1 @c Copyright (C) 1988,1989,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,
2 @c 2002, 2003, 2004 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 * Floating Point:: 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 * MIPS Coprocessors:: MIPS coprocessor support and how to customize it.
53 * PCH Target:: Validity checking for precompiled headers.
54 * C++ ABI:: Controlling C++ ABI changes.
55 * Misc:: Everything else.
58 @node Target Structure
59 @section The Global @code{targetm} Variable
61 @cindex target functions
63 @deftypevar {struct gcc_target} targetm
64 The target @file{.c} file must define the global @code{targetm} variable
65 which contains pointers to functions and data relating to the target
66 machine. The variable is declared in @file{target.h};
67 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
68 used to initialize the variable, and macros for the default initializers
69 for elements of the structure. The @file{.c} file should override those
70 macros for which the default definition is inappropriate. For example:
73 #include "target-def.h"
75 /* @r{Initialize the GCC target structure.} */
77 #undef TARGET_COMP_TYPE_ATTRIBUTES
78 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
80 struct gcc_target targetm = TARGET_INITIALIZER;
84 Where a macro should be defined in the @file{.c} file in this manner to
85 form part of the @code{targetm} structure, it is documented below as a
86 ``Target Hook'' with a prototype. Many macros will change in future
87 from being defined in the @file{.h} file to being part of the
88 @code{targetm} structure.
91 @section Controlling the Compilation Driver, @file{gcc}
93 @cindex controlling the compilation driver
95 @c prevent bad page break with this line
96 You can control the compilation driver.
98 @defmac SWITCH_TAKES_ARG (@var{char})
99 A C expression which determines whether the option @option{-@var{char}}
100 takes arguments. The value should be the number of arguments that
101 option takes--zero, for many options.
103 By default, this macro is defined as
104 @code{DEFAULT_SWITCH_TAKES_ARG}, which handles the standard options
105 properly. You need not define @code{SWITCH_TAKES_ARG} unless you
106 wish to add additional options which take arguments. Any redefinition
107 should call @code{DEFAULT_SWITCH_TAKES_ARG} and then check for
111 @defmac WORD_SWITCH_TAKES_ARG (@var{name})
112 A C expression which determines whether the option @option{-@var{name}}
113 takes arguments. The value should be the number of arguments that
114 option takes--zero, for many options. This macro rather than
115 @code{SWITCH_TAKES_ARG} is used for multi-character option names.
117 By default, this macro is defined as
118 @code{DEFAULT_WORD_SWITCH_TAKES_ARG}, which handles the standard options
119 properly. You need not define @code{WORD_SWITCH_TAKES_ARG} unless you
120 wish to add additional options which take arguments. Any redefinition
121 should call @code{DEFAULT_WORD_SWITCH_TAKES_ARG} and then check for
125 @defmac SWITCH_CURTAILS_COMPILATION (@var{char})
126 A C expression which determines whether the option @option{-@var{char}}
127 stops compilation before the generation of an executable. The value is
128 boolean, nonzero if the option does stop an executable from being
129 generated, zero otherwise.
131 By default, this macro is defined as
132 @code{DEFAULT_SWITCH_CURTAILS_COMPILATION}, which handles the standard
133 options properly. You need not define
134 @code{SWITCH_CURTAILS_COMPILATION} unless you wish to add additional
135 options which affect the generation of an executable. Any redefinition
136 should call @code{DEFAULT_SWITCH_CURTAILS_COMPILATION} and then check
137 for additional options.
140 @defmac SWITCHES_NEED_SPACES
141 A string-valued C expression which enumerates the options for which
142 the linker needs a space between the option and its argument.
144 If this macro is not defined, the default value is @code{""}.
147 @defmac TARGET_OPTION_TRANSLATE_TABLE
148 If defined, a list of pairs of strings, the first of which is a
149 potential command line target to the @file{gcc} driver program, and the
150 second of which is a space-separated (tabs and other whitespace are not
151 supported) list of options with which to replace the first option. The
152 target defining this list is responsible for assuring that the results
153 are valid. Replacement options may not be the @code{--opt} style, they
154 must be the @code{-opt} style. It is the intention of this macro to
155 provide a mechanism for substitution that affects the multilibs chosen,
156 such as one option that enables many options, some of which select
157 multilibs. Example nonsensical definition, where @code{-malt-abi},
158 @code{-EB}, and @code{-mspoo} cause different multilibs to be chosen:
161 #define TARGET_OPTION_TRANSLATE_TABLE \
162 @{ "-fast", "-march=fast-foo -malt-abi -I/usr/fast-foo" @}, \
163 @{ "-compat", "-EB -malign=4 -mspoo" @}
167 @defmac DRIVER_SELF_SPECS
168 A list of specs for the driver itself. It should be a suitable
169 initializer for an array of strings, with no surrounding braces.
171 The driver applies these specs to its own command line between loading
172 default @file{specs} files (but not command-line specified ones) and
173 choosing the multilib directory or running any subcommands. It
174 applies them in the order given, so each spec can depend on the
175 options added by earlier ones. It is also possible to remove options
176 using @samp{%<@var{option}} in the usual way.
178 This macro can be useful when a port has several interdependent target
179 options. It provides a way of standardizing the command line so
180 that the other specs are easier to write.
182 Do not define this macro if it does not need to do anything.
185 @defmac OPTION_DEFAULT_SPECS
186 A list of specs used to support configure-time default options (i.e.@:
187 @option{--with} options) in the driver. It should be a suitable initializer
188 for an array of structures, each containing two strings, without the
189 outermost pair of surrounding braces.
191 The first item in the pair is the name of the default. This must match
192 the code in @file{config.gcc} for the target. The second item is a spec
193 to apply if a default with this name was specified. The string
194 @samp{%(VALUE)} in the spec will be replaced by the value of the default
195 everywhere it occurs.
197 The driver will apply these specs to its own command line between loading
198 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
199 the same mechanism as @code{DRIVER_SELF_SPECS}.
201 Do not define this macro if it does not need to do anything.
205 A C string constant that tells the GCC driver program options to
206 pass to CPP@. It can also specify how to translate options you
207 give to GCC into options for GCC to pass to the CPP@.
209 Do not define this macro if it does not need to do anything.
212 @defmac CPLUSPLUS_CPP_SPEC
213 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
214 than C@. If you do not define this macro, then the value of
215 @code{CPP_SPEC} (if any) will be used instead.
219 A C string constant that tells the GCC driver program options to
220 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
222 It can also specify how to translate options you give to GCC into options
223 for GCC to pass to front ends.
225 Do not define this macro if it does not need to do anything.
229 A C string constant that tells the GCC driver program options to
230 pass to @code{cc1plus}. It can also specify how to translate options you
231 give to GCC into options for GCC to pass to the @code{cc1plus}.
233 Do not define this macro if it does not need to do anything.
234 Note that everything defined in CC1_SPEC is already passed to
235 @code{cc1plus} so there is no need to duplicate the contents of
236 CC1_SPEC in CC1PLUS_SPEC@.
240 A C string constant that tells the GCC driver program options to
241 pass to the assembler. It can also specify how to translate options
242 you give to GCC into options for GCC to pass to the assembler.
243 See the file @file{sun3.h} for an example of this.
245 Do not define this macro if it does not need to do anything.
248 @defmac ASM_FINAL_SPEC
249 A C string constant that tells the GCC driver program how to
250 run any programs which cleanup after the normal assembler.
251 Normally, this is not needed. See the file @file{mips.h} for
254 Do not define this macro if it does not need to do anything.
257 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
258 Define this macro, with no value, if the driver should give the assembler
259 an argument consisting of a single dash, @option{-}, to instruct it to
260 read from its standard input (which will be a pipe connected to the
261 output of the compiler proper). This argument is given after any
262 @option{-o} option specifying the name of the output file.
264 If you do not define this macro, the assembler is assumed to read its
265 standard input if given no non-option arguments. If your assembler
266 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
267 see @file{mips.h} for instance.
271 A C string constant that tells the GCC driver program options to
272 pass to the linker. It can also specify how to translate options you
273 give to GCC into options for GCC to pass to the linker.
275 Do not define this macro if it does not need to do anything.
279 Another C string constant used much like @code{LINK_SPEC}. The difference
280 between the two is that @code{LIB_SPEC} is used at the end of the
281 command given to the linker.
283 If this macro is not defined, a default is provided that
284 loads the standard C library from the usual place. See @file{gcc.c}.
288 Another C string constant that tells the GCC driver program
289 how and when to place a reference to @file{libgcc.a} into the
290 linker command line. This constant is placed both before and after
291 the value of @code{LIB_SPEC}.
293 If this macro is not defined, the GCC driver provides a default that
294 passes the string @option{-lgcc} to the linker.
297 @defmac REAL_LIBGCC_SPEC
298 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
299 @code{LIBGCC_SPEC} is not directly used by the driver program but is
300 instead modified to refer to different versions of @file{libgcc.a}
301 depending on the values of the command line flags @code{-static},
302 @code{-shared}, @code{-static-libgcc}, and @code{-shared-libgcc}. On
303 targets where these modifications are inappropriate, define
304 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
305 driver how to place a reference to @file{libgcc} on the link command
306 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
309 @defmac STARTFILE_SPEC
310 Another C string constant used much like @code{LINK_SPEC}. The
311 difference between the two is that @code{STARTFILE_SPEC} is used at
312 the very beginning of the command given to the linker.
314 If this macro is not defined, a default is provided that loads the
315 standard C startup file from the usual place. See @file{gcc.c}.
319 Another C string constant used much like @code{LINK_SPEC}. The
320 difference between the two is that @code{ENDFILE_SPEC} is used at
321 the very end of the command given to the linker.
323 Do not define this macro if it does not need to do anything.
326 @defmac THREAD_MODEL_SPEC
327 GCC @code{-v} will print the thread model GCC was configured to use.
328 However, this doesn't work on platforms that are multilibbed on thread
329 models, such as AIX 4.3. On such platforms, define
330 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
331 blanks that names one of the recognized thread models. @code{%*}, the
332 default value of this macro, will expand to the value of
333 @code{thread_file} set in @file{config.gcc}.
336 @defmac SYSROOT_SUFFIX_SPEC
337 Define this macro to add a suffix to the target sysroot when GCC is
338 configured with a sysroot. This will cause GCC to search for usr/lib,
339 et al, within sysroot+suffix.
342 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
343 Define this macro to add a headers_suffix to the target sysroot when
344 GCC is configured with a sysroot. This will cause GCC to pass the
345 updated sysroot+headers_suffix to CPP, causing it to search for
346 usr/include, et al, within sysroot+headers_suffix.
350 Define this macro to provide additional specifications to put in the
351 @file{specs} file that can be used in various specifications like
354 The definition should be an initializer for an array of structures,
355 containing a string constant, that defines the specification name, and a
356 string constant that provides the specification.
358 Do not define this macro if it does not need to do anything.
360 @code{EXTRA_SPECS} is useful when an architecture contains several
361 related targets, which have various @code{@dots{}_SPECS} which are similar
362 to each other, and the maintainer would like one central place to keep
365 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
366 define either @code{_CALL_SYSV} when the System V calling sequence is
367 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
370 The @file{config/rs6000/rs6000.h} target file defines:
373 #define EXTRA_SPECS \
374 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
376 #define CPP_SYS_DEFAULT ""
379 The @file{config/rs6000/sysv.h} target file defines:
383 "%@{posix: -D_POSIX_SOURCE @} \
384 %@{mcall-sysv: -D_CALL_SYSV @} \
385 %@{!mcall-sysv: %(cpp_sysv_default) @} \
386 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
388 #undef CPP_SYSV_DEFAULT
389 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
392 while the @file{config/rs6000/eabiaix.h} target file defines
393 @code{CPP_SYSV_DEFAULT} as:
396 #undef CPP_SYSV_DEFAULT
397 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
401 @defmac LINK_LIBGCC_SPECIAL
402 Define this macro if the driver program should find the library
403 @file{libgcc.a} itself and should not pass @option{-L} options to the
404 linker. If you do not define this macro, the driver program will pass
405 the argument @option{-lgcc} to tell the linker to do the search and will
406 pass @option{-L} options to it.
409 @defmac LINK_LIBGCC_SPECIAL_1
410 Define this macro if the driver program should find the library
411 @file{libgcc.a}. If you do not define this macro, the driver program will pass
412 the argument @option{-lgcc} to tell the linker to do the search.
413 This macro is similar to @code{LINK_LIBGCC_SPECIAL}, except that it does
414 not affect @option{-L} options.
417 @defmac LINK_GCC_C_SEQUENCE_SPEC
418 The sequence in which libgcc and libc are specified to the linker.
419 By default this is @code{%G %L %G}.
422 @defmac LINK_COMMAND_SPEC
423 A C string constant giving the complete command line need to execute the
424 linker. When you do this, you will need to update your port each time a
425 change is made to the link command line within @file{gcc.c}. Therefore,
426 define this macro only if you need to completely redefine the command
427 line for invoking the linker and there is no other way to accomplish
428 the effect you need. Overriding this macro may be avoidable by overriding
429 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
432 @defmac LINK_ELIMINATE_DUPLICATE_LDIRECTORIES
433 A nonzero value causes @command{collect2} to remove duplicate @option{-L@var{directory}} search
434 directories from linking commands. Do not give it a nonzero value if
435 removing duplicate search directories changes the linker's semantics.
438 @defmac MULTILIB_DEFAULTS
439 Define this macro as a C expression for the initializer of an array of
440 string to tell the driver program which options are defaults for this
441 target and thus do not need to be handled specially when using
442 @code{MULTILIB_OPTIONS}.
444 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
445 the target makefile fragment or if none of the options listed in
446 @code{MULTILIB_OPTIONS} are set by default.
447 @xref{Target Fragment}.
450 @defmac RELATIVE_PREFIX_NOT_LINKDIR
451 Define this macro to tell @command{gcc} that it should only translate
452 a @option{-B} prefix into a @option{-L} linker option if the prefix
453 indicates an absolute file name.
456 @defmac MD_EXEC_PREFIX
457 If defined, this macro is an additional prefix to try after
458 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
459 when the @option{-b} option is used, or the compiler is built as a cross
460 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
461 to the list of directories used to find the assembler in @file{configure.in}.
464 @defmac STANDARD_STARTFILE_PREFIX
465 Define this macro as a C string constant if you wish to override the
466 standard choice of @code{libdir} as the default prefix to
467 try when searching for startup files such as @file{crt0.o}.
468 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
469 is built as a cross compiler.
472 @defmac STANDARD_STARTFILE_PREFIX_1
473 Define this macro as a C string constant if you wish to override the
474 standard choice of @code{/lib} as a prefix to try after the default prefix
475 when searching for startup files such as @file{crt0.o}.
476 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
477 is built as a cross compiler.
480 @defmac STANDARD_STARTFILE_PREFIX_2
481 Define this macro as a C string constant if you wish to override the
482 standard choice of @code{/lib} as yet another prefix to try after the
483 default prefix when searching for startup files such as @file{crt0.o}.
484 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
485 is built as a cross compiler.
488 @defmac MD_STARTFILE_PREFIX
489 If defined, this macro supplies an additional prefix to try after the
490 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
491 @option{-b} option is used, or when the compiler is built as a cross
495 @defmac MD_STARTFILE_PREFIX_1
496 If defined, this macro supplies yet another prefix to try after the
497 standard prefixes. It is not searched when the @option{-b} option is
498 used, or when the compiler is built as a cross compiler.
501 @defmac INIT_ENVIRONMENT
502 Define this macro as a C string constant if you wish to set environment
503 variables for programs called by the driver, such as the assembler and
504 loader. The driver passes the value of this macro to @code{putenv} to
505 initialize the necessary environment variables.
508 @defmac LOCAL_INCLUDE_DIR
509 Define this macro as a C string constant if you wish to override the
510 standard choice of @file{/usr/local/include} as the default prefix to
511 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
512 comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
514 Cross compilers do not search either @file{/usr/local/include} or its
518 @defmac MODIFY_TARGET_NAME
519 Define this macro if you wish to define command-line switches that
520 modify the default target name.
522 For each switch, you can include a string to be appended to the first
523 part of the configuration name or a string to be deleted from the
524 configuration name, if present. The definition should be an initializer
525 for an array of structures. Each array element should have three
526 elements: the switch name (a string constant, including the initial
527 dash), one of the enumeration codes @code{ADD} or @code{DELETE} to
528 indicate whether the string should be inserted or deleted, and the string
529 to be inserted or deleted (a string constant).
531 For example, on a machine where @samp{64} at the end of the
532 configuration name denotes a 64-bit target and you want the @option{-32}
533 and @option{-64} switches to select between 32- and 64-bit targets, you would
537 #define MODIFY_TARGET_NAME \
538 @{ @{ "-32", DELETE, "64"@}, \
539 @{"-64", ADD, "64"@}@}
543 @defmac SYSTEM_INCLUDE_DIR
544 Define this macro as a C string constant if you wish to specify a
545 system-specific directory to search for header files before the standard
546 directory. @code{SYSTEM_INCLUDE_DIR} comes before
547 @code{STANDARD_INCLUDE_DIR} in the search order.
549 Cross compilers do not use this macro and do not search the directory
553 @defmac STANDARD_INCLUDE_DIR
554 Define this macro as a C string constant if you wish to override the
555 standard choice of @file{/usr/include} as the default prefix to
556 try when searching for header files.
558 Cross compilers ignore this macro and do not search either
559 @file{/usr/include} or its replacement.
562 @defmac STANDARD_INCLUDE_COMPONENT
563 The ``component'' corresponding to @code{STANDARD_INCLUDE_DIR}.
564 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
565 If you do not define this macro, no component is used.
568 @defmac INCLUDE_DEFAULTS
569 Define this macro if you wish to override the entire default search path
570 for include files. For a native compiler, the default search path
571 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
572 @code{SYSTEM_INCLUDE_DIR}, @code{GPLUSPLUS_INCLUDE_DIR}, and
573 @code{STANDARD_INCLUDE_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
574 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
575 and specify private search areas for GCC@. The directory
576 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
578 The definition should be an initializer for an array of structures.
579 Each array element should have four elements: the directory name (a
580 string constant), the component name (also a string constant), a flag
581 for C++-only directories,
582 and a flag showing that the includes in the directory don't need to be
583 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
584 the array with a null element.
586 The component name denotes what GNU package the include file is part of,
587 if any, in all uppercase letters. For example, it might be @samp{GCC}
588 or @samp{BINUTILS}. If the package is part of a vendor-supplied
589 operating system, code the component name as @samp{0}.
591 For example, here is the definition used for VAX/VMS:
594 #define INCLUDE_DEFAULTS \
596 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
597 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
598 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
605 Here is the order of prefixes tried for exec files:
609 Any prefixes specified by the user with @option{-B}.
612 The environment variable @code{GCC_EXEC_PREFIX}, if any.
615 The directories specified by the environment variable @code{COMPILER_PATH}.
618 The macro @code{STANDARD_EXEC_PREFIX}.
621 @file{/usr/lib/gcc/}.
624 The macro @code{MD_EXEC_PREFIX}, if any.
627 Here is the order of prefixes tried for startfiles:
631 Any prefixes specified by the user with @option{-B}.
634 The environment variable @code{GCC_EXEC_PREFIX}, if any.
637 The directories specified by the environment variable @code{LIBRARY_PATH}
638 (or port-specific name; native only, cross compilers do not use this).
641 The macro @code{STANDARD_EXEC_PREFIX}.
644 @file{/usr/lib/gcc/}.
647 The macro @code{MD_EXEC_PREFIX}, if any.
650 The macro @code{MD_STARTFILE_PREFIX}, if any.
653 The macro @code{STANDARD_STARTFILE_PREFIX}.
662 @node Run-time Target
663 @section Run-time Target Specification
664 @cindex run-time target specification
665 @cindex predefined macros
666 @cindex target specifications
668 @c prevent bad page break with this line
669 Here are run-time target specifications.
671 @defmac TARGET_CPU_CPP_BUILTINS ()
672 This function-like macro expands to a block of code that defines
673 built-in preprocessor macros and assertions for the target cpu, using
674 the functions @code{builtin_define}, @code{builtin_define_std} and
675 @code{builtin_assert}. When the front end
676 calls this macro it provides a trailing semicolon, and since it has
677 finished command line option processing your code can use those
680 @code{builtin_assert} takes a string in the form you pass to the
681 command-line option @option{-A}, such as @code{cpu=mips}, and creates
682 the assertion. @code{builtin_define} takes a string in the form
683 accepted by option @option{-D} and unconditionally defines the macro.
685 @code{builtin_define_std} takes a string representing the name of an
686 object-like macro. If it doesn't lie in the user's namespace,
687 @code{builtin_define_std} defines it unconditionally. Otherwise, it
688 defines a version with two leading underscores, and another version
689 with two leading and trailing underscores, and defines the original
690 only if an ISO standard was not requested on the command line. For
691 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
692 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
693 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
694 defines only @code{_ABI64}.
696 You can also test for the C dialect being compiled. The variable
697 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
698 or @code{clk_objective_c}. Note that if we are preprocessing
699 assembler, this variable will be @code{clk_c} but the function-like
700 macro @code{preprocessing_asm_p()} will return true, so you might want
701 to check for that first. If you need to check for strict ANSI, the
702 variable @code{flag_iso} can be used. The function-like macro
703 @code{preprocessing_trad_p()} can be used to check for traditional
707 @defmac TARGET_OS_CPP_BUILTINS ()
708 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
709 and is used for the target operating system instead.
712 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
713 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
714 and is used for the target object format. @file{elfos.h} uses this
715 macro to define @code{__ELF__}, so you probably do not need to define
719 @deftypevar {extern int} target_flags
720 This declaration should be present.
723 @cindex optional hardware or system features
724 @cindex features, optional, in system conventions
726 @defmac TARGET_@var{featurename}
727 This series of macros is to allow compiler command arguments to
728 enable or disable the use of optional features of the target machine.
729 For example, one machine description serves both the 68000 and
730 the 68020; a command argument tells the compiler whether it should
731 use 68020-only instructions or not. This command argument works
732 by means of a macro @code{TARGET_68020} that tests a bit in
735 Define a macro @code{TARGET_@var{featurename}} for each such option.
736 Its definition should test a bit in @code{target_flags}. It is
737 recommended that a helper macro @code{MASK_@var{featurename}}
738 is defined for each bit-value to test, and used in
739 @code{TARGET_@var{featurename}} and @code{TARGET_SWITCHES}. For
743 #define TARGET_MASK_68020 1
744 #define TARGET_68020 (target_flags & MASK_68020)
747 One place where these macros are used is in the condition-expressions
748 of instruction patterns. Note how @code{TARGET_68020} appears
749 frequently in the 68000 machine description file, @file{m68k.md}.
750 Another place they are used is in the definitions of the other
751 macros in the @file{@var{machine}.h} file.
754 @defmac TARGET_SWITCHES
755 This macro defines names of command options to set and clear
756 bits in @code{target_flags}. Its definition is an initializer
757 with a subgrouping for each command option.
759 Each subgrouping contains a string constant, that defines the option
760 name, a number, which contains the bits to set in
761 @code{target_flags}, and a second string which is the description
762 displayed by @option{--help}. If the number is negative then the bits specified
763 by the number are cleared instead of being set. If the description
764 string is present but empty, then no help information will be displayed
765 for that option, but it will not count as an undocumented option. The
766 actual option name is made by appending @samp{-m} to the specified name.
767 Non-empty description strings should be marked with @code{N_(@dots{})} for
768 @command{xgettext}. Please do not mark empty strings because the empty
769 string is reserved by GNU gettext. @code{gettext("")} returns the header entry
770 of the message catalog with meta information, not the empty string.
772 In addition to the description for @option{--help},
773 more detailed documentation for each option should be added to
776 One of the subgroupings should have a null string. The number in
777 this grouping is the default value for @code{target_flags}. Any
778 target options act starting with that value.
780 Here is an example which defines @option{-m68000} and @option{-m68020}
781 with opposite meanings, and picks the latter as the default:
784 #define TARGET_SWITCHES \
785 @{ @{ "68020", MASK_68020, "" @}, \
786 @{ "68000", -MASK_68020, \
787 N_("Compile for the 68000") @}, \
788 @{ "", MASK_68020, "" @}, \
793 @defmac TARGET_OPTIONS
794 This macro is similar to @code{TARGET_SWITCHES} but defines names of command
795 options that have values. Its definition is an initializer with a
796 subgrouping for each command option.
798 Each subgrouping contains a string constant, that defines the option
799 name, the address of a variable, a description string, and a value.
800 Non-empty description strings should be marked with @code{N_(@dots{})}
801 for @command{xgettext}. Please do not mark empty strings because the
802 empty string is reserved by GNU gettext. @code{gettext("")} returns the
803 header entry of the message catalog with meta information, not the empty
806 If the value listed in the table is @code{NULL}, then the variable, type
807 @code{char *}, is set to the variable part of the given option if the
808 fixed part matches. In other words, if the first part of the option
809 matches what's in the table, the variable will be set to point to the
810 rest of the option. This allows the user to specify a value for that
811 option. The actual option name is made by appending @samp{-m} to the
812 specified name. Again, each option should also be documented in
815 If the value listed in the table is non-@code{NULL}, then the option
816 must match the option in the table exactly (with @samp{-m}), and the
817 variable is set to point to the value listed in the table.
819 Here is an example which defines @option{-mshort-data-@var{number}}. If the
820 given option is @option{-mshort-data-512}, the variable @code{m88k_short_data}
821 will be set to the string @code{"512"}.
824 extern char *m88k_short_data;
825 #define TARGET_OPTIONS \
826 @{ @{ "short-data-", &m88k_short_data, \
827 N_("Specify the size of the short data section"), 0 @} @}
830 Here is a variant of the above that allows the user to also specify
831 just @option{-mshort-data} where a default of @code{"64"} is used.
834 extern char *m88k_short_data;
835 #define TARGET_OPTIONS \
836 @{ @{ "short-data-", &m88k_short_data, \
837 N_("Specify the size of the short data section"), 0 @} \
838 @{ "short-data", &m88k_short_data, "", "64" @},
842 Here is an example which defines @option{-mno-alu}, @option{-malu1}, and
843 @option{-malu2} as a three-state switch, along with suitable macros for
844 checking the state of the option (documentation is elided for brevity).
848 char *chip_alu = ""; /* Specify default here. */
851 extern char *chip_alu;
852 #define TARGET_OPTIONS \
853 @{ @{ "no-alu", &chip_alu, "", "" @}, \
854 @{ "alu1", &chip_alu, "", "1" @}, \
855 @{ "alu2", &chip_alu, "", "2" @}, @}
856 #define TARGET_ALU (chip_alu[0] != '\0')
857 #define TARGET_ALU1 (chip_alu[0] == '1')
858 #define TARGET_ALU2 (chip_alu[0] == '2')
862 @defmac TARGET_VERSION
863 This macro is a C statement to print on @code{stderr} a string
864 describing the particular machine description choice. Every machine
865 description should define @code{TARGET_VERSION}. For example:
869 #define TARGET_VERSION \
870 fprintf (stderr, " (68k, Motorola syntax)");
872 #define TARGET_VERSION \
873 fprintf (stderr, " (68k, MIT syntax)");
878 @defmac OVERRIDE_OPTIONS
879 Sometimes certain combinations of command options do not make sense on
880 a particular target machine. You can define a macro
881 @code{OVERRIDE_OPTIONS} to take account of this. This macro, if
882 defined, is executed once just after all the command options have been
885 Don't use this macro to turn on various extra optimizations for
886 @option{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for.
889 @defmac OPTIMIZATION_OPTIONS (@var{level}, @var{size})
890 Some machines may desire to change what optimizations are performed for
891 various optimization levels. This macro, if defined, is executed once
892 just after the optimization level is determined and before the remainder
893 of the command options have been parsed. Values set in this macro are
894 used as the default values for the other command line options.
896 @var{level} is the optimization level specified; 2 if @option{-O2} is
897 specified, 1 if @option{-O} is specified, and 0 if neither is specified.
899 @var{size} is nonzero if @option{-Os} is specified and zero otherwise.
901 You should not use this macro to change options that are not
902 machine-specific. These should uniformly selected by the same
903 optimization level on all supported machines. Use this macro to enable
904 machine-specific optimizations.
906 @strong{Do not examine @code{write_symbols} in
907 this macro!} The debugging options are not supposed to alter the
911 @defmac CAN_DEBUG_WITHOUT_FP
912 Define this macro if debugging can be performed even without a frame
913 pointer. If this macro is defined, GCC will turn on the
914 @option{-fomit-frame-pointer} option whenever @option{-O} is specified.
917 @node Per-Function Data
918 @section Defining data structures for per-function information.
919 @cindex per-function data
920 @cindex data structures
922 If the target needs to store information on a per-function basis, GCC
923 provides a macro and a couple of variables to allow this. Note, just
924 using statics to store the information is a bad idea, since GCC supports
925 nested functions, so you can be halfway through encoding one function
926 when another one comes along.
928 GCC defines a data structure called @code{struct function} which
929 contains all of the data specific to an individual function. This
930 structure contains a field called @code{machine} whose type is
931 @code{struct machine_function *}, which can be used by targets to point
932 to their own specific data.
934 If a target needs per-function specific data it should define the type
935 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
936 This macro should be used to initialize the function pointer
937 @code{init_machine_status}. This pointer is explained below.
939 One typical use of per-function, target specific data is to create an
940 RTX to hold the register containing the function's return address. This
941 RTX can then be used to implement the @code{__builtin_return_address}
942 function, for level 0.
944 Note---earlier implementations of GCC used a single data area to hold
945 all of the per-function information. Thus when processing of a nested
946 function began the old per-function data had to be pushed onto a
947 stack, and when the processing was finished, it had to be popped off the
948 stack. GCC used to provide function pointers called
949 @code{save_machine_status} and @code{restore_machine_status} to handle
950 the saving and restoring of the target specific information. Since the
951 single data area approach is no longer used, these pointers are no
954 @defmac INIT_EXPANDERS
955 Macro called to initialize any target specific information. This macro
956 is called once per function, before generation of any RTL has begun.
957 The intention of this macro is to allow the initialization of the
958 function pointer @code{init_machine_status}.
961 @deftypevar {void (*)(struct function *)} init_machine_status
962 If this function pointer is non-@code{NULL} it will be called once per
963 function, before function compilation starts, in order to allow the
964 target to perform any target specific initialization of the
965 @code{struct function} structure. It is intended that this would be
966 used to initialize the @code{machine} of that structure.
968 @code{struct machine_function} structures are expected to be freed by GC.
969 Generally, any memory that they reference must be allocated by using
970 @code{ggc_alloc}, including the structure itself.
974 @section Storage Layout
975 @cindex storage layout
977 Note that the definitions of the macros in this table which are sizes or
978 alignments measured in bits do not need to be constant. They can be C
979 expressions that refer to static variables, such as the @code{target_flags}.
980 @xref{Run-time Target}.
982 @defmac BITS_BIG_ENDIAN
983 Define this macro to have the value 1 if the most significant bit in a
984 byte has the lowest number; otherwise define it to have the value zero.
985 This means that bit-field instructions count from the most significant
986 bit. If the machine has no bit-field instructions, then this must still
987 be defined, but it doesn't matter which value it is defined to. This
988 macro need not be a constant.
990 This macro does not affect the way structure fields are packed into
991 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
994 @defmac BYTES_BIG_ENDIAN
995 Define this macro to have the value 1 if the most significant byte in a
996 word has the lowest number. This macro need not be a constant.
999 @defmac WORDS_BIG_ENDIAN
1000 Define this macro to have the value 1 if, in a multiword object, the
1001 most significant word has the lowest number. This applies to both
1002 memory locations and registers; GCC fundamentally assumes that the
1003 order of words in memory is the same as the order in registers. This
1004 macro need not be a constant.
1007 @defmac LIBGCC2_WORDS_BIG_ENDIAN
1008 Define this macro if @code{WORDS_BIG_ENDIAN} is not constant. This must be a
1009 constant value with the same meaning as @code{WORDS_BIG_ENDIAN}, which will be
1010 used only when compiling @file{libgcc2.c}. Typically the value will be set
1011 based on preprocessor defines.
1014 @defmac FLOAT_WORDS_BIG_ENDIAN
1015 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
1016 @code{TFmode} floating point numbers are stored in memory with the word
1017 containing the sign bit at the lowest address; otherwise define it to
1018 have the value 0. This macro need not be a constant.
1020 You need not define this macro if the ordering is the same as for
1021 multi-word integers.
1024 @defmac BITS_PER_UNIT
1025 Define this macro to be the number of bits in an addressable storage
1026 unit (byte). If you do not define this macro the default is 8.
1029 @defmac BITS_PER_WORD
1030 Number of bits in a word. If you do not define this macro, the default
1031 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
1034 @defmac MAX_BITS_PER_WORD
1035 Maximum number of bits in a word. If this is undefined, the default is
1036 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
1037 largest value that @code{BITS_PER_WORD} can have at run-time.
1040 @defmac UNITS_PER_WORD
1041 Number of storage units in a word; normally 4.
1044 @defmac MIN_UNITS_PER_WORD
1045 Minimum number of units in a word. If this is undefined, the default is
1046 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
1047 smallest value that @code{UNITS_PER_WORD} can have at run-time.
1050 @defmac POINTER_SIZE
1051 Width of a pointer, in bits. You must specify a value no wider than the
1052 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
1053 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
1054 a value the default is @code{BITS_PER_WORD}.
1057 @defmac POINTERS_EXTEND_UNSIGNED
1058 A C expression whose value is greater than zero if pointers that need to be
1059 extended from being @code{POINTER_SIZE} bits wide to @code{Pmode} are to
1060 be zero-extended and zero if they are to be sign-extended. If the value
1061 is less then zero then there must be an "ptr_extend" instruction that
1062 extends a pointer from @code{POINTER_SIZE} to @code{Pmode}.
1064 You need not define this macro if the @code{POINTER_SIZE} is equal
1065 to the width of @code{Pmode}.
1068 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
1069 A macro to update @var{m} and @var{unsignedp} when an object whose type
1070 is @var{type} and which has the specified mode and signedness is to be
1071 stored in a register. This macro is only called when @var{type} is a
1074 On most RISC machines, which only have operations that operate on a full
1075 register, define this macro to set @var{m} to @code{word_mode} if
1076 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
1077 cases, only integer modes should be widened because wider-precision
1078 floating-point operations are usually more expensive than their narrower
1081 For most machines, the macro definition does not change @var{unsignedp}.
1082 However, some machines, have instructions that preferentially handle
1083 either signed or unsigned quantities of certain modes. For example, on
1084 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
1085 sign-extend the result to 64 bits. On such machines, set
1086 @var{unsignedp} according to which kind of extension is more efficient.
1088 Do not define this macro if it would never modify @var{m}.
1091 @defmac PROMOTE_FUNCTION_MODE
1092 Like @code{PROMOTE_MODE}, but is applied to outgoing function arguments or
1093 function return values, as specified by @code{TARGET_PROMOTE_FUNCTION_ARGS}
1094 and @code{TARGET_PROMOTE_FUNCTION_RETURN}, respectively.
1096 The default is @code{PROMOTE_MODE}.
1099 @deftypefn {Target Hook} bool TARGET_PROMOTE_FUNCTION_ARGS (tree @var{fntype})
1100 This target hook should return @code{true} if the promotion described by
1101 @code{PROMOTE_FUNCTION_MODE} should be done for outgoing function
1105 @deftypefn {Target Hook} bool TARGET_PROMOTE_FUNCTION_RETURN (tree @var{fntype})
1106 This target hook should return @code{true} if the promotion described by
1107 @code{PROMOTE_FUNCTION_MODE} should be done for the return value of
1110 If this target hook returns @code{true}, @code{FUNCTION_VALUE} must
1111 perform the same promotions done by @code{PROMOTE_FUNCTON_MODE}.
1114 @defmac PARM_BOUNDARY
1115 Normal alignment required for function parameters on the stack, in
1116 bits. All stack parameters receive at least this much alignment
1117 regardless of data type. On most machines, this is the same as the
1121 @defmac STACK_BOUNDARY
1122 Define this macro to the minimum alignment enforced by hardware for the
1123 stack pointer on this machine. The definition is a C expression for the
1124 desired alignment (measured in bits). This value is used as a default
1125 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
1126 this should be the same as @code{PARM_BOUNDARY}.
1129 @defmac PREFERRED_STACK_BOUNDARY
1130 Define this macro if you wish to preserve a certain alignment for the
1131 stack pointer, greater than what the hardware enforces. The definition
1132 is a C expression for the desired alignment (measured in bits). This
1133 macro must evaluate to a value equal to or larger than
1134 @code{STACK_BOUNDARY}.
1137 @defmac FORCE_PREFERRED_STACK_BOUNDARY_IN_MAIN
1138 A C expression that evaluates true if @code{PREFERRED_STACK_BOUNDARY} is
1139 not guaranteed by the runtime and we should emit code to align the stack
1140 at the beginning of @code{main}.
1142 @cindex @code{PUSH_ROUNDING}, interaction with @code{PREFERRED_STACK_BOUNDARY}
1143 If @code{PUSH_ROUNDING} is not defined, the stack will always be aligned
1144 to the specified boundary. If @code{PUSH_ROUNDING} is defined and specifies
1145 a less strict alignment than @code{PREFERRED_STACK_BOUNDARY}, the stack may
1146 be momentarily unaligned while pushing arguments.
1149 @defmac FUNCTION_BOUNDARY
1150 Alignment required for a function entry point, in bits.
1153 @defmac BIGGEST_ALIGNMENT
1154 Biggest alignment that any data type can require on this machine, in bits.
1157 @defmac MINIMUM_ATOMIC_ALIGNMENT
1158 If defined, the smallest alignment, in bits, that can be given to an
1159 object that can be referenced in one operation, without disturbing any
1160 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1161 on machines that don't have byte or half-word store operations.
1164 @defmac BIGGEST_FIELD_ALIGNMENT
1165 Biggest alignment that any structure or union field can require on this
1166 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1167 structure and union fields only, unless the field alignment has been set
1168 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1171 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1172 An expression for the alignment of a structure field @var{field} if the
1173 alignment computed in the usual way (including applying of
1174 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1175 alignment) is @var{computed}. It overrides alignment only if the
1176 field alignment has not been set by the
1177 @code{__attribute__ ((aligned (@var{n})))} construct.
1180 @defmac MAX_OFILE_ALIGNMENT
1181 Biggest alignment supported by the object file format of this machine.
1182 Use this macro to limit the alignment which can be specified using the
1183 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1184 the default value is @code{BIGGEST_ALIGNMENT}.
1187 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1188 If defined, a C expression to compute the alignment for a variable in
1189 the static store. @var{type} is the data type, and @var{basic-align} is
1190 the alignment that the object would ordinarily have. The value of this
1191 macro is used instead of that alignment to align the object.
1193 If this macro is not defined, then @var{basic-align} is used.
1196 One use of this macro is to increase alignment of medium-size data to
1197 make it all fit in fewer cache lines. Another is to cause character
1198 arrays to be word-aligned so that @code{strcpy} calls that copy
1199 constants to character arrays can be done inline.
1202 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1203 If defined, a C expression to compute the alignment given to a constant
1204 that is being placed in memory. @var{constant} is the constant and
1205 @var{basic-align} is the alignment that the object would ordinarily
1206 have. The value of this macro is used instead of that alignment to
1209 If this macro is not defined, then @var{basic-align} is used.
1211 The typical use of this macro is to increase alignment for string
1212 constants to be word aligned so that @code{strcpy} calls that copy
1213 constants can be done inline.
1216 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1217 If defined, a C expression to compute the alignment for a variable in
1218 the local store. @var{type} is the data type, and @var{basic-align} is
1219 the alignment that the object would ordinarily have. The value of this
1220 macro is used instead of that alignment to align the object.
1222 If this macro is not defined, then @var{basic-align} is used.
1224 One use of this macro is to increase alignment of medium-size data to
1225 make it all fit in fewer cache lines.
1228 @defmac EMPTY_FIELD_BOUNDARY
1229 Alignment in bits to be given to a structure bit-field that follows an
1230 empty field such as @code{int : 0;}.
