1 @c Copyright (C) 1988,1989,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,2002
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 ``D float'' 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}, which
1332 is defined by the @command{configure} script, to determine whether the
1333 target machine has the same format as the host machine. If any other
1334 formats are actually in use on supported machines, new codes should be
1337 The ordering of the component words of floating point values stored in
1338 memory is controlled by @code{FLOAT_WORDS_BIG_ENDIAN}.
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 SHORT_TYPE_SIZE
1357 @item SHORT_TYPE_SIZE
1358 A C expression for the size in bits of the type @code{short} on the
1359 target machine. If you don't define this, the default is half a word.
1360 (If this would be less than one storage unit, it is rounded up to one
1363 @findex LONG_TYPE_SIZE
1364 @item LONG_TYPE_SIZE
1365 A C expression for the size in bits of the type @code{long} on the
1366 target machine. If you don't define this, the default is one word.
1368 @findex ADA_LONG_TYPE_SIZE
1369 @item ADA_LONG_TYPE_SIZE
1370 On some machines, the size used for the Ada equivalent of the type
1371 @code{long} by a native Ada compiler differs from that used by C. In
1372 that situation, define this macro to be a C expression to be used for
1373 the size of that type. If you don't define this, the default is the
1374 value of @code{LONG_TYPE_SIZE}.
1376 @findex MAX_LONG_TYPE_SIZE
1377 @item MAX_LONG_TYPE_SIZE
1378 Maximum number for the size in bits of the type @code{long} on the
1379 target machine. If this is undefined, the default is
1380 @code{LONG_TYPE_SIZE}. Otherwise, it is the constant value that is the
1381 largest value that @code{LONG_TYPE_SIZE} can have at run-time. This is
1384 @findex LONG_LONG_TYPE_SIZE
1385 @item LONG_LONG_TYPE_SIZE
1386 A C expression for the size in bits of the type @code{long long} on the
1387 target machine. If you don't define this, the default is two
1388 words. If you want to support GNU Ada on your machine, the value of this
1389 macro must be at least 64.
1391 @findex CHAR_TYPE_SIZE
1392 @item CHAR_TYPE_SIZE
1393 A C expression for the size in bits of the type @code{char} on the
1394 target machine. If you don't define this, the default is
1395 @code{BITS_PER_UNIT}.
1397 @findex MAX_CHAR_TYPE_SIZE
1398 @item MAX_CHAR_TYPE_SIZE
1399 Maximum number for the size in bits of the type @code{char} on the
1400 target machine. If this is undefined, the default is
1401 @code{CHAR_TYPE_SIZE}. Otherwise, it is the constant value that is the
1402 largest value that @code{CHAR_TYPE_SIZE} can have at run-time. This is
1405 @findex BOOL_TYPE_SIZE
1406 @item BOOL_TYPE_SIZE
1407 A C expression for the size in bits of the C++ type @code{bool} on the
1408 target machine. If you don't define this, the default is
1409 @code{CHAR_TYPE_SIZE}.
1411 @findex FLOAT_TYPE_SIZE
1412 @item FLOAT_TYPE_SIZE
1413 A C expression for the size in bits of the type @code{float} on the
1414 target machine. If you don't define this, the default is one word.
1416 @findex DOUBLE_TYPE_SIZE
1417 @item DOUBLE_TYPE_SIZE
1418 A C expression for the size in bits of the type @code{double} on the
1419 target machine. If you don't define this, the default is two
1422 @findex LONG_DOUBLE_TYPE_SIZE
1423 @item LONG_DOUBLE_TYPE_SIZE
1424 A C expression for the size in bits of the type @code{long double} on
1425 the target machine. If you don't define this, the default is two
1428 @findex MAX_LONG_DOUBLE_TYPE_SIZE
1429 Maximum number for the size in bits of the type @code{long double} on the
1430 target machine. If this is undefined, the default is
1431 @code{LONG_DOUBLE_TYPE_SIZE}. Otherwise, it is the constant value that is
1432 the largest value that @code{LONG_DOUBLE_TYPE_SIZE} can have at run-time.
1433 This is used in @code{cpp}.
1435 @findex INTEL_EXTENDED_IEEE_FORMAT
1436 Define this macro to be 1 if the target machine uses 80-bit floating-point
1437 values with 128-bit size and alignment. This is used in @file{real.c}.
1439 @findex WIDEST_HARDWARE_FP_SIZE
1440 @item WIDEST_HARDWARE_FP_SIZE
1441 A C expression for the size in bits of the widest floating-point format
1442 supported by the hardware. If you define this macro, you must specify a
1443 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1444 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1447 @findex DEFAULT_SIGNED_CHAR
1448 @item DEFAULT_SIGNED_CHAR
1449 An expression whose value is 1 or 0, according to whether the type
1450 @code{char} should be signed or unsigned by default. The user can
1451 always override this default with the options @option{-fsigned-char}
1452 and @option{-funsigned-char}.
1454 @findex DEFAULT_SHORT_ENUMS
1455 @item DEFAULT_SHORT_ENUMS
1456 A C expression to determine whether to give an @code{enum} type
1457 only as many bytes as it takes to represent the range of possible values
1458 of that type. A nonzero value means to do that; a zero value means all
1459 @code{enum} types should be allocated like @code{int}.
1461 If you don't define the macro, the default is 0.
1465 A C expression for a string describing the name of the data type to use
1466 for size values. The typedef name @code{size_t} is defined using the
1467 contents of the string.
1469 The string can contain more than one keyword. If so, separate them with
1470 spaces, and write first any length keyword, then @code{unsigned} if
1471 appropriate, and finally @code{int}. The string must exactly match one
1472 of the data type names defined in the function
1473 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1474 omit @code{int} or change the order---that would cause the compiler to
1477 If you don't define this macro, the default is @code{"long unsigned
1480 @findex PTRDIFF_TYPE
1482 A C expression for a string describing the name of the data type to use
1483 for the result of subtracting two pointers. The typedef name
1484 @code{ptrdiff_t} is defined using the contents of the string. See
1485 @code{SIZE_TYPE} above for more information.
1487 If you don't define this macro, the default is @code{"long int"}.
1491 A C expression for a string describing the name of the data type to use
1492 for wide characters. The typedef name @code{wchar_t} is defined using
1493 the contents of the string. See @code{SIZE_TYPE} above for more
1496 If you don't define this macro, the default is @code{"int"}.
1498 @findex WCHAR_TYPE_SIZE
1499 @item WCHAR_TYPE_SIZE
1500 A C expression for the size in bits of the data type for wide
1501 characters. This is used in @code{cpp}, which cannot make use of
1504 @findex MAX_WCHAR_TYPE_SIZE
1505 @item MAX_WCHAR_TYPE_SIZE
1506 Maximum number for the size in bits of the data type for wide
1507 characters. If this is undefined, the default is
1508 @code{WCHAR_TYPE_SIZE}. Otherwise, it is the constant value that is the
1509 largest value that @code{WCHAR_TYPE_SIZE} can have at run-time. This is
1512 @findex GCOV_TYPE_SIZE
1513 @item GCOV_TYPE_SIZE
1514 A C expression for the size in bits of the type used for gcov counters on the
1515 target machine. If you don't define this, the default is one
1516 @code{LONG_TYPE_SIZE} in case it is greater or equal to 64-bit and
1517 @code{LONG_LONG_TYPE_SIZE} otherwise. You may want to re-define the type to
1518 ensure atomicity for counters in multithreaded programs.
1522 A C expression for a string describing the name of the data type to
1523 use for wide characters passed to @code{printf} and returned from
1524 @code{getwc}. The typedef name @code{wint_t} is defined using the
1525 contents of the string. See @code{SIZE_TYPE} above for more
1528 If you don't define this macro, the default is @code{"unsigned int"}.
1532 A C expression for a string describing the name of the data type that
1533 can represent any value of any standard or extended signed integer type.
1534 The typedef name @code{intmax_t} is defined using the contents of the
1535 string. See @code{SIZE_TYPE} above for more information.
1537 If you don't define this macro, the default is the first of
1538 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1539 much precision as @code{long long int}.
1541 @findex UINTMAX_TYPE
1543 A C expression for a string describing the name of the data type that
1544 can represent any value of any standard or extended unsigned integer
1545 type. The typedef name @code{uintmax_t} is defined using the contents
1546 of the string. See @code{SIZE_TYPE} above for more information.
1548 If you don't define this macro, the default is the first of
1549 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1550 unsigned int"} that has as much precision as @code{long long unsigned
1553 @findex TARGET_PTRMEMFUNC_VBIT_LOCATION
1554 @item TARGET_PTRMEMFUNC_VBIT_LOCATION
1555 The C++ compiler represents a pointer-to-member-function with a struct
1562 ptrdiff_t vtable_index;
1569 The C++ compiler must use one bit to indicate whether the function that
1570 will be called through a pointer-to-member-function is virtual.
1571 Normally, we assume that the low-order bit of a function pointer must
1572 always be zero. Then, by ensuring that the vtable_index is odd, we can
1573 distinguish which variant of the union is in use. But, on some
1574 platforms function pointers can be odd, and so this doesn't work. In
1575 that case, we use the low-order bit of the @code{delta} field, and shift
1576 the remainder of the @code{delta} field to the left.
1578 GCC will automatically make the right selection about where to store
1579 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1580 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1581 set such that functions always start at even addresses, but the lowest
1582 bit of pointers to functions indicate whether the function at that
1583 address is in ARM or Thumb mode. If this is the case of your
1584 architecture, you should define this macro to
1585 @code{ptrmemfunc_vbit_in_delta}.
1587 In general, you should not have to define this macro. On architectures
1588 in which function addresses are always even, according to
1589 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1590 @code{ptrmemfunc_vbit_in_pfn}.
1592 @findex TARGET_VTABLE_USES_DESCRIPTORS
1593 @item TARGET_VTABLE_USES_DESCRIPTORS
1594 Normally, the C++ compiler uses function pointers in vtables. This
1595 macro allows the target to change to use ``function descriptors''
1596 instead. Function descriptors are found on targets for whom a
1597 function pointer is actually a small data structure. Normally the
1598 data structure consists of the actual code address plus a data
1599 pointer to which the function's data is relative.
1601 If vtables are used, the value of this macro should be the number
1602 of words that the function descriptor occupies.
1605 @node Escape Sequences
1606 @section Target Character Escape Sequences
1607 @cindex escape sequences
1609 By default, GCC assumes that the C character escape sequences take on
1610 their ASCII values for the target. If this is not correct, you must
1611 explicitly define all of the macros below.
1616 A C constant expression for the integer value for escape sequence
1621 A C constant expression for the integer value of the target escape
1622 character. As an extension, GCC evaluates the escape sequences
1623 @samp{\e} and @samp{\E} to this.
1627 @findex TARGET_NEWLINE
1630 @itemx TARGET_NEWLINE
1631 C constant expressions for the integer values for escape sequences
1632 @samp{\b}, @samp{\t} and @samp{\n}.
1640 C constant expressions for the integer values for escape sequences
1641 @samp{\v}, @samp{\f} and @samp{\r}.
1645 @section Register Usage
1646 @cindex register usage
1648 This section explains how to describe what registers the target machine
1649 has, and how (in general) they can be used.
1651 The description of which registers a specific instruction can use is
1652 done with register classes; see @ref{Register Classes}. For information
1653 on using registers to access a stack frame, see @ref{Frame Registers}.
1654 For passing values in registers, see @ref{Register Arguments}.
1655 For returning values in registers, see @ref{Scalar Return}.
1658 * Register Basics:: Number and kinds of registers.
1659 * Allocation Order:: Order in which registers are allocated.
1660 * Values in Registers:: What kinds of values each reg can hold.
1661 * Leaf Functions:: Renumbering registers for leaf functions.
1662 * Stack Registers:: Handling a register stack such as 80387.
1665 @node Register Basics
1666 @subsection Basic Characteristics of Registers
1668 @c prevent bad page break with this line
1669 Registers have various characteristics.
1672 @findex FIRST_PSEUDO_REGISTER
1673 @item FIRST_PSEUDO_REGISTER
1674 Number of hardware registers known to the compiler. They receive
1675 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1676 pseudo register's number really is assigned the number
1677 @code{FIRST_PSEUDO_REGISTER}.
1679 @item FIXED_REGISTERS
1680 @findex FIXED_REGISTERS
1681 @cindex fixed register
1682 An initializer that says which registers are used for fixed purposes
1683 all throughout the compiled code and are therefore not available for
1684 general allocation. These would include the stack pointer, the frame
1685 pointer (except on machines where that can be used as a general
1686 register when no frame pointer is needed), the program counter on
1687 machines where that is considered one of the addressable registers,
1688 and any other numbered register with a standard use.
1690 This information is expressed as a sequence of numbers, separated by
1691 commas and surrounded by braces. The @var{n}th number is 1 if
1692 register @var{n} is fixed, 0 otherwise.
1694 The table initialized from this macro, and the table initialized by
1695 the following one, may be overridden at run time either automatically,
1696 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1697 the user with the command options @option{-ffixed-@var{reg}},
1698 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1700 @findex CALL_USED_REGISTERS
1701 @item CALL_USED_REGISTERS
1702 @cindex call-used register
1703 @cindex call-clobbered register
1704 @cindex call-saved register
1705 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1706 clobbered (in general) by function calls as well as for fixed
1707 registers. This macro therefore identifies the registers that are not
1708 available for general allocation of values that must live across
1711 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1712 automatically saves it on function entry and restores it on function
1713 exit, if the register is used within the function.
1715 @findex CALL_REALLY_USED_REGISTERS
1716 @item CALL_REALLY_USED_REGISTERS
1717 @cindex call-used register
1718 @cindex call-clobbered register
1719 @cindex call-saved register
1720 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1721 that the entire set of @code{FIXED_REGISTERS} be included.
1722 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1723 This macro is optional. If not specified, it defaults to the value
1724 of @code{CALL_USED_REGISTERS}.
1726 @findex HARD_REGNO_CALL_PART_CLOBBERED
1727 @item HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1728 @cindex call-used register
1729 @cindex call-clobbered register
1730 @cindex call-saved register
1731 A C expression that is nonzero if it is not permissible to store a
1732 value of mode @var{mode} in hard register number @var{regno} across a
1733 call without some part of it being clobbered. For most machines this
1734 macro need not be defined. It is only required for machines that do not
1735 preserve the entire contents of a register across a call.
1737 @findex CONDITIONAL_REGISTER_USAGE
1739 @findex call_used_regs
1740 @item CONDITIONAL_REGISTER_USAGE
1741 Zero or more C statements that may conditionally modify five variables
1742 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1743 @code{reg_names}, and @code{reg_class_contents}, to take into account
1744 any dependence of these register sets on target flags. The first three
1745 of these are of type @code{char []} (interpreted as Boolean vectors).
1746 @code{global_regs} is a @code{const char *[]}, and
1747 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1748 called, @code{fixed_regs}, @code{call_used_regs},
1749 @code{reg_class_contents}, and @code{reg_names} have been initialized
1750 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1751 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1752 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1753 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1754 command options have been applied.
1756 You need not define this macro if it has no work to do.
1758 @cindex disabling certain registers
1759 @cindex controlling register usage
1760 If the usage of an entire class of registers depends on the target
1761 flags, you may indicate this to GCC by using this macro to modify
1762 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
1763 registers in the classes which should not be used by GCC@. Also define
1764 the macro @code{REG_CLASS_FROM_LETTER} to return @code{NO_REGS} if it
1765 is called with a letter for a class that shouldn't be used.
1767 (However, if this class is not included in @code{GENERAL_REGS} and all
1768 of the insn patterns whose constraints permit this class are
1769 controlled by target switches, then GCC will automatically avoid using
1770 these registers when the target switches are opposed to them.)
1772 @findex NON_SAVING_SETJMP
1773 @item NON_SAVING_SETJMP
1774 If this macro is defined and has a nonzero value, it means that
1775 @code{setjmp} and related functions fail to save the registers, or that
1776 @code{longjmp} fails to restore them. To compensate, the compiler
1777 avoids putting variables in registers in functions that use
1780 @findex INCOMING_REGNO
1781 @item INCOMING_REGNO (@var{out})
1782 Define this macro if the target machine has register windows. This C
1783 expression returns the register number as seen by the called function
1784 corresponding to the register number @var{out} as seen by the calling
1785 function. Return @var{out} if register number @var{out} is not an
1788 @findex OUTGOING_REGNO
1789 @item OUTGOING_REGNO (@var{in})
1790 Define this macro if the target machine has register windows. This C
1791 expression returns the register number as seen by the calling function
1792 corresponding to the register number @var{in} as seen by the called
1793 function. Return @var{in} if register number @var{in} is not an inbound
1797 @item LOCAL_REGNO (@var{regno})
1798 Define this macro if the target machine has register windows. This C
1799 expression returns true if the register is call-saved but is in the
1800 register window. Unlike most call-saved registers, such registers
1801 need not be explicitly restored on function exit or during non-local
1807 If the program counter has a register number, define this as that
1808 register number. Otherwise, do not define it.
1812 @node Allocation Order
1813 @subsection Order of Allocation of Registers
1814 @cindex order of register allocation
1815 @cindex register allocation order
1817 @c prevent bad page break with this line
1818 Registers are allocated in order.
1821 @findex REG_ALLOC_ORDER
1822 @item REG_ALLOC_ORDER
1823 If defined, an initializer for a vector of integers, containing the
1824 numbers of hard registers in the order in which GCC should prefer
1825 to use them (from most preferred to least).
1827 If this macro is not defined, registers are used lowest numbered first
1828 (all else being equal).
1830 One use of this macro is on machines where the highest numbered
1831 registers must always be saved and the save-multiple-registers
1832 instruction supports only sequences of consecutive registers. On such
1833 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
1834 the highest numbered allocable register first.
1836 @findex ORDER_REGS_FOR_LOCAL_ALLOC
1837 @item ORDER_REGS_FOR_LOCAL_ALLOC
1838 A C statement (sans semicolon) to choose the order in which to allocate
1839 hard registers for pseudo-registers local to a basic block.
1841 Store the desired register order in the array @code{reg_alloc_order}.
1842 Element 0 should be the register to allocate first; element 1, the next
1843 register; and so on.
1845 The macro body should not assume anything about the contents of
1846 @code{reg_alloc_order} before execution of the macro.
1848 On most machines, it is not necessary to define this macro.
1851 @node Values in Registers
1852 @subsection How Values Fit in Registers
1854 This section discusses the macros that describe which kinds of values
1855 (specifically, which machine modes) each register can hold, and how many
1856 consecutive registers are needed for a given mode.
1859 @findex HARD_REGNO_NREGS
1860 @item HARD_REGNO_NREGS (@var{regno}, @var{mode})
1861 A C expression for the number of consecutive hard registers, starting
1862 at register number @var{regno}, required to hold a value of mode
1865 On a machine where all registers are exactly one word, a suitable
1866 definition of this macro is
1869 #define HARD_REGNO_NREGS(REGNO, MODE) \
1870 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
1874 @findex HARD_REGNO_MODE_OK
1875 @item HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
1876 A C expression that is nonzero if it is permissible to store a value
1877 of mode @var{mode} in hard register number @var{regno} (or in several
1878 registers starting with that one). For a machine where all registers
1879 are equivalent, a suitable definition is
1882 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
1885 You need not include code to check for the numbers of fixed registers,
1886 because the allocation mechanism considers them to be always occupied.
1888 @cindex register pairs
1889 On some machines, double-precision values must be kept in even/odd
1890 register pairs. You can implement that by defining this macro to reject
1891 odd register numbers for such modes.
1893 The minimum requirement for a mode to be OK in a register is that the
1894 @samp{mov@var{mode}} instruction pattern support moves between the
1895 register and other hard register in the same class and that moving a
1896 value into the register and back out not alter it.
1898 Since the same instruction used to move @code{word_mode} will work for
1899 all narrower integer modes, it is not necessary on any machine for
1900 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
1901 you define patterns @samp{movhi}, etc., to take advantage of this. This
1902 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
1903 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
1906 Many machines have special registers for floating point arithmetic.
1907 Often people assume that floating point machine modes are allowed only
1908 in floating point registers. This is not true. Any registers that
1909 can hold integers can safely @emph{hold} a floating point machine
1910 mode, whether or not floating arithmetic can be done on it in those
1911 registers. Integer move instructions can be used to move the values.
1913 On some machines, though, the converse is true: fixed-point machine
1914 modes may not go in floating registers. This is true if the floating
1915 registers normalize any value stored in them, because storing a
1916 non-floating value there would garble it. In this case,
1917 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
1918 floating registers. But if the floating registers do not automatically
1919 normalize, if you can store any bit pattern in one and retrieve it
1920 unchanged without a trap, then any machine mode may go in a floating
1921 register, so you can define this macro to say so.
1923 The primary significance of special floating registers is rather that
1924 they are the registers acceptable in floating point arithmetic
1925 instructions. However, this is of no concern to
1926 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
1927 constraints for those instructions.
1929 On some machines, the floating registers are especially slow to access,
1930 so that it is better to store a value in a stack frame than in such a
1931 register if floating point arithmetic is not being done. As long as the
1932 floating registers are not in class @code{GENERAL_REGS}, they will not
1933 be used unless some pattern's constraint asks for one.
1935 @findex MODES_TIEABLE_P
1936 @item MODES_TIEABLE_P (@var{mode1}, @var{mode2})
1937 A C expression that is nonzero if a value of mode
1938 @var{mode1} is accessible in mode @var{mode2} without copying.
1940 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
1941 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
1942 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
1943 should be nonzero. If they differ for any @var{r}, you should define
1944 this macro to return zero unless some other mechanism ensures the
1945 accessibility of the value in a narrower mode.
1947 You should define this macro to return nonzero in as many cases as
1948 possible since doing so will allow GCC to perform better register
1951 @findex AVOID_CCMODE_COPIES
1952 @item AVOID_CCMODE_COPIES
1953 Define this macro if the compiler should avoid copies to/from @code{CCmode}
1954 registers. You should only define this macro if support for copying to/from
1955 @code{CCmode} is incomplete.
1958 @node Leaf Functions
1959 @subsection Handling Leaf Functions
1961 @cindex leaf functions
1962 @cindex functions, leaf
1963 On some machines, a leaf function (i.e., one which makes no calls) can run
1964 more efficiently if it does not make its own register window. Often this
1965 means it is required to receive its arguments in the registers where they
1966 are passed by the caller, instead of the registers where they would
1969 The special treatment for leaf functions generally applies only when
1970 other conditions are met; for example, often they may use only those
1971 registers for its own variables and temporaries. We use the term ``leaf
1972 function'' to mean a function that is suitable for this special
1973 handling, so that functions with no calls are not necessarily ``leaf
1976 GCC assigns register numbers before it knows whether the function is
1977 suitable for leaf function treatment. So it needs to renumber the
1978 registers in order to output a leaf function. The following macros
1982 @findex LEAF_REGISTERS
1983 @item LEAF_REGISTERS
1984 Name of a char vector, indexed by hard register number, which
1985 contains 1 for a register that is allowable in a candidate for leaf
1988 If leaf function treatment involves renumbering the registers, then the
1989 registers marked here should be the ones before renumbering---those that
1990 GCC would ordinarily allocate. The registers which will actually be
1991 used in the assembler code, after renumbering, should not be marked with 1
1994 Define this macro only if the target machine offers a way to optimize
1995 the treatment of leaf functions.
1997 @findex LEAF_REG_REMAP
1998 @item LEAF_REG_REMAP (@var{regno})
1999 A C expression whose value is the register number to which @var{regno}
2000 should be renumbered, when a function is treated as a leaf function.
2002 If @var{regno} is a register number which should not appear in a leaf
2003 function before renumbering, then the expression should yield @minus{}1, which
2004 will cause the compiler to abort.
2006 Define this macro only if the target machine offers a way to optimize the
2007 treatment of leaf functions, and registers need to be renumbered to do
2011 @findex current_function_is_leaf
2012 @findex current_function_uses_only_leaf_regs
2013 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2014 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2015 specially. They can test the C variable @code{current_function_is_leaf}
2016 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2017 set prior to local register allocation and is valid for the remaining
2018 compiler passes. They can also test the C variable
2019 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2020 functions which only use leaf registers.
2021 @code{current_function_uses_only_leaf_regs} is valid after reload and is
2022 only useful if @code{LEAF_REGISTERS} is defined.
2023 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2024 @c of the next paragraph?! --mew 2feb93
2026 @node Stack Registers
2027 @subsection Registers That Form a Stack
2029 There are special features to handle computers where some of the
2030 ``registers'' form a stack, as in the 80387 coprocessor for the 80386.
2031 Stack registers are normally written by pushing onto the stack, and are
2032 numbered relative to the top of the stack.
2034 Currently, GCC can only handle one group of stack-like registers, and
2035 they must be consecutively numbered.
2040 Define this if the machine has any stack-like registers.
2042 @findex FIRST_STACK_REG
2043 @item FIRST_STACK_REG
2044 The number of the first stack-like register. This one is the top
2047 @findex LAST_STACK_REG
2048 @item LAST_STACK_REG
2049 The number of the last stack-like register. This one is the bottom of
2053 @node Register Classes
2054 @section Register Classes
2055 @cindex register class definitions
2056 @cindex class definitions, register
2058 On many machines, the numbered registers are not all equivalent.
2059 For example, certain registers may not be allowed for indexed addressing;
2060 certain registers may not be allowed in some instructions. These machine
2061 restrictions are described to the compiler using @dfn{register classes}.
2063 You define a number of register classes, giving each one a name and saying
2064 which of the registers belong to it. Then you can specify register classes
2065 that are allowed as operands to particular instruction patterns.
2069 In general, each register will belong to several classes. In fact, one
2070 class must be named @code{ALL_REGS} and contain all the registers. Another
2071 class must be named @code{NO_REGS} and contain no registers. Often the
2072 union of two classes will be another class; however, this is not required.
2074 @findex GENERAL_REGS
2075 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2076 terribly special about the name, but the operand constraint letters
2077 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2078 the same as @code{ALL_REGS}, just define it as a macro which expands
2081 Order the classes so that if class @var{x} is contained in class @var{y}
2082 then @var{x} has a lower class number than @var{y}.
2084 The way classes other than @code{GENERAL_REGS} are specified in operand
2085 constraints is through machine-dependent operand constraint letters.
2086 You can define such letters to correspond to various classes, then use
2087 them in operand constraints.
2089 You should define a class for the union of two classes whenever some
2090 instruction allows both classes. For example, if an instruction allows
2091 either a floating point (coprocessor) register or a general register for a
2092 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2093 which includes both of them. Otherwise you will get suboptimal code.
2095 You must also specify certain redundant information about the register
2096 classes: for each class, which classes contain it and which ones are
2097 contained in it; for each pair of classes, the largest class contained
2100 When a value occupying several consecutive registers is expected in a
2101 certain class, all the registers used must belong to that class.
2102 Therefore, register classes cannot be used to enforce a requirement for
2103 a register pair to start with an even-numbered register. The way to
2104 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2106 Register classes used for input-operands of bitwise-and or shift
2107 instructions have a special requirement: each such class must have, for
2108 each fixed-point machine mode, a subclass whose registers can transfer that
2109 mode to or from memory. For example, on some machines, the operations for
2110 single-byte values (@code{QImode}) are limited to certain registers. When
2111 this is so, each register class that is used in a bitwise-and or shift
2112 instruction must have a subclass consisting of registers from which
2113 single-byte values can be loaded or stored. This is so that
2114 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2117 @findex enum reg_class
2118 @item enum reg_class
2119 An enumeral type that must be defined with all the register class names
2120 as enumeral values. @code{NO_REGS} must be first. @code{ALL_REGS}
2121 must be the last register class, followed by one more enumeral value,
2122 @code{LIM_REG_CLASSES}, which is not a register class but rather
2123 tells how many classes there are.
2125 Each register class has a number, which is the value of casting
2126 the class name to type @code{int}. The number serves as an index
2127 in many of the tables described below.
2129 @findex N_REG_CLASSES
2131 The number of distinct register classes, defined as follows:
2134 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2137 @findex REG_CLASS_NAMES
2138 @item REG_CLASS_NAMES
2139 An initializer containing the names of the register classes as C string
2140 constants. These names are used in writing some of the debugging dumps.
2142 @findex REG_CLASS_CONTENTS
2143 @item REG_CLASS_CONTENTS
2144 An initializer containing the contents of the register classes, as integers
2145 which are bit masks. The @var{n}th integer specifies the contents of class
2146 @var{n}. The way the integer @var{mask} is interpreted is that
2147 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2149 When the machine has more than 32 registers, an integer does not suffice.
2150 Then the integers are replaced by sub-initializers, braced groupings containing
2151 several integers. Each sub-initializer must be suitable as an initializer
2152 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2153 In this situation, the first integer in each sub-initializer corresponds to
2154 registers 0 through 31, the second integer to registers 32 through 63, and
2157 @findex REGNO_REG_CLASS
2158 @item REGNO_REG_CLASS (@var{regno})
2159 A C expression whose value is a register class containing hard register
2160 @var{regno}. In general there is more than one such class; choose a class
2161 which is @dfn{minimal}, meaning that no smaller class also contains the
2164 @findex BASE_REG_CLASS
2165 @item BASE_REG_CLASS
2166 A macro whose definition is the name of the class to which a valid
2167 base register must belong. A base register is one used in an address
2168 which is the register value plus a displacement.
2170 @findex MODE_BASE_REG_CLASS
2171 @item MODE_BASE_REG_CLASS (@var{mode})
2172 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2173 the selection of a base register in a mode depenedent manner. If
2174 @var{mode} is VOIDmode then it should return the same value as
2175 @code{BASE_REG_CLASS}.
2177 @findex INDEX_REG_CLASS
2178 @item INDEX_REG_CLASS
2179 A macro whose definition is the name of the class to which a valid
2180 index register must belong. An index register is one used in an
2181 address where its value is either multiplied by a scale factor or
2182 added to another register (as well as added to a displacement).
2184 @findex REG_CLASS_FROM_LETTER
2185 @item REG_CLASS_FROM_LETTER (@var{char})
2186 A C expression which defines the machine-dependent operand constraint
2187 letters for register classes. If @var{char} is such a letter, the
2188 value should be the register class corresponding to it. Otherwise,
2189 the value should be @code{NO_REGS}. The register letter @samp{r},
2190 corresponding to class @code{GENERAL_REGS}, will not be passed
2191 to this macro; you do not need to handle it.
2193 @findex REGNO_OK_FOR_BASE_P
2194 @item REGNO_OK_FOR_BASE_P (@var{num})
2195 A C expression which is nonzero if register number @var{num} is
2196 suitable for use as a base register in operand addresses. It may be
2197 either a suitable hard register or a pseudo register that has been
2198 allocated such a hard register.
2200 @findex REGNO_MODE_OK_FOR_BASE_P
2201 @item REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2202 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2203 that expression may examine the mode of the memory reference in
2204 @var{mode}. You should define this macro if the mode of the memory
2205 reference affects whether a register may be used as a base register. If
2206 you define this macro, the compiler will use it instead of
2207 @code{REGNO_OK_FOR_BASE_P}.
2209 @findex REGNO_OK_FOR_INDEX_P
2210 @item REGNO_OK_FOR_INDEX_P (@var{num})
2211 A C expression which is nonzero if register number @var{num} is
2212 suitable for use as an index register in operand addresses. It may be
2213 either a suitable hard register or a pseudo register that has been
2214 allocated such a hard register.
2216 The difference between an index register and a base register is that
2217 the index register may be scaled. If an address involves the sum of
2218 two registers, neither one of them scaled, then either one may be
2219 labeled the ``base'' and the other the ``index''; but whichever
2220 labeling is used must fit the machine's constraints of which registers
2221 may serve in each capacity. The compiler will try both labelings,
2222 looking for one that is valid, and will reload one or both registers
2223 only if neither labeling works.
2225 @findex PREFERRED_RELOAD_CLASS
2226 @item PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2227 A C expression that places additional restrictions on the register class
2228 to use when it is necessary to copy value @var{x} into a register in class
2229 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2230 another, smaller class. On many machines, the following definition is
2234 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2237 Sometimes returning a more restrictive class makes better code. For
2238 example, on the 68000, when @var{x} is an integer constant that is in range
2239 for a @samp{moveq} instruction, the value of this macro is always
2240 @code{DATA_REGS} as long as @var{class} includes the data registers.
2241 Requiring a data register guarantees that a @samp{moveq} will be used.
2243 If @var{x} is a @code{const_double}, by returning @code{NO_REGS}
2244 you can force @var{x} into a memory constant. This is useful on
2245 certain machines where immediate floating values cannot be loaded into
2246 certain kinds of registers.
2248 @findex PREFERRED_OUTPUT_RELOAD_CLASS
2249 @item PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2250 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2251 input reloads. If you don't define this macro, the default is to use
2252 @var{class}, unchanged.
2254 @findex LIMIT_RELOAD_CLASS
2255 @item LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2256 A C expression that places additional restrictions on the register class
2257 to use when it is necessary to be able to hold a value of mode
2258 @var{mode} in a reload register for which class @var{class} would
2261 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2262 there are certain modes that simply can't go in certain reload classes.
2264 The value is a register class; perhaps @var{class}, or perhaps another,
2267 Don't define this macro unless the target machine has limitations which
2268 require the macro to do something nontrivial.
2270 @findex SECONDARY_RELOAD_CLASS
2271 @findex SECONDARY_INPUT_RELOAD_CLASS
2272 @findex SECONDARY_OUTPUT_RELOAD_CLASS
2273 @item SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2274 @itemx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2275 @itemx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2276 Many machines have some registers that cannot be copied directly to or
2277 from memory or even from other types of registers. An example is the
2278 @samp{MQ} register, which on most machines, can only be copied to or
2279 from general registers, but not memory. Some machines allow copying all
2280 registers to and from memory, but require a scratch register for stores
2281 to some memory locations (e.g., those with symbolic address on the RT,
2282 and those with certain symbolic address on the Sparc when compiling
2283 PIC)@. In some cases, both an intermediate and a scratch register are
2286 You should define these macros to indicate to the reload phase that it may
2287 need to allocate at least one register for a reload in addition to the
2288 register to contain the data. Specifically, if copying @var{x} to a
2289 register @var{class} in @var{mode} requires an intermediate register,
2290 you should define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2291 largest register class all of whose registers can be used as
2292 intermediate registers or scratch registers.
2294 If copying a register @var{class} in @var{mode} to @var{x} requires an
2295 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2296 should be defined to return the largest register class required. If the
2297 requirements for input and output reloads are the same, the macro
2298 @code{SECONDARY_RELOAD_CLASS} should be used instead of defining both
2301 The values returned by these macros are often @code{GENERAL_REGS}.
2302 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2303 can be directly copied to or from a register of @var{class} in
2304 @var{mode} without requiring a scratch register. Do not define this
2305 macro if it would always return @code{NO_REGS}.
2307 If a scratch register is required (either with or without an
2308 intermediate register), you should define patterns for
2309 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2310 (@pxref{Standard Names}. These patterns, which will normally be
2311 implemented with a @code{define_expand}, should be similar to the
2312 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2315 Define constraints for the reload register and scratch register that
2316 contain a single register class. If the original reload register (whose
2317 class is @var{class}) can meet the constraint given in the pattern, the
2318 value returned by these macros is used for the class of the scratch
2319 register. Otherwise, two additional reload registers are required.
