default value, @code{virtual_stack_vars_rtx}, is correct for most
2662 machines. One reason you may need to define this macro is if
2663 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
2665 @findex RETURN_ADDR_RTX
2666 @item RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
2667 A C expression whose value is RTL representing the value of the return
2668 address for the frame @var{count} steps up from the current frame, after
2669 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
2670 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
2671 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
2673 The value of the expression must always be the correct address when
2674 @var{count} is zero, but may be @code{NULL_RTX} if there is not way to
2675 determine the return address of other frames.
2677 @findex RETURN_ADDR_IN_PREVIOUS_FRAME
2678 @item RETURN_ADDR_IN_PREVIOUS_FRAME
2679 Define this if the return address of a particular stack frame is accessed
2680 from the frame pointer of the previous stack frame.
2682 @findex INCOMING_RETURN_ADDR_RTX
2683 @item INCOMING_RETURN_ADDR_RTX
2684 A C expression whose value is RTL representing the location of the
2685 incoming return address at the beginning of any function, before the
2686 prologue. This RTL is either a @code{REG}, indicating that the return
2687 value is saved in @samp{REG}, or a @code{MEM} representing a location in
2690 You only need to define this macro if you want to support call frame
2691 debugging information like that provided by DWARF 2.
2693 If this RTL is a @code{REG}, you should also define
2694 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
2696 @findex INCOMING_FRAME_SP_OFFSET
2697 @item INCOMING_FRAME_SP_OFFSET
2698 A C expression whose value is an integer giving the offset, in bytes,
2699 from the value of the stack pointer register to the top of the stack
2700 frame at the beginning of any function, before the prologue. The top of
2701 the frame is defined to be the value of the stack pointer in the
2702 previous frame, just before the call instruction.
2704 You only need to define this macro if you want to support call frame
2705 debugging information like that provided by DWARF 2.
2707 @findex ARG_POINTER_CFA_OFFSET
2708 @item ARG_POINTER_CFA_OFFSET (@var{fundecl})
2709 A C expression whose value is an integer giving the offset, in bytes,
2710 from the argument pointer to the canonical frame address (cfa). The
2711 final value should coincide with that calculated by
2712 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
2713 during virtual register instantiation.
2715 The default value for this macro is @code{FIRST_PARM_OFFSET (fundecl)},
2716 which is correct for most machines; in general, the arguments are found
2717 immediately before the stack frame. Note that this is not the case on
2718 some targets that save registers into the caller's frame, such as SPARC
2719 and rs6000, and so such targets need to define this macro.
2721 You only need to define this macro if the default is incorrect, and you
2722 want to support call frame debugging information like that provided by
2727 Define this macro if the stack size for the target is very small. This
2728 has the effect of disabling gcc's built-in @samp{alloca}, though
2729 @samp{__builtin_alloca} is not affected.
2732 @node Exception Handling
2733 @subsection Exception Handling Support
2734 @cindex exception handling
2737 @findex EH_RETURN_DATA_REGNO
2738 @item EH_RETURN_DATA_REGNO (@var{N})
2739 A C expression whose value is the @var{N}th register number used for
2740 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
2741 @var{N} registers are usable.
2743 The exception handling library routines communicate with the exception
2744 handlers via a set of agreed upon registers. Ideally these registers
2745 should be call-clobbered; it is possible to use call-saved registers,
2746 but may negatively impact code size. The target must support at least
2747 2 data registers, but should define 4 if there are enough free registers.
2749 You must define this macro if you want to support call frame exception
2750 handling like that provided by DWARF 2.
2752 @findex EH_RETURN_STACKADJ_RTX
2753 @item EH_RETURN_STACKADJ_RTX
2754 A C expression whose value is RTL representing a location in which
2755 to store a stack adjustment to be applied before function return.
2756 This is used to unwind the stack to an exception handler's call frame.
2757 It will be assigned zero on code paths that return normally.
2759 Typically this is a call-clobbered hard register that is otherwise
2760 untouched by the epilogue, but could also be a stack slot.
2762 You must define this macro if you want to support call frame exception
2763 handling like that provided by DWARF 2.
2765 @findex EH_RETURN_HANDLER_RTX
2766 @item EH_RETURN_HANDLER_RTX
2767 A C expression whose value is RTL representing a location in which
2768 to store the address of an exception handler to which we should
2769 return. It will not be assigned on code paths that return normally.
2771 Typically this is the location in the call frame at which the normal
2772 return address is stored. For targets that return by popping an
2773 address off the stack, this might be a memory address just below
2774 the @emph{target} call frame rather than inside the current call
2775 frame. @code{EH_RETURN_STACKADJ_RTX} will have already been assigned,
2776 so it may be used to calculate the location of the target call frame.
2778 Some targets have more complex requirements than storing to an
2779 address calculable during initial code generation. In that case
2780 the @code{eh_return} instruction pattern should be used instead.
2782 If you want to support call frame exception handling, you must
2783 define either this macro or the @code{eh_return} instruction pattern.
2785 @findex ASM_PREFERRED_EH_DATA_FORMAT
2786 @item ASM_PREFERRED_EH_DATA_FORMAT(@var{code}, @var{global})
2787 This macro chooses the encoding of pointers embedded in the exception
2788 handling sections. If at all possible, this should be defined such
2789 that the exception handling section will not require dynamic relocations,
2790 and so may be read-only.
2792 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
2793 @var{global} is true if the symbol may be affected by dynamic relocations.
2794 The macro should return a combination of the @code{DW_EH_PE_*} defines
2795 as found in @file{dwarf2.h}.
2797 If this macro is not defined, pointers will not be encoded but
2798 represented directly.
2800 @findex ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX
2801 @item ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX(@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
2802 This macro allows the target to emit whatever special magic is required
2803 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
2804 Generic code takes care of pc-relative and indirect encodings; this must
2805 be defined if the target uses text-relative or data-relative encodings.
2807 This is a C statement that branches to @var{done} if the format was
2808 handled. @var{encoding} is the format chosen, @var{size} is the number
2809 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
2812 @findex MD_FALLBACK_FRAME_STATE_FOR
2813 @item MD_FALLBACK_FRAME_STATE_FOR(@var{context}, @var{fs}, @var{success})
2814 This macro allows the target to add cpu and operating system specific
2815 code to the call-frame unwinder for use when there is no unwind data
2816 available. The most common reason to implement this macro is to unwind
2817 through signal frames.
2819 This macro is called from @code{uw_frame_state_for} in @file{unwind-dw2.c}
2820 and @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
2821 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
2822 for the address of the code being executed and @code{context->cfa} for
2823 the stack pointer value. If the frame can be decoded, the register save
2824 addresses should be updated in @var{fs} and the macro should branch to
2825 @var{success}. If the frame cannot be decoded, the macro should do
2829 @node Stack Checking
2830 @subsection Specifying How Stack Checking is Done
2832 GCC will check that stack references are within the boundaries of
2833 the stack, if the @option{-fstack-check} is specified, in one of three ways:
2837 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
2838 will assume that you have arranged for stack checking to be done at
2839 appropriate places in the configuration files, e.g., in
2840 @code{TARGET_ASM_FUNCTION_PROLOGUE}. GCC will do not other special
2844 If @code{STACK_CHECK_BUILTIN} is zero and you defined a named pattern
2845 called @code{check_stack} in your @file{md} file, GCC will call that
2846 pattern with one argument which is the address to compare the stack
2847 value against. You must arrange for this pattern to report an error if
2848 the stack pointer is out of range.
2851 If neither of the above are true, GCC will generate code to periodically
2852 ``probe'' the stack pointer using the values of the macros defined below.
2855 Normally, you will use the default values of these macros, so GCC
2856 will use the third approach.
2859 @findex STACK_CHECK_BUILTIN
2860 @item STACK_CHECK_BUILTIN
2861 A nonzero value if stack checking is done by the configuration files in a
2862 machine-dependent manner. You should define this macro if stack checking
2863 is require by the ABI of your machine or if you would like to have to stack
2864 checking in some more efficient way than GCC's portable approach.
2865 The default value of this macro is zero.
2867 @findex STACK_CHECK_PROBE_INTERVAL
2868 @item STACK_CHECK_PROBE_INTERVAL
2869 An integer representing the interval at which GCC must generate stack
2870 probe instructions. You will normally define this macro to be no larger
2871 than the size of the ``guard pages'' at the end of a stack area. The
2872 default value of 4096 is suitable for most systems.
2874 @findex STACK_CHECK_PROBE_LOAD
2875 @item STACK_CHECK_PROBE_LOAD
2876 A integer which is nonzero if GCC should perform the stack probe
2877 as a load instruction and zero if GCC should use a store instruction.
2878 The default is zero, which is the most efficient choice on most systems.
2880 @findex STACK_CHECK_PROTECT
2881 @item STACK_CHECK_PROTECT
2882 The number of bytes of stack needed to recover from a stack overflow,
2883 for languages where such a recovery is supported. The default value of
2884 75 words should be adequate for most machines.
2886 @findex STACK_CHECK_MAX_FRAME_SIZE
2887 @item STACK_CHECK_MAX_FRAME_SIZE
2888 The maximum size of a stack frame, in bytes. GCC will generate probe
2889 instructions in non-leaf functions to ensure at least this many bytes of
2890 stack are available. If a stack frame is larger than this size, stack
2891 checking will not be reliable and GCC will issue a warning. The
2892 default is chosen so that GCC only generates one instruction on most
2893 systems. You should normally not change the default value of this macro.
2895 @findex STACK_CHECK_FIXED_FRAME_SIZE
2896 @item STACK_CHECK_FIXED_FRAME_SIZE
2897 GCC uses this value to generate the above warning message. It
2898 represents the amount of fixed frame used by a function, not including
2899 space for any callee-saved registers, temporaries and user variables.
2900 You need only specify an upper bound for this amount and will normally
2901 use the default of four words.
2903 @findex STACK_CHECK_MAX_VAR_SIZE
2904 @item STACK_CHECK_MAX_VAR_SIZE
2905 The maximum size, in bytes, of an object that GCC will place in the
2906 fixed area of the stack frame when the user specifies
2907 @option{-fstack-check}.
2908 GCC computed the default from the values of the above macros and you will
2909 normally not need to override that default.
2913 @node Frame Registers
2914 @subsection Registers That Address the Stack Frame
2916 @c prevent bad page break with this line
2917 This discusses registers that address the stack frame.
2920 @findex STACK_POINTER_REGNUM
2921 @item STACK_POINTER_REGNUM
2922 The register number of the stack pointer register, which must also be a
2923 fixed register according to @code{FIXED_REGISTERS}. On most machines,
2924 the hardware determines which register this is.
2926 @findex FRAME_POINTER_REGNUM
2927 @item FRAME_POINTER_REGNUM
2928 The register number of the frame pointer register, which is used to
2929 access automatic variables in the stack frame. On some machines, the
2930 hardware determines which register this is. On other machines, you can
2931 choose any register you wish for this purpose.
2933 @findex HARD_FRAME_POINTER_REGNUM
2934 @item HARD_FRAME_POINTER_REGNUM
2935 On some machines the offset between the frame pointer and starting
2936 offset of the automatic variables is not known until after register
2937 allocation has been done (for example, because the saved registers are
2938 between these two locations). On those machines, define
2939 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
2940 be used internally until the offset is known, and define
2941 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
2942 used for the frame pointer.
2944 You should define this macro only in the very rare circumstances when it
2945 is not possible to calculate the offset between the frame pointer and
2946 the automatic variables until after register allocation has been
2947 completed. When this macro is defined, you must also indicate in your
2948 definition of @code{ELIMINABLE_REGS} how to eliminate
2949 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
2950 or @code{STACK_POINTER_REGNUM}.
2952 Do not define this macro if it would be the same as
2953 @code{FRAME_POINTER_REGNUM}.
2955 @findex ARG_POINTER_REGNUM
2956 @item ARG_POINTER_REGNUM
2957 The register number of the arg pointer register, which is used to access
2958 the function's argument list. On some machines, this is the same as the
2959 frame pointer register. On some machines, the hardware determines which
2960 register this is. On other machines, you can choose any register you
2961 wish for this purpose. If this is not the same register as the frame
2962 pointer register, then you must mark it as a fixed register according to
2963 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
2964 (@pxref{Elimination}).
2966 @findex RETURN_ADDRESS_POINTER_REGNUM
2967 @item RETURN_ADDRESS_POINTER_REGNUM
2968 The register number of the return address pointer register, which is used to
2969 access the current function's return address from the stack. On some
2970 machines, the return address is not at a fixed offset from the frame
2971 pointer or stack pointer or argument pointer. This register can be defined
2972 to point to the return address on the stack, and then be converted by
2973 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
2975 Do not define this macro unless there is no other way to get the return
2976 address from the stack.
2978 @findex STATIC_CHAIN_REGNUM
2979 @findex STATIC_CHAIN_INCOMING_REGNUM
2980 @item STATIC_CHAIN_REGNUM
2981 @itemx STATIC_CHAIN_INCOMING_REGNUM
2982 Register numbers used for passing a function's static chain pointer. If
2983 register windows are used, the register number as seen by the called
2984 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
2985 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
2986 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
2989 The static chain register need not be a fixed register.
2991 If the static chain is passed in memory, these macros should not be
2992 defined; instead, the next two macros should be defined.
2994 @findex STATIC_CHAIN
2995 @findex STATIC_CHAIN_INCOMING
2997 @itemx STATIC_CHAIN_INCOMING
2998 If the static chain is passed in memory, these macros provide rtx giving
2999 @code{mem} expressions that denote where they are stored.
3000 @code{STATIC_CHAIN} and @code{STATIC_CHAIN_INCOMING} give the locations
3001 as seen by the calling and called functions, respectively. Often the former
3002 will be at an offset from the stack pointer and the latter at an offset from
3005 @findex stack_pointer_rtx
3006 @findex frame_pointer_rtx
3007 @findex arg_pointer_rtx
3008 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3009 @code{arg_pointer_rtx} will have been initialized prior to the use of these
3010 macros and should be used to refer to those items.
3012 If the static chain is passed in a register, the two previous macros should
3015 @findex DWARF_FRAME_REGISTERS
3016 @item DWARF_FRAME_REGISTERS
3017 This macro specifies the maximum number of hard registers that can be
3018 saved in a call frame. This is used to size data structures used in
3019 DWARF2 exception handling.
3021 Prior to GCC 3.0, this macro was needed in order to establish a stable
3022 exception handling ABI in the face of adding new hard registers for ISA
3023 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3024 in the number of hard registers. Nevertheless, this macro can still be
3025 used to reduce the runtime memory requirements of the exception handling
3026 routines, which can be substantial if the ISA contains a lot of
3027 registers that are not call-saved.
3029 If this macro is not defined, it defaults to
3030 @code{FIRST_PSEUDO_REGISTER}.
3032 @findex PRE_GCC3_DWARF_FRAME_REGISTERS
3033 @item PRE_GCC3_DWARF_FRAME_REGISTERS
3035 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3036 for backward compatibility in pre GCC 3.0 compiled code.
3038 If this macro is not defined, it defaults to
3039 @code{DWARF_FRAME_REGISTERS}.
3044 @subsection Eliminating Frame Pointer and Arg Pointer
3046 @c prevent bad page break with this line
3047 This is about eliminating the frame pointer and arg pointer.
3050 @findex FRAME_POINTER_REQUIRED
3051 @item FRAME_POINTER_REQUIRED
3052 A C expression which is nonzero if a function must have and use a frame
3053 pointer. This expression is evaluated in the reload pass. If its value is
3054 nonzero the function will have a frame pointer.
3056 The expression can in principle examine the current function and decide
3057 according to the facts, but on most machines the constant 0 or the
3058 constant 1 suffices. Use 0 when the machine allows code to be generated
3059 with no frame pointer, and doing so saves some time or space. Use 1
3060 when there is no possible advantage to avoiding a frame pointer.
3062 In certain cases, the compiler does not know how to produce valid code
3063 without a frame pointer. The compiler recognizes those cases and
3064 automatically gives the function a frame pointer regardless of what
3065 @code{FRAME_POINTER_REQUIRED} says. You don't need to worry about
3068 In a function that does not require a frame pointer, the frame pointer
3069 register can be allocated for ordinary usage, unless you mark it as a
3070 fixed register. See @code{FIXED_REGISTERS} for more information.
3072 @findex INITIAL_FRAME_POINTER_OFFSET
3073 @findex get_frame_size
3074 @item INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3075 A C statement to store in the variable @var{depth-var} the difference
3076 between the frame pointer and the stack pointer values immediately after
3077 the function prologue. The value would be computed from information
3078 such as the result of @code{get_frame_size ()} and the tables of
3079 registers @code{regs_ever_live} and @code{call_used_regs}.
3081 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3082 need not be defined. Otherwise, it must be defined even if
3083 @code{FRAME_POINTER_REQUIRED} is defined to always be true; in that
3084 case, you may set @var{depth-var} to anything.
3086 @findex ELIMINABLE_REGS
3087 @item ELIMINABLE_REGS
3088 If defined, this macro specifies a table of register pairs used to
3089 eliminate unneeded registers that point into the stack frame. If it is not
3090 defined, the only elimination attempted by the compiler is to replace
3091 references to the frame pointer with references to the stack pointer.
3093 The definition of this macro is a list of structure initializations, each
3094 of which specifies an original and replacement register.
3096 On some machines, the position of the argument pointer is not known until
3097 the compilation is completed. In such a case, a separate hard register
3098 must be used for the argument pointer. This register can be eliminated by
3099 replacing it with either the frame pointer or the argument pointer,
3100 depending on whether or not the frame pointer has been eliminated.
3102 In this case, you might specify:
3104 #define ELIMINABLE_REGS \
3105 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3106 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3107 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3110 Note that the elimination of the argument pointer with the stack pointer is
3111 specified first since that is the preferred elimination.
3113 @findex CAN_ELIMINATE
3114 @item CAN_ELIMINATE (@var{from-reg}, @var{to-reg})
3115 A C expression that returns nonzero if the compiler is allowed to try
3116 to replace register number @var{from-reg} with register number
3117 @var{to-reg}. This macro need only be defined if @code{ELIMINABLE_REGS}
3118 is defined, and will usually be the constant 1, since most of the cases
3119 preventing register elimination are things that the compiler already
3122 @findex INITIAL_ELIMINATION_OFFSET
3123 @item INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3124 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3125 specifies the initial difference between the specified pair of
3126 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3129 @findex LONGJMP_RESTORE_FROM_STACK
3130 @item LONGJMP_RESTORE_FROM_STACK
3131 Define this macro if the @code{longjmp} function restores registers from
3132 the stack frames, rather than from those saved specifically by
3133 @code{setjmp}. Certain quantities must not be kept in registers across
3134 a call to @code{setjmp} on such machines.
3137 @node Stack Arguments
3138 @subsection Passing Function Arguments on the Stack
3139 @cindex arguments on stack
3140 @cindex stack arguments
3142 The macros in this section control how arguments are passed
3143 on the stack. See the following section for other macros that
3144 control passing certain arguments in registers.
3147 @findex PROMOTE_PROTOTYPES
3148 @item PROMOTE_PROTOTYPES
3149 A C expression whose value is nonzero if an argument declared in
3150 a prototype as an integral type smaller than @code{int} should
3151 actually be passed as an @code{int}. In addition to avoiding
3152 errors in certain cases of mismatch, it also makes for better
3153 code on certain machines. If the macro is not defined in target
3154 header files, it defaults to 0.
3158 A C expression. If nonzero, push insns will be used to pass
3160 If the target machine does not have a push instruction, set it to zero.
3161 That directs GCC to use an alternate strategy: to
3162 allocate the entire argument block and then store the arguments into
3163 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3164 On some machines, the definition
3166 @findex PUSH_ROUNDING
3167 @item PUSH_ROUNDING (@var{npushed})
3168 A C expression that is the number of bytes actually pushed onto the
3169 stack when an instruction attempts to push @var{npushed} bytes.
3171 On some machines, the definition
3174 #define PUSH_ROUNDING(BYTES) (BYTES)
3178 will suffice. But on other machines, instructions that appear
3179 to push one byte actually push two bytes in an attempt to maintain
3180 alignment. Then the definition should be
3183 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3186 @findex ACCUMULATE_OUTGOING_ARGS
3187 @findex current_function_outgoing_args_size
3188 @item ACCUMULATE_OUTGOING_ARGS
3189 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3190 will be computed and placed into the variable
3191 @code{current_function_outgoing_args_size}. No space will be pushed
3192 onto the stack for each call; instead, the function prologue should
3193 increase the stack frame size by this amount.
3195 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3198 @findex REG_PARM_STACK_SPACE
3199 @item REG_PARM_STACK_SPACE (@var{fndecl})
3200 Define this macro if functions should assume that stack space has been
3201 allocated for arguments even when their values are passed in
3204 The value of this macro is the size, in bytes, of the area reserved for
3205 arguments passed in registers for the function represented by @var{fndecl},
3206 which can be zero if GCC is calling a library function.
3208 This space can be allocated by the caller, or be a part of the
3209 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3211 @c above is overfull. not sure what to do. --mew 5feb93 did
3212 @c something, not sure if it looks good. --mew 10feb93
3214 @findex MAYBE_REG_PARM_STACK_SPACE
3215 @findex FINAL_REG_PARM_STACK_SPACE
3216 @item MAYBE_REG_PARM_STACK_SPACE
3217 @itemx FINAL_REG_PARM_STACK_SPACE (@var{const_size}, @var{var_size})
3218 Define these macros in addition to the one above if functions might
3219 allocate stack space for arguments even when their values are passed
3220 in registers. These should be used when the stack space allocated
3221 for arguments in registers is not a simple constant independent of the
3222 function declaration.
3224 The value of the first macro is the size, in bytes, of the area that
3225 we should initially assume would be reserved for arguments passed in registers.
3227 The value of the second macro is the actual size, in bytes, of the area
3228 that will be reserved for arguments passed in registers. This takes two
3229 arguments: an integer representing the number of bytes of fixed sized
3230 arguments on the stack, and a tree representing the number of bytes of
3231 variable sized arguments on the stack.
3233 When these macros are defined, @code{REG_PARM_STACK_SPACE} will only be
3234 called for libcall functions, the current function, or for a function
3235 being called when it is known that such stack space must be allocated.
3236 In each case this value can be easily computed.
3238 When deciding whether a called function needs such stack space, and how
3239 much space to reserve, GCC uses these two macros instead of
3240 @code{REG_PARM_STACK_SPACE}.
3242 @findex OUTGOING_REG_PARM_STACK_SPACE
3243 @item OUTGOING_REG_PARM_STACK_SPACE
3244 Define this if it is the responsibility of the caller to allocate the area
3245 reserved for arguments passed in registers.
3247 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3248 whether the space for these arguments counts in the value of
3249 @code{current_function_outgoing_args_size}.
3251 @findex STACK_PARMS_IN_REG_PARM_AREA
3252 @item STACK_PARMS_IN_REG_PARM_AREA
3253 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3254 stack parameters don't skip the area specified by it.
3255 @c i changed this, makes more sens and it should have taken care of the
3256 @c overfull.. not as specific, tho. --mew 5feb93
3258 Normally, when a parameter is not passed in registers, it is placed on the
3259 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3260 suppresses this behavior and causes the parameter to be passed on the
3261 stack in its natural location.
3263 @findex RETURN_POPS_ARGS
3264 @item RETURN_POPS_ARGS (@var{fundecl}, @var{funtype}, @var{stack-size})
3265 A C expression that should indicate the number of bytes of its own
3266 arguments that a function pops on returning, or 0 if the
3267 function pops no arguments and the caller must therefore pop them all
3268 after the function returns.
3270 @var{fundecl} is a C variable whose value is a tree node that describes
3271 the function in question. Normally it is a node of type
3272 @code{FUNCTION_DECL} that describes the declaration of the function.
3273 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3275 @var{funtype} is a C variable whose value is a tree node that
3276 describes the function in question. Normally it is a node of type
3277 @code{FUNCTION_TYPE} that describes the data type of the function.
3278 From this it is possible to obtain the data types of the value and
3279 arguments (if known).
3281 When a call to a library function is being considered, @var{fundecl}
3282 will contain an identifier node for the library function. Thus, if
3283 you need to distinguish among various library functions, you can do so
3284 by their names. Note that ``library function'' in this context means
3285 a function used to perform arithmetic, whose name is known specially
3286 in the compiler and was not mentioned in the C code being compiled.
3288 @var{stack-size} is the number of bytes of arguments passed on the
3289 stack. If a variable number of bytes is passed, it is zero, and
3290 argument popping will always be the responsibility of the calling function.
3292 On the VAX, all functions always pop their arguments, so the definition
3293 of this macro is @var{stack-size}. On the 68000, using the standard
3294 calling convention, no functions pop their arguments, so the value of
3295 the macro is always 0 in this case. But an alternative calling
3296 convention is available in which functions that take a fixed number of
3297 arguments pop them but other functions (such as @code{printf}) pop
3298 nothing (the caller pops all). When this convention is in use,
3299 @var{funtype} is examined to determine whether a function takes a fixed
3300 number of arguments.
3303 @node Register Arguments
3304 @subsection Passing Arguments in Registers
3305 @cindex arguments in registers
3306 @cindex registers arguments
3308 This section describes the macros which let you control how various
3309 types of arguments are passed in registers or how they are arranged in
3313 @findex FUNCTION_ARG
3314 @item FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3315 A C expression that controls whether a function argument is passed
3316 in a register, and which register.
3318 The arguments are @var{cum}, which summarizes all the previous
3319 arguments; @var{mode}, the machine mode of the argument; @var{type},
3320 the data type of the argument as a tree node or 0 if that is not known
3321 (which happens for C support library functions); and @var{named},
3322 which is 1 for an ordinary argument and 0 for nameless arguments that
3323 correspond to @samp{@dots{}} in the called function's prototype.
3324 @var{type} can be an incomplete type if a syntax error has previously
3327 The value of the expression is usually either a @code{reg} RTX for the
3328 hard register in which to pass the argument, or zero to pass the
3329 argument on the stack.
3331 For machines like the VAX and 68000, where normally all arguments are
3332 pushed, zero suffices as a definition.
3334 The value of the expression can also be a @code{parallel} RTX@. This is
3335 used when an argument is passed in multiple locations. The mode of the
3336 of the @code{parallel} should be the mode of the entire argument. The
3337 @code{parallel} holds any number of @code{expr_list} pairs; each one
3338 describes where part of the argument is passed. In each
3339 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
3340 register in which to pass this part of the argument, and the mode of the
3341 register RTX indicates how large this part of the argument is. The
3342 second operand of the @code{expr_list} is a @code{const_int} which gives
3343 the offset in bytes into the entire argument of where this part starts.
3344 As a special exception the first @code{expr_list} in the @code{parallel}
3345 RTX may have a first operand of zero. This indicates that the entire
3346 argument is also stored on the stack.
3348 The last time this macro is called, it is called with @code{MODE ==
3349 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
3350 pattern as operands 2 and 3 respectively.
3352 @cindex @file{stdarg.h} and register arguments
3353 The usual way to make the ISO library @file{stdarg.h} work on a machine
3354 where some arguments are usually passed in registers, is to cause
3355 nameless arguments to be passed on the stack instead. This is done
3356 by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
3358 @cindex @code{MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
3359 @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
3360 You may use the macro @code{MUST_PASS_IN_STACK (@var{mode}, @var{type})}
3361 in the definition of this macro to determine if this argument is of a
3362 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
3363 is not defined and @code{FUNCTION_ARG} returns nonzero for such an
3364 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
3365 defined, the argument will be computed in the stack and then loaded into
3368 @findex MUST_PASS_IN_STACK
3369 @item MUST_PASS_IN_STACK (@var{mode}, @var{type})
3370 Define as a C expression that evaluates to nonzero if we do not know how
3371 to pass TYPE solely in registers. The file @file{expr.h} defines a
3372 definition that is usually appropriate, refer to @file{expr.h} for additional
3375 @findex FUNCTION_INCOMING_ARG
3376 @item FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3377 Define this macro if the target machine has ``register windows'', so
3378 that the register in which a function sees an arguments is not
3379 necessarily the same as the one in which the caller passed the
3382 For such machines, @code{FUNCTION_ARG} computes the register in which
3383 the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
3384 be defined in a similar fashion to tell the function being called
3385 where the arguments will arrive.
3387 If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
3388 serves both purposes.
3390 @findex FUNCTION_ARG_PARTIAL_NREGS
3391 @item FUNCTION_ARG_PARTIAL_NREGS (@var{cum}, @var{mode}, @var{type}, @var{named})
3392 A C expression for the number of words, at the beginning of an
3393 argument, that must be put in registers. The value must be zero for
3394 arguments that are passed entirely in registers or that are entirely
3395 pushed on the stack.
3397 On some machines, certain arguments must be passed partially in
3398 registers and partially in memory. On these machines, typically the
3399 first @var{n} words of arguments are passed in registers, and the rest
3400 on the stack. If a multi-word argument (a @code{double} or a
3401 structure) crosses that boundary, its first few words must be passed
3402 in registers and the rest must be pushed. This macro tells the
3403 compiler when this occurs, and how many of the words should go in
3406 @code{FUNCTION_ARG} for these arguments should return the first
3407 register to be used by the caller for this argument; likewise
3408 @code{FUNCTION_INCOMING_ARG}, for the called function.
3410 @findex FUNCTION_ARG_PASS_BY_REFERENCE
3411 @item FUNCTION_ARG_PASS_BY_REFERENCE (@var{cum}, @var{mode}, @var{type}, @var{named})
3412 A C expression that indicates when an argument must be passed by reference.
3413 If nonzero for an argument, a copy of that argument is made in memory and a
3414 pointer to the argument is passed instead of the argument itself.