1232 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1235 @defmac STRUCTURE_SIZE_BOUNDARY
1236 Number of bits which any structure or union's size must be a multiple of.
1237 Each structure or union's size is rounded up to a multiple of this.
1239 If you do not define this macro, the default is the same as
1240 @code{BITS_PER_UNIT}.
1243 @defmac STRICT_ALIGNMENT
1244 Define this macro to be the value 1 if instructions will fail to work
1245 if given data not on the nominal alignment. If instructions will merely
1246 go slower in that case, define this macro as 0.
1249 @defmac PCC_BITFIELD_TYPE_MATTERS
1250 Define this if you wish to imitate the way many other C compilers handle
1251 alignment of bit-fields and the structures that contain them.
1253 The behavior is that the type written for a named bit-field (@code{int},
1254 @code{short}, or other integer type) imposes an alignment for the entire
1255 structure, as if the structure really did contain an ordinary field of
1256 that type. In addition, the bit-field is placed within the structure so
1257 that it would fit within such a field, not crossing a boundary for it.
1259 Thus, on most machines, a named bit-field whose type is written as
1260 @code{int} would not cross a four-byte boundary, and would force
1261 four-byte alignment for the whole structure. (The alignment used may
1262 not be four bytes; it is controlled by the other alignment parameters.)
1264 An unnamed bit-field will not affect the alignment of the containing
1267 If the macro is defined, its definition should be a C expression;
1268 a nonzero value for the expression enables this behavior.
1270 Note that if this macro is not defined, or its value is zero, some
1271 bit-fields may cross more than one alignment boundary. The compiler can
1272 support such references if there are @samp{insv}, @samp{extv}, and
1273 @samp{extzv} insns that can directly reference memory.
1275 The other known way of making bit-fields work is to define
1276 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1277 Then every structure can be accessed with fullwords.
1279 Unless the machine has bit-field instructions or you define
1280 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1281 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1283 If your aim is to make GCC use the same conventions for laying out
1284 bit-fields as are used by another compiler, here is how to investigate
1285 what the other compiler does. Compile and run this program:
1304 printf ("Size of foo1 is %d\n",
1305 sizeof (struct foo1));
1306 printf ("Size of foo2 is %d\n",
1307 sizeof (struct foo2));
1312 If this prints 2 and 5, then the compiler's behavior is what you would
1313 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1316 @defmac BITFIELD_NBYTES_LIMITED
1317 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1318 to aligning a bit-field within the structure.
1321 @deftypefn {Target Hook} bool TARGET_ALIGN_ANON_BITFIELDS (void)
1322 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1323 whether unnamed bitfields affect the alignment of the containing
1324 structure. The hook should return true if the structure should inherit
1325 the alignment requirements of an unnamed bitfield's type.
1328 @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode})
1329 Return 1 if a structure or array containing @var{field} should be accessed using
1332 If @var{field} is the only field in the structure, @var{mode} is its
1333 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1334 case where structures of one field would require the structure's mode to
1335 retain the field's mode.
1337 Normally, this is not needed. See the file @file{c4x.h} for an example
1338 of how to use this macro to prevent a structure having a floating point
1339 field from being accessed in an integer mode.
1342 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1343 Define this macro as an expression for the alignment of a type (given
1344 by @var{type} as a tree node) if the alignment computed in the usual
1345 way is @var{computed} and the alignment explicitly specified was
1348 The default is to use @var{specified} if it is larger; otherwise, use
1349 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1352 @defmac MAX_FIXED_MODE_SIZE
1353 An integer expression for the size in bits of the largest integer
1354 machine mode that should actually be used. All integer machine modes of
1355 this size or smaller can be used for structures and unions with the
1356 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1357 (DImode)} is assumed.
1360 @defmac VECTOR_MODE_SUPPORTED_P (@var{mode})
1361 Define this macro to be nonzero if the port is prepared to handle insns
1362 involving vector mode @var{mode}. At the very least, it must have move
1363 patterns for this mode.
1366 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1367 If defined, an expression of type @code{enum machine_mode} that
1368 specifies the mode of the save area operand of a
1369 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1370 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1371 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1372 having its mode specified.
1374 You need not define this macro if it always returns @code{Pmode}. You
1375 would most commonly define this macro if the
1376 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1380 @defmac STACK_SIZE_MODE
1381 If defined, an expression of type @code{enum machine_mode} that
1382 specifies the mode of the size increment operand of an
1383 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1385 You need not define this macro if it always returns @code{word_mode}.
1386 You would most commonly define this macro if the @code{allocate_stack}
1387 pattern needs to support both a 32- and a 64-bit mode.
1390 @defmac TARGET_FLOAT_FORMAT
1391 A code distinguishing the floating point format of the target machine.
1392 There are four defined values:
1395 @item IEEE_FLOAT_FORMAT
1396 This code indicates IEEE floating point. It is the default; there is no
1397 need to define @code{TARGET_FLOAT_FORMAT} when the format is IEEE@.
1399 @item VAX_FLOAT_FORMAT
1400 This code indicates the ``F float'' (for @code{float}) and ``D float''
1401 or ``G float'' formats (for @code{double}) used on the VAX and PDP-11@.
1403 @item IBM_FLOAT_FORMAT
1404 This code indicates the format used on the IBM System/370.
1406 @item C4X_FLOAT_FORMAT
1407 This code indicates the format used on the TMS320C3x/C4x.
1410 If your target uses a floating point format other than these, you must
1411 define a new @var{name}_FLOAT_FORMAT code for it, and add support for
1412 it to @file{real.c}.
1414 The ordering of the component words of floating point values stored in
1415 memory is controlled by @code{FLOAT_WORDS_BIG_ENDIAN}.
1418 @defmac MODE_HAS_NANS (@var{mode})
1419 When defined, this macro should be true if @var{mode} has a NaN
1420 representation. The compiler assumes that NaNs are not equal to
1421 anything (including themselves) and that addition, subtraction,
1422 multiplication and division all return NaNs when one operand is
1425 By default, this macro is true if @var{mode} is a floating-point
1426 mode and the target floating-point format is IEEE@.
1429 @defmac MODE_HAS_INFINITIES (@var{mode})
1430 This macro should be true if @var{mode} can represent infinity. At
1431 present, the compiler uses this macro to decide whether @samp{x - x}
1432 is always defined. By default, the macro is true when @var{mode}
1433 is a floating-point mode and the target format is IEEE@.
1436 @defmac MODE_HAS_SIGNED_ZEROS (@var{mode})
1437 True if @var{mode} distinguishes between positive and negative zero.
1438 The rules are expected to follow the IEEE standard:
1442 @samp{x + x} has the same sign as @samp{x}.
1445 If the sum of two values with opposite sign is zero, the result is
1446 positive for all rounding modes expect towards @minus{}infinity, for
1447 which it is negative.
1450 The sign of a product or quotient is negative when exactly one
1451 of the operands is negative.
1454 The default definition is true if @var{mode} is a floating-point
1455 mode and the target format is IEEE@.
1458 @defmac MODE_HAS_SIGN_DEPENDENT_ROUNDING (@var{mode})
1459 If defined, this macro should be true for @var{mode} if it has at
1460 least one rounding mode in which @samp{x} and @samp{-x} can be
1461 rounded to numbers of different magnitude. Two such modes are
1462 towards @minus{}infinity and towards +infinity.
1464 The default definition of this macro is true if @var{mode} is
1465 a floating-point mode and the target format is IEEE@.
1468 @defmac ROUND_TOWARDS_ZERO
1469 If defined, this macro should be true if the prevailing rounding
1470 mode is towards zero. A true value has the following effects:
1474 @code{MODE_HAS_SIGN_DEPENDENT_ROUNDING} will be false for all modes.
1477 @file{libgcc.a}'s floating-point emulator will round towards zero
1478 rather than towards nearest.
1481 The compiler's floating-point emulator will round towards zero after
1482 doing arithmetic, and when converting from the internal float format to
1486 The macro does not affect the parsing of string literals. When the
1487 primary rounding mode is towards zero, library functions like
1488 @code{strtod} might still round towards nearest, and the compiler's
1489 parser should behave like the target's @code{strtod} where possible.
1491 Not defining this macro is equivalent to returning zero.
1494 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1495 This macro should return true if floats with @var{size}
1496 bits do not have a NaN or infinity representation, but use the largest
1497 exponent for normal numbers instead.
1499 Defining this macro to true for @var{size} causes @code{MODE_HAS_NANS}
1500 and @code{MODE_HAS_INFINITIES} to be false for @var{size}-bit modes.
1501 It also affects the way @file{libgcc.a} and @file{real.c} emulate
1502 floating-point arithmetic.
1504 The default definition of this macro returns false for all sizes.
1507 @deftypefn {Target Hook} bool TARGET_VECTOR_OPAQUE_P (tree @var{type})
1508 This target hook should return @code{true} a vector is opaque. That
1509 is, if no cast is needed when copying a vector value of type
1510 @var{type} into another vector lvalue of the same size. Vector opaque
1511 types cannot be initialized. The default is that there are no such
1515 @deftypefn {Target Hook} bool TARGET_MS_BITFIELD_LAYOUT_P (tree @var{record_type})
1516 This target hook returns @code{true} if bit-fields in the given
1517 @var{record_type} are to be laid out following the rules of Microsoft
1518 Visual C/C++, namely: (i) a bit-field won't share the same storage
1519 unit with the previous bit-field if their underlying types have
1520 different sizes, and the bit-field will be aligned to the highest
1521 alignment of the underlying types of itself and of the previous
1522 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1523 the whole enclosing structure, even if it is unnamed; except that
1524 (iii) a zero-sized bit-field will be disregarded unless it follows
1525 another bit-field of nonzero size. If this hook returns @code{true},
1526 other macros that control bit-field layout are ignored.
1528 When a bit-field is inserted into a packed record, the whole size
1529 of the underlying type is used by one or more same-size adjacent
1530 bit-fields (that is, if its long:3, 32 bits is used in the record,
1531 and any additional adjacent long bit-fields are packed into the same
1532 chunk of 32 bits. However, if the size changes, a new field of that
1533 size is allocated). In an unpacked record, this is the same as using
1534 alignment, but not equivalent when packing.
1536 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1537 the latter will take precedence. If @samp{__attribute__((packed))} is
1538 used on a single field when MS bit-fields are in use, it will take
1539 precedence for that field, but the alignment of the rest of the structure
1540 may affect its placement.
1543 @deftypefn {Target Hook} {const char *} TARGET_MANGLE_FUNDAMENTAL_TYPE (tree @var{type})
1544 If your target defines any fundamental types, define this hook to
1545 return the appropriate encoding for these types as part of a C++
1546 mangled name. The @var{type} argument is the tree structure
1547 representing the type to be mangled. The hook may be applied to trees
1548 which are not target-specific fundamental types; it should return
1549 @code{NULL} for all such types, as well as arguments it does not
1550 recognize. If the return value is not @code{NULL}, it must point to
1551 a statically-allocated string constant.
1553 Target-specific fundamental types might be new fundamental types or
1554 qualified versions of ordinary fundamental types. Encode new
1555 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1556 is the name used for the type in source code, and @var{n} is the
1557 length of @var{name} in decimal. Encode qualified versions of
1558 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1559 @var{name} is the name used for the type qualifier in source code,
1560 @var{n} is the length of @var{name} as above, and @var{code} is the
1561 code used to represent the unqualified version of this type. (See
1562 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1563 codes.) In both cases the spaces are for clarity; do not include any
1564 spaces in your string.
1566 The default version of this hook always returns @code{NULL}, which is
1567 appropriate for a target that does not define any new fundamental
1572 @section Layout of Source Language Data Types
1574 These macros define the sizes and other characteristics of the standard
1575 basic data types used in programs being compiled. Unlike the macros in
1576 the previous section, these apply to specific features of C and related
1577 languages, rather than to fundamental aspects of storage layout.
1579 @defmac INT_TYPE_SIZE
1580 A C expression for the size in bits of the type @code{int} on the
1581 target machine. If you don't define this, the default is one word.
1584 @defmac SHORT_TYPE_SIZE
1585 A C expression for the size in bits of the type @code{short} on the
1586 target machine. If you don't define this, the default is half a word.
1587 (If this would be less than one storage unit, it is rounded up to one
1591 @defmac LONG_TYPE_SIZE
1592 A C expression for the size in bits of the type @code{long} on the
1593 target machine. If you don't define this, the default is one word.
1596 @defmac ADA_LONG_TYPE_SIZE
1597 On some machines, the size used for the Ada equivalent of the type
1598 @code{long} by a native Ada compiler differs from that used by C. In
1599 that situation, define this macro to be a C expression to be used for
1600 the size of that type. If you don't define this, the default is the
1601 value of @code{LONG_TYPE_SIZE}.
1604 @defmac LONG_LONG_TYPE_SIZE
1605 A C expression for the size in bits of the type @code{long long} on the
1606 target machine. If you don't define this, the default is two
1607 words. If you want to support GNU Ada on your machine, the value of this
1608 macro must be at least 64.
1611 @defmac CHAR_TYPE_SIZE
1612 A C expression for the size in bits of the type @code{char} on the
1613 target machine. If you don't define this, the default is
1614 @code{BITS_PER_UNIT}.
1617 @defmac BOOL_TYPE_SIZE
1618 A C expression for the size in bits of the C++ type @code{bool} and
1619 C99 type @code{_Bool} on the target machine. If you don't define
1620 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1623 @defmac FLOAT_TYPE_SIZE
1624 A C expression for the size in bits of the type @code{float} on the
1625 target machine. If you don't define this, the default is one word.
1628 @defmac DOUBLE_TYPE_SIZE
1629 A C expression for the size in bits of the type @code{double} on the
1630 target machine. If you don't define this, the default is two
1634 @defmac LONG_DOUBLE_TYPE_SIZE
1635 A C expression for the size in bits of the type @code{long double} on
1636 the target machine. If you don't define this, the default is two
1640 @defmac TARGET_FLT_EVAL_METHOD
1641 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1642 assuming, if applicable, that the floating-point control word is in its
1643 default state. If you do not define this macro the value of
1644 @code{FLT_EVAL_METHOD} will be zero.
1647 @defmac WIDEST_HARDWARE_FP_SIZE
1648 A C expression for the size in bits of the widest floating-point format
1649 supported by the hardware. If you define this macro, you must specify a
1650 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1651 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1655 @defmac DEFAULT_SIGNED_CHAR
1656 An expression whose value is 1 or 0, according to whether the type
1657 @code{char} should be signed or unsigned by default. The user can
1658 always override this default with the options @option{-fsigned-char}
1659 and @option{-funsigned-char}.
1662 @deftypefn {Target Hook} bool TARGET_DEFAULT_SHORT_ENUMS (void)
1663 This target hook should return true if the compiler should give an
1664 @code{enum} type only as many bytes as it takes to represent the range
1665 of possible values of that type. It should return false if all
1666 @code{enum} types should be allocated like @code{int}.
1668 The default is to return false.
1672 A C expression for a string describing the name of the data type to use
1673 for size values. The typedef name @code{size_t} is defined using the
1674 contents of the string.
1676 The string can contain more than one keyword. If so, separate them with
1677 spaces, and write first any length keyword, then @code{unsigned} if
1678 appropriate, and finally @code{int}. The string must exactly match one
1679 of the data type names defined in the function
1680 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1681 omit @code{int} or change the order---that would cause the compiler to
1684 If you don't define this macro, the default is @code{"long unsigned
1688 @defmac PTRDIFF_TYPE
1689 A C expression for a string describing the name of the data type to use
1690 for the result of subtracting two pointers. The typedef name
1691 @code{ptrdiff_t} is defined using the contents of the string. See
1692 @code{SIZE_TYPE} above for more information.
1694 If you don't define this macro, the default is @code{"long int"}.
1698 A C expression for a string describing the name of the data type to use
1699 for wide characters. The typedef name @code{wchar_t} is defined using
1700 the contents of the string. See @code{SIZE_TYPE} above for more
1703 If you don't define this macro, the default is @code{"int"}.
1706 @defmac WCHAR_TYPE_SIZE
1707 A C expression for the size in bits of the data type for wide
1708 characters. This is used in @code{cpp}, which cannot make use of
1713 A C expression for a string describing the name of the data type to
1714 use for wide characters passed to @code{printf} and returned from
1715 @code{getwc}. The typedef name @code{wint_t} is defined using the
1716 contents of the string. See @code{SIZE_TYPE} above for more
1719 If you don't define this macro, the default is @code{"unsigned int"}.
1723 A C expression for a string describing the name of the data type that
1724 can represent any value of any standard or extended signed integer type.
1725 The typedef name @code{intmax_t} is defined using the contents of the
1726 string. See @code{SIZE_TYPE} above for more information.
1728 If you don't define this macro, the default is the first of
1729 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1730 much precision as @code{long long int}.
1733 @defmac UINTMAX_TYPE
1734 A C expression for a string describing the name of the data type that
1735 can represent any value of any standard or extended unsigned integer
1736 type. The typedef name @code{uintmax_t} is defined using the contents
1737 of the string. See @code{SIZE_TYPE} above for more information.
1739 If you don't define this macro, the default is the first of
1740 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1741 unsigned int"} that has as much precision as @code{long long unsigned
1745 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1746 The C++ compiler represents a pointer-to-member-function with a struct
1753 ptrdiff_t vtable_index;
1760 The C++ compiler must use one bit to indicate whether the function that
1761 will be called through a pointer-to-member-function is virtual.
1762 Normally, we assume that the low-order bit of a function pointer must
1763 always be zero. Then, by ensuring that the vtable_index is odd, we can
1764 distinguish which variant of the union is in use. But, on some
1765 platforms function pointers can be odd, and so this doesn't work. In
1766 that case, we use the low-order bit of the @code{delta} field, and shift
1767 the remainder of the @code{delta} field to the left.
1769 GCC will automatically make the right selection about where to store
1770 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1771 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1772 set such that functions always start at even addresses, but the lowest
1773 bit of pointers to functions indicate whether the function at that
1774 address is in ARM or Thumb mode. If this is the case of your
1775 architecture, you should define this macro to
1776 @code{ptrmemfunc_vbit_in_delta}.
1778 In general, you should not have to define this macro. On architectures
1779 in which function addresses are always even, according to
1780 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1781 @code{ptrmemfunc_vbit_in_pfn}.
1784 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1785 Normally, the C++ compiler uses function pointers in vtables. This
1786 macro allows the target to change to use ``function descriptors''
1787 instead. Function descriptors are found on targets for whom a
1788 function pointer is actually a small data structure. Normally the
1789 data structure consists of the actual code address plus a data
1790 pointer to which the function's data is relative.
1792 If vtables are used, the value of this macro should be the number
1793 of words that the function descriptor occupies.
1796 @defmac TARGET_VTABLE_ENTRY_ALIGN
1797 By default, the vtable entries are void pointers, the so the alignment
1798 is the same as pointer alignment. The value of this macro specifies
1799 the alignment of the vtable entry in bits. It should be defined only
1800 when special alignment is necessary. */
1803 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1804 There are a few non-descriptor entries in the vtable at offsets below
1805 zero. If these entries must be padded (say, to preserve the alignment
1806 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1807 of words in each data entry.
1810 @node Escape Sequences
1811 @section Target Character Escape Sequences
1812 @cindex escape sequences
1814 By default, GCC assumes that the C character escape sequences and other
1815 characters take on their ASCII values for the target. If this is not
1816 correct, you must explicitly define all of the macros below. All of
1817 them must evaluate to constants; they are used in @code{case}
1823 @findex TARGET_DIGIT0
1826 @findex TARGET_NEWLINE
1829 @multitable {@code{TARGET_NEWLINE}} {Escape} {ASCII character}
1830 @item Macro @tab Escape @tab ASCII character
1831 @item @code{TARGET_BELL} @tab @kbd{\a} @tab @code{07}, @code{BEL}
1832 @item @code{TARGET_BS} @tab @kbd{\b} @tab @code{08}, @code{BS}
1833 @item @code{TARGET_CR} @tab @kbd{\r} @tab @code{0D}, @code{CR}
1834 @item @code{TARGET_DIGIT0} @tab @kbd{0} @tab @code{30}, @code{ZERO}
1835 @item @code{TARGET_ESC} @tab @kbd{\e}, @kbd{\E} @tab @code{1B}, @code{ESC}
1836 @item @code{TARGET_FF} @tab @kbd{\f} @tab @code{0C}, @code{FF}
1837 @item @code{TARGET_NEWLINE} @tab @kbd{\n} @tab @code{0A}, @code{LF}
1838 @item @code{TARGET_TAB} @tab @kbd{\t} @tab @code{09}, @code{HT}
1839 @item @code{TARGET_VT} @tab @kbd{\v} @tab @code{0B}, @code{VT}
1843 Note that the @kbd{\e} and @kbd{\E} escapes are GNU extensions, not
1844 part of the C standard.
1847 @section Register Usage
1848 @cindex register usage
1850 This section explains how to describe what registers the target machine
1851 has, and how (in general) they can be used.
1853 The description of which registers a specific instruction can use is
1854 done with register classes; see @ref{Register Classes}. For information
1855 on using registers to access a stack frame, see @ref{Frame Registers}.
1856 For passing values in registers, see @ref{Register Arguments}.
1857 For returning values in registers, see @ref{Scalar Return}.
1860 * Register Basics:: Number and kinds of registers.
1861 * Allocation Order:: Order in which registers are allocated.
1862 * Values in Registers:: What kinds of values each reg can hold.
1863 * Leaf Functions:: Renumbering registers for leaf functions.
1864 * Stack Registers:: Handling a register stack such as 80387.
1867 @node Register Basics
1868 @subsection Basic Characteristics of Registers
1870 @c prevent bad page break with this line
1871 Registers have various characteristics.
1873 @defmac FIRST_PSEUDO_REGISTER
1874 Number of hardware registers known to the compiler. They receive
1875 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1876 pseudo register's number really is assigned the number
1877 @code{FIRST_PSEUDO_REGISTER}.
1880 @defmac FIXED_REGISTERS
1881 @cindex fixed register
1882 An initializer that says which registers are used for fixed purposes
1883 all throughout the compiled code and are therefore not available for
1884 general allocation. These would include the stack pointer, the frame
1885 pointer (except on machines where that can be used as a general
1886 register when no frame pointer is needed), the program counter on
1887 machines where that is considered one of the addressable registers,
1888 and any other numbered register with a standard use.
1890 This information is expressed as a sequence of numbers, separated by
1891 commas and surrounded by braces. The @var{n}th number is 1 if
1892 register @var{n} is fixed, 0 otherwise.
1894 The table initialized from this macro, and the table initialized by
1895 the following one, may be overridden at run time either automatically,
1896 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1897 the user with the command options @option{-ffixed-@var{reg}},
1898 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1901 @defmac CALL_USED_REGISTERS
1902 @cindex call-used register
1903 @cindex call-clobbered register
1904 @cindex call-saved register
1905 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1906 clobbered (in general) by function calls as well as for fixed
1907 registers. This macro therefore identifies the registers that are not
1908 available for general allocation of values that must live across
1911 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1912 automatically saves it on function entry and restores it on function
1913 exit, if the register is used within the function.
1916 @defmac CALL_REALLY_USED_REGISTERS
1917 @cindex call-used register
1918 @cindex call-clobbered register
1919 @cindex call-saved register
1920 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1921 that the entire set of @code{FIXED_REGISTERS} be included.
1922 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1923 This macro is optional. If not specified, it defaults to the value
1924 of @code{CALL_USED_REGISTERS}.
1927 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1928 @cindex call-used register
1929 @cindex call-clobbered register
1930 @cindex call-saved register
1931 A C expression that is nonzero if it is not permissible to store a
1932 value of mode @var{mode} in hard register number @var{regno} across a
1933 call without some part of it being clobbered. For most machines this
1934 macro need not be defined. It is only required for machines that do not
1935 preserve the entire contents of a register across a call.
1939 @findex call_used_regs
1942 @findex reg_class_contents
1943 @defmac CONDITIONAL_REGISTER_USAGE
1944 Zero or more C statements that may conditionally modify five variables
1945 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1946 @code{reg_names}, and @code{reg_class_contents}, to take into account
1947 any dependence of these register sets on target flags. The first three
1948 of these are of type @code{char []} (interpreted as Boolean vectors).
1949 @code{global_regs} is a @code{const char *[]}, and
1950 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1951 called, @code{fixed_regs}, @code{call_used_regs},
1952 @code{reg_class_contents}, and @code{reg_names} have been initialized
1953 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1954 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1955 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1956 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1957 command options have been applied.
1959 You need not define this macro if it has no work to do.
1961 @cindex disabling certain registers
1962 @cindex controlling register usage
1963 If the usage of an entire class of registers depends on the target
1964 flags, you may indicate this to GCC by using this macro to modify
1965 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
1966 registers in the classes which should not be used by GCC@. Also define
1967 the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT}
1968 to return @code{NO_REGS} if it
1969 is called with a letter for a class that shouldn't be used.
1971 (However, if this class is not included in @code{GENERAL_REGS} and all
1972 of the insn patterns whose constraints permit this class are
1973 controlled by target switches, then GCC will automatically avoid using
1974 these registers when the target switches are opposed to them.)
1977 @defmac NON_SAVING_SETJMP
1978 If this macro is defined and has a nonzero value, it means that
1979 @code{setjmp} and related functions fail to save the registers, or that
1980 @code{longjmp} fails to restore them. To compensate, the compiler
1981 avoids putting variables in registers in functions that use
1985 @defmac INCOMING_REGNO (@var{out})
1986 Define this macro if the target machine has register windows. This C
1987 expression returns the register number as seen by the called function
1988 corresponding to the register number @var{out} as seen by the calling
1989 function. Return @var{out} if register number @var{out} is not an
1993 @defmac OUTGOING_REGNO (@var{in})
1994 Define this macro if the target machine has register windows. This C
1995 expression returns the register number as seen by the calling function
1996 corresponding to the register number @var{in} as seen by the called
1997 function. Return @var{in} if register number @var{in} is not an inbound
2001 @defmac LOCAL_REGNO (@var{regno})
2002 Define this macro if the target machine has register windows. This C
2003 expression returns true if the register is call-saved but is in the
2004 register window. Unlike most call-saved registers, such registers
2005 need not be explicitly restored on function exit or during non-local
2010 If the program counter has a register number, define this as that
2011 register number. Otherwise, do not define it.
2014 @node Allocation Order
2015 @subsection Order of Allocation of Registers
2016 @cindex order of register allocation
2017 @cindex register allocation order
2019 @c prevent bad page break with this line
2020 Registers are allocated in order.
2022 @defmac REG_ALLOC_ORDER
2023 If defined, an initializer for a vector of integers, containing the
2024 numbers of hard registers in the order in which GCC should prefer
2025 to use them (from most preferred to least).
2027 If this macro is not defined, registers are used lowest numbered first
2028 (all else being equal).
2030 One use of this macro is on machines where the highest numbered
2031 registers must always be saved and the save-multiple-registers
2032 instruction supports only sequences of consecutive registers. On such
2033 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
2034 the highest numbered allocable register first.
2037 @defmac ORDER_REGS_FOR_LOCAL_ALLOC
2038 A C statement (sans semicolon) to choose the order in which to allocate
2039 hard registers for pseudo-registers local to a basic block.
2041 Store the desired register order in the array @code{reg_alloc_order}.
2042 Element 0 should be the register to allocate first; element 1, the next
2043 register; and so on.
2045 The macro body should not assume anything about the contents of
2046 @code{reg_alloc_order} before execution of the macro.
2048 On most machines, it is not necessary to define this macro.
2051 @node Values in Registers
2052 @subsection How Values Fit in Registers
2054 This section discusses the macros that describe which kinds of values
2055 (specifically, which machine modes) each register can hold, and how many
2056 consecutive registers are needed for a given mode.
2058 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2059 A C expression for the number of consecutive hard registers, starting
2060 at register number @var{regno}, required to hold a value of mode
2063 On a machine where all registers are exactly one word, a suitable
2064 definition of this macro is
2067 #define HARD_REGNO_NREGS(REGNO, MODE) \
2068 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2073 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2074 Define this macro if the natural size of registers that hold values
2075 of mode @var{mode} is not the word size. It is a C expression that
2076 should give the natural size in bytes for the specified mode. It is
2077 used by the register allocator to try to optimize its results. This
2078 happens for example on SPARC 64-bit where the natural size of
2079 floating-point registers is still 32-bit.
2082 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2083 A C expression that is nonzero if it is permissible to store a value
2084 of mode @var{mode} in hard register number @var{regno} (or in several
2085 registers starting with that one). For a machine where all registers
2086 are equivalent, a suitable definition is
2089 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2092 You need not include code to check for the numbers of fixed registers,
2093 because the allocation mechanism considers them to be always occupied.
2095 @cindex register pairs
2096 On some machines, double-precision values must be kept in even/odd
2097 register pairs. You can implement that by defining this macro to reject
2098 odd register numbers for such modes.
2100 The minimum requirement for a mode to be OK in a register is that the
2101 @samp{mov@var{mode}} instruction pattern support moves between the
2102 register and other hard register in the same class and that moving a
2103 value into the register and back out not alter it.
2105 Since the same instruction used to move @code{word_mode} will work for
2106 all narrower integer modes, it is not necessary on any machine for
2107 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2108 you define patterns @samp{movhi}, etc., to take advantage of this. This
2109 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2110 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2113 Many machines have special registers for floating point arithmetic.
2114 Often people assume that floating point machine modes are allowed only
2115 in floating point registers. This is not true. Any registers that
2116 can hold integers can safely @emph{hold} a floating point machine
2117 mode, whether or not floating arithmetic can be done on it in those
2118 registers. Integer move instructions can be used to move the values.
2120 On some machines, though, the converse is true: fixed-point machine
2121 modes may not go in floating registers. This is true if the floating
2122 registers normalize any value stored in them, because storing a
2123 non-floating value there would garble it. In this case,
2124 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2125 floating registers. But if the floating registers do not automatically
2126 normalize, if you can store any bit pattern in one and retrieve it
2127 unchanged without a trap, then any machine mode may go in a floating
2128 register, so you can define this macro to say so.
2130 The primary significance of special floating registers is rather that
2131 they are the registers acceptable in floating point arithmetic
2132 instructions. However, this is of no concern to
2133 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2134 constraints for those instructions.
2136 On some machines, the floating registers are especially slow to access,
2137 so that it is better to store a value in a stack frame than in such a
2138 register if floating point arithmetic is not being done. As long as the
2139 floating registers are not in class @code{GENERAL_REGS}, they will not
2140 be used unless some pattern's constraint asks for one.
2143 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2144 A C expression that is nonzero if it is OK to rename a hard register
2145 @var{from} to another hard register @var{to}.
2147 One common use of this macro is to prevent renaming of a register to
2148 another register that is not saved by a prologue in an interrupt
2151 The default is always nonzero.
2154 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2155 A C expression that is nonzero if a value of mode
2156 @var{mode1} is accessible in mode @var{mode2} without copying.
2158 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2159 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2160 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2161 should be nonzero. If they differ for any @var{r}, you should define
2162 this macro to return zero unless some other mechanism ensures the
2163 accessibility of the value in a narrower mode.
2165 You should define this macro to return nonzero in as many cases as
2166 possible since doing so will allow GCC to perform better register
2170 @defmac AVOID_CCMODE_COPIES
2171 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2172 registers. You should only define this macro if support for copying to/from
2173 @code{CCmode} is incomplete.
2176 @node Leaf Functions
2177 @subsection Handling Leaf Functions
2179 @cindex leaf functions
2180 @cindex functions, leaf
2181 On some machines, a leaf function (i.e., one which makes no calls) can run
2182 more efficiently if it does not make its own register window. Often this
2183 means it is required to receive its arguments in the registers where they
2184 are passed by the caller, instead of the registers where they would
2187 The special treatment for leaf functions generally applies only when
2188 other conditions are met; for example, often they may use only those
2189 registers for its own variables and temporaries. We use the term ``leaf
2190 function'' to mean a function that is suitable for this special
2191 handling, so that functions with no calls are not necessarily ``leaf
2194 GCC assigns register numbers before it knows whether the function is
2195 suitable for leaf function treatment. So it needs to renumber the
2196 registers in order to output a leaf function. The following macros
2199 @defmac LEAF_REGISTERS
2200 Name of a char vector, indexed by hard register number, which
2201 contains 1 for a register that is allowable in a candidate for leaf
2204 If leaf function treatment involves renumbering the registers, then the
2205 registers marked here should be the ones before renumbering---those that
2206 GCC would ordinarily allocate. The registers which will actually be
2207 used in the assembler code, after renumbering, should not be marked with 1
2210 Define this macro only if the target machine offers a way to optimize
2211 the treatment of leaf functions.
2214 @defmac LEAF_REG_REMAP (@var{regno})
2215 A C expression whose value is the register number to which @var{regno}
2216 should be renumbered, when a function is treated as a leaf function.
2218 If @var{regno} is a register number which should not appear in a leaf
2219 function before renumbering, then the expression should yield @minus{}1, which
2220 will cause the compiler to abort.
2222 Define this macro only if the target machine offers a way to optimize the
2223 treatment of leaf functions, and registers need to be renumbered to do
2227 @findex current_function_is_leaf
2228 @findex current_function_uses_only_leaf_regs
2229 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2230 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2231 specially. They can test the C variable @code{current_function_is_leaf}
2232 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2233 set prior to local register allocation and is valid for the remaining
2234 compiler passes. They can also test the C variable
2235 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2236 functions which only use leaf registers.
2237 @code{current_function_uses_only_leaf_regs} is valid after all passes
2238 that modify the instructions have been run and is only useful if
2239 @code{LEAF_REGISTERS} is defined.
2240 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2241 @c of the next paragraph?! --mew 2feb93
2243 @node Stack Registers
2244 @subsection Registers That Form a Stack
2246 There are special features to handle computers where some of the
2247 ``registers'' form a stack. Stack registers are normally written by
2248 pushing onto the stack, and are numbered relative to the top of the
2251 Currently, GCC can only handle one group of stack-like registers, and
2252 they must be consecutively numbered. Furthermore, the existing
2253 support for stack-like registers is specific to the 80387 floating
2254 point coprocessor. If you have a new architecture that uses
2255 stack-like registers, you will need to do substantial work on
2256 @file{reg-stack.c} and write your machine description to cooperate
2257 with it, as well as defining these macros.
2260 Define this if the machine has any stack-like registers.
2263 @defmac FIRST_STACK_REG
2264 The number of the first stack-like register. This one is the top
2268 @defmac LAST_STACK_REG
2269 The number of the last stack-like register. This one is the bottom of
2273 @node Register Classes
2274 @section Register Classes
2275 @cindex register class definitions
2276 @cindex class definitions, register
2278 On many machines, the numbered registers are not all equivalent.
2279 For example, certain registers may not be allowed for indexed addressing;
2280 certain registers may not be allowed in some instructions. These machine
2281 restrictions are described to the compiler using @dfn{register classes}.
2283 You define a number of register classes, giving each one a name and saying
2284 which of the registers belong to it. Then you can specify register classes
2285 that are allowed as operands to particular instruction patterns.
2289 In general, each register will belong to several classes. In fact, one
2290 class must be named @code{ALL_REGS} and contain all the registers. Another
2291 class must be named @code{NO_REGS} and contain no registers. Often the
2292 union of two classes will be another class; however, this is not required.
2294 @findex GENERAL_REGS
2295 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2296 terribly special about the name, but the operand constraint letters
2297 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2298 the same as @code{ALL_REGS}, just define it as a macro which expands
2301 Order the classes so that if class @var{x} is contained in class @var{y}
2302 then @var{x} has a lower class number than @var{y}.
2304 The way classes other than @code{GENERAL_REGS} are specified in operand
2305 constraints is through machine-dependent operand constraint letters.
2306 You can define such letters to correspond to various classes, then use
2307 them in operand constraints.
2309 You should define a class for the union of two classes whenever some
2310 instruction allows both classes. For example, if an instruction allows
2311 either a floating point (coprocessor) register or a general register for a
2312 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2313 which includes both of them. Otherwise you will get suboptimal code.
2315 You must also specify certain redundant information about the register
2316 classes: for each class, which classes contain it and which ones are
2317 contained in it; for each pair of classes, the largest class contained
2320 When a value occupying several consecutive registers is expected in a
2321 certain class, all the registers used must belong to that class.
2322 Therefore, register classes cannot be used to enforce a requirement for
2323 a register pair to start with an even-numbered register. The way to
2324 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2326 Register classes used for input-operands of bitwise-and or shift
2327 instructions have a special requirement: each such class must have, for
2328 each fixed-point machine mode, a subclass whose registers can transfer that
2329 mode to or from memory. For example, on some machines, the operations for
2330 single-byte values (@code{QImode}) are limited to certain registers. When
2331 this is so, each register class that is used in a bitwise-and or shift
2332 instruction must have a subclass consisting of registers from which
2333 single-byte values can be loaded or stored. This is so that
2334 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2336 @deftp {Data type} {enum reg_class}
2337 An enumerated type that must be defined with all the register class names
2338 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2339 must be the last register class, followed by one more enumerated value,
2340 @code{LIM_REG_CLASSES}, which is not a register class but rather
2341 tells how many classes there are.
2343 Each register class has a number, which is the value of casting
2344 the class name to type @code{int}. The number serves as an index
2345 in many of the tables described below.
2348 @defmac N_REG_CLASSES
2349 The number of distinct register classes, defined as follows:
2352 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2356 @defmac REG_CLASS_NAMES
2357 An initializer containing the names of the register classes as C string
2358 constants. These names are used in writing some of the debugging dumps.
2361 @defmac REG_CLASS_CONTENTS
2362 An initializer containing the contents of the register classes, as integers
2363 which are bit masks. The @var{n}th integer specifies the contents of class
2364 @var{n}. The way the integer @var{mask} is interpreted is that
2365 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2367 When the machine has more than 32 registers, an integer does not suffice.
2368 Then the integers are replaced by sub-initializers, braced groupings containing
2369 several integers. Each sub-initializer must be suitable as an initializer
2370 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2371 In this situation, the first integer in each sub-initializer corresponds to
2372 registers 0 through 31, the second integer to registers 32 through 63, and
2376 @defmac REGNO_REG_CLASS (@var{regno})
2377 A C expression whose value is a register class containing hard register
2378 @var{regno}. In general there is more than one such class; choose a class
2379 which is @dfn{minimal}, meaning that no smaller class also contains the
2383 @defmac BASE_REG_CLASS
2384 A macro whose definition is the name of the class to which a valid
2385 base register must belong. A base register is one used in an address
2386 which is the register value plus a displacement.
2389 @defmac MODE_BASE_REG_CLASS (@var{mode})
2390 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2391 the selection of a base register in a mode dependent manner. If
2392 @var{mode} is VOIDmode then it should return the same value as
2393 @code{BASE_REG_CLASS}.
2396 @defmac INDEX_REG_CLASS
2397 A macro whose definition is the name of the class to which a valid
2398 index register must belong. An index register is one used in an
2399 address where its value is either multiplied by a scale factor or
2400 added to another register (as well as added to a displacement).
2403 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2404 For the constraint at the start of @var{str}, which starts with the letter
2405 @var{c}, return the length. This allows you to have register class /
2406 constant / extra constraints that are longer than a single letter;
2407 you don't need to define this macro if you can do with single-letter
2408 constraints only. The definition of this macro should use
2409 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2410 to handle specially.
2411 There are some sanity checks in genoutput.c that check the constraint lengths
2412 for the md file, so you can also use this macro to help you while you are
2413 transitioning from a byzantine single-letter-constraint scheme: when you
2414 return a negative length for a constraint you want to re-use, genoutput
2415 will complain about every instance where it is used in the md file.
2418 @defmac REG_CLASS_FROM_LETTER (@var{char})
2419 A C expression which defines the machine-dependent operand constraint
2420 letters for register classes. If @var{char} is such a letter, the
2421 value should be the register class corresponding to it. Otherwise,
2422 the value should be @code{NO_REGS}. The register letter @samp{r},
2423 corresponding to class @code{GENERAL_REGS}, will not be passed
2424 to this macro; you do not need to handle it.