2320 Their classes are obtained from the constraints in the insn pattern.
2322 @var{x} might be a pseudo-register or a @code{subreg} of a
2323 pseudo-register, which could either be in a hard register or in memory.
2324 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2325 in memory and the hard register number if it is in a register.
2327 These macros should not be used in the case where a particular class of
2328 registers can only be copied to memory and not to another class of
2329 registers. In that case, secondary reload registers are not needed and
2330 would not be helpful. Instead, a stack location must be used to perform
2331 the copy and the @code{mov@var{m}} pattern should use memory as an
2332 intermediate storage. This case often occurs between floating-point and
2335 @findex SECONDARY_MEMORY_NEEDED
2336 @item SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2337 Certain machines have the property that some registers cannot be copied
2338 to some other registers without using memory. Define this macro on
2339 those machines to be a C expression that is nonzero if objects of mode
2340 @var{m} in registers of @var{class1} can only be copied to registers of
2341 class @var{class2} by storing a register of @var{class1} into memory
2342 and loading that memory location into a register of @var{class2}.
2344 Do not define this macro if its value would always be zero.
2346 @findex SECONDARY_MEMORY_NEEDED_RTX
2347 @item SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2348 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2349 allocates a stack slot for a memory location needed for register copies.
2350 If this macro is defined, the compiler instead uses the memory location
2351 defined by this macro.
2353 Do not define this macro if you do not define
2354 @code{SECONDARY_MEMORY_NEEDED}.
2356 @findex SECONDARY_MEMORY_NEEDED_MODE
2357 @item SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2358 When the compiler needs a secondary memory location to copy between two
2359 registers of mode @var{mode}, it normally allocates sufficient memory to
2360 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2361 load operations in a mode that many bits wide and whose class is the
2362 same as that of @var{mode}.
2364 This is right thing to do on most machines because it ensures that all
2365 bits of the register are copied and prevents accesses to the registers
2366 in a narrower mode, which some machines prohibit for floating-point
2369 However, this default behavior is not correct on some machines, such as
2370 the DEC Alpha, that store short integers in floating-point registers
2371 differently than in integer registers. On those machines, the default
2372 widening will not work correctly and you must define this macro to
2373 suppress that widening in some cases. See the file @file{alpha.h} for
2376 Do not define this macro if you do not define
2377 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2378 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2380 @findex SMALL_REGISTER_CLASSES
2381 @item SMALL_REGISTER_CLASSES
2382 On some machines, it is risky to let hard registers live across arbitrary
2383 insns. Typically, these machines have instructions that require values
2384 to be in specific registers (like an accumulator), and reload will fail
2385 if the required hard register is used for another purpose across such an
2388 Define @code{SMALL_REGISTER_CLASSES} to be an expression with a nonzero
2389 value on these machines. When this macro has a nonzero value, the
2390 compiler will try to minimize the lifetime of hard registers.
2392 It is always safe to define this macro with a nonzero value, but if you
2393 unnecessarily define it, you will reduce the amount of optimizations
2394 that can be performed in some cases. If you do not define this macro
2395 with a nonzero value when it is required, the compiler will run out of
2396 spill registers and print a fatal error message. For most machines, you
2397 should not define this macro at all.
2399 @findex CLASS_LIKELY_SPILLED_P
2400 @item CLASS_LIKELY_SPILLED_P (@var{class})
2401 A C expression whose value is nonzero if pseudos that have been assigned
2402 to registers of class @var{class} would likely be spilled because
2403 registers of @var{class} are needed for spill registers.
2405 The default value of this macro returns 1 if @var{class} has exactly one
2406 register and zero otherwise. On most machines, this default should be
2407 used. Only define this macro to some other expression if pseudos
2408 allocated by @file{local-alloc.c} end up in memory because their hard
2409 registers were needed for spill registers. If this macro returns nonzero
2410 for those classes, those pseudos will only be allocated by
2411 @file{global.c}, which knows how to reallocate the pseudo to another
2412 register. If there would not be another register available for
2413 reallocation, you should not change the definition of this macro since
2414 the only effect of such a definition would be to slow down register
2417 @findex CLASS_MAX_NREGS
2418 @item CLASS_MAX_NREGS (@var{class}, @var{mode})
2419 A C expression for the maximum number of consecutive registers
2420 of class @var{class} needed to hold a value of mode @var{mode}.
2422 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2423 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2424 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2425 @var{mode})} for all @var{regno} values in the class @var{class}.
2427 This macro helps control the handling of multiple-word values
2430 @item CLASS_CANNOT_CHANGE_MODE
2431 If defined, a C expression for a class that contains registers for
2432 which the compiler may not change modes arbitrarily.
2434 @item CLASS_CANNOT_CHANGE_MODE_P(@var{from}, @var{to})
2435 A C expression that is true if, for a register in
2436 @code{CLASS_CANNOT_CHANGE_MODE}, the requested mode punning is invalid.
2438 For the example, loading 32-bit integer or floating-point objects into
2439 floating-point registers on the Alpha extends them to 64-bits.
2440 Therefore loading a 64-bit object and then storing it as a 32-bit object
2441 does not store the low-order 32-bits, as would be the case for a normal
2442 register. Therefore, @file{alpha.h} defines @code{CLASS_CANNOT_CHANGE_MODE}
2443 as @code{FLOAT_REGS} and @code{CLASS_CANNOT_CHANGE_MODE_P} restricts
2444 mode changes to same-size modes.
2446 Compare this to IA-64, which extends floating-point values to 82-bits,
2447 and stores 64-bit integers in a different format than 64-bit doubles.
2448 Therefore @code{CLASS_CANNOT_CHANGE_MODE_P} is always true.
2451 Three other special macros describe which operands fit which constraint
2455 @findex CONST_OK_FOR_LETTER_P
2456 @item CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2457 A C expression that defines the machine-dependent operand constraint
2458 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2459 particular ranges of integer values. If @var{c} is one of those
2460 letters, the expression should check that @var{value}, an integer, is in
2461 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2462 not one of those letters, the value should be 0 regardless of
2465 @findex CONST_DOUBLE_OK_FOR_LETTER_P
2466 @item CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2467 A C expression that defines the machine-dependent operand constraint
2468 letters that specify particular ranges of @code{const_double} values
2469 (@samp{G} or @samp{H}).
2471 If @var{c} is one of those letters, the expression should check that
2472 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2473 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2474 letters, the value should be 0 regardless of @var{value}.
2476 @code{const_double} is used for all floating-point constants and for
2477 @code{DImode} fixed-point constants. A given letter can accept either
2478 or both kinds of values. It can use @code{GET_MODE} to distinguish
2479 between these kinds.
2481 @findex EXTRA_CONSTRAINT
2482 @item EXTRA_CONSTRAINT (@var{value}, @var{c})
2483 A C expression that defines the optional machine-dependent constraint
2484 letters that can be used to segregate specific types of operands, usually
2485 memory references, for the target machine. Any letter that is not
2486 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER}
2487 may be used. Normally this macro will not be defined.
2489 If it is required for a particular target machine, it should return 1
2490 if @var{value} corresponds to the operand type represented by the
2491 constraint letter @var{c}. If @var{c} is not defined as an extra
2492 constraint, the value returned should be 0 regardless of @var{value}.
2494 For example, on the ROMP, load instructions cannot have their output
2495 in r0 if the memory reference contains a symbolic address. Constraint
2496 letter @samp{Q} is defined as representing a memory address that does
2497 @emph{not} contain a symbolic address. An alternative is specified with
2498 a @samp{Q} constraint on the input and @samp{r} on the output. The next
2499 alternative specifies @samp{m} on the input and a register class that
2500 does not include r0 on the output.
2503 @node Stack and Calling
2504 @section Stack Layout and Calling Conventions
2505 @cindex calling conventions
2507 @c prevent bad page break with this line
2508 This describes the stack layout and calling conventions.
2512 * Exception Handling::
2517 * Register Arguments::
2519 * Aggregate Return::
2527 @subsection Basic Stack Layout
2528 @cindex stack frame layout
2529 @cindex frame layout
2531 @c prevent bad page break with this line
2532 Here is the basic stack layout.
2535 @findex STACK_GROWS_DOWNWARD
2536 @item STACK_GROWS_DOWNWARD
2537 Define this macro if pushing a word onto the stack moves the stack
2538 pointer to a smaller address.
2540 When we say, ``define this macro if @dots{},'' it means that the
2541 compiler checks this macro only with @code{#ifdef} so the precise
2542 definition used does not matter.
2544 @findex STACK_PUSH_CODE
2545 @item STACK_PUSH_CODE
2547 This macro defines the operation used when something is pushed
2548 on the stack. In RTL, a push operation will be
2549 @code{(set (mem (STACK_PUSH_CODE (reg sp))) ...)}
2551 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
2552 and @code{POST_INC}. Which of these is correct depends on
2553 the stack direction and on whether the stack pointer points
2554 to the last item on the stack or whether it points to the
2555 space for the next item on the stack.
2557 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
2558 defined, which is almost always right, and @code{PRE_INC} otherwise,
2559 which is often wrong.
2561 @findex FRAME_GROWS_DOWNWARD
2562 @item FRAME_GROWS_DOWNWARD
2563 Define this macro if the addresses of local variable slots are at negative
2564 offsets from the frame pointer.
2566 @findex ARGS_GROW_DOWNWARD
2567 @item ARGS_GROW_DOWNWARD
2568 Define this macro if successive arguments to a function occupy decreasing
2569 addresses on the stack.
2571 @findex STARTING_FRAME_OFFSET
2572 @item STARTING_FRAME_OFFSET
2573 Offset from the frame pointer to the first local variable slot to be allocated.
2575 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
2576 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
2577 Otherwise, it is found by adding the length of the first slot to the
2578 value @code{STARTING_FRAME_OFFSET}.
2579 @c i'm not sure if the above is still correct.. had to change it to get
2580 @c rid of an overfull. --mew 2feb93
2582 @findex STACK_POINTER_OFFSET
2583 @item STACK_POINTER_OFFSET
2584 Offset from the stack pointer register to the first location at which
2585 outgoing arguments are placed. If not specified, the default value of
2586 zero is used. This is the proper value for most machines.
2588 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2589 the first location at which outgoing arguments are placed.
2591 @findex FIRST_PARM_OFFSET
2592 @item FIRST_PARM_OFFSET (@var{fundecl})
2593 Offset from the argument pointer register to the first argument's
2594 address. On some machines it may depend on the data type of the
2597 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2598 the first argument's address.
2600 @findex STACK_DYNAMIC_OFFSET
2601 @item STACK_DYNAMIC_OFFSET (@var{fundecl})
2602 Offset from the stack pointer register to an item dynamically allocated
2603 on the stack, e.g., by @code{alloca}.
2605 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
2606 length of the outgoing arguments. The default is correct for most
2607 machines. See @file{function.c} for details.
2609 @findex DYNAMIC_CHAIN_ADDRESS
2610 @item DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
2611 A C expression whose value is RTL representing the address in a stack
2612 frame where the pointer to the caller's frame is stored. Assume that
2613 @var{frameaddr} is an RTL expression for the address of the stack frame
2616 If you don't define this macro, the default is to return the value
2617 of @var{frameaddr}---that is, the stack frame address is also the
2618 address of the stack word that points to the previous frame.
2620 @findex SETUP_FRAME_ADDRESSES
2621 @item SETUP_FRAME_ADDRESSES
2622 If defined, a C expression that produces the machine-specific code to
2623 setup the stack so that arbitrary frames can be accessed. For example,
2624 on the Sparc, we must flush all of the register windows to the stack
2625 before we can access arbitrary stack frames. You will seldom need to
2628 @findex BUILTIN_SETJMP_FRAME_VALUE
2629 @item BUILTIN_SETJMP_FRAME_VALUE
2630 If defined, a C expression that contains an rtx that is used to store
2631 the address of the current frame into the built in @code{setjmp} buffer.
2632 The default value, @code{virtual_stack_vars_rtx}, is correct for most
2633 machines. One reason you may need to define this macro is if
2634 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
2636 @findex RETURN_ADDR_RTX
2637 @item RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
2638 A C expression whose value is RTL representing the value of the return
2639 address for the frame @var{count} steps up from the current frame, after
2640 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
2641 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
2642 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
2644 The value of the expression must always be the correct address when
2645 @var{count} is zero, but may be @code{NULL_RTX} if there is not way to
2646 determine the return address of other frames.
2648 @findex RETURN_ADDR_IN_PREVIOUS_FRAME
2649 @item RETURN_ADDR_IN_PREVIOUS_FRAME
2650 Define this if the return address of a particular stack frame is accessed
2651 from the frame pointer of the previous stack frame.
2653 @findex INCOMING_RETURN_ADDR_RTX
2654 @item INCOMING_RETURN_ADDR_RTX
2655 A C expression whose value is RTL representing the location of the
2656 incoming return address at the beginning of any function, before the
2657 prologue. This RTL is either a @code{REG}, indicating that the return
2658 value is saved in @samp{REG}, or a @code{MEM} representing a location in
2661 You only need to define this macro if you want to support call frame
2662 debugging information like that provided by DWARF 2.
2664 If this RTL is a @code{REG}, you should also define
2665 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
2667 @findex INCOMING_FRAME_SP_OFFSET
2668 @item INCOMING_FRAME_SP_OFFSET
2669 A C expression whose value is an integer giving the offset, in bytes,
2670 from the value of the stack pointer register to the top of the stack
2671 frame at the beginning of any function, before the prologue. The top of
2672 the frame is defined to be the value of the stack pointer in the
2673 previous frame, just before the call instruction.
2675 You only need to define this macro if you want to support call frame
2676 debugging information like that provided by DWARF 2.
2678 @findex ARG_POINTER_CFA_OFFSET
2679 @item ARG_POINTER_CFA_OFFSET (@var{fundecl})
2680 A C expression whose value is an integer giving the offset, in bytes,
2681 from the argument pointer to the canonical frame address (cfa). The
2682 final value should coincide with that calculated by
2683 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
2684 during virtual register instantiation.
2686 The default value for this macro is @code{FIRST_PARM_OFFSET (fundecl)},
2687 which is correct for most machines; in general, the arguments are found
2688 immediately before the stack frame. Note that this is not the case on
2689 some targets that save registers into the caller's frame, such as SPARC
2690 and rs6000, and so such targets need to define this macro.
2692 You only need to define this macro if the default is incorrect, and you
2693 want to support call frame debugging information like that provided by
2698 Define this macro if the stack size for the target is very small. This
2699 has the effect of disabling gcc's built-in @samp{alloca}, though
2700 @samp{__builtin_alloca} is not affected.
2703 @node Exception Handling
2704 @subsection Exception Handling Support
2705 @cindex exception handling
2708 @findex EH_RETURN_DATA_REGNO
2709 @item EH_RETURN_DATA_REGNO (@var{N})
2710 A C expression whose value is the @var{N}th register number used for
2711 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
2712 @var{N} registers are usable.
2714 The exception handling library routines communicate with the exception
2715 handlers via a set of agreed upon registers. Ideally these registers
2716 should be call-clobbered; it is possible to use call-saved registers,
2717 but may negatively impact code size. The target must support at least
2718 2 data registers, but should define 4 if there are enough free registers.
2720 You must define this macro if you want to support call frame exception
2721 handling like that provided by DWARF 2.
2723 @findex EH_RETURN_STACKADJ_RTX
2724 @item EH_RETURN_STACKADJ_RTX
2725 A C expression whose value is RTL representing a location in which
2726 to store a stack adjustment to be applied before function return.
2727 This is used to unwind the stack to an exception handler's call frame.
2728 It will be assigned zero on code paths that return normally.
2730 Typically this is a call-clobbered hard register that is otherwise
2731 untouched by the epilogue, but could also be a stack slot.
2733 You must define this macro if you want to support call frame exception
2734 handling like that provided by DWARF 2.
2736 @findex EH_RETURN_HANDLER_RTX
2737 @item EH_RETURN_HANDLER_RTX
2738 A C expression whose value is RTL representing a location in which
2739 to store the address of an exception handler to which we should
2740 return. It will not be assigned on code paths that return normally.
2742 Typically this is the location in the call frame at which the normal
2743 return address is stored. For targets that return by popping an
2744 address off the stack, this might be a memory address just below
2745 the @emph{target} call frame rather than inside the current call
2746 frame. @code{EH_RETURN_STACKADJ_RTX} will have already been assigned,
2747 so it may be used to calculate the location of the target call frame.
2749 Some targets have more complex requirements than storing to an
2750 address calculable during initial code generation. In that case
2751 the @code{eh_return} instruction pattern should be used instead.
2753 If you want to support call frame exception handling, you must
2754 define either this macro or the @code{eh_return} instruction pattern.
2756 @findex ASM_PREFERRED_EH_DATA_FORMAT
2757 @item ASM_PREFERRED_EH_DATA_FORMAT(@var{code}, @var{global})
2758 This macro chooses the encoding of pointers embedded in the exception
2759 handling sections. If at all possible, this should be defined such
2760 that the exception handling section will not require dynamic relocations,
2761 and so may be read-only.
2763 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
2764 @var{global} is true if the symbol may be affected by dynamic relocations.
2765 The macro should return a combination of the @code{DW_EH_PE_*} defines
2766 as found in @file{dwarf2.h}.
2768 If this macro is not defined, pointers will not be encoded but
2769 represented directly.
2771 @findex ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX
2772 @item ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX(@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
2773 This macro allows the target to emit whatever special magic is required
2774 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
2775 Generic code takes care of pc-relative and indirect encodings; this must
2776 be defined if the target uses text-relative or data-relative encodings.
2778 This is a C statement that branches to @var{done} if the format was
2779 handled. @var{encoding} is the format chosen, @var{size} is the number
2780 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
2783 @findex MD_FALLBACK_FRAME_STATE_FOR
2784 @item MD_FALLBACK_FRAME_STATE_FOR(@var{context}, @var{fs}, @var{success})
2785 This macro allows the target to add cpu and operating system specific
2786 code to the call-frame unwinder for use when there is no unwind data
2787 available. The most common reason to implement this macro is to unwind
2788 through signal frames.
2790 This macro is called from @code{uw_frame_state_for} in @file{unwind-dw2.c}
2791 and @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
2792 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
2793 for the address of the code being executed and @code{context->cfa} for
2794 the stack pointer value. If the frame can be decoded, the register save
2795 addresses should be updated in @var{fs} and the macro should branch to
2796 @var{success}. If the frame cannot be decoded, the macro should do
2800 @node Stack Checking
2801 @subsection Specifying How Stack Checking is Done
2803 GCC will check that stack references are within the boundaries of
2804 the stack, if the @option{-fstack-check} is specified, in one of three ways:
2808 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
2809 will assume that you have arranged for stack checking to be done at
2810 appropriate places in the configuration files, e.g., in
2811 @code{TARGET_ASM_FUNCTION_PROLOGUE}. GCC will do not other special
2815 If @code{STACK_CHECK_BUILTIN} is zero and you defined a named pattern
2816 called @code{check_stack} in your @file{md} file, GCC will call that
2817 pattern with one argument which is the address to compare the stack
2818 value against. You must arrange for this pattern to report an error if
2819 the stack pointer is out of range.
2822 If neither of the above are true, GCC will generate code to periodically
2823 ``probe'' the stack pointer using the values of the macros defined below.
2826 Normally, you will use the default values of these macros, so GCC
2827 will use the third approach.
2830 @findex STACK_CHECK_BUILTIN
2831 @item STACK_CHECK_BUILTIN
2832 A nonzero value if stack checking is done by the configuration files in a
2833 machine-dependent manner. You should define this macro if stack checking
2834 is require by the ABI of your machine or if you would like to have to stack
2835 checking in some more efficient way than GCC's portable approach.
2836 The default value of this macro is zero.
2838 @findex STACK_CHECK_PROBE_INTERVAL
2839 @item STACK_CHECK_PROBE_INTERVAL
2840 An integer representing the interval at which GCC must generate stack
2841 probe instructions. You will normally define this macro to be no larger
2842 than the size of the ``guard pages'' at the end of a stack area. The
2843 default value of 4096 is suitable for most systems.
2845 @findex STACK_CHECK_PROBE_LOAD
2846 @item STACK_CHECK_PROBE_LOAD
2847 A integer which is nonzero if GCC should perform the stack probe
2848 as a load instruction and zero if GCC should use a store instruction.
2849 The default is zero, which is the most efficient choice on most systems.
2851 @findex STACK_CHECK_PROTECT
2852 @item STACK_CHECK_PROTECT
2853 The number of bytes of stack needed to recover from a stack overflow,
2854 for languages where such a recovery is supported. The default value of
2855 75 words should be adequate for most machines.
2857 @findex STACK_CHECK_MAX_FRAME_SIZE
2858 @item STACK_CHECK_MAX_FRAME_SIZE
2859 The maximum size of a stack frame, in bytes. GCC will generate probe
2860 instructions in non-leaf functions to ensure at least this many bytes of
2861 stack are available. If a stack frame is larger than this size, stack
2862 checking will not be reliable and GCC will issue a warning. The
2863 default is chosen so that GCC only generates one instruction on most
2864 systems. You should normally not change the default value of this macro.
2866 @findex STACK_CHECK_FIXED_FRAME_SIZE
2867 @item STACK_CHECK_FIXED_FRAME_SIZE
2868 GCC uses this value to generate the above warning message. It
2869 represents the amount of fixed frame used by a function, not including
2870 space for any callee-saved registers, temporaries and user variables.
2871 You need only specify an upper bound for this amount and will normally
2872 use the default of four words.
2874 @findex STACK_CHECK_MAX_VAR_SIZE
2875 @item STACK_CHECK_MAX_VAR_SIZE
2876 The maximum size, in bytes, of an object that GCC will place in the
2877 fixed area of the stack frame when the user specifies
2878 @option{-fstack-check}.
2879 GCC computed the default from the values of the above macros and you will
2880 normally not need to override that default.
2884 @node Frame Registers
2885 @subsection Registers That Address the Stack Frame
2887 @c prevent bad page break with this line
2888 This discusses registers that address the stack frame.
2891 @findex STACK_POINTER_REGNUM
2892 @item STACK_POINTER_REGNUM
2893 The register number of the stack pointer register, which must also be a
2894 fixed register according to @code{FIXED_REGISTERS}. On most machines,
2895 the hardware determines which register this is.
2897 @findex FRAME_POINTER_REGNUM
2898 @item FRAME_POINTER_REGNUM
2899 The register number of the frame pointer register, which is used to
2900 access automatic variables in the stack frame. On some machines, the
2901 hardware determines which register this is. On other machines, you can
2902 choose any register you wish for this purpose.
2904 @findex HARD_FRAME_POINTER_REGNUM
2905 @item HARD_FRAME_POINTER_REGNUM
2906 On some machines the offset between the frame pointer and starting
2907 offset of the automatic variables is not known until after register
2908 allocation has been done (for example, because the saved registers are
2909 between these two locations). On those machines, define
2910 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
2911 be used internally until the offset is known, and define
2912 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
2913 used for the frame pointer.
2915 You should define this macro only in the very rare circumstances when it
2916 is not possible to calculate the offset between the frame pointer and
2917 the automatic variables until after register allocation has been
2918 completed. When this macro is defined, you must also indicate in your
2919 definition of @code{ELIMINABLE_REGS} how to eliminate
2920 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
2921 or @code{STACK_POINTER_REGNUM}.
2923 Do not define this macro if it would be the same as
2924 @code{FRAME_POINTER_REGNUM}.
2926 @findex ARG_POINTER_REGNUM
2927 @item ARG_POINTER_REGNUM
2928 The register number of the arg pointer register, which is used to access
2929 the function's argument list. On some machines, this is the same as the
2930 frame pointer register. On some machines, the hardware determines which
2931 register this is. On other machines, you can choose any register you
2932 wish for this purpose. If this is not the same register as the frame
2933 pointer register, then you must mark it as a fixed register according to
2934 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
2935 (@pxref{Elimination}).
2937 @findex RETURN_ADDRESS_POINTER_REGNUM
2938 @item RETURN_ADDRESS_POINTER_REGNUM
2939 The register number of the return address pointer register, which is used to
2940 access the current function's return address from the stack. On some
2941 machines, the return address is not at a fixed offset from the frame
2942 pointer or stack pointer or argument pointer. This register can be defined
2943 to point to the return address on the stack, and then be converted by
2944 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
2946 Do not define this macro unless there is no other way to get the return
2947 address from the stack.
2949 @findex STATIC_CHAIN_REGNUM
2950 @findex STATIC_CHAIN_INCOMING_REGNUM
2951 @item STATIC_CHAIN_REGNUM
2952 @itemx STATIC_CHAIN_INCOMING_REGNUM
2953 Register numbers used for passing a function's static chain pointer. If
2954 register windows are used, the register number as seen by the called
2955 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
2956 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
2957 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
2960 The static chain register need not be a fixed register.
2962 If the static chain is passed in memory, these macros should not be
2963 defined; instead, the next two macros should be defined.
2965 @findex STATIC_CHAIN
2966 @findex STATIC_CHAIN_INCOMING
2968 @itemx STATIC_CHAIN_INCOMING
2969 If the static chain is passed in memory, these macros provide rtx giving
2970 @code{mem} expressions that denote where they are stored.
2971 @code{STATIC_CHAIN} and @code{STATIC_CHAIN_INCOMING} give the locations
2972 as seen by the calling and called functions, respectively. Often the former
2973 will be at an offset from the stack pointer and the latter at an offset from
2976 @findex stack_pointer_rtx
2977 @findex frame_pointer_rtx
2978 @findex arg_pointer_rtx
2979 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
2980 @code{arg_pointer_rtx} will have been initialized prior to the use of these
2981 macros and should be used to refer to those items.
2983 If the static chain is passed in a register, the two previous macros should
2986 @findex DWARF_FRAME_REGISTERS
2987 @item DWARF_FRAME_REGISTERS
2988 This macro specifies the maximum number of hard registers that can be
2989 saved in a call frame. This is used to size data structures used in
2990 DWARF2 exception handling.
2992 Prior to GCC 3.0, this macro was needed in order to establish a stable
2993 exception handling ABI in the face of adding new hard registers for ISA
2994 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
2995 in the number of hard registers. Nevertheless, this macro can still be
2996 used to reduce the runtime memory requirements of the exception handling
2997 routines, which can be substantial if the ISA contains a lot of
2998 registers that are not call-saved.
3000 If this macro is not defined, it defaults to
3001 @code{FIRST_PSEUDO_REGISTER}.
3003 @findex PRE_GCC3_DWARF_FRAME_REGISTERS
3004 @item PRE_GCC3_DWARF_FRAME_REGISTERS
3006 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3007 for backward compatibility in pre GCC 3.0 compiled code.
3009 If this macro is not defined, it defaults to
3010 @code{DWARF_FRAME_REGISTERS}.
3015 @subsection Eliminating Frame Pointer and Arg Pointer
3017 @c prevent bad page break with this line
3018 This is about eliminating the frame pointer and arg pointer.
3021 @findex FRAME_POINTER_REQUIRED
3022 @item FRAME_POINTER_REQUIRED
3023 A C expression which is nonzero if a function must have and use a frame
3024 pointer. This expression is evaluated in the reload pass. If its value is
3025 nonzero the function will have a frame pointer.
3027 The expression can in principle examine the current function and decide
3028 according to the facts, but on most machines the constant 0 or the
3029 constant 1 suffices. Use 0 when the machine allows code to be generated
3030 with no frame pointer, and doing so saves some time or space. Use 1
3031 when there is no possible advantage to avoiding a frame pointer.
3033 In certain cases, the compiler does not know how to produce valid code
3034 without a frame pointer. The compiler recognizes those cases and
3035 automatically gives the function a frame pointer regardless of what
3036 @code{FRAME_POINTER_REQUIRED} says. You don't need to worry about
3039 In a function that does not require a frame pointer, the frame pointer
3040 register can be allocated for ordinary usage, unless you mark it as a
3041 fixed register. See @code{FIXED_REGISTERS} for more information.
3043 @findex INITIAL_FRAME_POINTER_OFFSET
3044 @findex get_frame_size
3045 @item INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3046 A C statement to store in the variable @var{depth-var} the difference
3047 between the frame pointer and the stack pointer values immediately after
3048 the function prologue. The value would be computed from information
3049 such as the result of @code{get_frame_size ()} and the tables of
3050 registers @code{regs_ever_live} and @code{call_used_regs}.
3052 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3053 need not be defined. Otherwise, it must be defined even if
3054 @code{FRAME_POINTER_REQUIRED} is defined to always be true; in that
3055 case, you may set @var{depth-var} to anything.
3057 @findex ELIMINABLE_REGS
3058 @item ELIMINABLE_REGS
3059 If defined, this macro specifies a table of register pairs used to
3060 eliminate unneeded registers that point into the stack frame. If it is not
3061 defined, the only elimination attempted by the compiler is to replace
3062 references to the frame pointer with references to the stack pointer.
3064 The definition of this macro is a list of structure initializations, each
3065 of which specifies an original and replacement register.
3067 On some machines, the position of the argument pointer is not known until
3068 the compilation is completed. In such a case, a separate hard register
3069 must be used for the argument pointer. This register can be eliminated by
3070 replacing it with either the frame pointer or the argument pointer,
3071 depending on whether or not the frame pointer has been eliminated.
3073 In this case, you might specify:
3075 #define ELIMINABLE_REGS \
3076 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3077 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3078 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3081 Note that the elimination of the argument pointer with the stack pointer is
3082 specified first since that is the preferred elimination.
3084 @findex CAN_ELIMINATE
3085 @item CAN_ELIMINATE (@var{from-reg}, @var{to-reg})
3086 A C expression that returns nonzero if the compiler is allowed to try
3087 to replace register number @var{from-reg} with register number
3088 @var{to-reg}. This macro need only be defined if @code{ELIMINABLE_REGS}
3089 is defined, and will usually be the constant 1, since most of the cases
3090 preventing register elimination are things that the compiler already
3093 @findex INITIAL_ELIMINATION_OFFSET
3094 @item INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3095 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3096 specifies the initial difference between the specified pair of
3097 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3101 @node Stack Arguments
3102 @subsection Passing Function Arguments on the Stack
3103 @cindex arguments on stack
3104 @cindex stack arguments
3106 The macros in this section control how arguments are passed
3107 on the stack. See the following section for other macros that
3108 control passing certain arguments in registers.
3111 @findex PROMOTE_PROTOTYPES
3112 @item PROMOTE_PROTOTYPES
3113 A C expression whose value is nonzero if an argument declared in
3114 a prototype as an integral type smaller than @code{int} should
3115 actually be passed as an @code{int}. In addition to avoiding
3116 errors in certain cases of mismatch, it also makes for better
3117 code on certain machines. If the macro is not defined in target
3118 header files, it defaults to 0.
3122 A C expression. If nonzero, push insns will be used to pass
3124 If the target machine does not have a push instruction, set it to zero.
3125 That directs GCC to use an alternate strategy: to
3126 allocate the entire argument block and then store the arguments into
3127 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3128 On some machines, the definition
3130 @findex PUSH_ROUNDING
3131 @item PUSH_ROUNDING (@var{npushed})
3132 A C expression that is the number of bytes actually pushed onto the
3133 stack when an instruction attempts to push @var{npushed} bytes.
3135 On some machines, the definition
3138 #define PUSH_ROUNDING(BYTES) (BYTES)
3142 will suffice. But on other machines, instructions that appear
3143 to push one byte actually push two bytes in an attempt to maintain
3144 alignment. Then the definition should be
3147 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3150 @findex ACCUMULATE_OUTGOING_ARGS
3151 @findex current_function_outgoing_args_size
3152 @item ACCUMULATE_OUTGOING_ARGS
3153 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3154 will be computed and placed into the variable
3155 @code{current_function_outgoing_args_size}. No space will be pushed
3156 onto the stack for each call; instead, the function prologue should
3157 increase the stack frame size by this amount.
3159 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3162 @findex REG_PARM_STACK_SPACE
3163 @item REG_PARM_STACK_SPACE (@var{fndecl})
3164 Define this macro if functions should assume that stack space has been
3165 allocated for arguments even when their values are passed in
3168 The value of this macro is the size, in bytes, of the area reserved for
3169 arguments passed in registers for the function represented by @var{fndecl},
3170 which can be zero if GCC is calling a library function.
3172 This space can be allocated by the caller, or be a part of the
3173 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3175 @c above is overfull. not sure what to do. --mew 5feb93 did
3176 @c something, not sure if it looks good. --mew 10feb93
3178 @findex MAYBE_REG_PARM_STACK_SPACE
3179 @findex FINAL_REG_PARM_STACK_SPACE
3180 @item MAYBE_REG_PARM_STACK_SPACE
3181 @itemx FINAL_REG_PARM_STACK_SPACE (@var{const_size}, @var{var_size})
3182 Define these macros in addition to the one above if functions might
3183 allocate stack space for arguments even when their values are passed
3184 in registers. These should be used when the stack space allocated
3185 for arguments in registers is not a simple constant independent of the
3186 function declaration.
3188 The value of the first macro is the size, in bytes, of the area that
3189 we should initially assume would be reserved for arguments passed in registers.
3191 The value of the second macro is the actual size, in bytes, of the area
3192 that will be reserved for arguments passed in registers. This takes two
3193 arguments: an integer representing the number of bytes of fixed sized
3194 arguments on the stack, and a tree representing the number of bytes of
3195 variable sized arguments on the stack.
3197 When these macros are defined, @code{REG_PARM_STACK_SPACE} will only be
3198 called for libcall functions, the current function, or for a function
3199 being called when it is known that such stack space must be allocated.
3200 In each case this value can be easily computed.
3202 When deciding whether a called function needs such stack space, and how
3203 much space to reserve, GCC uses these two macros instead of
3204 @code{REG_PARM_STACK_SPACE}.
3206 @findex OUTGOING_REG_PARM_STACK_SPACE
3207 @item OUTGOING_REG_PARM_STACK_SPACE
3208 Define this if it is the responsibility of the caller to allocate the area
3209 reserved for arguments passed in registers.
3211 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3212 whether the space for these arguments counts in the value of
3213 @code{current_function_outgoing_args_size}.
3215 @findex STACK_PARMS_IN_REG_PARM_AREA
3216 @item STACK_PARMS_IN_REG_PARM_AREA
3217 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3218 stack parameters don't skip the area specified by it.
3219 @c i changed this, makes more sens and it should have taken care of the
3220 @c overfull.. not as specific, tho. --mew 5feb93
3222 Normally, when a parameter is not passed in registers, it is placed on the
3223 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3224 suppresses this behavior and causes the parameter to be passed on the
3225 stack in its natural location.
3227 @findex RETURN_POPS_ARGS
3228 @item RETURN_POPS_ARGS (@var{fundecl}, @var{funtype}, @var{stack-size})
3229 A C expression that should indicate the number of bytes of its own
3230 arguments that a function pops on returning, or 0 if the
3231 function pops no arguments and the caller must therefore pop them all
3232 after the function returns.
3234 @var{fundecl} is a C variable whose value is a tree node that describes
3235 the function in question. Normally it is a node of type
3236 @code{FUNCTION_DECL} that describes the declaration of the function.
3237 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3239 @var{funtype} is a C variable whose value is a tree node that
3240 describes the function in question. Normally it is a node of type
3241 @code{FUNCTION_TYPE} that describes the data type of the function.