3415 The pointer is passed in whatever way is appropriate for passing a pointer
3418 On machines where @code{REG_PARM_STACK_SPACE} is not defined, a suitable
3419 definition of this macro might be
3421 #define FUNCTION_ARG_PASS_BY_REFERENCE\
3422 (CUM, MODE, TYPE, NAMED) \
3423 MUST_PASS_IN_STACK (MODE, TYPE)
3425 @c this is *still* too long. --mew 5feb93
3427 @findex FUNCTION_ARG_CALLEE_COPIES
3428 @item FUNCTION_ARG_CALLEE_COPIES (@var{cum}, @var{mode}, @var{type}, @var{named})
3429 If defined, a C expression that indicates when it is the called function's
3430 responsibility to make a copy of arguments passed by invisible reference.
3431 Normally, the caller makes a copy and passes the address of the copy to the
3432 routine being called. When @code{FUNCTION_ARG_CALLEE_COPIES} is defined and is
3433 nonzero, the caller does not make a copy. Instead, it passes a pointer to the
3434 ``live'' value. The called function must not modify this value. If it can be
3435 determined that the value won't be modified, it need not make a copy;
3436 otherwise a copy must be made.
3438 @findex FUNCTION_ARG_REG_LITTLE_ENDIAN
3439 @item FUNCTION_ARG_REG_LITTLE_ENDIAN
3440 If defined TRUE on a big-endian system then structure arguments passed
3441 (and returned) in registers are passed in a little-endian manner instead of
3442 the big-endian manner. On the HP-UX IA64 and PA64 platforms structures are
3443 aligned differently then integral values and setting this value to true will
3444 allow for the special handling of structure arguments and return values.
3446 @findex CUMULATIVE_ARGS
3447 @item CUMULATIVE_ARGS
3448 A C type for declaring a variable that is used as the first argument of
3449 @code{FUNCTION_ARG} and other related values. For some target machines,
3450 the type @code{int} suffices and can hold the number of bytes of
3453 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
3454 arguments that have been passed on the stack. The compiler has other
3455 variables to keep track of that. For target machines on which all
3456 arguments are passed on the stack, there is no need to store anything in
3457 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
3458 should not be empty, so use @code{int}.
3460 @findex INIT_CUMULATIVE_ARGS
3461 @item INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{indirect})
3462 A C statement (sans semicolon) for initializing the variable @var{cum}
3463 for the state at the beginning of the argument list. The variable has
3464 type @code{CUMULATIVE_ARGS}. The value of @var{fntype} is the tree node
3465 for the data type of the function which will receive the args, or 0
3466 if the args are to a compiler support library function. The value of
3467 @var{indirect} is nonzero when processing an indirect call, for example
3468 a call through a function pointer. The value of @var{indirect} is zero
3469 for a call to an explicitly named function, a library function call, or when
3470 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
3473 When processing a call to a compiler support library function,
3474 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
3475 contains the name of the function, as a string. @var{libname} is 0 when
3476 an ordinary C function call is being processed. Thus, each time this
3477 macro is called, either @var{libname} or @var{fntype} is nonzero, but
3478 never both of them at once.
3480 @findex INIT_CUMULATIVE_LIBCALL_ARGS
3481 @item INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
3482 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
3483 it gets a @code{MODE} argument instead of @var{fntype}, that would be
3484 @code{NULL}. @var{indirect} would always be zero, too. If this macro
3485 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
3486 0)} is used instead.
3488 @findex INIT_CUMULATIVE_INCOMING_ARGS
3489 @item INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
3490 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
3491 finding the arguments for the function being compiled. If this macro is
3492 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
3494 The value passed for @var{libname} is always 0, since library routines
3495 with special calling conventions are never compiled with GCC@. The
3496 argument @var{libname} exists for symmetry with
3497 @code{INIT_CUMULATIVE_ARGS}.
3498 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
3499 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
3501 @findex FUNCTION_ARG_ADVANCE
3502 @item FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
3503 A C statement (sans semicolon) to update the summarizer variable
3504 @var{cum} to advance past an argument in the argument list. The
3505 values @var{mode}, @var{type} and @var{named} describe that argument.
3506 Once this is done, the variable @var{cum} is suitable for analyzing
3507 the @emph{following} argument with @code{FUNCTION_ARG}, etc.
3509 This macro need not do anything if the argument in question was passed
3510 on the stack. The compiler knows how to track the amount of stack space
3511 used for arguments without any special help.
3513 @findex FUNCTION_ARG_PADDING
3514 @item FUNCTION_ARG_PADDING (@var{mode}, @var{type})
3515 If defined, a C expression which determines whether, and in which direction,
3516 to pad out an argument with extra space. The value should be of type
3517 @code{enum direction}: either @code{upward} to pad above the argument,
3518 @code{downward} to pad below, or @code{none} to inhibit padding.
3520 The @emph{amount} of padding is always just enough to reach the next
3521 multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control
3524 This macro has a default definition which is right for most systems.
3525 For little-endian machines, the default is to pad upward. For
3526 big-endian machines, the default is to pad downward for an argument of
3527 constant size shorter than an @code{int}, and upward otherwise.
3529 @findex PAD_VARARGS_DOWN
3530 @item PAD_VARARGS_DOWN
3531 If defined, a C expression which determines whether the default
3532 implementation of va_arg will attempt to pad down before reading the
3533 next argument, if that argument is smaller than its aligned space as
3534 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
3535 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
3537 @findex FUNCTION_ARG_BOUNDARY
3538 @item FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
3539 If defined, a C expression that gives the alignment boundary, in bits,
3540 of an argument with the specified mode and type. If it is not defined,
3541 @code{PARM_BOUNDARY} is used for all arguments.
3543 @findex FUNCTION_ARG_REGNO_P
3544 @item FUNCTION_ARG_REGNO_P (@var{regno})
3545 A C expression that is nonzero if @var{regno} is the number of a hard
3546 register in which function arguments are sometimes passed. This does
3547 @emph{not} include implicit arguments such as the static chain and
3548 the structure-value address. On many machines, no registers can be
3549 used for this purpose since all function arguments are pushed on the
3552 @findex LOAD_ARGS_REVERSED
3553 @item LOAD_ARGS_REVERSED
3554 If defined, the order in which arguments are loaded into their
3555 respective argument registers is reversed so that the last
3556 argument is loaded first. This macro only affects arguments
3557 passed in registers.
3562 @subsection How Scalar Function Values Are Returned
3563 @cindex return values in registers
3564 @cindex values, returned by functions
3565 @cindex scalars, returned as values
3567 This section discusses the macros that control returning scalars as
3568 values---values that can fit in registers.
3571 @findex TRADITIONAL_RETURN_FLOAT
3572 @item TRADITIONAL_RETURN_FLOAT
3573 Define this macro if @option{-traditional} should not cause functions
3574 declared to return @code{float} to convert the value to @code{double}.
3576 @findex FUNCTION_VALUE
3577 @item FUNCTION_VALUE (@var{valtype}, @var{func})
3578 A C expression to create an RTX representing the place where a
3579 function returns a value of data type @var{valtype}. @var{valtype} is
3580 a tree node representing a data type. Write @code{TYPE_MODE
3581 (@var{valtype})} to get the machine mode used to represent that type.
3582 On many machines, only the mode is relevant. (Actually, on most
3583 machines, scalar values are returned in the same place regardless of
3586 The value of the expression is usually a @code{reg} RTX for the hard
3587 register where the return value is stored. The value can also be a
3588 @code{parallel} RTX, if the return value is in multiple places. See
3589 @code{FUNCTION_ARG} for an explanation of the @code{parallel} form.
3591 If @code{PROMOTE_FUNCTION_RETURN} is defined, you must apply the same
3592 promotion rules specified in @code{PROMOTE_MODE} if @var{valtype} is a
3595 If the precise function being called is known, @var{func} is a tree
3596 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
3597 pointer. This makes it possible to use a different value-returning
3598 convention for specific functions when all their calls are
3601 @code{FUNCTION_VALUE} is not used for return vales with aggregate data
3602 types, because these are returned in another way. See
3603 @code{STRUCT_VALUE_REGNUM} and related macros, below.
3605 @findex FUNCTION_OUTGOING_VALUE
3606 @item FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func})
3607 Define this macro if the target machine has ``register windows''
3608 so that the register in which a function returns its value is not
3609 the same as the one in which the caller sees the value.
3611 For such machines, @code{FUNCTION_VALUE} computes the register in which
3612 the caller will see the value. @code{FUNCTION_OUTGOING_VALUE} should be
3613 defined in a similar fashion to tell the function where to put the
3616 If @code{FUNCTION_OUTGOING_VALUE} is not defined,
3617 @code{FUNCTION_VALUE} serves both purposes.
3619 @code{FUNCTION_OUTGOING_VALUE} is not used for return vales with
3620 aggregate data types, because these are returned in another way. See
3621 @code{STRUCT_VALUE_REGNUM} and related macros, below.
3623 @findex LIBCALL_VALUE
3624 @item LIBCALL_VALUE (@var{mode})
3625 A C expression to create an RTX representing the place where a library
3626 function returns a value of mode @var{mode}. If the precise function
3627 being called is known, @var{func} is a tree node
3628 (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
3629 pointer. This makes it possible to use a different value-returning
3630 convention for specific functions when all their calls are
3633 Note that ``library function'' in this context means a compiler
3634 support routine, used to perform arithmetic, whose name is known
3635 specially by the compiler and was not mentioned in the C code being
3638 The definition of @code{LIBRARY_VALUE} need not be concerned aggregate
3639 data types, because none of the library functions returns such types.
3641 @findex FUNCTION_VALUE_REGNO_P
3642 @item FUNCTION_VALUE_REGNO_P (@var{regno})
3643 A C expression that is nonzero if @var{regno} is the number of a hard
3644 register in which the values of called function may come back.
3646 A register whose use for returning values is limited to serving as the
3647 second of a pair (for a value of type @code{double}, say) need not be
3648 recognized by this macro. So for most machines, this definition
3652 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
3655 If the machine has register windows, so that the caller and the called
3656 function use different registers for the return value, this macro
3657 should recognize only the caller's register numbers.
3659 @findex APPLY_RESULT_SIZE
3660 @item APPLY_RESULT_SIZE
3661 Define this macro if @samp{untyped_call} and @samp{untyped_return}
3662 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
3663 saving and restoring an arbitrary return value.
3666 @node Aggregate Return
3667 @subsection How Large Values Are Returned
3668 @cindex aggregates as return values
3669 @cindex large return values
3670 @cindex returning aggregate values
3671 @cindex structure value address
3673 When a function value's mode is @code{BLKmode} (and in some other
3674 cases), the value is not returned according to @code{FUNCTION_VALUE}
3675 (@pxref{Scalar Return}). Instead, the caller passes the address of a
3676 block of memory in which the value should be stored. This address
3677 is called the @dfn{structure value address}.
3679 This section describes how to control returning structure values in
3683 @findex RETURN_IN_MEMORY
3684 @item RETURN_IN_MEMORY (@var{type})
3685 A C expression which can inhibit the returning of certain function
3686 values in registers, based on the type of value. A nonzero value says
3687 to return the function value in memory, just as large structures are
3688 always returned. Here @var{type} will be a C expression of type
3689 @code{tree}, representing the data type of the value.
3691 Note that values of mode @code{BLKmode} must be explicitly handled
3692 by this macro. Also, the option @option{-fpcc-struct-return}
3693 takes effect regardless of this macro. On most systems, it is
3694 possible to leave the macro undefined; this causes a default
3695 definition to be used, whose value is the constant 1 for @code{BLKmode}
3696 values, and 0 otherwise.
3698 Do not use this macro to indicate that structures and unions should always
3699 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
3702 @findex DEFAULT_PCC_STRUCT_RETURN
3703 @item DEFAULT_PCC_STRUCT_RETURN
3704 Define this macro to be 1 if all structure and union return values must be
3705 in memory. Since this results in slower code, this should be defined
3706 only if needed for compatibility with other compilers or with an ABI@.
3707 If you define this macro to be 0, then the conventions used for structure
3708 and union return values are decided by the @code{RETURN_IN_MEMORY} macro.
3710 If not defined, this defaults to the value 1.
3712 @findex STRUCT_VALUE_REGNUM
3713 @item STRUCT_VALUE_REGNUM
3714 If the structure value address is passed in a register, then
3715 @code{STRUCT_VALUE_REGNUM} should be the number of that register.
3717 @findex STRUCT_VALUE
3719 If the structure value address is not passed in a register, define
3720 @code{STRUCT_VALUE} as an expression returning an RTX for the place
3721 where the address is passed. If it returns 0, the address is passed as
3722 an ``invisible'' first argument.
3724 @findex STRUCT_VALUE_INCOMING_REGNUM
3725 @item STRUCT_VALUE_INCOMING_REGNUM
3726 On some architectures the place where the structure value address
3727 is found by the called function is not the same place that the
3728 caller put it. This can be due to register windows, or it could
3729 be because the function prologue moves it to a different place.
3731 If the incoming location of the structure value address is in a
3732 register, define this macro as the register number.
3734 @findex STRUCT_VALUE_INCOMING
3735 @item STRUCT_VALUE_INCOMING
3736 If the incoming location is not a register, then you should define
3737 @code{STRUCT_VALUE_INCOMING} as an expression for an RTX for where the
3738 called function should find the value. If it should find the value on
3739 the stack, define this to create a @code{mem} which refers to the frame
3740 pointer. A definition of 0 means that the address is passed as an
3741 ``invisible'' first argument.
3743 @findex PCC_STATIC_STRUCT_RETURN
3744 @item PCC_STATIC_STRUCT_RETURN
3745 Define this macro if the usual system convention on the target machine
3746 for returning structures and unions is for the called function to return
3747 the address of a static variable containing the value.
3749 Do not define this if the usual system convention is for the caller to
3750 pass an address to the subroutine.
3752 This macro has effect in @option{-fpcc-struct-return} mode, but it does
3753 nothing when you use @option{-freg-struct-return} mode.
3757 @subsection Caller-Saves Register Allocation
3759 If you enable it, GCC can save registers around function calls. This
3760 makes it possible to use call-clobbered registers to hold variables that
3761 must live across calls.
3764 @findex DEFAULT_CALLER_SAVES
3765 @item DEFAULT_CALLER_SAVES
3766 Define this macro if function calls on the target machine do not preserve
3767 any registers; in other words, if @code{CALL_USED_REGISTERS} has 1
3768 for all registers. When defined, this macro enables @option{-fcaller-saves}
3769 by default for all optimization levels. It has no effect for optimization
3770 levels 2 and higher, where @option{-fcaller-saves} is the default.
3772 @findex CALLER_SAVE_PROFITABLE
3773 @item CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
3774 A C expression to determine whether it is worthwhile to consider placing
3775 a pseudo-register in a call-clobbered hard register and saving and
3776 restoring it around each function call. The expression should be 1 when
3777 this is worth doing, and 0 otherwise.
3779 If you don't define this macro, a default is used which is good on most
3780 machines: @code{4 * @var{calls} < @var{refs}}.
3782 @findex HARD_REGNO_CALLER_SAVE_MODE
3783 @item HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
3784 A C expression specifying which mode is required for saving @var{nregs}
3785 of a pseudo-register in call-clobbered hard register @var{regno}. If
3786 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
3787 returned. For most machines this macro need not be defined since GCC
3788 will select the smallest suitable mode.
3791 @node Function Entry
3792 @subsection Function Entry and Exit
3793 @cindex function entry and exit
3797 This section describes the macros that output function entry
3798 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
3800 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
3801 If defined, a function that outputs the assembler code for entry to a
3802 function. The prologue is responsible for setting up the stack frame,
3803 initializing the frame pointer register, saving registers that must be
3804 saved, and allocating @var{size} additional bytes of storage for the
3805 local variables. @var{size} is an integer. @var{file} is a stdio
3806 stream to which the assembler code should be output.
3808 The label for the beginning of the function need not be output by this
3809 macro. That has already been done when the macro is run.
3811 @findex regs_ever_live
3812 To determine which registers to save, the macro can refer to the array
3813 @code{regs_ever_live}: element @var{r} is nonzero if hard register
3814 @var{r} is used anywhere within the function. This implies the function
3815 prologue should save register @var{r}, provided it is not one of the
3816 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
3817 @code{regs_ever_live}.)
3819 On machines that have ``register windows'', the function entry code does
3820 not save on the stack the registers that are in the windows, even if
3821 they are supposed to be preserved by function calls; instead it takes
3822 appropriate steps to ``push'' the register stack, if any non-call-used
3823 registers are used in the function.
3825 @findex frame_pointer_needed
3826 On machines where functions may or may not have frame-pointers, the
3827 function entry code must vary accordingly; it must set up the frame
3828 pointer if one is wanted, and not otherwise. To determine whether a
3829 frame pointer is in wanted, the macro can refer to the variable
3830 @code{frame_pointer_needed}. The variable's value will be 1 at run
3831 time in a function that needs a frame pointer. @xref{Elimination}.
3833 The function entry code is responsible for allocating any stack space
3834 required for the function. This stack space consists of the regions
3835 listed below. In most cases, these regions are allocated in the
3836 order listed, with the last listed region closest to the top of the
3837 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
3838 the highest address if it is not defined). You can use a different order
3839 for a machine if doing so is more convenient or required for
3840 compatibility reasons. Except in cases where required by standard
3841 or by a debugger, there is no reason why the stack layout used by GCC
3842 need agree with that used by other compilers for a machine.
3845 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
3846 If defined, a function that outputs assembler code at the end of a
3847 prologue. This should be used when the function prologue is being
3848 emitted as RTL, and you have some extra assembler that needs to be
3849 emitted. @xref{prologue instruction pattern}.
3852 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
3853 If defined, a function that outputs assembler code at the start of an
3854 epilogue. This should be used when the function epilogue is being
3855 emitted as RTL, and you have some extra assembler that needs to be
3856 emitted. @xref{epilogue instruction pattern}.
3859 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
3860 If defined, a function that outputs the assembler code for exit from a
3861 function. The epilogue is responsible for restoring the saved
3862 registers and stack pointer to their values when the function was
3863 called, and returning control to the caller. This macro takes the
3864 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
3865 registers to restore are determined from @code{regs_ever_live} and
3866 @code{CALL_USED_REGISTERS} in the same way.
3868 On some machines, there is a single instruction that does all the work
3869 of returning from the function. On these machines, give that
3870 instruction the name @samp{return} and do not define the macro
3871 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
3873 Do not define a pattern named @samp{return} if you want the
3874 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
3875 switches to control whether return instructions or epilogues are used,
3876 define a @samp{return} pattern with a validity condition that tests the
3877 target switches appropriately. If the @samp{return} pattern's validity
3878 condition is false, epilogues will be used.
3880 On machines where functions may or may not have frame-pointers, the
3881 function exit code must vary accordingly. Sometimes the code for these
3882 two cases is completely different. To determine whether a frame pointer
3883 is wanted, the macro can refer to the variable
3884 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
3885 a function that needs a frame pointer.
3887 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
3888 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
3889 The C variable @code{current_function_is_leaf} is nonzero for such a
3890 function. @xref{Leaf Functions}.
3892 On some machines, some functions pop their arguments on exit while
3893 others leave that for the caller to do. For example, the 68020 when
3894 given @option{-mrtd} pops arguments in functions that take a fixed
3895 number of arguments.
3897 @findex current_function_pops_args
3898 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
3899 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
3900 needs to know what was decided. The variable that is called
3901 @code{current_function_pops_args} is the number of bytes of its
3902 arguments that a function should pop. @xref{Scalar Return}.
3903 @c what is the "its arguments" in the above sentence referring to, pray
3904 @c tell? --mew 5feb93
3911 @findex current_function_pretend_args_size
3912 A region of @code{current_function_pretend_args_size} bytes of
3913 uninitialized space just underneath the first argument arriving on the
3914 stack. (This may not be at the very start of the allocated stack region
3915 if the calling sequence has pushed anything else since pushing the stack
3916 arguments. But usually, on such machines, nothing else has been pushed
3917 yet, because the function prologue itself does all the pushing.) This
3918 region is used on machines where an argument may be passed partly in
3919 registers and partly in memory, and, in some cases to support the
3920 features in @code{<varargs.h>} and @code{<stdarg.h>}.
3923 An area of memory used to save certain registers used by the function.
3924 The size of this area, which may also include space for such things as
3925 the return address and pointers to previous stack frames, is
3926 machine-specific and usually depends on which registers have been used
3927 in the function. Machines with register windows often do not require
3931 A region of at least @var{size} bytes, possibly rounded up to an allocation
3932 boundary, to contain the local variables of the function. On some machines,
3933 this region and the save area may occur in the opposite order, with the
3934 save area closer to the top of the stack.
3937 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
3938 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
3939 @code{current_function_outgoing_args_size} bytes to be used for outgoing
3940 argument lists of the function. @xref{Stack Arguments}.
3943 Normally, it is necessary for the macros
3944 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
3945 @code{TARGET_ASM_FUNCTION_EPILOGUE} to treat leaf functions specially.
3946 The C variable @code{current_function_is_leaf} is nonzero for such a
3949 @findex EXIT_IGNORE_STACK
3950 @item EXIT_IGNORE_STACK
3951 Define this macro as a C expression that is nonzero if the return
3952 instruction or the function epilogue ignores the value of the stack
3953 pointer; in other words, if it is safe to delete an instruction to
3954 adjust the stack pointer before a return from the function.
3956 Note that this macro's value is relevant only for functions for which
3957 frame pointers are maintained. It is never safe to delete a final
3958 stack adjustment in a function that has no frame pointer, and the
3959 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
3961 @findex EPILOGUE_USES
3962 @item EPILOGUE_USES (@var{regno})
3963 Define this macro as a C expression that is nonzero for registers that are
3964 used by the epilogue or the @samp{return} pattern. The stack and frame
3965 pointer registers are already be assumed to be used as needed.
3967 @findex DELAY_SLOTS_FOR_EPILOGUE
3968 @item DELAY_SLOTS_FOR_EPILOGUE
3969 Define this macro if the function epilogue contains delay slots to which
3970 instructions from the rest of the function can be ``moved''. The
3971 definition should be a C expression whose value is an integer
3972 representing the number of delay slots there.
3974 @findex ELIGIBLE_FOR_EPILOGUE_DELAY
3975 @item ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
3976 A C expression that returns 1 if @var{insn} can be placed in delay
3977 slot number @var{n} of the epilogue.
3979 The argument @var{n} is an integer which identifies the delay slot now
3980 being considered (since different slots may have different rules of
3981 eligibility). It is never negative and is always less than the number
3982 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
3983 If you reject a particular insn for a given delay slot, in principle, it
3984 may be reconsidered for a subsequent delay slot. Also, other insns may
3985 (at least in principle) be considered for the so far unfilled delay
3988 @findex current_function_epilogue_delay_list
3989 @findex final_scan_insn
3990 The insns accepted to fill the epilogue delay slots are put in an RTL
3991 list made with @code{insn_list} objects, stored in the variable
3992 @code{current_function_epilogue_delay_list}. The insn for the first
3993 delay slot comes first in the list. Your definition of the macro
3994 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
3995 outputting the insns in this list, usually by calling
3996 @code{final_scan_insn}.
3998 You need not define this macro if you did not define
3999 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4001 @findex ASM_OUTPUT_MI_THUNK
4002 @item ASM_OUTPUT_MI_THUNK (@var{file}, @var{thunk_fndecl}, @var{delta}, @var{function})
4003 A C compound statement that outputs the assembler code for a thunk
4004 function, used to implement C++ virtual function calls with multiple
4005 inheritance. The thunk acts as a wrapper around a virtual function,
4006 adjusting the implicit object parameter before handing control off to
4009 First, emit code to add the integer @var{delta} to the location that
4010 contains the incoming first argument. Assume that this argument
4011 contains a pointer, and is the one used to pass the @code{this} pointer
4012 in C++. This is the incoming argument @emph{before} the function prologue,
4013 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4014 all other incoming arguments.
4016 After the addition, emit code to jump to @var{function}, which is a
4017 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4018 not touch the return address. Hence returning from @var{FUNCTION} will
4019 return to whoever called the current @samp{thunk}.
4021 The effect must be as if @var{function} had been called directly with
4022 the adjusted first argument. This macro is responsible for emitting all
4023 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4024 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4026 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4027 have already been extracted from it.) It might possibly be useful on
4028 some targets, but probably not.
4030 If you do not define this macro, the target-independent code in the C++
4031 front end will generate a less efficient heavyweight thunk that calls
4032 @var{function} instead of jumping to it. The generic approach does
4033 not support varargs.
4037 @subsection Generating Code for Profiling
4038 @cindex profiling, code generation
4040 These macros will help you generate code for profiling.
4043 @findex FUNCTION_PROFILER
4044 @item FUNCTION_PROFILER (@var{file}, @var{labelno})
4045 A C statement or compound statement to output to @var{file} some
4046 assembler code to call the profiling subroutine @code{mcount}.
4049 The details of how @code{mcount} expects to be called are determined by
4050 your operating system environment, not by GCC@. To figure them out,
4051 compile a small program for profiling using the system's installed C
4052 compiler and look at the assembler code that results.
4054 Older implementations of @code{mcount} expect the address of a counter
4055 variable to be loaded into some register. The name of this variable is
4056 @samp{LP} followed by the number @var{labelno}, so you would generate
4057 the name using @samp{LP%d} in a @code{fprintf}.
4059 @findex PROFILE_HOOK
4061 A C statement or compound statement to output to @var{file} some assembly
4062 code to call the profiling subroutine @code{mcount} even the target does
4063 not support profiling.
4065 @findex NO_PROFILE_COUNTERS
4066 @item NO_PROFILE_COUNTERS
4067 Define this macro if the @code{mcount} subroutine on your system does
4068 not need a counter variable allocated for each function. This is true
4069 for almost all modern implementations. If you define this macro, you
4070 must not use the @var{labelno} argument to @code{FUNCTION_PROFILER}.
4072 @findex PROFILE_BEFORE_PROLOGUE
4073 @item PROFILE_BEFORE_PROLOGUE
4074 Define this macro if the code for function profiling should come before
4075 the function prologue. Normally, the profiling code comes after.
4078 @findex TARGET_ALLOWS_PROFILING_WITHOUT_FRAME_POINTER
4079 @item TARGET_ALLOWS_PROFILING_WITHOUT_FRAME_POINTER
4080 On some targets, it is impossible to use profiling when the frame
4081 pointer has been omitted. For example, on x86 GNU/Linux systems,
4082 the @code{mcount} routine provided by the GNU C Library finds the
4083 address of the routine that called the routine that called @code{mcount}
4084 by looking in the immediate caller's stack frame. If the immediate
4085 caller has no frame pointer, this lookup will fail.
4087 By default, GCC assumes that the target does allow profiling when the
4088 frame pointer is omitted. This macro should be defined to a C
4089 expression that evaluates to @code{false} if the target does not allow
4090 profiling when the frame pointer is omitted.
4095 @subsection Permitting tail calls
4099 @findex FUNCTION_OK_FOR_SIBCALL
4100 @item FUNCTION_OK_FOR_SIBCALL (@var{decl})
4101 A C expression that evaluates to true if it is ok to perform a sibling
4102 call to @var{decl} from the current function.
4104 It is not uncommon for limitations of calling conventions to prevent
4105 tail calls to functions outside the current unit of translation, or
4106 during PIC compilation. Use this macro to enforce these restrictions,
4107 as the @code{sibcall} md pattern can not fail, or fall over to a
4112 @section Implementing the Varargs Macros
4113 @cindex varargs implementation
4115 GCC comes with an implementation of @code{<varargs.h>} and
4116 @code{<stdarg.h>} that work without change on machines that pass arguments
4117 on the stack. Other machines require their own implementations of
4118 varargs, and the two machine independent header files must have
4119 conditionals to include it.
4121 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4122 the calling convention for @code{va_start}. The traditional
4123 implementation takes just one argument, which is the variable in which
4124 to store the argument pointer. The ISO implementation of
4125 @code{va_start} takes an additional second argument. The user is
4126 supposed to write the last named argument of the function here.
4128 However, @code{va_start} should not use this argument. The way to find
4129 the end of the named arguments is with the built-in functions described
4133 @findex __builtin_saveregs
4134 @item __builtin_saveregs ()
4135 Use this built-in function to save the argument registers in memory so
4136 that the varargs mechanism can access them. Both ISO and traditional
4137 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4138 you use @code{SETUP_INCOMING_VARARGS} (see below) instead.
4140 On some machines, @code{__builtin_saveregs} is open-coded under the
4141 control of the macro @code{EXPAND_BUILTIN_SAVEREGS}. On other machines,
4142 it calls a routine written in assembler language, found in
4145 Code generated for the call to @code{__builtin_saveregs} appears at the
4146 beginning of the function, as opposed to where the call to
4147 @code{__builtin_saveregs} is written, regardless of what the code is.
4148 This is because the registers must be saved before the function starts
4149 to use them for its own purposes.
4150 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4153 @findex __builtin_args_info
4154 @item __builtin_args_info (@var{category})
4155 Use this built-in function to find the first anonymous arguments in
4158 In general, a machine may have several categories of registers used for
4159 arguments, each for a particular category of data types. (For example,
4160 on some machines, floating-point registers are used for floating-point
4161 arguments while other arguments are passed in the general registers.)
4162 To make non-varargs functions use the proper calling convention, you
4163 have defined the @code{CUMULATIVE_ARGS} data type to record how many
4164 registers in each category have been used so far
4166 @code{__builtin_args_info} accesses the same data structure of type
4167 @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
4168 with it, with @var{category} specifying which word to access. Thus, the
4169 value indicates the first unused register in a given category.
4171 Normally, you would use @code{__builtin_args_info} in the implementation
4172 of @code{va_start}, accessing each category just once and storing the
4173 value in the @code{va_list} object. This is because @code{va_list} will
4174 have to update the values, and there is no way to alter the
4175 values accessed by @code{__builtin_args_info}.
4177 @findex __builtin_next_arg
4178 @item __builtin_next_arg (@var{lastarg})
4179 This is the equivalent of @code{__builtin_args_info}, for stack
4180 arguments. It returns the address of the first anonymous stack
4181 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
4182 returns the address of the location above the first anonymous stack
4183 argument. Use it in @code{va_start} to initialize the pointer for
4184 fetching arguments from the stack. Also use it in @code{va_start} to
4185 verify that the second parameter @var{lastarg} is the last named argument
4186 of the current function.