2427 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2428 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2429 passed in @var{str}, so that you can use suffixes to distinguish between
2433 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2434 A C expression which is nonzero if register number @var{num} is
2435 suitable for use as a base register in operand addresses. It may be
2436 either a suitable hard register or a pseudo register that has been
2437 allocated such a hard register.
2440 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2441 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2442 that expression may examine the mode of the memory reference in
2443 @var{mode}. You should define this macro if the mode of the memory
2444 reference affects whether a register may be used as a base register. If
2445 you define this macro, the compiler will use it instead of
2446 @code{REGNO_OK_FOR_BASE_P}.
2449 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2450 A C expression which is nonzero if register number @var{num} is
2451 suitable for use as an index register in operand addresses. It may be
2452 either a suitable hard register or a pseudo register that has been
2453 allocated such a hard register.
2455 The difference between an index register and a base register is that
2456 the index register may be scaled. If an address involves the sum of
2457 two registers, neither one of them scaled, then either one may be
2458 labeled the ``base'' and the other the ``index''; but whichever
2459 labeling is used must fit the machine's constraints of which registers
2460 may serve in each capacity. The compiler will try both labelings,
2461 looking for one that is valid, and will reload one or both registers
2462 only if neither labeling works.
2465 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2466 A C expression that places additional restrictions on the register class
2467 to use when it is necessary to copy value @var{x} into a register in class
2468 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2469 another, smaller class. On many machines, the following definition is
2473 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2476 Sometimes returning a more restrictive class makes better code. For
2477 example, on the 68000, when @var{x} is an integer constant that is in range
2478 for a @samp{moveq} instruction, the value of this macro is always
2479 @code{DATA_REGS} as long as @var{class} includes the data registers.
2480 Requiring a data register guarantees that a @samp{moveq} will be used.
2482 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2483 @var{class} is if @var{x} is a legitimate constant which cannot be
2484 loaded into some register class. By returning @code{NO_REGS} you can
2485 force @var{x} into a memory location. For example, rs6000 can load
2486 immediate values into general-purpose registers, but does not have an
2487 instruction for loading an immediate value into a floating-point
2488 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2489 @var{x} is a floating-point constant. If the constant can't be loaded
2490 into any kind of register, code generation will be better if
2491 @code{LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2492 of using @code{PREFERRED_RELOAD_CLASS}.
2495 @defmac PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2496 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2497 input reloads. If you don't define this macro, the default is to use
2498 @var{class}, unchanged.
2501 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2502 A C expression that places additional restrictions on the register class
2503 to use when it is necessary to be able to hold a value of mode
2504 @var{mode} in a reload register for which class @var{class} would
2507 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2508 there are certain modes that simply can't go in certain reload classes.
2510 The value is a register class; perhaps @var{class}, or perhaps another,
2513 Don't define this macro unless the target machine has limitations which
2514 require the macro to do something nontrivial.
2517 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2518 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2519 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2520 Many machines have some registers that cannot be copied directly to or
2521 from memory or even from other types of registers. An example is the
2522 @samp{MQ} register, which on most machines, can only be copied to or
2523 from general registers, but not memory. Some machines allow copying all
2524 registers to and from memory, but require a scratch register for stores
2525 to some memory locations (e.g., those with symbolic address on the RT,
2526 and those with certain symbolic address on the SPARC when compiling
2527 PIC)@. In some cases, both an intermediate and a scratch register are
2530 You should define these macros to indicate to the reload phase that it may
2531 need to allocate at least one register for a reload in addition to the
2532 register to contain the data. Specifically, if copying @var{x} to a
2533 register @var{class} in @var{mode} requires an intermediate register,
2534 you should define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2535 largest register class all of whose registers can be used as
2536 intermediate registers or scratch registers.
2538 If copying a register @var{class} in @var{mode} to @var{x} requires an
2539 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2540 should be defined to return the largest register class required. If the
2541 requirements for input and output reloads are the same, the macro
2542 @code{SECONDARY_RELOAD_CLASS} should be used instead of defining both
2545 The values returned by these macros are often @code{GENERAL_REGS}.
2546 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2547 can be directly copied to or from a register of @var{class} in
2548 @var{mode} without requiring a scratch register. Do not define this
2549 macro if it would always return @code{NO_REGS}.
2551 If a scratch register is required (either with or without an
2552 intermediate register), you should define patterns for
2553 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2554 (@pxref{Standard Names}. These patterns, which will normally be
2555 implemented with a @code{define_expand}, should be similar to the
2556 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2559 Define constraints for the reload register and scratch register that
2560 contain a single register class. If the original reload register (whose
2561 class is @var{class}) can meet the constraint given in the pattern, the
2562 value returned by these macros is used for the class of the scratch
2563 register. Otherwise, two additional reload registers are required.
2564 Their classes are obtained from the constraints in the insn pattern.
2566 @var{x} might be a pseudo-register or a @code{subreg} of a
2567 pseudo-register, which could either be in a hard register or in memory.
2568 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2569 in memory and the hard register number if it is in a register.
2571 These macros should not be used in the case where a particular class of
2572 registers can only be copied to memory and not to another class of
2573 registers. In that case, secondary reload registers are not needed and
2574 would not be helpful. Instead, a stack location must be used to perform
2575 the copy and the @code{mov@var{m}} pattern should use memory as an
2576 intermediate storage. This case often occurs between floating-point and
2580 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2581 Certain machines have the property that some registers cannot be copied
2582 to some other registers without using memory. Define this macro on
2583 those machines to be a C expression that is nonzero if objects of mode
2584 @var{m} in registers of @var{class1} can only be copied to registers of
2585 class @var{class2} by storing a register of @var{class1} into memory
2586 and loading that memory location into a register of @var{class2}.
2588 Do not define this macro if its value would always be zero.
2591 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2592 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2593 allocates a stack slot for a memory location needed for register copies.
2594 If this macro is defined, the compiler instead uses the memory location
2595 defined by this macro.
2597 Do not define this macro if you do not define
2598 @code{SECONDARY_MEMORY_NEEDED}.
2601 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2602 When the compiler needs a secondary memory location to copy between two
2603 registers of mode @var{mode}, it normally allocates sufficient memory to
2604 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2605 load operations in a mode that many bits wide and whose class is the
2606 same as that of @var{mode}.
2608 This is right thing to do on most machines because it ensures that all
2609 bits of the register are copied and prevents accesses to the registers
2610 in a narrower mode, which some machines prohibit for floating-point
2613 However, this default behavior is not correct on some machines, such as
2614 the DEC Alpha, that store short integers in floating-point registers
2615 differently than in integer registers. On those machines, the default
2616 widening will not work correctly and you must define this macro to
2617 suppress that widening in some cases. See the file @file{alpha.h} for
2620 Do not define this macro if you do not define
2621 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2622 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2625 @defmac SMALL_REGISTER_CLASSES
2626 On some machines, it is risky to let hard registers live across arbitrary
2627 insns. Typically, these machines have instructions that require values
2628 to be in specific registers (like an accumulator), and reload will fail
2629 if the required hard register is used for another purpose across such an
2632 Define @code{SMALL_REGISTER_CLASSES} to be an expression with a nonzero
2633 value on these machines. When this macro has a nonzero value, the
2634 compiler will try to minimize the lifetime of hard registers.
2636 It is always safe to define this macro with a nonzero value, but if you
2637 unnecessarily define it, you will reduce the amount of optimizations
2638 that can be performed in some cases. If you do not define this macro
2639 with a nonzero value when it is required, the compiler will run out of
2640 spill registers and print a fatal error message. For most machines, you
2641 should not define this macro at all.
2644 @defmac CLASS_LIKELY_SPILLED_P (@var{class})
2645 A C expression whose value is nonzero if pseudos that have been assigned
2646 to registers of class @var{class} would likely be spilled because
2647 registers of @var{class} are needed for spill registers.
2649 The default value of this macro returns 1 if @var{class} has exactly one
2650 register and zero otherwise. On most machines, this default should be
2651 used. Only define this macro to some other expression if pseudos
2652 allocated by @file{local-alloc.c} end up in memory because their hard
2653 registers were needed for spill registers. If this macro returns nonzero
2654 for those classes, those pseudos will only be allocated by
2655 @file{global.c}, which knows how to reallocate the pseudo to another
2656 register. If there would not be another register available for
2657 reallocation, you should not change the definition of this macro since
2658 the only effect of such a definition would be to slow down register
2662 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2663 A C expression for the maximum number of consecutive registers
2664 of class @var{class} needed to hold a value of mode @var{mode}.
2666 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2667 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2668 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2669 @var{mode})} for all @var{regno} values in the class @var{class}.
2671 This macro helps control the handling of multiple-word values
2675 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2676 If defined, a C expression that returns nonzero for a @var{class} for which
2677 a change from mode @var{from} to mode @var{to} is invalid.
2679 For the example, loading 32-bit integer or floating-point objects into
2680 floating-point registers on the Alpha extends them to 64 bits.
2681 Therefore loading a 64-bit object and then storing it as a 32-bit object
2682 does not store the low-order 32 bits, as would be the case for a normal
2683 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2687 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2688 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2689 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2693 Three other special macros describe which operands fit which constraint
2696 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2697 A C expression that defines the machine-dependent operand constraint
2698 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2699 particular ranges of integer values. If @var{c} is one of those
2700 letters, the expression should check that @var{value}, an integer, is in
2701 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2702 not one of those letters, the value should be 0 regardless of
2706 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2707 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2708 string passed in @var{str}, so that you can use suffixes to distinguish
2709 between different variants.
2712 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2713 A C expression that defines the machine-dependent operand constraint
2714 letters that specify particular ranges of @code{const_double} values
2715 (@samp{G} or @samp{H}).
2717 If @var{c} is one of those letters, the expression should check that
2718 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2719 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2720 letters, the value should be 0 regardless of @var{value}.
2722 @code{const_double} is used for all floating-point constants and for
2723 @code{DImode} fixed-point constants. A given letter can accept either
2724 or both kinds of values. It can use @code{GET_MODE} to distinguish
2725 between these kinds.
2728 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2729 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
2730 string passed in @var{str}, so that you can use suffixes to distinguish
2731 between different variants.
2734 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
2735 A C expression that defines the optional machine-dependent constraint
2736 letters that can be used to segregate specific types of operands, usually
2737 memory references, for the target machine. Any letter that is not
2738 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
2739 @code{REG_CLASS_FROM_CONSTRAINT}
2740 may be used. Normally this macro will not be defined.
2742 If it is required for a particular target machine, it should return 1
2743 if @var{value} corresponds to the operand type represented by the
2744 constraint letter @var{c}. If @var{c} is not defined as an extra
2745 constraint, the value returned should be 0 regardless of @var{value}.
2747 For example, on the ROMP, load instructions cannot have their output
2748 in r0 if the memory reference contains a symbolic address. Constraint
2749 letter @samp{Q} is defined as representing a memory address that does
2750 @emph{not} contain a symbolic address. An alternative is specified with
2751 a @samp{Q} constraint on the input and @samp{r} on the output. The next
2752 alternative specifies @samp{m} on the input and a register class that
2753 does not include r0 on the output.
2756 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
2757 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
2758 in @var{str}, so that you can use suffixes to distinguish between different
2762 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
2763 A C expression that defines the optional machine-dependent constraint
2764 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
2765 be treated like memory constraints by the reload pass.
2767 It should return 1 if the operand type represented by the constraint
2768 at the start of @var{str}, the first letter of which is the letter @var{c},
2769 comprises a subset of all memory references including
2770 all those whose address is simply a base register. This allows the reload
2771 pass to reload an operand, if it does not directly correspond to the operand
2772 type of @var{c}, by copying its address into a base register.
2774 For example, on the S/390, some instructions do not accept arbitrary
2775 memory references, but only those that do not make use of an index
2776 register. The constraint letter @samp{Q} is defined via
2777 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
2778 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
2779 a @samp{Q} constraint can handle any memory operand, because the
2780 reload pass knows it can be reloaded by copying the memory address
2781 into a base register if required. This is analogous to the way
2782 a @samp{o} constraint can handle any memory operand.
2785 @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str})
2786 A C expression that defines the optional machine-dependent constraint
2787 letters, amongst those accepted by @code{EXTRA_CONSTRAINT} /
2788 @code{EXTRA_CONSTRAINT_STR}, that should
2789 be treated like address constraints by the reload pass.
2791 It should return 1 if the operand type represented by the constraint
2792 at the start of @var{str}, which starts with the letter @var{c}, comprises
2793 a subset of all memory addresses including
2794 all those that consist of just a base register. This allows the reload
2795 pass to reload an operand, if it does not directly correspond to the operand
2796 type of @var{str}, by copying it into a base register.
2798 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
2799 be used with the @code{address_operand} predicate. It is treated
2800 analogously to the @samp{p} constraint.
2803 @node Stack and Calling
2804 @section Stack Layout and Calling Conventions
2805 @cindex calling conventions
2807 @c prevent bad page break with this line
2808 This describes the stack layout and calling conventions.
2812 * Exception Handling::
2817 * Register Arguments::
2819 * Aggregate Return::
2827 @subsection Basic Stack Layout
2828 @cindex stack frame layout
2829 @cindex frame layout
2831 @c prevent bad page break with this line
2832 Here is the basic stack layout.
2834 @defmac STACK_GROWS_DOWNWARD
2835 Define this macro if pushing a word onto the stack moves the stack
2836 pointer to a smaller address.
2838 When we say, ``define this macro if @dots{},'' it means that the
2839 compiler checks this macro only with @code{#ifdef} so the precise
2840 definition used does not matter.
2843 @defmac STACK_PUSH_CODE
2844 This macro defines the operation used when something is pushed
2845 on the stack. In RTL, a push operation will be
2846 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
2848 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
2849 and @code{POST_INC}. Which of these is correct depends on
2850 the stack direction and on whether the stack pointer points
2851 to the last item on the stack or whether it points to the
2852 space for the next item on the stack.
2854 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
2855 defined, which is almost always right, and @code{PRE_INC} otherwise,
2856 which is often wrong.
2859 @defmac FRAME_GROWS_DOWNWARD
2860 Define this macro if the addresses of local variable slots are at negative
2861 offsets from the frame pointer.
2864 @defmac ARGS_GROW_DOWNWARD
2865 Define this macro if successive arguments to a function occupy decreasing
2866 addresses on the stack.
2869 @defmac STARTING_FRAME_OFFSET
2870 Offset from the frame pointer to the first local variable slot to be allocated.
2872 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
2873 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
2874 Otherwise, it is found by adding the length of the first slot to the
2875 value @code{STARTING_FRAME_OFFSET}.
2876 @c i'm not sure if the above is still correct.. had to change it to get
2877 @c rid of an overfull. --mew 2feb93
2880 @defmac STACK_ALIGNMENT_NEEDED
2881 Define to zero to disable final alignment of the stack during reload.
2882 The nonzero default for this macro is suitable for most ports.
2884 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
2885 is a register save block following the local block that doesn't require
2886 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
2887 stack alignment and do it in the backend.
2890 @defmac STACK_POINTER_OFFSET
2891 Offset from the stack pointer register to the first location at which
2892 outgoing arguments are placed. If not specified, the default value of
2893 zero is used. This is the proper value for most machines.
2895 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2896 the first location at which outgoing arguments are placed.
2899 @defmac FIRST_PARM_OFFSET (@var{fundecl})
2900 Offset from the argument pointer register to the first argument's
2901 address. On some machines it may depend on the data type of the
2904 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2905 the first argument's address.
2908 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
2909 Offset from the stack pointer register to an item dynamically allocated
2910 on the stack, e.g., by @code{alloca}.
2912 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
2913 length of the outgoing arguments. The default is correct for most
2914 machines. See @file{function.c} for details.
2917 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
2918 A C expression whose value is RTL representing the address in a stack
2919 frame where the pointer to the caller's frame is stored. Assume that
2920 @var{frameaddr} is an RTL expression for the address of the stack frame
2923 If you don't define this macro, the default is to return the value
2924 of @var{frameaddr}---that is, the stack frame address is also the
2925 address of the stack word that points to the previous frame.
2928 @defmac SETUP_FRAME_ADDRESSES
2929 If defined, a C expression that produces the machine-specific code to
2930 setup the stack so that arbitrary frames can be accessed. For example,
2931 on the SPARC, we must flush all of the register windows to the stack
2932 before we can access arbitrary stack frames. You will seldom need to
2936 @deftypefn {Target Hook} bool TARGET_BUILTIN_SETJMP_FRAME_VALUE ()
2937 This target hook should return an rtx that is used to store
2938 the address of the current frame into the built in @code{setjmp} buffer.
2939 The default value, @code{virtual_stack_vars_rtx}, is correct for most
2940 machines. One reason you may need to define this target hook is if
2941 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
2944 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
2945 A C expression whose value is RTL representing the value of the return
2946 address for the frame @var{count} steps up from the current frame, after
2947 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
2948 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
2949 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
2951 The value of the expression must always be the correct address when
2952 @var{count} is zero, but may be @code{NULL_RTX} if there is not way to
2953 determine the return address of other frames.
2956 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
2957 Define this if the return address of a particular stack frame is accessed
2958 from the frame pointer of the previous stack frame.
2961 @defmac INCOMING_RETURN_ADDR_RTX
2962 A C expression whose value is RTL representing the location of the
2963 incoming return address at the beginning of any function, before the
2964 prologue. This RTL is either a @code{REG}, indicating that the return
2965 value is saved in @samp{REG}, or a @code{MEM} representing a location in
2968 You only need to define this macro if you want to support call frame
2969 debugging information like that provided by DWARF 2.
2971 If this RTL is a @code{REG}, you should also define
2972 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
2975 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
2976 A C expression whose value is an integer giving a DWARF 2 column
2977 number that may be used as an alternate return column. This should
2978 be defined only if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
2979 general register, but an alternate column needs to be used for
2983 @defmac INCOMING_FRAME_SP_OFFSET
2984 A C expression whose value is an integer giving the offset, in bytes,
2985 from the value of the stack pointer register to the top of the stack
2986 frame at the beginning of any function, before the prologue. The top of
2987 the frame is defined to be the value of the stack pointer in the
2988 previous frame, just before the call instruction.
2990 You only need to define this macro if you want to support call frame
2991 debugging information like that provided by DWARF 2.
2994 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
2995 A C expression whose value is an integer giving the offset, in bytes,
2996 from the argument pointer to the canonical frame address (cfa). The
2997 final value should coincide with that calculated by
2998 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
2999 during virtual register instantiation.
3001 The default value for this macro is @code{FIRST_PARM_OFFSET (fundecl)},
3002 which is correct for most machines; in general, the arguments are found
3003 immediately before the stack frame. Note that this is not the case on
3004 some targets that save registers into the caller's frame, such as SPARC
3005 and rs6000, and so such targets need to define this macro.
3007 You only need to define this macro if the default is incorrect, and you
3008 want to support call frame debugging information like that provided by
3012 @node Exception Handling
3013 @subsection Exception Handling Support
3014 @cindex exception handling
3016 @defmac EH_RETURN_DATA_REGNO (@var{N})
3017 A C expression whose value is the @var{N}th register number used for
3018 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3019 @var{N} registers are usable.
3021 The exception handling library routines communicate with the exception
3022 handlers via a set of agreed upon registers. Ideally these registers
3023 should be call-clobbered; it is possible to use call-saved registers,
3024 but may negatively impact code size. The target must support at least
3025 2 data registers, but should define 4 if there are enough free registers.
3027 You must define this macro if you want to support call frame exception
3028 handling like that provided by DWARF 2.
3031 @defmac EH_RETURN_STACKADJ_RTX
3032 A C expression whose value is RTL representing a location in which
3033 to store a stack adjustment to be applied before function return.
3034 This is used to unwind the stack to an exception handler's call frame.
3035 It will be assigned zero on code paths that return normally.
3037 Typically this is a call-clobbered hard register that is otherwise
3038 untouched by the epilogue, but could also be a stack slot.
3040 Do not define this macro if the stack pointer is saved and restored
3041 by the regular prolog and epilog code in the call frame itself; in
3042 this case, the exception handling library routines will update the
3043 stack location to be restored in place. Otherwise, you must define
3044 this macro if you want to support call frame exception handling like
3045 that provided by DWARF 2.
3048 @defmac EH_RETURN_HANDLER_RTX
3049 A C expression whose value is RTL representing a location in which
3050 to store the address of an exception handler to which we should
3051 return. It will not be assigned on code paths that return normally.
3053 Typically this is the location in the call frame at which the normal
3054 return address is stored. For targets that return by popping an
3055 address off the stack, this might be a memory address just below
3056 the @emph{target} call frame rather than inside the current call
3057 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3058 been assigned, so it may be used to calculate the location of the
3061 Some targets have more complex requirements than storing to an
3062 address calculable during initial code generation. In that case
3063 the @code{eh_return} instruction pattern should be used instead.
3065 If you want to support call frame exception handling, you must
3066 define either this macro or the @code{eh_return} instruction pattern.
3069 @defmac RETURN_ADDR_OFFSET
3070 If defined, an integer-valued C expression for which rtl will be generated
3071 to add it to the exception handler address before it is searched in the
3072 exception handling tables, and to subtract it again from the address before
3073 using it to return to the exception handler.
3076 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3077 This macro chooses the encoding of pointers embedded in the exception
3078 handling sections. If at all possible, this should be defined such
3079 that the exception handling section will not require dynamic relocations,
3080 and so may be read-only.
3082 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3083 @var{global} is true if the symbol may be affected by dynamic relocations.
3084 The macro should return a combination of the @code{DW_EH_PE_*} defines
3085 as found in @file{dwarf2.h}.
3087 If this macro is not defined, pointers will not be encoded but
3088 represented directly.
3091 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3092 This macro allows the target to emit whatever special magic is required
3093 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3094 Generic code takes care of pc-relative and indirect encodings; this must
3095 be defined if the target uses text-relative or data-relative encodings.
3097 This is a C statement that branches to @var{done} if the format was
3098 handled. @var{encoding} is the format chosen, @var{size} is the number
3099 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3103 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs}, @var{success})
3104 This macro allows the target to add cpu and operating system specific
3105 code to the call-frame unwinder for use when there is no unwind data
3106 available. The most common reason to implement this macro is to unwind
3107 through signal frames.
3109 This macro is called from @code{uw_frame_state_for} in @file{unwind-dw2.c}
3110 and @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3111 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3112 for the address of the code being executed and @code{context->cfa} for
3113 the stack pointer value. If the frame can be decoded, the register save
3114 addresses should be updated in @var{fs} and the macro should branch to
3115 @var{success}. If the frame cannot be decoded, the macro should do
3118 For proper signal handling in Java this macro is accompanied by
3119 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3122 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3123 This macro allows the target to add operating system specific code to the
3124 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3125 usually used for signal or interrupt frames.
3127 This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}.
3128 @var{context} is an @code{_Unwind_Context};
3129 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3130 for the abi and context in the @code{.unwabi} directive. If the
3131 @code{.unwabi} directive can be handled, the register save addresses should
3132 be updated in @var{fs}.
3135 @defmac TARGET_USES_WEAK_UNWIND_INFO
3136 A C expression that evaluates to true if the target requires unwind
3137 info to be given comdat linkage. Define it to be @code{1} if comdat
3138 linkage is necessary. The default is @code{0}.
3141 @node Stack Checking
3142 @subsection Specifying How Stack Checking is Done
3144 GCC will check that stack references are within the boundaries of
3145 the stack, if the @option{-fstack-check} is specified, in one of three ways:
3149 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3150 will assume that you have arranged for stack checking to be done at
3151 appropriate places in the configuration files, e.g., in
3152 @code{TARGET_ASM_FUNCTION_PROLOGUE}. GCC will do not other special
3156 If @code{STACK_CHECK_BUILTIN} is zero and you defined a named pattern
3157 called @code{check_stack} in your @file{md} file, GCC will call that
3158 pattern with one argument which is the address to compare the stack
3159 value against. You must arrange for this pattern to report an error if
3160 the stack pointer is out of range.
3163 If neither of the above are true, GCC will generate code to periodically
3164 ``probe'' the stack pointer using the values of the macros defined below.
3167 Normally, you will use the default values of these macros, so GCC
3168 will use the third approach.
3170 @defmac STACK_CHECK_BUILTIN
3171 A nonzero value if stack checking is done by the configuration files in a
3172 machine-dependent manner. You should define this macro if stack checking
3173 is require by the ABI of your machine or if you would like to have to stack
3174 checking in some more efficient way than GCC's portable approach.
3175 The default value of this macro is zero.
3178 @defmac STACK_CHECK_PROBE_INTERVAL
3179 An integer representing the interval at which GCC must generate stack
3180 probe instructions. You will normally define this macro to be no larger
3181 than the size of the ``guard pages'' at the end of a stack area. The
3182 default value of 4096 is suitable for most systems.
3185 @defmac STACK_CHECK_PROBE_LOAD
3186 A integer which is nonzero if GCC should perform the stack probe
3187 as a load instruction and zero if GCC should use a store instruction.
3188 The default is zero, which is the most efficient choice on most systems.
3191 @defmac STACK_CHECK_PROTECT
3192 The number of bytes of stack needed to recover from a stack overflow,
3193 for languages where such a recovery is supported. The default value of
3194 75 words should be adequate for most machines.
3197 @defmac STACK_CHECK_MAX_FRAME_SIZE
3198 The maximum size of a stack frame, in bytes. GCC will generate probe
3199 instructions in non-leaf functions to ensure at least this many bytes of
3200 stack are available. If a stack frame is larger than this size, stack
3201 checking will not be reliable and GCC will issue a warning. The
3202 default is chosen so that GCC only generates one instruction on most
3203 systems. You should normally not change the default value of this macro.
3206 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3207 GCC uses this value to generate the above warning message. It
3208 represents the amount of fixed frame used by a function, not including
3209 space for any callee-saved registers, temporaries and user variables.
3210 You need only specify an upper bound for this amount and will normally
3211 use the default of four words.
3214 @defmac STACK_CHECK_MAX_VAR_SIZE
3215 The maximum size, in bytes, of an object that GCC will place in the
3216 fixed area of the stack frame when the user specifies
3217 @option{-fstack-check}.
3218 GCC computed the default from the values of the above macros and you will
3219 normally not need to override that default.
3223 @node Frame Registers
3224 @subsection Registers That Address the Stack Frame
3226 @c prevent bad page break with this line
3227 This discusses registers that address the stack frame.
3229 @defmac STACK_POINTER_REGNUM
3230 The register number of the stack pointer register, which must also be a
3231 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3232 the hardware determines which register this is.
3235 @defmac FRAME_POINTER_REGNUM
3236 The register number of the frame pointer register, which is used to
3237 access automatic variables in the stack frame. On some machines, the
3238 hardware determines which register this is. On other machines, you can
3239 choose any register you wish for this purpose.
3242 @defmac HARD_FRAME_POINTER_REGNUM
3243 On some machines the offset between the frame pointer and starting
3244 offset of the automatic variables is not known until after register
3245 allocation has been done (for example, because the saved registers are
3246 between these two locations). On those machines, define
3247 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3248 be used internally until the offset is known, and define
3249 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3250 used for the frame pointer.
3252 You should define this macro only in the very rare circumstances when it
3253 is not possible to calculate the offset between the frame pointer and
3254 the automatic variables until after register allocation has been
3255 completed. When this macro is defined, you must also indicate in your
3256 definition of @code{ELIMINABLE_REGS} how to eliminate
3257 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3258 or @code{STACK_POINTER_REGNUM}.
3260 Do not define this macro if it would be the same as
3261 @code{FRAME_POINTER_REGNUM}.
3264 @defmac ARG_POINTER_REGNUM
3265 The register number of the arg pointer register, which is used to access
3266 the function's argument list. On some machines, this is the same as the
3267 frame pointer register. On some machines, the hardware determines which
3268 register this is. On other machines, you can choose any register you
3269 wish for this purpose. If this is not the same register as the frame
3270 pointer register, then you must mark it as a fixed register according to
3271 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3272 (@pxref{Elimination}).
3275 @defmac RETURN_ADDRESS_POINTER_REGNUM
3276 The register number of the return address pointer register, which is used to
3277 access the current function's return address from the stack. On some
3278 machines, the return address is not at a fixed offset from the frame
3279 pointer or stack pointer or argument pointer. This register can be defined
3280 to point to the return address on the stack, and then be converted by
3281 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3283 Do not define this macro unless there is no other way to get the return
3284 address from the stack.
3287 @defmac STATIC_CHAIN_REGNUM
3288 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3289 Register numbers used for passing a function's static chain pointer. If
3290 register windows are used, the register number as seen by the called
3291 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3292 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3293 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3296 The static chain register need not be a fixed register.
3298 If the static chain is passed in memory, these macros should not be
3299 defined; instead, the next two macros should be defined.
3302 @defmac STATIC_CHAIN
3303 @defmacx STATIC_CHAIN_INCOMING
3304 If the static chain is passed in memory, these macros provide rtx giving
3305 @code{mem} expressions that denote where they are stored.
3306 @code{STATIC_CHAIN} and @code{STATIC_CHAIN_INCOMING} give the locations
3307 as seen by the calling and called functions, respectively. Often the former
3308 will be at an offset from the stack pointer and the latter at an offset from
3311 @findex stack_pointer_rtx
3312 @findex frame_pointer_rtx
3313 @findex arg_pointer_rtx
3314 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3315 @code{arg_pointer_rtx} will have been initialized prior to the use of these
3316 macros and should be used to refer to those items.
3318 If the static chain is passed in a register, the two previous macros should
3322 @defmac DWARF_FRAME_REGISTERS
3323 This macro specifies the maximum number of hard registers that can be
3324 saved in a call frame. This is used to size data structures used in
3325 DWARF2 exception handling.
3327 Prior to GCC 3.0, this macro was needed in order to establish a stable
3328 exception handling ABI in the face of adding new hard registers for ISA
3329 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3330 in the number of hard registers. Nevertheless, this macro can still be
3331 used to reduce the runtime memory requirements of the exception handling
3332 routines, which can be substantial if the ISA contains a lot of
3333 registers that are not call-saved.
3335 If this macro is not defined, it defaults to
3336 @code{FIRST_PSEUDO_REGISTER}.
3339 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3341 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3342 for backward compatibility in pre GCC 3.0 compiled code.
3344 If this macro is not defined, it defaults to
3345 @code{DWARF_FRAME_REGISTERS}.
3348 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3350 Define this macro if the target's representation for dwarf registers
3351 is different than the internal representation for unwind column.
3352 Given a dwarf register, this macro should return the internal unwind
3353 column number to use instead.
3355 See the PowerPC's SPE target for an example.
3358 @defmac DWARF_FRAME_REGNUM (@var{regno})
3360 Define this macro if the target's representation for dwarf registers
3361 used in .eh_frame or .debug_frame is different from that used in other
3362 debug info sections. Given a GCC hard register number, this macro
3363 should return the .eh_frame register number. The default is
3364 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3368 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3370 Define this macro to map register numbers held in the call frame info
3371 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3372 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3373 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3374 return @code{@var{regno}}.
3379 @subsection Eliminating Frame Pointer and Arg Pointer
3381 @c prevent bad page break with this line
3382 This is about eliminating the frame pointer and arg pointer.
3384 @defmac FRAME_POINTER_REQUIRED
3385 A C expression which is nonzero if a function must have and use a frame
3386 pointer. This expression is evaluated in the reload pass. If its value is
3387 nonzero the function will have a frame pointer.
3389 The expression can in principle examine the current function and decide
3390 according to the facts, but on most machines the constant 0 or the
3391 constant 1 suffices. Use 0 when the machine allows code to be generated
3392 with no frame pointer, and doing so saves some time or space. Use 1
3393 when there is no possible advantage to avoiding a frame pointer.
3395 In certain cases, the compiler does not know how to produce valid code
3396 without a frame pointer. The compiler recognizes those cases and
3397 automatically gives the function a frame pointer regardless of what
3398 @code{FRAME_POINTER_REQUIRED} says. You don't need to worry about
3401 In a function that does not require a frame pointer, the frame pointer
3402 register can be allocated for ordinary usage, unless you mark it as a
3403 fixed register. See @code{FIXED_REGISTERS} for more information.
3406 @findex get_frame_size
3407 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3408 A C statement to store in the variable @var{depth-var} the difference
3409 between the frame pointer and the stack pointer values immediately after
3410 the function prologue. The value would be computed from information
3411 such as the result of @code{get_frame_size ()} and the tables of
3412 registers @code{regs_ever_live} and @code{call_used_regs}.
3414 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3415 need not be defined. Otherwise, it must be defined even if
3416 @code{FRAME_POINTER_REQUIRED} is defined to always be true; in that
3417 case, you may set @var{depth-var} to anything.
3420 @defmac ELIMINABLE_REGS
3421 If defined, this macro specifies a table of register pairs used to
3422 eliminate unneeded registers that point into the stack frame. If it is not
3423 defined, the only elimination attempted by the compiler is to replace
3424 references to the frame pointer with references to the stack pointer.
3426 The definition of this macro is a list of structure initializations, each
3427 of which specifies an original and replacement register.
3429 On some machines, the position of the argument pointer is not known until
3430 the compilation is completed. In such a case, a separate hard register
3431 must be used for the argument pointer. This register can be eliminated by
3432 replacing it with either the frame pointer or the argument pointer,
3433 depending on whether or not the frame pointer has been eliminated.
3435 In this case, you might specify:
3437 #define ELIMINABLE_REGS \
3438 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3439 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3440 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3443 Note that the elimination of the argument pointer with the stack pointer is
3444 specified first since that is the preferred elimination.
3447 @defmac CAN_ELIMINATE (@var{from-reg}, @var{to-reg})
3448 A C expression that returns nonzero if the compiler is allowed to try
3449 to replace register number @var{from-reg} with register number
3450 @var{to-reg}. This macro need only be defined if @code{ELIMINABLE_REGS}
3451 is defined, and will usually be the constant 1, since most of the cases
3452 preventing register elimination are things that the compiler already
3456 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3457 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3458 specifies the initial difference between the specified pair of
3459 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3463 @node Stack Arguments
3464 @subsection Passing Function Arguments on the Stack
3465 @cindex arguments on stack
3466 @cindex stack arguments
3468 The macros in this section control how arguments are passed
3469 on the stack. See the following section for other macros that
3470 control passing certain arguments in registers.
3472 @deftypefn {Target Hook} bool TARGET_PROMOTE_PROTOTYPES (tree @var{fntype})
3473 This target hook returns @code{true} if an argument declared in a
3474 prototype as an integral type smaller than @code{int} should actually be
3475 passed as an @code{int}. In addition to avoiding errors in certain
3476 cases of mismatch, it also makes for better code on certain machines.
3477 The default is to not promote prototypes.
3481 A C expression. If nonzero, push insns will be used to pass
3483 If the target machine does not have a push instruction, set it to zero.
3484 That directs GCC to use an alternate strategy: to
3485 allocate the entire argument block and then store the arguments into
3486 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3489 @defmac PUSH_ARGS_REVERSED
3490 A C expression. If nonzero, function arguments will be evaluated from
3491 last to first, rather than from first to last. If this macro is not
3492 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3493 and args grow in opposite directions, and 0 otherwise.
3496 @defmac PUSH_ROUNDING (@var{npushed})
3497 A C expression that is the number of bytes actually pushed onto the
3498 stack when an instruction attempts to push @var{npushed} bytes.
3500 On some machines, the definition
3503 #define PUSH_ROUNDING(BYTES) (BYTES)
3507 will suffice. But on other machines, instructions that appear
3508 to push one byte actually push two bytes in an attempt to maintain
3509 alignment. Then the definition should be
3512 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3516 @findex current_function_outgoing_args_size
3517 @defmac ACCUMULATE_OUTGOING_ARGS
3518 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3519 will be computed and placed into the variable
3520 @code{current_function_outgoing_args_size}. No space will be pushed
3521 onto the stack for each call; instead, the function prologue should
3522 increase the stack frame size by this amount.
3524 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3528 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3529 Define this macro if functions should assume that stack space has been
3530 allocated for arguments even when their values are passed in
3533 The value of this macro is the size, in bytes, of the area reserved for
3534 arguments passed in registers for the function represented by @var{fndecl},
3535 which can be zero if GCC is calling a library function.
3537 This space can be allocated by the caller, or be a part of the
3538 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3541 @c above is overfull. not sure what to do. --mew 5feb93 did
3542 @c something, not sure if it looks good. --mew 10feb93
3544 @defmac OUTGOING_REG_PARM_STACK_SPACE
3545 Define this if it is the responsibility of the caller to allocate the area
3546 reserved for arguments passed in registers.
3548 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3549 whether the space for these arguments counts in the value of
3550 @code{current_function_outgoing_args_size}.
3553 @defmac STACK_PARMS_IN_REG_PARM_AREA
3554 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3555 stack parameters don't skip the area specified by it.
3556 @c i changed this, makes more sens and it should have taken care of the
3557 @c overfull.. not as specific, tho. --mew 5feb93
3559 Normally, when a parameter is not passed in registers, it is placed on the
3560 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3561 suppresses this behavior and causes the parameter to be passed on the
3562 stack in its natural location.
3565 @defmac RETURN_POPS_ARGS (@var{fundecl}, @var{funtype}, @var{stack-size})
3566 A C expression that should indicate the number of bytes of its own
3567 arguments that a function pops on returning, or 0 if the
3568 function pops no arguments and the caller must therefore pop them all
3569 after the function returns.
3571 @var{fundecl} is a C variable whose value is a tree node that describes
3572 the function in question. Normally it is a node of type
3573 @code{FUNCTION_DECL} that describes the declaration of the function.
3574 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3576 @var{funtype} is a C variable whose value is a tree node that
3577 describes the function in question. Normally it is a node of type
3578 @code{FUNCTION_TYPE} that describes the data type of the function.
3579 From this it is possible to obtain the data types of the value and
3580 arguments (if known).
3582 When a call to a library function is being considered, @var{fundecl}
3583 will contain an identifier node for the library function. Thus, if
3584 you need to distinguish among various library functions, you can do so
3585 by their names. Note that ``library function'' in this context means
3586 a function used to perform arithmetic, whose name is known specially
3587 in the compiler and was not mentioned in the C code being compiled.
3589 @var{stack-size} is the number of bytes of arguments passed on the
3590 stack. If a variable number of bytes is passed, it is zero, and
3591 argument popping will always be the responsibility of the calling function.
3593 On the VAX, all functions always pop their arguments, so the definition
3594 of this macro is @var{stack-size}. On the 68000, using the standard
3595 calling convention, no functions pop their arguments, so the value of
3596 the macro is always 0 in this case. But an alternative calling
3597 convention is available in which functions that take a fixed number of
3598 arguments pop them but other functions (such as @code{printf}) pop
3599 nothing (the caller pops all). When this convention is in use,
3600 @var{funtype} is examined to determine whether a function takes a fixed
3601 number of arguments.
3604 @defmac CALL_POPS_ARGS (@var{cum})
3605 A C expression that should indicate the number of bytes a call sequence
3606 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3607 when compiling a function call.
3609 @var{cum} is the variable in which all arguments to the called function
3610 have been accumulated.
3612 On certain architectures, such as the SH5, a call trampoline is used
3613 that pops certain registers off the stack, depending on the arguments
3614 that have been passed to the function. Since this is a property of the
3615 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3619 @node Register Arguments
3620 @subsection Passing Arguments in Registers
3621 @cindex arguments in registers
3622 @cindex registers arguments
3624 This section describes the macros which let you control how various
3625 types of arguments are passed in registers or how they are arranged in
3628 @defmac FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3629 A C expression that controls whether a function argument is passed
3630 in a register, and which register.
3632 The arguments are @var{cum}, which summarizes all the previous
3633 arguments; @var{mode}, the machine mode of the argument; @var{type},
3634 the data type of the argument as a tree node or 0 if that is not known
3635 (which happens for C support library functions); and @var{named},
3636 which is 1 for an ordinary argument and 0 for nameless arguments that
3637 correspond to @samp{@dots{}} in the called function's prototype.
3638 @var{type} can be an incomplete type if a syntax error has previously
3641 The value of the expression is usually either a @code{reg} RTX for the
3642 hard register in which to pass the argument, or zero to pass the
3643 argument on the stack.