3242 From this it is possible to obtain the data types of the value and
3243 arguments (if known).
3245 When a call to a library function is being considered, @var{fundecl}
3246 will contain an identifier node for the library function. Thus, if
3247 you need to distinguish among various library functions, you can do so
3248 by their names. Note that ``library function'' in this context means
3249 a function used to perform arithmetic, whose name is known specially
3250 in the compiler and was not mentioned in the C code being compiled.
3252 @var{stack-size} is the number of bytes of arguments passed on the
3253 stack. If a variable number of bytes is passed, it is zero, and
3254 argument popping will always be the responsibility of the calling function.
3256 On the VAX, all functions always pop their arguments, so the definition
3257 of this macro is @var{stack-size}. On the 68000, using the standard
3258 calling convention, no functions pop their arguments, so the value of
3259 the macro is always 0 in this case. But an alternative calling
3260 convention is available in which functions that take a fixed number of
3261 arguments pop them but other functions (such as @code{printf}) pop
3262 nothing (the caller pops all). When this convention is in use,
3263 @var{funtype} is examined to determine whether a function takes a fixed
3264 number of arguments.
3267 @node Register Arguments
3268 @subsection Passing Arguments in Registers
3269 @cindex arguments in registers
3270 @cindex registers arguments
3272 This section describes the macros which let you control how various
3273 types of arguments are passed in registers or how they are arranged in
3277 @findex FUNCTION_ARG
3278 @item FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3279 A C expression that controls whether a function argument is passed
3280 in a register, and which register.
3282 The arguments are @var{cum}, which summarizes all the previous
3283 arguments; @var{mode}, the machine mode of the argument; @var{type},
3284 the data type of the argument as a tree node or 0 if that is not known
3285 (which happens for C support library functions); and @var{named},
3286 which is 1 for an ordinary argument and 0 for nameless arguments that
3287 correspond to @samp{@dots{}} in the called function's prototype.
3288 @var{type} can be an incomplete type if a syntax error has previously
3291 The value of the expression is usually either a @code{reg} RTX for the
3292 hard register in which to pass the argument, or zero to pass the
3293 argument on the stack.
3295 For machines like the VAX and 68000, where normally all arguments are
3296 pushed, zero suffices as a definition.
3298 The value of the expression can also be a @code{parallel} RTX@. This is
3299 used when an argument is passed in multiple locations. The mode of the
3300 of the @code{parallel} should be the mode of the entire argument. The
3301 @code{parallel} holds any number of @code{expr_list} pairs; each one
3302 describes where part of the argument is passed. In each
3303 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
3304 register in which to pass this part of the argument, and the mode of the
3305 register RTX indicates how large this part of the argument is. The
3306 second operand of the @code{expr_list} is a @code{const_int} which gives
3307 the offset in bytes into the entire argument of where this part starts.
3308 As a special exception the first @code{expr_list} in the @code{parallel}
3309 RTX may have a first operand of zero. This indicates that the entire
3310 argument is also stored on the stack.
3312 The last time this macro is called, it is called with @code{MODE ==
3313 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
3314 pattern as operands 2 and 3 respectively.
3316 @cindex @file{stdarg.h} and register arguments
3317 The usual way to make the ISO library @file{stdarg.h} work on a machine
3318 where some arguments are usually passed in registers, is to cause
3319 nameless arguments to be passed on the stack instead. This is done
3320 by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
3322 @cindex @code{MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
3323 @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
3324 You may use the macro @code{MUST_PASS_IN_STACK (@var{mode}, @var{type})}
3325 in the definition of this macro to determine if this argument is of a
3326 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
3327 is not defined and @code{FUNCTION_ARG} returns nonzero for such an
3328 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
3329 defined, the argument will be computed in the stack and then loaded into
3332 @findex MUST_PASS_IN_STACK
3333 @item MUST_PASS_IN_STACK (@var{mode}, @var{type})
3334 Define as a C expression that evaluates to nonzero if we do not know how
3335 to pass TYPE solely in registers. The file @file{expr.h} defines a
3336 definition that is usually appropriate, refer to @file{expr.h} for additional
3339 @findex FUNCTION_INCOMING_ARG
3340 @item FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3341 Define this macro if the target machine has ``register windows'', so
3342 that the register in which a function sees an arguments is not
3343 necessarily the same as the one in which the caller passed the
3346 For such machines, @code{FUNCTION_ARG} computes the register in which
3347 the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
3348 be defined in a similar fashion to tell the function being called
3349 where the arguments will arrive.
3351 If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
3352 serves both purposes.
3354 @findex FUNCTION_ARG_PARTIAL_NREGS
3355 @item FUNCTION_ARG_PARTIAL_NREGS (@var{cum}, @var{mode}, @var{type}, @var{named})
3356 A C expression for the number of words, at the beginning of an
3357 argument, that must be put in registers. The value must be zero for
3358 arguments that are passed entirely in registers or that are entirely
3359 pushed on the stack.
3361 On some machines, certain arguments must be passed partially in
3362 registers and partially in memory. On these machines, typically the
3363 first @var{n} words of arguments are passed in registers, and the rest
3364 on the stack. If a multi-word argument (a @code{double} or a
3365 structure) crosses that boundary, its first few words must be passed
3366 in registers and the rest must be pushed. This macro tells the
3367 compiler when this occurs, and how many of the words should go in
3370 @code{FUNCTION_ARG} for these arguments should return the first
3371 register to be used by the caller for this argument; likewise
3372 @code{FUNCTION_INCOMING_ARG}, for the called function.
3374 @findex FUNCTION_ARG_PASS_BY_REFERENCE
3375 @item FUNCTION_ARG_PASS_BY_REFERENCE (@var{cum}, @var{mode}, @var{type}, @var{named})
3376 A C expression that indicates when an argument must be passed by reference.
3377 If nonzero for an argument, a copy of that argument is made in memory and a
3378 pointer to the argument is passed instead of the argument itself.
3379 The pointer is passed in whatever way is appropriate for passing a pointer
3382 On machines where @code{REG_PARM_STACK_SPACE} is not defined, a suitable
3383 definition of this macro might be
3385 #define FUNCTION_ARG_PASS_BY_REFERENCE\
3386 (CUM, MODE, TYPE, NAMED) \
3387 MUST_PASS_IN_STACK (MODE, TYPE)
3389 @c this is *still* too long. --mew 5feb93
3391 @findex FUNCTION_ARG_CALLEE_COPIES
3392 @item FUNCTION_ARG_CALLEE_COPIES (@var{cum}, @var{mode}, @var{type}, @var{named})
3393 If defined, a C expression that indicates when it is the called function's
3394 responsibility to make a copy of arguments passed by invisible reference.
3395 Normally, the caller makes a copy and passes the address of the copy to the
3396 routine being called. When @code{FUNCTION_ARG_CALLEE_COPIES} is defined and is
3397 nonzero, the caller does not make a copy. Instead, it passes a pointer to the
3398 ``live'' value. The called function must not modify this value. If it can be
3399 determined that the value won't be modified, it need not make a copy;
3400 otherwise a copy must be made.
3402 @findex FUNCTION_ARG_REG_LITTLE_ENDIAN
3403 @item FUNCTION_ARG_REG_LITTLE_ENDIAN
3404 If defined TRUE on a big-endian system then structure arguments passed
3405 (and returned) in registers are passed in a little-endian manner instead of
3406 the big-endian manner. On the HP-UX IA64 and PA64 platforms structures are
3407 aligned differently then integral values and setting this value to true will
3408 allow for the special handling of structure arguments and return values.
3410 @findex CUMULATIVE_ARGS
3411 @item CUMULATIVE_ARGS
3412 A C type for declaring a variable that is used as the first argument of
3413 @code{FUNCTION_ARG} and other related values. For some target machines,
3414 the type @code{int} suffices and can hold the number of bytes of
3417 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
3418 arguments that have been passed on the stack. The compiler has other
3419 variables to keep track of that. For target machines on which all
3420 arguments are passed on the stack, there is no need to store anything in
3421 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
3422 should not be empty, so use @code{int}.
3424 @findex INIT_CUMULATIVE_ARGS
3425 @item INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{indirect})
3426 A C statement (sans semicolon) for initializing the variable @var{cum}
3427 for the state at the beginning of the argument list. The variable has
3428 type @code{CUMULATIVE_ARGS}. The value of @var{fntype} is the tree node
3429 for the data type of the function which will receive the args, or 0
3430 if the args are to a compiler support library function. The value of
3431 @var{indirect} is nonzero when processing an indirect call, for example
3432 a call through a function pointer. The value of @var{indirect} is zero
3433 for a call to an explicitly named function, a library function call, or when
3434 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
3437 When processing a call to a compiler support library function,
3438 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
3439 contains the name of the function, as a string. @var{libname} is 0 when
3440 an ordinary C function call is being processed. Thus, each time this
3441 macro is called, either @var{libname} or @var{fntype} is nonzero, but
3442 never both of them at once.
3444 @findex INIT_CUMULATIVE_LIBCALL_ARGS
3445 @item INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
3446 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
3447 it gets a @code{MODE} argument instead of @var{fntype}, that would be
3448 @code{NULL}. @var{indirect} would always be zero, too. If this macro
3449 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
3450 0)} is used instead.
3452 @findex INIT_CUMULATIVE_INCOMING_ARGS
3453 @item INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
3454 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
3455 finding the arguments for the function being compiled. If this macro is
3456 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
3458 The value passed for @var{libname} is always 0, since library routines
3459 with special calling conventions are never compiled with GCC@. The
3460 argument @var{libname} exists for symmetry with
3461 @code{INIT_CUMULATIVE_ARGS}.
3462 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
3463 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
3465 @findex FUNCTION_ARG_ADVANCE
3466 @item FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
3467 A C statement (sans semicolon) to update the summarizer variable
3468 @var{cum} to advance past an argument in the argument list. The
3469 values @var{mode}, @var{type} and @var{named} describe that argument.
3470 Once this is done, the variable @var{cum} is suitable for analyzing
3471 the @emph{following} argument with @code{FUNCTION_ARG}, etc.
3473 This macro need not do anything if the argument in question was passed
3474 on the stack. The compiler knows how to track the amount of stack space
3475 used for arguments without any special help.
3477 @findex FUNCTION_ARG_PADDING
3478 @item FUNCTION_ARG_PADDING (@var{mode}, @var{type})
3479 If defined, a C expression which determines whether, and in which direction,
3480 to pad out an argument with extra space. The value should be of type
3481 @code{enum direction}: either @code{upward} to pad above the argument,
3482 @code{downward} to pad below, or @code{none} to inhibit padding.
3484 The @emph{amount} of padding is always just enough to reach the next
3485 multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control
3488 This macro has a default definition which is right for most systems.
3489 For little-endian machines, the default is to pad upward. For
3490 big-endian machines, the default is to pad downward for an argument of
3491 constant size shorter than an @code{int}, and upward otherwise.
3493 @findex PAD_VARARGS_DOWN
3494 @item PAD_VARARGS_DOWN
3495 If defined, a C expression which determines whether the default
3496 implementation of va_arg will attempt to pad down before reading the
3497 next argument, if that argument is smaller than its aligned space as
3498 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
3499 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
3501 @findex FUNCTION_ARG_BOUNDARY
3502 @item FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
3503 If defined, a C expression that gives the alignment boundary, in bits,
3504 of an argument with the specified mode and type. If it is not defined,
3505 @code{PARM_BOUNDARY} is used for all arguments.
3507 @findex FUNCTION_ARG_REGNO_P
3508 @item FUNCTION_ARG_REGNO_P (@var{regno})
3509 A C expression that is nonzero if @var{regno} is the number of a hard
3510 register in which function arguments are sometimes passed. This does
3511 @emph{not} include implicit arguments such as the static chain and
3512 the structure-value address. On many machines, no registers can be
3513 used for this purpose since all function arguments are pushed on the
3516 @findex LOAD_ARGS_REVERSED
3517 @item LOAD_ARGS_REVERSED
3518 If defined, the order in which arguments are loaded into their
3519 respective argument registers is reversed so that the last
3520 argument is loaded first. This macro only affects arguments
3521 passed in registers.
3526 @subsection How Scalar Function Values Are Returned
3527 @cindex return values in registers
3528 @cindex values, returned by functions
3529 @cindex scalars, returned as values
3531 This section discusses the macros that control returning scalars as
3532 values---values that can fit in registers.
3535 @findex TRADITIONAL_RETURN_FLOAT
3536 @item TRADITIONAL_RETURN_FLOAT
3537 Define this macro if @option{-traditional} should not cause functions
3538 declared to return @code{float} to convert the value to @code{double}.
3540 @findex FUNCTION_VALUE
3541 @item FUNCTION_VALUE (@var{valtype}, @var{func})
3542 A C expression to create an RTX representing the place where a
3543 function returns a value of data type @var{valtype}. @var{valtype} is
3544 a tree node representing a data type. Write @code{TYPE_MODE
3545 (@var{valtype})} to get the machine mode used to represent that type.
3546 On many machines, only the mode is relevant. (Actually, on most
3547 machines, scalar values are returned in the same place regardless of
3550 The value of the expression is usually a @code{reg} RTX for the hard
3551 register where the return value is stored. The value can also be a
3552 @code{parallel} RTX, if the return value is in multiple places. See
3553 @code{FUNCTION_ARG} for an explanation of the @code{parallel} form.
3555 If @code{PROMOTE_FUNCTION_RETURN} is defined, you must apply the same
3556 promotion rules specified in @code{PROMOTE_MODE} if @var{valtype} is a
3559 If the precise function being called is known, @var{func} is a tree
3560 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
3561 pointer. This makes it possible to use a different value-returning
3562 convention for specific functions when all their calls are
3565 @code{FUNCTION_VALUE} is not used for return vales with aggregate data
3566 types, because these are returned in another way. See
3567 @code{STRUCT_VALUE_REGNUM} and related macros, below.
3569 @findex FUNCTION_OUTGOING_VALUE
3570 @item FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func})
3571 Define this macro if the target machine has ``register windows''
3572 so that the register in which a function returns its value is not
3573 the same as the one in which the caller sees the value.
3575 For such machines, @code{FUNCTION_VALUE} computes the register in which
3576 the caller will see the value. @code{FUNCTION_OUTGOING_VALUE} should be
3577 defined in a similar fashion to tell the function where to put the
3580 If @code{FUNCTION_OUTGOING_VALUE} is not defined,
3581 @code{FUNCTION_VALUE} serves both purposes.
3583 @code{FUNCTION_OUTGOING_VALUE} is not used for return vales with
3584 aggregate data types, because these are returned in another way. See
3585 @code{STRUCT_VALUE_REGNUM} and related macros, below.
3587 @findex LIBCALL_VALUE
3588 @item LIBCALL_VALUE (@var{mode})
3589 A C expression to create an RTX representing the place where a library
3590 function returns a value of mode @var{mode}. If the precise function
3591 being called is known, @var{func} is a tree node
3592 (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
3593 pointer. This makes it possible to use a different value-returning
3594 convention for specific functions when all their calls are
3597 Note that ``library function'' in this context means a compiler
3598 support routine, used to perform arithmetic, whose name is known
3599 specially by the compiler and was not mentioned in the C code being
3602 The definition of @code{LIBRARY_VALUE} need not be concerned aggregate
3603 data types, because none of the library functions returns such types.
3605 @findex FUNCTION_VALUE_REGNO_P
3606 @item FUNCTION_VALUE_REGNO_P (@var{regno})
3607 A C expression that is nonzero if @var{regno} is the number of a hard
3608 register in which the values of called function may come back.
3610 A register whose use for returning values is limited to serving as the
3611 second of a pair (for a value of type @code{double}, say) need not be
3612 recognized by this macro. So for most machines, this definition
3616 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
3619 If the machine has register windows, so that the caller and the called
3620 function use different registers for the return value, this macro
3621 should recognize only the caller's register numbers.
3623 @findex APPLY_RESULT_SIZE
3624 @item APPLY_RESULT_SIZE
3625 Define this macro if @samp{untyped_call} and @samp{untyped_return}
3626 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
3627 saving and restoring an arbitrary return value.
3630 @node Aggregate Return
3631 @subsection How Large Values Are Returned
3632 @cindex aggregates as return values
3633 @cindex large return values
3634 @cindex returning aggregate values
3635 @cindex structure value address
3637 When a function value's mode is @code{BLKmode} (and in some other
3638 cases), the value is not returned according to @code{FUNCTION_VALUE}
3639 (@pxref{Scalar Return}). Instead, the caller passes the address of a
3640 block of memory in which the value should be stored. This address
3641 is called the @dfn{structure value address}.
3643 This section describes how to control returning structure values in
3647 @findex RETURN_IN_MEMORY
3648 @item RETURN_IN_MEMORY (@var{type})
3649 A C expression which can inhibit the returning of certain function
3650 values in registers, based on the type of value. A nonzero value says
3651 to return the function value in memory, just as large structures are
3652 always returned. Here @var{type} will be a C expression of type
3653 @code{tree}, representing the data type of the value.
3655 Note that values of mode @code{BLKmode} must be explicitly handled
3656 by this macro. Also, the option @option{-fpcc-struct-return}
3657 takes effect regardless of this macro. On most systems, it is
3658 possible to leave the macro undefined; this causes a default
3659 definition to be used, whose value is the constant 1 for @code{BLKmode}
3660 values, and 0 otherwise.
3662 Do not use this macro to indicate that structures and unions should always
3663 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
3666 @findex DEFAULT_PCC_STRUCT_RETURN
3667 @item DEFAULT_PCC_STRUCT_RETURN
3668 Define this macro to be 1 if all structure and union return values must be
3669 in memory. Since this results in slower code, this should be defined
3670 only if needed for compatibility with other compilers or with an ABI@.
3671 If you define this macro to be 0, then the conventions used for structure
3672 and union return values are decided by the @code{RETURN_IN_MEMORY} macro.
3674 If not defined, this defaults to the value 1.
3676 @findex STRUCT_VALUE_REGNUM
3677 @item STRUCT_VALUE_REGNUM
3678 If the structure value address is passed in a register, then
3679 @code{STRUCT_VALUE_REGNUM} should be the number of that register.
3681 @findex STRUCT_VALUE
3683 If the structure value address is not passed in a register, define
3684 @code{STRUCT_VALUE} as an expression returning an RTX for the place
3685 where the address is passed. If it returns 0, the address is passed as
3686 an ``invisible'' first argument.
3688 @findex STRUCT_VALUE_INCOMING_REGNUM
3689 @item STRUCT_VALUE_INCOMING_REGNUM
3690 On some architectures the place where the structure value address
3691 is found by the called function is not the same place that the
3692 caller put it. This can be due to register windows, or it could
3693 be because the function prologue moves it to a different place.
3695 If the incoming location of the structure value address is in a
3696 register, define this macro as the register number.
3698 @findex STRUCT_VALUE_INCOMING
3699 @item STRUCT_VALUE_INCOMING
3700 If the incoming location is not a register, then you should define
3701 @code{STRUCT_VALUE_INCOMING} as an expression for an RTX for where the
3702 called function should find the value. If it should find the value on
3703 the stack, define this to create a @code{mem} which refers to the frame
3704 pointer. A definition of 0 means that the address is passed as an
3705 ``invisible'' first argument.
3707 @findex PCC_STATIC_STRUCT_RETURN
3708 @item PCC_STATIC_STRUCT_RETURN
3709 Define this macro if the usual system convention on the target machine
3710 for returning structures and unions is for the called function to return
3711 the address of a static variable containing the value.
3713 Do not define this if the usual system convention is for the caller to
3714 pass an address to the subroutine.
3716 This macro has effect in @option{-fpcc-struct-return} mode, but it does
3717 nothing when you use @option{-freg-struct-return} mode.
3721 @subsection Caller-Saves Register Allocation
3723 If you enable it, GCC can save registers around function calls. This
3724 makes it possible to use call-clobbered registers to hold variables that
3725 must live across calls.
3728 @findex DEFAULT_CALLER_SAVES
3729 @item DEFAULT_CALLER_SAVES
3730 Define this macro if function calls on the target machine do not preserve
3731 any registers; in other words, if @code{CALL_USED_REGISTERS} has 1
3732 for all registers. When defined, this macro enables @option{-fcaller-saves}
3733 by default for all optimization levels. It has no effect for optimization
3734 levels 2 and higher, where @option{-fcaller-saves} is the default.
3736 @findex CALLER_SAVE_PROFITABLE
3737 @item CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
3738 A C expression to determine whether it is worthwhile to consider placing
3739 a pseudo-register in a call-clobbered hard register and saving and
3740 restoring it around each function call. The expression should be 1 when
3741 this is worth doing, and 0 otherwise.
3743 If you don't define this macro, a default is used which is good on most
3744 machines: @code{4 * @var{calls} < @var{refs}}.
3746 @findex HARD_REGNO_CALLER_SAVE_MODE
3747 @item HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
3748 A C expression specifying which mode is required for saving @var{nregs}
3749 of a pseudo-register in call-clobbered hard register @var{regno}. If
3750 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
3751 returned. For most machines this macro need not be defined since GCC
3752 will select the smallest suitable mode.
3755 @node Function Entry
3756 @subsection Function Entry and Exit
3757 @cindex function entry and exit
3761 This section describes the macros that output function entry
3762 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
3764 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
3765 If defined, a function that outputs the assembler code for entry to a
3766 function. The prologue is responsible for setting up the stack frame,
3767 initializing the frame pointer register, saving registers that must be
3768 saved, and allocating @var{size} additional bytes of storage for the
3769 local variables. @var{size} is an integer. @var{file} is a stdio
3770 stream to which the assembler code should be output.
3772 The label for the beginning of the function need not be output by this
3773 macro. That has already been done when the macro is run.
3775 @findex regs_ever_live
3776 To determine which registers to save, the macro can refer to the array
3777 @code{regs_ever_live}: element @var{r} is nonzero if hard register
3778 @var{r} is used anywhere within the function. This implies the function
3779 prologue should save register @var{r}, provided it is not one of the
3780 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
3781 @code{regs_ever_live}.)
3783 On machines that have ``register windows'', the function entry code does
3784 not save on the stack the registers that are in the windows, even if
3785 they are supposed to be preserved by function calls; instead it takes
3786 appropriate steps to ``push'' the register stack, if any non-call-used
3787 registers are used in the function.
3789 @findex frame_pointer_needed
3790 On machines where functions may or may not have frame-pointers, the
3791 function entry code must vary accordingly; it must set up the frame
3792 pointer if one is wanted, and not otherwise. To determine whether a
3793 frame pointer is in wanted, the macro can refer to the variable
3794 @code{frame_pointer_needed}. The variable's value will be 1 at run
3795 time in a function that needs a frame pointer. @xref{Elimination}.
3797 The function entry code is responsible for allocating any stack space
3798 required for the function. This stack space consists of the regions
3799 listed below. In most cases, these regions are allocated in the
3800 order listed, with the last listed region closest to the top of the
3801 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
3802 the highest address if it is not defined). You can use a different order
3803 for a machine if doing so is more convenient or required for
3804 compatibility reasons. Except in cases where required by standard
3805 or by a debugger, there is no reason why the stack layout used by GCC
3806 need agree with that used by other compilers for a machine.
3809 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
3810 If defined, a function that outputs assembler code at the end of a
3811 prologue. This should be used when the function prologue is being
3812 emitted as RTL, and you have some extra assembler that needs to be
3813 emitted. @xref{prologue instruction pattern}.
3816 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
3817 If defined, a function that outputs assembler code at the start of an
3818 epilogue. This should be used when the function epilogue is being
3819 emitted as RTL, and you have some extra assembler that needs to be
3820 emitted. @xref{epilogue instruction pattern}.
3823 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
3824 If defined, a function that outputs the assembler code for exit from a
3825 function. The epilogue is responsible for restoring the saved
3826 registers and stack pointer to their values when the function was
3827 called, and returning control to the caller. This macro takes the
3828 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
3829 registers to restore are determined from @code{regs_ever_live} and
3830 @code{CALL_USED_REGISTERS} in the same way.
3832 On some machines, there is a single instruction that does all the work
3833 of returning from the function. On these machines, give that
3834 instruction the name @samp{return} and do not define the macro
3835 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
3837 Do not define a pattern named @samp{return} if you want the
3838 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
3839 switches to control whether return instructions or epilogues are used,
3840 define a @samp{return} pattern with a validity condition that tests the
3841 target switches appropriately. If the @samp{return} pattern's validity
3842 condition is false, epilogues will be used.
3844 On machines where functions may or may not have frame-pointers, the
3845 function exit code must vary accordingly. Sometimes the code for these
3846 two cases is completely different. To determine whether a frame pointer
3847 is wanted, the macro can refer to the variable
3848 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
3849 a function that needs a frame pointer.
3851 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
3852 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
3853 The C variable @code{current_function_is_leaf} is nonzero for such a
3854 function. @xref{Leaf Functions}.
3856 On some machines, some functions pop their arguments on exit while
3857 others leave that for the caller to do. For example, the 68020 when
3858 given @option{-mrtd} pops arguments in functions that take a fixed
3859 number of arguments.
3861 @findex current_function_pops_args
3862 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
3863 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
3864 needs to know what was decided. The variable that is called
3865 @code{current_function_pops_args} is the number of bytes of its
3866 arguments that a function should pop. @xref{Scalar Return}.
3867 @c what is the "its arguments" in the above sentence referring to, pray
3868 @c tell? --mew 5feb93
3875 @findex current_function_pretend_args_size
3876 A region of @code{current_function_pretend_args_size} bytes of
3877 uninitialized space just underneath the first argument arriving on the
3878 stack. (This may not be at the very start of the allocated stack region
3879 if the calling sequence has pushed anything else since pushing the stack
3880 arguments. But usually, on such machines, nothing else has been pushed
3881 yet, because the function prologue itself does all the pushing.) This
3882 region is used on machines where an argument may be passed partly in
3883 registers and partly in memory, and, in some cases to support the
3884 features in @code{<varargs.h>} and @code{<stdarg.h>}.
3887 An area of memory used to save certain registers used by the function.
3888 The size of this area, which may also include space for such things as
3889 the return address and pointers to previous stack frames, is
3890 machine-specific and usually depends on which registers have been used
3891 in the function. Machines with register windows often do not require
3895 A region of at least @var{size} bytes, possibly rounded up to an allocation
3896 boundary, to contain the local variables of the function. On some machines,
3897 this region and the save area may occur in the opposite order, with the
3898 save area closer to the top of the stack.
3901 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
3902 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
3903 @code{current_function_outgoing_args_size} bytes to be used for outgoing
3904 argument lists of the function. @xref{Stack Arguments}.
3907 Normally, it is necessary for the macros
3908 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
3909 @code{TARGET_ASM_FUNCTION_EPILOGUE} to treat leaf functions specially.
3910 The C variable @code{current_function_is_leaf} is nonzero for such a
3913 @findex EXIT_IGNORE_STACK
3914 @item EXIT_IGNORE_STACK
3915 Define this macro as a C expression that is nonzero if the return
3916 instruction or the function epilogue ignores the value of the stack
3917 pointer; in other words, if it is safe to delete an instruction to
3918 adjust the stack pointer before a return from the function.
3920 Note that this macro's value is relevant only for functions for which
3921 frame pointers are maintained. It is never safe to delete a final
3922 stack adjustment in a function that has no frame pointer, and the
3923 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
3925 @findex EPILOGUE_USES
3926 @item EPILOGUE_USES (@var{regno})
3927 Define this macro as a C expression that is nonzero for registers that are
3928 used by the epilogue or the @samp{return} pattern. The stack and frame
3929 pointer registers are already be assumed to be used as needed.
3931 @findex DELAY_SLOTS_FOR_EPILOGUE
3932 @item DELAY_SLOTS_FOR_EPILOGUE
3933 Define this macro if the function epilogue contains delay slots to which
3934 instructions from the rest of the function can be ``moved''. The
3935 definition should be a C expression whose value is an integer
3936 representing the number of delay slots there.
3938 @findex ELIGIBLE_FOR_EPILOGUE_DELAY
3939 @item ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
3940 A C expression that returns 1 if @var{insn} can be placed in delay
3941 slot number @var{n} of the epilogue.
3943 The argument @var{n} is an integer which identifies the delay slot now
3944 being considered (since different slots may have different rules of
3945 eligibility). It is never negative and is always less than the number
3946 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
3947 If you reject a particular insn for a given delay slot, in principle, it
3948 may be reconsidered for a subsequent delay slot. Also, other insns may
3949 (at least in principle) be considered for the so far unfilled delay
3952 @findex current_function_epilogue_delay_list
3953 @findex final_scan_insn
3954 The insns accepted to fill the epilogue delay slots are put in an RTL
3955 list made with @code{insn_list} objects, stored in the variable
3956 @code{current_function_epilogue_delay_list}. The insn for the first
3957 delay slot comes first in the list. Your definition of the macro
3958 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
3959 outputting the insns in this list, usually by calling
3960 @code{final_scan_insn}.
3962 You need not define this macro if you did not define
3963 @code{DELAY_SLOTS_FOR_EPILOGUE}.
3965 @findex ASM_OUTPUT_MI_THUNK
3966 @item ASM_OUTPUT_MI_THUNK (@var{file}, @var{thunk_fndecl}, @var{delta}, @var{function})
3967 A C compound statement that outputs the assembler code for a thunk
3968 function, used to implement C++ virtual function calls with multiple
3969 inheritance. The thunk acts as a wrapper around a virtual function,
3970 adjusting the implicit object parameter before handing control off to
3973 First, emit code to add the integer @var{delta} to the location that
3974 contains the incoming first argument. Assume that this argument
3975 contains a pointer, and is the one used to pass the @code{this} pointer
3976 in C++. This is the incoming argument @emph{before} the function prologue,
3977 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
3978 all other incoming arguments.
3980 After the addition, emit code to jump to @var{function}, which is a
3981 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
3982 not touch the return address. Hence returning from @var{FUNCTION} will
3983 return to whoever called the current @samp{thunk}.
3985 The effect must be as if @var{function} had been called directly with
3986 the adjusted first argument. This macro is responsible for emitting all
3987 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
3988 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
3990 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
3991 have already been extracted from it.) It might possibly be useful on
3992 some targets, but probably not.
3994 If you do not define this macro, the target-independent code in the C++
3995 front end will generate a less efficient heavyweight thunk that calls
3996 @var{function} instead of jumping to it. The generic approach does
3997 not support varargs.
4001 @subsection Generating Code for Profiling
4002 @cindex profiling, code generation
4004 These macros will help you generate code for profiling.
4007 @findex FUNCTION_PROFILER
4008 @item FUNCTION_PROFILER (@var{file}, @var{labelno})
4009 A C statement or compound statement to output to @var{file} some
4010 assembler code to call the profiling subroutine @code{mcount}.
4013 The details of how @code{mcount} expects to be called are determined by
4014 your operating system environment, not by GCC@. To figure them out,
4015 compile a small program for profiling using the system's installed C
4016 compiler and look at the assembler code that results.
4018 Older implementations of @code{mcount} expect the address of a counter
4019 variable to be loaded into some register. The name of this variable is
4020 @samp{LP} followed by the number @var{labelno}, so you would generate
4021 the name using @samp{LP%d} in a @code{fprintf}.
4023 @findex PROFILE_HOOK
4025 A C statement or compound statement to output to @var{file} some assembly
4026 code to call the profiling subroutine @code{mcount} even the target does
4027 not support profiling.
4029 @findex NO_PROFILE_COUNTERS
4030 @item NO_PROFILE_COUNTERS
4031 Define this macro if the @code{mcount} subroutine on your system does
4032 not need a counter variable allocated for each function. This is true
4033 for almost all modern implementations. If you define this macro, you
4034 must not use the @var{labelno} argument to @code{FUNCTION_PROFILER}.
4036 @findex PROFILE_BEFORE_PROLOGUE
4037 @item PROFILE_BEFORE_PROLOGUE
4038 Define this macro if the code for function profiling should come before
4039 the function prologue. Normally, the profiling code comes after.
4042 @findex TARGET_ALLOWS_PROFILING_WITHOUT_FRAME_POINTER
4043 @item TARGET_ALLOWS_PROFILING_WITHOUT_FRAME_POINTER
4044 On some targets, it is impossible to use profiling when the frame
4045 pointer has been omitted. For example, on x86 GNU/Linux systems,
4046 the @code{mcount} routine provided by the GNU C Library finds the
4047 address of the routine that called the routine that called @code{mcount}
4048 by looking in the immediate caller's stack frame. If the immediate
4049 caller has no frame pointer, this lookup will fail.
4051 By default, GCC assumes that the target does allow profiling when the
4052 frame pointer is omitted. This macro should be defined to a C
4053 expression that evaluates to @code{false} if the target does not allow
4054 profiling when the frame pointer is omitted.
4059 @subsection Permitting tail calls
4063 @findex FUNCTION_OK_FOR_SIBCALL
4064 @item FUNCTION_OK_FOR_SIBCALL (@var{decl})
4065 A C expression that evaluates to true if it is ok to perform a sibling
4066 call to @var{decl} from the current function.
4068 It is not uncommon for limitations of calling conventions to prevent
4069 tail calls to functions outside the current unit of translation, or
4070 during PIC compilation. Use this macro to enforce these restrictions,
4071 as the @code{sibcall} md pattern can not fail, or fall over to a
4076 @section Implementing the Varargs Macros
4077 @cindex varargs implementation
4079 GCC comes with an implementation of @code{<varargs.h>} and
4080 @code{<stdarg.h>} that work without change on machines that pass arguments
4081 on the stack. Other machines require their own implementations of
4082 varargs, and the two machine independent header files must have
4083 conditionals to include it.
4085 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4086 the calling convention for @code{va_start}. The traditional
4087 implementation takes just one argument, which is the variable in which
4088 to store the argument pointer. The ISO implementation of
4089 @code{va_start} takes an additional second argument. The user is
4090 supposed to write the last named argument of the function here.
4092 However, @code{va_start} should not use this argument. The way to find
4093 the end of the named arguments is with the built-in functions described
4097 @findex __builtin_saveregs
4098 @item __builtin_saveregs ()
4099 Use this built-in function to save the argument registers in memory so
4100 that the varargs mechanism can access them. Both ISO and traditional
4101 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4102 you use @code{SETUP_INCOMING_VARARGS} (see below) instead.
4104 On some machines, @code{__builtin_saveregs} is open-coded under the
4105 control of the macro @code{EXPAND_BUILTIN_SAVEREGS}. On other machines,
4106 it calls a routine written in assembler language, found in
4109 Code generated for the call to @code{__builtin_saveregs} appears at the
4110 beginning of the function, as opposed to where the call to
4111 @code{__builtin_saveregs} is written, regardless of what the code is.
4112 This is because the registers must be saved before the function starts
4113 to use them for its own purposes.
4114 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4117 @findex __builtin_args_info
4118 @item __builtin_args_info (@var{category})
4119 Use this built-in function to find the first anonymous arguments in
4122 In general, a machine may have several categories of registers used for
4123 arguments, each for a particular category of data types. (For example,
4124 on some machines, floating-point registers are used for floating-point
4125 arguments while other arguments are passed in the general registers.)
4126 To make non-varargs functions use the proper calling convention, you
4127 have defined the @code{CUMULATIVE_ARGS} data type to record how many
4128 registers in each category have been used so far
4130 @code{__builtin_args_info} accesses the same data structure of type
4131 @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
4132 with it, with @var{category} specifying which word to access. Thus, the
4133 value indicates the first unused register in a given category.