4188 @findex __builtin_classify_type
4189 @item __builtin_classify_type (@var{object})
4190 Since each machine has its own conventions for which data types are
4191 passed in which kind of register, your implementation of @code{va_arg}
4192 has to embody these conventions. The easiest way to categorize the
4193 specified data type is to use @code{__builtin_classify_type} together
4194 with @code{sizeof} and @code{__alignof__}.
4196 @code{__builtin_classify_type} ignores the value of @var{object},
4197 considering only its data type. It returns an integer describing what
4198 kind of type that is---integer, floating, pointer, structure, and so on.
4200 The file @file{typeclass.h} defines an enumeration that you can use to
4201 interpret the values of @code{__builtin_classify_type}.
4204 These machine description macros help implement varargs:
4207 @findex EXPAND_BUILTIN_SAVEREGS
4208 @item EXPAND_BUILTIN_SAVEREGS ()
4209 If defined, is a C expression that produces the machine-specific code
4210 for a call to @code{__builtin_saveregs}. This code will be moved to the
4211 very beginning of the function, before any parameter access are made.
4212 The return value of this function should be an RTX that contains the
4213 value to use as the return of @code{__builtin_saveregs}.
4215 @findex SETUP_INCOMING_VARARGS
4216 @item SETUP_INCOMING_VARARGS (@var{args_so_far}, @var{mode}, @var{type}, @var{pretend_args_size}, @var{second_time})
4217 This macro offers an alternative to using @code{__builtin_saveregs} and
4218 defining the macro @code{EXPAND_BUILTIN_SAVEREGS}. Use it to store the
4219 anonymous register arguments into the stack so that all the arguments
4220 appear to have been passed consecutively on the stack. Once this is
4221 done, you can use the standard implementation of varargs that works for
4222 machines that pass all their arguments on the stack.
4224 The argument @var{args_so_far} is the @code{CUMULATIVE_ARGS} data
4225 structure, containing the values that are obtained after processing the
4226 named arguments. The arguments @var{mode} and @var{type} describe the
4227 last named argument---its machine mode and its data type as a tree node.
4229 The macro implementation should do two things: first, push onto the
4230 stack all the argument registers @emph{not} used for the named
4231 arguments, and second, store the size of the data thus pushed into the
4232 @code{int}-valued variable whose name is supplied as the argument
4233 @var{pretend_args_size}. The value that you store here will serve as
4234 additional offset for setting up the stack frame.
4236 Because you must generate code to push the anonymous arguments at
4237 compile time without knowing their data types,
4238 @code{SETUP_INCOMING_VARARGS} is only useful on machines that have just
4239 a single category of argument register and use it uniformly for all data
4242 If the argument @var{second_time} is nonzero, it means that the
4243 arguments of the function are being analyzed for the second time. This
4244 happens for an inline function, which is not actually compiled until the
4245 end of the source file. The macro @code{SETUP_INCOMING_VARARGS} should
4246 not generate any instructions in this case.
4248 @findex STRICT_ARGUMENT_NAMING
4249 @item STRICT_ARGUMENT_NAMING
4250 Define this macro to be a nonzero value if the location where a function
4251 argument is passed depends on whether or not it is a named argument.
4253 This macro controls how the @var{named} argument to @code{FUNCTION_ARG}
4254 is set for varargs and stdarg functions. If this macro returns a
4255 nonzero value, the @var{named} argument is always true for named
4256 arguments, and false for unnamed arguments. If it returns a value of
4257 zero, but @code{SETUP_INCOMING_VARARGS} is defined, then all arguments
4258 are treated as named. Otherwise, all named arguments except the last
4259 are treated as named.
4261 You need not define this macro if it always returns zero.
4263 @findex PRETEND_OUTGOING_VARARGS_NAMED
4264 @item PRETEND_OUTGOING_VARARGS_NAMED
4265 If you need to conditionally change ABIs so that one works with
4266 @code{SETUP_INCOMING_VARARGS}, but the other works like neither
4267 @code{SETUP_INCOMING_VARARGS} nor @code{STRICT_ARGUMENT_NAMING} was
4268 defined, then define this macro to return nonzero if
4269 @code{SETUP_INCOMING_VARARGS} is used, zero otherwise.
4270 Otherwise, you should not define this macro.
4274 @section Trampolines for Nested Functions
4275 @cindex trampolines for nested functions
4276 @cindex nested functions, trampolines for
4278 A @dfn{trampoline} is a small piece of code that is created at run time
4279 when the address of a nested function is taken. It normally resides on
4280 the stack, in the stack frame of the containing function. These macros
4281 tell GCC how to generate code to allocate and initialize a
4284 The instructions in the trampoline must do two things: load a constant
4285 address into the static chain register, and jump to the real address of
4286 the nested function. On CISC machines such as the m68k, this requires
4287 two instructions, a move immediate and a jump. Then the two addresses
4288 exist in the trampoline as word-long immediate operands. On RISC
4289 machines, it is often necessary to load each address into a register in
4290 two parts. Then pieces of each address form separate immediate
4293 The code generated to initialize the trampoline must store the variable
4294 parts---the static chain value and the function address---into the
4295 immediate operands of the instructions. On a CISC machine, this is
4296 simply a matter of copying each address to a memory reference at the
4297 proper offset from the start of the trampoline. On a RISC machine, it
4298 may be necessary to take out pieces of the address and store them
4302 @findex TRAMPOLINE_TEMPLATE
4303 @item TRAMPOLINE_TEMPLATE (@var{file})
4304 A C statement to output, on the stream @var{file}, assembler code for a
4305 block of data that contains the constant parts of a trampoline. This
4306 code should not include a label---the label is taken care of
4309 If you do not define this macro, it means no template is needed
4310 for the target. Do not define this macro on systems where the block move
4311 code to copy the trampoline into place would be larger than the code
4312 to generate it on the spot.
4314 @findex TRAMPOLINE_SECTION
4315 @item TRAMPOLINE_SECTION
4316 The name of a subroutine to switch to the section in which the
4317 trampoline template is to be placed (@pxref{Sections}). The default is
4318 a value of @samp{readonly_data_section}, which places the trampoline in
4319 the section containing read-only data.
4321 @findex TRAMPOLINE_SIZE
4322 @item TRAMPOLINE_SIZE
4323 A C expression for the size in bytes of the trampoline, as an integer.
4325 @findex TRAMPOLINE_ALIGNMENT
4326 @item TRAMPOLINE_ALIGNMENT
4327 Alignment required for trampolines, in bits.
4329 If you don't define this macro, the value of @code{BIGGEST_ALIGNMENT}
4330 is used for aligning trampolines.
4332 @findex INITIALIZE_TRAMPOLINE
4333 @item INITIALIZE_TRAMPOLINE (@var{addr}, @var{fnaddr}, @var{static_chain})
4334 A C statement to initialize the variable parts of a trampoline.
4335 @var{addr} is an RTX for the address of the trampoline; @var{fnaddr} is
4336 an RTX for the address of the nested function; @var{static_chain} is an
4337 RTX for the static chain value that should be passed to the function
4340 @findex TRAMPOLINE_ADJUST_ADDRESS
4341 @item TRAMPOLINE_ADJUST_ADDRESS (@var{addr})
4342 A C statement that should perform any machine-specific adjustment in
4343 the address of the trampoline. Its argument contains the address that
4344 was passed to @code{INITIALIZE_TRAMPOLINE}. In case the address to be
4345 used for a function call should be different from the address in which
4346 the template was stored, the different address should be assigned to
4347 @var{addr}. If this macro is not defined, @var{addr} will be used for
4350 @findex ALLOCATE_TRAMPOLINE
4351 @item ALLOCATE_TRAMPOLINE (@var{fp})
4352 A C expression to allocate run-time space for a trampoline. The
4353 expression value should be an RTX representing a memory reference to the
4354 space for the trampoline.
4356 @cindex @code{TARGET_ASM_FUNCTION_EPILOGUE} and trampolines
4357 @cindex @code{TARGET_ASM_FUNCTION_PROLOGUE} and trampolines
4358 If this macro is not defined, by default the trampoline is allocated as
4359 a stack slot. This default is right for most machines. The exceptions
4360 are machines where it is impossible to execute instructions in the stack
4361 area. On such machines, you may have to implement a separate stack,
4362 using this macro in conjunction with @code{TARGET_ASM_FUNCTION_PROLOGUE}
4363 and @code{TARGET_ASM_FUNCTION_EPILOGUE}.
4365 @var{fp} points to a data structure, a @code{struct function}, which
4366 describes the compilation status of the immediate containing function of
4367 the function which the trampoline is for. Normally (when
4368 @code{ALLOCATE_TRAMPOLINE} is not defined), the stack slot for the
4369 trampoline is in the stack frame of this containing function. Other
4370 allocation strategies probably must do something analogous with this
4374 Implementing trampolines is difficult on many machines because they have
4375 separate instruction and data caches. Writing into a stack location
4376 fails to clear the memory in the instruction cache, so when the program
4377 jumps to that location, it executes the old contents.
4379 Here are two possible solutions. One is to clear the relevant parts of
4380 the instruction cache whenever a trampoline is set up. The other is to
4381 make all trampolines identical, by having them jump to a standard
4382 subroutine. The former technique makes trampoline execution faster; the
4383 latter makes initialization faster.
4385 To clear the instruction cache when a trampoline is initialized, define
4386 the following macros which describe the shape of the cache.
4389 @findex INSN_CACHE_SIZE
4390 @item INSN_CACHE_SIZE
4391 The total size in bytes of the cache.
4393 @findex INSN_CACHE_LINE_WIDTH
4394 @item INSN_CACHE_LINE_WIDTH
4395 The length in bytes of each cache line. The cache is divided into cache
4396 lines which are disjoint slots, each holding a contiguous chunk of data
4397 fetched from memory. Each time data is brought into the cache, an
4398 entire line is read at once. The data loaded into a cache line is
4399 always aligned on a boundary equal to the line size.
4401 @findex INSN_CACHE_DEPTH
4402 @item INSN_CACHE_DEPTH
4403 The number of alternative cache lines that can hold any particular memory
4407 Alternatively, if the machine has system calls or instructions to clear
4408 the instruction cache directly, you can define the following macro.
4411 @findex CLEAR_INSN_CACHE
4412 @item CLEAR_INSN_CACHE (@var{beg}, @var{end})
4413 If defined, expands to a C expression clearing the @emph{instruction
4414 cache} in the specified interval. If it is not defined, and the macro
4415 @code{INSN_CACHE_SIZE} is defined, some generic code is generated to clear the
4416 cache. The definition of this macro would typically be a series of
4417 @code{asm} statements. Both @var{beg} and @var{end} are both pointer
4421 To use a standard subroutine, define the following macro. In addition,
4422 you must make sure that the instructions in a trampoline fill an entire
4423 cache line with identical instructions, or else ensure that the
4424 beginning of the trampoline code is always aligned at the same point in
4425 its cache line. Look in @file{m68k.h} as a guide.
4428 @findex TRANSFER_FROM_TRAMPOLINE
4429 @item TRANSFER_FROM_TRAMPOLINE
4430 Define this macro if trampolines need a special subroutine to do their
4431 work. The macro should expand to a series of @code{asm} statements
4432 which will be compiled with GCC@. They go in a library function named
4433 @code{__transfer_from_trampoline}.
4435 If you need to avoid executing the ordinary prologue code of a compiled
4436 C function when you jump to the subroutine, you can do so by placing a
4437 special label of your own in the assembler code. Use one @code{asm}
4438 statement to generate an assembler label, and another to make the label
4439 global. Then trampolines can use that label to jump directly to your
4440 special assembler code.
4444 @section Implicit Calls to Library Routines
4445 @cindex library subroutine names
4446 @cindex @file{libgcc.a}
4448 @c prevent bad page break with this line
4449 Here is an explanation of implicit calls to library routines.
4452 @findex MULSI3_LIBCALL
4453 @item MULSI3_LIBCALL
4454 A C string constant giving the name of the function to call for
4455 multiplication of one signed full-word by another. If you do not
4456 define this macro, the default name is used, which is @code{__mulsi3},
4457 a function defined in @file{libgcc.a}.
4459 @findex DIVSI3_LIBCALL
4460 @item DIVSI3_LIBCALL
4461 A C string constant giving the name of the function to call for
4462 division of one signed full-word by another. If you do not define
4463 this macro, the default name is used, which is @code{__divsi3}, a
4464 function defined in @file{libgcc.a}.
4466 @findex UDIVSI3_LIBCALL
4467 @item UDIVSI3_LIBCALL
4468 A C string constant giving the name of the function to call for
4469 division of one unsigned full-word by another. If you do not define
4470 this macro, the default name is used, which is @code{__udivsi3}, a
4471 function defined in @file{libgcc.a}.
4473 @findex MODSI3_LIBCALL
4474 @item MODSI3_LIBCALL
4475 A C string constant giving the name of the function to call for the
4476 remainder in division of one signed full-word by another. If you do
4477 not define this macro, the default name is used, which is
4478 @code{__modsi3}, a function defined in @file{libgcc.a}.
4480 @findex UMODSI3_LIBCALL
4481 @item UMODSI3_LIBCALL
4482 A C string constant giving the name of the function to call for the
4483 remainder in division of one unsigned full-word by another. If you do
4484 not define this macro, the default name is used, which is
4485 @code{__umodsi3}, a function defined in @file{libgcc.a}.
4487 @findex MULDI3_LIBCALL
4488 @item MULDI3_LIBCALL
4489 A C string constant giving the name of the function to call for
4490 multiplication of one signed double-word by another. If you do not
4491 define this macro, the default name is used, which is @code{__muldi3},
4492 a function defined in @file{libgcc.a}.
4494 @findex DIVDI3_LIBCALL
4495 @item DIVDI3_LIBCALL
4496 A C string constant giving the name of the function to call for
4497 division of one signed double-word by another. If you do not define
4498 this macro, the default name is used, which is @code{__divdi3}, a
4499 function defined in @file{libgcc.a}.
4501 @findex UDIVDI3_LIBCALL
4502 @item UDIVDI3_LIBCALL
4503 A C string constant giving the name of the function to call for
4504 division of one unsigned full-word by another. If you do not define
4505 this macro, the default name is used, which is @code{__udivdi3}, a
4506 function defined in @file{libgcc.a}.
4508 @findex MODDI3_LIBCALL
4509 @item MODDI3_LIBCALL
4510 A C string constant giving the name of the function to call for the
4511 remainder in division of one signed double-word by another. If you do
4512 not define this macro, the default name is used, which is
4513 @code{__moddi3}, a function defined in @file{libgcc.a}.
4515 @findex UMODDI3_LIBCALL
4516 @item UMODDI3_LIBCALL
4517 A C string constant giving the name of the function to call for the
4518 remainder in division of one unsigned full-word by another. If you do
4519 not define this macro, the default name is used, which is
4520 @code{__umoddi3}, a function defined in @file{libgcc.a}.
4522 @findex INIT_TARGET_OPTABS
4523 @item INIT_TARGET_OPTABS
4524 Define this macro as a C statement that declares additional library
4525 routines renames existing ones. @code{init_optabs} calls this macro after
4526 initializing all the normal library routines.
4528 @findex FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
4529 @item FLOAT_LIB_COMPARE_RETURNS_BOOL
4530 Define this macro as a C statement that returns nonzero if a call to
4531 the floating point comparison library function will return a boolean
4532 value that indicates the result of the comparison. It should return
4533 zero if one of gcc's own libgcc functions is called.
4535 Most ports don't need to define this macro.
4538 @cindex @code{EDOM}, implicit usage
4540 The value of @code{EDOM} on the target machine, as a C integer constant
4541 expression. If you don't define this macro, GCC does not attempt to
4542 deposit the value of @code{EDOM} into @code{errno} directly. Look in
4543 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
4546 If you do not define @code{TARGET_EDOM}, then compiled code reports
4547 domain errors by calling the library function and letting it report the
4548 error. If mathematical functions on your system use @code{matherr} when
4549 there is an error, then you should leave @code{TARGET_EDOM} undefined so
4550 that @code{matherr} is used normally.
4552 @findex GEN_ERRNO_RTX
4553 @cindex @code{errno}, implicit usage
4555 Define this macro as a C expression to create an rtl expression that
4556 refers to the global ``variable'' @code{errno}. (On certain systems,
4557 @code{errno} may not actually be a variable.) If you don't define this
4558 macro, a reasonable default is used.
4560 @findex TARGET_MEM_FUNCTIONS
4561 @cindex @code{bcopy}, implicit usage
4562 @cindex @code{memcpy}, implicit usage
4563 @cindex @code{memmove}, implicit usage
4564 @cindex @code{bzero}, implicit usage
4565 @cindex @code{memset}, implicit usage
4566 @item TARGET_MEM_FUNCTIONS
4567 Define this macro if GCC should generate calls to the ISO C
4568 (and System V) library functions @code{memcpy}, @code{memmove} and
4569 @code{memset} rather than the BSD functions @code{bcopy} and @code{bzero}.
4571 @findex LIBGCC_NEEDS_DOUBLE
4572 @item LIBGCC_NEEDS_DOUBLE
4573 Define this macro if @code{float} arguments cannot be passed to library
4574 routines (so they must be converted to @code{double}). This macro
4575 affects both how library calls are generated and how the library
4576 routines in @file{libgcc.a} accept their arguments. It is useful on
4577 machines where floating and fixed point arguments are passed
4578 differently, such as the i860.
4580 @findex NEXT_OBJC_RUNTIME
4581 @item NEXT_OBJC_RUNTIME
4582 Define this macro to generate code for Objective-C message sending using
4583 the calling convention of the NeXT system. This calling convention
4584 involves passing the object, the selector and the method arguments all
4585 at once to the method-lookup library function.
4587 The default calling convention passes just the object and the selector
4588 to the lookup function, which returns a pointer to the method.
4591 @node Addressing Modes
4592 @section Addressing Modes
4593 @cindex addressing modes
4595 @c prevent bad page break with this line
4596 This is about addressing modes.
4599 @findex HAVE_PRE_INCREMENT
4600 @findex HAVE_PRE_DECREMENT
4601 @findex HAVE_POST_INCREMENT
4602 @findex HAVE_POST_DECREMENT
4603 @item HAVE_PRE_INCREMENT
4604 @itemx HAVE_PRE_DECREMENT
4605 @itemx HAVE_POST_INCREMENT
4606 @itemx HAVE_POST_DECREMENT
4607 A C expression that is nonzero if the machine supports pre-increment,
4608 pre-decrement, post-increment, or post-decrement addressing respectively.
4610 @findex HAVE_POST_MODIFY_DISP
4611 @findex HAVE_PRE_MODIFY_DISP
4612 @item HAVE_PRE_MODIFY_DISP
4613 @itemx HAVE_POST_MODIFY_DISP
4614 A C expression that is nonzero if the machine supports pre- or
4615 post-address side-effect generation involving constants other than
4616 the size of the memory operand.
4618 @findex HAVE_POST_MODIFY_REG
4619 @findex HAVE_PRE_MODIFY_REG
4620 @item HAVE_PRE_MODIFY_REG
4621 @itemx HAVE_POST_MODIFY_REG
4622 A C expression that is nonzero if the machine supports pre- or
4623 post-address side-effect generation involving a register displacement.
4625 @findex CONSTANT_ADDRESS_P
4626 @item CONSTANT_ADDRESS_P (@var{x})
4627 A C expression that is 1 if the RTX @var{x} is a constant which
4628 is a valid address. On most machines, this can be defined as
4629 @code{CONSTANT_P (@var{x})}, but a few machines are more restrictive
4630 in which constant addresses are supported.
4633 @code{CONSTANT_P} accepts integer-values expressions whose values are
4634 not explicitly known, such as @code{symbol_ref}, @code{label_ref}, and
4635 @code{high} expressions and @code{const} arithmetic expressions, in
4636 addition to @code{const_int} and @code{const_double} expressions.
4638 @findex MAX_REGS_PER_ADDRESS
4639 @item MAX_REGS_PER_ADDRESS
4640 A number, the maximum number of registers that can appear in a valid
4641 memory address. Note that it is up to you to specify a value equal to
4642 the maximum number that @code{GO_IF_LEGITIMATE_ADDRESS} would ever
4645 @findex GO_IF_LEGITIMATE_ADDRESS
4646 @item GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
4647 A C compound statement with a conditional @code{goto @var{label};}
4648 executed if @var{x} (an RTX) is a legitimate memory address on the
4649 target machine for a memory operand of mode @var{mode}.
4651 It usually pays to define several simpler macros to serve as
4652 subroutines for this one. Otherwise it may be too complicated to
4655 This macro must exist in two variants: a strict variant and a
4656 non-strict one. The strict variant is used in the reload pass. It
4657 must be defined so that any pseudo-register that has not been
4658 allocated a hard register is considered a memory reference. In
4659 contexts where some kind of register is required, a pseudo-register
4660 with no hard register must be rejected.
4662 The non-strict variant is used in other passes. It must be defined to
4663 accept all pseudo-registers in every context where some kind of
4664 register is required.
4666 @findex REG_OK_STRICT
4667 Compiler source files that want to use the strict variant of this
4668 macro define the macro @code{REG_OK_STRICT}. You should use an
4669 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant
4670 in that case and the non-strict variant otherwise.
4672 Subroutines to check for acceptable registers for various purposes (one
4673 for base registers, one for index registers, and so on) are typically
4674 among the subroutines used to define @code{GO_IF_LEGITIMATE_ADDRESS}.
4675 Then only these subroutine macros need have two variants; the higher
4676 levels of macros may be the same whether strict or not.
4678 Normally, constant addresses which are the sum of a @code{symbol_ref}
4679 and an integer are stored inside a @code{const} RTX to mark them as
4680 constant. Therefore, there is no need to recognize such sums
4681 specifically as legitimate addresses. Normally you would simply
4682 recognize any @code{const} as legitimate.
4684 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
4685 sums that are not marked with @code{const}. It assumes that a naked
4686 @code{plus} indicates indexing. If so, then you @emph{must} reject such
4687 naked constant sums as illegitimate addresses, so that none of them will
4688 be given to @code{PRINT_OPERAND_ADDRESS}.
4690 @cindex @code{ENCODE_SECTION_INFO} and address validation
4691 On some machines, whether a symbolic address is legitimate depends on
4692 the section that the address refers to. On these machines, define the
4693 macro @code{ENCODE_SECTION_INFO} to store the information into the
4694 @code{symbol_ref}, and then check for it here. When you see a
4695 @code{const}, you will have to look inside it to find the
4696 @code{symbol_ref} in order to determine the section. @xref{Assembler
4699 @findex saveable_obstack
4700 The best way to modify the name string is by adding text to the
4701 beginning, with suitable punctuation to prevent any ambiguity. Allocate
4702 the new name in @code{saveable_obstack}. You will have to modify
4703 @code{ASM_OUTPUT_LABELREF} to remove and decode the added text and
4704 output the name accordingly, and define @code{STRIP_NAME_ENCODING} to
4705 access the original name string.
4707 You can check the information stored here into the @code{symbol_ref} in
4708 the definitions of the macros @code{GO_IF_LEGITIMATE_ADDRESS} and
4709 @code{PRINT_OPERAND_ADDRESS}.
4711 @findex REG_OK_FOR_BASE_P
4712 @item REG_OK_FOR_BASE_P (@var{x})
4713 A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
4714 RTX) is valid for use as a base register. For hard registers, it
4715 should always accept those which the hardware permits and reject the
4716 others. Whether the macro accepts or rejects pseudo registers must be
4717 controlled by @code{REG_OK_STRICT} as described above. This usually
4718 requires two variant definitions, of which @code{REG_OK_STRICT}
4719 controls the one actually used.
4721 @findex REG_MODE_OK_FOR_BASE_P
4722 @item REG_MODE_OK_FOR_BASE_P (@var{x}, @var{mode})
4723 A C expression that is just like @code{REG_OK_FOR_BASE_P}, except that
4724 that expression may examine the mode of the memory reference in
4725 @var{mode}. You should define this macro if the mode of the memory
4726 reference affects whether a register may be used as a base register. If
4727 you define this macro, the compiler will use it instead of
4728 @code{REG_OK_FOR_BASE_P}.
4730 @findex REG_OK_FOR_INDEX_P
4731 @item REG_OK_FOR_INDEX_P (@var{x})
4732 A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
4733 RTX) is valid for use as an index register.
4735 The difference between an index register and a base register is that
4736 the index register may be scaled. If an address involves the sum of
4737 two registers, neither one of them scaled, then either one may be
4738 labeled the ``base'' and the other the ``index''; but whichever
4739 labeling is used must fit the machine's constraints of which registers
4740 may serve in each capacity. The compiler will try both labelings,
4741 looking for one that is valid, and will reload one or both registers
4742 only if neither labeling works.
4744 @findex FIND_BASE_TERM
4745 @item FIND_BASE_TERM (@var{x})
4746 A C expression to determine the base term of address @var{x}.
4747 This macro is used in only one place: `find_base_term' in alias.c.
4749 It is always safe for this macro to not be defined. It exists so
4750 that alias analysis can understand machine-dependent addresses.
4752 The typical use of this macro is to handle addresses containing
4753 a label_ref or symbol_ref within an UNSPEC@.
4755 @findex LEGITIMIZE_ADDRESS
4756 @item LEGITIMIZE_ADDRESS (@var{x}, @var{oldx}, @var{mode}, @var{win})
4757 A C compound statement that attempts to replace @var{x} with a valid
4758 memory address for an operand of mode @var{mode}. @var{win} will be a
4759 C statement label elsewhere in the code; the macro definition may use
4762 GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{win});
4766 to avoid further processing if the address has become legitimate.
4768 @findex break_out_memory_refs
4769 @var{x} will always be the result of a call to @code{break_out_memory_refs},
4770 and @var{oldx} will be the operand that was given to that function to produce
4773 The code generated by this macro should not alter the substructure of
4774 @var{x}. If it transforms @var{x} into a more legitimate form, it
4775 should assign @var{x} (which will always be a C variable) a new value.
4777 It is not necessary for this macro to come up with a legitimate
4778 address. The compiler has standard ways of doing so in all cases. In
4779 fact, it is safe for this macro to do nothing. But often a
4780 machine-dependent strategy can generate better code.
4782 @findex LEGITIMIZE_RELOAD_ADDRESS
4783 @item LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
4784 A C compound statement that attempts to replace @var{x}, which is an address
4785 that needs reloading, with a valid memory address for an operand of mode
4786 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
4787 It is not necessary to define this macro, but it might be useful for
4788 performance reasons.
4790 For example, on the i386, it is sometimes possible to use a single
4791 reload register instead of two by reloading a sum of two pseudo
4792 registers into a register. On the other hand, for number of RISC
4793 processors offsets are limited so that often an intermediate address
4794 needs to be generated in order to address a stack slot. By defining
4795 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
4796 generated for adjacent some stack slots can be made identical, and thus
4799 @emph{Note}: This macro should be used with caution. It is necessary
4800 to know something of how reload works in order to effectively use this,
4801 and it is quite easy to produce macros that build in too much knowledge
4802 of reload internals.
4804 @emph{Note}: This macro must be able to reload an address created by a
4805 previous invocation of this macro. If it fails to handle such addresses
4806 then the compiler may generate incorrect code or abort.
4809 The macro definition should use @code{push_reload} to indicate parts that
4810 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
4811 suitable to be passed unaltered to @code{push_reload}.
4813 The code generated by this macro must not alter the substructure of
4814 @var{x}. If it transforms @var{x} into a more legitimate form, it
4815 should assign @var{x} (which will always be a C variable) a new value.
4816 This also applies to parts that you change indirectly by calling
4819 @findex strict_memory_address_p
4820 The macro definition may use @code{strict_memory_address_p} to test if
4821 the address has become legitimate.
4824 If you want to change only a part of @var{x}, one standard way of doing
4825 this is to use @code{copy_rtx}. Note, however, that is unshares only a
4826 single level of rtl. Thus, if the part to be changed is not at the
4827 top level, you'll need to replace first the top level.
4828 It is not necessary for this macro to come up with a legitimate
4829 address; but often a machine-dependent strategy can generate better code.
4831 @findex GO_IF_MODE_DEPENDENT_ADDRESS
4832 @item GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
4833 A C statement or compound statement with a conditional @code{goto
4834 @var{label};} executed if memory address @var{x} (an RTX) can have
4835 different meanings depending on the machine mode of the memory
4836 reference it is used for or if the address is valid for some modes
4839 Autoincrement and autodecrement addresses typically have mode-dependent
4840 effects because the amount of the increment or decrement is the size
4841 of the operand being addressed. Some machines have other mode-dependent
4842 addresses. Many RISC machines have no mode-dependent addresses.
4844 You may assume that @var{addr} is a valid address for the machine.
4846 @findex LEGITIMATE_CONSTANT_P
4847 @item LEGITIMATE_CONSTANT_P (@var{x})
4848 A C expression that is nonzero if @var{x} is a legitimate constant for
4849 an immediate operand on the target machine. You can assume that
4850 @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
4851 @samp{1} is a suitable definition for this macro on machines where
4852 anything @code{CONSTANT_P} is valid.
4855 @node Condition Code
4856 @section Condition Code Status
4857 @cindex condition code status
4859 @c prevent bad page break with this line
4860 This describes the condition code status.
4863 The file @file{conditions.h} defines a variable @code{cc_status} to
4864 describe how the condition code was computed (in case the interpretation of
4865 the condition code depends on the instruction that it was set by). This
4866 variable contains the RTL expressions on which the condition code is
4867 currently based, and several standard flags.
4869 Sometimes additional machine-specific flags must be defined in the machine
4870 description header file. It can also add additional machine-specific
4871 information by defining @code{CC_STATUS_MDEP}.
4874 @findex CC_STATUS_MDEP
4875 @item CC_STATUS_MDEP
4876 C code for a data type which is used for declaring the @code{mdep}
4877 component of @code{cc_status}. It defaults to @code{int}.