3645 For machines like the VAX and 68000, where normally all arguments are
3646 pushed, zero suffices as a definition.
3648 The value of the expression can also be a @code{parallel} RTX@. This is
3649 used when an argument is passed in multiple locations. The mode of the
3650 @code{parallel} should be the mode of the entire argument. The
3651 @code{parallel} holds any number of @code{expr_list} pairs; each one
3652 describes where part of the argument is passed. In each
3653 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
3654 register in which to pass this part of the argument, and the mode of the
3655 register RTX indicates how large this part of the argument is. The
3656 second operand of the @code{expr_list} is a @code{const_int} which gives
3657 the offset in bytes into the entire argument of where this part starts.
3658 As a special exception the first @code{expr_list} in the @code{parallel}
3659 RTX may have a first operand of zero. This indicates that the entire
3660 argument is also stored on the stack.
3662 The last time this macro is called, it is called with @code{MODE ==
3663 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
3664 pattern as operands 2 and 3 respectively.
3666 @cindex @file{stdarg.h} and register arguments
3667 The usual way to make the ISO library @file{stdarg.h} work on a machine
3668 where some arguments are usually passed in registers, is to cause
3669 nameless arguments to be passed on the stack instead. This is done
3670 by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
3672 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
3673 @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
3674 You may use the hook @code{targetm.calls.must_pass_in_stack}
3675 in the definition of this macro to determine if this argument is of a
3676 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
3677 is not defined and @code{FUNCTION_ARG} returns nonzero for such an
3678 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
3679 defined, the argument will be computed in the stack and then loaded into
3683 @deftypefn {Target Hook} bool TARGET_MUST_PASS_IN_STACK (enum machine_mode @var{mode}, tree @var{type})
3684 This target hook should return @code{true} if we should not pass @var{type}
3685 solely in registers. The file @file{expr.h} defines a
3686 definition that is usually appropriate, refer to @file{expr.h} for additional
3690 @defmac FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3691 Define this macro if the target machine has ``register windows'', so
3692 that the register in which a function sees an arguments is not
3693 necessarily the same as the one in which the caller passed the
3696 For such machines, @code{FUNCTION_ARG} computes the register in which
3697 the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
3698 be defined in a similar fashion to tell the function being called
3699 where the arguments will arrive.
3701 If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
3702 serves both purposes.
3705 @defmac FUNCTION_ARG_PARTIAL_NREGS (@var{cum}, @var{mode}, @var{type}, @var{named})
3706 A C expression for the number of words, at the beginning of an
3707 argument, that must be put in registers. The value must be zero for
3708 arguments that are passed entirely in registers or that are entirely
3709 pushed on the stack.
3711 On some machines, certain arguments must be passed partially in
3712 registers and partially in memory. On these machines, typically the
3713 first @var{n} words of arguments are passed in registers, and the rest
3714 on the stack. If a multi-word argument (a @code{double} or a
3715 structure) crosses that boundary, its first few words must be passed
3716 in registers and the rest must be pushed. This macro tells the
3717 compiler when this occurs, and how many of the words should go in
3720 @code{FUNCTION_ARG} for these arguments should return the first
3721 register to be used by the caller for this argument; likewise
3722 @code{FUNCTION_INCOMING_ARG}, for the called function.
3725 @deftypefn {Target Hook} bool TARGET_PASS_BY_REFERENCE (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
3726 This target hook should return @code{true} if an argument at the
3727 position indicated by @var{cum} should be passed by reference. This
3728 predicate is queried after target independent reasons for being
3729 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
3731 If the hook returns true, a copy of that argument is made in memory and a
3732 pointer to the argument is passed instead of the argument itself.
3733 The pointer is passed in whatever way is appropriate for passing a pointer
3737 @defmac FUNCTION_ARG_CALLEE_COPIES (@var{cum}, @var{mode}, @var{type}, @var{named})
3738 If defined, a C expression that indicates when it is the called function's
3739 responsibility to make a copy of arguments passed by invisible reference.
3740 Normally, the caller makes a copy and passes the address of the copy to the
3741 routine being called. When @code{FUNCTION_ARG_CALLEE_COPIES} is defined and is
3742 nonzero, the caller does not make a copy. Instead, it passes a pointer to the
3743 ``live'' value. The called function must not modify this value. If it can be
3744 determined that the value won't be modified, it need not make a copy;
3745 otherwise a copy must be made.
3748 @defmac CUMULATIVE_ARGS
3749 A C type for declaring a variable that is used as the first argument of
3750 @code{FUNCTION_ARG} and other related values. For some target machines,
3751 the type @code{int} suffices and can hold the number of bytes of
3754 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
3755 arguments that have been passed on the stack. The compiler has other
3756 variables to keep track of that. For target machines on which all
3757 arguments are passed on the stack, there is no need to store anything in
3758 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
3759 should not be empty, so use @code{int}.
3762 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
3763 A C statement (sans semicolon) for initializing the variable
3764 @var{cum} for the state at the beginning of the argument list. The
3765 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
3766 is the tree node for the data type of the function which will receive
3767 the args, or 0 if the args are to a compiler support library function.
3768 For direct calls that are not libcalls, @var{fndecl} contain the
3769 declaration node of the function. @var{fndecl} is also set when
3770 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
3771 being compiled. @var{n_named_args} is set to the number of named
3772 arguments, including a structure return address if it is passed as a
3773 parameter, when making a call. When processing incoming arguments,
3774 @var{n_named_args} is set to -1.
3776 When processing a call to a compiler support library function,
3777 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
3778 contains the name of the function, as a string. @var{libname} is 0 when
3779 an ordinary C function call is being processed. Thus, each time this
3780 macro is called, either @var{libname} or @var{fntype} is nonzero, but
3781 never both of them at once.
3784 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
3785 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
3786 it gets a @code{MODE} argument instead of @var{fntype}, that would be
3787 @code{NULL}. @var{indirect} would always be zero, too. If this macro
3788 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
3789 0)} is used instead.
3792 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
3793 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
3794 finding the arguments for the function being compiled. If this macro is
3795 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
3797 The value passed for @var{libname} is always 0, since library routines
3798 with special calling conventions are never compiled with GCC@. The
3799 argument @var{libname} exists for symmetry with
3800 @code{INIT_CUMULATIVE_ARGS}.
3801 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
3802 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
3805 @defmac FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
3806 A C statement (sans semicolon) to update the summarizer variable
3807 @var{cum} to advance past an argument in the argument list. The
3808 values @var{mode}, @var{type} and @var{named} describe that argument.
3809 Once this is done, the variable @var{cum} is suitable for analyzing
3810 the @emph{following} argument with @code{FUNCTION_ARG}, etc.
3812 This macro need not do anything if the argument in question was passed
3813 on the stack. The compiler knows how to track the amount of stack space
3814 used for arguments without any special help.
3817 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
3818 If defined, a C expression which determines whether, and in which direction,
3819 to pad out an argument with extra space. The value should be of type
3820 @code{enum direction}: either @code{upward} to pad above the argument,
3821 @code{downward} to pad below, or @code{none} to inhibit padding.
3823 The @emph{amount} of padding is always just enough to reach the next
3824 multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control
3827 This macro has a default definition which is right for most systems.
3828 For little-endian machines, the default is to pad upward. For
3829 big-endian machines, the default is to pad downward for an argument of
3830 constant size shorter than an @code{int}, and upward otherwise.
3833 @defmac PAD_VARARGS_DOWN
3834 If defined, a C expression which determines whether the default
3835 implementation of va_arg will attempt to pad down before reading the
3836 next argument, if that argument is smaller than its aligned space as
3837 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
3838 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
3841 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
3842 Specify padding for the last element of a block move between registers and
3843 memory. @var{first} is nonzero if this is the only element. Defining this
3844 macro allows better control of register function parameters on big-endian
3845 machines, without using @code{PARALLEL} rtl. In particular,
3846 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
3847 registers, as there is no longer a "wrong" part of a register; For example,
3848 a three byte aggregate may be passed in the high part of a register if so
3852 @defmac FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
3853 If defined, a C expression that gives the alignment boundary, in bits,
3854 of an argument with the specified mode and type. If it is not defined,
3855 @code{PARM_BOUNDARY} is used for all arguments.
3858 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
3859 A C expression that is nonzero if @var{regno} is the number of a hard
3860 register in which function arguments are sometimes passed. This does
3861 @emph{not} include implicit arguments such as the static chain and
3862 the structure-value address. On many machines, no registers can be
3863 used for this purpose since all function arguments are pushed on the
3867 @deftypefn {Target Hook} bool TARGET_SPLIT_COMPLEX_ARG (tree @var{type})
3868 This hook should return true if parameter of type @var{type} are passed
3869 as two scalar parameters. By default, GCC will attempt to pack complex
3870 arguments into the target's word size. Some ABIs require complex arguments
3871 to be split and treated as their individual components. For example, on
3872 AIX64, complex floats should be passed in a pair of floating point
3873 registers, even though a complex float would fit in one 64-bit floating
3876 The default value of this hook is @code{NULL}, which is treated as always
3880 @deftypefn {Target Hook} tree TARGET_GIMPLIFY_VA_ARG_EXPR (tree @var{valist}, tree @var{type}, tree *@var{pre_p}, tree *@var{post_p})
3881 This hook performs target-specific gimplification of
3882 @code{VA_ARG_EXPR}. The first two parameters correspond to the
3883 arguments to @code{va_arg}; the latter two are as in
3884 @code{gimplify.c:gimplify_expr}.
3888 @subsection How Scalar Function Values Are Returned
3889 @cindex return values in registers
3890 @cindex values, returned by functions
3891 @cindex scalars, returned as values
3893 This section discusses the macros that control returning scalars as
3894 values---values that can fit in registers.
3896 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
3897 A C expression to create an RTX representing the place where a
3898 function returns a value of data type @var{valtype}. @var{valtype} is
3899 a tree node representing a data type. Write @code{TYPE_MODE
3900 (@var{valtype})} to get the machine mode used to represent that type.
3901 On many machines, only the mode is relevant. (Actually, on most
3902 machines, scalar values are returned in the same place regardless of
3905 The value of the expression is usually a @code{reg} RTX for the hard
3906 register where the return value is stored. The value can also be a
3907 @code{parallel} RTX, if the return value is in multiple places. See
3908 @code{FUNCTION_ARG} for an explanation of the @code{parallel} form.
3910 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply the same
3911 promotion rules specified in @code{PROMOTE_MODE} if @var{valtype} is a
3914 If the precise function being called is known, @var{func} is a tree
3915 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
3916 pointer. This makes it possible to use a different value-returning
3917 convention for specific functions when all their calls are
3920 @code{FUNCTION_VALUE} is not used for return vales with aggregate data
3921 types, because these are returned in another way. See
3922 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
3925 @defmac FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func})
3926 Define this macro if the target machine has ``register windows''
3927 so that the register in which a function returns its value is not
3928 the same as the one in which the caller sees the value.
3930 For such machines, @code{FUNCTION_VALUE} computes the register in which
3931 the caller will see the value. @code{FUNCTION_OUTGOING_VALUE} should be
3932 defined in a similar fashion to tell the function where to put the
3935 If @code{FUNCTION_OUTGOING_VALUE} is not defined,
3936 @code{FUNCTION_VALUE} serves both purposes.
3938 @code{FUNCTION_OUTGOING_VALUE} is not used for return vales with
3939 aggregate data types, because these are returned in another way. See
3940 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
3943 @defmac LIBCALL_VALUE (@var{mode})
3944 A C expression to create an RTX representing the place where a library
3945 function returns a value of mode @var{mode}. If the precise function
3946 being called is known, @var{func} is a tree node
3947 (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
3948 pointer. This makes it possible to use a different value-returning
3949 convention for specific functions when all their calls are
3952 Note that ``library function'' in this context means a compiler
3953 support routine, used to perform arithmetic, whose name is known
3954 specially by the compiler and was not mentioned in the C code being
3957 The definition of @code{LIBRARY_VALUE} need not be concerned aggregate
3958 data types, because none of the library functions returns such types.
3961 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
3962 A C expression that is nonzero if @var{regno} is the number of a hard
3963 register in which the values of called function may come back.
3965 A register whose use for returning values is limited to serving as the
3966 second of a pair (for a value of type @code{double}, say) need not be
3967 recognized by this macro. So for most machines, this definition
3971 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
3974 If the machine has register windows, so that the caller and the called
3975 function use different registers for the return value, this macro
3976 should recognize only the caller's register numbers.
3979 @defmac APPLY_RESULT_SIZE
3980 Define this macro if @samp{untyped_call} and @samp{untyped_return}
3981 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
3982 saving and restoring an arbitrary return value.
3985 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MSB (tree @var{type})
3986 This hook should return true if values of type @var{type} are returned
3987 at the most significant end of a register (in other words, if they are
3988 padded at the least significant end). You can assume that @var{type}
3989 is returned in a register; the caller is required to check this.
3991 Note that the register provided by @code{FUNCTION_VALUE} must be able
3992 to hold the complete return value. For example, if a 1-, 2- or 3-byte
3993 structure is returned at the most significant end of a 4-byte register,
3994 @code{FUNCTION_VALUE} should provide an @code{SImode} rtx.
3997 @node Aggregate Return
3998 @subsection How Large Values Are Returned
3999 @cindex aggregates as return values
4000 @cindex large return values
4001 @cindex returning aggregate values
4002 @cindex structure value address
4004 When a function value's mode is @code{BLKmode} (and in some other
4005 cases), the value is not returned according to @code{FUNCTION_VALUE}
4006 (@pxref{Scalar Return}). Instead, the caller passes the address of a
4007 block of memory in which the value should be stored. This address
4008 is called the @dfn{structure value address}.
4010 This section describes how to control returning structure values in
4013 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MEMORY (tree @var{type}, tree @var{fntype})
4014 This target hook should return a nonzero value to say to return the
4015 function value in memory, just as large structures are always returned.
4016 Here @var{type} will be the data type of the value, and @var{fntype}
4017 will be the type of the function doing the returning, or @code{NULL} for
4020 Note that values of mode @code{BLKmode} must be explicitly handled
4021 by this function. Also, the option @option{-fpcc-struct-return}
4022 takes effect regardless of this macro. On most systems, it is
4023 possible to leave the hook undefined; this causes a default
4024 definition to be used, whose value is the constant 1 for @code{BLKmode}
4025 values, and 0 otherwise.
4027 Do not use this hook to indicate that structures and unions should always
4028 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4032 @defmac DEFAULT_PCC_STRUCT_RETURN
4033 Define this macro to be 1 if all structure and union return values must be
4034 in memory. Since this results in slower code, this should be defined
4035 only if needed for compatibility with other compilers or with an ABI@.
4036 If you define this macro to be 0, then the conventions used for structure
4037 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4040 If not defined, this defaults to the value 1.
4043 @deftypefn {Target Hook} rtx TARGET_STRUCT_VALUE_RTX (tree @var{fndecl}, int @var{incoming})
4044 This target hook should return the location of the structure value
4045 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4046 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4047 be @code{NULL}, for libcalls. You do not need to define this target
4048 hook if the address is always passed as an ``invisible'' first
4051 On some architectures the place where the structure value address
4052 is found by the called function is not the same place that the
4053 caller put it. This can be due to register windows, or it could
4054 be because the function prologue moves it to a different place.
4055 @var{incoming} is @code{true} when the location is needed in
4056 the context of the called function, and @code{false} in the context of
4059 If @var{incoming} is @code{true} and the address is to be found on the
4060 stack, return a @code{mem} which refers to the frame pointer.
4063 @defmac PCC_STATIC_STRUCT_RETURN
4064 Define this macro if the usual system convention on the target machine
4065 for returning structures and unions is for the called function to return
4066 the address of a static variable containing the value.
4068 Do not define this if the usual system convention is for the caller to
4069 pass an address to the subroutine.
4071 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4072 nothing when you use @option{-freg-struct-return} mode.
4076 @subsection Caller-Saves Register Allocation
4078 If you enable it, GCC can save registers around function calls. This
4079 makes it possible to use call-clobbered registers to hold variables that
4080 must live across calls.
4082 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4083 A C expression to determine whether it is worthwhile to consider placing
4084 a pseudo-register in a call-clobbered hard register and saving and
4085 restoring it around each function call. The expression should be 1 when
4086 this is worth doing, and 0 otherwise.
4088 If you don't define this macro, a default is used which is good on most
4089 machines: @code{4 * @var{calls} < @var{refs}}.
4092 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4093 A C expression specifying which mode is required for saving @var{nregs}
4094 of a pseudo-register in call-clobbered hard register @var{regno}. If
4095 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4096 returned. For most machines this macro need not be defined since GCC
4097 will select the smallest suitable mode.
4100 @node Function Entry
4101 @subsection Function Entry and Exit
4102 @cindex function entry and exit
4106 This section describes the macros that output function entry
4107 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4109 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4110 If defined, a function that outputs the assembler code for entry to a
4111 function. The prologue is responsible for setting up the stack frame,
4112 initializing the frame pointer register, saving registers that must be
4113 saved, and allocating @var{size} additional bytes of storage for the
4114 local variables. @var{size} is an integer. @var{file} is a stdio
4115 stream to which the assembler code should be output.
4117 The label for the beginning of the function need not be output by this
4118 macro. That has already been done when the macro is run.
4120 @findex regs_ever_live
4121 To determine which registers to save, the macro can refer to the array
4122 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4123 @var{r} is used anywhere within the function. This implies the function
4124 prologue should save register @var{r}, provided it is not one of the
4125 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4126 @code{regs_ever_live}.)
4128 On machines that have ``register windows'', the function entry code does
4129 not save on the stack the registers that are in the windows, even if
4130 they are supposed to be preserved by function calls; instead it takes
4131 appropriate steps to ``push'' the register stack, if any non-call-used
4132 registers are used in the function.
4134 @findex frame_pointer_needed
4135 On machines where functions may or may not have frame-pointers, the
4136 function entry code must vary accordingly; it must set up the frame
4137 pointer if one is wanted, and not otherwise. To determine whether a
4138 frame pointer is in wanted, the macro can refer to the variable
4139 @code{frame_pointer_needed}. The variable's value will be 1 at run
4140 time in a function that needs a frame pointer. @xref{Elimination}.
4142 The function entry code is responsible for allocating any stack space
4143 required for the function. This stack space consists of the regions
4144 listed below. In most cases, these regions are allocated in the
4145 order listed, with the last listed region closest to the top of the
4146 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4147 the highest address if it is not defined). You can use a different order
4148 for a machine if doing so is more convenient or required for
4149 compatibility reasons. Except in cases where required by standard
4150 or by a debugger, there is no reason why the stack layout used by GCC
4151 need agree with that used by other compilers for a machine.
4154 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
4155 If defined, a function that outputs assembler code at the end of a
4156 prologue. This should be used when the function prologue is being
4157 emitted as RTL, and you have some extra assembler that needs to be
4158 emitted. @xref{prologue instruction pattern}.
4161 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
4162 If defined, a function that outputs assembler code at the start of an
4163 epilogue. This should be used when the function epilogue is being
4164 emitted as RTL, and you have some extra assembler that needs to be
4165 emitted. @xref{epilogue instruction pattern}.
4168 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4169 If defined, a function that outputs the assembler code for exit from a
4170 function. The epilogue is responsible for restoring the saved
4171 registers and stack pointer to their values when the function was
4172 called, and returning control to the caller. This macro takes the
4173 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4174 registers to restore are determined from @code{regs_ever_live} and
4175 @code{CALL_USED_REGISTERS} in the same way.
4177 On some machines, there is a single instruction that does all the work
4178 of returning from the function. On these machines, give that
4179 instruction the name @samp{return} and do not define the macro
4180 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4182 Do not define a pattern named @samp{return} if you want the
4183 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4184 switches to control whether return instructions or epilogues are used,
4185 define a @samp{return} pattern with a validity condition that tests the
4186 target switches appropriately. If the @samp{return} pattern's validity
4187 condition is false, epilogues will be used.
4189 On machines where functions may or may not have frame-pointers, the
4190 function exit code must vary accordingly. Sometimes the code for these
4191 two cases is completely different. To determine whether a frame pointer
4192 is wanted, the macro can refer to the variable
4193 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4194 a function that needs a frame pointer.
4196 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4197 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4198 The C variable @code{current_function_is_leaf} is nonzero for such a
4199 function. @xref{Leaf Functions}.
4201 On some machines, some functions pop their arguments on exit while
4202 others leave that for the caller to do. For example, the 68020 when
4203 given @option{-mrtd} pops arguments in functions that take a fixed
4204 number of arguments.
4206 @findex current_function_pops_args
4207 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4208 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4209 needs to know what was decided. The variable that is called
4210 @code{current_function_pops_args} is the number of bytes of its
4211 arguments that a function should pop. @xref{Scalar Return}.
4212 @c what is the "its arguments" in the above sentence referring to, pray
4213 @c tell? --mew 5feb93
4216 @deftypefn {Target Hook} bool TARGET_LATE_RTL_PROLOGUE_EPILOGUE
4217 If set to @code{true}, it instructs the compiler to emit the RTL prologue
4218 and epilogue later in the game than usual, namely after all passes that
4219 modify the instructions (and not merely reorder them) have been run. In
4220 particular, the C variable @code{current_function_uses_only_leaf_regs} is
4221 valid at that point. This can be used on machines that have "register
4222 windows" to optimize away the regular "push" on the register stack.
4223 @xref{Leaf Functions}.
4228 @findex current_function_pretend_args_size
4229 A region of @code{current_function_pretend_args_size} bytes of
4230 uninitialized space just underneath the first argument arriving on the
4231 stack. (This may not be at the very start of the allocated stack region
4232 if the calling sequence has pushed anything else since pushing the stack
4233 arguments. But usually, on such machines, nothing else has been pushed
4234 yet, because the function prologue itself does all the pushing.) This
4235 region is used on machines where an argument may be passed partly in
4236 registers and partly in memory, and, in some cases to support the
4237 features in @code{<stdarg.h>}.
4240 An area of memory used to save certain registers used by the function.
4241 The size of this area, which may also include space for such things as
4242 the return address and pointers to previous stack frames, is
4243 machine-specific and usually depends on which registers have been used
4244 in the function. Machines with register windows often do not require
4248 A region of at least @var{size} bytes, possibly rounded up to an allocation
4249 boundary, to contain the local variables of the function. On some machines,
4250 this region and the save area may occur in the opposite order, with the
4251 save area closer to the top of the stack.
4254 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4255 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4256 @code{current_function_outgoing_args_size} bytes to be used for outgoing
4257 argument lists of the function. @xref{Stack Arguments}.
4260 @defmac EXIT_IGNORE_STACK
4261 Define this macro as a C expression that is nonzero if the return
4262 instruction or the function epilogue ignores the value of the stack
4263 pointer; in other words, if it is safe to delete an instruction to
4264 adjust the stack pointer before a return from the function. The
4267 Note that this macro's value is relevant only for functions for which
4268 frame pointers are maintained. It is never safe to delete a final
4269 stack adjustment in a function that has no frame pointer, and the
4270 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4273 @defmac EPILOGUE_USES (@var{regno})
4274 Define this macro as a C expression that is nonzero for registers that are
4275 used by the epilogue or the @samp{return} pattern. The stack and frame
4276 pointer registers are already be assumed to be used as needed.
4279 @defmac EH_USES (@var{regno})
4280 Define this macro as a C expression that is nonzero for registers that are
4281 used by the exception handling mechanism, and so should be considered live
4282 on entry to an exception edge.
4285 @defmac DELAY_SLOTS_FOR_EPILOGUE
4286 Define this macro if the function epilogue contains delay slots to which
4287 instructions from the rest of the function can be ``moved''. The
4288 definition should be a C expression whose value is an integer
4289 representing the number of delay slots there.
4292 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4293 A C expression that returns 1 if @var{insn} can be placed in delay
4294 slot number @var{n} of the epilogue.
4296 The argument @var{n} is an integer which identifies the delay slot now
4297 being considered (since different slots may have different rules of
4298 eligibility). It is never negative and is always less than the number
4299 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4300 If you reject a particular insn for a given delay slot, in principle, it
4301 may be reconsidered for a subsequent delay slot. Also, other insns may
4302 (at least in principle) be considered for the so far unfilled delay
4305 @findex current_function_epilogue_delay_list
4306 @findex final_scan_insn
4307 The insns accepted to fill the epilogue delay slots are put in an RTL
4308 list made with @code{insn_list} objects, stored in the variable
4309 @code{current_function_epilogue_delay_list}. The insn for the first
4310 delay slot comes first in the list. Your definition of the macro
4311 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4312 outputting the insns in this list, usually by calling
4313 @code{final_scan_insn}.
4315 You need not define this macro if you did not define
4316 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4319 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_MI_THUNK (FILE *@var{file}, tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, tree @var{function})
4320 A function that outputs the assembler code for a thunk
4321 function, used to implement C++ virtual function calls with multiple
4322 inheritance. The thunk acts as a wrapper around a virtual function,
4323 adjusting the implicit object parameter before handing control off to
4326 First, emit code to add the integer @var{delta} to the location that
4327 contains the incoming first argument. Assume that this argument
4328 contains a pointer, and is the one used to pass the @code{this} pointer
4329 in C++. This is the incoming argument @emph{before} the function prologue,
4330 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4331 all other incoming arguments.
4333 After the addition, emit code to jump to @var{function}, which is a
4334 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4335 not touch the return address. Hence returning from @var{FUNCTION} will
4336 return to whoever called the current @samp{thunk}.
4338 The effect must be as if @var{function} had been called directly with
4339 the adjusted first argument. This macro is responsible for emitting all
4340 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4341 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4343 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4344 have already been extracted from it.) It might possibly be useful on
4345 some targets, but probably not.
4347 If you do not define this macro, the target-independent code in the C++
4348 front end will generate a less efficient heavyweight thunk that calls
4349 @var{function} instead of jumping to it. The generic approach does
4350 not support varargs.
4353 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_MI_VCALL_THUNK (FILE *@var{file}, tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, int @var{vcall_offset}, tree @var{function})
4354 A function like @code{TARGET_ASM_OUTPUT_MI_THUNK}, except that if
4355 @var{vcall_offset} is nonzero, an additional adjustment should be made
4356 after adding @code{delta}. In particular, if @var{p} is the
4357 adjusted pointer, the following adjustment should be made:
4360 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4364 If this function is defined, it will always be used in place of
4365 @code{TARGET_ASM_OUTPUT_MI_THUNK}.
4369 @subsection Generating Code for Profiling
4370 @cindex profiling, code generation
4372 These macros will help you generate code for profiling.
4374 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4375 A C statement or compound statement to output to @var{file} some
4376 assembler code to call the profiling subroutine @code{mcount}.
4379 The details of how @code{mcount} expects to be called are determined by
4380 your operating system environment, not by GCC@. To figure them out,
4381 compile a small program for profiling using the system's installed C
4382 compiler and look at the assembler code that results.
4384 Older implementations of @code{mcount} expect the address of a counter
4385 variable to be loaded into some register. The name of this variable is
4386 @samp{LP} followed by the number @var{labelno}, so you would generate
4387 the name using @samp{LP%d} in a @code{fprintf}.
4390 @defmac PROFILE_HOOK
4391 A C statement or compound statement to output to @var{file} some assembly
4392 code to call the profiling subroutine @code{mcount} even the target does
4393 not support profiling.
4396 @defmac NO_PROFILE_COUNTERS
4397 Define this macro if the @code{mcount} subroutine on your system does
4398 not need a counter variable allocated for each function. This is true
4399 for almost all modern implementations. If you define this macro, you
4400 must not use the @var{labelno} argument to @code{FUNCTION_PROFILER}.
4403 @defmac PROFILE_BEFORE_PROLOGUE
4404 Define this macro if the code for function profiling should come before
4405 the function prologue. Normally, the profiling code comes after.
4409 @subsection Permitting tail calls
4412 @deftypefn {Target Hook} bool TARGET_FUNCTION_OK_FOR_SIBCALL (tree @var{decl}, tree @var{exp})
4413 True if it is ok to do sibling call optimization for the specified
4414 call expression @var{exp}. @var{decl} will be the called function,
4415 or @code{NULL} if this is an indirect call.
4417 It is not uncommon for limitations of calling conventions to prevent
4418 tail calls to functions outside the current unit of translation, or
4419 during PIC compilation. The hook is used to enforce these restrictions,
4420 as the @code{sibcall} md pattern can not fail, or fall over to a
4421 ``normal'' call. The criteria for successful sibling call optimization
4422 may vary greatly between different architectures.
4426 @section Implementing the Varargs Macros
4427 @cindex varargs implementation
4429 GCC comes with an implementation of @code{<varargs.h>} and
4430 @code{<stdarg.h>} that work without change on machines that pass arguments
4431 on the stack. Other machines require their own implementations of
4432 varargs, and the two machine independent header files must have
4433 conditionals to include it.
4435 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4436 the calling convention for @code{va_start}. The traditional
4437 implementation takes just one argument, which is the variable in which
4438 to store the argument pointer. The ISO implementation of
4439 @code{va_start} takes an additional second argument. The user is
4440 supposed to write the last named argument of the function here.
4442 However, @code{va_start} should not use this argument. The way to find
4443 the end of the named arguments is with the built-in functions described
4446 @defmac __builtin_saveregs ()
4447 Use this built-in function to save the argument registers in memory so
4448 that the varargs mechanism can access them. Both ISO and traditional
4449 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4450 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4452 On some machines, @code{__builtin_saveregs} is open-coded under the
4453 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4454 other machines, it calls a routine written in assembler language,
4455 found in @file{libgcc2.c}.
4457 Code generated for the call to @code{__builtin_saveregs} appears at the
4458 beginning of the function, as opposed to where the call to
4459 @code{__builtin_saveregs} is written, regardless of what the code is.
4460 This is because the registers must be saved before the function starts
4461 to use them for its own purposes.
4462 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4466 @defmac __builtin_args_info (@var{category})
4467 Use this built-in function to find the first anonymous arguments in
4470 In general, a machine may have several categories of registers used for
4471 arguments, each for a particular category of data types. (For example,
4472 on some machines, floating-point registers are used for floating-point
4473 arguments while other arguments are passed in the general registers.)
4474 To make non-varargs functions use the proper calling convention, you
4475 have defined the @code{CUMULATIVE_ARGS} data type to record how many
4476 registers in each category have been used so far
4478 @code{__builtin_args_info} accesses the same data structure of type
4479 @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
4480 with it, with @var{category} specifying which word to access. Thus, the
4481 value indicates the first unused register in a given category.
4483 Normally, you would use @code{__builtin_args_info} in the implementation
4484 of @code{va_start}, accessing each category just once and storing the
4485 value in the @code{va_list} object. This is because @code{va_list} will
4486 have to update the values, and there is no way to alter the
4487 values accessed by @code{__builtin_args_info}.
4490 @defmac __builtin_next_arg (@var{lastarg})
4491 This is the equivalent of @code{__builtin_args_info}, for stack
4492 arguments. It returns the address of the first anonymous stack
4493 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
4494 returns the address of the location above the first anonymous stack
4495 argument. Use it in @code{va_start} to initialize the pointer for
4496 fetching arguments from the stack. Also use it in @code{va_start} to
4497 verify that the second parameter @var{lastarg} is the last named argument
4498 of the current function.
4501 @defmac __builtin_classify_type (@var{object})
4502 Since each machine has its own conventions for which data types are
4503 passed in which kind of register, your implementation of @code{va_arg}
4504 has to embody these conventions. The easiest way to categorize the
4505 specified data type is to use @code{__builtin_classify_type} together
4506 with @code{sizeof} and @code{__alignof__}.
4508 @code{__builtin_classify_type} ignores the value of @var{object},
4509 considering only its data type. It returns an integer describing what
4510 kind of type that is---integer, floating, pointer, structure, and so on.
4512 The file @file{typeclass.h} defines an enumeration that you can use to
4513 interpret the values of @code{__builtin_classify_type}.
4516 These machine description macros help implement varargs:
4518 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN_SAVEREGS (void)
4519 If defined, this hook produces the machine-specific code for a call to
4520 @code{__builtin_saveregs}. This code will be moved to the very
4521 beginning of the function, before any parameter access are made. The
4522 return value of this function should be an RTX that contains the value
4523 to use as the return of @code{__builtin_saveregs}.
4526 @deftypefn {Target Hook} void TARGET_SETUP_INCOMING_VARARGS (CUMULATIVE_ARGS *@var{args_so_far}, enum machine_mode @var{mode}, tree @var{type}, int *@var{pretend_args_size}, int @var{second_time})
4527 This target hook offers an alternative to using
4528 @code{__builtin_saveregs} and defining the hook
4529 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
4530 register arguments into the stack so that all the arguments appear to
4531 have been passed consecutively on the stack. Once this is done, you can
4532 use the standard implementation of varargs that works for machines that
4533 pass all their arguments on the stack.
4535 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
4536 structure, containing the values that are obtained after processing the
4537 named arguments. The arguments @var{mode} and @var{type} describe the
4538 last named argument---its machine mode and its data type as a tree node.
4540 The target hook should do two things: first, push onto the stack all the
4541 argument registers @emph{not} used for the named arguments, and second,
4542 store the size of the data thus pushed into the @code{int}-valued
4543 variable pointed to by @var{pretend_args_size}. The value that you
4544 store here will serve as additional offset for setting up the stack
4547 Because you must generate code to push the anonymous arguments at
4548 compile time without knowing their data types,
4549 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
4550 have just a single category of argument register and use it uniformly
4553 If the argument @var{second_time} is nonzero, it means that the
4554 arguments of the function are being analyzed for the second time. This
4555 happens for an inline function, which is not actually compiled until the
4556 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
4557 not generate any instructions in this case.
4560 @deftypefn {Target Hook} bool TARGET_STRICT_ARGUMENT_NAMING (CUMULATIVE_ARGS *@var{ca})
4561 Define this hook to return @code{true} if the location where a function
4562 argument is passed depends on whether or not it is a named argument.
4564 This hook controls how the @var{named} argument to @code{FUNCTION_ARG}
4565 is set for varargs and stdarg functions. If this hook returns
4566 @code{true}, the @var{named} argument is always true for named
4567 arguments, and false for unnamed arguments. If it returns @code{false},
4568 but @code{TARGET_PRETEND_OUTOGOING_VARARGS_NAMED} returns @code{true},
4569 then all arguments are treated as named. Otherwise, all named arguments
4570 except the last are treated as named.
4572 You need not define this hook if it always returns zero.
4575 @deftypefn {Target Hook} bool TARGET_PRETEND_OUTGOING_VARARGS_NAMED
4576 If you need to conditionally change ABIs so that one works with
4577 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
4578 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
4579 defined, then define this hook to return @code{true} if
4580 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
4581 Otherwise, you should not define this hook.
4585 @section Trampolines for Nested Functions
4586 @cindex trampolines for nested functions
4587 @cindex nested functions, trampolines for
4589 A @dfn{trampoline} is a small piece of code that is created at run time
4590 when the address of a nested function is taken. It normally resides on
4591 the stack, in the stack frame of the containing function. These macros
4592 tell GCC how to generate code to allocate and initialize a
4595 The instructions in the trampoline must do two things: load a constant
4596 address into the static chain register, and jump to the real address of
4597 the nested function. On CISC machines such as the m68k, this requires
4598 two instructions, a move immediate and a jump. Then the two addresses
4599 exist in the trampoline as word-long immediate operands. On RISC
4600 machines, it is often necessary to load each address into a register in
4601 two parts. Then pieces of each address form separate immediate
4604 The code generated to initialize the trampoline must store the variable
4605 parts---the static chain value and the function address---into the
4606 immediate operands of the instructions. On a CISC machine, this is
4607 simply a matter of copying each address to a memory reference at the
4608 proper offset from the start of the trampoline. On a RISC machine, it
4609 may be necessary to take out pieces of the address and store them
4612 @defmac TRAMPOLINE_TEMPLATE (@var{file})
4613 A C statement to output, on the stream @var{file}, assembler code for a
4614 block of data that contains the constant parts of a trampoline. This
4615 code should not include a label---the label is taken care of
4618 If you do not define this macro, it means no template is needed
4619 for the target. Do not define this macro on systems where the block move
4620 code to copy the trampoline into place would be larger than the code
4621 to generate it on the spot.
4624 @defmac TRAMPOLINE_SECTION
4625 The name of a subroutine to switch to the section in which the
4626 trampoline template is to be placed (@pxref{Sections}). The default is
4627 a value of @samp{readonly_data_section}, which places the trampoline in
4628 the section containing read-only data.
4631 @defmac TRAMPOLINE_SIZE
4632 A C expression for the size in bytes of the trampoline, as an integer.
4635 @defmac TRAMPOLINE_ALIGNMENT
4636 Alignment required for trampolines, in bits.
4638 If you don't define this macro, the value of @code{BIGGEST_ALIGNMENT}
4639 is used for aligning trampolines.
4642 @defmac INITIALIZE_TRAMPOLINE (@var{addr}, @var{fnaddr}, @var{static_chain})
4643 A C statement to initialize the variable parts of a trampoline.
4644 @var{addr} is an RTX for the address of the trampoline; @var{fnaddr} is
4645 an RTX for the address of the nested function; @var{static_chain} is an
4646 RTX for the static chain value that should be passed to the function
4650 @defmac TRAMPOLINE_ADJUST_ADDRESS (@var{addr})
4651 A C statement that should perform any machine-specific adjustment in
4652 the address of the trampoline. Its argument contains the address that
4653 was passed to @code{INITIALIZE_TRAMPOLINE}. In case the address to be
4654 used for a function call should be different from the address in which
4655 the template was stored, the different address should be assigned to
4656 @var{addr}. If this macro is not defined, @var{addr} will be used for
4659 @cindex @code{TARGET_ASM_FUNCTION_EPILOGUE} and trampolines
4660 @cindex @code{TARGET_ASM_FUNCTION_PROLOGUE} and trampolines
4661 If this macro is not defined, by default the trampoline is allocated as
4662 a stack slot. This default is right for most machines. The exceptions
4663 are machines where it is impossible to execute instructions in the stack
4664 area. On such machines, you may have to implement a separate stack,
4665 using this macro in conjunction with @code{TARGET_ASM_FUNCTION_PROLOGUE}
4666 and @code{TARGET_ASM_FUNCTION_EPILOGUE}.
4668 @var{fp} points to a data structure, a @code{struct function}, which
4669 describes the compilation status of the immediate containing function of
4670 the function which the trampoline is for. The stack slot for the
4671 trampoline is in the stack frame of this containing function. Other
4672 allocation strategies probably must do something analogous with this
4676 Implementing trampolines is difficult on many machines because they have
4677 separate instruction and data caches. Writing into a stack location
4678 fails to clear the memory in the instruction cache, so when the program
4679 jumps to that location, it executes the old contents.
4681 Here are two possible solutions. One is to clear the relevant parts of
4682 the instruction cache whenever a trampoline is set up. The other is to
4683 make all trampolines identical, by having them jump to a standard
4684 subroutine. The former technique makes trampoline execution faster; the
4685 latter makes initialization faster.
4687 To clear the instruction cache when a trampoline is initialized, define
4688 the following macro.
4690 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
4691 If defined, expands to a C expression clearing the @emph{instruction
4692 cache} in the specified interval. The definition of this macro would
4693 typically be a series of @code{asm} statements. Both @var{beg} and
4694 @var{end} are both pointer expressions.
4697 The operating system may also require the stack to be made executable
4698 before calling the trampoline. To implement this requirement, define
4699 the following macro.
4701 @defmac ENABLE_EXECUTE_STACK
4702 Define this macro if certain operations must be performed before executing
4703 code located on the stack. The macro should expand to a series of C
4704 file-scope constructs (e.g. functions) and provide a unique entry point
4705 named @code{__enable_execute_stack}. The target is responsible for
4706 emitting calls to the entry point in the code, for example from the
4707 @code{INITIALIZE_TRAMPOLINE} macro.
4710 To use a standard subroutine, define the following macro. In addition,
4711 you must make sure that the instructions in a trampoline fill an entire
4712 cache line with identical instructions, or else ensure that the
4713 beginning of the trampoline code is always aligned at the same point in
4714 its cache line. Look in @file{m68k.h} as a guide.