4135 Normally, you would use @code{__builtin_args_info} in the implementation
4136 of @code{va_start}, accessing each category just once and storing the
4137 value in the @code{va_list} object. This is because @code{va_list} will
4138 have to update the values, and there is no way to alter the
4139 values accessed by @code{__builtin_args_info}.
4141 @findex __builtin_next_arg
4142 @item __builtin_next_arg (@var{lastarg})
4143 This is the equivalent of @code{__builtin_args_info}, for stack
4144 arguments. It returns the address of the first anonymous stack
4145 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
4146 returns the address of the location above the first anonymous stack
4147 argument. Use it in @code{va_start} to initialize the pointer for
4148 fetching arguments from the stack. Also use it in @code{va_start} to
4149 verify that the second parameter @var{lastarg} is the last named argument
4150 of the current function.
4152 @findex __builtin_classify_type
4153 @item __builtin_classify_type (@var{object})
4154 Since each machine has its own conventions for which data types are
4155 passed in which kind of register, your implementation of @code{va_arg}
4156 has to embody these conventions. The easiest way to categorize the
4157 specified data type is to use @code{__builtin_classify_type} together
4158 with @code{sizeof} and @code{__alignof__}.
4160 @code{__builtin_classify_type} ignores the value of @var{object},
4161 considering only its data type. It returns an integer describing what
4162 kind of type that is---integer, floating, pointer, structure, and so on.
4164 The file @file{typeclass.h} defines an enumeration that you can use to
4165 interpret the values of @code{__builtin_classify_type}.
4168 These machine description macros help implement varargs:
4171 @findex EXPAND_BUILTIN_SAVEREGS
4172 @item EXPAND_BUILTIN_SAVEREGS ()
4173 If defined, is a C expression that produces the machine-specific code
4174 for a call to @code{__builtin_saveregs}. This code will be moved to the
4175 very beginning of the function, before any parameter access are made.
4176 The return value of this function should be an RTX that contains the
4177 value to use as the return of @code{__builtin_saveregs}.
4179 @findex SETUP_INCOMING_VARARGS
4180 @item SETUP_INCOMING_VARARGS (@var{args_so_far}, @var{mode}, @var{type}, @var{pretend_args_size}, @var{second_time})
4181 This macro offers an alternative to using @code{__builtin_saveregs} and
4182 defining the macro @code{EXPAND_BUILTIN_SAVEREGS}. Use it to store the
4183 anonymous register arguments into the stack so that all the arguments
4184 appear to have been passed consecutively on the stack. Once this is
4185 done, you can use the standard implementation of varargs that works for
4186 machines that pass all their arguments on the stack.
4188 The argument @var{args_so_far} is the @code{CUMULATIVE_ARGS} data
4189 structure, containing the values that are obtained after processing the
4190 named arguments. The arguments @var{mode} and @var{type} describe the
4191 last named argument---its machine mode and its data type as a tree node.
4193 The macro implementation should do two things: first, push onto the
4194 stack all the argument registers @emph{not} used for the named
4195 arguments, and second, store the size of the data thus pushed into the
4196 @code{int}-valued variable whose name is supplied as the argument
4197 @var{pretend_args_size}. The value that you store here will serve as
4198 additional offset for setting up the stack frame.
4200 Because you must generate code to push the anonymous arguments at
4201 compile time without knowing their data types,
4202 @code{SETUP_INCOMING_VARARGS} is only useful on machines that have just
4203 a single category of argument register and use it uniformly for all data
4206 If the argument @var{second_time} is nonzero, it means that the
4207 arguments of the function are being analyzed for the second time. This
4208 happens for an inline function, which is not actually compiled until the
4209 end of the source file. The macro @code{SETUP_INCOMING_VARARGS} should
4210 not generate any instructions in this case.
4212 @findex STRICT_ARGUMENT_NAMING
4213 @item STRICT_ARGUMENT_NAMING
4214 Define this macro to be a nonzero value if the location where a function
4215 argument is passed depends on whether or not it is a named argument.
4217 This macro controls how the @var{named} argument to @code{FUNCTION_ARG}
4218 is set for varargs and stdarg functions. If this macro returns a
4219 nonzero value, the @var{named} argument is always true for named
4220 arguments, and false for unnamed arguments. If it returns a value of
4221 zero, but @code{SETUP_INCOMING_VARARGS} is defined, then all arguments
4222 are treated as named. Otherwise, all named arguments except the last
4223 are treated as named.
4225 You need not define this macro if it always returns zero.
4227 @findex PRETEND_OUTGOING_VARARGS_NAMED
4228 @item PRETEND_OUTGOING_VARARGS_NAMED
4229 If you need to conditionally change ABIs so that one works with
4230 @code{SETUP_INCOMING_VARARGS}, but the other works like neither
4231 @code{SETUP_INCOMING_VARARGS} nor @code{STRICT_ARGUMENT_NAMING} was
4232 defined, then define this macro to return nonzero if
4233 @code{SETUP_INCOMING_VARARGS} is used, zero otherwise.
4234 Otherwise, you should not define this macro.
4238 @section Trampolines for Nested Functions
4239 @cindex trampolines for nested functions
4240 @cindex nested functions, trampolines for
4242 A @dfn{trampoline} is a small piece of code that is created at run time
4243 when the address of a nested function is taken. It normally resides on
4244 the stack, in the stack frame of the containing function. These macros
4245 tell GCC how to generate code to allocate and initialize a
4248 The instructions in the trampoline must do two things: load a constant
4249 address into the static chain register, and jump to the real address of
4250 the nested function. On CISC machines such as the m68k, this requires
4251 two instructions, a move immediate and a jump. Then the two addresses
4252 exist in the trampoline as word-long immediate operands. On RISC
4253 machines, it is often necessary to load each address into a register in
4254 two parts. Then pieces of each address form separate immediate
4257 The code generated to initialize the trampoline must store the variable
4258 parts---the static chain value and the function address---into the
4259 immediate operands of the instructions. On a CISC machine, this is
4260 simply a matter of copying each address to a memory reference at the
4261 proper offset from the start of the trampoline. On a RISC machine, it
4262 may be necessary to take out pieces of the address and store them
4266 @findex TRAMPOLINE_TEMPLATE
4267 @item TRAMPOLINE_TEMPLATE (@var{file})
4268 A C statement to output, on the stream @var{file}, assembler code for a
4269 block of data that contains the constant parts of a trampoline. This
4270 code should not include a label---the label is taken care of
4273 If you do not define this macro, it means no template is needed
4274 for the target. Do not define this macro on systems where the block move
4275 code to copy the trampoline into place would be larger than the code
4276 to generate it on the spot.
4278 @findex TRAMPOLINE_SECTION
4279 @item TRAMPOLINE_SECTION
4280 The name of a subroutine to switch to the section in which the
4281 trampoline template is to be placed (@pxref{Sections}). The default is
4282 a value of @samp{readonly_data_section}, which places the trampoline in
4283 the section containing read-only data.
4285 @findex TRAMPOLINE_SIZE
4286 @item TRAMPOLINE_SIZE
4287 A C expression for the size in bytes of the trampoline, as an integer.
4289 @findex TRAMPOLINE_ALIGNMENT
4290 @item TRAMPOLINE_ALIGNMENT
4291 Alignment required for trampolines, in bits.
4293 If you don't define this macro, the value of @code{BIGGEST_ALIGNMENT}
4294 is used for aligning trampolines.
4296 @findex INITIALIZE_TRAMPOLINE
4297 @item INITIALIZE_TRAMPOLINE (@var{addr}, @var{fnaddr}, @var{static_chain})
4298 A C statement to initialize the variable parts of a trampoline.
4299 @var{addr} is an RTX for the address of the trampoline; @var{fnaddr} is
4300 an RTX for the address of the nested function; @var{static_chain} is an
4301 RTX for the static chain value that should be passed to the function
4304 @findex TRAMPOLINE_ADJUST_ADDRESS
4305 @item TRAMPOLINE_ADJUST_ADDRESS (@var{addr})
4306 A C statement that should perform any machine-specific adjustment in
4307 the address of the trampoline. Its argument contains the address that
4308 was passed to @code{INITIALIZE_TRAMPOLINE}. In case the address to be
4309 used for a function call should be different from the address in which
4310 the template was stored, the different address should be assigned to
4311 @var{addr}. If this macro is not defined, @var{addr} will be used for
4314 @findex ALLOCATE_TRAMPOLINE
4315 @item ALLOCATE_TRAMPOLINE (@var{fp})
4316 A C expression to allocate run-time space for a trampoline. The
4317 expression value should be an RTX representing a memory reference to the
4318 space for the trampoline.
4320 @cindex @code{TARGET_ASM_FUNCTION_EPILOGUE} and trampolines
4321 @cindex @code{TARGET_ASM_FUNCTION_PROLOGUE} and trampolines
4322 If this macro is not defined, by default the trampoline is allocated as
4323 a stack slot. This default is right for most machines. The exceptions
4324 are machines where it is impossible to execute instructions in the stack
4325 area. On such machines, you may have to implement a separate stack,
4326 using this macro in conjunction with @code{TARGET_ASM_FUNCTION_PROLOGUE}
4327 and @code{TARGET_ASM_FUNCTION_EPILOGUE}.
4329 @var{fp} points to a data structure, a @code{struct function}, which
4330 describes the compilation status of the immediate containing function of
4331 the function which the trampoline is for. Normally (when
4332 @code{ALLOCATE_TRAMPOLINE} is not defined), the stack slot for the
4333 trampoline is in the stack frame of this containing function. Other
4334 allocation strategies probably must do something analogous with this
4338 Implementing trampolines is difficult on many machines because they have
4339 separate instruction and data caches. Writing into a stack location
4340 fails to clear the memory in the instruction cache, so when the program
4341 jumps to that location, it executes the old contents.
4343 Here are two possible solutions. One is to clear the relevant parts of
4344 the instruction cache whenever a trampoline is set up. The other is to
4345 make all trampolines identical, by having them jump to a standard
4346 subroutine. The former technique makes trampoline execution faster; the
4347 latter makes initialization faster.
4349 To clear the instruction cache when a trampoline is initialized, define
4350 the following macros which describe the shape of the cache.
4353 @findex INSN_CACHE_SIZE
4354 @item INSN_CACHE_SIZE
4355 The total size in bytes of the cache.
4357 @findex INSN_CACHE_LINE_WIDTH
4358 @item INSN_CACHE_LINE_WIDTH
4359 The length in bytes of each cache line. The cache is divided into cache
4360 lines which are disjoint slots, each holding a contiguous chunk of data
4361 fetched from memory. Each time data is brought into the cache, an
4362 entire line is read at once. The data loaded into a cache line is
4363 always aligned on a boundary equal to the line size.
4365 @findex INSN_CACHE_DEPTH
4366 @item INSN_CACHE_DEPTH
4367 The number of alternative cache lines that can hold any particular memory
4371 Alternatively, if the machine has system calls or instructions to clear
4372 the instruction cache directly, you can define the following macro.
4375 @findex CLEAR_INSN_CACHE
4376 @item CLEAR_INSN_CACHE (@var{beg}, @var{end})
4377 If defined, expands to a C expression clearing the @emph{instruction
4378 cache} in the specified interval. If it is not defined, and the macro
4379 @code{INSN_CACHE_SIZE} is defined, some generic code is generated to clear the
4380 cache. The definition of this macro would typically be a series of
4381 @code{asm} statements. Both @var{beg} and @var{end} are both pointer
4385 To use a standard subroutine, define the following macro. In addition,
4386 you must make sure that the instructions in a trampoline fill an entire
4387 cache line with identical instructions, or else ensure that the
4388 beginning of the trampoline code is always aligned at the same point in
4389 its cache line. Look in @file{m68k.h} as a guide.
4392 @findex TRANSFER_FROM_TRAMPOLINE
4393 @item TRANSFER_FROM_TRAMPOLINE
4394 Define this macro if trampolines need a special subroutine to do their
4395 work. The macro should expand to a series of @code{asm} statements
4396 which will be compiled with GCC@. They go in a library function named
4397 @code{__transfer_from_trampoline}.
4399 If you need to avoid executing the ordinary prologue code of a compiled
4400 C function when you jump to the subroutine, you can do so by placing a
4401 special label of your own in the assembler code. Use one @code{asm}
4402 statement to generate an assembler label, and another to make the label
4403 global. Then trampolines can use that label to jump directly to your
4404 special assembler code.
4408 @section Implicit Calls to Library Routines
4409 @cindex library subroutine names
4410 @cindex @file{libgcc.a}
4412 @c prevent bad page break with this line
4413 Here is an explanation of implicit calls to library routines.
4416 @findex MULSI3_LIBCALL
4417 @item MULSI3_LIBCALL
4418 A C string constant giving the name of the function to call for
4419 multiplication of one signed full-word by another. If you do not
4420 define this macro, the default name is used, which is @code{__mulsi3},
4421 a function defined in @file{libgcc.a}.
4423 @findex DIVSI3_LIBCALL
4424 @item DIVSI3_LIBCALL
4425 A C string constant giving the name of the function to call for
4426 division of one signed full-word by another. If you do not define
4427 this macro, the default name is used, which is @code{__divsi3}, a
4428 function defined in @file{libgcc.a}.
4430 @findex UDIVSI3_LIBCALL
4431 @item UDIVSI3_LIBCALL
4432 A C string constant giving the name of the function to call for
4433 division of one unsigned full-word by another. If you do not define
4434 this macro, the default name is used, which is @code{__udivsi3}, a
4435 function defined in @file{libgcc.a}.
4437 @findex MODSI3_LIBCALL
4438 @item MODSI3_LIBCALL
4439 A C string constant giving the name of the function to call for the
4440 remainder in division of one signed full-word by another. If you do
4441 not define this macro, the default name is used, which is
4442 @code{__modsi3}, a function defined in @file{libgcc.a}.
4444 @findex UMODSI3_LIBCALL
4445 @item UMODSI3_LIBCALL
4446 A C string constant giving the name of the function to call for the
4447 remainder in division of one unsigned full-word by another. If you do
4448 not define this macro, the default name is used, which is
4449 @code{__umodsi3}, a function defined in @file{libgcc.a}.
4451 @findex MULDI3_LIBCALL
4452 @item MULDI3_LIBCALL
4453 A C string constant giving the name of the function to call for
4454 multiplication of one signed double-word by another. If you do not
4455 define this macro, the default name is used, which is @code{__muldi3},
4456 a function defined in @file{libgcc.a}.
4458 @findex DIVDI3_LIBCALL
4459 @item DIVDI3_LIBCALL
4460 A C string constant giving the name of the function to call for
4461 division of one signed double-word by another. If you do not define
4462 this macro, the default name is used, which is @code{__divdi3}, a
4463 function defined in @file{libgcc.a}.
4465 @findex UDIVDI3_LIBCALL
4466 @item UDIVDI3_LIBCALL
4467 A C string constant giving the name of the function to call for
4468 division of one unsigned full-word by another. If you do not define
4469 this macro, the default name is used, which is @code{__udivdi3}, a
4470 function defined in @file{libgcc.a}.
4472 @findex MODDI3_LIBCALL
4473 @item MODDI3_LIBCALL
4474 A C string constant giving the name of the function to call for the
4475 remainder in division of one signed double-word by another. If you do
4476 not define this macro, the default name is used, which is
4477 @code{__moddi3}, a function defined in @file{libgcc.a}.
4479 @findex UMODDI3_LIBCALL
4480 @item UMODDI3_LIBCALL
4481 A C string constant giving the name of the function to call for the
4482 remainder in division of one unsigned full-word by another. If you do
4483 not define this macro, the default name is used, which is
4484 @code{__umoddi3}, a function defined in @file{libgcc.a}.
4486 @findex INIT_TARGET_OPTABS
4487 @item INIT_TARGET_OPTABS
4488 Define this macro as a C statement that declares additional library
4489 routines renames existing ones. @code{init_optabs} calls this macro after
4490 initializing all the normal library routines.
4492 @findex FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
4493 @item FLOAT_LIB_COMPARE_RETURNS_BOOL
4494 Define this macro as a C statement that returns nonzero if a call to
4495 the floating point comparison library function will return a boolean
4496 value that indicates the result of the comparison. It should return
4497 zero if one of gcc's own libgcc functions is called.
4499 Most ports don't need to define this macro.
4502 @cindex @code{EDOM}, implicit usage
4504 The value of @code{EDOM} on the target machine, as a C integer constant
4505 expression. If you don't define this macro, GCC does not attempt to
4506 deposit the value of @code{EDOM} into @code{errno} directly. Look in
4507 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
4510 If you do not define @code{TARGET_EDOM}, then compiled code reports
4511 domain errors by calling the library function and letting it report the
4512 error. If mathematical functions on your system use @code{matherr} when
4513 there is an error, then you should leave @code{TARGET_EDOM} undefined so
4514 that @code{matherr} is used normally.
4516 @findex GEN_ERRNO_RTX
4517 @cindex @code{errno}, implicit usage
4519 Define this macro as a C expression to create an rtl expression that
4520 refers to the global ``variable'' @code{errno}. (On certain systems,
4521 @code{errno} may not actually be a variable.) If you don't define this
4522 macro, a reasonable default is used.
4524 @findex TARGET_MEM_FUNCTIONS
4525 @cindex @code{bcopy}, implicit usage
4526 @cindex @code{memcpy}, implicit usage
4527 @cindex @code{memmove}, implicit usage
4528 @cindex @code{bzero}, implicit usage
4529 @cindex @code{memset}, implicit usage
4530 @item TARGET_MEM_FUNCTIONS
4531 Define this macro if GCC should generate calls to the ISO C
4532 (and System V) library functions @code{memcpy}, @code{memmove} and
4533 @code{memset} rather than the BSD functions @code{bcopy} and @code{bzero}.
4535 @findex LIBGCC_NEEDS_DOUBLE
4536 @item LIBGCC_NEEDS_DOUBLE
4537 Define this macro if @code{float} arguments cannot be passed to library
4538 routines (so they must be converted to @code{double}). This macro
4539 affects both how library calls are generated and how the library
4540 routines in @file{libgcc.a} accept their arguments. It is useful on
4541 machines where floating and fixed point arguments are passed
4542 differently, such as the i860.
4544 @findex NEXT_OBJC_RUNTIME
4545 @item NEXT_OBJC_RUNTIME
4546 Define this macro to generate code for Objective-C message sending using
4547 the calling convention of the NeXT system. This calling convention
4548 involves passing the object, the selector and the method arguments all
4549 at once to the method-lookup library function.
4551 The default calling convention passes just the object and the selector
4552 to the lookup function, which returns a pointer to the method.
4555 @node Addressing Modes
4556 @section Addressing Modes
4557 @cindex addressing modes
4559 @c prevent bad page break with this line
4560 This is about addressing modes.
4563 @findex HAVE_PRE_INCREMENT
4564 @findex HAVE_PRE_DECREMENT
4565 @findex HAVE_POST_INCREMENT
4566 @findex HAVE_POST_DECREMENT
4567 @item HAVE_PRE_INCREMENT
4568 @itemx HAVE_PRE_DECREMENT
4569 @itemx HAVE_POST_INCREMENT
4570 @itemx HAVE_POST_DECREMENT
4571 A C expression that is nonzero if the machine supports pre-increment,
4572 pre-decrement, post-increment, or post-decrement addressing respectively.
4574 @findex HAVE_POST_MODIFY_DISP
4575 @findex HAVE_PRE_MODIFY_DISP
4576 @item HAVE_PRE_MODIFY_DISP
4577 @itemx HAVE_POST_MODIFY_DISP
4578 A C expression that is nonzero if the machine supports pre- or
4579 post-address side-effect generation involving constants other than
4580 the size of the memory operand.
4582 @findex HAVE_POST_MODIFY_REG
4583 @findex HAVE_PRE_MODIFY_REG
4584 @item HAVE_PRE_MODIFY_REG
4585 @itemx HAVE_POST_MODIFY_REG
4586 A C expression that is nonzero if the machine supports pre- or
4587 post-address side-effect generation involving a register displacement.
4589 @findex CONSTANT_ADDRESS_P
4590 @item CONSTANT_ADDRESS_P (@var{x})
4591 A C expression that is 1 if the RTX @var{x} is a constant which
4592 is a valid address. On most machines, this can be defined as
4593 @code{CONSTANT_P (@var{x})}, but a few machines are more restrictive
4594 in which constant addresses are supported.
4597 @code{CONSTANT_P} accepts integer-values expressions whose values are
4598 not explicitly known, such as @code{symbol_ref}, @code{label_ref}, and
4599 @code{high} expressions and @code{const} arithmetic expressions, in
4600 addition to @code{const_int} and @code{const_double} expressions.
4602 @findex MAX_REGS_PER_ADDRESS
4603 @item MAX_REGS_PER_ADDRESS
4604 A number, the maximum number of registers that can appear in a valid
4605 memory address. Note that it is up to you to specify a value equal to
4606 the maximum number that @code{GO_IF_LEGITIMATE_ADDRESS} would ever
4609 @findex GO_IF_LEGITIMATE_ADDRESS
4610 @item GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
4611 A C compound statement with a conditional @code{goto @var{label};}
4612 executed if @var{x} (an RTX) is a legitimate memory address on the
4613 target machine for a memory operand of mode @var{mode}.
4615 It usually pays to define several simpler macros to serve as
4616 subroutines for this one. Otherwise it may be too complicated to
4619 This macro must exist in two variants: a strict variant and a
4620 non-strict one. The strict variant is used in the reload pass. It
4621 must be defined so that any pseudo-register that has not been
4622 allocated a hard register is considered a memory reference. In
4623 contexts where some kind of register is required, a pseudo-register
4624 with no hard register must be rejected.
4626 The non-strict variant is used in other passes. It must be defined to
4627 accept all pseudo-registers in every context where some kind of
4628 register is required.
4630 @findex REG_OK_STRICT
4631 Compiler source files that want to use the strict variant of this
4632 macro define the macro @code{REG_OK_STRICT}. You should use an
4633 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant
4634 in that case and the non-strict variant otherwise.
4636 Subroutines to check for acceptable registers for various purposes (one
4637 for base registers, one for index registers, and so on) are typically
4638 among the subroutines used to define @code{GO_IF_LEGITIMATE_ADDRESS}.
4639 Then only these subroutine macros need have two variants; the higher
4640 levels of macros may be the same whether strict or not.
4642 Normally, constant addresses which are the sum of a @code{symbol_ref}
4643 and an integer are stored inside a @code{const} RTX to mark them as
4644 constant. Therefore, there is no need to recognize such sums
4645 specifically as legitimate addresses. Normally you would simply
4646 recognize any @code{const} as legitimate.
4648 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
4649 sums that are not marked with @code{const}. It assumes that a naked
4650 @code{plus} indicates indexing. If so, then you @emph{must} reject such
4651 naked constant sums as illegitimate addresses, so that none of them will
4652 be given to @code{PRINT_OPERAND_ADDRESS}.
4654 @cindex @code{ENCODE_SECTION_INFO} and address validation
4655 On some machines, whether a symbolic address is legitimate depends on
4656 the section that the address refers to. On these machines, define the
4657 macro @code{ENCODE_SECTION_INFO} to store the information into the
4658 @code{symbol_ref}, and then check for it here. When you see a
4659 @code{const}, you will have to look inside it to find the
4660 @code{symbol_ref} in order to determine the section. @xref{Assembler
4663 @findex saveable_obstack
4664 The best way to modify the name string is by adding text to the
4665 beginning, with suitable punctuation to prevent any ambiguity. Allocate
4666 the new name in @code{saveable_obstack}. You will have to modify
4667 @code{ASM_OUTPUT_LABELREF} to remove and decode the added text and
4668 output the name accordingly, and define @code{STRIP_NAME_ENCODING} to
4669 access the original name string.
4671 You can check the information stored here into the @code{symbol_ref} in
4672 the definitions of the macros @code{GO_IF_LEGITIMATE_ADDRESS} and
4673 @code{PRINT_OPERAND_ADDRESS}.
4675 @findex REG_OK_FOR_BASE_P
4676 @item REG_OK_FOR_BASE_P (@var{x})
4677 A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
4678 RTX) is valid for use as a base register. For hard registers, it
4679 should always accept those which the hardware permits and reject the
4680 others. Whether the macro accepts or rejects pseudo registers must be
4681 controlled by @code{REG_OK_STRICT} as described above. This usually
4682 requires two variant definitions, of which @code{REG_OK_STRICT}
4683 controls the one actually used.
4685 @findex REG_MODE_OK_FOR_BASE_P
4686 @item REG_MODE_OK_FOR_BASE_P (@var{x}, @var{mode})
4687 A C expression that is just like @code{REG_OK_FOR_BASE_P}, except that
4688 that expression may examine the mode of the memory reference in
4689 @var{mode}. You should define this macro if the mode of the memory
4690 reference affects whether a register may be used as a base register. If
4691 you define this macro, the compiler will use it instead of
4692 @code{REG_OK_FOR_BASE_P}.
4694 @findex REG_OK_FOR_INDEX_P
4695 @item REG_OK_FOR_INDEX_P (@var{x})
4696 A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
4697 RTX) is valid for use as an index register.
4699 The difference between an index register and a base register is that
4700 the index register may be scaled. If an address involves the sum of
4701 two registers, neither one of them scaled, then either one may be
4702 labeled the ``base'' and the other the ``index''; but whichever
4703 labeling is used must fit the machine's constraints of which registers
4704 may serve in each capacity. The compiler will try both labelings,
4705 looking for one that is valid, and will reload one or both registers
4706 only if neither labeling works.
4708 @findex FIND_BASE_TERM
4709 @item FIND_BASE_TERM (@var{x})
4710 A C expression to determine the base term of address @var{x}.
4711 This macro is used in only one place: `find_base_term' in alias.c.
4713 It is always safe for this macro to not be defined. It exists so
4714 that alias analysis can understand machine-dependent addresses.
4716 The typical use of this macro is to handle addresses containing
4717 a label_ref or symbol_ref within an UNSPEC@.
4719 @findex LEGITIMIZE_ADDRESS
4720 @item LEGITIMIZE_ADDRESS (@var{x}, @var{oldx}, @var{mode}, @var{win})
4721 A C compound statement that attempts to replace @var{x} with a valid
4722 memory address for an operand of mode @var{mode}. @var{win} will be a
4723 C statement label elsewhere in the code; the macro definition may use
4726 GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{win});
4730 to avoid further processing if the address has become legitimate.
4732 @findex break_out_memory_refs
4733 @var{x} will always be the result of a call to @code{break_out_memory_refs},
4734 and @var{oldx} will be the operand that was given to that function to produce
4737 The code generated by this macro should not alter the substructure of
4738 @var{x}. If it transforms @var{x} into a more legitimate form, it
4739 should assign @var{x} (which will always be a C variable) a new value.
4741 It is not necessary for this macro to come up with a legitimate
4742 address. The compiler has standard ways of doing so in all cases. In
4743 fact, it is safe for this macro to do nothing. But often a
4744 machine-dependent strategy can generate better code.
4746 @findex LEGITIMIZE_RELOAD_ADDRESS
4747 @item LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
4748 A C compound statement that attempts to replace @var{x}, which is an address
4749 that needs reloading, with a valid memory address for an operand of mode
4750 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
4751 It is not necessary to define this macro, but it might be useful for
4752 performance reasons.
4754 For example, on the i386, it is sometimes possible to use a single
4755 reload register instead of two by reloading a sum of two pseudo
4756 registers into a register. On the other hand, for number of RISC
4757 processors offsets are limited so that often an intermediate address
4758 needs to be generated in order to address a stack slot. By defining
4759 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
4760 generated for adjacent some stack slots can be made identical, and thus
4763 @emph{Note}: This macro should be used with caution. It is necessary
4764 to know something of how reload works in order to effectively use this,
4765 and it is quite easy to produce macros that build in too much knowledge
4766 of reload internals.
4768 @emph{Note}: This macro must be able to reload an address created by a
4769 previous invocation of this macro. If it fails to handle such addresses
4770 then the compiler may generate incorrect code or abort.
4773 The macro definition should use @code{push_reload} to indicate parts that
4774 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
4775 suitable to be passed unaltered to @code{push_reload}.
4777 The code generated by this macro must not alter the substructure of
4778 @var{x}. If it transforms @var{x} into a more legitimate form, it
4779 should assign @var{x} (which will always be a C variable) a new value.
4780 This also applies to parts that you change indirectly by calling
4783 @findex strict_memory_address_p
4784 The macro definition may use @code{strict_memory_address_p} to test if
4785 the address has become legitimate.
4788 If you want to change only a part of @var{x}, one standard way of doing
4789 this is to use @code{copy_rtx}. Note, however, that is unshares only a
4790 single level of rtl. Thus, if the part to be changed is not at the
4791 top level, you'll need to replace first the top level.
4792 It is not necessary for this macro to come up with a legitimate
4793 address; but often a machine-dependent strategy can generate better code.
4795 @findex GO_IF_MODE_DEPENDENT_ADDRESS
4796 @item GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
4797 A C statement or compound statement with a conditional @code{goto
4798 @var{label};} executed if memory address @var{x} (an RTX) can have
4799 different meanings depending on the machine mode of the memory
4800 reference it is used for or if the address is valid for some modes
4803 Autoincrement and autodecrement addresses typically have mode-dependent
4804 effects because the amount of the increment or decrement is the size
4805 of the operand being addressed. Some machines have other mode-dependent
4806 addresses. Many RISC machines have no mode-dependent addresses.
4808 You may assume that @var{addr} is a valid address for the machine.
4810 @findex LEGITIMATE_CONSTANT_P
4811 @item LEGITIMATE_CONSTANT_P (@var{x})
4812 A C expression that is nonzero if @var{x} is a legitimate constant for
4813 an immediate operand on the target machine. You can assume that
4814 @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
4815 @samp{1} is a suitable definition for this macro on machines where
4816 anything @code{CONSTANT_P} is valid.
4819 @node Condition Code
4820 @section Condition Code Status
4821 @cindex condition code status
4823 @c prevent bad page break with this line
4824 This describes the condition code status.
4827 The file @file{conditions.h} defines a variable @code{cc_status} to
4828 describe how the condition code was computed (in case the interpretation of
4829 the condition code depends on the instruction that it was set by). This
4830 variable contains the RTL expressions on which the condition code is
4831 currently based, and several standard flags.
4833 Sometimes additional machine-specific flags must be defined in the machine
4834 description header file. It can also add additional machine-specific
4835 information by defining @code{CC_STATUS_MDEP}.
4838 @findex CC_STATUS_MDEP
4839 @item CC_STATUS_MDEP
4840 C code for a data type which is used for declaring the @code{mdep}
4841 component of @code{cc_status}. It defaults to @code{int}.
4843 This macro is not used on machines that do not use @code{cc0}.
4845 @findex CC_STATUS_MDEP_INIT
4846 @item CC_STATUS_MDEP_INIT
4847 A C expression to initialize the @code{mdep} field to ``empty''.
4848 The default definition does nothing, since most machines don't use
4849 the field anyway. If you want to use the field, you should probably
4850 define this macro to initialize it.
4852 This macro is not used on machines that do not use @code{cc0}.
4854 @findex NOTICE_UPDATE_CC
4855 @item NOTICE_UPDATE_CC (@var{exp}, @var{insn})
4856 A C compound statement to set the components of @code{cc_status}
4857 appropriately for an insn @var{insn} whose body is @var{exp}. It is
4858 this macro's responsibility to recognize insns that set the condition
4859 code as a byproduct of other activity as well as those that explicitly
4862 This macro is not used on machines that do not use @code{cc0}.
4864 If there are insns that do not set the condition code but do alter
4865 other machine registers, this macro must check to see whether they
4866 invalidate the expressions that the condition code is recorded as
4867 reflecting. For example, on the 68000, insns that store in address
4868 registers do not set the condition code, which means that usually
4869 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
4870 insns. But suppose that the previous insn set the condition code
4871 based on location @samp{a4@@(102)} and the current insn stores a new
4872 value in @samp{a4}. Although the condition code is not changed by
4873 this, it will no longer be true that it reflects the contents of
4874 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
4875 @code{cc_status} in this case to say that nothing is known about the
4876 condition code value.
4878 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
4879 with the results of peephole optimization: insns whose patterns are
4880 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
4881 constants which are just the operands. The RTL structure of these
4882 insns is not sufficient to indicate what the insns actually do. What
4883 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
4884 @code{CC_STATUS_INIT}.
4886 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
4887 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
4888 @samp{cc}. This avoids having detailed information about patterns in
4889 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
4891 @findex EXTRA_CC_MODES
4892 @item EXTRA_CC_MODES
4893 A list of additional modes for condition code values in registers
4894 (@pxref{Jump Patterns}). This macro should expand to a sequence of
4895 calls of the macro @code{CC} separated by white space. @code{CC} takes
4896 two arguments. The first is the enumeration name of the mode, which
4897 should begin with @samp{CC} and end with @samp{mode}. The second is a C
4898 string giving the printable name of the mode; it should be the same as
4899 the first argument, but with the trailing @samp{mode} removed.
4901 You should only define this macro if additional modes are required.
4903 A sample definition of @code{EXTRA_CC_MODES} is:
4905 #define EXTRA_CC_MODES \
4906 CC(CC_NOOVmode, "CC_NOOV") \
4907 CC(CCFPmode, "CCFP") \
4908 CC(CCFPEmode, "CCFPE")
4911 @findex SELECT_CC_MODE
4912 @item SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
4913 Returns a mode from class @code{MODE_CC} to be used when comparison
4914 operation code @var{op} is applied to rtx @var{x} and @var{y}. For
4915 example, on the Sparc, @code{SELECT_CC_MODE} is defined as (see
4916 @pxref{Jump Patterns} for a description of the reason for this
4920 #define SELECT_CC_MODE(OP,X,Y) \
4921 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
4922 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
4923 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
4924 || GET_CODE (X) == NEG) \
4925 ? CC_NOOVmode : CCmode))
4928 You need not define this macro if @code{EXTRA_CC_MODES} is not defined.
4930 @findex CANONICALIZE_COMPARISON
4931 @item CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
4932 On some machines not all possible comparisons are defined, but you can
4933 convert an invalid comparison into a valid one. For example, the Alpha
4934 does not have a @code{GT} comparison, but you can use an @code{LT}
4935 comparison instead and swap the order of the operands.
4937 On such machines, define this macro to be a C statement to do any
4938 required conversions. @var{code} is the initial comparison code
4939 and @var{op0} and @var{op1} are the left and right operands of the
4940 comparison, respectively. You should modify @var{code}, @var{op0}, and
4941 @var{op1} as required.