4879 This macro is not used on machines that do not use @code{cc0}.
4881 @findex CC_STATUS_MDEP_INIT
4882 @item CC_STATUS_MDEP_INIT
4883 A C expression to initialize the @code{mdep} field to ``empty''.
4884 The default definition does nothing, since most machines don't use
4885 the field anyway. If you want to use the field, you should probably
4886 define this macro to initialize it.
4888 This macro is not used on machines that do not use @code{cc0}.
4890 @findex NOTICE_UPDATE_CC
4891 @item NOTICE_UPDATE_CC (@var{exp}, @var{insn})
4892 A C compound statement to set the components of @code{cc_status}
4893 appropriately for an insn @var{insn} whose body is @var{exp}. It is
4894 this macro's responsibility to recognize insns that set the condition
4895 code as a byproduct of other activity as well as those that explicitly
4898 This macro is not used on machines that do not use @code{cc0}.
4900 If there are insns that do not set the condition code but do alter
4901 other machine registers, this macro must check to see whether they
4902 invalidate the expressions that the condition code is recorded as
4903 reflecting. For example, on the 68000, insns that store in address
4904 registers do not set the condition code, which means that usually
4905 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
4906 insns. But suppose that the previous insn set the condition code
4907 based on location @samp{a4@@(102)} and the current insn stores a new
4908 value in @samp{a4}. Although the condition code is not changed by
4909 this, it will no longer be true that it reflects the contents of
4910 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
4911 @code{cc_status} in this case to say that nothing is known about the
4912 condition code value.
4914 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
4915 with the results of peephole optimization: insns whose patterns are
4916 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
4917 constants which are just the operands. The RTL structure of these
4918 insns is not sufficient to indicate what the insns actually do. What
4919 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
4920 @code{CC_STATUS_INIT}.
4922 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
4923 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
4924 @samp{cc}. This avoids having detailed information about patterns in
4925 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
4927 @findex EXTRA_CC_MODES
4928 @item EXTRA_CC_MODES
4929 A list of additional modes for condition code values in registers
4930 (@pxref{Jump Patterns}). This macro should expand to a sequence of
4931 calls of the macro @code{CC} separated by white space. @code{CC} takes
4932 two arguments. The first is the enumeration name of the mode, which
4933 should begin with @samp{CC} and end with @samp{mode}. The second is a C
4934 string giving the printable name of the mode; it should be the same as
4935 the first argument, but with the trailing @samp{mode} removed.
4937 You should only define this macro if additional modes are required.
4939 A sample definition of @code{EXTRA_CC_MODES} is:
4941 #define EXTRA_CC_MODES \
4942 CC(CC_NOOVmode, "CC_NOOV") \
4943 CC(CCFPmode, "CCFP") \
4944 CC(CCFPEmode, "CCFPE")
4947 @findex SELECT_CC_MODE
4948 @item SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
4949 Returns a mode from class @code{MODE_CC} to be used when comparison
4950 operation code @var{op} is applied to rtx @var{x} and @var{y}. For
4951 example, on the Sparc, @code{SELECT_CC_MODE} is defined as (see
4952 @pxref{Jump Patterns} for a description of the reason for this
4956 #define SELECT_CC_MODE(OP,X,Y) \
4957 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
4958 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
4959 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
4960 || GET_CODE (X) == NEG) \
4961 ? CC_NOOVmode : CCmode))
4964 You need not define this macro if @code{EXTRA_CC_MODES} is not defined.
4966 @findex CANONICALIZE_COMPARISON
4967 @item CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
4968 On some machines not all possible comparisons are defined, but you can
4969 convert an invalid comparison into a valid one. For example, the Alpha
4970 does not have a @code{GT} comparison, but you can use an @code{LT}
4971 comparison instead and swap the order of the operands.
4973 On such machines, define this macro to be a C statement to do any
4974 required conversions. @var{code} is the initial comparison code
4975 and @var{op0} and @var{op1} are the left and right operands of the
4976 comparison, respectively. You should modify @var{code}, @var{op0}, and
4977 @var{op1} as required.
4979 GCC will not assume that the comparison resulting from this macro is
4980 valid but will see if the resulting insn matches a pattern in the
4983 You need not define this macro if it would never change the comparison
4986 @findex REVERSIBLE_CC_MODE
4987 @item REVERSIBLE_CC_MODE (@var{mode})
4988 A C expression whose value is one if it is always safe to reverse a
4989 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
4990 can ever return @var{mode} for a floating-point inequality comparison,
4991 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
4993 You need not define this macro if it would always returns zero or if the
4994 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
4995 For example, here is the definition used on the Sparc, where floating-point
4996 inequality comparisons are always given @code{CCFPEmode}:
4999 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
5002 @findex REVERSE_CONDITION (@var{code}, @var{mode})
5003 A C expression whose value is reversed condition code of the @var{code} for
5004 comparison done in CC_MODE @var{mode}. The macro is used only in case
5005 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
5006 machine has some non-standard way how to reverse certain conditionals. For
5007 instance in case all floating point conditions are non-trapping, compiler may
5008 freely convert unordered compares to ordered one. Then definition may look
5012 #define REVERSE_CONDITION(CODE, MODE) \
5013 ((MODE) != CCFPmode ? reverse_condition (CODE) \
5014 : reverse_condition_maybe_unordered (CODE))
5017 @findex REVERSE_CONDEXEC_PREDICATES_P
5018 @item REVERSE_CONDEXEC_PREDICATES_P (@var{code1}, @var{code2})
5019 A C expression that returns true if the conditional execution predicate
5020 @var{code1} is the inverse of @var{code2} and vice versa. Define this to
5021 return 0 if the target has conditional execution predicates that cannot be
5022 reversed safely. If no expansion is specified, this macro is defined as
5026 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
5027 ((x) == reverse_condition (y))
5033 @section Describing Relative Costs of Operations
5034 @cindex costs of instructions
5035 @cindex relative costs
5036 @cindex speed of instructions
5038 These macros let you describe the relative speed of various operations
5039 on the target machine.
5043 @item CONST_COSTS (@var{x}, @var{code}, @var{outer_code})
5044 A part of a C @code{switch} statement that describes the relative costs
5045 of constant RTL expressions. It must contain @code{case} labels for
5046 expression codes @code{const_int}, @code{const}, @code{symbol_ref},
5047 @code{label_ref} and @code{const_double}. Each case must ultimately
5048 reach a @code{return} statement to return the relative cost of the use
5049 of that kind of constant value in an expression. The cost may depend on
5050 the precise value of the constant, which is available for examination in
5051 @var{x}, and the rtx code of the expression in which it is contained,
5052 found in @var{outer_code}.
5054 @var{code} is the expression code---redundant, since it can be
5055 obtained with @code{GET_CODE (@var{x})}.
5058 @findex COSTS_N_INSNS
5059 @item RTX_COSTS (@var{x}, @var{code}, @var{outer_code})
5060 Like @code{CONST_COSTS} but applies to nonconstant RTL expressions.
5061 This can be used, for example, to indicate how costly a multiply
5062 instruction is. In writing this macro, you can use the construct
5063 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
5064 instructions. @var{outer_code} is the code of the expression in which
5065 @var{x} is contained.
5067 This macro is optional; do not define it if the default cost assumptions
5068 are adequate for the target machine.
5070 @findex DEFAULT_RTX_COSTS
5071 @item DEFAULT_RTX_COSTS (@var{x}, @var{code}, @var{outer_code})
5072 This macro, if defined, is called for any case not handled by the
5073 @code{RTX_COSTS} or @code{CONST_COSTS} macros. This eliminates the need
5074 to put case labels into the macro, but the code, or any functions it
5075 calls, must assume that the RTL in @var{x} could be of any type that has
5076 not already been handled. The arguments are the same as for
5077 @code{RTX_COSTS}, and the macro should execute a return statement giving
5078 the cost of any RTL expressions that it can handle. The default cost
5079 calculation is used for any RTL for which this macro does not return a
5082 This macro is optional; do not define it if the default cost assumptions
5083 are adequate for the target machine.
5085 @findex ADDRESS_COST
5086 @item ADDRESS_COST (@var{address})
5087 An expression giving the cost of an addressing mode that contains
5088 @var{address}. If not defined, the cost is computed from
5089 the @var{address} expression and the @code{CONST_COSTS} values.
5091 For most CISC machines, the default cost is a good approximation of the
5092 true cost of the addressing mode. However, on RISC machines, all
5093 instructions normally have the same length and execution time. Hence
5094 all addresses will have equal costs.
5096 In cases where more than one form of an address is known, the form with
5097 the lowest cost will be used. If multiple forms have the same, lowest,
5098 cost, the one that is the most complex will be used.
5100 For example, suppose an address that is equal to the sum of a register
5101 and a constant is used twice in the same basic block. When this macro
5102 is not defined, the address will be computed in a register and memory
5103 references will be indirect through that register. On machines where
5104 the cost of the addressing mode containing the sum is no higher than
5105 that of a simple indirect reference, this will produce an additional
5106 instruction and possibly require an additional register. Proper
5107 specification of this macro eliminates this overhead for such machines.
5109 Similar use of this macro is made in strength reduction of loops.
5111 @var{address} need not be valid as an address. In such a case, the cost
5112 is not relevant and can be any value; invalid addresses need not be
5113 assigned a different cost.
5115 On machines where an address involving more than one register is as
5116 cheap as an address computation involving only one register, defining
5117 @code{ADDRESS_COST} to reflect this can cause two registers to be live
5118 over a region of code where only one would have been if
5119 @code{ADDRESS_COST} were not defined in that manner. This effect should
5120 be considered in the definition of this macro. Equivalent costs should
5121 probably only be given to addresses with different numbers of registers
5122 on machines with lots of registers.
5124 This macro will normally either not be defined or be defined as a
5127 @findex REGISTER_MOVE_COST
5128 @item REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
5129 A C expression for the cost of moving data of mode @var{mode} from a
5130 register in class @var{from} to one in class @var{to}. The classes are
5131 expressed using the enumeration values such as @code{GENERAL_REGS}. A
5132 value of 2 is the default; other values are interpreted relative to
5135 It is not required that the cost always equal 2 when @var{from} is the
5136 same as @var{to}; on some machines it is expensive to move between
5137 registers if they are not general registers.
5139 If reload sees an insn consisting of a single @code{set} between two
5140 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
5141 classes returns a value of 2, reload does not check to ensure that the
5142 constraints of the insn are met. Setting a cost of other than 2 will
5143 allow reload to verify that the constraints are met. You should do this
5144 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
5146 @findex MEMORY_MOVE_COST
5147 @item MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
5148 A C expression for the cost of moving data of mode @var{mode} between a
5149 register of class @var{class} and memory; @var{in} is zero if the value
5150 is to be written to memory, nonzero if it is to be read in. This cost
5151 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
5152 registers and memory is more expensive than between two registers, you
5153 should define this macro to express the relative cost.
5155 If you do not define this macro, GCC uses a default cost of 4 plus
5156 the cost of copying via a secondary reload register, if one is
5157 needed. If your machine requires a secondary reload register to copy
5158 between memory and a register of @var{class} but the reload mechanism is
5159 more complex than copying via an intermediate, define this macro to
5160 reflect the actual cost of the move.
5162 GCC defines the function @code{memory_move_secondary_cost} if
5163 secondary reloads are needed. It computes the costs due to copying via
5164 a secondary register. If your machine copies from memory using a
5165 secondary register in the conventional way but the default base value of
5166 4 is not correct for your machine, define this macro to add some other
5167 value to the result of that function. The arguments to that function
5168 are the same as to this macro.
5172 A C expression for the cost of a branch instruction. A value of 1 is
5173 the default; other values are interpreted relative to that.
5176 Here are additional macros which do not specify precise relative costs,
5177 but only that certain actions are more expensive than GCC would
5181 @findex SLOW_BYTE_ACCESS
5182 @item SLOW_BYTE_ACCESS
5183 Define this macro as a C expression which is nonzero if accessing less
5184 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
5185 faster than accessing a word of memory, i.e., if such access
5186 require more than one instruction or if there is no difference in cost
5187 between byte and (aligned) word loads.
5189 When this macro is not defined, the compiler will access a field by
5190 finding the smallest containing object; when it is defined, a fullword
5191 load will be used if alignment permits. Unless bytes accesses are
5192 faster than word accesses, using word accesses is preferable since it
5193 may eliminate subsequent memory access if subsequent accesses occur to
5194 other fields in the same word of the structure, but to different bytes.
5196 @findex SLOW_ZERO_EXTEND
5197 @item SLOW_ZERO_EXTEND
5198 Define this macro if zero-extension (of a @code{char} or @code{short}
5199 to an @code{int}) can be done faster if the destination is a register
5200 that is known to be zero.
5202 If you define this macro, you must have instruction patterns that
5203 recognize RTL structures like this:
5206 (set (strict_low_part (subreg:QI (reg:SI @dots{}) 0)) @dots{})
5210 and likewise for @code{HImode}.
5212 @findex SLOW_UNALIGNED_ACCESS
5213 @item SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
5214 Define this macro to be the value 1 if memory accesses described by the
5215 @var{mode} and @var{alignment} parameters have a cost many times greater
5216 than aligned accesses, for example if they are emulated in a trap
5219 When this macro is nonzero, the compiler will act as if
5220 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
5221 moves. This can cause significantly more instructions to be produced.
5222 Therefore, do not set this macro nonzero if unaligned accesses only add a
5223 cycle or two to the time for a memory access.
5225 If the value of this macro is always zero, it need not be defined. If
5226 this macro is defined, it should produce a nonzero value when
5227 @code{STRICT_ALIGNMENT} is nonzero.
5229 @findex DONT_REDUCE_ADDR
5230 @item DONT_REDUCE_ADDR
5231 Define this macro to inhibit strength reduction of memory addresses.
5232 (On some machines, such strength reduction seems to do harm rather
5237 The threshold of number of scalar memory-to-memory move insns, @emph{below}
5238 which a sequence of insns should be generated instead of a
5239 string move insn or a library call. Increasing the value will always
5240 make code faster, but eventually incurs high cost in increased code size.
5242 Note that on machines where the corresponding move insn is a
5243 @code{define_expand} that emits a sequence of insns, this macro counts
5244 the number of such sequences.
5246 If you don't define this, a reasonable default is used.
5248 @findex MOVE_BY_PIECES_P
5249 @item MOVE_BY_PIECES_P (@var{size}, @var{alignment})
5250 A C expression used to determine whether @code{move_by_pieces} will be used to
5251 copy a chunk of memory, or whether some other block move mechanism
5252 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
5253 than @code{MOVE_RATIO}.
5255 @findex MOVE_MAX_PIECES
5256 @item MOVE_MAX_PIECES
5257 A C expression used by @code{move_by_pieces} to determine the largest unit
5258 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
5260 @findex USE_LOAD_POST_INCREMENT
5261 @item USE_LOAD_POST_INCREMENT (@var{mode})
5262 A C expression used to determine whether a load postincrement is a good
5263 thing to use for a given mode. Defaults to the value of
5264 @code{HAVE_POST_INCREMENT}.
5266 @findex USE_LOAD_POST_DECREMENT
5267 @item USE_LOAD_POST_DECREMENT (@var{mode})
5268 A C expression used to determine whether a load postdecrement is a good
5269 thing to use for a given mode. Defaults to the value of
5270 @code{HAVE_POST_DECREMENT}.
5272 @findex USE_LOAD_PRE_INCREMENT
5273 @item USE_LOAD_PRE_INCREMENT (@var{mode})
5274 A C expression used to determine whether a load preincrement is a good
5275 thing to use for a given mode. Defaults to the value of
5276 @code{HAVE_PRE_INCREMENT}.
5278 @findex USE_LOAD_PRE_DECREMENT
5279 @item USE_LOAD_PRE_DECREMENT (@var{mode})
5280 A C expression used to determine whether a load predecrement is a good
5281 thing to use for a given mode. Defaults to the value of
5282 @code{HAVE_PRE_DECREMENT}.
5284 @findex USE_STORE_POST_INCREMENT
5285 @item USE_STORE_POST_INCREMENT (@var{mode})
5286 A C expression used to determine whether a store postincrement is a good
5287 thing to use for a given mode. Defaults to the value of
5288 @code{HAVE_POST_INCREMENT}.
5290 @findex USE_STORE_POST_DECREMENT
5291 @item USE_STORE_POST_DECREMENT (@var{mode})
5292 A C expression used to determine whether a store postdecrement is a good
5293 thing to use for a given mode. Defaults to the value of
5294 @code{HAVE_POST_DECREMENT}.
5296 @findex USE_STORE_PRE_INCREMENT
5297 @item USE_STORE_PRE_INCREMENT (@var{mode})
5298 This macro is used to determine whether a store preincrement is a good
5299 thing to use for a given mode. Defaults to the value of
5300 @code{HAVE_PRE_INCREMENT}.
5302 @findex USE_STORE_PRE_DECREMENT
5303 @item USE_STORE_PRE_DECREMENT (@var{mode})
5304 This macro is used to determine whether a store predecrement is a good
5305 thing to use for a given mode. Defaults to the value of
5306 @code{HAVE_PRE_DECREMENT}.
5308 @findex NO_FUNCTION_CSE
5309 @item NO_FUNCTION_CSE
5310 Define this macro if it is as good or better to call a constant
5311 function address than to call an address kept in a register.
5313 @findex NO_RECURSIVE_FUNCTION_CSE
5314 @item NO_RECURSIVE_FUNCTION_CSE
5315 Define this macro if it is as good or better for a function to call
5316 itself with an explicit address than to call an address kept in a
5321 @section Adjusting the Instruction Scheduler
5323 The instruction scheduler may need a fair amount of machine-specific
5324 adjustment in order to produce good code. GCC provides several target
5325 hooks for this purpose. It is usually enough to define just a few of
5326 them: try the first ones in this list first.
5328 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
5329 This hook returns the maximum number of instructions that can ever issue
5330 at the same time on the target machine. The default is one. This value
5331 must be constant over the entire compilation. If you need it to vary
5332 depending on what the instructions are, you must use
5333 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
5336 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more})
5337 This hook is executed by the scheduler after it has scheduled an insn
5338 from the ready list. It should return the number of insns which can
5339 still be issued in the current cycle. Normally this is
5340 @samp{@w{@var{more} - 1}}. You should define this hook if some insns
5341 take more machine resources than others, so that fewer insns can follow
5342 them in the same cycle. @var{file} is either a null pointer, or a stdio
5343 stream to write any debug output to. @var{verbose} is the verbose level
5344 provided by @option{-fsched-verbose-@var{n}}. @var{insn} is the
5345 instruction that was scheduled.
5348 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost})
5349 This function corrects the value of @var{cost} based on the relationship
5350 between @var{insn} and @var{dep_insn} through the dependence @var{link}.
5351 It should return the new value. The default is to make no adjustment to
5352 @var{cost}. This can be used for example to specify to the scheduler
5353 that an output- or anti-dependence does not incur the same cost as a
5357 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority})
5358 This hook adjusts the integer scheduling priority @var{priority} of
5359 @var{insn}. It should return the new priority. Reduce the priority to
5360 execute @var{insn} earlier, increase the priority to execute @var{insn}
5361 later. Do not define this hook if you do not need to adjust the
5362 scheduling priorities of insns.
5365 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
5366 This hook is executed by the scheduler after it has scheduled the ready
5367 list, to allow the machine description to reorder it (for example to
5368 combine two small instructions together on @samp{VLIW} machines).
5369 @var{file} is either a null pointer, or a stdio stream to write any
5370 debug output to. @var{verbose} is the verbose level provided by
5371 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
5372 list of instructions that are ready to be scheduled. @var{n_readyp} is
5373 a pointer to the number of elements in the ready list. The scheduler
5374 reads the ready list in reverse order, starting with
5375 @var{ready}[@var{*n_readyp}-1] and going to @var{ready}[0]. @var{clock}
5376 is the timer tick of the scheduler. You may modify the ready list and
5377 the number of ready insns. The return value is the number of insns that
5378 can issue this cycle; normally this is just @code{issue_rate}. See also
5379 @samp{TARGET_SCHED_REORDER2}.
5382 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_ready}, @var{clock})
5383 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
5384 function is called whenever the scheduler starts a new cycle. This one
5385 is called once per iteration over a cycle, immediately after
5386 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
5387 return the number of insns to be scheduled in the same cycle. Defining
5388 this hook can be useful if there are frequent situations where
5389 scheduling one insn causes other insns to become ready in the same
5390 cycle. These other insns can then be taken into account properly.
5393 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
5394 This hook is executed by the scheduler at the beginning of each block of
5395 instructions that are to be scheduled. @var{file} is either a null
5396 pointer, or a stdio stream to write any debug output to. @var{verbose}
5397 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
5398 @var{max_ready} is the maximum number of insns in the current scheduling
5399 region that can be live at the same time. This can be used to allocate
5400 scratch space if it is needed, e.g. by @samp{TARGET_SCHED_REORDER}.
5403 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
5404 This hook is executed by the scheduler at the end of each block of
5405 instructions that are to be scheduled. It can be used to perform
5406 cleanup of any actions done by the other scheduling hooks. @var{file}
5407 is either a null pointer, or a stdio stream to write any debug output
5408 to. @var{verbose} is the verbose level provided by
5409 @option{-fsched-verbose-@var{n}}.
5412 @deftypefn {Target Hook} rtx TARGET_SCHED_CYCLE_DISPLAY (int @var{clock}, rtx @var{last})
5413 This hook is called in verbose mode only, at the beginning of each pass
5414 over a basic block. It should insert an insn into the chain after
5415 @var{last}, which has no effect, but records the value @var{clock} in
5416 RTL dumps and assembly output. Define this hook only if you need this
5417 level of detail about what the scheduler is doing.
5421 @section Dividing the Output into Sections (Texts, Data, @dots{})
5422 @c the above section title is WAY too long. maybe cut the part between
5423 @c the (...)? --mew 10feb93
5425 An object file is divided into sections containing different types of
5426 data. In the most common case, there are three sections: the @dfn{text
5427 section}, which holds instructions and read-only data; the @dfn{data
5428 section}, which holds initialized writable data; and the @dfn{bss
5429 section}, which holds uninitialized data. Some systems have other kinds
5432 The compiler must tell the assembler when to switch sections. These
5433 macros control what commands to output to tell the assembler this. You
5434 can also define additional sections.
5437 @findex TEXT_SECTION_ASM_OP
5438 @item TEXT_SECTION_ASM_OP
5439 A C expression whose value is a string, including spacing, containing the
5440 assembler operation that should precede instructions and read-only data.
5441 Normally @code{"\t.text"} is right.
5443 @findex TEXT_SECTION
5445 A C statement that switches to the default section containing instructions.
5446 Normally this is not needed, as simply defining @code{TEXT_SECTION_ASM_OP}
5447 is enough. The MIPS port uses this to sort all functions after all data
5450 @findex DATA_SECTION_ASM_OP
5451 @item DATA_SECTION_ASM_OP
5452 A C expression whose value is a string, including spacing, containing the
5453 assembler operation to identify the following data as writable initialized
5454 data. Normally @code{"\t.data"} is right.
5456 @findex SHARED_SECTION_ASM_OP
5457 @item SHARED_SECTION_ASM_OP
5458 If defined, a C expression whose value is a string, including spacing,
5459 containing the assembler operation to identify the following data as
5460 shared data. If not defined, @code{DATA_SECTION_ASM_OP} will be used.
5462 @findex BSS_SECTION_ASM_OP
5463 @item BSS_SECTION_ASM_OP
5464 If defined, a C expression whose value is a string, including spacing,
5465 containing the assembler operation to identify the following data as
5466 uninitialized global data. If not defined, and neither
5467 @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined,
5468 uninitialized global data will be output in the data section if
5469 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
5472 @findex SHARED_BSS_SECTION_ASM_OP
5473 @item SHARED_BSS_SECTION_ASM_OP
5474 If defined, a C expression whose value is a string, including spacing,
5475 containing the assembler operation to identify the following data as
5476 uninitialized global shared data. If not defined, and
5477 @code{BSS_SECTION_ASM_OP} is, the latter will be used.
5479 @findex INIT_SECTION_ASM_OP
5480 @item INIT_SECTION_ASM_OP
5481 If defined, a C expression whose value is a string, including spacing,
5482 containing the assembler operation to identify the following data as
5483 initialization code. If not defined, GCC will assume such a section does
5486 @findex FINI_SECTION_ASM_OP
5487 @item FINI_SECTION_ASM_OP
5488 If defined, a C expression whose value is a string, including spacing,
5489 containing the assembler operation to identify the following data as
5490 finalization code. If not defined, GCC will assume such a section does
5493 @findex CRT_CALL_STATIC_FUNCTION
5494 @item CRT_CALL_STATIC_FUNCTION
5495 If defined, a C statement that calls the function named as the sole
5496 argument of this macro. This is used in @file{crtstuff.c} if
5497 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls to
5498 initialization and finalization functions from the init and fini
5499 sections. By default, this macro is a simple function call. Some
5500 ports need hand-crafted assembly code to avoid dependencies on
5501 registers initialized in the function prologue or to ensure that
5502 constant pools don't end up too far way in the text section.
5504 @findex EXTRA_SECTIONS
5507 @item EXTRA_SECTIONS
5508 A list of names for sections other than the standard two, which are
5509 @code{in_text} and @code{in_data}. You need not define this macro
5510 on a system with no other sections (that GCC needs to use).
5512 @findex EXTRA_SECTION_FUNCTIONS
5513 @findex text_section
5514 @findex data_section
5515 @item EXTRA_SECTION_FUNCTIONS
5516 One or more functions to be defined in @file{varasm.c}. These
5517 functions should do jobs analogous to those of @code{text_section} and
5518 @code{data_section}, for your additional sections. Do not define this
5519 macro if you do not define @code{EXTRA_SECTIONS}.
5521 @findex READONLY_DATA_SECTION
5522 @item READONLY_DATA_SECTION
5523 On most machines, read-only variables, constants, and jump tables are
5524 placed in the text section. If this is not the case on your machine,
5525 this macro should be defined to be the name of a function (either
5526 @code{data_section} or a function defined in @code{EXTRA_SECTIONS}) that
5527 switches to the section to be used for read-only items.
5529 If these items should be placed in the text section, this macro should
5532 @findex SELECT_SECTION
5533 @item SELECT_SECTION (@var{exp}, @var{reloc}, @var{align})
5534 A C statement or statements to switch to the appropriate section for
5535 output of @var{exp}. You can assume that @var{exp} is either a
5536 @code{VAR_DECL} node or a constant of some sort. @var{reloc}
5537 indicates whether the initial value of @var{exp} requires link-time
5538 relocations. Bit 1 is set when variable contains local relocations
5539 only, while bit 2 is set for global relocations.
5540 Select the section by calling @code{text_section} or one
5541 of the alternatives for other sections. @var{align} is the constant
5544 Do not define this macro if you put all read-only variables and
5545 constants in the read-only data section (usually the text section).
5547 @findex SELECT_RTX_SECTION
5548 @item SELECT_RTX_SECTION (@var{mode}, @var{rtx}, @var{align})
5549 A C statement or statements to switch to the appropriate section for
5550 output of @var{rtx} in mode @var{mode}. You can assume that @var{rtx}
5551 is some kind of constant in RTL@. The argument @var{mode} is redundant
5552 except in the case of a @code{const_int} rtx. Select the section by
5553 calling @code{text_section} or one of the alternatives for other
5554 sections. @var{align} is the constant alignment in bits.
5556 Do not define this macro if you put all constants in the read-only
5559 @findex JUMP_TABLES_IN_TEXT_SECTION
5560 @item JUMP_TABLES_IN_TEXT_SECTION
5561 Define this macro to be an expression with a nonzero value if jump
5562 tables (for @code{tablejump} insns) should be output in the text
5563 section, along with the assembler instructions. Otherwise, the
5564 readonly data section is used.
5566 This macro is irrelevant if there is no separate readonly data section.
5568 @findex ENCODE_SECTION_INFO
5569 @item ENCODE_SECTION_INFO (@var{decl})
5570 Define this macro if references to a symbol or a constant must be
5571 treated differently depending on something about the variable or
5572 function named by the symbol (such as what section it is in).
5574 The macro definition, if any, is executed under two circumstances. One
5575 is immediately after the rtl for @var{decl} that represents a variable
5576 or a function has been created and stored in @code{DECL_RTL
5577 (@var{decl})}. The value of the rtl will be a @code{mem} whose address
5578 is a @code{symbol_ref}. The other is immediately after the rtl for
5579 @var{decl} that represents a constant has been created and stored in
5580 @code{TREE_CST_RTL (@var{decl})}. The macro is called once for each
5581 distinct constant in a source file.
5583 @cindex @code{SYMBOL_REF_FLAG}, in @code{ENCODE_SECTION_INFO}
5584 The usual thing for this macro to do is to record a flag in the
5585 @code{symbol_ref} (such as @code{SYMBOL_REF_FLAG}) or to store a
5586 modified name string in the @code{symbol_ref} (if one bit is not enough
5589 @findex STRIP_NAME_ENCODING
5590 @item STRIP_NAME_ENCODING (@var{var}, @var{sym_name})
5591 Decode @var{sym_name} and store the real name part in @var{var}, sans
5592 the characters that encode section info. Define this macro if
5593 @code{ENCODE_SECTION_INFO} alters the symbol's name string.
5595 @findex UNIQUE_SECTION
5596 @item UNIQUE_SECTION (@var{decl}, @var{reloc})
5597 A C statement to build up a unique section name, expressed as a
5598 @code{STRING_CST} node, and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
5599 @var{reloc} indicates whether the initial value of @var{exp} requires
5600 link-time relocations. If you do not define this macro, GCC will use
5601 the symbol name prefixed by @samp{.} as the section name. Note - this
5602 macro can now be called for uninitialized data items as well as
5603 initialized data and functions.
5607 @section Position Independent Code
5608 @cindex position independent code
5611 This section describes macros that help implement generation of position
5612 independent code. Simply defining these macros is not enough to
5613 generate valid PIC; you must also add support to the macros
5614 @code{GO_IF_LEGITIMATE_ADDRESS} and @code{PRINT_OPERAND_ADDRESS}, as
5615 well as @code{LEGITIMIZE_ADDRESS}. You must modify the definition of
5616 @samp{movsi} to do something appropriate when the source operand
5617 contains a symbolic address. You may also need to alter the handling of
5618 switch statements so that they use relative addresses.