4716 @defmac TRANSFER_FROM_TRAMPOLINE
4717 Define this macro if trampolines need a special subroutine to do their
4718 work. The macro should expand to a series of @code{asm} statements
4719 which will be compiled with GCC@. They go in a library function named
4720 @code{__transfer_from_trampoline}.
4722 If you need to avoid executing the ordinary prologue code of a compiled
4723 C function when you jump to the subroutine, you can do so by placing a
4724 special label of your own in the assembler code. Use one @code{asm}
4725 statement to generate an assembler label, and another to make the label
4726 global. Then trampolines can use that label to jump directly to your
4727 special assembler code.
4731 @section Implicit Calls to Library Routines
4732 @cindex library subroutine names
4733 @cindex @file{libgcc.a}
4735 @c prevent bad page break with this line
4736 Here is an explanation of implicit calls to library routines.
4738 @defmac DECLARE_LIBRARY_RENAMES
4739 This macro, if defined, should expand to a piece of C code that will get
4740 expanded when compiling functions for libgcc.a. It can be used to
4741 provide alternate names for GCC's internal library functions if there
4742 are ABI-mandated names that the compiler should provide.
4745 @findex init_one_libfunc
4746 @findex set_optab_libfunc
4747 @deftypefn {Target Hook} void TARGET_INIT_LIBFUNCS (void)
4748 This hook should declare additional library routines or rename
4749 existing ones, using the functions @code{set_optab_libfunc} and
4750 @code{init_one_libfunc} defined in @file{optabs.c}.
4751 @code{init_optabs} calls this macro after initializing all the normal
4754 The default is to do nothing. Most ports don't need to define this hook.
4757 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
4758 This macro should return @code{true} if the library routine that
4759 implements the floating point comparison operator @var{comparison} in
4760 mode @var{mode} will return a boolean, and @var{false} if it will
4763 GCC's own floating point libraries return tristates from the
4764 comparison operators, so the default returns false always. Most ports
4765 don't need to define this macro.
4768 @defmac TARGET_LIBGCC_FUNCS
4769 This macro should evaluate to @code{true} if the standard GCC library
4770 names (like @code{__modsi3}) should be used for functions provided in
4771 @file{libgcc.a}. If this macro evaluates to @code{false}, then the
4772 target must explictily set the names of all desired library functions
4773 itself using the @code{TARGET_INIT_LIBFUNCS} hook.
4776 @defmac TARGET_LIB_INT_CMP_BIASED
4777 This macro should evaluate to @code{true} if the integer comparison
4778 functions (like @code{__cmpdi2}) return 0 to indicate that the first
4779 operand is smaller than the second, 1 to indicate that they are equal,
4780 and 2 to indicate that the first operand is greater than the second.
4781 If this macro evalutes to @code{false} the comparison functions return
4782 -1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
4783 in @file{libgcc.a}, you do not need to define this macro.
4786 @cindex US Software GOFAST, floating point emulation library
4787 @cindex floating point emulation library, US Software GOFAST
4788 @cindex GOFAST, floating point emulation library
4789 @findex gofast_maybe_init_libfuncs
4790 @defmac US_SOFTWARE_GOFAST
4791 Define this macro if your system C library uses the US Software GOFAST
4792 library to provide floating point emulation.
4794 In addition to defining this macro, your architecture must set
4795 @code{TARGET_INIT_LIBFUNCS} to @code{gofast_maybe_init_libfuncs}, or
4796 else call that function from its version of that hook. It is defined
4797 in @file{config/gofast.h}, which must be included by your
4798 architecture's @file{@var{cpu}.c} file. See @file{sparc/sparc.c} for
4801 If this macro is defined, the
4802 @code{TARGET_FLOAT_LIB_COMPARE_RETURNS_BOOL} target hook must return
4803 false for @code{SFmode} and @code{DFmode} comparisons.
4806 @cindex @code{EDOM}, implicit usage
4809 The value of @code{EDOM} on the target machine, as a C integer constant
4810 expression. If you don't define this macro, GCC does not attempt to
4811 deposit the value of @code{EDOM} into @code{errno} directly. Look in
4812 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
4815 If you do not define @code{TARGET_EDOM}, then compiled code reports
4816 domain errors by calling the library function and letting it report the
4817 error. If mathematical functions on your system use @code{matherr} when
4818 there is an error, then you should leave @code{TARGET_EDOM} undefined so
4819 that @code{matherr} is used normally.
4822 @cindex @code{errno}, implicit usage
4823 @defmac GEN_ERRNO_RTX
4824 Define this macro as a C expression to create an rtl expression that
4825 refers to the global ``variable'' @code{errno}. (On certain systems,
4826 @code{errno} may not actually be a variable.) If you don't define this
4827 macro, a reasonable default is used.
4830 @cindex C99 math functions, implicit usage
4831 @defmac TARGET_C99_FUNCTIONS
4832 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
4833 @code{sinf} and similarly for other functions defined by C99 standard. The
4834 default is nonzero that should be proper value for most modern systems, however
4835 number of existing systems lacks support for these functions in the runtime so
4836 they needs this macro to be redefined to 0.
4839 @defmac NEXT_OBJC_RUNTIME
4840 Define this macro to generate code for Objective-C message sending using
4841 the calling convention of the NeXT system. This calling convention
4842 involves passing the object, the selector and the method arguments all
4843 at once to the method-lookup library function.
4845 The default calling convention passes just the object and the selector
4846 to the lookup function, which returns a pointer to the method.
4849 @node Addressing Modes
4850 @section Addressing Modes
4851 @cindex addressing modes
4853 @c prevent bad page break with this line
4854 This is about addressing modes.
4856 @defmac HAVE_PRE_INCREMENT
4857 @defmacx HAVE_PRE_DECREMENT
4858 @defmacx HAVE_POST_INCREMENT
4859 @defmacx HAVE_POST_DECREMENT
4860 A C expression that is nonzero if the machine supports pre-increment,
4861 pre-decrement, post-increment, or post-decrement addressing respectively.
4864 @defmac HAVE_PRE_MODIFY_DISP
4865 @defmacx HAVE_POST_MODIFY_DISP
4866 A C expression that is nonzero if the machine supports pre- or
4867 post-address side-effect generation involving constants other than
4868 the size of the memory operand.
4871 @defmac HAVE_PRE_MODIFY_REG
4872 @defmacx HAVE_POST_MODIFY_REG
4873 A C expression that is nonzero if the machine supports pre- or
4874 post-address side-effect generation involving a register displacement.
4877 @defmac CONSTANT_ADDRESS_P (@var{x})
4878 A C expression that is 1 if the RTX @var{x} is a constant which
4879 is a valid address. On most machines, this can be defined as
4880 @code{CONSTANT_P (@var{x})}, but a few machines are more restrictive
4881 in which constant addresses are supported.
4884 @defmac CONSTANT_P (@var{x})
4885 @code{CONSTANT_P}, which is defined by target-independent code,
4886 accepts integer-values expressions whose values are not explicitly
4887 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
4888 expressions and @code{const} arithmetic expressions, in addition to
4889 @code{const_int} and @code{const_double} expressions.
4892 @defmac MAX_REGS_PER_ADDRESS
4893 A number, the maximum number of registers that can appear in a valid
4894 memory address. Note that it is up to you to specify a value equal to
4895 the maximum number that @code{GO_IF_LEGITIMATE_ADDRESS} would ever
4899 @defmac GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
4900 A C compound statement with a conditional @code{goto @var{label};}
4901 executed if @var{x} (an RTX) is a legitimate memory address on the
4902 target machine for a memory operand of mode @var{mode}.
4904 It usually pays to define several simpler macros to serve as
4905 subroutines for this one. Otherwise it may be too complicated to
4908 This macro must exist in two variants: a strict variant and a
4909 non-strict one. The strict variant is used in the reload pass. It
4910 must be defined so that any pseudo-register that has not been
4911 allocated a hard register is considered a memory reference. In
4912 contexts where some kind of register is required, a pseudo-register
4913 with no hard register must be rejected.
4915 The non-strict variant is used in other passes. It must be defined to
4916 accept all pseudo-registers in every context where some kind of
4917 register is required.
4919 @findex REG_OK_STRICT
4920 Compiler source files that want to use the strict variant of this
4921 macro define the macro @code{REG_OK_STRICT}. You should use an
4922 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant
4923 in that case and the non-strict variant otherwise.
4925 Subroutines to check for acceptable registers for various purposes (one
4926 for base registers, one for index registers, and so on) are typically
4927 among the subroutines used to define @code{GO_IF_LEGITIMATE_ADDRESS}.
4928 Then only these subroutine macros need have two variants; the higher
4929 levels of macros may be the same whether strict or not.
4931 Normally, constant addresses which are the sum of a @code{symbol_ref}
4932 and an integer are stored inside a @code{const} RTX to mark them as
4933 constant. Therefore, there is no need to recognize such sums
4934 specifically as legitimate addresses. Normally you would simply
4935 recognize any @code{const} as legitimate.
4937 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
4938 sums that are not marked with @code{const}. It assumes that a naked
4939 @code{plus} indicates indexing. If so, then you @emph{must} reject such
4940 naked constant sums as illegitimate addresses, so that none of them will
4941 be given to @code{PRINT_OPERAND_ADDRESS}.
4943 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
4944 On some machines, whether a symbolic address is legitimate depends on
4945 the section that the address refers to. On these machines, define the
4946 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
4947 into the @code{symbol_ref}, and then check for it here. When you see a
4948 @code{const}, you will have to look inside it to find the
4949 @code{symbol_ref} in order to determine the section. @xref{Assembler
4953 @defmac REG_OK_FOR_BASE_P (@var{x})
4954 A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
4955 RTX) is valid for use as a base register. For hard registers, it
4956 should always accept those which the hardware permits and reject the
4957 others. Whether the macro accepts or rejects pseudo registers must be
4958 controlled by @code{REG_OK_STRICT} as described above. This usually
4959 requires two variant definitions, of which @code{REG_OK_STRICT}
4960 controls the one actually used.
4963 @defmac REG_MODE_OK_FOR_BASE_P (@var{x}, @var{mode})
4964 A C expression that is just like @code{REG_OK_FOR_BASE_P}, except that
4965 that expression may examine the mode of the memory reference in
4966 @var{mode}. You should define this macro if the mode of the memory
4967 reference affects whether a register may be used as a base register. If
4968 you define this macro, the compiler will use it instead of
4969 @code{REG_OK_FOR_BASE_P}.
4972 @defmac REG_OK_FOR_INDEX_P (@var{x})
4973 A C expression that is nonzero if @var{x} (assumed to be a @code{reg}
4974 RTX) is valid for use as an index register.
4976 The difference between an index register and a base register is that
4977 the index register may be scaled. If an address involves the sum of
4978 two registers, neither one of them scaled, then either one may be
4979 labeled the ``base'' and the other the ``index''; but whichever
4980 labeling is used must fit the machine's constraints of which registers
4981 may serve in each capacity. The compiler will try both labelings,
4982 looking for one that is valid, and will reload one or both registers
4983 only if neither labeling works.
4986 @defmac FIND_BASE_TERM (@var{x})
4987 A C expression to determine the base term of address @var{x}.
4988 This macro is used in only one place: `find_base_term' in alias.c.
4990 It is always safe for this macro to not be defined. It exists so
4991 that alias analysis can understand machine-dependent addresses.
4993 The typical use of this macro is to handle addresses containing
4994 a label_ref or symbol_ref within an UNSPEC@.
4997 @defmac LEGITIMIZE_ADDRESS (@var{x}, @var{oldx}, @var{mode}, @var{win})
4998 A C compound statement that attempts to replace @var{x} with a valid
4999 memory address for an operand of mode @var{mode}. @var{win} will be a
5000 C statement label elsewhere in the code; the macro definition may use
5003 GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{win});
5007 to avoid further processing if the address has become legitimate.
5009 @findex break_out_memory_refs
5010 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5011 and @var{oldx} will be the operand that was given to that function to produce
5014 The code generated by this macro should not alter the substructure of
5015 @var{x}. If it transforms @var{x} into a more legitimate form, it
5016 should assign @var{x} (which will always be a C variable) a new value.
5018 It is not necessary for this macro to come up with a legitimate
5019 address. The compiler has standard ways of doing so in all cases. In
5020 fact, it is safe to omit this macro. But often a
5021 machine-dependent strategy can generate better code.
5024 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5025 A C compound statement that attempts to replace @var{x}, which is an address
5026 that needs reloading, with a valid memory address for an operand of mode
5027 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5028 It is not necessary to define this macro, but it might be useful for
5029 performance reasons.
5031 For example, on the i386, it is sometimes possible to use a single
5032 reload register instead of two by reloading a sum of two pseudo
5033 registers into a register. On the other hand, for number of RISC
5034 processors offsets are limited so that often an intermediate address
5035 needs to be generated in order to address a stack slot. By defining
5036 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5037 generated for adjacent some stack slots can be made identical, and thus
5040 @emph{Note}: This macro should be used with caution. It is necessary
5041 to know something of how reload works in order to effectively use this,
5042 and it is quite easy to produce macros that build in too much knowledge
5043 of reload internals.
5045 @emph{Note}: This macro must be able to reload an address created by a
5046 previous invocation of this macro. If it fails to handle such addresses
5047 then the compiler may generate incorrect code or abort.
5050 The macro definition should use @code{push_reload} to indicate parts that
5051 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5052 suitable to be passed unaltered to @code{push_reload}.
5054 The code generated by this macro must not alter the substructure of
5055 @var{x}. If it transforms @var{x} into a more legitimate form, it
5056 should assign @var{x} (which will always be a C variable) a new value.
5057 This also applies to parts that you change indirectly by calling
5060 @findex strict_memory_address_p
5061 The macro definition may use @code{strict_memory_address_p} to test if
5062 the address has become legitimate.
5065 If you want to change only a part of @var{x}, one standard way of doing
5066 this is to use @code{copy_rtx}. Note, however, that is unshares only a
5067 single level of rtl. Thus, if the part to be changed is not at the
5068 top level, you'll need to replace first the top level.
5069 It is not necessary for this macro to come up with a legitimate
5070 address; but often a machine-dependent strategy can generate better code.
5073 @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
5074 A C statement or compound statement with a conditional @code{goto
5075 @var{label};} executed if memory address @var{x} (an RTX) can have
5076 different meanings depending on the machine mode of the memory
5077 reference it is used for or if the address is valid for some modes
5080 Autoincrement and autodecrement addresses typically have mode-dependent
5081 effects because the amount of the increment or decrement is the size
5082 of the operand being addressed. Some machines have other mode-dependent
5083 addresses. Many RISC machines have no mode-dependent addresses.
5085 You may assume that @var{addr} is a valid address for the machine.
5088 @defmac LEGITIMATE_CONSTANT_P (@var{x})
5089 A C expression that is nonzero if @var{x} is a legitimate constant for
5090 an immediate operand on the target machine. You can assume that
5091 @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
5092 @samp{1} is a suitable definition for this macro on machines where
5093 anything @code{CONSTANT_P} is valid.
5096 @node Condition Code
5097 @section Condition Code Status
5098 @cindex condition code status
5100 @c prevent bad page break with this line
5101 This describes the condition code status.
5104 The file @file{conditions.h} defines a variable @code{cc_status} to
5105 describe how the condition code was computed (in case the interpretation of
5106 the condition code depends on the instruction that it was set by). This
5107 variable contains the RTL expressions on which the condition code is
5108 currently based, and several standard flags.
5110 Sometimes additional machine-specific flags must be defined in the machine
5111 description header file. It can also add additional machine-specific
5112 information by defining @code{CC_STATUS_MDEP}.
5114 @defmac CC_STATUS_MDEP
5115 C code for a data type which is used for declaring the @code{mdep}
5116 component of @code{cc_status}. It defaults to @code{int}.
5118 This macro is not used on machines that do not use @code{cc0}.
5121 @defmac CC_STATUS_MDEP_INIT
5122 A C expression to initialize the @code{mdep} field to ``empty''.
5123 The default definition does nothing, since most machines don't use
5124 the field anyway. If you want to use the field, you should probably
5125 define this macro to initialize it.
5127 This macro is not used on machines that do not use @code{cc0}.
5130 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5131 A C compound statement to set the components of @code{cc_status}
5132 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5133 this macro's responsibility to recognize insns that set the condition
5134 code as a byproduct of other activity as well as those that explicitly
5137 This macro is not used on machines that do not use @code{cc0}.
5139 If there are insns that do not set the condition code but do alter
5140 other machine registers, this macro must check to see whether they
5141 invalidate the expressions that the condition code is recorded as
5142 reflecting. For example, on the 68000, insns that store in address
5143 registers do not set the condition code, which means that usually
5144 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5145 insns. But suppose that the previous insn set the condition code
5146 based on location @samp{a4@@(102)} and the current insn stores a new
5147 value in @samp{a4}. Although the condition code is not changed by
5148 this, it will no longer be true that it reflects the contents of
5149 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5150 @code{cc_status} in this case to say that nothing is known about the
5151 condition code value.
5153 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5154 with the results of peephole optimization: insns whose patterns are
5155 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5156 constants which are just the operands. The RTL structure of these
5157 insns is not sufficient to indicate what the insns actually do. What
5158 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5159 @code{CC_STATUS_INIT}.
5161 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5162 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5163 @samp{cc}. This avoids having detailed information about patterns in
5164 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5167 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5168 Returns a mode from class @code{MODE_CC} to be used when comparison
5169 operation code @var{op} is applied to rtx @var{x} and @var{y}. For
5170 example, on the SPARC, @code{SELECT_CC_MODE} is defined as (see
5171 @pxref{Jump Patterns} for a description of the reason for this
5175 #define SELECT_CC_MODE(OP,X,Y) \
5176 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5177 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5178 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5179 || GET_CODE (X) == NEG) \
5180 ? CC_NOOVmode : CCmode))
5183 You should define this macro if and only if you define extra CC modes
5184 in @file{@var{machine}-modes.def}.
5187 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
5188 On some machines not all possible comparisons are defined, but you can
5189 convert an invalid comparison into a valid one. For example, the Alpha
5190 does not have a @code{GT} comparison, but you can use an @code{LT}
5191 comparison instead and swap the order of the operands.
5193 On such machines, define this macro to be a C statement to do any
5194 required conversions. @var{code} is the initial comparison code
5195 and @var{op0} and @var{op1} are the left and right operands of the
5196 comparison, respectively. You should modify @var{code}, @var{op0}, and
5197 @var{op1} as required.
5199 GCC will not assume that the comparison resulting from this macro is
5200 valid but will see if the resulting insn matches a pattern in the
5203 You need not define this macro if it would never change the comparison
5207 @defmac REVERSIBLE_CC_MODE (@var{mode})
5208 A C expression whose value is one if it is always safe to reverse a
5209 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
5210 can ever return @var{mode} for a floating-point inequality comparison,
5211 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
5213 You need not define this macro if it would always returns zero or if the
5214 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
5215 For example, here is the definition used on the SPARC, where floating-point
5216 inequality comparisons are always given @code{CCFPEmode}:
5219 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
5223 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
5224 A C expression whose value is reversed condition code of the @var{code} for
5225 comparison done in CC_MODE @var{mode}. The macro is used only in case
5226 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
5227 machine has some non-standard way how to reverse certain conditionals. For
5228 instance in case all floating point conditions are non-trapping, compiler may
5229 freely convert unordered compares to ordered one. Then definition may look
5233 #define REVERSE_CONDITION(CODE, MODE) \
5234 ((MODE) != CCFPmode ? reverse_condition (CODE) \
5235 : reverse_condition_maybe_unordered (CODE))
5239 @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{code1}, @var{code2})
5240 A C expression that returns true if the conditional execution predicate
5241 @var{code1} is the inverse of @var{code2} and vice versa. Define this to
5242 return 0 if the target has conditional execution predicates that cannot be
5243 reversed safely. If no expansion is specified, this macro is defined as
5247 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
5248 ((x) == reverse_condition (y))
5252 @deftypefn {Target Hook} bool TARGET_FIXED_CONDITION_CODE_REGS (unsigned int *, unsigned int *)
5253 On targets which do not use @code{(cc0)}, and which use a hard
5254 register rather than a pseudo-register to hold condition codes, the
5255 regular CSE passes are often not able to identify cases in which the
5256 hard register is set to a common value. Use this hook to enable a
5257 small pass which optimizes such cases. This hook should return true
5258 to enable this pass, and it should set the integers to which its
5259 arguments point to the hard register numbers used for condition codes.
5260 When there is only one such register, as is true on most systems, the
5261 integer pointed to by the second argument should be set to
5262 @code{INVALID_REGNUM}.
5264 The default version of this hook returns false.
5267 @deftypefn {Target Hook} enum machine_mode TARGET_CC_MODES_COMPATIBLE (enum machine_mode, enum machine_mode)
5268 On targets which use multiple condition code modes in class
5269 @code{MODE_CC}, it is sometimes the case that a comparison can be
5270 validly done in more than one mode. On such a system, define this
5271 target hook to take two mode arguments and to return a mode in which
5272 both comparisons may be validly done. If there is no such mode,
5273 return @code{VOIDmode}.
5275 The default version of this hook checks whether the modes are the
5276 same. If they are, it returns that mode. If they are different, it
5277 returns @code{VOIDmode}.
5281 @section Describing Relative Costs of Operations
5282 @cindex costs of instructions
5283 @cindex relative costs
5284 @cindex speed of instructions
5286 These macros let you describe the relative speed of various operations
5287 on the target machine.
5289 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
5290 A C expression for the cost of moving data of mode @var{mode} from a
5291 register in class @var{from} to one in class @var{to}. The classes are
5292 expressed using the enumeration values such as @code{GENERAL_REGS}. A
5293 value of 2 is the default; other values are interpreted relative to
5296 It is not required that the cost always equal 2 when @var{from} is the
5297 same as @var{to}; on some machines it is expensive to move between
5298 registers if they are not general registers.
5300 If reload sees an insn consisting of a single @code{set} between two
5301 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
5302 classes returns a value of 2, reload does not check to ensure that the
5303 constraints of the insn are met. Setting a cost of other than 2 will
5304 allow reload to verify that the constraints are met. You should do this
5305 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
5308 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
5309 A C expression for the cost of moving data of mode @var{mode} between a
5310 register of class @var{class} and memory; @var{in} is zero if the value
5311 is to be written to memory, nonzero if it is to be read in. This cost
5312 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
5313 registers and memory is more expensive than between two registers, you
5314 should define this macro to express the relative cost.
5316 If you do not define this macro, GCC uses a default cost of 4 plus
5317 the cost of copying via a secondary reload register, if one is
5318 needed. If your machine requires a secondary reload register to copy
5319 between memory and a register of @var{class} but the reload mechanism is
5320 more complex than copying via an intermediate, define this macro to
5321 reflect the actual cost of the move.
5323 GCC defines the function @code{memory_move_secondary_cost} if
5324 secondary reloads are needed. It computes the costs due to copying via
5325 a secondary register. If your machine copies from memory using a
5326 secondary register in the conventional way but the default base value of
5327 4 is not correct for your machine, define this macro to add some other
5328 value to the result of that function. The arguments to that function
5329 are the same as to this macro.
5333 A C expression for the cost of a branch instruction. A value of 1 is
5334 the default; other values are interpreted relative to that.
5337 Here are additional macros which do not specify precise relative costs,
5338 but only that certain actions are more expensive than GCC would
5341 @defmac SLOW_BYTE_ACCESS
5342 Define this macro as a C expression which is nonzero if accessing less
5343 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
5344 faster than accessing a word of memory, i.e., if such access
5345 require more than one instruction or if there is no difference in cost
5346 between byte and (aligned) word loads.
5348 When this macro is not defined, the compiler will access a field by
5349 finding the smallest containing object; when it is defined, a fullword
5350 load will be used if alignment permits. Unless bytes accesses are
5351 faster than word accesses, using word accesses is preferable since it
5352 may eliminate subsequent memory access if subsequent accesses occur to
5353 other fields in the same word of the structure, but to different bytes.
5356 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
5357 Define this macro to be the value 1 if memory accesses described by the
5358 @var{mode} and @var{alignment} parameters have a cost many times greater
5359 than aligned accesses, for example if they are emulated in a trap
5362 When this macro is nonzero, the compiler will act as if
5363 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
5364 moves. This can cause significantly more instructions to be produced.
5365 Therefore, do not set this macro nonzero if unaligned accesses only add a
5366 cycle or two to the time for a memory access.
5368 If the value of this macro is always zero, it need not be defined. If
5369 this macro is defined, it should produce a nonzero value when
5370 @code{STRICT_ALIGNMENT} is nonzero.
5374 The threshold of number of scalar memory-to-memory move insns, @emph{below}
5375 which a sequence of insns should be generated instead of a
5376 string move insn or a library call. Increasing the value will always
5377 make code faster, but eventually incurs high cost in increased code size.
5379 Note that on machines where the corresponding move insn is a
5380 @code{define_expand} that emits a sequence of insns, this macro counts
5381 the number of such sequences.
5383 If you don't define this, a reasonable default is used.
5386 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
5387 A C expression used to determine whether @code{move_by_pieces} will be used to
5388 copy a chunk of memory, or whether some other block move mechanism
5389 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
5390 than @code{MOVE_RATIO}.
5393 @defmac MOVE_MAX_PIECES
5394 A C expression used by @code{move_by_pieces} to determine the largest unit
5395 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
5399 The threshold of number of scalar move insns, @emph{below} which a sequence
5400 of insns should be generated to clear memory instead of a string clear insn
5401 or a library call. Increasing the value will always make code faster, but
5402 eventually incurs high cost in increased code size.
5404 If you don't define this, a reasonable default is used.
5407 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
5408 A C expression used to determine whether @code{clear_by_pieces} will be used
5409 to clear a chunk of memory, or whether some other block clear mechanism
5410 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
5411 than @code{CLEAR_RATIO}.
5414 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
5415 A C expression used to determine whether @code{store_by_pieces} will be
5416 used to set a chunk of memory to a constant value, or whether some other
5417 mechanism will be used. Used by @code{__builtin_memset} when storing
5418 values other than constant zero and by @code{__builtin_strcpy} when
5419 when called with a constant source string.
5420 Defaults to to 1 if @code{move_by_pieces_ninsns} returns less
5421 than @code{MOVE_RATIO}.
5424 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
5425 A C expression used to determine whether a load postincrement is a good
5426 thing to use for a given mode. Defaults to the value of
5427 @code{HAVE_POST_INCREMENT}.
5430 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
5431 A C expression used to determine whether a load postdecrement is a good
5432 thing to use for a given mode. Defaults to the value of
5433 @code{HAVE_POST_DECREMENT}.
5436 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
5437 A C expression used to determine whether a load preincrement is a good
5438 thing to use for a given mode. Defaults to the value of
5439 @code{HAVE_PRE_INCREMENT}.
5442 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
5443 A C expression used to determine whether a load predecrement is a good
5444 thing to use for a given mode. Defaults to the value of
5445 @code{HAVE_PRE_DECREMENT}.
5448 @defmac USE_STORE_POST_INCREMENT (@var{mode})
5449 A C expression used to determine whether a store postincrement is a good
5450 thing to use for a given mode. Defaults to the value of
5451 @code{HAVE_POST_INCREMENT}.
5454 @defmac USE_STORE_POST_DECREMENT (@var{mode})
5455 A C expression used to determine whether a store postdecrement is a good
5456 thing to use for a given mode. Defaults to the value of
5457 @code{HAVE_POST_DECREMENT}.
5460 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
5461 This macro is used to determine whether a store preincrement is a good
5462 thing to use for a given mode. Defaults to the value of
5463 @code{HAVE_PRE_INCREMENT}.
5466 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
5467 This macro is used to determine whether a store predecrement is a good
5468 thing to use for a given mode. Defaults to the value of
5469 @code{HAVE_PRE_DECREMENT}.
5472 @defmac NO_FUNCTION_CSE
5473 Define this macro if it is as good or better to call a constant
5474 function address than to call an address kept in a register.
5477 @defmac RANGE_TEST_NON_SHORT_CIRCUIT
5478 Define this macro if a non-short-circuit operation produced by
5479 @samp{fold_range_test ()} is optimal. This macro defaults to true if
5480 @code{BRANCH_COST} is greater than or equal to the value 2.
5483 @deftypefn {Target Hook} bool TARGET_RTX_COSTS (rtx @var{x}, int @var{code}, int @var{outer_code}, int *@var{total})
5484 This target hook describes the relative costs of RTL expressions.
5486 The cost may depend on the precise form of the expression, which is
5487 available for examination in @var{x}, and the rtx code of the expression
5488 in which it is contained, found in @var{outer_code}. @var{code} is the
5489 expression code---redundant, since it can be obtained with
5490 @code{GET_CODE (@var{x})}.
5492 In implementing this hook, you can use the construct
5493 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
5496 On entry to the hook, @code{*@var{total}} contains a default estimate
5497 for the cost of the expression. The hook should modify this value as
5498 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
5499 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
5500 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
5502 When optimizing for code size, i.e@. when @code{optimize_size} is
5503 non-zero, this target hook should be used to estimate the relative
5504 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
5506 The hook returns true when all subexpressions of @var{x} have been
5507 processed, and false when @code{rtx_cost} should recurse.
5510 @deftypefn {Target Hook} int TARGET_ADDRESS_COST (rtx @var{address})
5511 This hook computes the cost of an addressing mode that contains
5512 @var{address}. If not defined, the cost is computed from
5513 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
5515 For most CISC machines, the default cost is a good approximation of the
5516 true cost of the addressing mode. However, on RISC machines, all
5517 instructions normally have the same length and execution time. Hence
5518 all addresses will have equal costs.
5520 In cases where more than one form of an address is known, the form with
5521 the lowest cost will be used. If multiple forms have the same, lowest,
5522 cost, the one that is the most complex will be used.
5524 For example, suppose an address that is equal to the sum of a register
5525 and a constant is used twice in the same basic block. When this macro
5526 is not defined, the address will be computed in a register and memory
5527 references will be indirect through that register. On machines where
5528 the cost of the addressing mode containing the sum is no higher than
5529 that of a simple indirect reference, this will produce an additional
5530 instruction and possibly require an additional register. Proper
5531 specification of this macro eliminates this overhead for such machines.
5533 This hook is never called with an invalid address.
5535 On machines where an address involving more than one register is as
5536 cheap as an address computation involving only one register, defining
5537 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
5538 be live over a region of code where only one would have been if
5539 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
5540 should be considered in the definition of this macro. Equivalent costs
5541 should probably only be given to addresses with different numbers of
5542 registers on machines with lots of registers.
5546 @section Adjusting the Instruction Scheduler
5548 The instruction scheduler may need a fair amount of machine-specific
5549 adjustment in order to produce good code. GCC provides several target
5550 hooks for this purpose. It is usually enough to define just a few of
5551 them: try the first ones in this list first.
5553 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
5554 This hook returns the maximum number of instructions that can ever
5555 issue at the same time on the target machine. The default is one.
5556 Although the insn scheduler can define itself the possibility of issue
5557 an insn on the same cycle, the value can serve as an additional
5558 constraint to issue insns on the same simulated processor cycle (see
5559 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
5560 This value must be constant over the entire compilation. If you need
5561 it to vary depending on what the instructions are, you must use
5562 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
5564 You could define this hook to return the value of the macro
5565 @code{MAX_DFA_ISSUE_RATE}.
5568 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more})
5569 This hook is executed by the scheduler after it has scheduled an insn
5570 from the ready list. It should return the number of insns which can
5571 still be issued in the current cycle. The default is
5572 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
5573 @code{USE}, which normally are not counted against the issue rate.
5574 You should define this hook if some insns take more machine resources
5575 than others, so that fewer insns can follow them in the same cycle.
5576 @var{file} is either a null pointer, or a stdio stream to write any
5577 debug output to. @var{verbose} is the verbose level provided by
5578 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
5582 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost})
5583 This function corrects the value of @var{cost} based on the
5584 relationship between @var{insn} and @var{dep_insn} through the
5585 dependence @var{link}. It should return the new value. The default
5586 is to make no adjustment to @var{cost}. This can be used for example
5587 to specify to the scheduler using the traditional pipeline description
5588 that an output- or anti-dependence does not incur the same cost as a
5589 data-dependence. If the scheduler using the automaton based pipeline
5590 description, the cost of anti-dependence is zero and the cost of
5591 output-dependence is maximum of one and the difference of latency
5592 times of the first and the second insns. If these values are not
5593 acceptable, you could use the hook to modify them too. See also
5594 @pxref{Processor pipeline description}.
5597 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority})
5598 This hook adjusts the integer scheduling priority @var{priority} of
5599 @var{insn}. It should return the new priority. Reduce the priority to
5600 execute @var{insn} earlier, increase the priority to execute @var{insn}
5601 later. Do not define this hook if you do not need to adjust the
5602 scheduling priorities of insns.
5605 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
5606 This hook is executed by the scheduler after it has scheduled the ready
5607 list, to allow the machine description to reorder it (for example to
5608 combine two small instructions together on @samp{VLIW} machines).
5609 @var{file} is either a null pointer, or a stdio stream to write any
5610 debug output to. @var{verbose} is the verbose level provided by
5611 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
5612 list of instructions that are ready to be scheduled. @var{n_readyp} is
5613 a pointer to the number of elements in the ready list. The scheduler
5614 reads the ready list in reverse order, starting with
5615 @var{ready}[@var{*n_readyp}-1] and going to @var{ready}[0]. @var{clock}
5616 is the timer tick of the scheduler. You may modify the ready list and
5617 the number of ready insns. The return value is the number of insns that
5618 can issue this cycle; normally this is just @code{issue_rate}. See also
5619 @samp{TARGET_SCHED_REORDER2}.
5622 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_ready}, @var{clock})
5623 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
5624 function is called whenever the scheduler starts a new cycle. This one
5625 is called once per iteration over a cycle, immediately after
5626 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
5627 return the number of insns to be scheduled in the same cycle. Defining
5628 this hook can be useful if there are frequent situations where
5629 scheduling one insn causes other insns to become ready in the same
5630 cycle. These other insns can then be taken into account properly.
5633 @deftypefn {Target Hook} void TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK (rtx @var{head}, rtx @var{tail})
5634 This hook is called after evaluation forward dependencies of insns in
5635 chain given by two parameter values (@var{head} and @var{tail}
5636 correspondingly) but before insns scheduling of the insn chain. For
5637 example, it can be used for better insn classification if it requires
5638 analysis of dependencies. This hook can use backward and forward
5639 dependencies of the insn scheduler because they are already
5643 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
5644 This hook is executed by the scheduler at the beginning of each block of
5645 instructions that are to be scheduled. @var{file} is either a null
5646 pointer, or a stdio stream to write any debug output to. @var{verbose}
5647 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
5648 @var{max_ready} is the maximum number of insns in the current scheduling
5649 region that can be live at the same time. This can be used to allocate
5650 scratch space if it is needed, e.g. by @samp{TARGET_SCHED_REORDER}.
5653 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
5654 This hook is executed by the scheduler at the end of each block of
5655 instructions that are to be scheduled. It can be used to perform
5656 cleanup of any actions done by the other scheduling hooks. @var{file}
5657 is either a null pointer, or a stdio stream to write any debug output
5658 to. @var{verbose} is the verbose level provided by
5659 @option{-fsched-verbose-@var{n}}.
5662 @deftypefn {Target Hook} void TARGET_SCHED_INIT_GLOBAL (FILE *@var{file}, int @var{verbose}, int @var{old_max_uid})
5663 This hook is executed by the scheduler after function level initializations.
5664 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
5665 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
5666 @var{old_max_uid} is the maximum insn uid when scheduling begins.
5669 @deftypefn {Target Hook} void TARGET_SCHED_FINISH_GLOBAL (FILE *@var{file}, int @var{verbose})
5670 This is the cleanup hook corresponding to TARGET_SCHED_INIT_GLOBAL.
5671 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
5672 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
5675 @deftypefn {Target Hook} int TARGET_SCHED_DFA_PRE_CYCLE_INSN (void)
5676 The hook returns an RTL insn. The automaton state used in the
5677 pipeline hazard recognizer is changed as if the insn were scheduled
5678 when the new simulated processor cycle starts. Usage of the hook may
5679 simplify the automaton pipeline description for some @acronym{VLIW}
5680 processors. If the hook is defined, it is used only for the automaton
5681 based pipeline description. The default is not to change the state
5682 when the new simulated processor cycle starts.
5685 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN (void)
5686 The hook can be used to initialize data used by the previous hook.
5689 @deftypefn {Target Hook} int TARGET_SCHED_DFA_POST_CYCLE_INSN (void)
5690 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
5691 to changed the state as if the insn were scheduled when the new
5692 simulated processor cycle finishes.
5695 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN (void)
5696 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
5697 used to initialize data used by the previous hook.
5700 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD (void)
5701 This hook controls better choosing an insn from the ready insn queue
5702 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
5703 chooses the first insn from the queue. If the hook returns a positive
5704 value, an additional scheduler code tries all permutations of
5705 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
5706 subsequent ready insns to choose an insn whose issue will result in
5707 maximal number of issued insns on the same cycle. For the
5708 @acronym{VLIW} processor, the code could actually solve the problem of
5709 packing simple insns into the @acronym{VLIW} insn. Of course, if the
5710 rules of @acronym{VLIW} packing are described in the automaton.
5712 This code also could be used for superscalar @acronym{RISC}
5713 processors. Let us consider a superscalar @acronym{RISC} processor
5714 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
5715 @var{B}, some insns can be executed only in pipelines @var{B} or
5716 @var{C}, and one insn can be executed in pipeline @var{B}. The
5717 processor may issue the 1st insn into @var{A} and the 2nd one into
5718 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
5719 until the next cycle. If the scheduler issues the 3rd insn the first,
5720 the processor could issue all 3 insns per cycle.
5722 Actually this code demonstrates advantages of the automaton based
5723 pipeline hazard recognizer. We try quickly and easy many insn
5724 schedules to choose the best one.
5726 The default is no multipass scheduling.
5729 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD (rtx)
5731 This hook controls what insns from the ready insn queue will be
5732 considered for the multipass insn scheduling. If the hook returns
5733 zero for insn passed as the parameter, the insn will be not chosen to
5736 The default is that any ready insns can be chosen to be issued.
5739 @deftypefn {Target Hook} int TARGET_SCHED_DFA_NEW_CYCLE (FILE *, int, rtx, int, int, int *)
5741 This hook is called by the insn scheduler before issuing insn passed
5742 as the third parameter on given cycle. If the hook returns nonzero,
5743 the insn is not issued on given processors cycle. Instead of that,
5744 the processor cycle is advanced. If the value passed through the last
5745 parameter is zero, the insn ready queue is not sorted on the new cycle
5746 start as usually. The first parameter passes file for debugging
5747 output. The second one passes the scheduler verbose level of the
5748 debugging output. The forth and the fifth parameter values are
5749 correspondingly processor cycle on which the previous insn has been
5750 issued and the current processor cycle.
5753 @deftypefn {Target Hook} bool TARGET_SCHED_IS_COSTLY_DEPENDENCE (rtx @var{insn1}, rtx @var{insn2}, rtx @var{dep_link}, int @var{dep_cost}, int @var{distance})
5754 This hook is used to define which dependences are considered costly by
5755 the target, so costly that it is not advisable to schedule the insns that
5756 are involved in the dependence too close to one another. The parameters
5757 to this hook are as follows: The second parameter @var{insn2} is dependent
5758 upon the first parameter @var{insn1}. The dependence between @var{insn1}
5759 and @var{insn2} is represented by the third parameter @var{dep_link}. The
5760 fourth parameter @var{cost} is the cost of the dependence, and the fifth
5761 parameter @var{distance} is the distance in cycles between the two insns.
5762 The hook returns @code{true} if considering the distance between the two
5763 insns the dependence between them is considered costly by the target,
5764 and @code{false} otherwise.
5766 Defining this hook can be useful in multiple-issue out-of-order machines,
5767 where (a) it's practically hopeless to predict the actual data/resource
5768 delays, however: (b) there's a better chance to predict the actual grouping
5769 that will be formed, and (c) correctly emulating the grouping can be very
5770 important. In such targets one may want to allow issuing dependent insns
5771 closer to one another - i.e, closer than the dependence distance; however,
5772 not in cases of "costly dependences", which this hooks allows to define.