4943 GCC will not assume that the comparison resulting from this macro is
4944 valid but will see if the resulting insn matches a pattern in the
4947 You need not define this macro if it would never change the comparison
4950 @findex REVERSIBLE_CC_MODE
4951 @item REVERSIBLE_CC_MODE (@var{mode})
4952 A C expression whose value is one if it is always safe to reverse a
4953 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
4954 can ever return @var{mode} for a floating-point inequality comparison,
4955 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
4957 You need not define this macro if it would always returns zero or if the
4958 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
4959 For example, here is the definition used on the Sparc, where floating-point
4960 inequality comparisons are always given @code{CCFPEmode}:
4963 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
4966 @findex REVERSE_CONDITION (@var{code}, @var{mode})
4967 A C expression whose value is reversed condition code of the @var{code} for
4968 comparison done in CC_MODE @var{mode}. The macro is used only in case
4969 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
4970 machine has some non-standard way how to reverse certain conditionals. For
4971 instance in case all floating point conditions are non-trapping, compiler may
4972 freely convert unordered compares to ordered one. Then definition may look
4976 #define REVERSE_CONDITION(CODE, MODE) \
4977 ((MODE) != CCFPmode ? reverse_condition (CODE) \
4978 : reverse_condition_maybe_unordered (CODE))
4981 @findex REVERSE_CONDEXEC_PREDICATES_P
4982 @item REVERSE_CONDEXEC_PREDICATES_P (@var{code1}, @var{code2})
4983 A C expression that returns true if the conditional execution predicate
4984 @var{code1} is the inverse of @var{code2} and vice versa. Define this to
4985 return 0 if the target has conditional execution predicates that cannot be
4986 reversed safely. If no expansion is specified, this macro is defined as
4990 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
4991 ((x) == reverse_condition (y))
4997 @section Describing Relative Costs of Operations
4998 @cindex costs of instructions
4999 @cindex relative costs
5000 @cindex speed of instructions
5002 These macros let you describe the relative speed of various operations
5003 on the target machine.
5007 @item CONST_COSTS (@var{x}, @var{code}, @var{outer_code})
5008 A part of a C @code{switch} statement that describes the relative costs
5009 of constant RTL expressions. It must contain @code{case} labels for
5010 expression codes @code{const_int}, @code{const}, @code{symbol_ref},
5011 @code{label_ref} and @code{const_double}. Each case must ultimately
5012 reach a @code{return} statement to return the relative cost of the use
5013 of that kind of constant value in an expression. The cost may depend on
5014 the precise value of the constant, which is available for examination in
5015 @var{x}, and the rtx code of the expression in which it is contained,
5016 found in @var{outer_code}.
5018 @var{code} is the expression code---redundant, since it can be
5019 obtained with @code{GET_CODE (@var{x})}.
5022 @findex COSTS_N_INSNS
5023 @item RTX_COSTS (@var{x}, @var{code}, @var{outer_code})
5024 Like @code{CONST_COSTS} but applies to nonconstant RTL expressions.
5025 This can be used, for example, to indicate how costly a multiply
5026 instruction is. In writing this macro, you can use the construct
5027 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
5028 instructions. @var{outer_code} is the code of the expression in which
5029 @var{x} is contained.
5031 This macro is optional; do not define it if the default cost assumptions
5032 are adequate for the target machine.
5034 @findex DEFAULT_RTX_COSTS
5035 @item DEFAULT_RTX_COSTS (@var{x}, @var{code}, @var{outer_code})
5036 This macro, if defined, is called for any case not handled by the
5037 @code{RTX_COSTS} or @code{CONST_COSTS} macros. This eliminates the need
5038 to put case labels into the macro, but the code, or any functions it
5039 calls, must assume that the RTL in @var{x} could be of any type that has
5040 not already been handled. The arguments are the same as for
5041 @code{RTX_COSTS}, and the macro should execute a return statement giving
5042 the cost of any RTL expressions that it can handle. The default cost
5043 calculation is used for any RTL for which this macro does not return a
5046 This macro is optional; do not define it if the default cost assumptions
5047 are adequate for the target machine.
5049 @findex ADDRESS_COST
5050 @item ADDRESS_COST (@var{address})
5051 An expression giving the cost of an addressing mode that contains
5052 @var{address}. If not defined, the cost is computed from
5053 the @var{address} expression and the @code{CONST_COSTS} values.
5055 For most CISC machines, the default cost is a good approximation of the
5056 true cost of the addressing mode. However, on RISC machines, all
5057 instructions normally have the same length and execution time. Hence
5058 all addresses will have equal costs.
5060 In cases where more than one form of an address is known, the form with
5061 the lowest cost will be used. If multiple forms have the same, lowest,
5062 cost, the one that is the most complex will be used.
5064 For example, suppose an address that is equal to the sum of a register
5065 and a constant is used twice in the same basic block. When this macro
5066 is not defined, the address will be computed in a register and memory
5067 references will be indirect through that register. On machines where
5068 the cost of the addressing mode containing the sum is no higher than
5069 that of a simple indirect reference, this will produce an additional
5070 instruction and possibly require an additional register. Proper
5071 specification of this macro eliminates this overhead for such machines.
5073 Similar use of this macro is made in strength reduction of loops.
5075 @var{address} need not be valid as an address. In such a case, the cost
5076 is not relevant and can be any value; invalid addresses need not be
5077 assigned a different cost.
5079 On machines where an address involving more than one register is as
5080 cheap as an address computation involving only one register, defining
5081 @code{ADDRESS_COST} to reflect this can cause two registers to be live
5082 over a region of code where only one would have been if
5083 @code{ADDRESS_COST} were not defined in that manner. This effect should
5084 be considered in the definition of this macro. Equivalent costs should
5085 probably only be given to addresses with different numbers of registers
5086 on machines with lots of registers.
5088 This macro will normally either not be defined or be defined as a
5091 @findex REGISTER_MOVE_COST
5092 @item REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
5093 A C expression for the cost of moving data of mode @var{mode} from a
5094 register in class @var{from} to one in class @var{to}. The classes are
5095 expressed using the enumeration values such as @code{GENERAL_REGS}. A
5096 value of 2 is the default; other values are interpreted relative to
5099 It is not required that the cost always equal 2 when @var{from} is the
5100 same as @var{to}; on some machines it is expensive to move between
5101 registers if they are not general registers.
5103 If reload sees an insn consisting of a single @code{set} between two
5104 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
5105 classes returns a value of 2, reload does not check to ensure that the
5106 constraints of the insn are met. Setting a cost of other than 2 will
5107 allow reload to verify that the constraints are met. You should do this
5108 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
5110 @findex MEMORY_MOVE_COST
5111 @item MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
5112 A C expression for the cost of moving data of mode @var{mode} between a
5113 register of class @var{class} and memory; @var{in} is zero if the value
5114 is to be written to memory, nonzero if it is to be read in. This cost
5115 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
5116 registers and memory is more expensive than between two registers, you
5117 should define this macro to express the relative cost.
5119 If you do not define this macro, GCC uses a default cost of 4 plus
5120 the cost of copying via a secondary reload register, if one is
5121 needed. If your machine requires a secondary reload register to copy
5122 between memory and a register of @var{class} but the reload mechanism is
5123 more complex than copying via an intermediate, define this macro to
5124 reflect the actual cost of the move.
5126 GCC defines the function @code{memory_move_secondary_cost} if
5127 secondary reloads are needed. It computes the costs due to copying via
5128 a secondary register. If your machine copies from memory using a
5129 secondary register in the conventional way but the default base value of
5130 4 is not correct for your machine, define this macro to add some other
5131 value to the result of that function. The arguments to that function
5132 are the same as to this macro.
5136 A C expression for the cost of a branch instruction. A value of 1 is
5137 the default; other values are interpreted relative to that.
5140 Here are additional macros which do not specify precise relative costs,
5141 but only that certain actions are more expensive than GCC would
5145 @findex SLOW_BYTE_ACCESS
5146 @item SLOW_BYTE_ACCESS
5147 Define this macro as a C expression which is nonzero if accessing less
5148 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
5149 faster than accessing a word of memory, i.e., if such access
5150 require more than one instruction or if there is no difference in cost
5151 between byte and (aligned) word loads.
5153 When this macro is not defined, the compiler will access a field by
5154 finding the smallest containing object; when it is defined, a fullword
5155 load will be used if alignment permits. Unless bytes accesses are
5156 faster than word accesses, using word accesses is preferable since it
5157 may eliminate subsequent memory access if subsequent accesses occur to
5158 other fields in the same word of the structure, but to different bytes.
5160 @findex SLOW_UNALIGNED_ACCESS
5161 @item SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
5162 Define this macro to be the value 1 if memory accesses described by the
5163 @var{mode} and @var{alignment} parameters have a cost many times greater
5164 than aligned accesses, for example if they are emulated in a trap
5167 When this macro is nonzero, the compiler will act as if
5168 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
5169 moves. This can cause significantly more instructions to be produced.
5170 Therefore, do not set this macro nonzero if unaligned accesses only add a
5171 cycle or two to the time for a memory access.
5173 If the value of this macro is always zero, it need not be defined. If
5174 this macro is defined, it should produce a nonzero value when
5175 @code{STRICT_ALIGNMENT} is nonzero.
5177 @findex DONT_REDUCE_ADDR
5178 @item DONT_REDUCE_ADDR
5179 Define this macro to inhibit strength reduction of memory addresses.
5180 (On some machines, such strength reduction seems to do harm rather
5185 The threshold of number of scalar memory-to-memory move insns, @emph{below}
5186 which a sequence of insns should be generated instead of a
5187 string move insn or a library call. Increasing the value will always
5188 make code faster, but eventually incurs high cost in increased code size.
5190 Note that on machines where the corresponding move insn is a
5191 @code{define_expand} that emits a sequence of insns, this macro counts
5192 the number of such sequences.
5194 If you don't define this, a reasonable default is used.
5196 @findex MOVE_BY_PIECES_P
5197 @item MOVE_BY_PIECES_P (@var{size}, @var{alignment})
5198 A C expression used to determine whether @code{move_by_pieces} will be used to
5199 copy a chunk of memory, or whether some other block move mechanism
5200 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
5201 than @code{MOVE_RATIO}.
5203 @findex MOVE_MAX_PIECES
5204 @item MOVE_MAX_PIECES
5205 A C expression used by @code{move_by_pieces} to determine the largest unit
5206 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
5208 @findex USE_LOAD_POST_INCREMENT
5209 @item USE_LOAD_POST_INCREMENT (@var{mode})
5210 A C expression used to determine whether a load postincrement is a good
5211 thing to use for a given mode. Defaults to the value of
5212 @code{HAVE_POST_INCREMENT}.
5214 @findex USE_LOAD_POST_DECREMENT
5215 @item USE_LOAD_POST_DECREMENT (@var{mode})
5216 A C expression used to determine whether a load postdecrement is a good
5217 thing to use for a given mode. Defaults to the value of
5218 @code{HAVE_POST_DECREMENT}.
5220 @findex USE_LOAD_PRE_INCREMENT
5221 @item USE_LOAD_PRE_INCREMENT (@var{mode})
5222 A C expression used to determine whether a load preincrement is a good
5223 thing to use for a given mode. Defaults to the value of
5224 @code{HAVE_PRE_INCREMENT}.
5226 @findex USE_LOAD_PRE_DECREMENT
5227 @item USE_LOAD_PRE_DECREMENT (@var{mode})
5228 A C expression used to determine whether a load predecrement is a good
5229 thing to use for a given mode. Defaults to the value of
5230 @code{HAVE_PRE_DECREMENT}.
5232 @findex USE_STORE_POST_INCREMENT
5233 @item USE_STORE_POST_INCREMENT (@var{mode})
5234 A C expression used to determine whether a store postincrement is a good
5235 thing to use for a given mode. Defaults to the value of
5236 @code{HAVE_POST_INCREMENT}.
5238 @findex USE_STORE_POST_DECREMENT
5239 @item USE_STORE_POST_DECREMENT (@var{mode})
5240 A C expression used to determine whether a store postdecrement is a good
5241 thing to use for a given mode. Defaults to the value of
5242 @code{HAVE_POST_DECREMENT}.
5244 @findex USE_STORE_PRE_INCREMENT
5245 @item USE_STORE_PRE_INCREMENT (@var{mode})
5246 This macro is used to determine whether a store preincrement is a good
5247 thing to use for a given mode. Defaults to the value of
5248 @code{HAVE_PRE_INCREMENT}.
5250 @findex USE_STORE_PRE_DECREMENT
5251 @item USE_STORE_PRE_DECREMENT (@var{mode})
5252 This macro is used to determine whether a store predecrement is a good
5253 thing to use for a given mode. Defaults to the value of
5254 @code{HAVE_PRE_DECREMENT}.
5256 @findex NO_FUNCTION_CSE
5257 @item NO_FUNCTION_CSE
5258 Define this macro if it is as good or better to call a constant
5259 function address than to call an address kept in a register.
5261 @findex NO_RECURSIVE_FUNCTION_CSE
5262 @item NO_RECURSIVE_FUNCTION_CSE
5263 Define this macro if it is as good or better for a function to call
5264 itself with an explicit address than to call an address kept in a
5269 @section Adjusting the Instruction Scheduler
5271 The instruction scheduler may need a fair amount of machine-specific
5272 adjustment in order to produce good code. GCC provides several target
5273 hooks for this purpose. It is usually enough to define just a few of
5274 them: try the first ones in this list first.
5276 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
5277 This hook returns the maximum number of instructions that can ever issue
5278 at the same time on the target machine. The default is one. This value
5279 must be constant over the entire compilation. If you need it to vary
5280 depending on what the instructions are, you must use
5281 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
5284 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more})
5285 This hook is executed by the scheduler after it has scheduled an insn
5286 from the ready list. It should return the number of insns which can
5287 still be issued in the current cycle. Normally this is
5288 @samp{@w{@var{more} - 1}}. You should define this hook if some insns
5289 take more machine resources than others, so that fewer insns can follow
5290 them in the same cycle. @var{file} is either a null pointer, or a stdio
5291 stream to write any debug output to. @var{verbose} is the verbose level
5292 provided by @option{-fsched-verbose-@var{n}}. @var{insn} is the
5293 instruction that was scheduled.
5296 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost})
5297 This function corrects the value of @var{cost} based on the relationship
5298 between @var{insn} and @var{dep_insn} through the dependence @var{link}.
5299 It should return the new value. The default is to make no adjustment to
5300 @var{cost}. This can be used for example to specify to the scheduler
5301 that an output- or anti-dependence does not incur the same cost as a
5305 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority})
5306 This hook adjusts the integer scheduling priority @var{priority} of
5307 @var{insn}. It should return the new priority. Reduce the priority to
5308 execute @var{insn} earlier, increase the priority to execute @var{insn}
5309 later. Do not define this hook if you do not need to adjust the
5310 scheduling priorities of insns.
5313 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
5314 This hook is executed by the scheduler after it has scheduled the ready
5315 list, to allow the machine description to reorder it (for example to
5316 combine two small instructions together on @samp{VLIW} machines).
5317 @var{file} is either a null pointer, or a stdio stream to write any
5318 debug output to. @var{verbose} is the verbose level provided by
5319 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
5320 list of instructions that are ready to be scheduled. @var{n_readyp} is
5321 a pointer to the number of elements in the ready list. The scheduler
5322 reads the ready list in reverse order, starting with
5323 @var{ready}[@var{*n_readyp}-1] and going to @var{ready}[0]. @var{clock}
5324 is the timer tick of the scheduler. You may modify the ready list and
5325 the number of ready insns. The return value is the number of insns that
5326 can issue this cycle; normally this is just @code{issue_rate}. See also
5327 @samp{TARGET_SCHED_REORDER2}.
5330 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_ready}, @var{clock})
5331 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
5332 function is called whenever the scheduler starts a new cycle. This one
5333 is called once per iteration over a cycle, immediately after
5334 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
5335 return the number of insns to be scheduled in the same cycle. Defining
5336 this hook can be useful if there are frequent situations where
5337 scheduling one insn causes other insns to become ready in the same
5338 cycle. These other insns can then be taken into account properly.
5341 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
5342 This hook is executed by the scheduler at the beginning of each block of
5343 instructions that are to be scheduled. @var{file} is either a null
5344 pointer, or a stdio stream to write any debug output to. @var{verbose}
5345 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
5346 @var{max_ready} is the maximum number of insns in the current scheduling
5347 region that can be live at the same time. This can be used to allocate
5348 scratch space if it is needed, e.g. by @samp{TARGET_SCHED_REORDER}.
5351 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
5352 This hook is executed by the scheduler at the end of each block of
5353 instructions that are to be scheduled. It can be used to perform
5354 cleanup of any actions done by the other scheduling hooks. @var{file}
5355 is either a null pointer, or a stdio stream to write any debug output
5356 to. @var{verbose} is the verbose level provided by
5357 @option{-fsched-verbose-@var{n}}.
5360 @deftypefn {Target Hook} rtx TARGET_SCHED_CYCLE_DISPLAY (int @var{clock}, rtx @var{last})
5361 This hook is called in verbose mode only, at the beginning of each pass
5362 over a basic block. It should insert an insn into the chain after
5363 @var{last}, which has no effect, but records the value @var{clock} in
5364 RTL dumps and assembly output. Define this hook only if you need this
5365 level of detail about what the scheduler is doing.
5369 @section Dividing the Output into Sections (Texts, Data, @dots{})
5370 @c the above section title is WAY too long. maybe cut the part between
5371 @c the (...)? --mew 10feb93
5373 An object file is divided into sections containing different types of
5374 data. In the most common case, there are three sections: the @dfn{text
5375 section}, which holds instructions and read-only data; the @dfn{data
5376 section}, which holds initialized writable data; and the @dfn{bss
5377 section}, which holds uninitialized data. Some systems have other kinds
5380 The compiler must tell the assembler when to switch sections. These
5381 macros control what commands to output to tell the assembler this. You
5382 can also define additional sections.
5385 @findex TEXT_SECTION_ASM_OP
5386 @item TEXT_SECTION_ASM_OP
5387 A C expression whose value is a string, including spacing, containing the
5388 assembler operation that should precede instructions and read-only data.
5389 Normally @code{"\t.text"} is right.
5391 @findex TEXT_SECTION
5393 A C statement that switches to the default section containing instructions.
5394 Normally this is not needed, as simply defining @code{TEXT_SECTION_ASM_OP}
5395 is enough. The MIPS port uses this to sort all functions after all data
5398 @findex DATA_SECTION_ASM_OP
5399 @item DATA_SECTION_ASM_OP
5400 A C expression whose value is a string, including spacing, containing the
5401 assembler operation to identify the following data as writable initialized
5402 data. Normally @code{"\t.data"} is right.
5404 @findex SHARED_SECTION_ASM_OP
5405 @item SHARED_SECTION_ASM_OP
5406 If defined, a C expression whose value is a string, including spacing,
5407 containing the assembler operation to identify the following data as
5408 shared data. If not defined, @code{DATA_SECTION_ASM_OP} will be used.
5410 @findex BSS_SECTION_ASM_OP
5411 @item BSS_SECTION_ASM_OP
5412 If defined, a C expression whose value is a string, including spacing,
5413 containing the assembler operation to identify the following data as
5414 uninitialized global data. If not defined, and neither
5415 @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined,
5416 uninitialized global data will be output in the data section if
5417 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
5420 @findex SHARED_BSS_SECTION_ASM_OP
5421 @item SHARED_BSS_SECTION_ASM_OP
5422 If defined, a C expression whose value is a string, including spacing,
5423 containing the assembler operation to identify the following data as
5424 uninitialized global shared data. If not defined, and
5425 @code{BSS_SECTION_ASM_OP} is, the latter will be used.
5427 @findex INIT_SECTION_ASM_OP
5428 @item INIT_SECTION_ASM_OP
5429 If defined, a C expression whose value is a string, including spacing,
5430 containing the assembler operation to identify the following data as
5431 initialization code. If not defined, GCC will assume such a section does
5434 @findex FINI_SECTION_ASM_OP
5435 @item FINI_SECTION_ASM_OP
5436 If defined, a C expression whose value is a string, including spacing,
5437 containing the assembler operation to identify the following data as
5438 finalization code. If not defined, GCC will assume such a section does
5441 @findex CRT_CALL_STATIC_FUNCTION
5442 @item CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
5443 If defined, an ASM statement that switches to a different section
5444 via @var{section_op}, calls @var{function}, and switches back to
5445 the text section. This is used in @file{crtstuff.c} if
5446 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
5447 to initialization and finalization functions from the init and fini
5448 sections. By default, this macro uses a simple function call. Some
5449 ports need hand-crafted assembly code to avoid dependencies on
5450 registers initialized in the function prologue or to ensure that
5451 constant pools don't end up too far way in the text section.
5453 @findex FORCE_CODE_SECTION_ALIGN
5454 @item FORCE_CODE_SECTION_ALIGN
5455 If defined, an ASM statement that aligns a code section to some
5456 arbitrary boundary. This is used to force all fragments of the
5457 @code{.init} and @code{.fini} sections to have to same alignment
5458 and thus prevent the linker from having to add any padding.
5460 @findex EXTRA_SECTIONS
5463 @item EXTRA_SECTIONS
5464 A list of names for sections other than the standard two, which are
5465 @code{in_text} and @code{in_data}. You need not define this macro
5466 on a system with no other sections (that GCC needs to use).
5468 @findex EXTRA_SECTION_FUNCTIONS
5469 @findex text_section
5470 @findex data_section
5471 @item EXTRA_SECTION_FUNCTIONS
5472 One or more functions to be defined in @file{varasm.c}. These
5473 functions should do jobs analogous to those of @code{text_section} and
5474 @code{data_section}, for your additional sections. Do not define this
5475 macro if you do not define @code{EXTRA_SECTIONS}.
5477 @findex READONLY_DATA_SECTION
5478 @item READONLY_DATA_SECTION
5479 On most machines, read-only variables, constants, and jump tables are
5480 placed in the text section. If this is not the case on your machine,
5481 this macro should be defined to be the name of a function (either
5482 @code{data_section} or a function defined in @code{EXTRA_SECTIONS}) that
5483 switches to the section to be used for read-only items.
5485 If these items should be placed in the text section, this macro should
5488 @findex SELECT_SECTION
5489 @item SELECT_SECTION (@var{exp}, @var{reloc}, @var{align})
5490 A C statement or statements to switch to the appropriate section for
5491 output of @var{exp}. You can assume that @var{exp} is either a
5492 @code{VAR_DECL} node or a constant of some sort. @var{reloc}
5493 indicates whether the initial value of @var{exp} requires link-time
5494 relocations. Bit 1 is set when variable contains local relocations
5495 only, while bit 2 is set for global relocations.
5496 Select the section by calling @code{text_section} or one
5497 of the alternatives for other sections. @var{align} is the constant
5500 Do not define this macro if you put all read-only variables and
5501 constants in the read-only data section (usually the text section).
5503 @findex SELECT_RTX_SECTION
5504 @item SELECT_RTX_SECTION (@var{mode}, @var{rtx}, @var{align})
5505 A C statement or statements to switch to the appropriate section for
5506 output of @var{rtx} in mode @var{mode}. You can assume that @var{rtx}
5507 is some kind of constant in RTL@. The argument @var{mode} is redundant
5508 except in the case of a @code{const_int} rtx. Select the section by
5509 calling @code{text_section} or one of the alternatives for other
5510 sections. @var{align} is the constant alignment in bits.
5512 Do not define this macro if you put all constants in the read-only
5515 @findex JUMP_TABLES_IN_TEXT_SECTION
5516 @item JUMP_TABLES_IN_TEXT_SECTION
5517 Define this macro to be an expression with a nonzero value if jump
5518 tables (for @code{tablejump} insns) should be output in the text
5519 section, along with the assembler instructions. Otherwise, the
5520 readonly data section is used.
5522 This macro is irrelevant if there is no separate readonly data section.
5524 @findex ENCODE_SECTION_INFO
5525 @item ENCODE_SECTION_INFO (@var{decl})
5526 Define this macro if references to a symbol or a constant must be
5527 treated differently depending on something about the variable or
5528 function named by the symbol (such as what section it is in).
5530 The macro definition, if any, is executed under two circumstances. One
5531 is immediately after the rtl for @var{decl} that represents a variable
5532 or a function has been created and stored in @code{DECL_RTL
5533 (@var{decl})}. The value of the rtl will be a @code{mem} whose address
5534 is a @code{symbol_ref}. The other is immediately after the rtl for
5535 @var{decl} that represents a constant has been created and stored in
5536 @code{TREE_CST_RTL (@var{decl})}. The macro is called once for each
5537 distinct constant in a source file.
5539 @cindex @code{SYMBOL_REF_FLAG}, in @code{ENCODE_SECTION_INFO}
5540 The usual thing for this macro to do is to record a flag in the
5541 @code{symbol_ref} (such as @code{SYMBOL_REF_FLAG}) or to store a
5542 modified name string in the @code{symbol_ref} (if one bit is not enough
5545 @findex STRIP_NAME_ENCODING
5546 @item STRIP_NAME_ENCODING (@var{var}, @var{sym_name})
5547 Decode @var{sym_name} and store the real name part in @var{var}, sans
5548 the characters that encode section info. Define this macro if
5549 @code{ENCODE_SECTION_INFO} alters the symbol's name string.
5551 @findex UNIQUE_SECTION
5552 @item UNIQUE_SECTION (@var{decl}, @var{reloc})
5553 A C statement to build up a unique section name, expressed as a
5554 @code{STRING_CST} node, and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
5555 @var{reloc} indicates whether the initial value of @var{exp} requires
5556 link-time relocations. If you do not define this macro, GCC will use
5557 the symbol name prefixed by @samp{.} as the section name. Note - this
5558 macro can now be called for uninitialized data items as well as
5559 initialized data and functions.
5563 @section Position Independent Code
5564 @cindex position independent code
5567 This section describes macros that help implement generation of position
5568 independent code. Simply defining these macros is not enough to
5569 generate valid PIC; you must also add support to the macros
5570 @code{GO_IF_LEGITIMATE_ADDRESS} and @code{PRINT_OPERAND_ADDRESS}, as
5571 well as @code{LEGITIMIZE_ADDRESS}. You must modify the definition of
5572 @samp{movsi} to do something appropriate when the source operand
5573 contains a symbolic address. You may also need to alter the handling of
5574 switch statements so that they use relative addresses.
5575 @c i rearranged the order of the macros above to try to force one of
5576 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
5579 @findex PIC_OFFSET_TABLE_REGNUM
5580 @item PIC_OFFSET_TABLE_REGNUM
5581 The register number of the register used to address a table of static
5582 data addresses in memory. In some cases this register is defined by a
5583 processor's ``application binary interface'' (ABI)@. When this macro
5584 is defined, RTL is generated for this register once, as with the stack
5585 pointer and frame pointer registers. If this macro is not defined, it
5586 is up to the machine-dependent files to allocate such a register (if
5587 necessary). Note that this register must be fixed when in use (e.g.@:
5588 when @code{flag_pic} is true).
5590 @findex PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
5591 @item PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
5592 Define this macro if the register defined by
5593 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define
5594 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
5596 @findex FINALIZE_PIC
5598 By generating position-independent code, when two different programs (A
5599 and B) share a common library (libC.a), the text of the library can be
5600 shared whether or not the library is linked at the same address for both
5601 programs. In some of these environments, position-independent code
5602 requires not only the use of different addressing modes, but also
5603 special code to enable the use of these addressing modes.
5605 The @code{FINALIZE_PIC} macro serves as a hook to emit these special
5606 codes once the function is being compiled into assembly code, but not
5607 before. (It is not done before, because in the case of compiling an
5608 inline function, it would lead to multiple PIC prologues being
5609 included in functions which used inline functions and were compiled to
5612 @findex LEGITIMATE_PIC_OPERAND_P
5613 @item LEGITIMATE_PIC_OPERAND_P (@var{x})
5614 A C expression that is nonzero if @var{x} is a legitimate immediate
5615 operand on the target machine when generating position independent code.
5616 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
5617 check this. You can also assume @var{flag_pic} is true, so you need not
5618 check it either. You need not define this macro if all constants
5619 (including @code{SYMBOL_REF}) can be immediate operands when generating
5620 position independent code.
5623 @node Assembler Format
5624 @section Defining the Output Assembler Language
5626 This section describes macros whose principal purpose is to describe how
5627 to write instructions in assembler language---rather than what the
5631 * File Framework:: Structural information for the assembler file.
5632 * Data Output:: Output of constants (numbers, strings, addresses).
5633 * Uninitialized Data:: Output of uninitialized variables.
5634 * Label Output:: Output and generation of labels.
5635 * Initialization:: General principles of initialization
5636 and termination routines.
5637 * Macros for Initialization::
5638 Specific macros that control the handling of
5639 initialization and termination routines.
5640 * Instruction Output:: Output of actual instructions.
5641 * Dispatch Tables:: Output of jump tables.
5642 * Exception Region Output:: Output of exception region code.
5643 * Alignment Output:: Pseudo ops for alignment and skipping data.
5646 @node File Framework
5647 @subsection The Overall Framework of an Assembler File
5648 @cindex assembler format
5649 @cindex output of assembler code
5651 @c prevent bad page break with this line
5652 This describes the overall framework of an assembler file.
5655 @findex ASM_FILE_START
5656 @item ASM_FILE_START (@var{stream})
5657 A C expression which outputs to the stdio stream @var{stream}
5658 some appropriate text to go at the start of an assembler file.
5660 Normally this macro is defined to output a line containing
5661 @samp{#NO_APP}, which is a comment that has no effect on most
5662 assemblers but tells the GNU assembler that it can save time by not
5663 checking for certain assembler constructs.
5665 On systems that use SDB, it is necessary to output certain commands;
5666 see @file{attasm.h}.
5668 @findex ASM_FILE_END
5669 @item ASM_FILE_END (@var{stream})
5670 A C expression which outputs to the stdio stream @var{stream}
5671 some appropriate text to go at the end of an assembler file.
5673 If this macro is not defined, the default is to output nothing
5674 special at the end of the file. Most systems don't require any
5677 On systems that use SDB, it is necessary to output certain commands;
5678 see @file{attasm.h}.
5680 @findex ASM_COMMENT_START
5681 @item ASM_COMMENT_START
5682 A C string constant describing how to begin a comment in the target
5683 assembler language. The compiler assumes that the comment will end at
5684 the end of the line.
5688 A C string constant for text to be output before each @code{asm}
5689 statement or group of consecutive ones. Normally this is
5690 @code{"#APP"}, which is a comment that has no effect on most
5691 assemblers but tells the GNU assembler that it must check the lines
5692 that follow for all valid assembler constructs.
5696 A C string constant for text to be output after each @code{asm}
5697 statement or group of consecutive ones. Normally this is
5698 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
5699 time-saving assumptions that are valid for ordinary compiler output.
5701 @findex ASM_OUTPUT_SOURCE_FILENAME
5702 @item ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
5703 A C statement to output COFF information or DWARF debugging information
5704 which indicates that filename @var{name} is the current source file to
5705 the stdio stream @var{stream}.
5707 This macro need not be defined if the standard form of output
5708 for the file format in use is appropriate.
5710 @findex OUTPUT_QUOTED_STRING
5711 @item OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
5712 A C statement to output the string @var{string} to the stdio stream
5713 @var{stream}. If you do not call the function @code{output_quoted_string}
5714 in your config files, GCC will only call it to output filenames to
5715 the assembler source. So you can use it to canonicalize the format
5716 of the filename using this macro.
5718 @findex ASM_OUTPUT_SOURCE_LINE
5719 @item ASM_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
5720 A C statement to output DBX or SDB debugging information before code
5721 for line number @var{line} of the current source file to the
5722 stdio stream @var{stream}.
5724 This macro need not be defined if the standard form of debugging
5725 information for the debugger in use is appropriate.
5727 @findex ASM_OUTPUT_IDENT
5728 @item ASM_OUTPUT_IDENT (@var{stream}, @var{string})
5729 A C statement to output something to the assembler file to handle a
5730 @samp{#ident} directive containing the text @var{string}. If this
5731 macro is not defined, nothing is output for a @samp{#ident} directive.
5733 @findex OBJC_PROLOGUE
5735 A C statement to output any assembler statements which are required to
5736 precede any Objective-C object definitions or message sending. The
5737 statement is executed only when compiling an Objective-C program.
5740 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, unsigned int @var{align})
5741 Output assembly directives to switch to section @var{name}. The section
5742 should have attributes as specified by @var{flags}, which is a bit mask
5743 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{align}
5744 is nonzero, it contains an alignment in bytes to be used for the section,
5745 otherwise some target default should be used. Only targets that must
5746 specify an alignment within the section directive need pay attention to
5747 @var{align} -- we will still use @code{ASM_OUTPUT_ALIGN}.
5750 @deftypefn {Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
5751 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
5754 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
5755 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
5756 based on a variable or function decl, a section name, and whether or not the
5757 declaration's initializer may contain runtime relocations. @var{decl} may be
5758 null, in which case read-write data should be assumed.
5760 The default version if this function handles choosing code vs data,
5761 read-only vs read-write data, and @code{flag_pic}. You should only
5762 need to override this if your target has special flags that might be
5763 set via @code{__attribute__}.
5768 @subsection Output of Data
5771 @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP
5772 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
5773 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
5774 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
5775 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
5776 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
5777 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
5778 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
5779 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
5780 These hooks specify assembly directives for creating certain kinds
5781 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
5782 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
5783 aligned two-byte object, and so on. Any of the hooks may be
5784 @code{NULL}, indicating that no suitable directive is available.
5786 The compiler will print these strings at the start of a new line,
5787 followed immediately by the object's initial value. In most cases,
5788 the string should contain a tab, a pseudo-op, and then another tab.
5791 @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p})
5792 The @code{assemble_integer} function uses this hook to output an
5793 integer object. @var{x} is the object's value, @var{size} is its size
5794 in bytes and @var{aligned_p} indicates whether it is aligned. The
5795 function should return @code{true} if it was able to output the
5796 object. If it returns false, @code{assemble_integer} will try to
5797 split the object into smaller parts.
5799 The default implementation of this hook will use the
5800 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
5801 when the relevant string is @code{NULL}.
5805 @findex OUTPUT_ADDR_CONST_EXTRA
5806 @item OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
5807 A C statement to recognize @var{rtx} patterns that
5808 @code{output_addr_const} can't deal with, and output assembly code to
5809 @var{stream} corresponding to the pattern @var{x}. This may be used to
5810 allow machine-dependent @code{UNSPEC}s to appear within constants.
5812 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
5813 @code{goto fail}, so that a standard error message is printed. If it
5814 prints an error message itself, by calling, for example,
5815 @code{output_operand_lossage}, it may just complete normally.
5817 @findex ASM_OUTPUT_ASCII
5818 @item ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
5819 A C statement to output to the stdio stream @var{stream} an assembler
5820 instruction to assemble a string constant containing the @var{len}
5821 bytes at @var{ptr}. @var{ptr} will be a C expression of type
5822 @code{char *} and @var{len} a C expression of type @code{int}.
5824 If the assembler has a @code{.ascii} pseudo-op as found in the
5825 Berkeley Unix assembler, do not define the macro
5826 @code{ASM_OUTPUT_ASCII}.
5828 @findex ASM_OUTPUT_FDESC
5829 @item ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
5830 A C statement to output word @var{n} of a function descriptor for
5831 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
5832 is defined, and is otherwise unused.
5834 @findex CONSTANT_POOL_BEFORE_FUNCTION
5835 @item CONSTANT_POOL_BEFORE_FUNCTION
5836 You may define this macro as a C expression. You should define the
5837 expression to have a nonzero value if GCC should output the constant
5838 pool for a function before the code for the function, or a zero value if
5839 GCC should output the constant pool after the function. If you do
5840 not define this macro, the usual case, GCC will output the constant
5841 pool before the function.