5619 @c i rearranged the order of the macros above to try to force one of
5620 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
5623 @findex PIC_OFFSET_TABLE_REGNUM
5624 @item PIC_OFFSET_TABLE_REGNUM
5625 The register number of the register used to address a table of static
5626 data addresses in memory. In some cases this register is defined by a
5627 processor's ``application binary interface'' (ABI)@. When this macro
5628 is defined, RTL is generated for this register once, as with the stack
5629 pointer and frame pointer registers. If this macro is not defined, it
5630 is up to the machine-dependent files to allocate such a register (if
5633 @findex PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
5634 @item PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
5635 Define this macro if the register defined by
5636 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define
5637 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
5639 @findex FINALIZE_PIC
5641 By generating position-independent code, when two different programs (A
5642 and B) share a common library (libC.a), the text of the library can be
5643 shared whether or not the library is linked at the same address for both
5644 programs. In some of these environments, position-independent code
5645 requires not only the use of different addressing modes, but also
5646 special code to enable the use of these addressing modes.
5648 The @code{FINALIZE_PIC} macro serves as a hook to emit these special
5649 codes once the function is being compiled into assembly code, but not
5650 before. (It is not done before, because in the case of compiling an
5651 inline function, it would lead to multiple PIC prologues being
5652 included in functions which used inline functions and were compiled to
5655 @findex LEGITIMATE_PIC_OPERAND_P
5656 @item LEGITIMATE_PIC_OPERAND_P (@var{x})
5657 A C expression that is nonzero if @var{x} is a legitimate immediate
5658 operand on the target machine when generating position independent code.
5659 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
5660 check this. You can also assume @var{flag_pic} is true, so you need not
5661 check it either. You need not define this macro if all constants
5662 (including @code{SYMBOL_REF}) can be immediate operands when generating
5663 position independent code.
5666 @node Assembler Format
5667 @section Defining the Output Assembler Language
5669 This section describes macros whose principal purpose is to describe how
5670 to write instructions in assembler language---rather than what the
5674 * File Framework:: Structural information for the assembler file.
5675 * Data Output:: Output of constants (numbers, strings, addresses).
5676 * Uninitialized Data:: Output of uninitialized variables.
5677 * Label Output:: Output and generation of labels.
5678 * Initialization:: General principles of initialization
5679 and termination routines.
5680 * Macros for Initialization::
5681 Specific macros that control the handling of
5682 initialization and termination routines.
5683 * Instruction Output:: Output of actual instructions.
5684 * Dispatch Tables:: Output of jump tables.
5685 * Exception Region Output:: Output of exception region code.
5686 * Alignment Output:: Pseudo ops for alignment and skipping data.
5689 @node File Framework
5690 @subsection The Overall Framework of an Assembler File
5691 @cindex assembler format
5692 @cindex output of assembler code
5694 @c prevent bad page break with this line
5695 This describes the overall framework of an assembler file.
5698 @findex ASM_FILE_START
5699 @item ASM_FILE_START (@var{stream})
5700 A C expression which outputs to the stdio stream @var{stream}
5701 some appropriate text to go at the start of an assembler file.
5703 Normally this macro is defined to output a line containing
5704 @samp{#NO_APP}, which is a comment that has no effect on most
5705 assemblers but tells the GNU assembler that it can save time by not
5706 checking for certain assembler constructs.
5708 On systems that use SDB, it is necessary to output certain commands;
5709 see @file{attasm.h}.
5711 @findex ASM_FILE_END
5712 @item ASM_FILE_END (@var{stream})
5713 A C expression which outputs to the stdio stream @var{stream}
5714 some appropriate text to go at the end of an assembler file.
5716 If this macro is not defined, the default is to output nothing
5717 special at the end of the file. Most systems don't require any
5720 On systems that use SDB, it is necessary to output certain commands;
5721 see @file{attasm.h}.
5723 @findex ASM_COMMENT_START
5724 @item ASM_COMMENT_START
5725 A C string constant describing how to begin a comment in the target
5726 assembler language. The compiler assumes that the comment will end at
5727 the end of the line.
5731 A C string constant for text to be output before each @code{asm}
5732 statement or group of consecutive ones. Normally this is
5733 @code{"#APP"}, which is a comment that has no effect on most
5734 assemblers but tells the GNU assembler that it must check the lines
5735 that follow for all valid assembler constructs.
5739 A C string constant for text to be output after each @code{asm}
5740 statement or group of consecutive ones. Normally this is
5741 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
5742 time-saving assumptions that are valid for ordinary compiler output.
5744 @findex ASM_OUTPUT_SOURCE_FILENAME
5745 @item ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
5746 A C statement to output COFF information or DWARF debugging information
5747 which indicates that filename @var{name} is the current source file to
5748 the stdio stream @var{stream}.
5750 This macro need not be defined if the standard form of output
5751 for the file format in use is appropriate.
5753 @findex OUTPUT_QUOTED_STRING
5754 @item OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
5755 A C statement to output the string @var{string} to the stdio stream
5756 @var{stream}. If you do not call the function @code{output_quoted_string}
5757 in your config files, GCC will only call it to output filenames to
5758 the assembler source. So you can use it to canonicalize the format
5759 of the filename using this macro.
5761 @findex ASM_OUTPUT_SOURCE_LINE
5762 @item ASM_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
5763 A C statement to output DBX or SDB debugging information before code
5764 for line number @var{line} of the current source file to the
5765 stdio stream @var{stream}.
5767 This macro need not be defined if the standard form of debugging
5768 information for the debugger in use is appropriate.
5770 @findex ASM_OUTPUT_IDENT
5771 @item ASM_OUTPUT_IDENT (@var{stream}, @var{string})
5772 A C statement to output something to the assembler file to handle a
5773 @samp{#ident} directive containing the text @var{string}. If this
5774 macro is not defined, nothing is output for a @samp{#ident} directive.
5776 @findex OBJC_PROLOGUE
5778 A C statement to output any assembler statements which are required to
5779 precede any Objective-C object definitions or message sending. The
5780 statement is executed only when compiling an Objective-C program.
5783 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, unsigned int @var{align})
5784 Output assembly directives to switch to section @var{name}. The section
5785 should have attributes as specified by @var{flags}, which is a bit mask
5786 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{align}
5787 is nonzero, it contains an alignment in bytes to be used for the section,
5788 otherwise some target default should be used. Only targets that must
5789 specify an alignment within the section directive need pay attention to
5790 @var{align} -- we will still use @code{ASM_OUTPUT_ALIGN}.
5793 @deftypefn {Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
5794 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
5797 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
5798 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
5799 based on a variable or function decl, a section name, and whether or not the
5800 declaration's initializer may contain runtime relocations. @var{decl} may be
5801 null, in which case read-write data should be assumed.
5803 The default version if this function handles choosing code vs data,
5804 read-only vs read-write data, and @code{flag_pic}. You should only
5805 need to override this if your target has special flags that might be
5806 set via @code{__attribute__}.
5811 @subsection Output of Data
5813 @c prevent bad page break with this line
5814 This describes data output.
5817 @findex ASM_OUTPUT_LONG_DOUBLE
5818 @findex ASM_OUTPUT_DOUBLE
5819 @findex ASM_OUTPUT_FLOAT
5820 @item ASM_OUTPUT_LONG_DOUBLE (@var{stream}, @var{value})
5821 @itemx ASM_OUTPUT_DOUBLE (@var{stream}, @var{value})
5822 @itemx ASM_OUTPUT_FLOAT (@var{stream}, @var{value})
5823 @itemx ASM_OUTPUT_THREE_QUARTER_FLOAT (@var{stream}, @var{value})
5824 @itemx ASM_OUTPUT_SHORT_FLOAT (@var{stream}, @var{value})
5825 @itemx ASM_OUTPUT_BYTE_FLOAT (@var{stream}, @var{value})
5826 A C statement to output to the stdio stream @var{stream} an assembler
5827 instruction to assemble a floating-point constant of @code{TFmode},
5828 @code{DFmode}, @code{SFmode}, @code{TQFmode}, @code{HFmode}, or
5829 @code{QFmode}, respectively, whose value is @var{value}. @var{value}
5830 will be a C expression of type @code{REAL_VALUE_TYPE}. Macros such as
5831 @code{REAL_VALUE_TO_TARGET_DOUBLE} are useful for writing these
5834 @findex ASM_OUTPUT_QUADRUPLE_INT
5835 @findex ASM_OUTPUT_DOUBLE_INT
5836 @findex ASM_OUTPUT_INT
5837 @findex ASM_OUTPUT_SHORT
5838 @findex ASM_OUTPUT_CHAR
5839 @findex output_addr_const
5840 @item ASM_OUTPUT_QUADRUPLE_INT (@var{stream}, @var{exp})
5841 @itemx ASM_OUTPUT_DOUBLE_INT (@var{stream}, @var{exp})
5842 @itemx ASM_OUTPUT_INT (@var{stream}, @var{exp})
5843 @itemx ASM_OUTPUT_SHORT (@var{stream}, @var{exp})
5844 @itemx ASM_OUTPUT_CHAR (@var{stream}, @var{exp})
5845 A C statement to output to the stdio stream @var{stream} an assembler
5846 instruction to assemble an integer of 16, 8, 4, 2 or 1 bytes,
5847 respectively, whose value is @var{value}. The argument @var{exp} will
5848 be an RTL expression which represents a constant value. Use
5849 @samp{output_addr_const (@var{stream}, @var{exp})} to output this value
5850 as an assembler expression.
5852 For sizes larger than @code{UNITS_PER_WORD}, if the action of a macro
5853 would be identical to repeatedly calling the macro corresponding to
5854 a size of @code{UNITS_PER_WORD}, once for each word, you need not define
5857 @findex OUTPUT_ADDR_CONST_EXTRA
5858 @item OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
5859 A C statement to recognize @var{rtx} patterns that
5860 @code{output_addr_const} can't deal with, and output assembly code to
5861 @var{stream} corresponding to the pattern @var{x}. This may be used to
5862 allow machine-dependent @code{UNSPEC}s to appear within constants.
5864 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
5865 @code{goto fail}, so that a standard error message is printed. If it
5866 prints an error message itself, by calling, for example,
5867 @code{output_operand_lossage}, it may just complete normally.
5869 @findex ASM_OUTPUT_BYTE
5870 @item ASM_OUTPUT_BYTE (@var{stream}, @var{value})
5871 A C statement to output to the stdio stream @var{stream} an assembler
5872 instruction to assemble a single byte containing the number @var{value}.
5876 A C string constant, including spacing, giving the pseudo-op to use for a
5877 sequence of single-byte constants. If this macro is not defined, the
5878 default is @code{"\t.byte\t"}.
5880 @findex UNALIGNED_SHORT_ASM_OP
5881 @findex UNALIGNED_INT_ASM_OP
5882 @findex UNALIGNED_DOUBLE_INT_ASM_OP
5883 @item UNALIGNED_SHORT_ASM_OP
5884 @itemx UNALIGNED_INT_ASM_OP
5885 @itemx UNALIGNED_DOUBLE_INT_ASM_OP
5886 A C string constant, including spacing, giving the pseudo-op to use
5887 to assemble 16-, 32-, and 64-bit integers respectively @emph{without}
5888 adding implicit padding or alignment. These macros are required if
5889 DWARF 2 frame unwind is used. On ELF systems, these will default
5890 to @code{.2byte}, @code{.4byte}, and @code{.8byte}.
5892 @findex ASM_OUTPUT_ASCII
5893 @item ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
5894 A C statement to output to the stdio stream @var{stream} an assembler
5895 instruction to assemble a string constant containing the @var{len}
5896 bytes at @var{ptr}. @var{ptr} will be a C expression of type
5897 @code{char *} and @var{len} a C expression of type @code{int}.
5899 If the assembler has a @code{.ascii} pseudo-op as found in the
5900 Berkeley Unix assembler, do not define the macro
5901 @code{ASM_OUTPUT_ASCII}.
5903 @findex ASM_OUTPUT_FDESC
5904 @item ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
5905 A C statement to output word @var{n} of a function descriptor for
5906 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
5907 is defined, and is otherwise unused.
5909 @findex CONSTANT_POOL_BEFORE_FUNCTION
5910 @item CONSTANT_POOL_BEFORE_FUNCTION
5911 You may define this macro as a C expression. You should define the
5912 expression to have a nonzero value if GCC should output the constant
5913 pool for a function before the code for the function, or a zero value if
5914 GCC should output the constant pool after the function. If you do
5915 not define this macro, the usual case, GCC will output the constant
5916 pool before the function.
5918 @findex ASM_OUTPUT_POOL_PROLOGUE
5919 @item ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
5920 A C statement to output assembler commands to define the start of the
5921 constant pool for a function. @var{funname} is a string giving
5922 the name of the function. Should the return type of the function
5923 be required, it can be obtained via @var{fundecl}. @var{size}
5924 is the size, in bytes, of the constant pool that will be written
5925 immediately after this call.
5927 If no constant-pool prefix is required, the usual case, this macro need
5930 @findex ASM_OUTPUT_SPECIAL_POOL_ENTRY
5931 @item ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
5932 A C statement (with or without semicolon) to output a constant in the
5933 constant pool, if it needs special treatment. (This macro need not do
5934 anything for RTL expressions that can be output normally.)
5936 The argument @var{file} is the standard I/O stream to output the
5937 assembler code on. @var{x} is the RTL expression for the constant to
5938 output, and @var{mode} is the machine mode (in case @var{x} is a
5939 @samp{const_int}). @var{align} is the required alignment for the value
5940 @var{x}; you should output an assembler directive to force this much
5943 The argument @var{labelno} is a number to use in an internal label for
5944 the address of this pool entry. The definition of this macro is
5945 responsible for outputting the label definition at the proper place.
5946 Here is how to do this:
5949 ASM_OUTPUT_INTERNAL_LABEL (@var{file}, "LC", @var{labelno});
5952 When you output a pool entry specially, you should end with a
5953 @code{goto} to the label @var{jumpto}. This will prevent the same pool
5954 entry from being output a second time in the usual manner.
5956 You need not define this macro if it would do nothing.
5958 @findex CONSTANT_AFTER_FUNCTION_P
5959 @item CONSTANT_AFTER_FUNCTION_P (@var{exp})
5960 Define this macro as a C expression which is nonzero if the constant
5961 @var{exp}, of type @code{tree}, should be output after the code for a
5962 function. The compiler will normally output all constants before the
5963 function; you need not define this macro if this is OK@.
5965 @findex ASM_OUTPUT_POOL_EPILOGUE
5966 @item ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
5967 A C statement to output assembler commands to at the end of the constant
5968 pool for a function. @var{funname} is a string giving the name of the
5969 function. Should the return type of the function be required, you can
5970 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
5971 constant pool that GCC wrote immediately before this call.
5973 If no constant-pool epilogue is required, the usual case, you need not
5976 @findex IS_ASM_LOGICAL_LINE_SEPARATOR
5977 @item IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C})
5978 Define this macro as a C expression which is nonzero if @var{C} is
5979 used as a logical line separator by the assembler.
5981 If you do not define this macro, the default is that only
5982 the character @samp{;} is treated as a logical line separator.
5985 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
5986 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
5987 These target hooks are C string constants, describing the syntax in the
5988 assembler for grouping arithmetic expressions. If not overridden, they
5989 default to normal parentheses, which is correct for most assemblers.
5992 These macros are provided by @file{real.h} for writing the definitions
5993 of @code{ASM_OUTPUT_DOUBLE} and the like:
5996 @item REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
5997 @itemx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
5998 @itemx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
5999 @findex REAL_VALUE_TO_TARGET_SINGLE
6000 @findex REAL_VALUE_TO_TARGET_DOUBLE
6001 @findex REAL_VALUE_TO_TARGET_LONG_DOUBLE
6002 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the target's
6003 floating point representation, and store its bit pattern in the array of
6004 @code{long int} whose address is @var{l}. The number of elements in the
6005 output array is determined by the size of the desired target floating
6006 point data type: 32 bits of it go in each @code{long int} array
6007 element. Each array element holds 32 bits of the result, even if
6008 @code{long int} is wider than 32 bits on the host machine.
6010 The array element values are designed so that you can print them out
6011 using @code{fprintf} in the order they should appear in the target
6014 @item REAL_VALUE_TO_DECIMAL (@var{x}, @var{format}, @var{string})
6015 @findex REAL_VALUE_TO_DECIMAL
6016 This macro converts @var{x}, of type @code{REAL_VALUE_TYPE}, to a
6017 decimal number and stores it as a string into @var{string}.
6018 You must pass, as @var{string}, the address of a long enough block
6019 of space to hold the result.
6021 The argument @var{format} is a @code{printf}-specification that serves
6022 as a suggestion for how to format the output string.
6025 @node Uninitialized Data
6026 @subsection Output of Uninitialized Variables
6028 Each of the macros in this section is used to do the whole job of
6029 outputting a single uninitialized variable.
6032 @findex ASM_OUTPUT_COMMON
6033 @item ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
6034 A C statement (sans semicolon) to output to the stdio stream
6035 @var{stream} the assembler definition of a common-label named
6036 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
6037 is the size rounded up to whatever alignment the caller wants.
6039 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
6040 output the name itself; before and after that, output the additional
6041 assembler syntax for defining the name, and a newline.
6043 This macro controls how the assembler definitions of uninitialized
6044 common global variables are output.
6046 @findex ASM_OUTPUT_ALIGNED_COMMON
6047 @item ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
6048 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
6049 separate, explicit argument. If you define this macro, it is used in
6050 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
6051 handling the required alignment of the variable. The alignment is specified
6052 as the number of bits.
6054 @findex ASM_OUTPUT_ALIGNED_DECL_COMMON
6055 @item ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
6056 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
6057 variable to be output, if there is one, or @code{NULL_TREE} if there
6058 is no corresponding variable. If you define this macro, GCC will use it
6059 in place of both @code{ASM_OUTPUT_COMMON} and
6060 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
6061 the variable's decl in order to chose what to output.
6063 @findex ASM_OUTPUT_SHARED_COMMON
6064 @item ASM_OUTPUT_SHARED_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
6065 If defined, it is similar to @code{ASM_OUTPUT_COMMON}, except that it
6066 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_COMMON}
6069 @findex ASM_OUTPUT_BSS
6070 @item ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
6071 A C statement (sans semicolon) to output to the stdio stream
6072 @var{stream} the assembler definition of uninitialized global @var{decl} named
6073 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
6074 is the size rounded up to whatever alignment the caller wants.
6076 Try to use function @code{asm_output_bss} defined in @file{varasm.c} when
6077 defining this macro. If unable, use the expression
6078 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
6079 before and after that, output the additional assembler syntax for defining
6080 the name, and a newline.
6082 This macro controls how the assembler definitions of uninitialized global
6083 variables are output. This macro exists to properly support languages like
6084 C++ which do not have @code{common} data. However, this macro currently
6085 is not defined for all targets. If this macro and
6086 @code{ASM_OUTPUT_ALIGNED_BSS} are not defined then @code{ASM_OUTPUT_COMMON}
6087 or @code{ASM_OUTPUT_ALIGNED_COMMON} or
6088 @code{ASM_OUTPUT_ALIGNED_DECL_COMMON} is used.
6090 @findex ASM_OUTPUT_ALIGNED_BSS
6091 @item ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
6092 Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a
6093 separate, explicit argument. If you define this macro, it is used in
6094 place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in
6095 handling the required alignment of the variable. The alignment is specified
6096 as the number of bits.
6098 Try to use function @code{asm_output_aligned_bss} defined in file
6099 @file{varasm.c} when defining this macro.
6101 @findex ASM_OUTPUT_SHARED_BSS
6102 @item ASM_OUTPUT_SHARED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
6103 If defined, it is similar to @code{ASM_OUTPUT_BSS}, except that it
6104 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_BSS}
6107 @findex ASM_OUTPUT_LOCAL
6108 @item ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
6109 A C statement (sans semicolon) to output to the stdio stream
6110 @var{stream} the assembler definition of a local-common-label named
6111 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
6112 is the size rounded up to whatever alignment the caller wants.
6114 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
6115 output the name itself; before and after that, output the additional
6116 assembler syntax for defining the name, and a newline.
6118 This macro controls how the assembler definitions of uninitialized
6119 static variables are output.
6121 @findex ASM_OUTPUT_ALIGNED_LOCAL
6122 @item ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
6123 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
6124 separate, explicit argument. If you define this macro, it is used in
6125 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
6126 handling the required alignment of the variable. The alignment is specified
6127 as the number of bits.
6129 @findex ASM_OUTPUT_ALIGNED_DECL_LOCAL
6130 @item ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
6131 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
6132 variable to be output, if there is one, or @code{NULL_TREE} if there
6133 is no corresponding variable. If you define this macro, GCC will use it
6134 in place of both @code{ASM_OUTPUT_DECL} and
6135 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
6136 the variable's decl in order to chose what to output.
6138 @findex ASM_OUTPUT_SHARED_LOCAL
6139 @item ASM_OUTPUT_SHARED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
6140 If defined, it is similar to @code{ASM_OUTPUT_LOCAL}, except that it
6141 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_LOCAL}
6146 @subsection Output and Generation of Labels
6148 @c prevent bad page break with this line
6149 This is about outputting labels.
6152 @findex ASM_OUTPUT_LABEL
6153 @findex assemble_name
6154 @item ASM_OUTPUT_LABEL (@var{stream}, @var{name})
6155 A C statement (sans semicolon) to output to the stdio stream
6156 @var{stream} the assembler definition of a label named @var{name}.
6157 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
6158 output the name itself; before and after that, output the additional
6159 assembler syntax for defining the name, and a newline.
6161 @findex ASM_DECLARE_FUNCTION_NAME
6162 @item ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
6163 A C statement (sans semicolon) to output to the stdio stream
6164 @var{stream} any text necessary for declaring the name @var{name} of a
6165 function which is being defined. This macro is responsible for
6166 outputting the label definition (perhaps using
6167 @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the
6168 @code{FUNCTION_DECL} tree node representing the function.
6170 If this macro is not defined, then the function name is defined in the
6171 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
6173 @findex ASM_DECLARE_FUNCTION_SIZE
6174 @item ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
6175 A C statement (sans semicolon) to output to the stdio stream
6176 @var{stream} any text necessary for declaring the size of a function
6177 which is being defined. The argument @var{name} is the name of the
6178 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
6179 representing the function.
6181 If this macro is not defined, then the function size is not defined.
6183 @findex ASM_DECLARE_OBJECT_NAME
6184 @item ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
6185 A C statement (sans semicolon) to output to the stdio stream
6186 @var{stream} any text necessary for declaring the name @var{name} of an
6187 initialized variable which is being defined. This macro must output the
6188 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
6189 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
6191 If this macro is not defined, then the variable name is defined in the
6192 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
6194 @findex ASM_DECLARE_REGISTER_GLOBAL
6195 @item ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
6196 A C statement (sans semicolon) to output to the stdio stream
6197 @var{stream} any text necessary for claiming a register @var{regno}
6198 for a global variable @var{decl} with name @var{name}.
6200 If you don't define this macro, that is equivalent to defining it to do
6203 @findex ASM_FINISH_DECLARE_OBJECT
6204 @item ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
6205 A C statement (sans semicolon) to finish up declaring a variable name
6206 once the compiler has processed its initializer fully and thus has had a
6207 chance to determine the size of an array when controlled by an
6208 initializer. This is used on systems where it's necessary to declare
6209 something about the size of the object.
6211 If you don't define this macro, that is equivalent to defining it to do
6214 @findex ASM_GLOBALIZE_LABEL
6215 @item ASM_GLOBALIZE_LABEL (@var{stream}, @var{name})
6216 A C statement (sans semicolon) to output to the stdio stream
6217 @var{stream} some commands that will make the label @var{name} global;
6218 that is, available for reference from other files. Use the expression
6219 @code{assemble_name (@var{stream}, @var{name})} to output the name
6220 itself; before and after that, output the additional assembler syntax
6221 for making that name global, and a newline.
6223 @findex ASM_WEAKEN_LABEL
6224 @item ASM_WEAKEN_LABEL
6225 A C statement (sans semicolon) to output to the stdio stream
6226 @var{stream} some commands that will make the label @var{name} weak;
6227 that is, available for reference from other files but only used if
6228 no other definition is available. Use the expression
6229 @code{assemble_name (@var{stream}, @var{name})} to output the name
6230 itself; before and after that, output the additional assembler syntax
6231 for making that name weak, and a newline.
6233 If you don't define this macro, GCC will not support weak
6234 symbols and you should not define the @code{SUPPORTS_WEAK} macro.
6236 @findex SUPPORTS_WEAK
6238 A C expression which evaluates to true if the target supports weak symbols.
6240 If you don't define this macro, @file{defaults.h} provides a default
6241 definition. If @code{ASM_WEAKEN_LABEL} is defined, the default
6242 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
6243 you want to control weak symbol support with a compiler flag such as
6246 @findex MAKE_DECL_ONE_ONLY (@var{decl})
6247 @item MAKE_DECL_ONE_ONLY
6248 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
6249 public symbol such that extra copies in multiple translation units will
6250 be discarded by the linker. Define this macro if your object file
6251 format provides support for this concept, such as the @samp{COMDAT}
6252 section flags in the Microsoft Windows PE/COFF format, and this support
6253 requires changes to @var{decl}, such as putting it in a separate section.
6255 @findex SUPPORTS_ONE_ONLY
6256 @item SUPPORTS_ONE_ONLY
6257 A C expression which evaluates to true if the target supports one-only
6260 If you don't define this macro, @file{varasm.c} provides a default
6261 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
6262 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
6263 you want to control one-only symbol support with a compiler flag, or if
6264 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
6265 be emitted as one-only.
6267 @findex ASM_OUTPUT_EXTERNAL
6268 @item ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
6269 A C statement (sans semicolon) to output to the stdio stream
6270 @var{stream} any text necessary for declaring the name of an external
6271 symbol named @var{name} which is referenced in this compilation but
6272 not defined. The value of @var{decl} is the tree node for the
6275 This macro need not be defined if it does not need to output anything.
6276 The GNU assembler and most Unix assemblers don't require anything.
6278 @findex ASM_OUTPUT_EXTERNAL_LIBCALL
6279 @item ASM_OUTPUT_EXTERNAL_LIBCALL (@var{stream}, @var{symref})
6280 A C statement (sans semicolon) to output on @var{stream} an assembler
6281 pseudo-op to declare a library function name external. The name of the
6282 library function is given by @var{symref}, which has type @code{rtx} and
6283 is a @code{symbol_ref}.
6285 This macro need not be defined if it does not need to output anything.
6286 The GNU assembler and most Unix assemblers don't require anything.
6288 @findex ASM_OUTPUT_LABELREF
6289 @item ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
6290 A C statement (sans semicolon) to output to the stdio stream
6291 @var{stream} a reference in assembler syntax to a label named
6292 @var{name}. This should add @samp{_} to the front of the name, if that
6293 is customary on your operating system, as it is in most Berkeley Unix
6294 systems. This macro is used in @code{assemble_name}.
6296 @ignore @c Seems not to exist anymore.
6297 @findex ASM_OUTPUT_LABELREF_AS_INT
6298 @item ASM_OUTPUT_LABELREF_AS_INT (@var{file}, @var{label})
6299 Define this macro for systems that use the program @command{collect2}.
6300 The definition should be a C statement to output a word containing
6301 a reference to the label @var{label}.
6304 @findex ASM_OUTPUT_SYMBOL_REF
6305 @item ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
6306 A C statement (sans semicolon) to output a reference to
6307 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
6308 will be used to output the name of the symbol. This macro may be used
6309 to modify the way a symbol is referenced depending on information
6310 encoded by @code{ENCODE_SECTION_INFO}.
6312 @findex ASM_OUTPUT_LABEL_REF
6313 @item ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
6314 A C statement (sans semicolon) to output a reference to @var{buf}, the
6315 result of ASM_GENERATE_INTERNAL_LABEL. If not defined,
6316 @code{assemble_name} will be used to output the name of the symbol.
6317 This macro is not used by @code{output_asm_label}, or the @code{%l}
6318 specifier that calls it; the intention is that this macro should be set
6319 when it is necessary to output a label differently when its address
6322 @findex ASM_OUTPUT_INTERNAL_LABEL
6323 @item ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{prefix}, @var{num})
6324 A C statement to output to the stdio stream @var{stream} a label whose
6325 name is made from the string @var{prefix} and the number @var{num}.
6327 It is absolutely essential that these labels be distinct from the labels
6328 used for user-level functions and variables. Otherwise, certain programs
6329 will have name conflicts with internal labels.
6331 It is desirable to exclude internal labels from the symbol table of the
6332 object file. Most assemblers have a naming convention for labels that
6333 should be excluded; on many systems, the letter @samp{L} at the
6334 beginning of a label has this effect. You should find out what
6335 convention your system uses, and follow it.
6337 The usual definition of this macro is as follows:
6340 fprintf (@var{stream}, "L%s%d:\n", @var{prefix}, @var{num})
6343 @findex ASM_OUTPUT_DEBUG_LABEL
6344 @item ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
6345 A C statement to output to the stdio stream @var{stream} a debug info
6346 label whose name is made from the string @var{prefix} and the number
6347 @var{num}. This is useful for VLIW targets, where debug info labels
6348 may need to be treated differently than branch target labels. On some
6349 systems, branch target labels must be at the beginning of instruction
6350 bundles, but debug info labels can occur in the middle of instruction
6353 If this macro is not defined, then @code{ASM_OUTPUT_INTERNAL_LABEL} will be
6356 @findex ASM_OUTPUT_ALTERNATE_LABEL_NAME
6357 @item ASM_OUTPUT_ALTERNATE_LABEL_NAME (@var{stream}, @var{string})
6358 A C statement to output to the stdio stream @var{stream} the string
6361 The default definition of this macro is as follows:
6364 fprintf (@var{stream}, "%s:\n", LABEL_ALTERNATE_NAME (INSN))
6367 @findex ASM_GENERATE_INTERNAL_LABEL
6368 @item ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
6369 A C statement to store into the string @var{string} a label whose name
6370 is made from the string @var{prefix} and the number @var{num}.