5775 Macros in the following table are generated by the program
5776 @file{genattr} and can be useful for writing the hooks.
5778 @defmac MAX_DFA_ISSUE_RATE
5779 The macro definition is generated in the automaton based pipeline
5780 description interface. Its value is calculated from the automaton
5781 based pipeline description and is equal to maximal number of all insns
5782 described in constructions @samp{define_insn_reservation} which can be
5783 issued on the same processor cycle.
5787 @section Dividing the Output into Sections (Texts, Data, @dots{})
5788 @c the above section title is WAY too long. maybe cut the part between
5789 @c the (...)? --mew 10feb93
5791 An object file is divided into sections containing different types of
5792 data. In the most common case, there are three sections: the @dfn{text
5793 section}, which holds instructions and read-only data; the @dfn{data
5794 section}, which holds initialized writable data; and the @dfn{bss
5795 section}, which holds uninitialized data. Some systems have other kinds
5798 The compiler must tell the assembler when to switch sections. These
5799 macros control what commands to output to tell the assembler this. You
5800 can also define additional sections.
5802 @defmac TEXT_SECTION_ASM_OP
5803 A C expression whose value is a string, including spacing, containing the
5804 assembler operation that should precede instructions and read-only data.
5805 Normally @code{"\t.text"} is right.
5808 @defmac HOT_TEXT_SECTION_NAME
5809 If defined, a C string constant for the name of the section containing most
5810 frequently executed functions of the program. If not defined, GCC will provide
5811 a default definition if the target supports named sections.
5814 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
5815 If defined, a C string constant for the name of the section containing unlikely
5816 executed functions in the program.
5819 @defmac DATA_SECTION_ASM_OP
5820 A C expression whose value is a string, including spacing, containing the
5821 assembler operation to identify the following data as writable initialized
5822 data. Normally @code{"\t.data"} is right.
5825 @defmac READONLY_DATA_SECTION_ASM_OP
5826 A C expression whose value is a string, including spacing, containing the
5827 assembler operation to identify the following data as read-only initialized
5831 @defmac READONLY_DATA_SECTION
5832 A macro naming a function to call to switch to the proper section for
5833 read-only data. The default is to use @code{READONLY_DATA_SECTION_ASM_OP}
5834 if defined, else fall back to @code{text_section}.
5836 The most common definition will be @code{data_section}, if the target
5837 does not have a special read-only data section, and does not put data
5838 in the text section.
5841 @defmac BSS_SECTION_ASM_OP
5842 If defined, a C expression whose value is a string, including spacing,
5843 containing the assembler operation to identify the following data as
5844 uninitialized global data. If not defined, and neither
5845 @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined,
5846 uninitialized global data will be output in the data section if
5847 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
5851 @defmac INIT_SECTION_ASM_OP
5852 If defined, a C expression whose value is a string, including spacing,
5853 containing the assembler operation to identify the following data as
5854 initialization code. If not defined, GCC will assume such a section does
5858 @defmac FINI_SECTION_ASM_OP
5859 If defined, a C expression whose value is a string, including spacing,
5860 containing the assembler operation to identify the following data as
5861 finalization code. If not defined, GCC will assume such a section does
5865 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
5866 If defined, an ASM statement that switches to a different section
5867 via @var{section_op}, calls @var{function}, and switches back to
5868 the text section. This is used in @file{crtstuff.c} if
5869 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
5870 to initialization and finalization functions from the init and fini
5871 sections. By default, this macro uses a simple function call. Some
5872 ports need hand-crafted assembly code to avoid dependencies on
5873 registers initialized in the function prologue or to ensure that
5874 constant pools don't end up too far way in the text section.
5877 @defmac FORCE_CODE_SECTION_ALIGN
5878 If defined, an ASM statement that aligns a code section to some
5879 arbitrary boundary. This is used to force all fragments of the
5880 @code{.init} and @code{.fini} sections to have to same alignment
5881 and thus prevent the linker from having to add any padding.
5886 @defmac EXTRA_SECTIONS
5887 A list of names for sections other than the standard two, which are
5888 @code{in_text} and @code{in_data}. You need not define this macro
5889 on a system with no other sections (that GCC needs to use).
5892 @findex text_section
5893 @findex data_section
5894 @defmac EXTRA_SECTION_FUNCTIONS
5895 One or more functions to be defined in @file{varasm.c}. These
5896 functions should do jobs analogous to those of @code{text_section} and
5897 @code{data_section}, for your additional sections. Do not define this
5898 macro if you do not define @code{EXTRA_SECTIONS}.
5901 @defmac JUMP_TABLES_IN_TEXT_SECTION
5902 Define this macro to be an expression with a nonzero value if jump
5903 tables (for @code{tablejump} insns) should be output in the text
5904 section, along with the assembler instructions. Otherwise, the
5905 readonly data section is used.
5907 This macro is irrelevant if there is no separate readonly data section.
5910 @deftypefn {Target Hook} void TARGET_ASM_SELECT_SECTION (tree @var{exp}, int @var{reloc}, unsigned HOST_WIDE_INT @var{align})
5911 Switches to the appropriate section for output of @var{exp}. You can
5912 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
5913 some sort. @var{reloc} indicates whether the initial value of @var{exp}
5914 requires link-time relocations. Bit 0 is set when variable contains
5915 local relocations only, while bit 1 is set for global relocations.
5916 Select the section by calling @code{data_section} or one of the
5917 alternatives for other sections. @var{align} is the constant alignment
5920 The default version of this function takes care of putting read-only
5921 variables in @code{readonly_data_section}.
5924 @deftypefn {Target Hook} void TARGET_ASM_UNIQUE_SECTION (tree @var{decl}, int @var{reloc})
5925 Build up a unique section name, expressed as a @code{STRING_CST} node,
5926 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
5927 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
5928 the initial value of @var{exp} requires link-time relocations.
5930 The default version of this function appends the symbol name to the
5931 ELF section name that would normally be used for the symbol. For
5932 example, the function @code{foo} would be placed in @code{.text.foo}.
5933 Whatever the actual target object format, this is often good enough.
5936 @deftypefn {Target Hook} void TARGET_ASM_SELECT_RTX_SECTION (enum machine_mode @var{mode}, rtx @var{x}, unsigned HOST_WIDE_INT @var{align})
5937 Switches to the appropriate section for output of constant pool entry
5938 @var{x} in @var{mode}. You can assume that @var{x} is some kind of
5939 constant in RTL@. The argument @var{mode} is redundant except in the
5940 case of a @code{const_int} rtx. Select the section by calling
5941 @code{readonly_data_section} or one of the alternatives for other
5942 sections. @var{align} is the constant alignment in bits.
5944 The default version of this function takes care of putting symbolic
5945 constants in @code{flag_pic} mode in @code{data_section} and everything
5946 else in @code{readonly_data_section}.
5949 @deftypefn {Target Hook} void TARGET_ENCODE_SECTION_INFO (tree @var{decl}, rtx @var{rtl}, int @var{new_decl_p})
5950 Define this hook if references to a symbol or a constant must be
5951 treated differently depending on something about the variable or
5952 function named by the symbol (such as what section it is in).
5954 The hook is executed immediately after rtl has been created for
5955 @var{decl}, which may be a variable or function declaration or
5956 an entry in the constant pool. In either case, @var{rtl} is the
5957 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
5958 in this hook; that field may not have been initialized yet.
5960 In the case of a constant, it is safe to assume that the rtl is
5961 a @code{mem} whose address is a @code{symbol_ref}. Most decls
5962 will also have this form, but that is not guaranteed. Global
5963 register variables, for instance, will have a @code{reg} for their
5964 rtl. (Normally the right thing to do with such unusual rtl is
5967 The @var{new_decl_p} argument will be true if this is the first time
5968 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
5969 be false for subsequent invocations, which will happen for duplicate
5970 declarations. Whether or not anything must be done for the duplicate
5971 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
5972 @var{new_decl_p} is always true when the hook is called for a constant.
5974 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
5975 The usual thing for this hook to do is to record flags in the
5976 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
5977 Historically, the name string was modified if it was necessary to
5978 encode more than one bit of information, but this practice is now
5979 discouraged; use @code{SYMBOL_REF_FLAGS}.
5981 The default definition of this hook, @code{default_encode_section_info}
5982 in @file{varasm.c}, sets a number of commonly-useful bits in
5983 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
5984 before overriding it.
5987 @deftypefn {Target Hook} const char *TARGET_STRIP_NAME_ENCODING (const char *name)
5988 Decode @var{name} and return the real name part, sans
5989 the characters that @code{TARGET_ENCODE_SECTION_INFO}
5993 @deftypefn {Target Hook} bool TARGET_IN_SMALL_DATA_P (tree @var{exp})
5994 Returns true if @var{exp} should be placed into a ``small data'' section.
5995 The default version of this hook always returns false.
5998 @deftypevar {Target Hook} bool TARGET_HAVE_SRODATA_SECTION
5999 Contains the value true if the target places read-only
6000 ``small data'' into a separate section. The default value is false.
6003 @deftypefn {Target Hook} bool TARGET_BINDS_LOCAL_P (tree @var{exp})
6004 Returns true if @var{exp} names an object for which name resolution
6005 rules must resolve to the current ``module'' (dynamic shared library
6006 or executable image).
6008 The default version of this hook implements the name resolution rules
6009 for ELF, which has a looser model of global name binding than other
6010 currently supported object file formats.
6013 @deftypevar {Target Hook} bool TARGET_HAVE_TLS
6014 Contains the value true if the target supports thread-local storage.
6015 The default value is false.
6020 @section Position Independent Code
6021 @cindex position independent code
6024 This section describes macros that help implement generation of position
6025 independent code. Simply defining these macros is not enough to
6026 generate valid PIC; you must also add support to the macros
6027 @code{GO_IF_LEGITIMATE_ADDRESS} and @code{PRINT_OPERAND_ADDRESS}, as
6028 well as @code{LEGITIMIZE_ADDRESS}. You must modify the definition of
6029 @samp{movsi} to do something appropriate when the source operand
6030 contains a symbolic address. You may also need to alter the handling of
6031 switch statements so that they use relative addresses.
6032 @c i rearranged the order of the macros above to try to force one of
6033 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
6035 @defmac PIC_OFFSET_TABLE_REGNUM
6036 The register number of the register used to address a table of static
6037 data addresses in memory. In some cases this register is defined by a
6038 processor's ``application binary interface'' (ABI)@. When this macro
6039 is defined, RTL is generated for this register once, as with the stack
6040 pointer and frame pointer registers. If this macro is not defined, it
6041 is up to the machine-dependent files to allocate such a register (if
6042 necessary). Note that this register must be fixed when in use (e.g.@:
6043 when @code{flag_pic} is true).
6046 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
6047 Define this macro if the register defined by
6048 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define
6049 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
6052 @defmac FINALIZE_PIC
6053 By generating position-independent code, when two different programs (A
6054 and B) share a common library (libC.a), the text of the library can be
6055 shared whether or not the library is linked at the same address for both
6056 programs. In some of these environments, position-independent code
6057 requires not only the use of different addressing modes, but also
6058 special code to enable the use of these addressing modes.
6060 The @code{FINALIZE_PIC} macro serves as a hook to emit these special
6061 codes once the function is being compiled into assembly code, but not
6062 before. (It is not done before, because in the case of compiling an
6063 inline function, it would lead to multiple PIC prologues being
6064 included in functions which used inline functions and were compiled to
6068 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
6069 A C expression that is nonzero if @var{x} is a legitimate immediate
6070 operand on the target machine when generating position independent code.
6071 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
6072 check this. You can also assume @var{flag_pic} is true, so you need not
6073 check it either. You need not define this macro if all constants
6074 (including @code{SYMBOL_REF}) can be immediate operands when generating
6075 position independent code.
6078 @node Assembler Format
6079 @section Defining the Output Assembler Language
6081 This section describes macros whose principal purpose is to describe how
6082 to write instructions in assembler language---rather than what the
6086 * File Framework:: Structural information for the assembler file.
6087 * Data Output:: Output of constants (numbers, strings, addresses).
6088 * Uninitialized Data:: Output of uninitialized variables.
6089 * Label Output:: Output and generation of labels.
6090 * Initialization:: General principles of initialization
6091 and termination routines.
6092 * Macros for Initialization::
6093 Specific macros that control the handling of
6094 initialization and termination routines.
6095 * Instruction Output:: Output of actual instructions.
6096 * Dispatch Tables:: Output of jump tables.
6097 * Exception Region Output:: Output of exception region code.
6098 * Alignment Output:: Pseudo ops for alignment and skipping data.
6101 @node File Framework
6102 @subsection The Overall Framework of an Assembler File
6103 @cindex assembler format
6104 @cindex output of assembler code
6106 @c prevent bad page break with this line
6107 This describes the overall framework of an assembly file.
6109 @deftypefn {Target Hook} void TARGET_ASM_FILE_START ()
6110 @findex default_file_start
6111 Output to @code{asm_out_file} any text which the assembler expects to
6112 find at the beginning of a file. The default behavior is controlled
6113 by two flags, documented below. Unless your target's assembler is
6114 quite unusual, if you override the default, you should call
6115 @code{default_file_start} at some point in your target hook. This
6116 lets other target files rely on these variables.
6119 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_APP_OFF
6120 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
6121 printed as the very first line in the assembly file, unless
6122 @option{-fverbose-asm} is in effect. (If that macro has been defined
6123 to the empty string, this variable has no effect.) With the normal
6124 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
6125 assembler that it need not bother stripping comments or extra
6126 whitespace from its input. This allows it to work a bit faster.
6128 The default is false. You should not set it to true unless you have
6129 verified that your port does not generate any extra whitespace or
6130 comments that will cause GAS to issue errors in NO_APP mode.
6133 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_FILE_DIRECTIVE
6134 If this flag is true, @code{output_file_directive} will be called
6135 for the primary source file, immediately after printing
6136 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
6137 this to be done. The default is false.
6140 @deftypefn {Target Hook} void TARGET_ASM_FILE_END ()
6141 Output to @code{asm_out_file} any text which the assembler expects
6142 to find at the end of a file. The default is to output nothing.
6145 @deftypefun void file_end_indicate_exec_stack ()
6146 Some systems use a common convention, the @samp{.note.GNU-stack}
6147 special section, to indicate whether or not an object file relies on
6148 the stack being executable. If your system uses this convention, you
6149 should define @code{TARGET_ASM_FILE_END} to this function. If you
6150 need to do other things in that hook, have your hook function call
6154 @defmac ASM_COMMENT_START
6155 A C string constant describing how to begin a comment in the target
6156 assembler language. The compiler assumes that the comment will end at
6157 the end of the line.
6161 A C string constant for text to be output before each @code{asm}
6162 statement or group of consecutive ones. Normally this is
6163 @code{"#APP"}, which is a comment that has no effect on most
6164 assemblers but tells the GNU assembler that it must check the lines
6165 that follow for all valid assembler constructs.
6169 A C string constant for text to be output after each @code{asm}
6170 statement or group of consecutive ones. Normally this is
6171 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
6172 time-saving assumptions that are valid for ordinary compiler output.
6175 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
6176 A C statement to output COFF information or DWARF debugging information
6177 which indicates that filename @var{name} is the current source file to
6178 the stdio stream @var{stream}.
6180 This macro need not be defined if the standard form of output
6181 for the file format in use is appropriate.
6184 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
6185 A C statement to output the string @var{string} to the stdio stream
6186 @var{stream}. If you do not call the function @code{output_quoted_string}
6187 in your config files, GCC will only call it to output filenames to
6188 the assembler source. So you can use it to canonicalize the format
6189 of the filename using this macro.
6192 @defmac ASM_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
6193 A C statement to output DBX or SDB debugging information before code
6194 for line number @var{line} of the current source file to the
6195 stdio stream @var{stream}. @var{counter} is the number of time the
6196 macro was invoked, including the current invocation; it is intended
6197 to generate unique labels in the assembly output.
6199 This macro need not be defined if the standard form of debugging
6200 information for the debugger in use is appropriate.
6203 @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string})
6204 A C statement to output something to the assembler file to handle a
6205 @samp{#ident} directive containing the text @var{string}. If this
6206 macro is not defined, nothing is output for a @samp{#ident} directive.
6209 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, unsigned int @var{align})
6210 Output assembly directives to switch to section @var{name}. The section
6211 should have attributes as specified by @var{flags}, which is a bit mask
6212 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{align}
6213 is nonzero, it contains an alignment in bytes to be used for the section,
6214 otherwise some target default should be used. Only targets that must
6215 specify an alignment within the section directive need pay attention to
6216 @var{align} -- we will still use @code{ASM_OUTPUT_ALIGN}.
6219 @deftypefn {Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
6220 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
6223 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
6224 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
6225 based on a variable or function decl, a section name, and whether or not the
6226 declaration's initializer may contain runtime relocations. @var{decl} may be
6227 null, in which case read-write data should be assumed.
6229 The default version if this function handles choosing code vs data,
6230 read-only vs read-write data, and @code{flag_pic}. You should only
6231 need to override this if your target has special flags that might be
6232 set via @code{__attribute__}.
6237 @subsection Output of Data
6240 @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP
6241 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
6242 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
6243 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
6244 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
6245 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
6246 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
6247 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
6248 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
6249 These hooks specify assembly directives for creating certain kinds
6250 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
6251 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
6252 aligned two-byte object, and so on. Any of the hooks may be
6253 @code{NULL}, indicating that no suitable directive is available.
6255 The compiler will print these strings at the start of a new line,
6256 followed immediately by the object's initial value. In most cases,
6257 the string should contain a tab, a pseudo-op, and then another tab.
6260 @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p})
6261 The @code{assemble_integer} function uses this hook to output an
6262 integer object. @var{x} is the object's value, @var{size} is its size
6263 in bytes and @var{aligned_p} indicates whether it is aligned. The
6264 function should return @code{true} if it was able to output the
6265 object. If it returns false, @code{assemble_integer} will try to
6266 split the object into smaller parts.
6268 The default implementation of this hook will use the
6269 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
6270 when the relevant string is @code{NULL}.
6273 @defmac OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
6274 A C statement to recognize @var{rtx} patterns that
6275 @code{output_addr_const} can't deal with, and output assembly code to
6276 @var{stream} corresponding to the pattern @var{x}. This may be used to
6277 allow machine-dependent @code{UNSPEC}s to appear within constants.
6279 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
6280 @code{goto fail}, so that a standard error message is printed. If it
6281 prints an error message itself, by calling, for example,
6282 @code{output_operand_lossage}, it may just complete normally.
6285 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
6286 A C statement to output to the stdio stream @var{stream} an assembler
6287 instruction to assemble a string constant containing the @var{len}
6288 bytes at @var{ptr}. @var{ptr} will be a C expression of type
6289 @code{char *} and @var{len} a C expression of type @code{int}.
6291 If the assembler has a @code{.ascii} pseudo-op as found in the
6292 Berkeley Unix assembler, do not define the macro
6293 @code{ASM_OUTPUT_ASCII}.
6296 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
6297 A C statement to output word @var{n} of a function descriptor for
6298 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
6299 is defined, and is otherwise unused.
6302 @defmac CONSTANT_POOL_BEFORE_FUNCTION
6303 You may define this macro as a C expression. You should define the
6304 expression to have a nonzero value if GCC should output the constant
6305 pool for a function before the code for the function, or a zero value if
6306 GCC should output the constant pool after the function. If you do
6307 not define this macro, the usual case, GCC will output the constant
6308 pool before the function.
6311 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
6312 A C statement to output assembler commands to define the start of the
6313 constant pool for a function. @var{funname} is a string giving
6314 the name of the function. Should the return type of the function
6315 be required, it can be obtained via @var{fundecl}. @var{size}
6316 is the size, in bytes, of the constant pool that will be written
6317 immediately after this call.
6319 If no constant-pool prefix is required, the usual case, this macro need
6323 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
6324 A C statement (with or without semicolon) to output a constant in the
6325 constant pool, if it needs special treatment. (This macro need not do
6326 anything for RTL expressions that can be output normally.)
6328 The argument @var{file} is the standard I/O stream to output the
6329 assembler code on. @var{x} is the RTL expression for the constant to
6330 output, and @var{mode} is the machine mode (in case @var{x} is a
6331 @samp{const_int}). @var{align} is the required alignment for the value
6332 @var{x}; you should output an assembler directive to force this much
6335 The argument @var{labelno} is a number to use in an internal label for
6336 the address of this pool entry. The definition of this macro is
6337 responsible for outputting the label definition at the proper place.
6338 Here is how to do this:
6341 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
6344 When you output a pool entry specially, you should end with a
6345 @code{goto} to the label @var{jumpto}. This will prevent the same pool
6346 entry from being output a second time in the usual manner.
6348 You need not define this macro if it would do nothing.
6351 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
6352 A C statement to output assembler commands to at the end of the constant
6353 pool for a function. @var{funname} is a string giving the name of the
6354 function. Should the return type of the function be required, you can
6355 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
6356 constant pool that GCC wrote immediately before this call.
6358 If no constant-pool epilogue is required, the usual case, you need not
6362 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C})
6363 Define this macro as a C expression which is nonzero if @var{C} is
6364 used as a logical line separator by the assembler.
6366 If you do not define this macro, the default is that only
6367 the character @samp{;} is treated as a logical line separator.
6370 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
6371 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
6372 These target hooks are C string constants, describing the syntax in the
6373 assembler for grouping arithmetic expressions. If not overridden, they
6374 default to normal parentheses, which is correct for most assemblers.
6377 These macros are provided by @file{real.h} for writing the definitions
6378 of @code{ASM_OUTPUT_DOUBLE} and the like:
6380 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
6381 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
6382 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
6383 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the target's
6384 floating point representation, and store its bit pattern in the variable
6385 @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE}, this variable should
6386 be a simple @code{long int}. For the others, it should be an array of
6387 @code{long int}. The number of elements in this array is determined by
6388 the size of the desired target floating point data type: 32 bits of it
6389 go in each @code{long int} array element. Each array element holds 32
6390 bits of the result, even if @code{long int} is wider than 32 bits on the
6393 The array element values are designed so that you can print them out
6394 using @code{fprintf} in the order they should appear in the target
6398 @node Uninitialized Data
6399 @subsection Output of Uninitialized Variables
6401 Each of the macros in this section is used to do the whole job of
6402 outputting a single uninitialized variable.
6404 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
6405 A C statement (sans semicolon) to output to the stdio stream
6406 @var{stream} the assembler definition of a common-label named
6407 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
6408 is the size rounded up to whatever alignment the caller wants.
6410 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
6411 output the name itself; before and after that, output the additional
6412 assembler syntax for defining the name, and a newline.
6414 This macro controls how the assembler definitions of uninitialized
6415 common global variables are output.
6418 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
6419 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
6420 separate, explicit argument. If you define this macro, it is used in
6421 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
6422 handling the required alignment of the variable. The alignment is specified
6423 as the number of bits.
6426 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
6427 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
6428 variable to be output, if there is one, or @code{NULL_TREE} if there
6429 is no corresponding variable. If you define this macro, GCC will use it
6430 in place of both @code{ASM_OUTPUT_COMMON} and
6431 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
6432 the variable's decl in order to chose what to output.
6435 @defmac ASM_OUTPUT_SHARED_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
6436 If defined, it is similar to @code{ASM_OUTPUT_COMMON}, except that it
6437 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_COMMON}
6441 @defmac ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
6442 A C statement (sans semicolon) to output to the stdio stream
6443 @var{stream} the assembler definition of uninitialized global @var{decl} named
6444 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
6445 is the size rounded up to whatever alignment the caller wants.
6447 Try to use function @code{asm_output_bss} defined in @file{varasm.c} when
6448 defining this macro. If unable, use the expression
6449 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
6450 before and after that, output the additional assembler syntax for defining
6451 the name, and a newline.
6453 This macro controls how the assembler definitions of uninitialized global
6454 variables are output. This macro exists to properly support languages like
6455 C++ which do not have @code{common} data. However, this macro currently
6456 is not defined for all targets. If this macro and
6457 @code{ASM_OUTPUT_ALIGNED_BSS} are not defined then @code{ASM_OUTPUT_COMMON}
6458 or @code{ASM_OUTPUT_ALIGNED_COMMON} or
6459 @code{ASM_OUTPUT_ALIGNED_DECL_COMMON} is used.
6462 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
6463 Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a
6464 separate, explicit argument. If you define this macro, it is used in
6465 place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in
6466 handling the required alignment of the variable. The alignment is specified
6467 as the number of bits.
6469 Try to use function @code{asm_output_aligned_bss} defined in file
6470 @file{varasm.c} when defining this macro.
6473 @defmac ASM_OUTPUT_SHARED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
6474 If defined, it is similar to @code{ASM_OUTPUT_BSS}, except that it
6475 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_BSS}
6479 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
6480 A C statement (sans semicolon) to output to the stdio stream
6481 @var{stream} the assembler definition of a local-common-label named
6482 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
6483 is the size rounded up to whatever alignment the caller wants.
6485 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
6486 output the name itself; before and after that, output the additional
6487 assembler syntax for defining the name, and a newline.
6489 This macro controls how the assembler definitions of uninitialized
6490 static variables are output.
6493 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
6494 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
6495 separate, explicit argument. If you define this macro, it is used in
6496 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
6497 handling the required alignment of the variable. The alignment is specified
6498 as the number of bits.
6501 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
6502 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
6503 variable to be output, if there is one, or @code{NULL_TREE} if there
6504 is no corresponding variable. If you define this macro, GCC will use it
6505 in place of both @code{ASM_OUTPUT_DECL} and
6506 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
6507 the variable's decl in order to chose what to output.
6510 @defmac ASM_OUTPUT_SHARED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
6511 If defined, it is similar to @code{ASM_OUTPUT_LOCAL}, except that it
6512 is used when @var{name} is shared. If not defined, @code{ASM_OUTPUT_LOCAL}
6517 @subsection Output and Generation of Labels
6519 @c prevent bad page break with this line
6520 This is about outputting labels.
6522 @findex assemble_name
6523 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
6524 A C statement (sans semicolon) to output to the stdio stream
6525 @var{stream} the assembler definition of a label named @var{name}.
6526 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
6527 output the name itself; before and after that, output the additional
6528 assembler syntax for defining the name, and a newline. A default
6529 definition of this macro is provided which is correct for most systems.
6533 A C string containing the appropriate assembler directive to specify the
6534 size of a symbol, without any arguments. On systems that use ELF, the
6535 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
6536 systems, the default is not to define this macro.
6538 Define this macro only if it is correct to use the default definitions
6539 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
6540 for your system. If you need your own custom definitions of those
6541 macros, or if you do not need explicit symbol sizes at all, do not
6545 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
6546 A C statement (sans semicolon) to output to the stdio stream
6547 @var{stream} a directive telling the assembler that the size of the
6548 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
6549 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
6553 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
6554 A C statement (sans semicolon) to output to the stdio stream
6555 @var{stream} a directive telling the assembler to calculate the size of
6556 the symbol @var{name} by subtracting its address from the current
6559 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
6560 provided. The default assumes that the assembler recognizes a special
6561 @samp{.} symbol as referring to the current address, and can calculate
6562 the difference between this and another symbol. If your assembler does
6563 not recognize @samp{.} or cannot do calculations with it, you will need
6564 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
6568 A C string containing the appropriate assembler directive to specify the
6569 type of a symbol, without any arguments. On systems that use ELF, the
6570 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
6571 systems, the default is not to define this macro.
6573 Define this macro only if it is correct to use the default definition of
6574 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
6575 custom definition of this macro, or if you do not need explicit symbol
6576 types at all, do not define this macro.
6579 @defmac TYPE_OPERAND_FMT
6580 A C string which specifies (using @code{printf} syntax) the format of
6581 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
6582 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
6583 the default is not to define this macro.
6585 Define this macro only if it is correct to use the default definition of
6586 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
6587 custom definition of this macro, or if you do not need explicit symbol
6588 types at all, do not define this macro.
6591 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
6592 A C statement (sans semicolon) to output to the stdio stream
6593 @var{stream} a directive telling the assembler that the type of the
6594 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
6595 that string is always either @samp{"function"} or @samp{"object"}, but
6596 you should not count on this.
6598 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
6599 definition of this macro is provided.
6602 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
6603 A C statement (sans semicolon) to output to the stdio stream
6604 @var{stream} any text necessary for declaring the name @var{name} of a
6605 function which is being defined. This macro is responsible for
6606 outputting the label definition (perhaps using
6607 @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the
6608 @code{FUNCTION_DECL} tree node representing the function.
6610 If this macro is not defined, then the function name is defined in the
6611 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
6613 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
6617 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
6618 A C statement (sans semicolon) to output to the stdio stream
6619 @var{stream} any text necessary for declaring the size of a function
6620 which is being defined. The argument @var{name} is the name of the
6621 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
6622 representing the function.
6624 If this macro is not defined, then the function size is not defined.
6626 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
6630 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
6631 A C statement (sans semicolon) to output to the stdio stream
6632 @var{stream} any text necessary for declaring the name @var{name} of an
6633 initialized variable which is being defined. This macro must output the
6634 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
6635 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
6637 If this macro is not defined, then the variable name is defined in the
6638 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
6640 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
6641 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
6644 @defmac ASM_DECLARE_CONSTANT_NAME (@var{stream}, @var{name}, @var{exp}, @var{size})
6645 A C statement (sans semicolon) to output to the stdio stream
6646 @var{stream} any text necessary for declaring the name @var{name} of a
6647 constant which is being defined. This macro is responsible for
6648 outputting the label definition (perhaps using
6649 @code{ASM_OUTPUT_LABEL}). The argument @var{exp} is the
6650 value of the constant, and @var{size} is the size of the constant
6651 in bytes. @var{name} will be an internal label.
6653 If this macro is not defined, then the @var{name} is defined in the
6654 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
6656 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
6660 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
6661 A C statement (sans semicolon) to output to the stdio stream
6662 @var{stream} any text necessary for claiming a register @var{regno}
6663 for a global variable @var{decl} with name @var{name}.
6665 If you don't define this macro, that is equivalent to defining it to do
6669 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
6670 A C statement (sans semicolon) to finish up declaring a variable name
6671 once the compiler has processed its initializer fully and thus has had a
6672 chance to determine the size of an array when controlled by an
6673 initializer. This is used on systems where it's necessary to declare
6674 something about the size of the object.
6676 If you don't define this macro, that is equivalent to defining it to do
6679 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
6680 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
6683 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_LABEL (FILE *@var{stream}, const char *@var{name})
6684 This target hook is a function to output to the stdio stream
6685 @var{stream} some commands that will make the label @var{name} global;
6686 that is, available for reference from other files.
6688 The default implementation relies on a proper definition of
6689 @code{GLOBAL_ASM_OP}.
6692 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
6693 A C statement (sans semicolon) to output to the stdio stream
6694 @var{stream} some commands that will make the label @var{name} weak;
6695 that is, available for reference from other files but only used if
6696 no other definition is available. Use the expression
6697 @code{assemble_name (@var{stream}, @var{name})} to output the name
6698 itself; before and after that, output the additional assembler syntax
6699 for making that name weak, and a newline.
6701 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
6702 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
6706 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
6707 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
6708 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
6709 or variable decl. If @var{value} is not @code{NULL}, this C statement
6710 should output to the stdio stream @var{stream} assembler code which
6711 defines (equates) the weak symbol @var{name} to have the value
6712 @var{value}. If @var{value} is @code{NULL}, it should output commands
6713 to make @var{name} weak.
6716 @defmac SUPPORTS_WEAK
6717 A C expression which evaluates to true if the target supports weak symbols.
6719 If you don't define this macro, @file{defaults.h} provides a default
6720 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
6721 is defined, the default definition is @samp{1}; otherwise, it is
6722 @samp{0}. Define this macro if you want to control weak symbol support
6723 with a compiler flag such as @option{-melf}.
6726 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
6727 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
6728 public symbol such that extra copies in multiple translation units will
6729 be discarded by the linker. Define this macro if your object file
6730 format provides support for this concept, such as the @samp{COMDAT}
6731 section flags in the Microsoft Windows PE/COFF format, and this support
6732 requires changes to @var{decl}, such as putting it in a separate section.
6735 @defmac SUPPORTS_ONE_ONLY
6736 A C expression which evaluates to true if the target supports one-only
6739 If you don't define this macro, @file{varasm.c} provides a default
6740 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
6741 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
6742 you want to control one-only symbol support with a compiler flag, or if
6743 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
6744 be emitted as one-only.
6747 @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_VISIBILITY (tree @var{decl}, const char *@var{visibility})
6748 This target hook is a function to output to @var{asm_out_file} some
6749 commands that will make the symbol(s) associated with @var{decl} have
6750 hidden, protected or internal visibility as specified by @var{visibility}.
6753 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
6754 A C expression that evaluates to true if the target's linker expects
6755 that weak symbols do not appear in a static archive's table of contents.
6756 The default is @code{0}.
6758 Leaving weak symbols out of an archive's table of contents means that,
6759 if a symbol will only have a definition in one translation unit and
6760 will have undefined references from other translation units, that
6761 symbol should not be weak. Defining this macro to be nonzero will
6762 thus have the effect that certain symbols that would normally be weak
6763 (explicit template instantiations, and vtables for polymorphic classes
6764 with noninline key methods) will instead be nonweak.
6766 The C++ ABI requires this macro to be zero. Define this macro for
6767 targets where full C++ ABI compliance is impossible and where linker
6768 restrictions require weak symbols to be left out of a static archive's
6772 @defmac TARGET_SUPPORTS_HIDDEN
6773 A C expression that evaluates to true if the target supports hidden
6774 visibility. By default this expression is true if and only if
6775 @code{HAS_GAS_HIDDEN} is defined. Set this macro if the
6776 @code{HAS_GAS_HIDDEN} macro gives the wrong answer for this
6777 target. (For example, if the target's mechanism for supporting
6778 hidden visibility is not the same as GAS's.)
6781 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
6782 A C statement (sans semicolon) to output to the stdio stream
6783 @var{stream} any text necessary for declaring the name of an external
6784 symbol named @var{name} which is referenced in this compilation but
6785 not defined. The value of @var{decl} is the tree node for the
6788 This macro need not be defined if it does not need to output anything.
6789 The GNU assembler and most Unix assemblers don't require anything.
6792 @deftypefn {Target Hook} void TARGET_ASM_EXTERNAL_LIBCALL (rtx @var{symref})
6793 This target hook is a function to output to @var{asm_out_file} an assembler
6794 pseudo-op to declare a library function name external. The name of the
6795 library function is given by @var{symref}, which is a @code{symbol_ref}.
6798 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
6799 A C statement (sans semicolon) to output to the stdio stream
6800 @var{stream} a reference in assembler syntax to a label named
6801 @var{name}. This should add @samp{_} to the front of the name, if that
6802 is customary on your operating system, as it is in most Berkeley Unix
6803 systems. This macro is used in @code{assemble_name}.
6806 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
6807 A C statement (sans semicolon) to output a reference to
6808 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
6809 will be used to output the name of the symbol. This macro may be used
6810 to modify the way a symbol is referenced depending on information
6811 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
6814 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
6815 A C statement (sans semicolon) to output a reference to @var{buf}, the
6816 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
6817 @code{assemble_name} will be used to output the name of the symbol.
6818 This macro is not used by @code{output_asm_label}, or the @code{%l}
6819 specifier that calls it; the intention is that this macro should be set
6820 when it is necessary to output a label differently when its address is
6824 @deftypefn {Target Hook} void TARGET_ASM_INTERNAL_LABEL (FILE *@var{stream}, const char *@var{prefix}, unsigned long @var{labelno})
6825 A function to output to the stdio stream @var{stream} a label whose
6826 name is made from the string @var{prefix} and the number @var{labelno}.
6828 It is absolutely essential that these labels be distinct from the labels
6829 used for user-level functions and variables. Otherwise, certain programs
6830 will have name conflicts with internal labels.
6832 It is desirable to exclude internal labels from the symbol table of the
6833 object file. Most assemblers have a naming convention for labels that
6834 should be excluded; on many systems, the letter @samp{L} at the
6835 beginning of a label has this effect. You should find out what
6836 convention your system uses, and follow it.
6838 The default version of this function utilizes ASM_GENERATE_INTERNAL_LABEL.
6841 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
6842 A C statement to output to the stdio stream @var{stream} a debug info
6843 label whose name is made from the string @var{prefix} and the number
6844 @var{num}. This is useful for VLIW targets, where debug info labels
6845 may need to be treated differently than branch target labels. On some
6846 systems, branch target labels must be at the beginning of instruction
6847 bundles, but debug info labels can occur in the middle of instruction
6850 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
6854 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
6855 A C statement to store into the string @var{string} a label whose name
6856 is made from the string @var{prefix} and the number @var{num}.
6858 This string, when output subsequently by @code{assemble_name}, should
6859 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
6860 with the same @var{prefix} and @var{num}.
6862 If the string begins with @samp{*}, then @code{assemble_name} will
6863 output the rest of the string unchanged. It is often convenient for
6864 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
6865 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
6866 to output the string, and may change it. (Of course,
6867 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
6868 you should know what it does on your machine.)
6871 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
6872 A C expression to assign to @var{outvar} (which is a variable of type
6873 @code{char *}) a newly allocated string made from the string
6874 @var{name} and the number @var{number}, with some suitable punctuation
6875 added. Use @code{alloca} to get space for the string.
6877 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
6878 produce an assembler label for an internal static variable whose name is
6879 @var{name}. Therefore, the string must be such as to result in valid
6880 assembler code. The argument @var{number} is different each time this
6881 macro is executed; it prevents conflicts between similarly-named
6882 internal static variables in different scopes.
6884 Ideally this string should not be a valid C identifier, to prevent any
6885 conflict with the user's own symbols. Most assemblers allow periods
6886 or percent signs in assembler symbols; putting at least one of these
6887 between the name and the number will suffice.
6889 If this macro is not defined, a default definition will be provided
6890 which is correct for most systems.
6893 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
6894 A C statement to output to the stdio stream @var{stream} assembler code
6895 which defines (equates) the symbol @var{name} to have the value @var{value}.
6898 If @code{SET_ASM_OP} is defined, a default definition is provided which is
6899 correct for most systems.
6902 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
6903 A C statement to output to the stdio stream @var{stream} assembler code
6904 which defines (equates) the symbol whose tree node is @var{decl_of_name}
6905 to have the value of the tree node @var{decl_of_value}. This macro will
6906 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
6907 the tree nodes are available.
6910 If @code{SET_ASM_OP} is defined, a default definition is provided which is
6911 correct for most systems.
6914 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
6915 A C statement to output to the stdio stream @var{stream} assembler code
6916 which defines (equates) the weak symbol @var{name} to have the value
6917 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
6918 an undefined weak symbol.
6920 Define this macro if the target only supports weak aliases; define
6921 @code{ASM_OUTPUT_DEF} instead if possible.
6924 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
6925 Define this macro to override the default assembler names used for
6926 Objective-C methods.
6928 The default name is a unique method number followed by the name of the
6929 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
6930 the category is also included in the assembler name (e.g.@:
6933 These names are safe on most systems, but make debugging difficult since
6934 the method's selector is not present in the name. Therefore, particular
6935 systems define other ways of computing names.
6937 @var{buf} is an expression of type @code{char *} which gives you a
6938 buffer in which to store the name; its length is as long as
6939 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
6940 50 characters extra.
6942 The argument @var{is_inst} specifies whether the method is an instance
6943 method or a class method; @var{class_name} is the name of the class;
6944 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
6945 in a category); and @var{sel_name} is the name of the selector.
6947 On systems where the assembler can handle quoted names, you can use this
6948 macro to provide more human-readable names.
6951 @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
6952 A C statement (sans semicolon) to output to the stdio stream
6953 @var{stream} commands to declare that the label @var{name} is an
6954 Objective-C class reference. This is only needed for targets whose
6955 linkers have special support for NeXT-style runtimes.
6958 @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
6959 A C statement (sans semicolon) to output to the stdio stream
6960 @var{stream} commands to declare that the label @var{name} is an
6961 unresolved Objective-C class reference. This is only needed for targets
6962 whose linkers have special support for NeXT-style runtimes.