5843 @findex ASM_OUTPUT_POOL_PROLOGUE
5844 @item ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
5845 A C statement to output assembler commands to define the start of the
5846 constant pool for a function. @var{funname} is a string giving
5847 the name of the function. Should the return type of the function
5848 be required, it can be obtained via @var{fundecl}. @var{size}
5849 is the size, in bytes, of the constant pool that will be written
5850 immediately after this call.
5852 If no constant-pool prefix is required, the usual case, this macro need
5855 @findex ASM_OUTPUT_SPECIAL_POOL_ENTRY
5856 @item ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
5857 A C statement (with or without semicolon) to output a constant in the
5858 constant pool, if it needs special treatment. (This macro need not do
5859 anything for RTL expressions that can be output normally.)
5861 The argument @var{file} is the standard I/O stream to output the
5862 assembler code on. @var{x} is the RTL expression for the constant to
5863 output, and @var{mode} is the machine mode (in case @var{x} is a
5864 @samp{const_int}). @var{align} is the required alignment for the value
5865 @var{x}; you should output an assembler directive to force this much
5868 The argument @var{labelno} is a number to use in an internal label for
5869 the address of this pool entry. The definition of this macro is
5870 responsible for outputting the label definition at the proper place.
5871 Here is how to do this:
5874 ASM_OUTPUT_INTERNAL_LABEL (@var{file}, "LC", @var{labelno});
5877 When you output a pool entry specially, you should end with a
5878 @code{goto} to the label @var{jumpto}. This will prevent the same pool
5879 entry from being output a second time in the usual manner.
5881 You need not define this macro if it would do nothing.
5883 @findex CONSTANT_AFTER_FUNCTION_P
5884 @item CONSTANT_AFTER_FUNCTION_P (@var{exp})
5885 Define this macro as a C expression which is nonzero if the constant
5886 @var{exp}, of type @code{tree}, should be output after the code for a
5887 function. The compiler will normally output all constants before the
5888 function; you need not define this macro if this is OK@.
5890 @findex ASM_OUTPUT_POOL_EPILOGUE
5891 @item ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
5892 A C statement to output assembler commands to at the end of the constant
5893 pool for a function. @var{funname} is a string giving the name of the
5894 function. Should the return type of the function be required, you can
5895 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
5896 constant pool that GCC wrote immediately before this call.
5898 If no constant-pool epilogue is required, the usual case, you need not
5901 @findex IS_ASM_LOGICAL_LINE_SEPARATOR
5902 @item IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C})
5903 Define this macro as a C expression which is nonzero if @var{C} is
5904 used as a logical line separator by the assembler.
5906 If you do not define this macro, the default is that only
5907 the character @samp{;} is treated as a logical line separator.
5910 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
5911 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
5912 These target hooks are C string constants, describing the syntax in the
5913 assembler for grouping arithmetic expressions. If not overridden, they
5914 default to normal parentheses, which is correct for most assemblers.
5917 These macros are provided by @file{real.h} for writing the definitions
5918 of @code{ASM_OUTPUT_DOUBLE} and the like:
5921 @item REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
5922 @itemx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
5923 @itemx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
5924 @findex REAL_VALUE_TO_TARGET_SINGLE
5925 @findex REAL_VALUE_TO_TARGET_DOUBLE
5926 @findex REAL_VALUE_TO_TARGET_LONG_DOUBLE
5927 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the target's
5928 floating point representation, and store its bit pattern in the array of
5929 @code{long int} whose address is @var{l}. The number of elements in the
5930 output array is determined by the size of the desired target floating
5931 point data type: 32 bits of it go in each @code{long int} array
5932 element. Each array element holds 32 bits of the result, even if
5933 @code{long int} is wider than 32 bits on the host machine.
5935 The array element values are designed so that you can print them out
5936 using @code{fprintf} in the order they should appear in the target
5939 @item REAL_VALUE_TO_DECIMAL (@var{x}, @var{format}, @var{string})
5940 @findex REAL_VALUE_TO_DECIMAL
5941 This macro converts @var{x}, of type @code{REAL_VALUE_TYPE}, to a
5942 decimal number and stores it as a string into @var{string}.
5943 You must pass, as @var{string}, the address of a long enough block
5944 of space to hold the result.
5946 The argument @var{format} is a @code{printf}-specification that serves
5947 as a suggestion for how to format the output string.
5950 @node Uninitialized Data
5951 @subsection Output of Uninitialized Variables
5953 Each of the macros in this section is used to do the whole job of
5954 outputting a single uninitialized variable.
5957 @findex ASM_OUTPUT_COMMON
5958 @item ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
5959 A C statement (sans semicolon) to output to the stdio stream
5960 @var{stream} the assembler definition of a common-label named
5961 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
5962 is the size rounded up to whatever alignment the caller wants.
5964 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5965 output the name itself; before and after that, output the additional
5966 assembler syntax for defining the name, and a newline.
5968 This macro controls how the assembler definitions of uninitialized
5969 common global variables are output.
5971 @findex ASM_OUTPUT_ALIGNED_COMMON
5972 @item ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
5973 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
5974 separate, explicit argument. If you define this macro, it is used in
5975 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
5976 handling the required alignment of the variable. The alignment is specified
5977 as the number of bits.
5979 @findex ASM_OUTPUT_ALIGNED_DECL_COMMON
5980 @item ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5981 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
5982 variable to be output, if there is one, or @code{NULL_TREE} if there
5983 is no corresponding variable. If you define this macro, GCC will use it
5984 in place of both @code{ASM_OUTPUT_COMMON} and
5985 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
5986 the variable's decl in order to chose what to output.
5988 @findex ASM_OUTPUT_SHARED_COMMON
5989 @item ASM_OUTPUT_SHARED_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
5990 If defined, it is similar to @code{ASM_OUTPUT_COMMON}, except that it
5991 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_COMMON}
5994 @findex ASM_OUTPUT_BSS
5995 @item ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
5996 A C statement (sans semicolon) to output to the stdio stream
5997 @var{stream} the assembler definition of uninitialized global @var{decl} named
5998 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
5999 is the size rounded up to whatever alignment the caller wants.
6001 Try to use function @code{asm_output_bss} defined in @file{varasm.c} when
6002 defining this macro. If unable, use the expression
6003 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
6004 before and after that, output the additional assembler syntax for defining
6005 the name, and a newline.
6007 This macro controls how the assembler definitions of uninitialized global
6008 variables are output. This macro exists to properly support languages like
6009 C++ which do not have @code{common} data. However, this macro currently
6010 is not defined for all targets. If this macro and
6011 @code{ASM_OUTPUT_ALIGNED_BSS} are not defined then @code{ASM_OUTPUT_COMMON}
6012 or @code{ASM_OUTPUT_ALIGNED_COMMON} or
6013 @code{ASM_OUTPUT_ALIGNED_DECL_COMMON} is used.
6015 @findex ASM_OUTPUT_ALIGNED_BSS
6016 @item ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
6017 Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a
6018 separate, explicit argument. If you define this macro, it is used in
6019 place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in
6020 handling the required alignment of the variable. The alignment is specified
6021 as the number of bits.
6023 Try to use function @code{asm_output_aligned_bss} defined in file
6024 @file{varasm.c} when defining this macro.
6026 @findex ASM_OUTPUT_SHARED_BSS
6027 @item ASM_OUTPUT_SHARED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
6028 If defined, it is similar to @code{ASM_OUTPUT_BSS}, except that it
6029 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_BSS}
6032 @findex ASM_OUTPUT_LOCAL
6033 @item ASM_OUTPUT_LOCAL (@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 local-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 static variables are output.
6046 @findex ASM_OUTPUT_ALIGNED_LOCAL
6047 @item ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
6048 Like @code{ASM_OUTPUT_LOCAL} 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_LOCAL}, 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_LOCAL
6055 @item ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
6056 Like @code{ASM_OUTPUT_ALIGNED_DECL} 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_DECL} and
6060 @code{ASM_OUTPUT_ALIGNED_DECL}. 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_LOCAL
6064 @item ASM_OUTPUT_SHARED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
6065 If defined, it is similar to @code{ASM_OUTPUT_LOCAL}, except that it
6066 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_LOCAL}
6071 @subsection Output and Generation of Labels
6073 @c prevent bad page break with this line
6074 This is about outputting labels.
6077 @findex ASM_OUTPUT_LABEL
6078 @findex assemble_name
6079 @item ASM_OUTPUT_LABEL (@var{stream}, @var{name})
6080 A C statement (sans semicolon) to output to the stdio stream
6081 @var{stream} the assembler definition of a label named @var{name}.
6082 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
6083 output the name itself; before and after that, output the additional
6084 assembler syntax for defining the name, and a newline.
6086 @findex ASM_DECLARE_FUNCTION_NAME
6087 @item ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
6088 A C statement (sans semicolon) to output to the stdio stream
6089 @var{stream} any text necessary for declaring the name @var{name} of a
6090 function which is being defined. This macro is responsible for
6091 outputting the label definition (perhaps using
6092 @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the
6093 @code{FUNCTION_DECL} tree node representing the function.
6095 If this macro is not defined, then the function name is defined in the
6096 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
6098 @findex ASM_DECLARE_FUNCTION_SIZE
6099 @item ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
6100 A C statement (sans semicolon) to output to the stdio stream
6101 @var{stream} any text necessary for declaring the size of a function
6102 which is being defined. The argument @var{name} is the name of the
6103 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
6104 representing the function.
6106 If this macro is not defined, then the function size is not defined.
6108 @findex ASM_DECLARE_OBJECT_NAME
6109 @item ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
6110 A C statement (sans semicolon) to output to the stdio stream
6111 @var{stream} any text necessary for declaring the name @var{name} of an
6112 initialized variable which is being defined. This macro must output the
6113 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
6114 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
6116 If this macro is not defined, then the variable name is defined in the
6117 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
6119 @findex ASM_DECLARE_REGISTER_GLOBAL
6120 @item ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
6121 A C statement (sans semicolon) to output to the stdio stream
6122 @var{stream} any text necessary for claiming a register @var{regno}
6123 for a global variable @var{decl} with name @var{name}.
6125 If you don't define this macro, that is equivalent to defining it to do
6128 @findex ASM_FINISH_DECLARE_OBJECT
6129 @item ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
6130 A C statement (sans semicolon) to finish up declaring a variable name
6131 once the compiler has processed its initializer fully and thus has had a
6132 chance to determine the size of an array when controlled by an
6133 initializer. This is used on systems where it's necessary to declare
6134 something about the size of the object.
6136 If you don't define this macro, that is equivalent to defining it to do
6139 @findex ASM_GLOBALIZE_LABEL
6140 @item ASM_GLOBALIZE_LABEL (@var{stream}, @var{name})
6141 A C statement (sans semicolon) to output to the stdio stream
6142 @var{stream} some commands that will make the label @var{name} global;
6143 that is, available for reference from other files. Use the expression
6144 @code{assemble_name (@var{stream}, @var{name})} to output the name
6145 itself; before and after that, output the additional assembler syntax
6146 for making that name global, and a newline.
6148 @findex ASM_WEAKEN_LABEL
6149 @item ASM_WEAKEN_LABEL
6150 A C statement (sans semicolon) to output to the stdio stream
6151 @var{stream} some commands that will make the label @var{name} weak;
6152 that is, available for reference from other files but only used if
6153 no other definition is available. Use the expression
6154 @code{assemble_name (@var{stream}, @var{name})} to output the name
6155 itself; before and after that, output the additional assembler syntax
6156 for making that name weak, and a newline.
6158 If you don't define this macro, GCC will not support weak
6159 symbols and you should not define the @code{SUPPORTS_WEAK} macro.
6161 @findex SUPPORTS_WEAK
6163 A C expression which evaluates to true if the target supports weak symbols.
6165 If you don't define this macro, @file{defaults.h} provides a default
6166 definition. If @code{ASM_WEAKEN_LABEL} is defined, the default
6167 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
6168 you want to control weak symbol support with a compiler flag such as
6171 @findex MAKE_DECL_ONE_ONLY (@var{decl})
6172 @item MAKE_DECL_ONE_ONLY
6173 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
6174 public symbol such that extra copies in multiple translation units will
6175 be discarded by the linker. Define this macro if your object file
6176 format provides support for this concept, such as the @samp{COMDAT}
6177 section flags in the Microsoft Windows PE/COFF format, and this support
6178 requires changes to @var{decl}, such as putting it in a separate section.
6180 @findex SUPPORTS_ONE_ONLY
6181 @item SUPPORTS_ONE_ONLY
6182 A C expression which evaluates to true if the target supports one-only
6185 If you don't define this macro, @file{varasm.c} provides a default
6186 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
6187 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
6188 you want to control one-only symbol support with a compiler flag, or if
6189 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
6190 be emitted as one-only.
6192 @findex ASM_OUTPUT_EXTERNAL
6193 @item ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
6194 A C statement (sans semicolon) to output to the stdio stream
6195 @var{stream} any text necessary for declaring the name of an external
6196 symbol named @var{name} which is referenced in this compilation but
6197 not defined. The value of @var{decl} is the tree node for the
6200 This macro need not be defined if it does not need to output anything.
6201 The GNU assembler and most Unix assemblers don't require anything.
6203 @findex ASM_OUTPUT_EXTERNAL_LIBCALL
6204 @item ASM_OUTPUT_EXTERNAL_LIBCALL (@var{stream}, @var{symref})
6205 A C statement (sans semicolon) to output on @var{stream} an assembler
6206 pseudo-op to declare a library function name external. The name of the
6207 library function is given by @var{symref}, which has type @code{rtx} and
6208 is a @code{symbol_ref}.
6210 This macro need not be defined if it does not need to output anything.
6211 The GNU assembler and most Unix assemblers don't require anything.
6213 @findex ASM_OUTPUT_LABELREF
6214 @item ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
6215 A C statement (sans semicolon) to output to the stdio stream
6216 @var{stream} a reference in assembler syntax to a label named
6217 @var{name}. This should add @samp{_} to the front of the name, if that
6218 is customary on your operating system, as it is in most Berkeley Unix
6219 systems. This macro is used in @code{assemble_name}.
6221 @findex ASM_OUTPUT_SYMBOL_REF
6222 @item ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
6223 A C statement (sans semicolon) to output a reference to
6224 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
6225 will be used to output the name of the symbol. This macro may be used
6226 to modify the way a symbol is referenced depending on information
6227 encoded by @code{ENCODE_SECTION_INFO}.
6229 @findex ASM_OUTPUT_LABEL_REF
6230 @item ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
6231 A C statement (sans semicolon) to output a reference to @var{buf}, the
6232 result of ASM_GENERATE_INTERNAL_LABEL. If not defined,
6233 @code{assemble_name} will be used to output the name of the symbol.
6234 This macro is not used by @code{output_asm_label}, or the @code{%l}
6235 specifier that calls it; the intention is that this macro should be set
6236 when it is necessary to output a label differently when its address
6239 @findex ASM_OUTPUT_INTERNAL_LABEL
6240 @item ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{prefix}, @var{num})
6241 A C statement to output to the stdio stream @var{stream} a label whose
6242 name is made from the string @var{prefix} and the number @var{num}.
6244 It is absolutely essential that these labels be distinct from the labels
6245 used for user-level functions and variables. Otherwise, certain programs
6246 will have name conflicts with internal labels.
6248 It is desirable to exclude internal labels from the symbol table of the
6249 object file. Most assemblers have a naming convention for labels that
6250 should be excluded; on many systems, the letter @samp{L} at the
6251 beginning of a label has this effect. You should find out what
6252 convention your system uses, and follow it.
6254 The usual definition of this macro is as follows:
6257 fprintf (@var{stream}, "L%s%d:\n", @var{prefix}, @var{num})
6260 @findex ASM_OUTPUT_DEBUG_LABEL
6261 @item ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
6262 A C statement to output to the stdio stream @var{stream} a debug info
6263 label whose name is made from the string @var{prefix} and the number
6264 @var{num}. This is useful for VLIW targets, where debug info labels
6265 may need to be treated differently than branch target labels. On some
6266 systems, branch target labels must be at the beginning of instruction
6267 bundles, but debug info labels can occur in the middle of instruction
6270 If this macro is not defined, then @code{ASM_OUTPUT_INTERNAL_LABEL} will be
6273 @findex ASM_OUTPUT_ALTERNATE_LABEL_NAME
6274 @item ASM_OUTPUT_ALTERNATE_LABEL_NAME (@var{stream}, @var{string})
6275 A C statement to output to the stdio stream @var{stream} the string
6278 The default definition of this macro is as follows:
6281 fprintf (@var{stream}, "%s:\n", LABEL_ALTERNATE_NAME (INSN))
6284 @findex ASM_GENERATE_INTERNAL_LABEL
6285 @item ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
6286 A C statement to store into the string @var{string} a label whose name
6287 is made from the string @var{prefix} and the number @var{num}.
6289 This string, when output subsequently by @code{assemble_name}, should
6290 produce the output that @code{ASM_OUTPUT_INTERNAL_LABEL} would produce
6291 with the same @var{prefix} and @var{num}.
6293 If the string begins with @samp{*}, then @code{assemble_name} will
6294 output the rest of the string unchanged. It is often convenient for
6295 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
6296 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
6297 to output the string, and may change it. (Of course,
6298 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
6299 you should know what it does on your machine.)
6301 @findex ASM_FORMAT_PRIVATE_NAME
6302 @item ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
6303 A C expression to assign to @var{outvar} (which is a variable of type
6304 @code{char *}) a newly allocated string made from the string
6305 @var{name} and the number @var{number}, with some suitable punctuation
6306 added. Use @code{alloca} to get space for the string.
6308 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
6309 produce an assembler label for an internal static variable whose name is
6310 @var{name}. Therefore, the string must be such as to result in valid
6311 assembler code. The argument @var{number} is different each time this
6312 macro is executed; it prevents conflicts between similarly-named
6313 internal static variables in different scopes.
6315 Ideally this string should not be a valid C identifier, to prevent any
6316 conflict with the user's own symbols. Most assemblers allow periods
6317 or percent signs in assembler symbols; putting at least one of these
6318 between the name and the number will suffice.
6320 @findex ASM_OUTPUT_DEF
6321 @item ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
6322 A C statement to output to the stdio stream @var{stream} assembler code
6323 which defines (equates) the symbol @var{name} to have the value @var{value}.
6326 If @code{SET_ASM_OP} is defined, a default definition is provided which is
6327 correct for most systems.
6329 @findex ASM_OUTPUT_DEF_FROM_DECLS
6330 @item ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
6331 A C statement to output to the stdio stream @var{stream} assembler code
6332 which defines (equates) the symbol whose tree node is @var{decl_of_name}
6333 to have the value of the tree node @var{decl_of_value}. This macro will
6334 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
6335 the tree nodes are available.
6337 @findex ASM_OUTPUT_DEFINE_LABEL_DIFFERENCE_SYMBOL
6338 @item ASM_OUTPUT_DEFINE_LABEL_DIFFERENCE_SYMBOL (@var{stream}, @var{symbol}, @var{high}, @var{low})
6339 A C statement to output to the stdio stream @var{stream} assembler code
6340 which defines (equates) the symbol @var{symbol} to have a value equal to
6341 the difference of the two symbols @var{high} and @var{low},
6342 i.e.@: @var{high} minus @var{low}. GCC guarantees that the symbols @var{high}
6343 and @var{low} are already known by the assembler so that the difference
6344 resolves into a constant.
6347 If @code{SET_ASM_OP} is defined, a default definition is provided which is
6348 correct for most systems.
6350 @findex ASM_OUTPUT_WEAK_ALIAS
6351 @item ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
6352 A C statement to output to the stdio stream @var{stream} assembler code
6353 which defines (equates) the weak symbol @var{name} to have the value
6354 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
6355 an undefined weak symbol.
6357 Define this macro if the target only supports weak aliases; define
6358 @code{ASM_OUTPUT_DEF} instead if possible.
6360 @findex OBJC_GEN_METHOD_LABEL
6361 @item OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
6362 Define this macro to override the default assembler names used for
6363 Objective-C methods.
6365 The default name is a unique method number followed by the name of the
6366 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
6367 the category is also included in the assembler name (e.g.@:
6370 These names are safe on most systems, but make debugging difficult since
6371 the method's selector is not present in the name. Therefore, particular
6372 systems define other ways of computing names.
6374 @var{buf} is an expression of type @code{char *} which gives you a
6375 buffer in which to store the name; its length is as long as
6376 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
6377 50 characters extra.
6379 The argument @var{is_inst} specifies whether the method is an instance
6380 method or a class method; @var{class_name} is the name of the class;
6381 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
6382 in a category); and @var{sel_name} is the name of the selector.
6384 On systems where the assembler can handle quoted names, you can use this
6385 macro to provide more human-readable names.
6387 @findex ASM_DECLARE_CLASS_REFERENCE
6388 @item ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
6389 A C statement (sans semicolon) to output to the stdio stream
6390 @var{stream} commands to declare that the label @var{name} is an
6391 Objective-C class reference. This is only needed for targets whose
6392 linkers have special support for NeXT-style runtimes.
6394 @findex ASM_DECLARE_UNRESOLVED_REFERENCE
6395 @item ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
6396 A C statement (sans semicolon) to output to the stdio stream
6397 @var{stream} commands to declare that the label @var{name} is an
6398 unresolved Objective-C class reference. This is only needed for targets
6399 whose linkers have special support for NeXT-style runtimes.
6402 @node Initialization
6403 @subsection How Initialization Functions Are Handled
6404 @cindex initialization routines
6405 @cindex termination routines
6406 @cindex constructors, output of
6407 @cindex destructors, output of
6409 The compiled code for certain languages includes @dfn{constructors}
6410 (also called @dfn{initialization routines})---functions to initialize
6411 data in the program when the program is started. These functions need
6412 to be called before the program is ``started''---that is to say, before
6413 @code{main} is called.
6415 Compiling some languages generates @dfn{destructors} (also called
6416 @dfn{termination routines}) that should be called when the program
6419 To make the initialization and termination functions work, the compiler
6420 must output something in the assembler code to cause those functions to
6421 be called at the appropriate time. When you port the compiler to a new
6422 system, you need to specify how to do this.
6424 There are two major ways that GCC currently supports the execution of
6425 initialization and termination functions. Each way has two variants.
6426 Much of the structure is common to all four variations.
6428 @findex __CTOR_LIST__
6429 @findex __DTOR_LIST__
6430 The linker must build two lists of these functions---a list of
6431 initialization functions, called @code{__CTOR_LIST__}, and a list of
6432 termination functions, called @code{__DTOR_LIST__}.
6434 Each list always begins with an ignored function pointer (which may hold
6435 0, @minus{}1, or a count of the function pointers after it, depending on
6436 the environment). This is followed by a series of zero or more function
6437 pointers to constructors (or destructors), followed by a function
6438 pointer containing zero.
6440 Depending on the operating system and its executable file format, either
6441 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
6442 time and exit time. Constructors are called in reverse order of the
6443 list; destructors in forward order.
6445 The best way to handle static constructors works only for object file
6446 formats which provide arbitrarily-named sections. A section is set
6447 aside for a list of constructors, and another for a list of destructors.
6448 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
6449 object file that defines an initialization function also puts a word in
6450 the constructor section to point to that function. The linker
6451 accumulates all these words into one contiguous @samp{.ctors} section.
6452 Termination functions are handled similarly.
6454 This method will be chosen as the default by @file{target-def.h} if
6455 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
6456 support arbitrary sections, but does support special designated
6457 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
6458 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
6460 When arbitrary sections are available, there are two variants, depending
6461 upon how the code in @file{crtstuff.c} is called. On systems that
6462 support a @dfn{.init} section which is executed at program startup,
6463 parts of @file{crtstuff.c} are compiled into that section. The
6464 program is linked by the @code{gcc} driver like this:
6467 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
6470 The prologue of a function (@code{__init}) appears in the @code{.init}
6471 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
6472 for the function @code{__fini} in the @dfn{.fini} section. Normally these
6473 files are provided by the operating system or by the GNU C library, but
6474 are provided by GCC for a few targets.
6476 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
6477 compiled from @file{crtstuff.c}. They contain, among other things, code
6478 fragments within the @code{.init} and @code{.fini} sections that branch
6479 to routines in the @code{.text} section. The linker will pull all parts
6480 of a section together, which results in a complete @code{__init} function
6481 that invokes the routines we need at startup.
6483 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
6486 If no init section is available, when GCC compiles any function called
6487 @code{main} (or more accurately, any function designated as a program
6488 entry point by the language front end calling @code{expand_main_function}),
6489 it inserts a procedure call to @code{__main} as the first executable code
6490 after the function prologue. The @code{__main} function is defined
6491 in @file{libgcc2.c} and runs the global constructors.
6493 In file formats that don't support arbitrary sections, there are again
6494 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
6495 and an `a.out' format must be used. In this case,
6496 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
6497 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
6498 and with the address of the void function containing the initialization
6499 code as its value. The GNU linker recognizes this as a request to add
6500 the value to a @dfn{set}; the values are accumulated, and are eventually
6501 placed in the executable as a vector in the format described above, with
6502 a leading (ignored) count and a trailing zero element.
6503 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
6504 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
6505 the compilation of @code{main} to call @code{__main} as above, starting
6506 the initialization process.
6508 The last variant uses neither arbitrary sections nor the GNU linker.
6509 This is preferable when you want to do dynamic linking and when using
6510 file formats which the GNU linker does not support, such as `ECOFF'@. In
6511 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
6512 termination functions are recognized simply by their names. This requires
6513 an extra program in the linkage step, called @command{collect2}. This program
6514 pretends to be the linker, for use with GCC; it does its job by running
6515 the ordinary linker, but also arranges to include the vectors of
6516 initialization and termination functions. These functions are called
6517 via @code{__main} as described above. In order to use this method,
6518 @code{use_collect2} must be defined in the target in @file{config.gcc}.
6521 The following section describes the specific macros that control and
6522 customize the handling of initialization and termination functions.
6525 @node Macros for Initialization
6526 @subsection Macros Controlling Initialization Routines
6528 Here are the macros that control how the compiler handles initialization
6529 and termination functions:
6532 @findex INIT_SECTION_ASM_OP
6533 @item INIT_SECTION_ASM_OP
6534 If defined, a C string constant, including spacing, for the assembler
6535 operation to identify the following data as initialization code. If not
6536 defined, GCC will assume such a section does not exist. When you are
6537 using special sections for initialization and termination functions, this
6538 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
6539 run the initialization functions.
6541 @item HAS_INIT_SECTION
6542 @findex HAS_INIT_SECTION
6543 If defined, @code{main} will not call @code{__main} as described above.
6544 This macro should be defined for systems that control start-up code
6545 on a symbol-by-symbol basis, such as OSF/1, and should not
6546 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
6548 @item LD_INIT_SWITCH
6549 @findex LD_INIT_SWITCH
6550 If defined, a C string constant for a switch that tells the linker that
6551 the following symbol is an initialization routine.
6553 @item LD_FINI_SWITCH
6554 @findex LD_FINI_SWITCH
6555 If defined, a C string constant for a switch that tells the linker that
6556 the following symbol is a finalization routine.
6558 @item COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
6559 If defined, a C statement that will write a function that can be
6560 automatically called when a shared library is loaded. The function
6561 should call @var{func}, which takes no arguments. If not defined, and
6562 the object format requires an explicit initialization function, then a
6563 function called @code{_GLOBAL__DI} will be generated.
6565 This function and the following one are used by collect2 when linking a
6566 shared library that needs constructors or destructors, or has DWARF2
6567 exception tables embedded in the code.
6569 @item COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
6570 If defined, a C statement that will write a function that can be
6571 automatically called when a shared library is unloaded. The function
6572 should call @var{func}, which takes no arguments. If not defined, and
6573 the object format requires an explicit finalization function, then a
6574 function called @code{_GLOBAL__DD} will be generated.
6577 @findex INVOKE__main
6578 If defined, @code{main} will call @code{__main} despite the presence of
6579 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
6580 where the init section is not actually run automatically, but is still
6581 useful for collecting the lists of constructors and destructors.
6583 @item SUPPORTS_INIT_PRIORITY
6584 @findex SUPPORTS_INIT_PRIORITY
6585 If nonzero, the C++ @code{init_priority} attribute is supported and the
6586 compiler should emit instructions to control the order of initialization
6587 of objects. If zero, the compiler will issue an error message upon
6588 encountering an @code{init_priority} attribute.
6591 @deftypefn {Target Hook} bool TARGET_HAVE_CTORS_DTORS
6592 This value is true if the target supports some ``native'' method of
6593 collecting constructors and destructors to be run at startup and exit.
6594 It is false if we must use @command{collect2}.
6597 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
6598 If defined, a function that outputs assembler code to arrange to call
6599 the function referenced by @var{symbol} at initialization time.
6601 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
6602 no arguments and with no return value. If the target supports initialization
6603 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
6604 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
6606 If this macro is not defined by the target, a suitable default will
6607 be chosen if (1) the target supports arbitrary section names, (2) the
6608 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
6612 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
6613 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
6614 functions rather than initialization functions.
6617 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
6618 generated for the generated object file will have static linkage.
6620 If your system uses @command{collect2} as the means of processing
6621 constructors, then that program normally uses @command{nm} to scan
6622 an object file for constructor functions to be called.
6624 On certain kinds of systems, you can define these macros to make
6625 @command{collect2} work faster (and, in some cases, make it work at all):
6628 @findex OBJECT_FORMAT_COFF
6629 @item OBJECT_FORMAT_COFF
6630 Define this macro if the system uses COFF (Common Object File Format)
6631 object files, so that @command{collect2} can assume this format and scan
6632 object files directly for dynamic constructor/destructor functions.
6634 @findex OBJECT_FORMAT_ROSE
6635 @item OBJECT_FORMAT_ROSE
6636 Define this macro if the system uses ROSE format object files, so that
6637 @command{collect2} can assume this format and scan object files directly
6638 for dynamic constructor/destructor functions.
6640 These macros are effective only in a native compiler; @command{collect2} as
6641 part of a cross compiler always uses @command{nm} for the target machine.
6643 @findex REAL_NM_FILE_NAME
6644 @item REAL_NM_FILE_NAME
6645 Define this macro as a C string constant containing the file name to use
6646 to execute @command{nm}. The default is to search the path normally for
6649 If your system supports shared libraries and has a program to list the
6650 dynamic dependencies of a given library or executable, you can define
6651 these macros to enable support for running initialization and
6652 termination functions in shared libraries:
6656 Define this macro to a C string constant containing the name of the program
6657 which lists dynamic dependencies, like @command{"ldd"} under SunOS 4.
6659 @findex PARSE_LDD_OUTPUT
6660 @item PARSE_LDD_OUTPUT (@var{ptr})
6661 Define this macro to be C code that extracts filenames from the output
6662 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
6663 of type @code{char *} that points to the beginning of a line of output
6664 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
6665 code must advance @var{ptr} to the beginning of the filename on that
6666 line. Otherwise, it must set @var{ptr} to @code{NULL}.
6669 @node Instruction Output
6670 @subsection Output of Assembler Instructions
6672 @c prevent bad page break with this line
6673 This describes assembler instruction output.
6676 @findex REGISTER_NAMES
6677 @item REGISTER_NAMES
6678 A C initializer containing the assembler's names for the machine
6679 registers, each one as a C string constant. This is what translates
6680 register numbers in the compiler into assembler language.
6682 @findex ADDITIONAL_REGISTER_NAMES
6683 @item ADDITIONAL_REGISTER_NAMES
6684 If defined, a C initializer for an array of structures containing a name
6685 and a register number. This macro defines additional names for hard
6686 registers, thus allowing the @code{asm} option in declarations to refer
6687 to registers using alternate names.
6689 @findex ASM_OUTPUT_OPCODE
6690 @item ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
6691 Define this macro if you are using an unusual assembler that
6692 requires different names for the machine instructions.
6694 The definition is a C statement or statements which output an
6695 assembler instruction opcode to the stdio stream @var{stream}. The
6696 macro-operand @var{ptr} is a variable of type @code{char *} which
6697 points to the opcode name in its ``internal'' form---the form that is
6698 written in the machine description. The definition should output the
6699 opcode name to @var{stream}, performing any translation you desire, and
6700 increment the variable @var{ptr} to point at the end of the opcode
6701 so that it will not be output twice.
6703 In fact, your macro definition may process less than the entire opcode
6704 name, or more than the opcode name; but if you want to process text
6705 that includes @samp{%}-sequences to substitute operands, you must take
6706 care of the substitution yourself. Just be sure to increment
6707 @var{ptr} over whatever text should not be output normally.
6709 @findex recog_data.operand
6710 If you need to look at the operand values, they can be found as the
6711 elements of @code{recog_data.operand}.
6713 If the macro definition does nothing, the instruction is output
6716 @findex FINAL_PRESCAN_INSN
6717 @item FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
6718 If defined, a C statement to be executed just prior to the output of
6719 assembler code for @var{insn}, to modify the extracted operands so
6720 they will be output differently.
6722 Here the argument @var{opvec} is the vector containing the operands
6723 extracted from @var{insn}, and @var{noperands} is the number of
6724 elements of the vector which contain meaningful data for this insn.
6725 The contents of this vector are what will be used to convert the insn
6726 template into assembler code, so you can change the assembler output
6727 by changing the contents of the vector.
6729 This macro is useful when various assembler syntaxes share a single
6730 file of instruction patterns; by defining this macro differently, you
6731 can cause a large class of instructions to be output differently (such
6732 as with rearranged operands). Naturally, variations in assembler
6733 syntax affecting individual insn patterns ought to be handled by
6734 writing conditional output routines in those patterns.
6736 If this macro is not defined, it is equivalent to a null statement.
6738 @findex FINAL_PRESCAN_LABEL
6739 @item FINAL_PRESCAN_LABEL
6740 If defined, @code{FINAL_PRESCAN_INSN} will be called on each
6741 @code{CODE_LABEL}. In that case, @var{opvec} will be a null pointer and
6742 @var{noperands} will be zero.
6744 @findex PRINT_OPERAND
6745 @item PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
6746 A C compound statement to output to stdio stream @var{stream} the
6747 assembler syntax for an instruction operand @var{x}. @var{x} is an
6750 @var{code} is a value that can be used to specify one of several ways
6751 of printing the operand. It is used when identical operands must be
6752 printed differently depending on the context. @var{code} comes from
6753 the @samp{%} specification that was used to request printing of the
6754 operand. If the specification was just @samp{%@var{digit}} then
6755 @var{code} is 0; if the specification was @samp{%@var{ltr}
6756 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
6759 If @var{x} is a register, this macro should print the register's name.
6760 The names can be found in an array @code{reg_names} whose type is
6761 @code{char *[]}. @code{reg_names} is initialized from
6762 @code{REGISTER_NAMES}.
6764 When the machine description has a specification @samp{%@var{punct}}
6765 (a @samp{%} followed by a punctuation character), this macro is called
6766 with a null pointer for @var{x} and the punctuation character for
6769 @findex PRINT_OPERAND_PUNCT_VALID_P
6770 @item PRINT_OPERAND_PUNCT_VALID_P (@var{code})
6771 A C expression which evaluates to true if @var{code} is a valid
6772 punctuation character for use in the @code{PRINT_OPERAND} macro. If
6773 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
6774 punctuation characters (except for the standard one, @samp{%}) are used
6777 @findex PRINT_OPERAND_ADDRESS
6778 @item PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
6779 A C compound statement to output to stdio stream @var{stream} the
6780 assembler syntax for an instruction operand that is a memory reference
6781 whose address is @var{x}. @var{x} is an RTL expression.