6372 This string, when output subsequently by @code{assemble_name}, should
6373 produce the output that @code{ASM_OUTPUT_INTERNAL_LABEL} would produce
6374 with the same @var{prefix} and @var{num}.
6376 If the string begins with @samp{*}, then @code{assemble_name} will
6377 output the rest of the string unchanged. It is often convenient for
6378 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
6379 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
6380 to output the string, and may change it. (Of course,
6381 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
6382 you should know what it does on your machine.)
6384 @findex ASM_FORMAT_PRIVATE_NAME
6385 @item ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
6386 A C expression to assign to @var{outvar} (which is a variable of type
6387 @code{char *}) a newly allocated string made from the string
6388 @var{name} and the number @var{number}, with some suitable punctuation
6389 added. Use @code{alloca} to get space for the string.
6391 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
6392 produce an assembler label for an internal static variable whose name is
6393 @var{name}. Therefore, the string must be such as to result in valid
6394 assembler code. The argument @var{number} is different each time this
6395 macro is executed; it prevents conflicts between similarly-named
6396 internal static variables in different scopes.
6398 Ideally this string should not be a valid C identifier, to prevent any
6399 conflict with the user's own symbols. Most assemblers allow periods
6400 or percent signs in assembler symbols; putting at least one of these
6401 between the name and the number will suffice.
6403 @findex ASM_OUTPUT_DEF
6404 @item ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
6405 A C statement to output to the stdio stream @var{stream} assembler code
6406 which defines (equates) the symbol @var{name} to have the value @var{value}.
6409 If @code{SET_ASM_OP} is defined, a default definition is provided which is
6410 correct for most systems.
6412 @findex ASM_OUTPUT_DEF_FROM_DECLS
6413 @item ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
6414 A C statement to output to the stdio stream @var{stream} assembler code
6415 which defines (equates) the symbol whose tree node is @var{decl_of_name}
6416 to have the value of the tree node @var{decl_of_value}. This macro will
6417 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
6418 the tree nodes are available.
6420 @findex ASM_OUTPUT_DEFINE_LABEL_DIFFERENCE_SYMBOL
6421 @item ASM_OUTPUT_DEFINE_LABEL_DIFFERENCE_SYMBOL (@var{stream}, @var{symbol}, @var{high}, @var{low})
6422 A C statement to output to the stdio stream @var{stream} assembler code
6423 which defines (equates) the symbol @var{symbol} to have a value equal to
6424 the difference of the two symbols @var{high} and @var{low},
6425 i.e.@: @var{high} minus @var{low}. GCC guarantees that the symbols @var{high}
6426 and @var{low} are already known by the assembler so that the difference
6427 resolves into a constant.
6430 If @code{SET_ASM_OP} is defined, a default definition is provided which is
6431 correct for most systems.
6433 @findex ASM_OUTPUT_WEAK_ALIAS
6434 @item ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
6435 A C statement to output to the stdio stream @var{stream} assembler code
6436 which defines (equates) the weak symbol @var{name} to have the value
6437 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
6438 an undefined weak symbol.
6440 Define this macro if the target only supports weak aliases; define
6441 @code{ASM_OUTPUT_DEF} instead if possible.
6443 @findex OBJC_GEN_METHOD_LABEL
6444 @item OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
6445 Define this macro to override the default assembler names used for
6446 Objective-C methods.
6448 The default name is a unique method number followed by the name of the
6449 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
6450 the category is also included in the assembler name (e.g.@:
6453 These names are safe on most systems, but make debugging difficult since
6454 the method's selector is not present in the name. Therefore, particular
6455 systems define other ways of computing names.
6457 @var{buf} is an expression of type @code{char *} which gives you a
6458 buffer in which to store the name; its length is as long as
6459 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
6460 50 characters extra.
6462 The argument @var{is_inst} specifies whether the method is an instance
6463 method or a class method; @var{class_name} is the name of the class;
6464 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
6465 in a category); and @var{sel_name} is the name of the selector.
6467 On systems where the assembler can handle quoted names, you can use this
6468 macro to provide more human-readable names.
6470 @findex ASM_DECLARE_CLASS_REFERENCE
6471 @item ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
6472 A C statement (sans semicolon) to output to the stdio stream
6473 @var{stream} commands to declare that the label @var{name} is an
6474 Objective-C class reference. This is only needed for targets whose
6475 linkers have special support for NeXT-style runtimes.
6477 @findex ASM_DECLARE_UNRESOLVED_REFERENCE
6478 @item ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
6479 A C statement (sans semicolon) to output to the stdio stream
6480 @var{stream} commands to declare that the label @var{name} is an
6481 unresolved Objective-C class reference. This is only needed for targets
6482 whose linkers have special support for NeXT-style runtimes.
6485 @node Initialization
6486 @subsection How Initialization Functions Are Handled
6487 @cindex initialization routines
6488 @cindex termination routines
6489 @cindex constructors, output of
6490 @cindex destructors, output of
6492 The compiled code for certain languages includes @dfn{constructors}
6493 (also called @dfn{initialization routines})---functions to initialize
6494 data in the program when the program is started. These functions need
6495 to be called before the program is ``started''---that is to say, before
6496 @code{main} is called.
6498 Compiling some languages generates @dfn{destructors} (also called
6499 @dfn{termination routines}) that should be called when the program
6502 To make the initialization and termination functions work, the compiler
6503 must output something in the assembler code to cause those functions to
6504 be called at the appropriate time. When you port the compiler to a new
6505 system, you need to specify how to do this.
6507 There are two major ways that GCC currently supports the execution of
6508 initialization and termination functions. Each way has two variants.
6509 Much of the structure is common to all four variations.
6511 @findex __CTOR_LIST__
6512 @findex __DTOR_LIST__
6513 The linker must build two lists of these functions---a list of
6514 initialization functions, called @code{__CTOR_LIST__}, and a list of
6515 termination functions, called @code{__DTOR_LIST__}.
6517 Each list always begins with an ignored function pointer (which may hold
6518 0, @minus{}1, or a count of the function pointers after it, depending on
6519 the environment). This is followed by a series of zero or more function
6520 pointers to constructors (or destructors), followed by a function
6521 pointer containing zero.
6523 Depending on the operating system and its executable file format, either
6524 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
6525 time and exit time. Constructors are called in reverse order of the
6526 list; destructors in forward order.
6528 The best way to handle static constructors works only for object file
6529 formats which provide arbitrarily-named sections. A section is set
6530 aside for a list of constructors, and another for a list of destructors.
6531 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
6532 object file that defines an initialization function also puts a word in
6533 the constructor section to point to that function. The linker
6534 accumulates all these words into one contiguous @samp{.ctors} section.
6535 Termination functions are handled similarly.
6537 This method will be chosen as the default by @file{target-def.h} if
6538 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
6539 support arbitrary sections, but does support special designated
6540 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
6541 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
6543 When arbitrary sections are available, there are two variants, depending
6544 upon how the code in @file{crtstuff.c} is called. On systems that
6545 support a @dfn{.init} section which is executed at program startup,
6546 parts of @file{crtstuff.c} are compiled into that section. The
6547 program is linked by the @code{gcc} driver like this:
6550 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
6553 The prologue of a function (@code{__init}) appears in the @code{.init}
6554 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
6555 for the function @code{__fini} in the @dfn{.fini} section. Normally these
6556 files are provided by the operating system or by the GNU C library, but
6557 are provided by GCC for a few targets.
6559 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
6560 compiled from @file{crtstuff.c}. They contain, among other things, code
6561 fragments within the @code{.init} and @code{.fini} sections that branch
6562 to routines in the @code{.text} section. The linker will pull all parts
6563 of a section together, which results in a complete @code{__init} function
6564 that invokes the routines we need at startup.
6566 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
6569 If no init section is available, when GCC compiles any function called
6570 @code{main} (or more accurately, any function designated as a program
6571 entry point by the language front end calling @code{expand_main_function}),
6572 it inserts a procedure call to @code{__main} as the first executable code
6573 after the function prologue. The @code{__main} function is defined
6574 in @file{libgcc2.c} and runs the global constructors.
6576 In file formats that don't support arbitrary sections, there are again
6577 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
6578 and an `a.out' format must be used. In this case,
6579 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
6580 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
6581 and with the address of the void function containing the initialization
6582 code as its value. The GNU linker recognizes this as a request to add
6583 the value to a @dfn{set}; the values are accumulated, and are eventually
6584 placed in the executable as a vector in the format described above, with
6585 a leading (ignored) count and a trailing zero element.
6586 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
6587 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
6588 the compilation of @code{main} to call @code{__main} as above, starting
6589 the initialization process.
6591 The last variant uses neither arbitrary sections nor the GNU linker.
6592 This is preferable when you want to do dynamic linking and when using
6593 file formats which the GNU linker does not support, such as `ECOFF'@. In
6594 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
6595 termination functions are recognized simply by their names. This requires
6596 an extra program in the linkage step, called @command{collect2}. This program
6597 pretends to be the linker, for use with GCC; it does its job by running
6598 the ordinary linker, but also arranges to include the vectors of
6599 initialization and termination functions. These functions are called
6600 via @code{__main} as described above. In order to use this method,
6601 @code{use_collect2} must be defined in the target in @file{config.gcc}.
6604 The following section describes the specific macros that control and
6605 customize the handling of initialization and termination functions.
6608 @node Macros for Initialization
6609 @subsection Macros Controlling Initialization Routines
6611 Here are the macros that control how the compiler handles initialization
6612 and termination functions:
6615 @findex INIT_SECTION_ASM_OP
6616 @item INIT_SECTION_ASM_OP
6617 If defined, a C string constant, including spacing, for the assembler
6618 operation to identify the following data as initialization code. If not
6619 defined, GCC will assume such a section does not exist. When you are
6620 using special sections for initialization and termination functions, this
6621 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
6622 run the initialization functions.
6624 @item HAS_INIT_SECTION
6625 @findex HAS_INIT_SECTION
6626 If defined, @code{main} will not call @code{__main} as described above.
6627 This macro should be defined for systems that control start-up code
6628 on a symbol-by-symbol basis, such as OSF/1, and should not
6629 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
6631 @item LD_INIT_SWITCH
6632 @findex LD_INIT_SWITCH
6633 If defined, a C string constant for a switch that tells the linker that
6634 the following symbol is an initialization routine.
6636 @item LD_FINI_SWITCH
6637 @findex LD_FINI_SWITCH
6638 If defined, a C string constant for a switch that tells the linker that
6639 the following symbol is a finalization routine.
6642 @findex INVOKE__main
6643 If defined, @code{main} will call @code{__main} despite the presence of
6644 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
6645 where the init section is not actually run automatically, but is still
6646 useful for collecting the lists of constructors and destructors.
6648 @item SUPPORTS_INIT_PRIORITY
6649 @findex SUPPORTS_INIT_PRIORITY
6650 If nonzero, the C++ @code{init_priority} attribute is supported and the
6651 compiler should emit instructions to control the order of initialization
6652 of objects. If zero, the compiler will issue an error message upon
6653 encountering an @code{init_priority} attribute.
6656 @deftypefn {Target Hook} bool TARGET_HAVE_CTORS_DTORS
6657 This value is true if the target supports some ``native'' method of
6658 collecting constructors and destructors to be run at startup and exit.
6659 It is false if we must use @command{collect2}.
6662 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
6663 If defined, a function that outputs assembler code to arrange to call
6664 the function referenced by @var{symbol} at initialization time.
6666 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
6667 no arguments and with no return value. If the target supports initialization
6668 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
6669 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
6671 If this macro is not defined by the target, a suitable default will
6672 be chosen if (1) the target supports arbitrary section names, (2) the
6673 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
6677 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
6678 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
6679 functions rather than initialization functions.
6682 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
6683 generated for the generated object file will have static linkage.
6685 If your system uses @command{collect2} as the means of processing
6686 constructors, then that program normally uses @command{nm} to scan
6687 an object file for constructor functions to be called.
6689 On certain kinds of systems, you can define these macros to make
6690 @command{collect2} work faster (and, in some cases, make it work at all):
6693 @findex OBJECT_FORMAT_COFF
6694 @item OBJECT_FORMAT_COFF
6695 Define this macro if the system uses COFF (Common Object File Format)
6696 object files, so that @command{collect2} can assume this format and scan
6697 object files directly for dynamic constructor/destructor functions.
6699 @findex OBJECT_FORMAT_ROSE
6700 @item OBJECT_FORMAT_ROSE
6701 Define this macro if the system uses ROSE format object files, so that
6702 @command{collect2} can assume this format and scan object files directly
6703 for dynamic constructor/destructor functions.
6705 These macros are effective only in a native compiler; @command{collect2} as
6706 part of a cross compiler always uses @command{nm} for the target machine.
6708 @findex REAL_NM_FILE_NAME
6709 @item REAL_NM_FILE_NAME
6710 Define this macro as a C string constant containing the file name to use
6711 to execute @command{nm}. The default is to search the path normally for
6714 If your system supports shared libraries and has a program to list the
6715 dynamic dependencies of a given library or executable, you can define
6716 these macros to enable support for running initialization and
6717 termination functions in shared libraries:
6721 Define this macro to a C string constant containing the name of the program
6722 which lists dynamic dependencies, like @command{"ldd"} under SunOS 4.
6724 @findex PARSE_LDD_OUTPUT
6725 @item PARSE_LDD_OUTPUT (@var{ptr})
6726 Define this macro to be C code that extracts filenames from the output
6727 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
6728 of type @code{char *} that points to the beginning of a line of output
6729 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
6730 code must advance @var{ptr} to the beginning of the filename on that
6731 line. Otherwise, it must set @var{ptr} to @code{NULL}.
6734 @node Instruction Output
6735 @subsection Output of Assembler Instructions
6737 @c prevent bad page break with this line
6738 This describes assembler instruction output.
6741 @findex REGISTER_NAMES
6742 @item REGISTER_NAMES
6743 A C initializer containing the assembler's names for the machine
6744 registers, each one as a C string constant. This is what translates
6745 register numbers in the compiler into assembler language.
6747 @findex ADDITIONAL_REGISTER_NAMES
6748 @item ADDITIONAL_REGISTER_NAMES
6749 If defined, a C initializer for an array of structures containing a name
6750 and a register number. This macro defines additional names for hard
6751 registers, thus allowing the @code{asm} option in declarations to refer
6752 to registers using alternate names.
6754 @findex ASM_OUTPUT_OPCODE
6755 @item ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
6756 Define this macro if you are using an unusual assembler that
6757 requires different names for the machine instructions.
6759 The definition is a C statement or statements which output an
6760 assembler instruction opcode to the stdio stream @var{stream}. The
6761 macro-operand @var{ptr} is a variable of type @code{char *} which
6762 points to the opcode name in its ``internal'' form---the form that is
6763 written in the machine description. The definition should output the
6764 opcode name to @var{stream}, performing any translation you desire, and
6765 increment the variable @var{ptr} to point at the end of the opcode
6766 so that it will not be output twice.
6768 In fact, your macro definition may process less than the entire opcode
6769 name, or more than the opcode name; but if you want to process text
6770 that includes @samp{%}-sequences to substitute operands, you must take
6771 care of the substitution yourself. Just be sure to increment
6772 @var{ptr} over whatever text should not be output normally.
6774 @findex recog_operand
6775 If you need to look at the operand values, they can be found as the
6776 elements of @code{recog_operand}.
6778 If the macro definition does nothing, the instruction is output
6781 @findex FINAL_PRESCAN_INSN
6782 @item FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
6783 If defined, a C statement to be executed just prior to the output of
6784 assembler code for @var{insn}, to modify the extracted operands so
6785 they will be output differently.
6787 Here the argument @var{opvec} is the vector containing the operands
6788 extracted from @var{insn}, and @var{noperands} is the number of
6789 elements of the vector which contain meaningful data for this insn.
6790 The contents of this vector are what will be used to convert the insn
6791 template into assembler code, so you can change the assembler output
6792 by changing the contents of the vector.
6794 This macro is useful when various assembler syntaxes share a single
6795 file of instruction patterns; by defining this macro differently, you
6796 can cause a large class of instructions to be output differently (such
6797 as with rearranged operands). Naturally, variations in assembler
6798 syntax affecting individual insn patterns ought to be handled by
6799 writing conditional output routines in those patterns.
6801 If this macro is not defined, it is equivalent to a null statement.
6803 @findex FINAL_PRESCAN_LABEL
6804 @item FINAL_PRESCAN_LABEL
6805 If defined, @code{FINAL_PRESCAN_INSN} will be called on each
6806 @code{CODE_LABEL}. In that case, @var{opvec} will be a null pointer and
6807 @var{noperands} will be zero.
6809 @findex PRINT_OPERAND
6810 @item PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
6811 A C compound statement to output to stdio stream @var{stream} the
6812 assembler syntax for an instruction operand @var{x}. @var{x} is an
6815 @var{code} is a value that can be used to specify one of several ways
6816 of printing the operand. It is used when identical operands must be
6817 printed differently depending on the context. @var{code} comes from
6818 the @samp{%} specification that was used to request printing of the
6819 operand. If the specification was just @samp{%@var{digit}} then
6820 @var{code} is 0; if the specification was @samp{%@var{ltr}
6821 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
6824 If @var{x} is a register, this macro should print the register's name.
6825 The names can be found in an array @code{reg_names} whose type is
6826 @code{char *[]}. @code{reg_names} is initialized from
6827 @code{REGISTER_NAMES}.
6829 When the machine description has a specification @samp{%@var{punct}}
6830 (a @samp{%} followed by a punctuation character), this macro is called
6831 with a null pointer for @var{x} and the punctuation character for
6834 @findex PRINT_OPERAND_PUNCT_VALID_P
6835 @item PRINT_OPERAND_PUNCT_VALID_P (@var{code})
6836 A C expression which evaluates to true if @var{code} is a valid
6837 punctuation character for use in the @code{PRINT_OPERAND} macro. If
6838 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
6839 punctuation characters (except for the standard one, @samp{%}) are used
6842 @findex PRINT_OPERAND_ADDRESS
6843 @item PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
6844 A C compound statement to output to stdio stream @var{stream} the
6845 assembler syntax for an instruction operand that is a memory reference
6846 whose address is @var{x}. @var{x} is an RTL expression.
6848 @cindex @code{ENCODE_SECTION_INFO} usage
6849 On some machines, the syntax for a symbolic address depends on the
6850 section that the address refers to. On these machines, define the macro
6851 @code{ENCODE_SECTION_INFO} to store the information into the
6852 @code{symbol_ref}, and then check for it here. @xref{Assembler Format}.
6854 @findex DBR_OUTPUT_SEQEND
6855 @findex dbr_sequence_length
6856 @item DBR_OUTPUT_SEQEND(@var{file})
6857 A C statement, to be executed after all slot-filler instructions have
6858 been output. If necessary, call @code{dbr_sequence_length} to
6859 determine the number of slots filled in a sequence (zero if not
6860 currently outputting a sequence), to decide how many no-ops to output,
6863 Don't define this macro if it has nothing to do, but it is helpful in
6864 reading assembly output if the extent of the delay sequence is made
6865 explicit (e.g.@: with white space).
6867 @findex final_sequence
6868 Note that output routines for instructions with delay slots must be
6869 prepared to deal with not being output as part of a sequence
6870 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
6871 found.) The variable @code{final_sequence} is null when not
6872 processing a sequence, otherwise it contains the @code{sequence} rtx
6875 @findex REGISTER_PREFIX
6876 @findex LOCAL_LABEL_PREFIX
6877 @findex USER_LABEL_PREFIX
6878 @findex IMMEDIATE_PREFIX
6880 @item REGISTER_PREFIX
6881 @itemx LOCAL_LABEL_PREFIX
6882 @itemx USER_LABEL_PREFIX
6883 @itemx IMMEDIATE_PREFIX
6884 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
6885 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
6886 @file{final.c}). These are useful when a single @file{md} file must
6887 support multiple assembler formats. In that case, the various @file{tm.h}
6888 files can define these macros differently.
6890 @item ASM_FPRINTF_EXTENSIONS(@var{file}, @var{argptr}, @var{format})
6891 @findex ASM_FPRINTF_EXTENSIONS
6892 If defined this macro should expand to a series of @code{case}
6893 statements which will be parsed inside the @code{switch} statement of
6894 the @code{asm_fprintf} function. This allows targets to define extra
6895 printf formats which may useful when generating their assembler
6896 statements. Note that upper case letters are reserved for future
6897 generic extensions to asm_fprintf, and so are not available to target
6898 specific code. The output file is given by the parameter @var{file}.
6899 The varargs input pointer is @var{argptr} and the rest of the format
6900 string, starting the character after the one that is being switched
6901 upon, is pointed to by @var{format}.
6903 @findex ASSEMBLER_DIALECT
6904 @item ASSEMBLER_DIALECT
6905 If your target supports multiple dialects of assembler language (such as
6906 different opcodes), define this macro as a C expression that gives the
6907 numeric index of the assembler language dialect to use, with zero as the
6910 If this macro is defined, you may use constructs of the form
6912 @samp{@{option0|option1|option2@dots{}@}}
6915 in the output templates of patterns (@pxref{Output Template}) or in the
6916 first argument of @code{asm_fprintf}. This construct outputs
6917 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
6918 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
6919 within these strings retain their usual meaning. If there are fewer
6920 alternatives within the braces than the value of
6921 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
6923 If you do not define this macro, the characters @samp{@{}, @samp{|} and
6924 @samp{@}} do not have any special meaning when used in templates or
6925 operands to @code{asm_fprintf}.
6927 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
6928 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
6929 the variations in assembler language syntax with that mechanism. Define
6930 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
6931 if the syntax variant are larger and involve such things as different
6932 opcodes or operand order.
6934 @findex ASM_OUTPUT_REG_PUSH
6935 @item ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
6936 A C expression to output to @var{stream} some assembler code
6937 which will push hard register number @var{regno} onto the stack.
6938 The code need not be optimal, since this macro is used only when
6941 @findex ASM_OUTPUT_REG_POP
6942 @item ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
6943 A C expression to output to @var{stream} some assembler code
6944 which will pop hard register number @var{regno} off of the stack.
6945 The code need not be optimal, since this macro is used only when
6949 @node Dispatch Tables
6950 @subsection Output of Dispatch Tables
6952 @c prevent bad page break with this line
6953 This concerns dispatch tables.
6956 @cindex dispatch table
6957 @findex ASM_OUTPUT_ADDR_DIFF_ELT
6958 @item ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
6959 A C statement to output to the stdio stream @var{stream} an assembler
6960 pseudo-instruction to generate a difference between two labels.
6961 @var{value} and @var{rel} are the numbers of two internal labels. The
6962 definitions of these labels are output using
6963 @code{ASM_OUTPUT_INTERNAL_LABEL}, and they must be printed in the same
6964 way here. For example,
6967 fprintf (@var{stream}, "\t.word L%d-L%d\n",
6968 @var{value}, @var{rel})
6971 You must provide this macro on machines where the addresses in a
6972 dispatch table are relative to the table's own address. If defined, GCC
6973 will also use this macro on all machines when producing PIC@.
6974 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
6975 mode and flags can be read.
6977 @findex ASM_OUTPUT_ADDR_VEC_ELT
6978 @item ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
6979 This macro should be provided on machines where the addresses
6980 in a dispatch table are absolute.
6982 The definition should be a C statement to output to the stdio stream
6983 @var{stream} an assembler pseudo-instruction to generate a reference to
6984 a label. @var{value} is the number of an internal label whose
6985 definition is output using @code{ASM_OUTPUT_INTERNAL_LABEL}.
6989 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
6992 @findex ASM_OUTPUT_CASE_LABEL
6993 @item ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
6994 Define this if the label before a jump-table needs to be output
6995 specially. The first three arguments are the same as for
6996 @code{ASM_OUTPUT_INTERNAL_LABEL}; the fourth argument is the
6997 jump-table which follows (a @code{jump_insn} containing an
6998 @code{addr_vec} or @code{addr_diff_vec}).
7000 This feature is used on system V to output a @code{swbeg} statement
7003 If this macro is not defined, these labels are output with
7004 @code{ASM_OUTPUT_INTERNAL_LABEL}.
7006 @findex ASM_OUTPUT_CASE_END
7007 @item ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
7008 Define this if something special must be output at the end of a
7009 jump-table. The definition should be a C statement to be executed
7010 after the assembler code for the table is written. It should write
7011 the appropriate code to stdio stream @var{stream}. The argument
7012 @var{table} is the jump-table insn, and @var{num} is the label-number
7013 of the preceding label.
7015 If this macro is not defined, nothing special is output at the end of
7019 @node Exception Region Output
7020 @subsection Assembler Commands for Exception Regions
7022 @c prevent bad page break with this line
7024 This describes commands marking the start and the end of an exception
7028 @findex ASM_OUTPUT_EH_REGION_BEG
7029 @item ASM_OUTPUT_EH_REGION_BEG ()
7030 A C expression to output text to mark the start of an exception region.
7032 This macro need not be defined on most platforms.
7034 @findex ASM_OUTPUT_EH_REGION_END
7035 @item ASM_OUTPUT_EH_REGION_END ()
7036 A C expression to output text to mark the end of an exception region.
7038 This macro need not be defined on most platforms.
7040 @findex EH_FRAME_SECTION_NAME
7041 @item EH_FRAME_SECTION_NAME
7042 If defined, a C string constant for the name of the section containing
7043 exception handling frame unwind information. If not defined, GCC will
7044 provide a default definition if the target supports named sections.
7045 @file{crtstuff.c} uses this macro to switch to the appropriate section.
7047 You should define this symbol if your target supports DWARF 2 frame
7048 unwind information and the default definition does not work.
7050 @findex EH_FRAME_IN_DATA_SECTION
7051 @item EH_FRAME_IN_DATA_SECTION
7052 If defined, DWARF 2 frame unwind information will be placed in the
7053 data section even though the target supports named sections. This
7054 might be necessary, for instance, if the system linker does garbage
7055 collection and sections cannot be marked as not to be collected.
7057 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
7060 @findex OMIT_EH_TABLE
7061 @item OMIT_EH_TABLE ()
7062 A C expression that is nonzero if the normal exception table output
7065 This macro need not be defined on most platforms.
7067 @findex EH_TABLE_LOOKUP
7068 @item EH_TABLE_LOOKUP ()
7069 Alternate runtime support for looking up an exception at runtime and
7070 finding the associated handler, if the default method won't work.
7072 This macro need not be defined on most platforms.
7074 @findex DOESNT_NEED_UNWINDER
7075 @item DOESNT_NEED_UNWINDER
7076 A C expression that decides whether or not the current function needs to
7077 have a function unwinder generated for it. See the file @file{except.c}
7078 for details on when to define this, and how.
7080 @findex MASK_RETURN_ADDR
7081 @item MASK_RETURN_ADDR
7082 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
7083 that it does not contain any extraneous set bits in it.
7085 @findex DWARF2_UNWIND_INFO
7086 @item DWARF2_UNWIND_INFO
7087 Define this macro to 0 if your target supports DWARF 2 frame unwind
7088 information, but it does not yet work with exception handling.
7089 Otherwise, if your target supports this information (if it defines
7090 @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP}
7091 or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of
7094 If this macro is defined to 1, the DWARF 2 unwinder will be the default
7095 exception handling mechanism; otherwise, @code{setjmp}/@code{longjmp} will be used by
7098 If this macro is defined to anything, the DWARF 2 unwinder will be used
7099 instead of inline unwinders and @code{__unwind_function} in the non-@code{setjmp} case.
7101 @findex DWARF_CIE_DATA_ALIGNMENT
7102 @item DWARF_CIE_DATA_ALIGNMENT
7103 This macro need only be defined if the target might save registers in the
7104 function prologue at an offset to the stack pointer that is not aligned to
7105 @code{UNITS_PER_WORD}. The definition should be the negative minimum
7106 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
7107 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
7108 the target supports DWARF 2 frame unwind information.
7112 @deftypefn {Target Hook} void TARGET_ASM_EXCEPTION_SECTION ()
7113 If defined, a function that switches to the section in which the main
7114 exception table is to be placed (@pxref{Sections}). The default is a
7115 function that switches to a section named @code{.gcc_except_table} on
7116 machines that support named sections via
7117 @code{TARGET_ASM_NAMED_SECTION}, otherwise if @option{-fpic} or
7118 @option{-fPIC} is in effect, the @code{data_section}, otherwise the
7119 @code{readonly_data_section}.
7122 @deftypefn {Target Hook} void TARGET_ASM_EH_FRAME_SECTION ()
7123 If defined, a function that switches to the section in which the DWARF 2
7124 frame unwind information to be placed (@pxref{Sections}). The default
7125 is a function that outputs a standard GAS section directive, if
7126 @code{EH_FRAME_SECTION_NAME} is defined, or else a data section
7127 directive followed by a synthetic label.
7130 @node Alignment Output
7131 @subsection Assembler Commands for Alignment
7133 @c prevent bad page break with this line
7134 This describes commands for alignment.
7138 @item JUMP_ALIGN (@var{label})
7139 The alignment (log base 2) to put in front of @var{label}, which is
7140 a common destination of jumps and has no fallthru incoming edge.
7142 This macro need not be defined if you don't want any special alignment
7143 to be done at such a time. Most machine descriptions do not currently
7146 Unless it's necessary to inspect the @var{label} parameter, it is better
7147 to set the variable @var{align_jumps} in the target's
7148 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7149 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
7151 @findex LABEL_ALIGN_AFTER_BARRIER
7152 @item LABEL_ALIGN_AFTER_BARRIER (@var{label})
7153 The alignment (log base 2) to put in front of @var{label}, which follows
7156 This macro need not be defined if you don't want any special alignment
7157 to be done at such a time. Most machine descriptions do not currently
7160 @findex LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
7161 @item LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
7162 The maximum number of bytes to skip when applying
7163 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
7164 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7167 @item LOOP_ALIGN (@var{label})
7168 The alignment (log base 2) to put in front of @var{label}, which follows
7169 a @code{NOTE_INSN_LOOP_BEG} note.