6965 @node Initialization
6966 @subsection How Initialization Functions Are Handled
6967 @cindex initialization routines
6968 @cindex termination routines
6969 @cindex constructors, output of
6970 @cindex destructors, output of
6972 The compiled code for certain languages includes @dfn{constructors}
6973 (also called @dfn{initialization routines})---functions to initialize
6974 data in the program when the program is started. These functions need
6975 to be called before the program is ``started''---that is to say, before
6976 @code{main} is called.
6978 Compiling some languages generates @dfn{destructors} (also called
6979 @dfn{termination routines}) that should be called when the program
6982 To make the initialization and termination functions work, the compiler
6983 must output something in the assembler code to cause those functions to
6984 be called at the appropriate time. When you port the compiler to a new
6985 system, you need to specify how to do this.
6987 There are two major ways that GCC currently supports the execution of
6988 initialization and termination functions. Each way has two variants.
6989 Much of the structure is common to all four variations.
6991 @findex __CTOR_LIST__
6992 @findex __DTOR_LIST__
6993 The linker must build two lists of these functions---a list of
6994 initialization functions, called @code{__CTOR_LIST__}, and a list of
6995 termination functions, called @code{__DTOR_LIST__}.
6997 Each list always begins with an ignored function pointer (which may hold
6998 0, @minus{}1, or a count of the function pointers after it, depending on
6999 the environment). This is followed by a series of zero or more function
7000 pointers to constructors (or destructors), followed by a function
7001 pointer containing zero.
7003 Depending on the operating system and its executable file format, either
7004 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
7005 time and exit time. Constructors are called in reverse order of the
7006 list; destructors in forward order.
7008 The best way to handle static constructors works only for object file
7009 formats which provide arbitrarily-named sections. A section is set
7010 aside for a list of constructors, and another for a list of destructors.
7011 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
7012 object file that defines an initialization function also puts a word in
7013 the constructor section to point to that function. The linker
7014 accumulates all these words into one contiguous @samp{.ctors} section.
7015 Termination functions are handled similarly.
7017 This method will be chosen as the default by @file{target-def.h} if
7018 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
7019 support arbitrary sections, but does support special designated
7020 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
7021 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
7023 When arbitrary sections are available, there are two variants, depending
7024 upon how the code in @file{crtstuff.c} is called. On systems that
7025 support a @dfn{.init} section which is executed at program startup,
7026 parts of @file{crtstuff.c} are compiled into that section. The
7027 program is linked by the @command{gcc} driver like this:
7030 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
7033 The prologue of a function (@code{__init}) appears in the @code{.init}
7034 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
7035 for the function @code{__fini} in the @dfn{.fini} section. Normally these
7036 files are provided by the operating system or by the GNU C library, but
7037 are provided by GCC for a few targets.
7039 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
7040 compiled from @file{crtstuff.c}. They contain, among other things, code
7041 fragments within the @code{.init} and @code{.fini} sections that branch
7042 to routines in the @code{.text} section. The linker will pull all parts
7043 of a section together, which results in a complete @code{__init} function
7044 that invokes the routines we need at startup.
7046 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
7049 If no init section is available, when GCC compiles any function called
7050 @code{main} (or more accurately, any function designated as a program
7051 entry point by the language front end calling @code{expand_main_function}),
7052 it inserts a procedure call to @code{__main} as the first executable code
7053 after the function prologue. The @code{__main} function is defined
7054 in @file{libgcc2.c} and runs the global constructors.
7056 In file formats that don't support arbitrary sections, there are again
7057 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
7058 and an `a.out' format must be used. In this case,
7059 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
7060 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
7061 and with the address of the void function containing the initialization
7062 code as its value. The GNU linker recognizes this as a request to add
7063 the value to a @dfn{set}; the values are accumulated, and are eventually
7064 placed in the executable as a vector in the format described above, with
7065 a leading (ignored) count and a trailing zero element.
7066 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
7067 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
7068 the compilation of @code{main} to call @code{__main} as above, starting
7069 the initialization process.
7071 The last variant uses neither arbitrary sections nor the GNU linker.
7072 This is preferable when you want to do dynamic linking and when using
7073 file formats which the GNU linker does not support, such as `ECOFF'@. In
7074 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
7075 termination functions are recognized simply by their names. This requires
7076 an extra program in the linkage step, called @command{collect2}. This program
7077 pretends to be the linker, for use with GCC; it does its job by running
7078 the ordinary linker, but also arranges to include the vectors of
7079 initialization and termination functions. These functions are called
7080 via @code{__main} as described above. In order to use this method,
7081 @code{use_collect2} must be defined in the target in @file{config.gcc}.
7084 The following section describes the specific macros that control and
7085 customize the handling of initialization and termination functions.
7088 @node Macros for Initialization
7089 @subsection Macros Controlling Initialization Routines
7091 Here are the macros that control how the compiler handles initialization
7092 and termination functions:
7094 @defmac INIT_SECTION_ASM_OP
7095 If defined, a C string constant, including spacing, for the assembler
7096 operation to identify the following data as initialization code. If not
7097 defined, GCC will assume such a section does not exist. When you are
7098 using special sections for initialization and termination functions, this
7099 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
7100 run the initialization functions.
7103 @defmac HAS_INIT_SECTION
7104 If defined, @code{main} will not call @code{__main} as described above.
7105 This macro should be defined for systems that control start-up code
7106 on a symbol-by-symbol basis, such as OSF/1, and should not
7107 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
7110 @defmac LD_INIT_SWITCH
7111 If defined, a C string constant for a switch that tells the linker that
7112 the following symbol is an initialization routine.
7115 @defmac LD_FINI_SWITCH
7116 If defined, a C string constant for a switch that tells the linker that
7117 the following symbol is a finalization routine.
7120 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
7121 If defined, a C statement that will write a function that can be
7122 automatically called when a shared library is loaded. The function
7123 should call @var{func}, which takes no arguments. If not defined, and
7124 the object format requires an explicit initialization function, then a
7125 function called @code{_GLOBAL__DI} will be generated.
7127 This function and the following one are used by collect2 when linking a
7128 shared library that needs constructors or destructors, or has DWARF2
7129 exception tables embedded in the code.
7132 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
7133 If defined, a C statement that will write a function that can be
7134 automatically called when a shared library is unloaded. The function
7135 should call @var{func}, which takes no arguments. If not defined, and
7136 the object format requires an explicit finalization function, then a
7137 function called @code{_GLOBAL__DD} will be generated.
7140 @defmac INVOKE__main
7141 If defined, @code{main} will call @code{__main} despite the presence of
7142 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
7143 where the init section is not actually run automatically, but is still
7144 useful for collecting the lists of constructors and destructors.
7147 @defmac SUPPORTS_INIT_PRIORITY
7148 If nonzero, the C++ @code{init_priority} attribute is supported and the
7149 compiler should emit instructions to control the order of initialization
7150 of objects. If zero, the compiler will issue an error message upon
7151 encountering an @code{init_priority} attribute.
7154 @deftypefn {Target Hook} bool TARGET_HAVE_CTORS_DTORS
7155 This value is true if the target supports some ``native'' method of
7156 collecting constructors and destructors to be run at startup and exit.
7157 It is false if we must use @command{collect2}.
7160 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
7161 If defined, a function that outputs assembler code to arrange to call
7162 the function referenced by @var{symbol} at initialization time.
7164 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
7165 no arguments and with no return value. If the target supports initialization
7166 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
7167 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
7169 If this macro is not defined by the target, a suitable default will
7170 be chosen if (1) the target supports arbitrary section names, (2) the
7171 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
7175 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
7176 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
7177 functions rather than initialization functions.
7180 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
7181 generated for the generated object file will have static linkage.
7183 If your system uses @command{collect2} as the means of processing
7184 constructors, then that program normally uses @command{nm} to scan
7185 an object file for constructor functions to be called.
7187 On certain kinds of systems, you can define this macro to make
7188 @command{collect2} work faster (and, in some cases, make it work at all):
7190 @defmac OBJECT_FORMAT_COFF
7191 Define this macro if the system uses COFF (Common Object File Format)
7192 object files, so that @command{collect2} can assume this format and scan
7193 object files directly for dynamic constructor/destructor functions.
7195 This macro is effective only in a native compiler; @command{collect2} as
7196 part of a cross compiler always uses @command{nm} for the target machine.
7199 @defmac COLLECT_PARSE_FLAG (@var{flag})
7200 Define this macro to be C code that examines @command{collect2} command
7201 line option @var{flag} and performs special actions if
7202 @command{collect2} needs to behave differently depending on @var{flag}.
7205 @defmac REAL_NM_FILE_NAME
7206 Define this macro as a C string constant containing the file name to use
7207 to execute @command{nm}. The default is to search the path normally for
7210 If your system supports shared libraries and has a program to list the
7211 dynamic dependencies of a given library or executable, you can define
7212 these macros to enable support for running initialization and
7213 termination functions in shared libraries:
7217 Define this macro to a C string constant containing the name of the program
7218 which lists dynamic dependencies, like @command{"ldd"} under SunOS 4.
7221 @defmac PARSE_LDD_OUTPUT (@var{ptr})
7222 Define this macro to be C code that extracts filenames from the output
7223 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
7224 of type @code{char *} that points to the beginning of a line of output
7225 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
7226 code must advance @var{ptr} to the beginning of the filename on that
7227 line. Otherwise, it must set @var{ptr} to @code{NULL}.
7230 @node Instruction Output
7231 @subsection Output of Assembler Instructions
7233 @c prevent bad page break with this line
7234 This describes assembler instruction output.
7236 @defmac REGISTER_NAMES
7237 A C initializer containing the assembler's names for the machine
7238 registers, each one as a C string constant. This is what translates
7239 register numbers in the compiler into assembler language.
7242 @defmac ADDITIONAL_REGISTER_NAMES
7243 If defined, a C initializer for an array of structures containing a name
7244 and a register number. This macro defines additional names for hard
7245 registers, thus allowing the @code{asm} option in declarations to refer
7246 to registers using alternate names.
7249 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
7250 Define this macro if you are using an unusual assembler that
7251 requires different names for the machine instructions.
7253 The definition is a C statement or statements which output an
7254 assembler instruction opcode to the stdio stream @var{stream}. The
7255 macro-operand @var{ptr} is a variable of type @code{char *} which
7256 points to the opcode name in its ``internal'' form---the form that is
7257 written in the machine description. The definition should output the
7258 opcode name to @var{stream}, performing any translation you desire, and
7259 increment the variable @var{ptr} to point at the end of the opcode
7260 so that it will not be output twice.
7262 In fact, your macro definition may process less than the entire opcode
7263 name, or more than the opcode name; but if you want to process text
7264 that includes @samp{%}-sequences to substitute operands, you must take
7265 care of the substitution yourself. Just be sure to increment
7266 @var{ptr} over whatever text should not be output normally.
7268 @findex recog_data.operand
7269 If you need to look at the operand values, they can be found as the
7270 elements of @code{recog_data.operand}.
7272 If the macro definition does nothing, the instruction is output
7276 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
7277 If defined, a C statement to be executed just prior to the output of
7278 assembler code for @var{insn}, to modify the extracted operands so
7279 they will be output differently.
7281 Here the argument @var{opvec} is the vector containing the operands
7282 extracted from @var{insn}, and @var{noperands} is the number of
7283 elements of the vector which contain meaningful data for this insn.
7284 The contents of this vector are what will be used to convert the insn
7285 template into assembler code, so you can change the assembler output
7286 by changing the contents of the vector.
7288 This macro is useful when various assembler syntaxes share a single
7289 file of instruction patterns; by defining this macro differently, you
7290 can cause a large class of instructions to be output differently (such
7291 as with rearranged operands). Naturally, variations in assembler
7292 syntax affecting individual insn patterns ought to be handled by
7293 writing conditional output routines in those patterns.
7295 If this macro is not defined, it is equivalent to a null statement.
7298 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
7299 A C compound statement to output to stdio stream @var{stream} the
7300 assembler syntax for an instruction operand @var{x}. @var{x} is an
7303 @var{code} is a value that can be used to specify one of several ways
7304 of printing the operand. It is used when identical operands must be
7305 printed differently depending on the context. @var{code} comes from
7306 the @samp{%} specification that was used to request printing of the
7307 operand. If the specification was just @samp{%@var{digit}} then
7308 @var{code} is 0; if the specification was @samp{%@var{ltr}
7309 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
7312 If @var{x} is a register, this macro should print the register's name.
7313 The names can be found in an array @code{reg_names} whose type is
7314 @code{char *[]}. @code{reg_names} is initialized from
7315 @code{REGISTER_NAMES}.
7317 When the machine description has a specification @samp{%@var{punct}}
7318 (a @samp{%} followed by a punctuation character), this macro is called
7319 with a null pointer for @var{x} and the punctuation character for
7323 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
7324 A C expression which evaluates to true if @var{code} is a valid
7325 punctuation character for use in the @code{PRINT_OPERAND} macro. If
7326 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
7327 punctuation characters (except for the standard one, @samp{%}) are used
7331 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
7332 A C compound statement to output to stdio stream @var{stream} the
7333 assembler syntax for an instruction operand that is a memory reference
7334 whose address is @var{x}. @var{x} is an RTL expression.
7336 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
7337 On some machines, the syntax for a symbolic address depends on the
7338 section that the address refers to. On these machines, define the hook
7339 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
7340 @code{symbol_ref}, and then check for it here. @xref{Assembler
7344 @findex dbr_sequence_length
7345 @defmac DBR_OUTPUT_SEQEND (@var{file})
7346 A C statement, to be executed after all slot-filler instructions have
7347 been output. If necessary, call @code{dbr_sequence_length} to
7348 determine the number of slots filled in a sequence (zero if not
7349 currently outputting a sequence), to decide how many no-ops to output,
7352 Don't define this macro if it has nothing to do, but it is helpful in
7353 reading assembly output if the extent of the delay sequence is made
7354 explicit (e.g.@: with white space).
7357 @findex final_sequence
7358 Note that output routines for instructions with delay slots must be
7359 prepared to deal with not being output as part of a sequence
7360 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
7361 found.) The variable @code{final_sequence} is null when not
7362 processing a sequence, otherwise it contains the @code{sequence} rtx
7366 @defmac REGISTER_PREFIX
7367 @defmacx LOCAL_LABEL_PREFIX
7368 @defmacx USER_LABEL_PREFIX
7369 @defmacx IMMEDIATE_PREFIX
7370 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
7371 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
7372 @file{final.c}). These are useful when a single @file{md} file must
7373 support multiple assembler formats. In that case, the various @file{tm.h}
7374 files can define these macros differently.
7377 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
7378 If defined this macro should expand to a series of @code{case}
7379 statements which will be parsed inside the @code{switch} statement of
7380 the @code{asm_fprintf} function. This allows targets to define extra
7381 printf formats which may useful when generating their assembler
7382 statements. Note that uppercase letters are reserved for future
7383 generic extensions to asm_fprintf, and so are not available to target
7384 specific code. The output file is given by the parameter @var{file}.
7385 The varargs input pointer is @var{argptr} and the rest of the format
7386 string, starting the character after the one that is being switched
7387 upon, is pointed to by @var{format}.
7390 @defmac ASSEMBLER_DIALECT
7391 If your target supports multiple dialects of assembler language (such as
7392 different opcodes), define this macro as a C expression that gives the
7393 numeric index of the assembler language dialect to use, with zero as the
7396 If this macro is defined, you may use constructs of the form
7398 @samp{@{option0|option1|option2@dots{}@}}
7401 in the output templates of patterns (@pxref{Output Template}) or in the
7402 first argument of @code{asm_fprintf}. This construct outputs
7403 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
7404 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
7405 within these strings retain their usual meaning. If there are fewer
7406 alternatives within the braces than the value of
7407 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
7409 If you do not define this macro, the characters @samp{@{}, @samp{|} and
7410 @samp{@}} do not have any special meaning when used in templates or
7411 operands to @code{asm_fprintf}.
7413 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
7414 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
7415 the variations in assembler language syntax with that mechanism. Define
7416 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
7417 if the syntax variant are larger and involve such things as different
7418 opcodes or operand order.
7421 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
7422 A C expression to output to @var{stream} some assembler code
7423 which will push hard register number @var{regno} onto the stack.
7424 The code need not be optimal, since this macro is used only when
7428 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
7429 A C expression to output to @var{stream} some assembler code
7430 which will pop hard register number @var{regno} off of the stack.
7431 The code need not be optimal, since this macro is used only when
7435 @node Dispatch Tables
7436 @subsection Output of Dispatch Tables
7438 @c prevent bad page break with this line
7439 This concerns dispatch tables.
7441 @cindex dispatch table
7442 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
7443 A C statement to output to the stdio stream @var{stream} an assembler
7444 pseudo-instruction to generate a difference between two labels.
7445 @var{value} and @var{rel} are the numbers of two internal labels. The
7446 definitions of these labels are output using
7447 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
7448 way here. For example,
7451 fprintf (@var{stream}, "\t.word L%d-L%d\n",
7452 @var{value}, @var{rel})
7455 You must provide this macro on machines where the addresses in a
7456 dispatch table are relative to the table's own address. If defined, GCC
7457 will also use this macro on all machines when producing PIC@.
7458 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
7459 mode and flags can be read.
7462 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
7463 This macro should be provided on machines where the addresses
7464 in a dispatch table are absolute.
7466 The definition should be a C statement to output to the stdio stream
7467 @var{stream} an assembler pseudo-instruction to generate a reference to
7468 a label. @var{value} is the number of an internal label whose
7469 definition is output using @code{(*targetm.asm_out.internal_label)}.
7473 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
7477 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
7478 Define this if the label before a jump-table needs to be output
7479 specially. The first three arguments are the same as for
7480 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
7481 jump-table which follows (a @code{jump_insn} containing an
7482 @code{addr_vec} or @code{addr_diff_vec}).
7484 This feature is used on system V to output a @code{swbeg} statement
7487 If this macro is not defined, these labels are output with
7488 @code{(*targetm.asm_out.internal_label)}.
7491 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
7492 Define this if something special must be output at the end of a
7493 jump-table. The definition should be a C statement to be executed
7494 after the assembler code for the table is written. It should write
7495 the appropriate code to stdio stream @var{stream}. The argument
7496 @var{table} is the jump-table insn, and @var{num} is the label-number
7497 of the preceding label.
7499 If this macro is not defined, nothing special is output at the end of
7503 @deftypefn {Target Hook} void TARGET_ASM_EMIT_UNWIND_LABEL (@var{stream}, @var{decl}, @var{for_eh}, @var{empty})
7504 This target hook emits a label at the beginning of each FDE. It
7505 should be defined on targets where FDEs need special labels, and it
7506 should write the appropriate label, for the FDE associated with the
7507 function declaration @var{decl}, to the stdio stream @var{stream}.
7508 The third argument, @var{for_eh}, is a boolean: true if this is for an
7509 exception table. The fourth argument, @var{empty}, is a boolean:
7510 true if this is a placeholder label for an omitted FDE.
7512 The default is that FDEs are not given nonlocal labels.
7515 @deftypefn {Taget Hook} void TARGET_UNWIND_EMIT (FILE * @var{stream}, rtx @var{insn})
7516 This target hook emits and assembly directives required to unwind the
7517 given instruction. This is only used when TARGET_UNWIND_INFO is set.
7520 @node Exception Region Output
7521 @subsection Assembler Commands for Exception Regions
7523 @c prevent bad page break with this line
7525 This describes commands marking the start and the end of an exception
7528 @defmac EH_FRAME_SECTION_NAME
7529 If defined, a C string constant for the name of the section containing
7530 exception handling frame unwind information. If not defined, GCC will
7531 provide a default definition if the target supports named sections.
7532 @file{crtstuff.c} uses this macro to switch to the appropriate section.
7534 You should define this symbol if your target supports DWARF 2 frame
7535 unwind information and the default definition does not work.
7538 @defmac EH_FRAME_IN_DATA_SECTION
7539 If defined, DWARF 2 frame unwind information will be placed in the
7540 data section even though the target supports named sections. This
7541 might be necessary, for instance, if the system linker does garbage
7542 collection and sections cannot be marked as not to be collected.
7544 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
7548 @defmac MASK_RETURN_ADDR
7549 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
7550 that it does not contain any extraneous set bits in it.
7553 @defmac DWARF2_UNWIND_INFO
7554 Define this macro to 0 if your target supports DWARF 2 frame unwind
7555 information, but it does not yet work with exception handling.
7556 Otherwise, if your target supports this information (if it defines
7557 @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP}
7558 or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of
7561 If this macro is defined to 1, the DWARF 2 unwinder will be the default
7562 exception handling mechanism; otherwise, @code{setjmp}/@code{longjmp} will be used by
7565 If this macro is defined to anything, the DWARF 2 unwinder will be used
7566 instead of inline unwinders and @code{__unwind_function} in the non-@code{setjmp} case.
7569 @defmac TARGET_UNWIND_INFO
7570 Define this macro if your target has ABI specified unwind tables. Usually
7571 these will be output by @code{TARGET_UNWIND_EMIT}.
7574 @defmac MUST_USE_SJLJ_EXCEPTIONS
7575 This macro need only be defined if @code{DWARF2_UNWIND_INFO} is
7576 runtime-variable. In that case, @file{except.h} cannot correctly
7577 determine the corresponding definition of
7578 @code{MUST_USE_SJLJ_EXCEPTIONS}, so the target must provide it directly.
7581 @defmac DWARF_CIE_DATA_ALIGNMENT
7582 This macro need only be defined if the target might save registers in the
7583 function prologue at an offset to the stack pointer that is not aligned to
7584 @code{UNITS_PER_WORD}. The definition should be the negative minimum
7585 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
7586 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
7587 the target supports DWARF 2 frame unwind information.
7590 @deftypefn {Target Hook} void TARGET_ASM_EXCEPTION_SECTION ()
7591 If defined, a function that switches to the section in which the main
7592 exception table is to be placed (@pxref{Sections}). The default is a
7593 function that switches to a section named @code{.gcc_except_table} on
7594 machines that support named sections via
7595 @code{TARGET_ASM_NAMED_SECTION}, otherwise if @option{-fpic} or
7596 @option{-fPIC} is in effect, the @code{data_section}, otherwise the
7597 @code{readonly_data_section}.
7600 @deftypefn {Target Hook} void TARGET_ASM_EH_FRAME_SECTION ()
7601 If defined, a function that switches to the section in which the DWARF 2
7602 frame unwind information to be placed (@pxref{Sections}). The default
7603 is a function that outputs a standard GAS section directive, if
7604 @code{EH_FRAME_SECTION_NAME} is defined, or else a data section
7605 directive followed by a synthetic label.
7608 @deftypevar {Target Hook} bool TARGET_TERMINATE_DW2_EH_FRAME_INFO
7609 Contains the value true if the target should add a zero word onto the
7610 end of a Dwarf-2 frame info section when used for exception handling.
7611 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
7615 @deftypefn {Target Hook} rtx TARGET_DWARF_REGISTER_SPAN (rtx @var{reg})
7616 Given a register, this hook should return a parallel of registers to
7617 represent where to find the register pieces. Define this hook if the
7618 register and its mode are represented in Dwarf in non-contiguous
7619 locations, or if the register should be represented in more than one
7620 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
7621 If not defined, the default is to return @code{NULL_RTX}.
7624 @node Alignment Output
7625 @subsection Assembler Commands for Alignment
7627 @c prevent bad page break with this line
7628 This describes commands for alignment.
7630 @defmac JUMP_ALIGN (@var{label})
7631 The alignment (log base 2) to put in front of @var{label}, which is
7632 a common destination of jumps and has no fallthru incoming edge.
7634 This macro need not be defined if you don't want any special alignment
7635 to be done at such a time. Most machine descriptions do not currently
7638 Unless it's necessary to inspect the @var{label} parameter, it is better
7639 to set the variable @var{align_jumps} in the target's
7640 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7641 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
7644 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
7645 The alignment (log base 2) to put in front of @var{label}, which follows
7648 This macro need not be defined if you don't want any special alignment
7649 to be done at such a time. Most machine descriptions do not currently
7653 @defmac LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
7654 The maximum number of bytes to skip when applying
7655 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
7656 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7659 @defmac LOOP_ALIGN (@var{label})
7660 The alignment (log base 2) to put in front of @var{label}, which follows
7661 a @code{NOTE_INSN_LOOP_BEG} note.
7663 This macro need not be defined if you don't want any special alignment
7664 to be done at such a time. Most machine descriptions do not currently
7667 Unless it's necessary to inspect the @var{label} parameter, it is better
7668 to set the variable @code{align_loops} in the target's
7669 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7670 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
7673 @defmac LOOP_ALIGN_MAX_SKIP
7674 The maximum number of bytes to skip when applying @code{LOOP_ALIGN}.
7675 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7678 @defmac LABEL_ALIGN (@var{label})
7679 The alignment (log base 2) to put in front of @var{label}.
7680 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
7681 the maximum of the specified values is used.
7683 Unless it's necessary to inspect the @var{label} parameter, it is better
7684 to set the variable @code{align_labels} in the target's
7685 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
7686 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
7689 @defmac LABEL_ALIGN_MAX_SKIP
7690 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}.
7691 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
7694 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
7695 A C statement to output to the stdio stream @var{stream} an assembler
7696 instruction to advance the location counter by @var{nbytes} bytes.
7697 Those bytes should be zero when loaded. @var{nbytes} will be a C
7698 expression of type @code{int}.
7701 @defmac ASM_NO_SKIP_IN_TEXT
7702 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
7703 text section because it fails to put zeros in the bytes that are skipped.
7704 This is true on many Unix systems, where the pseudo--op to skip bytes
7705 produces no-op instructions rather than zeros when used in the text
7709 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
7710 A C statement to output to the stdio stream @var{stream} an assembler
7711 command to advance the location counter to a multiple of 2 to the
7712 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
7715 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
7716 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
7717 for padding, if necessary.
7720 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
7721 A C statement to output to the stdio stream @var{stream} an assembler
7722 command to advance the location counter to a multiple of 2 to the
7723 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
7724 satisfy the alignment request. @var{power} and @var{max_skip} will be
7725 a C expression of type @code{int}.
7729 @node Debugging Info
7730 @section Controlling Debugging Information Format
7732 @c prevent bad page break with this line
7733 This describes how to specify debugging information.
7736 * All Debuggers:: Macros that affect all debugging formats uniformly.
7737 * DBX Options:: Macros enabling specific options in DBX format.
7738 * DBX Hooks:: Hook macros for varying DBX format.
7739 * File Names and DBX:: Macros controlling output of file names in DBX format.
7740 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
7741 * VMS Debug:: Macros for VMS debug format.
7745 @subsection Macros Affecting All Debugging Formats
7747 @c prevent bad page break with this line
7748 These macros affect all debugging formats.
7750 @defmac DBX_REGISTER_NUMBER (@var{regno})
7751 A C expression that returns the DBX register number for the compiler
7752 register number @var{regno}. In the default macro provided, the value
7753 of this expression will be @var{regno} itself. But sometimes there are
7754 some registers that the compiler knows about and DBX does not, or vice
7755 versa. In such cases, some register may need to have one number in the
7756 compiler and another for DBX@.
7758 If two registers have consecutive numbers inside GCC, and they can be
7759 used as a pair to hold a multiword value, then they @emph{must} have
7760 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
7761 Otherwise, debuggers will be unable to access such a pair, because they
7762 expect register pairs to be consecutive in their own numbering scheme.
7764 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
7765 does not preserve register pairs, then what you must do instead is
7766 redefine the actual register numbering scheme.
7769 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
7770 A C expression that returns the integer offset value for an automatic
7771 variable having address @var{x} (an RTL expression). The default
7772 computation assumes that @var{x} is based on the frame-pointer and
7773 gives the offset from the frame-pointer. This is required for targets
7774 that produce debugging output for DBX or COFF-style debugging output
7775 for SDB and allow the frame-pointer to be eliminated when the
7776 @option{-g} options is used.
7779 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
7780 A C expression that returns the integer offset value for an argument
7781 having address @var{x} (an RTL expression). The nominal offset is
7785 @defmac PREFERRED_DEBUGGING_TYPE
7786 A C expression that returns the type of debugging output GCC should
7787 produce when the user specifies just @option{-g}. Define
7788 this if you have arranged for GCC to support more than one format of
7789 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
7790 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
7791 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
7793 When the user specifies @option{-ggdb}, GCC normally also uses the
7794 value of this macro to select the debugging output format, but with two
7795 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
7796 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
7797 defined, GCC uses @code{DBX_DEBUG}.
7799 The value of this macro only affects the default debugging output; the
7800 user can always get a specific type of output by using @option{-gstabs},
7801 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
7805 @subsection Specific Options for DBX Output
7807 @c prevent bad page break with this line
7808 These are specific options for DBX output.
7810 @defmac DBX_DEBUGGING_INFO
7811 Define this macro if GCC should produce debugging output for DBX
7812 in response to the @option{-g} option.
7815 @defmac XCOFF_DEBUGGING_INFO
7816 Define this macro if GCC should produce XCOFF format debugging output
7817 in response to the @option{-g} option. This is a variant of DBX format.
7820 @defmac DEFAULT_GDB_EXTENSIONS
7821 Define this macro to control whether GCC should by default generate
7822 GDB's extended version of DBX debugging information (assuming DBX-format
7823 debugging information is enabled at all). If you don't define the
7824 macro, the default is 1: always generate the extended information
7825 if there is any occasion to.
7828 @defmac DEBUG_SYMS_TEXT
7829 Define this macro if all @code{.stabs} commands should be output while
7830 in the text section.
7833 @defmac ASM_STABS_OP
7834 A C string constant, including spacing, naming the assembler pseudo op to
7835 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
7836 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
7837 applies only to DBX debugging information format.
7840 @defmac ASM_STABD_OP
7841 A C string constant, including spacing, naming the assembler pseudo op to
7842 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
7843 value is the current location. If you don't define this macro,
7844 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
7848 @defmac ASM_STABN_OP
7849 A C string constant, including spacing, naming the assembler pseudo op to
7850 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
7851 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
7852 macro applies only to DBX debugging information format.
7855 @defmac DBX_NO_XREFS
7856 Define this macro if DBX on your system does not support the construct
7857 @samp{xs@var{tagname}}. On some systems, this construct is used to
7858 describe a forward reference to a structure named @var{tagname}.
7859 On other systems, this construct is not supported at all.
7862 @defmac DBX_CONTIN_LENGTH
7863 A symbol name in DBX-format debugging information is normally
7864 continued (split into two separate @code{.stabs} directives) when it
7865 exceeds a certain length (by default, 80 characters). On some
7866 operating systems, DBX requires this splitting; on others, splitting
7867 must not be done. You can inhibit splitting by defining this macro
7868 with the value zero. You can override the default splitting-length by
7869 defining this macro as an expression for the length you desire.
7872 @defmac DBX_CONTIN_CHAR
7873 Normally continuation is indicated by adding a @samp{\} character to
7874 the end of a @code{.stabs} string when a continuation follows. To use
7875 a different character instead, define this macro as a character
7876 constant for the character you want to use. Do not define this macro
7877 if backslash is correct for your system.
7880 @defmac DBX_STATIC_STAB_DATA_SECTION
7881 Define this macro if it is necessary to go to the data section before
7882 outputting the @samp{.stabs} pseudo-op for a non-global static
7886 @defmac DBX_TYPE_DECL_STABS_CODE
7887 The value to use in the ``code'' field of the @code{.stabs} directive
7888 for a typedef. The default is @code{N_LSYM}.
7891 @defmac DBX_STATIC_CONST_VAR_CODE
7892 The value to use in the ``code'' field of the @code{.stabs} directive
7893 for a static variable located in the text section. DBX format does not
7894 provide any ``right'' way to do this. The default is @code{N_FUN}.
7897 @defmac DBX_REGPARM_STABS_CODE
7898 The value to use in the ``code'' field of the @code{.stabs} directive
7899 for a parameter passed in registers. DBX format does not provide any
7900 ``right'' way to do this. The default is @code{N_RSYM}.
7903 @defmac DBX_REGPARM_STABS_LETTER
7904 The letter to use in DBX symbol data to identify a symbol as a parameter
7905 passed in registers. DBX format does not customarily provide any way to
7906 do this. The default is @code{'P'}.
7909 @defmac DBX_MEMPARM_STABS_LETTER
7910 The letter to use in DBX symbol data to identify a symbol as a stack
7911 parameter. The default is @code{'p'}.
7914 @defmac DBX_FUNCTION_FIRST
7915 Define this macro if the DBX information for a function and its
7916 arguments should precede the assembler code for the function. Normally,
7917 in DBX format, the debugging information entirely follows the assembler
7921 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
7922 Define this macro if the value of a symbol describing the scope of a
7923 block (@code{N_LBRAC} or @code{N_RBRAC}) should be relative to the start
7924 of the enclosing function. Normally, GCC uses an absolute address.
7927 @defmac DBX_USE_BINCL
7928 Define this macro if GCC should generate @code{N_BINCL} and
7929 @code{N_EINCL} stabs for included header files, as on Sun systems. This
7930 macro also directs GCC to output a type number as a pair of a file
7931 number and a type number within the file. Normally, GCC does not
7932 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
7933 number for a type number.
7937 @subsection Open-Ended Hooks for DBX Format
7939 @c prevent bad page break with this line
7940 These are hooks for DBX format.
7942 @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
7943 Define this macro to say how to output to @var{stream} the debugging
7944 information for the start of a scope level for variable names. The
7945 argument @var{name} is the name of an assembler symbol (for use with
7946 @code{assemble_name}) whose value is the address where the scope begins.
7949 @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
7950 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
7953 @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl})
7954 Define this macro if the target machine requires special handling to
7955 output an @code{N_FUN} entry for the function @var{decl}.
7958 @defmac DBX_OUTPUT_FUNCTION_END (@var{stream}, @var{function})
7959 Define this macro if the target machine requires special output at the
7960 end of the debugging information for a function. The definition should
7961 be a C statement (sans semicolon) to output the appropriate information
7962 to @var{stream}. @var{function} is the @code{FUNCTION_DECL} node for
7966 @defmac NO_DBX_FUNCTION_END
7967 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
7968 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
7969 On those machines, define this macro to turn this feature off without
7970 disturbing the rest of the gdb extensions.
7973 @node File Names and DBX
7974 @subsection File Names in DBX Format
7976 @c prevent bad page break with this line
7977 This describes file names in DBX format.
7979 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
7980 A C statement to output DBX debugging information to the stdio stream
7981 @var{stream} which indicates that file @var{name} is the main source
7982 file---the file specified as the input file for compilation.
7983 This macro is called only once, at the beginning of compilation.
7985 This macro need not be defined if the standard form of output
7986 for DBX debugging information is appropriate.
7989 @defmac DBX_OUTPUT_MAIN_SOURCE_DIRECTORY (@var{stream}, @var{name})
7990 A C statement to output DBX debugging information to the stdio stream
7991 @var{stream} which indicates that the current directory during
7992 compilation is named @var{name}.
7994 This macro need not be defined if the standard form of output
7995 for DBX debugging information is appropriate.
7998 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
7999 A C statement to output DBX debugging information at the end of
8000 compilation of the main source file @var{name}.
8002 If you don't define this macro, nothing special is output at the end
8003 of compilation, which is correct for most machines.
8008 @subsection Macros for SDB and DWARF Output
8010 @c prevent bad page break with this line
8011 Here are macros for SDB and DWARF output.
8013 @defmac SDB_DEBUGGING_INFO
8014 Define this macro if GCC should produce COFF-style debugging output
8015 for SDB in response to the @option{-g} option.
8018 @defmac DWARF2_DEBUGGING_INFO
8019 Define this macro if GCC should produce dwarf version 2 format
8020 debugging output in response to the @option{-g} option.
8022 To support optional call frame debugging information, you must also
8023 define @code{INCOMING_RETURN_ADDR_RTX} and either set
8024 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
8025 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
8026 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
8029 @defmac DWARF2_FRAME_INFO
8030 Define this macro to a nonzero value if GCC should always output
8031 Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO}
8032 (@pxref{Exception Region Output} is nonzero, GCC will output this
8033 information not matter how you define @code{DWARF2_FRAME_INFO}.
8036 @defmac DWARF2_GENERATE_TEXT_SECTION_LABEL
8037 By default, the Dwarf 2 debugging information generator will generate a
8038 label to mark the beginning of the text section. If it is better simply
8039 to use the name of the text section itself, rather than an explicit label,
8040 to indicate the beginning of the text section, define this macro to zero.
8043 @defmac DWARF2_ASM_LINE_DEBUG_INFO
8044 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
8045 line debug info sections. This will result in much more compact line number
8046 tables, and hence is desirable if it works.
8049 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
8050 A C statement to issue assembly directives that create a difference
8051 between the two given labels, using an integer of the given size.
8054 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label})
8055 A C statement to issue assembly directives that create a
8056 section-relative reference to the given label, using an integer of the
8060 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
8061 A C statement to issue assembly directives that create a self-relative
8062 reference to the given label, using an integer of the given size.
8065 @defmac PUT_SDB_@dots{}
8066 Define these macros to override the assembler syntax for the special
8067 SDB assembler directives. See @file{sdbout.c} for a list of these
8068 macros and their arguments. If the standard syntax is used, you need
8069 not define them yourself.
8073 Some assemblers do not support a semicolon as a delimiter, even between
8074 SDB assembler directives. In that case, define this macro to be the
8075 delimiter to use (usually @samp{\n}). It is not necessary to define
8076 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
8080 @defmac SDB_GENERATE_FAKE
8081 Define this macro to override the usual method of constructing a dummy
8082 name for anonymous structure and union types. See @file{sdbout.c} for
8086 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
8087 Define this macro to allow references to unknown structure,
8088 union, or enumeration tags to be emitted. Standard COFF does not
8089 allow handling of unknown references, MIPS ECOFF has support for
8093 @defmac SDB_ALLOW_FORWARD_REFERENCES
8094 Define this macro to allow references to structure, union, or
8095 enumeration tags that have not yet been seen to be handled. Some
8096 assemblers choke if forward tags are used, while some require it.
8101 @subsection Macros for VMS Debug Format
8103 @c prevent bad page break with this line
8104 Here are macros for VMS debug format.
8106 @defmac VMS_DEBUGGING_INFO
8107 Define this macro if GCC should produce debugging output for VMS
8108 in response to the @option{-g} option. The default behavior for VMS
8109 is to generate minimal debug info for a traceback in the absence of
8110 @option{-g} unless explicitly overridden with @option{-g0}. This
8111 behavior is controlled by @code{OPTIMIZATION_OPTIONS} and
8112 @code{OVERRIDE_OPTIONS}.
8115 @node Floating Point
8116 @section Cross Compilation and Floating Point
8117 @cindex cross compilation and floating point
8118 @cindex floating point and cross compilation
8120 While all modern machines use twos-complement representation for integers,
8121 there are a variety of representations for floating point numbers. This
8122 means that in a cross-compiler the representation of floating point numbers
8123 in the compiled program may be different from that used in the machine
8124 doing the compilation.
8126 Because different representation systems may offer different amounts of
8127 range and precision, all floating point constants must be represented in
8128 the target machine's format. Therefore, the cross compiler cannot
8129 safely use the host machine's floating point arithmetic; it must emulate
8130 the target's arithmetic. To ensure consistency, GCC always uses
8131 emulation to work with floating point values, even when the host and
8132 target floating point formats are identical.
8134 The following macros are provided by @file{real.h} for the compiler to
8135 use. All parts of the compiler which generate or optimize
8136 floating-point calculations must use these macros. They may evaluate
8137 their operands more than once, so operands must not have side effects.
8139 @defmac REAL_VALUE_TYPE
8140 The C data type to be used to hold a floating point value in the target
8141 machine's format. Typically this is a @code{struct} containing an
8142 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
8146 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
8147 Compares for equality the two values, @var{x} and @var{y}. If the target
8148 floating point format supports negative zeroes and/or NaNs,
8149 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
8150 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
8153 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
8154 Tests whether @var{x} is less than @var{y}.
8157 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
8158 Truncates @var{x} to a signed integer, rounding toward zero.
8161 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
8162 Truncates @var{x} to an unsigned integer, rounding toward zero. If
8163 @var{x} is negative, returns zero.