6783 @cindex @code{ENCODE_SECTION_INFO} usage
6784 On some machines, the syntax for a symbolic address depends on the
6785 section that the address refers to. On these machines, define the macro
6786 @code{ENCODE_SECTION_INFO} to store the information into the
6787 @code{symbol_ref}, and then check for it here. @xref{Assembler Format}.
6789 @findex DBR_OUTPUT_SEQEND
6790 @findex dbr_sequence_length
6791 @item DBR_OUTPUT_SEQEND(@var{file})
6792 A C statement, to be executed after all slot-filler instructions have
6793 been output. If necessary, call @code{dbr_sequence_length} to
6794 determine the number of slots filled in a sequence (zero if not
6795 currently outputting a sequence), to decide how many no-ops to output,
6798 Don't define this macro if it has nothing to do, but it is helpful in
6799 reading assembly output if the extent of the delay sequence is made
6800 explicit (e.g.@: with white space).
6802 @findex final_sequence
6803 Note that output routines for instructions with delay slots must be
6804 prepared to deal with not being output as part of a sequence
6805 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
6806 found.) The variable @code{final_sequence} is null when not
6807 processing a sequence, otherwise it contains the @code{sequence} rtx
6810 @findex REGISTER_PREFIX
6811 @findex LOCAL_LABEL_PREFIX
6812 @findex USER_LABEL_PREFIX
6813 @findex IMMEDIATE_PREFIX
6815 @item REGISTER_PREFIX
6816 @itemx LOCAL_LABEL_PREFIX
6817 @itemx USER_LABEL_PREFIX
6818 @itemx IMMEDIATE_PREFIX
6819 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
6820 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
6821 @file{final.c}). These are useful when a single @file{md} file must
6822 support multiple assembler formats. In that case, the various @file{tm.h}
6823 files can define these macros differently.
6825 @item ASM_FPRINTF_EXTENSIONS(@var{file}, @var{argptr}, @var{format})
6826 @findex ASM_FPRINTF_EXTENSIONS
6827 If defined this macro should expand to a series of @code{case}
6828 statements which will be parsed inside the @code{switch} statement of
6829 the @code{asm_fprintf} function. This allows targets to define extra
6830 printf formats which may useful when generating their assembler
6831 statements. Note that upper case letters are reserved for future
6832 generic extensions to asm_fprintf, and so are not available to target
6833 specific code. The output file is given by the parameter @var{file}.
6834 The varargs input pointer is @var{argptr} and the rest of the format
6835 string, starting the character after the one that is being switched
6836 upon, is pointed to by @var{format}.
6838 @findex ASSEMBLER_DIALECT
6839 @item ASSEMBLER_DIALECT
6840 If your target supports multiple dialects of assembler language (such as
6841 different opcodes), define this macro as a C expression that gives the
6842 numeric index of the assembler language dialect to use, with zero as the
6845 If this macro is defined, you may use constructs of the form
6847 @samp{@{option0|option1|option2@dots{}@}}
6850 in the output templates of patterns (@pxref{Output Template}) or in the
6851 first argument of @code{asm_fprintf}. This construct outputs
6852 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
6853 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
6854 within these strings retain their usual meaning. If there are fewer
6855 alternatives within the braces than the value of
6856 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
6858 If you do not define this macro, the characters @samp{@{}, @samp{|} and
6859 @samp{@}} do not have any special meaning when used in templates or
6860 operands to @code{asm_fprintf}.
6862 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
6863 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
6864 the variations in assembler language syntax with that mechanism. Define
6865 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
6866 if the syntax variant are larger and involve such things as different
6867 opcodes or operand order.
6869 @findex ASM_OUTPUT_REG_PUSH
6870 @item ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
6871 A C expression to output to @var{stream} some assembler code
6872 which will push hard register number @var{regno} onto the stack.
6873 The code need not be optimal, since this macro is used only when
6876 @findex ASM_OUTPUT_REG_POP
6877 @item ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
6878 A C expression to output to @var{stream} some assembler code
6879 which will pop hard register number @var{regno} off of the stack.
6880 The code need not be optimal, since this macro is used only when
6884 @node Dispatch Tables
6885 @subsection Output of Dispatch Tables
6887 @c prevent bad page break with this line
6888 This concerns dispatch tables.
6891 @cindex dispatch table
6892 @findex ASM_OUTPUT_ADDR_DIFF_ELT
6893 @item ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
6894 A C statement to output to the stdio stream @var{stream} an assembler
6895 pseudo-instruction to generate a difference between two labels.
6896 @var{value} and @var{rel} are the numbers of two internal labels. The
6897 definitions of these labels are output using
6898 @code{ASM_OUTPUT_INTERNAL_LABEL}, and they must be printed in the same
6899 way here. For example,
6902 fprintf (@var{stream}, "\t.word L%d-L%d\n",
6903 @var{value}, @var{rel})
6906 You must provide this macro on machines where the addresses in a
6907 dispatch table are relative to the table's own address. If defined, GCC
6908 will also use this macro on all machines when producing PIC@.
6909 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
6910 mode and flags can be read.
6912 @findex ASM_OUTPUT_ADDR_VEC_ELT
6913 @item ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
6914 This macro should be provided on machines where the addresses
6915 in a dispatch table are absolute.
6917 The definition should be a C statement to output to the stdio stream
6918 @var{stream} an assembler pseudo-instruction to generate a reference to
6919 a label. @var{value} is the number of an internal label whose
6920 definition is output using @code{ASM_OUTPUT_INTERNAL_LABEL}.
6924 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
6927 @findex ASM_OUTPUT_CASE_LABEL
6928 @item ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
6929 Define this if the label before a jump-table needs to be output
6930 specially. The first three arguments are the same as for
6931 @code{ASM_OUTPUT_INTERNAL_LABEL}; the fourth argument is the
6932 jump-table which follows (a @code{jump_insn} containing an
6933 @code{addr_vec} or @code{addr_diff_vec}).
6935 This feature is used on system V to output a @code{swbeg} statement
6938 If this macro is not defined, these labels are output with
6939 @code{ASM_OUTPUT_INTERNAL_LABEL}.
6941 @findex ASM_OUTPUT_CASE_END
6942 @item ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
6943 Define this if something special must be output at the end of a
6944 jump-table. The definition should be a C statement to be executed
6945 after the assembler code for the table is written. It should write
6946 the appropriate code to stdio stream @var{stream}. The argument
6947 @var{table} is the jump-table insn, and @var{num} is the label-number
6948 of the preceding label.
6950 If this macro is not defined, nothing special is output at the end of
6954 @node Exception Region Output
6955 @subsection Assembler Commands for Exception Regions
6957 @c prevent bad page break with this line
6959 This describes commands marking the start and the end of an exception
6963 @findex EH_FRAME_SECTION_NAME
6964 @item EH_FRAME_SECTION_NAME
6965 If defined, a C string constant for the name of the section containing
6966 exception handling frame unwind information. If not defined, GCC will
6967 provide a default definition if the target supports named sections.
6968 @file{crtstuff.c} uses this macro to switch to the appropriate section.
6970 You should define this symbol if your target supports DWARF 2 frame
6971 unwind information and the default definition does not work.
6973 @findex EH_FRAME_IN_DATA_SECTION
6974 @item EH_FRAME_IN_DATA_SECTION
6975 If defined, DWARF 2 frame unwind information will be placed in the
6976 data section even though the target supports named sections. This
6977 might be necessary, for instance, if the system linker does garbage
6978 collection and sections cannot be marked as not to be collected.
6980 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
6983 @findex MASK_RETURN_ADDR
6984 @item MASK_RETURN_ADDR
6985 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
6986 that it does not contain any extraneous set bits in it.
6988 @findex DWARF2_UNWIND_INFO
6989 @item DWARF2_UNWIND_INFO
6990 Define this macro to 0 if your target supports DWARF 2 frame unwind
6991 information, but it does not yet work with exception handling.
6992 Otherwise, if your target supports this information (if it defines
6993 @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP}
6994 or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of
6997 If this macro is defined to 1, the DWARF 2 unwinder will be the default
6998 exception handling mechanism; otherwise, @code{setjmp}/@code{longjmp} will be used by
7001 If this macro is defined to anything, the DWARF 2 unwinder will be used
7002 instead of inline unwinders and @code{__unwind_function} in the non-@code{setjmp} case.
7004 @findex DWARF_CIE_DATA_ALIGNMENT
7005 @item DWARF_CIE_DATA_ALIGNMENT
7006 This macro need only be defined if the target might save registers in the
7007 function prologue at an offset to the stack pointer that is not aligned to
7008 @code{UNITS_PER_WORD}. The definition should be the negative minimum
7009 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
7010 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
7011 the target supports DWARF 2 frame unwind information.
7015 @deftypefn {Target Hook} void TARGET_ASM_EXCEPTION_SECTION ()
7016 If defined, a function that switches to the section in which the main
7017 exception table is to be placed (@pxref{Sections}). The default is a
7018 function that switches to a section named @code{.gcc_except_table} on
7019 machines that support named sections via
7020 @code{TARGET_ASM_NAMED_SECTION}, otherwise if @option{-fpic} or
7021 @option{-fPIC} is in effect, the @code{data_section}, otherwise the
7022 @code{readonly_data_section}.
7025 @deftypefn {Target Hook} void TARGET_ASM_EH_FRAME_SECTION ()
7026 If defined, a function that switches to the section in which the DWARF 2
7027 frame unwind information to be placed (@pxref{Sections}). The default
7028 is a function that outputs a standard GAS section directive, if
7029 @code{EH_FRAME_SECTION_NAME} is defined, or else a data section
7030 directive followed by a synthetic label.
7033 @node Alignment Output
7034 @subsection Assembler Commands for Alignment
7036 @c prevent bad page break with this line
7037 This describes commands for alignment.
7041 @item JUMP_ALIGN (@var{label})
7042 The alignment (log base 2) to put in front of @var{label}, which is
7043 a common destination of jumps and has no fallthru incoming edge.
7045 This macro need not be defined if you don't want any special alignment
7046 to be done at such a time. Most machine descriptions do not currently
7049 Unless it's necessary to inspect the @var{label} parameter, it is better
7050 to set the variable @var{align_jumps} in the target's
7051 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7052 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
7054 @findex LABEL_ALIGN_AFTER_BARRIER
7055 @item LABEL_ALIGN_AFTER_BARRIER (@var{label})
7056 The alignment (log base 2) to put in front of @var{label}, which follows
7059 This macro need not be defined if you don't want any special alignment
7060 to be done at such a time. Most machine descriptions do not currently
7063 @findex LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
7064 @item LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
7065 The maximum number of bytes to skip when applying
7066 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
7067 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7070 @item LOOP_ALIGN (@var{label})
7071 The alignment (log base 2) to put in front of @var{label}, which follows
7072 a @code{NOTE_INSN_LOOP_BEG} note.
7074 This macro need not be defined if you don't want any special alignment
7075 to be done at such a time. Most machine descriptions do not currently
7078 Unless it's necessary to inspect the @var{label} parameter, it is better
7079 to set the variable @code{align_loops} in the target's
7080 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7081 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
7083 @findex LOOP_ALIGN_MAX_SKIP
7084 @item LOOP_ALIGN_MAX_SKIP
7085 The maximum number of bytes to skip when applying @code{LOOP_ALIGN}.
7086 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7089 @item LABEL_ALIGN (@var{label})
7090 The alignment (log base 2) to put in front of @var{label}.
7091 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
7092 the maximum of the specified values is used.
7094 Unless it's necessary to inspect the @var{label} parameter, it is better
7095 to set the variable @code{align_labels} in the target's
7096 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7097 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
7099 @findex LABEL_ALIGN_MAX_SKIP
7100 @item LABEL_ALIGN_MAX_SKIP
7101 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}.
7102 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7104 @findex ASM_OUTPUT_SKIP
7105 @item ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
7106 A C statement to output to the stdio stream @var{stream} an assembler
7107 instruction to advance the location counter by @var{nbytes} bytes.
7108 Those bytes should be zero when loaded. @var{nbytes} will be a C
7109 expression of type @code{int}.
7111 @findex ASM_NO_SKIP_IN_TEXT
7112 @item ASM_NO_SKIP_IN_TEXT
7113 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
7114 text section because it fails to put zeros in the bytes that are skipped.
7115 This is true on many Unix systems, where the pseudo--op to skip bytes
7116 produces no-op instructions rather than zeros when used in the text
7119 @findex ASM_OUTPUT_ALIGN
7120 @item ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
7121 A C statement to output to the stdio stream @var{stream} an assembler
7122 command to advance the location counter to a multiple of 2 to the
7123 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
7125 @findex ASM_OUTPUT_MAX_SKIP_ALIGN
7126 @item ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
7127 A C statement to output to the stdio stream @var{stream} an assembler
7128 command to advance the location counter to a multiple of 2 to the
7129 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
7130 satisfy the alignment request. @var{power} and @var{max_skip} will be
7131 a C expression of type @code{int}.
7135 @node Debugging Info
7136 @section Controlling Debugging Information Format
7138 @c prevent bad page break with this line
7139 This describes how to specify debugging information.
7142 * All Debuggers:: Macros that affect all debugging formats uniformly.
7143 * DBX Options:: Macros enabling specific options in DBX format.
7144 * DBX Hooks:: Hook macros for varying DBX format.
7145 * File Names and DBX:: Macros controlling output of file names in DBX format.
7146 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
7147 * VMS Debug:: Macros for VMS debug format.
7151 @subsection Macros Affecting All Debugging Formats
7153 @c prevent bad page break with this line
7154 These macros affect all debugging formats.
7157 @findex DBX_REGISTER_NUMBER
7158 @item DBX_REGISTER_NUMBER (@var{regno})
7159 A C expression that returns the DBX register number for the compiler
7160 register number @var{regno}. In the default macro provided, the value
7161 of this expression will be @var{regno} itself. But sometimes there are
7162 some registers that the compiler knows about and DBX does not, or vice
7163 versa. In such cases, some register may need to have one number in the
7164 compiler and another for DBX@.
7166 If two registers have consecutive numbers inside GCC, and they can be
7167 used as a pair to hold a multiword value, then they @emph{must} have
7168 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
7169 Otherwise, debuggers will be unable to access such a pair, because they
7170 expect register pairs to be consecutive in their own numbering scheme.
7172 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
7173 does not preserve register pairs, then what you must do instead is
7174 redefine the actual register numbering scheme.
7176 @findex DEBUGGER_AUTO_OFFSET
7177 @item DEBUGGER_AUTO_OFFSET (@var{x})
7178 A C expression that returns the integer offset value for an automatic
7179 variable having address @var{x} (an RTL expression). The default
7180 computation assumes that @var{x} is based on the frame-pointer and
7181 gives the offset from the frame-pointer. This is required for targets
7182 that produce debugging output for DBX or COFF-style debugging output
7183 for SDB and allow the frame-pointer to be eliminated when the
7184 @option{-g} options is used.
7186 @findex DEBUGGER_ARG_OFFSET
7187 @item DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
7188 A C expression that returns the integer offset value for an argument
7189 having address @var{x} (an RTL expression). The nominal offset is
7192 @findex PREFERRED_DEBUGGING_TYPE
7193 @item PREFERRED_DEBUGGING_TYPE
7194 A C expression that returns the type of debugging output GCC should
7195 produce when the user specifies just @option{-g}. Define
7196 this if you have arranged for GCC to support more than one format of
7197 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
7198 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
7199 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
7201 When the user specifies @option{-ggdb}, GCC normally also uses the
7202 value of this macro to select the debugging output format, but with two
7203 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined and
7204 @code{LINKER_DOES_NOT_WORK_WITH_DWARF2} is not defined, GCC uses the
7205 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
7206 defined, GCC uses @code{DBX_DEBUG}.
7208 The value of this macro only affects the default debugging output; the
7209 user can always get a specific type of output by using @option{-gstabs},
7210 @option{-gcoff}, @option{-gdwarf-1}, @option{-gdwarf-2}, @option{-gxcoff},
7215 @subsection Specific Options for DBX Output
7217 @c prevent bad page break with this line
7218 These are specific options for DBX output.
7221 @findex DBX_DEBUGGING_INFO
7222 @item DBX_DEBUGGING_INFO
7223 Define this macro if GCC should produce debugging output for DBX
7224 in response to the @option{-g} option.
7226 @findex XCOFF_DEBUGGING_INFO
7227 @item XCOFF_DEBUGGING_INFO
7228 Define this macro if GCC should produce XCOFF format debugging output
7229 in response to the @option{-g} option. This is a variant of DBX format.
7231 @findex DEFAULT_GDB_EXTENSIONS
7232 @item DEFAULT_GDB_EXTENSIONS
7233 Define this macro to control whether GCC should by default generate
7234 GDB's extended version of DBX debugging information (assuming DBX-format
7235 debugging information is enabled at all). If you don't define the
7236 macro, the default is 1: always generate the extended information
7237 if there is any occasion to.
7239 @findex DEBUG_SYMS_TEXT
7240 @item DEBUG_SYMS_TEXT
7241 Define this macro if all @code{.stabs} commands should be output while
7242 in the text section.
7244 @findex ASM_STABS_OP
7246 A C string constant, including spacing, naming the assembler pseudo op to
7247 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
7248 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
7249 applies only to DBX debugging information format.
7251 @findex ASM_STABD_OP
7253 A C string constant, including spacing, naming the assembler pseudo op to
7254 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
7255 value is the current location. If you don't define this macro,
7256 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
7259 @findex ASM_STABN_OP
7261 A C string constant, including spacing, naming the assembler pseudo op to
7262 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
7263 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
7264 macro applies only to DBX debugging information format.
7266 @findex DBX_NO_XREFS
7268 Define this macro if DBX on your system does not support the construct
7269 @samp{xs@var{tagname}}. On some systems, this construct is used to
7270 describe a forward reference to a structure named @var{tagname}.
7271 On other systems, this construct is not supported at all.
7273 @findex DBX_CONTIN_LENGTH
7274 @item DBX_CONTIN_LENGTH
7275 A symbol name in DBX-format debugging information is normally
7276 continued (split into two separate @code{.stabs} directives) when it
7277 exceeds a certain length (by default, 80 characters). On some
7278 operating systems, DBX requires this splitting; on others, splitting
7279 must not be done. You can inhibit splitting by defining this macro
7280 with the value zero. You can override the default splitting-length by
7281 defining this macro as an expression for the length you desire.
7283 @findex DBX_CONTIN_CHAR
7284 @item DBX_CONTIN_CHAR
7285 Normally continuation is indicated by adding a @samp{\} character to
7286 the end of a @code{.stabs} string when a continuation follows. To use
7287 a different character instead, define this macro as a character
7288 constant for the character you want to use. Do not define this macro
7289 if backslash is correct for your system.
7291 @findex DBX_STATIC_STAB_DATA_SECTION
7292 @item DBX_STATIC_STAB_DATA_SECTION
7293 Define this macro if it is necessary to go to the data section before
7294 outputting the @samp{.stabs} pseudo-op for a non-global static
7297 @findex DBX_TYPE_DECL_STABS_CODE
7298 @item DBX_TYPE_DECL_STABS_CODE
7299 The value to use in the ``code'' field of the @code{.stabs} directive
7300 for a typedef. The default is @code{N_LSYM}.
7302 @findex DBX_STATIC_CONST_VAR_CODE
7303 @item DBX_STATIC_CONST_VAR_CODE
7304 The value to use in the ``code'' field of the @code{.stabs} directive
7305 for a static variable located in the text section. DBX format does not
7306 provide any ``right'' way to do this. The default is @code{N_FUN}.
7308 @findex DBX_REGPARM_STABS_CODE
7309 @item DBX_REGPARM_STABS_CODE
7310 The value to use in the ``code'' field of the @code{.stabs} directive
7311 for a parameter passed in registers. DBX format does not provide any
7312 ``right'' way to do this. The default is @code{N_RSYM}.
7314 @findex DBX_REGPARM_STABS_LETTER
7315 @item DBX_REGPARM_STABS_LETTER
7316 The letter to use in DBX symbol data to identify a symbol as a parameter
7317 passed in registers. DBX format does not customarily provide any way to
7318 do this. The default is @code{'P'}.
7320 @findex DBX_MEMPARM_STABS_LETTER
7321 @item DBX_MEMPARM_STABS_LETTER
7322 The letter to use in DBX symbol data to identify a symbol as a stack
7323 parameter. The default is @code{'p'}.
7325 @findex DBX_FUNCTION_FIRST
7326 @item DBX_FUNCTION_FIRST
7327 Define this macro if the DBX information for a function and its
7328 arguments should precede the assembler code for the function. Normally,
7329 in DBX format, the debugging information entirely follows the assembler
7332 @findex DBX_LBRAC_FIRST
7333 @item DBX_LBRAC_FIRST
7334 Define this macro if the @code{N_LBRAC} symbol for a block should
7335 precede the debugging information for variables and functions defined in
7336 that block. Normally, in DBX format, the @code{N_LBRAC} symbol comes
7339 @findex DBX_BLOCKS_FUNCTION_RELATIVE
7340 @item DBX_BLOCKS_FUNCTION_RELATIVE
7341 Define this macro if the value of a symbol describing the scope of a
7342 block (@code{N_LBRAC} or @code{N_RBRAC}) should be relative to the start
7343 of the enclosing function. Normally, GCC uses an absolute address.
7345 @findex DBX_USE_BINCL
7347 Define this macro if GCC should generate @code{N_BINCL} and
7348 @code{N_EINCL} stabs for included header files, as on Sun systems. This
7349 macro also directs GCC to output a type number as a pair of a file
7350 number and a type number within the file. Normally, GCC does not
7351 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
7352 number for a type number.
7356 @subsection Open-Ended Hooks for DBX Format
7358 @c prevent bad page break with this line
7359 These are hooks for DBX format.
7362 @findex DBX_OUTPUT_LBRAC
7363 @item DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
7364 Define this macro to say how to output to @var{stream} the debugging
7365 information for the start of a scope level for variable names. The
7366 argument @var{name} is the name of an assembler symbol (for use with
7367 @code{assemble_name}) whose value is the address where the scope begins.
7369 @findex DBX_OUTPUT_RBRAC
7370 @item DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
7371 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
7373 @findex DBX_OUTPUT_ENUM
7374 @item DBX_OUTPUT_ENUM (@var{stream}, @var{type})
7375 Define this macro if the target machine requires special handling to
7376 output an enumeration type. The definition should be a C statement
7377 (sans semicolon) to output the appropriate information to @var{stream}
7378 for the type @var{type}.
7380 @findex DBX_OUTPUT_FUNCTION_END
7381 @item DBX_OUTPUT_FUNCTION_END (@var{stream}, @var{function})
7382 Define this macro if the target machine requires special output at the
7383 end of the debugging information for a function. The definition should
7384 be a C statement (sans semicolon) to output the appropriate information
7385 to @var{stream}. @var{function} is the @code{FUNCTION_DECL} node for
7388 @findex DBX_OUTPUT_STANDARD_TYPES
7389 @item DBX_OUTPUT_STANDARD_TYPES (@var{syms})
7390 Define this macro if you need to control the order of output of the
7391 standard data types at the beginning of compilation. The argument
7392 @var{syms} is a @code{tree} which is a chain of all the predefined
7393 global symbols, including names of data types.
7395 Normally, DBX output starts with definitions of the types for integers
7396 and characters, followed by all the other predefined types of the
7397 particular language in no particular order.
7399 On some machines, it is necessary to output different particular types
7400 first. To do this, define @code{DBX_OUTPUT_STANDARD_TYPES} to output
7401 those symbols in the necessary order. Any predefined types that you
7402 don't explicitly output will be output afterward in no particular order.
7404 Be careful not to define this macro so that it works only for C@. There
7405 are no global variables to access most of the built-in types, because
7406 another language may have another set of types. The way to output a
7407 particular type is to look through @var{syms} to see if you can find it.
7413 for (decl = syms; decl; decl = TREE_CHAIN (decl))
7414 if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)),
7416 dbxout_symbol (decl);
7422 This does nothing if the expected type does not exist.
7424 See the function @code{init_decl_processing} in @file{c-decl.c} to find
7425 the names to use for all the built-in C types.
7427 Here is another way of finding a particular type:
7429 @c this is still overfull. --mew 10feb93
7433 for (decl = syms; decl; decl = TREE_CHAIN (decl))
7434 if (TREE_CODE (decl) == TYPE_DECL
7435 && (TREE_CODE (TREE_TYPE (decl))
7437 && TYPE_PRECISION (TREE_TYPE (decl)) == 16
7438 && TYPE_UNSIGNED (TREE_TYPE (decl)))
7440 /* @r{This must be @code{unsigned short}.} */
7441 dbxout_symbol (decl);
7447 @findex NO_DBX_FUNCTION_END
7448 @item NO_DBX_FUNCTION_END
7449 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
7450 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
7451 On those machines, define this macro to turn this feature off without
7452 disturbing the rest of the gdb extensions.
7456 @node File Names and DBX
7457 @subsection File Names in DBX Format
7459 @c prevent bad page break with this line
7460 This describes file names in DBX format.
7463 @findex DBX_WORKING_DIRECTORY
7464 @item DBX_WORKING_DIRECTORY
7465 Define this if DBX wants to have the current directory recorded in each
7468 Note that the working directory is always recorded if GDB extensions are
7471 @findex DBX_OUTPUT_MAIN_SOURCE_FILENAME
7472 @item DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
7473 A C statement to output DBX debugging information to the stdio stream
7474 @var{stream} which indicates that file @var{name} is the main source
7475 file---the file specified as the input file for compilation.
7476 This macro is called only once, at the beginning of compilation.
7478 This macro need not be defined if the standard form of output
7479 for DBX debugging information is appropriate.
7481 @findex DBX_OUTPUT_MAIN_SOURCE_DIRECTORY
7482 @item DBX_OUTPUT_MAIN_SOURCE_DIRECTORY (@var{stream}, @var{name})
7483 A C statement to output DBX debugging information to the stdio stream
7484 @var{stream} which indicates that the current directory during
7485 compilation is named @var{name}.
7487 This macro need not be defined if the standard form of output
7488 for DBX debugging information is appropriate.
7490 @findex DBX_OUTPUT_MAIN_SOURCE_FILE_END
7491 @item DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
7492 A C statement to output DBX debugging information at the end of
7493 compilation of the main source file @var{name}.
7495 If you don't define this macro, nothing special is output at the end
7496 of compilation, which is correct for most machines.
7498 @findex DBX_OUTPUT_SOURCE_FILENAME
7499 @item DBX_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7500 A C statement to output DBX debugging information to the stdio stream
7501 @var{stream} which indicates that file @var{name} is the current source
7502 file. This output is generated each time input shifts to a different
7503 source file as a result of @samp{#include}, the end of an included file,
7504 or a @samp{#line} command.
7506 This macro need not be defined if the standard form of output
7507 for DBX debugging information is appropriate.
7512 @subsection Macros for SDB and DWARF Output
7514 @c prevent bad page break with this line
7515 Here are macros for SDB and DWARF output.
7518 @findex SDB_DEBUGGING_INFO
7519 @item SDB_DEBUGGING_INFO
7520 Define this macro if GCC should produce COFF-style debugging output
7521 for SDB in response to the @option{-g} option.
7523 @findex DWARF_DEBUGGING_INFO
7524 @item DWARF_DEBUGGING_INFO
7525 Define this macro if GCC should produce dwarf format debugging output
7526 in response to the @option{-g} option.
7528 @findex DWARF2_DEBUGGING_INFO
7529 @item DWARF2_DEBUGGING_INFO
7530 Define this macro if GCC should produce dwarf version 2 format
7531 debugging output in response to the @option{-g} option.
7533 To support optional call frame debugging information, you must also
7534 define @code{INCOMING_RETURN_ADDR_RTX} and either set
7535 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
7536 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
7537 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
7539 @findex DWARF2_FRAME_INFO
7540 @item DWARF2_FRAME_INFO
7541 Define this macro to a nonzero value if GCC should always output
7542 Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO}
7543 (@pxref{Exception Region Output} is nonzero, GCC will output this
7544 information not matter how you define @code{DWARF2_FRAME_INFO}.
7546 @findex LINKER_DOES_NOT_WORK_WITH_DWARF2
7547 @item LINKER_DOES_NOT_WORK_WITH_DWARF2
7548 Define this macro if the linker does not work with Dwarf version 2.
7549 Normally, if the user specifies only @option{-ggdb} GCC will use Dwarf
7550 version 2 if available; this macro disables this. See the description
7551 of the @code{PREFERRED_DEBUGGING_TYPE} macro for more details.
7553 @findex DWARF2_GENERATE_TEXT_SECTION_LABEL
7554 @item DWARF2_GENERATE_TEXT_SECTION_LABEL
7555 By default, the Dwarf 2 debugging information generator will generate a
7556 label to mark the beginning of the text section. If it is better simply
7557 to use the name of the text section itself, rather than an explicit label,
7558 to indicate the beginning of the text section, define this macro to zero.
7560 @findex DWARF2_ASM_LINE_DEBUG_INFO
7561 @item DWARF2_ASM_LINE_DEBUG_INFO
7562 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
7563 line debug info sections. This will result in much more compact line number
7564 tables, and hence is desirable if it works.
7566 @findex PUT_SDB_@dots{}
7567 @item PUT_SDB_@dots{}
7568 Define these macros to override the assembler syntax for the special
7569 SDB assembler directives. See @file{sdbout.c} for a list of these
7570 macros and their arguments. If the standard syntax is used, you need
7571 not define them yourself.
7575 Some assemblers do not support a semicolon as a delimiter, even between
7576 SDB assembler directives. In that case, define this macro to be the
7577 delimiter to use (usually @samp{\n}). It is not necessary to define
7578 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
7581 @findex SDB_GENERATE_FAKE
7582 @item SDB_GENERATE_FAKE
7583 Define this macro to override the usual method of constructing a dummy
7584 name for anonymous structure and union types. See @file{sdbout.c} for
7587 @findex SDB_ALLOW_UNKNOWN_REFERENCES
7588 @item SDB_ALLOW_UNKNOWN_REFERENCES
7589 Define this macro to allow references to unknown structure,
7590 union, or enumeration tags to be emitted. Standard COFF does not
7591 allow handling of unknown references, MIPS ECOFF has support for
7594 @findex SDB_ALLOW_FORWARD_REFERENCES
7595 @item SDB_ALLOW_FORWARD_REFERENCES
7596 Define this macro to allow references to structure, union, or
7597 enumeration tags that have not yet been seen to be handled. Some
7598 assemblers choke if forward tags are used, while some require it.
7603 @subsection Macros for VMS Debug Format
7605 @c prevent bad page break with this line
7606 Here are macros for VMS debug format.
7609 @findex VMS_DEBUGGING_INFO
7610 @item VMS_DEBUGGING_INFO
7611 Define this macro if GCC should produce debugging output for VMS
7612 in response to the @option{-g} option. The default behavior for VMS
7613 is to generate minimal debug info for a traceback in the absence of
7614 @option{-g} unless explicitly overridden with @option{-g0}. This
7615 behavior is controlled by @code{OPTIMIZATION_OPTIONS} and
7616 @code{OVERRIDE_OPTIONS}.
7619 @node Cross-compilation
7620 @section Cross Compilation and Floating Point
7621 @cindex cross compilation and floating point
7622 @cindex floating point and cross compilation
7624 While all modern machines use 2's complement representation for integers,
7625 there are a variety of representations for floating point numbers. This
7626 means that in a cross-compiler the representation of floating point numbers
7627 in the compiled program may be different from that used in the machine
7628 doing the compilation.
7631 Because different representation systems may offer different amounts of
7632 range and precision, the cross compiler cannot safely use the host
7633 machine's floating point arithmetic. Therefore, floating point constants
7634 must be represented in the target machine's format. This means that the
7635 cross compiler cannot use @code{atof} to parse a floating point constant;
7636 it must have its own special routine to use instead. Also, constant
7637 folding must emulate the target machine's arithmetic (or must not be done
7640 The macros in the following table should be defined only if you are cross
7641 compiling between different floating point formats.
7643 Otherwise, don't define them. Then default definitions will be set up which
7644 use @code{double} as the data type, @code{==} to test for equality, etc.
7646 You don't need to worry about how many times you use an operand of any
7647 of these macros. The compiler never uses operands which have side effects.
7650 @findex REAL_VALUE_TYPE
7651 @item REAL_VALUE_TYPE
7652 A macro for the C data type to be used to hold a floating point value
7653 in the target machine's format. Typically this would be a
7654 @code{struct} containing an array of @code{int}.
7656 @findex REAL_VALUES_EQUAL
7657 @item REAL_VALUES_EQUAL (@var{x}, @var{y})
7658 A macro for a C expression which compares for equality the two values,
7659 @var{x} and @var{y}, both of type @code{REAL_VALUE_TYPE}.
7661 @findex REAL_VALUES_LESS
7662 @item REAL_VALUES_LESS (@var{x}, @var{y})
7663 A macro for a C expression which tests whether @var{x} is less than
7664 @var{y}, both values being of type @code{REAL_VALUE_TYPE} and
7665 interpreted as floating point numbers in the target machine's
7668 @findex REAL_VALUE_LDEXP
7670 @item REAL_VALUE_LDEXP (@var{x}, @var{scale})
7671 A macro for a C expression which performs the standard library
7672 function @code{ldexp}, but using the target machine's floating point
7673 representation. Both @var{x} and the value of the expression have
7674 type @code{REAL_VALUE_TYPE}. The second argument, @var{scale}, is an
7677 @findex REAL_VALUE_FIX
7678 @item REAL_VALUE_FIX (@var{x})
7679 A macro whose definition is a C expression to convert the target-machine
7680 floating point value @var{x} to a signed integer. @var{x} has type
7681 @code{REAL_VALUE_TYPE}.
7683 @findex REAL_VALUE_UNSIGNED_FIX
7684 @item REAL_VALUE_UNSIGNED_FIX (@var{x})
7685 A macro whose definition is a C expression to convert the target-machine
7686 floating point value @var{x} to an unsigned integer. @var{x} has type
7687 @code{REAL_VALUE_TYPE}.
7689 @findex REAL_VALUE_RNDZINT
7690 @item REAL_VALUE_RNDZINT (@var{x})
7691 A macro whose definition is a C expression to round the target-machine
7692 floating point value @var{x} towards zero to an integer value (but still
7693 as a floating point number). @var{x} has type @code{REAL_VALUE_TYPE},
7694 and so does the value.
7696 @findex REAL_VALUE_UNSIGNED_RNDZINT
7697 @item REAL_VALUE_UNSIGNED_RNDZINT (@var{x})
7698 A macro whose definition is a C expression to round the target-machine
7699 floating point value @var{x} towards zero to an unsigned integer value
7700 (but still represented as a floating point number). @var{x} has type
7701 @code{REAL_VALUE_TYPE}, and so does the value.
7703 @findex REAL_VALUE_ATOF
7704 @item REAL_VALUE_ATOF (@var{string}, @var{mode})
7705 A macro for a C expression which converts @var{string}, an expression of
7706 type @code{char *}, into a floating point number in the target machine's
7707 representation for mode @var{mode}. The value has type
7708 @code{REAL_VALUE_TYPE}.
7710 @findex REAL_INFINITY
7712 Define this macro if infinity is a possible floating point value, and
7713 therefore division by 0 is legitimate.