7171 This macro need not be defined if you don't want any special alignment
7172 to be done at such a time. Most machine descriptions do not currently
7175 Unless it's necessary to inspect the @var{label} parameter, it is better
7176 to set the variable @code{align_loops} in the target's
7177 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7178 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
7180 @findex LOOP_ALIGN_MAX_SKIP
7181 @item LOOP_ALIGN_MAX_SKIP
7182 The maximum number of bytes to skip when applying @code{LOOP_ALIGN}.
7183 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7186 @item LABEL_ALIGN (@var{label})
7187 The alignment (log base 2) to put in front of @var{label}.
7188 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
7189 the maximum of the specified values is used.
7191 Unless it's necessary to inspect the @var{label} parameter, it is better
7192 to set the variable @code{align_labels} in the target's
7193 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7194 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
7196 @findex LABEL_ALIGN_MAX_SKIP
7197 @item LABEL_ALIGN_MAX_SKIP
7198 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}.
7199 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7201 @findex ASM_OUTPUT_SKIP
7202 @item ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
7203 A C statement to output to the stdio stream @var{stream} an assembler
7204 instruction to advance the location counter by @var{nbytes} bytes.
7205 Those bytes should be zero when loaded. @var{nbytes} will be a C
7206 expression of type @code{int}.
7208 @findex ASM_NO_SKIP_IN_TEXT
7209 @item ASM_NO_SKIP_IN_TEXT
7210 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
7211 text section because it fails to put zeros in the bytes that are skipped.
7212 This is true on many Unix systems, where the pseudo--op to skip bytes
7213 produces no-op instructions rather than zeros when used in the text
7216 @findex ASM_OUTPUT_ALIGN
7217 @item ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
7218 A C statement to output to the stdio stream @var{stream} an assembler
7219 command to advance the location counter to a multiple of 2 to the
7220 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
7222 @findex ASM_OUTPUT_MAX_SKIP_ALIGN
7223 @item ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
7224 A C statement to output to the stdio stream @var{stream} an assembler
7225 command to advance the location counter to a multiple of 2 to the
7226 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
7227 satisfy the alignment request. @var{power} and @var{max_skip} will be
7228 a C expression of type @code{int}.
7232 @node Debugging Info
7233 @section Controlling Debugging Information Format
7235 @c prevent bad page break with this line
7236 This describes how to specify debugging information.
7239 * All Debuggers:: Macros that affect all debugging formats uniformly.
7240 * DBX Options:: Macros enabling specific options in DBX format.
7241 * DBX Hooks:: Hook macros for varying DBX format.
7242 * File Names and DBX:: Macros controlling output of file names in DBX format.
7243 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
7244 * VMS Debug:: Macros for VMS debug format.
7248 @subsection Macros Affecting All Debugging Formats
7250 @c prevent bad page break with this line
7251 These macros affect all debugging formats.
7254 @findex DBX_REGISTER_NUMBER
7255 @item DBX_REGISTER_NUMBER (@var{regno})
7256 A C expression that returns the DBX register number for the compiler
7257 register number @var{regno}. In the default macro provided, the value
7258 of this expression will be @var{regno} itself. But sometimes there are
7259 some registers that the compiler knows about and DBX does not, or vice
7260 versa. In such cases, some register may need to have one number in the
7261 compiler and another for DBX@.
7263 If two registers have consecutive numbers inside GCC, and they can be
7264 used as a pair to hold a multiword value, then they @emph{must} have
7265 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
7266 Otherwise, debuggers will be unable to access such a pair, because they
7267 expect register pairs to be consecutive in their own numbering scheme.
7269 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
7270 does not preserve register pairs, then what you must do instead is
7271 redefine the actual register numbering scheme.
7273 @findex DEBUGGER_AUTO_OFFSET
7274 @item DEBUGGER_AUTO_OFFSET (@var{x})
7275 A C expression that returns the integer offset value for an automatic
7276 variable having address @var{x} (an RTL expression). The default
7277 computation assumes that @var{x} is based on the frame-pointer and
7278 gives the offset from the frame-pointer. This is required for targets
7279 that produce debugging output for DBX or COFF-style debugging output
7280 for SDB and allow the frame-pointer to be eliminated when the
7281 @option{-g} options is used.
7283 @findex DEBUGGER_ARG_OFFSET
7284 @item DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
7285 A C expression that returns the integer offset value for an argument
7286 having address @var{x} (an RTL expression). The nominal offset is
7289 @findex PREFERRED_DEBUGGING_TYPE
7290 @item PREFERRED_DEBUGGING_TYPE
7291 A C expression that returns the type of debugging output GCC should
7292 produce when the user specifies just @option{-g}. Define
7293 this if you have arranged for GCC to support more than one format of
7294 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
7295 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
7296 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
7298 When the user specifies @option{-ggdb}, GCC normally also uses the
7299 value of this macro to select the debugging output format, but with two
7300 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined and
7301 @code{LINKER_DOES_NOT_WORK_WITH_DWARF2} is not defined, GCC uses the
7302 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
7303 defined, GCC uses @code{DBX_DEBUG}.
7305 The value of this macro only affects the default debugging output; the
7306 user can always get a specific type of output by using @option{-gstabs},
7307 @option{-gcoff}, @option{-gdwarf-1}, @option{-gdwarf-2}, @option{-gxcoff},
7312 @subsection Specific Options for DBX Output
7314 @c prevent bad page break with this line
7315 These are specific options for DBX output.
7318 @findex DBX_DEBUGGING_INFO
7319 @item DBX_DEBUGGING_INFO
7320 Define this macro if GCC should produce debugging output for DBX
7321 in response to the @option{-g} option.
7323 @findex XCOFF_DEBUGGING_INFO
7324 @item XCOFF_DEBUGGING_INFO
7325 Define this macro if GCC should produce XCOFF format debugging output
7326 in response to the @option{-g} option. This is a variant of DBX format.
7328 @findex DEFAULT_GDB_EXTENSIONS
7329 @item DEFAULT_GDB_EXTENSIONS
7330 Define this macro to control whether GCC should by default generate
7331 GDB's extended version of DBX debugging information (assuming DBX-format
7332 debugging information is enabled at all). If you don't define the
7333 macro, the default is 1: always generate the extended information
7334 if there is any occasion to.
7336 @findex DEBUG_SYMS_TEXT
7337 @item DEBUG_SYMS_TEXT
7338 Define this macro if all @code{.stabs} commands should be output while
7339 in the text section.
7341 @findex ASM_STABS_OP
7343 A C string constant, including spacing, naming the assembler pseudo op to
7344 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
7345 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
7346 applies only to DBX debugging information format.
7348 @findex ASM_STABD_OP
7350 A C string constant, including spacing, naming the assembler pseudo op to
7351 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
7352 value is the current location. If you don't define this macro,
7353 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
7356 @findex ASM_STABN_OP
7358 A C string constant, including spacing, naming the assembler pseudo op to
7359 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
7360 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
7361 macro applies only to DBX debugging information format.
7363 @findex DBX_NO_XREFS
7365 Define this macro if DBX on your system does not support the construct
7366 @samp{xs@var{tagname}}. On some systems, this construct is used to
7367 describe a forward reference to a structure named @var{tagname}.
7368 On other systems, this construct is not supported at all.
7370 @findex DBX_CONTIN_LENGTH
7371 @item DBX_CONTIN_LENGTH
7372 A symbol name in DBX-format debugging information is normally
7373 continued (split into two separate @code{.stabs} directives) when it
7374 exceeds a certain length (by default, 80 characters). On some
7375 operating systems, DBX requires this splitting; on others, splitting
7376 must not be done. You can inhibit splitting by defining this macro
7377 with the value zero. You can override the default splitting-length by
7378 defining this macro as an expression for the length you desire.
7380 @findex DBX_CONTIN_CHAR
7381 @item DBX_CONTIN_CHAR
7382 Normally continuation is indicated by adding a @samp{\} character to
7383 the end of a @code{.stabs} string when a continuation follows. To use
7384 a different character instead, define this macro as a character
7385 constant for the character you want to use. Do not define this macro
7386 if backslash is correct for your system.
7388 @findex DBX_STATIC_STAB_DATA_SECTION
7389 @item DBX_STATIC_STAB_DATA_SECTION
7390 Define this macro if it is necessary to go to the data section before
7391 outputting the @samp{.stabs} pseudo-op for a non-global static
7394 @findex DBX_TYPE_DECL_STABS_CODE
7395 @item DBX_TYPE_DECL_STABS_CODE
7396 The value to use in the ``code'' field of the @code{.stabs} directive
7397 for a typedef. The default is @code{N_LSYM}.
7399 @findex DBX_STATIC_CONST_VAR_CODE
7400 @item DBX_STATIC_CONST_VAR_CODE
7401 The value to use in the ``code'' field of the @code{.stabs} directive
7402 for a static variable located in the text section. DBX format does not
7403 provide any ``right'' way to do this. The default is @code{N_FUN}.
7405 @findex DBX_REGPARM_STABS_CODE
7406 @item DBX_REGPARM_STABS_CODE
7407 The value to use in the ``code'' field of the @code{.stabs} directive
7408 for a parameter passed in registers. DBX format does not provide any
7409 ``right'' way to do this. The default is @code{N_RSYM}.
7411 @findex DBX_REGPARM_STABS_LETTER
7412 @item DBX_REGPARM_STABS_LETTER
7413 The letter to use in DBX symbol data to identify a symbol as a parameter
7414 passed in registers. DBX format does not customarily provide any way to
7415 do this. The default is @code{'P'}.
7417 @findex DBX_MEMPARM_STABS_LETTER
7418 @item DBX_MEMPARM_STABS_LETTER
7419 The letter to use in DBX symbol data to identify a symbol as a stack
7420 parameter. The default is @code{'p'}.
7422 @findex DBX_FUNCTION_FIRST
7423 @item DBX_FUNCTION_FIRST
7424 Define this macro if the DBX information for a function and its
7425 arguments should precede the assembler code for the function. Normally,
7426 in DBX format, the debugging information entirely follows the assembler
7429 @findex DBX_LBRAC_FIRST
7430 @item DBX_LBRAC_FIRST
7431 Define this macro if the @code{N_LBRAC} symbol for a block should
7432 precede the debugging information for variables and functions defined in
7433 that block. Normally, in DBX format, the @code{N_LBRAC} symbol comes
7436 @findex DBX_BLOCKS_FUNCTION_RELATIVE
7437 @item DBX_BLOCKS_FUNCTION_RELATIVE
7438 Define this macro if the value of a symbol describing the scope of a
7439 block (@code{N_LBRAC} or @code{N_RBRAC}) should be relative to the start
7440 of the enclosing function. Normally, GCC uses an absolute address.
7442 @findex DBX_USE_BINCL
7444 Define this macro if GCC should generate @code{N_BINCL} and
7445 @code{N_EINCL} stabs for included header files, as on Sun systems. This
7446 macro also directs GCC to output a type number as a pair of a file
7447 number and a type number within the file. Normally, GCC does not
7448 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
7449 number for a type number.
7453 @subsection Open-Ended Hooks for DBX Format
7455 @c prevent bad page break with this line
7456 These are hooks for DBX format.
7459 @findex DBX_OUTPUT_LBRAC
7460 @item DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
7461 Define this macro to say how to output to @var{stream} the debugging
7462 information for the start of a scope level for variable names. The
7463 argument @var{name} is the name of an assembler symbol (for use with
7464 @code{assemble_name}) whose value is the address where the scope begins.
7466 @findex DBX_OUTPUT_RBRAC
7467 @item DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
7468 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
7470 @findex DBX_OUTPUT_ENUM
7471 @item DBX_OUTPUT_ENUM (@var{stream}, @var{type})
7472 Define this macro if the target machine requires special handling to
7473 output an enumeration type. The definition should be a C statement
7474 (sans semicolon) to output the appropriate information to @var{stream}
7475 for the type @var{type}.
7477 @findex DBX_OUTPUT_FUNCTION_END
7478 @item DBX_OUTPUT_FUNCTION_END (@var{stream}, @var{function})
7479 Define this macro if the target machine requires special output at the
7480 end of the debugging information for a function. The definition should
7481 be a C statement (sans semicolon) to output the appropriate information
7482 to @var{stream}. @var{function} is the @code{FUNCTION_DECL} node for
7485 @findex DBX_OUTPUT_STANDARD_TYPES
7486 @item DBX_OUTPUT_STANDARD_TYPES (@var{syms})
7487 Define this macro if you need to control the order of output of the
7488 standard data types at the beginning of compilation. The argument
7489 @var{syms} is a @code{tree} which is a chain of all the predefined
7490 global symbols, including names of data types.
7492 Normally, DBX output starts with definitions of the types for integers
7493 and characters, followed by all the other predefined types of the
7494 particular language in no particular order.
7496 On some machines, it is necessary to output different particular types
7497 first. To do this, define @code{DBX_OUTPUT_STANDARD_TYPES} to output
7498 those symbols in the necessary order. Any predefined types that you
7499 don't explicitly output will be output afterward in no particular order.
7501 Be careful not to define this macro so that it works only for C@. There
7502 are no global variables to access most of the built-in types, because
7503 another language may have another set of types. The way to output a
7504 particular type is to look through @var{syms} to see if you can find it.
7510 for (decl = syms; decl; decl = TREE_CHAIN (decl))
7511 if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)),
7513 dbxout_symbol (decl);
7519 This does nothing if the expected type does not exist.
7521 See the function @code{init_decl_processing} in @file{c-decl.c} to find
7522 the names to use for all the built-in C types.
7524 Here is another way of finding a particular type:
7526 @c this is still overfull. --mew 10feb93
7530 for (decl = syms; decl; decl = TREE_CHAIN (decl))
7531 if (TREE_CODE (decl) == TYPE_DECL
7532 && (TREE_CODE (TREE_TYPE (decl))
7534 && TYPE_PRECISION (TREE_TYPE (decl)) == 16
7535 && TYPE_UNSIGNED (TREE_TYPE (decl)))
7537 /* @r{This must be @code{unsigned short}.} */
7538 dbxout_symbol (decl);
7544 @findex NO_DBX_FUNCTION_END
7545 @item NO_DBX_FUNCTION_END
7546 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
7547 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
7548 On those machines, define this macro to turn this feature off without
7549 disturbing the rest of the gdb extensions.
7553 @node File Names and DBX
7554 @subsection File Names in DBX Format
7556 @c prevent bad page break with this line
7557 This describes file names in DBX format.
7560 @findex DBX_WORKING_DIRECTORY
7561 @item DBX_WORKING_DIRECTORY
7562 Define this if DBX wants to have the current directory recorded in each
7565 Note that the working directory is always recorded if GDB extensions are
7568 @findex DBX_OUTPUT_MAIN_SOURCE_FILENAME
7569 @item DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
7570 A C statement to output DBX debugging information to the stdio stream
7571 @var{stream} which indicates that file @var{name} is the main source
7572 file---the file specified as the input file for compilation.
7573 This macro is called only once, at the beginning of compilation.
7575 This macro need not be defined if the standard form of output
7576 for DBX debugging information is appropriate.
7578 @findex DBX_OUTPUT_MAIN_SOURCE_DIRECTORY
7579 @item DBX_OUTPUT_MAIN_SOURCE_DIRECTORY (@var{stream}, @var{name})
7580 A C statement to output DBX debugging information to the stdio stream
7581 @var{stream} which indicates that the current directory during
7582 compilation is named @var{name}.
7584 This macro need not be defined if the standard form of output
7585 for DBX debugging information is appropriate.
7587 @findex DBX_OUTPUT_MAIN_SOURCE_FILE_END
7588 @item DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
7589 A C statement to output DBX debugging information at the end of
7590 compilation of the main source file @var{name}.
7592 If you don't define this macro, nothing special is output at the end
7593 of compilation, which is correct for most machines.
7595 @findex DBX_OUTPUT_SOURCE_FILENAME
7596 @item DBX_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7597 A C statement to output DBX debugging information to the stdio stream
7598 @var{stream} which indicates that file @var{name} is the current source
7599 file. This output is generated each time input shifts to a different
7600 source file as a result of @samp{#include}, the end of an included file,
7601 or a @samp{#line} command.
7603 This macro need not be defined if the standard form of output
7604 for DBX debugging information is appropriate.
7609 @subsection Macros for SDB and DWARF Output
7611 @c prevent bad page break with this line
7612 Here are macros for SDB and DWARF output.
7615 @findex SDB_DEBUGGING_INFO
7616 @item SDB_DEBUGGING_INFO
7617 Define this macro if GCC should produce COFF-style debugging output
7618 for SDB in response to the @option{-g} option.
7620 @findex DWARF_DEBUGGING_INFO
7621 @item DWARF_DEBUGGING_INFO
7622 Define this macro if GCC should produce dwarf format debugging output
7623 in response to the @option{-g} option.
7625 @findex DWARF2_DEBUGGING_INFO
7626 @item DWARF2_DEBUGGING_INFO
7627 Define this macro if GCC should produce dwarf version 2 format
7628 debugging output in response to the @option{-g} option.
7630 To support optional call frame debugging information, you must also
7631 define @code{INCOMING_RETURN_ADDR_RTX} and either set
7632 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
7633 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
7634 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
7636 @findex DWARF2_FRAME_INFO
7637 @item DWARF2_FRAME_INFO
7638 Define this macro to a nonzero value if GCC should always output
7639 Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO}
7640 (@pxref{Exception Region Output} is nonzero, GCC will output this
7641 information not matter how you define @code{DWARF2_FRAME_INFO}.
7643 @findex LINKER_DOES_NOT_WORK_WITH_DWARF2
7644 @item LINKER_DOES_NOT_WORK_WITH_DWARF2
7645 Define this macro if the linker does not work with Dwarf version 2.
7646 Normally, if the user specifies only @option{-ggdb} GCC will use Dwarf
7647 version 2 if available; this macro disables this. See the description
7648 of the @code{PREFERRED_DEBUGGING_TYPE} macro for more details.
7650 @findex DWARF2_GENERATE_TEXT_SECTION_LABEL
7651 @item DWARF2_GENERATE_TEXT_SECTION_LABEL
7652 By default, the Dwarf 2 debugging information generator will generate a
7653 label to mark the beginning of the text section. If it is better simply
7654 to use the name of the text section itself, rather than an explicit label,
7655 to indicate the beginning of the text section, define this macro to zero.
7657 @findex DWARF2_ASM_LINE_DEBUG_INFO
7658 @item DWARF2_ASM_LINE_DEBUG_INFO
7659 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
7660 line debug info sections. This will result in much more compact line number
7661 tables, and hence is desirable if it works.
7663 @findex PUT_SDB_@dots{}
7664 @item PUT_SDB_@dots{}
7665 Define these macros to override the assembler syntax for the special
7666 SDB assembler directives. See @file{sdbout.c} for a list of these
7667 macros and their arguments. If the standard syntax is used, you need
7668 not define them yourself.
7672 Some assemblers do not support a semicolon as a delimiter, even between
7673 SDB assembler directives. In that case, define this macro to be the
7674 delimiter to use (usually @samp{\n}). It is not necessary to define
7675 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
7678 @findex SDB_GENERATE_FAKE
7679 @item SDB_GENERATE_FAKE
7680 Define this macro to override the usual method of constructing a dummy
7681 name for anonymous structure and union types. See @file{sdbout.c} for
7684 @findex SDB_ALLOW_UNKNOWN_REFERENCES
7685 @item SDB_ALLOW_UNKNOWN_REFERENCES
7686 Define this macro to allow references to unknown structure,
7687 union, or enumeration tags to be emitted. Standard COFF does not
7688 allow handling of unknown references, MIPS ECOFF has support for
7691 @findex SDB_ALLOW_FORWARD_REFERENCES
7692 @item SDB_ALLOW_FORWARD_REFERENCES
7693 Define this macro to allow references to structure, union, or
7694 enumeration tags that have not yet been seen to be handled. Some
7695 assemblers choke if forward tags are used, while some require it.
7700 @subsection Macros for VMS Debug Format
7702 @c prevent bad page break with this line
7703 Here are macros for VMS debug format.
7706 @findex VMS_DEBUGGING_INFO
7707 @item VMS_DEBUGGING_INFO
7708 Define this macro if GCC should produce debugging output for VMS
7709 in response to the @option{-g} option. The default behavior for VMS
7710 is to generate minimal debug info for a traceback in the absence of
7711 @option{-g} unless explicitly overridden with @option{-g0}. This
7712 behavior is controlled by @code{OPTIMIZATION_OPTIONS} and
7713 @code{OVERRIDE_OPTIONS}.
7716 @node Cross-compilation
7717 @section Cross Compilation and Floating Point
7718 @cindex cross compilation and floating point
7719 @cindex floating point and cross compilation
7721 While all modern machines use 2's complement representation for integers,
7722 there are a variety of representations for floating point numbers. This
7723 means that in a cross-compiler the representation of floating point numbers
7724 in the compiled program may be different from that used in the machine
7725 doing the compilation.
7728 Because different representation systems may offer different amounts of
7729 range and precision, the cross compiler cannot safely use the host
7730 machine's floating point arithmetic. Therefore, floating point constants
7731 must be represented in the target machine's format. This means that the
7732 cross compiler cannot use @code{atof} to parse a floating point constant;
7733 it must have its own special routine to use instead. Also, constant
7734 folding must emulate the target machine's arithmetic (or must not be done
7737 The macros in the following table should be defined only if you are cross
7738 compiling between different floating point formats.
7740 Otherwise, don't define them. Then default definitions will be set up which
7741 use @code{double} as the data type, @code{==} to test for equality, etc.
7743 You don't need to worry about how many times you use an operand of any
7744 of these macros. The compiler never uses operands which have side effects.
7747 @findex REAL_VALUE_TYPE
7748 @item REAL_VALUE_TYPE
7749 A macro for the C data type to be used to hold a floating point value
7750 in the target machine's format. Typically this would be a
7751 @code{struct} containing an array of @code{int}.
7753 @findex REAL_VALUES_EQUAL
7754 @item REAL_VALUES_EQUAL (@var{x}, @var{y})
7755 A macro for a C expression which compares for equality the two values,
7756 @var{x} and @var{y}, both of type @code{REAL_VALUE_TYPE}.
7758 @findex REAL_VALUES_LESS
7759 @item REAL_VALUES_LESS (@var{x}, @var{y})
7760 A macro for a C expression which tests whether @var{x} is less than
7761 @var{y}, both values being of type @code{REAL_VALUE_TYPE} and
7762 interpreted as floating point numbers in the target machine's
7765 @findex REAL_VALUE_LDEXP
7767 @item REAL_VALUE_LDEXP (@var{x}, @var{scale})
7768 A macro for a C expression which performs the standard library
7769 function @code{ldexp}, but using the target machine's floating point
7770 representation. Both @var{x} and the value of the expression have
7771 type @code{REAL_VALUE_TYPE}. The second argument, @var{scale}, is an
7774 @findex REAL_VALUE_FIX
7775 @item REAL_VALUE_FIX (@var{x})
7776 A macro whose definition is a C expression to convert the target-machine
7777 floating point value @var{x} to a signed integer. @var{x} has type
7778 @code{REAL_VALUE_TYPE}.
7780 @findex REAL_VALUE_UNSIGNED_FIX
7781 @item REAL_VALUE_UNSIGNED_FIX (@var{x})
7782 A macro whose definition is a C expression to convert the target-machine
7783 floating point value @var{x} to an unsigned integer. @var{x} has type
7784 @code{REAL_VALUE_TYPE}.
7786 @findex REAL_VALUE_RNDZINT
7787 @item REAL_VALUE_RNDZINT (@var{x})
7788 A macro whose definition is a C expression to round the target-machine
7789 floating point value @var{x} towards zero to an integer value (but still
7790 as a floating point number). @var{x} has type @code{REAL_VALUE_TYPE},
7791 and so does the value.
7793 @findex REAL_VALUE_UNSIGNED_RNDZINT
7794 @item REAL_VALUE_UNSIGNED_RNDZINT (@var{x})
7795 A macro whose definition is a C expression to round the target-machine
7796 floating point value @var{x} towards zero to an unsigned integer value
7797 (but still represented as a floating point number). @var{x} has type
7798 @code{REAL_VALUE_TYPE}, and so does the value.
7800 @findex REAL_VALUE_ATOF
7801 @item REAL_VALUE_ATOF (@var{string}, @var{mode})
7802 A macro for a C expression which converts @var{string}, an expression of
7803 type @code{char *}, into a floating point number in the target machine's
7804 representation for mode @var{mode}. The value has type
7805 @code{REAL_VALUE_TYPE}.
7807 @findex REAL_INFINITY
7809 Define this macro if infinity is a possible floating point value, and
7810 therefore division by 0 is legitimate.
7812 @findex REAL_VALUE_ISINF
7814 @item REAL_VALUE_ISINF (@var{x})
7815 A macro for a C expression which determines whether @var{x}, a floating
7816 point value, is infinity. The value has type @code{int}.
7817 By default, this is defined to call @code{isinf}.
7819 @findex REAL_VALUE_ISNAN
7821 @item REAL_VALUE_ISNAN (@var{x})
7822 A macro for a C expression which determines whether @var{x}, a floating
7823 point value, is a ``nan'' (not-a-number). The value has type
7824 @code{int}. By default, this is defined to call @code{isnan}.
7827 @cindex constant folding and floating point
7828 Define the following additional macros if you want to make floating
7829 point constant folding work while cross compiling. If you don't
7830 define them, cross compilation is still possible, but constant folding
7831 will not happen for floating point values.
7834 @findex REAL_ARITHMETIC
7835 @item REAL_ARITHMETIC (@var{output}, @var{code}, @var{x}, @var{y})
7836 A macro for a C statement which calculates an arithmetic operation of
7837 the two floating point values @var{x} and @var{y}, both of type
7838 @code{REAL_VALUE_TYPE} in the target machine's representation, to
7839 produce a result of the same type and representation which is stored
7840 in @var{output} (which will be a variable).
7842 The operation to be performed is specified by @var{code}, a tree code
7843 which will always be one of the following: @code{PLUS_EXPR},
7844 @code{MINUS_EXPR}, @code{MULT_EXPR}, @code{RDIV_EXPR},
7845 @code{MAX_EXPR}, @code{MIN_EXPR}.
7847 @cindex overflow while constant folding
7848 The expansion of this macro is responsible for checking for overflow.
7849 If overflow happens, the macro expansion should execute the statement
7850 @code{return 0;}, which indicates the inability to perform the
7851 arithmetic operation requested.
7853 @findex REAL_VALUE_NEGATE
7854 @item REAL_VALUE_NEGATE (@var{x})
7855 A macro for a C expression which returns the negative of the floating
7856 point value @var{x}. Both @var{x} and the value of the expression
7857 have type @code{REAL_VALUE_TYPE} and are in the target machine's
7858 floating point representation.
7860 There is no way for this macro to report overflow, since overflow
7861 can't happen in the negation operation.
7863 @findex REAL_VALUE_TRUNCATE
7864 @item REAL_VALUE_TRUNCATE (@var{mode}, @var{x})
7865 A macro for a C expression which converts the floating point value
7866 @var{x} to mode @var{mode}.
7868 Both @var{x} and the value of the expression are in the target machine's
7869 floating point representation and have type @code{REAL_VALUE_TYPE}.
7870 However, the value should have an appropriate bit pattern to be output
7871 properly as a floating constant whose precision accords with mode
7874 There is no way for this macro to report overflow.
7876 @findex REAL_VALUE_TO_INT
7877 @item REAL_VALUE_TO_INT (@var{low}, @var{high}, @var{x})
7878 A macro for a C expression which converts a floating point value
7879 @var{x} into a double-precision integer which is then stored into
7880 @var{low} and @var{high}, two variables of type @var{int}.
7882 @item REAL_VALUE_FROM_INT (@var{x}, @var{low}, @var{high}, @var{mode})
7883 @findex REAL_VALUE_FROM_INT
7884 A macro for a C expression which converts a double-precision integer
7885 found in @var{low} and @var{high}, two variables of type @var{int},
7886 into a floating point value which is then stored into @var{x}.
7887 The value is in the target machine's representation for mode @var{mode}
7888 and has the type @code{REAL_VALUE_TYPE}.
7891 @node Mode Switching
7892 @section Mode Switching Instructions
7893 @cindex mode switching
7894 The following macros control mode switching optimizations:
7897 @findex OPTIMIZE_MODE_SWITCHING
7898 @item OPTIMIZE_MODE_SWITCHING (@var{entity})
7899 Define this macro if the port needs extra instructions inserted for mode
7900 switching in an optimizing compilation.
7902 For an example, the SH4 can perform both single and double precision
7903 floating point operations, but to perform a single precision operation,
7904 the FPSCR PR bit has to be cleared, while for a double precision
7905 operation, this bit has to be set. Changing the PR bit requires a general
7906 purpose register as a scratch register, hence these FPSCR sets have to
7907 be inserted before reload, i.e.@: you can't put this into instruction emitting
7908 or @code{MACHINE_DEPENDENT_REORG}.
7910 You can have multiple entities that are mode-switched, and select at run time
7911 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
7912 return nonzero for any @var{entity} that needs mode-switching.
7913 If you define this macro, you also have to define
7914 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
7915 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
7916 @code{NORMAL_MODE} is optional.
7918 @findex NUM_MODES_FOR_MODE_SWITCHING
7919 @item NUM_MODES_FOR_MODE_SWITCHING
7920 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
7921 initializer for an array of integers. Each initializer element
7922 N refers to an entity that needs mode switching, and specifies the number
7923 of different modes that might need to be set for this entity.
7924 The position of the initializer in the initializer - starting counting at
7925 zero - determines the integer that is used to refer to the mode-switched
7927 In macros that take mode arguments / yield a mode result, modes are
7928 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
7929 switch is needed / supplied.
7932 @item MODE_NEEDED (@var{entity}, @var{insn})
7933 @var{entity} is an integer specifying a mode-switched entity. If
7934 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
7935 return an integer value not larger than the corresponding element in
7936 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
7937 be switched into prior to the execution of @var{insn}.