8166 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
8167 Converts @var{string} into a floating point number in the target machine's
8168 representation for mode @var{mode}. This routine can handle both
8169 decimal and hexadecimal floating point constants, using the syntax
8170 defined by the C language for both.
8173 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
8174 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
8177 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
8178 Determines whether @var{x} represents infinity (positive or negative).
8181 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
8182 Determines whether @var{x} represents a ``NaN'' (not-a-number).
8185 @deftypefn Macro void REAL_ARITHMETIC (REAL_VALUE_TYPE @var{output}, enum tree_code @var{code}, REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
8186 Calculates an arithmetic operation on the two floating point values
8187 @var{x} and @var{y}, storing the result in @var{output} (which must be a
8190 The operation to be performed is specified by @var{code}. Only the
8191 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
8192 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
8194 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
8195 target's floating point format cannot represent infinity, it will call
8196 @code{abort}. Callers should check for this situation first, using
8197 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
8200 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
8201 Returns the negative of the floating point value @var{x}.
8204 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
8205 Returns the absolute value of @var{x}.
8208 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x})
8209 Truncates the floating point value @var{x} to fit in @var{mode}. The
8210 return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an
8211 appropriate bit pattern to be output asa floating constant whose
8212 precision accords with mode @var{mode}.
8215 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
8216 Converts a floating point value @var{x} into a double-precision integer
8217 which is then stored into @var{low} and @var{high}. If the value is not
8218 integral, it is truncated.
8221 @deftypefn Macro void REAL_VALUE_FROM_INT (REAL_VALUE_TYPE @var{x}, HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, enum machine_mode @var{mode})
8222 Converts a double-precision integer found in @var{low} and @var{high},
8223 into a floating point value which is then stored into @var{x}. The
8224 value is truncated to fit in mode @var{mode}.
8227 @node Mode Switching
8228 @section Mode Switching Instructions
8229 @cindex mode switching
8230 The following macros control mode switching optimizations:
8232 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
8233 Define this macro if the port needs extra instructions inserted for mode
8234 switching in an optimizing compilation.
8236 For an example, the SH4 can perform both single and double precision
8237 floating point operations, but to perform a single precision operation,
8238 the FPSCR PR bit has to be cleared, while for a double precision
8239 operation, this bit has to be set. Changing the PR bit requires a general
8240 purpose register as a scratch register, hence these FPSCR sets have to
8241 be inserted before reload, i.e.@: you can't put this into instruction emitting
8242 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
8244 You can have multiple entities that are mode-switched, and select at run time
8245 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
8246 return nonzero for any @var{entity} that needs mode-switching.
8247 If you define this macro, you also have to define
8248 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
8249 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
8250 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
8254 @defmac NUM_MODES_FOR_MODE_SWITCHING
8255 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
8256 initializer for an array of integers. Each initializer element
8257 N refers to an entity that needs mode switching, and specifies the number
8258 of different modes that might need to be set for this entity.
8259 The position of the initializer in the initializer - starting counting at
8260 zero - determines the integer that is used to refer to the mode-switched
8262 In macros that take mode arguments / yield a mode result, modes are
8263 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
8264 switch is needed / supplied.
8267 @defmac MODE_NEEDED (@var{entity}, @var{insn})
8268 @var{entity} is an integer specifying a mode-switched entity. If
8269 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
8270 return an integer value not larger than the corresponding element in
8271 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
8272 be switched into prior to the execution of @var{insn}.
8275 @defmac MODE_AFTER (@var{mode}, @var{insn})
8276 If this macro is defined, it is evaluated for every @var{insn} during
8277 mode switching. It determines the mode that an insn results in (if
8278 different from the incoming mode).
8281 @defmac MODE_ENTRY (@var{entity})
8282 If this macro is defined, it is evaluated for every @var{entity} that needs
8283 mode switching. It should evaluate to an integer, which is a mode that
8284 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
8285 is defined then @code{MODE_EXIT} must be defined.
8288 @defmac MODE_EXIT (@var{entity})
8289 If this macro is defined, it is evaluated for every @var{entity} that needs
8290 mode switching. It should evaluate to an integer, which is a mode that
8291 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
8292 is defined then @code{MODE_ENTRY} must be defined.
8295 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
8296 This macro specifies the order in which modes for @var{entity} are processed.
8297 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
8298 lowest. The value of the macro should be an integer designating a mode
8299 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
8300 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
8301 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
8304 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
8305 Generate one or more insns to set @var{entity} to @var{mode}.
8306 @var{hard_reg_live} is the set of hard registers live at the point where
8307 the insn(s) are to be inserted.
8310 @node Target Attributes
8311 @section Defining target-specific uses of @code{__attribute__}
8312 @cindex target attributes
8313 @cindex machine attributes
8314 @cindex attributes, target-specific
8316 Target-specific attributes may be defined for functions, data and types.
8317 These are described using the following target hooks; they also need to
8318 be documented in @file{extend.texi}.
8320 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
8321 If defined, this target hook points to an array of @samp{struct
8322 attribute_spec} (defined in @file{tree.h}) specifying the machine
8323 specific attributes for this target and some of the restrictions on the
8324 entities to which these attributes are applied and the arguments they
8328 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
8329 If defined, this target hook is a function which returns zero if the attributes on
8330 @var{type1} and @var{type2} are incompatible, one if they are compatible,
8331 and two if they are nearly compatible (which causes a warning to be
8332 generated). If this is not defined, machine-specific attributes are
8333 supposed always to be compatible.
8336 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
8337 If defined, this target hook is a function which assigns default attributes to
8338 newly defined @var{type}.
8341 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
8342 Define this target hook if the merging of type attributes needs special
8343 handling. If defined, the result is a list of the combined
8344 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
8345 that @code{comptypes} has already been called and returned 1. This
8346 function may call @code{merge_attributes} to handle machine-independent
8350 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
8351 Define this target hook if the merging of decl attributes needs special
8352 handling. If defined, the result is a list of the combined
8353 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
8354 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
8355 when this is needed are when one attribute overrides another, or when an
8356 attribute is nullified by a subsequent definition. This function may
8357 call @code{merge_attributes} to handle machine-independent merging.
8359 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
8360 If the only target-specific handling you require is @samp{dllimport}
8361 for Microsoft Windows targets, you should define the macro
8362 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
8363 will then define a function called
8364 @code{merge_dllimport_decl_attributes} which can then be defined as
8365 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
8366 add @code{handle_dll_attribute} in the attribute table for your port
8367 to perform initial processing of the @samp{dllimport} and
8368 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
8369 @file{i386/i386.c}, for example.
8372 @defmac TARGET_DECLSPEC
8373 Define this macro to a non-zero value if you want to treat
8374 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
8375 default, this behavior is enabled only for targets that define
8376 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
8377 of @code{__declspec} is via a built-in macro, but you should not rely
8378 on this implementation detail.
8381 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
8382 Define this target hook if you want to be able to add attributes to a decl
8383 when it is being created. This is normally useful for back ends which
8384 wish to implement a pragma by using the attributes which correspond to
8385 the pragma's effect. The @var{node} argument is the decl which is being
8386 created. The @var{attr_ptr} argument is a pointer to the attribute list
8387 for this decl. The list itself should not be modified, since it may be
8388 shared with other decls, but attributes may be chained on the head of
8389 the list and @code{*@var{attr_ptr}} modified to point to the new
8390 attributes, or a copy of the list may be made if further changes are
8394 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (tree @var{fndecl})
8396 This target hook returns @code{true} if it is ok to inline @var{fndecl}
8397 into the current function, despite its having target-specific
8398 attributes, @code{false} otherwise. By default, if a function has a
8399 target specific attribute attached to it, it will not be inlined.
8402 @node MIPS Coprocessors
8403 @section Defining coprocessor specifics for MIPS targets.
8404 @cindex MIPS coprocessor-definition macros
8406 The MIPS specification allows MIPS implementations to have as many as 4
8407 coprocessors, each with as many as 32 private registers. GCC supports
8408 accessing these registers and transferring values between the registers
8409 and memory using asm-ized variables. For example:
8412 register unsigned int cp0count asm ("c0r1");
8418 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
8419 names may be added as described below, or the default names may be
8420 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
8422 Coprocessor registers are assumed to be epilogue-used; sets to them will
8423 be preserved even if it does not appear that the register is used again
8424 later in the function.
8426 Another note: according to the MIPS spec, coprocessor 1 (if present) is
8427 the FPU. One accesses COP1 registers through standard mips
8428 floating-point support; they are not included in this mechanism.
8430 There is one macro used in defining the MIPS coprocessor interface which
8431 you may want to override in subtargets; it is described below.
8433 @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES
8434 A comma-separated list (with leading comma) of pairs describing the
8435 alternate names of coprocessor registers. The format of each entry should be
8437 @{ @var{alternatename}, @var{register_number}@}
8443 @section Parameters for Precompiled Header Validity Checking
8444 @cindex parameters, precompiled headers
8446 @deftypefn {Target Hook} void * TARGET_GET_PCH_VALIDITY (size_t * @var{sz})
8447 Define this hook if your target needs to check a different collection
8448 of flags than the default, which is every flag defined by
8449 @code{TARGET_SWITCHES} and @code{TARGET_OPTIONS}. It should return
8450 some data which will be saved in the PCH file and presented to
8451 @code{TARGET_PCH_VALID_P} later; it should set @code{SZ} to the size
8455 @deftypefn {Target Hook} const char * TARGET_PCH_VALID_P (const void * @var{data}, size_t @var{sz})
8456 Define this hook if your target needs to check a different collection of
8457 flags than the default, which is every flag defined by @code{TARGET_SWITCHES}
8458 and @code{TARGET_OPTIONS}. It is given data which came from
8459 @code{TARGET_GET_PCH_VALIDITY} (in this version of this compiler, so there
8460 is no need for extensive validity checking). It returns @code{NULL} if
8461 it is safe to load a PCH file with this data, or a suitable error message
8462 if not. The error message will be presented to the user, so it should
8467 @section C++ ABI parameters
8468 @cindex parameters, c++ abi
8470 @deftypefn {Target Hook} tree TARGET_CXX_GUARD_TYPE (void)
8471 Define this hook to override the integer type used for guard variables.
8472 These are used to implement one-time construction of static objects. The
8473 default is long_long_integer_type_node.
8476 @deftypefn {Target Hook} bool TARGET_CXX_GUARD_MASK_BIT (void)
8477 This hook determines how guard variables are used. It should return
8478 @code{false} (the default) if first byte should be used. A return value of
8479 @code{true} indicates the least significant bit should be used.
8482 @deftypefn {Target Hook} tree TARGET_CXX_GET_COOKIE_SIZE (tree @var{type})
8483 This hook returns the size of the cookie to use when allocating an array
8484 whose elements have the indicated @var{type}. Assumes that it is already
8485 known that a cookie is needed. The default is
8486 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
8487 IA64/Generic C++ ABI.
8490 @deftypefn {Target Hook} bool TARGET_CXX_COOKIE_HAS_SIZE (void)
8491 This hook should return @code{true} if the element size should be stored in
8492 array cookies. The default is to return @code{false}.
8495 @deftypefn {Target Hook} int TARGET_CXX_IMPORT_EXPORT_CLASS (tree @var{type}, int @var{import_export})
8496 If defined by a backend this hook allows the decision made to export
8497 class @var{type} to be overruled. Upon entry @var{import_export}
8498 will contain 1 if the class is going to be exported, -1 if it is going
8499 to be imported and 0 otherwise. This function should return the
8500 modified value and perform any other actions necessary to support the
8501 backend's targeted operating system.
8504 @deftypefn {Target Hook} bool TARGET_CXX_CDTOR_RETURNS_THIS (void)
8505 This hook should return @code{true} if constructors and destructors return
8506 the address of the object created/destroyed. The default is to return
8511 @section Miscellaneous Parameters
8512 @cindex parameters, miscellaneous
8514 @c prevent bad page break with this line
8515 Here are several miscellaneous parameters.
8517 @defmac PREDICATE_CODES
8518 Define this if you have defined special-purpose predicates in the file
8519 @file{@var{machine}.c}. This macro is called within an initializer of an
8520 array of structures. The first field in the structure is the name of a
8521 predicate and the second field is an array of rtl codes. For each
8522 predicate, list all rtl codes that can be in expressions matched by the
8523 predicate. The list should have a trailing comma. Here is an example
8524 of two entries in the list for a typical RISC machine:
8527 #define PREDICATE_CODES \
8528 @{"gen_reg_rtx_operand", @{SUBREG, REG@}@}, \
8529 @{"reg_or_short_cint_operand", @{SUBREG, REG, CONST_INT@}@},
8532 Defining this macro does not affect the generated code (however,
8533 incorrect definitions that omit an rtl code that may be matched by the
8534 predicate can cause the compiler to malfunction). Instead, it allows
8535 the table built by @file{genrecog} to be more compact and efficient,
8536 thus speeding up the compiler. The most important predicates to include
8537 in the list specified by this macro are those used in the most insn
8540 For each predicate function named in @code{PREDICATE_CODES}, a
8541 declaration will be generated in @file{insn-codes.h}.
8544 @defmac HAS_LONG_COND_BRANCH
8545 Define this boolean macro to indicate whether or not your architecture
8546 has conditional branches that can span all of memory. It is used in
8547 conjunction with an optimization that partitions hot and cold basic
8548 blocks into separate sections of the executable. If this macro is
8549 set to false, gcc will convert any conditional branches that attempt
8550 to cross between sections into unconditional branches or indirect jumps.
8553 @defmac HAS_LONG_UNCOND_BRANCH
8554 Define this boolean macro to indicate whether or not your architecture
8555 has unconditional branches that can span all of memory. It is used in
8556 conjunction with an optimization that partitions hot and cold basic
8557 blocks into separate sections of the executable. If this macro is
8558 set to false, gcc will convert any unconditional branches that attempt
8559 to cross between sections into indirect jumps.
8562 @defmac SPECIAL_MODE_PREDICATES
8563 Define this if you have special predicates that know special things
8564 about modes. Genrecog will warn about certain forms of
8565 @code{match_operand} without a mode; if the operand predicate is
8566 listed in @code{SPECIAL_MODE_PREDICATES}, the warning will be
8569 Here is an example from the IA-32 port (@code{ext_register_operand}
8570 specially checks for @code{HImode} or @code{SImode} in preparation
8571 for a byte extraction from @code{%ah} etc.).
8574 #define SPECIAL_MODE_PREDICATES \
8575 "ext_register_operand",
8579 @defmac CASE_VECTOR_MODE
8580 An alias for a machine mode name. This is the machine mode that
8581 elements of a jump-table should have.
8584 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
8585 Optional: return the preferred mode for an @code{addr_diff_vec}
8586 when the minimum and maximum offset are known. If you define this,
8587 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
8588 To make this work, you also have to define @code{INSN_ALIGN} and
8589 make the alignment for @code{addr_diff_vec} explicit.
8590 The @var{body} argument is provided so that the offset_unsigned and scale
8591 flags can be updated.
8594 @defmac CASE_VECTOR_PC_RELATIVE
8595 Define this macro to be a C expression to indicate when jump-tables
8596 should contain relative addresses. You need not define this macro if
8597 jump-tables never contain relative addresses, or jump-tables should
8598 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
8602 @defmac CASE_DROPS_THROUGH
8603 Define this if control falls through a @code{case} insn when the index
8604 value is out of range. This means the specified default-label is
8605 actually ignored by the @code{case} insn proper.
8608 @defmac CASE_VALUES_THRESHOLD
8609 Define this to be the smallest number of different values for which it
8610 is best to use a jump-table instead of a tree of conditional branches.
8611 The default is four for machines with a @code{casesi} instruction and
8612 five otherwise. This is best for most machines.
8615 @defmac CASE_USE_BIT_TESTS
8616 Define this macro to be a C expression to indicate whether C switch
8617 statements may be implemented by a sequence of bit tests. This is
8618 advantageous on processors that can efficiently implement left shift
8619 of 1 by the number of bits held in a register, but inappropriate on
8620 targets that would require a loop. By default, this macro returns
8621 @code{true} if the target defines an @code{ashlsi3} pattern, and
8622 @code{false} otherwise.
8625 @defmac WORD_REGISTER_OPERATIONS
8626 Define this macro if operations between registers with integral mode
8627 smaller than a word are always performed on the entire register.
8628 Most RISC machines have this property and most CISC machines do not.
8631 @defmac LOAD_EXTEND_OP (@var{mem_mode})
8632 Define this macro to be a C expression indicating when insns that read
8633 memory in @var{mem_mode}, an integral mode narrower than a word, set the
8634 bits outside of @var{mem_mode} to be either the sign-extension or the
8635 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
8636 of @var{mem_mode} for which the
8637 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
8638 @code{NIL} for other modes.
8640 This macro is not called with @var{mem_mode} non-integral or with a width
8641 greater than or equal to @code{BITS_PER_WORD}, so you may return any
8642 value in this case. Do not define this macro if it would always return
8643 @code{NIL}. On machines where this macro is defined, you will normally
8644 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
8646 You may return a non-@code{NIL} value even if for some hard registers
8647 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
8648 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
8649 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
8650 integral mode larger than this but not larger than @code{word_mode}.
8652 You must return @code{NIL} if for some hard registers that allow this
8653 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
8654 @code{word_mode}, but that they can change to another integral mode that
8655 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
8658 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
8659 Define this macro if loading short immediate values into registers sign
8663 @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC
8664 Define this macro if the same instructions that convert a floating
8665 point number to a signed fixed point number also convert validly to an
8670 The maximum number of bytes that a single instruction can move quickly
8671 between memory and registers or between two memory locations.
8674 @defmac MAX_MOVE_MAX
8675 The maximum number of bytes that a single instruction can move quickly
8676 between memory and registers or between two memory locations. If this
8677 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
8678 constant value that is the largest value that @code{MOVE_MAX} can have
8682 @defmac SHIFT_COUNT_TRUNCATED
8683 A C expression that is nonzero if on this machine the number of bits
8684 actually used for the count of a shift operation is equal to the number
8685 of bits needed to represent the size of the object being shifted. When
8686 this macro is nonzero, the compiler will assume that it is safe to omit
8687 a sign-extend, zero-extend, and certain bitwise `and' instructions that
8688 truncates the count of a shift operation. On machines that have
8689 instructions that act on bit-fields at variable positions, which may
8690 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
8691 also enables deletion of truncations of the values that serve as
8692 arguments to bit-field instructions.
8694 If both types of instructions truncate the count (for shifts) and
8695 position (for bit-field operations), or if no variable-position bit-field
8696 instructions exist, you should define this macro.
8698 However, on some machines, such as the 80386 and the 680x0, truncation
8699 only applies to shift operations and not the (real or pretended)
8700 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
8701 such machines. Instead, add patterns to the @file{md} file that include
8702 the implied truncation of the shift instructions.
8704 You need not define this macro if it would always have the value of zero.
8707 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
8708 A C expression which is nonzero if on this machine it is safe to
8709 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
8710 bits (where @var{outprec} is smaller than @var{inprec}) by merely
8711 operating on it as if it had only @var{outprec} bits.
8713 On many machines, this expression can be 1.
8715 @c rearranged this, removed the phrase "it is reported that". this was
8716 @c to fix an overfull hbox. --mew 10feb93
8717 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
8718 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
8719 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
8720 such cases may improve things.
8723 @defmac STORE_FLAG_VALUE
8724 A C expression describing the value returned by a comparison operator
8725 with an integral mode and stored by a store-flag instruction
8726 (@samp{s@var{cond}}) when the condition is true. This description must
8727 apply to @emph{all} the @samp{s@var{cond}} patterns and all the
8728 comparison operators whose results have a @code{MODE_INT} mode.
8730 A value of 1 or @minus{}1 means that the instruction implementing the
8731 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
8732 and 0 when the comparison is false. Otherwise, the value indicates
8733 which bits of the result are guaranteed to be 1 when the comparison is
8734 true. This value is interpreted in the mode of the comparison
8735 operation, which is given by the mode of the first operand in the
8736 @samp{s@var{cond}} pattern. Either the low bit or the sign bit of
8737 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
8740 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
8741 generate code that depends only on the specified bits. It can also
8742 replace comparison operators with equivalent operations if they cause
8743 the required bits to be set, even if the remaining bits are undefined.
8744 For example, on a machine whose comparison operators return an
8745 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
8746 @samp{0x80000000}, saying that just the sign bit is relevant, the
8750 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
8757 (ashift:SI @var{x} (const_int @var{n}))
8761 where @var{n} is the appropriate shift count to move the bit being
8762 tested into the sign bit.
8764 There is no way to describe a machine that always sets the low-order bit
8765 for a true value, but does not guarantee the value of any other bits,
8766 but we do not know of any machine that has such an instruction. If you
8767 are trying to port GCC to such a machine, include an instruction to
8768 perform a logical-and of the result with 1 in the pattern for the
8769 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
8771 Often, a machine will have multiple instructions that obtain a value
8772 from a comparison (or the condition codes). Here are rules to guide the
8773 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
8778 Use the shortest sequence that yields a valid definition for
8779 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
8780 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
8781 comparison operators to do so because there may be opportunities to
8782 combine the normalization with other operations.
8785 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
8786 slightly preferred on machines with expensive jumps and 1 preferred on
8790 As a second choice, choose a value of @samp{0x80000001} if instructions
8791 exist that set both the sign and low-order bits but do not define the
8795 Otherwise, use a value of @samp{0x80000000}.
8798 Many machines can produce both the value chosen for
8799 @code{STORE_FLAG_VALUE} and its negation in the same number of
8800 instructions. On those machines, you should also define a pattern for
8801 those cases, e.g., one matching
8804 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
8807 Some machines can also perform @code{and} or @code{plus} operations on
8808 condition code values with less instructions than the corresponding
8809 @samp{s@var{cond}} insn followed by @code{and} or @code{plus}. On those
8810 machines, define the appropriate patterns. Use the names @code{incscc}
8811 and @code{decscc}, respectively, for the patterns which perform
8812 @code{plus} or @code{minus} operations on condition code values. See
8813 @file{rs6000.md} for some examples. The GNU Superoptizer can be used to
8814 find such instruction sequences on other machines.
8816 If this macro is not defined, the default value, 1, is used. You need
8817 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
8818 instructions, or if the value generated by these instructions is 1.
8821 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
8822 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
8823 returned when comparison operators with floating-point results are true.
8824 Define this macro on machine that have comparison operations that return
8825 floating-point values. If there are no such operations, do not define
8829 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
8830 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
8831 A C expression that evaluates to true if the architecture defines a value
8832 for @code{clz} or @code{ctz} with a zero operand. If so, @var{value}
8833 should be set to this value. If this macro is not defined, the value of
8834 @code{clz} or @code{ctz} is assumed to be undefined.
8836 This macro must be defined if the target's expansion for @code{ffs}
8837 relies on a particular value to get correct results. Otherwise it
8838 is not necessary, though it may be used to optimize some corner cases.
8840 Note that regardless of this macro the ``definedness'' of @code{clz}
8841 and @code{ctz} at zero do @emph{not} extend to the builtin functions
8842 visible to the user. Thus one may be free to adjust the value at will
8843 to match the target expansion of these operations without fear of
8848 An alias for the machine mode for pointers. On most machines, define
8849 this to be the integer mode corresponding to the width of a hardware
8850 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
8851 On some machines you must define this to be one of the partial integer
8852 modes, such as @code{PSImode}.
8854 The width of @code{Pmode} must be at least as large as the value of
8855 @code{POINTER_SIZE}. If it is not equal, you must define the macro
8856 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
8860 @defmac FUNCTION_MODE
8861 An alias for the machine mode used for memory references to functions
8862 being called, in @code{call} RTL expressions. On most machines this
8863 should be @code{QImode}.
8866 @defmac STDC_0_IN_SYSTEM_HEADERS
8867 In normal operation, the preprocessor expands @code{__STDC__} to the
8868 constant 1, to signify that GCC conforms to ISO Standard C@. On some
8869 hosts, like Solaris, the system compiler uses a different convention,
8870 where @code{__STDC__} is normally 0, but is 1 if the user specifies
8871 strict conformance to the C Standard.
8873 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
8874 convention when processing system header files, but when processing user
8875 files @code{__STDC__} will always expand to 1.
8878 @defmac NO_IMPLICIT_EXTERN_C
8879 Define this macro if the system header files support C++ as well as C@.
8880 This macro inhibits the usual method of using system header files in
8881 C++, which is to pretend that the file's contents are enclosed in
8882 @samp{extern "C" @{@dots{}@}}.
8887 @defmac REGISTER_TARGET_PRAGMAS ()
8888 Define this macro if you want to implement any target-specific pragmas.
8889 If defined, it is a C expression which makes a series of calls to
8890 @code{c_register_pragma} for each pragma. The macro may also do any
8891 setup required for the pragmas.
8893 The primary reason to define this macro is to provide compatibility with
8894 other compilers for the same target. In general, we discourage
8895 definition of target-specific pragmas for GCC@.
8897 If the pragma can be implemented by attributes then you should consider
8898 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
8900 Preprocessor macros that appear on pragma lines are not expanded. All
8901 @samp{#pragma} directives that do not match any registered pragma are
8902 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
8905 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
8907 Each call to @code{c_register_pragma} establishes one pragma. The
8908 @var{callback} routine will be called when the preprocessor encounters a
8912 #pragma [@var{space}] @var{name} @dots{}
8915 @var{space} is the case-sensitive namespace of the pragma, or
8916 @code{NULL} to put the pragma in the global namespace. The callback
8917 routine receives @var{pfile} as its first argument, which can be passed
8918 on to cpplib's functions if necessary. You can lex tokens after the
8919 @var{name} by calling @code{c_lex}. Tokens that are not read by the
8920 callback will be silently ignored. The end of the line is indicated by
8921 a token of type @code{CPP_EOF}
8923 For an example use of this routine, see @file{c4x.h} and the callback
8924 routines defined in @file{c4x-c.c}.
8926 Note that the use of @code{c_lex} is specific to the C and C++
8927 compilers. It will not work in the Java or Fortran compilers, or any
8928 other language compilers for that matter. Thus if @code{c_lex} is going
8929 to be called from target-specific code, it must only be done so when
8930 building the C and C++ compilers. This can be done by defining the
8931 variables @code{c_target_objs} and @code{cxx_target_objs} in the
8932 target entry in the @file{config.gcc} file. These variables should name
8933 the target-specific, language-specific object file which contains the
8934 code that uses @code{c_lex}. Note it will also be necessary to add a
8935 rule to the makefile fragment pointed to by @code{tmake_file} that shows
8936 how to build this object file.
8941 @defmac HANDLE_SYSV_PRAGMA
8942 Define this macro (to a value of 1) if you want the System V style
8943 pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name>
8944 [=<value>]} to be supported by gcc.
8946 The pack pragma specifies the maximum alignment (in bytes) of fields
8947 within a structure, in much the same way as the @samp{__aligned__} and
8948 @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets
8949 the behavior to the default.
8951 A subtlety for Microsoft Visual C/C++ style bit-field packing
8952 (e.g. -mms-bitfields) for targets that support it:
8953 When a bit-field is inserted into a packed record, the whole size
8954 of the underlying type is used by one or more same-size adjacent
8955 bit-fields (that is, if its long:3, 32 bits is used in the record,
8956 and any additional adjacent long bit-fields are packed into the same
8957 chunk of 32 bits. However, if the size changes, a new field of that
8960 If both MS bit-fields and @samp{__attribute__((packed))} are used,
8961 the latter will take precedence. If @samp{__attribute__((packed))} is
8962 used on a single field when MS bit-fields are in use, it will take
8963 precedence for that field, but the alignment of the rest of the structure
8964 may affect its placement.
8966 The weak pragma only works if @code{SUPPORTS_WEAK} and
8967 @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation
8968 of specifically named weak labels, optionally with a value.
8973 @defmac HANDLE_PRAGMA_PACK_PUSH_POP
8974 Define this macro (to a value of 1) if you want to support the Win32
8975 style pragmas @samp{#pragma pack(push,@var{n})} and @samp{#pragma
8976 pack(pop)}. The @samp{pack(push,@var{n})} pragma specifies the maximum alignment
8977 (in bytes) of fields within a structure, in much the same way as the
8978 @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A
8979 pack value of zero resets the behavior to the default. Successive
8980 invocations of this pragma cause the previous values to be stacked, so
8981 that invocations of @samp{#pragma pack(pop)} will return to the previous
8985 @defmac DOLLARS_IN_IDENTIFIERS
8986 Define this macro to control use of the character @samp{$} in
8987 identifier names for the C family of languages. 0 means @samp{$} is
8988 not allowed by default; 1 means it is allowed. 1 is the default;
8989 there is no need to define this macro in that case.
8992 @defmac NO_DOLLAR_IN_LABEL
8993 Define this macro if the assembler does not accept the character
8994 @samp{$} in label names. By default constructors and destructors in
8995 G++ have @samp{$} in the identifiers. If this macro is defined,
8996 @samp{.} is used instead.
8999 @defmac NO_DOT_IN_LABEL
9000 Define this macro if the assembler does not accept the character
9001 @samp{.} in label names. By default constructors and destructors in G++
9002 have names that use @samp{.}. If this macro is defined, these names
9003 are rewritten to avoid @samp{.}.
9006 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
9007 Define this macro as a C expression that is nonzero if it is safe for the
9008 delay slot scheduler to place instructions in the delay slot of @var{insn},
9009 even if they appear to use a resource set or clobbered in @var{insn}.
9010 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
9011 every @code{call_insn} has this behavior. On machines where some @code{insn}
9012 or @code{jump_insn} is really a function call and hence has this behavior,
9013 you should define this macro.
9015 You need not define this macro if it would always return zero.
9018 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
9019 Define this macro as a C expression that is nonzero if it is safe for the
9020 delay slot scheduler to place instructions in the delay slot of @var{insn},
9021 even if they appear to set or clobber a resource referenced in @var{insn}.
9022 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
9023 some @code{insn} or @code{jump_insn} is really a function call and its operands
9024 are registers whose use is actually in the subroutine it calls, you should
9025 define this macro. Doing so allows the delay slot scheduler to move
9026 instructions which copy arguments into the argument registers into the delay
9029 You need not define this macro if it would always return zero.
9032 @defmac MULTIPLE_SYMBOL_SPACES
9033 Define this macro if in some cases global symbols from one translation
9034 unit may not be bound to undefined symbols in another translation unit
9035 without user intervention. For instance, under Microsoft Windows
9036 symbols must be explicitly imported from shared libraries (DLLs).
9039 @deftypefn {Target Hook} tree TARGET_MD_ASM_CLOBBERS (tree @var{clobbers})
9040 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
9041 any hard regs the port wishes to automatically clobber for all asms.
9042 It should return the result of the last @code{tree_cons} used to add a
9046 @defmac MATH_LIBRARY
9047 Define this macro as a C string constant for the linker argument to link
9048 in the system math library, or @samp{""} if the target does not have a
9049 separate math library.
9051 You need only define this macro if the default of @samp{"-lm"} is wrong.
9054 @defmac LIBRARY_PATH_ENV
9055 Define this macro as a C string constant for the environment variable that
9056 specifies where the linker should look for libraries.
9058 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
9062 @defmac TARGET_HAS_F_SETLKW
9063 Define this macro if the target supports file locking with fcntl / F_SETLKW@.
9064 Note that this functionality is part of POSIX@.
9065 Defining @code{TARGET_HAS_F_SETLKW} will enable the test coverage code
9066 to use file locking when exiting a program, which avoids race conditions
9067 if the program has forked.
9070 @defmac MAX_CONDITIONAL_EXECUTE
9072 A C expression for the maximum number of instructions to execute via
9073 conditional execution instructions instead of a branch. A value of
9074 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
9075 1 if it does use cc0.
9078 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
9079 Used if the target needs to perform machine-dependent modifications on the
9080 conditionals used for turning basic blocks into conditionally executed code.
9081 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
9082 contains information about the currently processed blocks. @var{true_expr}
9083 and @var{false_expr} are the tests that are used for converting the
9084 then-block and the else-block, respectively. Set either @var{true_expr} or
9085 @var{false_expr} to a null pointer if the tests cannot be converted.
9088 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
9089 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
9090 if-statements into conditions combined by @code{and} and @code{or} operations.
9091 @var{bb} contains the basic block that contains the test that is currently
9092 being processed and about to be turned into a condition.
9095 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
9096 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
9097 be converted to conditional execution format. @var{ce_info} points to
9098 a data structure, @code{struct ce_if_block}, which contains information
9099 about the currently processed blocks.
9102 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
9103 A C expression to perform any final machine dependent modifications in
9104 converting code to conditional execution. The involved basic blocks
9105 can be found in the @code{struct ce_if_block} structure that is pointed
9106 to by @var{ce_info}.
9109 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
9110 A C expression to cancel any machine dependent modifications in
9111 converting code to conditional execution. The involved basic blocks
9112 can be found in the @code{struct ce_if_block} structure that is pointed
9113 to by @var{ce_info}.
9116 @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info})
9117 A C expression to initialize any extra fields in a @code{struct ce_if_block}
9118 structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro.
9121 @defmac IFCVT_EXTRA_FIELDS
9122 If defined, it should expand to a set of field declarations that will be
9123 added to the @code{struct ce_if_block} structure. These should be initialized
9124 by the @code{IFCVT_INIT_EXTRA_FIELDS} macro.
9127 @deftypefn {Target Hook} void TARGET_MACHINE_DEPENDENT_REORG ()
9128 If non-null, this hook performs a target-specific pass over the
9129 instruction stream. The compiler will run it at all optimization levels,
9130 just before the point at which it normally does delayed-branch scheduling.
9132 The exact purpose of the hook varies from target to target. Some use
9133 it to do transformations that are necessary for correctness, such as
9134 laying out in-function constant pools or avoiding hardware hazards.
9135 Others use it as an opportunity to do some machine-dependent optimizations.
9137 You need not implement the hook if it has nothing to do. The default
9141 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS ()
9142 Define this hook if you have any machine-specific built-in functions
9143 that need to be defined. It should be a function that performs the
9146 Machine specific built-in functions can be useful to expand special machine
9147 instructions that would otherwise not normally be generated because
9148 they have no equivalent in the source language (for example, SIMD vector
9149 instructions or prefetch instructions).
9151 To create a built-in function, call the function
9152 @code{lang_hooks.builtin_function}
9153 which is defined by the language front end. You can use any type nodes set
9154 up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2};
9155 only language front ends that use those two functions will call
9156 @samp{TARGET_INIT_BUILTINS}.
9159 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore})
9161 Expand a call to a machine specific built-in function that was set up by
9162 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
9163 function call; the result should go to @var{target} if that is
9164 convenient, and have mode @var{mode} if that is convenient.
9165 @var{subtarget} may be used as the target for computing one of
9166 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
9167 ignored. This function should return the result of the call to the
9171 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
9173 Take a branch insn in @var{branch1} and another in @var{branch2}.
9174 Return true if redirecting @var{branch1} to the destination of
9175 @var{branch2} is possible.
9177 On some targets, branches may have a limited range. Optimizing the
9178 filling of delay slots can result in branches being redirected, and this
9179 may in turn cause a branch offset to overflow.
9182 @defmac ALLOCATE_INITIAL_VALUE (@var{hard_reg})
9184 When the initial value of a hard register has been copied in a pseudo
9185 register, it is often not necessary to actually allocate another register
9186 to this pseudo register, because the original hard register or a stack slot
9187 it has been saved into can be used. @code{ALLOCATE_INITIAL_VALUE}, if
9188 defined, is called at the start of register allocation once for each
9189 hard register that had its initial value copied by using
9190 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
9191 Possible values are @code{NULL_RTX}, if you don't want
9192 to do any special allocation, a @code{REG} rtx---that would typically be
9193 the hard register itself, if it is known not to be clobbered---or a
9195 If you are returning a @code{MEM}, this is only a hint for the allocator;
9196 it might decide to use another register anyways.
9197 You may use @code{current_function_leaf_function} in the definition of the
9198 macro, functions that use @code{REG_N_SETS}, to determine if the hard
9199 register in question will not be clobbered.
9202 @defmac TARGET_OBJECT_SUFFIX
9203 Define this macro to be a C string representing the suffix for object
9204 files on your target machine. If you do not define this macro, GCC will
9205 use @samp{.o} as the suffix for object files.
9208 @defmac TARGET_EXECUTABLE_SUFFIX
9209 Define this macro to be a C string representing the suffix to be
9210 automatically added to executable files on your target machine. If you
9211 do not define this macro, GCC will use the null string as the suffix for
9215 @defmac COLLECT_EXPORT_LIST
9216 If defined, @code{collect2} will scan the individual object files
9217 specified on its command line and create an export list for the linker.
9218 Define this macro for systems like AIX, where the linker discards
9219 object files that are not referenced from @code{main} and uses export
9223 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
9224 Define this macro to a C expression representing a variant of the
9225 method call @var{mdecl}, if Java Native Interface (JNI) methods
9226 must be invoked differently from other methods on your target.
9227 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
9228 the @code{stdcall} calling convention and this macro is then
9229 defined as this expression:
9232 build_type_attribute_variant (@var{mdecl},
9234 (get_identifier ("stdcall"),
9239 @deftypefn {Target Hook} bool TARGET_CANNOT_MODIFY_JUMPS_P (void)
9240 This target hook returns @code{true} past the point in which new jump
9241 instructions could be created. On machines that require a register for
9242 every jump such as the SHmedia ISA of SH5, this point would typically be
9243 reload, so this target hook should be defined to a function such as:
9247 cannot_modify_jumps_past_reload_p ()
9249 return (reload_completed || reload_in_progress);
9254 @deftypefn {Target Hook} int TARGET_BRANCH_TARGET_REGISTER_CLASS (void)
9255 This target hook returns a register class for which branch target register
9256 optimizations should be applied. All registers in this class should be
9257 usable interchangeably. After reload, registers in this class will be
9258 re-allocated and loads will be hoisted out of loops and be subjected
9259 to inter-block scheduling.
9262 @deftypefn {Target Hook} bool TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED (bool @var{after_prologue_epilogue_gen})
9263 Branch target register optimization will by default exclude callee-saved
9265 that are not already live during the current function; if this target hook
9266 returns true, they will be included. The target code must than make sure
9267 that all target registers in the class returned by
9268 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
9269 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
9270 epilogues have already been generated. Note, even if you only return
9271 true when @var{after_prologue_epilogue_gen} is false, you still are likely
9272 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
9273 to reserve space for caller-saved target registers.
9276 @defmac POWI_MAX_MULTS
9277 If defined, this macro is interpreted as a signed integer C expression
9278 that specifies the maximum number of floating point multiplications
9279 that should be emitted when expanding exponentiation by an integer
9280 constant inline. When this value is defined, exponentiation requiring
9281 more than this number of multiplications is implemented by calling the
9282 system library's @code{pow}, @code{powf} or @code{powl} routines.
9283 The default value places no upper bound on the multiplication count.
9286 @deftypefn Macro void TARGET_EXTRA_INCLUDES (int @var{stdinc})
9287 This target hook should register any extra include files for the
9288 target. The parameter @var{stdinc} indicates if normal include files
9292 @deftypefn Macro void TARGET_OPTF (char *@var{path})
9293 This target hook should register special include paths for the target.
9294 The parameter @var{path} is the include to register. On Darwin
9295 systems, this is used for Framework includes, which have semantics
9296 that are different from @option{-I}.
9299 @deftypefn {Target Hook} bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
9300 This target hook returns @code{true} if it is safe to use a local alias
9301 for a virtual function @var{fndecl} when constructing thunks,
9302 @code{false} otherwise. By default, the hook returns @code{true} for all
9303 functions, if a target supports aliases (ie. defines
9304 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
9307 @defmac TARGET_FORMAT_TYPES
9308 If defined, this macro is the name of a global variable containing
9309 target-specific format checking information for the @option{-Wformat}
9310 option. The default is to have no target-specific format checks.
9313 @defmac TARGET_N_FORMAT_TYPES
9314 If defined, this macro is the number of entries in
9315 @code{TARGET_FORMAT_TYPES}.