7715 @findex REAL_VALUE_ISINF
7717 @item REAL_VALUE_ISINF (@var{x})
7718 A macro for a C expression which determines whether @var{x}, a floating
7719 point value, is infinity. The value has type @code{int}.
7720 By default, this is defined to call @code{isinf}.
7722 @findex REAL_VALUE_ISNAN
7724 @item REAL_VALUE_ISNAN (@var{x})
7725 A macro for a C expression which determines whether @var{x}, a floating
7726 point value, is a ``nan'' (not-a-number). The value has type
7727 @code{int}. By default, this is defined to call @code{isnan}.
7730 @cindex constant folding and floating point
7731 Define the following additional macros if you want to make floating
7732 point constant folding work while cross compiling. If you don't
7733 define them, cross compilation is still possible, but constant folding
7734 will not happen for floating point values.
7737 @findex REAL_ARITHMETIC
7738 @item REAL_ARITHMETIC (@var{output}, @var{code}, @var{x}, @var{y})
7739 A macro for a C statement which calculates an arithmetic operation of
7740 the two floating point values @var{x} and @var{y}, both of type
7741 @code{REAL_VALUE_TYPE} in the target machine's representation, to
7742 produce a result of the same type and representation which is stored
7743 in @var{output} (which will be a variable).
7745 The operation to be performed is specified by @var{code}, a tree code
7746 which will always be one of the following: @code{PLUS_EXPR},
7747 @code{MINUS_EXPR}, @code{MULT_EXPR}, @code{RDIV_EXPR},
7748 @code{MAX_EXPR}, @code{MIN_EXPR}.
7750 @cindex overflow while constant folding
7751 The expansion of this macro is responsible for checking for overflow.
7752 If overflow happens, the macro expansion should execute the statement
7753 @code{return 0;}, which indicates the inability to perform the
7754 arithmetic operation requested.
7756 @findex REAL_VALUE_NEGATE
7757 @item REAL_VALUE_NEGATE (@var{x})
7758 A macro for a C expression which returns the negative of the floating
7759 point value @var{x}. Both @var{x} and the value of the expression
7760 have type @code{REAL_VALUE_TYPE} and are in the target machine's
7761 floating point representation.
7763 There is no way for this macro to report overflow, since overflow
7764 can't happen in the negation operation.
7766 @findex REAL_VALUE_TRUNCATE
7767 @item REAL_VALUE_TRUNCATE (@var{mode}, @var{x})
7768 A macro for a C expression which converts the floating point value
7769 @var{x} to mode @var{mode}.
7771 Both @var{x} and the value of the expression are in the target machine's
7772 floating point representation and have type @code{REAL_VALUE_TYPE}.
7773 However, the value should have an appropriate bit pattern to be output
7774 properly as a floating constant whose precision accords with mode
7777 There is no way for this macro to report overflow.
7779 @findex REAL_VALUE_TO_INT
7780 @item REAL_VALUE_TO_INT (@var{low}, @var{high}, @var{x})
7781 A macro for a C expression which converts a floating point value
7782 @var{x} into a double-precision integer which is then stored into
7783 @var{low} and @var{high}, two variables of type @var{int}.
7785 @item REAL_VALUE_FROM_INT (@var{x}, @var{low}, @var{high}, @var{mode})
7786 @findex REAL_VALUE_FROM_INT
7787 A macro for a C expression which converts a double-precision integer
7788 found in @var{low} and @var{high}, two variables of type @var{int},
7789 into a floating point value which is then stored into @var{x}.
7790 The value is in the target machine's representation for mode @var{mode}
7791 and has the type @code{REAL_VALUE_TYPE}.
7794 @node Mode Switching
7795 @section Mode Switching Instructions
7796 @cindex mode switching
7797 The following macros control mode switching optimizations:
7800 @findex OPTIMIZE_MODE_SWITCHING
7801 @item OPTIMIZE_MODE_SWITCHING (@var{entity})
7802 Define this macro if the port needs extra instructions inserted for mode
7803 switching in an optimizing compilation.
7805 For an example, the SH4 can perform both single and double precision
7806 floating point operations, but to perform a single precision operation,
7807 the FPSCR PR bit has to be cleared, while for a double precision
7808 operation, this bit has to be set. Changing the PR bit requires a general
7809 purpose register as a scratch register, hence these FPSCR sets have to
7810 be inserted before reload, i.e.@: you can't put this into instruction emitting
7811 or @code{MACHINE_DEPENDENT_REORG}.
7813 You can have multiple entities that are mode-switched, and select at run time
7814 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
7815 return nonzero for any @var{entity} that needs mode-switching.
7816 If you define this macro, you also have to define
7817 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
7818 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
7819 @code{NORMAL_MODE} is optional.
7821 @findex NUM_MODES_FOR_MODE_SWITCHING
7822 @item NUM_MODES_FOR_MODE_SWITCHING
7823 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
7824 initializer for an array of integers. Each initializer element
7825 N refers to an entity that needs mode switching, and specifies the number
7826 of different modes that might need to be set for this entity.
7827 The position of the initializer in the initializer - starting counting at
7828 zero - determines the integer that is used to refer to the mode-switched
7830 In macros that take mode arguments / yield a mode result, modes are
7831 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
7832 switch is needed / supplied.
7835 @item MODE_NEEDED (@var{entity}, @var{insn})
7836 @var{entity} is an integer specifying a mode-switched entity. If
7837 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
7838 return an integer value not larger than the corresponding element in
7839 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
7840 be switched into prior to the execution of @var{insn}.
7843 @item NORMAL_MODE (@var{entity})
7844 If this macro is defined, it is evaluated for every @var{entity} that needs
7845 mode switching. It should evaluate to an integer, which is a mode that
7846 @var{entity} is assumed to be switched to at function entry and exit.
7848 @findex MODE_PRIORITY_TO_MODE
7849 @item MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
7850 This macro specifies the order in which modes for @var{entity} are processed.
7851 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
7852 lowest. The value of the macro should be an integer designating a mode
7853 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
7854 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
7855 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
7857 @findex EMIT_MODE_SET
7858 @item EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
7859 Generate one or more insns to set @var{entity} to @var{mode}.
7860 @var{hard_reg_live} is the set of hard registers live at the point where
7861 the insn(s) are to be inserted.
7864 @node Target Attributes
7865 @section Defining target-specific uses of @code{__attribute__}
7866 @cindex target attributes
7867 @cindex machine attributes
7868 @cindex attributes, target-specific
7870 Target-specific attributes may be defined for functions, data and types.
7871 These are described using the following target hooks; they also need to
7872 be documented in @file{extend.texi}.
7874 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
7875 If defined, this target hook points to an array of @samp{struct
7876 attribute_spec} (defined in @file{tree.h}) specifying the machine
7877 specific attributes for this target and some of the restrictions on the
7878 entities to which these attributes are applied and the arguments they
7882 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
7883 If defined, this target hook is a function which returns zero if the attributes on
7884 @var{type1} and @var{type2} are incompatible, one if they are compatible,
7885 and two if they are nearly compatible (which causes a warning to be
7886 generated). If this is not defined, machine-specific attributes are
7887 supposed always to be compatible.
7890 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
7891 If defined, this target hook is a function which assigns default attributes to
7892 newly defined @var{type}.
7895 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
7896 Define this target hook if the merging of type attributes needs special
7897 handling. If defined, the result is a list of the combined
7898 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
7899 that @code{comptypes} has already been called and returned 1. This
7900 function may call @code{merge_attributes} to handle machine-independent
7904 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
7905 Define this target hook if the merging of decl attributes needs special
7906 handling. If defined, the result is a list of the combined
7907 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
7908 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
7909 when this is needed are when one attribute overrides another, or when an
7910 attribute is nullified by a subsequent definition. This function may
7911 call @code{merge_attributes} to handle machine-independent merging.
7913 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
7914 If the only target-specific handling you require is @samp{dllimport} for
7915 Windows targets, you should define the macro
7916 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. This links in a function
7917 called @code{merge_dllimport_decl_attributes} which can then be defined
7918 as the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. This is done
7919 in @file{i386/cygwin.h} and @file{i386/i386.c}, for example.
7922 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
7923 Define this target hook if you want to be able to add attributes to a decl
7924 when it is being created. This is normally useful for back ends which
7925 wish to implement a pragma by using the attributes which correspond to
7926 the pragma's effect. The @var{node} argument is the decl which is being
7927 created. The @var{attr_ptr} argument is a pointer to the attribute list
7928 for this decl. The list itself should not be modified, since it may be
7929 shared with other decls, but attributes may be chained on the head of
7930 the list and @code{*@var{attr_ptr}} modified to point to the new
7931 attributes, or a copy of the list may be made if further changes are
7935 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (tree @var{fndecl})
7937 This target hook returns @code{true} if it is ok to inline @var{fndecl}
7938 into the current function, despite its having target-specific
7939 attributes, @code{false} otherwise. By default, if a function has a
7940 target specific attribute attached to it, it will not be inlined.
7944 @section Miscellaneous Parameters
7945 @cindex parameters, miscellaneous
7947 @c prevent bad page break with this line
7948 Here are several miscellaneous parameters.
7951 @item PREDICATE_CODES
7952 @findex PREDICATE_CODES
7953 Define this if you have defined special-purpose predicates in the file
7954 @file{@var{machine}.c}. This macro is called within an initializer of an
7955 array of structures. The first field in the structure is the name of a
7956 predicate and the second field is an array of rtl codes. For each
7957 predicate, list all rtl codes that can be in expressions matched by the
7958 predicate. The list should have a trailing comma. Here is an example
7959 of two entries in the list for a typical RISC machine:
7962 #define PREDICATE_CODES \
7963 @{"gen_reg_rtx_operand", @{SUBREG, REG@}@}, \
7964 @{"reg_or_short_cint_operand", @{SUBREG, REG, CONST_INT@}@},
7967 Defining this macro does not affect the generated code (however,
7968 incorrect definitions that omit an rtl code that may be matched by the
7969 predicate can cause the compiler to malfunction). Instead, it allows
7970 the table built by @file{genrecog} to be more compact and efficient,
7971 thus speeding up the compiler. The most important predicates to include
7972 in the list specified by this macro are those used in the most insn
7975 For each predicate function named in @code{PREDICATE_CODES}, a
7976 declaration will be generated in @file{insn-codes.h}.
7978 @item SPECIAL_MODE_PREDICATES
7979 @findex SPECIAL_MODE_PREDICATES
7980 Define this if you have special predicates that know special things
7981 about modes. Genrecog will warn about certain forms of
7982 @code{match_operand} without a mode; if the operand predicate is
7983 listed in @code{SPECIAL_MODE_PREDICATES}, the warning will be
7986 Here is an example from the IA-32 port (@code{ext_register_operand}
7987 specially checks for @code{HImode} or @code{SImode} in preparation
7988 for a byte extraction from @code{%ah} etc.).
7991 #define SPECIAL_MODE_PREDICATES \
7992 "ext_register_operand",
7995 @findex CASE_VECTOR_MODE
7996 @item CASE_VECTOR_MODE
7997 An alias for a machine mode name. This is the machine mode that
7998 elements of a jump-table should have.
8000 @findex CASE_VECTOR_SHORTEN_MODE
8001 @item CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
8002 Optional: return the preferred mode for an @code{addr_diff_vec}
8003 when the minimum and maximum offset are known. If you define this,
8004 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
8005 To make this work, you also have to define INSN_ALIGN and
8006 make the alignment for @code{addr_diff_vec} explicit.
8007 The @var{body} argument is provided so that the offset_unsigned and scale
8008 flags can be updated.
8010 @findex CASE_VECTOR_PC_RELATIVE
8011 @item CASE_VECTOR_PC_RELATIVE
8012 Define this macro to be a C expression to indicate when jump-tables
8013 should contain relative addresses. If jump-tables never contain
8014 relative addresses, then you need not define this macro.
8016 @findex CASE_DROPS_THROUGH
8017 @item CASE_DROPS_THROUGH
8018 Define this if control falls through a @code{case} insn when the index
8019 value is out of range. This means the specified default-label is
8020 actually ignored by the @code{case} insn proper.
8022 @findex CASE_VALUES_THRESHOLD
8023 @item CASE_VALUES_THRESHOLD
8024 Define this to be the smallest number of different values for which it
8025 is best to use a jump-table instead of a tree of conditional branches.
8026 The default is four for machines with a @code{casesi} instruction and
8027 five otherwise. This is best for most machines.
8029 @findex WORD_REGISTER_OPERATIONS
8030 @item WORD_REGISTER_OPERATIONS
8031 Define this macro if operations between registers with integral mode
8032 smaller than a word are always performed on the entire register.
8033 Most RISC machines have this property and most CISC machines do not.
8035 @findex LOAD_EXTEND_OP
8036 @item LOAD_EXTEND_OP (@var{mode})
8037 Define this macro to be a C expression indicating when insns that read
8038 memory in @var{mode}, an integral mode narrower than a word, set the
8039 bits outside of @var{mode} to be either the sign-extension or the
8040 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
8041 of @var{mode} for which the
8042 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
8043 @code{NIL} for other modes.
8045 This macro is not called with @var{mode} non-integral or with a width
8046 greater than or equal to @code{BITS_PER_WORD}, so you may return any
8047 value in this case. Do not define this macro if it would always return
8048 @code{NIL}. On machines where this macro is defined, you will normally
8049 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
8051 @findex SHORT_IMMEDIATES_SIGN_EXTEND
8052 @item SHORT_IMMEDIATES_SIGN_EXTEND
8053 Define this macro if loading short immediate values into registers sign
8056 @findex FIXUNS_TRUNC_LIKE_FIX_TRUNC
8057 @item FIXUNS_TRUNC_LIKE_FIX_TRUNC
8058 Define this macro if the same instructions that convert a floating
8059 point number to a signed fixed point number also convert validly to an
8064 The maximum number of bytes that a single instruction can move quickly
8065 between memory and registers or between two memory locations.
8067 @findex MAX_MOVE_MAX
8069 The maximum number of bytes that a single instruction can move quickly
8070 between memory and registers or between two memory locations. If this
8071 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
8072 constant value that is the largest value that @code{MOVE_MAX} can have
8075 @findex SHIFT_COUNT_TRUNCATED
8076 @item SHIFT_COUNT_TRUNCATED
8077 A C expression that is nonzero if on this machine the number of bits
8078 actually used for the count of a shift operation is equal to the number
8079 of bits needed to represent the size of the object being shifted. When
8080 this macro is nonzero, the compiler will assume that it is safe to omit
8081 a sign-extend, zero-extend, and certain bitwise `and' instructions that
8082 truncates the count of a shift operation. On machines that have
8083 instructions that act on bit-fields at variable positions, which may
8084 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
8085 also enables deletion of truncations of the values that serve as
8086 arguments to bit-field instructions.
8088 If both types of instructions truncate the count (for shifts) and
8089 position (for bit-field operations), or if no variable-position bit-field
8090 instructions exist, you should define this macro.
8092 However, on some machines, such as the 80386 and the 680x0, truncation
8093 only applies to shift operations and not the (real or pretended)
8094 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
8095 such machines. Instead, add patterns to the @file{md} file that include
8096 the implied truncation of the shift instructions.
8098 You need not define this macro if it would always have the value of zero.
8100 @findex TRULY_NOOP_TRUNCATION
8101 @item TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
8102 A C expression which is nonzero if on this machine it is safe to
8103 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
8104 bits (where @var{outprec} is smaller than @var{inprec}) by merely
8105 operating on it as if it had only @var{outprec} bits.
8107 On many machines, this expression can be 1.
8109 @c rearranged this, removed the phrase "it is reported that". this was
8110 @c to fix an overfull hbox. --mew 10feb93
8111 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
8112 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
8113 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
8114 such cases may improve things.
8116 @findex STORE_FLAG_VALUE
8117 @item STORE_FLAG_VALUE
8118 A C expression describing the value returned by a comparison operator
8119 with an integral mode and stored by a store-flag instruction
8120 (@samp{s@var{cond}}) when the condition is true. This description must
8121 apply to @emph{all} the @samp{s@var{cond}} patterns and all the
8122 comparison operators whose results have a @code{MODE_INT} mode.
8124 A value of 1 or @minus{}1 means that the instruction implementing the
8125 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
8126 and 0 when the comparison is false. Otherwise, the value indicates
8127 which bits of the result are guaranteed to be 1 when the comparison is
8128 true. This value is interpreted in the mode of the comparison
8129 operation, which is given by the mode of the first operand in the
8130 @samp{s@var{cond}} pattern. Either the low bit or the sign bit of
8131 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
8134 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
8135 generate code that depends only on the specified bits. It can also
8136 replace comparison operators with equivalent operations if they cause
8137 the required bits to be set, even if the remaining bits are undefined.
8138 For example, on a machine whose comparison operators return an
8139 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
8140 @samp{0x80000000}, saying that just the sign bit is relevant, the
8144 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
8151 (ashift:SI @var{x} (const_int @var{n}))
8155 where @var{n} is the appropriate shift count to move the bit being
8156 tested into the sign bit.
8158 There is no way to describe a machine that always sets the low-order bit
8159 for a true value, but does not guarantee the value of any other bits,
8160 but we do not know of any machine that has such an instruction. If you
8161 are trying to port GCC to such a machine, include an instruction to
8162 perform a logical-and of the result with 1 in the pattern for the
8163 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
8165 Often, a machine will have multiple instructions that obtain a value
8166 from a comparison (or the condition codes). Here are rules to guide the
8167 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
8172 Use the shortest sequence that yields a valid definition for
8173 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
8174 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
8175 comparison operators to do so because there may be opportunities to
8176 combine the normalization with other operations.
8179 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
8180 slightly preferred on machines with expensive jumps and 1 preferred on
8184 As a second choice, choose a value of @samp{0x80000001} if instructions
8185 exist that set both the sign and low-order bits but do not define the
8189 Otherwise, use a value of @samp{0x80000000}.
8192 Many machines can produce both the value chosen for
8193 @code{STORE_FLAG_VALUE} and its negation in the same number of
8194 instructions. On those machines, you should also define a pattern for
8195 those cases, e.g., one matching
8198 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
8201 Some machines can also perform @code{and} or @code{plus} operations on
8202 condition code values with less instructions than the corresponding
8203 @samp{s@var{cond}} insn followed by @code{and} or @code{plus}. On those
8204 machines, define the appropriate patterns. Use the names @code{incscc}
8205 and @code{decscc}, respectively, for the patterns which perform
8206 @code{plus} or @code{minus} operations on condition code values. See
8207 @file{rs6000.md} for some examples. The GNU Superoptizer can be used to
8208 find such instruction sequences on other machines.
8210 You need not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
8213 @findex FLOAT_STORE_FLAG_VALUE
8214 @item FLOAT_STORE_FLAG_VALUE (@var{mode})
8215 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
8216 returned when comparison operators with floating-point results are true.
8217 Define this macro on machine that have comparison operations that return
8218 floating-point values. If there are no such operations, do not define
8223 An alias for the machine mode for pointers. On most machines, define
8224 this to be the integer mode corresponding to the width of a hardware
8225 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
8226 On some machines you must define this to be one of the partial integer
8227 modes, such as @code{PSImode}.
8229 The width of @code{Pmode} must be at least as large as the value of
8230 @code{POINTER_SIZE}. If it is not equal, you must define the macro
8231 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
8234 @findex FUNCTION_MODE
8236 An alias for the machine mode used for memory references to functions
8237 being called, in @code{call} RTL expressions. On most machines this
8238 should be @code{QImode}.
8240 @findex INTEGRATE_THRESHOLD
8241 @item INTEGRATE_THRESHOLD (@var{decl})
8242 A C expression for the maximum number of instructions above which the
8243 function @var{decl} should not be inlined. @var{decl} is a
8244 @code{FUNCTION_DECL} node.
8246 The default definition of this macro is 64 plus 8 times the number of
8247 arguments that the function accepts. Some people think a larger
8248 threshold should be used on RISC machines.
8250 @findex STDC_0_IN_SYSTEM_HEADERS
8251 @item STDC_0_IN_SYSTEM_HEADERS
8252 In normal operation, the preprocessor expands @code{__STDC__} to the
8253 constant 1, to signify that GCC conforms to ISO Standard C@. On some
8254 hosts, like Solaris, the system compiler uses a different convention,
8255 where @code{__STDC__} is normally 0, but is 1 if the user specifies
8256 strict conformance to the C Standard.
8258 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
8259 convention when processing system header files, but when processing user
8260 files @code{__STDC__} will always expand to 1.
8262 @findex SCCS_DIRECTIVE
8263 @item SCCS_DIRECTIVE
8264 Define this if the preprocessor should ignore @code{#sccs} directives
8265 and print no error message.
8267 @findex NO_IMPLICIT_EXTERN_C
8268 @item NO_IMPLICIT_EXTERN_C
8269 Define this macro if the system header files support C++ as well as C@.
8270 This macro inhibits the usual method of using system header files in
8271 C++, which is to pretend that the file's contents are enclosed in
8272 @samp{extern "C" @{@dots{}@}}.
8274 @findex HANDLE_PRAGMA
8275 @item HANDLE_PRAGMA (@var{getc}, @var{ungetc}, @var{name})
8276 This macro is no longer supported. You must use
8277 @code{REGISTER_TARGET_PRAGMAS} instead.
8279 @findex REGISTER_TARGET_PRAGMAS
8282 @item REGISTER_TARGET_PRAGMAS (@var{pfile})
8283 Define this macro if you want to implement any target-specific pragmas.
8284 If defined, it is a C expression which makes a series of calls to
8285 @code{cpp_register_pragma} for each pragma, with @var{pfile} passed as
8286 the first argument to to these functions. The macro may also do any
8287 setup required for the pragmas.
8289 The primary reason to define this macro is to provide compatibility with
8290 other compilers for the same target. In general, we discourage
8291 definition of target-specific pragmas for GCC@.
8293 If the pragma can be implemented by attributes then you should consider
8294 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
8296 Preprocessor macros that appear on pragma lines are not expanded. All
8297 @samp{#pragma} directives that do not match any registered pragma are
8298 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
8300 @deftypefun void cpp_register_pragma (cpp_reader *@var{pfile}, const char *@var{space}, const char *@var{name}, void (*@var{callback}) (cpp_reader *))
8302 Each call to @code{cpp_register_pragma} establishes one pragma. The
8303 @var{callback} routine will be called when the preprocessor encounters a
8307 #pragma [@var{space}] @var{name} @dots{}
8310 @var{space} is the case-sensitive namespace of the pragma, or
8311 @code{NULL} to put the pragma in the global namespace. The callback
8312 routine receives @var{pfile} as its first argument, which can be passed
8313 on to cpplib's functions if necessary. You can lex tokens after the
8314 @var{name} by calling @code{c_lex}. Tokens that are not read by the
8315 callback will be silently ignored. The end of the line is indicated by
8316 a token of type @code{CPP_EOF}.
8318 For an example use of this routine, see @file{c4x.h} and the callback
8319 routines defined in @file{c4x-c.c}.
8321 Note that the use of @code{c_lex} is specific to the C and C++
8322 compilers. It will not work in the Java or Fortran compilers, or any
8323 other language compilers for that matter. Thus if @code{c_lex} is going
8324 to be called from target-specific code, it must only be done so when
8325 building the C and C++ compilers. This can be done by defining the
8326 variables @code{c_target_objs} and @code{cxx_target_objs} in the
8327 target entry in the @file{config.gcc} file. These variables should name
8328 the target-specific, language-specific object file which contains the
8329 code that uses @code{c_lex}. Note it will also be necessary to add a
8330 rule to the makefile fragment pointed to by @code{tmake_file} that shows
8331 how to build this object file.
8334 @findex HANDLE_SYSV_PRAGMA
8337 @item HANDLE_SYSV_PRAGMA
8338 Define this macro (to a value of 1) if you want the System V style
8339 pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name>
8340 [=<value>]} to be supported by gcc.
8342 The pack pragma specifies the maximum alignment (in bytes) of fields
8343 within a structure, in much the same way as the @samp{__aligned__} and
8344 @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets
8345 the behavior to the default.
8347 The weak pragma only works if @code{SUPPORTS_WEAK} and
8348 @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation
8349 of specifically named weak labels, optionally with a value.
8351 @findex HANDLE_PRAGMA_PACK_PUSH_POP
8354 @item HANDLE_PRAGMA_PACK_PUSH_POP
8355 Define this macro (to a value of 1) if you want to support the Win32
8356 style pragmas @samp{#pragma pack(push,@var{n})} and @samp{#pragma
8357 pack(pop)}. The @samp{pack(push,@var{n})} pragma specifies the maximum alignment
8358 (in bytes) of fields within a structure, in much the same way as the
8359 @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A
8360 pack value of zero resets the behavior to the default. Successive
8361 invocations of this pragma cause the previous values to be stacked, so
8362 that invocations of @samp{#pragma pack(pop)} will return to the previous
8365 @findex DOLLARS_IN_IDENTIFIERS
8366 @item DOLLARS_IN_IDENTIFIERS
8367 Define this macro to control use of the character @samp{$} in identifier
8368 names. 0 means @samp{$} is not allowed by default; 1 means it is allowed.
8369 1 is the default; there is no need to define this macro in that case.
8370 This macro controls the compiler proper; it does not affect the preprocessor.
8372 @findex NO_DOLLAR_IN_LABEL
8373 @item NO_DOLLAR_IN_LABEL
8374 Define this macro if the assembler does not accept the character
8375 @samp{$} in label names. By default constructors and destructors in
8376 G++ have @samp{$} in the identifiers. If this macro is defined,
8377 @samp{.} is used instead.
8379 @findex NO_DOT_IN_LABEL
8380 @item NO_DOT_IN_LABEL
8381 Define this macro if the assembler does not accept the character
8382 @samp{.} in label names. By default constructors and destructors in G++
8383 have names that use @samp{.}. If this macro is defined, these names
8384 are rewritten to avoid @samp{.}.
8386 @findex DEFAULT_MAIN_RETURN
8387 @item DEFAULT_MAIN_RETURN
8388 Define this macro if the target system expects every program's @code{main}
8389 function to return a standard ``success'' value by default (if no other
8390 value is explicitly returned).
8392 The definition should be a C statement (sans semicolon) to generate the
8393 appropriate rtl instructions. It is used only when compiling the end of
8398 Define this if the target system lacks the function @code{atexit}
8399 from the ISO C standard. If this macro is defined, a default definition
8400 will be provided to support C++. If @code{ON_EXIT} is not defined,
8401 a default @code{exit} function will also be provided.
8405 Define this macro if the target has another way to implement atexit
8406 functionality without replacing @code{exit}. For instance, SunOS 4 has
8407 a similar @code{on_exit} library function.
8409 The definition should be a functional macro which can be used just like
8410 the @code{atexit} function.
8414 Define this if your @code{exit} function needs to do something
8415 besides calling an external function @code{_cleanup} before
8416 terminating with @code{_exit}. The @code{EXIT_BODY} macro is
8417 only needed if @code{NEED_ATEXIT} is defined and @code{ON_EXIT} is not
8420 @findex INSN_SETS_ARE_DELAYED
8421 @item INSN_SETS_ARE_DELAYED (@var{insn})
8422 Define this macro as a C expression that is nonzero if it is safe for the
8423 delay slot scheduler to place instructions in the delay slot of @var{insn},
8424 even if they appear to use a resource set or clobbered in @var{insn}.
8425 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
8426 every @code{call_insn} has this behavior. On machines where some @code{insn}
8427 or @code{jump_insn} is really a function call and hence has this behavior,
8428 you should define this macro.
8430 You need not define this macro if it would always return zero.
8432 @findex INSN_REFERENCES_ARE_DELAYED
8433 @item INSN_REFERENCES_ARE_DELAYED (@var{insn})
8434 Define this macro as a C expression that is nonzero if it is safe for the
8435 delay slot scheduler to place instructions in the delay slot of @var{insn},
8436 even if they appear to set or clobber a resource referenced in @var{insn}.
8437 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
8438 some @code{insn} or @code{jump_insn} is really a function call and its operands
8439 are registers whose use is actually in the subroutine it calls, you should
8440 define this macro. Doing so allows the delay slot scheduler to move
8441 instructions which copy arguments into the argument registers into the delay
8444 You need not define this macro if it would always return zero.
8446 @findex MACHINE_DEPENDENT_REORG
8447 @item MACHINE_DEPENDENT_REORG (@var{insn})
8448 In rare cases, correct code generation requires extra machine
8449 dependent processing between the second jump optimization pass and
8450 delayed branch scheduling. On those machines, define this macro as a C
8451 statement to act on the code starting at @var{insn}.
8453 @findex MULTIPLE_SYMBOL_SPACES
8454 @item MULTIPLE_SYMBOL_SPACES
8455 Define this macro if in some cases global symbols from one translation
8456 unit may not be bound to undefined symbols in another translation unit
8457 without user intervention. For instance, under Microsoft Windows
8458 symbols must be explicitly imported from shared libraries (DLLs).
8460 @findex MD_ASM_CLOBBERS
8461 @item MD_ASM_CLOBBERS (@var{clobbers})
8462 A C statement that adds to @var{clobbers} @code{STRING_CST} trees for
8463 any hard regs the port wishes to automatically clobber for all asms.
8465 @findex MAX_INTEGER_COMPUTATION_MODE
8466 @item MAX_INTEGER_COMPUTATION_MODE
8467 Define this to the largest integer machine mode which can be used for
8468 operations other than load, store and copy operations.
8470 You need only define this macro if the target holds values larger than
8471 @code{word_mode} in general purpose registers. Most targets should not define
8474 @findex MATH_LIBRARY
8476 Define this macro as a C string constant for the linker argument to link
8477 in the system math library, or @samp{""} if the target does not have a
8478 separate math library.
8480 You need only define this macro if the default of @samp{"-lm"} is wrong.
8482 @findex LIBRARY_PATH_ENV
8483 @item LIBRARY_PATH_ENV
8484 Define this macro as a C string constant for the environment variable that
8485 specifies where the linker should look for libraries.
8487 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
8490 @findex TARGET_HAS_F_SETLKW
8491 @item TARGET_HAS_F_SETLKW
8492 Define this macro if the target supports file locking with fcntl / F_SETLKW@.
8493 Note that this functionality is part of POSIX@.
8494 Defining @code{TARGET_HAS_F_SETLKW} will enable the test coverage code
8495 to use file locking when exiting a program, which avoids race conditions
8496 if the program has forked.
8498 @findex MAX_CONDITIONAL_EXECUTE
8499 @item MAX_CONDITIONAL_EXECUTE
8501 A C expression for the maximum number of instructions to execute via
8502 conditional execution instructions instead of a branch. A value of
8503 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
8504 1 if it does use cc0.
8506 @findex IFCVT_MODIFY_TESTS
8507 @item IFCVT_MODIFY_TESTS
8508 A C expression to modify the tests in @code{TRUE_EXPR}, and
8509 @code{FALSE_EXPR} for use in converting insns in @code{TEST_BB},
8510 @code{THEN_BB}, @code{ELSE_BB}, and @code{JOIN_BB} basic blocks to
8511 conditional execution. Set either @code{TRUE_EXPR} or @code{FALSE_EXPR}
8512 to a null pointer if the tests cannot be converted.
8514 @findex IFCVT_MODIFY_INSN
8515 @item IFCVT_MODIFY_INSN
8516 A C expression to modify the @code{PATTERN} of an @code{INSN} that is to
8517 be converted to conditional execution format.
8519 @findex IFCVT_MODIFY_FINAL
8520 @item IFCVT_MODIFY_FINAL
8521 A C expression to perform any final machine dependent modifications in
8522 converting code to conditional execution in the basic blocks
8523 @code{TEST_BB}, @code{THEN_BB}, @code{ELSE_BB}, and @code{JOIN_BB}.
8525 @findex IFCVT_MODIFY_CANCEL
8526 @item IFCVT_MODIFY_CANCEL
8527 A C expression to cancel any machine dependent modifications in
8528 converting code to conditional execution in the basic blocks
8529 @code{TEST_BB}, @code{THEN_BB}, @code{ELSE_BB}, and @code{JOIN_BB}.
8532 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS ()
8533 Define this hook if you have any machine-specific built-in functions
8534 that need to be defined. It should be a function that performs the
8537 Machine specific built-in functions can be useful to expand special machine
8538 instructions that would otherwise not normally be generated because
8539 they have no equivalent in the source language (for example, SIMD vector
8540 instructions or prefetch instructions).
8542 To create a built-in function, call the function @code{builtin_function}
8543 which is defined by the language front end. You can use any type nodes set
8544 up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2};
8545 only language front ends that use those two functions will call
8546 @samp{TARGET_INIT_BUILTINS}.
8549 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore})
8551 Expand a call to a machine specific built-in function that was set up by
8552 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
8553 function call; the result should go to @var{target} if that is
8554 convenient, and have mode @var{mode} if that is convenient.
8555 @var{subtarget} may be used as the target for computing one of
8556 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
8557 ignored. This function should return the result of the call to the
8562 @findex MD_CAN_REDIRECT_BRANCH
8563 @item MD_CAN_REDIRECT_BRANCH(@var{branch1}, @var{branch2})
8565 Take a branch insn in @var{branch1} and another in @var{branch2}.
8566 Return true if redirecting @var{branch1} to the destination of
8567 @var{branch2} is possible.
8569 On some targets, branches may have a limited range. Optimizing the
8570 filling of delay slots can result in branches being redirected, and this
8571 may in turn cause a branch offset to overflow.
8573 @findex ALLOCATE_INITIAL_VALUE
8574 @item ALLOCATE_INITIAL_VALUE(@var{hard_reg})
8576 When the initial value of a hard register has been copied in a pseudo
8577 register, it is often not necessary to actually allocate another register
8578 to this pseudo register, because the original hard register or a stack slot
8579 it has been saved into can be used. @code{ALLOCATE_INITIAL_VALUE}, if
8580 defined, is called at the start of register allocation once for each
8581 hard register that had its initial value copied by using
8582 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
8583 Possible values are @code{NULL_RTX}, if you don't want
8584 to do any special allocation, a @code{REG} rtx---that would typically be
8585 the hard register itself, if it is known not to be clobbered---or a
8587 If you are returning a @code{MEM}, this is only a hint for the allocator;
8588 it might decide to use another register anyways.
8589 You may use @code{current_function_leaf_function} in the definition of the
8590 macro, functions that use @code{REG_N_SETS}, to determine if the hard
8591 register in question will not be clobbered.
8593 @findex TARGET_OBJECT_SUFFIX
8594 @item TARGET_OBJECT_SUFFIX
8595 Define this macro to be a C string representing the suffix for object
8596 files on your target machine. If you do not define this macro, GCC will
8597 use @samp{.o} as the suffix for object files.
8599 @findex TARGET_EXECUTABLE_SUFFIX
8600 @item TARGET_EXECUTABLE_SUFFIX
8601 Define this macro to be a C string representing the suffix to be
8602 automatically added to executable files on your target machine. If you
8603 do not define this macro, GCC will use the null string as the suffix for
8606 @findex COLLECT_EXPORT_LIST
8607 @item COLLECT_EXPORT_LIST
8608 If defined, @code{collect2} will scan the individual object files
8609 specified on its command line and create an export list for the linker.
8610 Define this macro for systems like AIX, where the linker discards
8611 object files that are not referenced from @code{main} and uses export