7940 @item NORMAL_MODE (@var{entity})
7941 If this macro is defined, it is evaluated for every @var{entity} that needs
7942 mode switching. It should evaluate to an integer, which is a mode that
7943 @var{entity} is assumed to be switched to at function entry and exit.
7945 @findex MODE_PRIORITY_TO_MODE
7946 @item MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
7947 This macro specifies the order in which modes for @var{entity} are processed.
7948 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
7949 lowest. The value of the macro should be an integer designating a mode
7950 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
7951 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
7952 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
7954 @findex EMIT_MODE_SET
7955 @item EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
7956 Generate one or more insns to set @var{entity} to @var{mode}.
7957 @var{hard_reg_live} is the set of hard registers live at the point where
7958 the insn(s) are to be inserted.
7961 @node Target Attributes
7962 @section Defining target-specific uses of @code{__attribute__}
7963 @cindex target attributes
7964 @cindex machine attributes
7965 @cindex attributes, target-specific
7967 Target-specific attributes may be defined for functions, data and types.
7968 These are described using the following target hooks; they also need to
7969 be documented in @file{extend.texi}.
7971 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
7972 If defined, this target hook points to an array of @samp{struct
7973 attribute_spec} (defined in @file{tree.h}) specifying the machine
7974 specific attributes for this target and some of the restrictions on the
7975 entities to which these attributes are applied and the arguments they
7979 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
7980 If defined, this target hook is a function which returns zero if the attributes on
7981 @var{type1} and @var{type2} are incompatible, one if they are compatible,
7982 and two if they are nearly compatible (which causes a warning to be
7983 generated). If this is not defined, machine-specific attributes are
7984 supposed always to be compatible.
7987 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
7988 If defined, this target hook is a function which assigns default attributes to
7989 newly defined @var{type}.
7992 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
7993 Define this target hook if the merging of type attributes needs special
7994 handling. If defined, the result is a list of the combined
7995 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
7996 that @code{comptypes} has already been called and returned 1. This
7997 function may call @code{merge_attributes} to handle machine-independent
8001 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
8002 Define this target hook if the merging of decl attributes needs special
8003 handling. If defined, the result is a list of the combined
8004 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
8005 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
8006 when this is needed are when one attribute overrides another, or when an
8007 attribute is nullified by a subsequent definition. This function may
8008 call @code{merge_attributes} to handle machine-independent merging.
8010 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
8011 If the only target-specific handling you require is @samp{dllimport} for
8012 Windows targets, you should define the macro
8013 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. This links in a function
8014 called @code{merge_dllimport_decl_attributes} which can then be defined
8015 as the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. This is done
8016 in @file{i386/cygwin.h} and @file{i386/i386.c}, for example.
8019 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
8020 Define this target hook if you want to be able to add attributes to a decl
8021 when it is being created. This is normally useful for back ends which
8022 wish to implement a pragma by using the attributes which correspond to
8023 the pragma's effect. The @var{node} argument is the decl which is being
8024 created. The @var{attr_ptr} argument is a pointer to the attribute list
8025 for this decl. The list itself should not be modified, since it may be
8026 shared with other decls, but attributes may be chained on the head of
8027 the list and @code{*@var{attr_ptr}} modified to point to the new
8028 attributes, or a copy of the list may be made if further changes are
8032 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (tree @var{fndecl})
8034 This target hook returns @code{true} if it is ok to inline @var{fndecl}
8035 into the current function, despite its having target-specific
8036 attributes, @code{false} otherwise. By default, if a function has a
8037 target specific attribute attached to it, it will not be inlined.
8041 @section Miscellaneous Parameters
8042 @cindex parameters, miscellaneous
8044 @c prevent bad page break with this line
8045 Here are several miscellaneous parameters.
8048 @item PREDICATE_CODES
8049 @findex PREDICATE_CODES
8050 Define this if you have defined special-purpose predicates in the file
8051 @file{@var{machine}.c}. This macro is called within an initializer of an
8052 array of structures. The first field in the structure is the name of a
8053 predicate and the second field is an array of rtl codes. For each
8054 predicate, list all rtl codes that can be in expressions matched by the
8055 predicate. The list should have a trailing comma. Here is an example
8056 of two entries in the list for a typical RISC machine:
8059 #define PREDICATE_CODES \
8060 @{"gen_reg_rtx_operand", @{SUBREG, REG@}@}, \
8061 @{"reg_or_short_cint_operand", @{SUBREG, REG, CONST_INT@}@},
8064 Defining this macro does not affect the generated code (however,
8065 incorrect definitions that omit an rtl code that may be matched by the
8066 predicate can cause the compiler to malfunction). Instead, it allows
8067 the table built by @file{genrecog} to be more compact and efficient,
8068 thus speeding up the compiler. The most important predicates to include
8069 in the list specified by this macro are those used in the most insn
8072 For each predicate function named in @code{PREDICATE_CODES}, a
8073 declaration will be generated in @file{insn-codes.h}.
8075 @item SPECIAL_MODE_PREDICATES
8076 @findex SPECIAL_MODE_PREDICATES
8077 Define this if you have special predicates that know special things
8078 about modes. Genrecog will warn about certain forms of
8079 @code{match_operand} without a mode; if the operand predicate is
8080 listed in @code{SPECIAL_MODE_PREDICATES}, the warning will be
8083 Here is an example from the IA-32 port (@code{ext_register_operand}
8084 specially checks for @code{HImode} or @code{SImode} in preparation
8085 for a byte extraction from @code{%ah} etc.).
8088 #define SPECIAL_MODE_PREDICATES \
8089 "ext_register_operand",
8092 @findex CASE_VECTOR_MODE
8093 @item CASE_VECTOR_MODE
8094 An alias for a machine mode name. This is the machine mode that
8095 elements of a jump-table should have.
8097 @findex CASE_VECTOR_SHORTEN_MODE
8098 @item CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
8099 Optional: return the preferred mode for an @code{addr_diff_vec}
8100 when the minimum and maximum offset are known. If you define this,
8101 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
8102 To make this work, you also have to define INSN_ALIGN and
8103 make the alignment for @code{addr_diff_vec} explicit.
8104 The @var{body} argument is provided so that the offset_unsigned and scale
8105 flags can be updated.
8107 @findex CASE_VECTOR_PC_RELATIVE
8108 @item CASE_VECTOR_PC_RELATIVE
8109 Define this macro to be a C expression to indicate when jump-tables
8110 should contain relative addresses. If jump-tables never contain
8111 relative addresses, then you need not define this macro.
8113 @findex CASE_DROPS_THROUGH
8114 @item CASE_DROPS_THROUGH
8115 Define this if control falls through a @code{case} insn when the index
8116 value is out of range. This means the specified default-label is
8117 actually ignored by the @code{case} insn proper.
8119 @findex CASE_VALUES_THRESHOLD
8120 @item CASE_VALUES_THRESHOLD
8121 Define this to be the smallest number of different values for which it
8122 is best to use a jump-table instead of a tree of conditional branches.
8123 The default is four for machines with a @code{casesi} instruction and
8124 five otherwise. This is best for most machines.
8126 @findex WORD_REGISTER_OPERATIONS
8127 @item WORD_REGISTER_OPERATIONS
8128 Define this macro if operations between registers with integral mode
8129 smaller than a word are always performed on the entire register.
8130 Most RISC machines have this property and most CISC machines do not.
8132 @findex LOAD_EXTEND_OP
8133 @item LOAD_EXTEND_OP (@var{mode})
8134 Define this macro to be a C expression indicating when insns that read
8135 memory in @var{mode}, an integral mode narrower than a word, set the
8136 bits outside of @var{mode} to be either the sign-extension or the
8137 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
8138 of @var{mode} for which the
8139 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
8140 @code{NIL} for other modes.
8142 This macro is not called with @var{mode} non-integral or with a width
8143 greater than or equal to @code{BITS_PER_WORD}, so you may return any
8144 value in this case. Do not define this macro if it would always return
8145 @code{NIL}. On machines where this macro is defined, you will normally
8146 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
8148 @findex SHORT_IMMEDIATES_SIGN_EXTEND
8149 @item SHORT_IMMEDIATES_SIGN_EXTEND
8150 Define this macro if loading short immediate values into registers sign
8153 @findex IMPLICIT_FIX_EXPR
8154 @item IMPLICIT_FIX_EXPR
8155 An alias for a tree code that should be used by default for conversion
8156 of floating point values to fixed point. Normally,
8157 @code{FIX_ROUND_EXPR} is used.
8159 @findex FIXUNS_TRUNC_LIKE_FIX_TRUNC
8160 @item FIXUNS_TRUNC_LIKE_FIX_TRUNC
8161 Define this macro if the same instructions that convert a floating
8162 point number to a signed fixed point number also convert validly to an
8165 @findex EASY_DIV_EXPR
8167 An alias for a tree code that is the easiest kind of division to
8168 compile code for in the general case. It may be
8169 @code{TRUNC_DIV_EXPR}, @code{FLOOR_DIV_EXPR}, @code{CEIL_DIV_EXPR} or
8170 @code{ROUND_DIV_EXPR}. These four division operators differ in how
8171 they round the result to an integer. @code{EASY_DIV_EXPR} is used
8172 when it is permissible to use any of those kinds of division and the
8173 choice should be made on the basis of efficiency.
8177 The maximum number of bytes that a single instruction can move quickly
8178 between memory and registers or between two memory locations.
8180 @findex MAX_MOVE_MAX
8182 The maximum number of bytes that a single instruction can move quickly
8183 between memory and registers or between two memory locations. If this
8184 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
8185 constant value that is the largest value that @code{MOVE_MAX} can have
8188 @findex SHIFT_COUNT_TRUNCATED
8189 @item SHIFT_COUNT_TRUNCATED
8190 A C expression that is nonzero if on this machine the number of bits
8191 actually used for the count of a shift operation is equal to the number
8192 of bits needed to represent the size of the object being shifted. When
8193 this macro is nonzero, the compiler will assume that it is safe to omit
8194 a sign-extend, zero-extend, and certain bitwise `and' instructions that
8195 truncates the count of a shift operation. On machines that have
8196 instructions that act on bit-fields at variable positions, which may
8197 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
8198 also enables deletion of truncations of the values that serve as
8199 arguments to bit-field instructions.
8201 If both types of instructions truncate the count (for shifts) and
8202 position (for bit-field operations), or if no variable-position bit-field
8203 instructions exist, you should define this macro.
8205 However, on some machines, such as the 80386 and the 680x0, truncation
8206 only applies to shift operations and not the (real or pretended)
8207 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
8208 such machines. Instead, add patterns to the @file{md} file that include
8209 the implied truncation of the shift instructions.
8211 You need not define this macro if it would always have the value of zero.
8213 @findex TRULY_NOOP_TRUNCATION
8214 @item TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
8215 A C expression which is nonzero if on this machine it is safe to
8216 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
8217 bits (where @var{outprec} is smaller than @var{inprec}) by merely
8218 operating on it as if it had only @var{outprec} bits.
8220 On many machines, this expression can be 1.
8222 @c rearranged this, removed the phrase "it is reported that". this was
8223 @c to fix an overfull hbox. --mew 10feb93
8224 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
8225 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
8226 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
8227 such cases may improve things.
8229 @findex STORE_FLAG_VALUE
8230 @item STORE_FLAG_VALUE
8231 A C expression describing the value returned by a comparison operator
8232 with an integral mode and stored by a store-flag instruction
8233 (@samp{s@var{cond}}) when the condition is true. This description must
8234 apply to @emph{all} the @samp{s@var{cond}} patterns and all the
8235 comparison operators whose results have a @code{MODE_INT} mode.
8237 A value of 1 or @minus{}1 means that the instruction implementing the
8238 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
8239 and 0 when the comparison is false. Otherwise, the value indicates
8240 which bits of the result are guaranteed to be 1 when the comparison is
8241 true. This value is interpreted in the mode of the comparison
8242 operation, which is given by the mode of the first operand in the
8243 @samp{s@var{cond}} pattern. Either the low bit or the sign bit of
8244 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
8247 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
8248 generate code that depends only on the specified bits. It can also
8249 replace comparison operators with equivalent operations if they cause
8250 the required bits to be set, even if the remaining bits are undefined.
8251 For example, on a machine whose comparison operators return an
8252 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
8253 @samp{0x80000000}, saying that just the sign bit is relevant, the
8257 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
8264 (ashift:SI @var{x} (const_int @var{n}))
8268 where @var{n} is the appropriate shift count to move the bit being
8269 tested into the sign bit.
8271 There is no way to describe a machine that always sets the low-order bit
8272 for a true value, but does not guarantee the value of any other bits,
8273 but we do not know of any machine that has such an instruction. If you
8274 are trying to port GCC to such a machine, include an instruction to
8275 perform a logical-and of the result with 1 in the pattern for the
8276 comparison operators and let us know
8278 (@pxref{Bug Reporting,,How to Report Bugs}).
8281 (@pxref{Bug Reporting,,How to Report Bugs,gcc.info,Using GCC}).
8284 Often, a machine will have multiple instructions that obtain a value
8285 from a comparison (or the condition codes). Here are rules to guide the
8286 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
8291 Use the shortest sequence that yields a valid definition for
8292 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
8293 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
8294 comparison operators to do so because there may be opportunities to
8295 combine the normalization with other operations.
8298 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
8299 slightly preferred on machines with expensive jumps and 1 preferred on
8303 As a second choice, choose a value of @samp{0x80000001} if instructions
8304 exist that set both the sign and low-order bits but do not define the
8308 Otherwise, use a value of @samp{0x80000000}.
8311 Many machines can produce both the value chosen for
8312 @code{STORE_FLAG_VALUE} and its negation in the same number of
8313 instructions. On those machines, you should also define a pattern for
8314 those cases, e.g., one matching
8317 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
8320 Some machines can also perform @code{and} or @code{plus} operations on
8321 condition code values with less instructions than the corresponding
8322 @samp{s@var{cond}} insn followed by @code{and} or @code{plus}. On those
8323 machines, define the appropriate patterns. Use the names @code{incscc}
8324 and @code{decscc}, respectively, for the patterns which perform
8325 @code{plus} or @code{minus} operations on condition code values. See
8326 @file{rs6000.md} for some examples. The GNU Superoptizer can be used to
8327 find such instruction sequences on other machines.
8329 You need not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
8332 @findex FLOAT_STORE_FLAG_VALUE
8333 @item FLOAT_STORE_FLAG_VALUE (@var{mode})
8334 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
8335 returned when comparison operators with floating-point results are true.
8336 Define this macro on machine that have comparison operations that return
8337 floating-point values. If there are no such operations, do not define
8342 An alias for the machine mode for pointers. On most machines, define
8343 this to be the integer mode corresponding to the width of a hardware
8344 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
8345 On some machines you must define this to be one of the partial integer
8346 modes, such as @code{PSImode}.
8348 The width of @code{Pmode} must be at least as large as the value of
8349 @code{POINTER_SIZE}. If it is not equal, you must define the macro
8350 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
8353 @findex FUNCTION_MODE
8355 An alias for the machine mode used for memory references to functions
8356 being called, in @code{call} RTL expressions. On most machines this
8357 should be @code{QImode}.
8359 @findex INTEGRATE_THRESHOLD
8360 @item INTEGRATE_THRESHOLD (@var{decl})
8361 A C expression for the maximum number of instructions above which the
8362 function @var{decl} should not be inlined. @var{decl} is a
8363 @code{FUNCTION_DECL} node.
8365 The default definition of this macro is 64 plus 8 times the number of
8366 arguments that the function accepts. Some people think a larger
8367 threshold should be used on RISC machines.
8369 @findex STDC_0_IN_SYSTEM_HEADERS
8370 @item STDC_0_IN_SYSTEM_HEADERS
8371 In normal operation, the preprocessor expands @code{__STDC__} to the
8372 constant 1, to signify that GCC conforms to ISO Standard C@. On some
8373 hosts, like Solaris, the system compiler uses a different convention,
8374 where @code{__STDC__} is normally 0, but is 1 if the user specifies
8375 strict conformance to the C Standard.
8377 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
8378 convention when processing system header files, but when processing user
8379 files @code{__STDC__} will always expand to 1.
8381 @findex SCCS_DIRECTIVE
8382 @item SCCS_DIRECTIVE
8383 Define this if the preprocessor should ignore @code{#sccs} directives
8384 and print no error message.
8386 @findex NO_IMPLICIT_EXTERN_C
8387 @item NO_IMPLICIT_EXTERN_C
8388 Define this macro if the system header files support C++ as well as C@.
8389 This macro inhibits the usual method of using system header files in
8390 C++, which is to pretend that the file's contents are enclosed in
8391 @samp{extern "C" @{@dots{}@}}.
8393 @findex HANDLE_PRAGMA
8394 @item HANDLE_PRAGMA (@var{getc}, @var{ungetc}, @var{name})
8395 This macro is no longer supported. You must use
8396 @code{REGISTER_TARGET_PRAGMAS} instead.
8398 @findex REGISTER_TARGET_PRAGMAS
8401 @item REGISTER_TARGET_PRAGMAS (@var{pfile})
8402 Define this macro if you want to implement any target-specific pragmas.
8403 If defined, it is a C expression which makes a series of calls to
8404 @code{cpp_register_pragma} for each pragma, with @var{pfile} passed as
8405 the first argument to to these functions. The macro may also do any
8406 setup required for the pragmas.
8408 The primary reason to define this macro is to provide compatibility with
8409 other compilers for the same target. In general, we discourage
8410 definition of target-specific pragmas for GCC@.
8412 If the pragma can be implemented by attributes then you should consider
8413 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
8415 Preprocessor macros that appear on pragma lines are not expanded. All
8416 @samp{#pragma} directives that do not match any registered pragma are
8417 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
8419 @deftypefun void cpp_register_pragma (cpp_reader *@var{pfile}, const char *@var{space}, const char *@var{name}, void (*@var{callback}) (cpp_reader *))
8421 Each call to @code{cpp_register_pragma} establishes one pragma. The
8422 @var{callback} routine will be called when the preprocessor encounters a
8426 #pragma [@var{space}] @var{name} @dots{}
8429 @var{space} is the case-sensitive namespace of the pragma, or
8430 @code{NULL} to put the pragma in the global namespace. The callback
8431 routine receives @var{pfile} as its first argument, which can be passed
8432 on to cpplib's functions if necessary. You can lex tokens after the
8433 @var{name} by calling @code{c_lex}. Tokens that are not read by the
8434 callback will be silently ignored. The end of the line is indicated by
8435 a token of type @code{CPP_EOF}.
8437 For an example use of this routine, see @file{c4x.h} and the callback
8438 routines defined in @file{c4x-c.c}.
8440 Note that the use of @code{c_lex} is specific to the C and C++
8441 compilers. It will not work in the Java or Fortran compilers, or any
8442 other language compilers for that matter. Thus if @code{c_lex} is going
8443 to be called from target-specific code, it must only be done so when
8444 building the C and C++ compilers. This can be done by defining the
8445 variables @code{c_target_objs} and @code{cxx_target_objs} in the
8446 target entry in the @file{config.gcc} file. These variables should name
8447 the target-specific, language-specific object file which contains the
8448 code that uses @code{c_lex}. Note it will also be necessary to add a
8449 rule to the makefile fragment pointed to by @code{tmake_file} that shows
8450 how to build this object file.
8453 @findex HANDLE_SYSV_PRAGMA
8456 @item HANDLE_SYSV_PRAGMA
8457 Define this macro (to a value of 1) if you want the System V style
8458 pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name>
8459 [=<value>]} to be supported by gcc.
8461 The pack pragma specifies the maximum alignment (in bytes) of fields
8462 within a structure, in much the same way as the @samp{__aligned__} and
8463 @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets
8464 the behavior to the default.
8466 The weak pragma only works if @code{SUPPORTS_WEAK} and
8467 @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation
8468 of specifically named weak labels, optionally with a value.
8470 @findex HANDLE_PRAGMA_PACK_PUSH_POP
8473 @item HANDLE_PRAGMA_PACK_PUSH_POP
8474 Define this macro (to a value of 1) if you want to support the Win32
8475 style pragmas @samp{#pragma pack(push,@var{n})} and @samp{#pragma
8476 pack(pop)}. The @samp{pack(push,@var{n})} pragma specifies the maximum alignment
8477 (in bytes) of fields within a structure, in much the same way as the
8478 @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A
8479 pack value of zero resets the behavior to the default. Successive
8480 invocations of this pragma cause the previous values to be stacked, so
8481 that invocations of @samp{#pragma pack(pop)} will return to the previous
8484 @findex DOLLARS_IN_IDENTIFIERS
8485 @item DOLLARS_IN_IDENTIFIERS
8486 Define this macro to control use of the character @samp{$} in identifier
8487 names. 0 means @samp{$} is not allowed by default; 1 means it is allowed.
8488 1 is the default; there is no need to define this macro in that case.
8489 This macro controls the compiler proper; it does not affect the preprocessor.
8491 @findex NO_DOLLAR_IN_LABEL
8492 @item NO_DOLLAR_IN_LABEL
8493 Define this macro if the assembler does not accept the character
8494 @samp{$} in label names. By default constructors and destructors in
8495 G++ have @samp{$} in the identifiers. If this macro is defined,
8496 @samp{.} is used instead.
8498 @findex NO_DOT_IN_LABEL
8499 @item NO_DOT_IN_LABEL
8500 Define this macro if the assembler does not accept the character
8501 @samp{.} in label names. By default constructors and destructors in G++
8502 have names that use @samp{.}. If this macro is defined, these names
8503 are rewritten to avoid @samp{.}.
8505 @findex DEFAULT_MAIN_RETURN
8506 @item DEFAULT_MAIN_RETURN
8507 Define this macro if the target system expects every program's @code{main}
8508 function to return a standard ``success'' value by default (if no other
8509 value is explicitly returned).
8511 The definition should be a C statement (sans semicolon) to generate the
8512 appropriate rtl instructions. It is used only when compiling the end of
8517 Define this if the target system lacks the function @code{atexit}
8518 from the ISO C standard. If this macro is defined, a default definition
8519 will be provided to support C++. If @code{ON_EXIT} is not defined,
8520 a default @code{exit} function will also be provided.
8524 Define this macro if the target has another way to implement atexit
8525 functionality without replacing @code{exit}. For instance, SunOS 4 has
8526 a similar @code{on_exit} library function.
8528 The definition should be a functional macro which can be used just like
8529 the @code{atexit} function.
8533 Define this if your @code{exit} function needs to do something
8534 besides calling an external function @code{_cleanup} before
8535 terminating with @code{_exit}. The @code{EXIT_BODY} macro is
8536 only needed if @code{NEED_ATEXIT} is defined and @code{ON_EXIT} is not
8539 @findex INSN_SETS_ARE_DELAYED
8540 @item INSN_SETS_ARE_DELAYED (@var{insn})
8541 Define this macro as a C expression that is nonzero if it is safe for the
8542 delay slot scheduler to place instructions in the delay slot of @var{insn},
8543 even if they appear to use a resource set or clobbered in @var{insn}.
8544 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
8545 every @code{call_insn} has this behavior. On machines where some @code{insn}
8546 or @code{jump_insn} is really a function call and hence has this behavior,
8547 you should define this macro.
8549 You need not define this macro if it would always return zero.
8551 @findex INSN_REFERENCES_ARE_DELAYED
8552 @item INSN_REFERENCES_ARE_DELAYED (@var{insn})
8553 Define this macro as a C expression that is nonzero if it is safe for the
8554 delay slot scheduler to place instructions in the delay slot of @var{insn},
8555 even if they appear to set or clobber a resource referenced in @var{insn}.
8556 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
8557 some @code{insn} or @code{jump_insn} is really a function call and its operands
8558 are registers whose use is actually in the subroutine it calls, you should
8559 define this macro. Doing so allows the delay slot scheduler to move
8560 instructions which copy arguments into the argument registers into the delay
8563 You need not define this macro if it would always return zero.
8565 @findex MACHINE_DEPENDENT_REORG
8566 @item MACHINE_DEPENDENT_REORG (@var{insn})
8567 In rare cases, correct code generation requires extra machine
8568 dependent processing between the second jump optimization pass and
8569 delayed branch scheduling. On those machines, define this macro as a C
8570 statement to act on the code starting at @var{insn}.
8572 @findex MULTIPLE_SYMBOL_SPACES
8573 @item MULTIPLE_SYMBOL_SPACES
8574 Define this macro if in some cases global symbols from one translation
8575 unit may not be bound to undefined symbols in another translation unit
8576 without user intervention. For instance, under Microsoft Windows
8577 symbols must be explicitly imported from shared libraries (DLLs).
8579 @findex MD_ASM_CLOBBERS
8580 @item MD_ASM_CLOBBERS (@var{clobbers})
8581 A C statement that adds to @var{clobbers} @code{STRING_CST} trees for
8582 any hard regs the port wishes to automatically clobber for all asms.
8584 @findex MAX_INTEGER_COMPUTATION_MODE
8585 @item MAX_INTEGER_COMPUTATION_MODE
8586 Define this to the largest integer machine mode which can be used for
8587 operations other than load, store and copy operations.
8589 You need only define this macro if the target holds values larger than
8590 @code{word_mode} in general purpose registers. Most targets should not define
8593 @findex MATH_LIBRARY
8595 Define this macro as a C string constant for the linker argument to link
8596 in the system math library, or @samp{""} if the target does not have a
8597 separate math library.
8599 You need only define this macro if the default of @samp{"-lm"} is wrong.
8601 @findex LIBRARY_PATH_ENV
8602 @item LIBRARY_PATH_ENV
8603 Define this macro as a C string constant for the environment variable that
8604 specifies where the linker should look for libraries.
8606 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
8609 @findex TARGET_HAS_F_SETLKW
8610 @item TARGET_HAS_F_SETLKW
8611 Define this macro if the target supports file locking with fcntl / F_SETLKW@.
8612 Note that this functionality is part of POSIX@.
8613 Defining @code{TARGET_HAS_F_SETLKW} will enable the test coverage code
8614 to use file locking when exiting a program, which avoids race conditions
8615 if the program has forked.
8617 @findex MAX_CONDITIONAL_EXECUTE
8618 @item MAX_CONDITIONAL_EXECUTE
8620 A C expression for the maximum number of instructions to execute via
8621 conditional execution instructions instead of a branch. A value of
8622 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
8623 1 if it does use cc0.
8625 @findex IFCVT_MODIFY_TESTS
8626 @item IFCVT_MODIFY_TESTS
8627 A C expression to modify the tests in @code{TRUE_EXPR}, and
8628 @code{FALSE_EXPR} for use in converting insns in @code{TEST_BB},
8629 @code{THEN_BB}, @code{ELSE_BB}, and @code{JOIN_BB} basic blocks to
8630 conditional execution. Set either @code{TRUE_EXPR} or @code{FALSE_EXPR}
8631 to a null pointer if the tests cannot be converted.
8633 @findex IFCVT_MODIFY_INSN
8634 @item IFCVT_MODIFY_INSN
8635 A C expression to modify the @code{PATTERN} of an @code{INSN} that is to
8636 be converted to conditional execution format.
8638 @findex IFCVT_MODIFY_FINAL
8639 @item IFCVT_MODIFY_FINAL
8640 A C expression to perform any final machine dependent modifications in
8641 converting code to conditional execution in the basic blocks
8642 @code{TEST_BB}, @code{THEN_BB}, @code{ELSE_BB}, and @code{JOIN_BB}.
8644 @findex IFCVT_MODIFY_CANCEL
8645 @item IFCVT_MODIFY_CANCEL
8646 A C expression to cancel any machine dependent modifications in
8647 converting code to conditional execution in the basic blocks
8648 @code{TEST_BB}, @code{THEN_BB}, @code{ELSE_BB}, and @code{JOIN_BB}.
8651 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS ()
8652 Define this hook if you have any machine-specific built-in functions
8653 that need to be defined. It should be a function that performs the
8656 Machine specific built-in functions can be useful to expand special machine
8657 instructions that would otherwise not normally be generated because
8658 they have no equivalent in the source language (for example, SIMD vector
8659 instructions or prefetch instructions).
8661 To create a built-in function, call the function @code{builtin_function}
8662 which is defined by the language front end. You can use any type nodes set
8663 up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2};
8664 only language front ends that use those two functions will call
8665 @samp{TARGET_INIT_BUILTINS}.
8668 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore})
8670 Expand a call to a machine specific built-in function that was set up by
8671 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
8672 function call; the result should go to @var{target} if that is
8673 convenient, and have mode @var{mode} if that is convenient.
8674 @var{subtarget} may be used as the target for computing one of
8675 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
8676 ignored. This function should return the result of the call to the
8681 @findex MD_CAN_REDIRECT_BRANCH
8682 @item MD_CAN_REDIRECT_BRANCH(@var{branch1}, @var{branch2})
8684 Take a branch insn in @var{branch1} and another in @var{branch2}.
8685 Return true if redirecting @var{branch1} to the destination of
8686 @var{branch2} is possible.
8688 On some targets, branches may have a limited range. Optimizing the
8689 filling of delay slots can result in branches being redirected, and this
8690 may in turn cause a branch offset to overflow.
8692 @findex ALLOCATE_INITIAL_VALUE
8693 @item ALLOCATE_INITIAL_VALUE(@var{hard_reg})
8695 When the initial value of a hard register has been copied in a pseudo
8696 register, it is often not necessary to actually allocate another register
8697 to this pseudo register, because the original hard register or a stack slot
8698 it has been saved into can be used. @code{ALLOCATE_INITIAL_VALUE}, if
8699 defined, is called at the start of register allocation once for each
8700 hard register that had its initial value copied by using
8701 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
8702 Possible values are @code{NULL_RTX}, if you don't want
8703 to do any special allocation, a @code{REG} rtx---that would typically be
8704 the hard register itself, if it is known not to be clobbered---or a
8706 If you are returning a @code{MEM}, this is only a hint for the allocator;
8707 it might decide to use another register anyways.
8708 You may use @code{current_function_leaf_function} in the definition of the
8709 macro, functions that use @code{REG_N_SETS}, to determine if the hard
8710 register in question will not be clobbered.